Transcription factors serve critical roles in the progressive development of general body plan, organ commitment, and finally, specific cell types. Comparison of the biological roles of a series of individual members within a family permits some generalizations to be made regarding the developmental events that are likely to be regulated by a particular class of transcription factors. Here, we evidence that the developmental functions of the family of transcription factors characterized by the POU DNA binding motif exerts roles in mammalian development. The POU domain family of transcription factors was defined following the observation that the products of three mammalian genes, Pit-1, Oct-1, and Oct-2, and the protein encoded by the C. elegans gene unc-86, shared a region of homology, known as the POU domain1-9. The POU domain is a bipartite DNA binding domain10-13, consisting of two highly conserved regions, tethered by a variable linker. The approximately 75 amino acid N-terminal region was called the POU-specific domain and the C-terminal 60 amino acid region, the POU-homeodomain. High- affinity site-specific DNA binding by POU domain transcription factors requires both the POU-specific and the POU-homeodomain11-15. Resolution of the crystal structures of Oct-1 and Pit-1 POU domains bound to DNA as a monomer and homodimer, respectively, confirmed several of the in vitro findings regarding interactions of this bipartite DNA binding domain with DNA and has provided important information regarding the flexibility and versatility of POU domain proteins. Overall the crystal structure of a monomer of the Oct-1 POU domain bound to the octamer element was similar to that predicted by the NMR solution structures of the POU specific domain and the POU-homeodomain in solution, with the POU-specific domain consists of four alpha helices, with the second and third helices forming a structure similar to the helix-turn-helix motif of the λ and 434 repressors; several of the DNA base contacts are also conserved. A homodimer of the Pit-1 POU domain was crystallized bound to a Pit-1 dimer DNA element that is closely related to a site in the proximal promoter of the prolactin gene. The structure of the Pit-1 POU domain on DNA is very similar to that of Oct-1, and the Pit-1 POU-homeodomain/DNA structure is strikingly similar to that of other homeodomains, including the Oct-1 POU-homeodomain. The DNA contacts made by the Pit-1 POU-specific domain are also similar to those of Oct-1 and conserved with many made by the prokaryotic repressors. In the Oct-1 crystal, the POU-specific domain recognizes a GCAT half-site, while the corresponding sequence recognized by the Pit-1 POU-specific domain, GTAT, is on the opposing strand. As a result, the orientation of the Pit-1 POU-specific domain relative to the POU-homeodomain is flipped, as compared to the Oct-1 crystal structure, indicating the remarkable flexibility of the POU-specific domain in adapting to variations in sequence within the site. Also in contrast to the Oct-1 monomer structure is the observation that the POU- specific and POU-homeodomain of each Pit-1 molecule make major groove contacts on the same face of the DNA, consistent with the constraints imposed by its 15 amino acid linker. As a result, the Pit-1 POU domain homodimer essentially surrounds its DNA binding site. In the Pit-1 POU domain homodimer the dimerization interface is formed between the C-terminal end of helix 3 of the POU-homeodomain of one Pit-1 molecule and the N-terminus of helix 1 and the loop between helices 3 and 4 of the POU-specific domain of the other Pit- 1 molecule. In contrast to other homeodomain crystal structures, the C- terminus of helix 3 in the Pit-1 POU-homeodomain has an extended structure. Because each member of the class III POU domain gene family exhibits a distinct, yet overlapping, pattern of expression in the developing and mature nervous system, it is tempting to consider the possibility that combinatorial codes of specific POU III proteins are responsible for determining specific neuronal phenotypes.