Gα13 Switch Region 2 Relieves Talin Autoinhibition to Activate αIIbβ3 Integrin*

Integrins function as bi-directional signaling transducers that regulate cell-cell and cell-matrix signals across the membrane. A key modulator of integrin activation is talin, a large cytoskeletal protein that exists in an autoinhibited state in quiescent cells. Talin is a large 235-kDa protein composed of an N-terminal 45-kDa FERM (4.1, ezrin-, radixin-, and moesin-related protein) domain, also known as the talin head domain, and a series of helical bundles known as the rod domain. The talin head domain consists of four distinct lobes designated as F0–F3. Integrin binding and activation are mediated through the F3 region, a critically regulated domain in talin. Regulation of the F3 lobe is accomplished through autoinhibition via anti-parallel dimerization. In the anti-parallel dimerization model, the rod domain region of one talin molecule binds to the F3 lobe on an adjacent talin molecule, thus achieving the state of autoinhibition. Platelet functionality requires integrin activation for adherence and thrombus formation, and thus regulation of talin presents a critical node where pharmacological intervention is possible. A major mechanism of integrin activation in platelets is through heterotrimeric G protein signaling regulating hemostasis and thrombosis. Here, we provide evidence that switch region 2 (SR2) of the ubiquitously expressed G protein (Gα13) directly interacts with talin, relieves its state of autoinhibition, and triggers integrin activation. Biochemical analysis of Gα13 shows SR2 binds directly to the F3 lobe of talin's head domain and competes with the rod domain for binding. Intramolecular FRET analysis shows Gα13 can relieve autoinhibition in a cellular milieu. Finally, a myristoylated SR2 peptide shows demonstrable decrease in thrombosis in vivo. Altogether, we present a mechanistic basis for the regulation of talin through Gα13.


Integrins function as bi-directional signaling transducers that regulate cell-cell and cell-matrix signals across the membrane. A key modulator of integrin activation is talin, a large cytoskeletal protein that exists in an autoinhibited state in quiescent cells. Talin is a large 235-kDa protein composed of an N-terminal 45-kDa FERM (4.1, ezrin-, radixin-, and moesin-related protein) domain, also known as the talin head domain, and a series of helical bundles known as the rod domain. The talin head domain consists of four distinct lobes designated as F0 -F3.
Integrin binding and activation are mediated through the F3 region, a critically regulated domain in talin. Regulation of the F3 lobe is accomplished through autoinhibition via anti-parallel dimerization. In the anti-parallel dimerization model, the rod domain region of one talin molecule binds to the F3 lobe on an adjacent talin molecule, thus achieving the state of autoinhibition. Platelet functionality requires integrin activation for adherence and thrombus formation, and thus regulation of talin presents a critical node where pharmacological intervention is possible. A major mechanism of integrin activation in platelets is through heterotrimeric G protein signaling regulating hemostasis and thrombosis. Here, we provide evidence that switch region 2 (SR2) of the ubiquitously expressed G protein (G␣13) directly interacts with talin, relieves its state of autoinhibition, and triggers integrin activation. Biochemical analysis of G␣ 13 shows SR2 binds directly to the F3 lobe of talin's head domain and competes with the rod domain for binding. Intramolecular FRET analysis shows G␣ 13 can relieve autoinhibition in a cellular milieu. Finally, a myristoylated SR2 peptide shows demonstrable decrease in thrombosis in vivo. Altogether, we present a mechanistic basis for the regulation of talin through G␣ 13 .
Integrins are transmembrane receptors that regulate dynamic cell/cell and cell/matrix interactions (1,2). Environmental cues regulate a multitude of cellular processes through integrins, such as cell proliferation, shape, adhesion, and migration. Because of the well established importance of integrin signaling in hemostasis and thrombosis, the mechanisms of integrin regulation remain an area of intense investigation. Upon activation, platelet integrins switch from a low affinity to a high affinity state, thus permitting the binding to multivalent ligands such as von Willebrand factor and fibrinogen to enable clotting (3)(4)(5)(6)(7). The most abundant platelet integrin complex, ␣IIb␤3, is activated through the cytoskeletal protein talin (1, 8 -12). Talin is a large 235-kDa protein composed of an N-terminal 45-kDa FERM 2 domain, also known as the talin head domain (THD). THD is connected to a large rod domain composed of a series of ␣-helical bundles. The talin head domain consists of four distinct lobes or regions designated as F0 -F3. Integrin binding and activation are mediated through the F3 region, a critically regulated domain in talin. Regulation of the F3 lobe is accomplished through autoinhibition via anti-parallel dimerization, although some evidence suggests an autoinhibited monomer conformation (13)(14)(15)(16)(17). In the anti-parallel dimerization model, the rod domain region of one talin molecule binds to the positively charged face of the F3 lobe on an adjacent talin molecule, thus achieving the state of autoinhibition (15).
The primary mechanism for disabling talin autoinhibition is thought to occur through localized distribution of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) at the plasma membrane surface (11,18,19). This process is mediated by PIP kinase I␥ via engaging the Cal-DAG/Rap1a/RIAM pathway. PIP 2 can bind to several sites within the talin head/FERM domain and is believed to displace the strong negative interface of the rod domain by attracting the positive charges of the head domain in a so-called "push-pull" mechanism (20). The RIAM (Rap1-GTP-interacting adapter molecule) effector, however, has been shown to recruit talin to the membrane as well as to specifically bind to the F3 lobe of the FERM domain in an area that relieves rod binding and to promote ␤3 binding and integrin activation (13). Recently, however, RIAM has been shown to be dispensable for integrin activation in platelets, suggesting the existence of a separate activator that disrupts talin autoinhibition (21,22). In this study, we present a model whereby G␣ 13 -mediated activation of talin, through the relief of autoinhibition, regulates integrin signaling.
A major mechanism of integrin activation in platelets is through heterotrimeric G protein signaling. In this pathway, platelet agonists bind directly to a variety of GPCRs, triggering rapid recruitment and activation of circulating platelets to the site of vascular injury. Heterotrimeric G proteins consist of a complex of ␣, ␤, and ␥ subunits bound to their cognate GPCRs. Subsequent to GPCR activation, the three conformationally sensitive switch regions of the G protein ␣ subunit (SR1-3) permit the exchange of GDP for GTP and dissociation from the ␤ and ␥ subunits (23,24). Both the ␣ and ␤␥ subunits can interact with downstream molecules resulting in a signaling cascade. Although the switch regions are necessary for G protein activation, it is known that these regions also have a secondary function, which enables their specific recognition of the downstream effectors (25)(26)(27). Previously, the G protein subunit G␣ 13 has been shown to engage its SR1 to directly bind to p115RhoGEF thus increasing RhoA activity leading to cytoskeletal rearrangements (25). Recently, our group reported that G␣ 13 SR2 (where SR2 is one of three conformationally sensitive switch regions present in G␣ subunits) directly binds to talin and thereby modulates platelet integrin ␣IIb␤3 activation (27). This finding provided the first evidence for a G␣ 13 /talin interaction and thus represented a novel regulatory pathway. In this study, we propose a mechanistic basis for the ability of G␣ 13 SR2 to activate talin, i.e. by relieving talin autoinhibition. This new mechanism for talin activation by G␣ 13 not only has important implications regarding platelet integrin activation and in vivo thrombosis, but also underscores the critical importance of G␣ 13 in numerous developmental and pathological states because G␣ 13 and talin are ubiquitously expressed.
Biotinylated fragments of THD were examined by Western blotting (Fig. 1C, left panel). For ease of interpretation, the smallest band at ϳ5 kDa containing b-G␣ 13 SR2 Pep and CNBr digested THD was analyzed (Fig. 1C, arrow #1). Based on the predicted molecular mass, this band could originate from 5 of 15 potential methionine-cleaved peptides present in the THD (Table 1). THD peptides that contain a lysine residue could potentially cross-link to b-G␣ 13 SR2 Pep via BS 3 primary amine cross-linking agent and are thus viable candidates for binding. The predicted molecular mass of b-G␣ 13 SR2 Pep is ϳ2.3 kDa, and therefore the sum of the THD CNBr product molecular mass and the 2.3-kDa b-G␣ 13 SR2 Pep should be ϳ5 kDa. On this basis, all candidate peptides containing lysine residues and molecular masses greater than 2 kDa but less than 4 kDa were considered potential binding sites (Table 1). It can be seen that only THD fragments 2, 6, 11, 13, and 14 fit these criteria for potential binding. To supplement CNBr mapping, an additional thrombin digestion of cross-linked THD and b-G␣ 13 SR2 pep was performed. Thrombin cleaves THD at amino acid 328 (see Thrombin Mapping in Table 2), resulting in two fragments of ϳ38 and ϳ7.6 kDa. However, data shown in Fig. 1C revealed full size THD as well as a band of an ϳ38-kDa but not a 7.6-kDa band, suggesting that SR2 does not bind to the C-terminal side of the F3 lobe (Table 2).
To further screen the binding of b-G␣ 13 SR2 Pep to THD, a dot blot analysis was performed using b-G␣ 13 SR2 Pep and talin head domain constructs lacking the F2 (⌬F2), F3 (⌬F3), and F2-3 (⌬F2-3) regions. Data shown in Fig. 1D, along with the loading controls ( Fig. 1E), illustrate that b-G␣ 13 SR2 Pep requires a sequence contained within the F2-3 lobes for binding to THD (left panel). Furthermore, separate dot blots (Fig. 1E, right panel) demonstrated that F3 is indeed required for SR2 binding. Based on these findings, the CNBr fragments 2, 6, and 11 were ruled out because they do not incorporate F3 (Table 1). Further dot blot analysis also eliminated a portion of F2 and F3 lobes that is represented by the CNBr 13 fragment. Taken together, these data indicate that the candidate amino acid residues reside within the 310 -328 amino acid region of THD.
This putative binding region within THD was further validated by ELISA measurements by quantifying b-G␣ 13 SR2 Pep binding to recombinant THD carrying F3 mutations at residues Met-319 and Leu-325. These residues have been examined by others and shown to play a role in maintaining talin autoinhibition as well as talin-mediated clot retraction (10,15,28). As a negative control, amino acid Glu-335 that does not fall within the predicted 310 -328 binding region for b-G␣ 13 SR2 Pep was also mutated. In these experiments, b-G␣ 13 SR2 Pep was used to coat the ELISA plate, and binding to THD constructs (analyte) was measured using a head domain-specific antibody. The results shown in Fig. 1F demonstrate that the talin residues 319 and 325, but not 335, are important for G␣ 13 SR2 binding THD.
Finally, the putative SR2 binding region within THD was confirmed using the mass spectrometry-based mapping of the cross-linked THD complex with biotin-G␣ 13 SR2 Pep . The unique peptides obtained after mass spectrometry with the highest fidelity contained amino acids 307-338. This sequence encompasses completely with the 310 -328 region identified using the collective data derived from the CNBr peptide mapping, thrombin peptide mapping, deletion mutagenesis, and point mutagenesis experiments (Table 2). Taken together, these results identify a specific G␣ 13 SR2-binding site within the THD domain. Importantly, this binding site includes the critical Met-319 and Leu-325 residues that are present on a flexible region in the F3 lobe that directly engages the rod domain and promotes talin autoinhibition (Fig. 1G) (15, 16, 28 -30). Guided by these findings, we elected to examine the mechanism by which G␣ 13 modulates talin autoinhibition. G␣ 13 SR2 Alters Talin Conformation-A schematic diagram illustrates critical regions within talin's full-length structure that participate in talin's dynamic state of autoinhibition ( Fig.  2A). Specifically, when talin F3 and R9 ( Fig. 2A, light red) bind to FIGURE 1. G␣ 13 SR2 binds talin head domain F3. A, schematic depiction of THD cross-linking to biotinylated G␣ 13 SR2 Pep and subsequent fragmenting using either chemical or enzymatic digestion. Biotinylation detection allowed peptide mapping to elucidate a novel THD-binding site. B, Coomassie Blue stain of recombinant THD (left panel) and streptavidin-HRP detection of cross-linked THD of either G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep . C, cyanogen bromide (CNBr) cleavage of THD cross-linked to G␣ 13 SR2 Pep (left panel) and thrombin digestion (right panel). The black arrow on the left panel highlights the smallest detectable fragment for further analysis. The black arrows on the right panel show biotinylated peptide is present on the large THD thrombin fragment but not a smaller fragment. D, dot blot with ligands THD or THD lacking domains ⌬F2, ⌬F3, or both ⌬F2-3. Biotinylated G␣ 13 SR2 pep analyte was bound differentially to each ligand, as detected by streptavidin-HRP. E, loading controls for each dot blot ligand. F, ELISA of biotinylated G␣ 13 SR2 Pep ligand bound to THD WT or point mutants. G, RCSB PDB structure 2KGX depicting THD F3 (green) binding to R9 (yellow). Pink color was added to highlight binding site of SR2 elucidated through peptide mapping and mass spectrometry. Blue residues (aa 319 and 325) are point mutants that are necessary to bind G␣ 13 SR2 Pep , although the red residue (335) is not. Statistical significance was determined using a Student's unpaired t test where *, p Ͻ 0.05; AU, arbitrary units.

TABLE 1 Chemical cleavage of talin
CNBr cleavage yielded 15 unique peptide fragments from THD. Table contains the predicted sequence, molecular mass, cross-linked molecular mass, lysine residues. The presence of THD segment within each peptide sequence is indicated. Five of the 15 peptides match the criteria for potential binding sites of SR2 to THD.

Mapping of G␣ 13 SR2 and talin-binding interface
Data provide a list of multiple techniques used to map the SR2/THD binding interface and their respective locations. The final row (green) contains the overlapping region revealed by each technique. one another, the crucial integrin-binding site present within the F3 lobe is obscured, thereby creating the autoinhibited conformation (5,8,12,30,31). Importantly, the G␣ 13 SR2-talin-binding site (see Fig. 1G) overlaps with the talin autoinhibitory binding site created by the interaction of F3 and R9. This juxtaposition of the G␣ 13 SR2 and F3-R9 binding regions suggests G␣ 13 may play a role in relieving autoinhibition by inducing a conformational change in talin.
To examine the possibility that G␣ 13 can relieve talin autoinhibition, a competitive binding assay was performed. GST fusion proteins containing the SR2 sequence (GST-G␣ 13 SR2) or scrambled sequence (GST-G␣ 13 SR2 Random ) were incubated with recombinant THD. Recombinant talin rod domain con-taining the autoinhibitory region (R9 -12) or a non-inhibitory protein (BSA) was added to their respective samples. If R9 -12 and G␣ 13 SR2 compete for an overlapping binding region on THD, then a decrease in THD binding to GST-G␣ 13 SR2 should be observed (Fig. 2B). Alternatively, a decrease in THD binding to GST-G␣ 13 SR2 would also be observed if the binding to GST-G␣ 13 SR2 to THD caused a conformational change in THD such that it could no longer bind R9 -12. However, if R9 -12 and G␣ 13 SR2 do not share a binding site on THD, or if G␣ 13 SR2 does not allosterically interfere with THD-R9 -12 binding, then a ternary complex would be expected to form between R9 -12 and THD. The data shown in Fig. 2C (lane 1) demonstrate that GST-G␣ 13 SR2 binds directly to THD, although GST- helical domains, and 1 dimerization domain at the C terminus. Domains F3 and R9 bind to one another to create an autoinhibited state and are shown in light red. R13 cannot bind talin's FERM domain and is shown in teal. Linker regions between R7-8, R8 -9, and R9 -10 are highlighted in red as reference for future FRET experiments. B, schematic depiction of a competitive binding assay between GST-G␣ 13 SR2 and R9 -12 for THD F3 binding. GST-G␣ 13 SR2 binds to THD, and upon the addition of R9 -12, the possible binding scenarios are the formation of a ternary complex (non-overlapping binding site), no R9 -12 binding due to allosteric changes to talin's conformation, or competitive binding if R9 -12 and SR2 share an overlapping binding site on THD. C, binding assay demonstrating GST-G␣ 13 SR2 can bind directly to THD but not GST-G␣ 13 SR2 Random (lanes 1 and 2), but is sterically occluded when increasing amounts of recombinant R9 -12 were added at either 50, 25, or 5 M (lanes [3][4][5]. The addition of equal concentrations of BSA does not sterically occlude GST/G␣ 13 SR2/THD interaction (lanes 6 -8) demonstrating specificity of R9 -12 binding. Finally, no concomitant GST-G␣ 13 SR2/R9 -12 complex was formed demonstrating specific competition. D, Western blotting analysis showing the effect of limited proteolysis on full-length talin after G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep or PIP 2 combination treatment. 1st to 4th lanes show 87.5 M G␣ 13 SR2 Pep (green) or G␣ 13 SR2 RandomPep (black) were incubated with full-length talin and either 8 or 0.32 ng of the protease Endo-Asp-N (red). 5th to 8th lanes show the same experiment with the addition of 45 M PIP 2 . 9th lane is full-length talin input. Loss of talin signal is attributed to proteolysis resulting from conformational change. E, GST pulldown assay where GST-THD or GST alone are incubated with G␣ 13  G␣ 13 SR2 Random did not (lane 2). Recombinant R9 -12 was added at either 50, 25, or 5 M as shown in lanes 3-5 (Fig. 2C), respectively. A dose-dependent decrease in the THD binding to GST-G␣ 13 SR2 was observed without any detectable R9 -12 binding to GST-G␣ 13 SR2. These results demonstrate G␣ 13 SR2 significantly reduces the ability of R9 -12 to bind to THD, suggesting that G␣ 13 SR2 can interfere with the autoinhibited conformation of talin.
Although these data suggest G␣ 13 may activate talin, few experimental options are available that can test the effect of ligands on the full-length talin molecule. To investigate this model further, we devised an assay to evaluate the conformational state of full-length talin isolated from human platelets. To examine whether G␣ 13 binding triggers conformational changes in full-length talin, a limited proteolysis protection assay was performed. The limited proteolysis approach utilizes a small amount of protease that has little detectable effect on structurally stable proteins. The dimer fraction of platelet talin is reported to exhibit a dumbbell dimer conformation in an autoinhibited state (14,16); thus, if G␣ 13 SR2 perturbs this state, talin may become more vulnerable to limited proteolytic degradation. To validate this model, the dimeric fraction of platelet talin was purified and then pre-treated with G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep. Next, the complex was exposed to decreasing amounts of the protease Endo-AspN, and proteolysis was analyzed by Western blotting (Fig. 2D). The results show that with increasing concentrations of protease, there is an increase in the level of talin degradation when treated with G␣ 13 SR2 Pep but not with G␣ 13 SR2 RandomPep (Fig. 2D, left panel). This observation suggests that talin is more sensitized to proteolysis after treatment with the G␣ 13 SR2 Pep likely due to a change in its conformation.
To further corroborate these findings, we examined a known activator of talin, the phospholipid PIP 2 . PIP 2 accumulation in the plasma membrane adjacent to the integrin complex is thought to repel the negatively charged autoinhibitory rod domain and simultaneously pull the positively charged face of talin's head domain, in a push-pull mechanism to facilitate talin activation (20). As expected, the addition of monomeric PIP 2 at a concentration below the critical micelle concentration was able to increase G␣ 13 SR2-facilitated talin degradation (Fig. 2D, middle panel). The addition of PIP 2 and G␣ 13 SR2 Pep had a greater effect on talin degradation than G␣ 13 SR2 Pep alone or G␣ 13 SR2 RandomPep and PIP 2 (Fig. 2D). Because G␣ 13 SR2 RandomPep had no significant effect on talin degradation, all observed proteolysis (Fig. 2D, lanes 7 and 8) is presumably attributable to PIP 2 . These data are consistent with the notion that G␣ 13 SR2 triggers a conformational change in talin, which is significantly more efficient when added in combination with PIP 2 . This observation suggests a possible cooperative effect whereby the combination of G␣ 13 and PIP 2 may enhance integrin activation.
Direct Binding of Full-length G␣ 13 to Talin-Although G␣ 13 SR2 Pep has demonstrable binding to THD, full-length G␣ 13 binding to both THD has yet to be examined outside of the cellular milieu. To further characterize G␣ 13 interaction with talin, recombinant full-length G␣ 13 purified from Sf9 insect cells was used, as described previously (32). G␣ 13 pre-treated with GDP (inactive) bound significantly less to GST-THD than GTP␥S-treated G␣ 13 (active), consistent with previous findings (Fig. 2E) (25). In both cases, no significant binding to GST was observed. An ELISA was used to estimate the affinity between G␣ 13 and THD, yielding a dissociation constant (K d ) of 13.4 M. However, the physiological relevance of this affinity will require further characterization of their respective interactions in a milieu more closely resembling the plasma membrane microenvironment.
To further characterize the competition between G␣ 13 and talin domains, we used a GST-THD fusion construct as bait, and we examined its interaction with full-length G␣ 13 in the presence of talin R9 -12 module (see model in Fig. 2F). To obtain a sufficient amount of purified G␣ 13 required for this assay, we expressed a chimeric G␣ 13 subunit containing the N-terminal domain of G␣ i1 (see under "Materials and Methods"), as described previously (33). G␣ 13 in an active GTPbound state interacts with GST-THD (Fig. 2G, lane 1). We then measured G␣ 13 binding to THD in the presence of the talin R9 -12 module. G␣ 13 binding to GST-THD was measured in competition with talin R9 -12 at increasing concentrations of 5, 10, 20, 30, and 50 M talin R9 -12 ( Fig. 2G, lanes 2-6). Moreover, biochemical interaction of inactive G␣ 13 (30 M) bound to GDP was measured with GST-THD in the presence of 30 M talin R9 -12 (Fig. 2G, lane 7). These results demonstrate that as the binding of talin R9 -12 increases, the binding of G␣ 13 to THD decreases. In contrast, inactive GDP-bound G␣ 13 has little effect on talin R9 -12 binding to THD (Fig. 2G, lane 7). These data suggest that G␣ 13 binding to THD occurs at the expense of talin rod binding in a GTP-dependent manner. Taken together, these results (Figs. 1 and 2) demonstrate a GTP-dependent binding of G␣ 13 SR2 to THD, which competes with the talin rod domain regions responsible for maintaining an autoinhibited state.
Endogenous G␣ 13 Changes Talin Conformation in a Cellular Milieu-Although talin activity is often measured indirectly through integrin activation, few options exist that can examine its conformational state directly. Because of the importance of talin in the regulation of integrin activation, we sought to develop a technique that could directly assess gross changes in talin's conformational state. Although it is unclear whether GFP-talin exists as an autoinhibited dimer in CHOA5 cells (where CHOA5 is Chinese hamster ovary cells constitutively expressing the platelet integrin ␣IIb␤3), it has been shown that GFP-talin causes a low basal level of integrin activation as compared with the GFP-talin head domain (15). Furthermore, a single amino acid mutation of talin's head domain, M319A, increases integrin activation compared with both wild type fulllength GFP-talin and GFP-THD (15,31). Additionally, it was shown that the co-transfection of GFP-talin with G␣ 13 , but not SR2 mutants R227A or R232A, increased integrin activation (27). These observations suggest that GFP-talin is partially, if not completely, autoinhibited when overexpressed in this model cell system. To characterize the autoinhibited state of talin in the cellular milieu, we devised a FRET-based assay to assess the effect of G␣ 13 on integrin activation. Detailed technical aspects of this experimental approach are discussed under "Materials and Methods." Briefly, CCPGCC amino acids were engineered into three separate GFP-talin constructs at amino acids 1457, 1655, and 1822 (Fig. 3A). In an autoinhibited state, GFP can act as a donor molecule to a cell-permeant fluorescent dye known as ReAsH that binds CCPGCC (Fig. 3B). The FRET signal should significantly diminish in an activated/elongated state (Fig. 3B). The specificity of ReAsH binding was evaluated by immunofluorescence microscopy. Cells transfected with GFP-talin without a CCPGCC motif showed no significant red fluorescence background signal (Fig. 3C, left panels) as compared with GFP-talin with a CCPGCC motif at amino acid 1822 (GFP-talin-1822) (Fig. 3C, right panels). These observations demonstrated successful engineering of the CCPGCC motif in talin and highly specific binding of the dye under these conditions.
Next, we elected to measure talin's conformational state using flow cytometry. This technique was employed due to its sensitivity to even weak fluorescence signals while simultaneously examining large populations of cells to increase the likelihood of detecting FRET-positive cells. Flow cytometry utilized only the 488-nm laser to eliminate potential inappropriate activation of ReAsH. The 488-nm laser is sufficient to excite GFP, with only a 2% spectral overlap with the excitation spectrum of ReAsH. The gating of CHOA5 cells, shown in Fig. 3D, indicates the distinction whereby R1 represents the general population of CHOA5 cells, and R2 and R3 show FITC-or PE-positive cells, respectively. To confirm that no ReAsH signal was detected in the absence of FRET, a tubulin construct that contains the CCPGCC motif, and can therefore bind to the red fluorescence dye, was utilized. ReAsH emission was expected to emit at ϳ615 nm, which was detected using a Texas red filter (PE). Because only the 488-nm laser was utilized, no red fluorescence signal should occur under these conditions. The  Fig. 2A). B, schematic depicting GFP-talin in an autoinhibited state that permits GFP/ReAsH (red) FRET pairing. Activation relieves autoinhibition and decreases ReAsH FRET signal (gray). C, GFP-talin containing CCPGCC insert at amino acid 1822 (GFP-talin-1822; right panels) or GFP-talin (left panels). Scale bars, 20 m. D, flow cytometry-based detection of FRET signal. CHOA5 cell populations (R1 gate) were evaluated for GFP signal (R2 gate) or ReAsH signal (R3 gate) using the FITC or PE channel, respectively, using the 488-nm laser for specific excitation of GFP only. E, tubulin-ReAsH construct (1st panel) demonstrates that no PE signal emitting from ReAsH is visible with the 488-nm laser alone; therefore, any signal observed in the PE channel results from GFP-ReAsH fret pairing. GFP-talin (2nd panel) shows no PE signal. 3rd to 5th panels show an increasing level of ReAsH emission in the PE channel, suggesting that by moving the location of the CCPGCC motif toward the C terminus there is an increasing FRET pairing. F, quantification of FRET signal observed. G, change in FRET signal from GFP-talin-1822 or F3 point mutations M319A or L325K. H, percent of FRET ϩ cells with GFP-talin-1822 co-expressed with vehicle or with RIAM. I, percent of FRET ϩ cells with GFP-talin-1822 co-expressed with vehicle, G␣ 13 , or G␣ 13 R227A. Statistical significance was determined using a Student's unpaired t test where *, p Ͻ 0.05. results showed no detectable PE signal in cells expressing tubulin containing a CCPGCC motif (Fig. 3E, panel #1). Thus, any signal detected with the GFP-talin constructs was attributed to FRET pairing under these conditions.
To account for any nonspecific ReAsH signal, a second control, GFP-talin (without a CCPGCC motif), pre-incubated with ReAsH dye was used to gate all GFP ϩ /FRET Ϫ cells (Fig. 3E,  panel #2). GFP-talin signal (R2) and any signal that appeared on the PE axis (R3) were considered emitting from FRET pairing. Next, each GFP-talin construct containing a CCPGCC motif was tested for FRET signal. Data shown in Figs. 3E (panels #3-5) and F clearly demonstrate a trend of increasing FRET signal as the CCPGCC motif is incorporated toward the C terminus of talin. The largest FRET signal was detected when the CCPGCC motif was inserted at amino acid 1822 of talin. This sequence is located immediately after the autoinhibitory R9 domain ( Fig. 2A), thus supporting the hypothesis that talin is in an autoinhibited state permitting FRET pairing under these conditions. To further investigate the specificity of this interaction, GFP-talin-1822 was mutated at residues M319A and L325K. The M319A mutation has been shown to increase integrin activation nearly 3-fold as compared with native talin sequence, suggesting that this amino acid is critical for maintaining autoinhibition, and when mutated it permits hyperactivation of talin. The results shown in Fig. 3G demonstrate a 30% decrease in FRET signal in the GFP-talin-1822-M319A construct, although GFP-talin-1822-L325K showed no significant decrease in the FRET signal. Collectively, these data indicate that GFP-talin undergoes FRET pairing when autoinhibited and can be specifically disrupted by mutating a single amino acid that is critical at the F3/R9 interface.
Next, we investigated talin activation through protein/protein interactions rather than relying only on point mutations. This approach was accomplished by co-transfecting RIAM, a known activator of talin, with GFP-talin-1822. RIAM binding to the F3 lobe of the talin FERM domain is sufficient to relieve its autoinhibition, and therefore RIAM was used to validate the assay (13). Co-transfection of RIAM resulted in a 50% reduction of FRET signal (Fig. 3H), suggesting a conformational change in talin. Previous studies have shown an increase in integrin activation when RIAM and talin are co-expressed (13,34,35), which is consistent with the observed decrease in the FRET signal.
To examine the effect of G␣ 13 on talin, GFP-talin-1822 was co-transfected with either wild type G␣ 13 or G␣ 13 SR2 mutant R227A (Fig. 3I). Similar to RIAM, wild type G␣ 13 showed a 50% decrease in the FRET signal suggesting a conformational change in talin that suppressed FRET pairing. The specificity of G␣ 13 /talin interaction was further validated through the G␣ 13 SR2 R227A mutation. In this case, the FRET signal was slightly reduced but was not significantly different from the GFP-talin-1822 control (Fig. 3I). Together, these data suggest the following: 1) GFP-talin exists in an autoinhibited state in the CHOA5 cell system; 2) FRET pairing is attained when the red fluorescence acceptor molecule is placed close to the autoinhibitory R9 region; 3) the conformation of talin is altered by single amino acid changes that are known to relieve autoinhibition; and 4) co-expression of talin-activating molecules such as RIAM and G␣ 13 trigger a decrease in FRET signal suggesting a critical role of these molecules in talin regulation. The FRETbased methodology, as outlined here, offers a novel experimental tool for detecting gross GFP-talin conformational changes in a cellular milieu and can be used to screen potential activators of talin. In summary, we conclude that G␣ 13 plays a significant role in regulating the dynamic conformation of talin.
Endogenous G␣ 13 Facilitates Talin Activation in CHOA5 Cells-Our results thus far suggest that G␣ 13 plays a functional role in relieving talin autoinhibition. Next, we assessed the role of G␣ 13 and GFP-talin interaction in cell morphology. Patterned on the FRET assay, the CHOA5 cells were used as the model system that most closely approximates integrin activation in platelets. CHOA5 cells were transfected with G␣ 13 , G␣ 13 R227A, GFP-talin, GFP-THD, or a combination of G␣ 13 and GFP-talin. After 24 h, the cells were plated on fibrinogencoated slides for 15 min and imaged using immunofluorescence microscopy. Data in Fig. 4A (top three panels) show that transfection of G␣ 13 ϩ GFP, G␣ 13 (R227A) ϩ GFP, and GFPtalin yielded no significant difference in cell spreading. Upon co-expression of G␣ 13 ϩ GFP-talin, a significant increase in cell spreading occurred, which was reduced upon co-expression of G␣ 13 (R227A) ϩ GFP-talin. These results show a significant increase in spreading only when G␣ 13 and GFP-talin are expressed together and that this cell spreading is SR2-dependent. Next, we examined the effect of GFP-talin-Met-319, which earlier showed a decreased FRET signal indicating an active state. As expected, this point mutation alone, talin/Met-319, increased cell spreading similar to the level observed with G␣ 13 ϩ GFP-talin. To investigate whether G␣ 13 could enhance this phenotype, G␣ 13 and GFP-talin-M319A were co-transfected. No increased spreading was observed indicative of talin conformation that is already constitutively active. Finally, we examined the effect of GFP-THD on cell spreading. An increase in cell spreading was observed that is similar to that of G␣ 13 ϩ GFP-talin as well as GFP-talin-M319A. Similar to the effect of GFP-talin-M319A, co-transfection of G␣ 13 showed no significant impact on cell spreading consistent with the constitutively active state of GFP-THD. Altogether, these data demonstrate the following: 1) G␣ 13 -mediated activation of talin leads to increased cell spreading, although the G␣ 13 SR2 mutant cannot, and 2) talin M319A and THD are constitutively active for cell spreading and do not benefit from the co-expression of G␣ 13 .
To examine the specific localization of the G␣ 13 -talin-integrin complex, immunofluorescence analysis was performed. G␣ 13 or G␣ 13 R227A were transfected with GFP-talin into CHOA5 cells (Fig. 4, C and D). Subsequently, co-localization of ␤3 integrin and GFP-talin (Fig. 4C) or G␣ 13 and GFP-talin (Fig.  4D) was examined. The ␤3 integrin (red) and GFP-talin (green) co-localized (yellow) in punctate foci throughout the cell when wild type G␣ 13 was expressed (Fig. 4C, top panels). However, upon expression of G␣ 13 R227A (Fig. 4C, middle panels), ␤3 integrin did not co-localize with GFP-talin as evident by the diffuse nonspecific overlapping signal. An enlarged view of the data shown in Fig. 4C (panels A and B) is highlighted in the bottom left and right panels, respectively. These data indicate the functional importance of G␣ 13 SR2 in determining the intracellular localization of the talin-␤3 integrin complex. It can also be seen that GFP-talin (green) and G␣ 13 (red) co-localized (yellow) when G␣ 13 was co-transfected (Fig. 4D, top panels). When G␣ 13 R227A was co-transfected (middle panels), there was significantly less overlap between G␣ 13 R227A and GFPtalin (Fig. 4D), indicating a decreased co-localization. An enlarged view of the data shown in Fig. 4D (panels A and B) is highlighted in the bottom left and right panels, respectively. Collectively, these data demonstrate the following: 1) G␣ 13 colocalizes with talin directly; 2) G␣ 13 SR2 mutant R227A prevents talin/G␣ 13 co-localization; and 3) G␣ 13 promotes talin localization to integrin ␣IIb␤3. The striking cooperation between G␣ 13 and talin led us to evaluate several ex vivo and in vivo models of talin-integrin function to juxtapose in vitro data in the context of physiological significance. G␣ 13 SR2 Modulates ex Vivo Platelet Adhesion, Clot Formation, and in Vivo Thrombosis-Platelets serve as a useful model system for assessing the talin activity because aggregation, integrin activation, and clot retraction rely on activated talin molecules. Our previous work demonstrated that myristoylated SR2 peptide (G␣ 13 SR2 Pep ) inhibits platelet aggregation in response to all agonists tested (27). This global inhibition of platelet aggregation was attributed to inhibition of ␣IIb␤3 integrin activation (27). Because the biochemical evidence indicates that talin activation can be inhibited by G␣ 13 SR2, talindependent physiological processes are expected to be affected by G␣ 13 SR2 inhibition as well. To further characterize the effects of G␣ 13 SR2 on platelet physiology, in vitro adhesion and thrombus formation induced by collagen were assessed. Platelets in whole human blood were pre-treated with G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep , perfused over collagen, and analyzed for platelet thrombus formation. A significant reduction in the size and number of platelet thrombi was observed in the G␣ 13 SR2 Pep -treated blood relative to the G␣ 13 SR2 RandomPeptreated blood under physiological conditions of shear stress (Fig. 5A). These results are consistent with our previous data showing platelet aggregation induced by collagen-related peptide is inhibited by G␣ 13 SR2 Pep (27).
A second talin-dependent physiological process we investigated is clot retraction. Clot retraction requires talin to link the cytoskeleton to integrins in a bidirectional signaling process (28,36). This signaling mechanism provides mechanical response required to reduce the overall size of clots thus restoring normal blood flow. Interestingly, a point mutation in the F3 lobe of THD at amino acid 325 has been shown to cause impaired clot retraction (28). As this critical residue overlaps with the putative binding site of G␣ 13 SR2 Pep to THD (Fig. 1), it might be expected that G␣ 13 SR2 would play a functional role in regulating clot retraction. To examine this possibility, platelet-rich plasma was pre-treated with G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep , and clotting was induced with thrombin ( Fig. 5B). Thrombus retraction occurred in an upward direction in vitro, and the degree to which it retracted was measured by the volume of liquid below. A significant reduction in clot retraction occurred in G␣ 13 SR2 Pep -but not in G␣ 13 FIGURE 4. G␣ 13 and talin co-localize to integrins. A, immunofluorescence images of CHOA5 cells plated on fibrinogen-coated slides. Phalloiden-594 (red) was used to highlight the degree of spreading, and Hoechst 33342 (blue) was used to identify single cells. Scale bars, 10 m. B, quantification of relative spreading of each CHOA5 cells. AU, arbitrary units. C, DNA (blue), ␤3 integrin (red), and GFP-talin (green) detected in CHOA5 cells in which G␣ 13 (top panels) or G␣ 13 R227A (middle panels) are co-expressed with GFP-talin. Co-localization between GFP-talin and ␤3 integrin appears as yellow. Enlarged merged images are shown below. D, DNA (blue), G␣ 13 (red), and GFP-talin (green) detected in CHOA5 cells in which G␣ 13 (top panels) or G␣ 13 R227A (middle panels) are co-expressed with GFP-talin. Co-localization between GFP-talin and ␤3 integrin appears as yellow. Enlarged merged images are shown below. Scale bars for C and D indicate 10 m. Statistical significance was determined using a Student's unpaired t test where *, p Ͻ 0.05, and **, p Ͻ 0.01. SR2 RandomPep -treated platelets (Fig. 5B) . The observed clot retraction defect induced by G␣ 13 SR2 Pep is consistent with previously reported mutations in talin as well as the notion that an unperturbed talin-integrin linkage is necessary to promote effective mechanical stress for regulating the clot retraction. Taken together, G␣ 13 SR2 Pep is able to significantly reduce platelet binding to collagen under physiological shear stress, to reduce platelet clot retraction, and is a potent inhibitor of platelet aggregation as reported previously. These in vitro data compelled us to examine the functional role of SR2 peptide in vivo.
Talin-mediated integrin activation is directly involved in primary thrombus formation and the prevention of blood loss upon vascular damage. Furthermore, point mutation in the talin F3 lobe at amino acid Leu-325 is known to cause impaired integrin activation and increased bleeding times (28). Based on this observation, we examined the effect of G␣ 13 SR2 Pep on bleeding times in mice pre-treated with either vehicle, aspirin, G␣ 13 SR2 Pep , or G␣ 13 SR2 RandomPep (Fig. 5C). The results showed a modest but significant increase in bleeding when treated with G␣ 13 SR2 Pep as compared with the vehicle or myristoylated G␣ 13 SR2 RandomPep (Fig. 5C). Interestingly, the G␣ 13 SR2 Pep -induced increase in the bleeding time is relatively less than that observed with aspirin alone, even though previous experiments have shown that G␣ 13 SR2 Pep is more effective in blocking platelet aggregation than aspirin (27). This functional divergence in pharmacological activity suggests that the ability to target talin may serve as a more effective therapeutic modality than aspirin, which produces a more severe bleeding phenotype.
Finally, to test the effect of G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep on thrombus formation in vivo, we employed a well established laser-induced thrombosis model (37,38). Importantly, the G␣ 13 SR2 RandomPep -treated mice generated larger thrombi than mice treated with G␣ 13 SR2 Pep (Fig. 5D and supplemental Movies 1 and 2). Thrombus formation in vivo was analyzed over 140 s, and representative pictures of the first 90 s are shown (Fig. 5D). In summary, these data suggest that G␣ 13 SR2 Pep blocks talin activation resulting in defective platelet adhesion to collagen and clot retraction, increased bleeding time, and decreased in vivo thrombus formation.

Discussion
G protein signaling through G␣ 13 has been an area of intense interest recently because of its broad physiological and pathological implications. G␣ 12 and G␣ 13 are often paired together as having similar properties; however, it is quite clear that G␣ 13 plays a far more critical role in numerous developmental and pathological processes. For example, systemic knock-out of GNA13 is lethal at the embryonic stage, but GNA12 is not (39). Similarly, conditional knock-out of G␣ 13 produces hemostatic defects, although the knock-out of G␣ 12 is without such effect (40). Therefore, crucial functional and signaling differences must exist between the two closely related G proteins. In this context, it is known that G␣ 13 plays an important role in the regulation of RhoA activity in a variety of cells, notably through its SR1. In platelets, G␣ 13 /RhoA activity is primarily modulated through protein-activated receptor GPCR signaling (25), but this pathway alone fails to explain the global importance of G␣ 13 in a broad spectrum of biological processes, including neurogenesis, embryogenesis, hemostasis, vascular defects, and impaired chemokinesis, among others (39,40). Furthermore, unlike most other heterotrimeric G proteins, G␣ 13 is ubiquitously expressed in all cell lines and in all mammalian species. This conservation of G␣ 13 , and in particular the conservation of its switch regions, throughout the evolutionary process indicates that G␣ 13 functions in an extremely important capacity in cell development and disease processes. Because the importance of integrin signaling is also of global biological significance, and integrin signaling and G␣ 13 signaling share many biological profiles, we recently investigated whether there is a common mechanistic link between each signaling process. In this regard, we recently reported that G␣ 13 directly binds to another protein that is also embryonically essential and ubiquitously expressed, i.e. talin. Specifically, it was found that the highly conserved switch region 2 of G␣ 13 forms a complex with THD, also called the FERM domain, and the disruption of this interaction resulted in global inhibition of human platelet aggregation. Although these findings provided the first evidence for a novel G␣ 13 -talin signaling pathway, they did not define the underlying mechanism by which G␣ 13 functions to activate talinmediated integrin signaling. In this study, we provide a molecular mechanism by which SR2 of G␣ 13 binds to the talin head domain and disrupts the autoinhibited state of talin (Fig. 6).
Our first series of experiments demonstrated that G␣ 13 SR2 Pep recognized a single binding site on THD that is critical for talin autoinhibition. Point mutagenesis tests further confirmed these findings. Next, the GST-SR2 fusion proteins were found to sterically interfere with talin's rod domain R9 -12, and similar results were obtained using the full-length recombinant G␣ 13 . Finally, we demonstrated that G␣ 13 SR2 Pep could sensi-tize full-length talin to limited proteolysis, which increased upon the addition of PIP 2 , suggesting a possible cooperative effect of two modulators. These observations are similar to those attributed to RIAM, which binds and interferes at the F3/Rod autoinhibitory domain on THD, suggesting G␣ 13 may act in a parallel or perhaps a compensatory manner (13,21). If the G␣ 13 /RIAM-binding sites on talin overlap, it would be reasonable to suggest that the SR2 peptide would not only inhibit G␣ 13 from binding and activating talin but prevent RIAM from doing so as well. Future studies into their respective pathways, presumably requiring conditional knock-out mouse models, may clarify some of these questions. Interestingly, G␣ 13 binding to a FERM domain is not without precedent. G␣ 13 binds to the N-terminal half of radixin, which contains a highly similar sequence to THD (41). G␣ 13 binding to radixin also induces a conformational change; therefore, G␣ 13 may have a dynamic role in the regulation of FERM domain containing proteins.
In this study, several experimental approaches were used to investigate the functional role of G␣ 13 in regulating talin conformational changes. In one case, an N-terminal GFP-tagged full-length talin with a calpain-deficient mutation (L432G) was utilized (42). The calpain-deficient mutation was necessary to clarify the open conformation of talin separated by the head and rod domains due to potential proteolysis by calpain. Moreover, a tetracysteine CCPGCC motif capable of binding to an arsenical dye ReAsH was introduced into the talin rod domain. ReAsH was selected due to the relatively low probability of perturbing talin function with the insertion of small amino acid motifs. Additionally, ReAsH was a preferable alternative to adding a second fluorophore at the C terminus of talin often employed by other techniques such as the bimolecular fluorescence complementation. Finally, ReAsH provides more room for error in the placement of a fluorescence probe than the traditional FRET pairs that display activity at ϳ10 Å, thus increasing the chances of detecting signal without placing the probe in a location that could perturb the functionality of talin. Although this assay is designed to examine gross talin conformation, it cannot distinguish between the multitude of inhibited states such as originating from the "donut" model elucidated through small angle X-ray scattering (16), the "dumbbell model" (14), or the "triple domain" composed of the autoinhibitory rod domains (17). Talin is dynamically regulated by a multitude of proteins, and our FRET assay used in this study demonstrates the importance of G␣ 13 in these processes, more specifically through switch region 2. Importantly, these data agree with our previous integrin activation studies (27). A decrease in FRET signal was observed after co-transfection of GFP-talin-1822 and G␣ 13 as compared with the G␣ 13 R227A mutant (Fig. 3I). The reduced FRET signal indicates a switch to an active talin conformation, which agrees with our previous study where G␣ 13 co-transfection with GFP talin has a greater degree of integrin activation than G␣ 13 R227A (27).
To demonstrate the effect of full-length G␣ 13 in the cellular milieu, the CHOA5 cells were co-transfected with G␣ 13 and GFP-talin. CHOA5 cells expressing both G␣ 13 and GFP-talin showed a relatively larger surface area as compared with cells expressing each construct individually (Fig. 4, A and B). The effect of G␣ 13 and GFP-talin co-expression was comparable FIGURE 6. Schematic representation of G␣ 13 -talin-integrin activation pathway. G␣ 13 directly binds to the F3 domain of talin to relieve its autoinhibited state. Upon activation, talin can bind and activate integrin resulting in conformational changes that convert integrin from a low affinity to high affinity state.
with a constitutively active version of talin, GFP-THD. The SR2 mutant, R227A, showed a phenotype similar to either GFPtalin or G␣ 13 alone. These results are consistent with our previous findings where PAC1 (where PAC1 is the antibody that binds to the active form of platelet integrin ␣IIb␤3) binding to CHOA5 cells, reflecting integrin activation, increased after cotransfection of G␣ 13 and GFP-talin but not R227A or R232A SR2 mutants (27). The observed effects of G␣ 13 and talin on cell morphology were also in agreement with their co-localization pattern. Altogether, these data suggest that expression of G␣ 13 can increase cell spreading but only when full-length talin is co-expressed. In platelets, the copy number of talin to G␣ 13 is ϳ8:1; therefore, this overexpression system does not serve to elucidate precisely what occurs in vivo but rather establishes that there is an observable co-localization of G␣ 13 with talin that yields increased integrin activation and spreading.
The critical role of G␣ 13 SR2 in the regulation of talin autoinhibition led us to evaluate its translational potential in platelets as a model system. Treatment of whole blood with myristoylated SR2 peptide showed a significant, but not complete, inhibition of platelet adhesion and aggregate size when perfused over collagen as a substrate under physiological conditions of sheer stress (Fig. 5A). These results provide evidence that platelet adhesion and recruitment on a collagen surface occur through G␣ 13 SR2-mediated events. Similarly, the clot retraction in thrombin-stimulated platelets showed a significant defect when G␣ 13 SR2-talin signaling was blocked with G␣ 13 SR2 Pep, which is consistent with the notion that G␣ 13 SR2 plays an important role in this talin-dependent process. Furthermore, in vivo studies in mice demonstrated that infusion of G␣ 13 SR2 Pep did, as might be expected, increase bleeding time. However, it is noteworthy that this increase was significantly less than that produced by aspirin infusion, even though ex vivo platelet aggregation studies revealed the reverse profile, i.e. profound inhibition of aggregation with G␣ 13 SR2 Pep and only modest inhibition with aspirin (27). This divergence in pharmacological activity suggests that therapeutic targeting of G␣ 13 SR2 may produce an improvement over some prevailing anti-thrombotic methodologies. Finally, the laser-induced thrombosis in mice treated with G␣ 13 SR2 Pep showed a significant decrease in both thrombus size and footprint on the vessel wall (Fig. 5D). This observation is consistent with our previous in vitro collagen results regarding the critical contributions of G␣ 13 SR2talin signaling to both platelet adhesion and platelet recruitment.
While reporting our initial G␣ 13 and talin binding studies (27), we demonstrated that G␣ 13 SR2 is linked to talin-mediated integrin activation via the inside-out signaling pathway. Additional results also suggested that G␣ 13 SR2-talin signaling may be involved in the process cellular adhesion (27). The present results are consistent with a role for G␣ 13 SR2 in each of these processes, because they provide evidence that G␣ 13 SR2-talin signaling participates in both platelet adhesion and platelet recruitment to the site of vascular damage. Our model differs from the previous studies (43)(44)(45) where the focus was primarily on the outside-in signaling mediated through the direct interaction of G␣ 13 SR1 with ␤3 integrin. It is to be noted that in the previous study (45), a 6-amino acid peptide (FEEERA) derived from ␤3 integrin was used as an inhibitor of integrin activation and laser-induced thrombosis in vivo. In fact, only a 5-amino acid (EEERA) peptide derived from ␤3 integrin was sufficient to bind G␣ 13 (45). However, a simple database BLAST search of EEERA sequence reveals hundreds of identical hits, including several in proteins expressed in platelets. In contrast, in this study, we used a unique 17-amino acid peptide (SR2) derived from G␣ 13 as an inhibitor of multiple platelet functions, including aggregation, adhesion, clot formation, bleeding time, and laser-induced thrombosis in vivo. Future studies will be required to reconcile the functional role of G␣ 13 in inside-out and outside-in signaling pathways, with a particular emphasis on the validity of highly charged and relatively small 5-amino acid peptides of ␤3 integrin as specific binders of G␣ 13 and their potential utility as anti-thrombotic leads with reduced bleeding diathesis.
In summary, our findings unveil a novel molecular mechanism of G␣ 13 as a regulator of talin activation by relieving its autoinhibition. These studies suggest that the G␣ 13 SR2-talin activation pathway may represent a global integrin signaling process. Because of the highly conserved nature of G␣ 13 , talin, and integrins, the broad implications of this regulatory mechanism may thus extend to a multitude of cell types and pathologies.

Materials and Methods
Protein Expression and Purification-GST-G␣ 13 SR2 or GST-G␣ 13 SR2 Random fusion proteins were generated as described previously (27). THD constructs were cloned using pET15b as a backbone containing amino acids 1-400. QuikChange mutagenesis was employed to construct THD mutants M319A, L325K, and E335R. Inverse PCR deletion mutagenesis was used to create THD ⌬F2, ⌬F3, and ⌬F2-3 constructs. For the recombinant talin rod domain constructs, R9 -12 (residues 1654 -2344) and R13 (residues 2338 -2541) were cloned into pLIC-His vectors as described previously (46,47). Similarly, the GST-THD was cloned into pLIC-GST vector. All THD and rod domain constructs were purified using BL21 strain Escherichia coli, 0.2 mM isopropyl 1-thio-␤-D-galactopyranoside induction for 3 h at 37°C, lysis using tip sonication, and purification using a nickel-nitrilotriacetic acid FastFlow column (GE Healthcare). Recombinant G␣ 13 was purified as described previously using Sf9 insect cells (32). Baculovirus constructs containing murine G␣ 13 , G␤, and His 6 -tagged G␥ constructs were generously provided by Dr. Tohru Kozasa. Briefly, three baculoviruses containing G␣ 13 , G␤, and G␥ were used to co-infect logarithmic phase Sf9 cells in Sf900II (GE Healthcare) media for 48 h. The membrane fractions were isolated as described previously (32) and purified using a nickel-nitrilotriacetic acid FastFlow column, and G␣ 13 was eluted using AlF 4 -GDP (where AlF 4 -GDP is tetrahedral aluminum fluoride that forms a complex with GDP bound by G␣ subunits that mimic the transition state of GTP hydrolysis). Purification of full-length native talin from human platelets was performed, as described previously (48), with the following modifications: ion exchange chromatography employed Source15Q resin (GE Healthcare) for anion exchange, followed by Mono S resin (GE Healthcare) for cation exchange, followed by S200 resin gel filtration to isolate talin dimer fraction. The S200 column was calibrated using a high molecular mass gel filtration calibration kit, using ferritin (440 kDa) as a marker for dimeric talin (GE Healthcare).
THD Cross-linking and Peptide Mapping-Recombinant THD at a concentration of 5 mg/ml was cross-linked to biotinylated SR2 peptide (VGGQRSERKRWFECFDS, b-G␣ 13 SR2 pep ) or SR2 scrambled peptide (GCRKEVFS-DRQWFGSRE, b-G␣ 13 SR2 RandomPep ) (27) using BS 3 homo-bifunctional amine cross-linker (ThermoFisher Scientific). The reaction was quenched in 50 mM Tris and resolved with a precast SDS-12% PAGE (Bio-Rad). Specific cross-linking was determined using a streptavidin-HRP antibody and resolved with a chemiluminescent substrate (Pierce). Cyanogen bromide cleavage was performed as described previously (49). Briefly, 5 mg/ml THD cross-linked with b-SR2 was added to a solution of 70% formic acid containing 0.1 M cyanogen bromide and incubated for 24 h at room temperature. After TCA precipitation, fragments were resolved using a pre-cast 4 -20% gradient SDS-PAGE. Bands containing the cross-linked peptide were examined using streptavidin-HRP. Because CNBr cleavage occurs at the C terminus of methionine residues, we used ExPASy peptide cutter protein cleavage sites prediction tool to map the potential sites. Additionally, thrombin (0.01 units/ml) was used to cleave THD, which is specific for one site in the F3 domain. Samples prepared for mass spectrometry were crosslinked as described above using THD, b-G␣ 13 SR2 pep , and BS 3 . Biotinylation modifies the primary amine of the SR2 peptide, leaving a single lysine as a substrate for amine cross-linking. The reaction was resolved using a pre-cast SDS-12% PAGE, and the band that was ϳ2.0 kDa larger than THD treated with BS 3 alone was excised using a sterile scalpel. Samples were examined at the Taplin Mass Spectrometry Core Facility at Harvard Medical School. Unique THD sequences cross-linked to SR2 were identified using a Sequest algorithm, tracking the sequence "BS 3 -KR." Quantification of THD Mutants by ELISA-b-G␣ 13 SR2 Pep or b-G␣ 13 SR2 RandomPep was added to Nunc-Immulon wells (Dynatech) at a concentration of 250 nM in 10 mM NaHCO 3 / Na2CO 3 , pH 9.4, in a volume of 100 l, and incubated overnight at 4°C. Wells were washed with blocking buffer (PBS, 2% BSA, 0.05% Tween 20). Even distribution of peptide was verified using streptavidin-HRP and TMB substrate (Pierce) detection. Wells were blocked for 1 h at room temperature, followed by the addition of 2.5 M THD constructs WT, M319A, L325K, and E335R. Each well was washed, and signal was detected using an in-house antibody specific against the head domain of talin (Rb79).
GST Peptide Binding and Competition Assays-GST fusion GST-G␣ 13 SR2 or GST-G␣ 13 SR2 Random at a concentration of 2.5 M was incubated with GSH resin (GE Healthcare) in the binding buffer (20 mM Tris-HCl, pH 8, 150 mM NaCl, 1.0 mM MgCl 2 , 1.0 mM DTT, 0.01% Tween 20). Subsequently, 5 M THD was added in the presence of BSA or recombinant talin rod domain R9 -12. Samples were washed in the binding buffer and boiled in SDS sample buffer for 10 min and resolved using SDS-10% PAGE. Equal loading of GST fusion proteins was determined with Ponceau staining, and the THD amount was determined using the Rb79 antibody, although the rod domain was detected using a talin 8d4 antibody (Sigma). Binding of full-length G␣ 13 was accomplished with GST fusion protein expressing talin's head domain (aa 1-400) or GST alone. GST-THD (5 g) or equimolar amounts of GST were incubated with 1 g of G␣ 13 in binding buffer (above) with either 30 M GDP or GTP␥S on ice for 1 h. Each sample was added to 10 l of GSH resin pre-equilibrated in binding buffer and bound for 1 h at 4°C with gentle agitation. Samples were washed four times in binding buffer and boiled in SDS sample buffer. Equal loading was confirmed via Ponceau stain, and G␣ 13 binding was confirmed via G␣ 13 mAb (1:500) (Santa Cruz Biotechnology).
GST-THD, G␣ 13 , and Rod Domain Binding and Competition Assays-To obtain sufficient quantities of full-length G␣ 13 required for competition binding assays with the talin rod domain, we utilized G␣ i/13 constructs generously provided by Drs. Kozasa and Kreutz. Briefly, the N terminus of G␣ 13 is replaced by G␣ i1 that permits the generation of soluble G␣ subunits without co-expression of cognate ␤ and ␥ subunits, which in turn dramatically increased the yield (33). The binding assay was performed using GST-THD, talin R9 -12, and G␣ 13 that were dialyzed overnight in the binding buffer (20 mM Tris-HCl, pH 8, 1.0 mM EDTA, 10 mM ␤-mercaptoethanol, 300 mM NaCl, 10 mM MgCl 2 , 30 M GDP). Prior to the binding assay, G␣ 13 was incubated with either 60 M GTP␥S or GDP for 1 h on ice.
G␣ 13 (30 M), in a total reaction volume of 250 l, was incubated with 30 g of GST-THD for 2 h at 4°C. Talin R9 -12 was added in increasing concentrations (0, 5, 10, 20, 30, and 50 M). G␣ 13 (30 M) bound to GDP was incubated with GST-THD, and 30 M talin R9 -12 added subsequently. Each reaction was incubated for 1 h at 4°C, followed by incubation with 30 l of GSH resin pre-equilibrated in the binding buffer with 0.1% BSA for 2 h at 4°C. Beads were washed 5 times with the binding buffer containing 0.01% Triton X-100. Each sample was boiled in SDS-PAGE sample buffer, resolved using a 4 -20% gradient gel, and probed using either G␣ 13 pAb (Santa Cruz Biotechnology) or anti-rod domain antibody Rb89.
Limited Proteolysis of Talin Bound to Peptides-Full-length talin (dimer fraction) at a concentration of 5 M was pre-incubated with 87.5 M G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep and/or 45 M PIP 2 (Avanti Polar Lipids) for 30 min at room temperature. Subsequently, either Endo LysC or Endo AspN was added at 8.0, 1.6, or 0.32 ng into a total reaction volume of 50 l in PBS for 30 min at room temperature. Samples were treated with Complete Protease Inhibitor Mixture (Roche Applied Science), boiled in SDS loading buffer, and resolved using 4 -20% gradient SDS-PAGE (Bio-Rad). Talin was probed using anti-talin head domain antibody TA205 (Sigma).
CHOA5 Cell Culture and Immunofluorescence-Cells and reagents were obtained from the following sources: CHO cells stably expressing the platelet integrin ␣IIb␤3 (CHOA5 cells) were kindly provided by Dr. Mark Ginsberg; cells were transfected with GFP-C1-talin (kindly provided by Dr. Jun Qin); GFP-C1 and GFP-C1-THD (kindly provided by Dr. Edward Plow); pcDNA4-RIAM was a gift from Vicki Boussiotis (Addgene plasmid 32803); and murine pCMV-Sport6-G␣ 13 or pCMV-Sport6-G␣ 13 (R227A), originally obtained from the laboratory of Dr. Guy Le Breton. Cell spreading was performed using fibrinogen-coated slides (100 g/ml) blocked with 1% BSA for 1 h at room temperature. CHOA5 cells transfected with respective plasmids were plated onto fibrinogen-coated slides for 15 min at 37°C. Subsequently, cells were fixed in 4% paraformaldehyde for 10 min and treated with phalloidin (594) according to the manufacturer's specifications (Molecular Probes) and Hoechst 33342. For protein localization analysis, G␣ 13 was detected using a G␣ 13 mouse monoclonal antibody (Santa Cruz Biotechnology). ␤3 integrin was detected using the goat polyclonal antibody (Santa Cruz Biotechnology). A Nikon Eclipse TE2000-E microscope was used for ϫ60 to 100 magnification for fluorescence microscopy. MetaMorph series software (Molecular Devices) was used to record images. Green fluorescence detected using FITC-conjugated mouse antibody (Molecular Probes) was employed for the detection of G␣ 13 , and AF633conjugated mouse or goat antibodies were utilized for G␣ 13 and ␤3 integrin detection, respectively (Molecular Probes).
FRET Constructs-GFP-talin was used as the backbone for all FRET experiments. GFP-talin has been validated by several groups (13,15,27,42); therefore, GFP served as the donor molecule. The FRET acceptor molecule, a cell-permeant red fluorescence dye known as ReAsH (ThermoFisher Scientific), can bind to a 6-amino acid sequence designated as a tetracysteine motif or "CCPGCC" (50). The tetracysteine motif was engineered into three separate constructs of GFP-talin at defined areas of the rod domain (Fig. 3A). To avoid perturbation of the structure of talin's rod domain, this motif was incorporated into flexible linker regions connecting the helical bundles at amino acids 1457, 1655, and 1822. The rationale for selecting these regions was to place FRET acceptor molecules at locations near the autoinhibitory R9 domain. The assumption was that GFPtalin is autoinhibited by an adjacent talin molecule; thus, depending on the location of the FRET acceptor, there should be varying levels of FRET pairing. This idea could conceivably apply to an autoinhibited monomer as well. One additional consideration was the cleavage of talin by calpain. Calpain-2 is believed to cleave talin at an unstructured linker region located between the head domain and the rod domain, which can regulate the formation and turnover of focal adhesions (42). Talin proteolytic cleavage would thus reflect a decrease in the FRET signal due to the separation of the rod and head domains rather than a gross conformational change. Therefore, to avoid false-positive signals, a calpain-cleavage immune construct designated here as GFP-talin-L432G was utilized for each experiment.
CHOA5 cells were transfected with each construct individually using the DNA transfection reagent (BioTools) and incubated for 24 h. ReAsH dye was added to a final concentration of 2.0 M and washed with BAL buffer according to the manufacturer's instructions. An LSRII flow cytometer (BD Biosciences) was used to analyze the FRET signal. Untransfected cells stained with ReAsH dye were used as a negative control to assess any nonspecific dye binding. It should be noted that ␣-tubulin1b (pC2xTC1-␣-tubulin, which was a gift from Amy Palmer (Addgene plasmid 36326)) was transfected because it possesses a CCPGCC motif and can bind the ReAsH dye but has no donor molecule (such as GFP). Consequently, it should have no FRET signal and can be used to validate that activation of ReAsH signal does not occur using the 488 nm (FITC) laser alone. Finally, GFP-C1-talin-L432G with no CCPGCC motif was incubated with ReAsH dye and used to gate GFP ϩ /FRET Ϫ cells, because no acceptor molecule is present.
FRET Analysis and Flow Cytometry-Each construct (GFP-C1-talin-L432G-1457; GFP-C1-talin-L432G-1655; GFP-C1talin-L432G-1822) was analyzed using the 488-nm laser, which should only excite the GFP (donor) molecule. The R 0 (where R 0 indicates the Förster distance at which half the energy is transferred from a donor to an acceptor molecule) of ReAsH is ϳ54 Å, thus enabling a wide range for error in the placement of our ReAsH-binding motifs (50). However, the overall length of fully extended talin molecules is roughly 500 Å, thus decreasing the possibility of FRET occurring in an active elongated state. Any signal detected in the PE channel was therefore attributed to ReAsH excitation due to FRET pairing. Fixation of cells greatly diminished the FRET signal in our hands; therefore, equal expression of constructs (G␣ 13 and G␣ 13 R227A) was verified using Western blotting. It should be noted that FRET analysis did not use conventional calculation methodology (51,52). Traditional calculations examine the ratio of median fluorescence intensities of FRET-positive and -negative populations. We observed that samples with the fewer FRET-positive cells had the highest MFI, as only double-positive cells expressing the highest levels of GFP-talin-CCPGCC constructs showed FRET to occur. Moreover, cells with the highest ratio of FRET-positive cells showed FRET in cells with lower MFI, suggesting that even low levels of GFP-talin-CCPGCC constructs were capable of emitting higher FRET signals. If MFI is the only metric, it would appear that the samples with the fewest FRET-positive cells will show higher efficiency, which is misleading. Thus, our results only examined the ratio of FRET-positive cells expressing the GFPtalin-CCPGCC constructs to FRET Ϫ cells, which express the GFPtalin-CCPGCC: PE ϩ FITC ϩ /PE ϩ GFP ϩ ϩ PE Ϫ GFP ϩ .
Peptide Preparations-All peptides were synthesized, and HPLC was purified (Ͼ95% pure) by the Research Resource Center at the University of Illinois, Chicago. Peptides were dissolved in DMSO for all in vitro experiments. For all in vivo assays utilizing myristoylated peptides, a 50 mM stock solution was first prepared by dissolving each peptide in 70% DMSO and 30% PEG300. The stock solution was then diluted with 0.9% saline to bring the final concentration to 3.7 mM, and a 100-l solution was administered to each mouse, bringing down the total final DMSO concentration to 0.2%.
Collagen Binding and Clot Retraction-Human platelets were provided by Tufts Medical Center Blood Bank and American Red Cross, Dedham, MA. One milliliter of citrated whole blood was treated with 150 M G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep for 30 min at room temperature with gentle agitation. Samples were perfused over Chrono-par collagen (Chrono-log Corp.)-coated slides at a rate of 1000/s using a GlycoTech parallel plate flow chamber (GlycoTech Corp.). The chamber was washed with PBS, pH 7.4, and examined at ϫ10 magnification. Five fields were taken per experiment, and each experiment was performed in triplicate. Clot retraction was performed using platelet-rich plasma obtained from blood using the sodium citrate (3.2%) anti-coagulant (BD Biosciences). Platelet count was normalized to 2 ϫ 10 8 /ml using platelet-poor plasma. 100 l of platelet solution was added to HEPES/ Tyrode's buffer, as well as 2 l of packed RBCs for color contrast, to aid visualization. RBC addition had no effect on clot retraction in parallel experiments. G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep was dis-solved in DMSO and incubated with each sample for 15 min prior to thrombin (1 unit/ml) addition. The clot-free volume was removed and measured for volume calculation.
Tail Bleeding Analysis-Tail bleeding was performed as described previously (53). Briefly, mice (ϳ25 g each) were anesthetized with sodium pentobarbital, and 100 l of either vehicle (n ϭ 11), 3.7 mM aspirin (n ϭ 9), 3.7 mM SR2, or SR2 random peptide (n ϭ 14) was injected through the tail vein. After 5 min, an amputation was performed at 0.5 cm from the tip of the tail. The tail was placed into a pre-warmed saline bath at 37°C, and the bleeding times were measured.
Induction of in Vivo Thrombosis-Laser-induced thrombosis in mice was performed as described previously (37). Briefly, G␣ 13 SR2 Pep or G␣ 13 SR2 RandomPep (5 m peptide for every mouse kilogram) was injected through a jugular vein cannula followed by laser-induced thrombosis of an exposed cremaster muscle artery. Mouse platelets were labeled using a plateletspecific ␤3 integrin antibody conjugated to AlexaFluor 647. Laser-induced injuries triggering platelet accumulation at the site of injury and fluorescent intensity over time were recorded. Three mice were analyzed under each condition with 10 -15 observations per mouse. Scale bars (10 m) and intensities of platelet signal (ranging from 308 to 7668 arbitrary units) are equal between G␣ 13 SR2 Pep and G␣ 13 SR2 RandomPep conditions.
Statistics-Statistical significance for all methods, with the exception of laser-induced thrombosis, was determined using a Student's unpaired t test where *, p Ͻ 0.05, and **, p Ͻ 0.01. Laser-induced thrombosis significance was determined using the Mann-Whitney test on median fluorescent intensity values of thrombus formation over time.