Mechanical signaling in a pulmonary microvascular endothelial cell monolayer

Abstract

The mechanical microenvironment of an endothelial cell includes a stable protein scaffold on the basal side, flowing blood on the apical side and contractile cells on the lateral sides. Interaction with the protein scaffold and flowing blood modulates the ability of endothelial cells to migrate, align and maintain barrier function. Interaction with neighbors provides the endothelial monolayer unique “collective” properties. However, the nature of local mechanical signaling – i.e., the local functional consequence of a cell interacting with its contractile neighbors – remains unclear. Using an advancing sheet of pulmonary microvascular endothelial cells, here we examine the mechanical properties of an individual cell and its neighboring region. By combining Monolayer Stress Microscopy (MSM) with a novel analysis, we assessed several mechanical properties of an individual cell and its neighboring region. Across the monolayer, mechanical properties of the neighboring region defined multicellular “subdivisions” wherein constituent cells were exposed to a similar mechanical microenvironment. Adjacent subdivisions were separated by a narrow interface where adjoining cells were exposed to remarkably different mechanical microenvironments. Comparison of temporal fluctuations in mechanical properties of individual cells and those of their neighboring regions suggested three distinct intercellular mechanical signaling processes. These processes indicated that change in size, shape and speed of individual cells is associated with change in contractile forces in their neighboring regions. In summary, we present a novel approach to assess the mechanical interactions of individual cells with their contractile neighbors and identify potential functional consequences of such interactions.

Introduction

Decellularized lung scaffolds have wide applications in pre-clinical tissue engineering studies and in lung transplantation model [1]. During reendothelialization of a decellularized lung scaffold, endothelial cells are expected to attach, migrate, cover the scaffold vascular surface and establish a restrictive barrier [1,2]. To cover the vascular surface, endothelial cells migrate while maintaining physical contact with their neighbors [3,4]. Such collective endothelial migration emerges from the ability of each cell to sense and respond to chemical and mechanical signals within its microenvironment [5-8]. For an endothelial cell, an important component of the microenvironment is its neighboring cells. Junctions between the neighboring cells enable transmission of mechanical signals, such as cellular contractile forces, over a long distance [7]. Across the monolayer, long-distance force transmission creates correlated force patterns that can regulate endothelial barrier function [9]. However, the nature of mechanical signals from immediate neighbors, and the endothelial response to those mechanical signals, remains unclear.

Extensive heterogeneity has been described in endothelial cells along the pulmonary vascular bed [10]. Unlike endothelial cells from pulmonary arteries and pulmonary veins, the endothelial cells from pulmonary microvessels have the capability to cover the entire pulmonary vasculature of a decellularized lung scaffold [2]. Here, we focused on pulmonary microvascular endothelial cell (PMVEC) monolayers cultured on collagen-coated hydrogel of stiffness resembling an in vivo condition [11]. Using MSM we quantified subcellular mechanical stress and physical motion across the monolayer [7,12]. Using novel data analysis, we quantified several mechanical properties of individual cells and their neighboring regions. To assess cellular morphology, we quantified spread area, orientation, and circularity. To assess the state of mechanical stress, we quantified cytoskeletal tension, the orientation of maximum cytoskeletal tension, cytoskeletal tension anisotropy, and mechanical stress transmitted to the extracellular matrix (i.e., substrate traction). To assess motion, we quantified speed and the direction of motion. Individual endothelial cells within the monolayer appeared to belong to one of the two categories, either as part of an extended subdivision with neighboring cells receiving similar mechanical signals or as part of a narrow strip where neighboring cells received remarkably dissimilar mechanical signals. Surprisingly, changes in the size, shape, and speed of an individual cell were associated with changes in the mechanical stresses in the neighboring region.

Materials and Methods

Cell culture

Rat pulmonary microvascular endothelial cells (rat 1, passage 11) were acquired from the cell culture core of the Center for Lung Biology at the University of South Alabama and cultured in Dulbecco's Modified Eagle Medium (Invitrogen, 11965) containing 10% fetal bovine serum (Atlanta Biologicals, S11550H) in a standard tissue culture environment (37°C, 95% air, and 5% CO₂) [10,13]. The data was acquired from cellular passages 12 through 16. The data include time-lapse sequences, three of which were more than 940 minutes each, and five were more than 30 minutes each. The frequency of time-lapse was 0.2 Hz.

Polyacrylamide hydrogel preparation

Cells were seeded on a collagen-coated (Corning, 354236) polyacrylamide gels of 1250 Pa shear modulus (i.e., 3750 Pa Young’s modulus) and approximately 100 μm thickness with fluorescent beads (0.5 μm in diameter, Molecular Probes, F8812) embedded immediately underneath the top surface of the gel (Fig. 1a) [14,15]. The hydrogels were prepared in 35 mm glass-bottom dishes, and the images were acquired using an inverted wide-field fluorescence microscope (Leica, DMI 6000B) and confocal microscope (Nikon A1R).

Figure 1. Quantitative assessment of the morphology, motion, and mechanical stresses of advancing PAECs and their neighboring regions.

a. Schematic of the in vitro cell migration assay used to culture and visualize the PAECs. b. For each cell within the monolayer (red area) and its immediate neighboring region (blue area), cellular motion and forces were quantified at a subcellular level.

Assessment of subcellular forces and motion

To quantify mechanical stresses across the interface between the cell and the hydrogel (i.e., substrate traction), we used Fourier Transform Traction Microscopy (FTTM) [15,16]. To quantify mechanical stresses across the cell sheet, we used Monolayer Stress Microscopy (MSM) [7,12]. To quantify cellular motion, we used Particle Image Velocimetry (PIV) [7]. The cross-correlation window size of approximately 15.8 μm × 15.8 μm for PIV data analysis, and approximately 7.9 μm × 7.9 μm for FTTM and MSM data analysis. Spacing between adjacent windows was approximately 7.9 μm for PIV data analysis and approximately 2.6 μm for FTTM and MSM data analysis.

Assessment of mechanical properties of the cell

To examine mechanical and biochemical signals in the cell simultaneously, we have developed an Integrative Cellular Signaling Toolkit (iCST). iCST uses the FTTM, MSM and PIV framework to respectively obtain subcellular values of the state of mechanical stress and migration properties of the cell [7,12,15,16]. Data acquisition component of iCST is automated using a custom software based on μManager-2.0 [17]. Data analysis and visualization component of iCST is implemented in NIH-ImageJ [18]. Further details on segmentation algorithms used in iCST are out of the scope of this paper and will be part of a separate manuscript. The definition and the approach to quantify the mechanical properties of each cell and its neighboring region are described below.

For each cell, using image segmentation capabilities of iCST, we quantified morphological properties such as area, orientation, and circularity. Cellular circularity (C) was defined as C = 4π(A/P²) where A is the spread area and P is the perimeter of the segmented cell.

To quantify the subcellular state of mechanical stress, we started with the outcome of FTTM – i.e., subcellular, two-dimensional components of shear stress defined with respect to a laboratory frame of reference, $T_i=\left[\begin{array}{c}\hfill T_x\hfill \\ \hfill T_y\hfill \end{array}\right]$ [15,16], the outcome of MSM – i.e., subcellular, thickness-averaged, two-dimensional components of mechanical stress defined with respect to a laboratory frame of reference, ${\sigma }_{ij}=\left[\begin{array}{cc}\hfill {\sigma }_{xx}\hfill & \hfill {\sigma }_{xy}\hfill \\ \hfill {\sigma }_{xy}\hfill & \hfill {\sigma }_{yy}\hfill \end{array}\right]$ [7,12]. At each numerical grid point, these stress components and the orientation of the laboratory frame of reference were used to obtain local maximum, σ_max, and minimum, σ_min, principal stresses and maximum principal orientation, θ [7]. Using these properties, we defined subcellular substrate traction as $\sqrt{T_x^2+T_y^2}$, maximum tension orientation as θ, cytoskeletal tension as (σ_max + σ_min)/2, tension anisotropy as (σ_max − σ_min/2.

To quantify the subcellular state of motion, we started with the outcome of PIV – i.e., subcellular two-dimensional components of velocity in the laboratory frame of reference, $v_i=\left[\begin{array}{c}\hfill v_x\hfill \\ \hfill v_y\hfill \end{array}\right]$ [7]. Using these velocity components, we defined local speed as $\sqrt{v_x^2+v_y^2}$, and local orientation of motion as $\text{arctan}{\left(\frac{v_y}{v_x}\right)}_{.}$.

To quantify the state of mechanical stress and motion of an entire cell (Fig. 2d-i), we considered the corresponding subcellular values within the cellular boundary and used the median of these values as the mechanical property of the cell (Fig. 1b, the red-colored area shows an example of a region within the cellular boundary). In the same way, to quantify the state of mechanical stress and motion of the neighboring region (Fig. 2j-o), we considered the corresponding subcellular values within the neighboring region and used the median of these values as the mechanical property of the neighboring region (Fig. 1b, the blue-colored area shows an example of the neighboring region).

Figure 2. Maps of the mechanical properties of individual PMVECs and their neighboring regions.

Panels a-i show patterns of mechanical properties for individual cells. These mechanical properties include a. cellular spread area, b. cellular orientation, c. cellular circularity, d. cytoskeletal tension, e. maximum tension orientation, f. tension anisotropy, g. cellular speed, h. direction of cellular motion, i. substrate traction of individual cells. Panels d-i indicate instantaneous median values of the mechanical properties within the cell boundary (Fig. 1b, red area). Panels j-o show patterns of mechanical signals from the neighboring region of each cell. The mechanical signals include j. cytoskeletal tension, k. maximum tension orientation, l. tension anisotropy, m. cellular speed, n. direction of cellular motion, o. substrate tractions within the neighboring region of individual cells. Panels j-o indicate instantaneous median values of the mechanical properties within the neighboring region of the cell (Fig. 1b, blue area). All color maps are based on the same color spectrum shown at the bottom. This spectrum defines a linear range bound by minimum:maximum values for each panel as follows a. 290:3220, b. 0:180, c. 0.3:0.9, d. −22:58, e. 7:172, f. 4:75, g. 0:0.5, h. −180:180, i. 3:40, j. −3:40, k. 30:150, l. 9:30, m. 0.1:0.3, n. −97:43, and o. 6:16. Size of each image is 770 μm × 650 μm.

Leaders and followers

Cells in the monolayer moved back and forth from being at migration front to a couple of cell distances away. Cells in this frontal region, which spanned roughly 100 μm, were visibly more spread than remaining cells (Fig. 2a). The cells in this frontal region were defined as leader cells (or leaders). In order to have a non-overlapping neighboring region, the follower cells (or followers) were defined as the cells that were more than 200 μm away from the migration front. From all the leaders and followers, the top 20 of the most consistently detected cells were chosen for distribution plots (Fig. 3).

Figure 3. The cumulative probability distribution of the mechanical properties of individual PMVECs and their neighboring regions.

Panels show cumulative probability distribution of a. cellular spread area, b. cellular orientation, c. cellular circularity, d. cytoskeletal tension, e. maximum tension orientation, f. tension anisotropy, g. cellular speed, h. direction of cellular motion, i. substrate traction. Morphological properties of the cells in the neighboring region were not quantified; hence, panels a-c do not contain data with blue lines. Migration front was oriented approximately vertical (90°).

The cumulative probability distribution of mechanical properties

Cumulative probability distribution plots were made from the image sequences that were more than 940 minutes long. To generate the plot for distribution of cytoskeletal tension in the leaders (Fig. 3d, thick red line), we randomly chose 100 of more than 188 instances from each sequence and considered 20 cells per instance. This data set was used to generate the cumulative distribution plot for 6000 leader cell instances. The same procedure was used to generate distribution plots for other mechanical properties and for the followers and neighboring regions of leaders and followers.

Results and Discussion

Heterogeneous patterns of the mechanical properties of PMVECs

Cells closer to the migration front had a larger spread area than those farther from the front (Fig. 2a). By contrast, cellular orientation and circularity showed no apparent spatial pattern (Fig. 2b,c). Monolayer contained small clusters that were engaged in a local tug-of-war, where tension was highest at the center of the cluster and fell off to a small value at the edge of the cluster (Fig. 2d) [7,15]. Over a large contiguous area, maximum cytoskeletal tension orientation was approximately perpendicular to the migration front (Fig. 2e, 0°, and 180° are both horizontal orientations). Cytoskeletal tension was slightly more anisotropic in the cells near the migration front compared to that in the cells away from the front (Fig. 2f). Fast-moving cells existed across all the observed area; nevertheless, they represented a majority of the cells located close to the migration front (Fig. 2g). Cells moved remarkably perpendicular to the migration front and towards the open space (Fig. 2h). Interestingly, substrate tractions and tension anisotropy appeared to form a similar pattern (Fig. 2f,i). Note that the patterns described here are formed by the median values of the subcellular properties. In that, they reveal low-frequency spatial fluctuations across the monolayer.

Larger but sharper patterns of the mechanical properties of the neighboring regions

Maps of the mechanical properties of the neighboring region appeared to form multicellular “subdivisions” where all constituent cells experienced a similar mechanical microenvironment (Fig. 2j-o). Moreover, adjacent subdivisions were separated by a narrow interface where adjoining cells experienced a remarkably different mechanical microenvironment. These patterns would be similar to a median-filtered version of the patterns formed by the corresponding properties of individual cells (Fig. 2d-i).

Distribution of the mechanical properties

Change in the physical properties is slower in leaders and their neighboring regions than in followers and their neighboring regions (Fig. 3a-i, thin curves are smoother than thick curves). Leaders were larger than the followers (Fig. 3a, the thick line is consistently below the thin line). At several instances, leaders and followers had an approximately horizontal orientation (Fig. 3b, the segment closer to 0° is steepest). Morphology was less circular in leaders than in followers (Fig. 3c, the thick line is consistently above the thin line). Negative cytoskeletal tension was observed more frequently in followers and their neighboring regions than in leaders and their neighboring regions (Fig. 3d, the segments with negative tension have thin curves steeper than thick curves). High cytoskeletal tension was more frequent in followers and their neighboring regions than in leaders and their neighboring regions (Fig. 3d, the thick curves have steeper low and intermediate tension segments whereas thin curves have steeper high tension segments). At numerous instances, highest tension orientation was approximately horizontal (Fig. 3e, the thick lines have a long steep segment closer to 0° whereas the thin lines have a long steep segment closer to 180°). The tension was more anisotropic in leaders than in followers (Fig. 3f, the thick red curve is consistently below the thin red curve). Leaders moved faster than followers (Fig. 3g, the thick red curve is consistently below the thin red curve). At numerous instances, the cells were moving directly towards the open space (Fig. 3h, all curves are steepest close to 0°). Instances of large tractions were more frequent in followers and their neighboring regions than in leaders and their neighboring regions (Fig. 3i, the segments corresponding to large tractions were steepest for the thick red curve).

For the leaders, highest tension orientation being approximately perpendicular to the migration front is consistent with kenotaxis, which is the finding that cellular monolayer has a mechanical drive to fill open space (Fig. 3e, the thick red curve has a large steep segment close to 0°) [8]. The motion of cells being largely aligned with the highest tension orientation is consistent with plithotaxis, which is the finding that cells in a monolayer have the tendency to move along local maximum principal stress orientation (Fig. 3e,h, curves in both graphs have steepest segments around horizontal orientations) [7]. Compared to the leaders, the followers had a more circular shape, more slower motion, and they were more extensively surrounded by other cells (Fig. 3c,g). Simultaneous occurrence of these three conditions was consistent with followers being closer to the jamming transition than leaders, which is the state of the cell where motion being constrained by neighbors and has a structural signature of the shape being more circular [19].

A potential association between the mechanical properties of an individual cell and those of its neighboring region

Comparison of temporal fluctuations in mechanical properties of the cells and their neighboring regions suggested three new mechanical signaling processes. Each of these signaling processes was reminiscent of mechanical behavior of other cellular systems, an elastic sheet or a viscous film. Accordingly, we call these signals as relaxation, fluidization, and anchoring.

Relaxation signal was reminiscent of early epithelial-to-mesenchymal transition in Madin-Darby canine kidney type II cells behaving like an elastic sheet [20]. According to a similar elastic response in the neighboring region of a cell, a decrease in tension would be associated with a reduction in area in the neighboring region. If the motion of the outer boundary of the region is constrained, then the inner boundary will move closer to the outer boundary. Such elastic relaxation of the inner boundary would increase the area of the embedded cell. Similarly, increased tension in the neighboring region will, in turn, decrease the area of the embedded cell. In response to an increase in tension anisotropy in the neighboring region, the elastic relaxation of the inner boundary will be more anisotropic, thereby reducing circularity of the embedded cell. Similarly, decreased tension anisotropy in the neighboring region will in turn increase circularity of the embedded cell.

Fluidization signal described in Table 1 was reminiscent of cyclic stretch-induced cellular fluidization where contractile actin-cytoskeletal machinery disengages and the pool of globular actin increases [21]. Dysfunctional contractile machinery induces cellular spreading [20]. Taken together, these findings and current observations suggest a novel testable hypothesis that fluctuations in cytoskeletal tension or tension anisotropy in the neighboring region disengage cytoskeletal machinery of the embedded cell to induce uniform spreading of the cell.

Table 1. Intrinsic mechanical signaling between individual PMVECs and their neighboring regions.

There were a few exceptions to these signaling processes where behavior was opposite to that described in the table (Fig. 4a-f, dashed lines).

Mechanical signal	Process in the cell	Process in the neighboring region	Relationship	Observation
Relaxation	Changes in spread area	Changes in cytoskeletal tension	Inverse association	Fig. 4a,d, vertically aligned pairs of arrows drawn using continuous lines
	Changes in circularity	Changes in tension anisotropy	Inverse association	Fig. 4b,e, vertically aligned pairs of arrows drawn using continuous lines
Fluidization	Increase in spread area	Fluctuation in cytoskeletal tension	Association	Fig. 4a,d, vertically aligned pairs of a box and an arrow drawn using continuous lines
	Increase in circularity	Fluctuation in tension anisotropy	Association	Fig. 4b,e, vertically aligned pairs of a box and an arrow drawn using continuous lines
Anchoring	Changes in speed	Changes in tension anisotropy	Inverse association	Fig. 4c,f, vertically aligned pairs of arrows drawn using continuous lines

Anchoring signal was reminiscent of viscous flow around an anchored object [8]. The anchor will both increase tension anisotropy in the vicinity of the object and limit movement of the object. In cellular monolayer, “anchoring” represents an increase in substrate tractions. Indeed, the regions where substrate tractions are high were also the regions where tension was highly anisotropic (Fig. 2f,i,l,o). Such “anchoring”-induced increase in tension anisotropy will also reduce cellular speed. Lifting of such an anchor will both reduce tension anisotropy in the neighboring region and allow an increase in the speed of the cell.

Detailed assessments of the state of the actin cytoskeleton and focal adhesions are needed to examine structural implications of the relaxation, fluidization and anchoring signals.

Mechanical interaction at the length scale of a cell

Forces transmitted across neighboring cells have a strong influence on both biological and physical properties of the monolayer [4,7-9,14,20]. Alignment of such forces over length scales much larger than the size of a cell produces collective processes such as tug-of-war, plithotaxis, kenotaxis and jamming [7-9,15,19,20]. By contrast, the magnitude of such forces at a length scales much smaller than the size of a cell alters stability and structure of force-transmitting molecules [22,23]. However, at the size of the cell, the influence of intercellular interaction remains unclear. Not only the influence, but how to assess it is also unclear. We show that the median value of the subcellular forces and motion present an insightful picture of the intercellular interaction at the size of an individual cell.

Important questions that remain unanswered are what accounts for systematic regional differences in cell shape, size, and speed? Is there a systematic mechanical signal in the neighboring region that coincides or precedes a cellular division event? In connection with the subdivisions formed by the properties of neighboring regions, how is the behavior of a cell located at the middle of the subdivision different from that of a cell located at the interface of adjacent subdivisions? Finally, how do physiologic conditions (i.e., humoral factors, fluid shear stress, and gas concentration appropriate to microvessels) affect the mechanical signals in the neighboring region and the associated cellular response?

In conclusion, median values of the subcellular mechanical properties seemed to provide an insightful representation of mechanical signaling between an advancing PMVEC and its neighboring region. Across the monolayer, neighboring regions exposed adjoining cells to either remarkably similar or remarkably different microenvironment. Comparison of temporal fluctuations in mechanical properties of a cell with those in its neighboring region suggests that local intercellular interaction may be largely regulated by three mechanical signaling processes: relaxation, fluidization, and anchoring. Further assessment of these signaling processes may provide a unique mechanistic approach to tissue engineering.

Figure 4. Temporal fluctuation in the mechanical properties of an individual PMVEC and its neighboring region.

a-c Properties of a leader cell. d-f Properties of a follower cell. a, d Temporal fluctuations in spread area of a cell (Δ markers, left vertical axis) and cytoskeletal tension in its neighboring region (+ markers, right vertical axis). b, e Temporal fluctuations in circularity of a cell (Δ markers, left vertical axis) and tension anisotropy in its neighboring region (+ markers, right vertical axis). c, f Temporal fluctuations in the speed of a cell (Δ markers, left vertical axis) and tension anisotropy in its neighboring region (+ markers, right vertical axis). For each vertical axis, the choice of the range was was guided by the visual clarity of the temporal fluctuations. In each panel, vertically aligned pair of arrows indicate similar (dashed lines) or opposite (continuous lines) fluctuations; and vertically aligned pair of a boxes and an arrow indicate phases of rapid fluctuations in the mechanical properties of neighboring region and associated decrease (dashed lines) or increase (continuous lines) in cellular size (a, d) and circularity (b, e). The blue curves represent a local average over twelve (c) and eight (f) points. Red and green colors are used for visual clarity in highlighting successive phases of temporal fluctuations.

Highlights.

Cellular morphology and motion are associated with tension fluctuations in neighbors