We have investigated the role of Smad family proteins, known to be important cytoplasmic mediators of signals from the transforming growth factor-beta (TGF-beta) receptor serine/threonine kinases, in TGF-beta-dependent differentiation of hematopoietic cells, using as a model the human promyelocytic leukemia cell line, HL-60. TGF-beta-dependent differentiation of these cells to monocytes, but not retinoic acid-dependent differentiation to granulocytes, was accompanied by rapid phosphorylation and nuclear translocation of Smad2 and Smad3. Vitamin D(3) also induced phosphorylation of Smad2/3 and monocytic differentiation; however the effects were indirect, dependent on its ability to induce expression of TGF-beta1. Simultaneous treatment of these cells with TGF-beta1 and all-trans-retinoic acid (ATRA), which leads to almost equal numbers of granulocytes and monocytes, significantly reduced the level of phospho-Smad2/3 and its nuclear accumulation, compared with that in cells treated with TGF-beta1 alone. TGF-beta1 and ATRA activate P42/44 mitogen-activated protein (MAP) kinase with nearly identical kinetics, ruling out its involvement in these effects on Smad phosphorylation. Addition of the inhibitor-of-protein serine/threonine phosphatases, okadaic acid, blocks the ATRA-mediated reduction in TGF-beta-induced phospho-Smad2 and shifts the differentiation toward monocytic end points. In HL-60R mutant cells, which harbor a defective retinoic acid receptor-alpha (RAR-alpha), ATRA is unable to reduce levels of TGF-beta-induced phospho-Smad2/3, coincident with its inability to differentiate these cells along granulocytic pathways. Together, these data suggest a new level of cross-talk between ATRA and TGF-beta, whereby a putative RAR-alpha-dependent phosphatase activity limits the levels of phospho-Smad2/3 induced by TGF-beta, ultimately reducing the levels of nuclear Smad complexes mediating the TGF-beta-dependent differentiation of the cells to monocytic end points.