Abstract : Cellular progression within the cell cycle is governed by a family of protein kinase complexes : the CDK/cyclin complexes. Their activity is regulated, notably by activating dephosphorylation brought about by CDC25 tyrosine-threonine dual specificity phosphatases. Three CDC25 phosphatases have been identified in mammalian cells, and CDC25B has been studied in the course of this work. In human adenocarcinoma (HeLa) cells, we showed that CDC25B is either strictly nuclear or cytoplasmic, or pancellular, suggesting active nucleo-cytoplasmic shuttling. Using site-directed mutagenesis and transient expression of wild-type or mutated proteins, we have identified a nuclear localization signal (NLS) and a nuclear export signal (NES) and showed that cytoplasmic sequestering of CDC25B requires its interaction with 14-3-3 proteins. In order to characterize the mechanisms regulating CDC25B function in vivo, we undertook a search for new interacting partners of this phosphatase. Starting from biochemical observations, e.g. the phosphorylation of CDC25B by Eg3 kinase in vitro, we established, in a cell line expressing conditionally CDC25B, that these proteins interact as well in vivo. A series of CDC25B mutant proteins allowed us to identify (i) the interaction site with protein Eg3, (ii) a phosphorylation site. Phosphorylation by Eg3 did not modify the phosphatase activity of CDC25B. However we showed by flow cytometry analysis that in vivo, CDC25B expression can cancel the piling up of cells in G2 phase caused by overexpression of Eg3. These last results strongly suggest a functional interaction between CDC25B and Eg3, whose mechanisms in vivo will require further studies. The isolation of new protein partners of CDC25B has been undertaken based on co-purification using U2OS cell lines conditionally expressing HA-tagged CDC25B, or recombinant protein MBP-CDC25B. Technical problems have as yet prevented the isolation of partners, but preliminary results are encouraging.