c-Yes, a member of the Src tyrosine kinase family, is found highly activated in colon carcinoma but its importance relative to c-Src has remained unclear. Here we show that, in HT29 colon carcinoma cells, silencing of c-Yes, but not of c-Src, selectively leads to an increase of cell clustering associated with a localisation of β-catenin at cell membranes and a reduction of expression of β-catenin target genes. c-Yes silencing induced an increase in apoptosis, inhibition of growth in soft-agar and in mouse xenografts, inhibition of cell migration and loss of the capacity to generate liver metastases in mice. Re-introduction of c-Yes, but not c -Src, restores transforming properties of c-Yes depleted cells. Moreover, we found that c-Yes kinase activity is required for its role in β-catenin localisation and growth in soft agar, whereas kinase activity is dispensable for its role in cell migration. We conclude that c-Yes regulates specific oncogenic signalling pathways important for colon cancer progression that is not shared with c-Src.

Corresponding authors: LP (Tel +33 22323 4473, luc.paillard@univ-rennes1.fr); and YA (Tel +33 22323 4475, yann.audic@univ-rennes1.fr).Running title CELF1, ELAVL1, and splice isoform production 1 5 10 15 20 ABSTRACT ELAVL1 and CELF1 are RNA-binding proteins that are involved in alternative splicing control. To explore their functional relationship, we looked for mRNAs that are differentially spliced following the depletion of CELF1, ELAVL1, or both. We found that these proteins control the usage of their target exons in the same direction. Double depletion has a greater effect than individual depletions, showing that CELF1 and ELAVL1 exert additive control. To confirm these results, we carried out RT-PCR on the alternative cassette exons of several mRNAs, including CD44, WNK1, PHACTR2, MICAL3, SPTBN1, and PPP3CB. Using FRET, we found that CELF1 and ELAVL1 directly interact in cell nuclei. We demonstrated that the combined levels of CELF1 and ELAVL1 are a valuable biomarker in several cancers, even when their individual levels may yield very limited information. CD44 alternative splicing probably accounts in part for the effects of CELF1 and ELAVL1 levels on patient survival. These data point to strong functional interactions between CELF1 and ELAVL1 in the control of mRNA isoform production, resulting in significant impacts on human pathology.

No abstract

In an era when numerous disease-associated genes have been identified, determining the molecular mechanisms of complex diseases is still difficult. The CELF1 region was identified by genome-wide association studies as an Alzheimer's disease (AD) risk locus. Using transcriptomics and cross-linking and immunoprecipitation sequencing (CLIP-seq), we unexpectedly found that CELF1, an RNA-binding protein, binds to KLC1 RNA and regulates its splicing. Analysis of two brain banks revealed that CELF1 expression is correlated with inclusion of KLC1 exons downstream of the CELF1-binding region identified by CLIP-seq. In AD, low CELF1 levels result in high levels of KLC1 splice variant E (KLC1_vE), an amyloid- β (A β) pathology-driving gene product. Cell culture experiments confirmed regulation of KLC1_vE by CELF1. Analysis of mouse strains with different propensities for A β accumulation confirmed that Klc1_vE drives A β pathology. Using omics methods, we revealed the molecular pathway of a complex disease supported by human and mouse genetics.