The probability of retroviral vector insertion but not the probability of forming a common insertion site depends on the expression level of the affected gene. (A) Array data from enriched hematopoietic progenitors containing both long-term and short-term repopulating cells (LSK cells, freshly isolated) were divided into 10 equal bins according to relative gene expression levels. The curves show the number of genes marked by RVISs in the different bins. Irrespective of the selection conditions (primary recipient, secondary recipient, or leukemia), the probability of RVIS is highest in the 40% most highly expressed genes. (B) Similar results were obtained when examining array data from LSK cells that were cultured for 2 days. (C) The selection for insertions in highly expressed genes is most pronounced for class 1 genes. (D) Expression levels of all genes detected by the arrays of LSK cells versus all RVIS genes of the IDDb, showing that the latter clearly have a much higher expression. The CRVIS genes of the IDDb are superimposed, showing that these do not cluster in the highest expression levels. Labeled genes represent CRVISs that were hit 3 times or more. Genes below the dotted line are not expressed in LSK cells.