The Effects of T-DNA Integration Sites on Transgene Expression in Arabidopsis

Abstract

Several recent investigations of T-DNA integration sites in Arabidopsis thaliana have reported "cold spots" of integration, especially near centromeric regions. These observations have contributed to the ongoing debate over whether T-DNA integration is random, or if integration preferentially occurs in transcriptionally active regions. When transgenic plants are identified by selecting or screening for transgene activity, transformants with integrations into genomic regions that suppress transcription, such as heterochromatin, may not be identified. This phenomenon, which we call selection bias, may explain the perceived non-random distribution of T-DNA integration in previous studies. In order to investigate this possibility, I have characterized the sites of T-DNA integration in the genomes of transgenic plants identified by pooled PCR, a procedure that does not required expression of the transgene and is therefore free of selection bias.

Over 100 transgenic Arabidopsis plants were identified by PCR and compared to kanamycin-selected transformants from the same T1 seed pool. A higher perceived transformation efficiency and a higher frequency of transgene silencing were observed in the PCR-identified lines. Together, the data suggest approximately 30% of transformation events may result in non-expressing transgenes that would preclude identification by selection. Genomic integration sites in PCR-identified lines were compared to those in existing T-DNA integration databases. In PCR-identified lines with silenced transgenes, the integration sites mapped to regions significantly underrepresented by T-DNA integrations in studies where transformants were identified by selection. The data presented here suggest that selection bias can account for at least some of the observed non-random integration of T-DNA into the Arabidopsis genome.