||The neurofibromatosis 1 gene (NFl) spans approximately 350 kb of genomic DNA. To develop a comprehensive screen for NFl mutations, I constructed a genomic DNA contig that included the entire NFl cDNA sequence. Using these genomic clones, I identified all intron-exon boundaries of the NFl coding region and established that the gene comprises 59 exons. The flanking intronic sequences obtained enabled me and others to design intron-based oligonucleotide primers to amplify NFl exons. I also discovered a locus homologous to NFl on chromosome 15. In spite of its high degree of sequence homology with NF1, frame-shifting deletions in some exons of the chromosome 15 locus destroy the intact coding sequence. NFl encodes a p21ras GTPase-activating protein (GAP). Since p21ras-GTP is a major regulator of cellular proliferation, mutant neurofibromins with diminished GAP activity might deregulate the ras-GTP level and contribute to the development of tumors. With this hypothesis in mind, I found an amino acid substitution (altering Lys-1423, in the NFl GAP-related-domain [NF1-GRD]) in three tumor samples: a colorectal adenocarcinoma, bone marrow from a myelodysplastic syndrome patient, and an anaplastic astrocytoma. I further proved that this substitution causes a drastic decrease in GAP activity. Since I also found somatic mutations in NFl in some breast cancer samples, I constructed several NF1 -deficient breast epithelial cell lines using an antisense-cDNA method. However, I have found no evidence that these cells are tumorigenic. To further explore the tumor-suppressive role of NF1, I overexpressed full-length NF 1 cDNA, as well as wild type and mutant (K1423E) NF1-GRD, in a colon carcinoma cell line, and demonstrated that overexpression of all three constructs suppressed the tumor phenotype. I also demonstrated that Raf-1 kinase (a downstream effector of p21ras) activity was reduced in cells with overexpressed neurofibromin and that reintroducing an activated Raf-1 kinase domain can reverse tumor suppression. I therefore concluded that tumor suppression by neurofibromin is not due to GAP activity, but rather to tight binding to p21ras-GTP.