The Arabidopsis BELL1 and KNOX TALE homeodomain proteins interact through a domain conserved between plants and animals

Abstract

Interactions between TALE (three-amino acid loop extension) homeodomain proteins play important roles in the development of both fungi and animals. Although in plants, two different subclasses of TALE proteins include important developmental regulators, the existence of interactions between plant TALE proteins has remained unexplored. We have used the yeast two-hybrid system to demonstrate that the Arabidopsis BELL1 (BEL1) homeodomain protein can selectively heterodimerize with specific KNAT homeodomain proteins. Interaction is mediated by BEL1 sequences N terminal to the homeodomain and KNAT sequences including the MEINOX domain. These findings validate the hypothesis that the MEINOX domain has been conserved between plants and animals as an interaction domain for developmental regulators. In yeast, BEL1 and KNAT proteins can activate transcription only as a heterodimeric complex, suggesting a role for such complexes in planta. Finally, overlapping patterns of BEL1 and SHOOT MERISTEMLESS (STM) expression within the inflorescence meristem suggest a role for the BEL1-STM complex in maintaining the indeterminacy of the inflorescence meristem.

Figure 1.

Interactions between BEL1 and KNAT Proteins via the Yeast Two-Hybrid System. (A) Scheme of intact and truncated BEL1 sequences used as bait in two-hybrid screens and quantitative assays. The patterned box (HD) represents the homeodomain, whereas the other boxes represent the SKY and BELL domains. The putative amphipathic α-helix, represented by the cross-hatched box, comprises the N-terminal portion of the BELL domain. Numbers in superscript indicate the BEL1 amino acids encoded by each clone. Intact BEL1 was used for two-hybrid screen A, whereas BEL^1-396 was used for two-hybrid screen B. (B) β-Galactosidase activity resulting from the combinations of BEL1 bait constructs and pTA-KNAT prey clones. pTA-based “prey” clones are indicated at the bottom of the histogram, and pDB-based “bait” clones are indicated to the right of the histogram. Enzyme activity resulting from the combination of the yeast pDB-SFN4 and pTA-SFN1 clones is included as a positive control for interacting proteins, combinations with pDB-CRUCIFERIN are included as a negative control, and pDB-BEL1^447-611 is included as a clone that induces β-galactosidase activity autonomously. Asterisks denote tests that were not performed. For each prey category, the bait combinations are presented from left to right as follows: pDB-CRUCIFERIN, pDB-BEL1, pDB-BEL1^1-396, pDB-BEL1^447-611, and pDB-SNF4. (C) GST pulldown experiments demonstrating specific BEL1/KNAT interactions in vitro. The table indicates which combinations of translation products were used in each lane. The bottom panel shows an SDS-PAGE gel of the various translation products before or after incubation with glutathione–agarose beads as indicated (asterisks denote lanes in which translation products were not incubated with beads). In vitro translated GST-BEL1 displays a major, presumably full-length product and several minor truncated products (lane 1). STM and KNAT1 display single products (lanes 2 and 3, respectively), whereas KNAT3 displays two major products (lane 4). No STM, KNAT3, or KNAT1 products are retained by the beads alone (lanes 8 to 10, respectively). GST-BEL1, but not KNAT3, is retained by the beads when these translation products are incubated together (lane 7). STM and KNAT1, however, are retained by the beads in combination with GST-BEL1 (lanes 5 and 6, respectively), indicating specific interactions between BEL1 and STM or KNAT1. To confirm that BEL1 interacts with STM, the experiment was conducted in the reverse orientation. In vitro translated BEL1 (lane 11) and GST-STM (lane 12) display major, presumably full-length products and several minor truncated products. BEL1 was not retained by the glutathione–agarose beads alone (data not shown, but comparable to the results shown for STM, KNAT1, and KNAT3), but it was retained in combination with GST-STM (lane 13).

Figure 2.

Deletion Analysis to Identify Protein–Protein Interaction Domains of BEL1. Schemes of the BEL1 sequences expressed as C-terminal fusions to the GAL4 TA are depicted as black boxes at left. The patterned box (HD) represents the homeodomain, whereas the other boxes represent the SKY and BELL domains. Numbers in superscript indicate the BEL1 amino acids encoded by the deletion construct. β-Galactosidase activities resulting from the combinations of individual BEL1 deletion constructs with pDB-CRUCIFERIN, pDB-KNAT1, and pDB-KNAT3 are presented in the histogram at right.

Figure 3.

Sequence Analyses of KNAT cDNA Clones Isolated in Two-Hybrid Screens Using BEL1 as Bait. Prey plasmids were isolated with either pDB-BEL1 (pTA-A clones) or pDB-BEL^1-396 (pTA-B clones) as bait. Predicted polypeptides are depicted schematically with N-terminal and C-terminal amino acids indicated at left and right of the schemes, respectively. Boxes with diagonal hatches represent the MEINOX domain, gray boxes represent the ELK domain, and black boxes represent the homeodomain (HD). Only a subset of the isolated KNAT1 and KNAT3 clones are depicted because several were identified by restriction digest analysis.

Figure 4.

Activation of Transcription by BEL1–KNAT Heterodimers. β-Galactosidase activity resulting from the combinations of STM, KNAT1, KNAT3, and KNAT5 bait constructs with NLS-BEL1 are shown. Bars indicate ±sd.

Figure 5.

BEL1 Function in Situ Hybridization Analyses of the Expression Patterns of BEL1, STM, and KNAT1 in the Inflorescence Apex of Arabidopsis. (A) Wild-type Arabidopsis inflorescence apex from which lateral flowers are produced indeterminately. (B) Bel1 mutant apex that has terminated in a flower. (C) Bel1 terminal flower at a slightly later stage than in (B). (D) BEL1 transcript is detected in developing ovules in a pattern that predicts the position of integument formation. (E) No BEL1 transcript is detected in the inflorescence of a plant homozygous for the bel1-3 allele. (F) Low levels of BEL1 mRNA can be detected in the inflorescence apex and developing floral primordia. (G) STM transcript is detected across the apical meristem and in the apex of the developing floral primordium. (H) Close-up of floral primordium shown at left in (F). (I) KNAT1 is transcribed in the early floral primordium in what will become the pedicel. As the pedicel develops, this expression persists in the cortex next to the vascular tissue. ov, ovule; fp, floral primordium; im, inflorescence meristem. Bars in (A) to (C) = 1 mm; bars in (D) to (I) = 30 μm.

Figure 6.

Model Depicting the Evolution of Interactions between TALE Homeodomain Proteins (after Bürglin, 1998). It has been hypothesized that an ancestral MEINOX–TALE homeodomain (HD) protein was capable of forming homodimers and interacting with another, typical homeodomain protein. Before the emergence of plants and animals, this ancestral MEINOX gene duplicated, and the diverging proteins retained the ability to interact with each other. In animals, MEINOX proteins (represented by the MEIS class) retained the ability to interact with TALE homeodomain proteins that evolved into the PBC family and typical homeodomain proteins that evolved into the HOX family. In plants, the MEINOX proteins have been conserved as KNOX proteins that have retained the ability to interact with the BEL1 homeodomain protein. The existence of BEL1–KNAT interactions with typical homeodomain proteins is hypothetical.