Abstract
Retroviruses and retrovirus-derived vectors integrate nonrandomly into the genomes of host cells with specific preferences for transcribed genes, gene-rich regions, and CpG islands. However, the genomic features that influence the transcriptional activities of integrated retroviruses or retroviral vectors are poorly understood. We report here the cloning and characterization of avian sarcoma virus integration sites from chicken tumors. Growing progressively, dependent on high and stable expression of the transduced v-src oncogene, these tumors represent clonal expansions of cells bearing transcriptionally active replication-defective proviruses. Therefore, integration sites in our study distinguished genomic loci favorable for the expression of integrated retroviruses and gene transfer vectors. Analysis of integration sites from avian sarcoma virus-induced tumors showed strikingly nonrandom distribution, with proviruses found prevalently within or close to transcription units, particularly in genes broadly expressed in multiple tissues but not in tissue-specifically expressed genes. We infer that proviruses integrated in these genomic areas efficiently avoid transcriptional silencing and remain active for a long time during the growth of tumors. Defining the differences between unselected retroviral integration sites and sites selected for long-terminal-repeat-driven gene expression is relevant for retrovirus-mediated gene transfer and has ramifications for gene therapy.