In the present study we have developed and validated a novel class of conditional gene trap vectors that activate gene expression at their insertion sites. Activation is achieved by Oct4 transcription factor binding to ectopic OPEs inserted into the target genes via gene trap vectors. By activating gene expression, these vectors make more genes accessible to trapping including genes that are poorly represented in the gene trap libraries. Since the increase in trappability also included antisense transcripts and intergenic regions, enhanced gene trapping will improve genome annotation and also aid the functional analysis of the over 40 000 antisense transcripts that are thought to be expressed in mouse genome (21
The enhanced vectors are highly mutagenic, report bona fide gene trap events and have no effect on genes neighboring the insertion sites. Thus, enhanced alleles may be used directly for generating knockout mice as they are unlikely to complicate interpretation of the resulting phenotypes.
Recently published quantitative gene expression data in E14 ESCs (22
), makes it possible to estimate the number of genes rendered accessible to trapping by the use of the enhanced vectors. Using minimum expression level to define trappability, numbers of trappable and untrappable genes can be estimated and compared for the different vectors (). As shown in , out of the 7435 genes of the Nord et al.
) data set, the nonenhanced FlipRosaβgeo vector would be predicted to trap 5170 genes (70% of genes in the Nord et al.
data set). This value for percentage of genes accessible to gene trapping is similar to a previous estimate for genome-wide trappability, which was based on 8000 full-length ENSEMBL genes trapped in the existing resources (5
). By comparison, the use of the eFlipRosaβgeo* vector would be predicted to increase the number of trappable genes in the Nord et al.
data set by about 15% (n
= 1108; ). When extrapolated to the entire genome, this suggests that nearly 85% of annotated protein-coding genes may be accessible through the use of enhanced gene trap vectors. This overall increase in trappability appears to be distributed among functional categories of protein coding genes, as the distribution of genes trapped with or without enhancer is similar among the GeneOntology-defined subclasses (Supplementary Table 3
Finally, the OPE is likely to find wider use beyond gene trapping. For example, enhanced gene trapping cassettes may increase the effficiency of ‘targeted’ trapping, which also requires gene expression (23
). In addition, the OPE may improve the performance of exogenous promoters that drive the expression of selectable marker genes in conventional gene targeting vectors.
In conclusion, the results described here are highly relevant to the worldwide large-scale ESC mutagenesis programs started recently under the umbrella of the International Knock Out Mouse Consortium (IKMC) (24
). The IKMC programs employ a combination of gene trapping and gene targeting in the effort to knock out every single gene in the mouse genome, and an optimal balance between the two technologies is sought in order to apply the most efficient mutagenesis strategy (24
). Because trapping is cheaper and generally involves less work, targeted mutagenesis is normally reserved for genes that are least accessible via trapping. Accordingly, genes well represented in gene trap libraries are generally excluded from gene targeting. Based on the results presented here however, we predict that the enhanced gene trapping approach will significantly increase the pool of genes accessible via trapping or targeting and thus reduce the overall effort and costs of the ongoing mouse mutagenesis programs. Furthermore, these results provide evidence that although gene-trapping vectors have been widely used for almost two decades, the evolution of these vectors is still ongoing and further vector innovations have the potential to significantly impact the accessibility of genes for functional analyses.