Lentiviral vector and gamma-retroviral vector integration site distributions in the murine genome. Vector integration sites for pooled samples were compared to their matched random control sites. In the matching procedure (20), each unique integration site was matched with 10 control sites in the genome randomly selected in silico that were constrained to lie the same distance from an MseI recognition site as the experimentally determined integration site. Comparison of experimentally determined integration sites to the matched random controls thus ‘washed out’ any possible biases introduced by the use of MseI cleavage. A value above 1 indicates favored integration relative to random control sites; a value below 1 indicates disfavored integration. (A) Frequency of integration in transcription units. MEF: control integration sites from cultured murine embryonic fibroblasts. Healthy: control healthy mice. Tumor: mice with lymphoproliferation. (B) Frequency of integration near transcription start site. (C) Frequency of integration near CpG islands. (D) Frequency of integration in transcription units. In this analysis the integration site data sets are pooled for all tissues from each mouse. (E) Frequency of integration in transcription units for integration sites pooled by tissue of origin (samples from liver are not included in this analysis due to the low number of integration sites). Comparisons between the lentiviral and gamma-retroviral vectors in each of the panels achieved P <0.0001 (Fisher's exact test). Comprehensive analysis of integration frequency relative to many genomic features can be found in Supplementary Data 1–3.

Severe biases in recovery of integration sites arising due to use of different restriction enzymes. (A) Venn diagram indicating the overlap between integration sites isolated by cleaving genomic DNA with MseI versus Tsp509I. (B) Collector's curve (rarefaction) analysis of integration sites recovered after MseI cleavage. Repeated samples of integration site subsets were used to evaluate whether further sampling would likely yield additional integration sites (rarefaction analysis), as indicated by whether the curve has reached a plateau value. The y-axis indicates the number of integration site sequences detected, the x-axis the number of integration sites in the subset analyzed. ‘Sob’ (Species-observed) indicates rarefaction on the original data. The Chao1 estimator was used to estimate the number of undetected integration sites from frequency of isolation information. ‘Chao1’ indicates collector's curve analysis on Chao1 estimates for sequence subsets. (C) Collector's curve analysis for integration sites recovered after cleavage of genomic DNA with Tsp509I. Markings as in (B).

The DNA bar coding strategy. Each LTR primer used in ligation-mediated PCR contained a unique DNA barcode that specified the mouse and tissue of origin. Each barcode consists of a unique 4-bp nucleotide sequence, inserted between the sequencing primer binding site and the LTR specific primer segment. Thus all sequencing reads begin with the 4-bp barcode identifiers. A sample primer with a bar code is shown at the bottom of the diagram.

Correlations of clone counts between tissues in mice with abnormal lymphoproliferation (A and B) or healthy controls (C and D). The mouse studied is indicated on the figure, along with the genes nearest the integration site. Each point represents an individual integration site; the values on the x- and y-axes indicate the number of sequences for each clone. The square root of clone counts for each tissue is plotted to allow very high counts to be displayed conveniently. Integration sites at known proto-oncogenes are indicated by the larger red lettering. The r-values indicate the Spearman correlations for counts between tissues. For detailed analysis of all mice see Supplementary Data 4. We note that it is probable that not all potential cancer-related genes have been identified. Also, for any given insertion, additional studies are required to establish whether the integration event up-regulates Cancergene transcription.

Bias in recovery of integration sites due to the distance between integration sites and the closest MseI sites. The distribution of counts of identical sequence reads is shown as a function of distance to the nearest restriction site. The y-axis represents the percentage of integration sites with the indicated frequencies of isolation, and the x-axis is the range of distances between integration sites and their nearby MseI restriction sites used for binning.