Abstract : The human disease neurofibromatosis type 1 is one of the most common genetic disorders because it has a de novo incidence of one in 3500 individuals. The NF1 gene has been identified and it has been demonstrated to be a tumor suppressor. It encodes a protein called neurofibromin (Nf1), which is ubiquitously expressed but most abundant in neurons, astrocytes, Schwann cells and oligodendrocytes. It is of great interest to understand the regulation of this protein. A model study has been developped in yeast in order to characterize the molecular basis of the interaction between Ira2p, the Nf1 homologue, and Tfs1p, a protein which inhibits it. We identified the molecular determinants of Tfs1p involved in the Tfs1p/Ira2p interaction: the accessibility of the surface pocket, its Nterminal region and the specific electrostatic properties of a large surface region containing these two elements. The physiological role of Tfs1p in cells has also been studied. Systematic studies suggest that when Tfs1p is over produced its most important function is the inhibition of Ira2p. We have shown that this inhibitory role is implicated in a negative feedback control of the stress response. On the basis of this study, we wanted to know if the Ira2p/Tfs1p interaction was conserved in human between Nf1 and RKIP, the human homologue of Tfs1p. In parallel, we have performed a two-hybrid screen between Nf1 and a cDNA library of human brain in order to identify new regulators of its functions. 1464 candidates have been isolated. Actually, 8% of the candidates have been identified and we have brought into light three interesting targets.