Cell spread area and traction forces determine myosin-II-based cortex thickness regulation

Abstract

Actomyosin network under the plasma membrane of cells forms a cortical layer that regulates cellular deformations during different processes. What regulates the cortex? Characterized by its thickness, it is believed to be regulated by actin dynamics, filament-length regulators and myosin motor proteins. However, its regulation by cellular morphology (e.g. cell spread area) or mechanical microenvironment (e.g. substrate stiffness) has remained largely unexplored. In this study, super- and high-resolution imaging of actin in CHO cells demonstrates that at high spread areas (> 450 μm²), the cortex is thinner, better separated as layers, and sensitive to deactivation of myosin II motors or reduction of substrate stiffness (and traction forces). In less spread cells (< 400 μm²) such perturbations do not elicit a response. Myosin IIA's mechanosensing is limited here due to its lowered actin-bound fraction and higher turnover rate. Cofilin, in line with its competitive inhibitory role, is found to be overexpressed in these cells. To establish the causal relation, we initiate a spread area drop by de-adhesion and find enhanced actin dynamics and fragmentation along with oscillations and increase in thickness. This is more correlated to the reduction of traction forces than the endocytosis-based reduction in cell volume. Cortex thickness control by spread area is also found be true during differentiation of THP-1 monocytes to macrophages. Thus, we propose that spread area regulates cortex and its thickness by traction-based mechanosensing of myosin II.

1. Introduction

Actin filaments under the plasma membrane, anchored to the it through Ezrin, Radixin, and Moesin proteins [1], align parallel to the membrane - although disorderedly. They get crosslinked by myosin II and actin-binding proteins [2,3] and form a sub-membranous layer separate from the cytoplasmic actin. In the absence of any other specialized structures in this layer, it is termed as the actin cortex. The thickness of this layer has been reported to be ~100 nm by techniques like–electron microscopy [4,5], and sub-resolution image analysis [4] and is proposed to be established by the polymerization and depolymerization of actin [7,8]. Together with the turnover rates of actin, myosin II, and cross-linkers like α-actinin and formins [9], cortex remodelling timescale is expected to be ~1 min. Its role in mechanical protection of cells [1,10] and the fact that loss of its lateral homogeneity can cause cell shape oscillations [11] or its own fragmentation [12] points to the need of its tight regulation. Currently, however, thickness regulation of the cortex is best studied in geometrically round cells [6,13] so as to avoid structures like stress-fibres. Such studies have revealed regulation by proteins like actin length-regulators and myosin II motor activity [13]. However, studying cells with varied spread areas would allow investigation of the role of a wider range of cellular architectural signals (cell shape, spread area, geometry) in shaping the cortex.

Just as mechanical properties of cells depend on its spread area, mechanical properties of sub-membranous actin layer (at the apical surface) are also reported to depend on the state of adhesion of the cell [14]. Stress fibres are more prominent at the basal plane in these cells [15] and traction forces higher [16] than cells with low spread area. The actomyosin network at the basal plane is expected to feel enhanced lateral forces due to which myosin II binding to actin may also be favoured [17]. How does the spread-area-mediated mechanical cues originating at the basal plane affect the actin-layer thickness and density-profile at higher planes?

This central question is addressed herein by separately evaluating the effect of spread area, traction forces/substrate stiffness and myosin II motor activity on cortex thickness in interphase CHO cells – micropatterned, de-adhered as well as non-patterned. Super-resolution microscopy as well as high-resolution techniques (confocal, high-NA epifluorescence) are employed at planes far from basal stress fibres for cortex thickness quantification. Using TIRF (Total Internal Reflection Fluorescence) microscopy we address the concomitant alteration in the dynamics of the basal actin network. FRAP (Fluorescence Recovery After Photobleaching) allows monitoring molecular level dynamics of motor proteins at the cortex while FACS (Flow Assisted Cell Scan) allows a population level evaluation of changes in levels of actin-severing protein Cofilin. Together, we reveal the key players and propose the mechanism that cause the differential regulation of cortex thickness by cell spread area.

2. Results

2.1. Micropatterning enables measurement of thickness of cortical actin layer in high spread area cells

Measurement of cortex thickness requires imaging actin at cell edges free of stress fibres [18]. Such cortical regions are clearly visible at the equatorial plane of rounded cells – a geometry favoured at low spread areas. Well spread cells, however, with their usual tent-like morphology, have such regions very close to the plane rich with stress fibres. To obtain high spread areas as well as clear cortical regions we grow CHO cells on two rectangular (R1005 and R3015) and one line-shaped micropattern (Line10) (Fig. 1 A) resulting in cell populations of three different values of average cell spread area (Fig. 1 B). High spread area is obtained by Line10 pattern which allows cells to spread to desired area while maintaining the cell width and yielding clear cortical regions devoid of stress fibres at planes ~2 μm from the basal plane (Fig. S1). The choice of the plane is made to minimize overestimation due to vertical bending of the membrane as explored by confocal microscopy (Fig. S1 A-E). Super resolution imaging (using Stimulated Emission Depletion (STED) microscopy (Fig. S2)) is thus performed at higher planes (~2.4 μm for Line10 and R3015; ~5 μm for R1005) for all three patterns (Fig. 1 C). The thickness estimation of the sub-membranous actin is then performed by analysing straightened-out cortex sections (Fig. 1 D, E). Normal line scans (white line, Fig. 1 E) of labelled actin are fit with multi-peaked Gaussian functions (Fig. 1 E right). The central Gaussian captures the profile of the cortex and the others help in accounting for contribution from other accumulations/layers near or far from the cortex (Fig. S2). This is followed by calculation of full width at half maxima (FWHM) of the central Gaussian, performed as described with signal-to-noise (S/N) ratio maintained > 10 so that thickness calculation is not affected (Fig. S3). In Fig. S4 we demonstrate that purely cytoplasmic pool of fluorescent proteins (EGFP in Fig. S4) do not artifactually present a cortex-like fluorescence profile at the cell edge of either well spread or rounded cells. Therefore, peaked profiles of fluorescent actin must really represent the cortex.

Fig. 1.

Super-resolution measurement of cortical actin thickness on altering spread area. CHO cells grown on large (Line10, 10 μm width line), medium (R3015, 30 × 15 μm rectangle) and small (R1005, 10 × 5 μm rectangle) micropatterns, fixed, stained with Phalloidin Alexa Fluor 568 and imaged. (A) Confocal images of basal actin layer of cells gown on Line10, R3015 and R1005 micropattern (top) (image of photomasks used in micropatterning shown in inset). (B) Spread area of CHO cells measured from basal plane image (bottom), (n = 15 cells in each condition). Scale, 10 and 3 μm. (C) Representative STED images of micropatterned CHO cells imaged at indicated distances above the basal plane. Scale bar, 10 μm. (D) Linearized and zoomed-in view of the cortex (at marked with yellow ROI in (top)). (E) A typical cortex showing distinct layer parallel to cortex. Scale bar, 2 μm. Intensity profile of the linearized cortex (at white line) with its fit to a 3-term Gaussian function (F) peak-to-peak distance (separation) from gaussian fits (G) Thickness measured from STED imaging of cells on Line10, R3015 and R1005 patterns. Each data of scatter box plot shows average thickness of cell measured from at-least 10 ROIs per cell and approximately 250 line-scan (2 μm) per ROI. (n = 15, 12, 15 cells respectively). (H) Cartoon depicting other layers (light green) in the vicinity of cortex (dark green) of cells on Line10 expand, merge and come closer in cells on R1005. Dashed lines are guides to the eye for centres of the layers. ** p < 0.001, Wilcoxon based Mann-Whitney U test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

2.2. Cell spread area affects cortex profile and thickness

Imaging cells on different patterns with STED microscopy (Fig. 1 C-E), we first report the cross-sectional profile of the cortex to be significantly heterogenous as evident from images of distinct parallel layers (Fig. 1 E) and also from the fact that multi-peaked Gaussians are needed for fitting the cortical profile. On comparing the distance between the first and second peak of Gaussian fits, we find that in Line 10 cells, the average distance between the layers in ~117 nm which reduces to ~63 nm for cells with reduced spread area (R1005) (Fig. 1 F left). The same is true for the distance between the first and third peaks (Fig. 1 F right). Using FWHM of first Gaussian as a measure of the thickness of the primary cortex (green arrow, Fig. 1 E) we find that in Line10 cells thickness is ~197 ± 17 nm which increases in cells with reduced spread areas (221 ± 19 nm (R3015) and 265 ± 13 nm (R1005)) (Fig. 1 G). Taken together, we believe the cortex is transversely heterogenous with overlapping structures that expand and further merge on spread area reduction – thereby increasing cortex thickness and reducing first-peak to second-peak distance (Fig. 1 H). Note that with a PSF of ~80 nm, STED allows a close mapping of cross-sectional cortex profiles and enables estimation of the thickness matching closely numbers reported by other groups [6].

Next, we use confocal microscopy, with a working PSF of ~370 nm, and find that the thickening due to reduction of spread area could be significantly captured (Fig. 2 A-C). This is further validated by using thickness measurements at all planes and comparing the best imaged (minimum thickness) planes between representative cells on Line10 and R1005 patterns (Fig. S1 D - E). We find that thickness increases from ~434 ± 43 nm (Line 10) to ~487 ± 23 nm (R3015) and ~591 ± 39 nm (R1005) (Table S2). On analysing non-patterned well-spread cells (Fig. 2 D), the criteria of distance from basal membrane was lowered owing to the higher spread area. Thickening of actin-layer on spread area reduction was still true (Fig. 2 D-E) comparing well-spread cells to patterned ones. We also study patterned (R3015 and R1005) cells with two spreading times (1 h and 24 h) showing increased spread area with time for each pattern. In parallel, the thickness decreases (Fig. 2 D-E). Including the spreading/plating times thus highlights that remodelling occurs even beyond the first hour. Note that surface area and volume of cells are also reduced on reducing spread area (Fig. 2 E – middle and right). The relative trends in thickness were found even in high-NA epifluorescence imaging (Fig. S5).

Fig. 2.

Thickening of cortex and sub-membranous actin layer by high-resolution imaging. (A) Confocal xyz slice view of Phalloidin Alexa Fluor 568 stained CHO cells grown on Line10 and (B) R1005 pattern (captured at 2.8 ± 0.3 μm and 6.9 ± 1 μm above the basal plane). (C) Box plot of thickness and spread area measured from confocal imaging shown in (B). (n > 12 cells). Scale bars, 10 μm. (D) Similar confocal xyz slice view of actin stained CHO cells on Control, Line10, R3015, R1005 (plated for 24 h), R1005_1hr R3015_1hr (plated for 1 h). Distance from basal planes during imaging mentioned in the insets. Scale bars, 10 μm. (E) Box plot showing thickness, spread area and volume of cell grown on Control, Line_10, R3015_1hr, R3015_24hr, R1005_24hr and R1005_1hr. (12, 10, 12, 8, 12, 11 and 12 cells respectively). ** p < 0.001, One-way ANOVA using Bonferroni post hoc method.

In order to understand why thickness as well as thickening is overestimated by confocal microscopy (in contrast to STED), we go back to comparisons involving distance between peaks of multi-peak Gaussian fits (Fig. S2). We propose that in Line10 cells the distance between layers (~160 ± 110 nm) is partly resolvable by confocal microscopy but in R1005 cells, this distance further reduces and becomes more unresolvable by fits to confocal images. Thus, R1005 confocal profile merge the first two layers and hence estimate a thicker cortex than reality. This is first depicted through the STED and confocal profile (measured on the same microscope) of the same region showing two peaks in STED but only one in confocal (Fig. S2 A). We next use data acquired in separate microscopes as presented in Figs. 1 and 2 and tabulate the peak-to-peak distances. While for Line10 cells distance between peak-1 and peak-2 measured in confocal is similar to as measured by STED ~160 nm, it is not so for R1005 (Fig. S2 B). In fact, for these cells, the distance between peak-1 and peak-2 from confocal data is same as the distance between peak-1 and peak-3 from STED data. Hence confocal cannot demarcate the first two layers and identifies the third one observed in STED as the second layer. We next assume the cortex to actually have a step-like actin density. We convolute the imaging PSFs with this profile and obtain the expected FWHM of these profiles as the width of the step function (thickness of cortex) is varied (Fig. S2 C). The overestimation of thickness and thickening is well depicted.

While the cortex thus can be cross-sectionally heterogeneous, laterally too, thickness is expected to have variations. Is it tightly regulated and what role does actomyosin dynamics have? Thus, we next test the effect of perturbing actin/myosin in well-spread cells.

2.3. Perturbations to actin and myosin II activity thicken cortex in high spread area cells

High-NA widefield imaging of fixed cells is used to test the effect of enhanced actin depolymerization by cytochalasin D (Cyto D), latrunculin B (Lat B), effect of ATP depletion and the effect of myosin II motor activity blocker [19], blebbistatin (Blebb). Cells (Fig. S6 A, B) display visible heterogeneities in the sub-membranous actin layer on these drug-treatments consistent with results proposing myosin II motor activity contracting the weakened cortex into clumps [20]. Actin intensity profile spreads out, the peak decreases and cortex thickness increases (Fig. S6 C, D) on Cyto D and Lat B treatments. The average FWHM of actin profile increases on reducing motor activity of myosin II (Fig. S6 B) and so does the lateral variability (Fig. S6 E). This is consistent with myosin II's role in creating lateral coherence in the cortical structure [21]. We also follow single cells expressing LifeAct-mCherry after administering Cyto D (10 μM) (Fig. S7A), Supplementary Movies S1, S2). Rupture in the lateral network of the sub-membranous actin was observed after the cell edge straightens (Fig. S7 B -D). Thickness fluctuates after addition of Cyto D and reduces (Fig. S7 E) before rupturing. Post rupture, the edge relaxes back, actin network flows away at 0.5–1.5 μm/min (Fig. S7 C) and shows enhanced final thickness (Fig. S7 E). Therefore, CytoD transiently decreases layer thickness but finally the network reaches a less intense and more spread out state. However, on Blebb or ATP depletion, the network spreads out without becoming less intense (Fig. S6 C).

To test the effect of reducing myosin II activity on the cortical layer, we revert to STED microscopy (Fig. 3 A, B) and also ask if this is true for an already thickened cortex in cells with reduced spread area. At super-resolution, in patterned cells, we confirm that for cells with high spread area, cortex thickness increases on Blebb treatment (Fig. 3 C-D). But cells with low spread area do not show this remodelling significantly when the cell-averaged thickness values are compared (Fig. 3 C-D). This observation suggests that spread area not only affects thickness but also its regulation.

Fig. 3.

Spread area alters effect of myosin II on cortex, its turnover and the expression of cofilin. (A) Representative STED image of a control and blebbestatin treated CHO cell grown on Line10 and R1005 patterned stained with Phalloidin Alexa Fluor 568. Scale bar, 10 μm. (B) Linearized and zoomed-in view of the cortex (at marked with yellow ROI in (A)) and its color-code. Scale bar, 2 μm. (C) Box plots of spread area of basal plane calculated form control and 50 μM Blebb treated cell grown on Line10, R3015 and R1005 pattern. (n = 100, 115, 250, 292, 76, 214 cells from condition in plot left to right). (D) Cortex thickness calculated form control and Blebb treated cell grown on Line10, R3015 and R1005 pattern. (n = 15, 12, 9, 13, 15, 11 cells from condition in plot left to right). (E) Myosin IIA transfected CHO cells on Line10 and R1005 pattern captured at 2.1 and 5.4 μm above the basal plane. Scale 10 μm. (F) Top: Confocal image sequence of 8 × 8 μm region (as marked in dotted ROI in (E)) followed through pre-bleach and post-bleach time points. Circular red ROIs (radius = 0.5 μm) were chosen to bleach and obtained fluorescence recovery curve. Scale bar, 2 μm. (F, top left) Bottom left: Fluorescence recovery intensity (Rec; from bleach region marked as red in (F, top) and photobleaching intensity (Bleach; captured from 3 μm rectangle away from recovery) as function of time from individual cell and normalized by setting prebleach value to 1 for Line10 (averaged over 41 cells) and R1005 (averaged 31 cells) conditions. Error bars denote SDs. Right: For fitting, besides background subtraction and prebleach normalization, the postbleach intensity is also normalized to 0. Normalized data is fit to two-term exponential (F(t) = F₁(1 − exp(−t / τ₁)) + F₂(1 − exp (−t / τ₂)) to separate fluorescence recovery due to protein turnover from much slower timescales due to other response. Box plot of fluorescence recovery time (τ₁) for Line10 and R1005 condition are 6.8 ± 2.1 s and 3.6 ± 1.2 s respectively (mean ± SD; N = 41 and 31 cells from two independent experiments) while τ₂ is 62 ± 13 s and 22 ± 5 s respectively. (G) Typical cofilin (cof) and phospho-cofilin (p-cof) levels in cells grown for 24 h and 5 h conditions measured by flow cytometry. Figure labels 24 h and 5 h represent fluorescence intensity of protein of interest detected by primary and compatible secondary antibody; Sec. AB_24hr and Sec. AB_5hr labels represent signal obtained from samples incubated with secondary antibody only. (H) Fold increase in protein expression of cells on 5 h w.r.t to 24 h condition. Data represents mean ± SD from three independent experiment. ** p < 0.001 and N.S. denotes non-significant change. Test used: Wilcoxon based Mann-Whitney U test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

To understand this further, we next check myosin IIA's bound fraction and turnover-rate using FRAP at the cortex of cells expressing mCherry-Myosin IIA (Fig. 3 E-F). We find that in R1005-patterned cells the mobile fraction of myosin IIA is higher than in Line10-patterned cells (Fig. 3 F, lower-left) and display faster turnover of myosin IIA (Fig. 3 F, lower-right). Note that the diffusive timescales are much smaller (expected to be ~0.1 s) and the timescales probed here represent slower process of binding and unbinding of myosin IIA with actin filaments [9].

Cofilin is believed to lead to competitive antagonist inactivation of myosin II [22] and also implicated in mechanosensing [23]. Using flow cytometry, we find cofilin levels to be 2-fold higher in cells in low spread area (cells plated on glass for 5 h) than ones with higher spread area (cells plated on glass for 24 h) (Fig. 3 G). The spread area difference between the two conditions are confirmed by microscopy (Fig. S8 A-B). The change in cofilin and phospho-cofilin (phosphorylated cofilin) levels is checked to also show up in micropatterned cells (R1005, Line10) by microscopy and flow cytometry (Fig. S8 E-G). Note the folds change of cofilin is more than that of phospho-cof implying an average increase in cofilin activity in cells with low spread area.

Myosin II is postulated to have load-dependent turnover [17] while cofilin too is believed to be activated by enhanced stress in the actin network [24]. Therefore, we next check if traction forces have any role in this control of cortex thickness by spread area.

2.4. Increasing traction forces reduces thickness

Spread area reduction has been reported to reduce maximum traction stress [25] and strain energy [26]. Is it through altered traction forces that spread area affects cortex thickness? To address this, we first show that, as reported earlier, reducing spread area reduces traction forces (Fig. S9 A - C). Next, we create conditions where spread area is held constant, but substrate stiffness is altered resulting in altered traction forces.

We create micropatterns on PAG (poly acrylamide gel), plate cells and measure traction stresses and cortex thickness separately (Fig. 4 A). Spread area, thus controlled, did not differ between substrates of varying stiffness (Fig. 4 B). An increase in traction stresses is measured for Line10 with increasing substrate stiffness but for cells on R1005 this is not significantly altered (Fig. 4 C, D). Interestingly, for cells of Line10, we thus could recreate a condition in which spread area is maintained but cortex thickness significantly varies. This occurs on increasing gel stiffness and only for cells on Line10 (Fig. 4 E). Cells on R1005 neither had altered traction forces nor did their cortex thickness change (Table S3). This data strongly suggests basal traction forces affect cortex thickness at higher planes.

Fig. 4.

Thickening of cortex on reducing traction force. CHO cells with labelled (Phalloidin Alexa Flour 568) actin on patterned polyacrylamide gel of varying stiffness. (A) Actin cross section, displacement and traction field vector and traction stress magnitude of cells spread for 24 h. Dotted line in traction stress images indicate cell periphery. Scale bars, 10 μm. Box plots of (B) Spread area, (C) Average traction stress (D) Maximum traction stress and (E) Thickness of sub-membranous actin, plotted for cells on two different patterns and three different substrate stiffness as indicated. (25 < N < 60, see Table S2). Error bars represent SD. * p < 0.05, ** p < 0.001, Wilcoxon based Mann-Whitney U test were performed for all statistical test except thickness (one-way ANOVA using Bonferroni post hoc method). N.S. denotes non-significant change.

To directly assay this in same cells undergoing change in traction forces, we use CHO cells expressing LifeAct-mCherry and image them before and after 60 min of EDTA-mediated de-adhesion - measuring both - thickness of sub-membranous actin layer and traction stress. Although cells are still partially adhered after 60 min of EDTA treatment (Fig. 5 A), the traction stress drops substantially accompanied by thickening of the cortex (Fig. 5 B, C). We find that combining well-spread and EDTA treated cells, thickness is strongly correlated to traction stress (r = −0.96, p ≪ 0.001; σ = −0.87, p ≪ 0.001) (Fig. 5 B). Importantly, even within well-spread cells, a strong correlation of thickness to traction stress exists (r = −0.91, p = 0.03; σ = −0.9, p = 0.04). Our previous results are thus confirmed and a reduction in traction forces by de-adhesion also results in increased thickness of the actin-layer.

Fig. 5.

De-adhesion of cell reduces traction stress and cell volume and thickens the actin layer. Thickness and traction stress of CHO cells expressing LifeAct-mCherry were measured on uniformly fibronectin coated polyacrylamide gel. (A) Traction field vector and traction stress magnitude of a cell spread for 24 h (left) and then after 1 h in EDTA solution (right). Dotted line in traction stress images cell periphery. Zoomed views are shown in inset (at ROI). Scale bars, 10 μm. (B) Scatter plot showing correlation of average traction stress with thickness (r = −0.96, p ≪ 0.001; σ = −0.87, p ≪ 0.001) (n = 5 cells for control and 5 for EDTA). (C) Percentage change in traction stress and thickness upon de-adhesion. (D) Spread area, volume and cortex thickness of same LifeAct-mCherry transfected cells before (Control) and after EDTA treatment (n = 32 cells (Control), n = 32 cells (EDTA) from three independent experiments). (E) Scatter plot showing correlation of thickness vs. spread area (r = −0.61, p ≪ 0.001; σ = 0.7, p ≪ 0.001; left) and volume (r = −0.76, p ≪ 0.001; σ = 0.8, p ≪ 0.001; middle) and normalized distributions of thickness for both conditions for cells represented in (D). (F) Left: Actin and nucleus (DAPI) labelled confocal image of CHO cell in interphase, mitotic and 1 h EDTA condition; Right: Scatter box plot of thickness of interphase, mitotic and 1 h EDTA cells (n = 30 cells for each condition and 10 ROIs per cells and 31 line-scans per ROIs, measured from three independent experiments). Error bars represent SD. * p < 0.05, ** p < 0.001, Wilcoxon based Mann-Whitney U test were performed for all statistical test except thickness (one-way ANOVA using Bonferroni post hoc method).

Independently, also measuring in same cells before and after EDTA treatment, we confirm that spread area and volume are reduced while thickness of sub-membranous actin-layer increases (Fig. 5 D) and becomes more variable (Fig. 5 E). Trypsinization yields similar response (Fig. S10 A-B) while measuring FWHM of intensity profiles of cross-linkers α-Actinin and Filamin A also shows similar thickening on de-adhesion (Fig. S10 C–F). We also confirm that cell spreading – expected to enhance traction stress with time – results in thinning of the sub-membranous actin layer (Fig. S11).

We next check if volume and spread area are similarly correlated to thickness. We find that fewer cells have overlapping spread area (dashed line, Fig. 5 E) than volume indicating a higher dependence of thickness on spread area. Thickness is also less correlated to volume (r = −0.61, p ≪ 0.001; σ = 0.7, p ≪ 0.001) than spread area (r = −0.76, p ≪ 0.001; σ = 0.8, p ≪ 0.001) as calculated from Fig. 5 E – and in both cases weaker than the correlation observed with traction stress. As a control, we show that we can measure cortex thickness to be lower for mitotic cells than rounded cells in line with earlier reports [13]. Comparisons (Fig. 5 F) also show that interphase cells with high spread area have a thinner actin-layer than mitotic cells.

Next, we confirm that the volume reduction on de-adhesion is endocytosis mediated (Fig. 6 A, B), as expected [27]. On blocking dynamin-dependent endocytosis by Dynasore, EDTA treatment reduces spread area (Fig. 6 C) but volume reduction is not significant anymore (Fig. 6 D). However, thickening of the actin layer by de-adhesion is not prevented (Fig. 6 E). The estimated cortical volume (product of surface area and thickness of actin layer) reduces significantly in control cells but on Dynasore treated cells, the cortical volume remains same (Table S1). This shows that volume change can remove cortical constituents – but the thickness is determined by traction forces.

Fig. 6.

Transient endocytosis-mediated volume reduction not essential for thickening of actin-layer on de-adhesion. (A) Slice view of FM 1-43 FX imaging of CHO cells grown on Line10, EDTA treatment of Line10 cells and cells grown on R1005 micropattern. Scale bar, 10 μm. (B) Probability distribution of per cell FM 1-43 FX intensity for the three cells (left) and their mean and SD (right) (n = 73, 54 and 40 cells for Line10, EDTA and R1005 respectively). (C) Normalized histogram of endocytosed FM intensity of control and EDTA treatment condition. (D) Spread area, (E) volume and (F) Actin-layer thickness of CHO cells expressing LifeAct-m-Cherry in indicated conditions. (n = 12 cells (Control (1), EDTA), and 08 cells for (Control (2), Dyna and Dyna + EDTA)). Error bars represent SD. * p < 0.05, ** p < 0.001, Wilcoxon based Mann-Whitney U test were performed for all statistical test except thickness (one-way ANOVA using Bonferroni post hoc method.

Since traction forces, therefore, emerge as most correlated with thickness and can influence thickness even when spread area is maintained, we next study the basal plane where through stress fibres, traction forces are known to build up. Using live assays we initiate directly a change in spread and follow the fate of the basal actin network in addition to studying the timescales of cortex thickening at cell edges.

2.5. Thickness fluctuates while basal cortex gets fragmented and dynamic on de-adhesion

Single cells are taken through EDTA-mediated de-adhesion and the actin at cell edge visualized by transiently expressing LifeAct-mCherry in these cells. We observe that the spread area, volume and surface area reduce exponentially (Fig. 7 A) while thickness increases (Fig. 7 A) but oscillates peaking at ~10–20 min and 30–40 min - the peaks identified by multi-term Gaussian fits.

Fig. 7.

Time evolution of changes in actomyosin network on de-adhesion. (A) Spread area, thickness and volume of LifeAct-mCherry expressing CHO cells were measured from z stacks captured during EDTA mediated de-adhesion. Solid lines represent exponential fits to spread area and volume or multi-term Gaussian fit to thickness. Average time constants obtained from exponential fits of spread area, volume, surface area evolution and peaking times obtained from Gaussian fits to thickness evolution (n = 14 cells). Error bars represent SD. * p < 0.05, ** p < 0.001, Wilcoxon based Mann-Whitney U test were performed for all statistical test except thickness (one-way ANOVA using Bonferroni post hoc method). (B) TIRF image of basal cortex of a live CHO cell transiently expressing LifeAct-mCherry. Scale bars, 10 μm. (C) Evolution of a representative region lacking stress fibres. (D) Evolution of a representative region enriched with stress fibres showing buckled (marked with yellow arrow) and fragmented of cortical fibres upon de-adhesion. Scale bars (C-D), 2 μm. Numbers indicated in (C-D) represent time elapsed in min after administration of EDTA. (E) Top: Evolution of total intensity, coherence and orientation parameters through de-adhesion in regions depicted in D (left) and yellow ROI of C (right). Bottom: Number of 0.3–3 μm size filaments in region with stress fibres and region lacking stress fibres. Arrows indicate increasing/decreasing trends. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

To assess the other morphological alterations of the actin network, we image the basal actin-layer using TIRF microscopy (Fig. 7 B-D, Movie S3). In live CHO cells, TIRF imaging reveals both – stress fibres (Fig. 7 B, D) and smaller filamentous structures (Fig. 7 B, C) which are then analyzed for quantifying the changes (Fig. S12).

Quantification of parameters - orientation, coherence and total intensity is performed for the same-sized ROI in every frame in the image sequence [28]. On de-adhesion, the total intensity and coherence decreases while orientation shows more fluctuations (Fig. 7 E, S12 B-D). The increasing (and fluctuating) trend of frame-to-frame variation of orientation parameter (Fig. S12 D) signifying enhanced orientational-mobility of actin filaments. We find the opposite trend in control cells that re-spread (Fig. S 12 E-F, Movie S4). In de-adhering cells, object detection (Fig. S12 G) reveals a population of filamentous structures ranging from 0.3 to 3 μm at the initial timepoint (Fig. S12 H). Particle analysis performed on filamentous structures displays fluctuations in object-length and object-density with progress of de-adhesion (Fig. 7 E). Stress fibres disassemble in the first 5–15 min (Fig. 7 E) and display curved morphology (Fig. 7 D, yellow arrows), events of buckling (Fig. S12 A) and fragmented filaments with increased density and mobility (Movie S5).

The enhanced dynamics (Movie S6) of these filamentous structures sized ~0.3–3 μm, on quantification, yield some varying trends, for e.g., in number density (Fig. 7 E, S12 I; decrease followed by increase: 6/12 cases, second slope: 1.1 ± 0.75 per min per 75 μm²; steady increase: 5/12 cases, slope: 0.67 ± 0.45 per min per 75 μm²; minor reduction: 1/12 cases, slope: −0.85 per min per 75 μm²). However, fluctuations in length of these filamentous structures, are consistently seen on quantification (Fig. 7E, S12 J).

This validates the causal relationship of how de-adhesion or a sudden loss of traction forces results in the contracting and crumpling of the network perhaps thereby freeing up myosin II. We next seek to understand if spread area and cortex thickness are similarly related in a system where a change in spread area is not artificially induced but triggered by differentiation.

2.6. Differentiation increases spread area and reduces thickness in THP1 cells

THP-1 cells were plated on Line10 patterns and either untreated or treated with PMA (Phorbol 12-myristate 13-acetate) – a differentiating agent [29] resulting in their transition from monocytes to macrophages. On this micropattern, in the absence of PMA treatment, cells are rounded (Fig. 8 A). Differentiated cells, or cells treated with PMA, however, display a well spread morphology (Fig. 8 B) with significant and clear difference in their spread area despite being plated on the same micropattern (Line10). We compare the cortex in the two populations (Fig. 8 C-D) and find that differentiation leads to increase in spread area (Fig. 8 D left) and a concomitant decrease in cortex thickness (Fig. 8 D right). Therefore, the dependence of cortex thickness on cell spread area in true even when the spread area change is induced not by artificial means like micropatterning or EDTA-mediated de-adhesion but a natural process like monocyte to macrophage differentiation.

Fig. 8.

Differentiation of THP-1 cells increases spread area and reduces cortex thickness. (A) Representative image of Phalloidin Alexa Fluor 568 staining the F-actin of a THP-1 cell grown on Line10 micropattern and (B) a THP-1 cell treated with PMA (THP-1 + PMA) also grown on Line10 pattern captured at 5.8 and 2.2 μm above basal plane respectively. Scale bar: 10 μm. (C) Linearized and zoomed-in view and color-coded image of the cortex (at marked with yellow ROI in (B)). Scale bar: 2 μm. (D) Spread area (measured at basal plane) and thickness (measured at 2.2 and 5.8 μm above basal plane) of both conditions. (n = 40 and 34 cells from THP-1 on Line 10 and THP-1 + PMA on Line 10 condition captured from three independent experiments). ** p < 0.001. Test used: Wilcoxon based Mann-Whitney U test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3. Discussion

Currently, cortex thickness is believed to be regulated by filament length regulators, motor proteins, actin turnover rates etc. This study demonstrates another key regulator – cell spread area. Two important methods were advanced for addressing this. Visualizing a clear cortical region is a challenge in well spread cells and is achieved here by micropatterning. This revealed the varying efficiency of myosin II in cortex thickness regulation – the central message of this paper and enabled cortex measurements in undifferentiated and differentiated THP-1 cells. The other important advance undertaken is incorporating multi-Gaussian functions instead of a single Gaussians [30,31] for fitting the cross-sectional intensity profile of sub-membranous actin to obtain the FWHM (used as a measure of thickness). This provides better estimation of the cortex thickness and allows understanding numbers emerging from STED and confocal images (Fig. S2).

STED imaging of cortex reveals more than cortex thickness. The presence of intermittent parallel layers near the “wall” of cortex and merging of these structures (lesser distance (in R1005 than Line10) between peaks (Fig. 1 F)) is new in the field, to the best of our knowledge, and falls well in line with the observation of cofilin's expression and its proposed role in network fluidization [32].

A recurrent question in biology is how cells sense their mechano-environment. Having shown that cells with low spread area have thicker cortex, we pinpoint the mechanism to be myosin-II-based and accompanied by actin-severing protein cofilin [13]. The first reason that we tend to conclude myosin II to be affected first is because of the existing theory of its load-dependent turning over [17], our observation of traction forces being best correlated to thickness than other physical parameters (surface area, volume) (Fig. 5) and our measurement of indeed enhanced myosin IIA turnover in rounded cells (Fig. 3). Blebb, an inhibitor of myosin II's motor activity, increases cortex thickness only in well spread cells (experiencing high traction forces) (Fig. 3). Therefore, the simplest model suggested here is reducing traction forces releases myosin II from the actin network and thus thickening the cortex and making it less sensitive to substrate stiffness (traction forces) or myosin inhibitors. Cofilin can be triggered by actin remodelling [24] although the complete mechanism therein needs to be separately investigated. Since spread area not only alters thickness but also the efficiency of myosin II's regulation of thickness, we believe it may indeed be a part of cellular mechanosensing.

In CHO cells, we find that cells choose to spread less on Line10 patterns than they do on glass (Fig. 2E middle) but have similar thickness (Fig. 2E left) suggesting a saturating effect of myosin II's contraction of the cortex. The high spread area regime (> 450 μm²) is perhaps chosen when cells can be highly mechanically sensitive through their cortex. Evaluating the physiological implication of this observation, cell spread area and biochemical state are indeed intricately linked in differentiation to metastasis. Interestingly, stem cells at low spread areas are known to preserve their stemness better [33] while in THP-1 monocytes, 3D geometry is reported to initiate adhesion and sufficient to trigger differentiation which can also be chemically triggered by PMA [34]. We show that reduction in cortex thickness accompanies PMA-triggered differentiation of THP-1 cells and the increase in cell spread area (Fig. 8). This suggests that architectural factors like cell spread area may control cortex thickness and thus help cells sense spatial gradients or temporal alterations in the microenvironment. This could accordingly retune programs relevant in differentiation, development [35], proliferation [36] or cancer metastasis [37], each of which has spatial patterns closely linked to the mechano-environment.

Finally, on a technical note, this paper puts forward the significance of maintaining similar cell spread area during studies addressing effect of cell-state on cortex mechanics. As an example, the reported increasing of tension and lowering of thickness in mitotic cells [13] can be understood only when compared to rounded interphase cells. Well-spread cells, as we show, have a thinner cortex than mitotic cells (Fig. 5 F). Extending the argument, we believe that the thinner cortex on Cyto D administration (expected to transiently increase the myosin II to actin ratio [3]) - coinciding with straightening of the cell edge (Fig. S6) - actually reflects higher internal stress acting laterally on the actin-layer prior to rupture. Note that the retractions velocities (0.5–1.5 μm/min, Fig. S6) fall in the expected range [13]. Thus, we argue that thickness monitored in similarly spread cells may inversely reflect the loading of the cortex.

4. Conclusion

The actomyosin cortex is a hub where mechanosensing is best achieved. Cell spread area directly alters the forces of interaction of cells with their mechanical environment and affects various factors in the cell including its volume. Herein we dissect out the plethora of internal cellular response – altered volume, surface area, endocytosis, cofilin expression, myosin II's turnover rate and actin cortex thickness. We conclude that cell spread area regulates cortex thickness and the efficiency of its regulation by myosin II by altering myosin II's turnover rate, in line with propositions of its load-dependent binding. Also true during THP-1 differentiation, this control could be a common design possibly in play in processes where mechanics defines cell fate changes.

5. Experimental procedures

5.1. Cell culture, immunostaining and actin labelling

CHO cells were cultured in Dulbecco's Modified Eagle's Medium (Gibco, Life Technologies, USA) supplemented with 10% Fetal Bovine Serum (Gibco). Cells were grown for 24 h before experiments unless otherwise specified. Fixation was performed using 4% paraformaldehyde (PFA) (Sigma) for 15 min at 37 °C followed washing thrice with PBS (Sigma). Immunostaining was carried out using 2% BSA as blocking buffer for 2 h followed by 1:100 dilution of Anti-Filamin A primary antibody (Abcam, UK) for 1 h and then PBS wash. Cells were then incubated with 1:400 dilution goat anti rabbit IgG Alexa Fluor 488 conjugated secondary antibody (Abcam, UK) along with 1:500 Phalloidin Alexa Fluor 568 (Molecular Probes, Life Technologies) for 45 min.

THP-1 cells were cultured in RPMI media-1640 media supplemented with 10% heat inactivated FBS (Gibco) and 5% Penstep (Gibco). Chemical stimulation of THP-1 cells on Line10 was done with 50 nM PMA for 48 h.

F-actin labelling was performed by incubating fixed cells either with 1:200 dilution of Phalloidin Alexa Fluor 488 or Phalloidin Alexa Fluor 568 (Molecular Probes, Life Technologies) or 33 nM Rhodamine-phalloidin (Molecular Probes, Life Technologies) for 45 min in the dark. 1 μg/ml DAPI (Sigma) was used along with Phalloidin to label DNA and selecting mitotic cells for imaging. Coverslips were either mounted on slides with mowiol (Sigma) or cells on glass-bottom petri dishes were fixed and immersed in PBS for imaging.

5.2. Plasmids and transfection

Plasmids used were pDEST/Life-Act-mCherry-N1 (a gift from Rowin Shaw, Addgene, plasmid # 40908) and mEmerald-LifeAct-7 (a gift from Micheal Davidson, Addgene, plasmid # 54148) for actin [38], mCherry-Myosin-IIA-C-18 (a gift from Micheal Davidson, Addgene, plasmid # 55105) for myosin II, mCherry-Alfa_actinin-19 (a gift from Micheal Davidson, Addgene plasmid # 54975) for α-actinin. pEGFP-C1 (Clontech) for cytosolic marker.

CHO cells were plated on customized 35 mm glass-bottom imaging dishes for at least 16 h before transfection. All transfections were performed using 0.5–0.75 μg plasmid DNA by lipofection (Lipofectamine 3000, Life Technologies) according to manufacturer's instructions.

5.3. Perturbations using drugs

To perturb actin, cells were treated with 5 μM Cyto D (Sigma) [39,40], 5 μM Lat B (Sigma) [40] and 5 μM or 50 μM Blebb (Sigma) [19] for 30 min before fixation and staining. 5 μM Cyto D was used for live time-lapse imaging. For ATP depletion (ATP dep) cells were incubated with mixture of 10 mM 2-Deoxy-d-glucose and 10 mM NaN₃ for 30 min in glucose free M1 medium. 80 μM Dynasore (Sigma) [41] were used to prevent actin dynamics. 10 μM Cyto D was used for assaying the temporal response of the cortex.

5.4. Surface micropatterning

Glass coverslips were etched with 19:1 mixture of ethanol and acetic acid for 30 min followed by washing twice and incubating for 15 min with ethanol. Air dried coverslips were exposed with deep UV in the UV-Ozone cleaner (Jelight Company, USA) for 5 min. Clean coverslips were incubated with 0.2 mg/ml PLL-g-PEG (SuSos, Switzerland) (prepared in 10 mM HEPES (Sigma), pH 8.5) for 2 h inside a moisture box. Photo-masks (JD Photo Data, UK) already cleaned in the UV-Ozone cleaner for 5 min were used for patterning the PLL-g-PEG coated coverslips for 10 min. Coverslips with water drops were placed on the photomask and excess water soaked off to aid adherence. Patterned coverslips were removed from photomask by floating them off using water. Finally, patterned coverslips were incubated with 20 μl of 25 μg/μl fibronectin (Sigma) (prepared in 100 mM NaHCO₃, pH 8.6) for 45 min inside a moisture chamber. Patterned coverslips were washed twice with PBS before seeding cells. CHO cells were detached from culture using 0.02% EDTA (Calbiochem, USA). The cell suspension was centrifuged to discard EDTA. 2 × 10⁵ cells were seeded per 25 mm coverslip and unattached cells were removed 30 min after plating using equilibrated media. The micropatterning protocol was adapted from [42,43].

5.5. Cell de-adhesion

For measuring cortex thickness and volume of single live cells through de-adhesion, LifeAct-mCherry transfected CHO cells were treated either with Trypsin EDTA (0.25%) (Gibco, Life Technologies, USA) or with EDTA solution and followed. EDTA solution was prepared in PBS by ensuring final concentrations at 1 mM EDTA and 20 mM glucose (Sigma). Imaging media (DMEM +10% FBS) was replaced by EDTA solution and z-stacks were captured either before and after 1 h of EDTA addition or every 5 min after addition of EDTA. For understanding the effect of Dyna on de-adhesion, same cells were assayed in three phases - without drugs, with drugs in media and with drugs in EDTA solution at time points 0, 30 and 90 min.

5.6. Endocytosis quantification

CHO cells were grown on Line10 and R1005 patterns. Post 24 h of cell seeding petri dishes were kept on ice for 10 min before incubation for 10 min with 5 μM FM 1–43 FX (Molecular Probes, Life Technologies) resuspended in PBS or EDTA solution. Cells were fixed for 15 min using 3.7% PFA. For Line10 and R1005, PBS supplemented with glucose (20 mM) was used as a medium, while 1 mM EDTA in PBS supplemented with glucose for EDTA condition. Excess free FM 1–43 FX was washed twice with PBS before fixation. Total amount of endocytosed FM 1–43 FX was calculated by integrating the fluorescence intensity of each cell over whole stacks captured with 0.5 μm z-step-size.

5.7. Polyacrylamide hydrogel preparation

Traction force microscopy was performed on two kind of hydrogels - micropatterned and non-patterned. For non-patterned hydrogel preparation, clean 22 × 22 mm glass coverslips were activated by incubating with 3-(Trimethoxysilyl) propyl methacrylate (Sigma) for 5 min, washed with water, followed by 0.5% glutaraldehyde for 1 h. In parallel, clean coverslips (12 mm diameter) were charged by covering with 0.025% poly-l-lysine solution (Sigma) for 1 h, air dried and covered with 0.02% solution of 100 nm fluorescent beads (Molecular Probes, Life Technologies) for 1 h. Gel solution containing 0.5% ammonium persulfate (Calbiochem), 0.05% tetramethylethylenediamine (Sigma), 2 mg/ml NH-acrylate were mixed with 7.25% acrylamide and 0.16% Bis-acrylamide to create gel with ~13.6 kPa stiffness [44]. 7 μl of solution was placed on functionalized glass-bottom coverslip and overlaid with the bead-coated-coverslip. Finally, the top coverslip was removed and mounted on a custom-made glass-bottom dish and incubated with 100 μg/ml fibronectin solution until cell seeding.

Patterned hydrogel was obtained by following patterning from UV glass pattern method describes in [45]. PLL-g-PEG coated circular coverglass were micropatterned using photomask and coated with fibronectin as described earlier in surface coating section. Polyacrylamide hydrogel were prepared with final working concentration of 7.5%, 12%, 12% acrylamide (BioRad, Hercules, CA) and 0.1%, 0.15%, 0.45% Bis-acrylamide (BioRad) to achieve 2.8, 16.3 and 43.3 KPa stiffness respectively [46]. A gel solution was polymerized by mixing 0.01% of 100 nm fluorescent beads (Molecular Probes, Life Technologies), 0.5% ammonium persulfate (Calbiochem), 0.05% tetramethylethylenediamine (Sigma) with appropriate acrylamide and Bis-acrylamide mixture. 7 μl of final polymerizing gel mixture were sandwiched between micropatterned circular coverglass (10 mm diameter) and salinized coverslips (22 × 22 mm). Micropatterned circular coverglass was removed, micropatterned hydrogel containing coverslips was mounted on custom-made glass-bottom dish with silicon grease followed by cell seeding.

5.8. Microscopy

Imaging was performed on an inverted microscope (IX81, Olympus Corporation, Japan) using 100× TIRF objective (N.A 1.49) and a CMOS camera (ORCA-Flash 4.0, Hamamatsu Photonics, Japan) with a 1 pixel = 65 nm. During epi-fluorescence imaging, same acquisition settings were used through each experiment ensuring that the signal-to-noise of the actin cortex was ~10. TIRF images were captured using DPSS lasers of 488 nm and 561 nm wavelengths and at penetration depth of 70 nm. Traction force imaging was performed by 60 × 1.49NA objective with pixel size 108 nm. Bright field image and corresponding fluorescence image of beads embedded beneath the cell was captured. Cells were removed by replacing media with trypsin-EDTA and image of beads in relaxed condition was taken. For all live cell imaging cells were maintained at 37 °C throughout by using an onstage as well as cage incubator (Okolab, Italy).

Confocal imaging was done on Laser scanning confocal microscope (Carl Zeiss, LSM 710) with 100× oil objective lens (NA 1.46).

All STED imaging was performed on commercial system Dual Color STED Microscope (Abberior Instruments, GmbH) equipped with Olympus UPlanSApo 100 × 1.4 NA objective. F-actin labelling molecules Phalloidin Alexa Fluor 568 were excited with 561 nm pulse laser, depleted using 775 nm STED laser with repetition 40 MHz and pulse width 90 ns. 605/50 filter was used in emission detection. Image was capture using Imspector software (Abberior Instruments, GmbH) using pixel size 20 nm.

5.9. FRAP

FRAP experiments were executed with 100 × 1.49 objective on laser scanning confocal microscope (Fluoview, FV3000, Olympus, Berlin, Germany) with 561 nm laser and spectral detector. In FRAP experiments, circular region (radius = 0.5 μm) focusing at cell edge was bleached with 100% laser power in one iteration using tornado mode. Pixel dwell time for bleaching was 12 μs/pixel and typical duration was 527 ms. Bleaching and imaging plane for Line10 and R1005 condition was 2.1 and 5.4 μm above the basal plane respectively. Before bleaching 5 frame prebleach image were captured. Recovery area of 8 × 8 μm² was capture at interval of approximately 3 frame per second over 150 frames.

5.10. Flow cytometry

CHO cells were either grown on micropattern (Line10 and R1005) or just glass coverslips in 35 mm dish either for 24 h (Line10, R1005 and 24 h conditions) or 5 h (5 h condition). Cells were fixed with 4% PFA for 15 min followed by trypsin-EDTA mediated detachment from the glass surface. Cells were centrifuged at 1500 rpm for 5 min and then permeabilized with 0.5% tween 20 solution for 15 min followed by incubation with 100 μl rabbit anti-cofilin (Abcam, UK) or rabbit anti-S3-p-cofilin (Abcam, UK) (1:100 dilution) for 2 h. Final incubation was with Alexa Fluor 488 conjugated anti-rabbit IgG (1:400 dilution) for 30 min in dark. Cells were washed with PBS and resuspended in 400 μl PBS then analyzed by flowcytometry (BD FACSVerse, San, Jose CA). Same PMT voltage, threshold and other parameter were used throughout all conditions. Immunolabeling were confirmed by comparing PMT voltage of immune stained cells with unstained sample incubated with secondary antibody. 10,000 cells were analyzed in each condition. Alexa Fluor 488 was excited with 488 nm laser and emission signal is collected with 527/32 band pass filter.

5.11. Extracting cortex thickness, surface area and volume

Image analysis was performed using softwares - Image J/Fiji (https://imagej.net/Fiji) and MATLAB (MathWorks). For calculating the FWHM of cortex, the analysis was performed in two parts – straightening image of the cell edges across cortex using Fiji and fitting intensity line-scans perpendicular to cortex to gaussian for FWHM calculation in MATLAB. Firstly, in Fiji, segmented lines of 2 μm (100 pixels in STED or 12 pixels in confocal or 31 pixels in epifluorescence) length across cell cortex were drawn, these lines were fitted to non-uniform cubic spline, following straightening 5 um wide region providing 2 μm × 5 μm image segment of straighten cortex. Secondly, Fluorescence line-scans perpendicular to the cortex were fitted using 3- and 4-term Gaussian function and the better fit (higher R²) used for further analysis. FWHM of the Gaussian whose peak was located closest to the centre of the line was used as a measure of thickness at that cross-section. FWHM averaged across ROIs or cells was termed the average thickness and the standard deviation (SD) termed the variability of thickness. Image plane considered for cortex thickness measurement is mentioned in figures otherwise maximum projection plane was consider for analysis to avoid over estimation. For quantification of actin intensity, images of actin labelled cells were acquired using same concentration of Rhodamine-phalloidin and same acquisition parameters. Peak intensity values of line-scans were measured after background subtraction for both fixed and live cells. For relative changes in actin intensity in live cells, same cells were followed, and same acquisition settings were used before and after EDTA administration. For spread area, volume and surface area quantification, stacks of actin labelled cells were captured from bottom to top using 0.43 (Confocal) or 0.5 μm step size. Manual ROIs along the boundary of each stack were drawn. Total projected area at the bottom plane were considered as a spread area of cells. Volume of each cells were computed by calculating the summation of the cell-areas obtained for each image of the stack and multiplying it by the z-step-size. Similarly, surface area was calculated by summing up the cell-perimeters obtained for each image of the stack and multiplying by z-step-size.

5.12. Filaments orientation and traction stress analysis

Average parameters termed energy, orientation and coherence reflect the total intensity, the orientation and coherence in pixel intensity distributions [28] that were calculated for each frame of image sequences using Orientation J plugin of Image J. SD of orientation for 5 consecutive frames (using sliding window algorithm) was calculated and termed SD(Orientation). Single filament analysis was performed by first normalizing images (0.3% saturated pixels) in Fiji and then detecting objects by background subtraction followed by local thresholding using Otsu thresholding algorithm using MATLAB.

Stressed and relaxed bead-images were aligned at subpixel to account for any drift using Align Slice – an ImageJ plugin. Bead displacement fields were computed using Particle Image Velocimetry (PIV) as describe previously [47] using the same iteration scheme for all cells. The iteration scheme used were 128/256 - pixel (1 pixel = 108 nm) (Interrogation window size/Search window) window for 1st pass, followed by 64/128 in 2nd pass and finally 32/64 in 3rd pass at a threshold of 0.60 (cross-correlation coefficient). Post PIV processing were performed by Normalized Mean Test (NMT) and parameters were NMT noise 0.3 and NMT threshold of 3.0. The traction stress field and magnitude were calculated by the Fourier Transformed Traction Force Cytometry (FTTC) [47] using stiffness 2.8, 13.6, 16.3 and 34.3 KPa, Poisson ratio 0.5 and regularization factor 9 × 10⁻¹¹. Average, maximum displacement and average, maximum traction stress was calculated for each cell by overlaying cell ROI captured from brightfield image.

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbamcr.2019.07.011.