Missense mutations of the ephrin receptor EPHA1 associated with Alzheimer's disease disrupt receptor signaling functions

Abstract

Missense mutations in the EPHA1 receptor tyrosine kinase have been identified in Alzheimer's patients. To gain insight into their potential role in disease pathogenesis, we investigated the effects of four of these mutations. We show that the P460L mutation in the second fibronectin type III (FN2) domain drastically reduces EPHA1 cell surface localization while increasing tyrosine phosphorylation of the cell surface-localized receptor. The R791H mutation in the kinase domain abolishes EPHA1 tyrosine phosphorylation, indicating abrogation of kinase-dependent signaling. Furthermore, both mutations decrease EPHA1 phosphorylation on S906 in the kinase-SAM linker region, suggesting impairment of a noncanonical form of signaling regulated by serine/threonine kinases. The R492Q mutation, also in the FN2 domain, has milder effects than the P460L mutation while the R926C mutation in the SAM domain increases S906 phosphorylation. We also found that EPHA1 undergoes constitutive proteolytic cleavage in the FN2 domain, generating a soluble 55 kDa N-terminal fragment containing the ligand-binding domain and a transmembrane 60 kDa C-terminal fragment. The 60 kDa WT fragment is phosphorylated on both tyrosine residues and S906, suggesting signaling functions. The P460L mutant 60 kDa fragment undergoes proteasomal degradation and the R791H mutant fragment lacks tyrosine phosphorylation and has decreased S906 phosphorylation. These findings advance our understanding of EPHA1 signaling mechanisms and support the notion that alterations in EPHA1 signaling due to missense mutations contribute to Alzheimer's disease pathogenesis.

Keywords: Alzheimer’s disease; Eph receptor; N-linked glycosylation; phosphoserine; phosphotyrosine; proteolytic cleavage; receptor modification; receptor tyrosine kinase.

Figure 1

The fully N-glycosylated form of EPHA1 is greatly reduced by the P460L mutation. HEK293 cells were transiently transfected with constructs encoding EGFP, EPHA1 WT, or the indicated EPHA1 mutants. A, cell lysates were probed by immunoblotting with the EPHA1 SAM antibody. B, cell lysates were immediately frozen (−) or incubated at 50 °C without or with Peptide:N-glycosidase F (PNGase F). The immunoblots were probed for EPHA1. C, Flag immunoprecipitates (IPs) and cell lysates were probed with WGA conjugated to HRP or by immunoblotting for EPHA1. D, cell lysates were probed by immunoblotting for EPHA1. E, schematic showing the positions of the two identified EPHA1 N-glycosylation sites.

Figure 2

The P460L mutation and impaired N-glycosylation decrease EPHA1 cell surface localization.A, proteins on the surface of HEK293 cells transiently expressing EPHA1 WT or the indicated mutants were biotinylated and pulled down with streptavidin beads. Pulled down proteins (cell surface) and cell lysates were probed as indicated. IGF1 receptor β (IGF1Rβ) was detected as a control that should have similar cell surface localization in all samples. The graph shows cell surface expression of the indicated mutants normalized to EPHA1 WT based on quantification of immunoblots. The bars show means and SE from quantification of five to seven experiments, with the individual data points shown as black dots. ∗∗∗, p < 0.001 for the comparison with WT using one-way ANOVA and Dunnett’s multiple comparisons test. B, cell surface biotinylation experiment similar to (A) comparing the indicated mutants. C, HEK293 cells stably expressing EPHA1 WT or the P460L mutant were treated with cycloheximide for the indicated times and probed by immunoblotting with the EPHA1 SAM antibody and a β-tubulin antibody as a loading control. The graph shows EPHA1 levels normalized to EPHA1 level at time 0 of cycloheximide treatment, quantified from the immunoblots from four experiments. The error bars represent SEs. ∗, p < 0.05 by unpaired Student’s t test for the comparison between EPHA1 WT and L460L at the 14 h time point.

Figure 3

The P460L mutation reduces the abundance of an EPHA1 60 kDa C-terminal fragment. HEK293 cells were transiently transfected with constructs encoding EGFP, EPHA1WT, or the indicated EPHA1 mutants. A, cell lysates were probed with the EPHA1 SAM domain antibody; the long exposure (long exp.) reveals a fragment of ∼60 kDa in cells expressing EPHA1 WT and some of the mutants. B, the 60 kDa band is also detected in Flag immunoprecipitates. The graphs show means and SEs from quantification of five experiments, with the individual data points shown as black dots. ∗, p < 0.05 and ∗∗∗∗, p < 0.0001 for the comparison with EPHA1 WT by one-way ANOVA and Dunnett’s multiple comparison test. C, cell lysates were probed with antibodies recognizing the EPHA1 SAM domain or the Flag epitope at the N terminus of EPHA1. D, the EPHA1 60 kDa fragment is detected in Flag immunoprecipitates carried out in both 0.5% TX-100 buffer and modified RIPA buffer that contains 0.1% SDS. E, the EPHA1 60 kDa fragment is not detected in Flag immunoprecipitates carried out in modified RIPA buffer containing 1% SDS and heated to disrupt protein interactions. EPHA1 immunoblots of Flag immunoprecipitates (Flag IPs) and lysates are shown. F, longer exposure of the lower part of the EPHA1 immunoblot shown in Fig. 1B highlights the EPHA1 C-terminal 60 kDa fragment, which exhibits a lower size following deglycosylation. G, the 60 kDa fragment has reduced size as a consequence of the N478 mutation that prevents glycosylation at this site. H, schematic illustrating the EPHA1 60 kDa C-terminal fragment. A red asterisk in (A,C,D,E,F,G) marks the EPHA1 60 kDa fragment.

Figure 4

EPHA1 WT and the P460L mutant are both cleaved in the extracellular region. HEK293 cells were transiently transfected with constructs encoding EGFP, EPHA1 WT, or the indicated EPHA1 mutants. A, EPHA1 extracellular fragments are detected in the culture medium. Conditioned culture medium was probed by immunoblotting with an antibody that recognizes the EPHA1 extracellular region (EPHA1 R&D). Lysates were probed with the EPHA1 SAM antibody and different exposures are shown for the full-length receptor and the less abundant 60 kDa and 45 kDa fragments. Amido black protein stain verifies protein loading. B, schematic illustrating two sets of N- and C-terminal fragments generated by two proteolytic cleavages in different parts of the EPHA1 extracellular region. C, the P460L cytoplasmic fragment is prone to proteasomal degradation. Cells were treated with MG132 or chloroquine before lysis and probed by immunoblotting with the EPHA1 SAM antibody. The lysates were also probed for β-catenin as a loading control and to verify successful MG132 treatment, as indicated by the nondegraded ubiquitinated β-catenin bands, which have a higher molecular weight. D, proteasomal degradation of the 60 kDa fragment does not require EPHA1 kinase activity, since the 60 kDa fragment is degraded even when the P460L mutation is combined with the kinase-inactivating K656R mutation. Red asterisks mark the N- and C-terminal fragments generated by EPHA1 cleavage within the FN2 domain; orange asterisks mark the N- and C-terminal fragments likely generated by EPHA1 cleavage near the plasma membrane. pTyr, phosphotyrosine.

Figure 5

The EPHA1 ectodomain is cleaved by matrix metalloproteinases. HEK293 cells were transiently transfected with constructs encoding EGFP, EPHA1 WT, or the indicated EPHA1 mutants. A and B, after transfection, the cells were incubated without (−) or with (+) the broad-spectrum MMP inhibitor GM6001 (A) or a MMP 9/MMP13 inhibitor (B). Conditioned medium and cell lysates were probed as indicated. C, Flag immunoprecipitates from conditioned cell culture medium were probed with an antibody to the EPHA1 extracellular region. EPHA1 A25-Q462 is a Flag-tagged engineered EPHA1 secreted form. D, cell lysates were probed with the EPHA1 SAM antibody. L463-D976 is a Flag-tagged engineered EPHA1 truncated form. E, conditioned medium and lysates from cells expressing EPHA1 with mutations in predicted FN2 domain MMP proteolytic cleavage motifs were probed as in (A) and (B). F, effects of EPHA1 FN2 domain mutations on the relative abundance of the 55 kDa versus the 70 kDa fragment in the culture medium. Different amounts of medium were loaded to enable detection of the two bands even when they are present at very low levels. The graphs show means and SEs from the quantification of immunoblots from multiple experiments, with individual data points shown as black dots. Three experiments were quantified in both panels in A; five experiments in both panels in B; three to eight (left panel) and four to seven (right panel) experiments in E; and eight (left panel), two to five (middle panel), and one to three (right panel) experiments in F. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001; and ∗∗∗∗, p < 0.0001 for comparisons with WT (left and middle graphs) or P460L (right graph) by one-way ANOVA and Dunnett’s multiple comparison test. Red asterisks mark the N- and C-terminal fragments generated by EPHA1 cleavage within the FN2 domain; orange asterisks mark the N- and C-terminal fragments generated by EPHA1 cleavage near the plasma membrane.

Figure 6

Some Alzheimer’s mutations disrupt EPHA1 tyrosine and S906 phosphorylation.A, Flag immunoprecipitates (IPs) from transiently transfected HEK293 cells were probed by immunoblotting with the indicated antibodies. The graphs show means and SEs for the quantifications of EPHA1 tyrosine phosphorylation (pTyr, n = 5) and S906 phosphorylation (pS906, n = 6) normalized to EPHA1 levels in the upper of the two bands detected with the EPHA1 SAM antibody (which corresponds to the phosphorylated band). Individual data points are shown as black dots. ∗, p < 0.05 and ∗∗, p < 0.01 for the comparison with EPHA1 WT using one-way ANOVA and Dunnett’s multiple comparisons test. B, HEK293 cells stably expressing Flag-tagged EPHA1 WT or P460L mutant were stimulated with ephrinA1-Fc for the indicated times and Flag immunoprecipitates were probed by immunoblotting for phosphotyrosine (pTyr) or EPHA1. C and D, quantifications of the immunoblots in (B) and others (n = 5). The basal EPHA1 tyrosine phosphorylation (pTyr) signal was normalized to EPHA1 levels in the absence of ephrinA1-Fc stimulation and then further normalized to the pTyr/EPHA1 WT signal at 30 min of ephrinA1-Fc stimulation. The graph in (C) shows means and SEs and the individual data points are shown as black dots. ∗, p < 0.05 by unpaired Student’s t test. The graph in (D) shows means ± SEs from five experiments. E, experiment similar to A, but using a longer exposure to visualize the EPHA1 60 kDa fragment co-immunoprecipitated with full-length EPHA1 as in Fig. 3, D and E. The graphs show means and SEs for the quantifications of EPHA1 tyrosine phosphorylation (pTyr, n = 5) and S906 phosphorylation (pS906, n = 5) normalized to EPHA1 levels. The P460L mutant fragment could not be reliably quantified given its very low level. ∗∗∗∗, p < 0.0001 for the comparison with WT using one-way ANOVA and Dunnett’s multiple comparisons test. F, HEK293 cells were transiently transfected with Flag-tagged EPHA1 WT (comprising residues A25-D976), the Flag-tagged engineered EPHA1 L448-D976 and L463-D976 truncated forms, and EGFP as a control. Immunoblotting for tyrosine phosphorylation (pTyr), S906 phosphorylation, and the Flag tag demonstrates high phosphorylation of the EPHA1 engineered truncated forms.

Figure 7

Schematic illustrating the effects of the EPHA1 Alzheimer’s mutations on EPHA1 signaling features.A, EPHA1 WT traffics to the cell surface, where it can signal through tyrosine phosphorylation (pTyr, red circles) and S906 phosphorylation (pS906, blue circle) and can be cleaved by metalloproteases (MMPs). Cleavage generates truncated fragments released from the cells (a 55 kDa fragment, shown, and a 70 kDa fragment, not shown) or remaining associated with the cells (60 kDa fragment, shown, and 45 kDa fragment, not shown). The 60 kDa cell-associated fragment is phosphorylated on both tyrosine residues and S906, suggesting signaling ability. B, the P460L mutation in the FN2 domain decreases EPHA1 levels at the plasma membrane, likely because FN2 domain misfolding impairs EPHA1 trafficking to the cell surface. Full-length EPHA1 P460L on the cell surface is more tyrosine phosphorylated and less phosphorylated on S906 than EPHA1 WT. Matrix metalloproteases cleave the EPHA1 P460L FN2 domain more than EPHA1 WT, but the P460L cell-associated 60 kDa fragment undergoes proteasomal degradation. The R492Q mutation, also in FN2, has effects similar to the P460L mutation but less pronounced. C, the R791H kinase domain mutation abolishes kinase activity, and thus autophosphorylation on tyrosine residues, and also decreases S906 phosphorylation. D, the R926C mutation in the SAM domain increases S906 phosphorylation. Bold or darker font indicates more pronounced effects. The P460L mutant FN2 domain is shown with a different shape than WT to suggest misfolding.