FAK, Vinculin, and Talin Control Mechanosensitive YAP Nuclear Localization

Abstract

Focal adhesions (FAs) are nanoscale complexes containing clustered integrin receptors and intracellular structural and signaling proteins that function as principal sites of mechanotransduction in part via promoting the nuclear translocation and activation of the transcriptional coactivator yes-associated protein (YAP). Knockdown of FA proteins such as focal adhesion kinase (FAK), talin, and vinculin can prevent YAP nuclear localization. However, the mechanism(s) of action remain poorly understood. Herein, we investigated the role of different functional domains in vinculin, talin, and FAK in regulating YAP nuclear localization. Using genetic or pharmacological inhibition of fibroblasts and human mesenchymal stem cells (hMSCs) adhering to deformable substrates, we find that disruption of vinculin-talin binding versus talin-FAK binding reduces YAP nuclear localization and transcriptional activity via different mechanisms. Disruption of vinculin-talin binding or knockdown of talin-1 reduces nuclear size, traction forces, and YAP nuclear localization. In contrast, disruption of the talin binding site on FAK or elimination of FAK catalytic activity did not alter nuclear size yet still prevented YAP nuclear localization and activity. These data support both nuclear tension-dependent and independent models for matrix stiffness-regulated YAP nuclear localization. Our results highlight the importance of vinculin-talin-FAK interactions at FAs of adherent cells, controlling YAP nuclear localization and activity.

Introduction

Cells perceive mechanical cues in their environment through interactions with other cells and the extracellular matrix (ECM) [1, 2]. Understanding cell-ECM interactions is important to biomaterials design, tissue engineering, and regenerative medicine [3, 4]. Adhesive interactions between cells and the ECM are primarily mediated by integrins [5]. As a response to mechanical or biochemical signals, integrins undergo conformational changes, transitioning from a bent to an extended state, which enhance their ligand binding affinity [5]. Integrins then cluster, which leads to integrin-mediated adhesion complexes maturing into focal adhesion (FA) nanocomplexes. This process involves the recruitment of many structural and signaling proteins and actin polymerization components [6–8]. FAs link the ECM and the actin cytoskeleton to transmit bidirectional contractile forces via alterations in actin polymerization and actomyosin contractility. This positions FAs as critical mechanosensing structures.

Studies utilizing contractility inhibitors, deformable substrates, and optical tweezers have provided mechanistic insights into the assembly and turnover of FAs, as well as the identification of important mechanosensitive FA proteins [7, 9]. Among these proteins, talin, vinculin, and focal adhesion kinase (FAK) play central roles in FA functions. Talin and vinculin are key structural proteins that participate in force transmission through binding to the actin cytoskeleton. Talin bridges integrins and actin filaments [10]. By binding to integrins, talin modulates integrin conformation to increase binding affinity for ECM ligands. Talin then initiates mechanotransduction by undergoing a conformational change in its rod domain at a specific tension threshold [11]. These conformational changes in talin expose binding sites for other FA proteins, such as paxillin and vinculin, which contribute to FA growth [10–12]. Vinculin binds to residues in the talin rod domain, and this facilitates actin filament association with integrin-talin complexes [13]. By doing so, vinculin promotes integrin clustering, FA growth, and force generation while exposing additional cryptic binding domains on talin [13–16].

In contrast to talin and vinculin, FAK functions as a signaling scaffolding protein tyrosine kinase. FAK can impact cytoskeletal reorganization in part by binding and tyrosine phosphorylation of paxillin, binding to Rho GEFs and Rho GAPs, or activation of signaling cascades controlling the actin cytoskeleton [17, 18]. FAK plays a central role in FA turnover, phosphorylation of cadherins, and enhancement of growth factor receptor signaling [9]. Through integrin-stimulated FAK autophosphorylation and the recruitment of c-Src kinase into a multiprotein signaling complex with FAK, FAK converts mechanical stimuli into downstream signals [9, 19]. Moreover, FAK and talin are not only binding partners but can recruit each other to nascent adhesions [20, 21]. Together, talin, vinculin, and FAK play critical regulatory roles and serve as key players in cell mechano-transduction.

Yes-associated protein (YAP) is a transcriptional coregulator that translates mechanical and cell area/spatial cues into changes in gene expression [22–25]. YAP regulates various biological responses, including proliferation, survival, differentiation, myofibroblast activation, and matrix remodeling [23, 26–29]. YAP transcriptional activity is partially regulated by cytoskeletal structure and tension, which impact YAP nuclear translocation and activation [24, 30]. Several models have been proposed for how YAP functions as a mechanosensitive transcriptional regulator. Elosegui-Artola et al. showed that cytoskeletal tension deforms the nucleus (as measured by changes in nuclear area) resulting in nuclear pore opening and increased influx of YAP into the nucleus to activate gene expression [31]. Meng et al. found that RAP2 mediates YAP nuclear localization through downstream inhibition of RhoA and activation of LATS1/2 kinases [32]. Alternatively, FAK-Src signaling connects YAP-dependent colonic epithelium reprogramming with ECM remodeling [33], a FAK-YAP-mTOR signaling axis regulates stem cell renewal during development [34], a FAK-P130cas-YAP signaling complex regulates radiosensitivity in non-small cell lung cancer [35], and FAK activates YAP via MOB1 phosphorylation resulting in Hippo pathway inhibition [36]. The interplay between these models is yet to be understood. Vinculin, talin, and FAK are all important to YAP-mediated transcriptional regulation, as vinculin, talin, or FAK knockout prevents YAP nuclear localization and transcriptional activity [32, 37–39]. However, it is not clear how interactions among these FA proteins influence YAP nuclear localization.

Here, we investigated the effects of perturbing functional domains and interactions between vinculin, talin, and FAK on the nuclear localization and transcriptional activity of YAP in fibroblasts and mesenchymal stromal cells adhering to deformable substrates. Using pharmacological inhibitors, short hairpin RNAs, and mutations known to modulate key FA proteins, we find that disrupting talin-vinculin or talin-FAK binding impaired nuclear size and YAP nuclear localization. Deleting vinculin, disrupting talin-vinculin interactions, knocking down talin-1, or preventing FAK Y397 phosphorylation also reduced nuclear size and YAP nuclear localization. However, disrupting the talin binding site on FAK or blocking intrinsic FAK activity prevented YAP activation but did not alter nuclear size. Taken together, our results support both nuclear tension-dependent and independent models for matrix stiffness-regulated YAP nuclear localization and transcriptional activation on deformable substrates.

Materials and Methods

Cells and Reagents

Mouse embryonic fibroblasts (MEFs) expressing enhanced green fluorescent protein (eGFP)-vinculin constructs were generated by transducing vinculin-null MEFs with retroviral constructs containing eGFP-vinculin WT or eGFP-vinculin A50I, followed by flow cytometry sorting for eGFP expression as described [14]. MEFs expressing eGFP-FAK WT, eGFP-FAK Y397F, eGFP-FAK K454R, and eGFP-FAK E1015A were derived by transducing FAK-null MEFs (isolated from p53−/− mice [40]) with lentiviral constructs containing FAK point mutations and sorted for eGFP expression as pooled cell populations [21]. Primary FAK K454R fibroblasts were isolated from knocked-in mice as described [41].

Talin-1 was depleted in NIH3T3 fibroblasts using talin-1 shRNA lentivirus (VectorBuilder). Briefly, 3T3 fibroblasts were transduced with anti-talin-1 or scrambled shRNA in DMEM containing 8 μg/μL polybrene, centrifuged at 1200 × g for 30 minutes in a swinging bucket rotor, and then incubated for 24 hours at 37 °C and 5% CO₂. Media was changed, and pooled cell populations were obtained by sorting for red fluorescent protein (RFP) co-expressed by the lentivirus. MEFs and NIH3T3 fibroblasts were maintained in DMEM containing L-glutamine (Gibco), 10% fetal bovine serum (FBS) (ATCC), 1% sodium pyruvate (Corning), and 1% penicillin-streptomycin (Corning).

Human mesenchymal stromal cells (hMSCs) were acquired from the NIH Resource Center at Texas A&M University (TAMU; College Station, TX). hMSCs were obtained from healthy, willing participants via bone marrow aspirate under IRB-approved protocols and isolated by plastic adherence. Cells were certified as hMSCs in accordance with ISCT standards by surface marker and differentiation characterization by the manufacturer [42]. hMSCs were cultured in αMEM (Gibco), 16.5% MSC qualified FBS (Gibco), 2–4 mM of L-glutamine (Corning), 100 U/mL of penicillin (Corning), 100 μg/mL streptomycin (Corning).

Micropillar Array Devices (mPADs)

mPADs were fabricated using polydimethylsiloxane (PDMS) replica molding from silicon masters [43, 44]. To make negative molds, 1:10 PDMS prepolymer (Dow Chemical) was cast on silicon masters in a foil-lined dish. The PDMS was baked at 110 °C for 1 hour. The negative molds were carefully peeled from the masters. Excess PDMS was trimmed from the negative molds, then the molds were exposed to an oxygen plasma for 5 seconds (Plasma-Preen; Terra Universal) and silanized for 4 hours under vacuum with 1–2 drops of trichloro(1H,1H,2H,2H-perfluorooctyl)silane (Sigma-Aldrich). A 1:10 PDMS prepolymer droplet was placed on each newly silanized negative mold. The covered negative molds were degassed in a vacuum desiccator for 25 minutes, and each mold was flipped over onto plasma-cleaned, circular 25 mm diameter, #1 glass coverslips using curved tip tweezers. Next, mPADs were baked at 110 °C for 20 hours. After being removed from the oven and cooled, mPADs were soaked in ethanol for 5 minutes, and the negative molds were removed with tweezers to prevent shearing of posts. mPADs were placed face down into another dish of ethanol and sonicated for 5 minutes to recover collapsed posts. mPADs were super-critically dried (Samdri-PVT-3D; Tousimis) with liquid CO₂ to preserve post integrity.

Microcontact Printing and Cell Seeding

Flat slabs of 1:20 PDMS were cut to the size of mPADs, sonicated in ethanol, and dried with a stream of N₂. 100 μL of a solution of fibronectin (Gibco, 50 μg/mL) in phosphate buffered saline (PBS) (Corning) for YAP nuclear localization studies or fibronectin (50 μg/mL) and Alexa Fluor 647 (AF647)-labeled fibrinogen (Invitrogen, 27 μg/mL) in PBS for force measurement studies was pipetted onto each stamp and left to incubate for 1 hour at room temperature. AF647-labeled fibrinogen was used only as a fiduciary marker to track micropost top deflections. During this incubation, mPADs were trimmed of excess PDMS to create a flat surface for stamping, then surface-oxidized for 10 minutes (UVO-Model 342; Jelight). After incubation, the fibronectin/fibrinogen-coated stamps were submerged in dH₂O and dried with a stream of N₂. The fibronectin/fibrinogen-coated surface was placed into contact with the mPAD surface for 30 seconds. The stamps were then removed in ethanol, and the inked mPADs were placed in PBS and moved into a tissue culture hood. mPADs were passivated with 0.2% Pluronics F-127 (Sigma- Aldrich) solution for 1 hour at room temperature. This solution was replaced with PBS. Cells were seeded on mPADs in 2 mL at 20,000 cells/mL for MEFs and 10,000 cells/mL for hMSCs (40,000 cells/mPAD for MEFs and NIH3T3 cells, 20,000 cells/mPAD for hMSCs). After 15 minutes, unattached cells were aspirated, and the media was replaced with 2 mL of fresh media. This procedure results in a cell density of 6000 cells/cm² for MEFs and NIH3T3 cells and 1600 cells/cm² for hMSCs for each device. Cells were allowed to spread overnight.

Pharmacological Inhibition

Cells were treated with inhibitors or vehicle control (dimethylsulfoxide, DMSO). Blebbistatin (Sigma) was used at 50 μM for 2 hours. PF-573228 (EMD Millipore) was used at 2 μM for 2 hours. Cells were treated with Y-27632 (STEMCELL Technologies) at 10 μM. PY-60 (MedChemExpress) was used at 10 μM, and verteporfin (Tocris) at 2 μM. For Y-27632, PY-60, and verteporfin, cells were treated for 72 hours. For LATS1/2 inhibition, cells were treated with TRULI (MedChemExpress) at 10 μM for 24 hours. The DMSO volume fraction for all inhibitors was 0.1% (v/v).

Immunofluorescence Staining

mPADs were fixed in a warm 1:1 mixture of cytoskeleton-stabilizing buffer (CSK buffer, pH 7.0: 0.5% Triton X-100 (VWR), 10 mM PIPES (Alfa Aesar) buffer, 50 mM NaCl (Fisher Scientific), 150 mM sucrose (Fisher Scientific), 3 mM MgCl2 (Quality Biological), Halt^™ protease inhibitor cocktail (ThermoFisher, 1:250 dilution)), Halt^™ phosphatase inhibitor cocktail (ThermoFisher, 1:400 dilution)) and 10% paraformaldehyde (Electron Microscopy Sciences) diluted in PBS, for 10 minutes at 37 °C. Cells were then permeabilized in CSK buffer with 0.5% Triton X-100 for 5 minutes, incubated in 0.1 M glycine (Amresco) to quench free aldehydes for 5 minutes, blocked in 33% goat serum (Gibco) in PBS for 1 hour, and incubated with primary antibodies at 4 °C in PBS with 33% goat serum and 0.02% Tween-20 (Sigma-Aldrich). On the following day, samples were incubated with fluorescent dye-labeled secondary antibodies in PBS with 33% goat serum and 0.02% Tween-20 at room temperature for 1 hour. Samples were stained with DAPI for 5 minutes (NucBlue^™, ThermoFisher). In between the primary stain, secondary stain, DAPI stain, and imaging, samples were washed via rocking in wash buffer (PBS, 0.02% Tween-20) 3 times for 5 minutes each at room temperature. The following antibodies were used: mouse isotype control IgG_2a (Santa Cruz, sc-3878; 1:100), mouse anti-YAP IgG_2a (Santa Cruz, sc-101199; 1:200), goat anti-mouse IgG-TRITC (Abcam, ab6786; 1:200), goat anti-mouse IgG-AF647 (Abcam, ab150115; 1:200).

Western Blotting

Cells were washed with ice-cold PBS containing calcium and magnesium and lysed in cold radioimmunoprecipitation assay buffer (Sigma-Aldrich) supplemented with Pierce Protease Inhibitor (ThermoFisher) and Phosphatase Inhibitor (ThermoFisher) cocktails for 15 minutes. Lysates were collected via scrapping into chilled microcentrifuge tubes, sonicated briefly, incubated at 4 °C for 15 minutes, and then centrifugated at 14,000 × g for 15 minutes. Lysate protein concentration was determined using the Pierce 660nm Protein Assay kit (ThermoFisher) with bovine serum albumin as a standard. Equal amounts of protein were boiled (95 °C) in Laemmli sample buffer (2% SDS, 10% glycerol, 100 mM DTT, 60 mM Tris·HCl (pH 6.8), and 0.001% bromophenol blue) for 5 minutes and separated by SDS-PAGE. Proteins were transferred by electrophoresis onto 0.45 μm PVDF membranes (EMD Millipore). All blocking, washing, and staining steps were done on a tube roller. Transferred membranes were blocked with LI-COR Blocking Buffer (LI-COR Biosciences) for 1 hour at room temperature. Membranes underwent overnight staining with primary antibodies at 4 °C. Staining with secondary antibodies was done at room temperature for 1 hour. In between staining and blocking steps, membranes were washed (PBS, 0.02% Tween-20) 3 times for 5 minutes and were imaged with an LI-COR Odyssey Imager (LI-COR Biosciences). The following antibodies were used: mouse anti-talin IgG₁ (Sigma, 8d4; 1:200), rabbit anti-β-actin IgG (Cell Signaling, D6A8, 1:100), goat anti-rabbit-IgG-IRDye680RD (LI-COR Biosciences, 925–68071; 1:10000), and goat anti-mouse- IgG-IRDye800CW (LI-COR Biosciences, 926–32210; 1:10000).

Hydrogel-functionalized coverslips

Glass coverslips (30 mm) were functionalized with thiol groups by incubating in 5% v/v (3-mercaptopropyl) trimethoxysilane (Sigma Aldrich) and toluene (VWR) under vacuum overnight. After washing with toluene and isopropyl alcohol, the coverslips were dried with N₂, baked for 2 hours at 110°C, and then stored protected from light. Four-arm polyethylene glycol amide norbornene (5 kDa, PEG4aNB, JenKem) was reconstituted in PBS containing 20 mM HEPES (pH 7.4). Cell adhesive peptide, cycloRGDFC (Biosynth), was dissolved in PBS containing 20 mM HEPES. Crosslinker, 1,4-dithiothreitol (DTT, Sigma), was dissolved in PBS containing 20 mM HEPES. Photoinitiator, lithium phenyl-2,4,6-trimethylbenzoylphophinate (LAP, Tokyo Chemical Industry), was dissolved in PBS containing 20 mM HEPES. Every component except PEG4aNB was sterile filtered; for PEG4aNB the solvent was sterilized before dissolving. To make the gel solution, dissolved PEG4aNB, cycloRGDFC, DTT, and LAP were mixed together. Either 10 μL droplets or 100 μL droplets were placed on parafilm. The 100 μL droplets were sandwiched between a thiolated coverslip and parafilm. The gel solution droplets were polymerized using a UV-A oven (ELC-500 UV Curing Chamber, Electro-Lite Corporation) for 30 seconds. Hydrogels were allowed to swell in PBS or media overnight at 37°C. For immunofluorescence staining, NIH3T3 fibroblasts were seeded overnight at 2000 cells/cm².

Rheological properties were measured using a cone and plate rheometer (2° cone with 10 mm diameter, MCR 302, Anton-Paar). Storage and loss modulus measurements were obtained by averaging the storage modulus over an oscillatory frequency range of 1–10 Hz at 2% strain at 37 °C (the linear viscoelastic range was determined using oscillatory strain amplitude sweep).

YAP Reporter Plasmid and Transfection

The YAP reporter plasmid pCDNA-eCFP-5xMCAT(SV)-49-(N)mCherry was derived from pGL3–5xMCAT(SV)-49-Luciferase [45–47]. pCDNA-ZSgreen-5xMCAT(SV)-49-(N)mCherry, a gift from Dr. John Lamar, was modified by replacing ZSgreen with eCFP. In the YAP reporter, mCherry is expressed via a YAP responsive promoter (5xMCAT), and eCFP is constitutively driven by the SV40 promoter. Plasmid transfection was accomplished by mixing Opti-MEM I Reduced Serum Medium (Thermo Fisher), endotoxin-free plasmid DNA, and TransIT-X2 reagent (Mirus) at room temperature with a 4:1 ratio of TransIT-X2 reagent volume (μL) to plasmid DNA mass (μg) for 20 min. Transfection complexes were added dropwise onto cells for 16 h at 37°C.

YAP Activity Assay

Tissue culture plastic was incubated with fibronectin (10 μg/mL) for 30 minutes at room temperature and washed once with PBS. Cells cultured onto fibronectin-coated tissue culture plastic or PEG hydrogels attached to thiolated glass coverslips were seeded overnight at 1500 cells/cm² for FAK-null MEFs expressing GFP-FAK WT or GFP-FAK E1015A, 2000 cells/cm² for NIH3T3 cells, and 2500 cells/cm² for FAK-null parental MEFs or FAK-null MEFs expressing GFP-FAK K454R and GFP-FAK Y397F. On day 1, cells were cultured with either media or media containing inhibitors. On day 2, the cells were transfected with plasmid. On day 3, cells were fed again with either media or media containing inhibitors. On day 4, cells were lifted off the plate, using 2 mL/well of 0.25% trypsin with 2.21 mM EDTA (Corning), suspended in Hanks’ Balanced Salt Solution (Gibco) with 2% FBS, and passed through a 70 μm filter into a FACS tube and placed on ice. Flow cytometry was conducted using a Cytoflex S Flow Cytometer (Beckman Coulter) and was analyzed using FCS Express v6 (De Novo Software, Glendale, CA). The following gating strategy was implemented (Figure S1). The cell population was selected from forward scatter (FS) vs. side scatter (SS) plots. Singlets were selected from FS height vs. FS area plots, and singlets were selected again from SS height vs. SS area plots. Transfected cells were then selected from an eCFP area vs. SS area plot using a non-transfected control. Based on these selected cells a YAP activity ratio was calculated using Equation 1.

$$\mathit{\text{YAP}}\mathit{\text{Activity}}\mathit{\text{Ratio}}\text{=}\frac{\mathit{\text{Geometric}}\mathit{\text{mCherry}}\mathit{\text{MFI}}}{\mathit{\text{Geometric}}\mathit{\text{eCFP}}\mathit{\text{MFI}}}$$ (1)

Confocal Imaging

Confocal microscopy with a Zeiss LSM 710 NLO confocal scan head mounted on an AxioObserver Z1 inverted microscope stage was used for fixed and live cell imaging. During imaging, cells were maintained in a stage incubator that controlled humidity, temperature, and CO₂ levels. Images were taken with high magnification objectives (20X Plan-Apochromat, NA 0.8; 63X oil-immersion Plan Apochromat, NA 1.4).

Quantification of YAP Nuclear Localization

The mean fluorescence intensity of YAP immunostained cells was calculated using a published MATLAB macro [48]. YAP nuclear localization was quantified through the YAP N/C ratio using Equation 2. To determine the location and size of the cell nucleus, a DAPI nuclear counterstain was used in conjunction with the YAP immunostaining.

$$\mathit{\text{YAP}}\mathit{\text{N/C}}\mathit{\text{Ratio}}\text{=}\frac{\mathit{\text{MFI}}\mathit{\text{of}}\mathit{\text{YAP}}\mathit{\text{in}}\mathit{\text{Nucleus}}}{\mathit{\text{MFI}}\mathit{\text{of}}\mathit{\text{YAP}}\mathit{\text{in}}\mathit{\text{Cytosolic}}\mathit{\text{Ring}}\mathit{\text{Around}}\mathit{\text{the}}\mathit{\text{Nucleus}}}$$ (2)

Traction Force Measurement

The top surfaces of the microposts were coated with fibronectin and AF647-conjugated fibrinogen (a fiduciary marker to track the microposts). Microscopy images were acquired, and deflections of micropost top surfaces were measured using a custom MATLAB macro [49]. Traction force, F, was calculated using Bernoulli-Euler beam theory, Equation 3, in which E, D, L, and δ are Youngs’ modulus, post diameter, post height, and post deflection, respectively.

$$F=\delta \frac{3\pi E{D}^4}{64L^3}$$ (3)

For this study, mPADs with the following characteristics were used:

• 1.83 μm post diameter, 4 μm center-center distance, 18.19 nN/μm spring constant, and 14.22 kPa effective modulus.

• 1.83 μm post diameter, 4 μm center-center distance, 7.22 nN/μm spring constant, and 5.65 kPa effective modulus.

Cell Area, Cell Polarity, Nucleus Area, and Nucleus Polarity Measurements

Cell area and polarity measurements were made on acquired images using MATLAB. Cell area was calculated from a user drawn cell area mask. The cell area mask was then fit to an ellipse which has the same normalized second central moment as the cell area mask. This was also used with a nuclear area mask generated from a DAPI nuclear counter stain. From these ellipses, the minor and major axes were determined and used to calculate cell and nucleus polarity with Equations 4 and 5.

$$\mathit{\text{Cell}}\mathit{\text{Polarity}}\text{=}\frac{\mathit{\text{Cell}}\mathit{\text{Major}}\mathit{\text{Axis}}}{\mathit{\text{Cell}}\mathit{\text{Minor}}\mathit{\text{Axis}}}$$ (4)

$$\mathit{\text{Nucleus}}\mathit{\text{Polarity}}\text{=}\frac{\mathit{\text{Nucleus}}\mathit{\text{Major}}\mathit{\text{Axis}}}{\mathit{\text{Nucleus}}\mathit{\text{Minor}}\mathit{\text{Axis}}}$$ (5)

Statistical Analyses

Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software Inc., La Jolla, CA). Statistical tests and P values are reported in the corresponding figures. To test differences between groups, a two-sided unpaired t-test, one-way ANOVA, or their non-parametric equivalent was conducted. To examine the relationship between the total traction force magnitude and cell spread area, a simple linear regression was conducted. Data is presented as mean ± SD.

Results and Discussion

Micropillar Array Rigidity Influences YAP Nuclear Localization

A linear relationship between force and increased FAK tyrosine phosphorylation (an early mechanosignaling event) at FAs occurs on substrates with rigidities greater than or equal to 14.22 kPa and a loss of this relationship occurs at rigidities at or below 5.65 kPa [50]. Disruption of talin-vinculin and talin-FAK binding, perturbing key functional domains, or deleting theses FA proteins abrogated this linear force-FAK signaling relationship. Herein, we sought to further investigate the functional domain contributions of vinculin, talin, and FAK in promoting YAP nuclear localization for adherent cells on deformable (14.22 kPa) ECM-coated pillars.

To examine whether YAP nuclear localization was impacted by substrate rigidity (i.e., mechanotransduction), fibronectin-coated micropillar arrays with stiffness values of 14.22 kPa and 5.65 kPa were utilized. We observed that YAP nuclear localization was increased when NIH3T3 fibroblasts were cultured on 14.22 kPa micropillar arrays as compared to 5.65 kPa micropillar arrays (Figure 1A, B). This stiffness-dependent difference in YAP nuclear localization was also seen on continuous synthetic hydrogels (Figure S2A–D). In addition, nuclear area was also larger on stiff substrates (Figure S2E), but no differences in nuclear polarity were observed (Figure S2F). These results are expected as rigidity-dependent YAP nuclear localization has been reported on flat hydrogels [31], and these results show equivalent stiffness-dependent YAP nuclear localization for both continuous substrates and discrete micropillar arrays. In addition, for cells adhering to micropillar arrays, we examined the effects of FAK kinase activity and contractility by treating fibroblasts with a FAK kinase inhibitor, PF-573228 [2 μM for 2 hours], or ROCK inhibitor, Y-27632 [10 μM for 2 hours]. Both PF-573228 and Y-27632 reduced YAP nuclear localization compared to controls for cells adhering to stiff, but not soft, micropillar arrays (Figure 1A, B). No changes in nucleus area were seen (Figure 1C), although less polarized nuclei were observed upon ROCK inhibition on soft (5.65 kPa) micropillar arrays (Figure 1D).

Fig. 1 |. Micropillar Array Rigidity Influences YAP Nuclear Localization.

(A) Representative confocal microscopy images of NIH 3T3 cells on 5.65 kPa and 14.22 kPa micropillar substrates. (B) YAP N/C (mean ± SD). Ordinary one-way ANOVA with Šídák’s multiple comparisons test. (C) Nucleus Area (mean ± SD). Ordinary one-way ANOVA with Šídák’s multiple comparisons test. (D) Nucleus Polarity (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. n=28–35.

Disruption of Vinculin-Talin Binding Reduces YAP Nuclear Localization

Although previous studies demonstrated that vinculin expression impacts YAP nuclear localization [39, 51], it is unclear whether vinculin-talin binding is required. We used two vinculin constructs expressed in vinculin-null cells: WT and A50I (Figure 2A). Full length vinculin A50I contains a point mutation in the vinculin head domain that prevents talin binding, and A50I also stabilizes an auto-inhibited vinculin conformation [16, 50, 52]. Vinculin-null and A50I vinculin-expressing cells exhibited reduced levels of YAP nuclear localization compared to cells expressing WT vinculin (Figure 2B, C). Interestingly, compared to wildtype cells, vinculin-null and A50I vinculin-expressing cells exhibited significantly smaller nuclei (Figure 2D) but no differences in nuclear polarity (Figure 2E).

Fig. 2 |. Disrupting Talin-Vinculin Binding Impairs YAP Nuclear Localization.

(A) Schematic of vinculin constructs used. (B) Representative confocal microscopy images of vinculin null MEFs expressing various vinculin constructs on 14.22 kPa micropillar substrates. (C) YAP N/C Ratio (mean ± SD). Brown-Forsythe and Welch ANOVA tests with Dunnett’s T3 multiple comparisons test. (D) Nucleus Area (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (E) Nucleus Polarity (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. n=27–29. (F) Representative confocal microscopy images of micropillars with cell traction forces overlayed. (G) Total Traction Force Magnitude (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (H) Cell Spread Area (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (I) Force per Cell Spread Area (mean ± SD). Ordinary One-Way ANOVA with Tukey’s multiple comparisons test. n=24–25. Scale bar is 20 μm.

The reductions in YAP nuclear localization and nuclear area correlated with reduced traction forces as vinculin-null and A50I-expressing cells generated lower cell traction forces compared to WT vinculin-expressing cells (Figure 2F–I and S3B). Only the deletion of vinculin reduced cell spread area (Figure 2H and S3A), while no differences in cell polarity (Figure S3C) were observed, supporting the notion that reductions in nuclear size and traction forces from disrupting vinculin-talin interactions occur independent of cell morphology changes. The traction force-cell area linear relationship was maintained in the three vinculin cell lines (Figure S4), suggesting that vinculin expression or vinculin-talin interactions are not essential for coupling of traction force to a spread cell morphology. As the talin binding site on vinculin affects nuclear size and cell traction forces but not overall cell morphology or polarity, these data support the notion that vinculin modulation of YAP nuclear localization is associated with changes in nucleus morphology arising from increased traction forces.

Talin-1 Promotes YAP Nuclear Localization on Stiff Substrates

Although a role for talin in force transmission and signal transduction to YAP is known [11], we sought to extend the results obtained using the vinculin A50I mutation by evaluating effects of talin-1 knockdown on fibroblasts cultured on 14.22 kPa mPADs. We targeted talin-1 as it is expressed in all cell types, while talin-2 is not expressed ubiquitously [53]. Western blots of treated NIH3T3 cells showed that shRNA treatment resulted in an 84% reduction in talin-1 protein compared to control shRNA treatment (Figure S5). Talin knockdown was associated with reduced YAP nuclear localization (Figure 3A, B), reduced nuclear area (Figure 3C), and reduced nuclear polarity (Figure 3D), and cells with talin-1 knocked down exhibited reduced traction forces (Figure 3E, F, H, and S6B) and were less spread (Figure 3G and S6A) compared control shRNA treatment. However, cells treated with talin-1 shRNA still exhibited a linear relationship between traction force and cell area (Figure S7). This result indicates that talin-1 is not necessary for the linear coupling between traction force and cell area, but it is possible that talin-2 expression is sufficient to maintain this relationship. Lastly, there was no difference in cell polarity between treated and control groups (Figure S6C). These observations support the model that changes in nucleus morphology from actomyosin contractility alter YAP nuclear localization. By knocking down talin-1, the link between integrins and the actin cytoskeleton is impaired, which reduces cell traction forces leading to reduced nucleus size and lower YAP nuclear localization.

Fig. 3 |. Talin-1 Knockdown Reduces YAP Nuclear Localization and Cell Traction Forces.

(A) Representative confocal microscopy images of NIH 3T3 cells transduced with lentiviral particles containing scramble or talin-1 shRNAs on 14.22 kPa micropillar substrates. (B) YAP N/C Ratio (mean ± SD). Unpaired t test. (C) Nucleus Area (mean ± SD). Mann-Whitney test. (D) Nucleus Polarity (mean ± SD). Unpaired t test. n=30. (E) Representative confocal microscopy images of NIH 3T3 cells with cell traction forces overlayed. (F) Total Traction Force Magnitude (mean ± SD). Mann-Whitney test. (G) Cell Spread Area (mean ± SD). Mann-Whitney test. (H) Force per Cell Spread Area (mean ± SD). Unpaired t test. n=24–25. Scale bar is 20 μm.

FAK Small Molecule Inhibition Reduces YAP Nuclear Localization

As FAK binds talin at FAs [21] and FAK can regulate traction forces [50, 54], we performed experiments with human mesenchymal stromal cells (hMSCs) and small molecule inhibitors to test the role of intrinsic FAK activity in promoting YAP nuclear localization [32]. We used hMSCs because they have been used extensively in mechanobiology studies, are being evaluated in multiple clinical trials, and have many similarities to fibroblasts in their phenotype and genotype [55–58]. hMSCs were seeded onto micropillar arrays and treated with PF-573228 [2 μM for 2 hours], an inhibitor of FAK activity, or blebbistatin [50 μM for 2 hours], an actomyosin contractility inhibitor. Compared to cells treated with vehicle control, PF-572338 or blebbistatin reduced YAP nuclear localization (Figure 4A, B). Additionally, blebbistatin treatment reduced nuclear area, whereas PF-572338 did not alter nuclear area compared to control (Figure 4C). Neither treatment affected nuclear polarity (Figure 4D), and parallel results were obtained when hMSCs were plated onto fibronectin-coated glass (Figure S8). These results confirm that actomyosin contractility and FAK activity are both important for YAP nuclear localization. As FAK inhibition blocked YAP nuclear localization independent of effects on nuclear size, our results support both nuclear tension-dependent (vinculin) and independent (FAK) models for matrix stiffness-regulated YAP nuclear localization.

Fig. 4 |. Inhibition of FAK and Myosin Contractility Disrupt YAP Nuclear Localization.

(A) Representative confocal microscopy images of hMSCs on 14.22 kPa micropillar substrates. (B) YAP N/C Ratio (mean ± SD). Ordinary one-way ANOVA with Šídák’s multiple comparisons test. (C) Nucleus Area (mean ± SD). Brown-Forsythe and Welch ANOVA tests with Dunnett’s T3 multiple comparisons test. (D) Nucleus Polarity (mean ± SD). Ordinary one-way ANOVA with Šídák’s multiple comparisons test. n=23–24 for all conditions. Scale bar is 20 μm.

FAK Controls YAP Nuclear Localization and YAP Transcriptional Activity Independent of Changes in Nuclear Tension

To assess the role of FAK functional domains in the control of YAP nuclear localization, FAK-null fibroblasts stably re-expressing wildtype (WT) FAK, autophosphorylation site mutated (Y397F) FAK, catalytically inactive (K454R) FAK, and a talin binding site mutant of FAK (E1015A) (Figure 5A) were cultured on micropillar arrays and evaluated for effects on YAP nuclear localization (Figure 5B). Only WT FAK functioned to promote YAP nuclear localization (Figure 5C). Interestingly, small differences in nuclear area (Figure 5D) or nuclear polarity (Figure 5E) were observed between FAK null, FAK WT, or FAK mutant expressing cells. This is in contrast with the importance of FAK activity, FAK Y397 phosphorylation, and the talin binding site on FAK in the regulation of matrix-stimulated traction forces [50]. We previously found that expression of WT, Y397F, and K454R FAK in FAK-null fibroblasts led to increased traction forces compared to FAK-null controls, and expression of WT, Y397F, and E1015A FAK in FAK-null fibroblasts resulted increased spread area [50]. Together, these results support the notion that FAK does not control YAP nuclear localization by changes in nuclear morphology and size.

Fig. 5 |. Impairing FAK catalytic activity, FAK activation, and talin-FAK interactions Impacts YAP Nuclear Localization and YAP Transcriptional Activity.

(A) FAK schematic showing location of point mutations in the various FAK constructs. (B) Representative confocal microscopy images of FAK Null MEFs expressing various WT, E1015A, K454R, and Y397F FAK constructs on 14.22 kPa micropillar substrates. (C) YAP N/C Ratio (mean ± SD). Brown-Forsythe and Welch ANOVA tests with Dunnett’s T3 multiple comparisons test. (D) Nucleus Area (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (E) Nucleus Polarity (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (F) YAP Activity Ratio of FAK null MEFS expressing WT, E1015A, K454R, Y397F FAK, ordinary one-way ANOVA with multiple comparisons, n=4–5.

To determine whether FAK-mediated YAP nuclear localization was associated with increased YAP transcriptional activity, a mCherry fluorescent YAP reporter with constitutive eCFP expression was used to provide a normalized metric of YAP transcriptional activity (YAP activity ratio) as quantified by flow cytometry (Figure S9A, B). This YAP reporter was validated by transfection of NIH3T3 fibroblasts plated on fibronectin and treated with a ROCK inhibitor (Y-27632 [10 μM for 72 hours]), a YAP inhibitor (verteporfin [2 μM for 72 hours]), or an activator of YAP transcriptional activity (PY-60 [10 μM for 72 hours] [59]). As expected, Y-27632 and verteporfin inhibited YAP transcriptional activity, whereas PY-60 significantly increased the YAP activity ratio (Figure S9C). We next assessed the YAP activity ratio for cells cultured on continuous hydrogel substrates with similar mechanical properties as the mPADs used (Figure S9D). WT FAK-expressing cells displayed higher YAP activity ratio compared to control FAK-null and FAK mutant-expressing cells (Figure 5F). In the Hippo kinase cascade, LATS1/2 phosphorylation of YAP controls YAP nuclear localization [60]. For YAP mechanotransduction, soft substrates have been found to induce LATS1/2 activation [32]. Therefore, FAK regulation of YAP nuclear localization may be dependent on the activity of LATS1/2. In MCF-10A cells, FAK inhibition via PF-573228 has been shown to inhibit YAP nuclear localization in a LATS1/2-dependent fashion [61]. To determine whether LATS1/2 contributed to YAP nuclear localization in the context of FAK expression, we treated WT FAK-expressing and FAK-null fibroblasts with a LATS1/2 inhibitor, TRULI [10 μM for 24 hours] [62] or DMSO. Pharmacological inhibition of LATS1/2, which prevents Hippo pathway regulation of YAP, increased YAP nuclear localization for both FAK-expressing and FAK-null cells (Figure 6A and B). FAK has been shown to modulate YAP nuclear localization through the Hippo pathway by inhibiting MOB1 and LATS1/2 as well as Hippo-pathway independent actions [36], so the mechanisms by which FAK regulates YAP nuclear localization in our experimental system remain unclear. Interestingly, inhibition of LATS1/2 resulted in increases in nuclear area but not polarity for both FAK-expressing and FAK-null cells (Figure 6C and D), suggesting contributions from nuclear tension and deformation [63]. These results show that FAK deletion, prevention of FAK Y397 phosphorylation, genetic inactivation of FAK activity, and disruption of talin binding to FAK all reduce YAP transcriptional activity. Together, these results highlight the importance of nuclear tension-independent FAK signals in modulating YAP transcriptional activity.

Fig. 6 |. Inhibition of LATS1/2 Promotes YAP Nuclear Localization.

(A) Representative confocal microscopy images of FAK Null MEFs expressing WT FAK construct on 14.22 kPa pillar substrates. (B) YAP N/C (mean ± SD). Ordinary one-way ANOVA with Šídák’s multiple comparisons test. (C) Nucleus Area (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. (D) Nucleus Polarity (mean ± SD). Kruskal-Wallis test with Dunn’s multiple comparisons test. n=17–28.

The Focal Adhesion-YAP signaling axis is complex and yet to be fully understood. This study advances our understanding of this signaling pathway by demonstrating an important role for the FA proteins vinculin, talin, and FAK in YAP nuclear localization in adherent mechanosensitive cells. Disrupting talin-vinculin and talin-FAK binding and eliminating FAK catalytic activity reduce YAP nuclear localization and transcriptional activity. The data we present here suggests that FA interactions can impact YAP nuclear localization in a nuclear tension dependent (vinculin and talin) and a nuclear tension independent manner (FAK). Potentially, vinculin and talin interactions alter the force balance on the nucleus causing its deformation, which permits YAP nuclear influx. Although several models have been proposed to describe the Focal Adhesion-YAP signaling axis, there is no consensus and there is likely interplay between them because of the many interactions that exist among FA proteins. For example, while the deletion and perturbation of vinculin does not alter pY397 levels of FAK, talin-2 depletion or inhibition of actomyosin contractility by blebbistatin, reduces Y397 phosphorylation of FAK [50, 64, 65]. This suggests that FAK has a context dependent role and understanding how these pathways intertwine is needed to completely understand the FA-YAP pathway. This study provides new insights into the role of talin-vinculin interactions, talin-FAK interactions, and FAK catalytic activity in mechanosensing of substrate stiffness. Due to the importance of FA and YAP for many cellular and tissue processes, these findings generate additional insights for improved design of future biomaterial and regenerative medicine therapies.

Conclusion

We investigated the role of different functional domains in vinculin, talin, and FAK in regulating YAP nuclear localization. Using genetic manipulations or pharmacological inhibition of fibroblasts and human mesenchymal stem cells (hMSCs) adhering to deformable substrates, we found that disruption of vinculin-talin binding versus talin-FAK binding reduces YAP nuclear localization and transcriptional activity via different mechanisms. Disruption of vinculin-talin binding or knockdown of talin-1 also reduces nuclear size, traction forces, and YAP nuclear localization. In contrast, disruption of the talin binding site on FAK or elimination of FAK catalytic activity did not alter nuclear size yet still prevented YAP nuclear localization and activity. These data support both nuclear tension-dependent and independent models for matrix stiffness-regulated YAP nuclear localization. Our results highlight the importance of vinculin-talin-FAK interactions at FAs of adherent cells, controlling YAP nuclear localization and activity.

Data Availability

Data files are available from the corresponding author upon reasonable request.