The flower development of angiosperms is controlled by floral homeotic MIKCC-type MADS-domain transcription factors (MADS-TFs) that activate or repress target genes by forming floral organ specific DNA-bound heterotetrameric complexes termed floral quartets. The ability to form floral quartets highly differs between floral homeotic MADS-TFs of certain subfamilies. However, to date relatively little is known about how these subfamily-specific interaction patterns of floral homeotic proteins evolved during angiosperm evolution and which sequence determinants account for the different interaction capabilities. Based on interaction studies of floral homeotic proteins from early diverging angiosperms I could show that the interactions governing flower development in core eudicots are also present in these distantly related species. However, especially AP3- and PI-like proteins from early diverging angiosperms possess additional interactions compared to their orthologs from core eudicots which form obligate heterodimers only. The more diverse interactions among floral homeotic proteins from early diverging angiosperms suggest a shift from promiscuity to specificity in the protein-protein interaction network during early angiosperm evolution. By comprehensive amino acid sequence analyses of MADS-TFs I demonstrated that the structure of the protein-protein interacting keratin-like domain (K-domain) is most likely highly similar among all subfamilies of floral homeotic proteins. Amino acid substitutions within the K-domain of the floral homeotic hub protein SEP3 revealed that highly conserved leucine residues at interacting sites are essential mediators of floral quartet-like complex formation. The absence of leucine residues at homologous amino acid positions in non-hubs such as AP3- and PI-like proteins probably accounts for their less promiscuous interactions. Beside the highly specific protein-protein interactions among floral homeotic proteins I studied another interaction of the K-domain. The phytoplasma effector protein SAP54 targets the K-domain to specifically bind MADS-TFs of certain subfamilies and destines them for degradation. Based on amino acid sequence analyses and structural predictions I provided preliminary evidence that SAP54 folds into a structure similar to that of the K-domain. Based on my findings I hypothesized that SAP54 evolved via convergent sequence and structural evolution to mimic the K-domain of its MADS-TF targets.