Structural basis for specific gene regulation by Auxin Response Factors

Auxin is a plant hormone that triggers a broad variety of responses during plant development. These responses range from correct cell division patterns during embryogenesis to formation and growth of different organs. Due to its importance for plant growth and development, many aspects of the biology of auxin have been studied. In Chapter 2, we use Arabidopsis embryogenesis as a stage to describe generalities about its biosynthesis, transport, components of its signaling pathway and transcriptional control of some know target genes. As most of the players involved in transcriptional regulation in response to auxin have been identified, the question of how the same signal can elicit so many different responses remains open. In this thesis we approach this issue by focusing on the ultimate effectors of the auxin signaling pathway: the ARF family of transcription factors. In Chapter 3 we present the crystal structure of the DNA binding Domain (DBD) of two divergent members of the family: ARF1 and ARF5. Careful observation of the structures, followed by in vitro and in vivo experiments led to the following conclusions: 1) ARF DBDs dimerize through a conserved alpha-helix, and bind cooperatively to an inverted repeat of the canonical TGTCTC AuxRE. Dimerization of this domain is important for high-affinity DNA binding and in vivo activity. 2) Monomeric ARFs have the same binding preference for the DNA sequence TGTCGG (determined by protein binding microarray). 3) DNA-contacting residues are almost completely conserved within the ARF family members. 4) The distance between the AuxREs may play a role for binding of specific ARF dimers as for example, ARF5 can accommodate and bind to different spacing (6-9 bp) compared to ARF1 which is more rigid (7-8 bp). In Chapter 4 we follow up on the observations made. First we again used structural biology to determine the reason of the high binding affinity to the TGTCGG sequence compared to the previously identified canonical TGTCTC element. We found that in complex with TGTCGG, His137 (ARF1) could rotate and make hydrogen bonds with either G5 or G6, as well as a hydrogen bond with the C opposing to G6. This rotation is not possible when in complex with TGTCTC and there the same histidine can make only one hydrogen bond with the G opposing to C6. We conclude then that this histidine plays a role in determining the strength of binding to TGTCNN elements and that this also reflects in its specific transcriptional activity as mutating the corresponding histidine in ARF5 renders a semi-functional protein in vivo (Chapter 3). The next observation we followed up in Chapter 4 is the biological meaning of ARF DBDcooperative binding to DNA. We identified AuxRE inverted repeats (IR) in the promoter of the TMO5 gene and mutated them. This brought the expression of the gene to very low levels despite the presence of other multiple single AuxREs. Thus, the single inverted AuxRE repeat in the TMO5 promoter is essential for ARF5 binding and gene regulation. Importantly, mutating only a single AuxRE element within the inverted repeat led to very pronounced loss of activity, consistent with requirement of both AuxRE sites for high-affinity ARF5 binding. We then concluded that IR AuxREs have a significant effect in gene regulation by ARFs. Next we search the genome for bipartite AuxREs that correlated to auxin response and found two main elements: inverted repeat with 8 bases of spacing (IR8) and direct repeat with 5 bases of spacing (DR5). As this kind of bipartite AuxREs are rarer to find than single AuxREs, we tested their presence in promoters as predictors of auxin responsiveness by qPCR. We found that about 75% of the selected genes containing either IR8 or DR5 responded to auxin. The expression study also show that genes containing the DR5 sequence were only up-regulated when regulated. Interestingly, Surface Plasmon Resonance study showed that only class A (activator) ARFs can bind the DR5 sequence cooperatively. As the structural differences of ARFs DBDs are subtle, we then asked if specific gene targeting is determined by this domain alone. In Chapter 5 we used a DBD swap experiment and conclude that the DBD is necessary for specific gene targeting but not sufficient and the other domains of an ARF also contribute in its specific activity. In Chapter 5 we expand our focus from the DBD to the other ARF domains, Middle Region (MR) and C-terminal (CT). As ARFs have protein-protein interaction interfaces in all three domains, we expressed the isolated domains of ARF5 and perform immuno-precipitation followed by tandem mass-spectrometry. Although the procedure needs optimization, some interactions expected for each domain could be identified. The DBD showed to interact with the general transcription machinery and the CT could interact with another ARF and 3 Aux/IAA. These interactions seem to be specific as the Aux/IAA recovered are not the most abundant in the sampled tissue. Finally, in Chapter 6 all the obtained results are put in a broader context and new questions derived from our results are proposed.