Synergistic phase separation of two pathways promotes integrin clustering and nascent adhesion formation

  1. Lindsay B Case  Is a corresponding author
  2. Milagros De Pasquale
  3. Lisa Henry
  4. Michael K Rosen  Is a corresponding author
  1. Department of Biophysics, Howard Hughes Medical Institute, The University of Texas Southwestern Medical Center, United States
  2. Department of Biology, Massachusetts Institute of Technology, United States

Abstract

Integrin adhesion complexes (IACs) are integrin-based plasma-membrane-associated compartments where cells sense environmental cues. The physical mechanisms and molecular interactions that mediate initial IAC formation are unclear. We found that both p130Cas (‘Cas’) and Focal adhesion kinase (‘FAK’) undergo liquid-liquid phase separation in vitro under physiologic conditions. Cas- and FAK- driven phase separation is sufficient to reconstitute kindlin-dependent integrin clustering in vitro with recombinant mammalian proteins. In vitro condensates and IACs in mouse embryonic fibroblasts (MEFs) exhibit similar sensitivities to environmental perturbations including changes in temperature and pH. Furthermore, mutations that inhibit or enhance phase separation in vitro reduce or increase the number of IACs in MEFs, respectively. Finally, we find that the Cas and FAK pathways act synergistically to promote phase separation, integrin clustering, IAC formation and partitioning of key components in vitro and in cells. We propose that Cas- and FAK-driven phase separation provides an intracellular trigger for integrin clustering and nascent IAC formation.

Editor's evaluation

This paper identifies phase separation as underlying mechanism that imitates/contributes to the formation of Integrin-containing adhesion sites. The authors demonstrate that p130Cas and FAK undergo phase separation, concentrate kindlin and eventually integrins. The experiments and findings are well described and controlled. This paper is an important contribution to the understanding of how integrins and focal adhesion proteins cluster to form cell attachment sites.

https://doi.org/10.7554/eLife.72588.sa0

Introduction

Integrin-mediated adhesion complexes (IACs) are plasma membrane-associated compartments that provide specific adhesion between cells and their surroundings (Case and Waterman, 2015; Chastney et al., 2021). IACs serve as sites of force transmission between the actin cytoskeleton and the extracellular matrix (ECM) to drive tissue morphogenesis, cell movement, and ECM remodeling. Additionally, they serve as signaling hubs where cells sense biochemical and physical cues in their environment to regulate the cell cycle, differentiation and death. Thus, IACs mediate an array of functions involving biochemical and physical interactions between the cell and its environment.

Integrins are heterodimeric transmembrane receptors that specifically bind to ligands within the ECM or to receptors on adjacent cells (Hynes, 2002). Both chains of the α/β heterodimer have a large extracellular domain that imparts ligand specificity and a small cytoplasmic domain (20–50 amino acids) that mediates intracellular signaling. Since integrins lack catalytic activity and do not directly bind actin (Hynes, 2002), downstream signaling requires the macromolecular assembly of integrins with cytoplasmic adaptor proteins, signaling molecules, and actin binding proteins to form IACs. Thus, molecular interactions that regulate integrin clustering and initial IAC formation control integrin-dependent signaling (Hotchin and Hall, 1995; Miyamoto et al., 1995; Robertson et al., 2015; Theodosiou et al., 2016).

IACs initially form as puncta (~120 nm diameter) within the lamellipodia, termed nascent adhesions. Nascent adhesion formation is dependent on ligand binding, integrin conformational activation, and mechanical forces imparted by retrograde actin flow (Changede et al., 2015; Choi et al., 2008). Although nascent adhesion formation and cell spreading are impaired on soft substrates (where mechanical forces are reduced), this can be rescued by inducing the integrin conformational change with Mn2+, suggesting that during the initial steps of nascent adhesion assembly force is only required to activate integrins (Oakes et al., 2018). Kindlin and talin are adaptor proteins that directly bind the β integrin cytoplasmic domain (Sun et al., 2019). Kindlin binding is required for integrin clustering and nascent adhesion assembly, although it is not clear how kindlin orchestrates this higher order assembly (Theodosiou et al., 2016; Ye et al., 2013). Many proteins appear to simultaneously assemble into nascent adhesions, including α5β1 integrin, αVβ3 integrin, talin, kindlin, focal adhesion kinase (FAK), paxillin, and p130Cas (Bachir et al., 2014; Changede et al., 2015; Choi et al., 2008; Donato et al., 2010; Lawson et al., 2012; Yu et al., 2011). After initial formation, nascent adhesions undergo dramatic growth and compositional maturation dependent on talin, increased forces, and bundled actin filaments (Choi et al., 2008; Kuo et al., 2011; Oakes et al., 2012; Schwarz and Gardel, 2012). While IAC maturation is reasonably well understood, the specific molecular interactions that regulate initial nascent adhesion formation are unclear. In this study, we investigate how specific protein-protein interactions contribute to nascent adhesion formation.

Liquid-liquid phase separation driven by weak interactions between multivalent molecules has emerged as an important mechanism that can drive the formation of micron-sized, membraneless cellular compartments, termed ‘biomolecular condensates’ (Banani et al., 2017; Case et al., 2019a; Shin and Brangwynne, 2017). At the plasma membrane, phase separation can promote the assembly of transmembrane proteins and their cytoplasmic binding partners into dynamic, micron-sized clusters (Banjade and Rosen, 2014; Beutel et al., 2019; Case et al., 2019a; Su et al., 2016; Zeng et al., 2016). Phase separation is highly dependent on molecular valence (Li et al., 2012), and increasing the valence of interactions reduces the concentration threshold required for phase separation. Condensates often possess liquid-like material properties, can contain hundreds of molecular constituents, and form over a wide range of molecular stoichiometries (Banani et al., 2017; Case et al., 2019b). Condensate composition can vary dramatically and change rapidly in response to signals (Banani et al., 2016; Markmiller et al., 2018; Youn et al., 2018).

IACs are discrete, micron-sized structures that exhibit many characteristics of phase separated compartments. They are highly enriched in multivalent adaptor proteins whose valence can be modulated by phosphorylation and mechanical force (Pellicena and Miller, 2001; Schiller and Fässler, 2013; Yao et al., 2016), concomitant with changes in IAC size. IACs often exhibit liquid-like material properties such as rapid constituent exchange (Supplementary file 1; Lavelin et al., 2013; Pasapera et al., 2010; Stutchbury et al., 2017) and the ability to fuse (Berginski et al., 2011; Changede et al., 2015). IACs contain hundreds of different proteins (Schiller and Fässler, 2013), are stoichiometrically undefined (Bachir et al., 2014), and their composition can change dramatically under different cellular conditions (Horton et al., 2015; Kuo et al., 2011; Schiller and Fässler, 2013). Because of their properties and molecular composition, we hypothesized that multivalent oligomerization and phase separation of IAC-associated proteins might contribute to integrin clustering and nascent adhesion assembly.

Here, we investigated biochemical interactions among the earliest proteins to arrive at nascent adhesions. We found that several of these undergo liquid-liquid phase separation when combined in vitro at physiologic concentrations and under physiologic buffer conditions. Multivalent interactions between phosphorylated p130Cas (‘pCas’), Nck and N-WASP are sufficient for phase separation. ‘FAK’ also phase separates, and multivalent interactions between FAK and paxillin enhance this behavior. pCas and FAK undergo phase separation synergistically to form droplets that more strongly concentrate kindlin, and consequently integrin. Using a novel experimental platform to reconstitute β1 integrin clustering on supported phospholipid bilayers, we found that kindlin likely plays a central role in nascent adhesion assembly by coupling the pCas and FAK pathways to integrins. In vitro condensates and cellular IACs exhibit similar sensitivity to temperature and pH. Further, mutations in Cas or FAK that inhibit phase separation in vitro reduce the number IACs in cells, while paxillin mutations that enhance phase separation in vitro increase the number of IACs in cells. Finally, we find that the Cas and FAK pathways act synergistically to promote phase separation, integrin clustering, nascent adhesion formation and partitioning of key components in vitro and in cells. We propose that pCas- and FAK-driven phase separation provides an intracellular trigger for integrin clustering and nascent adhesion formation.

Results

Multivalent interactions promote phase separation of p130Cas under physiologic conditions

The adaptor protein Cas (p130Cas/BCAR1) is present in nascent adhesions (Donato et al., 2010) and has been implicated in promoting IAC formation (Meenderink et al., 2010). Cas contains an N-terminal SH3 domain, a central disordered ‘substrate’ domain, a Src-binding domain, and a C-terminal Cas-family homology domain (Figure 1—figure supplement 1; Meenderink et al., 2010). Within the substrate domain, there are 15 YXXP motifs that, when phosphorylated, can bind to the SH2 domain-containing adaptor proteins, Crk-II and Nck (Iwahara et al., 2004; Pellicena and Miller, 2001; Schlaepfer et al., 1997). In cells, Cas is robustly phosphorylated on 10 of its YXXP motifs, and proper cell migration requires a minimum of four phosphorylation sites (Shin et al., 2004). Crk-II contains two SH3 domains which can bind adaptor proteins such as SOS and C3G that contain multiple proline-rich motifs (PRMs) (Birge et al., 2009). Similarly, Nck contains three SH3 domains which can bind adaptor proteins such as N-WASP, SOS, and Abl that contain multiple PRMs (Li et al., 2001). In other signaling pathways, phosphorylation of proteins on multiple tyrosine residues and subsequent binding by multivalent adaptor proteins can promote phase separation (Banjade and Rosen, 2014; Kim et al., 2019; Su et al., 2016). For example, multivalent interactions between phosphorylated Nephrin, Nck, and N-WASP or phosphorylated Lat, Grb2 and SOS drive phase separation to form droplets in solution and clusters at membranes (Banjade and Rosen, 2014; Banjade et al., 2015; Kim et al., 2019; Li et al., 2012; Su et al., 2016). Thus, we sought to determine if the multiply-phosphorylated Cas substrate domain could promote phase separation, similar to phosphorylated Nephrin and Lat. Nck has been localized within IACs (Goicoechea et al., 2002; Horton et al., 2015), N-WASP colocalizes with Cas within the lamellipodia where nascent adhesions form (Zhang et al., 2014), and both Nck and N-WASP have been implicated in regulating cell adhesion to fibronectin (Misra et al., 2007; Ruusala et al., 2008). Thus, we chose to use Nck as the SH2/SH3 domain containing adaptor protein and N-WASP as the PRM containing adaptor protein for our studies (Figure 1a, Supplementary file 2). To assess phase separation in vitro, we purified recombinant proteins (Figure 1—figure supplement 2), phosphorylated Cas on tyrosine to an average of ~17–19 sites/molecule (pCas, Figure 1—figure supplement 3), and assessed phase separation by both measuring solution turbidity (Absorbance at 350 nm) and visualizing droplets with fluorescence microscopy. For microscopy measurements, we conjugated specific proteins with Alexa fluorophores, as indicated in figure legends. Although purified pCas is more highly phosphorylated than in cells, the valency for Nck is likely similar (five of the six Nck-binding motifs are robustly phosphorylated in cells) (Shin et al., 2004). For all in vitro experiments, we used buffer containing 50 mM HEPES (pH 7.3), 50 mM KCl, 1 mM TCEP, and 0.1% BSA. No crowding agents were used in any experiments. In mammalian cells, the concentration of Nck is ~200 nM, that of N-WASP (and its homolog, WASP) ranges from 150 nM to 10 μ M, and that of Cas is ~70 nM (Supplementary file 3, Hein et al., 2015; Higgs and Pollard, 2000; Isaac et al., 2010; Roybal et al., 2016). We combined Nck and N-WASP at physiologic concentrations (200 nM Nck/1 μ M N-WASP and 500 nM Nck/1 μ M N-WASP) and titrated increasing concentrations of unphosphorylated or phosphorylated Cas. We specifically observed an increase in solution turbidity at physiologic concentrations (10–200 nM) of pCas (Figure 1b). Spinning disk confocal fluorescence microscopy confirmed that the increase in solution turbidity was due to formation of condensed foci at physiologic concentrations (Figure 1—figure supplement 4). We increased protein concentrations to 1 μ M to ensure that the foci were larger than the point spread function of the microscope, revealing that they are spherical (Figure 1c) and suggesting that they behave as liquid droplets. We measured the partition coefficient (PC) of molecules into the droplets (Intensity inside droplet/[Intensity in bulk solution]; Figure 1d, Figure 1—figure supplement 5, see Materials and methods) and performed fluorescence recovery after photobleaching (FRAP) analysis (Figure 1—figure supplement 6). We found that pCas, Nck and N-WASP rapidly exchanged between droplets and bulk solution (Figure 1e; Supplementary file 4; pCas t1/2 fast = 2 s; pCas t1/2 slow = 39s; Nck t1/2 = 8 s; N-WASP t1/2 fast = 1 s; N-WASP t1/2 slow = 37 s), although a fraction of Nck and pCas molecules did not recover in the 100 s timeframe of the experiment, suggesting a slower phase of recovery and/or an immobile fraction (Supplementary file 4). In cells, IAC-associated proteins, including Cas, often contain a population of fast exchanging molecules (t1/2 < 30 s) as well as an immobile fraction (up to 50% immobile for some proteins; Supplementary file 1). Thus, similar to cellular IACs, droplets exhibit liquid-like material properties with some solid-like elements as well. We conclude that interactions between multiply-phosphorylated Cas, Nck, and N-WASP are sufficient to promote liquid-liquid phase separation at physiologic protein concentrations.

Figure 1 with 6 supplements see all
p130Cas, Nck, and N-WASP undergo liquid-liquid phase separation.

(A) Molecular interactions of IAC proteins, KD values indicated where known. Details and references in Supplementary file 1. (B) Solution turbidity measurements. Nck (200 nM, green; or 500 nM, magenta+ black) and N-WASP (1 µM) were combined with increasing concentrations of phosphorylated Cas (pCas, green+ magenta) or unphosphorylated Cas (Cas, black). Each point represents the mean ± SEM of three independent measurements. (C) Spinning disk confocal fluorescence microscopy images of droplets. Nck (1 μM, 15% Alexa568 labeled) and N-WASP (1 μM, 15% Alexa647 labeled) were combined ± pCas (1 μM, unlabeled). (D) Quantification of constituent partitioning into droplets. Each grey point represents an individual measurement, and the mean indicated by black line. Each condition contains at least 75 measurements from two or more independent experiments. (E) Fluorescence Recovery After Photobleaching (FRAP) measurements of droplets. Droplets formed from 1 μM each of Nck (15% Alexa568 labeled), N-WASP (15% Alexa647 labeled) and pCas (5%–647 labeled). Each point represents the mean ± SEM of at least six independent measurements. Recovery curves were fit with a single exponential (Nck) or biexponential (pCas, N-WASP) model and the fits are overlayed on the graph (black line). Detailed fit information in Supplementary file 4. All scale bars = 10 μm.

FAK and paxillin phase separate under physiologic conditions

Like many condensates, IAC assembly is robust to changes in composition. Although Cas knockout (KO) fibroblasts still form IACs, they exhibit decreased adhesion assembly rates (Meenderink et al., 2010). These observations suggest a potential role for Cas in nascent adhesion assembly that is at least partially compensated for by additional proteins (Honda et al., 1998). Thus, we sought to identify additional interactions that may promote integrin clustering at IACs, perhaps through phase separation (Figure 2a, Supplementary file 2). Talin, kindlin, paxillin, and FAK regulate nascent adhesion assembly, are present in nascent adhesions, and rapidly exchange between IACs and the cytoplasm (Supplementary file 1; Bachir et al., 2014; Changede et al., 2015; Choi et al., 2008; Meenderink et al., 2010; Swaminathan et al., 2016; Theodosiou et al., 2016). Full-length talin is autoinhibited (Dedden et al., 2019), but the talin head domain (talinH) is sufficient to rescue nascent adhesion formation in talin depleted cells (Changede et al., 2015). We expressed and purified recombinant talinH, kindlin, paxillin, and FAK (Figure 1—figure supplement 2) and screened for phase separation in vitro by measuring solution turbidity. We observed an increase in solution turbidity with increasing concentrations of FAK, starting at ~100 nM, but not with increasing concentrations of talinH, kindlin, or paxillin (Figure 2b, Figure 2—figure supplement 1a). Note that we purify FAK in buffer containing 300 mM NaCl, preventing self-assembly prior to dilution into the experimental buffer. Using spinning disk confocal fluorescence microscopy, we observed small FAK foci at 40 nM concentration (Figure 2—figure supplement 2a-b), the upper end of FAK concentrations in mammalian cells (5–40 nM) (Brami-Cherrier et al., 2014; Hein et al., 2015; Supplementary file 3). With 1 μ M FAK (1% Alexa-647-labeled), we observe larger, micron-scale spherical droplets (Figure 2c) that exchange molecules with bulk solution as assessed by FRAP (Figure 2g. Figure 2—figure supplement 3a; Supplementary file 4; t1/2 fast = 6 s; t1/2 slow = 56 s), although a fraction of molecules (38%) did not recover. Thus, droplets exhibit liquid-like material properties, albeit with some solid-like elements as well. We conclude that FAK undergoes liquid-liquid phase separation in vitro under physiologic conditions.

Figure 2 with 3 supplements see all
FAK and paxillin undergo liquid-liquid phase separation.

(A) Molecular interactions of IAC proteins, KD values indicated where known. Details and references in Supplementary file 1. (B) Solution turbidity measurements with increasing concentrations of FAK. (C) Spinning disk confocal fluorescence microscopy images of droplets formed from 1 μM FAK (1% Alexa647 labeled). The same image is displayed with two different contrast settings for comparison to panel E. (D) Solution turbidity measurements. 250 nM FAK was combined with increasing concentrations of talinH (magenta), kindlin (black), or paxillin (green). (E) Spinning disk confocal fluorescence microscopy images of droplets formed from 1 μM FAK (1% Alexa647 labeled) and 1 μM FAK paxillin (5% Alexa546 labeled). For comparison, contrast settings of FAK image are identical to those in 2 C, low contrast. (F) Quantification of constituent partitioning into droplets. Each condition contains at least 750 measurements from four or more independent experiments. Significance tested with student’s t-test. (G) Fluorescence Recovery After Photobleaching (FRAP) measurements of droplets. Droplets formed from 1 μM of FAK (1% Alexa647 labeled, cyan) or 1 μM each of FAK (1% Alexa647 labeled, dark blue) and paxillin (5% Alexa546 labeled, magenta). Each point represents the mean ± SEM of at least 15 independent measurements. Recovery curves were fit with a biexponential model and the fits are overlayed on the graph (black line). Detailed fit information in Supplementary file 4. In (B) and (D), each point represents the mean ± SEM of three independent measurements. In (F), each gray point represents an individual measurement, and the mean is indicated by black line. All scale bars = 10 μm.

Paxillin enhances FAK phase separation

Next, we sought to identify additional interactions that might enhance FAK phase separation. FAK can interact directly with paxillin and talin (Lawson et al., 2012; Thomas et al., 1999). Thus, we added increasing concentrations of paxillin, talinH, or kindlin to 250 nM FAK. We found that paxillin, but not talinH or kindlin, increased solution turbidity (Figure 2d). Turbidity of a solution containing 250 nM FAK and paxillin was not increased by addition of talinH or kindlin (Figure 2—figure supplement 1b). When paxillin (5% Alexa546) and FAK (1% Alexa647) were combined at 1 μ M concentration, micron-sized spherical droplets containing both proteins were formed (Figure 2e). In the presence of equimolar paxillin, FAK enrichment in the droplets increased twofold (Figure 2f). Both FAK and paxillin exchange between droplets and bulk solution in FRAP analyses (Figure 2g, Figure 2—figure supplement 3b-c; Supplementary file 4); FAK t1/2 fast = 10 s; FAK t1/2 slow = 98 s; paxillin t1/2 fast = 7 s, paxillin t1/2 slow = 78 s, although FAK recovery is slowed relative to droplets containing FAK alone. Thus, similar to cellular IACs (Supplementary file 1), droplets formed with FAK and paxillin exhibit predominantly liquid-like material properties with some solid-like elements. Finally, we found that paxillin enhances droplet formation even at low nanomolar concentrations of FAK and paxillin (Figure 2—figure supplement 2c-d). We conclude that interactions with paxillin enhance FAK phase separation.

The pCas and FAK pathways phase separate synergistically

Our results suggest that pCas-dependent and FAK-dependent phase separation can promote higher order assembly of IAC-associated proteins. Furthermore, pCas directly interacts with both paxillin and FAK (Wisniewska et al., 2005; Zhang et al., 2017) and Cas, FAK, and paxillin have been observed to associate as a complex in the cytoplasm (Hoffmann et al., 2014), suggesting these two pathways could function together to regulate IAC formation (Figure 3a, Figure 1—figure supplement 1, Supplementary file 2). Thus, we sought to better understand the relationship between pCas- and FAK- driven phase separation in vitro. When we combined 1 μ M each of pCas, Nck, N-WASP, FAK, and paxillin (‘pCas + FAK mix’), we observed a single class of droplets containing all molecules (Figure 3b). Moreover, partitioning of all molecules into droplets increased two- to fivefold with the pCas + FAK mix (Figure 3c vs Figures 1d and 2f). Next, we combined lower concentrations of proteins (either 250 nM of each or the cytoplasmic concentrations (Supplementary file 3)) and measured solution turbidity in buffers with increasing concentrations of salt. Turbidity was observed at salt concentrations up to 100 mM (cytoplasmic protein concentrations) or 150 mM (uniform 250 nM protein concentration), and generally decreased with increasing salt, consistent with phase separation driven by electrostatic interactions (Figure 3—figure supplement 1). We conclude that the pCas and FAK pathways undergo phase separation synergistically to form droplets that more strongly concentrate cytoplasmic adaptor proteins.

Figure 3 with 2 supplements see all
The pCas and FAK pathways phase separate synergistically.

(A) Molecular interactions and known KD values, including between the two pathways. Details and references in Supplementary file 1. (B) Spinning disk confocal fluorescence microscopy images of droplets formed with the pCas + FAK mix. TOP: 1 μM each of Nck (15% alexa647), N-WASP (15% Alexa568), pCas, FAK, paxillin (15% Alexa488); MIDDLE: 1 μM each of Nck, N-WASP, pCas, FAK (1% Alexa647 labeled), paxillin (1% Alexa488 labeled); BOTTOM: 1 μM each of Nck (15% Alexa488 labeled), N-WASP, pCas (1% Alexa647 labeled), FAK, paxillin; (C) Quantification of constituent partitioning into droplets. Each condition contains at least 750 measurements from four or more independent experiments. (D) Quantification of integrin-GFP partitioning into droplets formed with the pCas + FAK mix. 1 μM each of Nck, N-WASP, pCas, FAK, paxillin, and B1 Integrin (15% GFP labeled) with either 1 μM Kin or 1 μM TlnH. Each condition contains at least 70 measurements from two or more independent experiments. (E) Spinning disk confocal fluorescence microscopy images of droplets. Top: droplets formed with the pCas mix (1 μM each of Nck [15% Alexa647 labeled], N-WASP, pCas, and kindlin [5% Alexa568 labeled]). Middle: Droplets formed with the FAK mix (1 μM each of FAK (1% Alexa647 labeled), paxillin, and kindlin (5% Alexa568 labeled)). Bottom: Droplets formed with the pCas + FAK mix (1 μM each of Nck, N-WASP, pCas, FAK (1% Alexa647 labeled), paxillin, and kindlin (5% Alexa568 labeled)). Contrast settings of kindlin images are matched in all images. (F) Quantification of kindlin partitioning. Each condition contains at least 60 measurements from two or more independent experiments. (G) Spinning disk confocal fluorescence microscopy images of droplets. Top: Droplets formed with the pCas mix (1 μM each of Nck [15% Alexa647 labeled], N-WASP, pCas, kindlin, and integrin [15% GFP labeled]). Middle: Droplets formed with the FAK mix (1 μM each of FAK [1% Alexa647 labeled], paxillin, kindlin and integrin [15% GFP labeled]). Bottom: Droplets formed with the pCas + FAK mix (1 μM each of Nck, N-WASP, pCas, FAK [1% Alexa647 labeled], paxillin, kindlin, and integrin [15% GFP labeled]). Contrast settings of integrin images are matched in all images. (H) Quantification of integrin partitioning. Each condition contains at least 70 measurements from two or more independent experiments. In (D), (F) and (H) significance tested by one-way ANOVA followed by a Tukey multiple comparison test. All scale bars = 5 μm.

Kindlin recruits integrin into pCas+FAK droplets

Next, we sought to determine whether phase separated droplets could recruit the β1 integrin cytoplasmic tail. Droplets formed from 1 μ M pCas, Nck, N-WASP, FAK, and paxillin did not recruit β1 integrin (PC = 1.2), and this was not changed by addition of 1 µM talinH (PC = 1.2). However, adding 1 μ M kindlin, significantly increased integrin partitioning in droplets (PC = 3.0, Figure 3d). Thus, kindlin specifically couples phase separation of pCas and FAK to the integrin cytoplasmic tail.

To better understand how kindlin recruits integrin into droplets, we measured kindlin partitioning into droplets formed from the pCas mix, FAK mix, and pCas + FAK mix (Figure 3e–f). We found that kindlin weakly partitioned into droplets formed with either the pCas mix or the FAK mix, but its partitioning was significantly increased in droplets formed with the pCas + FAK mix (Figure 3f). Although kindlin can directly bind paxillin (albeit with low affinity, KD = 200 µM) (Böttcher et al., 2017; Zhu et al., 2019), the protein has not been reported to interact with Cas, Nck, or N-WASP (Dong et al., 2016). Thus, we next sought to determine how kindlin partitions into droplets containing only pCas, Nck, and N-WASP (the pCas mix). First, we measured kindlin partitioning into previously described droplets formed by phosphorylated Nephrin, Nck and N-WASP but lacking pCas (Li et al., 2012). Κindlin did not enrich in these droplets (PC = 1), suggesting that kindlin does not interact strongly with pNephrin, Nck, or N-WASP (Figure 3—figure supplement 2a-b). Thus, we hypothesized that kindlin partitions into droplets containing pCas, Nck and N-WASP by binding pCas. To determine if Kindlin could directly interact with Cas, we performed a pull-down experiment. We found that His-tagged kindlin can pull-down both unphosphorylated and phosphorylated Cas (Figure 3—figure supplement 2c-d). We conclude that kindlin enriches in droplets formed with the pCas mix through direct interactions with Cas that do not require Cas phosphorylation (Figure 3a). A low affinity interaction with both Paxillin and Cas is consistent with a smaller partition coefficient.

Next, we measured β1 integrin tail partitioning into droplets formed with the pCas mix, FAK mix, and pCas + FAK mix (all containing kindlin). Similar to kindlin, β1 integrin weakly partitions into droplets formed with either the pCas mix or the FAK mix, but partitioning significantly increases when both pathways are combined to form droplets with the pCas + FAK mix (Figure 3g–h). We conclude that pCas and FAK undergo phase separation synergistically to form droplets that more strongly concentrate kindlin, and consequently integrin.

pCas- and FAK-dependent phase separation synergistically promote integrin clustering on membranes

Since IACs are membrane-associated condensates, we next tested whether pCas- and FAK- dependent phase separation were sufficient to promote integrin clustering on supported phospholipid bilayers. The α and β integrin cytoplasmic tails separate upon activation (Kim et al., 2003; Wegener and Campbell, 2008), and talin and kindlin bind to the latter (Li et al., 2017). Thus, the β integrin cytoplasmic domain is the minimal fragment required to examine talin- or kindlin- dependent integrin clustering. Although the cell surface expression of integrin receptors varies under different conditions, integrin densities between 300–1500 molecules/μm2 have been observed in cells (Rossier et al., 2012; Wiseman et al., 2004). We attached His10-tagged β1 integrin cytoplasmic domain (His-β1) at a density of ~1000 molecules/μm2 on phospholipid bilayers composed of 98% phosphatidylcholine (POPC) plus 2% Ni-NTA lipids (Figure 4a; Su et al., 2017). Using Total Internal Reflection Fluorescence (TIRF) microscopy, we confirmed our methodology consistently generated fluid phospholipid bilayers (Figure 4—figure supplement 1) and that His-β1 was uniformly distributed and rapidly diffusing on the bilayer (Figure 4b–c ‘control’).

Figure 4 with 2 supplements see all
Phase separation is sufficient to reconstitute kindlin-dependent integrin clustering on supported phospholipid bilayers.

(A) Cartoon describing phospholipid bilayer reconstitution. (B) Total Internal Reflection Fluorescence (TIRF) microscopy images. Different combinations of proteins (pCas mix: 1 μM each of Nck, N-WASP, Cas and kindlin; FAK mix: 200 nM FAK, and 1 μM each of paxillin and kindlin; pCas + FAK mix: 200 nM FAK and 1 μ M each of Nck, N-WASP, Cas, paxillin and kindlin; Control: Buffer only) were added to membrane-bound integrin (15% Alexa488 labeled, ~ 1000 molecules/μm2). (C) Fluorescence Recovery After Photobleaching (FRAP) measurements of his-Integrin in clusters. Each point represents the mean ± SEM of at least 12 independent measurements. The t1/2 was calculated from a single exponential fit; fit overlayed on graph (black line). (D) Quantification of integrin clusters. Each gray point represents a single field of view, black lines represent mean ± SEM. (E) Quantification of integrin intensity within clusters formed with the pCas + FAK mix or in the surrounding unclustered regions (‘Uncluster’). Each point represents a single measurement, black lines represent mean ± SEM. Each condition contains at least 150 measurements from two independent experiments. (F, G) Quantification of droplets visible in TIRF field. Experiments were done on PEG-coated glass (PEG), empty phospholipid bilayers (lipids) and phospholipid bilayers containing unlabeled membrane-bound integrin (Lipid+ Integrin). Droplets containing Nck-Alexa647 were visualized with TIRF illumination. 1 μM concentration: 1 μ M pCas, 1 μM Nck (15% Alexa647 labeled), 1 μM N-WASP, 1 μM Kin, 1 μM Paxillin, 200 nM FAK; physiological concentration: 70 n M pCas, 180 nM Nck (15% Alexa647 labeled), 140 nM N-WASP, 60 nM Kin, 60 nM Paxillin, 40 nM FAK. Each point represents the mean ± SEM of at least 17 measurements from three independent experiments. Significance tested by one-way ANOVA followed by a Tukey multiple comparison test. All scale bars = 10 μm.

After addition of 1 μM each of pCas, Nck, N-WASP and kindlin (‘pCas mix’), we observe the formation of micron-sized integrin clusters (~1100 clusters/mm2, Figure 4b–d). Similarly, after addition of 200 nM FAK plus 1 μM each of paxillin and kindlin (“FAK mix”), we observe the formation of micron-sized integrin clusters (~500 clusters/mm2, Figure 4b–d). However, when we combined 200 nM FAK and 1 μ M each of pCas, Nck, N-WASP, paxillin, and kindlin (‘pCas + FAK mix’), we observe more than a 10-fold increase in the number of integrin clusters (12,200 clusters/mm2) compared with either the pCas mix or FAK mix alone (Figure 4b–d). Unexpectedly, we found that FAK interacts with Ni-NTA lipid and competes with His-tagged proteins for membrane binding (data not shown). To reduce this effect, we lowered the FAK concentration to 200 nM for experiments on supported phospholipid bilayers. We conclude that pCas- and FAK-driven phase separation synergistically promote integrin clustering on phospholipid bilayers.

FRAP analysis demonstrates that integrin rapidly exchanges between clusters and the surrounding membrane (Figure 4c), indicating that the clusters are dynamic assemblies. Furthermore, without the complete mixture of proteins in either the pCas mix (Figure 4—figure supplement 2 a-b) or FAK mix (Figure 4—figure supplement 2 c-d), clusters fail to form. Thus, both kindlin and molecules that promote phase separation are required for integrin clustering on bilayers.

Next, we compared fluorescence intensity of integrin inside clusters formed with the pCas + FAK mix with that in the surrounding bilayer. We found that integrin intensity within pCas + FAK clusters increased an average of twofold compared with unclustered integrins (Figure 4e). In both MEFs and CHO cells plated on fibronectin, α5β1 integrin is 1.3–2 times more concentrated in nascent adhesions compared with the surrounding regions of the membrane (Wiseman et al., 2004). Thus, kindlin coupled to pCas- and FAK-driven phase separation is sufficient to reconstitute physiologically relevant enrichment of integrin within clusters.

These higher protein concentrations are sufficient for the pCas + FAK mix to form droplets in solution (Figure 3b). Thus, we sought to determine whether integrin clustering occurred through pre-formed droplets settling on the membrane, or if condensates were also nucleated on the membrane. We formed droplets with 1 μ M pCas, 1 μM Nck (15% Alexa647 labeled), 1 μM N-WASP, 1 μM Kin, 1 μM Paxillin and 200 nM FAK (‘1 μM concentration’) and quantified the droplet density in the TIRF field on either PEG-coated glass (PEG), empty phospholipid bilayers (lipids) and phospholipid bilayers containing unlabeled membrane-bound integrin (Lipid+ Integrin). As expected, droplets nucleated in solution did settle over time, indicated by the increased droplet density on both PEG and empty lipids (Figure 4f). However, there was a significant increase in droplet density at 1 min and 5 min timepoints on bilayers containing integrin, suggesting integrin can promote de novo nucleation on bilayers. Furthermore, when we combined proteins at lower, more physiological concentrations (70 nM pCas, 180 nM Nck (15% Alexa647 labeled), 140 nM N-WASP, 60 nM Kin, 60 nM Paxillin, 40 nM FAK) we observed very few droplets on PEG (Figure 4g). There is a slight increase in droplet density on empty lipids, possibly due to interaction of FAK with the Ni-NTA lipids. Even with these physiological protein concentrations, we observe a significant increase in droplet density at 5 min and 10 min timepoints on bilayers containing integrin (Figure 4g). Thus, the β1 integrin cytoplasmic tail significantly increases the rate of condensate formation on bilayers, likely by increasing the local protein concentration at the membrane and thus accelerating condensate nucleation (Snead et al., 2021).

In vitro droplets and cellular IACs similarly respond to environmental perturbations

Since phase separation is sensitive to solvent conditions, we sought to compare how in vitro droplets and cellular IACs responded to changes in environment. First, we determined the effect of temperature on in vitro phase separation. We combined 250 nM pCas, Nck, N-WASP, FAK, and paxillin, incubated at different temperatures for 30 min, and measured solution turbidity. We observed phase separation at 4 °C, 22°C, and 37°C, although there was a small but significant decrease in solution turbidity at 4 °C (Figure 5a). The decrease in phase separation at lower temperatures is somewhat unusual, but not unheard of Jiang et al., 2015; Vrhovski et al., 1997, as for most protein systems phase separation is enhanced as temperature decreases (Nott et al., 2015; Yoshizawa et al., 2018). To determine how IACs respond to a transient change in temperature, cells were plated on fibronectin for 3 hr and then incubated at 4 °C, 22 °C, or 37 °C for 10 min, followed by fixation and immunostaining for endogenous paxillin. IACs were observed at all temperatures, but there was a decrease in total adhesion area at 4 °C (Figure 5b–c), mirroring the temperature dependence of the in vitro droplets.

Integrin adhesion complexes are sensitive to solvent perturbations that alter phase separation.

(A) Solution turbidity measurements of solution containing 250 nM each of Nck, N-WASP, pCas, FAK, and Paxillin in buffer containing 50 mM Hepes pH 7.3, 50 mM KCl and 0.1% BSA. Solution incubated at indicated temperatures of 30 min prior to measurements. (B) Total adhesion area per cell quantified from spinning disk images of endogenous paxillin. Cells were incubated at indicated temperatures for 10 min prior to fixation and paxillin immunostaining. (C) Representative spinning disk confocal microscopy images. (D) Turbidity measurements of solution containing 250 nM each of Nck, N-WASP, pCas, FAK, and Paxillin in buffer containing 50 mM KCl, 0.1% BSA and either 50 mM of Hepes or Mes at the indicated pH. (E) Total adhesion area per cell quantified from spinning disk confocal microscopy images of endogenous paxillin. Cells were incubated for 10 min prior to fixation and paxillin immunostaining in buffer containing 10 μ M nigericin, 10 μ M valinomycin, 150 mM NaCl, 50 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 2 mM Glutamax, and either 50 mM Mes pH 6.5 or 50 mM Hepes pH 7.5. Significance tested with unpaired t-test. (F) Representative spinning disk confocal microscopy images. All scale bars = 20 μm.

Next, we determined the effect of pH on phase separation. In cells, IACs are sensitive to intracellular pH (Choi et al., 2013). Mutations in Nhe1 that decrease intracellular pH from a typical resting value of ~7.5–7.0 cause an increase in IAC size and number (Denker and Barber, 2002; Srivastava et al., 2008). We tested the effect of buffer pH on phase separation in vitro and found that turbidity of solutions containing 250 nM pCas, Nck, N-WASP, FAK, and paxillin increased with increasing acidity. We measured a twofold decrease in turbidity between pH 6.5 and pH 7.5 (Figure 5c). To transiently alter intracellular pH, we treated cells with buffer containing 10 µM Nigericin+ valinomycin, ionophores that equilibrate the extracellular and intracellular pH (Triandafillou et al., 2020). We incubated cells with buffers for 10 min followed by fixation and immunostaining. We observed a twofold decrease in total adhesion area in cells treated with pH 7.5 compared with pH 6.5 (Figure 5d–e). Thus, solution pH has parallel effects in vitro and in cells, with increasing pH causing decreased pCas+ FAK phase separation and decreased total adhesion area. We conclude that solvent perturbations similarly alter phase separation in vitro and IACs in cells.

In vitro droplets and cellular IACs respond similarly to genetic perturbations

We have identified two distinct sets of molecular interactions, one pCas-dependent and one FAK-dependent, that are sufficient to promote phase separation and drive kindlin-dependent clustering of integrins in vitro. Next, we tested whether mutations that perturb Cas or FAK phase separation in vitro similarly alter IACs in cells.

Preventing Cas phosphorylation reduces droplet formation and paxillin partitioning in vitro

As noted above, unphosphorylated Cas does not phase separate in the presence of Nck and N-WASP (Figure 1b). To parallel this material in cells, all 15 tyrosines within the Cas substrate domain were mutated to phenylalanine (Y15F), which prevents phosphorylation of Cas in cells (Donato et al., 2010). Using solution turbidity measurements and microscopy, we first confirmed that recombinant CasY15F protein does not phase separate in the presence of Nck and N-WASP (Figure 6a; Figure 6—figure supplement 1). Next, we determined the effect of CasY15F on in vitro droplet formation and paxillin partitioning in the Cas + FAK mix. CasY15F can still interact with FAK through its N-terminal SH2 domain and paxillin through its C-terminal Cas homology domain (Figure 6b; Figure 1—figure supplement 1). We find that CasY15F results in fewer, smaller droplets compared with pCas (Figure 6c). Paxillin partitioning is also decreased ~1.4 fold in droplets containing CasY15F (Figure 6d).

Figure 6 with 3 supplements see all
Integrin adhesion complexes are sensitive to genetic perturbations that alter Cas-dependent phase separation.

(A) In vitro solution turbidity measurements. 500 nM Nck and 1 µM N-WASP were combined with increasing concentrations of phosphorylated Cas (pCas, magenta) or Y15F Cas (black). Each point represents the mean ± SEM of three independent measurements. (B) Representation of in vitro experiments in C-D. (C) Spinning disk confocal fluorescence microscopy images of droplets formed with the Cas + FAK mix. TOP: 1 μM each of pCas, Nck, N-WASP, Kin, FAK, and paxillin (15% Alexa546); BOTTOM: 1 μM each of Y15F Cas, Nck, N-WASP, Kin, FAK, and paxillin (15% Alexa546). (D) Quantification of paxillin partitioning into droplets. Each condition contains at least 40 measurements from two independent experiments. (E) Spinning disk fluorescence microscopy images of MEFs with immunostaining for endogenous paxillin. (F) Quantification of number of adhesions. (G) Quantification of paxillin partitioning into adhesions (partition coefficient). Significance tested by one-way ANOVA followed by a Tukey multiple comparison test. In (C) scalebar = 5 μm. In (E) scalebar = 10 μm. Note that the pCas protein used in Figure 6 is a different batch than protein used in Figures 15 (See Figure 1—figure supplement 3).

Preventing Cas phosphorylation reduces the number of IACs and paxillin partitioning in cells

Next, we sought to compare these in vitro measurements to the effect of CasY15F on IAC formation and paxillin partitioning in cells. Cas and FAK are important for both IAC formation and disassembly (Donato et al., 2010; Swaminathan et al., 2016). To distinguish between potentially confounding functions during these opposing processes, we quantified the number of IACs during initial cell spreading when formation dominates. We plated cells on fibronectin-coated glass and allowed them to spread and form IACs for 20 min. After fixation and immunostaining for endogenous paxillin, we counted the number of IACs (‘number of adhesions’, Figure 6—figure supplement 2; Horzum et al., 2014).

We first examined the role of Cas in regulating the number of IACs. We found that Cas KO MEFs (Figure 6—figure supplement 3a) formed significantly fewer IACs than WT MEFs, consistent with a potential defect in nascent adhesion formation (Figure 6e–f). We found that expressing WT Cas in Cas KO MEFs restored WT numbers of IACs, while expressing Cas Y15F failed to rescue the number of IACs (Figure 6e–f). Furthermore, the partitioning of endogenous paxillin into adhesions ([Intensity inside adhesions]/[Intensity in cytoplasm]) was decreased in Cas KO MEFs, and partitioning was rescued with WT Cas but not Cas Y15F (Figure 6g). Paxillin partitioning was ~1.4-fold higher in cells expressing WT Cas compared with Cas Y15F, consistent with the effect of reduced phase separation observed in vitro (Figure 6d). Thus, phosphorylation of Cas is required for Cas-dependent regulation of IACs during cell spreading.

Preventing FAK oligomerization inhibits phase separation and reduces the number of IACs in cells

FAK contains several molecular features that could underlie its ability to phase separate. There are at least two distinct sets of FAK-FAK intermolecular interactions that could promote higher-order oligomerization (Brami-Cherrier et al., 2014). FAK can dimerize through association of two FAK FERM domains, and the FERM:FERM interaction is then further stabilized by an additional interaction between a basic patch in the FERM domain and the C-terminal FAT domain (Figure 2a). Importantly, these self-interactions can occur in the presence of paxillin, suggesting that higher order FAK oligomerization is compatible with paxillin binding (Brami-Cherrier et al., 2014). FAK also contains an intrinsically disordered region (IDR) between the kinase domain and FAT domain that has sequence features common to IDRs that phase separate (Figure 7—figure supplement 1a; Vernon et al., 2018).

We sought to determine whether perturbing any of these molecular features would reduce FAK phase separation in vitro. Pyk2 is closely related to FAK, but the Pyk2 IDR is predicted to have a lower phase separation propensity than the FAK IDR (Figure 7—figure supplement 1a). Fus is an RNA-binding protein with an IDR that is sufficient for phase separation in vitro (Lin et al., 2015). Swapping the FAK IDR for either the Pyk2 IDR or the FUS IDR did not dramatically alter FAK phase separation in vitro (Figure 7—figure supplement 1a-b). Furthermore, a C-terminal fragment containing the IDR and FAT domain did not undergo phase separation (Figure 7—figure supplement 1c-d). Thus, the IDR is not a dominant driver of FAK phase separation. In contrast, a single point mutation in the FERM domain (W226A) that weakens the FERM-FERM interaction (Brami-Cherrier et al., 2014) was sufficient to dramatically reduce FAK phase separation measured by solution turbidity (Figure 7a) or microscopy (Figure 7—figure supplement 2). Furthermore, titrating the FAK C-terminus inhibits phase separation of the full-length protein, suggesting that the FERM-FAT interaction is also important for phase separation (Figure 7—figure supplement 1e). We conclude that FAK phase separation is primarily driven by higher order oligomerization (Figure 7—figure supplement 1f).

Figure 7 with 3 supplements see all
Integrin adhesion complexes are sensitive to genetic perturbations that alter FAK-dependent phase separation.

(A) In vitro solution turbidity measurements with increasing concentrations of recombinant WT FAK or W266A FAK (to inhibit dimerization). WT FAK data is duplicated from Figure 1B for comparison. Each point represents the mean ± SEM of three independent measurements. (B) Representation of in vitro experiments in C-D. (C) Spinning disk confocal fluorescence microscopy images of droplets formed with the pCas+ FAK mix. Top: 1 μM each of pCas, Nck, N-WASP, Kin, FAK, and paxillin (15% Alexa546); Bottom: 1 μM each of pCas, Nck, N-WASP, Kin, FAK W266A, and paxillin (15% Alexa546). (D) Quantification of paxillin partitioning into droplets. Each condition contains at least 40 measurements from two independent experiments. (E) Spinning disk fluorescence microscopy images of MEFs transiently expressing GFP-FAK variants with immunostaining for endogenous paxillin. (F) Quantification of number of adhesions. (G) Quantification of paxillin partitioning into adhesions (partition coefficient). (H) Spinning disk fluorescence microscopy images of MEFs transiently expressing GFP-FAK variants. (I) Quantification of GFP-FAK partitioning into adhesions (partition coefficient). (J) In vitro solution turbidity measurements of paxillin variants. 250 nM recombinant FAK was combined with increasing concentrations of recombinant paxillin variants. (K) Spinning disk confocal fluorescence microscopy images of MEF cells transiently expressing GFP-paxillin variants. (L) Quantification of number of adhesions. (M) Quantification of GFP-paxillin partitioning into adhesions (partition coefficient). In (F), (G), (I), (L) and (M) each grey point represents a measurement from one cell, and the mean ± SEM mean is indicated by black lines. Data from at least 35 cells from two or more independent experiments. In (D), (F) and (G) significance tested by one-way ANOVA followed by a Tukey multiple comparison test. In (I), (L) and (M) significance tested by an unpaired t-test. In (C) scalebar = 5 μm. All other scale bars = 10 μm. Note that the pCas protein used in Figure 7 is a different batch than protein used in Figures 15 (See Figure 1—figure supplement 3).

Next, we further characterized the effect of FAK W266A in vitro. Unlike 1 μM WT FAK (Figure 2c), 1 μM FAK W266A (1% Alexa-546) does not form droplets on its own (Figure 7—figure supplement 2). In the pCas+ FAK mix, FAK W266A results in fewer, smaller droplets (Figure 7c) in which paxillin is partitioned ~1.2 more weakly compared with WT FAK (Figure 7d).

We next examined the role of FAK in regulating the number of IACs during cell spreading. We found that FAK KO MEFs (Figure 6—figure supplement 3b) formed significantly fewer IACs and were noticeably smaller than WT MEFs, consistent with the role of FAK in nascent adhesion assembly (Figure 6d–e; Swaminathan et al., 2016). While FAK KO MEFs expressing GFP-WT FAK remained noticeably smaller than WT MEFs, GFP-WT FAK still partially rescued the number of IACs. In contrast, FAK KO MEFS expressing GFP-FAK W266A had no change in the number of IACs compared with FAK KO MEFs. Furthermore, the partitioning of endogenous paxillin into adhesions was decreased in FAK KO MEFs, and partitioning was fully rescued with GFP-WT FAK but not GFP-FAK W266A (Figure 7g). Paxillin partitioning was ~1.4-fold higher in cells expressing GFP-WT FAK compared with GFP-FAK W266A, consistent with our observations in vitro (Figure 7d). Thus, FAK oligomerization is required for FAK-dependent regulation of IACs during cell spreading.

We next measured FAK partitioning into IACs. We found that the W266A mutation causes a 50% reduction in FAK partitioning (Figure 6f–g). While W266A FAK cannot drive phase separation, it may still partition into IACs through additional protein-protein interactions, for example binding of the FAK proline-rich motifs to the Cas SH3 domain (Figure 1—figure supplement 1). In the ‘scaffold/client’ description of condensate composition, the mutation converts FAK from behaving more scaffold-like to more client-like (Banani et al., 2016; Ditlev et al., 2018; Xing et al., 2020). Thus, a decrease in FAK partitioning is consistent with loss of FAK-dependent phase separation. Together, these cellular data demonstrate that impairing the phase separation of either Cas or FAK corelates with a partial reduction in the number of IACs observed after 20 min of cell spreading.

Increasing paxillin valence increases the number of IACs in cells

Next, we sought to identify mutations that might increase phase separation at IACs. Since interactions between paxillin and FAK enhance FAK-dependent phase separation in vitro (Figure 2d–f), we sought to understand the molecular basis of this enhancement. The FAK C-terminal FAT domain contains two distinct binding sites for at least two paxillin LD motifs (LD2 and LD4; Gao et al., 2004; Scheswohl et al., 2008; Thomas et al., 1999). To test if multivalent interaction between paxillin and FAK were necessary to enhance phase separation, we engineered paxillin mutants (Figure 7—figure supplement 3). To weaken paxillin-FAK binding, we mutated a key Asp residue in each of the five LD motifs to reduce the affinity for FAK (‘Paxillin LD mutant’; Thomas et al., 1999). To increase paxillin valence, we duplicated the N-terminus (Residues 1–321) to double the number of LD motifs (‘Double Paxillin’, DP). We added increasing concentrations of each paxillin mutant to 250 nM FAK and measured solution turbidity (Figure 7j). We found that the paxillin LD mutant enhances FAK phase separation less than WT, while the DP mutant acts more strongly than WT. We conclude that multivalent interactions between paxillin LD motifs and the FAK C-term FAT domain enhance FAK phase separation.

Next, we tested whether increasing paxillin valence could alter the number of IACs formed during cell spreading. We expressed GFP-WT paxillin or GFP-DP paxillin in WT MEFs. Cells expressing DP paxillin formed twice as many IACs as cells expressing equal levels of WT paxillin (Figure 7k–l). Thus, doubling the number of paxillin LD motifs is sufficient to increase the number of IACs. Furthermore, we found that DP paxillin was more strongly partitioned into IACs compared with WT paxillin (Figure 7m), consistent with an increase in paxillin-dependent phase separation. Thus, enhancing paxillin-dependent phase separation corelates with an increase in the number of IACs. Together, these experiments demonstrate three distinct genetic perturbations that similarly alter phase separation behavior in vitro and IAC number in cells.

pCas and FAK synergistically promote nascent adhesion assembly

Since our in vitro data demonstrate that pCas and FAK synergistically promote phase separation and kindlin-dependent integrin clustering, we sought to determine if FAK and Cas act synergistically to promote nascent adhesion formation in cells. To assess nascent adhesion assembly more specifically during cell spreading, we fixed cells after only 5 min of spreading on fibronection, immunostained for endogenous paxillin, and imaged cells with TIRF. At this 5-min time point, cells were still predominantly undergoing isotropic cell spreading and IACs remained small and un-elongated. To simultaneously reduce Cas and FAK protein levels in cells, we first used siRNA to knock down Cas in WT or FAK KO MEFs (Figure 8—figure supplement 1a). We found that the number of nascent adhesions formed in 5 min was significantly reduced in cells expressing Cas siRNA, while FAK KO MEFs did not significantly differ from WT MEFs (Figure 8—figure supplement 1b-c). However, western blot analysis showed that Cas protein levels are elevated in FAK KO MEFs (Figure 8—figure supplement 1a) and can be returned to WT levels with the re-expression of WT-FAK-GFP (Figure 6—figure supplement 3b). Additionally, FAK KO MEFs were more resistant to Cas knockdown with siRNA (Figure 8—figure supplement 1a). While this genetic compensation supports our biochemical evidence suggesting that FAK and Cas may be functionally linked, this approach was not sufficient to simultaneously remove FAK and Cas protein from cells.

Next, we tried an alternative approach to remove FAK and Cas protein from cells by using siRNA to knock down FAK in WT or Cas KO MEFs. Western blot analysis showed this strategy was more effective at achieving low protein levels of both Cas and FAK (Figure 8a). Using this approach, we found that loss of either Cas or FAK partially reduces the number of nascent adhesions compared with WT MEFs (Figure 8b–c). Simultaneous loss of both Cas and FAK causes a further reduction in the number of nascent adhesions compared with the loss of either protein alone (Figure 8b–c). When both Cas and FAK are absent, cells form few nascent adhesions and fail to spread in 5 min. Fixation and immunostaining of cells after 20 min of spreading showed similar results (Figure 8—figure supplement 2), suggesting that decreased nascent adhesion assembly in the absence of Cas and FAK leads to significantly fewer IACs and reduced cell spreading even after 20 min.

Figure 8 with 2 supplements see all
Cas and FAK synergistically promote nascent adhesion assembly during cell spreading.

(A) Western blot analysis of FAK knockdown. WT (+) or Cas knockout (-) MEFs were treated with nontargeting (+) or FAK (si) siRNA for 48 hr and lysates were blotted with Cas, FAK or actin antibodies. (B) Total Internal Reflection Fluorescence (TIRF) microscopy images of MEFs fixed after 5 min of spreading with immunostaining for endogenous paxillin. Scalebar = 5 μm. (C) Quantification of number of adhesions. Each grey point represents a measurement from one cell, and the mean ± SEM mean is indicated by black lines. Data from at least 60 cells from two or more independent experiments. Significance tested by one-way ANOVA followed by a Tukey multiple comparison test.

Figure 8—source data 1

Uncropped gel images and unprocessed gel.tif files from Figure 8a.

https://cdn.elifesciences.org/articles/72588/elife-72588-fig8-data1-v2.zip

Together, these data suggest that Cas and FAK have functionally linked roles in promoting nascent adhesion assembly. Consistent with the in vitro reconstitution, maximal nascent adhesion assembly occurs in cells expressing both Cas and FAK. Furthermore, genetic studies have found that Cas and FAK are synthetic lethal in Drosophila (Tikhmyanova et al., 2010). Neither the FAK mutant nor the Cas mutant flies have a reported phenotype but combining the Cas and FAK mutations is 100% embryonic lethal with a phenotype similar to the β integrin mutant. Our data suggest that the synthetic lethality in flies may reflect the cooperativity in phase separation, integrin clustering, and nascent adhesion assembly that we have seen biochemically and in cultured cells.

Discussion

Combining purified recombinant proteins on phospholipid bilayers, we successfully developed an in vitro reconstitution of nascent adhesions. We found that a mixture of seven proteins is sufficient to reconstitute integrin clustering on phospholipid bilayers and recapitulate the integrin enrichment observed within nascent adhesions in cells. These proteins assemble through phase separation at physiologic concentrations and buffer conditions. This phase separation is necessary and sufficient for kindlin-dependent integrin clustering on phospholipid bilayers. Our data suggest that kindlin plays a central role in nascent adhesion assembly by coupling the Cas and FAK pathways to integrins. We find that both environmental perturbations and genetic perturbations can similarly alter phase separation in vitro and the number of IACs in cells. Finally, we find that pCas-dependent and FAK-dependent phase separation synergistically promote integrin clustering on phospholipid bilayers, and that while Cas and FAK can independently promote nascent adhesion assembly, maximum assembly occurs in cells expressing both proteins. Together, these data lead to a model wherein multivalency-driven phase separation of two connected pathways—one based on pCas and the other based on FAK, both coupled to integrins through kindlin—underlies formation of nascent adhesions in cells.

Although kindlin is required for robust IAC formation (Theodosiou et al., 2016), the specific role of kindlin in nascent adhesion formation has remained enigmatic. We find that kindlin is required to couple Cas- and FAK-dependent phase separation to integrins through direct interactions with Cas and paxillin. Observations from our biochemical reconstitution are consistent with previous observations that integrin associates with kindlin and not talin during nascent adhesion assembly (Bachir et al., 2014), that kindlin does not regulate integrin affinity but rather promotes integrin clustering (Ye et al., 2013), and that the kindlin-paxillin interaction is required for efficient nascent adhesion formation in cells (Theodosiou et al., 2016).

Although protein-driven phase separation is consistent with many features of cellular IACs, phase separation likely cooperates with additional factors to regulate nascent adhesion formation in cells. Ligand binding to the extracellular domain of integrin, talin binding to the cytoplasmic domain of integrin, and retrograde actin flow are all required for nascent adhesion formation in cells. Our data suggest that while the ECM-integrin-talin-actin linkage may be critical for integrin conformational activation, it may not be required for subsequent macromolecular assembly of adaptor proteins on the integrin cytoplasmic tail. Furthermore, many IAC-associated proteins can interact with negatively charged phosphoinositides. FAK and N-WASP preferentially bind PI(4,5)P2 via basic amino acids (Goñi et al., 2014; Papayannopoulos et al., 2005), while kindlin preferentially binds PI(3,4,5)P3 via its PH domain (Liu et al., 2011). We find that phosphoinositides are not necessary to reconstitute clusters on phospholipid bilayers that resemble nascent adhesions. Consistent with our observations, reduction of PI(4,5)P2 synthesis at IACs in cells leads to the formation of small IACs that contain paxillin and kindlin, but lack talin and vinculin (Legate et al., 2011). Furthermore, talin has been shown to directly interact with and activate PIP5K1C (Di Paolo et al., 2002), and PIP4K2A was identified in as a constituent of IACs whose local concentration potentially changes during adhesion assembly and growth (Horton et al., 2015). Thus, PI(4,5)P2 may be locally generated during IAC formation or maturation, and additional interactions between phosphoinositides and cytoplasmic adaptor proteins could be important to reduce the threshold concentration of phase separation during nascent adhesion assembly or subsequent adhesion maturation and growth (Mitchison, 2020). Relatedly, coupling between phase separation of lipids and phase separation of membrane-associated proteins was recently described in vitro for reconstituted clusters of T cell receptor signaling proteins (Chung et al., 2021). Extracellular ligand binding (Changede et al., 2015), steric exclusion (Paszek et al., 2009), differential lipid composition (Gaus et al., 2006; Sharma et al., 2004; Son et al., 2017), and the actomyosin cytoskeleton (Changede et al., 2015; Kalappurakkal et al., 2019) likely function together with pCas- and FAK-dependent phase separation to regulate different aspects of integrin activation, nascent adhesion formation and adhesion maturation.

We have focused our investigations here on a minimal set of proteins that are known to assemble at integrins coincident with the initiation of nascent adhesion assembly and which play important functional roles in generating the cellular structures. Yet phase separation at mammalian IACs, particularly during maturation where many more proteins arrive at the structures, likely involves even more proteins. Of the 60 proteins consistently found in proteomic analysis of IACs, 46 have either three or more repeated modular binding domains and/or long intrinsically disordered regions (Supplementary file 5; Horton et al., 2015). The multivalent nature of these proteins suggests that they could contribute to higher order assembly and phase separation within IACs. Indeed, LIMD1, a multivalent protein containing a long intrinsically disordered region (Supplementary file 5), was recently shown to undergo both phase separation and force-dependent localization to adhesions during maturation (Wang et al., 2021). As IAC composition changes during maturation, many different multivalent proteins become enriched (Horton et al., 2015; Kuo et al., 2011; Schiller and Fässler, 2013). Future work will be needed to understand how the changing composition might alter the physical properties of the condensed IAC phase and whether additional multivalent proteins, such as LIMD1, become more important at different stages of IAC formation and maturation.

Mechanical forces play a critical role in regulating IAC composition and function. During nascent adhesion assembly, forces are simply required to promote integrin activation (Oakes et al., 2018). However, subsequent IAC growth and compositional maturation are dependent on forces transmitted across IACs between the extracellular matrix and the actin cytoskeleton (Choi et al., 2008; Kuo et al., 2011; Oakes et al., 2012; Schwarz and Gardel, 2012). Future work will be required to understand how the actin cytoskeleton interacts with and potentially reorganizes the condensed IAC phase. In other systems, actin can regulate condensates by acting as a physical barrier to prevent fusion (Feric et al., 2016), by providing mechanical force to promote movement across the membrane surface (Ditlev et al., 2019; Kim et al., 2019), or by acting as a scaffold along which condensates can wet or assemble (Case et al., 2019b; Ditlev et al., 2019; Su et al., 2016). Understanding how forces are transmitted across the condensed IAC phase will require an understanding of the emergent physical properties of the phase. In vitro reconstitution will be a useful tool to directly address these questions.

Mature IACs (called focal adhesions) are organized into vertical layers, with paxillin and FAK localized near the membrane and actin and actin binding proteins localized >50 nm above the membrane (Kanchanawong et al., 2010). At mature focal adhesions, the plasma membrane and actin cytoskeleton provide an inherent polarity; the membrane is physically connected to the ECM through integrins on one side and actin generates forces on the other. These forces are transmitted across the components of the mature focal adhesion. While nascent adhesions may behave like an isotropic liquid condensate, the actin cytoskeleton could induce molecular reorganizations to produce order in the direction perpendicular to the membrane and generate a state more akin to a liquid crystal. While it is not known when this layered organization emerges during IAC assembly and maturation, talin plays a critical role in controlling the distance between actin and the plasma membrane (Liu et al., 2015). Additionally, talin undergoes force dependent conformational changes that extend the protein and expose additional vinculin binding sites (del Rio et al., 2009). Thus, force-dependent talin extension likely plays an important role in patterning the mature focal adhesion. Many other condensates interact with actin, and the layered organization of mature focal adhesions may be more common than previously appreciated (Beutel et al., 2019; Bresler et al., 2004; Zeng et al., 2016).

In conclusion, we have reconstituted minimal macromolecular integrin clusters that have similar composition and features to cellular nascent adhesions. Our biochemical and cellular data provide evidence that Cas- and FAK-dependent phase separation promotes integrin clustering and nascent adhesion assembly. A phase separation model is consistent with many well-documented characteristics of IACs. In other systems, phase separation of receptors enhances downstream signaling (Case et al., 2019b; Huang et al., 2019; Su et al., 2016). Thus, phase separation may provide an important framework for understanding the regulation of signaling and force transmission at IACs.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Escherichia coli)BL21(DE3) Competent E. coliNew England BioLabsCat#C2527IChemically competent E. coli.
Strain, strain background (Escherichia coli)MAX Efficiency DH10Bac Competent CellsThermo FisherCat#10361012Chemically competent E. coli.
Cell line (Mus musculus)Ptk2+/+, Trp53-/- MEFs; WT MEFsATCCCRL-2645; RRID: CVCL_8955
Cell line (Mus musculus)Ptk2-/-, Trp53-/- MEFs; FAK KO MEFsATCCCRL-2644; RRID:CVCL_8954
Cell line (Mus musculus)Bcar1-/- MEFs + Control Vector; Cas KO control MEFsMeenderink et al., 2010N/A
Cell line (Mus musculus)Bcar1-/- MEFs + CasWT Vector; Cas KO CasWT MEFsMeenderink et al., 2010N/A
cell line (Mus musculus)Bcar1-/- MEFs + Cas15 F Vector; Cas KO CasY15F MEFsMeenderink et al., 2010N/A
Cell line (Spodoptera frugiperda)Sf9 CellsGibcoCat#11496015
AntibodyMouse monoclonal anti-Paxillin (Clone 349/Paxillin)BD BiosciencesCat#610051; RRID: AB_397463(1:100)
AntibodyMouse monoclonal anti-FAK (Clone 4.47)MilliporeCat#05–537; RRID: AB_2173817(1:1000)
antibodyRabbit monoclonal anti-p130 Cas (E1L9H)Cell Signaling TechnologyCat#13846; RRID: AB_2798328(1:1,000)
AntibodyRabbit polyclonal anti-beta ActinAbcamCat#ab8227; RRID: AB_2305186(1:5000)
AntibodyGoat polyclonal anti-Mouse IgG, Alexa Fluor 568Thermo FisherCat#A-11004; RRID: AB_2534072(1:250)
AntibodyGoat polyclonal anti-Mouse IgG, HRPSanta Cruz BiotechnologyCat#sc-2005; RRID: AB_631736(1:10,000)
AntibodyGoat polyclonal anti-Rabbit IgG, HRPSanta Cruz BiotechnologyCat# sc-2030; RRID: AB_631747(1:10,000)
Sequence-based reagentON-TARGETplus Mouse Bcar1 (12927) siRNA - SMARTpoolHorizon Discovery (Dharmacon)Cat#L-041961-00-0005Commercially available
Sequence-based reagentON-TARGETplus Non-targeting Control PoolHorizon Discovery (Dharmacon)Cat#D-001810-10-05Commercially available
Sequence-based reagentON-TARGETplus Mouse Ptk2 siRNA - SMARTpoolHorizon Discovery (Dharmacon)Cat#L-041099-00-0005Commercially available
Chemical compound, drugAlexa Fluor 647 C2 MaleimideInvitrogenCat#A20347
Chemical compound, drugAlexa Fluor 488 C5 MaleimideInvitrogenCat#A10254
Chemical compound, drugAlexa Fluor 568 C5 MaleimideInvitrogenCat#A20341
Chemical compound, drugSNAP-Surface Alexa Fluor 546New England BioLabsCat#S9132S
Chemical compound, drugSNAP-Surface Alexa Fluor 488New England BioLabsCat#S9129S
Chemical compound, drugIntracellular pH Calibration Buffer Kit (Nigericin/Valinomycin)InvitrogenCat#P35379
Software, algorithmMATLABMathworks
Software, algorithmFIJIhttps://fiji.scRRID:SCR_002285
OtherPOPCAvanti Polar LipidsCat# 850457C
OtherPEG5000-PEAvanti Polar LipidsCat# 880230C
OtherDGS-NTA-NiAvanti Polar LipidsCat# 790404C
OtherNBD PCAvanti Polar LipidsCat# 810130C

Protein expression, purification, and modification

Information on different recombinant protein constructs is provided in Supplementary file 6.

Lck purification:

Request a detailed protocol

His6-Lck was expressed from baculovirus in Spodoptera frugiperta (Sf9) cells. Cells were collected by centrifugation and lysed by douncing on ice in 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 5 mM βME, 0.01% NP-40, 1 mM PMSF, 1 μg/ml antipain, 1 μg/ml benzamidine, and 1 μg/ml leupeptin. Centrifugation-cleared lysate was applied to Ni-NTA agarose beads (Qiagen), washed with 20 mM Tris-HCl (pH 7.5), 1 M NaCl, 20 mM imidazole (pH 7.5), 5 mM βME, and 10% glycerol, and then eluted with 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, 200 mM imidazole (pH 7.5), 5 mM βME, and 10% glycerol. The elute was applied to a Source 15 Q anion exchange column and eluted with a gradient of 100 → 300 mM NaCl in 25 mM HEPES (pH 7.5) and 2 mM βME. Collected fractions were concentrated (Amicon 10 K, Millipore) and applied to an SD75 column in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 1 mM βME.

p130Cas purification

Request a detailed protocol

His6-p130Cas was expressed from baculovirus in Spodoptera frugiperta (Sf9) cells. Cells were collected by centrifugation and lysed by douncing on ice in 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 500 mM NaCl, 10% glycerol and 5 mM βME + cOmplete, EDTA-free Protease Inhibitor tablet (Roche). Centrifugation-cleared lysate was applied to Ni-NTA agarose beads (Qiagen), washed with 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 500 mM NaCl, 10% glycerol and 5 mM βME, and then eluted with 20 mM Tris-HCl (pH 8.0), 400 mM Imidazole (pH 8.0), 500 mM NaCl, 10% glycerol and 5 mM βME. The his tag was removed using TEV protease treatment for 16 hr at 4 °C. Cleaved protein was applied to a Source 15 Q anion exchange column and eluted with a gradient of 100 → 300 mM NaCl in 20 mM Immidazole (pH 7.0), 1 mM DTT and 10% glycerol. Collected fractions were concentrated (Amicon 50 K, Millipore) and applied to an SD200 column in 25 mM Hepes pH7.5, 150 mM NaCl and 10% glycerol. Cas was concentrated using Amicon Ultra Centrifugal Filter units (Millipore) to >400 μ M, mixed with 100 mM HEPES (pH 7.5), 100 mM NaCl, 15 mM ATP, 20 mM MgCl2, 2 mM DTT, and 150 nM active Lck, and incubated for 16 hrs at 30°C. Phosphorylated Cas was resolved on a Mono Q anion exchange column using a shallow 100 mM → 350 mM NaCl gradient in 40 mM Imidazole (pH 7.0), 1 mM DTT and 10% glycerol. Fully phosphorylated Cas was additionally purified by size exclusion chromatography to remove any protein aggregates using a Superdex 200 prepgrade column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM βME, and 10% glycerol. Cas phosphorylation was confirmed by size shift in SDS-PAGE gel and quantified with mass spectrometry.

Nephrin purification

Request a detailed protocol

BL21(DE3) cells expressing MBP-His8-Neprhin were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM imidazole (pH 8.0), 150 mM NaCl, 5 mM βME, 0.1% NP-40, 10% glycerol, 1 mM PMSF, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Ni-NTA agarose (Qiagen), washed with 50 mM imidazole (pH 8.0), 150 mM NaCl, 5 mM βME, 0.01% NP-40, and 10% glycerol, and eluted with 500 mM imidazole (pH 8.0), 150 mM NaCl, 5 mM βME, 0.01% NP-40, and 10% glycerol. The MBP tag was removed using TEV protease treatment for 16 hr at 4 °C (for non his-tagged nephrin, the his6-tag was also removed with precision protease treatment). Cleaved protein was applied to a Source 15 Q anion exchange column and eluted with a gradient of 150 mM→350 mM NaCl in 20 mM Immidazole (pH 8.0) and 2 mM DTT followed by size exclusion chromatography using a Superdex 200 prepgrade column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 2 mM DTT. Nephrin was concentrated using Amicon Ultra Centrifugal Filter units (Millipore) to >400 μ M, mixed with 100 mM HEPES (pH 7.5), 100 mM NaCl, 15 mM ATP, 20 mM MgCl2, 2 mM DTT, and 150 nM active Lck, and incubated for 16 hr at 30 °C. Phosphorylated Nephrin (pNephrin) was resolved on a Mono Q anion exchange column using a shallow 100 mM → 350 mM NaCl gradient in 20 mM imidazole (pH 8.0) and 2 mM DTT to separate differentially phosphorylated species of Neprhin. Fully phosphorylated pNephrin was additionally purified by size exclusion chromatography to remove any protein aggregates using a Superdex 200 prepgrade column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM βME, and 10% glycerol. Complete Nephrin phosphorylation was confirmed by mass spectrometry.

Nck purification

Request a detailed protocol

BL21(DE3) cells expressing GST-Nck were collected by centrifugation and lysed by sonication in 25 mM Tris-HCl (pH 8.0), 200 mM NaCl, 2 mM EDTA (pH 8.0), 1 mM DTT, 1 mM PMSF, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Glutathione Sepharose 4B (GE Healthcare) and washed with 20 mM Tris-HCl (pH 8.0), 200 mM NaCl, and 1 mM DTT. GST was cleaved from protein by TEV protease treatment for 16 hr at 4 °C. Cleaved protein was applied to a Source 15 Q anion exchange column and eluted with a gradient of 5 → 250 mM NaCl in 20 mM imidazole (pH 7.0) and 1 mM DTT. Eluted protein was pooled and applied to a Source 15 S cation exchange column and eluted with a gradient of 0 → 500 mM NaCl in 20 mM imidazole (pH 7.0) and 1 mM DTT. Eluted protein was concentrated using Amicon Ultra 10 k concentrators and further purified by size exclusion chromatography to remove any protein aggregates using a Superdex 75 prepgrade column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 1 mM βME.

N-WASP purification

Request a detailed protocol

BL21(DE3) cells expressing His6-N-WASP were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM imidazole (pH 7.0), 300 mM KCl, 5 mM βME, 0.01% NP-40, 1 mM PMSF, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. The cleared lysate was applied to Ni-NTA agarose (Qiagen), washed with 50 mM imidazole (pH 7.0), 300 mM KCl, 5 mM βME, and eluted with 300 mM imidazole (pH 7.0), 100 mM KCl, and 5 mM βME. The elute was further purified over a Source 15 Q column using a gradient of 250 → 450 mM NaCl in 20 mM imidazole (pH 7.0), and 1 mM DTT. The His6-tag was removed by TEV protease at 4 °C for 16 hr (for His-N-WASP, no TEV treatment occurred). Cleaved N-WASP (or uncleaved for His-N-WASP) was then applied to a Source 15 S column using a gradient of 110 → 410 mM NaCl in 20 mM imidazole (pH 7.0), 1 mM DTT. Fractions containing N-WASP were concentrated using an Amicon Ultra 10 k concentrator (Millipore) and further purified by size exclusion chromatography to remove any protein aggregates using a Superdex 200 prepgrade column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM KCl, 1 mM βME, and 10% glycerol.

Kindlin purification:

Request a detailed protocol

BL21(DE3) cells expressing His6-sumo-kindlin were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 150 mM NaCl, 0.01% NP-40, 5 mM βME, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. The cleared lysate was applied to Ni-NTA agarose (Qiagen) and washed with 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 150 mM NaCl, 0.01% NP-40, 5 mM βME. Kindlin was eluted with 20 mM Tris-HCl (pH 8.0), 300 mM Imidazole (pH 8.0), 150 mM NaCl, 0.01% NP-40, 5 mM βME. The elute was further purified over a Source 15 Q column using a gradient of 0 → 300 mM NaCl in 20 mM Tris-HCl (pH 8.0), 1 mM DTT. The Collected fractions were pooled and the His6-sumo tag was removed by Ulp1 sumo protease for 2 hr at room temperature. Protein was concentrated using an Amicon Ultra 10 k concentrators and further purified by size exclusion chromatography to remove any protein aggregates using a Superdex 200 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 300 mM NaCl, 10% glycerol, and 1 mM βME.

Talin head purification:

Request a detailed protocol

BL21(DE3) cells expressing talin head-His6 were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM Tris-HCl (pH 8.0), 5 mM Imidazole (pH 8.0), 500 mM NaCl, 1% TritonX, 5 mM βME, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Ni-NTA agarose (Qiagen), washed with 20 mM Tris-HCl (pH 8.0), 30 mM Imidazole (pH 8.0), 500 mM NaCl, 5 mM βME, 1 μg/ml benzamidine, and eluted with 20 mM Tris-HCl (pH 8.0), 100 mM Imidazole (pH 8.0), 500 mM NaCl, 5 mM βME, 1 μg/ml benzamidine. The his6-tag was removed using TEV protease treatment for 16 hr at 4 °C. Cleaved protein was applied to a Source 15 S cation exchange column and eluted with a gradient of 150 mM→500 mM NaCl in 20 mM Immidazole (pH 7.0), 10% glycerol, and 2 mM DTT followed by size exclusion chromatography to remove any protein aggregates using a Superdex 75 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 2 mM βME.

Integrin purification

Request a detailed protocol

BL21(DE3) cells expressing integrin (MBP-his10-Integrin, MBP-his6-Integrin, or MBP-his6-Integrin-GFP) were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 150 mM NaCl, 0.01% NP-40, 10% glycerol, 5 mM βME, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Ni-NTA agarose (Qiagen), washed with 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 150 mM NaCl, 10% glycerol, 5 mM βME, and eluted with 20 mM Tris-HCl (pH 8.0), 300 mM Imidazole (pH 8.0), 150 mM NaCl, 10% glycerol, 5 mM βME. Eluate was applied to a Source 15Q anion exchange column and eluted with a gradient of 0–300 mM NaCl in 20 mM Tris-HCl (pH 8.0), 10% glycerol and 1 mM DTT. The MBP-tag or MBP-his6-tag was removed using TEV protease treatment for 16 hrs at 4 °C. Cleaved protein was applied to a Source 15 S cation exchange column and eluted with a gradient of 150 mM→500 mM NaCl in 20 mM Bis-Tris (pH 6.0), 10% glycerol, and 1 mM DTT followed by size exclusion chromatography to remove any protein aggregates using a Superdex 75 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 1 M NaCl, 10% glycerol, 1 mM βME.

FAK purification

View detailed protocol

His6-FAK was expressed from baculovirus in Spodoptera frugiperta (Sf9) cells. Cells were collected by centrifugation and lysed by douncing on ice in 25 mM HEPES (pH 7.5), 20 mM Imidazole (pH 7.5), 500 mM NaCl, 10% glycerol and 5 mM βME + cOmplete(TM), EDTA-free Protease Inhibitor tablet. Centrifugation-cleared lysate was applied to Ni-NTA agarose beads (Qiagen), washed with 25 mM HEPES (pH 7.5), 20 mM Imidazole (pH 7.5), 500 mM NaCl, 10% glycerol, 5 mM βME, 1 μg/ml benzamidine, and eluted with 25 mM HEPES (pH 7.5), 400 mM Imidazole (pH 7.5), 1 M NaCl, 10% glycerol, 5 mM βME, 1 μg/ml benzamidine. The his6-tag was removed using TEV protease treatment for 16 hr at 4 °C. Cleaved protein was further purified with size exclusion chromatography to remove any protein aggregates applied to a Superdex 200 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 300 mM NaCl, and 1 mM βME.

Paxillin purification

Request a detailed protocol

BL21(DE3) cells expressing GST-Paxillin were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM Tris-HCl (pH 8.0), 300 mM NaCl, 0.01% NP-40, 10% glycerol, 1 mM DTT, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Glutathione Sepharose 4B (GE Healthcare) and washed with 20 mM Tris-HCl (pH 8.0), 300 mM NaCl, 0.01% NP-40, 10% glycerol, 1 mM DTT, 1 μg/ml benzamidine. GST was cleaved from protein by TEV protease treatment for 16 hr at 4 °C. Cleaved protein was applied to a Source 15 Q anion exchange column and eluted with a gradient of 5 → 500 mM NaCl in 20 mM imidazole (pH 8.0), 10% glycerol and 1 mM DTT. Eluted protein was concentrated using Amicon Ultra 10 k concentrators and further purified by size exclusion chromatography to remove any protein aggregates using a Superdex 200 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 1 mM βME.

FAK c-term purification

Request a detailed protocol

BL21(DE3) cells expressing His6-sumo-FAK c-term were collected by centrifugation and lysed by cell disruption (Emulsiflex-C5, Avestin) in 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 300 mM NaCl, 0.01% NP-40, 5 mM βME, 1 μg/ml antipain, 1 μg/ml benzamidine, 1 μg/ml leupeptin, and 1 μg/ml pepstatin. Centrifugation-cleared lysate was applied to Ni-NTA agarose (Qiagen), washed 20 mM Tris-HCl (pH 8.0), 300 mM Imidazole (pH 8.0), 300 mM NaCl, 0.01% NP-40, 5 mM βME, and eluted with 20 mM Tris-HCl (pH 8.0), 20 mM Imidazole (pH 8.0), 300 mM NaCl, 5 mM βME,. Eluate was applied to a Source 15 Q anion exchange column and eluted with a gradient of 150 mM→300 mM NaCl in 20 mM Tris-HCl (pH 8.0), 10% glycerol, and 1 mM DTT. Collected fractions were pooled and the His6-sumo tag was removed by Ulp1 sumo protease for 2 hr at room temperature. Protein was concentrated using an Amicon Ultra 10 k concentrators and further purified by size exclusion chromatography to remove any protein aggregates using a Superdex 75 column (GE Healthcare) in 25 mM HEPES (pH 7.5), 300 mM NaCl, and 1 mM DTT.

Fluorophore conjugation

Request a detailed protocol

For conjugation with Maleimide chemistry (Nck, N-WASP, Integrin, FAK) recombinant proteins to be labeled with Alexa fluorophores were concentrated using Amicon Ultra Centrifugal Filter units (Millipore) to ~100 μM. 5 mM βME was added to reduce cysteine residues followed by buffer exchange using a HiTrap 26/10 Desalting column (GE Healthcare) in 25 mM HEPES (pH 7.5) and 150 mM NaCl. Fractions containing protein were collected and concentrated to 100 μM. 500 μM Alexa Fluor 647 C2 Maleimide (ThermoFisher, for N-WASP), Alexa Fluor 488 C5 Maleimide (ThermoFisher, for his8-pNephrin, N-WASP, and his8-N-WASP) or Alexa Fluor 568 C5 Maleimide (ThermoFisher, for Nck) was added, and the reaction was incubated with gentle mixing at 4 °C for 16 hr. The reaction was quenched with 1 μl 14.3 M βME followed by final buffer exchange using size exclusion chromatography (GE Healthcare) in 25 mM HEPES (pH 7.5), 150 mM NaCl, and 1 mM βME. We consistently achieve >98% labeling efficiency.

For conjugation of SNAP-tagged proteins (paxillin, kindlin) recombinant proteins to be labeled with fluorophores were concentrated using Amicon Ultra Centrifugal Filter units (Millipore) to ~50 μM. 1 mM DTT and twofold molar excess SNAP-tag Substrate (SNAP-Surface Alexa488 or SNAP-Surface Alexa546) were added. The mixture was incubated for 30 min at 37 °C. Labeled proteins were purified by final buffer exchange using size exclusion chromatography (Superdex 200 column, GE Healthcare) in the appropriate buffer (i.e. final buffer in the above purification protocol). We consistently achieve >98% labeling efficiency.

Final protein concentration and degree of labeling were calculated from the protein absorbance using the following formulas:

  • Concentration (M) = A280 – (A494 × 0.11)/ ExtCoef

  • Alexa 488

  • Degree of labeling = A494/(71,000 x [protein conc.])

  • Alexa 568

  • Concentration (M) = A280 – (A577 × 0.46)/ExtCoef

  • Degree of labeling = A577/(91,300 x [protein conc.])

  • Alexa 647

  • Concentration (M) = A280 – (A650 × 0.03)/ExtCoef

  • Degree of labeling = A650/(239,000 x [protein conc.])

Mass spectrometry analysis of p130Cas phosphorylation

Request a detailed protocol

Protein samples were analyzed by LC/MS using a Sciex X500B Q-ToF mass spectrometer running Sciex OS v.1.6.1, coupled to an Agilent 1,290 Infinity II HPLC. Samples were injected onto a POROS R1 reverse-phase column (2.1 × 30 mm, 20 µm particle size, 4000 Å pore size), desalted, and the amount of buffer B was manually increased stepwise until the protein eluted off the column. Buffer A contained 0.1% formic acid in water and buffer B contained 0.1% formic acid in acetonitrile. The mobile phase flow rate was 300 µL/min. The acquired mass spectra for the protein of interest were deconvoluted using BioPharmaView v. 3.0.1 software (Sciex) in order to obtain the molecular weight.

Turbidity measurements

Request a detailed protocol

Unlabeled proteins were diluted in 50 mM HEPES (pH 7.3), 50 mM KCl, 1 mM TCEP (‘Buffer A’), and 0.1% (1 mg/mL BSA), unless a different buffer is indicated in figure legends. After a 30–60 min incubation, solution was transferred to Quartz cuvette and Absorbance at 350 nm was measured in a Spectrophotometer (Agilent or Thermo Fisher). While we use 1 mg/mL BSA (0.1%) to prevent nonspecific interactions with surfaces, we note that this is well below the BSA concentration necessary to induce crowding (typically 100 mg/mL or greater). For temperature-controlled measurements, Absorbance at 350 nm was measured using a Cary 100 UV-Visible spectrophotometer equipped with a Peltier thermal controller (Agilent Technologies, Australia). The reaction mixture was incubated in a microcentrifuge tube at the desired temperature for 30 min, and then placed into a pre-equilibrated cuvette and spectrophotometer for measurement.

3D droplet assays

Request a detailed protocol

384-well glass-bottomed plates (Brooks) were washed with 5% Hellmanex III (Hëlma Analytics) for 3.5 hrs at 55 °C and thoroughly rinsed with MilliQ H2O. Plates were then washed with 1 M NaOH for 1 hr at 55 °C and thoroughly rinsed with MilliQ H2O. 50 μ L of 20 mg/mL mPEG silane MW 5 k (Creative PEGworks) in 95% EtOH was added to each well. The plate was covered in parafilm and incubated overnight at room temperature. The plate was thoroughly rinsed with MilliQ H2O, dried, and sealed with foil. Prior to experiment, individual wells were washed 3 X with MilliQ H2O, blocked with Buffer A containing 0.1% BSA for 30 min at room temperature. For droplet experiments, proteins were combined at 1 μM concentration in Buffer A containing 0.1% BSA and a glucose/glucose oxidase/catalase O2-scavenging system. 80 μL of the mixture was added to the well and incubated for 30 min prior to imaging.

Small unilamellar vesicle preparation

Request a detailed protocol

The general protocol we follow to make small unilamellar vesicles (SUVs) and supported phospholipid bilayers is described in Su et al., 2017. Synthetic 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC), 1,2-dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl] (nickel salt, DGS-NTA-Ni), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)–5000] (ammonium salt) (PEG5000 PE) were purchased from Avanti Polar Lipids as chloroform suspension. Using glass Hamilton syringes, lipids were mixed to make a chloroform suspension containing 98% POPC, 2% DGS-NTA-Ni and 0.1% PEG5000 PE. Chloroform was evaporated with gentle stream of Argon Gas, desiccated in a vacuum overnight, and resuspended in PBS (pH 7.3) with vortexing. To promote the formation of small unilamellar vesicles (SUVs), the lipid solution was repeatedly frozen in liquid N2 and thawed using a 37°C water bath until the solution cleared (~35 freeze-thaw cycles). SUV-containing solution was centrifuged at 33,500 g for 45 min at 4°C to remove large vesicles. Cleared supernatant containing SUVs was collected and stored at 4°C covered with Argon for up to two weeks.

Reconstitution on supported phospholipid bilayers

Request a detailed protocol

Briefly, 96-well glass-bottomed plates (Brooks) were washed with 5% Hellmanex III (Hëlma Analytics) for 3.5 hr at 55°C. The plate was thoroughly rinsed with MilliQ H2O, dried, and sealed with foil. Prior to experiment, individual wells were washed with 6 M NaOH for 30 min at 50°C two times, and thoroughly rinsed with MilliQ H2O followed by equilibration with 50 mM HEPES (pH 7.3), 50 mM KCl, and 1 mM TCEP (‘Buffer A’). Twelve μL SUVs were added to cleaned wells covered by Buffer A and incubated for 40 min hr at 40°C to allow SUVs to collapse on glass and fuse to form the bilayer. Bilayers were washed three times with Buffer A to remove excess SUVs, and then blocked with Buffer A containing 0.1% BSA for 30 min at room temperature. His-tagged proteins (10 nM his10-Integrin) were mixed in Buffer A containing 0.1% BSA, added to phospholipid bilayers, and incubated for 2 hr. Bilayers were then washed with Buffer A containing 0.1% BSA to remove unbound His-tagged proteins. Additional proteins were added to the well if required for the experiment. Microscopy experiments were performed in the presence of a glucose/glucose oxidase/catalase O2-scavenging system to reduce photodamage and photobleaching. Unexpectedly, we found that FAK interacts with Nickel and competes with his-tagged proteins for Ni-lipid binding (data not shown). To reduce this effect, we lowered FAK concentration to 200 nM for experiments on supported phospholipid bilayers. Prior to any experiment, the fluidity of bilayers was indirectly assessed by imaging integrin Alexa-488. We photobleached a 5 micron region, and experiments were only performed if FRAP t1/2 < 10 s. To confirm that our methods consistently give fluid, uniform bilayers, we also assessed bilayer fluidity directly in bilayers containing 1% PC-NBD, a fluorescent lipid (Figure 4—figure supplement 1).

Microscopy

Request a detailed protocol

TIRF images were captured using a TIRF/iLAS2 TIRF/FRAP module (Biovision) mounted on a Leica DMI6000 microscope base equipped with a plan apo 100 × 1.49 NA TIRF objective and a 405/488/561/647 nm Laser Quad Band Set filter cube for TIRF applications (Chroma). Illumination was provided by an integrated laser engine equipped with multiple laser lines (405 nm-100mw/445 nm-75mw/488 nm-150mw/514 nm-40mw/561 nm-150mw/637 nm-140mw/730 nm-40mw, Spectral). Confocal images were captured using a Yokogawa spinning disk and a 405/488/561/647 nm Laser Quad Band Set filter cube (Chroma) with a plan apo 63 × 1.40 NA objective. Images were acquired using a Hamamatsu ImagEMX2 EM-CCD camera.

In Figure 7, TIRF images were captured using a Leica TIRF-module mounted on a Leica DMi8 equipped with a plan apo 100 × 1.47 NA TIRF objective and a Quad Band set filter cube for TIRF applications. Illumination was provided by an integrated laser system equipped with multiple laser lines (405 nm-50mw/488 nm-150mw/561 nm-120mw/638 nm-150mW). Images were acquired using LASX software a Hamamatsu Flash 4.0 V3 CMOS camera.

Quantification of microscope PSF

Request a detailed protocol

Measurements were performed in 384-well glass-bottomed plates (Brooks) prepared identically to the 3D droplet assays. Five μL of 0.1 μm diameter TetraSpeck beads (Fisher) were diluted in 120 μL ethanol (final density of ~ 7.5 x 109 particles/mL). A total of 100 μL was added to a well and incubated for 10 min. Ethanol was removed and the well was rinsed 3X with Buffer A containing 0.1% BSA. Of Buffer A containing 0.1% BSA, 200 μL was added and images were acquired in all channels with identical acquisition settings to 3D droplet assays. Using spinning disk fluorescence microscopy, Z-stacks of beads were acquired. The PSF was measured for each channel in Image J. For each bead, the Z-plane with the highest intensity was identified (Figure 1—figure supplement 5a). A linescan through the bead was plotted and the full width half max (FWHM) was determined from a Gaussian fit (Figure 1—figure supplement 5). For each channel, 20 beads were measured and the mean FWHM was determined. A single-plane PSF with the calculated FWHM was generated using the Gaussian PSF 3D ImageJ plugin. Images with circles of known diameter (1 – 50 pixels) were generated. The images were convolved with the PSF. The convolved images were analyzes using MATLAB (Mathworks). A mask was generated by a global image threshold using Otsu’s method. The diameter and mean intensity inside each masked region were measured. The original circle intensity (255) was divided by the measured intensity of the convolved image and plotted against the measured diameter of the convolved image (Figure 1—figure supplement 5d). The data were fit with a single exponential association. This equation was used to correct the intensity of small droplets.

Measuring droplet partition coefficient

Request a detailed protocol

Quantitative image analysis was performed using Matlab (Mathworks). Background images were collected and subtracted from all images before processing. To correct for uneven illumination and detector sensitivity, pixel intensities across a solution containing dye were normalized to the maximum intensity of the image to obtain pixel-by-pixel correction factors (in a 0–1 range). Experimental images were then corrected by dividing by these factors. A mask of droplets was generated from the 647-channel (Either Nck-Alexa647 or FAK-Alexa647) by a global image threshold using Otsu’s method. The diameter of each masked droplet was measured. For each individual masked droplet, the mean fluorescence intensity inside the masked region was calculated for each channel (Alexa488, Alexa546, Alexa647, although the number of channels in each experiment varied). The mask was dilated and inverted to calculate the mean intensity in the bulk solution (outside of droplets). Droplet intensity was plotted against droplet diameter. If intensity increased with increasing diameter, the intensity was corrected using the PSF correction as in Figure 1—figure supplement 5. Droplets with a diameter less than 12 pixels were discarded, since the intensity cannot be accurately measured or corrected. Partition coefficient was calculated for each droplet by dividing droplet mean intensity by the bulk mean intensity.

Droplet FRAP measurements

Request a detailed protocol

FRAP analysis of droplets was performed using spinning disk confocal microscopy as described above. Droplets with a diameter of ~4-5 microns were used for FRAP measurements. A 19-pixel (4.8 μm) diameter circular region of interest was drawn around the droplet, and the entire droplet was bleached with 405 laser illumination. Images were acquired at 1s interval for 90 sec. Quantification of droplet intensity was performed using ImageJ. The mean intensity inside the ROI was measured, and an identical sized region in the bulk solution was used to correct for photobleaching during imaging. Background and photobleaching corrected intensities were used for normalization.

  • Normalized Intensity = (Intensity – Intensitypostbleach)/(Intensityprebleach – Intensitypostbleach).

  • Intensityprebleach = average of three measurements in the ROI before 405 laser.

  • Intensitypostbleach = intensity within the ROI immediately after 405 laser (at t=0).

Measurements were repeated at least 8 times, and the curves were averaged and fit using Prism software. To determine the best fit, a single exponential and double exponential fit were statistically compared with an extra sum-of-squares F Test. The values of the best fit are displayed in Supplementary file 4. For several molecules, the percent recovery could not be accurately fit from these data, as the recovery did not sufficiently plateau within the 90 second experiment. These are listed as nd.

His-kindlin pull-down

Request a detailed protocol

His6-sumo-kindlin was purified as described above, skipping the Ulp1 cleavage step to keep the his6 tag attached. For the pull-down assay, we made a wash Buffer (20 mM Tris pH 7.5, 150 mM NaCl, 10 mM Immidazole) and elution buffer (20 mM Tris pH 7.5, 150 mM NaCl, 300 mM Immidazole). We added 50 μ L of resuspended Ni-NTA Resin to Pierce spin columns (Thermo Scientific). Resin was washed 5 x with 400 μ L of wash buffer, removing buffer with centrifugation at 10,000 x g after each wash. A 10 μ M solution of his-kindlin was prepared in wash buffer. 100 μ L of 10 μ M Kin or 100 μ L of wash buffer control was added to resin and incubated with gentle shaking for 1 hr at 4°C. After 1 hr, flowthrough was collected by centrifugation at 10,000x g. Resin was washed 5 x with 400 uL of wash buffer, removing buffer with centrifugation at 10,000 x g after each wash. A 10 μ M solution of Cas or pCas was prepared in wash buffer. 100 μ L of 10 μ M Cas, 10 μ M pCas or wash buffer control was added to resin and incubated with gentle shaking for 2 hr at 4 °C. The flowthrough was collected by centrifugation at 10,000 x g. Resin was washed 5 x with 400 μ L of wash buffer, removing buffer with centrifugation at 10,000 x g after each wash. Finally, 250 μ L elution buffer was added to resin and incubated for 5 min at room temperature. Eluted proteins were collected by centrifugation at 10,000 x g. The final elution was mixed with equal volume 2 X SDS-PAGE loading buffer and boiled for 10 min to denature proteins. Samples were loaded onto a 10% bis-Acrylamide gel for SDS-PAGE (200 V for 40 min). Gels were stained with Coomassie Blue and imaged. The following conditions were tested: (1) 10 μ Μ his-kindlin+ buffer control; (2) Buffer control +10 μ M Cas; (3) 10 μ Μ his-kindlin +10 μ M Cas; (4) Buffer control +10 μ M pCas; (5) his-kindlin +10 μ M pCas.

Cell culture and transfection

Request a detailed protocol

Mouse Embryonic Fibroblasts (MEFs) were grown in DMEM supplemented with 10% FBS, 2 mM GlutaMAX (Gibco), 100 U/mL penicillin, and 100 µg/mL Streptomycin. Ptk2 +/+ and Ptk2 -/- MEFs were obtained from ATCC. Bcar1-/- MEFs with Control, CasWT, or CasY15F vectors stably expressed were obtained from Steve Hanks (Vanderbilt) and Larisa Ryzhova (Maine Medical Center). MEF phenotypes were confirmed with westernblot analysis of Cas or FAK protein (Figure 6—figure supplement 3) and cells were routinely tested for mycoplasm with the Lonza mycoplasma detection kit. For expression of GFP-tagged Paxillin or FAK variants, cells were transiently transfected with 0.5 μg DNA of EGFP-tagged proteins using lipofectamine 2000 and incubated for 24 hr prior to experiment. For siRNA experiments, cells were transfected using lipofectamine RNAiMAX with pooled oligos (Nontargeting, BCAR1 or PTK2, Dharmacon) and incubated for 48 hr prior to experiment. Details of plasmids used for expression in cells found in Supplementary file 6.

Quantification of total adhesion area

Request a detailed protocol

To determine the transient effect of solvent perturbations on adhesion size and/or number, we quantified total adhesion area after a 10-min incubation in specified buffers or media. Cells were plated on glass coverslips coated with 10 μg/mL fibronectin and incubated overnight. The experimental buffer or media was added, and cells were incubated for 10 mins. For temperature experiments, the media was pre-incubated at 4 °C, 22 °C, or 37 °C prior to adding to cells. After a 10-min incubation, cells were fixed with 3% paraformaldehyde in Cytoskeleton Buffer (CB: 10 mM MES pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl2, 5 mM glucose) for 20 min at room temperature. Cells were then permeabilized for 8 min with CB +0.5% triton-X, quenched for 10 min with CB +0.1 M glycine, and rinsed 2 × 5 min in TBS-T. Coverslips were blocked with 2% BSA in TBS-T for 1 hour. Coverslips were incubated with primary antibody (Ms anti Paxillin, 1:100) in TBS-T with 2% BSA overnight at 4 °C. Coverslips were rinsed 3 × 5 min in TBS-T and incubated with secondary antibody (anti Ms Alexa568, 1:250) in in TBS-T with 2% BSA for 45 min at room temperature. Coverslips were rinsed 3 × 5 min with TBS-T and mounted on slides with Vectashield (Fischer Scientific, H1000NB). Cells were imaged with spinning disk fluorescence microscopy. To quantify total adhesion area, paxillin images were analyzed in ImageJ. Images were thresholded with a lower threshold level of 5500 and an upper threshold level of 65,535. The total thresholded area was then calculated.

Adhesion formation assay

View detailed protocol

Cells were trypsinized and plated on glassbottom Matek dishes (5-min timepoint; #1.5 glass) or glass coverslips (20-min timepoint; #1.5 glass) coated with 10 μg/mL fibronectin. For a 5-min time point, unbound cells were gently washed away with fresh media after 1 min. After another 4 min (total spreading time of 5 min), cells were fixed. For a 20-min time point, unbound cells were gently washed away with fresh media after 10 min. After another 10 min (total spreading time of 20 min), cells were fixed. For all experiments, cells were fixed with 3% paraformaldehyde in Cytoskeleton Buffer (CB: 10 mM MES pH 6.1, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl2, 5 mM glucose) for 20 min at room temperature. Cells were then permeabilized for 8 min with CB +0.5% triton-X, quenched for 10 min with CB +0.1 M glycine, and rinsed 2 × 5 min in TBS-T. Coverslips were blocked with 2% BSA in TBS-T for 1 hr. Coverslips were incubated with primary antibody (Ms anti Paxillin, 1:100) in TBS-T with 2% BSA overnight at 4 °C. Coverslips were rinsed 3 × 5 min in TBS-T and incubated with secondary antibody (anti Ms Alexa568, 1:250) in in TBS-T with 2% BSA for 45 min at room temperature. Coverslips were rinsed 3 × 5 min with TBS-T. For the 5-min timepoint, PBS was added to the dish and cells were imaged with TIRF microscopy using 561 illumination. For the 20-min timepoint coverslips were mounted on slides with Vectashield (Fischer Scientific, H1000NB) and imaged with spinning disk fluorescence microscopy. Images in the 488 channel (GFP-tagged proteins) and 561 channel (endogenous paxillin) were acquired.

Counting adhesions

Request a detailed protocol

To reduce experimental noise due to differences in expression levels, GFP-transfected cells were only analyzed if the mean GFP intensity within the cell fell within a defined range (intensity between 1000 and 5000 a.u. following background subtraction, at least 50% of imaged cells were retained with these cutoffs). Adhesions were segmented and counted with ImageJ macros based on a previously published ImageJ workflow (Horzum et al., 2014). First macro: run("Subtract Background...", "rolling = 50 sliding"); run("Enhance Local Contrast (CLAHE)", "blocksize = 19 histogram = 256 maximum = 6 mask=*None* fast_(less_accurate)"); run("Exp"); run("Enhance Contrast", "saturated = 0.35"); Then manually run LoG 3D plugin with sigma = 2. Final macro: setAutoThreshold("Default dark"); setOption("BlackBackground", false); run("Convert to Mask"); run("Invert"); run("Analyze Particles...", "size = 5–1000 circularity = 0.00–1.00 summarize"). The final mask was compared to the original image to visually confirm the results. We also validated the final results of automated analysis by manually segmenting adhesions (data not shown).

Measuring adhesion partition coefficient

Request a detailed protocol

Cells expressing GFP-FAK variants or GFP-Paxillin variants were further analyzed to quantify adhesion partitioning of GFP-tagged protein. The same set of images were analyzed for partitioning and counting adhesions. In ImageJ, the image was manually segmented by thresholding using Otsu’s method for dark background. The mean intensity within the threshold was measured. Then the thresholding was manually adjusted to segment the cytoplasm (i.e. exclude the adhesions) and the mean intensity was measured. The background was subtracted from intensity measurements, and the intensity within adhesions was divided by the intensity within the cytoplasm.

Western blot analysis

Request a detailed protocol

Cells were plated on 10 cm round tissue culture dish and grown until ~75% confluent. Media was removed and plate was gently washed with 1 mL cold PBS. 1 mL of cold PBS was added to the dish and cells were scraped for 30 sec. Cells were pelleted with centrifugation at 100 g for 4 min at 4°C. PBS was aspirated and cells were resuspended and lysed in 300 μL cold RIPA buffer (10 mM Tris-HCl (pH 7.6), 1 mM EDTA, 0.1% SDS, 0.1% Na-Deoxycholate, 1% TritonX-100)+ cOmplete, EDTA-free Protease Inhibitor tablet (Roche). Lysates were centrifuged for 10 min at 10,000 rpm and the supernatant was collected. 100 μL of supernatant was combined with 100 μL of 2 X SDS-PAGE loading buffer and boiled for 10 min to denature proteins. Samples were loaded onto a 10% bis-Acrylamide gel for SDS-PAGE (200 V for 40 min). Samples were transferred onto Immobilon-P PVDF Membrane (90 V for 90 min). Membrane was blocked with 5% BSA in TBS-T for 1 hr. Membrane was incubated with primary antibody in TBS-T +2% BSA overnight at 4 °C (Ms anti FAK: 1:1000; Rb anti Cas: 1:1000; Rb anti actin: 1:5000). Membrane was rinsed 3 × 5 min with TBS-T and incubated with secondary antibody 45 min at room temperature (HRP tagged antibodies, 1:10,000). Membrane rinsed 3 × 5 min with TBS-T. Membrane was treated with ECL (Immobilon Western chemiluminescence HRP substrate) for 1 min and visualized with the ChemiDoc MP imaging system (BioRad). Exposure was optimized for each blot.

Data availability

All imaging data has been deposited on Dryad under https://doi.org/10.5061/dryad.9p8cz8wj0.

The following data sets were generated
    1. Case LB
    2. Rosen MK
    (2022) Dryad Digital Repository
    Synergistic Phase Separation of Two Pathways Promotes Integrin Clustering and Nascent Adhesion Formation.
    https://doi.org/10.5061/dryad.9p8cz8wj0

References

Decision letter

  1. Reinhard Fässler
    Reviewing Editor; Max Planck Institute of Biochemistry, Germany
  2. Anna Akhmanova
    Senior Editor; Utrecht University, Netherlands
  3. Reinhard Fässler
    Reviewer; Max Planck Institute of Biochemistry, Germany
  4. Daniel Lietha
    Reviewer; CNIO, Spain

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Synergistic Phase Separation of Two Pathways Promotes Integrin Clustering and Focal Adhesion Formation" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Reinhard Fässler as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Senior Editor.

The reviewers spent quite a bit of time reviewing and then discussing this paper, in part because the paper lacks important control experiments and not all conclusions appear to have been directly tested. The reviewers considered the possibility of having you resubmit a revised manuscript after clarifying some of these issues but it was ultimately decided that there were likely too many experiments that needed to be done so it would be better to give you the necessary time to perform the missing experiments and resubmit a new manuscript to eLife, or to give you the opportunity to resubmit elsewhere.

Reviewer #1:

Cell-ECM adhesion sites contain clusters of integrins associated with plethora of cytoplasmic adaptor and signaling proteins. How integrins cluster to form small and large adhesion sites is not understood. Case and colleagues demonstrate that p130Cas and FAK commence adhesion assembly by undergoing phase separation, which in turn concentrates kindlin and eventually integrins. This work represents a major conceptual advance as it provides interesting evidence for a mechanism of how integrin adhesion site formation is probably initiated. Still, important control experiments and explanations (DLS experiments, protein concentrations, buffer composition, FRAP measurements, SLB studies, characterization of mutant proteins, etc.) are missing, and certain claims (p130Cas-kindlin binding, FAs versus nascent adhesions, etc) would require further exploration.

Case and colleagues report that multivalent interactions of p130Cas with Nck and N-WASP, and FAK with paxillin undergo phase separation in vitro. Furthermore, the authors demonstrate that p130Cas and FAK synergize to induce in vitro condensates, enrich the integrin-activating and clustering adaptor protein kindlin and induce nascent, cell-ECM adhesions. The paper is well written and interesting for a broad readership. The experiments are carried out by a lab with outstanding accomplishments in deciphering the formation of molecular condensates by liquid-liquid phase separation. Still, important control experiments and explanations (DLS experiments, protein concentrations, buffer composition, FRAP measurements, SLB studies, characterization of mutant proteins, etc.) are missing, and certain claims (p130Cas-kindlin binding, FAs versus nascent adhesions, etc.) require further exploration.

1) The authors should perform dynamic light scattering measurements of the protein stock solutions to exclude that the individual recombinant proteins (in particular FAK) form aggregates before applying them in their assays. Ideally, this is done over time and at different temperatures.

2) Phase diagrams demonstrating the influence of protein concentration versus pH or salt concentrations on the formation of condensates are missing.

3) The concentrations of the recombinant proteins used for in vitro phase separation experiments are far above their concentrations in cells.

4) It is not clear from the result and method sections whether the buffer used in the in vitro studies has the appropriate osmolarity.

5) The FRAP experiments of FAK and paxillin in phase droplets substantially differ from those reported in cellular adhesion sites. Therefore, the authors should determine the material properties (viscosity) of the droplets in vitro and ideally compare them to nascent adhesions in cells, e.g. by observing fusion events of droplets.

6) The influence of p130Cas on kindlin depends on an interaction between the two proteins. A direct binding has not been reported so far, which should therefore be demonstrated.

7) Since kindlin binds paxillin in Zn+2-dependent manner, the authors should compare the influence of EDTA and Zn2+ on FAK-mediated condensates and in particular kindlin enrichment.

8) The SLB experiments are carried out without phospholipids and strangely, in the presence of PEG and BSA. In light of the PH domain that is present in kindlin, this experimental set-up is not plausible.

9) The mutant p130Cas and FAK proteins need to be characterized and tested in phase separation experiments in vitro.

10) The authors conclude that 'that pCas- and FAK-driven phase separation provides an intracellular trigger for integrin clustering and nascent adhesion formation.' In contrast to this conclusion, the cellular studies investigate mature focal adhesion rather than nascent adhesion formation. Ideally, the cellular studies are carried out with cells undergoing isotropic spreading (around 5 min upon seeding, depending on cell type and ECM substrate) which is associated with the assembly (and turn over) of nascent adhesions.

Reviewer #2:

In the present work the authors report the formation of condensates of adhesion complex proteins. They show in a series of experiments that the condensation of pCas and FAK appears to orchestrate the phase separation and the binding of the these complexes to membranes via kindlin.

Although a bit sensational in the title the paper gives a nice proposal of activation of nascent adhesion formation. The title is a bit misleading since focal adhesions are not observed on the SLB and no actin filament binding is involved. A clear distinction between the here observed condensates and the complexes which are focal adhesion sites should be made throughout the manuscript. Maybe the condensates come closer to nascent adhesion sites.

The interpretation of two distinct phase separation pathways working together to promote integrin clustering seems to be to be an overinterpretation and not entirely consistent with literature. It seems unlikely that the p130Cas+NcK+N-WASP would be involved in FA formation, as N-WASP is an actin nucleator and the actin layer in FAs is spatially separated from FAK (https://www.nature.com/articles/nature09621 this paper was cited, but no comment was made to that discrepancy). Which leads to the structure of the complex – which is certainly beyond the scope of the paper but still somehow important to argue a function of the formed aggregates. Neither p130Cas, Nck or N-WASP are components of the consensus adhesome and are not found in nascent FAs.: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4663675/ The cited paper of Meenderink et al. 2010 states that p130Cas effect is most pronounced in FA disassembly. In this paper a pre-existing phosphorylation pattern is used to show complex formation with FAK. But it was shown that a FAK+Src complex leads to the phosphorylation of p130Cas, which is needed for complex formation (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2700377/). In summary, in the present study the protein system of Case et al. 2019 was used, basically only switching out the phosphotyrosine-protein, without really focusing on the context of FA formation process or if the second phase separation mix is even needed. It is not examined if only adding phosphorylated or dephosphorylated p130Cas to the FAK mix is sufficient for increasing the molecular condensates. This would be needed to claim real synergistic behavior – which is actually a still loosely used term here. I am even not sure if really just an increase of condensate size and protein content (fluorescence intensity) at 1µM with undefined stoichiometry is a strong point – or better how to relate this to the emergence of nascent adhesions – which have a well defined size range and are tightly controlled.

It is not clear how the FRAP measurements are done. The condensates are freely diffusing (also rotational) in 3D solution which would make it difficult to obtain proper FRAP analysis. The stable fluorescence before bleaching need to be reported and contribution of recovery by rotational and translational diffusion of the particles themselves need to be experimentally excluded. The presented curves do not seem to look like a single exponential, and the rapid recovery at t=0 and almost no recovery for pCas and NcK bothers quite a bit.

Unambiguous FRAP data and analysis would seem to be key for any claims on the existence of functional condensates.

Additional specific concerns:

– It is unclear to me how the FRAP measurements are done? The condensates are freely diffusing (also rotational!) in 3D solution which would make it difficult to obtain proper FRAP analysis. The stable fluorescence before bleaching need to be reported and contribution of recovery by rotational and translational diffusion of the particles themselves need to be experimentally excluded. The presented curves do not look like a single exponential, and the rapid recovery at t=0 and almost no recovery for pCas and NcK bothers quite a bit.

Unambiguous FRAP data and analysis seem to be key for any claims on the existence of functional condensates.

– In Figure 1b, the authors show the solution turbidity as a function of the concentration of pCas. Could the authors comment on why the turbidity increase from 0nM to 50nM and decrease from 50 nM on. Do the aggregates disappear or is this a different scattering regime? If they disappear, why are the microscopy measurements work at 1µM? why does the solution turbidity of pCas system does not decay above 100 nM thus a different trend than FAK+Paxillin system (Figure 2d)? Does the difference in multivalency matter or explain the differences?

– The pCas system in Figure 4b should have 1 µm of kindlin, or not? It seems it is missing from the caption

Figure 3: In this figure as well as in the text it should be clearer, that "pCas" or "FAK" stands for the whole condensation mix from above This should be explicitly stated and the color code introduced in 3a used in the subsequent mentions of theses protein mixes.

E There is a massive difference in Kindlin Intensities between pCas-droplets and FAK-droplets, but F shows about the same partition coefficient?

Figure 4: No, quality control of SLB is presented, as no fluorescent lipids are present in the SLBs, hence no images shown of SLB or FRAPs of SLB. SLBs are formed in 96 wells which makes the essential washing steps difficult. Well cleaning protocol seems to be insufficient to get clean SLBs from what is commonly state of the art. Control measurements on the quality of the lipid layer needs to be provided (e.g. FRAP of lipids). As a matter of fact, such configuration could also help to perform the FRAP measurements of the condensates.

Surprisingly, and contrary to most reports, no PIP2 is required in the membrane to mediate the adhesion of the complexes to the membrane.

In methods section the written name of the main lipid is DOPC, whereas the given abbreviation is POPC. FAK concentration was lowered, because there was unspecific binding to NiNTA, which is a bit worrisome for the interpretations and comparisons. Increased salt concentration could prevent unspecific binding, but might alter protein condensate dynamics at the same time. Added low concentration (<100mM) of imidazole may help here to prevent unspecific Ni-binding.

Figure 5: In humans, temperature normally varies around 1 {degree sign}C and pH varies 0.1. That means the observed different conditions are far from physiological. It seems a bit far to explain differences in focal adhesion size with the protein condensate properties, as at these conditions the phosphorylation state as well as the lipid composition should be massively altered: https://www.molbiolcell.org/doi/10.1091/mbc.e12-04-0296

Figure 6: The paper is very focused on an effect of phase separation on forming FAs, but the presented in vivo data looks at FAs at a static time point and does not take into account the FA dynamics.

Reviewer #3:

Case, Henry and Rosen investigate the influence of the Cas- and FAK-driven phase separation on the formation of integrin-based focal adhesions. Using a combination of in vitro reconstituted and cellular systems, the authors could show that phase separation is apparently sufficient to recreate kindlin-dependent integrin clustering. Interestingly, the authors were able to develop an in vitro system, which is able to reconstitute nascent focal adhesions on supported lipid bilayers consisting of seven proteins. The study is of great general interest, the data quality in most parts sufficient and the manuscript is well written so it is accessible for the large and interdisciplinary readership of eLife. The results of this manuscript have the potential to provide the community with interesting clues towards understanding focal adhesion formation based on phase separation of a multitude of interactor proteins.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Synergistic Phase Separation of Two Pathways Promotes Integrin Clustering and Integrin Adhesion Complex Formation" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Reinhard Fässler as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen Anna Akhmanova as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Daniel Lietha (Reviewer #3).

This newly submitted manuscript has been reviewed at eLife before. It was rejected due to the vast number of experiments that needed to be done. In the new submission, several concerns have been addressed. However, a few remain after the reviewers have discussed their reviews with one another. The Reviewing Editor has drafted this to help you prepare a revised submission.

Summary:

This paper provides evidence for phase separation as underlying mechanism to initiate the formation of integrin adhesion sites. The experiments show that p130Cas and FAK undergo phase separation, concentrate kindlin and eventually integrins. The experiments and findings are well described, while several controls are missing and certain conclusions not sufficiently substantiated by the experimental results.

Essential revisions:

1) The reviewers request experiments that show whether the membrane and/or integrins influence the formation of condensates seen by TIRF in Figure 4, or if the condensates would also form in absence of integrin/membrane (ie, comparable to in solution). To test this questions the existing quantifications in Figure 4 should be compared with (a) quantifications of condensates on membranes without integrin tails and (b) quantification of condensates without integrin tails and without lipids (i.e. on the glass surface or some inert coating).

2) The reviewers are not convinced that simply counting nascent adhesions provides a good correlation for the in vitro experiments with purified components. It would be better to measure partitioning between cytosol and nascent FAs. This should be done for FAK-WT vs FAK-W266A, Cas-WT vs Cas-Y15F, and should be straight forward with the available images.

3) in vitro, the components of the pCAS or FAK mix increase the partitioning of other components of the same mix into condensates. To correlate this with cells, partitioning of paxillin into nascent adhesions of FAK-/- , FAK-WT (and perhaps FAK-W266A) cells should be measured, and also partitioning of N-WASP or Nck into adhesions of Cas-/-, Cas-WT (and perhaps Cas-Y15F) cells.

4) The reviewers also request phase separation experiments with mutant Cas and FAK. It is important to confirm also in vitro that condensates are smaller when these mutations are introduced into the proteins. Cas carries 15 mutations, and this experiment will exclude at least in part that the cellular effects occur via phase separation-independent mechanisms.

https://doi.org/10.7554/eLife.72588.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

Case and colleagues report that multivalent interactions of p130Cas with Nck and N-WASP, and FAK with paxillin undergo phase separation in vitro. Furthermore, the authors demonstrate that p130Cas and FAK synergize to induce in vitro condensates, enrich the integrin-activating and clustering adaptor protein kindlin and induce nascent, cell-ECM adhesions. The paper is well written and interesting for a broad readership. The experiments are carried out by a lab with outstanding accomplishments in deciphering the formation of molecular condensates by liquid-liquid phase separation. Still, important control experiments and explanations (DLS experiments, protein concentrations, buffer composition, FRAP measurements, SLB studies, characterization of mutant proteins, etc.) are missing, and certain claims (p130Cas-kindlin binding, FAs versus nascent adhesions, etc.) require further exploration.

1) The authors should perform dynamic light scattering measurements of the protein stock solutions to exclude that the individual recombinant proteins (in particular FAK) form aggregates before applying them in their assays. Ideally, this is done over time and at different temperatures.

Our final purification step for every protein preparation is a size exclusion column, to separate potential aggregates from monomers. All our purified proteins, including kindlin and FAK, elute as a single, gaussian peak on the size exclusion column, indicating they are not aggregated (as aggregates would elute in the column void volume). We have optimized purification conditions and final buffers to ensure we have pure, unaggregated protein for our studies. We now more explicitly describe this throughout the methods: the protein is purified by…. “followed by size exclusion chromatography to remove any protein aggregates” …

For example, Author response image 1 is a FAK chromatogram from an SD200 column:

You can see the narrow peak of FAK, eluting after the void volume (~9 mL), followed by a smaller peak of a lower molecular weight contaminant (~60 KDa). Thus, our rigorous purification protocols ensure our starting materials are pure and not aggregated.

Author response image 1

2) Phase diagrams demonstrating the influence of protein concentration versus pH or salt concentrations on the formation of condensates are missing.

With a simple, one-component system, it is possible to map a phase diagram of protein concentration versus salt. However, with complex multi-component systems, this becomes less straightforward as changing the concentration of a single component can have non-intuitive effects. Even with only two proteins phase separating together, the simple features of single-component phase separation are not observed (this is explained in more detail in this recent review: Lyon et al., 2020 https://doi.org/10.1038/s41580-02000303-z).

In the revised manuscript, we performed additional experiments to characterize the dependence of droplet formation on salt, in addition to our previous experiments showing the dependence on pH (Figure 5). We titrated NaCl concentration for two different concentrations of proteins and measured solution turbidity (new Figure 3 —figure supplement 1). We find that turbidity decreases with increasing salt concentrations, consistent with phase separation driven by electrostatic interactions (such as the FAK FERM:FAT interaction). With a higher protein concentration of 250 nM each of pCas, Nck, N-WASP, FAK, Paxillin, and Kindlin, Abs350 approaches 0 around 300 mM NaCl. With a lower protein concentration using the cytoplasmic concentrations of each component, Abs350 approaches 0 around 150 mM NaCl.

3) The concentrations of the recombinant proteins used for in vitro phase separation experiments are far above their concentrations in cells.

Table S3 lists the measured cellular concentrations for all of the proteins used in our studies. In Figure 1 —figure supplement 3 and Figure 2 —figure supplement 2 we examined our systems at these cellular concentrations, and observed small droplets. We also now characterize the effect of salt on droplets formed with cellular concentrations of pCas, Nck, N-WASP, paxillin, and FAK (Figure 3 —figure supplement 1). However, the droplets formed at low concentrations are smaller than the resolution limit of our microscope, and thus not amenable to quantitative fluorescence imaging. Thus, in other parts of the manuscript we used higher concentrations to generate larger droplets, which enable quantification of partition coefficients and FRAP dynamics. Nevertheless, we are confident that the molecules undergo phase separation at physiologic concentrations/conditions, and that the behaviors we observe are reflective of those conditions.

4) It is not clear from the result and method sections whether the buffer used in the in vitro studies has the appropriate osmolarity.

We list the buffers used throughout the results, figure legends and methods section. The standard buffer for most in vitro assays is: 50 mM Hepes pH 7.3, 50 mM KCl, 0.1% BSA. Our buffers contain 50 mM KCl as the salt, which is commonly used for in vitro reconstitutions, especially those involving the actin cytoskeleton (see for example, Hansen et al., 2013: https://doi.org/10.1007/978-1-62703-538-5_9). One of our long-term goals is to examine coupling of actin to the phase separated integrin clusters. Thus, we used a buffer that will be compatible with future studies interrogating the role of actin polymers in regulating in vitro clusters.

As described above, we have performed additional experiments to test the effect of NaCl concentration on condensates. We see that droplet formation becomes less favored between 150-300 mM NaCl. We speculate that at cellular concentrations and osmolarity, spontaneous phase separation would be less likely to occur in the cytoplasm, which could ensure that macromolecular assembly occurs specifically at sites of integrin activation on the membrane.

Finally, we note that there is no universally agreed upon buffer to best represent the interior of the cell, and various biochemical reconstitutions vary widely in composition and osmolarity. For example, as distinct from the actin buffer above, studies of microtubule dynamics are typically performed in BRB80 buffer (80 mM PIPES, 1 mM MgCl2, 1 mM EGTA, pH 6.8; e.g. Hyman and Mitchison, J. Cell Biol., 110, 1607 (1990)), studies of transcription often use acetate buffers (e.g. 10 mM Hepes, 6.5 mM MgSO4, 5 mM magnesium acetate, 70 mM potassium acetate, 5 mM EGTA, 2.5 mM dithiothreitol,4 mM phosphoenolpyruvate,10% glycerol; Lue and Kornberg, PNAS, 84, 8839 (1987)), and reconstitution of vesicle trafficking uses buffers containing KCl and acetate (25 mM HEPES (pH 7.0), 15 mM KCI, 2.5 mM magnesium acetate, and 0.2 M sucrose; Orci, Glick and Rothman, Cell, 46, 1986). In all these cases, meaningful biochemical data that inform on cellular processes have been obtained. We feel that, especially with the new salt dependence data, our biochemical studies will be similarly informative on biology even if our buffers are only approximations of the cytoplasm.

5) The FRAP experiments of FAK and paxillin in phase droplets substantially differ from those reported in cellular adhesion sites. Therefore, the authors should determine the material properties (viscosity) of the droplets in vitro and ideally compare them to nascent adhesions in cells, e.g. by observing fusion events of droplets.

We agree that ideally one would be able to observe similar material properties of condensates reconstituted in vitro with their counterparts in cells. However, numerous considerations makes such comparisons difficult to perform and hard to interpret. First, we obviously are lacking components in our reconstitution that likely contribute strongly to material properties of adhesions in cells—attachment to the extracellular matrix, attachments to the actin cytoskeleton, multi-component attachments to lipids (see below), molecular crowding, etc. Adding these components is well beyond the scope of the present work, and moreover, it is unclear even which might play important roles in dictating material properties.

Second, viscosity cannot be measured simply by observing fusion events, since fusion rates depend on both viscosity and surface tension. Rheological measurements would be necessary. These would be extremely difficult to perform on the inside of the plasma membrane in a live cell, in a short-lived structure like a nascent adhesion. Further, given that our condensates are attached to membranes, membrane fluidity (at the site of adhesion) will play a significant role in their physical properties, and it is unclear how one might reproduce this in vitro.

Third, FRAP measurements in cells vary quite widely in different studies, and can even vary for identical condensates within a single cell. FRAP measurements within focal adhesions vary widely in the literature. FRAP measurements are also only easily done in larger adhesions that remain stable for several minutes (i.e. not in nascent adhesions). Because of these issues is not clear to us how one would measure the material properties of cellular nascent adhesions and meaningfully compare them to properties measured in vitro, and this challenge is beyond the scope of this study.

This being said, we believe that our FRAP experiments on in vitro condensates are, in fact, qualitatively consistent with those previously reported in cells. Although the in vitro droplets described in this study lack many of the core components of cellular adhesion sites, such as talin and vinculin, their FRAP dynamics are within an order of magnitude of reported cellular measurements. Our in vitro data, which fits best with a biexponential model, show FAK t1/2 fast = 6 s; FAK t1/2 slow = 56 s; FAK 62% recovery; paxillin t1/2 fast = 7 s; paxillin t1/2 slow = 78 s; pCas t1/2 fast = 2 s; pCas t1/2 slow = 39s. While our time-courses were too short to reliably determine a fraction recovery, it is clear qualitatively that the condensates also contain an immobile fraction. By comparison, a recent cellular study fit curves with a single exponential and found FAK t1/2 = 10s with 60% recovery, paxillin t1/2 = 16s with 65% recovery and Cas t1/2 = 14s with 70% recovery (Stutchbury et al., 2017). Thus, like our reconstituted droplets, cellular adhesions contain a dynamic pool of molecules that rapidly exchange with the cytoplasm and a more stable population of molecules that do not recover. in vitro condensates, while not perfect, do reasonably recapitulate the published properties of cellular adhesions. To help readers better understand and compare FRAP measurements, we have now included detailed information about the fit in supplemental table 4, and made more direct comparisons between our data and published cellular data.

For example, we now include the following discussion of FRAP data in the results:

“We found that pCas, Nck and N-WASP rapidly exchanged between droplets and bulk solution (Figure 1e; Table S4; pCas t1/2 fast = 2 s; pCas t1/2 slow = 39s; Nck t1/2 = 8 s; N-WASP t1/2 fast = 1 s; N-WASP t1/2 slow = 37 s ), although a fraction of Nck and pCas molecules did not recover in the 100 s timeframe of the experiment, suggesting a slower phase of recovery and/or an immobile fraction (Table S4). In cells, IAC-associated proteins, including Cas, often contain a population of fast exchanging molecules (t1/2 < 30 s) as well as an immobile fraction (up to 50% immobile for some proteins; Table S1). Thus, similar to cellular IACs, droplets exhibit liquid-like material properties with some solid-like elements as well.”

6) The influence of p130Cas on kindlin depends on an interaction between the two proteins. A direct binding has not been reported so far, which should therefore be demonstrated.

To address this concern, we have performed additional biochemical experiments. These demonstrate a direct interaction between kindlin and p130Cas. As the reviewer suggested, we performed a pull-down assay with purified recombinant proteins. We found that hisKindlin could pull down both unphosphorylated Cas and phosphorylated Cas (Figure 3 —figure supplement 2). These new data provide evidence that kindlin interacts directly with p130Cas, and that the interaction does not require phosphorylation of the Cas substrate domain.

7) Since kindlin binds paxillin in Zn+2-dependent manner, the authors should compare the influence of EDTA and Zn2+ on FAK-mediated condensates and in particular kindlin enrichment.

A recent structural study used NMR to investigate the structural basis of the paxillin – kindlin interaction (Zhu et al., 2019: https://doi.org/10.1016/j.str.2019.09.006). They found that the paxillin LIM4 domain binds directly to the Kindlin F0 domain. Furthermore, they showed that the LIM4 domain adopts a canonical LIM domain fold, with two classic zinc fingers. Zinc fingers require Zn2+ to maintain their structure, and EDTA is known to irreversibly denature and aggregate these domains (see for example doi: 10.1042/BJ20041096). Considering this data, treating these condensates with EDTA will most likely denature and aggregate paxillin, making any results difficult to interpret. Thus, we have not explored this issue further.

8) The SLB experiments are carried out without phospholipids and strangely, in the presence of PEG and BSA. In light of the PH domain that is present in kindlin, this experimental set-up is not plausible.

Our SLBs are phospholipid bilayers, and we apologize for not describing our bilayer composition in sufficient detail previously. We have edited the methods to hopefully be less opaque (including adding the lipid information to the resource table) and also now refer to the bilayers in the text and phospholipid bilayers. We use bilayers containing 98% POPC, 2% DGS-NTA(Ni), and 0.1% PE-PEG, and each component has a specific purpose:

POPC is a phosphatidylcholine (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) commonly used to make phospholipid bilayers, liposomes, and nanodiscs for biophysical/biochemical assays

(e.g. Pautot et al., 2005 https://doi.org/10.1038/nchembio737; Jiang et al., 2019 https://doi.org/10.1073/pnas.1916204116; Libby and Su, 2020, 10.1007/978-1-07160266-9_13; Tay and Du, 2017 https://doi.org/10.1016/bs.mie.2016.09.043; Ritchie et al., 2009, 10.1016/S0076-6879(09)64011-8; Carbone et al., PNAS, 2017, https://doi.org/10.1073/pnas.1710358114).

Author response image 2

DGS-NTA(Ni) (1,2-dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl] (nickel salt)) is a Ni2+-chelating lipid that enables attachment of His10-tagged proteins to the bilayer. In Author response image 3, we used it to attach the b-integrin tail.

Author response image 3

PE-PEG-5000 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-5000] is a phospholipid with a PEG moiety on the head group. Adding very small amounts of PEGylated phospholipids ensures the bilayers remain fluid and prevents their tight adhesion to the glass below. Again, doping with PEGylated lipid is widely used in biochemistry performed on supported lipid bilayers (Jiang et al., 2019 https://doi.org/10.1073/pnas.1916204116; Libby and Su, 2020, 10.1007/978-1-0716-0266-9_13; Carbone et al., PNAS, 2017, https://doi.org/10.1073/pnas.1710358114).

Author response image 4

We also include a small amount of BSA (0.1% = 1 mg/mL) in all our in vitro buffers to prevent non-specific interactions of proteins with plastic or lipid surfaces. Again, this is standard practice in a wide swath of biochemical analyses (Kwiatkowski et al., PNAS, 2010; Nair et al., Nat Protoc, 2010; Risca et al., PNAS, 2012; Acker and Auld, Perspectives in Science, 2014; Nyren, Analytical Biochemistry, 1987; Carbone et al., PNAS, 2017, https://doi.org/10.1073/pnas.1710358114). The buffers used in the bilayer assays is the same as all in vitro experiments. Importantly, 1 mg/mL BSA is well below the concentrations used to induce molecular crowding (typically 100 mg/mL or higher).Although our bilayers are a phospholipid bilayer, they do not mimic the lipid composition of the plasma membrane. We agree that the role of specific phospholipids, especially Phosphatidylinositol 4,5-bisphosphate (PIP2) and Phosphatidylinositol 3,4,5bisphosphate (PIP3) is of particular interest to integrin adhesions as FAK and full-length N-WASP can directly interact with PIP2 and kindlin directly interacts with PIP3. To address the Reviewers’ questions about the role of PIP2, we did try to make supported bilayers containing 3% PIP2 using our standard approach. However, these bilayers were not uniform and were immobile. Studies have shown that PIP2 lipids in supported bilayers are rapidly immobilized due to interactions with the underlying glass surface (https://dx.doi.org/10.1021/jacs.0c03800). We are actively developing new technical approaches to enable to incorporation of PIP2 into fluid lipid bilayers. However, this will involve careful optimization and validation of a new technical approach and is better suited for a separate, PIP2-focused study/methods paper.

Our data suggest that phosphoinositides are not explicitly required for macromolecular assembly at integrin adhesions. However, we predict that including PIP2 or PIP3 phospholipids would increase protein density at the membrane through direct binding to FAK (PIP2), N-WASP (PIP2) and kindlin (PIP3), likely promote clustering at lower protein concentrations. If so, this would represent a novel mechanism to control LLPS of membrane-associated condensate systems. Because of this potential, Dr. Case is developing new technologies to examine phosphatidyl inositols in her recently inaugurated independent lab. To address these issues in the revised manuscript we have elaborated on the potential contribution of phosphoinositides towards nascent adhesion assembly and/or focal adhesion maturation in the discussion.

We now include an expanded discussion about phosphoinositides to highlight these possibilities:

“Furthermore, many IAC-associated proteins can interact with negatively charged phosphoinositides. FAK and N-WASP preferentially bind PI(4,5)P2 via basic amino acids (Goni et al., 2014; Papayannopoulos et al., 2005), while kindlin preferentially binds PI(3,4,5)P3 via its PH domain (Liu et al., 2011). We find that phosphoinositides are not necessary to reconstitute clusters on phospholipid bilayers that resemble nascent adhesions. Consistent with our observations, reduction of PI(4,5)P2 synthesis at IACs in cells leads to the formation of small IACs that contain paxillin and kindlin, but lack talin and vinculin (Legate et al., 2011). Furthermore, talin has been shown to directly interact with and activate PIP5K1C (Di Paolo et al., 2002), and PIP4K2A was identified in as a constituent of IACs whose local concentration potentially changes during adhesion assembly and growth (Horton et al., 2015). Thus, PI(4,5)P2 may be locally generated during IAC formation or maturation, and additional interactions between phosphoinositides and cytoplasmic adaptor proteins could be important to reduce the threshold concentration of phase separation during nascent adhesion assembly or subsequent adhesion maturation and growth (Mitchison, 2020).”

9) The mutant p130Cas and FAK proteins need to be characterized and tested in phase separation experiments in vitro.

We performed the corresponding in vitro measurements in Figure 1b-c and Figure 6. While we did not purify Y15F p130Cas, we performed an analogous experiment in vitro using fully dephosphorylated p130Cas (which is functionally very similar to Y15F). We do not observe phase separation with unphosphrylated p130Cas (Figure 1b-c). Similarly, we found that purified FAK W266A failed to phase separate even at 5 µM concentration (assessed by turbidity, figure 6c). Since unphosphorylated Cas and W266A FAK fail to undergo phase separation, further in vitro characterization such as measuring PC or FRAP recovery are not possible (since droplets do not form). To make figure 6 more clear, we labeled the graphs in 6c and 6h with “in vitro”.

10) The authors conclude that 'that pCas- and FAK-driven phase separation provides an intracellular trigger for integrin clustering and nascent adhesion formation.' In contrast to this conclusion, the cellular studies investigate mature focal adhesion rather than nascent adhesion formation. Ideally, the cellular studies are carried out with cells undergoing isotropic spreading (around 5 min upon seeding, depending on cell type and ECM substrate) which is associated with the assembly (and turn over) of nascent adhesions.

Thank you for drawing our attention to this issue. The standard approach to examine nascent adhesion formation or adhesion dynamics is to perform live cell imaging with overexpressed Paxillin-GFP. However, we find that paxillin can enhance phase separation (Figure 2e-f), and thus overexpression would likely alter phase separation-dependent processes. Thus, we decided to use immunostaining to visualize cellular adhesions without altering endogenous protein levels. We initially performed experiments at a 20 min time point, but as the reviewer suggests this could confound effects due to nascent adhesion formation and subsequent maturation. Thus, we have repeated the experiments in the original Figure 7 at a 5-minute time point. We now report these data in a new Figure 7 and Figure 7 —figure supplement 1, and have moved the 20 minute timepoints to Figure 7 —figure supplement 2. At 5 minutes the MEFs were sufficiently adhered to enable fixation and immunostaining, but still appeared to be undergoing isotropic spreading. We also used TIRF microscopy instead of spinning disk microscopy to improve our resolution of nascent adhesions.

These new data were consistent with our initial 20-minute timepoint observations. That is, single knockout or knockdown of FAK or Cas only partially reduces the number of adhesions, while double knockdown leads to severe reduction in adhesion formation. With the 5 minute and 20-minute timepoints, we can more confidently conclude that the effects of Cas and FAK on adhesion number during cell spreading occur primarily due to defects in nascent adhesion formation.

Reviewer #2:

In the present work the authors report the formation of condensates of adhesion complex proteins. They show in a series of experiments that the condensation of pCas and FAK appears to orchestrate the phase separation and the binding of the these complexes to membranes via kindlin.

Although a bit sensational in the title the paper gives a nice proposal of activation of nascent adhesion formation. The title is a bit misleading since focal adhesions are not observed on the SLB and no actin filament binding is involved. A clear distinction between the here observed condensates and the complexes which are focal adhesion sites should be made throughout the manuscript. Maybe the condensates come closer to nascent adhesion sites.

We agree with the reviewer that we are reconstituting clusters that are similar to nascent adhesions. To make this clearer, we have tried to be more specific and consistent with our language throughout the paper. We were using the term focal adhesion to broadly refer to any integrin adhesion complex. We now refer to “integrin adhesion complexes (IACs)” to broadly refer to cellular integrin macromolecular complexes, “nascent adhesions” to refer to newly formed integrin adhesion complexes, and “mature focal adhesion” to refer to the stabilized, elongated integrin adhesion complexes associated with actin stress fibers. We hope it is clear from these changes that our goal in the study was to identify protein interactions involved in nascent adhesion assembly.

The interpretation of two distinct phase separation pathways working together to promote integrin clustering seems to be to be an overinterpretation and not entirely consistent with literature. It seems unlikely that the p130Cas+NcK+N-WASP would be involved in FA formation, as N-WASP is an actin nucleator and the actin layer in FAs is spatially separated from FAK (https://www.nature.com/articles/nature09621 this paper was cited, but no comment was made to that discrepancy). Which leads to the structure of the complex – which is certainly beyond the scope of the paper but still somehow important to argue a function of the formed aggregates.

While N-WASP is an actin nucleator, it is also a membrane-associated protein. Full length N-WASP binds to PIP2 phospholipids and is known to nucleate actin filaments specifically at the plasma membrane (Papayannopoulos et al., 2005; 10.1016/j.molcel.2004.11.054). N-WASP is also phosphorylated by FAK in an adhesion-dependent manner (Wu et al., 2004; 10.1074/jbc.M310739200) and colocalizes with phosphorylated Cas in lamellipodia (Zhang et al., 2014; 10.1242/jcs.134692), and these interactions regulate cell adhesion and spreading.

To our knowledge, no one has looked at the specific nanoscale architecture of nascent adhesions. iPALM experiments are biased towards large, mature adhesions, as confident localization requires measuring the position of 100s – 1000s of individual molecules within each adhesion. Thus, it is unknown when this layered architecture arises during adhesion formation and maturation. Evidence suggests that talin length determines the height of the FA-stress fiber interface (Liu et al., PNAS, 2015; https://doi.org/10.1073/pnas.1512025112). It is possible that axial stratification does not occur until talin starts stretching under mechanical force. While the initiation of the nanoscale organization is an interesting question, it is well beyond the scope of the current paper. We have included a more detailed discussion of adhesion nanoscale architecture, and how the condensates we observe may be modulated by force to generate mature focal adhesions, in the discussion:

“Mature IACs (called focal adhesions) are organized into vertical layers, with paxillin and FAK localized near the membrane and actin and actin binding proteins localized > 50 nm above the membrane (Kanchanawong et al., 2010). At mature focal adhesions, the plasma membrane and actin cytoskeleton provide an inherent polarity; the membrane is physically connected to the ECM through integrins on one side and actin generates forces on the other. These forces are transmitted across the components of the mature focal adhesion. While nascent adhesions may behave like an isotropic liquid condensate, the actin cytoskeleton could induce molecular reorganizations to produce order in the direction perpendicular to the membrane and generate a state more akin to a liquid crystal. While it is not known when this layered organization emerges during IAC assembly and maturation, talin plays a critical role in controlling the distance between actin and the plasma membrane (Liu et al., 2015). Additionally, talin undergoes force dependent conformational changes that extend the protein and expose additional vinculin binding sites (del Rio et al., 2009). Thus, force-dependent talin extension likely plays an important role in patterning the mature focal adhesion. Many other condensates interact with actin, and the layered organization of mature focal adhesions may be more common than previously appreciated (Beutel et al., 2019; Bresler et al., 2004; Zeng et al., 2016).”

Neither p130Cas, Nck or N-WASP are components of the consensus adhesome and are not found in nascent FAs.: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4663675/ The cited paper of Meenderink et al. 2010 states that p130Cas effect is most pronounced in FA disassembly. In this paper a pre-existing phosphorylation pattern is used to show complex formation with FAK. But it was shown that a FAK+Src complex leads to the phosphorylation of p130Cas, which is needed for complex formation (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2700377/). In summary, in the present study the protein system of Case et al. 2019 was used, basically only switching out the phosphotyrosine-protein, without really focusing on the context of FA formation process or if the second phase separation mix is even needed. It is not examined if only adding phosphorylated or dephosphorylated p130Cas to the FAK mix is sufficient for increasing the molecular condensates. This would be needed to claim real synergistic behavior – which is actually a still loosely used term here. I am even not sure if really just an increase of condensate size and protein content (fluorescence intensity) at 1µM with undefined stoichiometry is a strong point – or better how to relate this to the emergence of nascent adhesions – which have a well defined size range and are tightly controlled.

While the proteomics work leading to the development of the consensus adhesome is an immensely powerful resource, the techniques used to develop the consensus are not infallible. Mass spectrometry approaches are more likely to identify proteins that are highly concentrated, stable constituents of integrin adhesions. Lack of detection by these approaches does not de facto mean the protein does not play an important role in integrin adhesions. As the authors of that paper themselves make this point explicitly: “some well characterised IAC components were not enriched in all seven datasets (e.g. β3 integrin, FAK, kindlin, paxillin and talin) or were observed in the meta-adhesome but not the consensus adhesome (e.g. p130Cas and Src family kinases). These omissions may be due to cell-type-specific expression, cell-type-specific IAC maturation, protein abundance at IACs, preferential use of β1 integrin or non-specific detection in negative controls.”

Furthermore, Nck was detected in the Meta-Adhesome and has since also been identified by proximity ligase-based approaches (Dong et al., 2016; 10.1126/scisignal.aaf3572 and Chastney et al., 2020: https://doi.org/10.1083/jcb.202003038). While N-WASP was not detected by proteomics studies, it has been observed to colocalize with pCas in newly formed adhesions via fluorescence microscopy (Zhang et al., 2014; doi: 10.1242/jcs.134692). Furthermore, WASF2, a WASP family member that has 97% amino acid sequence identity to the N-WASP protein used in this study, was recently identified in Kindlin and Paxillin BioID experiments (Dong et al., 2016).

Nck and N-WASP are not the only multivalent adaptor proteins that interact with p130Cas. In cells, there are likely multiple adaptor proteins capable of interacting in a similar fashion with pCas. However, Nck and N-WASP are the adaptor proteins with strongest evidence showing potential role in nascent adhesion formation. We have edited the manuscript to include more of this nuance:

“Within the substrate domain, there are 15 YXXP motifs that, when phosphorylated, can bind to the SH2 domain-containing adaptor proteins, Crk-II and Nck (Iwahara et al., 2004; Pellicena and Miller, 2001; Schlaepfer et al., 1997). In cells, Cas is robustly phosphorylated on 10 of its YXXP motifs, and proper cell migration requires a minimum of four phosphorylation sites (Shin et al., 2004). CrkII contains two SH3 domains which can bind adaptor proteins such as SOS and C3G that contain multiple proline rich motifs (PRMs) (Birge et al., 2009). Similarly, Nck contains three SH3 domains which can bind adaptor proteins such as NWASP, SOS and Abl that contain multiple PRMs (Li et al., 2001). … Nck has been localized within IACs (Goicoechea et al., 2002; Horton et al., 2015), N-WASP colocalizes with Cas within the lamellipodia where nascent adhesions form (Zhang et al., 2014), and both Nck and N-WASP have been implicated in regulating cell adhesion to fibronectin (Misra et al., 2007; Ruusala et al., 2008). Thus, we chose to use Nck as the SH2/SH3 domain containing adaptor protein and N-WASP as the PRM containing adaptor protein for our studies (Figure 1a, Table S2).”

Regarding p130Cas, as the reviewer cites, Cas and FAK directly interact (Wisniewska et al., 2005), and FAK binding to the Cas SH3 domain helps recruit Src kinase to phosphorylate the Cas substrate domain (Hamasaki et al., 1996; Tachibana et al., 1997). However, the related kinase Pyk2 can also promote Src-dependent Cas phosphorylation (Astier et al., 1997), and Cas is robustly phosphorylated in FAK-/- MEFs (Hamasaki et al., 1996). Thus, the phosphorylation of Cas can occur by both FAK-dependent and FAK independent mechanisms. In this study, we do not attempt (or claim) to capture the detailed pathway by which Cas becomes phosphorylated, whether by FAK, Src or a complex of the two. Rather, we examine the consequences of that phosphorylation, which is the ability to assemble with other proteins in the adhesion structure and contribute to its formation. One strength of an in vitro approach is the ability to differentiate between the act of phosphorylation and the downstream consequences of phosphorylation. Since we phosphorylate Cas during purification and do not include ATP in our experiments, we know that once Cas is phosphorylated, kinase activity is not necessary for the macromolecular assembly we observe. While understanding the spatiotemporal dynamics of Cas phosphorylation during nascent adhesion assembly is important, to work out this pathway would be a project unto itself beyond the scope of the present work.

In summary, we believe that based on current literature data our reconstitution does contain a reasonable collection of molecules to support our claim that they capture important mechanistic aspects of nascent adhesion assembly.

It is not clear how the FRAP measurements are done. The condensates are freely diffusing (also rotational) in 3D solution which would make it difficult to obtain proper FRAP analysis. The stable fluorescence before bleaching need to be reported and contribution of recovery by rotational and translational diffusion of the particles themselves need to be experimentally excluded. The presented curves do not seem to look like a single exponential, and the rapid recovery at t=0 and almost no recovery for pCas and NcK bothers quite a bit.

Unambiguous FRAP data and analysis would seem to be key for any claims on the existence of functional condensates.

As detailed in our reply to the next question, in our FRAP experiments the droplets are not freely diffusing, but rather are weakly attached to the PEG-coated slide surface. So they do not move laterally in solution during FRAP bleaching or recovery; it is a reasonable assumption, then, that they are not rotating either. We also bleach the entire droplet and integrate its total volume during recovery. Because of this, corrections for motion are not necessary. We have included additional supplemental figures demonstrating representative images of droplets during FRAP recovery and raw intensity measurements. We also include pre-bleach timepoints in these examples to illustrate stable fluorescence. We note that the intensity at t=0 does not reflect rapid recovery, but rather how much of the molecules were bleached relative to pre-bleach intensity. To more appropriately display our data, the average FRAP recovery curve has been normalized such that the yaxis represents the percent recovery (with normalized intensity of 0 at t=0). We now clarify this point in the methods:

“Quantification of droplet intensity was performed using ImageJ. The mean intensity inside the ROI was measured, and an identical sized region in the bulk solution was used to correct for photobleaching during imaging. Background and photobleaching corrected intensities were used for normalization.

Normalized Intensity = (Intensity – Intensitypostbleach)/(Intprebleach – Intensitypostbleach).

Intprebleach = average of three measurements in the ROI before 405 laser. Intensitypostbleach = intensity within the ROI immediately after 405 laser (at t=0).”

At the referee’s suggestion we have reanalyzed the recovery curves using bi-exponentials, and found that indeed these fit better than single exponentials with the exception of Nck. We now report both time constants in the Figures 1E and 2G, and include more detailed information about the fit in a new supplemental table (Table S4). With the exception of NWASP, most proteins show a percentage of immobile (non-recovering) molecules, although for some molecules we are unable to accurately determine the percent recovery. We and others have observed this for other phase separated condensates both in vitro (Nephrin condensates: Banjade & Rosen, eLife, 2014; cGas condensates: Du and Chen, Science, 2020; Post-synaptic densities: Zeng et al, Cell, 2018) and in cells (Nephrin condensates: Kim et al., Mol. Biol. Cell, 2019; P-bodies: Xing et al., eLife, 2020; Stress Granules: Niewidok et al., 2018; doi: 10.1083/jcb.201709007; cGas condensates: Du and Chen, Science, 2020; Post-synaptic densities: Ying et al, Journal of Neuroscience, 2010). The functional significance, if any, of the immobile fraction remains unknown in virtually all cases.

In fact, the immobile fractions that we observe (between 60 – 45% immobile for different proteins) in our reconstitution are reasonably close to those observed for the same molecules in cellular IACs. For example, one recent study found that 40% of FAK molecules, 35% of paxillin molecules, and 30% of Cas molecules are immobile within IACs (Stutchbury et al., 2017). This behavior indicates that many condensates, including IACs, contain both dynamic and immobile elements; the physical and function relationships between these elements is an open question in the field.

Additional specific concerns:

- It is unclear to me how the FRAP measurements are done? The condensates are freely diffusing (also rotational!) in 3D solution which would make it difficult to obtain proper FRAP analysis. The stable fluorescence before bleaching need to be reported and contribution of recovery by rotational and translational diffusion of the particles themselves need to be experimentally excluded. The presented curves do not look like a single exponential, and the rapid recovery at t=0 and almost no recovery for pCas and NcK bothers quite a bit.

Unambiguous FRAP data and analysis seem to be key for any claims on the existence of functional condensates.

FRAP measurements of in vitro condensates is a standard approach commonly used by many labs to assess general material properties of condensates (Alberti et al., 2020 10.1016/j.cell.2018.12.035). To better illustrate our FRAP data, we have now included additional supplemental figures with representative individual FRAP measurements, including images of the droplet during recovery and raw fluorescence intensity measurements (Figure 1 —figure supplement 5 and Figure 2 —figure supplement 3). We note that the intensity at t=0 does not reflect rapid recovery, but rather how much of the molecules were bleached relative to pre-bleach intensity. To more appropriately display our data, the average FRAP recovery curve has been normalized such that the y-axis now represents the percent recovery (with 0% recovery at t=0).

We bleach the entire droplet (~5-micron diameter), and the droplets do not significantly move over the course of our experiment. We observe consistent droplet intensity in unbleached droplets during timelapse imaging of similar duration, making us confident that movement or rotation of the droplet is not greatly affecting our fluorescence intensity measurements (Over the course of a control timelapse, the average droplet intensity = 4252 with a standard deviation of 64, in the representative graph in Author response image 5).

Author response image 5

In the revised manuscript, we include three pre-bleach intensity measurements in the new representative FRAP graphs (Figure 1 —figure supplement 5 and Figure 2 —figure supplement 3). We have also included the fit of the data on our graphs in Figure 1e and 2 g. We also include a new supplemental table with detailed information about the fits (Table S4). We have added a more nuanced discussion of the FRAP data, including more discussion of the population of molecules that does not recover.For example, we now include the following discussion of FRAP data:

“We found that pCas, Nck and N-WASP rapidly exchanged between droplets and bulk solution (Figure 1e; Table S4; pCas t1/2 fast = 2 s; pCas t1/2 slow = 39s; Nck t1/2 = 8 s; N-WASP t1/2 fast = 1 s; N-WASP t1/2 slow = 37 s ), although a fraction of Nck and pCas molecules did not recover in the 100 s timeframe of the experiment, suggesting a slower phase of recovery and/or an immobile fraction (Table S4). In cells, IAC-associated proteins, including Cas, often contain a population of fast exchanging molecules (t1/2 < 30 s) as well as an immobile fraction (up to 50% immobile for some proteins; Table S1). Thus, similar to cellular IACs, droplets exhibit liquid-like material properties with some solid-like elements as well.”

– In Figure 1b, the authors show the solution turbidity as a function of the concentration of pCas. Could the authors comment on why the turbidity increase from 0nM to 50nM and decrease from 50 nM on. Do the aggregates disappear or is this a different scattering regime? If they disappear, why are the microscopy measurements work at 1µM? why does the solution turbidity of pCas system does not decay above 100 nM thus a different trend than FAK+Paxillin system (Figure 2d)? Does the difference in multivalency matter or explain the differences?

In previous work (Yoshizawa et al., Cell, 2018; Cinar and Winter, SciReports, 2020; Schuster et al., PNAS, 2020; Aleberti et al., Cell 2019) we and others have observed a strong correlation between turbidity values and the number and size of droplets observed by microscopy. Thus, turbidity is generally used as a measure of the total summed volume of all droplets in solution—essentially, the degree of phase separation in the system. So we believe that the bell-shaped curve in Figure 1b reflects increasing, and then decreasing phase separation as pCas concentration increases. The microscopy measurements were done with all three components at 1 µM (rather than just N-WASP, as in Figure 1b), where droplet formation is robust.

This behavior is expected for multi-component phase separating systems based on modular domain interactions (Li et al., Nature 2012; Case et al., Science 2019). Essentially, the relative concentrations of the domains must be balanced in order to produce oligomerization and consequent phase separation. When the concentrations become imbalanced as one component increases to large excess, the systems titrate out of the 2phase regime. As the referee intuits, this behavior is modulated by the valence of the interacting species, and differences in valence (e.g. pCas system vs FAK system), and also molecular drivers of phase separation (heterotypic interactions for pCas vs heterotypic + homotypic for FAK) can produce different behaviors. In the pCas system, pCas has 6 Nck SH2 binding sites and Nck has only 1 SH2 domain + 3 SH3 domains. Maximum phase separation likely occurs at an optimal ratio of the three components that enables a highly interconnected, oligomerized network. It is interesting that we observe highest phase separation near the physiological concentration of Cas (70 nM).

– The pCas system in Figure 4b should have 1 µm of kindlin, or not? It seems it is missing from the caption.

Yes everything in 4B should have kindlin (except control). This was an error in our figure legend, and has been corrected.

Figure 3: In this figure as well as in the text it should be clearer, that "pCas" or "FAK" stands for the whole condensation mix from above This should be explicitly stated and the color code introduced in 3a used in the subsequent mentions of theses protein mixes.

Thank you for these helpful suggestions. We now refer to “pCas mix” “FAK mix” and “pCas + FAK mix” to distinguish these terms from the individual proteins. We have also colorcoded Figure 3e-h and Figure 4b-I to match the color scheme introduce in Figure 3a. We hope these changes make our experiments and text clearer.

E There is a massive difference in Kindlin Intensities between pCas-droplets and FAK-droplets, but F shows about the same partition coefficient?

Both example images fall within the range of what we observe in the quantification shown in Figure 3f. However, to make the figure more intuitive, we have selected images more representative of the population average to display in Figure 3e.

Figure 4: No, quality control of SLB is presented, as no fluorescent lipids are present in the SLBs, hence no images shown of SLB or FRAPs of SLB. SLBs are formed in 96 wells which makes the essential washing steps difficult. Well cleaning protocol seems to be insufficient to get clean SLBs from what is commonly state of the art. Control measurements on the quality of the lipid layer needs to be provided (e.g. FRAP of lipids). As a matter of fact such configuration could also help to perform the FRAP measurements of the condensates.

Our lab and others have been using this approach to successfully generate fluid, POPC bilayers for many biophysical and cellular studies (See the detailed methods papers we follow, now cited in the manuscript: Su et al., 2017; https://doi.org/10.1007/978-1-49396881-7_5). While laborious, we do wash 96-well plates thoroughly, well by well. Additionally, we have performed many quality-control experiments throughout the course of these studies that are not reported. In fact, we find that integrin clustering does not occur if the bilayer is poorly made or immobile. Prior to any experiment, we confirm the fluidity and uniformity of the bilayer by imaging the distribution and dynamics of His10integrin. Bilayers must exhibit t1/2 recovery < 10 s, indirectly assessed by His10-integrin FRAP, to be used for clustering experiments. For clustering experiments, we do not include fluorescent lipids in the bilayer, as this could potentially change the chemistry of the bilayer. However, we frequently make bilayers with fluorescent lipids as a qualitycontrol step, to confirm our methods generate fluid, uniform bilayers.

To address the reviewer’s concerns, we included a new supplemental figure (Figure 4 —figure supplement 1) with data characterizing the bilayer with fluorescent lipids. We include 1% PC-NBD (a fluorescent phospholipid with excitation/emission at 464/531 nm) to directly visualize the lipid bilayer. While NBD is less photostable than Alexa fluorophores, we can correct for photobleaching that occurs during the timelapse. We consistently generate bilayers with FRAP t1/2 recovery < 10 s for a 5 µm diameter bleach ROI. This is now specifically described in the methods (under “Reconstitution of Supported Phospholipid Bilayers”):

“Prior to any experiment, the fluidity of bilayers was indirectly assessed by imaging integrin Alexa-488. We photobleached a 5-micron region, and experiments were only performed if FRAP t1/2 < 10 s. To confirm that our methods consistently give fluid, uniform bilayers, we also assessed bilayer fluidity directly in bilayers containing 1% PC-NBD, a fluorescent lipid (Figure 4, supplement 1).”

Surprisingly, and contrary to most reports, no PIP2 is required in the membrane to mediate the adhesion of the complexes to the membrane.

To clarify one point, in our system the integrin tail is attached to the bilayer through an Nterminal His10-tag, which binds with high affinity to the 2% Ni-NTA lipid in the bilayer. So the mechanism by which the condensates associate with the bilayer is through binding the membrane-attached integrin.

This being said, as described above, we anticipate that the adhesion of the condensate to the membrane would be stronger if we had incorporated PIP2 or PIP3 into the bilayer as well. This is because FAK and full-length N-WASP can directly bind PIP2 and kindlin directly binds PIP3. In cells, it may be that cooperative binding to phosphatidylinositols and integrin tails is needed for condensates to be bright enough to readily observe. To address the Reviewers’ questions about the role of PIP2, we did try to make supported bilayers containing 3% PIP2 using our standard approach. However, these bilayers were not uniform and were immobile. Recent studies have shown that PIP2 lipids in supported bilayers are rapidly immobilized due to interactions with the underlying glass surface (https://dx.doi.org/10.1021/jacs.0c03800). We are actively developing new technical approaches to enable to incorporation of PIP2 into fluid lipid bilayers. However, this will involve careful optimization and validation of a new technical approach and is better suited for a separate, PIP2-focused study/methods paper.

Alternatively, PIP2 may be needed primarily in later stages of adhesion formation/maturation. In cells, depletion of PIPKIγ from IACs (to prevent local PIP2 synthesis) lead to the formation of small IACs that contain paxillin and kindlin, but lack talin and vinculin (Legate et al., 2011). This is consistent with our findings that phosphoinositides are not required for macromolecules (including paxillin and kindlin) to assemble and cluster integrin tails in vitro, and our assertion that these condensates may best represent nascent adhesions. To our knowledge, no cellular studies have definitively investigated the specific timing of PIP2 generation during nascent adhesion assembly and focal adhesion growth. Talin has been shown to directly interact with and activate PIP5K1C (Paolo et al., 2002) and PIP4K2A was identified in as a constituent of integrin adhesion complexes whose local concentration potentially changes during adhesion assembly and growth (Horton et al., 2015). Thus, while PIP2 may influence adhesions by binding to adaptor proteins, adhesions likely also influence the local concentration of PIP2. This could positively reinforce adhesion formation and/or growth. We have elaborated on the potential contribution of PIP2 interactions towards nascent adhesion assembly and/or focal adhesion maturation in the discussion:

“Furthermore, many IAC-associated proteins can interact with negatively charged phosphoinositides. FAK and N-WASP preferentially bind PI(4,5)P2 via basic amino acids (Goni et al., 2014; Papayannopoulos et al., 2005), while kindlin preferentially binds PI(3,4,5)P3 via its PH domain (Liu et al., 2011). We find that phosphoinositides are not necessary to reconstitute clusters on phospholipid bilayers that resemble nascent adhesions. Consistent with our observations, reduction of PI(4,5)P2 synthesis at IACs in cells leads to the formation of small IACs that contain paxillin and kindlin, but lack talin and vinculin (Legate et al., 2011). Furthermore, talin has been shown to directly interact with and activate PIP5K1C (Di Paolo et al., 2002), and PIP4K2A was identified as a constituent of IACs whose local concentration potentially changes during adhesion assembly and growth (Horton et al., 2015). Thus, PI(4,5)P2 may be locally generated during IAC formation or maturation, and additional interactions between phosphoinositides and cytoplasmic adaptor proteins could be important to reduce the threshold concentration of phase separation during nascent adhesion assembly or subsequent adhesion maturation and growth (Mitchison, 2020).”

Addressing these various possibilities in vitro and in cells will require non-trivial technical innovations and is beyond the scope of our current study. We believe that these open questions do not detract from our findings, but simply say that there are additional aspects of the system to explore beyond this initial report.

In methods section the written name of the main lipid is DOPC, whereas the given abbreviation is POPC.

We apologize for this typo in the methods section. We used POPC, not DOPC, for all experiments. We have corrected the methods section and added more detailed information on the lipids used in the resource table.

FAK concentration was lowered, because there was unspecific binding to NiNTA, which is a bit worrisome for the interpretations and comparisons. Increased salt concentration could prevent unspecific binding, but might alter protein condensate dynamics at the same time. Added low concentration (<100mM) of imidazole may help here to prevent unspecific Ni-binding.

After the initial observation that FAK was interacting with Nickel, we screened many different buffers to reduce this effect (Assessed with Ni-NTA resin binding). The only addition that reduced the FAK-Ni interaction was 20 mM or 40 mM Imidazole. However, 20 mM Imidazole also disrupted the attachment of His10-Integrin to the bilayer (assessed with microscopy in Author response image 6):

Author response image 6

From these experiments, we concluded that using His10-integrin (instead of His8-integrin) in combination with 200 nM FAK was the most effective approach to ensure integrin remained attached to the bilayer while reducing the effect of nonspecific FAK-Ni interactions on the observations.

Author response image 7

Figure 5: In humans, temperature normally varies around 1 {degree sign}C and pH varies 0.1. That means the observed different conditions are far from physiological. It seems a bit far to explain differences in focal adhesion size with the protein condensate properties, as at these conditions the phosphorylation state as well as the lipid composition should be massively altered: https://www.molbiolcell.org/doi/10.1091/mbc.e12-04-0296

The main goal of Figure 5 is to use environmental perturbations to directly compare in vitro droplets and integrin adhesions, even if these perturbations are not physiologically relevant. Such non-physiologic treatments are widely used in cellular experiments—e.g. heat shock, osmotic shock, various drug or small molecule treatments. It is in this spirit that we perform these experiments. We specifically look at shorter timepoints (10 min) to increase the likelihood that effects are due to phase separation (which should be a fast response) and not cellular responses that rely on slower changes in transcription, translation, or other longer-term adaptations. The specific paper the reviewer cites is looking at timepoints hours (10-24 hr) after temperature shock, which is a very different experiment from the 10 minute treatments we use here.

Furthermore, we disagree that these pH differences are not physiological, since the local pH at FAs has been observed to vary in cells between ~ 7.1 – 7.7 due to the local concentration and activation of NHE1, a Na/H+ exchanger (Ludwig et al., 201310.1002/jcp.24293). Thus, of the possible environmental perturbations that can occur in cells, regulated fluctuations in pH are the most likely to be relevant to integrin adhesions.

Figure 6: The paper is very focused on an effect of phase separation on forming FAs, but the presented in vivo data looks at FAs at a static time point and does not take into account the FA dynamics.

The standard approach to examine integrin adhesion dynamics is to perform live cell imaging of Paxillin-GFP. However, we find that paxillin can enhance phase separation (Figure 2e-f), and thus overexpression of paxillin would likely alter phase separation dependent processes. Thus, we decided to use immunostaining to visualize adhesions and infer information about adhesion formation without altering endogenous protein levels. We initially performed experiments at a 20 min time point, but this could confound effects due to adhesion formation and adhesion maturation.

Thus, taking the suggestion of reviewers, we have repeated the experiments at a 5-minute time point. At 5 minutes the MEFs were sufficiently adhered to enable fixation and immunostaining, but still appeared to be undergoing isotropic spreading. We also used TIRF microscopy instead of spinning disk microscopy to improve our resolution of nascent adhesions. With our new data, we now have a 5 min (Figure 7b-c) and a 20 min (Figure 7 —figure supplement 2) snapshot of endogenous integrin adhesions during cell spreading, and find similar behavior in both. This new data strengthens our conclusion that Cas and Fak are required for integrin adhesion assembly.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential Revisions:

1) The reviewers request experiments that show whether the membrane and/or integrins influence the formation of condensates seen by TIRF in Figure 4, or if the condensates would also form in absence of integrin/membrane (ie, comparable to in solution). To test this questions the existing quantifications in Figure 4 should be compared with (a) quantifications of condensates on membranes without integrin tails and (b) quantification of condensates without integrin tails and without lipids (i.e. on the glass surface or some inert coating).

We have performed the requested experiment and the results are now included in Figure 4f-g. This experiment was extremely insightful, revealing that integrin specifically increases the rate of droplet appearance in the TIRF field compared with an empty lipid bilayer. The effect of integrin was even more pronounced when we repeated this experiment with physiological protein concentrations (70 nM pCas, 180 nM Nck (15% Alexa647 labeled), 140 nM N-WASP, 60 nM Kin, 60 nM Paxillin, 40 nM FAK). Thus, while condensates are observed in the TIRF field in the absence of integrin (especially with higher 1 μM protein concentrations), we also observe that integrin specifically increases condensate formation, likely through increasing the rate of nucleation on the membrane.

2) The reviewers are not convinced that simply counting nascent adhesions provides a good correlation for the in vitro experiments with purified components. It would be better to measure partitioning between cytosol and nascent FAs. This should be done for FAK-WT vs FAK-W266A, Cas-WT vs Cas-Y15F, and should be straight forward with the available images.

We have reanalyzed the images from these figures to measure paxillin partitioning in adhesions. These data are now in Figure 6g (Cas+/+, Cas-/-, Cas WT, CasY15F) and Figure 7g (FAK +/+, FAK-/-, WT FAK, FAK W266A).

3) in vitro, the components of the pCAS or FAK mix increase the partitioning of other components of the same mix into condensates. To correlate this with cells, partitioning of paxillin into nascent adhesions of FAK-/- , FAK-WT (and perhaps FAK-W266A) cells should be measured, and also partitioning of N-WASP or Nck into adhesions of Cas-/-, Cas-WT (and perhaps Cas-Y15F) cells.

With new in vitro measurements (see point 4 below), we find that paxillin partitioning decreases in pCas+FAK droplets with both FAK W266A and Cas Y15F proteins, likely because Cas interacts with both FAK and paxillin in addition to Nck. Thus, we can address this question by correlating paxillin partitioning in vitro and in cellular experiments. These comparisons can be found in Figure 6 (Cas Y15F) and Figure 7 (FAK W266A), and indeed show parallel responses in vitro and in cells to these perturbations.

We did also try to measure Nck partitioning into Cas-/- cells. However, this proved to be a confounding experiment. We did not have reliable antibodies to measure endogenous N-WASP or Nck at adhesions, so we transiently expressed Nck-mCherry in cells and labeled adhesions with paxillin antibodies. However, we found that cells overexpressing Nck often failed to spread and form adhesions, suggesting the Nck overexpression was somehow inhibiting adhesion formation or spreading. See two examples of this phenotype in Cas +/+ MEFs in Author response image 8:

Author response image 8

Despite this, we were able to find some cells expressing Nck-mCherry that formed adhesions (imaged in Author response image 9 and (Author response image 1) with spinning disk microscopy). We do see Nck colocalized within Paxillin-labeled adhesions in cells with WT Cas but not with Y15F Cas, although Nck has a higher cytosolic localization and lower adhesion partition compared with paxillin in WT Cas cells.

Author response image 9
Author response image 10

However, due to the unexpected and not understood effect of Nck expression on cell spreading and/or adhesion formation, we did not include these data in the manuscript, as they would require analysis of a small subpopulation of cells, which we feel are not reliable.

4) The reviewers also request phase separation experiments with mutant Cas and FAK. It is important to confirm also in vitro that condensates are smaller when these mutations are introduced into the proteins. Cas carries 15 mutations, and this experiment will exclude at least in part that the cellular effects occur via phase separation-independent mechanisms.

We have included new in vitro experiments with Cas Y15F. Just like unphosphorylated Cas, CasY15F did not undergo any phase separation when combined with Nck and N-WASP, which we have demonstrated with both solution turbidity (Figure 6a) and microscopy (Figure 6 —figure supplement 1). Similarly, FAK W266A does not undergo any phase separation on its own (Figure 7a, Figure 7 —figure supplement 2). We have also quantified the effects of mutant Cas and FAK on pCas+FAK droplets, which more specifically mirror the perturbations we do in cells. Both Cas Y15F and FAK W266A reduce phase separation in the Cas+FAK mix and significantly decrease paxillin partitioning into droplets. These new experiments provide a more direct correlation between in vitro experiments and cellular data.

https://doi.org/10.7554/eLife.72588.sa2

Article and author information

Author details

  1. Lindsay B Case

    1. Department of Biophysics, Howard Hughes Medical Institute, The University of Texas Southwestern Medical Center, Dallas, United States
    2. Department of Biology, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft, Writing – review and editing
    For correspondence
    lcase@mit.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3166-1138
  2. Milagros De Pasquale

    Department of Biology, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Investigation, Performed experiments during revision
    Competing interests
    No competing interests declared
  3. Lisa Henry

    Department of Biophysics, Howard Hughes Medical Institute, The University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Michael K Rosen

    Department of Biophysics, Howard Hughes Medical Institute, The University of Texas Southwestern Medical Center, Dallas, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing – review and editing
    For correspondence
    michael.rosen@utsouthwestern.edu
    Competing interests
    is a co-founder of Faze Medicines
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0775-7917

Funding

Damon Runyon Cancer Research Foundation (postdoctoral fellowship)

  • Lindsay B Case
  • Michael K Rosen

Damon Runyon Cancer Research Foundation (Dale Frey Scientist Award)

  • Lindsay B Case

Howard Hughes Medical Institute (Investigator)

  • Michael K Rosen

Welch Foundation (I-1544)

  • Michael K Rosen

Damon Runyon Cancer Research Foundation (DRG-2249-16)

  • Lindsay B Case
  • Michael K Rosen

Damon Runyon Cancer Research Foundation (DFS-38-20)

  • Lindsay B Case

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Steve Hanks (Vanderbilt) and Larisa Ryzhova (Maine Medical Center) for sharing Bcar1-/- cell lines, Robert Liddington (Sanford Burnham Institute) for sharing talin and FAK DNA plasmids, W Todd Miller (Stony Brook University) for sharing p130Cas DNA plasmid, Jun Qin (Cleveland Clinic) for sharing kindlin DNA plasmid, and David Corey and Jiaxin Hu (UT Southwestern Medical Center) for use of their Cary 100 UV-Visible spectrophotometer and Peltier thermal controller. We thank the UT Southwestern Proteomics Core Facility and the Koch Institute Biopolymers & Proteomics Facility for performing mass spectrometry analysis. L.B.C. was a Robert Black Fellow of the Damon Runyon Cancer Research Foundation (DRG-2249–16) and is further supported (in part) by the Damon Runyon Cancer Research Foundation (DFS-38–20). This work was also supported by the Howard Hughes Medical Institute and the Welch Foundation (grant I-1544).

Senior Editor

  1. Anna Akhmanova, Utrecht University, Netherlands

Reviewing Editor

  1. Reinhard Fässler, Max Planck Institute of Biochemistry, Germany

Reviewers

  1. Reinhard Fässler, Max Planck Institute of Biochemistry, Germany
  2. Daniel Lietha, CNIO, Spain

Publication history

  1. Received: July 28, 2021
  2. Accepted: January 10, 2022
  3. Accepted Manuscript published: January 20, 2022 (version 1)
  4. Version of Record published: January 26, 2022 (version 2)

Copyright

© 2022, Case et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,662
    Page views
  • 477
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Lindsay B Case
  2. Milagros De Pasquale
  3. Lisa Henry
  4. Michael K Rosen
(2022)
Synergistic phase separation of two pathways promotes integrin clustering and nascent adhesion formation
eLife 11:e72588.
https://doi.org/10.7554/eLife.72588

Further reading

    1. Biochemistry and Chemical Biology
    2. Cancer Biology
    Stefania Monterisi, Johanna Michl ... Pawel Swietach
    Research Article Updated

    Growth of cancer cells in vitro can be attenuated by genetically inactivating selected metabolic pathways. However, loss-of-function mutations in metabolic pathways are not negatively selected in human cancers, indicating that these genes are not essential in vivo. We hypothesize that spontaneous mutations in ‘metabolic genes’ will not necessarily produce functional defects because mutation-bearing cells may be rescued by metabolite exchange with neighboring wild-type cells via gap junctions. Using fluorescent substances to probe intercellular diffusion, we show that colorectal cancer (CRC) cells are coupled by gap junctions assembled from connexins, particularly Cx26. Cells with genetically inactivated components of pH regulation (SLC9A1), glycolysis (ALDOA), or mitochondrial respiration (NDUFS1) could be rescued through access to functional proteins in co-cultured wild-type cells. The effect of diffusive coupling was also observed in co-culture xenografts. Rescue was largely dependent on solute exchange via Cx26 channels, a uniformly and constitutively expressed isoform in CRCs. Due to diffusive coupling, the emergent phenotype is less heterogenous than its genotype, and thus an individual cell should not be considered as the unit under selection, at least for metabolite-handling processes. Our findings can explain why certain loss-of-function mutations in genes ascribed as ‘essential’ do not influence the growth of human cancers.

    1. Biochemistry and Chemical Biology
    2. Neuroscience
    Jinli Geng, Yingjun Tang ... Xiaodong Liu
    Research Article

    Dynamic Ca2+ signals reflect acute changes in membrane excitability (e.g. responses to stimuli), and also mediate intracellular signaling cascades that normally take longer time to manifest (e.g., regulations of transcription). In both cases, chronic Ca2+ imaging has been often desired, but largely hindered by unexpected cytotoxicity intrinsic to GCaMP, a popular series of genetically-encoded Ca2+ indicators. Here, we demonstrate the performance of GCaMP-X in chronic Ca2+ imaging with long-term probe expression in cortical neurons, which has been designed to eliminate the unwanted interactions between conventional GCaMP indicators and endogenous (apo)calmodulin-binding proteins. By expressing in live adult mice at high levels over an extended time frame, GCaMP-X indicators showed less damage and improved performance in two-photon imaging of acute Ca2+ responses to whisker deflection or spontaneous Ca2+ fluctuations. Chronic Ca2+ imaging data (³1 month) were acquired from cultured cortical neurons expressing GCaMP-X, unveiling that spontaneous/local Ca2+ transients would progressively develop into autonomous/global Ca2+ oscillations. Besides the morphological indices of neurite length and soma size, the major metrics of oscillatory Ca2+, including rate, amplitude and synchrony were also examined along with the multiple stages (from neonatal to mature) during neural development. Dysregulations of both neuritogenesis and Ca2+ oscillations were observed typically in 2-3 weeks, which were exacerbated by stronger or prolonged expression of GCaMP. In comparison, neurons expressing GCaMP-X exhibited significantly less damage. By varying the timepoints of virus infection or drug induction, GCaMP-X outperformed GCaMP similarly in cultured mature neurons. These data altogether highlight the unique importance of oscillatory Ca2+ to morphology and health of neurons, presumably underlying the differential performance between GCaMP-X and GCaMP. In summary, GCaMP-X provides a viable option for Ca2+ imaging applications involving long-time and/or high-level expression of Ca2+ probes.