Nanoscale dynamics of the cadherin–catenin complex bound to vinculin revealed by neutron spin echo spectroscopy

NSE Analysis.

In solution, protein dynamics occurs in the low Reynolds number Brownian dynamics regime. Here, dynamics depends upon the size of the object but not its mass (50). The NSE results are thus best analyzed in terms of the mobility tensors H. We generalize the remarkable Akcasu–Gurol formula (51) to include rotational diffusion (3). The effective diffusion coefficient D_eff(q) for a rigid body or for a body of multiple modules or domains is calculated as:

where k_BT is the temperature factor and b_j is the scattering length of a subunit j (b_j is taken without loss of generality as 1 for the hydrogenated proteins and 0 for the deuterated proteins). L_j = r_i × q is the torque vector for each atom, H^T is the translational mobility tensor (defining the velocity response v to a given applied force F by v = H^TF), and H^R is the rotational mobility tensor (defining the angular velocity response ω to a given applied torque τ by ω = H^Rτ).

It is essential to note that the mobility tensors define internal molecular motion. For a rigid body, all pairs of atoms are included in the numerator in Eq. 3. When a protein or protein complex (such as the VABE complex) is parsed into different separately moving modules or domains, only pairs of atoms both contained within a given module or domain are included in the numerator. The diffusion constant per atom is chosen uniformly for a given parsing for the fully hydrogenated complex so that D_eff(q = 0) equals the experimentally measured diffusion constant D₀ of the complex. This value of the diffusion constant per atom is retained for the deuterated proteins. The mobility tensors are then calculated using a given hypothesized parsing of the complex into separately moving modules, the experimentally measured diffusion constant D₀, and the molecular structural coordinates of the complex. The overall strategy then is to propose a hypothetical parsing of the complex into modules, calculate (via Eq. 3) the effective diffusion constant D_eff(q) for each hypothetical parsing, and test the hypotheses by comparing the results for each parsing with the experimental NSE data. Further details can be found in our reviews (12, 13, 47, 52) and are summarized in SI Appendix, Supplementary Materials 2.

We first parsed the VABE complex into a single rigid body and then, progressively two, three, and four module models. The models are displayed in Figs. 3 and 4. In Fig. 3, we show the parsing into three modules, with each module being a protein that builds up the complex. Module 1 is the entire protein vinculin, module 2 is the entire protein β-catenin, and module 3 is the entire protein α-catenin. In Fig. 3A, the three modules move in unison—thus, the whole complex is taken as a rigid body. In Fig. 3 B–D, two modules move in unison, while the third moves separately, as indicated by the colors in the figures. In Fig. 3E, all of the three modules move independently but are connected by soft spring linkers. The resultant D_eff(q) for each of the five models of the fully hydrogenated complex is shown in Fig. 3F. The model displayed in Fig. 3E with all three modules moving independently fits the experimental data best.

Our SANS studies (44) show that the ensemble-averaged structure of the VABE complex resembles a humanoid motorman with two flexible legs and a trunk mounted on a soft spring linker, Figs. 1A and 4. We next consider a parsing of the “motorman” VABE complex into four modules, as shown in Fig. 4, which are different from the 3-module parsing shown in Fig. 3:

Module 1 (hip of motorman), shown in blue in Fig. 4D, includes the V_h domain of vinculin, the VBS of α-catenin, and the M2M3 subdomains of α-catenin.

Module 2 (trunk of motorman), shown in cyan in Fig. 4D, includes β-catenin, EcadCT, the α-catenin N1N2 subdomains, and the linker between α-catenin N2 and α-catenin VBS.

Module 3 (one leg of motorman), shown in green in Fig. 4D, includes the linker between α-catenin M domain and α-catenin ABD, and the α-catenin ABD.

Module 4 (2nd leg of motorman) shown in magenta in Fig. 4D is the linker between V_h and V_t, and the vinculin ABD or V_t domain.

We performed calculations for the following models: 1) treating the complex as rigid, with modules 2, 3, and 4 moving in unison with module 1, which is the same rigid-body model shown in Fig. 3A; 2) two modules taken as moving in unison, with the other two modules moving independently, Fig. 4 A–C; and 3) all four modules moving independently, Fig. 4D. Fig. 4E displays the calculated D_eff(q) for the fully hydrogenated complex for the rigid-body and all hypothetical models shown in Fig. 4 A–D and shows that model D with all four modules moving independently fits the NSE data best. Our calculations thus show that the rigid body and two-module models are clearly excluded by the NSE data, However, the 3-module and the 4-module motorman models both fit the NSE data well. Additional calculations on the selectively deuterated complexes confirm that the 3-module and 4-module models both fit the experimental data, SI Appendix, Fig. S5. Thus, further subdivision is of limited utility.

Comparing the calculations with the NSE data suggests that the VABE complex is highly dynamic with activated motions in alpha-catenin, β-catenin, and in vinculin. Notably, the NSE measurements, interpreted by our calculations, reveal that upon recruiting vinculin to the cadherin–catenin complex, the motion of the entire α-catenin is activated. This is in stark contrast to the α-catenin homodimer, in which our previous NSE study showed that only the small disordered C-terminal tail attached to α-catenin ABD is moving, while the majority of α-catenin behaves as a rigid body (9). The activated dynamic motion of the whole α-catenin as part of the VABE complex is also strikingly different from that of α-catenin as part of the ABE complex, in which our NSE study revealed that the motion of the larger α-catenin ABD is activated (10).

The Cadherin–Catenin–vinculin Complex Is a Three-Way Motorman Entropic Spring That Is Flexible and Elastic.

Alpha-catenin and vinculin are force transducers between the cell adhesion sites and the actin cytoskeleton (24, 53). In-cell single-molecule fluorescence resonance energy transfer (smFRET) studies have demonstrated that α-catenin and vinculin undergo cycles of rapid and repeated conformational changes in response to altered tensions at cell–cell junctions (45), and at cell–matrix focal adhesion sites (36), respectively. Furthermore, a study by Shoyer et al. has revealed that the “loadable” or open conformation of vinculin is recruited to cell–cell junctions during collective cell migration (54). Perturbing the ability of vinculin to change its conformation, by phosphorylation mimetics at the vinculin V_t domain, alters the speed and long-range coordination of collective cell migration. In one of our previous studies using NSE (7), we noticed a related effect, in that phosphorylation mimetics changed the nanoscale dynamics of an adapter protein NHERF1 (but not its structure or conformation), which altered NHERF1 binding kinetics to Ezrin (7).

Collectively these smFRET studies suggest that α-catenin and vinculin are flexible and elastic as force transducers—so that these force transducers are entropic springs. Notably, the smFRET studies revealed the conformational changes in the linker region preceding the actin-binding modules of α-catenin and vinculin (36, 45), which correspond to the leg modules of the motorman shown in Fig. 4. The conformational changes measured by the above smFRET are on longer time scales than the nanoscale dynamics measured by our NSE experiments. Nevertheless, nanoscale dynamics inspires the longer temporal conformational dynamics observed in single-molecule experiments (55).

Vinculin is recruited to the AJ when cells are under increased mechanical tension. Our NSE analysis reveals that the vinculin-bound AJ complex is more dynamic than the core AJ cadherin–catenin complex alone. As a result, the vinculin-bound AJ is more flexible and more elastic to meet the demand of recurrent and enhanced cell–cell junction tensions. Flexibility and elasticity are important physical traits of cellular force transducers (56). Here, our dynamics analysis of the VABE complex details the modes of activated nanoscale motions in these force transducers as they combine to form the cadherin–catenin–vinculin complex. The activated nanoscale protein motions in the motorman-like VABE complex provide the molecular dynamics basis for the cadherin–catenin–vinculin complex to be flexible and elastic (57, 58), so as to respond to and to transduce a multitude of cell–cell junction forces, such as during collective cell migration.

Studies have demonstrated that α-catenin as part of the cadherin–catenin complex does not bind to the F-actin microfilament under zero force (26) and that mechanical tension is required to enable α-catenin as part of the cadherin–catenin complex to bind the F-actin microfilament (25). We previously showed that activated nanoscale domain motion in α-catenin ABD (46) as part of the ABE complex is correlated with increased conformational entropy in the ABE complex (10). We proposed that this increased conformational entropy serves as a negative allosteric regulatory mechanism, which impedes the α-catenin ABD as part of the cadherin–catenin complex from binding to F-actin under zero force. Mechanical tension can reduce the conformational entropy in α-catenin as part of the cadherin–catenin complex (25), so that the cadherin–catenin complex can bind to an F-actin microfilament under mechanical tension.

Upon recruiting vinculin to the cadherin–catenin complex, here, we find that α-catenin becomes even more dynamic, with activated dynamic motions in the whole α-catenin protein. This finding corroborates our previous biochemical and structural study that α-catenin as part of the VABE complex does not bind actin microfilaments under zero force, but the VABE complexes employ vinculin as the major actin-binding mode (44). The elevated nanoscale dynamics in the entire α-catenin as part of the VABE complex may contribute additional conformational entropy that further negatively regulates (or impedes) α-catenin from binding to actin microfilament, thus requiring an even larger magnitude of force for α-catenin to bind actin microfilament. Nevertheless, once bound to F-actin at high tension, the flexible α-catenin and vinculin in the VABE complex may have prolonged lifetimes bound to the cytoskeletal actin, which can then endure high tension before dissociating from the cytoskeletal actin microfilament. Thus, in the vinculin-bound AJ complex, the enhanced dynamics of the α-catenin is likely tuned for sensing and transducing higher magnitude forces.

Our NSE analysis shows that in the VABE complex, the activated nanoscale dynamics span the motorman trunk, hip, and the two legs, Fig. 4. Recent studies found that vinculin is essential for sustaining a “normal level of endogenous forces” at the AJs (59). The motorman model of the VABE complex provides a structural dynamics explanation for this alternative β-catenin-vinculin interaction: Under low force, the cadherin–catenin–vinculin complex employs the vinculin motorman leg to bind cytoskeletal actin, without the α-catenin leg participating in actin binding. The motorman trunk module that is composed of β-catenin and EcadCT transmits the force from the vinculin leg to the extracellular domain of cadherin. At high tension, both the motorman vinculin and α-catenin legs interact with cytoskeletal actin, and the motorman trunk module (β-catenin and EcadCT) integrates and transmits the forces from both legs to the extracellular domain of cadherin.

Studies find that β-catenin binds to vinculin directly (60). Our structural analysis indicates that in the VABE complex, a fraction of β-catenin has extended conformation with the N-terminal tail in contact to the V_h domain of vinculin (44). Such interaction between V_h and β-catenin N-terminal tail may contribute additively to the motion of the motorman trunk module

Vinculin, β-catenin, and α-catenin are critical adapter proteins that couple the AJs to the cytoskeletal actin of neighboring cells. The nanoscale dynamics of α-catenin, vinculin, and β-catenin in the cadherin–catenin–vinculin complex, as revealed in the NSE studies, inspires the longer timescale motions that are observed by smFRET (36, 45, 54). The control of protein flexibility and elasticity underscores the importance of nanoscale dynamics of the AJ complex in sensing, bearing, and transmitting mechanical tensions of multiple magnitudes from the cellular cytoskeleton to the transcellular cell–cell adhesion sites. Our motorman modular model of nanoscale dynamics of the VABE complex provides the structural dynamics mechanism to explain the force transmission functions of the AJ as vinculin is recruited to the AJ.

Protein Purification and Protein Complex Reconstitution.

The bacterial expression and fast protein liquid chromatography purification of full-length human αE-catenin, human β-catenin, the cytoplasmic domain of human EcadCT (residues 731 to 882), and the full-length human vinculin autoinhibition-disruption mutant VK3E have been described previously (44, 46). Deuterated αE-catenin, β-catenin, VK3E, and EcadCT were grown in 85% D₂O (vol/vol) M9 medium following the protocols described in refs. 47, and 48. The purification of deuterated proteins was the same as that of the hydrogenated proteins. Reconstitution of the hydrogenated ^hA^hB^hE and selectively deuterated ^hA^dB^dE complexes followed the same protocol as we described previously (46). Reconstitution of the fully hydrogenated ^hV^hA^hB^hE and selectively deuterated ^hV^dA^dB^dE and ^dV^hA^dB^dE complex was described in a previous publication (44).

Before the NSE experiments, a phosphate-buffered saline (PBS) tablet (ThermoFisher, Waltham, MA) for making 500 mL 1× PBS buffer was soaked in 2 mL 99.9% D₂O and vacuum dried at 80 °C for five cycles to exchange the H in the tablet to D. The D-exchanged PBS tablet was added to 500 mL 99.9% D₂O to make the D₂O PBS buffer. The protein complexes were then exchanged into the D₂O PBS buffer using a Vivaspin 20 centrifugal concentrator of 10 kDa molecular weight cut-off (Sartorius Stedim North America Inc, Bohemia, NY). The NSE measured protein concentrations were 10.86 mg/mL for the ^hV^hA^hB^hE complex, 13.1 mg/mL for the ^dV^hA^dB^dE complex, and 12.4 mg/mL for the ^hV^dA^dB^dE complex.

NSE Measurements.

The NSE experiments were performed with the upgraded high-resolution IN15 instrument at the ILL (2). We used three different wavelengths: 6, 10, and 13.5 Å covering 0.5 to 42, 0.2 to 194, and 0.4 to 477-ns time intervals. The covered Fourier time scales with the third power of the wavelength, but the incoming flux also drops roughly with the fourth power. The choice of the wavelength was made by optimizing the compromise between the resolution need and the incoming neutron flux. The beam monochromatization in each case was 15% full width at half-maximum as given by the neutron velocity selector. The samples were filled in quartz cells with 2-mm sample thickness, and the temperature was controlled at 10.0 ± 0.1 °C. Instrumental resolution was measured from the standard Grafoil (GrafTech, Lakewood, OH), which gives a strong, elastic, coherent small-angle scattering. The background was measured on the PBS D₂O buffer, and the sample spectra were corrected using the relative transmissions following the standard procedures. The sample environment box is equipped with a home-designed DLS apparatus where optical monomodal fibers bring in the laser light and collect scattered light at 90° scattering angle. All optics is placed below the neutron beam to avoid radiation damage. The correlator is a FLEX02-01D multitau digital correlator by Correlator.com LTD. During the NSE measurement, the DLS is run repeatedly to monitor for potential sample aging. In our case, both the scattered light intensity and the relaxation rate were constant within accuracy.

Data Management-ILL Neutrons for Society.

10.5291/ILL-DATA.8-04-942.