Mapping the protein regions responsible for the functional specificities of the Arabidopsis MADS domain organ - identity proteins ( flower development / domain swapping )

The Arabidopsis MADS domain proteins API, AP3, PI, andAG specify floral organ identity. All of these proteins contain a MADS domain required for DNA binding and dimerization; a region termed L (linker between MADS domain and K domain), which plays an important role in dimerization specificity; the K domain, named for its similarity to the coiled-coil domain of keratin; and a C-terminal region of unknown function. To determine which regions of these proteins are responsible for their abilities to specify different organs, we have made a number of chimeric MADS box genes. The in vivo function of these chimeric genes was investigated by ectopic expression in transgenic Arabidopsis plants. The four proteins fall into two classes on the basis of regions responsible for their functional specificities. The L region and K domain define the functional specificities ofAP3 and PI, while the MADS domain and L region define the functional specificities of AP1 and AG. One general question in the molecular analysis of development is how homeotic genes, which are key regulators of organ or body-segment identity, perform their functions. Many homeotic genes have been cloned and shown to encode DNAbinding proteins. While these proteins are thought to act by binding to the promoters of different genes and thus regulating the spatial and temporal expression patterns of these genes, the mechanism leading to the activation or repression of specific genes is often unknown, as is (for the most part) the identity and function of the downstream genes. In some cases, a group of closely related homeotic genes act to specify different developmental pathways in adjacent regions. Two examples are homeotic selector genes in insects and vertebrates, which encode DNA-binding homeodomain proteins that act in different segments to specify different fates (reviewed in refs. 1 and 2), and organ-identity genes in flowers, which encode DNA-binding MADS domain proteins that act to specify organ primordia as sepals, petals, stamens, or carpels (reviewed in refs. 3-5). In the case of homeotic genes that are members of gene families yet have different functions in vivo, a way to investigate the mode of action of their gene products is to identify the particular region(s) of the proteins that is responsible for the different organ-specifying activities. This can be done by making chimeric genes, fusing parts from proteins that are related but have different developmental functions, and then assaying the in vivo function of the chimeric proteins. If the DNA-binding regions are implicated, then the different functions of the proteins may be a result of intrinsic differences in sequence specificity conferred by their related but nonidentical DNA-binding domains. This would cause the proteins to 4063 bind different DNA sequences and regulate different sets of genes. Downstream genes could then be identified by determination of the binding-site selectivities of the proteins by using in vitro techniques, followed by identification of these sequences in the genome. Alternatively, the DNA-binding domains could all be equivalent, with other protein regions responsible for the different functions of the related proteins. In this case, an understanding of the specificity of these regulators will likely lie in understanding their interactions with other proteins, which may either direct the regulators to different genomic regions or dictate different functions of the homeotic proteins when they are bound to the same genomic sequences. Here, we have made chimeric proteins based on four different plant MADS domain proteins [APETALA1 (AP1), APETALA3 (AP3), PISTILLATA (PI), and AGAMOUS (AG)], each with a similar structure but different functions in floral organ specification. Several years ago a genetic model was developed in which the activities of these four proteins were assigned to three different classes, A, B, and C, with AP1 anA function protein, AP3 and PI required for B function, and AG required for C function (6-9). The proteins were proposed to function combinatorially in the specification of floral organs, such that A function specifies sepals, A function in combination with B function specifies petals, B and C functions together specify stamens, and C function specifies carpels. All of these proteins are members of the MADS domain family of transcription factors which includes proteins from yeast (MCM1), animals (SRF, MEF2), and plants (reviewed in ref. 10). They share a conserved 56-amino acid DNA-binding and dimerization motif called the MADS domain. The four plant proteins also share an additional region, the K domain, which has low similarity at the amino acid level but is predicted to form amphipathic a-helices in each protein (11, 12). The region between the MADS and K domains is called the L, or linker, domain. Ectopic expression of each of the four MADS box genes under the control of the constitutive 35S cauliflower mosaic virus promoter (p35S) produces a distinctive dominant gainof-function phenotype (13-17). The uniqueness of these ectopic expression phenotypes has allowed us to assay the ability of chimeric MADS box genes to act like their respective parental genes in transgenic Arabidopsis plants. We find that the functional specificities of these four proteins fall into two classes with the functional specificity ofAP3 and PI dependent on their L and K regions, while the specificity of AG and AP1 Abbreviations: AP1, APETALA1; AP3, APETALA3; PI, PISTILLATA; AG, AGAMOUS. *To whom reprint requests should be addressed. 4064 Plant Biology: Krizek and Meyerowitz resides in their MADS and L domains. These differences may result from different protein-protein interaction surfaces and/or different MADS-domain dimerization requirements but do not implicate different intrinsic DNA-binding properties in functional specificity. MATERIALS AND METHODS Construction of Chimeric Genes. The chimeric MADS box genes were constructed by using PCR mutagenesis, involving two rounds of PCR. In the first round, one outer primer (containing either a BamHI or an Xba I site) and one internal primer (containing sequences from both of the MADS box parental genes to be fused) were used. The products of these two PCR reactions were then used in a second PCR reaction with the outer primers. All constructs were ligated into the BamHI and Xba I sites of pGEM-3Z into which an 842-bp 35S promoter had previously been inserted in the Asp718 and BamHI sites. Correct sequences were confirmed by doublestranded sequencing of the recombinant plasmids. The 35S promoter-chimeric MADS genes were then cloned into the Asp718 andXba I sites of pCGN1547 (18) containing a 253-bp sequence from the 3' end of nopaline synthase in theXba I and Pst I sites.