Kindlin-2 Mediates Mechanical Activation of Cardiac Myofibroblasts

We identify the focal adhesion protein kindlin-2 as player in a novel mechanotransduction pathway that controls profibrotic cardiac fibroblast to myofibroblast activation. Kindlin-2 is co-upregulated with the myofibroblast marker α-smooth muscle actin (α-SMA) in fibrotic rat hearts and in human cardiac fibroblasts exposed to fibrosis-stiff culture substrates and pro-fibrotic TGF-β1. Stressing fibroblasts using ferromagnetic microbeads, stretchable silicone membranes, and cell contraction agonists all result in kindlin-2 translocation to the nucleus. Overexpression of full-length kindlin-2 but not of kindlin-2 missing a putative nuclear localization sequence (∆NLS kindlin-2) results in increased α-SMA promoter activity. Downregulating kindlin-2 with siRNA leads to decreased myofibroblast contraction and reduced α-SMA expression, which is dependent on CC(A/T)-rich GG(CArG) box elements in the α-SMA promoter. Lost myofibroblast features under kindlin-2 knockdown are rescued with wild-type but not ∆NLS kindlin-2, indicating that myofibroblast control by kindlin-2 requires its nuclear translocation. Because kindlin-2 can act as a mechanotransducer regulating the transcription of α-SMA, it is a potential target to interfere with myofibroblast activation in tissue fibrosis.


Introduction
The ability of cardiac fibroblasts to sense and control the mechanical properties of the extracellular matrix (ECM) is essential to adapt heart tissue to mechanical load (e.g., hypertension) and to repair injuries (e.g., after myocardial infarct) [1,2]. Aberrant mechanosensing results in activation of cardiac fibroblasts into α-smooth muscle actin (α-SMA)-expressing myofibroblasts that are characterized by excessive collagen secretion and contraction [3][4][5][6]. The outcome of myofibroblast dysfunction is fibrosis-accumulation of stiff collagen scar tissue. Stiff fibrotic tissue impairs heart distensibility, inverted microscope (Axiovert 135, Carl Zeiss). To induce cell contraction, cells were treated with 20 µM plasminogen activator inhibitor (PAR)-1 agonist TFLLRN (Anaspec, Fremont, CA, USA) for 1 h. To quantify the percentage of contractile cells, the numbers of wrinkling and total cells were counted for ≥10 random fields of view (~25 cells/field) per experimental condition in three experimental repeats.

Mechanical Cell Stimulation
To subject hCF to global strain, we used custom-made stretchable culture membranes that were functionalized with 2 µg/cm 2 fibronectin (Millipore) [39]. HCF were cultured at 1000 cells per cm 2 for 24 h before straining the cultures by 20% in a custom-built uniaxial stretching device for 1 h. To strain hCF at sites of FAs, we used magnetic microparticles (3 µm diameter, Fe 3 O 4 , Sigma, Oakville, ON, Canada), coated with 100 µg/mL fibronectin (Millipore) [35,41]. Fibronectin-coated microbeads were added to hCF cultures for 30 min, followed by three washes with medium to remove unbound microbeads. A ceramic permanent magnet was then applied to generate tensile forces perpendicular to the cells' growth plane via the attached microbeads for 1 h. Microbead-bound proteins were isolated by washing cells with PBS and scraping into extraction buffer (0.5% Triton X-100, 50 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 , 10 mM 1,4-Piperazinediethanesulfonic acid, pH 6.8), supplemented with protease inhibitors cocktail (Sigma). The suspension was sonicated and the microbead fraction was separated from the cytosolic fraction using the magnetic separation stand. Bead fractions were suspended in ice-cold extraction buffer, homogenized, and boiled in sample buffer for 5 min. Microbeads were pelleted and supernatant was collected for Western blotting.

Nuclear Extracts and Western Blotting
For Western blotting, we produced total cell extracts according to standard procedures, and nuclear and non-nuclear (cytosol and cytoskeleton) extracts using a nuclear extraction kit (Pierce, Thermo Fisher Scientific, Waltham, QC, Canada) according to the manufacturer's recommendation. Blots were probed with the same primary antibodies as described for microscopy. Horseradish peroxidase-conjugated secondary goat anti-mouse and goat anti-rabbit antibodies (Invitrogen, Burlington, ON, Canada) were used, followed by chemiluminescence (Invitrogen). Loading controls were vimentin for total cell Cells 2020, 9, 2702 4 of 22 lysates, lamin B for nuclear fractions, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for cytosol, and fibronectin for microbead fractions. Densitometry analysis was performed with Fiji software [43].
Luciferase reporter assays were performed by additionally transfecting cells with a mixture of α-SMA promoter luciferase reporter construct and constitutive renilla luciferase expression vector (Vector pRL-TK, Promega, Madison, WI, USA) as internal control. We used constructs encoding the wild-type and various mutant versions of the 765 bp rat α-SMA (ACTA2) promoter controlling the expression of firefly luciferase in pGL3-basic vector [45] Constructs harbored mutations inactivating: (1) the two CArG boxes (CArG-A and CArG-B); (2) the two Smad-binding elements SBE1, SBE2, and the TCE. After 48 h, cells were serum starved for another 3 h and lysed. Luciferase activity in lysates was determined using a dual luciferase reporter assay system kit (Promega) and a multi-well plate luminometer (Berthold Technologies, Harpenden, UK) according to the manufacturer's instructions. Firefly luciferase activity reporting α-SMA promoter activity was normalized to control renilla luciferase activity of the same sample.
For quantitative real time (qRT) PCR, RNA was isolated using the RNeasy mini kit (Qiagen, SA Biosciences, ON, Canada).
qRT-PCR was performed using Superscript Vilo (Invitrogen, Live Technologies, NY, USA) using RT SYBR Green PCR master mix (Qiagen, SA Biosciences, ON, Canada) and forward and reverse primers TGGACGGGATAAGGATGCCA and TGACATCGAGTTTTTCCACCAAC for kindlin-2 (FERMT2); CCCAGACACCAGGGAGTAATGG and TCTATCGGATACTTCAGGGTCA for α-SMA (ACTA2); AGGTCGGTGTGAACGGATTTG and TGTAGACCATGTAGTTGAGGTCA for GAPDH on a Step One Plus Real Time PCR System (Applied Biosciences, Live Technologies, Burlington, ON, Canada) according to the manufacturer's recommendations.

Statistical Analysis
All experiments were performed at least three times and data are presented as mean ±standard deviation (SD) or standard error of the mean (SEM), where applicable. As independent experiments, we considered data obtained from different batches of fibroblasts and/or different animals. We assessed differences between two groups with a one sample Student t-test and multiple groups using ANOVA followed by a post-hoc Tukey's multiple comparison test. Differences were statistically significant with p ≤ 0.05.

Kindlin-2 Expression Is Upregulated in Activated Cardiac Fibroblasts In Vitro and In Vivo
Because kindlin-2 controls the affinity of fibroblast integrins and mediates intracellular binding of the integrin cytoplasmic tail to contractile stress fibers, we investigated its role in hCF mechanosensing and myofibroblast activation in conditions of cardiac fibrosis. Expression of kindlin-2 in the mouse heart and contribution to heart development were described previously [31] but the physiological functions are yet unclear. We localized expression of kindlin-2 within the left ventricles of rats that underwent infra-renal abdominal aortic coarctation to develop hypertension and left ventricular heart fibrosis [37]. In normal left ventricles of control animals, kindlin-2 was expressed in vimentin-positive fibroblasts and α-SMA-positive vascular smooth muscle cells of the interstitial myocardium, in addition to cardiomyocytes ( Figure 1A) as previously reported [32,46]. In contrast, the disorganized interstitium of fibrotic left ventricles was characterized by accumulation of α-SMAand vimentin-positive myofibroblasts expressing high levels of kindlin-2 ( Figure 1A). These findings suggest a role for kindlin-2 in normal and excessive tissue repair leading to heart fibrosis.
Two main factors control myofibroblast activation: transforming growth factor beta (TGF-β1) and mechanical stress [47,48]. Fibroblasts cultured on conventional tissue culture plastic (TCP) for a total of 4 day exhibited significantly increased expression of kindlin-2 protein after 1 day (1.7-fold) and 4 day (3.2-fold) of TGF-β1 treatment compared to untreated controls ( Figure 1B). Concomitantly, the levels of phosphorylated Smad2/3, indicators of active TGF-β1 signaling, were elevated 1 day and 4 day post TGF-β1 treatment. Increase in kindlin-2 expression after 1 day TGF-β1 treatment was relatively higher than that of α-SMA, which was enhanced after 1 day by 1.3-fold and after 4 day by 3.0-fold following TGF-β1 treatment ( Figure 1B). These Western blot results were confirmed by immunofluorescence staining, with higher amounts of kindlin-2 and larger kindlin-2-positive FAs observed after 4 day of TGF-β1 treatment compared to control ( Figure 1C).
Next, we used silicone cell culture substrates with controlled stiffness to test whether the mechanical ECM conditions characteristic of heart fibrosis can contribute to increased kindlin-2 expression in myofibroblasts. The elastic modulus of heart muscle was previously determined to range between 10-15 kPa [38,49]. We and others have shown that growth on soft culture substrates (3-5 kPa) suppresses heart myofibroblast activation whereas 26-65 kPa stiff substrates approximating the stiffness of fibrotic heart activate cardiac myofibroblasts [49][50][51][52][53][54]. Expression of kindlin-2 and α-SMA in hCF cultured for 5 d on stiffness-tuned substrates increased 2.1-and 1.9-fold, respectively, from 3 kPa to 65 kPa and were highest in hCF on GPa-stiff TCP ( Figure 1D). Kindlin-2 localized to small FAs formed on soft and large FAs formed on stiff substrates at the stress fiber termini of cultured hCF ( Figure 1E). Collectively, these results show that kindlin-2 expression is co-regulated with myofibroblast activation. Because regulation of kindlin-2 by TGF-β1 has been established before in various cell types [36,[55][56][57][58][59], we continued to investigate the relationship between mechanical stress, kindlin-2, and myofibroblast activation.

Mechanical Stimulation of hCF Results in Kindlin-2 Accumulation in the Nucleus
The recruitment of specific cytosolic proteins to FAs under stress is an integral part of fibroblast mechanosensing [12,[60][61][62][63]. To assess whether kindlin-2 follows the same regimen, we applied tensile forces to fibronectin-coated ferromagnetic microbeads [35], bound to dorsal FAs of hCF cultured on TCP. Immunoblots of the microbead-associated protein fraction confirmed that the stress-responsive proteins β-cytoplasmic actin, vinculin, paxillin, and β1 integrin were recruited to FAs under strain ( Figure 2A, "microbead"). ventricular heart fibrosis [37]. In normal left ventricles of control animals, kindlin-2 was expressed in vimentin-positive fibroblasts and α-SMA-positive vascular smooth muscle cells of the interstitial myocardium, in addition to cardiomyocytes ( Figure 1A) as previously reported [32,46]. In contrast, the disorganized interstitium of fibrotic left ventricles was characterized by accumulation of α-SMAand vimentin-positive myofibroblasts expressing high levels of kindlin-2 ( Figure 1A). These findings suggest a role for kindlin-2 in normal and excessive tissue repair leading to heart fibrosis.  (A) Sections of normal and hypertrophic rat hearts were stained for kindlin-2 (green), α-smooth muscle actin (α-SMA) (red), and vimentin (blue), and were observed with confocal microscopy; average intensity projections of z-stacks are displayed. Vimentin-and α-SMA-positive myofibroblasts in the hypertensive heart strongly express kindlin-2 in fibrotic lesions. (B) Human fibroblasts were cultured on tissue culture plastic (TCP) for 5 day and treated for 1-4 day with transforming growth factor (TGF)-β1 (2 ng/mL) to assess expression of kindlin-2 and α-SMA by quantitative immunoblotting with indicated molecular weights. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as loading control and Smad2/3 to assess TGF-β1 downstream signaling. (C) After 4 day treatment with TGF-β1, cells were immunostained for kindlin-2 (green), α-SMA (red), and nuclei (4 ,6-diamidino-2-phenylindole, DAPI, blue) and confocal images were taken. (D,E) Likewise, human cardiac fibroblasts (hCF) were cultured on silicone culture substrates with elastic modulus of 3, 26, and 65 kPa and GPa-stiff TCP for 5 d, followed by analysis using (B) immunoblotting, densitometry and (C) immunofluorescence for kindlin-2 (green) and vinculin (red). All immunoblot bands were quantified by densitometry, first normalized to GAPDH loading control and then to TCP control. Shown are mean values from at least three independent experiments (data points) ±SD (* p < 0.05, ** p <0.01, *** p < 0.005, using ANOVA followed by a post-hoc Tukey's multiple comparison test). All scale bars: 20 µm. show optical z-section reconstruction along the indicted plane (blue line) (see also Supplementary Videos S1 and S2). Shown are mean values from at least three independent experiments (data points) ±SD (** p < 0.01, using one sample Student t-test). All scale bars: 20 µm.
Unexpectedly, application of strain reduced kindlin-2 levels in the microbead-associated fraction by 4.5-fold compared with non-strained control ( Figure 2A); total protein levels did not change (Figure 2A, "total"). Tubulin, fibronectin, and GAPDH were used as controls for loading and fractionation purity. Interestingly, the decrease of kindlin-2 in the strained microbead fraction ( Figure 2B) coincided with 3.5-fold enrichment of kindlin-2 in the nucleus of hCF as quantified from confocal immunofluorescence images ( Figure 2C). Nuclear localization of kindlin-2 was confirmed in three-dimensional reconstructions of confocal optical sections ( Figure 2D,E, Supplementary Videos S1 and S2).
Because the presence of beads was incompatible with centrifuge-based protocols for biochemical protein quantification of nuclear extracts, hCF were next grown on stretchable culture membranes for 4 day and then subjected to one single static strain of 20% to be assessed after different times. Immunostaining ( Figure 3A) and quantification of kindlin-2 staining intensity ( Figure 3B) revealed increased accumulation of kindlin-2 in the nucleus already 1 h after strain compared with non-strained controls. Kindlin-2 in the nucleus further increased 2 h after strain and decreased moderately after 5 h ( Figure 3A,B).
Immunoblotting of the nuclear, non-nuclear, and total cell fractions collected 1 h after 20% cell strain confirmed 2.0-fold increase of kindlin-2 in the nuclear fraction ( Figure 3C). Purity of the respective fractions was controlled by immunoblotting for GAPDH that was enriched in the non-nuclear fraction and present at low levels in the nuclear fraction as reported previously [56]. Lamin B was restricted to the nuclear fraction with negligible cross-contaminations ( Figure 3C). GAPDH content in the nuclear fraction did not change with strain. Notably, constitutive levels of kindlin-2 in the nucleus were low but present in non-strained hCF grown on stretchable membranes (modulus of 3 MPa) for days ( Figure 3A-C), confirming similar observations made with fibroblasts grown on TCP (GPa) and stiff silicone substrates (65 kPa) ( Figure 1).
Concomitantly, addition of TFLLRN resulted in a 2.5-fold increase of kindlin-2 protein in the nucleus, compared with vehicle-treated control ( Figure 4B). Collectively, three different approaches to mechanically stimulate hCF all demonstrated that kindlin-2 is mechanosensitive and shows enhanced translocation to the nucleus in acutely stressed fibroblasts.
Kindlin-2 siRNA transfection achieved 60-80% downregulation of kindlin-2, correlating with 40-50% reduction in α-SMA expression both at the protein and mRNA level ( Figure 5C,D). Concomitantly, cultures of hCF co-transfected with kindlin-2 siRNAs and siGlo to identify transfected cells by fluorescence microscopy, showed~2-fold reduced percentages of contractile hCF compared with scrambled siRNA controls ( Figure 5E). Hence, kindlin-2 expression contributes to maintaining the α-SMA-positive phenotype and contractile function of cardiac myofibroblasts. protein quantification of nuclear extracts, hCF were next grown on stretchable culture membranes for 4 d and then subjected to one single static strain of 20% to be assessed after different times. Immunostaining ( Figure 3A) and quantification of kindlin-2 staining intensity ( Figure 3B) revealed increased accumulation of kindlin-2 in the nucleus already 1 h after strain compared with nonstrained controls. Kindlin-2 in the nucleus further increased 2 h after strain and decreased moderately after 5 h ( Figure 3A,B). Cell straining leads to kindlin-2 accumulation in the nucleus. Human cardiac fibroblasts (hCF) were strained on silicone culture membranes once by 20% and assessed after different times. (A) 1 h, 2 h, and 5 h after unique cell strain (20% strain) or non-strained (control), hCF were immunostained for kindlin-2 (green) and lamin B (red) and scanning confocal images were taken; greyscale insets show kindlin-2. Scale bars: 15 µ m. (B) Levels of kindlin-2 in the nucleus (fluorescence intensity) were quantified by image analysis from confocal images. (C) Nuclear, non-nuclear, and total cell fractions were isolated and analyzed by immunoblotting and densitometry. Lamin B expression levels were used as a loading control for nuclear and total fractions; Glyceraldehyde 3phosphate dehydrogenase (GAPDH) was used to control similar loading of the non-nuclear and total fractions. Total fractions were diluted 1:2 to allow simultaneous blotting of all fractions without signal saturation. Shown are mean values from at least three independent experiments (data points) ±SD (* p < 0.05, ** p < 0.01, *** p < 0.005, using ANOVA followed by a post-hoc Tukey's multiple comparison test).
Immunoblotting of the nuclear, non-nuclear, and total cell fractions collected 1 h after 20% cell strain confirmed 2.0-fold increase of kindlin-2 in the nuclear fraction ( Figure 3C). Purity of the  ) were quantified by image analysis from confocal images. (C) Nuclear, non-nuclear, and total cell fractions were isolated and analyzed by immunoblotting and densitometry. Lamin B expression levels were used as a loading control for nuclear and total fractions; Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used to control similar loading of the non-nuclear and total fractions. Total fractions were diluted 1:2 to allow simultaneous blotting of all fractions without signal saturation. Shown are mean values from at least three independent experiments (data points) ±SD (* p < 0.05, ** p < 0.01, *** p < 0.005, using ANOVA followed by a post-hoc Tukey's multiple comparison test). Next, we assessed the possible link between hCF stress perception, nuclear translocation of kindlin-2, and α-SMA gene transcription. Downregulation of kindlin-2 with siRNA resulted in about 30% decrease in luciferase reporter activity under control of the full-length wild-type promoter of α-SMA compared with non-targeting siRNA control ( Figure 5F). Kindlin-2 has previously been shown to promote TGF-β1 signaling by binding to the TGF-β1 receptor and the TGF-β1 downstream co-transcription factor Smad3 [57]. To test whether the α-SMA-inducing function of kindlin-2 is mediated via TGF-β1/Smad3, we inactivated the TGF-β control element (TCE) and Smad-binding elements (SBEs) from the α-SMA-promoter in luciferase reporter constructs [45,[65][66][67]. Baseline reporter activity of the ∆SBE reporter was 30% lower compared to the wild-type promoter construct ( Figure 5F). Knockdown of kindlin-2 reduced the activity of SBE/TCE-inactive α-SMA reporter promoter by 30%, like the wild-type promoter ( Figure 5F). Hence, despite the absence of SBE/TCE, loss of kindlin-2 will thus affect α-SMA expression. In addition to SBE/TCE, the α-SMA promoter contains CArG boxes that bind serum response factor (SRF) together with myocardin-related transcription factor (MRTF/MLK-1) [35,[67][68][69][70][71]. In fibroblasts, mechanical stimulation and integrin engagement cause nuclear translocation of MRTF-A, which it turn enhances the transcriptional activity of SRF, thereby inducing α-SMA gene expression through the cis-elements [70,[72][73][74][75][76][77]. Kindlin-2 knockdown did not reduce luciferase expression under control of α-SMA promoter with inactive CArG boxes below baseline, which was 35% lower compared to the wild-type promoter construct ( Figure 5F). These results suggested that kindlin-2 promotes α-SMA transcription via CArG boxes in the α-SMA promoter.
Cells 2020, 9, x 9 of 22 respective fractions was controlled by immunoblotting for GAPDH that was enriched in the nonnuclear fraction and present at low levels in the nuclear fraction as reported previously [56]. Lamin B was restricted to the nuclear fraction with negligible cross-contaminations ( Figure 3C). GAPDH content in the nuclear fraction did not change with strain. Notably, constitutive levels of kindlin-2 in the nucleus were low but present in non-strained hCF grown on stretchable membranes (modulus of 3 MPa) for days ( Figure 3A-C), confirming similar observations made with fibroblasts grown on TCP (GPa) and stiff silicone substrates (65 kPa) (Figure 1). FAs are strained both in response to extracellularly and intracellularly applied force [12,63]. To increase intracellular force, we induced RhoA/Rho-associated kinase-mediated cell contraction using the protease-activated receptor-1 (PAR-1)-activating peptide TFLLRN [64]. When adding TFLLRN to cells grown on "wrinkling" silicone culture substrates, visible surface deformations increased 2.5fold within 50 min ( Figure 4A). Concomitantly, addition of TFLLRN resulted in a 2.5-fold increase of kindlin-2 protein in the nucleus, compared with vehicle-treated control ( Figure 4B). Collectively, three different approaches to mechanically stimulate hCF all demonstrated that kindlin-2 is mechanosensitive and shows enhanced translocation to the nucleus in acutely stressed fibroblasts.

Nuclear Kindlin-2 Plays a Role in Myofibroblast Activation
Kindlin-2 knockdown resulted in only moderate changes in hCF morphology or FA size after 48 h ( Figure 5A,B) despite others reporting severely reduced adhesion and spreading of kindlin-2 deficient cells [78]. Restricting myofibroblast adhesion and FA size has been shown to result in reduced levels of α-SMA expression by reducing intracellular stress [79]. To discriminate between possible adhesion-mediated effects of kindlin-2 in FAs and putative regulation of α-SMA gene transcription by nuclear kindlin-2, we overexpressed kindlin-2-GFP and a kindlin-2-GFP mutant lacking the putative nuclear location sequence (NLS) (kindlin-2-∆NLS-GFP) in human MRC-5 fibroblasts. MRC-5 were chosen for their low baseline expression of α-SMA and ease of transfection compared to hCF. Both kindlin-2 constructs co-localized with endogenous kindlin-2 in FAs ( Figure 6A) and kindin-2-GFP additionally localized to the nucleus and cytosol. Cells 2020, 9, x 10 of 22  In contrast, kindlin-2-∆NLS-GFP was almost completely excluded from the nucleus and accumulated in the cytosol ( Figure 6A,B), resulting in low ratios of nuclear versus cytosolic kindlin-2 (GFP-tagged plus endogenous) ( Figure 6C). Overexpression of kindin-2-GFP resulted in overall enhanced nuclear localization of kindlin-2 over cytosol ( Figure 6C).
Finally, we assessed the potential of kindlin-2-GFP and kindlin-2-∆NLS-GFP to rescue myofibroblast features that were lost in rat embryonic fibroblasts (REF) upon kindlin-2 knockdown. Lineage REF were chosen because they exhibit high α-SMA baseline levels like hCF, but are easier to transfect [80]. To be able to rescue knockdown effects and target rat kindlin-2, REF were transfected with a 3 -UTR-targeting kindlin-2 siRNA (K-2 siRNA III). K-2 siRNA III was different from the open reading frame-targeting kindlin-2 siRNA K-2 siRNA I and II used for human fibroblasts. K-2 siRNA III resulted in downregulation of kindlin-2 protein and mRNA, 48 h after transfection of REF ( Figure 7A).
In contrast, kindlin-2-ΔNLS-GFP was almost completely excluded from the nucleus and accumulated in the cytosol ( Figure 6A,B), resulting in low ratios of nuclear versus cytosolic kindlin-2 (GFP-tagged plus endogenous) ( Figure 6C). Overexpression of kindin-2-GFP resulted in overall enhanced nuclear localization of kindlin-2 over cytosol ( Figure 6C). Lineage REF were chosen because they exhibit high α-SMA baseline levels like hCF, but are easier to transfect [80]. To be able to rescue knockdown effects and target rat kindlin-2, REF were transfected with a 3′-UTR-targeting kindlin-2 siRNA (K-2 siRNA III). K-2 siRNA III was different from the open reading frame-targeting kindlin-2 siRNA K-2 siRNA I and II used for human fibroblasts. K-2 siRNA III resulted in downregulation of kindlin-2 protein and mRNA, 48 h after transfection of REF ( Figure  7A).

Discussion
By binding to integrins and the contractile actin cytoskeleton, kindlin-2 is ideally positioned to mediate mechanical signals that are perceived at sites of FAs. From our observation that kindlin-2 expression is upregulated in myofibroblasts upon mechanical overload and fibrosis of the heart, we hypothesized that kindlin-2 regulates the activation of cardiac myofibroblasts. In skin, kindlin-2 is expressed in epidermal keratinocytes and dermal fibroblasts, and it is upregulated upon myofibroblast activation during skin wound healing and in 4-6-week-old human cutaneous scars [42,81]. Kindlin-2 also contributes to kidney fibrosis by interfering with TGF-β1 signaling [58,82]. Our results show that overexpression of kindlin-2 in hCF, MRC-5, and REF results in increased transcription and expression of the myofibroblast marker α-SMA. Consistently, knockdown of kindlin-2 leads to reduced transcription and expression of α-SMA in cultured cardiac myofibroblasts and REF. A major novel finding of our study is that acute mechanical strain induces kindlin-2 translocation into the nuclei of hCF. Applying strain and force locally to FAs using ECM-coated microbeads, globally by stretching hCF on deformable substrates, or by inducing fibroblast contraction all increase kindlin-2 localization in the nucleus within one hour. Although kindlin-2 expression levels were reduced in hCF grown on soft compared to stiff culture substrates, nuclear kindlin-2 levels are comparably low in hCF grown for prolonged periods on stiff culture plastic. Conceivably, kindlin-2 is involved in acute rather than long-term stress responses of fibroblastic cells.
In contrast to the loss of kindlin-2, talin1 has previously been shown to be recruited to FAs under acute mechanical strain which is possibly explained by their different integrin-binding characteristics [97]. Kindlin-2 directly binds the membrane-distal NxxY motifs of the β1 and β3 integrin cytoplasmic tails via its FERM domain whereas talin1 binds to membrane-proximal NPxY motifs of β-integrin [16,22,26,103,104]. Several studies have indeed revealed distinct roles of talin1 and kindlin-2 in regulating integrin trafficking [34,103,105], integrin force-coupling [106], and cell signaling [97,107]. Given its central role in FAs it is not surprising that kindlin-2 has been implicated in mechanical cell communication with the environment, for instance by integrating the Rho pathway [108][109][110]. Considering our results, it is difficult to appreciate how kindlin-2 can contribute to stress-mediated integrin activation if stress removes it from ECM adhesions. However, our results demonstrate that a substantial fraction of kindlin-2 remains in the stress-bearing peripheral FAs after straining myofibroblasts. Similarly, diverse stress-dependent behavior has been shown for zyxin that is differentially localized in FAs, stress fibers and the nucleus, depending on the levels and location of applied mechanical stress [111,112].
The ability for nuclear translocation seems to be a prerequisite for kindlin-2 to regulate myofibroblast activation. Overexpression of a kindlin-2 mutant lacking the putative NLS at the N-terminus of kindlin-2 resulted in only moderately increased α-SMA transcription in contrast to a strong increase upon wild-type kindlin-2 transfection. Our results suggest that kindlin-2 interacts with factors in the nucleus that are known to control the activity of the α-SMA promoter via CArG boxes, e.g., MRTF and SRF. This novel role of kindlin-2 in transcriptional regulation is consistent with recent findings that kindlin-2 acts as a co-transcription factor with β-catenin and T-cell factor 4 to control transcription genes regulated by the Wnt pathway, including axin-2, cyclin D1, twist, lymphoid enhancer-binding factor 1, matrix metalloproteinase-2, secreted frizzled-related protein, and versican [84]. Moreover, kindlin-2 was recently shown to control renal tubular epithelial-to-mesenchymal differentiation via ERK1/2 and Akt signaling pathways [58]. Epithelial-to-mesenchymal transition is the first step in a process that ultimately culminates in the expression of α-SMA as a key indicator of the "myogenic" differentiation program under the conditions of tissue repair and fibrosis [113]. Another important transcriptional co-factor that was shown to regulate myofibroblast activation is MRTF-A in conjunction with SRF [70,73,74,114]. Conceivably, kindlin-2 primarily senses the state of force transduction through integrins whereas MRTF-A senses the state of actin polymerization; these two types of information may then converge on SRF to regulate transcription of the α-SMA promoter.
It is unlikely that kindlin-2 downregulation and overexpression affect myofibroblast activation exclusively due to a nuclear function. The same poly lysine motif in kindlin-2 that putatively controls nuclear localization was reported to promote binding of the F1 loop in kindlin to acidic membrane phospholipids and cooperation with talin1 in integrin activation [115]. However, our deletion mutant was able to recruit to FAs, in contrast to the mutant used in this previous study [115], and we did not observe dramatic loss of adhesion or alteration of vinculin-positive FAs in 48 h kindlin-2 knockdown experiments with three different fibroblasts: hCF, REF, and MRC-5. It is possible that adhesion of cells with low kindlin-2 levels is rescued by integrins that are less dependent on kindlin-2-mediated regulation. For instance, αvβ5 integrin is expressed in cardiac fibroblasts [51] and promotes cell spreading and adhesion to fibronectin and vitronectin among other ECM ligands. Recent studies suggest that kindlin-2 may not be required to bind to αvβ5 integrin to promote cell adhesion [116]. Nevertheless, it is likely that kindlin-2 expression levels will influence activation of other integrins (e.g., α5β1 integrin) and fibroblast adhesion strength [117] as shown by reduced spreading of suspended kindlin-2 knockdown cells [78]. The reduced ability of hCF and REF to wrinkle deformable substrates is possibly a combination of reduced force transmission due to changes in FAs and/or reduced contraction due to the loss of α-SMA in stress fibers. Our own research has shown that both are crucial in promoting long term myofibroblast activation [51,79,80].
We conclude that both functions of kindlin-2, as a novel mechanosensor that shuttles to the nucleus of mechanically strained cells and for FA protein-promoting integrin activation, have a central role in controlling myofibroblast activation. Whether this role makes kindlin-2 a suitable, i.e., sufficiently specific, target to treat conditions of fibrosis remains to be shown.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4409/9/12/2702/s1, Video S1: Human cardiac fibroblasts were incubated with fibronectin-coated magnetic microbeads for two hours. Cells were then fixed permeabilized, and immunostained for kindlin-2 (green) and lamin B (red). Z-stacks of confocal optical sections were obtained by structured illumination (Apotome, Zeiss) with 0.1 µm oversampling (left upper frame moves through the 114 sections) and used to reconstruct three-dimensional images. Reconstructions of kindlin-2 alone (lower left), lamin B alone (lower right) and merged channels (upper right) are rotated by 180 • . Video S2: Human cardiac fibroblasts were incubated with fibronectin-coated magnetic microbeads for one hour followed by tensile force application using a magnet for another hour. Cells were then fixed permeabilized, and immunostained for kindlin-2 (green) and lamin B (red). Z-stacks of confocal optical sections were obtained by structured illumination (Apotome, Zeiss) with 0.1 µm oversampling (left upper frame moves through the 120 sections) and used to reconstruct three-dimensional images. Reconstructions of kindlin-2 alone (lower left), lamin B alone (lower right) and merged channels (upper right) are rotated by 180 • .