Regulatory elements of the floral homeotic gene AGAMOUS identified by phylogenetic footprinting and shadowing

Abstract

In Arabidopsis thaliana, cis-regulatory sequences of the floral homeotic gene AGAMOUS (AG) are located in the second intron. This 3-kb intron contains binding sites for two direct activators of AG, LEAFY (LFY) and WUSCHEL (WUS), along with other putative regulatory elements. We have used phylogenetic footprinting and the related technique of phylogenetic shadowing to identify putative cis-regulatory elements in this intron. Among 29 Brassicaceae species, several other motifs, but not the LFY and WUS binding sites identified previously, are largely invariant. Using reporter gene analyses, we tested six of these motifs and found that they are all functionally important for the activity of AG regulatory sequences in A. thaliana. Although there is little obvious sequence similarity outside the Brassicaceae, the intron from cucumber AG has at least partial activity in A. thaliana. Our studies underscore the value of the comparative approach as a tool that complements gene-by-gene promoter dissection but also demonstrate that sequence-based studies alone are insufficient for a complete identification of cis-regulatory sites.

Figure 1.

Brassicaceae ITS DNA Phylogeny and AG Protein Phylogeny. (A) Rooted neighbor-joining distance tree of 5.8S rDNA ITS sequences using Akaike informational criterion log-likelihood DNA-substitution parameters. Bootstrap support (1000 replicates) is given next to the branches. This tree includes 26 of the 29 species analyzed for AG, plus additional species extracted from GenBank (see supplemental data online for accessions and sequences used). Outgroups were Aethionema, a basal Brassicaceae species, and Cleome, from the Capparaceae family, which some authors have included in the Brassicaceae sensu lato (Judd et al., 1994). The evolutionary distances between A. thaliana and other species used in this study are estimated between 5.8 million years (A. thaliana and A. arenosa) and 40 million years (Aethionema and the rest of the Brassicaceae) (Koch et al., 2001a). (B) Neighbor-joining distance tree of predicted AG protein sequences. Putative AG orthologs of Capsella rubella (CrAG), Capsella bursa-pastoris (CbpAG), Camelina sativa (CsaAG), Coronopus squamatus (CsAG), Lepidium phlebopetalum (LpAG), Eruca sativa (EsAG), Guillenia flavescens (GfAG), and Thlaspi arvense (TaAG) are from this study. Numbered suffixes designate AG proteins in cases of multiple genomic copies. SHP1 and SHP2 are the closest paralogs of AG in A. thaliana and serve as the outgroup to Brassicaceae AG, whereas the AG-like proteins from monocots and gymnosperms (ZAG1/2, ZMM1/2, OsMAD3a, SAG1a, and DAL2) serve as outgroups to AG homologs from dicotyledons (PTAG1/2, CUM1, TAG1, FAR, PLE, and PMADS3).

Figure 2.

Comparisons of AG Introns from Brassicaceae Species. (A) Sliding-window analysis. Flanking exon sequences are depicted as open boxes. KB14 and KB31 refer to complementary A. thaliana AG enhancers that confer similar expression patterns in reporter gene assays (Busch et al., 1999). Valleys indicate three regions of reduced sequence divergence and therefore high conservation. These regions also contain the only clusters of highly conserved blocks of at least 6 bp (>90% sequence identity across all positions and species), as indicated by diamonds on the line at top. (B) Conserved motifs. Invariant positions are shown in uppercase letters. For the consensus sequences of the aAGAAT box and CCAATCA box 1, Draba was excluded, because both motifs are largely deleted in the AG intron sequenced from this species. For the adjacent LFY and WUS binding sites (LBS/WBS), the core motifs to which LFY and WUS bind are underlined. Dots indicate insertions/deletions. Activity refers to effects seen when these sites are mutated in the context of A. thaliana sequences. (C) Observed distribution of invariant blocks compared with the distribution expected if individual invariant positions were arranged randomly within the sequence alignment. Green bars are shown in the foreground.

Figure 3.

Putative LFY Binding Site 4 in Brassicaceae AG Introns. Ovals denote the presence of LFY consensus binding sites (CCANTG[T/G]). The highlighted region indicates the position of LBS4 in A. thaliana (see supplemental data online). Some species have additional consensus motifs, whereas others lack them altogether. Species are ordered according to the phylogenetic relatedness of this region (data not shown). Numbers at bottom refer to the A. thaliana sequence.

Figure 4.

Requirement of Putative LFY Binding Sites 3 and 4 and CArG Boxes for the Activity of AG Enhancers. (A) and (D) KB14, wild-type 5′ AG enhancer reporter (Busch et al., 1999). (B) and (E) RH149, with LBS/WBS3 mutated. (C) and (F) RH141, with LBS4 mutated. (G) KB31, wild-type 3′ AG enhancer reporter (Busch et al., 1999). (H) MX144, with CArG box 1 mutated, shows ectopic GUS activity in the shoot apical meristem (asterisk). (I) MX144 in lfy-12. (J) MX215, with mutations in CArG box 1 and LBS1 and LBS2. (K) RH155, with CArG box 2 mutated. (L) RH174, with both CArG box 1 and 2 mutated. Sections of 5-bromo-4-chloro-3-indolyl-β-d-glucuronide–stained apices are shown. Staining intensities increase from orange to pink to purple. Arrows indicate staining in the center of early floral primordia between stages 3 and 6 and in the stamens, particularly in the developing filaments of stage-6 to -8 flowers. Bar in (A) = 50 μm for (A) to (C) and (G) to (L) and 100 μm for (D) to (F)

Figure 5.

Requirement of CCAATCA Boxes for the Maintenance of AG Expression. (A) and (E) KB31, wild-type 3′ AG enhancer reporter (Busch et al., 1999). (B) and (F) RH47, with CCAATCA box 1 deleted. (C) and (G) RH48, with CCAATCA box 2 deleted. (D) and (H) RH49, with both CCAATCA boxes deleted. Arrowheads in (E) to (H) indicate gynoecia from stage 8 on, and arrows indicate reporter gene activity at the base of the gynoecium. Asterisks indicate shoot apical meristems. Bar in (A) = 50 μm for (A) to (D) and 100 μm for (E) to (H).

Figure 6.

Locations of Putative cis-Regulatory Elements in AG Introns from Dicotyledons Outside the Brassicaceae. Dot plots of poplar paralogs, PTAG1 and PTAG2, and of orthologs from tomato and petunia, which are in different subfamilies of the Solanaceae, are shown at top. The window size was 25 bp, with a minimum of 50% identity and a step size of 1 character. Diagrams of AG introns with putative cis-regulatory motifs are shown at bottom. The position of the Tam3 transposon insertion, which results in ectopic expression of the PLE ovulata mutant (Bradley et al., 1993), is indicated by the arrowhead. Regions of extended similarity that include several motifs are highlighted in gray. The lengths of the introns are as follows: AG, 2999 bp; CUM1, 1993 bp; PMADS3, 4011 bp; PTAG1, 4864 bp; PTAG2, 3882 bp; PLE, 4087 bp; FAR, 2965 bp; and TAG1, 3251 bp.

Figure 7.

Conserved Motifs in Introns of AG Homologs from Outside the Brassicaceae. Asterisks indicate positions identical to the AG sequence in each alignment. At top, CCAATCA boxes (shaded) are separated by a more variable region. At bottom, the aAGAAT motif is part of a more extended region of similarity. Searches in the AliBaba2.1 (http://www.gene-regulation.com) transcription factor binding site database did not identify obvious candidates that could bind to the core aAGAAT motif.

Figure 8.

Activity of the CUM1 Intron in A. thaliana. (A) and (B) KB9, full-length A. thaliana AG intron reporter. (C) and (D) CUM1:GUS apices. Shoot apical meristems are indicated by asterisks, and numbers indicate floral stages (Smyth et al., 1990). The onset of expression (arrowhead in [C]) during the early stage (e3) is similar in AG:GUS and CUM1:GUS, but CUM1:GUS expression is not maintained as long, except for staining at the base of the gynoecium (D). Staining intensities increase from orange to pink to purple. g, gynoecium; st, stamens. Bar in (A) = 50 μm for (A) and (B) and 100 μm for (C) and (D).

Figure 9.

Summary of AG Regulation. Initiation and early AG expression in flowers requires all four LFY and LFY/WUS binding sites, as well as CArG box 2. The same LFY/WUS binding sites, along with a pair of CCAATCA boxes, also are required for the maintenance of AG expression in maturing carpels and stamens. In the shoot apical meristem (SAM), CArG box 1 mediates the repression of AG by a MADS domain protein(s). This repression appears to prevent the ectopic activation of AG by WUS, which is expressed in the shoot apical meristem (Mayer et al., 1998), and possibly by another unknown protein, factor Z.