Molecular mechanism for strengthening E-cadherin adhesion using a monoclonal antibody

Abstract

E-cadherin (Ecad) is an essential cell-cell adhesion protein with tumor suppression properties. The adhesive state of Ecad can be modified by the monoclonal antibody 19A11, which has potential applications in reducing cancer metastasis. Using X-ray crystallography, we determine the structure of 19A11 Fab bound to Ecad and show that the antibody binds to the first extracellular domain of Ecad near its primary adhesive motif: the strand-swap dimer interface. Molecular dynamics simulations and single-molecule atomic force microscopy demonstrate that 19A11 interacts with Ecad in two distinct modes: one that strengthens the strand-swap dimer and one that does not alter adhesion. We show that adhesion is strengthened by the formation of a salt bridge between 19A11 and Ecad, which in turn stabilizes the swapped β-strand and its complementary binding pocket. Our results identify mechanistic principles for engineering antibodies to enhance Ecad adhesion.

Keywords: 19A11; E-cadherin; adhesion; antibody; strand–swap dimer.

Fig. 1.

Structure of 19A11 bound to Ecad. (A) X-ray crystal structure of 19A11 Fab heavy chain (magenta) and light chain (orange) bound to Ecad EC1–EC2 domains (cyan). (B) Detailed view of the hydrogen bonds and salt bridges between 19A11 Fab heavy chain (magenta) and Ecad EC1 domain (cyan). (C) Detailed view of the interactions between 19A11 Fab light chain (orange) and Ecad EC1 domain (cyan). The distances between interacting atoms in (B and C) are shown in Å (black dashed lines).

Fig. 2.

Binding of 19A11 stabilizes both the Ecad β-strand and the W2 hydrophobic pocket. MD simulations were performed with (A) the Ecad strand–swap dimer (EcadA and EcadB) in the absence of 19A11 (0ab), (B) the Ecad strand–swap dimer with a single 19A11 Fab (abA) bound to EcadA (1ab), and (C) the Ecad strand–swap dimer with two 19A11 Fabs (abA and abB) bound to both Ecads (2ab). (D) Ecad–antibody binding interface. Two salt bridges are observed: E13–R99 and K14–D58. The 19A11 binding region is located between the β-strand and the W2 hydrophobic pocket (referred to as “Pocket”) on Ecad. (E) Average RMSF values for residues 1–30 of Ecad in the 2ab case (solid black), EcadA in the 1ab case (dashed green), EcadB in the 1ab case (dashed orange), and Ecad in the 0ab case (solid purple). The W2 position is highlighted using a vertical dashed red line. The lower RMSF values show that the binding of 19A11 stabilizes the β-strand and the W2 hydrophobic pocket of both Ecads in the 2ab case while it only stabilizes EcadA (which is bound to 19A11) in the 1ab case.

Fig. 3.

Salt bridges between 19A11 and Ecad stabilize the β-strand and the W2 hydrophobic pocket. (A–E) Violin plots of the distances between charged atoms in the E13–R99 and K14–D58 salt bridges measured during the last 40 ns of each 2ab MD simulation. The median distance is shown as a black line on each violin. Distances for EcadA and EcadB are shown in the left and right panels, respectively. Distances measured for E13–R99 interactions during the MD simulations are shown in red and charged atoms distances for K14-D58 are shown in blue. (A) simulation 1 (set 1), (B) simulation 2 (set 2), (C) simulation 3 (set 3), (D) simulation 4 (set 4), (E) simulation 5 (set 5). Both EcadA and EcadB form at least one salt bridge with the bound 19A11 in set 1 and set 2. However, only one of the Ecads formed a salt bridge with 19A11 in sets 3–5 (EcadB in set 3, EcadB in set 4, and EcadA in set 5). (F) Comparison of the average backbone RMSF values when an Ecad forms at least one salt bridge with its corresponding 19A11 (purple solid line), when an Ecad does not form at least one salt bridge with its corresponding 19A11 (black solid line), and in the absence of 19A11, i.e., 0ab (dashed cyan line). The RMSF for W2 is highlighted using a vertical dashed red line. The β-strand and the W2 hydrophobic pocket have a lower RMSF when a salt bridge is formed as compared to when no salt bridges are formed.

Fig. 4.

Adhesion strengthening requires two bound 19A11 antibodies to form salt bridges with partner Ecads. Constant-force SMD simulations with change in Ecad–Ecad interfacial area calculated from the ΔSASA, in the (A) 0ab condition, (B) the 1ab conditions, and (C) the 2ab conditions. (D) Distance between center of mass of W2 and the center of mass of the hydrophobic pockets in each of the constant-force SMD simulations. While the lifetimes of the Ecad–Ecad bonds are similar in the 0ab and 1ab and sets 3–5 of the 2ab condition, the lifetime of the Ecad–Ecad bond in sets 1–2 of the 2ab condition, where both interacting Ecads form at least one salt bridge with 19A11, are substantially longer (also see Movie S1).

Fig. 5.

Direct, single-molecule measurements of 19A11-mediated strengthening of Ecad homophilic adhesion. (A) Top Left: Scheme for AFM experiment carried out in the absence of 19A11 (−ab). Ecads were immobilized on an AFM tip and substrate functionalized with PEG tethers. Top Right: Scheme for AFM experiment with antibody (+ab). Both the AFM tip and substrate were incubated with 19A11. Bottom: Example force curve. Stretching of the PEG tether, which served as a “signature” of a single-molecule unbinding event, was fit to a WLC model (red line). Experiments were performed with WT and K14E Ecad in the absence (−ab) and presence (+ab) of 19A11. Histograms of the unbinding forces were generated by binning the data in each condition using the Freedman–Diaconis rule. The optimal number of Gaussian distributions for each fit was determined using BIC. This analysis prescribed one Gaussian distribution for (B, E, and F) and two Gaussian distributions for (C and D). (B) Probability density of Ecad–Ecad unbinding forces measured in the absence of 19A11. Forces are Gaussian distributed (red line) with a peak force of 49.4 ± 12.0 pN. (C) Probability density of Ecad–Ecad unbinding forces in the presence of 20 nM 19A11 was best fit by a bimodal Gaussian distribution. While the first peak at 48.6 ± 17.3 pN (green line) corresponds to a “native” Ecad unbinding force, the second peak at 73.1 ± 27.5 pN (blue line) corresponds to strengthened adhesion. (D) Increasing the concentration of 19A11 in solution to 150 nM yields a similar bimodal Gaussian distribution with peaks at 51.7 ± 14.5 pN (green line) and 69.6 ± 25.5 pN (blue line). This demonstrates that the bimodal distribution of forces does not occur due to low 19A11-Ecad binding affinity but rather because 19A11 binds to Ecad in two distinct modes. (E) Probability density of Ecad K14E–K14E unbinding forces measured in the absence of 19A11. Forces are Gaussian distributed (red line) with a peak force of 41.2 ± 17.3pN. (F) Probability density of Ecad K14E–K14E unbinding forces measured in the presence of 150 nM 19A11. Since the K14E mutant abolishes the formation of the K14–D58 salt bridge, 19A11 Fab can no longer strengthen Ecad adhesion. Forces are single Gaussian distributed (red line) with a peak force of 40.3 ± 24.1pN.