The regulatory regions of af9, p16INK4a, and p15INK4b were investigated in depth using computer algorithms designed to predict the locations of promoters and upstream transcription factor binding sites in DNA sequence. The evolution of these regions also was investigated through analysis of the repeat elements found in the sequence data.

In an effort to gain a deeper understanding of the primary structure of human genomic DNA regions implicated in specific instances of tumorigenesis, over a third of a megabase of human genomic DNA has been sequenced from four cosmids from human chromosome band 9p22, four cosmids from human chromosome band 9p21, a P1 clone from human chromosome band 3q26, and a single cosmid from human chromosome band 9q34. The 9p22 cosmids include over 80% of the af9 gene that is involved in the leukemogenic translocation with the MLL gene on human chromosome band 11q23 and ten breakpoints have been mapped to the second of the three contigs formed by these four cosmids. The entire coding sequence of two genes, p16INK4a and p15INK4b, is found within the four cosmids in chromosome band 9p21 and this chromosomal region frequently is deleted in cancers originating in many different tissues. Chromosome band 3q26 also is involved in leukemogenic translocations and the P1 clone sequenced as part of the research presented here includes a single exon from the MDS 1 gene. The TSC1 gene is found in human chromosome band 9q33 and although the single cosmid sequenced originally was thought to contain this gene, it was not found in this cosmid. This cosmid contains several Alu repeat elements and subclones from this cosmid were used in the latter portion of this dissertation research to investigate the ability of Alu repeat sequences to bind protein at the DNA level.

In all the sequence determined, four genes were found and the regulation of three of these genes was investigated through computer predicted transcription factor binding sites. The presence of various computer predicted promoters and transcription factor binding sites confirmed and extended the known tissue specific expression patterns for the human proto-oncogenes encoded in the sequenced regions.

A shotgun sequencing strategy was employed to determine the unambiguous nucleotide sequences reported in this dissertation. Here, randomly generated fragments from each genomic clone were subcloned initially into M13mp18 and the latter stages into pUCmp18. After isolation by a modified cleared lysate procedure, each subclone was end-sequenced employing cloning vector specific universal forward or reverse primers. The resulting nested fragment set was resolved by electrophoresis on ABI377 automated sequencers and the sequence data was assembled on Sun workstations using the phred and phrap software. Each cosmid was assembled into a single contiguous sequence and discrepancies resolved through primer-walking reactions with custom synthesized primers on subclones, PCR product, or the original genomic clone and all projects have reactions covering both orientations for over 95% of their length. The sequence of each cosmid was analyzed with GRAIL and GENSCAN to predict putative exons, RepeatMasker and Sputnik for determine the location of repetitive DNA, and BLAST to find regions with homology to other known genes.