Regulation of gene activityis fundamental for cells to control growth, differentiation, and adapt to environmental change. Genes encode information for producing biologically active molecules whose synthesis results from a series of linked processes. Transcription makes an RNA copy of the DNA and is one of the most important points in the control of gene activity. Genes are flexible units that can yield different quantities of RNA copies,as well as qualitatively different RNA copies, sometimes with significant ramifications for the final gene activity. The choice of transcription startsites (TSSs) location constitutes an important element for this flexibility of genes. Precise studies of TSSs in mammalian cells have also established that transcription occurs in both directions around genes. However, only transcription into the gene gives rise to stable RNA. In contrast, the divergent non-coding transcripts are rapidly degraded by an enzymatic complex called the exosome. In plants, few studies have attempted to identify TSSson a genome-wide level, and it is still unclear whether transcription is generally bidirectional or not. The objectives of this thesis were to (i) produce robust data on TSSs in the plant model Arabidopsis thaliana, either wild type or exosome mutants, (ii) leverage the data to clarify whether bidirectional transcription occurs, and (iii) investigate whether the choices of TSSs maybe of functional importance for rapid genetic reprogramming in connection with the plant immune response. We show that, contrary to mammals, bidirectional transcription is rare but not completely absent in Arabidopsis. We identify a likely explanation for this striking difference: Plants have more versatile machinery for the production of small regulatory RNA. In Arabidopsis, divergent transcripts are readily converted into small regulatory RNAs, potentially leading to aberrant repression of gene activity. This work has also uncovered exosome-sensitive antisense transcription in downstream parts of a considerable set of genes which was enriched in regulatory factor activities, therefore suggesting functional importance for this type of antisense transcription. The second part of this work reports that changes in TSSs occur quite frequently in genes induced during immune response activation. We identified several levels at which such changes could be functionally significant: For example,cases where the cellular location of the protein product may change, or cases suggesting differential efficiency of protein synthesis from the mature messenger RNA. Finally, this work has led to a new discovery of how genetic reprogramming works during the activation of the immune response: Our experimental design investigated gene activity very early after immune stimulation, and showed the induction of a rapid and transient wave of hitherto overlooked regulatory factors. Since these regulatory factors have the potential to activate many of the genes known to becharacteristic of the immunestate, our work provides novel insights into transcriptional reprogramming events taking place earlier than the well studied immune response in plants.