Structural insights into proteolysis-dependent and -independent suppression of the master regulator DELLA by the gibberellin receptor

Abstract

The perception of the phytohormone gibberellin (GA) by its nuclear receptor GIBBERELLIN INSENSITIVE DWARF1 (GID1) triggers polyubiquitination and proteasomal degradation of master growth regulators-DELLA proteins-mediated by the SCF^SLY1/GID2 E3 ubiquitin ligase complex. DELLA-encoding genes are known as 'Green Revolution' genes, as their dominant mutations lead to semidwarf cereal varieties with significantly higher yields due to reduced GA response. DELLAs function as central signaling hubs, coordinating diverse physiological responses by interacting with key transcription factors across multiple cellular pathways. While the DELLA domain mediates GA-GID1 binding, the mechanism of SCF^SLY1/GID2 recruitment remained unknown. Additionally, GA-GID1 binding can inhibit DELLA protein activity independently of its proteolysis, although the underlying mechanism was unclear. Here, we present the cryo-EM structures of GA₃-GID1A complexed with a full-length DELLA protein in Arabidopsis, RGA (REPRESSOR OF ga1-3), and the GA₃-GID1A-RGA-SLY1-ASK1 complex. We show that the DELLA domain of RGA functions as a molecular bridge to enhance its GRAS domain binding to GID1A through direct interactions with both the GRAS domain and GID1A. Disrupting either intramolecular (DELLA-GRAS) or intermolecular (GRAS-GID1A) interactions weakens RGA-GID1 binding. Contrary to prior models, SLY1 binds the GRAS domain's concave surface without inducing conformational changes. Combining AlphaFold modeling and yeast three-hybrid assays, we demonstrate that GID1 binding to the RGA GRAS domain blocks its interactions with INDETERMINATE DOMAIN (IDD) transcription factors, explaining how GA-GID1 relieves growth suppression independently of DELLA degradation.

Keywords: DELLA; gibberellin receptor; gibberellin signaling.

Figure 1.. Cryo-EM structure of the GA₃-GID1A-RGA complex.

(A) Cryo-EM density map and (B) cartoon representation of the GA₃-GID1A-RGA complex. GID1A is colored in orange, and the RGA DELLA and GRAS domains are colored in purple blue and deep purple respectively. GA₃ is shown in the space-filling model. Architecture organizations of GID1A and RGA are labeled. (C) Interface residues between DELLA-GRAS. (D-E) Interface residues between GID1A-GRAS. Bb, backbone; sc, side chain. (F) Dependence of the RGA DELLA-GRAS interaction on the presence of GA₃-GID1A in Y3H assays using split RGA. RGA-CT1 containing the GRAS domain was fused to the Gal4 BD domain as the bait, and RGA-NT1 containing the DELLA domain was fused to the Gal4 AD domain as the prey. GID1A was expressed using the pH3 vector. 100 μM GA₃ was included in the +GA media. Immunoblot analyses showed that similar levels of WT and mutant proteins (for GRAS and for DELLA) were present across different yeast cells (Fig. S3). Interaction of BD and AD fusion proteins in the PJ69–4A yeast cells was scored by the relative growth in –His media containing varying concentrations of 3-AT as labeled. “−” indicates empty vector, or no growth at 0 mM 3-AT. Max, detectable cell growth at the maximum 3-AT concentration. Two biological repeats showed similar results.

Figure 2.. Cryo-EM structure of the GA₃-GID1A-RGA-SLY1-ASK1 complex.

(A) Cryo-EM density map and (B) cartoon representation of the GA₃-GID1A-RGA-SLY1-ASK1 complex. GID1A, SLY1, and ASK1 are colored in orange, green, and brown, respectively and the RGA DELLA and GRAS domains are colored in purple blue and deep purple, respectively. GA₃ is shown in the space-filling model. (C) A zoomed-in view of the SLY1-ASK1 secondary structure interactions. (D) Interface of SLY1 and the RGA GRAS domain. (E) Y3H assays show that mutations in sly1-d disrupting its interaction with the RGA GRAS domain abolished RGA binding. sly1-d was fused to the Gal4 BD domain as the bait, and full-length RGA was fused to the Gal4 AD domain as the prey. GID1A was expressed using the pH3 vector. 100 μM GA₃ was included in the +GA media. Immunoblot analyses showed that similar levels of RGA and sly1 mutant proteins were present across different yeast cells (Fig. S7). Interaction of BD and AD fusion proteins in the PJ69–4A yeast cells was scored by the relative growth in –His media containing varying concentrations of 3-AT as labeled. “−” indicates empty vector, or no growth at 0 mM 3-AT. Max, detectable cell growth at the maximum 3-AT concentration. Two biological repeats showed similar results.

Figure 3.. Overexpression of DELLA domain partially rescued the dwarf phenotype and GA responsiveness of rga-CT1.

(A and B) Phenotypes of WT, single transgenic rga-CT1, and double transgenic rga-CT1 rga-NT1 Arabidopsis lines. In (A), photo of 25d-old representative plants. In (B), final plant heights. n=8–11. In the boxplot, center lines and box edges are medians and the lower/upper quartiles, respectively. Whiskers extend to the lowest and highest data points within 1.5x interquartile range (IQR) below and above the lower and upper quartiles, respectively. Different letters above the boxes represent significant differences (p < 0.01) as determined by two-tailed Tukey’s HSD mean separation test. (C) Co-expression of rga-NT1 resulted in a reduction in rga-CT1-Myc protein accumulation. Immunoblots containing total protein extracted from Arabidopsis seedlings were probed with anti-RGA polyclonal antibodies, which detected both endogenous RGA and rga-CT1-Myc. *, non-specific background band. Ponceau S (PS)-stained blot showing equal loading. (D) rga-NT1 led to reduced SCL3 and GA3ox1 mRNA levels without downregulating rga-CT1-Myc expression. (E)-(F) rga-NT1 caused an increased GA response in the double transgenic rga-CT1-Myc rga-NT1 line. Hypocotyl elongation assay was performed using seedlings grown on MS media containing 1 μM paclobutrazol (PAC, a GA biosynthesis inhibitor) and varying concentrations of GA₃ for 9d. In (E), photo showing representative seedlings. Bar = 5 mm. In (F), Average hypocotyl lengths. n = 11–12. Means ± SE. (G) rga-CT1-Myc protein showed GA-dependent degradation only in the presence of rga-NT1. Seedlings grown on MS media containing 1 μM PAC were treated with 10 μM GA₃ for 0h-24h as labeled. Immunoblots containing total protein extracted from Arabidopsis seedlings were probed with anti-RGA or anti-Myc antibodies. *, non-specific background band. (H) RT-qPCR showing mRNA levels of SCL3 and GA3ox1 were downregulated by GA more dramatically in the double transgenic rga-CT1-Myc rga-NT1 line than in rga-CT1-Myc. RNA was extracted from seedlings after mock or GA treatment as in (G). In (D) and (H), PP2A was used to normalize different samples. Means ± SE of three biological repeats are shown. The p values were calculated by a two-tailed Tukey’s HSD mean separation test. In (D), different letters above the bars represent significant differences (p < 0.01). In (A)-(C) and (E)-(F), two biological repeats showed similar results.

Figure 4.. GA₃-GID1A directly suppresses the RGA-transcription factor interaction without RGA degradation.

(A) Y3H assays showing IDD1-RGA interaction was attenuated by GID1 in a GA-dependent manner. A C-terminal IDD1 fragment (246–403) and a IDD1 peptide (357–395) spanning the motif A were fused to the Gal4 BD domain as the baits, and the full-length RGA was fused to the Gal4 AD domain as the prey. GID1A was expressed using the pH3 vector. 100 μM GA₃ was included in the +GA media. Immunoblot analyses showed that similar levels of IDD1 and RGA were present in the presence or absence of GID1A (Fig. S11). Interaction of BD and AD fusion proteins in the PJ69–4A yeast cells was scored by the relative growth in –His media containing varying concentrations of 3-AT as labeled. “−” indicates empty vector, or no growth at 0 mM 3-AT. Max, detectable cell growth at maximum 3-AT concentration. Two biological repeats showed similar results. (B) The consensus sequence of motif A in IDDs. (C) Cartoon representation of the AlphaFold3-predicted models of the IDD consensus peptide (cyan) bound to the RGA GRAS domain (green). The AlphaFold3 models are superimposed with the cryo-EM structure of the GA₃-GID1A-RGA complex, with GID1A colored in orange and the RGA DELLA and GRAS domains colored in purple blue and deep purple, respectively. LHR1, Leucine Heptad Repeat region (V218-P279). The predicted template modeling (pTM) and interface predicted template modeling (ipTM) scores are labeled. (D) A schematic model of two mechanisms for GA-GID1-mediated inhibition of DELLA protein. Proteolysis-dependent mechanism: Without GA-GID1 binding, the disordered DELLA domain may weaken GRAS-SLY1 interaction by docking transiently on the SLY1 binding interface of the GRAS domain. GA-GID1 promote DELLA protein degradation by binding to the DELLA domain, which converts to a stable structure and acts as a molecular bridge to enhance GID1-GRAS binding to recruit the SCF^SLY1 complex for polyubiquitination and degradation by the 26S proteasome. Proteolysis-independent mechanism: GA-GID1 binding to DELLA and GRAS domains directly competes with interaction of transcription factors, such as IDDs, with the DELLA protein. The solid line between GID1A and GRAS reflecting the interface loop preceding the GRAS domain. Panel (D) was partially created with BioRender.com.