|Home | About | Journals | Submit | Contact Us | Français|
Focal adhesions (FAs) undergo myosinII-mediated maturation wherein they grow and change composition to modulate integrin signaling for cell migration, growth and differentiation. To determine how FA composition is modulated by myosinII activity, we performed proteomic analysis of isolated FAs and compared protein abundance in FAs from cells with and without myosinII inhibition. We identified FA 905 proteins, 459 of which changed in FA abundance with myosinII inhibition, defining the myosinII-responsive FA proteome. FA abundance of 73% of proteins was enhanced by contractility, including those involved in Rho-mediated FA maturation and endocytosis- and calpain-dependent FA disassembly. 27% of proteins, including those involved in Rac-mediated lamellipodial protrusion, were enriched in FA by myosinII inhibition, establishing for the first time negative regulation of FA protein recruitment by contractility. We focused on the Rac guanine nucleotide exchange factor, β-PIX, documenting its role in negative regulation of FA maturation and promotion of lamellipodial protrusion, FA turnover to drive cell migration.
Integrin-mediated focal adhesions (FAs) are dynamic, multifunctional organelles that facilitate cell adhesion, force transmission and cytoskeleton regulation, as well as signaling that controls cell division, differentiation, or apoptosis1,2. The diverse FA functionality is reflected in their complex composition. Over 150 different proteins are known to associate with FAs making up the “Integrin Adhesome”3. The Adhesome includes cytoskeletal-binding and adapter proteins and enzymes including kinases, phosphatases, phospholipases and small GTPases and their modulators. Furthermore, FA composition is heterogeneous and dynamic, even across different FAs within a single cell4. However, no FA compositions have been ascribed to specific downstream functions.
One important regulator of FA composition is physical perturbation5,6. FAs are mechanosensitive and recruit proteins to grow, and change composition in response to tension in the process of FA maturation7–9. Tension driving FA maturation is supplied either by myosinII activity or external forces applied to the cell9,10. Forces on tension-sensitive FA proteins drive conformational changes that unmask binding sites for non-tension-sensitive proteins11. Thus, tension-mediated protein recruitment regulates FA composition to mediate outcome determination downstream of integrin-mediated adhesion12.
The cascade of FA compositional changes during myosinII-mediated FA maturation is being elucidated for a small fraction of the Adhesome. After integrin activation by talin13, the adapter paxillin and the tyrosine kinase FAK are recruited to nascent FA14–16. FA growth is accompanied by recruitment of vinculin17,18, which strengthens the integrin–talin-actin link19,20. This is followed by formin-mediated elongation of an actin/α-actinin bundle, where adapter proteins zyxin and tensin accumulate9,17,21. Tension on integrins also promotes FAK and Src activation22,23 which phosphorylate tyrosines on paxillin and p130cas forming binding sites for SH2 domain-containing proteins24. However, major differences in biological response downstream of integrin engagement are likely mediated by more significant FA compositional Here, we utilized proteomic analysis to characterize FA composition changes induced by myosinII inhibition. We find myosinII mediates major compositional changes in FA. We focus on a Rac1 guanine nucleotide exchange factor (GEF), β-Pix whose FA abundance was enhanced by myosinII inhibition and demonstrate its role in negative regulation of FA maturation by promoting Rac1 activation, lamellipodial protrusion, and nascent FA turnover.
To characterize FA composition and its modification by myosinII activity, we developed a FA isolation method for HFF1 fibroblasts, minimizing contamination by actin stress fibers25 (Fig. 1a). Cells were hypotonically shocked and strongly triturated to remove cell bodies and majority of the cytoskeleton, leaving substrate-bound FAs, together with a subset of thin and/or fragmented stress fibers for collection and analysis. Imaging cells expressing GFP-paxillin during hypotonic shock revealed no effects on FA morphology (Supplemental Fig. S1a). Anti-paxillin immunostaining of FA in intact cells, hypotonically shocked cells, and FAs on the substrate after trituration revealed no difference in size and spatial organization (Fig. 1b). Quantifying fluorescence density (intensity/μm2) of immunolocalized paxillin, vinculin, zyxin, talin, phospho-tyrosine, and VASP in FAs showed that neither hypotonic shock nor trituration altered their FA abundance compared to intact cells, and no proteins remained on the substrate after FA collection (Supplemental Fig. S1b). We also compared the concentration of FA proteins (paxillin, vinculin, talin, VASP, actin, phosphorylated paxillin, and phosphorylated FAK), soluble (GAPDH), cytoskeletal (tubulin), and membrane-associated components (Akt and FGFR) in isolated FA and cell body fractions by western blot. This confirmed that the isolation method concentrated FA components in the FA fraction and separated and concentrated soluble, cytoskeletal, and membrane-associated components in the cell body fraction (Fig. 1c). The presence of FA proteins in the cell body fraction agrees with their known localizations to non-FA structures. Therefore, our FA isolation method preserves native FA organization and size and association of known FA proteins.
Isolated HFF1 FAs were then analyzed by multidimensional protein identification technology mass spectrometry (MudPIT MS)26. FAs were collected, solubilized, denatured, and the abundant proteins actin and fibronectin removed by immunodepletion. Remaining proteins were reduced, alkylated, protease digested and separated by two-dimensional liquid chromatography prior to MS analysis. Tandem mass spectra were generated and matched to peptides in a human protein database using SEQUEST™27. Each protein is represented by multiple mass spectra (sequence counts) from different peptide fragments, and each peptide fragment can be detected multiple times. Protein abundance was thus quantified as the sum of all mass spectra identified per protein (spectrum counts).
We applied two criteria for considering a protein a reproducibly identified component of isolated FAs (Fig. 2a). To be included in a “detectable list” required presence of one or more sequence count and a false positive protein identification rate of less than 1%28 (Supplemental Methods). Inclusion in the “reproducible list” required inclusion in the detectable list in at least two out of three replicate runs. By these criteria, 1917 and 754 proteins were identified as detectible and reproducible components of isolated FA, respectively (Supplemental Table 1). Although the reproducibly identified proteins may include cytosolic and cytoskeletal contaminants, this first comprehensive analysis of FA composition likely includes many undiscovered FA proteins.
We classified the reproducibly identified proteins in isolated FA according to their cell biological function. Of the 150 proteins in the Integrin Adhesome3, 72 and 50 proteins were identified in our detectable and reproducible lists, respectively. Cell type or ECM composition differences, or protein susceptibility to digestion may underlie our inability to identify all Adhesome proteins. 283 proteins were grouped into an “expected FA list” based on further literature analysis. This included the 50 proteins in the Adhesome as well as proteins in the same family (24), known to interact with (142), or subunits of a macromolecular complex (47) of an Adhesome member, as well as 20 FA associated proteins that were not included in the Adhesome (Fig. 2b). Thus, nearly 40% (283 out of 754 total) of the reproducibly identified proteins could be expected to be present in FAs based on their known properties (Fig. 2c), providing further validation of our FA isolation and protein identification method. We illustrate our FA composition, with an interactome of proteins in our expected FA list (Supplemental Fig. S2), with protein functional classification color-coded and characterized interactions indicated by overlap (Supplemental Table 2 and http://dir.nhlbi.nih.gov/papers/lctm/focaladhesion/Home/index.html).
The remaining ~60% (471 out of 754 total) of reproducibly identified proteins in isolated FA (Supplemental Table 3) were either uncharacterized, had no known association with, function in, or interaction with FAs or proteins in the Adhesome. These may be undiscovered or weakly associated or transient FA proteins, or cytosolic and/or cytoskeletal contaminants. These were classified according to biological function: endomembrane systems/trafficking (129); uncharacterized (84); cytoskeleton (81); metabolism (70); plasma membrane channels/transporters (36); RNA/DNA processing (31); cytosolic signaling (25); or folding/degradation (15). The large proportion of cytoskeletal and endomembrane trafficking proteins is not surprising, given that part of actin stress fibers were isolated with FA, and all three cytoskeletal systems associate with FAs29, while endomembrane trafficking regulates FA turnover30. Similarly, membrane channels are included in the plasma membrane present in FA. However, the purpose of this study is not comprehensive documentation of bona fide FA proteins, but rather determination of FA composition changes mediated by myosinII activity.
We next characterized the effects of myosinII inhibition on protein composition and abundance in isolated FAs. MyosinII activity was inhibited with blebbistatin (50μM, 2hr). Immunolocalization of paxillin in intact cells showed that blebbistatin abolished large, central FA and induced small, peripheral FA, with FA of a similar size and spatial distribution remaining substrate-bound after isolation (Fig. 3a,b). MudPIT MS analysis of isolated FA from blebbistatin-treated cells revealed reproducible identification of 665 proteins (Supplemental Table 4), 517 of which (78%) were reproducibly identified in FA of control cells, for a total of 905 proteins from both conditions (237 proteins only in control FAs, 148 proteins only in blebbistatin-treated FAs, and 517 proteins in both conditions).
To evaluate the relative changes in protein abundance in FAs isolated from control and blebbistatin-treated cells, we developed a normalized, ratio-based metric, the “MyosinII Dependence Ratio” (MDR), that represents the dependence of a protein’s abundance in FA on myosinII activity. MDR was the ratio of normalized protein abundance (total spectrum counts) in control FAs to that in FAs from blebbistatin-treated cells (Fig. 3c, Supplemental Methods). Because digestion and detection efficiencies are independent of sample origin, spectrum counts accurately reflect relative change in protein abundance in similarly prepared samples31,32. Thus MDR<1 indicates myosinII activity inhibits the protein’s accumulation in FA, MDR close to 1 indicates myosinII activity-independent FA association, and MDR>1 indicates myosinII activity-dependent enrichment in FA.
We validated MudPIT MS-based MDR results by comparison to western blot and two-dimensional fluorescence difference gel electrophoresis analysis (2D-DIGE)33. Western analysis (20 different proteins) and 2D-DIGE (21 different proteins) comparison of protein abundance in isolated FA from control and blebbistatin-treated cells revealed corroboration rates of 95% (Supplemental Table 5) and 90% (Supplemental Fig. S3 and Table 6) with MDRs, respectively. Thus, MDR provides a reliable metric for determining how myosinII activity effects relative abundance of proteins in FAs.
To characterize how FA composition is modulated by myosinII activity, proteins were categorized according to their MDR magnitude. Of the 905 reproducibly identified proteins in FA of control and blebbistatin-treated cells, 459 proteins (51%) showed >/= 2-fold difference in abundance between conditions. These included 335 proteins with MDR >2, indicating FA enrichment by myosinII contraction. Surprisingly, 124 proteins exhibited MDR <0.5, indicating suppression of their FA abundance by myosinII activity (Fig. 4a). Many proteins exibited changes between 1.25~2-fold (MDR: 1.25–2 or 0.5–0.8). This represents either minor myosinII-dependent changes in FA abundance, or low confidence/low detectability by MudPIT methods. Thus, myosinII activity both positively and negatively effects FA enrichment of proteins to promote major changes in FA composition.
We created a myosinII-responsive FA interactome from proteins in the expected FA list by color-coding proteins according to MDR magnitude (Supplemental Fig. S4 and Table 7, http://dir.nhlbi.nih.gov/papers/lctm/focaladhesion/Home/index.html). The interactome illustrates the full range of MDR values, including proteins exhibiting minor/low confidence changes. This interactome suggests how myosinII activity may collectively modulate FA abundance of groups of proteins mediating distinct pathways.
Proteins known to be involved in pathways of myosinII-mediated FA maturation and stress fiber formation were enriched in FA of controls compared to blebbistatin-treated cells (Fig. 4b). This included minor FA enrichment of RhoA7 (MDR:1.29) and greater enrichment of enhancers of Rho activity including GEF-H134 (MDR:2.00), TRIP635 (MDR:3.86) and testin36 (MDR:6.65). Stress fiber formation regulators enriched in control FAs included the actin bundling proteins α-actinin37 (MDR:4.19), synaptopodin-238 (MDR:2.60), supervillin39,40; formin-241 (MDR:1.86), and several PDZ-LIM cytoskeletal adapters42 (zyxin, MDR:2.91; PDLI1, MDR:6.77; PDLI4, MDR:2.93; PDLI5, MDR:1.33; PDLI7, MDR:2.11; and FHL2, MDR:2.07).
MyosinII inhibition also reduced FA abundance of factors mediating integrin-actin linkage13,43–45 (Fig. 4c). FA accumulation of migfilin (MDR:9.72) and filamins (filamin-1, MDR:2.88; filamin-B, MDR:2.31; and filamin-C, MDR:2.50) was strongly inhibited by blebbistatin. Blebbistatin also weakly inhibited FA abundance of talin-1 (MDR:1.33) and vinculin (MDR:1.41) and its binding partner, vinexin46 (MDR:1.59), Thus, myosinII activity promotes FA accumulation of integrin-cytoskeletal linkers.
Disassembly of mature FA is regulated by calpain-mediated proteolysis47 and/or endocytosis of integrins30,48. Components of both of these pathways were strongly enriched in FA from control cells compared to blebbistatin-treated cells. This included calpain-2 and components of both clathrin-dependent (clathrin heavy chain-1, MDR:2.08; clathrin adapters Dab2, MDR:2.74 and numb, MDR:1.94) and caveolin-mediated, clathrin-independent pathways (caveolin-1, MDR:2.21; scaffolding proteins flotillin-1, MDR:12 and flotillin-2 MDR:1.48) (Fig. 4d). This suggests mature FAs recruit disassembly factors by a myosinII-dependent pathway.
Surprisingly, we also found proteins involved in Rac1-mediated lamellipodial protrusion were strongly enriched in FA of blebbistatin-treated cells compared to controls, indicating negative regulation of their FA recruitment by myosinII activity. Although Rac1 (MDR:2.47) itself was more enriched in FA of control cells, myosinII inhibition caused FA enrichment of Rac1 activators, effectors, and downstream targets (Fig. 4e). Rac1 activators included the Rac guaninine nucleotide exchange factor (GEF) β-PIX49 (MDR:0.27) and the Rac GEF modulator, EPS850 (MDR:0.48), while PKA51 and its A-kinase activating protein, MAP252, MIF53, and the RhoA inhibitor RhoGDI54 (MDR:0.19) have all my indirectly promote Rac1 activation. Rac1 effectors included IRSP5355 and N-WASP55,56 (MDR:0.26), both mediators of Arp2/3-dependent actin polymerization. Rac1 downstream targets involved in actin treadmilling included Arp2/3 ((ARC1A (MDR:0.24), ARC1B (MDR:0.68), ARPC3 (MDR:0.69), ARP5L (MDR:0.74)), cofilin57 (MDR:0.46) and its activator PP2A58 (MDR:0.21), and the actin monomer binding protein Cap159 (MDR:0.25).
To understand how myosinII activity negatively regulates Rac1 and lamellipodial protrusion, we focused on the Rac GEF β-PIX. Immunoblotting FA fractions validated myosinII’s negative regulation of β-PIX accumulation in FA, showing a 3.73-fold concentration of β-PIX in FA by myosinII inhibition compared to control, similar to the MDR (3.7-fold (1/MDR) (Fig. 5a)). Quantifying the ratio of fluorescence density of paxillin and β-PIX in FA showed that myosinII inhibition did not significantly effects paxillin FA density, but significantly increased (25%) FA density of β-PIX compared to control (Fig. 5b,c). β-PIX accumulation in FAs was also regulated by physiological downregulation of myosinII contractility60, since western blot of isolated FAs and immunofluorescence analysis revealed increased accumulation of β-PIX in FAs from cells plated on compliant compared to stiffer substrates(Fig. 5d,e,f). Thus, β-PIX concentration in FAs is negatively regulated by myosinII contractility.
To determine if FA maturation modulates FA abundance of β-PIX, we quantified the ratio of fluorescence density in immature (<2μm2) relative to mature (>2μm2) FA. This revealed 9% more paxillin in big compared small to FAs, but 25% less β-PIX (Fig. 5b,c). Analysis of mApple-paxillin and eGFP-β-PIX during FA maturation showed that paxillin and β-PIX concentrations and FA area all increased at similar rates as nascent FA assembled. Subsequently, FA area increase slowed, paxillin continued to accumulate, but β-PIX level did not increase and eventually decreased in large, paxillin-dense FA (Fig. 5g). Analysis of paxillin and β-PIX during perfusion of Y27632 to inhibit ROCK (10μM; blebbistatin could not be used due to phototoxicity) revealed that reducing contraction caused dissociation of paxillin and FA shrinkage, concomittant with association of β-PIX, followed by adhesion turnover (Fig. 5h and Supplemental Fig. S5). Thus, β-PIX concentrates in assembling nascent FAs in the absence of myosinII contractility, and it dissociates from FA during myosinII-meditated FA maturation.
We next examined the role of β-PIX in Rac1 activation and lamellipodial formation induced by myosin II inhibition61,62. Immunolocalization of paxillin and the lamellipodia marker, cortactin, in GFP-β-PIX-expressing cells showed that β-PIX localized in nascent FAs in lamellipodia63,64 induced by blebbistatin (Fig. 6a). Lentiviral expression of shRNAs to inhibit β-PIX expression abrogated the blebbistatin-induced enhancement of cell spreading and decreased the ratio of lamellipodia area to total cell area65 (Fig. 6b and Supplemental Fig. S7). Rac1-GTP pulldown assays revealed that in cells expressing control shRNAs, Rac1 activity was increased by blebbistatin treatment61 (Fig. 6c), while blebbistatin-induced Rac1 activation was abolished by β-PIX knockdown. This inhibitory effect was not induced by silencing the related FA-associated Rac GEF, α-PIX, and was rescued by re-expressing myc-tagged mouse β-PIX (Fig. 6c, Supplemental Fig. S6). Thus, β-PIX is required for lamellipodial formation, cell shape changes, and Rac1 activity induced by inhibition of myosinII.
To determine the role of β-PIX in FA maturation, we altered β-PIX levels and analyzed FA size. Overexpression of β-PIX abrogated mid-sized (2–6 μm2) FA, promoted small FA (<2 μm2) at the cell periphery, and significantly increased the proportion of small FA compared to controls (Fig. 7a,b). In contrast, silencing β-PIX abrogated small peripheral FA and increased mid-sized FA (Fig. 7a,b). Analysis of mApple-paxillin (Fig. 7c) revealed that β-PIX silencing had little effect on FA assembly rate (Fig. 7c) or maximal paxillin level (Fig. 7e), but slowed FA disassembly (Fig. 7c), significantly increasing FA lifetime compared to control (Fig. 7d). Silencing β-PIX also reduced migration velocity compared to controls, and abrogated the enhancement of cell velocity induced by inhibiting myosinII with blebbistatin (Fig. 7f, Supplemental Table 9). Thus, β-PIX drives rapid nascent FA turnover, negatively regulating FA maturation and enhancing cell migration.
We developed a proteomics approach to profile global FA composition changes in response to inhibition of myosinII activity to better understand myosinII-driven FA maturation. This method provides a powerful approach for dissecting FA composition changes in response to different stimuli or molecular perturbations25. Our FA proteome contained 905 proteins, supporting the notion that FA are complex organelles. Whether all these proteins participate in FA function is unknown. More importantly, we show that protein abundance of half (459/905) the FA proteome is altered by myosinII activity and half remain constant as core FA components or non-specific contaminants. Most FA compositional change is likely not caused by myosinII ATPase activity acting directly on FA proteins or indicative of protein tension sensitivity per se, but reflects downstream effects of myosinII-mediated mechanotransduction. Our results define the myosinII-responsive FA proteome, indicating substantial compositional differences between immature and mature FA. Inhibiting myosinII reduced FA abundance of 73% of the myosinII-responsive FA proteome, supporting the established role of myosinII as an enhancer of FA protein recruitment. However, surprisingly, inhibition of myosinII induced FA enrichment of 27% of the myosinII-responsive FA proteome, indicating for the first time that myosinII is an important negative regulator of FA protein recrutiment.
FA composition in myosinII-inhibited cells likely reflects that in nascent FA that form and turn over rapidly during lamellipodial protrusion in the absence of local myosinII activity66,67, and which are critical to cell migration17. Though our controls include FA of various maturation states, their FA composition is likely dominated by larger and more abundant matured FA in the lamella and cell center which are disassembling or specialized for ECM remodeling. Whether similar FA compositional changes occur is being investigated.
Our results suggest a model in which myosinII activity regulates FA maturation and turnover by collectively modulating the FA abundance of protein functional modules mediating specialization of FA for cell migration. FA abundance of inside-out integrin activation, (talins, kindlins) was not myosinII-sensitive, supporting the notion that nascent FA formation is contractility-independent17. Our results suggest that, independent of contractility, activated integrin complexes recruit Rac1 activators (β-PIX, EPS8, MIF, PKA) and Rac1 effectors (IRSP53, N-WASP) and their targets involved in dendritic actin treadmilling (Arp2/3, cofilin-1, CAP-1). This suggests a positive feedback mechanism in which lamellipodial protrusion is coupled to formation of nascent FA that recruit a Rac1 regulatory module which propagates further protrusion and nascent FA formation to drive cell migration.
Nascent FA subjected to myosinII contractility exhibit both dissociation and recruitment of protein functional modules that mediate FA strengthening, stress fiber formation, and mature FA disassembly (Fig. 7g). We find that myosinII promotes enrichment of proteins that strengthen the integrin-actin linkage (migfilin, filamins, vinculin). At the same time, contractility drives FA dissociation of the Rac1 regulatory module to terminate the nascent FA turnover loop and allow FA maturation. MyosinII-driven loss of the Rac1 module is mirrored by recruitment of RhoA, its activators (GEF-H1, TRIP6, testin), and downstream targets and stress fiber proteins including actin-bundling proteins (α-actinin, supervillin, formin-2, and synaptopodin-2) and cytoskeletal adapters (PDLI1, PDLI4, PDLI5, PDLI7, zyxin, and FHL2). Contraction also drives FA disassembly15 by recruiting calpain and clathrin-dependent and caveolin-mediated endocytosis protein modules..
To support the role of myosinII in negative regulation of a Rac1 regulatory module in FA, we focused on the Rac GEF β-PIX, and examined the myosinII-dependence of its role in nascent FA turnover and lamellipodial protrusion63,64. We find that β-PIX concentrates in nascent FA64, and dissociates from FA as they mature. We show that β-PIX is required for Rac1 activity and lamellipodial protrusion induced by myosinII inhibition, suggesting that β–PIX may enhance Rac1 activity locally in myosinII-free cell regions such as lamellipodia. Finally, we show that β-PIX is required for rapid nascent FA turnover, implicating it in negative regulation of FA maturation and enhancement of cell migration.
We thank J. Hildebrand (University of Pittsburgh), J. Peterson (Fox Chase Cancer Center), M. Davidson (Florida State), R.H. Chen (Academia Sinica) and M. Beckerle (University of Utah) for reagents, the NHLBI proteomics and flow cytometry facilities, William Shin and Ian Schneider for help with imaging and analysis, and Aaron Aslanian. C.M.W. is supported by NHLBI, J.R.Y. by NIH P41 RR011823.
AUTHOR CONTRIBUTIONSJ.C.K. and C.M.W designed experiments and wrote the paper; J.C.K. performed experiments, X.H. and J.R.Y performed MudPIT MS analysis; C.T.H. created the web site.