Analysis of the structural requirements for ERF-ERK interaction and their effect on ERF function

Abstract

ERF is the only transcription factor ofthe Ets family with two functional FXF motifs identified so far and this is the only example of two FXF motifs displaying different specificities with the two forms of ERK, the activated and the inactive. ERF is also the first protein shown to have different affinities for the active and inactive ERK, preferably associates with the activated ERK2-P rather than the inactive ERK2 with 4-fold higher affinity. ERF is an effector of the Ras/ERK pathway which is regulated through phosphorylation by ERKs affecting its localization and transcriptional repression. ERFinteracts with and is phosphorylated by ERK in vitro and in vivo. The specific efficient phosphorylation of ERF by ERKs requires its interaction through specific docking sites. ERF contains a substantial number of ERK interaction motifs and domains, yet only two docking sites were identified as functional ERK-docking sites. Two distinct FXF motifs have been identified within the ERK interaction domain (EID) of ERF, one (FSF motif) affecting association with the activated ERK2-P and the other (FKF motif) affecting association with the inactive and activated ERK2. Consistent with the deletion analysis the FSF motif is located within the phosphospecific region identified for association with the activated ERK while the FKF motif is located within the 30 amino acid stretch specific for the interaction with the inactive ERK. This is a unique feature of ERF whose distici functional FXF motifs recognize different forms ofERK2. Both FXF motifs are required for the efficient phosphorylation ofERF by ERK, with FSF playing the major role in promoting phosphorylation while the FKF on its own doesn’t seem to affect phosphorylation. The FSF docking site appears to direct phosphorylation of phosphoacceptor sites 3 and 4 (S248, S251). The phosphorylation deficient mutant Ml-7 where all putative phosphorylation sites are mutated to alanine could still associate with ERK2P indicating that phosphorylation ofERF at these sites does not affect the interaction. FSF mutations increased the nuclear localization of ERF. In contrast FKF mutations had no impact on the subcellular localization ofERF. Combination of both FXF mutations ie (FSF/FKF mutant) had an even more profound effect on the subcellular distribution of ERFincreasing its nuclear localization frequency. The increased nuclear localization ofthe ERKbinding deficient mutants is due to reduced phosphorylation indicating that the decreased association results in reduced phosphorylation. FXF mutations did not affect the nuclear export rate. All FXF mutants were excluded from the nucleus with similar rates after prolonged periods of induction suggesting that FXF mutations did not affect the nuclear export rate despite their implication in the import-export process. The binding deficient mutant of ERF-A66 where the phospho-specific FSF motif was removed exhibits reduced phosphorylation, increased nuclear localization and it had theability to partially affect cell proliferation as well as growth and morphology of Rastransformed cells. Expression of the binding deficient mutant missing the FSF motif in nontransformed NIH3T3 cells resulted in reduced BrdU incorporation and inhibition of cellular proliferation, whereas expression in ras-transformed cells resulted in partial reversion of the transformed phenotype. Our study strongly suggests that ERF is a high affinity substrate for ERKs and that this specific association of ERF with ERKs determine its biological function as an effector of the ras/ERK-signaling pathway.