MADS dynamics : gene regulation in flower development by changes in chromatin structure and MADS-domain protein binding

Abstract During the life cycle, a plant undergoes a series of developmental phase changes. The first phase change is the transition from the initial juvenile vegetative stage into the adult vegetative phase. During the juvenile phase plants produce leaves and axillary buds, whereas during the adult phase the initiation of reproductive structures occurs. The next developmental change is the switch from vegetative to reproductive growth, when the shoot apical meristem acquires the identity of an inflorescence meristem that will then produce floral meristems. Arabidopsis floral meristems produce four concentric whorls of floral organs: sepals, petals, stamens and carpels. Each developmental change is controlled by coordinated network of regulators, known as gene regulatory networks (GRNs), which determine the transcription of a specific set of genes. The aim of the study presented in this thesis was to understand the dynamics of GRNs during floral organ development in Arabidopsis and correlate the binding of key regulatory MADS domain transcription factors with the accessibility of the chromatin in a genome-wide context. In chapter 1 and 2 we reviewed the current knowledge on the regulation of transcription in the model plant Arabidopsis thaliana. In chapter 1 we mainly focus on how the view of the GRN underling flower development has changed during the last decades, while in chapter 2 we more broadly revised the mechanisms that control developmental switches in plants. The recent introduction of next-generation sequencing and genome-wide approaches has changed our view on gene regulation and GRNs. We moved from linear genetic interactions towards global highly connected gene networks. The high numbers of interactions that were detected in protein-DNA binding profiles revealed a much higher network complexity than previously anticipated and demonstrated that master regulators of development not only control another layer of regulators, but also genes encoding structural proteins, enzymes and signalling proteins. Moreover, most transcription factors bind to their own locus, highlighting that auto-regulatory loops are a common mechanism of regulation. The discovery of interactions between transcriptional master regulators with epigenetic factors provides new insights into general transcriptional regulatory mechanisms. Switches of developmental programmes and cell fates in complex organisms are controlled at the level of gene expression by the combined action of chromatin regulators and transcription factors. Although many master regulators of meristem and organ identities have been identified, it is still not well understood how they act at the molecular level and how they can switch an entire developmental program in which thousands of genes are involved. Using flower development as a model system, in chapters 3 and 4 we investigated general concepts of transcription regulation by analysing the dynamics of protein-DNA binding, chromatin accessibility and gene expression. Using an inducible system for synchronised flower formation, we characterised DNA-binding profiles of two MADS-domain transcription factors, APETALA1 (AP1) and SEPALLATA3 (SEP3), at three stages of flower development. Our study revealed that these MADS-domain proteins, select their binding sites, and thereby their target genes, in a partly stage-specific fashion. By combining the information from DNA-binding and gene expression data, we proposed models of stage-specific GRNs in flower development. Since developmental control of gene expression is tightly linked with dynamic changes in chromatin accessibility, we identified DNase I hypersensitive sites (DHSs, chapter 3) and we characterised nucleosome occupancy (chapter 4) at different stages of flower development. We observed dynamics in chromatin landscape manifested in increasing and decreasing DHSs as well as in changes in nucleosome occupancy and position. Next, we addressed the question how MADS-domain protein stage-specific binding is achieved at the molecular level in a chromatin context. In the nucleus the DNA is wrapped around histone octamers to form nucleosomes, which are then packed into highly dense structures, and hence transcription factor binding sites may not be easily accessible. A result of the combined analysis of MADS-domain binding and chromatin dynamics is that MADS-domain proteins bind prevalently to nucleosome depleted regions, and that binding of AP1 and SEP3 to DNA precedes opening of the chromatin, which suggests that these MADS-domain transcription factors may act as so-called “pioneer factors”. The isolation and analysis of developing flowers of specific stages increased the specificity of our genome-wide experiments, enabling the identification of novel actors in the GRN that regulates flower development. In this thesis we characterised the role of some novel regulators in more detail: in chapter 3 we focussed on the GROWTH REGULATING FACTOR (GRF) family genes; in chapter 5 we investigated the action of STERILE APETALA (SAP); and in chapter 6 we elucidated the regulation and the role of a member of the WUSCHEL-related homeobox (WOX) family, WOX12. GRF family genes are dynamically bound by AP1 and SEP3 at the different stages of flower development. All family members are bound by SEP3, while only a subset of the genes is bound by AP1. The defects in floral organs observed upon down-regulation of these genes highlight their role down-stream of MADS-domain transcription factors. In addition to AP1 and SEP3, SAP is also a target of other MADS-domain proteins, such as APETALA3 (AP3), PISTILLATA (PI), and AGAMOUS (AG). SAP is strongly expressed in meristems and loss of function of SAP causes strong aberrations in flowers, such as a reduction in petal and stamen numbers. We found that SAP interacts with proteins of the SCF ubiquitin ligase complex, suggesting that SAP could act in the ubiquitination pathway. WOX12 down-regulation leads to defects in floral organ identity specification with the formation of stamenoid-petals, while ectopic expression of WOX12 leads to an opposite effect: it causes the formation of petaloid-stamens in the third whorl. WOX12 acts downstream of AP1. Ectopic expression of WOX12 leads to reduction of AG expression, suggesting a role for WOX12 in the antagonistic interplay between the homeotic genes AP1 and AG. In chapter 7 we discuss the findings of this thesis. Taken together, the work performed in this thesis increased our knowledge on the GRN that regulates flower development and on the mode of action of MADS-domain transcription factors. We hypothesise that MADS-domain proteins may act as pioneer factors, proteins that access and remodel condensed chromatin. However, differently from other pioneer factors, MADS-domain transcription factors do not actively deplete nucleosomes, but instead they interact with chromatin remodelers to shape chromatin landscape. Given the important roles of MADS-domain proteins as master regulators of developmental switches, their pioneer behaviour represents an intriguing mode of action.