The promyelocytic leukemia-retinoic acid receptor alpha (PML-RARalpha) protein of acute promyelocytic leukemia (APL) is oncogenic in vivo. It has been hypothesized that the ability of PML-RARalpha to inhibit RARalpha function through PML-dependent aberrant recruitment of histone deacetylases (HDACs) and chromatin remodeling is the key initiating event for leukemogenesis. To elucidate the role of HDAC in this process, we have generated HDAC1-RARalpha fusion proteins and tested their activity and oncogenicity in vitro and in vivo in transgenic mice (TM). In parallel, we studied the in vivo leukemogenic potential of dominant negative (DN) and truncated RARalpha mutants, as well as that of PML-RARalpha mutants that are insensitive to retinoic acid. Surprisingly, although HDAC1-RARalpha did act as a bona fide DN RARalpha mutant in cellular in vitro and in cell culture, this fusion protein, as well as other DN RARalpha mutants, did not cause a block in myeloid differentiation in vivo in TM and were not leukemogenic. Comparative analysis of these TM and of TM/PML(-/-) and p53(-/-) compound mutants lends support to a model by which the RARalpha and PML blockade is necessary, but not sufficient, for leukemogenesis and the PML domain of the fusion protein provides unique functions that are required for leukemia initiation.