Direct observation of a coil-to-helix contraction triggered by vinculin binding to talin

Real-time detection of vinculin head binding to the talin R3 domain

To isolate individual vinculin-binding events, we study the talin R3 domain, a four-helix bundle located in the N-terminal region of the talin rod (fig. S1) (11). This domain has two vinculin-binding sites and remarkable low stability due to the four-threonine belt buried in its hydrophobic core (11, 22, 23); hence, it is ideally placed to play a key role in talin activation by force, recruiting a cluster of vinculin molecules at low forces, which would amplify the mechanical coupling and contribute to the maturation of the focal adhesion. Mutation of these four threonines to hydrophobic valine and isoleucine residues (T809I/T833V/T867V/T901I, the IVVI mutation; R3 IVVI from now on) increases the domain stability while leaving its vinculin-binding properties intact (11).

Here, we study vinculin binding both to the R3 WT (wild type) and R3 IVVI but focus on the characterization of the mutant because of its higher mechanical stability, which amplifies the mechanical signature for vinculin binding; however, the binding mechanism is completely equivalent on both domains. We use our custom-made magnetic tweezers to apply physiological forces to single R3 domains in the presence of vinculin head and measure its extension changes in real time with nanometer resolution. Our molecular construct contains the R3 domain (either the WT or IVVI mutant) followed by eight repeats of titin I91 domain as molecular handles, flanked by a HaloTag enzyme for covalent tethering to the glass surface, and biotin for anchoring to a streptavidin-coated superparamagnetic bead (Fig. 2A; see the Supplementary Materials for detailed methods). Forces between 0.1 and 120 pN are applied with a sub-piconewton resolution by generating a magnetic field with a pair of permanent magnets or a magnetic tape head (26, 27).

Figure 2B shows our single-molecule assay for vinculin head binding detection. Starting from folded R3 IVVI at 4 pN, we increase to 9 pN, where the R3 IVVI domain exhibits reversible folding/unfolding dynamics [see (27) and fig. S2 for the mechanical characterization of R3 IVVI and R3 WT]. The unfolding/folding transitions are resolved as ascending/descending ∼20-nm changes in the extension of R3 due to the transition between the folded state and an unstructured polymer. Previous force spectroscopy studies showed that vinculin head binding blocks talin refolding (25). Our experiments confirm this observation; in the presence of 20 nM vinculin head, R3 folding dynamics cease after a few seconds and the protein is locked on its unfolded state (Fig. 2B). This blocked state extends for several hours in contrast with the unaltered R3 folding dynamics observed in the absence of vinculin head (fig. S3). The improved resolution of our instrument and the use of a constant force allow us to observe the binding event as a short contraction of the unfolded polypeptide, which always precedes the arrest of R3 folding dynamics (Fig. 2B, inset, red arrow). This contraction indicates that vinculin head binding induces a conformational change on unfolded R3, likely a coil-to-helix transition required for vinculin head to firmly bind its substrate (Fig. 1). This bound state can be reversed by a high force pulse, where the opposite transition occurs and ∼3-nm upward steps are observed (blue arrow), after which the R3 domain recovers its ability to fold (fig. S4). Experiments with R3 WT in the presence of vinculin head reveal this same effect; however, at the coexistence force for R3 WT (5 pN), the binding contraction is too small to be detected and is only observed at forces >8 pN (fig. S5).

Cooperative binding of vinculin head to the talin R3 domain

Our measurements demonstrate that the mechanical fingerprint for vinculin head binding is a contraction of a few nanometers on the unfolded R3 polypeptide. Structural studies have determined that vinculin-binding sites are amphipathic six-turn α-helices buried in the core of talin domains by extensive hydrophobic interactions (28). Upon binding, this helix is inserted intimately into vinculin head, which displaces the initial interaction between the head and tail of vinculin, present in its autoinhibited state (29). Hence, this helical structure is required for vinculin recognition and binding, which might explain the need for the structural contraction we observe upon binding.

Figure 3A shows averaged recordings of individual vinculin head binding events on R3 IVVI, measured at different forces. The size of the contraction induced by binding increases with force, which is an expected observation in the transition from a random-coiled chain to a compact structure. In all our experiments, we observe a single contraction event that is sufficient to form the bound state, although the R3 domain has two vinculin-binding sites. By contrast, when dissociating mechanically at forces above 40 pN, we resolve, in most cases, two distinct steps with an extension of ∼3 nm (Fig. 3B). In some traces, we observe only a single unbinding step, likely because the first one occurred too fast to be resolved since two unbinding steps are required for complete dissociation (fig. S4). An analysis of the unbinding kinetics confirms that a fraction of unbinding events occurs within the resolution limit of our instrument (fig. S6).

Our data suggest that two vinculin head molecules bind simultaneously to unfolded R3, whereas each vinculin head unbinds independently. The magnitude of the binding and unbinding events scales with force following the freely jointed chain (FJC) model for polymer elasticity (30). We measure the size of the binding contraction as a function of force (Fig. 3C, red circles) and fit its dependence to the FJC model, obtaining a contour length change of ∆Lc = 7.33 ± 0.69 nm (Fig. 3C, red dashed line). Plotting the same FJC fit with half that contour length results in a complete agreement with the steps measured for unbinding (Fig. 3C, blue). This indicates that the R3 sequence sequestered by each binding contraction is twice that liberated by each unbinding step. Given that each vinculin-binding site contains 19 residues and that the extension of the formed helix is about 3.5 nm (11, 29), the expected contour length change for each coil-to-helix transition is 3.7 nm, in agreement with our measurements. An equivalent force scaling is measured in the binding/unbinding events on the R3 WT domain, which indicates that the structural transition triggered by binding is analogous in both the mutant and R3 WT domains (fig. S5). Together, our observations confirm that each binding contraction corresponds to the simultaneous reformation of two vinculin-binding helices, whereas each unbinding step is the uncoiling of a single helix.

From this evidence, it remains uncertain which is the binding pathway followed by vinculin head to reach the bound state. One of the possible scenarios could involve repeated fluctuations between the coil and helix states in talin, with a first-hitting binding mechanism upon encounter of the appropriate substrate conformation. However, an interesting observation from our single-molecule recordings is that binding is not instantaneous but occurs as a slow relaxation, which, at 9 pN, takes as long as 500 ms and that accelerates with force (fig. S7). This suggests that vinculin head binding requires a maturation process, perhaps initiated by a recognition of key residues on the unfolded R3 polypeptide, followed by a sequential contraction and reformation of the helices. However, the rationale for the force dependence observed in this maturation process remains inconclusive.

The simultaneous nature of the vinculin head binding event suggests a cooperative mechanism that should introduce a nonlinear concentration dependence on the binding kinetics. From the perspective of single-molecule enzyme kinetics (31, 32), this process can be modeled as

where F stands for folded R3, U stands for unfolded R3, U_b stands for the bound state, and Vh stands for the vinculin head domain. This process is governed by three kinetic rates: k_U and k_F are the unfolding and folding rates of R3, and $k_{\text{on}}^0$ is the pseudo–first-order rate constant, which depends on the concentration of vinculin head as $k_{\text{on}}^0={[Vh]}^2{k}_{\text{on}}$ because two molecules are required to acquire the bound state. Experimentally, we measure the waiting time (t_b) to observe the talin-bound state (Fig. 4A), and the binding rate k_b = 1/t_b can be derived analytically as (see section IX in the Supplementary Materials)

where K_M = (k_U + k_F)/k_on. This expression is the single-molecule analogous to a second-order Hill equation. In this case, the unfolding rate of talin plays the role of the maximum velocity of the reaction.

Figure 4A shows binding trajectories to R3 IVVI at 9 pN and different vinculin head concentrations. The waiting time t_b is determined from each recording as the time from the start of the 9-pN probe pulse until the contraction event after which R3 folding dynamics stop. Figure 4B shows the distributions of waiting times at three representative concentrations, fitted to the expression derived from the kinetic model (see section IX in the Supplementary Materials). Vinculin head binding kinetics are governed by two competing time scales: the folding/unfolding dynamics of R3 and the concentration-dependent on-rate $k_{\text{on}}^0$. At low concentrations, the on-rate is much slower than R3 folding kinetics, and the process is rate-limited by vinculin head association. However, as the concentration increases, both opposing processes become comparable and we observe a peaked distribution; binding cannot occur faster than the R3 domain unfolds. Unfortunately, it is not possible to do experiments at higher vinculin head concentrations because binding occurs so fast that the fingerprint for vinculin head binding is lost.

From these distributions, we calculate the binding rate k_b, which has a quadratic dependence with the concentration as described by Hill equation (Fig. 4C). This nonlinearity demonstrates the cooperative character of the vinculin head binding reaction. Fitting our data to Eq. 1, we obtain k_U = 0.65 ± 0.09 s⁻¹ and K_M = 8143 ± 1700 nM². The value of the unfolding rate agrees with that measured from R3 IVVI folding trajectories (fig. S2). From K_M, we obtain that half the maximum binding velocity occurs at ∼90 nM. Vinculin head has a strong tendency to aggregate, which might lead to small uncertainties in the determination of its concentration, which, however, do not affect the described mechanism (see fig. S8).

Our observations agree with previous evidence that suggested a binding affinity in the nanomolar range (29, 33). Those experiments reported values between 3 and 30 nM but were calculated using biochemical assays where vinculin head was left to interact with isolated vinculin-binding site helices. However, the physiological affinity of vinculin for talin must be understood as a force-dependent quantity, which, as we demonstrated here, triggers structural changes on the binding substrate that should depend strongly on the tension applied to talin.

Mechanical force regulates vinculin head binding

Force is an essential actor in the interaction between vinculin and talin, being required to unfold talin domains and expose its cryptic sites. In addition, and as previously reported (25), we have shown that vinculin head dissociates at high forces (>40 pN). This process arises likely from the destabilization of the reformed helices with force, which eventually will uncoil and expel vinculin head. In this sense, the binding reaction should be hampered by force, as the helices reform and contract a polymer that is mechanically stretched. This suggests that force could play a biphasic role in vinculin binding, first by establishing the threshold for talin unfolding but also by hindering the coil-to-helix contraction as force increases.

We measure vinculin head binding to R3 IVVI at different forces and over a fixed time window of 50 s, which readily demonstrates the biphasic effect of force on binding (Fig. 5A). At 8 pN, R3 IVVI explores the unfolded state with low probability, and vinculin head binds with slow kinetics. As force is increased to 9 and 10 pN, R3 IVVI unfolds more frequently, which results in faster vinculin head binding. However, at 15 and 20 pN, although R3 IVVI is always unfolded, the binding kinetics rapidly slow down with force until binding is blocked above 30 pN.

This negative effect of force on binding arises from the coil-to-helix contraction that is triggered by binding. The R3 polypeptide shortens out of the equilibrium extension imposed by force; hence, vinculin binding does mechanical work against the pulling force, and this energy penalty increases steeply with force. To demonstrate the proposed mechanism, we measure the binding probability over a 50-s time window (Fig. 5B) both on R3 WT and R3 IVVI. R3 WT has lower mechanical stability, showing equilibrium transitions between 4 and 6 pN, whereas R3 IVVI folds between 8 and 10 pN (fig. S2). This difference in mechanical stability results in a lower threshold force for binding for R3 WT, compared to R3 IVVI. The binding probability quickly increases from 4 pN and saturates at 5 pN because of the sharp dependence of the unfolding rates (black squares), whereas for R3 IVVI the same behavior is observed at a higher force of 8 pN (red circles). However, the inhibitory effect of force arising from the coil-to-helix contraction is analogous for both domains because the polymer properties of their binding sites are equal in both cases; the binding probability drops in the same fashion until binding is blocked at forces above 30 pN.

The mechanical work of binding can be estimated on a first approximation as ∆W ≈ F · ∆L(F), where ∆L is the force-dependent contraction measured in Fig. 3C. In this regard, we assume that the on-rate depends on the force as k⁰_on = Ae^{−∆W(F)/kT}. From this simple relation, we derive an analytical expression for the binding probability that incorporates the double effect of force on binding, controlled positively by the domain-dependent unfolding rate k_U, but negatively by the on-rate k⁰_on (see section X in the Supplementary Materials for derivation). We use this expression to describe accurately our experimental data using only two free parameters (Fig. 5B, dotted lines; see table S1 for parameter list). The agreement between the experimental data and our analytical description confirms the proposed mechanism for the mechanical regulation of the interaction between vinculin head and talin rod domains. Vinculin binding requires a coil-to-helix reformation, which is hampered by force. This, together with the force-induced exposure of the cryptic binding sites, defines the force regime over which vinculin binds. This mechanism can be directly extrapolated to other talin domains or to ligand binding reactions that occur under similar conditions. The hierarchical mechanical stability of the talin rod domains will define a range of threshold forces for binding; however, the negative force dependency arises from the entropic penalty of the coil-to-helix transition required for binding and, thus, can be expected to operate in a similar way for all vinculin-binding talin domains.

Magnetic tweezers setup

All experiments were done on our custom-made magnetic tweezers setup, as described before (26, 27). Single molecules were tethered to superparamagnetic Dynabeads M-270 beads (2.8 μm diameter). Calibrated forces were applied using either a voice-coil mounted pair of permanent magnets (Equipment Solutions) or a magnetic tape head (Brush Industries). Image processing was done by custom-written software written in C++/Qt, available upon request. All experiments are done in custom-made fluid chambers built by two sandwiched glass coverslips, separated by a laser-cut parafilm pattern. The fluid chambers are functionalized with the HaloTag ligand and reference beads as described before (26). Both the fluid chambers and magnetic beads are passivized using tris blocking buffer [20 mM tris-HCl (pH 7.4), 150 mM NaCl, 2 mM MgCl₂, and 1% (w/v) sulfhydryl-blocked bovine serum albumin]. All experiments are carried out in Hepes buffer [Hepes 10 mM (pH 7.2), NaCl 150 mM, EDTA 1 mM], 10 mM ascorbic acid, and the desired vinculin head concentration. See section XIII in the Supplementary Materials for further details.

Protein expression and purification

Polyprotein constructs are engineered using BamH I, Bgl II, and Kpn I restriction sites in pFN18a restriction vector, as described previously (26). Our protein construct contains the R3 IVVI, or R3 WT mouse talin domain, followed by eight titin I91 domains, and flanked by an N-terminal HaloTag enzyme and a C-terminal AviTag for biotinylation. Human vinculin head was expressed and purified following an analogous procedure, skipping the biotinylation process. See section XIII in the Supplementary Materials for further details.

Single-molecule data analysis

Our data acquisition software collects data as a binary file, visualized later with custom-written software in Igor Pro (WaveMetrics). All data are acquired at 1000 to 1600 frames per second. Data are smoothed with a fourth-order Savitzky-Golay filter using a box size of N = 101. The folding/unfolding states of talin are automatically detected using a double threshold algorithm. See section XIII in the Supplementary Materials for further details.