Rap1 GTPases bind effectors, such as RIAM, to enable talin1 to induce integrin activation. In addition, Rap1 binds directly to the talin1 F0 domain (F0); however, this interaction makes a limited contribution to integrin activation in CHO cells or platelets. Here, we show that talin1 F1 domain (F1) contains a previously undetected Rap1-binding site of similar affinity to that in F0. A structure-guided point mutant (R118E) in F1, which blocks Rap1 binding, abolishes the capacity of Rap1 to potentiate talin1-induced integrin activation. The capacity of F1 to mediate Rap1-dependent integrin activation depends on a unique loop in F1 that has a propensity to form a helix upon binding to membrane lipids. Basic membrane-facing residues of this helix are critical, as charge-reversal mutations led to dramatic suppression of talin1-dependent activation. Thus, a novel Rap1-binding site and a transient lipid-dependent helix in F1 work in tandem to enable a direct Rap1–talin1 interaction to cause integrin activation.
Integrin receptors are critical mediators of cellular adhesion, migration, and assembly of the extracellular matrix, thereby playing an indispensable role in development and in many pathological processes (Hynes, 2002). Regulation of the affinity of integrins for their ligands is central to their function. In particular, integrins in blood cells are usually expressed in a low-affinity form until intracellular signals initiated by agonists acting via distinct excitatory receptors induce a high-affinity state, a process operationally defined as integrin activation. Binding of talin1 to the cytoplasmic tail of integrin β1 (Tadokoro et al., 2003; Shattil et al., 2010), β2 (Simonson et al., 2006), β3 (Nieswandt et al., 2007; Petrich et al., 2007a,b; Haling et al., 2011), and β7 (Sun et al., 2018) is a critical final common step in integrin activation.
Talin1 is a large (270-kD) multidomain protein that links integrins to the actin cytoskeleton via its N-terminal head domain that binds to the β-integrin cytoplasmic tail and its C-terminal flexible rod domain that binds F-actin (Critchley and Gingras, 2008). The talin1 head domain (THD) comprises an atypical FERM (4.1 protein, ezrin, radixin, and moesin) domain, because it contains an additional F0 domain to the characteristic FERM F1, F2, and F3 domains and can adopt a linear arrangement rather than the cloverleaf structure typically observed in other FERM domains (Calderwood et al., 2013). Talin1 is autoinhibited in the cytosol due to the interaction of the THD with the rod domain, which prevents the interaction of THD with the integrin β cytoplasmic tail (Song et al., 2012). Our understanding of the signaling events that regulate talin1 recruitment to the plasma membrane and its association with integrin is incomplete.
Rap1 GTPases are perhaps the most completely studied relays of signals from cell surface receptors to integrin activation. Combined deficiency of both Rap1A and Rap1B isoforms in the megakaryocyte lineage showed an essential role for Rap1 signaling in platelet integrin activation and function (Stefanini et al., 2018). Although the Rap1 effector Rap1-GTP–interacting adapter molecule (RIAM) plays a major role in recruiting talin1 in leukocytes (Klapproth et al., 2015; Su et al., 2015; Lagarrigue et al., 2017), its role in talin1-dependent activation of platelet integrins has been unequivocally ruled out (Klapproth et al., 2015; Stritt et al., 2015; Su et al., 2015).
Studies using recombinant αIIbβ3-expressing A5 CHO cells revealed that Rap1 activity regulates THD-induced integrin αIIbβ3 activation (Lagarrigue et al., 2018), while activation by the F2F3 subdomain is Rap1 independent (Han et al., 2006). This result suggests that a direct interaction between Rap1 and the F0F1 domains facilitates integrin activation. Structural studies showed that Rap1b binds directly to F0 with low affinity (Goult et al., 2010). Furthermore, in Dictyostelium discoideum Rap1 directly interacts with the Ras-associating (RA) domain of talinB (Plak et al., 2016) to enable adhesion during D. discoideum morphogenesis. Similarly, a mutation in F0 that inhibits Rap1 binding is embryonic lethal in Drosophila melanogaster (Camp et al., 2018). Zhu et al. (2017) confirmed the direct Rap1–talin1 F0 interaction and showed that membrane-anchored Rap1b in vesicles has enhanced binding to THD, suggesting a mechanism of talin1 recruitment to integrins by Rap1. Quantitative proteomic analyses of murine platelets revealed the high abundance of Rap1b and talin1 (Zeiler et al., 2014). The abundance of the proteins at equal molar ratios and the lack of a known Rap1 effector with such a high abundance in platelets suggest that F0 may act as a direct effector of Rap1 to activate integrins in platelets. However, we recently reported a talin1 point mutation (R35E) in F0 domain that reduces Rap1 affinity by >25-fold does not impair the capacity of THD to activate integrins in A5 cells nor does it abolish effects of Rap1 activity on THD-induced activation. In accord with this result, we and others (Bromberger et al., 2018) found that mice with F0 mutations that disrupt Rap1 binding are viable, fertile, and apparently healthy and exhibit similar or slightly reduced extent and kinetics of αIIbβ3 activation (Lagarrigue et al., 2018). These mice did manifest a mild defect in platelet aggregation and hemostasis. Thus, the low-affinity Rap1–talin1 F0 interaction does not make a major contribution to integrin αIIbβ3 activation in mice and cannot account for the profound effects on activation in RIAM-deficient platelets of (1) loss of Rap1 activity (Stefanini et al., 2018), (2) deletion of talin1 (Nieswandt et al., 2007; Petrich et al., 2007b), (3) mutations of integrin β3 that block talin1 binding (Petrich et al., 2007a), and (4) mutations of talin1 that prevents binding to integrin β3 (Haling et al., 2011).
Here, we show that F1 contains a previously undiscovered Rap1-binding site of similar affinity to that in F0. A structure-guided mutant in F1, which blocks Rap1 binding, abolishes the capacity of Rap1 to potentiate talin1-induced integrin activation. Talin1 F1’s ability to mediate Rap1-dependent integrin activation requires a unique unstructured loop that transforms into an amphipathic helix upon binding to membrane lipids. Charge-reversal mutations of basic membrane-facing residues of this helix profoundly suppressed talin1-dependent activation. Thus, a novel Rap1 binding site and a transient lipid-dependent helix in F1 work in tandem to enable a direct Rap1–talin1 interaction to promote integrin activation.
Results and discussion
THD contains a second Rap1-binding site
Since integrin activation induced by THD(R35E) was still Rap1 dependent (Lagarrigue et al., 2018), we hypothesized that THD could have a second Rap1-binding site. We used nuclear magnetic resonance (NMR) spectroscopy to examine the interaction of 15N-labeled Rap1b with unlabeled THD. The 2D-HSQC of Rap1b showed dispersed resonances characteristic of a well-folded protein (Fig. 1 A). Addition of THD caused extensive broadening of Rap1b resonances (Fig. 1 B) due to its interaction with the larger THD. We previously showed that the F0 domain of talin1(R35E) does not detectably bind Rap1b (Lagarrigue et al., 2018); however, THD(R35E) caused broadening of the Rap1b resonances (Fig. 1 C), whereas THD(wt) (Fig. 1 B) caused a more profound effect. While R35E mutation reduces the affinity dramatically, residual binding might still be able to cause broadening. We therefore tested THD(ΔF0), comprising talin1 F1–F3 domains, and found that it bound Rap1 as assessed by intermediate broadening (Fig. 1 D), consistent with the reduced mass of THD(ΔF0). The residual Rap1-binding capacity of THD(R35E) and THD(ΔF0) combined with the lack of Rap1 dependence of F2F3-mediated integrin activation (Han et al., 2006) raised the possibility that THD has a second Rap1-binding site in the F1 domain.
To assess the contribution of each subdomain in THD in Rap1-regulated integrin activation, we generated THD deletions by removing the F0, F1, or F0F1 domains (Fig. 1 E) and tested effects on integrin activation in A5 cells. THD induced a similar extent of integrin activation as the addition of manganese (Fig. S1 A), indicating that THD expression provokes substantial integrin αIIbβ3 activation in A5 cells. Expression of all three THD mutants induced activation (Fig. 1 F, gray bars), with a markedly lower activation index for THD(ΔF1) and THD(ΔF0F1), both lacking the F1 domain. Inhibition of endogenous Rap1 activity by coexpression of Rap1GAP1 led to a decrease in both THD and THD(ΔF0)-mediated activation but did not affect activation with THD mutants lacking the F1 domain (Fig. 1 F, blue bars). Furthermore, expression of the constitutively activated Rap1(Q63E) mutant trended to increased THD(ΔF0)-mediated αIIbβ3 activation but had no evident effect on activation with THD mutants lacking the F1 domain (Fig. 1 F, red bars). Thus, the effect of Rap1 activity on integrin activation by THD(ΔF0) resembles that observed with THD(wt) (Lagarrigue et al., 2018). Immunoblotting confirmed similar level of expression for all THD mutants (Fig. 1 G). Because Bromberger et al. (2018) reported a slight reduction in platelet αIIbβ3 activation by introducing the three mutations K15A, R30A, and R35A in the F0 domain, we tested the effect of these mutations on THD-dependent αIIbβ3 integrin activation in A5 cells and observed no effect (Fig. S1, B–E). Taken together, these data suggest that talin1 F1 has a second Rap1-binding site important in integrin activation, a surprising conclusion in light of a previous report that the F1 domain does not bind Rap1b (Zhu et al., 2017).
F1 binds Rap1
Since both F0 and F1 have a ubiquitin fold (Goult et al., 2010), we aligned their amino acid sequence and noticed that the key positively charged residues for Rap1 binding are conserved: K15 and R35 in F0 and R98 and R118 in F1 (Fig. 2 A). Moreover, superimposition of the NMR structures of F0 with F1 showed that the position of these residues is maintained. We therefore generated a model of the talin1 F1–Rap1 complex by superimposing F1 with the F0 domain in complex with Rap1b (Zhu et al., 2017). We found the F1 interface with Rap1b to be very similar to that of F0, with positively charged R98 and R118 showing surface charge complementarity with the negatively charged groups on Rap1 surface (Fig. 2 B). Importantly, using the NMR structure of F1 alone (Goult et al., 2010), we found the inserted F1 loop to be positioned far away from the binding interface where it would not sterically hinder Rap1b binding.
To test the presence of a Rap1-binding site, we prepared 15N-labeled F1 for NMR studies. The 2D-sofast-HMQC (sfHMQC) of F1 showed a stable folded protein, with most resonances being highly dispersed (Fig. 2 C, blue). The presence of sharp intense signals located in the center of the spectra is consistent with a 35-residue loop that is unstructured and highly dynamic. Addition of Rap1b caused residue-specific chemical shift changes in F1 (Fig. 2 C, red), in contrast to the studies of Zhu et al. (2017) using the F1F2 double domain. As predicted, the nondispersed resonances from the loop were not affected by Rap1b. Furthermore, we mapped specific chemical shifts, including Arg98 and Arg118 (Fig. S2 A), that clustered to amino acids located in the predicted RA binding interface near the F1 β2 strand (Fig. S2 B). NMR titrations showed concentration-dependent chemical shift perturbations with fast exchange, that suggested a weak affinity of F1 for Rap1 based on several amino acids located at the binding interface (Fig. 2 D and Fig. S2, E–H, blue lines). As highlighted in Fig. 2 E, the chemical shift perturbations were small and made the determination of the affinity less accurate. After careful analysis, we found that the apparent Kd values for the F1–Rap1 interaction varied from the low-to-high micromolar range (summarized in Fig. S2 C). Analysis of our previous data on F0 also showed similar variations (Fig. S2 D; Goult et al., 2010; Lagarrigue et al., 2018), suggesting that this is an intrinsic property of F0 and F1. Thus, F1 contains an RA domain that can bind Rap1b to a similar extent as F0 (Goult et al., 2010; Zhu et al., 2017; Lagarrigue et al., 2018).
Dramatically reduced affinity of talin1(R118E) F1 domain binding to Rap1
Our molecular modeling suggested that talin1 Arg98 and Arg118 play an important role on the surface charge complementarity with Rap1b (Fig. 2 B), and the NMR data showed that their backbone amide chemical shift is affected by binding. We used the acidic Glu mutation because of the predicted stronger effects of charge repulsions between F1 and the acidic patch on the surface of Rap1b that mediates their interaction. We therefore purified recombinant 15N-labeled F1(R118E) mutant. Comparison of the NMR 2D-sfHMQC of the F1(wt) and F1(R118E) mutant showed only very subtle chemical shift changes (Fig. 2 F); thus, the mutant protein adopts a similar fold to F1(wt). Addition of GMP.PNP Rap1b to F1(R118E) produced no specific chemical shift changes (Fig. 2 G), suggesting that it does not bind to Rap1b. The minimal chemical shift perturbations detected at sevenfold excess of Rap1b with F1(R118E) indicated a substantial reduction in Rap1b affinity (Fig. 2 D and Fig. S2, E–H, red lines) as judged by negligible chemical shift for each previously perturbed residue. Thus, F1(R118E) is well folded and has a greatly reduced affinity for Rap1.
Talin1 F1–Rap1b interaction is important for THD-mediated αIIbβ3 activation in A5 cells
To assess the contribution of the talin1 F0 and F1 interactions with Rap1 on integrin activation, we introduced the mutations R35E and R118E and a double mutant (R35E,R118E) into the THD (Fig. 3 A) and tested effects on integrin activation in A5 cells. Expression of all THDs induced integrin activation, with THD(wt) and THD(R35E) producing comparable levels of activation. In sharp contrast, both constructs with F1 mutations, THD(R118E) and THD(R35E,R118E), exhibited a substantially reduced activation (Fig. 3 B, gray bars). Furthermore, we carefully monitored expression of THD-EGFP and its mutants by flow cytometry and observed a strong effect of the F1 mutants at equivalent levels of THD expression on a per-cell basis (Fig. 3 C). Immunoblotting confirmed similar level of global expression for all THD mutants (Fig. 3 D). More importantly, coexpression of both Rap1(Q63E) and Rap1GAP1 had no effect on integrin activation when the R118E mutation was introduced in the F1 domain (Fig. 3 B, red and blue bars). Also, THD(wt)-induced activation in the presence of Rap1GAP1 was very similar to activation with the THD(R118E) mutant alone, further supporting the conclusion that the F1–Rap1 interaction plays a crucial role in Rap1’s ability to potentiate THD-induced αIIbβ3 activation.
Talin1 F1 loop is important for Rap1 to increase THD-mediated αIIbβ3 activation
Since both the talin1 F0 and F1 domains contain a Rap1-binding site with similar affinity for Rap1b in vitro, and because only the THD(R118E) mutant in F1 had an effect on THD-mediated αIIbβ3 activation, we examined differences between the two domains. First, the F1 domain is located adjacent to the F2 domain, which contains a critical membrane-orientation patch required for activation (Anthis et al., 2009); however, in THD(ΔF1), the F0 is adjacent to F2 yet THD(ΔF1) exhibits little Rap1 dependence (Fig. 1 F). Second, the F1 domain contains an unstructured loop implicated in integrin activation. Goult et al. (2010) showed that this loop has helical propensity and forms a more stable helix on binding to phospholipids. Basic residues, predicted to reside on one surface of the helix, are required for binding to acidic phospholipids (Goult et al., 2010).
To assess the contribution of both the talin1 F1 loop and the interaction with Rap1 on integrin activation, we tested integrin activation by THD wherein the loop was removed, THD(ΔL) (Fig. 4 A). As previously observed (Goult et al., 2010), THD(ΔL) reduced integrin activation (Fig. 4 B, gray bars). Importantly, coexpression of both Rap1(Q63E) and Rap1GAP1 had a negligible effect on integrin activation by THD(ΔL) (Fig. 4 B, red and blue bars). Furthermore, the effect of the double mutant THD(ΔL,R118E) resembled that of THD(ΔL). Immunoblotting confirmed similar level of expression for all THD mutants (Fig. 4 C). 2D-sfHMQC NMR spectra of 15N-labeled talin1 F1(ΔL) exhibited highly dispersed resonances consistent with a stable folded protein (Fig. 4 D, blue). Addition of Rap1b caused residue-specific chemical shift changes in talin1 F1(ΔL) (Fig. 4 D, red) that mapped to amino acids located in the predicted RA–binding interface, including Thr96 and Thr99 (Fig. 4 E). Thus, the talin1 F1 loop is important for Rap1 dependence of THD-mediated integrin activation, but not for Rap1b binding, indicating that the F1 loop works in tandem with Rap1 binding in performing this function.
To further validate the proposed formation of a helix mechanism whereby the talin1 F1 loop plays a role in integrin activation, we introduced three charge-reversal mutations in basic residues on the surface of the predicted helix (R146E,R153E,K156E (3EL)) to create a charge repulsion with the negatively charged phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) of the plasma membrane (Fig. S3 A) that blocks lipid binding (Goult et al., 2010). THD(3EL) was even less active than THD(ΔL) in activating αIIbβ3 (Fig. S3 B). Immunoblotting confirmed similar level of expression for all THD mutants (Fig. S3 C).
Since we used the THD in our experiments, we also wanted to test the effect of the F1 mutants in full-length talin1. To ensure maximal Rap1 activation, we expressed Rap1 (Q63E) in A5 cells and noticed no increase in αIIbβ3 activation, consistent with the modest levels of talin in these cells (Fig. S3 D), as previously reported (Han et al., 2006). Coexpression of talin1 increased activation, as did the talin1(R35E) mutant. Importantly, both talin1(R118E) and talin1(ΔL) reduced integrin activation to a similar extent. Indeed, the effect of the combined talin1(ΔL,R35E,R118E) mutant was similar to that of talin1(R118E) alone, emphasizing the critical role of Rap1 binding F1 and suggesting that the Rap1 dependence of talin1-induced activation depends on the loop in F1, thus explaining why F1 is more important than F0 in this function.
Rap1 binding to THD F0 and F1 facilitates its recruitment to paxillin-containing adhesions
To further evaluate the role of the Rap1–talin1 interaction in integrin functions, we examined the recruitment of THD to adhesion structures in adherent 3T3 fibroblasts. THD localized to paxillin-positive adhesions (Fig. 5 A). A previous report revealed the role of Rap1 binding to the F0 domain in talin1 recruitment to adhesion sites (Zhu et al., 2017). Similarly, we found that colocalization of THD(R35E) mutant with paxillin was reduced (Fig. 5, A and B). Remarkably, THD(R118E) mutant showed an even more impaired recruitment to paxillin-positive adhesions (Fig. 5, A and B), resulting in more diffuse cytoplasmic localization. Thus, Rap1 binding to each Rap1 binding site in THD is important for recruitment to paxillin-containing adhesions.
The studies reported here provide new insight into how talin1 activates integrins, a process central to mammalian development and numerous physiological functions. Studies in A5 cells led to the conclusion that talin1 is required for integrin activation, established the importance of specific structural elements in the integrin and in THD, and identified the importance of lipid binding residues (Tadokoro et al., 2003; Wegener et al., 2007; Anthis et al., 2009; Goult et al., 2010). The A5 cells were also instrumental in studies that showed how Rap1, by engaging RIAM, enabled RIAM to recruit talin1 to integrins (Han et al., 2006; Lee et al., 2009), a key pathway in leukocyte trafficking and formation of the immune synapse (Lagarrigue et al., 2017). Nevertheless, Rap1 can activate integrins in the absence of RIAM or obvious candidate effectors. Goult’s pioneering studies suggested that talin1 F0 could enable talin1 to itself be an effector (Goult et al., 2010), an idea further advanced by Zhu et al. (2017); however, we found that the F0 domain made a minor contribution in mammals. We now find that the F1 domain contains a second Rap1-binding site and that this site functions in conjunction with a unique inserted loop to enable Rap1 to regulate activation. The proximity of the putative membrane-binding helix of the F1 loop to the geranyl-geranyl moiety of Rap1 bound to F1 provides a cogent model to explain the complementary role of these two membrane-binding sites in integrin activation (Fig. 5 C). Indeed, the now more complete picture of how THD interacts with the membrane shows that an extended series of membrane-binding sites (Fig. 5 D) serve to stabilize the weak interaction of talin1 F3 with the integrin β cytoplasmic domain to explain the membrane dependence of talin1-induced activation (Ye et al., 2010). Our studies show how Rap1 can add to those membrane-binding sites to enable activation.
Our studies raise several new questions. In mice, Rap1 binding to talin1 F0 is clearly dispensable for normal development and has only a modest effect on activation of αIIbβ3 and platelet function. In stark contrast, in D. discoideum (Plak et al., 2016) and Drosophila (Camp et al., 2018), blocking Rap1 binding to talin1 F0 has severe consequences for morphogenesis. We note that D. discoideum TalinB F1 has no loop, and in Drosophila talin, the loop sequence is not conserved, suggesting that evolution has endowed mammalian F1 with a loop capable of supporting most Rap1-dependent integrin function, even in the absence of F0’s contribution. Absent this enhanced function of F1, F0’s contribution may become indispensable. This hypothesis will be testable by future reverse genetic studies in which the developmental and physiological effects of Rap1-binding–defective F1 mutations can be assessed alone or in combination with Rap1-binding–defective F0 mutations. The future reverse genetic experiments will also enable studies to assess the importance of Rap1 binding to talin1 in cells, such as leukocytes, that possess sufficient MRL (MIG-10, RIAM, and Lamellipodin) proteins (e.g., RIAM) to provide an alternative mechanism by which a Rap1 effector can engage talin1, leading to its interaction with and activation of integrins. Finally, our findings will enable future studies to characterize the role of each of these Rap1-binding sites in activation in cells in which talin1 and Rap1 are in varying abundance and assess their roles in stability of integrin-mediated adhesions and resulting mechanotransduction.
Materials and methods
Integrin activation in CHO cells
Cells were cultured in DMEM (Corning) supplemented with 10% fetal bovine serum (Sigma-Aldrich) and 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco). The sequences encoding murine THD (aa 1–433) or full-length talin1 were cloned into pEGFP-N1 (Clontech). The sequences encoding human Rap1a(Q63E) and Rap1GAP1 were cloned into p3xFlag7.1(−) (Sigma-Aldrich). Transient transfection was performed using TransIT-LT1 Reagent (Mirus) according to the manufacturer’s protocol. The PAC-1–binding assay in A5 cells was conducted as previously described (Frojmovic et al., 1991). Briefly, cells were harvested by using trypsin 1 d after transfection and washed once in HBSS buffer (containing calcium and magnesium; Gibco) supplemented with 1% (wt/vol) bovine serum albumin (Sigma-Aldrich). PAC-1 IgM (ascites, 1:200) was incubated with cells for 30 min at room temperature before staining with the appropriate Alexa Fluor 647 secondary antibody (Life Technologies) for 30 min on ice. Cells were analyzed by flow cytometry using a FACS Accuri C6 Plus (BD Biosciences) and gated on EGFP-positive events. Integrin activation was defined as αIIbβ3-specific ligand binding corrected for αIIbβ3 expression and was calculated as 100 × ([MFI]i − MFI0)/ΔMFID57, where MFIi is the mean fluorescence intensity (MFI) of bound PAC-1; MFI0 is the MFI of bound PAC-1 in the presence of 10 µM Eptifibatide, and ΔMFID57 is the specific fluorescence intensity of anti-αIIbβ3 D57 antibody. PAC1 (Shattil et al., 1985) and D57 (O’Toole et al., 1994) antibodies have been previously described.
Protein expression and purification
The murine talin1 residues 1–400 (THD), 84–400 (F1–F3), 84–196 (F1), and 84–196 Δ141–170 (F1 Δloop) were cloned into the expression vector pETM-11 (His tagged; EMBL) and expressed in Escherichia coli BL21 Star (DE3) cultured in minimal medium for 15N-labeled samples for NMR or LB for unlabeled protein. Briefly, recombinant His-tagged proteins was purified by nickel-affinity chromatography, the His-tag removed by cleavage with Tobacco Etch Virus protease overnight, and the protein further purified by size exclusion chromatography using a Superdex-75 (26/600) column (GE Healthcare). The column was preequilibrated and run with (NMR buffer) 20 mM sodium phosphate, 50 mM NaCl, 3 mM MgCl2, and 2 mM DTT, pH 6.5. The human Rap1 isoform Rap1b (residues 1–167) cloned into pTAC vector in the E. coli strain CK600K was a generous gift from Alfred Wittinghofer (Max Planck Institute of Molecular Physiology, Dortmund, Germany). Untagged Rap1b was purified by ion exchange, followed by Superdex-75 (26/600) gel filtration as previously described (Gingras et al., 2016). The column was preequilibrated and run with NMR buffer.
NMR samples of all the protein constructs were prepared in NMR buffer containing 0.1 mM GMP.PNP and 5% (vol/vol) of 2H2O. All 2D [1H,15N]-sfHMQC spectra were recorded at 298 K. To test THD binding to Rap1b, 100 µM 15N-labeled GMP.PNP-bound Rap1b was incubated with 250 µM unlabeled THD or THD(ΔF0). Titration curves for the interaction of talin1 F1 with GMP.PNP-bound Rap1b were determined using 100 µM 15N-labeled F1 in NMR buffer. Chemical shift changes (Δδobs (HN,N)) were calculated using CcpNmr Analysis “follow shift changes” function, and analyzed with the one-site binding model to determine the apparent Kd value in Prism 5.0 (GraphPad Software; Vranken et al., 2005).
Western blotting (WB)
Cells were lysed in Laemmli sample buffer. Lysates were subjected to a 4–20% gradient SDS-PAGE. Polyclonal serum directed against EGFP was raised in rabbit (Abgent). Antibody against β-actin (AC-15) was obtained from Sigma-Aldrich. The appropriate IRDye/Alexa Fluor–coupled secondary antibodies were from LI-COR. Nitrocellulose membranes were scanned using an Odyssey CLx infrared imaging system (LI-COR), and blots were processed using Image Studio Lite software (LI-COR).
Flp-In 3T3 cells (Life Technologies) were cultured in DMEM supplemented with 10% fetal bovine serum and 100 U/ml penicillin and 100 µg/ml streptomycin. The sequence encoding mRuby2 fused to the C terminus of Paxillin was cloned into the pMSCV retroviral backbone (Addgene). Transduced mRuby2-positive cells were single-cell cloned by automated cell deposition unit using a BD FACSAria (BD Biosciences). A selected clone was then transfected with plasmids encoding EGFP fused to the C terminus of murine THD (aa 1–433). THD-EGFP WT or mutant sequences downstream of the Tet-On responsive element were cloned into the pcDNA5/FRT plasmid (Life Technologies) expressing the Tet-On transactivator from the mouse phosphoglycerate kinase promoter. Cells were cotransfected with pcDNA5 constructs and the pOG44 Flp recombinase expression vector using TransIT-LT1 (Mirus). Transformants were selected in presence of 200 µg/ml of Hygromycin B (Life Technologies). THD-EGFP expression was induced by adding 100 ng/ml doxycycline (Sigma-Aldrich) overnight. Cells were trypsinized and seeded on top of glass coverslips coated with 10 µg/ml human fibrinogen (Sigma-Aldrich). Cells were fixed 2 h later with 2% formaldehyde (Tousimis) for 20 min at 37°C, permeabilized with Triton X-100, and stained with DAPI and Phalloidin conjugated to Alexa Fluor 647 (Life Technologies). 3D stacks of images were acquired using a Zeiss laser scanning confocal microscope Airy Scan 880, equipped with a 63× (1.46 NA) objective (using a 0.17-μm step interval) and an automated piezo stage. Each image consisted of z-stacks of multiple frames that were first processed from a raw Airyscan image to a final integrated, corrected and deconvolved image that was then flattened as a maximum intensity projection using ZEN software (Zeiss). Only the optical two image slices that were closest to the coverslip were projected in order to analyze localized proteins that are on the basal surface of the cell: peripheral paxillin-rich adhesions. The Images were then further processed in Image Pro Premier 10 (IPP10; Media Cybernetics) in the following manner. Maximum intensity projection images were calibrated and autotraced/masked for paxillin-rich signals using a smart segmentation macro inherent in IPP10 that, based on control images, was able to define all relevant paxillin-rich adhesions in the cells and outline them. These masked outlines were used as regions of interest from which a colocalization module in IPP10 autocalculated and displayed on the image (pseudocolored in white) the amount of colocalization between paxiilin-rich zones (red) and THD-rich zones (green) as seen in last panel of Fig. 5 A. The relevant fluorescent signal range in the colocalization module was 60–256 levels of gray, which was well above background controls. All colocalization values from IPP10 were exported directly into Excel, where they were further processed. The experiment was repeated twice, examining on average 800 paxillin-rich adhesions over an average of 20 cells per category.
GMP.PNP was purchased from Sigma-Aldrich.
Statistical significance was assayed by a two-tailed t test for single comparisons or ANOVA for multiple comparisons with a Bonferroni post hoc test. A P value <0.05 was considered significant. Data distribution was assumed to be normal, but this was not formally tested.
Online supplemental material
Fig. S1 shows A5 cells stably expressing αIIbβ3 integrin transfected with cDNA encoding THD(wt)-EGFP or THD(TM)-EGFP or in the presence of MnCl2. Fig. S2 summarizes the effects of Rap1 binding on the F1 by NMR spectroscopy. Fig. S3 shows the effects of transfecting THD F1 loop mutants and full-length talin1 mutants in A5 cells.
This work was supported by the National Institutes of Health, National Heart, Lung, and Blood Institute (grant HL 139947 to M.H. Ginsberg, grant K01HL 133530-01 to M.A. Lopez-Ramirez, and grant HL 078784 to M.H. Ginsberg and K. Ley), and American Heart Association Career Development Award 18CDA34110228 (to F. Lagarrigue) and Grant-In-Aid 16GRNT29650005 (to A.R. Gingras). The Zeiss LSM880 was supported by National Institutes of Health grant S10OD021831.
The authors declare no competing financial interests.
Author contributions: A.R. Gingras, F. Lagarrigue, and M.H. Ginsberg conceived the study, designed experiments, interpreted data, and wrote the manuscript. A.R. Gingras, F. Lagarrigue, M.N. Cuevas, M. Zorovich, A.J. Valadez, W. McLaughlin, W. Kiosses, and N. Seban performed and analyzed experiments. M.A. Lopez-Ramirez and K. Ley provided vital reagents and critical expertise.
A.R. Gingras and F. Lagarrigue contributed equally to this paper.