• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of biophysjLink to Publisher's site
Biophys J. Oct 5, 2011; 101(7): 1772–1780.
PMCID: PMC3183806

Monomeric and Dimeric Conformation of the Vinculin Tail Five-Helix Bundle in Solution Studied by EPR Spectroscopy

Abstract

The cytoskeletal adaptor protein vinculin plays an important role in the control of cell adhesion and migration, linking the actin cytoskeleton to adhesion receptor complexes in cell adhesion sites. The conformation of the vinculin tail dimer, which is crucial for protein function, was analyzed using site-directed spin labeling in electron paramagnetic resonance spectroscopy. Interspin distances for a set of six singly and four doubly spin-labeled mutants of the tail domain of vinculin were determined and used as constraints for modeling of the vinculin tail dimer. A comparison of the results obtained by molecular dynamic simulations and a rotamer library approach reveals that the crystal structure of the vinculin tail monomer is essentially preserved in aqueous solution. The orientation of monomers within the dimer observed previously by x-ray crystallography agrees with the solution electron paramagnetic resonance data. Furthermore, the distance between positions 1033 is shown to increase by >3 nm upon interaction of the vinculin tail domain with F-actin.

Introduction

The highly-conserved, cytoskeletal adaptor protein vinculin plays an important role in the control of cell adhesion and migration, linking actin dynamics to receptor-based cell adhesion. Control of vinculin binding to the membrane-apposed cell adhesion complex affects mechanical stability and turnover of cell adhesions and thus the cell's migratory status (1–3). Vinculin consists of five α-helical bundle domains (Vd1–Vd5), which perform an autoinhibitory head (Vd1–Vd4) to tail (Vt or Vd5) interaction (4). Vinculin activation requires release of this intramolecular head-to-tail interaction. The Vt domain, which forms an antiparallel bundle of five amphipathic helices (Fig. 1 A) (5), interacts with several ligands including actin filaments (F-actin) and acidic phospholipids like phosphatidylinositol-4,5-bisphosphate or phosphatidylserine (6,7). The binding of F-actin and acidic phospholipids to the Vt domain is competitive (8) and was suggested to determine vinculin activation and inactivation in cell adhesions (1,5,9,10). Furthermore, phosphatidylinositol-4,5-bisphosphate signaling has been implicated in the regulation of the intramolecular head-to-tail bond (6,11) and lipid binding is accompanied by a conformational change in the Vt domain (5,12). Importantly, binding of F-actin to Vt stimulates a different rearrangement in the Vt domain that in turn promotes vinculin dimerization and results in actin filament bundling (5,13,14). Hence, control of vinculin tail conformation and dimerization appear to be crucial for protein function in cell adhesion sites. The crystal structure of Vt reveals a dimer (5), and NMR studies provide evidence for the presence of dimeric Vt in aqueous solution even in the absence of interaction partners (15).

Figure 1
(A) Vt structure with R1 side chains bound to positions 901, 922, 950, 957, 984, and 1033 (PDB-code: 1ST6). (B) EPR spectra of the spin-labeled Vt constructs obtained at room temperature. (C) Plot of left angle bracketH2right angle bracket−1 (inverse of the second ...

The chemical shift changes were found to be consistent with the dimer interface identified by x-ray crystallography; however, the exact dimer conformation could not be concluded from the NMR titration data (15). Furthermore, it remained unclear whether this conformation correlates with the Vt dimer induced in the presence of F-actin, because a model of F-actin cross-linking by Vt, derived from electron microscopy and computational docking, suggested a dimer distinct from that of pure Vt observed by NMR and x-ray crystallography (13,15). In cells, direct experimental evidence for dimerization of vinculin in the absence of (tail) ligands is missing. However, under physiological conditions, ligand-induced conformational changes and dimerization could be facilitated by or develop from transient interaction of Vt domains through sterically accessible dimerization surfaces (13). To elucidate the suggested differences in Vt dimer conformations in the presence and absence of ligands, we apply site-directed spin-labeling electron paramagnetic resonance (EPR) spectroscopy. In this work, we compare inter- and intramolecular distances derived from monomeric and dimeric Vt in solution in the absence of ligands with those of the crystal structure and provide experimental data for the F-actin-induced rearrangement of the dimer.

Site-directed spin-labeling EPR spectroscopy has evolved as a powerful technique to investigate protein structure and conformational changes under near-physiological conditions (16–18). The shape of continuous-wave (cw) EPR spectra recorded at room temperature is sensitive to the reorientational motion of the spin-label side chain providing information on the motional restriction of the nitroxide due to sterical interaction with the secondary and tertiary structure (18,19). In addition, solvent accessibility of the spin-label side chain and the polarity of the nitroxide microenvironment characterize the protein topology (18,20). Determination of distance distributions between two spin-label side chains using cw or pulse EPR approaches, such as double electron-electron resonance (DEER) spectroscopy, has proven to be valuable for the analysis of protein structures and conformational changes on the level of the backbone fold (21–25).

Due to the inherent flexibility of commonly used spin-label side chains, modeling of the spin-label side-chain orientations is necessary for the interpretation of interspin distances with respect to the protein secondary or tertiary structure. Molecular dynamics (MD) or Monte Carlo simulations and the so-called rotamer library analysis (RLA) (26–29) are feasible methods. MD simulations provide detailed insight into the spin-label side-chain motion (26); however, this method is time-consuming and requires large computational resources. The RLA (28) predicts spin-label conformations using a semidynamic structure model. The restriction to a limited set of possible spin-label side-chain conformations renders this method computationally much less demanding. In our approach, MD simulations and the RLA are used to predict internitroxide distances for doubly spin-labeled Vt mutants. The results are related to interspin distance distributions determined experimentally by cw EPR and DEER spectroscopy at X- and Q-band and show that the conformation of monomeric Vt in solution largely resembles that of Vt in the crystal. The DEER data of singly labeled Vt exhibit well-defined interspin distances distributions that allow for the identification of the conformation of the Vt dimer in solution. Furthermore, the comparison of MD simulation and the RLA reveals that the latter method can provide a fast estimation of the orientation distribution of nitroxide spin-label side chains.

Materials and Methods

Cloning and mutagenesis of vinculin tail constructs

Vinculin-encoding cDNAs used in this study are described in Chandrasekar et al. (11). Vt constructs in pQE-30 (Qiagen, Venlo, The Netherlands), encoding amino acids 858–1066, are equipped with N-terminal FLAG and His tags. The QuikChange Method (Stratagene, La Jolla, CA) was used to replace the three wild-type cysteines at positions 950, 972, and 985 by alanine. Subsequently, the triple alanine mutant of Vt was used to generate six mutants carrying a single cysteine, Vt-(A901C), Vt-(A922C), Vt-(A950C), Vt-(A957C), Vt-(V984C/C985V), and Vt-(S1033C), as well as four double cysteine mutants, Vt-(A901C/A957C), Vt-(A901C/S1033C), Vt-(A922C/A957C), and Vt-(V984C/C985V/S1033C). All mutant constructs were verified by DNA sequencing.

Protein expression, purification, and characterization

In the following, the six spin-labeled single cysteine Vt mutants are termed Vt901R1, Vt922R1, Vt950R1, Vt957R1, Vt984R1, and Vt1033R1, and the four double cysteine mutants are Vt901R1/957R1, Vt922R1/957R1, Vt901R1/1033R1, and Vt984R1/1033R1. Spin-label side-chain (R1) positions under investigation are illustrated in terms of their location in the Vt crystal structure in Fig. 1 A. Recombinant Vt mutants were expressed and purified as described before (30). Labeling with the (1-oxy-2,2,5,5-tetramethyl-pyrrolinyl-3-methyl) methanethiosulfonate spin label (MTSSL) was carried out overnight by incubating the proteins bound to Ni-NTA beads (Qiagen) with 1 mM MTSSL, for single cysteine mutants, or 2 mM MTSSL, for double cysteine mutants.

Unbound spin label was removed by washing the Ni-NTA beads several times with 50 mM phosphate buffer, pH7.2. Spin-labeled protein was eluted according to the manufacturer's instructions (Qiagen) and transferred into 20 mM phosphate (pH 7.2) buffer using PD10 desalting columns (GE Healthcare, Waukesha, WI). Spin-labeling efficiency was determined by comparison of double integrated EPR spectra with a standard and found to be between 75 and 90%. Protein integrity was confirmed using Coomassie blue-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels and circular dichroism spectroscopy. To verify protein function, F-actin bundling and lipid binding of spin-labeled vinculin tail mutants were confirmed as described earlier (11,30) (see Fig. S1 and Fig. S2 of the Supporting Material). Protein concentrations were determined using a BCA assay (Thermo Fisher Scientific, Waltham, MA).

EPR spectroscopy

Room temperature cw-EPR measurements were carried out on a MiniScope benchtop EPR spectrometer (MS200; Magnettech, Berlin, Germany) equipped with a rectangular TE102 resonator fluxed with gaseous nitrogen to keep the temperature stable. The microwave power was set to 10 mW and the B-field modulation to 0.15 mT. The sample solutions contain 40% (m/v) sucrose to reduce protein rotational diffusion. Ten microliters of sample volume containing protein concentrations of 80–120 μM was filled into EPR glass capillaries (0.9-mm inner diameter). The width of the central line and the spectral breadth (second moment) were used for the analysis of the spin-label mobility (18).

Low temperature cw-EPR measurements were carried out at 160 K using a homemade X-band EPR spectrometer equipped with an AEG H103 rectangular cavity. The microwave power was set to 0.2 mW, and a B-field modulation amplitude of 0.24 mT was used. Forty microliters of sample solution with a final protein concentration of 80–120 μM was filled into an EPR quartz capillary (3-mm inner diameter). A continuous flow cryostat model No. ESR 900 (Oxford Instruments, Abingdon, UK) allowed stabilization of the sample temperature. Spectra were recorded with a field sweep width of 12 mT. In a distance range below 2 nm, the dipolar interaction between two spin labels lead to a considerable increase of the line width of the cw-EPR spectra. Below 1 nm, Heisenberg exchange interaction may contribute (31). Values of the interspin distances were determined from a detailed line shape analysis. Therefore, simulated EPR powder spectra convoluted with a dipolar splitting function were fitted to the experimental data using DIPFIT (31). In addition, the Tikhonov regularization-based program SHORT DISTANCES, kindly provided by Christian Altenbach (32), was used, which determines best-fit parameters for the interspin distance and distance distribution. In both cases, isotropic distributions of mutual nitroxide orientations were assumed and contributions due to possible Heisenberg exchange interaction were disregarded.

Pulse EPR experiments (DEER) were performed at X-band frequencies (9.4 GHz) with an Elexsys 580 spectrometer equipped with a Flexline split ring resonator ER 4118XMS3 (Bruker, Rheinstetten, Germany). Forty microliters of sample solution with a final protein concentration of 80–120 μM containing 10% of deuterated glycerol was filled into an EPR quartz capillary (1-mm inner diameter). A continuous flow helium cryostat, model No. ESR900 (Oxford Instruments) and an ITC 503S (Oxford Instruments) were used for temperature controlling. All measurements were performed using the four-pulse DEER sequence: π/2(υobs) − τ1 – π (υobs) – t′ – π (υpump) – (τ1 + τ2t′) – π (υobs) − τ2echo (33). For the DEER pulses at the observer frequency, the left angle bracketxright angle bracket channels were used. A two-step phase cycling (+left angle bracketxright angle bracket, −left angle bracketxright angle bracket) was performed on π/2(υobs). Time t′ was varied, whereas τ1 and τ2 were kept constant, and the dipolar evolution time is given by t = t′ – τ1. Data were analyzed only for t > 0. The resonator was overcoupled.

The pump frequency υpump was set to the center of the resonator dip and coincided with the maximum of the nitroxide EPR absorption spectrum, whereas the observer frequency υobs was 65 MHz higher and coincided with the low field local maximum of the absorption spectrum. All measurements were performed at a temperature of 50 K with observer pulse lengths of 16 ns for π/2 and 32 ns for π-pulses and a pump pulse length of 12 ns. Deuterium modulation was averaged by adding traces at eight different τ1 values, starting at τ1,0 = 400 ns and incrementing by Δτ1 = 56 ns. Background correction of the DEER spectrum provided the dipolar evolution function, which is a superposition of functions oscillating with the dipolar coupling frequency ωdd depending on the distances of interacting spins. Interspin distance distributions were determined by fitting of the Fourier-transformed dipolar evolution function (Pake pattern) using a model of random oriented spins with respect to the dipolar axis and Tikhonov regularization as implemented in DEERAnalysis 2006 (34).

Q-band DEER spectra were recorded at 50 K on an Elexsys SuperQ FT-EPR Spectrometer (Bruker) equipped with a CF935 Helium flow-cryostat (Oxford Instruments) and a home-built EPR resonator. The pump and probe frequencies were set at 33.6037 GHz and 33.6237 GHz. The length of the pump pulse was 80 ns and the 90° and 180° detection pulses were of length 40 ns and 80 ns. The times between the pulses were set as follows: τ1 = τ2 = 400 ns, t′ = 1200 ns. The Q-band DEER data were analyzed in the same manner as the X-band data.

The rotamer library approach

To predict orientations and distances of protein-bound R1 side chains, a semidynamic structure model of the spin label was used. In this model, the conformational dynamics of the R1 side chain is represented by a discrete set of possible conformations. These conformational structures of the spin-label side chain, R1, were provided by a so-called rotamer library that contains 98 R1 rotamer structures (28). The sets of possible rotamers on the selected positions were determined by calculating the probability of a particular rotamer to occur in the protein structure (PDB code: 1ST6). Therefore, the free energy for each rotamer was computed considering only van der Waals interactions between atoms of the label and atoms of the protein using a forgive-factor of 0.01 (28). From the rotamer sets of two different residual positions, all possible distance combinations were calculated and their intensities were weighted by the product of the populations of the considered rotamers to determine distance distributions.

MD simulation

Modeled R1 side chains replaced the native amino acids at the investigated positions in the crystal structure (PDB code: 1ST6). The nitroxide side chain was reoriented manually to avoid steric overlaps with neighboring side chains according to the method described earlier (26,27). All MD simulations were performed with the GROMACS simulation suite. The force field ffG43b1 from GROMOS, which is integrated into GROMACS, was used for the simulations in vacuum. Energy minimization was performed by using steepest descent and conjugate gradients algorithms. Several MD simulations were performed with different spin-labeled sites at 600 K in vacuo with position restraints on the backbone atoms. The final production MD runs were carried out for 5 × 106 steps with a time increment of 2 fs. The current structure was saved every 0.2 ps giving a 10-ns trajectory file with 50,000 frames. Distances between nitroxide N-O group centers were extracted from each frame of the trajectory files and used for interspin distance calculations, leading to distance distribution histograms.

Results

Spin-label side-chain mobility

The reorientational motion of the nitroxide spin-label side chain provides information on the immediate environment of the spin-labeled position in the protein analyzed. The motion comprises contributions from 1), the Brownian rotary diffusion of the protein, 2), its backbone dynamics, and 3), bond rotational isomerization within the spin-label side chain. The latter is modulated by the protein secondary and tertiary structure in the vicinity of the spin-label binding site. The overall reorientational motion of the spin label is reflected in the shape of the room temperature cw-EPR spectrum due to the partial averaging of the anisotropic components of the g- and hyperfine tensors. The inverse line width of the central line (ΔH0−1) and the inverse second moment (left angle bracketH2right angle bracket−1) of room temperature EPR spectra were found to be suitable parameters to describe the complicated properties of the spin-label mobility and to provide information about the interaction between the R1 side chain and its environment. From experiments with numerous proteins, correlations between the mobility of R1 and protein topography have been established (35–37).

In all of these proteins, R1 residues are highly immobilized at buried sites, are immobilized or have complex multicomponent spectra at tertiary contact sites, and have intermediate mobility at helix surface sites. The apparent hyperfine splitting and the line widths of Vt901R1, Vt922R1, Vt950R1, Vt957R1, Vt984R1, and Vt1033R1 (Fig. 1 B) indicate site-dependent restrictions of the R1 side-chain mobility (compare to Fig. 1 A for R1 positions). The highest degree of motional restriction is found for Vt957R1, indicating a buried or helix contact site. A slightly higher mobility is revealed for Vt984R1. The EPR spectrum of Vt901R1 shows two components, which can either result from the presence of rotameric subpopulations or Vt conformations differing in their motional restrictions. The largest fraction shows the properties of helix contact sites, with only a small fraction revealing very high dynamics.

The spectral shapes of Vt922R1 and Vt950R1 resemble those of helix surface sites, as revealed by comparison with the extensively studied spectra of surface sites D72R1 and V131R1 of T4 lysozyme (38,39). A slightly higher restriction of the mobility is observed for Vt1033R1. Topological regions in the ΔH0−1 versus left angle bracketH2right angle bracket−1 plot (Fig. 1 C) indicate that Vt922R1, Vt950R1, and Vt1033R1 are located in the area characteristic for helix surface sites in agreement with the above spectral shape comparison. In contrast, Vt957R1 and Vt984R1 show clear features of R1 side chains subject to tertiary contact. This result is consistent with the Vt crystal structure, where the corresponding Cα-Cβ bonds are oriented toward neighboring helices. (R1 side-chain locations on the Vt crystal structure are illustrated in Fig. 1 A.) This applies also for Vt950R1, even though the EPR results suggest that position 950 should be surface-exposed. However, the latter is in line with the fact that position 950 is located close to the N-terminal end of helix three (H3).

Thus, the dynamics of the H3 N-terminal end might contribute to the observed high mobility of the bound spin label. The value of the inverse line width of Vt901R1 is governed by the mobile component that reflects a loop site. In contrast, the value of the inverse second moment is dominated by the immobile spectral component and refers to a tertiary contact or helix surface-related position. In the Vt crystal structure, position 901 is located at helix H1 with the Cα-Cβ bond oriented toward helix H5. A spin label attached at this position would experience tertiary contact to helix H5 and would not exhibit any mobile features in the EPR spectrum. The appearance of an additional mobile component in the EPR spectrum of Vt901R1 suggests that the structure for a small fraction of Vt molecules in aqueous solution differs from that in the crystal.

Intramolecular interspin distances

Interhelical distances were determined for four different Vt double cysteine mutants. In addition to Vt901R1/957R1 with spin labels on distant helices H1 and H3, three Vt variants, Vt922R1/957R1 (H2/H3), Vt901R1/1033R1 (H1/H5), and Vt984R1/1033R1 (H4/H5), were selected to analyze distances between neighboring helices. Schemes of the investigated spin-label positions and the corresponding low temperature (T = 160 K) cw-EPR spectra are shown in Fig. 2. The spectra of Vt922R1/957R1 (H2/H3), Vt901R1/1033R1 (H1/H5), and Vt984R1/1033R1 (H4/H5) exhibit significant broadening, indicating close proximity between the respective spin-labeled side chains. For Vt901R1/1033R1, the fit by Tikhonov regularization using the program SHORT DISTANCES agrees with the experimentally recorded spectrum and reveals a mean distance of 0.8 nm ± 0.1 nm. The simulated spectrum of Vt984R1/1033R1 shows small deviations with respect to the experimental one in the region of the low field maximum. The mean distance, extracted from the calculated distance distribution (0.7 nm ± 0.1 nm), is similar to the result obtained for Vt922R1/957R1 (0.8 nm ± 0.1 nm).

Figure 2
Interspin distances between R1 side chains in doubly spin-labeled Vt. (Left) Populated rotamers for Vt901R1/957R1 (H1/H3), Vt922R1/957R1 (H2/H3), Vt901R1/1033R1 (H1/H5), and Vt984R1/1033R1 (H4/H5) (stick representation) are shown attached to the Vt crystal ...

In general, the agreement between simulations and experiments are satisfactory, the largest deviations being obvious for Vt922R1/957R1. Interspin distances were also determined by using the fitting program DIPFIT, considering a Gaussian-shaped distance distribution (31). Calculated mean distance values are in agreement with those obtained by the program SHORT DISTANCES. However, in the interpretation of these data it has to be considered that the assumption of an isotropic distribution of nitroxide orientations might not be valid, because the R1 orientations might be restricted due to interactions with neighboring protein atoms (see below). Mobility measurements and RLA (see below) indicate motionally restricted and thus orientationally correlated spin-label pairs. Additionally, in the distance range discussed here, Heisenberg spin exchange is relevant (40), which depends on the mutual R1 orientation as well. Because Heisenberg exchange interaction is not considered in both of the applied fitting programs, the fitted interspin distance values might be understood as a lower limit. On the other hand, the spectra of Vt922R1/957R1, Vt901R1/1033R1, and Vt984R1/1033R1 are significantly broader than the spectrum of crystallized spin-labeled insulin (B2SL) where the nitroxide groups are 1 nm apart (31). Therefore, this value defines the upper distance limit for these three double mutants.

Dipolar evolution functions obtained for Vt901R1/957R1 (H1/H3) by Q- and X-band DEER measurements show six well-resolved modulation maxima within the time domain of 2200 ns (X-band) (Fig. 2) (see also Fig. S4). The distance distributions calculated using Tikhonov regularization of the X- and Q-band data yield mean values of 2.65 nm. The full width at half-maximum of the main distance peak of 0.2 nm suggests strong restriction of the reorientation of both spin-label side chains, providing evidence for specific spin-label orientations with respect to the dipolar axis (see below).

Modeling of spin-label side-chain orientations: RLA and MD simulations

Spin-label side-chain conformations for the spin-labeled sites of the Vt crystal structure were obtained using the RLA (see Materials and Methods). This approach yields only one significantly populated spin-label rotamer for positions 1033 and 922, whereas up to seven different rotamer conformations can be found for the other studied positions. For double mutants Vt901R1/1033R1, Vt922R1/957R1, and Vt984R1/1033R1, where clashes between R1 side chains can occur, the computation was done with the Vt1033R1 or Vt922R1 side chains oriented according to the previously found rotamers. For Vt984R1/1033R1, the number of populated rotamers for Vt984R1 (H4) is reduced from seven to three, indicating that the R1 side chain bound at position 1033 (H5) restricts the space accessible for R1 at position 984 (H4). For Vt901R1 (H1), three significantly populated rotamers were found in both absence and presence of R1 at position 1033 (H5).

The number of populated rotamers (two) for Vt957R1 (H3) does not change in absence or presence of the R1 side chain at position 922 (H2). Thus, the motional freedom of R1 on positions 957 (H3) and 901 (H1) is not influenced by a spin-label side chain at positions 922 (H2) or 1033 (H5). However, the total number of populated rotamer conformations is small in each case, highlighting the restrictions due to interactions with protein and/or other adjacent R1 side-chain atoms. The rotamers exhibiting the highest population of each position are shown in Fig. 2. Distances of N-O group centers were calculated and weighted with the populations of the corresponding rotamers. The distance distribution of Vt901/957R1 exhibits a mean distance of 2.8 nm and a full width at half-maximum of 0.1 nm. Vt922R1/957R1 (H2/H3) and Vt984R1/1033R1 (H4/H5) exhibit distance distributions with a mean distance below 1.0 nm, whereas Vt901R1/1033R1 (H1/H5) shows distances between 1.0 and 1.2 nm.

In addition to the RLA, the accessible space for the spin-label side chains was characterized by MD simulation trajectories of 10-ns length calculated at 600 K. Because the protein backbone was fixed, only the motional freedom of the side chains contributes to the observed interspin distance distributions shown in Fig. 2. The mean distance between N-O groups of Vt901R1/957R1 yields 2.73 nm. The small distribution width of 0.16 nm reflects the restricted flexibility of 901R1 and 957R1. From the distance distributions of Vt922R1/957R1 (H2/H3), Vt901R1/1033R1 (H1/H5), and Vt984R1/1033R1 (H4/H5), mean distances below 1.3 nm were obtained. Vt922R1/957R1 (H2/H3) reveals a narrow distance distribution with a mean distance of 0.95 nm. For Vt984R1/1033R1 (H4/H5), two distinct populations were observed with the most populated state at a mean distance of 0.73 nm. In fact, the MD trajectories suggest two distinct orientations for Vt1033R1 (data not shown). The distance distribution for Vt901R1/1033R1 (H1/H5) exhibits a mean distance of 1.27 nm, with a distribution width of 0.22 nm, indicating a higher motional freedom for Vt901R1 and Vt1033R1 compared to the other spin-label side chains studied.

Intermolecular interspin distances

It was previously shown that wild-type Vt can self-associate with monomer/dimer Kd values of ~300 μM (15). Monomer concentrations at ~100 μM as used in our DEER experiments would result in formation of ~25% of dimers. In fact, the DEER dipolar evolution functions of Vt singly spin-labeled mutants (Fig. 3) reveal dipolar modulations, indicating dipolar interaction due to Vt dimerization. The particular modulation depths vary between 0.1 and 0.2. This corresponds to a number of interacting spins of between 1.1 and 1.3, which reflects the fraction of dimers in the sample to be in the range from 10% to 30% in agreement with the above-mentioned Kd.

Figure 3
DEER spectroscopy of singly spin-labeled Vt constructs. (Left) Dipolar evolution functions (shaded) and the corresponding fits calculated with DEERAnalysis (solid). (Right) Experimental distance distributions, P(r), obtained by DEERAnalysis (solid) are ...

The distance distributions are well defined (Fig. 3). Additional small contributions in the distance plots occur most likely due to the low signal/noise ratio, especially in the cases of Vt922R1 (H2), Vt950R1 (H3), and Vt1033R1 (H5). For Vt922R1 and Vt957R1, the applied evolution time allows reliable distance quantification up to ~5 nm. Thus, the distances in these cases might be slightly larger than those obtained by Tikhonov regularization. The interspin distances for Vt1033R1 and those for Vt1033R1 within F-actin bundles (VtA1033R1) differ by >3 nm, pointing to distinct dimer conformations of Vt in the presence and absence of F-actin.

Determination of the molecular alignment within the Vt dimer

NMR studies (15) as well as the DEER measurements on singly labeled Vt domains performed in this study provide evidence for the presence of dimeric Vt in aqueous solution even in the absence of interaction partners. Experimental interspin distances of the helix-bound R1 side chains and calculated internitroxide distances using the RLA were used to determine the relative alignment of two Vt molecules. For spin-labeled sites of T4 lysozyme, it has been shown that calculated rotamers match the location of the x-ray rotamers reasonably well, although spin-label rotamer ensembles were predicted instead of the single rotamer seen in the x-ray structure (41). In addition, backbone flexibility is neglected in the RLA. Hence, in our comparison, relative populations of the different rotamers and thus the shape of the calculated distance distribution were disregarded, and only the calculated range of interspin distances accessible for the R1 side chains due to sterically allowed rotamers is being discussed.

In Fig. 3, results of the RLA calculations based on the dimer found in the Vt crystal structure (1QKR) are shown and compared to the experimental data. The distance vectors of respective spin-labeled sites are depicted in Fig. 4. The internitroxide distances calculated for Vt950R1, Vt957R1, Vt984R1, and Vt1033R1 of the Vt dimer crystal structure agree with the experimental data. For Vt922R1, the calculated interspin distances are located at the upper limit of the distance range accessible in our experiment (5.3 nm). At this limit, the population of the experimentally determined distance distribution is generally underestimated. With this in mind, calculated and experimental data for Vt922R1 do not disagree. A significant deviation between calculated and experimental distance distributions is obvious for Vt901R1. Although populated rotamers contribute to interspin distances of between 2.6 and 2.8 nm, these values are larger than the experimental value of 2.3 nm (Fig. 3). To accommodate the experimental data, helices H1 of both monomers have to be slightly rotated and shifted into the direction of the dimerization interface by 0.2–0.3 nm each. Such a shift decreasing the distance between positions 901 and 1033 within the monomers by 0.2–0.3 nm is well in agreement with the experimental data determined for the doubly labeled Vt901R1/1033R1 construct (see Fig. 2).

Figure 4
Crystal structure of the Vt dimer (1QKR) showing interspin distance vectors that connect Cβ positions of the spin-labeled sites used in this study.

A grid search with systematic variation of the relative Vt monomer orientations within the dimer was performed to investigate whether the solution discussed above is unique (see Materials and Methods). All calculations were performed with and without N- and C-terminal domains included in the clash calculations, to consider a potential rearrangement of the strap and the C-terminus upon Vt dimerization. The best fitting solution reveals a crosslike upside-down alignment of the two Vt molecules, dimer model 2 (see Fig. S5). In this model, one Vt molecule shows primarily involvement of helix H5 in the interaction surface, together with parts of helices H1 and H4. The contact region of the Vt partner molecule consists of the N-terminal strap, which is adjacent to helix H1, and a major part of helix H1 itself. This Vt dimer configuration thus differs significantly from that of the crystal structure. In model 2, the agreement of calculated interspin distance distributions for Vt922R1, Vt950R1, Vt957R1, Vt984R1, and Vt1033R1 with the experimental data is of similar quality to that obtained for the crystal dimer (see Fig. S6). For Vt901R1, however, the calculated mean interspin distance (2.1 nm) is less than the experimental one (2.3 nm), and a shift of helix H1, as required to accommodate the experimental data of Vt901R1/1033R1, would further increase the discrepancy.

In principle, selection of the most probable of these two dimer models could be done with the help of computational structure prediction as demonstrated recently (42). However, the disagreement of dimer model 2 with published NMR results (15), which do not show chemical shift perturbations in the strap nor in the C-terminal end of helix H5, and the agreement between our experimental data and the crystal structure, render this dispensable.

Discussion

Vinculin plays an important role in cell adhesion and migration modulating the anchorage of the actin cytoskeleton to the cell adhesion complex. In this respect, control of vinculin tail conformation and dimerization appears to be crucial for vinculin function. The crystal structure of Vt reveals a homodimer with each monomer consisting of an antiparallel bundle of five amphipathic helices (H1–H5) (5). NMR studies (15) provide further evidence for the presence of dimeric Vt in aqueous solution in the absence of interaction partners. In cells, however, activation of vinculin requires combinatorial input of several binding partners, restricting vinculin ligand binding to sites of cell adhesion (1,9,10,43). The Vt domain is released from its head domain only in close proximity to head and tail binding partners, which together determine the conformation (or allosteric state) of the vinculin protein. Therefore, the free vinculin tail as studied by biochemical and biophysical methods in the absence of binding partners, does not exist in cells, but may rather represent a transient intermediate state between interactions with competing Vt ligands, i.e., vinculin head, F-actin, and acidic phospholipids. Consistently, formation of Vt dimers as required for example in actin filament bundling, was proposed to be facilitated through interaction of Vt domains via sterically accessible dimerization surfaces (13). Vinculin behavior in cell adhesions highlights the biological relevance of Vt dimers.

To investigate the structure of monomeric and dimeric Vt in solution and to set the basis for an elucidation of the suggested difference in the conformations of dimeric Vt in the absence and presence of F-actin, we applied cw- and pulse EPR methods. Spin-label mobility and intramolecular interspin distances between R1 side chains bound to selected positions were studied. The analysis of the spin-label mobility does not provide any evidence for a significant difference between solution and crystal structure of Vt. MD simulation and rotamer calculations were performed to compare both computational methods and relate their predictive power to interspin distance results. For Vt901R1/957R1, as a probe for the relative arrangement of helices H1 and H3 within the Vt monomer, experimental and calculated interspin distances match nicely. For Vt922R1/957R1 and Vt984R1/1033R1, the distance distributions determined by MD simulation and by RLA resemble each other and result in interspin distances below 1 nm.

In this distance regime, Heisenberg exchange coupling contribute to the EPR spectral shape, which may interfere with the determination of the interspin distance. However, our experimental data provide strong evidence for interspin distances distributions centered below 1 nm, indicating the arrangement of H2-H3 and H4-H5 in solution being similar to the Vt crystal structure. For Vt901R1/1033R1, the distance ranges obtained by RLA and MD simulations are similar; however, the populations of rotamers calculated by the two methods differ. This discrepancy could be explained by the fact that in MD simulations protein side chains are allowed to reorient upon interactions with the R1 side chains, whereas all protein atoms were fixed in the RLA. Comparison of the dipolar broadened cw EPR spectrum with that of spin-labeled insulin B2SL (31) reveals an interspin distance value below 1 nm that deviates from the values simulated for the Vt crystal structure. Hence, H1 has to be marginally rotated and shifted in the direction of helix H5 to enable the convergence of the two R1 side chains. Taken together, the results for the here-investigated doubly labeled Vt constructs indicate that the five-helical-bundle arrangement proposed by the crystal structure of Vt is largely preserved in an aqueous environment.

DEER measurements on Vt single mutants show clear evidence for a Vt dimerization in solution. The fraction of dimers found for the concentrations used is in agreement with the previously determined Kd of 300 μM (15). Interspin distance distributions determined for the dimer in the crystal (1QKR) fit the experimental data, if the above-mentioned shift of H1 is taken into consideration. A grid search of monomer orientations, which fulfill the experimental distance constraints, shows that the structure of the dimer found in the crystal is one of two structures yielding reasonable agreement between calculated and experimental interspin distances. Previous NMR chemical shift data (15) revealed parts of helices H4 and H5 to be involved in the dimerization interface, in agreement with the dimer interface identified by x-ray crystallography.

Furthermore, EPR data available as of this writing show that the relative arrangement of helices H1, H2, and H3 in the Vt dimer agrees with that found in the crystal. Otherwise, a model of F-actin cross-linking by Vt, derived from electron microscopy and computational docking, suggested a dimer distinct from that of pure Vt observed by NMR and x-ray crystallography (13,15). The distinct interspin distance distributions determined for positions 1033 of the dimer in the absence (Vt1033R1) and presence of F-actin (VtA1033R1) reveal that in the F-actin bound state the distance between these sites is increased by >3 nm. Consistent with the EM-based model (13), the observed distance distribution excludes concomitant involvement of the helices H5 of both monomers in the dimerization interface of Vt dimers bound to F-actin. This result proves the usefulness of our approach for further elucidation of the Vt-dimer/F-actin complex at the level of the backbone fold.

Analysis of vinculin dynamics in adhesion sites (2,11,44) suggests a continuous competition of different binding partners, which requires fast and reversible modulation of Vt domain ligand binding. The interaction surface of the Vt dimer in solution sets an example for the homotypic interaction of Vt domains and should be related to the dimers interacting with F-actin and acidic phospholipids, as the vinculin protein shuttles between different binding states (of the tail) and inactivation (1). Therefore, the Vt dimer in solution is considered one of the biologically relevant Vt interaction structures. It is furthermore instrumental to the currently performed analysis of 1), F-actin and 2), lipid-induced Vt dimers, which together will promote elucidation of the vinculin conformations that modulate and control the anchorage of the actin cytoskeleton.

Conclusion

Combining experimental and computational techniques of high-resolution distance measurement, we observe that the crystal structure of monomeric Vt is essentially maintained in an aqueous solution, except for a small rearrangement of helices H1 and H5, which brings residues 901 (H1) and 1033 (H5) in closer proximity. The reasonable agreement of the computational results for several positions shows that MD simulation and the computation time-saving rotamer library approach are suitable techniques to predict spin-label side-chain orientations and flexibility. The combination of EPR spectroscopy and computational methods enabled the determination of the Vt dimer configuration in a near-physiological environment.

Acknowledgments

We thank Dr. Ulrike Hinssen (Hypermol, Bielefeld, Germany) for many useful discussions and Daniel Klose for additional RLA calculations. Prof. Wolfgang Lubitz (Max Planck Institute, Mülheim, Germany) is thanked for providing us with the opportunity to perform the Q-band DEER measurements. We also thank Dr. Yevhen Polyhach and Prof. Gunnar Jeschke (The Swiss Federal Institute of Technology, Zürich, Switzerland) for many fruitful discussions and for the professional introduction into the rotamer library approach.

We gratefully acknowledge financial support from the Deutsche Forschungsgemeinschaft (grant No. STE640/7 to H.-J.S. and grant No. IL43/2-4 to S.I.), The Interdisciplinary Centre for Clinical Research, Leipzig (grant No. N04 to W.H.Z.), and the PhD program of the State of Lower Saxony in Osnabrück.

Supporting Material

Document S1. Additional narrative with six figures:

References

1. Ziegler W.H., Liddington R.C., Critchley D.R. The structure and regulation of vinculin. Trends Cell Biol. 2006;16:453–460. [PubMed]
2. Humphries J.D., Wang P., Ballestrem C. Vinculin controls focal adhesion formation by direct interactions with talin and actin. J. Cell Biol. 2007;179:1043–1057. [PMC free article] [PubMed]
3. Mierke C.T., Kollmannsberger P., Fabry B. Vinculin facilitates cell invasion into three-dimensional collagen matrices. J. Biol. Chem. 2010;285:13121–13130. [PMC free article] [PubMed]
4. Bakolitsa C., Cohen D.M., Liddington R.C. Structural basis for vinculin activation at sites of cell adhesion. Nature. 2004;430:583–586. [PubMed]
5. Bakolitsa C., de Pereda J.M., Liddington R.C. Crystal structure of the vinculin tail suggests a pathway for activation. Cell. 1999;99:603–613. [PubMed]
6. Gilmore A.P., Burridge K. Regulation of vinculin binding to talin and actin by phosphatidyl-inositol-4-5-bisphosphate. Nature. 1996;381:531–535. [PubMed]
7. Hüttelmaier S., Mayboroda O., Rüdiger M. The interaction of the cell-contact proteins VASP and vinculin is regulated by phosphatidylinositol-4,5-bisphosphate. Curr. Biol. 1998;8:479–488. [PubMed]
8. Steimle P.A., Hoffert J.D., Craig S.W. Polyphosphoinositides inhibit the interaction of vinculin with actin filaments. J. Biol. Chem. 1999;274:18414–18420. [PubMed]
9. Chen H., Choudhury D.M., Craig S.W. Coincidence of actin filaments and talin is required to activate vinculin. J. Biol. Chem. 2006;281:40389–40398. [PubMed]
10. Cohen D.M., Kutscher B., Craig S.W. A conformational switch in vinculin drives formation and dynamics of a talin-vinculin complex at focal adhesions. J. Biol. Chem. 2006;281:16006–16015. [PubMed]
11. Chandrasekar I., Stradal T.E.B., Ziegler W.H. Vinculin acts as a sensor in lipid regulation of adhesion-site turnover. J. Cell Sci. 2005;118:1461–1472. [PubMed]
12. Palmer S.M., Playford M.P., Campbell S.L. Lipid binding to the tail domain of vinculin: specificity and the role of the N and C termini. J. Biol. Chem. 2009;284:7223–7231. [PMC free article] [PubMed]
13. Janssen M.E.W., Kim E., Hanein D. Three-dimensional structure of vinculin bound to actin filaments. Mol. Cell. 2006;21:271–281. [PubMed]
14. Hüttelmaier S., Bubeck P., Jockusch B.M. Characterization of two F-actin-binding and oligomerization sites in the cell-contact protein vinculin. Eur. J. Biochem. 1997;247:1136–1142. [PubMed]
15. Palmer S.M., Schaller M.D., Campbell S.L. Vinculin tail conformation and self-association is independent of pH and H906 protonation. Biochemistry. 2008;47:12467–12475. [PubMed]
16. Altenbach C., Marti T., Hubbell W.L. Transmembrane protein structure: spin labeling of bacteriorhodopsin mutants. Science. 1990;248:1088–1092. [PubMed]
17. Altenbach C., Yang K., Hubbell W.L. Structural features and light-dependent changes in the cytoplasmic interhelical E-F loop region of rhodopsin: a site-directed spin-labeling study. Biochemistry. 1996;35:12470–12478. [PubMed]
18. Bordignon E., Steinhoff H.-J. Membrane protein structure and dynamics studied by site-directed spin labeling ESR. In: Hemminga M.A., Berliner L.J., editors. Biological Magnetic Resonance 27—ESR Spectroscopy in Membrane Biophysics. Springer Science and Business Media; New York: 2007. pp. 129–164.
19. Pfeiffer M., Rink T., Steinhoff H.J. Site-directed spin-labeling reveals the orientation of the amino acid side-chains in the E-F loop of bacteriorhodopsin. J. Mol. Biol. 1999;287:163–171. [PubMed]
20. Plato M., Steinhoff H.-J., Möbius K. Molecular orbital study of polarity and hydrogen bonding effects on the g and hyperfine tensors of site directed NO spin labeled bacteriorhodopsin. Mol. Phys. 2002;100:3711–3721.
21. Altenbach C., Kusnetzow A.K., Hubbell W.L. High-resolution distance mapping in rhodopsin reveals the pattern of helix movement due to activation. Proc. Natl. Acad. Sci. USA. 2008;105:7439–7444. [PMC free article] [PubMed]
22. Jeschke G., Bender A., Paulsen H. Localization of the N-terminal domain in light-harvesting chlorophyll a/b protein by EPR measurements. J. Biol. Chem. 2005;280:18623–18630. [PubMed]
23. Jeschke G., Polyhach Y. Distance measurements on spin-labeled biomacromolecules by pulsed electron paramagnetic resonance. Phys. Chem. Chem. Phys. 2007;9:1895–1910. [PubMed]
24. Mchaourab H.S., Perozo E. Determination of protein folds and conformational dynamics using spin labeling EPR spectroscopy. In: Berliner L.J., Eaton S.S., Eaton G.R., editors. Distance Measurements in Biological Systems by EPR. Kluwer; New York: 2000. pp. 155–218.
25. Wegener A.A., Klare J.P., Steinhoff H.J. Structural insights into the early steps of receptor-transducer signal transfer in archaeal phototaxis. EMBO J. 2001;20:5312–5319. [PMC free article] [PubMed]
26. Beier C., Steinhoff H.-J. A structure-based simulation approach for electron paramagnetic resonance spectra using molecular and stochastic dynamics simulations. Biophys. J. 2006;91:2647–2664. [PMC free article] [PubMed]
27. Gajula M.N.V.P., Borovykh I.V., Steinhoff H.-J. Spin-labeled photosynthetic reaction centers from Rhodobacter sphaeroides studied by electron paramagnetic resonance spectroscopy and molecular dynamics simulations. Appl. Magn. Reson. 2007;31:167–178.
28. Polyhach Y., Godt A., Jeschke G. Spin pair geometry revealed by high-field DEER in the presence of conformational distributions. J. Magn. Reson. 2007;185:118–129. [PubMed]
29. Böhme S., Padmavathi P.V., Steinhoff H.J. Topology of the amphipathic helices of the colicin A pore-forming domain in E. coli lipid membranes studied by pulse EPR. Phys. Chem. Chem. Phys. 2009;11:6770–6777. [PubMed]
30. Ziegler W.H., Tigges U., Jockusch B.M. A lipid-regulated docking site on vinculin for protein kinase C. J. Biol. Chem. 2002;277:7396–7404. [PubMed]
31. Steinhoff H.-J., Radzwill N., Wollmer A. Determination of interspin distances between spin labels attached to insulin: comparison of electron paramagnetic resonance data with the x-ray structure. Biophys. J. 1997;73:3287–3298. [PMC free article] [PubMed]
32. Chiang Y.W., Zheng T.Y., Horng J.C. Determination of interspin distance distributions by cw-ESR is a single linear inverse problem. Biophys. J. 2009;97:930–936. [PMC free article] [PubMed]
33. Pannier M., Veit S., Spiess H.W. Dead-time free measurement of dipole-dipole interactions between electron spins. J. Magn. Reson. 2000;142:331–340. [PubMed]
34. Jeschke G., Chechik V., Jung H. DeerAnalysis2006—a comprehensive software package for analyzing pulsed ELDOR data. Appl. Magn. Reson. 2006;30:473–498.
35. Hubbell W.L., Mchaourab H.S., Lietzow M.A. Watching proteins move using site-directed spin labeling. Structure. 1996;4:779–783. [PubMed]
36. Mchaourab H.S., Lietzow M.A., Hubbell W.L. Motion of spin-labeled side chains in T4 lysozyme. Correlation with protein structure and dynamics. Biochemistry. 1996;35:7692–7704. [PubMed]
37. Isas J.M., Langen R., Hubbell W.L. Structure and dynamics of a helical hairpin and loop region in annexin 12: a site-directed spin labeling study. Biochemistry. 2002;41:1464–1473. [PubMed]
38. Columbus L., Kálai T., Hubbell W.L. Molecular motion of spin labeled side chains in α-helices: analysis by variation of side chain structure. Biochemistry. 2001;40:3828–3846. [PubMed]
39. Zhang Z., Fleissner M.R., Freed J.H. Multifrequency electron spin resonance study of the dynamics of spin labeled T4 lysozyme. J. Phys. Chem. B. 2010;114:5503–5521. [PMC free article] [PubMed]
40. Rabenstein M.D., Shin Y.-K. Determination of the distance between two spin labels attached to a macromolecule. Proc. Natl. Acad. Sci. USA. 1995;92:8239–8243. [PMC free article] [PubMed]
41. Polyhach Y., Bordignon E., Jeschke G. Rotamer libraries of spin labeled cysteines for protein studies. Phys. Chem. Chem. Phys. 2011;13:2356–2366. [PubMed]
42. Hirst S.J., Alexander N., Meiler J. RosettaEPR: an integrated tool for protein structure determination from sparse EPR data. J. Struct. Biol. 2011;173:506–514. [PMC free article] [PubMed]
43. Golji J., Lam J., Mofrad M.R. Vinculin activation is necessary for complete talin binding. Biophys. J. 2011;100:332–340. [PMC free article] [PubMed]
44. Chen H., Cohen D.M., Craig S.W. Spatial distribution and functional significance of activated vinculin in living cells. J. Cell Biol. 2005;169:459–470. [PMC free article] [PubMed]

Articles from Biophysical Journal are provided here courtesy of The Biophysical Society
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • MedGen
    MedGen
    Related information in MedGen
  • PubMed
    PubMed
    PubMed citations for these articles
  • Substance
    Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...