CLCA2 is a p53-inducible transmembrane protein that is frequently downregulated in breast cancer. CLCA2 is a 943 amino acid type I transmembrane protein that is cleaved near amino acid 700 to produce a diffusible 100kD product. The N-terminus contains a hydrolase-like domain with well-conserved HEXXH zinc binding amino acid motif that was proposed to cleave the precursor auto-proteolytically. We investigate the auto-proteolysis of CLCA2 precursor. Using membrane extracts or purified protein from CLCA2-transfected cells, we show here that CLCA2 cleavage is catalyzed by zinc and inhibited by metal chelator EDTA. Moreover, an E165Q mutation in the metal binding site abolished processing without affecting stability or trafficking. The mutant could be cleaved by co-transfected wild type CLCA2, showing that the mutation had not caused an un-cleavable conformation and suggesting that it occurs in trans. Wild type CLCA2 was able to cleave CLCA2 E165Q mutant in vitro only after denaturation and renaturation, suggesting that a conformational shift is required for cleavage. The efficiency of cleavage increased steeply with increasing concentration of precursor, consistent with trans proteolysis but not cis or cleavage by another agent. Accordingly, CLCA2 molecules bearing different epitope tags formed a stable complex that could be co-immunoprecipitated. Cleavage appears to be specific within isoforms; CLCA1 was unable to neither cleave CLCA2 nor form a stable complex with it. Furthermore, cleavage causes a conformational shift: an N-terminal antibody that immunoprecipitates the precursor fails to precipitate the N-terminal product unless it is first denatured with ionic detergent. We found that cleavage is enhanced by p53 induction due to DNA damage, implying that the cleavage has functional consequences for stress response. Moreover, we found that HEK and MCF10A cells expressing the E165Q mutant had a higher proliferation rate than cells expressing wild type CLCA2, suggesting that the metalloprotease activity contributes to the anti-proliferative effect of CLCA2. Physiologically, CLCA2 plays an essential role in epithelial differentiation. It is induced during epithelial differentiation in immortalized human mammary epithelial cells (HMEC), and its knockdown causes epithelial to mesenchymal transition (EMT). To determine how CLCA2 promotes epithelial differentiation, we searched for interactors using membrane dihybrid screening. We discovered a strong interaction with Epithelial V-like Antigen 1 (EVA1) and confirmed by co-immunoprecipitation. Like CLCA2, EVA1 is a type I transmembrane protein that is regulated by p53 family. EVA1 resembles tight junction proteins called Junctional Adhesion Molecules (JAMs) by structure but we found by confocal analysis that EVA1 is localized the lateral interface at cell-cell junctions. Analysis of transcriptional profiles revealed that EVA1 is frequently downregulated in breast tumors and breast cancer cell lines, especially those of mesenchymal phenotype, and upregulated during epithelial differentiation. Like CLCA2, knockdown of EVA1 resulted in rapid EMT in immortalized HMEC. The interacting domains were delimited by deletion analysis, revealing that both the proteins interact via their transmembrane segments (TMS). The interaction was specific, as other transmembrane proteins did not interact with CLCA2 or EVA1. We also found that CLCA2 binds to ZO-1 and beta-catenin at its c-terminus but EVA1 does not. Interestingly, we found that EVA1 does interact with ZO-1 in the presence of CLCA2, indicating that these three form a complex at the cell-cell junctions that allows stabilization of belt-like adherens junctions (AJ). On the other hand CLCA2 may also stabilize adherens junctions by sequestering beta-catenin at the cell-cell junctions. These results indicate that CLCA2 plays a key role in maintaining epithelial differentiation via multiple ways. Either by binding to beta-catenin or forming a complex with EVA1 and ZO-1, it plays a pivotal role in maintaining epithelial differentiation. This explains the downregulation of both CLCA2 and EVA1 during tumor progression.