The EphA2 receptor is activated through induction of distinct, ligand-dependent oligomeric structures

Abstract

The EphA2 receptor tyrosine kinase is capable of activating multiple diverse signaling pathways with roles in processes such as tissue homeostasis and cancer. EphA2 is known to form activated oligomers in the presence of ephrin-A ligands. Here, we characterize the lateral interactions between full-length EphA2 molecules in the plasma membrane in the presence of three types of ligands (dimeric ephrinA1-Fc, monomeric ephrinA1, and an engineered peptide ligand) as well as in the absence of ligand, using a quantitative FRET technique. The data show that EphA2 forms higher-order oligomers and two different types of dimers that all lead to increased EphA2 tyrosine phosphorylation, which is indicative of increased kinase-dependent signaling. We find that different ligands stabilize conformationally distinct oligomers that are assembled through two different interfaces. Our results suggest that these different oligomeric assemblies could have distinct signaling properties, contributing to the diverse activities of the EphA2 receptor.

Fig. 1

Dimeric ephrinA1-Fc induces EphA2 receptor clustering. a Portion of a HEK293T cell expressing EphA2-mTurq and EphA2-eYFP in hypo-osmotic medium, imaged when mTurq was excited. A plasma membrane region of homogeneous fluorescence, a few µm in length (yellow box), is analyzed to determine the EphA2-mTurq concentration, the EphA2-eYFP concentration and the FRET efficiency, as described in the Materials and methods. The scale bar is 5 μm. b FRET efficiency vs. acceptor (EphA2-eYFP) concentration. Each data point corresponds to one plasma membrane region. c Donor (EphA2-mTurq) concentration vs. acceptor (EphA2-eYFP) concentration in the selected membrane regions. In b and c, 275 cells were imaged in four independent experiments, yielding 858 data points. d Mean square error (MSE) vs. oligomer order. MSE is minimized for n > 4, indicating the presence of oligomers that are larger than dimers (i.e., clusters). e Clustered EphA2 receptor fraction as a function of total EphA2 concentration. The data were binned and the averages are shown along with the standard errors. The solid line represents the theoretical best fit to the data. f Mean FRET efficiencies vs. mean acceptor fractions, determined as shown in Supplementary Figure 1. The plot is based on more than 1000 data points. The dependence deviates from a linear function, supporting the conclusion that exposure to ephrinA1-Fc induces preferentially the formation of EphA2 clusters

Fig. 2

Interfaces involved in EphA2 receptor clustering induced by ephrinA1-Fc. a Crystal structure showing a lateral view of four EphA2 extracellular regions (gray) bound to four ephrinA1 molecules (light blue; PDB ID: 3MX0). The receptor tetramer is stabilized via two interfaces: the “clustering” interface (approximately outlined in orange), which includes contacts mediated by L223, L254, and V255 in the cysteine-rich domain, and the dimerization interface (approximately outlined in wine), which includes contacts mediated by G131 in the ligand-binding domain. b Comparison of raw FRET data for EphA2 wild-type and the L223R/L254R/V255R mutant in the presence of ephrinA1-Fc. In this experiment, 275 cells were imaged in four independent experiments to obtain 858 data points for the wild-type, and 196 cells were imaged in four independent experiments to obtain 563 data points for the L223R/L254R/V255R mutant. c Comparison of raw FRET data for EphA2 wild-type and the G131Y mutant in the presence of ephrinA1-Fc. A total of 618 cells were imaged in six independent experiments to yield 2310 data points for the G131Y mutant. d MSE vs. oligomer order for the L223R/L254R/V255R and G131Y mutants in the presence of ephrinA1-Fc. The MSE minimum for the L223R/L254R/V255R mutant occurs at n = 6. The MSE for the G131Y mutant is the same for n ≥ 2. As previously shown, these results indicate that the EphA2 receptor is preferentially assembled into clusters, although the presence of some dimers cannot be excluded. e Representation of the clustered fractions for EphA2 wild-type and the L223R/L254R/V255R and G131Y mutants as a function of total receptor concentration shows that mutation of both interfaces decreases the fraction of clustered EphA2

Fig. 3

Effect of the R103E mutation on EphA2 clustering induced by ephrinA1-Fc. a, b Side and top views of a crystal structure of four EphA2 molecules (gray) bound to four ephrinA1 molecules (light blue; PDB ID: 3MX0). The position of the R103E mutation in two of the EphA2 molecules is shown in magenta, and indicated by an arrow in one. c Comparison of raw FRET data for EphA2 wild-type and the R103E mutant in the presence of ephrinA1-Fc. 275 cells were imaged in four independent experiments to obtain 858 data points for the wild type. A total of 201 cells were imaged in three independent experiments to obtain 474 data points for R03E mutant. d MSE vs. oligomer order for EphA2 R103E in the presence of ephrinA1-Fc. The MSE value is the same for all n ≥ 2, indicating oligomerization with predominance of clusters. e Comparison of EphA2 wild-type and R103E mutant clustered fractions in the presence of saturating concentration of ephrinA1-Fc shows that the R103E mutation severely destabilizes the clusters

Fig. 4

Dimerization of EphA2 wild-type and the L223R/L254R/V255R, G131Y and R103E mutants in the absence of ligand binding. a MSE vs. oligomer order for EphA2 wild-type and the three mutants. MSE is minimized at n = 2 for all, indicating the presence of dimers. b Dimerization curves for EphA2 wild-type and the L223R/L254R/V255R and G131E mutants. The L223R/L254R/V255R mutations reduce dimerization, while the G131Y mutation has no effect. Thus, the unliganded dimer is stabilized through the “clustering interface”. c Dimerization curves for EphA2 wild-type and the R103E mutant. The R103E mutant exhibits a reduced dimerization propensity, despite the fact that this residue is not part of the clustering interface. d A representative Western blot comparing Y772 phosphorylation for EphA2 wild-type and the G131Y mutant. e Quantification from three independent experiments (shown as solid circles) shows no statistically significant difference (p > 0.05 from Student’s t-test). f A representative Western blot image comparing S897 phosphorylation for EphA2 wild-type and the G131Y mutant. g Quantification from four independent experiments (shown as solid circles) shows no statistically significant difference (p > 0.05 from Student’s t-test). h Representative Western blot images comparing Y772 phosphorylation for EphA2 wild-type and the R103E mutant. i Quantification from three independent experiments (shown as solid circles) shows that the R103E mutant has lower Y772 phosphorylation than EphA2 wild-type (***p < 0.001 from Student’s t-test). j A representative Western blot comparing S897 phosphorylation for EphA2 wild-type and the R103E mutant. k Quantification from three independent experiments (shown as solid circles) shows that the R103E mutant has higher S897 phosphorylation than EphA2 wild-type (*p < 0.05 from Student’s t-test). l Cell migration assays with HEK293T cells expressing EphA2 wild-type and the three mutants. The solid circles represent the the individual experiments. Cells expressing EphA2 wild-type and the G131Y mutant exhibit similar migratory ability. In contrast, cells expressing the L223R/L254R/V255R mutant and the R103E mutant migrate faster than wild-type. (**p < 0.01 from ANOVA, n.s. non-significant, p > 0.05). The bars in e, g, i and k represent the averages from different experiments with the standard errors

Fig. 5

EphA2 dimerization induced by the monomeric ephrinA1 ligand. a MSE vs. oligomer order for EphA2 wild-type and the L223R/L254R/V255R and G131Y mutants in the presence of 200 nM m-ephrinA1. The MSEs are all minimized for n = 2, indicating dimerization. b Comparison of EphA2 dimerization propensity in the presence and absence of m-ephrinA1 shows that m-ephrinA1 significantly enhances EphA2 dimerization. c Dimerization curves in the presence of m-ephrinA1 show that the dimerization propensity of EphA2 wild-type and the L223R/L254R/V255R mutant are the same, while the G131Y mutant has a reduced dimerization propensity, indicating the involvement of the dimerization interface. d A representative Western blot showing Y772 phosphorylation of EphA2 wild-type and the indicated mutants following a 15 min stimulation with m-ephrinA1 and FBS. e Quantification of Y772 phosphorylation from three to four independent measurements is shown as solid circles. The bars represent the averages and the standard errors. EphA2 wild-type and the L223R/L254R/V255R mutant exhibit similar levels of Y772 phosphorylation while the G131Y mutant shows significantly lower phosphorylation (**p < 0.01 from Student t-test with Bonferroni correction). f Representative Western blots showing the S897 phosphorylation of EphA2 wild-type and mutants following a 15 min stimulation with m-ephrinA1 and FBS. g Quantification of S897 phosphorylation from three to five independent measurements is shown as solid circles. The bars represent the averages with standard errors. The wild-type and the L223R/L254R/V255R mutant show similar S897 phosphorylation while the G131Y mutant shows significantly higher phosphorylation (*p < 0.02 from Student t-test with Bonferroni correction). h Migration of HEK293T cells expressing wild-type and mutant EphA2 in the presence of m-ephrinA1. The solid circles represent the data points from three to four independent measurements. The bars represent the averages with standard errors. Cells expressing EphA2 wild-type and the L223R/L254R/V255R mutant exibit similar migratory propensity while cells expressing the G131Y EphA2 mutant migrate faster. (**p < 0.01 from ANOVA)

Fig. 6

Effect of the R103E mutation on EphA2 dimerization in the presence of monomeric ephrinA1 ligand. a MSE vs. oligomer order for the EphA2 R103E mutant in the presence of m-ephrinA1. The MSE is miminized for n = 2, indicating dimerization. b Dimerization propensity of the EphA2 R103E mutant in the absence and in the presence of m-ephrinA1. The ligand slightly enhances dimerization, indicating that the EphA2 R103E mutation does not completely abrogate m-ephrinA1 binding. c Dimerization curves for EphA2 wild-type and the R103E mutant in the presence of m-ephrinA1. The R103E mutation decreases the stability of the dimers, most likely because it severely impairs m-ephrinA1 binding

Fig. 7

EphA2 dimerization in the presence of the monomeric YSA peptide ligand. a MSE vs. oligomer order for EphA2 wild-type and the L223R/L254R/V255R, G131Y, and R103E mutants. In all cases, the MSE minimum occurs at n = 2, indicating dimerization. b Dimerization curves show that the dimerization propensity of EphA2 wild-type and the G131Y mutant are the same, while the L223R/L254R/V255R mutant has reduced dimerization propensity. Data for the wild-type and the L223R/L254R/V255R mutant are from ref. . c Dimerization curves for EphA2 wild-type and the R103E mutant show that the mutant has reduced dimerization propensity

Fig. 8

Cartoon representation of the findings: EphA2 can associate into two types of dimers as well as clusters, depending of the nature of the activating ligand. TK tyrosine kinase, FP fluorescent protein