Homeotic genes specify the identity of different plant tissues and organs. Mutations in of these genes results in replacement of a specific organ or tissue by another plant body part or tissue, which is normally formed at a different position. An important group of homeotic genes in plants is represented by members of the MADS box family of transcription factors. The first members isolated from this large family were involved in specifying the identity of floral organs. Based on the genetic and molecular analysis of these genes the "ABC(DE)" model of floral development was established. Meanwhile, MADS box transcription factors with other kind of functions were identified, like the regulation of flowering time.
To gain insight in the molecular mechanisms underlying MADS box transcription factor functioning, the MADS box family of the model species Petunia hybrida has been analysed. MADS box proteins are active as homo- or heterodimers and even higher-order ternary and quaternary complex formation have been shown recently. Therefore, this study focuses mainly on the characterisation of protein-protein interactions. Initially, some new members of the petunia MADS box family were isolated from inflorescence and ovary cDNA-libraries, bringing the total number of identified petunia MADS box genes up-to 23. Members of this family are either named FLORAL BINDING PROTEINs ( FBPs ) or PETUNIA MADS ( pMADS ). The 23 members were all characterised by expression pattern analysis, phylogenetic comparison and yeast two-hybrid screenings for the identification of protein-protein interactions. The obtained results were compared with data from MADS box transcription factor families of other species, like Antirrhinum majus and Arabidopsis thaliana . This comparison demonstrated that MADS box genes and their functions are largely conserved among flowering plants and therefore, functions could be predicted for some unknown petunia and Arabidopsis MADS box proteins. Furthermore, some of the characterised 23 genes appeared to have one or more close relatives that share similar expression patterns and protein-protein interactions. Most likely these closely related genes are paralogues and functionally redundant. This phenomenon is more common among plant MADS box transcription factors and complicates their functional analysis because all redundant genes need to be suppressed in order to obtain a mutant phenotype. Cosuppression has proven to be a way to suppress multiple genes with a high percentage of sequence similarity and therefore, it is a suitable method to identify functions for redundant genes. This technology has been applied successfully to determine the function of the petunia MADS box transcription factor FBP10. Petunia transformants repressing this gene revealed to be unable to flower and therefore FBP10 was renamed to PETUNIA FLOWERING GENE (PFG) . Molecular analysis of these non-flowering plants demonstrated that besides PFG expression also the expression of FBP26 was repressed. However, most likely this repression is indirect and PFG and FBP26 are not functionally redundant.
Overexpression of a gene of interest is another way to analyse its function and this technology has been used to gain insight in the function of FBP20, which is a MADS box gene expressed during the vegetative phase of development. Plants overexpressing this gene had leaf-like petals and ectopic trichome formation on floral organs. Based on these phenotypic alterations the gene was renamed to UNSHAVEN ( UNS ). Surprisingly, overexpression of a truncated UNS protein lacking the MADS domain resulted in the same phenotypic changes in both petunia and Arabidopsis . Molecular analysis of the mutated plants in combination with yeast two-hybrid analyses revealed that the observed phenotype is most likely caused by a dominant-negative effect on another protein involved in maintenance of the floral identity instead of a gain of function of UNS. As a consequence, the native function of UNS is still unclear. Nevertheless, the obtained results show that conclusions concerning gene function drawn from overexpression studies should be handled with care.
To determine whether dominant-negative mutations can be used more generally to obtain MADS box transcription factor mutants, several mutated forms of PFG were overexpressed in petunia. In addition mutations for the MADS box transcription factor APETALA1 (AP1) from Arabidopsis were generated and tested for their ability to repress AP1 gene function in a dominant-negative manner. Unfortunately, overexpression of none of the generated constructs resulted in a dominant-negative effect, demonstrating that it is difficult to generate a universal dominant-negative strategy. Nevertheless, this study has elucidatedrevealed the importance of some motifs and domains in the AP1 protein for its functioning. From these experiments it was also clear that more knowledge about the function of specific MADS box protein domains and their interactions is essential to predict which domain needs to be modified to obtain a dominant-negative version. To obtain information about the interactions between MADS box transcription factors, yeast two hybrid-experiments can be performed, because these systems have proven to be powerful methods to determinegain knowledge about protein-protein interactions. However, no knowledge is obtained about the occurrence existence of these interactions in living plant cells and in which cell compartments the interactions take place. To get insight in the biological relevance of the identified interactions, in-planta analyses need to be performed. For this reason a few of the identified petunia MADS box protein dimers were analysed in leaf protoplasts by means of spectroscopy techniques based on Fluorescence Resonance Energy Transfer (FRET). All the in yeast identified heterodimers for the ovule specific MADS box protein FBP11 could be confirmed in living cells and, in addition, homodimerisation was observedtained for the MADS box proteins FBP2, FBP5 and FBP9. Subsequently, localisation studies were conductedperformed with full-length and truncated proteins lacking the supposed bipartite Nuclear Localisation Signal (NLS), making use of Confocal Laser Scanning Microscopy (CLSM). The obtained results demonstratedshowed that both partners of a MADS box protein dimer need to contain the NLS signal for nuclear localisation and furthermore, dimerisation appeared to be essential for this translocation from cytoplasm to nucleus. All these observations demonstrate again the importance of protein-protein interactions for MADS box protein functioning. The knowledge about these interactions does not only teach us about the molecular mechanisms underlying MADS box transcription factor functioning but gives also information about the presence of functional redundancy among members of this large family of transcription factors. Furthermore, a pilot experiment has shown that protein-protein interaction screenings can be exploited for the identification of functional homologues (orthologues) of specific MADS box proteins from species that are genetically not very well characterised. These findings demonstrate that interaction mapping provide an additional tool for functional genomics.