Journal of Biomechanical Science and Engineering The

Bulletin of the JSME

Journal of Biomechanical Science and EngineeringVol.12, No.2, 2017

© 2017 The Japan Society of Mechanical EngineersJ-STAGE Advance Publication date: 2 May, 2017Paper No.16-00670

[DOI: 10.1299/jbse.16-00670]

The opposite mechano-response of paxillin phosphorylation between subcellular and whole-cell levels is explained by a

minimal model of cell–substrate adhesions

Abstract Cell-substrate adhesions are a mechanosensitive protein complex that regulates various cellular functions. Molecular mechanisms underlying the physical force-dependent regulation remain elusive partly because endogenous forces are distributed in a spatially heterogeneous manner within cells, thus complicating the interpretation on the effect of forces. Here we use a micropatterning technique to focus spatially distributed intracellular contractile forces onto a particular subcellular area, with which mechanical and pharmacological effects are separately analyzed. Single human osteosarcoma U2OS cells were plated within square micropatterns, and phosphorylation of an adhesion-associated adaptor protein paxillin was analyzed. Paxillin, visualized with immunostaining, was highly accumulated in the proximity of the corners of the square micropatterns where cellular forces are concentrated, but the huge paxillin-labeled adhesions were less phosphorylated compared to those present elsewhere as a small patch. Pharmacological inhibition of the endogenous forces resulted in disassembly of the huge dephoshorylated paxillin clusters; in contrast, the small, highly phosphorylated paxillin patches persisted. Similar negative regulation of paxillin phosphorylation is also induced upon loading of exogenous forces. Unexpectedly, on the other hand, immunoblot of cell lysates showed a tendency of a reduction in paxillin phosphorylation upon the same pharmacological inhibition of the endogenous forces. Thus, the response of paxillin phosphorylation to mechanical forces was the opposite between the immunostaining and immunoblot data; i.e., the phosphorylation is reduced and enhanced at subcellular level and whole-cell level, respectively, in response to loading of mechanical forces. To reconcile the contradictory results, we submit a simple model that is consistent with not only the present but also previous reports on the regulation of paxillin. The model implies that similar opposite response can generally emerge if the protein activation is negatively regulated at a local place while the activation trigger alters the assembly of the protein to the local place.

Key words: Cell–substrate adhesion, Focal adhesions, Paxillin, Cellular contractility, Micropatterning, Mechanobiology

1. Introduction

Cell–substrate adhesions, or simply adhesions, are multiprotein complexes that anchor cells to the extracellular matrix (ECM) substrate. These complexes function not only as an adhesion site but also as a signaling site to regulate various cellular functions that include the proliferation, apoptosis, and differentiation of particularly mesenchymal cell types. The signaling regulation at adhesions is modulated by the level of endogenous cellular tension, or contractile force, applied to the individual adhesions (Chen, 2008). Such a cellular tension is constantly exerted at cell adhesions as they are connected to nonmuscle myosin-II (NMII)–actin contractile fibers or stress fibers within cells. Thus, a

1

Shinji DEGUCHI*,**, Akira C. SAITO*, Tsubasa S. MATSUI*,**, Wenjing HUANG*,*** and Masaaki SATO*,****

* Department of Bioengineering and Robotics, Tohoku University

** Present: Division of Bioengineering, Graduate School of Engineering Science, Osaka University

1-3 Machikaneyama, Toyonaka, Osaka 560-8531, Japan

E-mail: [email protected]

*** Present: Department of Bioengineering, The University of Tokyo

**** Present: Frontier Research Institute for Interdisciplinary Sciences, Tohoku University

Received: 12 December 2016; Revised: 17 February 2017; Accepted: 24 April 2017

2© 2017 The Japan Society of Mechanical Engineers

Deguchi, Saito, Matsui, Huang and Sato, Journal of Biomechanical Science and Engineering, Vol.12, No.2 (2017)

[DOI: 10.1299/jbse.16-00670]

change in cell shape alters the level of cellular tension, which in turn elicits a tension-dependent regulation of cellular functions (Théry et al., 2006; Deguchi et al., 2011).

Molecular mechanisms underlying the mechanosensitivity of adhesions have been extensively investigated. Paxillin, a signal transducing adaptor protein closely associated with the adhesion complexes, is phosphorylated at tyrosine 118 (pY118-paxillin) mainly by focal adhesion kinase (FAK) and Src (pp60c-src) to mediate cell migration and adhesion maturation (Petit et al., 2000; Turner, 2000; Zaidel-Bar et al., 2003, 2007; Webb et al., 2004; Pasapera et al., 2010; Kuo et al., 2011). Previously, quantification of immunofluorescence images showed that the ratio of pY118-paxillin to total paxillin within individual adhesions is negatively regulated by the NMII activity or cellular tension (Zaidel-Bar et al., 2007). Consistent with this observation, we reported that individual adhesions that can sense local micro-geometry of the substrate contain highly unphosphorylated paxillin in the presence of the NMII activity (Saito et al., 2014). However, it remains unclear whether the NMII-driven cellular tension induces the selective recruitment of unphosphorylated paxillin to adhesions, or induces dephosphorylation at adhesions. Experimentally separating the effects of mechanical and pharmacological effects on the paxillin phosphorylation will be key to better deciphering this physical force-related molecular process.

In the present study, we confine single cells within square-micropatterned substrates in which their intrinsic contractile forces are borne predominantly along the edges of the square, and the forces are particularly focused on the four corners. Taking advantage of this spatially concentrated mechanical loading realized with micropatterns, mechanical and pharmacological effects in the regulation of adhesions are separately analyzed. We observed localization of adhesions that contain large amounts of unphosphorylated paxillin at the mechanically stressed subcellular regions, as expected from the previous studies described above. Unexpectedly, however, pharmacological inhibition of the cellular tension reduced the averaged phosphorylation level of paxillin contained in cell lysates. We then submit a conceptual, but comprehensive model to reconcile the discrepancy between the regional and averaged regulations of paxillin phosphorylation.

2. Materials and Methods 2.1 Cell culture

Human osteosarcoma U2OS cells (HTB-96, ATCC) were cultured in high-glucose (4.5 g/L) DMEM (12800-017, Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (172012, SAFC Biosciences) and 1% penicillin–streptomycin (168-23191, Wako). Cells were maintained in a humidified 5% CO2 incubator at 37°C.

2.2 Micropatterning

The surface of a φ27-mm-glass-bottom culture dish (3970-035, Iwaki) was coated using a spin-coater with a polydimethylsiloxane (PDMS) prepolymer (Sylgard 184, Dow Corning Toray), which was in advance prepared according to the manufacturer’s instruction, i.e., at a weight ratio of 10: 1 (base to curing agent). The dish was then placed in an oven (60°C) overnight to cure the PDMS. With a plasma-shielding mask (Athene TEM grid, EM Japan) attached to the surface, the dish was exposed to oxygen plasma (SEDE-P, Meiwafosis) at 4 mA in current and 20 Pa in pressure for 3 min (Fig. 1A). After removing the mask by forceps, the PDMS substrate was treated with 0.2% Pluronic F-127 (P6866, Thermo Fisher Scientific) in PBS for 1 h and then with 0.1% gelatin (G9391, Sigma) in PBS for 1 day at 37°C, followed by the seeding of U2OS cells to finally have square-micropatterned cells. 2.3 Finite element method

Mechanical stress generated within micropatterned cells was computationally evaluated by a finite element method (FEM) using COMSOL Multiphysics (ver. 4.4) with a method essentially similar to our previous study (Deguchi et al., 2011). Briefly, we herein simply consider two components: cell and substrate (Fig. 2A). A single cell is approximated as a linear elastic material (with a Young’s modulus of 1 kPa and Poisson’s ratio of 0.3) because only the steady state under eventual force balance is considered in this analysis. The cell model is cuboid in initial shape (35 µm in edge length and 5 µm in height), with round corners mimicking the actual projected cell geometry; the curvature radius at the corners is one-twentieth of the planar edge, i.e., 1.75 µm. The substrate is approximated to be a virtually infinite (100 µm in edge length and 20 µm in height), planar elastic material with a Young’s modulus of 1 MPa and Poisson’s ratio of 0. No particular subcellular components such as the nucleus are considered, but cellular contraction is assumed to occur uniformly in the cytoplasm because contraction-inducing actin and NMII are ubiquitous proteins. The

2



[DOI: 10.1299/jbse.16-00670]

Fig. 1 Micropatterning of cells using plasma-shielding masks. (A) Schematic for the methodology. See the text for details. (B) A single U2OS cell confined within a square micropattern. Scale, 20 µm. (C) Kymograph is created along a cyan-boxed area in B. The cell edges at the initial and final time points are marked by cyan arrowheads. This immobile position of the arrowheads suggests that the cell outline is unchanged over time. Scale for time, 10 min; scale for space, 20 µm. (D) Immunolabeling of pY165-p130Cas and fluorescent phalloidin staining of F-actin. The cell is 35 µm in edge length.

Fig. 2 FEM to predict endogenous mechanical stress. (A) Model for a cuboid cell plated on a planar substrate. (B) Equivalent stress in the substrate and cell, the latter of which undergoes isotropic contraction. (C) Equivalent stress at the upper surface of the substrate, with a magnified view of a boxed area showing one of the corners.

3



[DOI: 10.1299/jbse.16-00670]

actomyosin contraction is represented by a decrease in the stress-free reference length in all directions and all discretized elements of the cell. As no slip at any of the interface between the cell and substrate is assumed, the cellular contraction on the practically rigid substrate results in isometric contraction. Because the level of cell prestrain ranges up to 0.3 (Deguchi et al., 2006; Deguchi and Sato, 2009), here we decreased the reference length to 77% (=1/(1+0.3)×100) of the original length. The resulting von Mises stress (or the equivalent stress) is computed. Mesh sensitivity was separately performed to ensure that the results are essentially independent of the size of the computational meshes. 2.4 Microscopy

Cells were fixed with 1% paraformaldehyde for 30 min, washed with PBS, permeabilized with 0.1% Triton X-100 for 5 min, treated with 10% normal goat serum in PBS for 1 h at room temperature, incubated with anti-paxillin (1:100, BD Biosciences) and pY118-paxillin (1:100, Cell Signaling Technology) or pY165-p130Cas (p130Cas phosphorylated at tyrosine 165; 1:100, Cell Signaling Technology) at 4ºC overnight, incubated with the appropriate fluorescently-labeled secondary antibodies. F-actin in fixed cells was stained with Alexa phalloidin (Molecular Probes). Epifluorescence images or phase-contrast images were captured using a camera (ORCA-R2, Hamamatsu) on a microscope (IX-71, Olympus). 2.5 Inhibitors

Inhibitors used are as follows: actin polymerization inhibitor Cytochalasin D (CytoD, Santa Cruz Biotechnology), NMII’s ATPase inhibitor (-)-blebbistatin (blebbistatin, Merck Millipore), FAK inhibitor FAK II (Merck Millipore), and Src inhibitors Src I (Sigma) and PP1 (protein phosphatase 1, Merck Millipore). Cells cultured for 1 day were treated with either 2 µM CytoD, 5 µM blebbistatin, 1 µM FAK II, 1 µM Src I and 1 µM PP1, or solvent vehicle alone (0.5% dimethyl sulfoxide, DMSO) for 30 min in a 5% CO2 incubator at 37°C for subsequent immunostaining or immunoblotting. 2.6 Microstretch experiment

Micropatterns were created in a PDMS chamber that was fabricated according to Saito et al. (2014). Briefly, the PDMS prepolymer described above was poured onto a custom-made acrylic mold to form a hollow chamber (hollow size: 20 mm in width, 20 mm in thickness, and 11 mm in height) and oven cured at 55°C for 2 h. After removing the mold, the resulting chamber was further oven cured at 125°C for 15 min. A separately prepared PDMS sheet with a thickness of 90 µm was attached to the bottom of the chamber and oven cured at 125°C for 15 min. The PDMS sheet was then treated with the oxygen plasma through the mask according to the method described above. Cells were plated in the chamber supporting the micropatterned substrate and were stretched uniaxially using a custom-made stretcher to reach 10% strain. The stretched cells were incubated in the incubator for 1 h for subsequent immunostaining. 2.7 Image analysis

Immunofluorescence (IF) image of paxillin is divided by that of pY118-paxillin at each pixel using image analysis software ImageJ (NIH). The output is defined as Fractional IF signal, which displays the spatial distribution of the level of paxillin phosphorylation. To quantify the effect of exogenous stretch on Fractional IF signal, vicinities of the four corners of square-shaped cells are chosen as a region of interest (ROI). Each of the ROIs is further divided into two triangles: one is in parallel with the direction of stretch, and the other is in perpendicular to that. Thus, the former and latter represent the response of Fractional IF signal at stretched (S) and unstretched (U) regions, respectively. Each triangular region may have different numbers of pixels. Thus, the top 30 Fractional IF signal values are extracted at each region to represent the local response of paxillin phosphorylation to the microstretch. 2.8 Immunoblotting

Cells were solubilized in SDS sample buffer, boiled for 5 min, and fractionated on polyacrylamide gels. Samples were transferred onto polyvinylidene difluoride membranes (Millipore) for immunoblotting. Membranes were incubated with an antibody to paxillin (1:250, Cell Signaling Technology), pY118-paxillin (1:500, Cell Signaling Technology), and β-actin (1:5000, Applied Biological Materials). The primary antibodies were detected using HRP-conjugated goat anti-mouse (1:10000, Bio-Rad Laboratories) and goat anti-rabbit secondary antibodies (1:5000,

4



[DOI: 10.1299/jbse.16-00670]

Bio-Rad Laboratories) and enhanced chemiluminescence (Immobilon Western, Millipore). Fractional immunoblot (IB) signal is then obtained as the ratio of pY118-paxillin expression to the total paxillin expression at various conditions with or without inhibitors. 3. Results 3.1 Cell–substrate adhesions are localized at mechanically stressed subcellular regions in a cellular tension-dependent manner

We employed a square-shaped micropattern (with an edge length of 35 µm) to artificially define the morphology of U2OS cells (Fig. 1B). Square-shaped single cells extended/shrank lamellipodia; but, on average, they retained the defined square, as confirmed by a kymograph constructed along an inspected line drawn perpendicular to one of the edges (Fig. 1C). Stress fibers, visualized with fluorescent phalloidin for actin staining, are markedly observed along the edges of the squares together with p130Cas, a marker of focal adhesions, located at the termini of the stress fibers (Fig. 1D).

Our previous study demonstrated that the subcellular position of adhesion localization in micropatterned cells is accurately predicted by an FEM analysis that considers isotropic contraction (Deguchi et al., 2011). We therein found the presence of a positive correlation in the subcellular position of hot spots between the intracellular mechanical stress and adhesions. In the present study, we applied the same analytical approach into predicting the distribution of intracellular mechanical stresses within the square micropatterned cells (Fig. 2A). An isotropic contraction, which mimics NMII-driven cellular shortening (Deguchi et al., 2011), was given to an elastic, flat cuboid (i.e., cell model) whose displacement is not permitted along the interface between the cuboid and underlying, another elastic material (i.e., substrate) with a stiffer elastic modulus. The resulting distribution of von Mises stress (i.e., equivalent stress) showed that the cell is highly distorted along the periphery (Fig. 2B) and more markedly at the corners of the square (Fig. 2C).

Consistent with the predicted stress magnitude, immunostaining showed that paxillin-labeled cell–substrate adhesions were localized along the edges of square cells (Fig. 3A). Adhesions were particularly accumulated close to the corners to have huge clusters that extend over several micrometers. At the edges apart from the corners, small paxillin patches were instead observed. To confirm reproducibility of these observations, we superimposed totally n = 9 different cell images into one to make a single stack image. This stack image, scaled for intensity, visualizes how frequently paxillin emerges at the same subcellular regions. Thus, the presence of high-intensity spots at the corners assures the universal feature of paxillin localization.

Because cellular contractile forces are artificially focused on the corner of square micropatterns (Fig. 2C), the close correlation between the paxillin localization and mechanical stress suggests that paxillin is recruited to subcellular sites subjected to high stresses. To address this hypothesis, we treated cells with cytochalasin D (CytoD) that inhibits actin polymerization or blebbistatin that inhibits the NMII activity (with an inhibition of phosphate release in the ATPase cycle) to in turn abolish the endogenous cellular tension (Fig. 3B). A limited amount of paxillin-labeled adhesions still remained around the edges after the treatment with CytoD or blebbistatin, but they largely disassembled into small patches, thus supporting the positive relationship between paxillin localization and force. 3.2 Paxillin phosphorylation is negatively regulated by endogenous cellular tension

We examined the phosphorylation of paxillin at tyrosine 118 in micropatterned cells with immunostaining. pY118-paxillin was observed apparently in a distribution similar to paxillin, and a stack of originally n = 9 images assured the reproducibility (Fig. 3A). Analysis of the images of all paxillin and phospho-paxillin allows for distinguishing whether the high phosphorylation at the corners is due to merely the high accumulation of paxillin itself or due to position-specifically enhanced phosphorylation. Here paxillin itself is shown in pseudocolored green, while pY118-paxillin in pseudocolored red (Fig. 3B). This visualization indicated that huge adhesion clusters around the corners of control cells (DMSO) are relatively green in color, while small patches at the edges other than the corners are red. Magnified views on huge adhesions at the corners further demonstrated that marked phosphorylation occurs only at the very outer rims of the huge adhesions (Fig. 3B). In CytoD or blebbistatin-treated cells, in contrast, all paxillin-labeled small patches located near the edges were red, suggesting that they are largely phosphorylated.

5



[DOI: 10.1299/jbse.16-00670]

Fig. 3 Paxillin phosphorylation is negatively regulated by endogenous cellular tension. (A) Fluorescence images of paxillin and its phosphorylation at tyrosin residue 118. Single represents a single, representative cell. Stack represents totally n = 9 different fluorescence images stacked together as a single image and scaled for intensity. Lighter colors correspond to more frequent appearance of high fluorescence intensities. (B) Pharmacological effects on paxillin phosphorylation. DMSO represents vehicle control. Enlarged images of boxed areas in the left panels are shown in the right panels. (C) Fractional IF signal is displayed to evaluate the gradient of endogenous cellular tension. Lighter colors correspond to more phosphorylated paxillin intensities. Changes along the boxed area are shown in red with a linear regression in blue. All the cells are 35 µm in edge length.

Additionally we examined the ratio of pY118-paxillin to all paxillin, or Fractional IF signal (Fig. 3C). Paxillin in huge adhesion clusters was low in phosphorylation level (i.e., Fractional IF signal). The phosphorylation level is gradually elevated from the corner (i.e., highly tensed subcellular region) toward the center (i.e., less tensed subcellular region) of the edge of square cells. Collectively, these results suggest that the phosphorylation of paxillin accumulated in an NMII or cellular tension-dependent manner is negatively regulated by the endogenous force. 3.3 Paxillin exposed to exogenous stretch is less phosphorylated

To corroborate the above conclusion that phosphorylation of paxillin is negatively regulated by the endogenous cellular tension, we made micropatterns on a silicone chamber that is uniaxially stretchable using a linear slider. Square-micropatterned cells are thus subjected to an exogenous force (Fig. 4A). At 1h after the application of 10% stretch (i.e., a planar displacement of 3.5 µm was applied to cells with a size of 35x35-µm2), cells were fixed for double immunostaining of paxillin and pY118-paxillin to quantify Fractional IF signal.

Both merge image (Fig. 4B) and, more clearly, Fractional IF signal (Fig. 4C) showed that the adhesions that are in parallel with the stretch direction are significantly dephosphorylated than those that are orthogonal to the stretch direction. The parallel adhesions underwent the exogenous stretch more directly compared to the orthogonal ones. These results demonstrate that externally stretched cell–substrate adhesions (Fig. 4C, S) are more prone to paxillin dephosphorylation compared to externally unstretched cell–substrate adhesions (Fig. 4C, U), thus again suggesting the presence of the force-elicited negative regulation of phosphorylation.

A� B�

C�

Paxillin�St

ack�

Sing

le�

pY118-paxillin�Merge (Paxillin + pY118-paxillin)�

Cyto

D�DM

SO�

Fractional IF signal = –––––––––––– �pY118-paxillin�Paxillin�

*� **�

Frac

tiona

l IF

signa

l (A.

U.)�

Position�

*� **� Bleb

bist

atin�

Merge

Inte

nsity�

Fig.3�

6



[DOI: 10.1299/jbse.16-00670]

Fig. 4 Paxillin phosphorylation is negatively regulated by exogenous cellular tension. (A) The upper two represent schematic views of microstretch experiment in which square-shaped cells are subjected to 10% uniaxial stretch. The left edge (L) is unchanged in position, while the right (R) is displaced rightward (R’), resulting in that the cell length lc is elongated to 1.1lc only within the stretched direction. The middle represents an initial state of a cell under static culture with F-actin visualized. The lower two represent cells undergoing 10% stretch: note that cells are different between the middle and lower images. (B) Merge of paxillin and pY118-paxillin images, and its magnification regarding the four corners. The arrow represents the direction of stretch. (C) Fractional IF signal, in which darker colors correspond to more phosphorylated paxillin intensities. Dot-graph represents the top 30 pixel signals examined at each ROI of stretched (S) and unstretched (U) regions. The line segments show the means. Data are normalized by the mean of U. The asterisk represents a statistically significant difference (unpaired Student’s t-test, p < 0.001). All the cells are 35 µm in edge length. 3.4 The total phosphorylation level of paxillin is reduced with pharmacological inhibition of cellular tension

To further investigate the negative paxillin regulation by force, we performed immunoblotting on cell lysates with or without CytoD treatment. Quantification of the ratio of pY118-paxillin to total paxillin (i.e., Fractional IB signal) showed that CytoD reduced the level of the phosphorylation ratio to on average 65% of vehicle control (Fig. 5A). Thus, curiously, the immunoblot results are sharply in contrast to the immunostaining results, in the latter of which the phosphorylation level was rather elevated (Fig. 3).

There is a possibility that the unexpected response of paxillin phosphorylation to CytoD emerges because the induced actin polymerization may affect other signaling pathways in addition to the removal of cellular tension. We next used a more specific inhibitor blebbistatin that inhibits the release of inorganic phosphate in the actin–NMII ATP hydrolysis to consequently suppress the NMII-driven cellular tension (Fig. 5B). Furthermore, because kinases that are responsible for paxillin phosphorylation are FAK (Bellis et al., 1995; Schaller and Parsons, 1995) and Src (Thomas et al., 1995; Petch et al., 1995), those cell samples were also prepared with or without pharmacological inhibition on FAK

7



[DOI: 10.1299/jbse.16-00670]

Fig. 5 Immunoblotting suggests a positive regulation of paxillin phosphorylation by cellular tension. (A, B) Fractional IB signal with (+) or without (-) CytoD (A) and FAK II, Src I + PP1, and blebbistatin (B). Data are normalized by the vehicle control. Mean ± standard deviation. The data on the middle lane filled with black are not discussed here. and/or Src. Our results showed that the phosphorylation level was reduced to 75% of control by blebbistatin treatment. The reduction was enhanced in the presence of the small molecule inhibitor for FAK or Src, at similar levels, to 48% or 54%, respectively. Simultaneous inhibitions of FAK and Src did not further reduce the phosphorylation level. 4. Discussion 4.1 Micropatterning allows for separating mechanical and pharmacological effects

Tyrosine phosphorylation of cell–substrate adhesions is regulated mechanically as well as biochemically and has been implicated in regulation of adhesion turnover to control cell migration. In the present study, we addressed the molecular mechanisms regarding the phosphorylation of paxillin, an adhesion-associated adaptor protein, using a micropatterning technique. This experimental tool allows for defining the morphology of individual cells and is useful particularly in mechanobiology studies as it provides the cells with an artificial gradient in endogenous forces (Fig. 2). Specifically, cells exert traction forces at cell–substrate adhesions, which are originated from NMII-induced contraction, so that these endogenous forces (that balance with the traction forces) can be focused on particular subcellular portions owing to the arbitrary defined cell morphology. This is technically in contrast to the case using normal cell culture, in which endogenous forces are distributed in a spatially heterogeneous manner within cells, thus often complicating interpretation of data.

Here we used square-shaped micropatterns, with which we aimed at focusing the forces particularly at the corners of the squares. Indeed, large-sized cell–substrate adhesions, or so-called focal adhesions (FAs), were localized at the target subcellular regions. Pharmacological inhibition of the endogenous cellular tension led to disappearance in the huge adhesion clusters. These results suggest the presence of a positive feedback system in which mechanical force and adhesion accumulation enhance each other, as consistent with the finding obtained from previous studies using various morphologies of cell micropatterns (Théry et al., 2006; Deguchi et al., 2011). 4.2 Paxillin phosphorylation responds to cellular force differently between the niche and global

Paxillin in the huge adhesion clusters was predominantly unphosphorylated as visualized in immunostaining (Fig. 3B, C). The same protein associated with small adhesion patches that are so-called focal complexes (FCs), whose association to the ECM is insensitive to the cellular tension inhibitor (Fig. 3B), was by contrast highly phosphorylated (Fig. 4).

CytoD - +FAK II - + - + - + - +

Src I + PP1 - - + + - - + +Blebbistatin - - - - + + + +

A�B�

1

0

Rel

ativ

e ex

pres

sion

of

(pY1

18-p

axilli

n)/(t

otal

pax

illin)

2

1

0

Rel

ativ

e ex

pres

sion

of

(pY1

18-p

axilli

n)/(t

otal

pax

illin)

Frac

tiona

l IB

signa

l (A.

U.) �

Frac

tiona

l IB

signa

l (A.

U.) �

Fractional IB signal = –––––––––––– �pY118-paxillin�Paxillin�

IB: pY118-paxillin

IB: total paxillin

IB: β-actin

CytoD - +IB: pY118-paxillin

IB: total paxillin

IB: β-actin

FAK II - + - + - + - +Src I + PP1 - - + + - - + +Blebbistatin - - - - + + + +

Fig.5�

8



[DOI: 10.1299/jbse.16-00670]

It is not quite clear concerning the regulation of paxillin phosphorylation or more specifically whether the force induces selective recruitment of unphosphorylated paxillin to adhesions or induces its dephosphorylation at adhesions. Our immunostaining data on individual adhesions apparently suggest that the cellular tension negatively modulates the phosphorylation of paxillin (Fig. 3). Likewise, extracellular tension also modulated the phosphorylation negatively (Fig. 4). On the other hand, the immunoblot data that analyzed averaged cellular responses suggest that the pharmacological inhibition of cellular tension rather reduces the phosphorylation level per each paxillin (Fig. 5). Consistent with our results, the average of paxillin phosphorylation was previously reported to decrease upon an NMII inhibition with blebbistatin treatment (Pasapera et al., 2010). Thus, the adhesion response to NMII inhibition, or loss of the intrinsic cellular tension, is the opposite to each other between the niche and global behaviors. 4.3 Model of paxillin phosphorylation to explain the apparent discrepancy between the immunostaining and immunoblot data

To reconcile the opposite responses, we submit a minimal model of paxillin accumulation and phosphorylation as follows, which is based not only on the present results but also previously reported quantitative data (Fig. 6): First, paxillin has two subpopulations within cells; i.e., one is diffusive (Fig. 6, light green), and the other is associated with anchored adhesions (Fig. 6, light red). Cell–substrate adhesions have mainly two forms; i.e., one is small FCs that are induced independently of NMII-generated cellular tension (Fig. 6, upper panel; Cellular tension –), and the other is large FAs that are induced by cellular tension (Fig. 6, lower panel; Cellular tension +). Based on this simple classification, we distinguish between the immunostaining (subcellular) and immunoblot (whole-cell) data.

Recruitment of paxillin to adhesions is reportedly independent of cellular tension (Zaidel-Bar et al., 2007); thus,

Fig. 6 Pie chart model to describe the opposite mechano-response of paxillin phosphorylation between immunostaining and immunoblot. See the text for details.

Cellular tension –(X– << 1-X–)

Diffusive paxillin

Anchored paxillin at FCs

Cellular tension +(X+ >> 1-X+)

Diffusive paxillin

Anchored paxillin at FAs

Diffusive paxillin

Anchored paxillin at FCs

1-X–

X–

X–

1-X–

=Y–

1-Y–

Phosphorylated paxillin

Unphosphorylated paxillin=

Z–

1-Z–

1-X+

X+

X+

1-Y+Y+

=

=

Anchored paxillin at FAs

Diffusive paxillin

1-X+

Z+

1-Z+

Total phosphorylated paxillin

[Fractional IF signal]

Total paxillin

= Y (at adhesions) or Z ≈ 0 (at cytoplasm)�

= XY+(1-X)Z ≈ XY [Fractional IB signal]

or

X = Anchored paxillin

Total paxillin

Y =Anchored phosphorylated paxillin

Total anchored paxillin

Z =Diffusive phosphorylated paxillin

Total diffusive paxillin

Phosphorylation in cellular tension +

Phosphorylation in cellular tension –

Y+= Y–

[Immunostaining]

= [Immunoblot]X+ Y+

X– Y– Fig.6

9



[DOI: 10.1299/jbse.16-00670]

paxillin is nearly universally associated with any types of adhesions regardless of FAs and FCs (and is consequently often used as an adhesion marker (Kuo et al., 2011)). Only limited amounts of FCs with a small size compared to FAs are observed within tension-inhibited cells, suggesting that the majority of paxillin in those tension-free conditions exist in a diffusive state (Fig. 6, upper panel). This is consistent with Kuo et al. (2011) that separately analyzed isolated adhesions and cell lysates, in which paxillin in blebbistatin-treated cells was located predominantly in the cytoplasm other than at adhesions. The phosphorylation of paxillin is mediated mainly by FAK and Src at adhesions (Turner, 2000; Webb et al., 2004). It is thus likely that the diffusive paxillin in the absence of cellular tension is unphosphorylated because the diffusive molecules are seldom at the FAK/Src-containing scaffolds. Thus, in the absence of cellular tension, paxillin would be predominantly diffusive in the cytoplasm (Fig. 6, upper panel, light green) and unphosphorylated (Fig. 6, upper panel, gray).

How about the case in the presence of cellular tension? Larger amounts of paxillin compared to the case in the absence of tension would become anchored to adhesion clusters because, upon the increased NMII activity, the adhesion size is increased to mature from FCs to FAs (Fig. 6, lower panel) (Gardel et al., 2008). This view that paxillin under tensed adhesions is predominantly anchored is supported by the study of Kuo and colleagues in which the relative paxillin expression was analyzed separately for isolated adhesion fractions and for other cell body fractions to find that 79% (= 3.74/(3.74+1)×100%; Fig. 1c of the paper) of total paxillin were present together with the isolated FAs.

To discuss paxillin phosphorylation at adhesions regardless of FCs and FAs, we consider FAK/Src recruitment as well because they are responsible for the phosphorylation. Pasapera et al. (2010) reported that both adhesion recruitment and tyrosine phosphorylation of FAK at tyrosine 397 proceed under cellular tension. Other studies reported that recruitment of FAK to adhesions is independent of cellular tension, and FAK undergoes phosphorylation at tyrosine 397 upon integrin engagement with the ECM (Schoenwaelder and Burridge, 1999; Zaidel-Bar et al., 2007; DeMali et al., 2003). Here we adopt the force-independent regulation of FAK because the majority of paxillin is highly phosphorylated in the absence of cellular tension as observed in the present immunostaining (Fig. 3) as well as elsewhere (Deguchi et al., 2011). Thus, it is likely that the activated FAK (together with Src (Webb et al., 2004)) phosphorylates the adaptor paxillin to act as scaffolds for phosphotyrosine-binding SH2 domain–containing proteins and consequently enhances maturation of the adhesions. Diffusive paxillin, which is mostly unphosphorylated, then starts to localize at FAK/Src-associated adhesions and then undergo the kinases-mediated phosphorylation.

Previously, negative regulation of paxillin phosphorylation by cellular tension was reported based on immunostaining with consideration on the involvement of FAK and Src (Zaidel-Bar et al., 2007). In the present study, we observed similar immunostaining patterns; notably, because we used micropatterns, the effect of force gradient was visualized with profound accumulation of FAs at the corners (Fig. 3). In the absence of cellular tension, paxillin is phosphorylated within small-sized FCs as the antibodies to paxillin and pY118-paxillin are precisely colocalized. Meanwhile, in the presence of cellular tension, colocalization of these antibodies was also seen all over FAs, but the phosphorylation level was locally high along the outer rims of the corners. At the inner parts within the individual FAs, the phosphorylation level is almost spatially uniform. The spatially concentrated phospho-paxillin at the marginal parts of the adhesions may be reasonable given that pY118-paxillin reportedly enhances lamellipodial protrusion at the cell periphery [Zaidel-Bar et al., 2007]. The spatially uniform paxillin of moderate phosphorylation level in the FAs other than the periphery suggests that cellular tension would negatively but moderately regulate the tyrosine phosphorylation.

Taken together, a simple mathematical expression is drawn as follows (Fig. 6). Here we denote X, Y, and Z as the proportions of anchored paxillin to total paxillin, anchored phospho-paxillin to total anchored paxillin, and diffusive phospho-paxillin to total diffusive paxillin, respectively. In immunostaining, we have discussed local change in the phosphorylation with a focus only on anchored paxillin; consequently, the ratio of total phospho-paxillin to total paxillin (i.e., Fractional IF signal) is expressed simply as Y. At the cytoplasm except for the adhesions, the same ratio (i.e., Fractional IF signal) is expressed as Z; but, as almost no paxillin phosphorylation was seen there, Z should nearly be zero. In immunoblot, the same ratio of total phospho-paxillin to total paxillin (i.e., Fractional IB signal) is represented by XY+(1-X)Z because not only anchored but also all paxillin in cell lysates are analyzed. Again with Z ≈ 0, it is reduced to XY. With a subscript of either + or – in the presence or absence of cellular tension, the ratio of the phosphorylation levels at the tensed cells to the tension-free cells is described as Y+/Y– for immunostaining data and (X+/X–)×(Y+/Y–) for immunoblot data.

We already described in this section that X+ is ~0.79 based the data of Waterman group (Kuo et al., 2011). Using

10



[DOI: 10.1299/jbse.16-00670]

immunoblot data reported by the same group (Fig. 6A of the paper of Pasapera et al., 2010), (X+/X–)×(Y+/Y–) is on average 4.2 (=1/0.24). We could not find direct data on X– in reports from the same group. Instead, according to a study on myotubes reported from Discher group (Sen et al., 2011), the immunostaining intensity of paxillin (perhaps in whole individual cells) decreased with blebbistatin treatment to ~22% of the original; similar tendencies were also seen in Schiller et al. (2011). Thus, we estimate that X+/X– is ~4.5 (=100/22). From these values, we now have an average Y+/Y– of 0.93 (=4.2/4.5), and X– is ~0.18 (= 0.79/4.5). These representative values indicate that, apparently, the phosphorylation level of paxillin decreases with removal of cellular tension in immunoblot (i.e., the reciprocal of (X+/X–)×(Y+/Y–) is 0.24 (= 1/4.2)), but it nevertheless increases with removal of cellular tension in immunostaining (i.e., the reciprocal of Y+/Y– is 1.1 (=1/0.93)). Thus, our simple model is able to explain the opposite response between the immunoblot and immunostaining. Note that assumptions concerning the unaddressed question, i.e., whether force induces selective recruitment of unphosphorylated paxillin or induces dephosphorylation of phosphopaxillin, are dispensable in the present model. 5. Conclusions

We submit a minimal model to address the essence of the apparent discrepancy in paxillin phosphorylation response to cellular tension analyzed by immunostaining and immunoblotting. In this model, the ratio of the phosphorylation levels at control cells to tension-removed cells is described by the following simple formulas: Y+/Y– for immunostaining data, and (X+/X–)×(Y+/Y–) for immunoblotting. The majority of paxillin are anchored to adhesions in control cell (X+), whereas not in tension-free cell (X–). For such anchored paxillin, they are prone to be phosphorylated at tension-free (Y–) compared to control (Y+). If the fraction of X+/X– is higher in magnitude than Y–/Y+, then the former becomes predominant to potentially reverse the mechano-response between immunostaining (Y+/Y–) and immunoblotting ((X+/X–)×(Y+/Y–)). Our estimates from experiments suggest that indeed X+/X– is much higher than unity, whereas Y+/Y– is slightly below unity. Thus, behaviors predicted from the model will be qualitatively consistent with the actual response of paxillin phosphorylation to the inhibition of CytoD or blebbistatin. The model also implies that similar opposite response between immunostaining and immunoblot can generally emerge if the activation of a protein of interest is negatively regulated at a local place while the trigger for the activation also alters the extent of the assembly of the protein to the particular local place. Acknowledgments

The authors thank Atsuki Kubotani and Hitomi Onodera for their technical assistance. This study was partly supported by KAKENHI grants (no. 20001007, 24680049, 25242039, 15H03004, 16K12872, and 16H05907). References Bellis, S.L., Miller, J.T., Turner, C.E., Characterization of tyrosine phosphorylation of paxillin in vitro by focal

adhesion kinase. J. Biol. Chem., Vol. 270 (1995), pp. 17437-17441. Chen, C.S., Mechanotransduction – a field pulling together? J. Cell Sci., Vol. 121 (2008), pp. 3285-3292. Deguchi, S., Ohashi, T., Sato, M., Tensile properties of single stress fibers isolated from cultured vascular smooth

muscle cells. J. Biomech., Vol. 39 (2006), pp. 2603-2610. Deguchi, S., Sato, M., Biomechanical properties of actin stress fibers of non-motile cells. Biorheology, Vol. 46 (2009),

pp. 93-105. Deguchi, S., Matsui, T.S., Iio, K., The position and size of individual focal adhesions are determined by intracellular

stress-dependent positive regulation. Cytoskeleton, Vol. 68 (2011), pp. 639-651. DeMali, K.A., Wennerberg, K., Burridge, K., Integrin signaling to the actin cytoskeleton. Curr. Opin. Cell Biol., Vol. 15

(2003), pp. 572-582. Gardel, M.L., Sabass, B., Ji, Li, Danuser, G., Schwarz, U.S., Waterman, C.M., Traction stress in focal adhesions

correlates biphasically with actin retrograde flow speed. J. Cell Biol., Vol. 183 (2008), pp. 999-1005. Kuo, J.C., Han, X., Hsiao, C.T., Yates, J.R. 3rd, Waterman, C.M., Analysis of the myosin-II-responsive focal adhesion

proteome reveals a role for β-Pix in negative regulation of focal adhesion maturation. Nat. Cell Biol., Vol. 13 (2011), pp. 383-393.

Pasapera, A.M., Schneider, I.C., Rericha, E., Schlaepfer, D.D., Waterman, C.M., Myosin II activity regulates vinculin recruitment to focal adhesions through FAK-mediated paxillin phosphorylation. J. Cell Biol., Vol. 188 (2010), pp.

11



[DOI: 10.1299/jbse.16-00670]

877-890. Petch, L.A., Bockholt, S.M., Bouton, A., Parsons, J.T., Burridge, K., Adhesion-induced tyrosine phosphorylation of the

p130 src substrate. J. Cell Sci., Vol. 108 (1995), pp. 1371-1379. Petit, V., Boyer, B., Lentz, D., Turner, C.E., Thiery, J.P., Vallés, A.M., Phosphorylation of tyrosine residues 31 and 118

on paxillin regulates cell migration through an association with CRK in NBT-II cells. J. Cell Biol., Vol. 148 (2000), pp. 957-970.

Saito, A.C., Matsui, T.S., Ohishi, T., Sato, M., Deguchi, S., Contact guidance of smooth muscle cells is associated with tension-mediated adhesion maturation. Exp. Cell Res., Vol. 327 (2014), pp. 1-11.

Schaller, M.D., Parsons, J.T., pp125FAK-dependent tyrosine phosphorylation of paxillin creates a high-affinity binding site for Crk. Mol. Cell Biol., Vol. 15 (1995), pp. 2635-2645.

Schiller, H.B., Friedel, C.C., Boulegue, C., Fässler, R., Quantitative proteomics of the integrin adhesome show a myosin II-dependent recruitment of LIM domain proteins. EMBO Rep., Vol. 12 (2011), pp. 259-266.

Schoenwaelder, S.M., Burridge, K., Bidirectional signaling between the cytoskeleton and integrins. Curr. Opin. Cell Biol., Vol. 11 (1999), pp. 274-286.

Sen, S., Tewari, M., Zajac, A., Barton, E., Sweeney, H.L., Discher, D.E., Upregulation of paxillin and focal adhesion signaling follows Dystroglycan Complex deletions and promotes a hypertensive state of differentiation. Eur. J. Cell Biol., Vol. 90 (2011), pp. 249-260.

Théry, M., Pépin, A., Dressaire, E., Chen, Y., Bornens, M., Cell distribution of stress fibres in response to the geometry of the adhesive environment. Cell Motil. Cytoskeleton, Vol. 63 (2006), pp. 341-355.

Thomas, S.M., Soriano, P., Inamoto, A., Specific and redundant roles of Src and Fyn in organizing the cytoskeleton. Nature, Vol. 376 (1995), pp. 267-271.

Turner, C.E., Paxillin interactions. J. Cell Sci., Vol. 113 (2000), pp. 4139-4140. Webb, D.J., Donais, K., Whitmore, L.A., Thomas, S.M., Turner, C.E., Parsons JT, Horwitz AF., FAK-Src signalling

through paxillin, ERK and MLCK regulates adhesion disassembly. Nat. Cell Biol., Vol. 6 (2004), pp. 154-161. Zaidel-Bar, R., Ballestrem, C., Kam, Z., Geiger, B., Early molecular events in the assembly of matrix adhesions at the

leading edge of migrating cells. J. Cell Sci., Vol. 116 (2003), pp. 4605-4613. Zaidel-Bar, R., Milo, R., Kam, Z., Geiger, B., A paxillin tyrosine phosphorylation switch regulates the assembly and

form of cell-matrix adhesions. J. Cell Sci., Vol. 120 (2007), pp. 137-148.

12

Documents

Journal of Biomechanical Science and Engineering The