What is the structure and mechanism of interaction of the CC-HH finger domains with nucleic acids? This question is addressed through a detailed structural analysis of the finger regions from several known nucleic acid binding proteins. A database of Zinc Finger proteins has been assembled for the purpose of statistical and structural modeling of the individual finger regions. The Zinc Finger Database was established to accumulate a complete collection of potential zinc finger gene sequences that could be rapidly searched by the members of the research community to prevent a major duplication of sequencing efforts. This collection contains both published and unpublished gene sequence data. The database is available for sequence comparisons on the DCRT Convex 240. The service to the research community provided is that new zinc finger gene sequences are e- mailed to the NIH, added to database, and a FASTA search of the results is returned without the alignments. A histogram of the statistical distribution of the search results and listing of the potential scores is included. When a match to an unpublished sequence occurs, then the name and contact information of the submitting author is returned so the concerned parties may correspond with each other. To date the collection contains 147 different entries. This large collection of functionally related sequences has provided excellent problem set for multiple sequence alignment and motif analysis tests. Statistical analysis of these data has revealed 5 repeat classes of CC-HH zinc finger domains ranging in length from 27 to 32 amino acids and 22 different repeat patterns. Furthermore, compositional statistics of the largest class of domains, 29 amino acid repeat length, reveals a remarkable conservation of serine or threonine when there is an arginine or glutamine in the DNA binding region of the finger domain. A correlation between the nucleic acid sequence was established and some domains have been observed. Constrained molecular dynamics simulations suggest that there are two general folding motifs for the zinc finger domains. Domains with sequences consistent with the consensus structure can adopt fold similar to the published x-ray data for the zf268 protein/dna co-crystal. While non- consensus structures adopt a flexible beta loop conformation, homology- based molecular models using the published x-ray crystallographic data are under construction. These homology-based structural models will be used for molecular docking experiments to explore the energetics of DNA sequence recognition.
