Interactions and Feedbacks in E-Cadherin Transcriptional Regulation

Abstract

Epithelial tissues rely on the adhesion between participating cells to retain their integrity. The transmembrane protein E-cadherin is the major protein that mediates homophilic adhesion between neighbouring cells and is, therefore, one of the critical components for epithelial integrity. E-cadherin downregulation has been described extensively as a prerequisite for epithelial-to-mesenchymal transition and is a hallmark in many types of cancer. Due to this clinical importance, research has been mostly focused on understanding the mechanisms leading to transcriptional repression of this adhesion molecule. However, in recent years it has become apparent that re-expression of E-cadherin is a major step in the progression of many cancers during metastasis. Here, we review the currently known molecular mechanisms of E-cadherin transcriptional activation and inhibition and highlight complex interactions between individual mechanisms. We then propose an additional mechanism, whereby the competition between adhesion complexes and heterochromatin protein-1 for binding to STAT92E fine-tunes the levels of E-cadherin expression in Drosophila but also regulates other genes promoting epithelial robustness. We base our hypothesis on both existing literature and our experimental evidence and suggest that such feedback between the cell surface and the nucleus presents a powerful paradigm for epithelial resilience.

Keywords: HP1; JAK/STAT; PAR-3; adhesion; heterochromatin.

FIGURE 1

Molecular mechanisms regulating E-cadherin gene expression. (A,B) Summary of the best known negative and positive regulators of E-cad in mammals (A) and Drosophila (B). The expression of E-cad gene (shotgun, shg, in Drosophila) is controlled by transcription factors (ellipses), which inhibit (red) or promote it (green) and interact with the E-cad promoter with varying affinities. Several further proteins modify chromatin at the promoter (squircles) silencing the E-cad expression (red) or increasing it (green). These proteins cooperate, compete, and regulate each other, establishing the intricated network that fine-tune the expression of E-cad. In panel (B) the non-conserved proteins were removed while conserved maintained their relative positions but their names were replaced with those of the Drosophila orthologs. The regulatory roles of transparentized proteins in panels (A,B) were not demonstrated in the respective systems. (C) The proposed model of E-cad modulation of its expression. A subpool of STAT92E translocates into the nucleus independently of the canonical JAK/STAT signalling, where it interacts with heterochromatin protein 1 (HP1). Within euchromatin, HP1 localises at specific loci corresponding to gene promoters, including that of the shg gene. Concurrently, STAT92E is recruited to the cell surface by E-cad and its binding partner Par-3. Therefore, elevated levels of E-cad outcompete STAT92E from the nucleus inhibiting its function in heterochromatin formation and promoter regulation. As the result, elevated E-cad at the cell surface reduces shg expression, ultimately restoring its levels. It is unclear whether this is accompanied by a change in chromatin organisation around shg promoter, indicated by ‘?.’

FIGURE 2

The crosstalk between E-cadherin, STAT92E and HP1. (A) A diagram of position effect variegation (PEV) in relation to chromatin compaction. (B,C) E-cad overexpression suppresses PEV: examples (B) and quantification (C). Scale bar – 0.2 mm, ***p = 0.0001 (unpaired t-test). (D) Overlapping box-and-violin plots showing the quantification of rBEADS normalised HP1 signal on larval-specific (red) and common (blue) promoters. The p-value above – the significance calculated using U-test. The value below the boxplots – the mean of the signal. (E) The distribution of the rBEADS normalised HP1 signal over transcription start sites (TSS) loci of larval-specific (red) and common (blue) genes. The vertical grey line represents the location of TSS, the plots span 1 kb upstream and downstream from annotated TSS. (F) A Venn diagram showing the overlaps between peak loci in the larval dataset and two independent embryonic (16–24 h) replicates. All three peak sets were filtered from peaks overlapping centromeric heterochromatin clusters and non-mappable, repetitive regions, defined based on GEM mappability analyses. modENCODE peak calls have been lifted from dm3 to dm6 genome assembly to match the analyses of larval HP1 peaks. The IDR method with 0.05 p-vale threshold was used to collapse the replicates to a common, high confidence peak set. (G) shotgun expression following overexpression of non-phosphorylatable STAT92E^Y704F measured using RT-qPCR and normalised to the expression of house-keeping gene RpL32. Four biological replicates are shown as distinct dots with three technical replicates each. *p = 0.02 (one-sample t-test comparing to 1). (H–J) E-cad-GFP expressed from the endogenous promoter is downregulated following E-cad overexpression (E-cad OS) from a heterologous UAS promoter. A representative image (H) shows endogenously expressed E-cad-GFP visualised with native GFP fluorescence (green, left; grey, middle) and total E-cad visualised with antibody staining (magenta, left; grey, right), scale bar – 10 μm. Levels of endogenously expressed E-cad-GFP are quantified in panel (I) and total E-cad levels in panel (J). **p = 0.0018 and ***p = 0.0003 (paired t-test).