Cadherin Diffusion in Supported Lipid Bilayers Exhibits Calcium-Dependent Dynamic Heterogeneity

Abstract

Ca 2þ ions are critical to cadherin ectodomain rigidity, which is required for the activation of adhesive functions. Therefore, changes in Ca 2þ concentration, both in vivo and in vitro, can affect cadherin conformation and function. We employed single-molecule tracking to measure the diffusion of cadherin ectodomains tethered to supported lipid bilayers at varying Ca 2þ concentrations. At a relatively high Ca 2þ concentration of 2 mM, cadherin molecules exhibited a fast diffusion coefficient that was identical to that of individual lipid molecules in the bilayer (D fast z 3 mm 2 /s). At lower Ca 2þ concentrations, where cadherin molecules were less rigid, the ensemble-average cadherin diffusion coefficient was systematically smaller. Individual cadherin trajectories were temporally heterogeneous, exhibiting alternating periods of fast and slow diffusion; the periods of slow diffusion (D slow z 0.1 mm 2 /s) were more prevalent at lower Ca 2þ concentration. These observations suggested that more flexible cadherin ectodomains at lower Ca 2þ concentration alternated between upright and lying-down conformations, where the latter interacted with more lipid molecules and experienced greater viscous drag.

Cadherin clusters stabilized by a combination of specific and nonspecific cis-interactions

“…As is apparent from Figure 1B–J , transport properties are often coupled to the FRET-state of a molecule. This is because the FRET-state reflects the oligomeric state of an E-cad molecule, and large oligomers diffuse slower than a monomer due to increased protein-lipid interactions, which is the primary source of drag ( Cai et al, 2016 ). In order to assess this hypothesis and confirm that the high-FRET state does in fact correlate with protein clusters involving a donor and one or more acceptors, the average short-time diffusion coefficient ( ) was determined for the high and low FRET-state populations independently.…”

“…Furthermore, due to inherent defects in supported lipid bilayers, a permanently immobile (or highly confined) population was observed in the donor channel ( Knight et al, 2010 ). Traditionally, a displacement-based trajectory filtering procedure or photobleaching is applied to remove these slowly diffusing trajectories in lipid bilayer studies ( Cai et al, 2016 ; Chaparro Sosa et al, 2018 ; Chung et al, 2016 ; Knight and Falke, 2009 ; Knight et al, 2010 ; Ziemba and Falke, 2013 ). However, we opted to instead use a median donor intensity trajectory exclusion criterion, as this removed many bright contaminants, donor aggregates, and donor E-cad labeled with multiple fluorophores, but did not accidentally remove slowly diffusing E-cad clusters.…”

“…Considering that mutant E-cad tethered to the bilayer diffuses the same as a single lipid at all surface coverages ( Figure 5—figure supplement 3 and Thompson et al, 2019 ), we can extract the effective size of E-cad clusters assuming additive friction factor contributions from each E-cad molecule in the cluster ( Cai et al, 2016 ; Knight et al, 2010 ; Thompson et al, 2019 ; Ziemba and Falke, 2013 ). The apparent trajectory friction factor, f , can be expanded as: where f i is the friction factor contribution due to each E-cad molecule in the cluster and N is the number of protein molecules in the cluster.…”

Engineering the Composition of Heterogeneous Lipid Bilayers to Stabilize Tethered Enzymes

“…Previous work has shown that the diffusion of LB‐tethered proteins is controlled by the hydrodynamic coupling between the tethered protein and the bilayer, which is a function of the strength of protein‐lipid interactions and the effective size of the protein‐bilayer interface. [ 39–42 ] Notably, we previously found that unfolded NfsB diffused more slowly than folded NfsB, [ 33 ] reflecting stronger interactions between the LB and the unfolded state of the enzyme, which led to increased friction due to greater contact between the unfolded state and the LB. As in our previous work, D̅ folded and D̅ unfolded were determined from complementary cumulative square displacement distributions for each state, which were fit to a Gaussian mixture model with two diffusive modes, and the mean diffusion coefficients were calculated as the weighted average of the two modes.…”

“…Lipids were dried under nitrogen gas, and subsequently rehydrated in buffer containing sodium phosphate (10 × 10 –3 m ) and sodium chloride (800 × 10 –3 m ) at a pH of 8.0 followed by extrusion to form unilamellar vesicles with diameter of 50 nm as described previously. [ 39 ]…”