Aristolochia fimbriata (Aristolochiaceae) is a member of an early diverging lineage of flowering plants and a promising candidate for evo-devo studies. Aristolochia flowers exhibit a unique floral synorganization that consists of a monosymmetric and petaloid calyx formed by three congenitally fused sepals, and a gynostemium formed by the congenital fusion between stamens and the stigmatic region of the carpels. This floral ground plan atypical in the magnoliids can be used to evaluate the role of floral organ identity MADS-box genes during early flower evolution. In this study, we present in situ hybridization experiments for the homologs of the canonical C-, D-, and E-class genes. Spatiotemporal expression of the C-class gene AfimAG is restricted to stamens, ovary, and ovules, suggesting a conserved stamen and carpel identity function, consistent with that reported in core-eudicots and monocots. The D-class gene AfimSTK is detected in the anthers, the stigmas, the ovary, the ovules, the fruit, and the seeds, suggesting conserved roles in ovule and seed identity and unique roles in stamens, ovary, and fruit development. In addition, AfimSTK expression patterns in areas of organ abscission and dehiscence zones suggest putative roles linked to senescence processes. We found that both E-class genes are expressed in the anthers and the ovary; however, AfimSEP2 exhibits higher expression compared to AfimSEP1. These findings provide a comprehensive picture of the ancestral expression patterns of the canonical MADS-box floral organ identity genes and the foundations for further comparative analyses in other magnoliids.

Floral identity MADS-box A, B, C, D, E, and AGL6 class genes are predominantly single copy in Magnoliids, and predate the whole genome duplication (WGD) events in monocots and eudicots. By comparison with the model species Arabidopsis thaliana, the expression patterns of B-, C-, and D-class genes in stamen, carpel, and ovules are conserved in Aristolochia fimbriata, whereas A-, E-class, and AGL6 genes have different expression patterns. Nevertheless, the interactions of these proteins that act through multimeric complexes remain poorly known in early divergent angiosperms. This study evaluates protein interactions among all floral MADS-box A. fimbriata proteins using the Yeast Two Hybrid System (Y2H). We found no homodimers and less heterodimers formed by AfimFUL when compared to AfimAGL6, which allowed us to suggest AGL6 homodimers in combination with AfimSEP2 as the most likely tetramer in sepal identity. We found AfimAP3-AfimPI obligate heterodimers and AfimAG-AfimSEP2 protein interactions intact suggesting conserved stamen and carpel tetrameric complexes in A. fimbriata. We observed a broader interaction partner set for AfimSEP2 than for its paralog AfimSEP1. We show conserved and exclusive MADS-box protein interactions in A. fimbriata in comparison with other eudicot and monocot model species in order to establish plesiomorphic MADS-box protein floral networks in angiosperms.

Background: In Aristolochia (Aristolochiaceae) flowers, the congenital fusion of the anthers and the commissural, stigmatic lobes forms a gynostemium. Although the molecular bases associated to the apical-basal gynoecium patterning have been described in eudicots, comparative expression studies of the style and stigma regulatory genes have never been performed in early divergent angiosperms possessing a gynostemium. Results: In this study, we assess the expression of five genes typically involved in gynoecium development in Aristolochia fimbriata. We found that all five genes (AfimCRC , AfimSPT, AfimNGA, AfimHEC1 and AfimHEC3) are expressed in the ovary, the placenta, the ovules and the transmitting tract. In addition, only AfimHEC3, AfimNGA and AfimSPT are temporarily expressed during the initiation of the stigma, while none of the genes studied is maintained during the elaboration of the stigmatic surfaces in the gynostemium. Conclusions: Expression patterns suggest that CRC , HEC, NGA and SPT homologs establish ovary and style identity in Aristolochia fimbriata. Only NGA,HEC3 and SPT genes may play a role in the early differentiation of the stigmatic lobes, but none of the genes studied seems to control late stigma differentiation in the gynostemium. The data gathered so far raises the possibility that such transient expression early on provides sufficient signal for late stigma differentiation or that unidentified late identity genes are controlling stigma development in the gynostemium. Our data does not rule out the possibility that stigmas could correspond to staminal filaments with convergent pollen-receptive surfaces.