E-cadherin destabilization accounts for the pathogenicity of missense mutations in hereditary diffuse gastric cancer

Abstract

E-cadherin is critical for the maintenance of tissue architecture due to its role in cell-cell adhesion. E-cadherin mutations are the genetic cause of Hereditary Diffuse Gastric Cancer (HDGC) and missense mutations represent a clinical burden, due to the uncertainty of their pathogenic role. In vitro and in vivo, most mutations lead to loss-of-function, although the causal factor is unknown for the majority. We hypothesized that destabilization could account for the pathogenicity of E-cadherin missense mutations in HDGC, and tested our hypothesis using in silico and in vitro tools. FoldX algorithm was used to calculate the impact of each mutation in E-cadherin native-state stability, and the analysis was complemented with evolutionary conservation, by SIFT. Interestingly, HDGC patients harbouring germline E-cadherin destabilizing mutants present a younger age at diagnosis or death, suggesting that the loss of native-state stability of E-cadherin accounts for the disease phenotype. To elucidate the biological relevance of E-cadherin destabilization in HDGC, we investigated a group of newly identified HDGC-associated mutations (E185V, S232C and L583R), of which L583R is predicted to be destabilizing. We show that this mutation is not functional in vitro, exhibits shorter half-life and is unable to mature, due to premature proteasome-dependent degradation, a phenotype reverted by stabilization with the artificial mutation L583I (structurally tolerated). Herein we report E-cadherin structural models suitable to predict the impact of the majority of cancer-associated missense mutations and we show that E-cadherin destabilization leads to loss-of-function in vitro and increased pathogenicity in vivo.

Figure 1. E-cadherin structural models.

A) Sequence alignment of the extracellular domains of human E-cad and xenopus EP-cad. The extracellular sequences were obtained in Uniprot with the corresponding references (human E-cadherin, P12830; xenopus EP-cadherin, P33148). M-Coffee regular was used to perform the alignment, a package that combines different alignment methods. Red brick regions are in perfect agreement across all the methods, green and yellow regions are regions of no agreement between the different alignment methods. The average consistency score obtained was 98, confirming the reliability of the alignment. The blue stars identify the aminoacids that were removed from the 1L3W structure before humanizing. The black arrow indicates the end of the structural model obtained. B) The human structure of domains EC1-EC2 (PDB 2O72, blue) was aligned with the same domains of the human model generated from the xenopus structure (PDB 1L3W, red). Image created with Pymol. C) Schematic representation of human E-cadherin domains, highlighting the coverage of the three different structural models obtained with FoldX (models of prodomain, extracellular domain and the Catenin Binding Domain).

Figure 2. In silico analysis of the impact of germline E-cadherin missense mutations.

A) Schematic representation of E-cadherin domains, mapping all the modelled germline mutations found in the setting of HDGC or EODGC. Above the scheme are the mutations that resulted in destabilization, as predicted by FoldX (ΔΔG>0.8 kcal/mol) and below the scheme all the non-destabilizing mutations (ΔΔG<0.8 kcal/mol). The newly identified mutations are underlined. B) FoldX and SIFT were used to evaluate the impact of the mutations present in A) and the predictions were classified as: True Positive (TP) when the software predicts high impact and the mutants exhibit in vitro loss of function; True Negative (TN) when the software predicts no impact and the mutant is functional in vitro; False Positive (FP) when the software predicts high impact but the mutants is functional in vitro; and False Negative (FN) when the software predicts no impact and the mutants exhibits in vitro loss of function. The results from both predictors result in 70% overlap with E-cadherin protein function tested in vitro (TP+TN). C) Box-plot representing the median and interquartile ranges of the native-state stability changes (ΔΔG) of the Destabilizing and Non-destabilizing mutations, as predicted by FoldX. D) Box-plot representing the median and interquartile ranges of ages of Gastric Cancer detection or associated death, corresponding to the Destabilizing and Non-destabilizing mutations carriers. All the data was collected from the literature. The group of patients harbouring destabilizing mutations is characterized by a clear younger age of diagnosis or death, suggesting the contribution of E-cadherin destabilization for the disease phenotype.

Figure 3. Functional impact of three new HDGC-associated E-cadherin missense mutations: E185V, S232C and L583R.

CHO cells were transiently (A) or stably (B–C) transfected with an empty vector (Mock) or WT, E185V, S232C, L583R E-cadherin cDNA. A) Functional aggregation assay was performed as described in Material and Methods. L583R cells show E-cadherin loss of function, resulting in a scattered pattern, resembling Mock cells. ΔΔG was calculated using FoldX algorithm and is 0 for the WT reference; B) Total cell lysates were prepared and E-cadherin was detected by Western Blot using anti-E-cadherin antibody. Anti-α-Tubulin antibody was used as a loading control. The expression of L583R is reduced and shifted to higher molecular weight, indicative of being retained as immature (approximately 130 kDa). C) E-cadherin expression in the Plasma Membrane (PM) was evaluated using Flow Cytometry, after staining with an extracellular anti-human E-cadherin antibody. L583R is less expressed in the PM.

Figure 4. ERAD is involved in the regulation of E-cadherin destabilizing mutations.

CHO cells were stably (A, C) or transiently (B, D, E) transfected with an empty vector (Mock) or WT, E185V, S232C, L583R, L583I E-cadherin cDNA. A) E-cadherin and Calnexin immunofluoresce was performed in stable CHO cells expressing WT and L583R. Calnexin was used as an ER marker. L583R is retained in the ER, as evaluated by the colocalization with calnexin (yellow and arrows). B) Protein synthesis was blocked with Cicloheximide for 8 h and 16 h, to analyse E-cadherin turnover. E-cadherin was detected by Western Blot using anti-E-cadherin antibody and anti-α-Tubulin antibody was used as a loading control. L583R exhibits higher turnover. C) Cells were incubated with proteasome inhibitor MG132 for 16 h, and total cell lysates were prepared and analyzed. Proteasomal degradation results on the accumulation of L583R to levels similar to WT, indicating that the proteasome is necessary for the mutant downregulation. D) Functional aggregation assay was performed as described in Material and Methods. Cells expressing the artificial mutant L583I recover E-cadherin adhesive function, resembling WT cells, in contrast to L583R, which are not able to perform adhesion. E) Protein synthesis was blocked with Cicloheximide for 8 h and 16 h, to analyse E-cadherin turnover. In contrast to the unstable L583R, the stable mutation (L583I) is resistant to protein synthesis blockage, exhibiting lower turnover, comparable to the WT protein. The two bands of E-cad in B) and E) correspond to mature (lower, 120 kDa) and immature (upper, 130 kDa) forms of the protein, and result from the overload of protein commonly observed upon transient transfections.