Membrane-associated DELLA degradation modulates growth under carbon/nitrogen imbalance

TL;DR

Plants adjust growth under carbon/nitrogen imbalance by degrading DELLA proteins at the membrane, avoiding harmful effects on development.

Contribution

A novel membrane-based mechanism for DELLA degradation under C/N stress, independent of the canonical GA-GID1 pathway.

Findings

RGA and GAI DELLA proteins are key to the C/N stress response through a membrane-associated mechanism.

ATL31 E3-ligase promotes DELLA ubiquitination and degradation at the membrane during C/N stress.

Enhanced ATL31 expression alone does not alter DELLA-related traits without stress.

Abstract

Crop yield and sustainability rely on the ability of plants to perceive and efficiently use nutrients. When high carbon (C) to nitrogen (N) ratios are perceived, plants trigger a specific response leading to reduced growth and enhanced anthocyanin accumulation. Here, using (Arabidopsis thaliana), we provide genetic, molecular and physiological evidence supporting a role for DELLA proteins to control growth under C/N stress through a non-nuclear mechanism that regulates their stability. C/N stress response specifically requires the RGA (REPRESSOR OF ga1-3 1) and GAI (GIBBERELLIC ACID INSENSITIVE) DELLA proteins, whose stability is reduced by a membrane-associated mechanism independent of the canonical gibberellic acid (GA)-GID1 (GIBBERELLIN INSENSITIVE DWARF1) pathway. Although C/N stress enhances DELLA accumulation by reducing GA levels, it also promotes their ubiquitination and degradation via interaction with the ATL31 E3-ligase at the membrane, even in the absence of GAs or when GA-resistant alleles are used. Moreover, phenotypic traits known to be altered by DELLA levels are not affected by enhanced ATL31 expression in the absence of stress. We propose that this mechanism fine-tunes DELLA-mediated C/N stress responses without adverse effects on plant development.

DELLA degradation at the membrane affects growth under nutrient stress.

Figures

GAI and RGA DELLA proteins mediate C/N stress responses. a) The indicated genotypes were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N), 200 mM glucose (200C), or/and 1 µM GA as indicated. b) Cotyledon color intensity scale (hue values) is based on the angle originating from the a* (green-red) and b* (blue-yellow) axes in polar coordinates and is used in c), e, and other figures. c) Hue values recorded for seedlings from the plates shown in a) were individually analyzed. Image J was used for image scoring (see Methods and Figure S2). For statistical analyses, high C/N ratio measurements were normalized to 0.3N. Different letters denote statistically significant differences (P < 0.01) using Welch's 1-way ANOVA with the Games-Howell test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated. d) The indicated genotypes were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N), 100 (100C), or 200 (200C) mM glucose or/and 0.05 or 1 µM GA. Mutant alleles used in all panels were ga1-3, rgl2-1, rga-t2, gai-t6, and rgl1-1. e) Hue values recorded for seedlings. For statistical analyses, high C/N measurements were normalized to the 0.3N plate. Different letters denote statistically significant differences (P < 0.01) using Welch's 1-way ANOVA with the Games-Howell test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated.

C/N stress promotes DELLA accumulation. a) pGAI:GAI-GFP and pRGA:GFP-RGA seedlings were grown for 5 d under nitrogen deficiency (0.3N), C/N stress conditions (100C:0.3N or 200C:0.3N) or nutritionally balanced conditions (200C:3N or 200C:30N). Proteins were then extracted and subjected to immunoblotting with an anti-GFP antibody. b) RT-qPCR analysis using RNA from Arabidopsis WT seedlings (Ler and Col-0) grown for 2 or 3 d under the indicated conditions. Values were normalized to UBC21 expression. GA3ox1 and UBC21 specific primers are listed in Table S1. Averages and SE of 2 replicates are shown. c) HACR-GAI seeds germinated for 3 d in control plates (0.3N) were transferred to control or C/N stress conditions (100C:0.3N or 200C:0.3N). After 24 and 48 h, seedlings were sprayed with 200 mM luciferin and luciferase activity was quantified using a NightOWL imaging system. Different letters denote statistically significant differences (P < 0.05) using 1-way ANOVA with the Tukey HSD test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated in Supplementary Data Set S1. d) Confocal microscopy images of Venus fluorescence alone and merged with bright field as observed in seedlings from b) after 48 h. Bars = 30 μm. e) Protein extracts from pGAI:GAI-3xYPET and pRGA:GFP-RGA seedlings grown in MS were incubated for the indicated times with protein extracts from WT (Ler) plants grown under control (0.3N) or C/N stress (200C:0.3N) conditions. Extracts from 5-d-old seedlings were used in all cases.

ATL31 interacts with DELLAs at the plasma membrane. a) Yeast strains containing the GAI or RGA CDS fused to the GAL4-BD (BD-; bait) were mated to strains containing the AD-Ø, AD-GFP or AD-ATL31C143SΔTM constructs. Diploid cells were grown on diploid (-L-W) and screening (-L-W-H) plates with or without 3-AT (mM). b) Co-IP assays with GFP-ATL31 and HA-DELLAs co-expressed in Nicotiana leaves. Soluble proteins before (input) and after (IP: α-GFP) immunoprecipitation with an anti-GFP antibody were immunoblotted with anti-GFPHRP (α-GFP) or an anti-HAHRP (α-HA) antibodies. The HA-GAI and RGA-HA proteins were detected by the anti-HAHRP antibody in protein extracts from leaves agroinfiltrated with HA-GAI/RGA-HA constructs (α-HA input panels, lanes 1 to 4). Likewise, the anti-GFPHRP antibody successfully detected the GFP-ATL31 fusion protein (α-GFP input panel, lanes 2 and 4). The GFP epitope fused to ATL31 was efficiently immunoprecipitated by an anti-GFP antibody (α-GFP IP panels, lanes 2 and 4). c) cCFP translational fusions to GRF6 (positive control), GAI and RGA were co-infiltrated with an ATL31C143S-nYFP construct in Nicotiana cells as indicated. Images were acquired using a confocal laser microscope. Bars = 100 μm. d) A GAI translational fusion with cCFP was co-infiltrated with a ATL31ΔTM-nYFP construct in Nicotiana cells as indicated. Images were acquired using a confocal laser microscope. Bars = 100 μm. e) FLIM analysis of GAI-GFP, ATL31-mCherry and ATL31C143S-mCherry in Nicotiana leaf cells. Fusion proteins were expressed from an estradiol-inducible promoter and imaged 3 to 4 d after infiltration. Micrographs display pixel-wise fitted average fluorescence lifetime values of GFP in representative individual cells expressing GAI-GFP with or without free mCherry, ATL31-mCherry and ATL31C143S-mCherry. Twenty-six to 35 different cells were analyzed per genotype. Different letters denote statistically significant differences (P < 0.05) using 1-way ANOVA with the Tukey HSD test for multiple comparisons.

ATL31 promotes cytosolic GAI and RGA degradation by the proteasome. a) Plasmolysis experiment for GAI-GFP. The indicated genotypes were grown for 4 d in MS/2 and then were treated with 5 µM PAC. After 48 h, the seedlings were immersed in a 1 M mannitol solution for 80 min. Images were acquired using a confocal laser microscope. Bars = 10 μm. b) Western blots of subcellular fractions isolated from pGAI:GAI-GFP and pRGA:GFP-RGA seedlings grown for 5 d in the indicated media. The GAI-GFP and GFP-RGA proteins were detected by anti-GFP antibodies. Histone H3 (anti-H3) was used as marker for the nuclear fraction, and UGPase (anti-UGPase) was used as marker for the cytosolic fraction. c) Ler, ATL31-GFP and ATL31C143S-GFP overexpressing seedlings were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen and 200 mM glucose. Proteins were then extracted and subjected to immunoblotting with an anti-RGA antibody. In parallel, Co-IP assay with anti-GFP was performed. Immunoprecipitation with an anti-GFP antibody was immunoblotted with anti-GFPHRP (α-GFP) or an anti-RGA (α-RGA) antibodies. d) Ubiquitination experiment for GAI-GFP. GAI-GFP, ATL31-mCherry and ATL31C143S-mCherry were expressed from an estradiol-inducible promoter in Nicotiana leaves and collected 3 d after infiltration. 50 µM MG132 was infiltrated 24 h before collection to prevent degradation. A Co-IP assay with anti-GFP was performed and immunoblotted with anti-GFPHRP (α-GFP) or an anti-UBQ (α-UBQ). e) Degradation assay for ATL31 target proteins. GFP-tagged GAI and RGA proteins were expressed alone or in combination with a 35S:ATL31 construct (ATL31ox) in Nicotiana leaves. Protein extracts were incubated in the presence or absence of 50 µM MG132 for 2 h and immunoblotted with an anti-GFPHRP antibody.

GA-resistant DELLAs can be degraded by ATL31 and do not enhance sensitivity to C/N stress. a) pGAI:GAI-3xYPET, pRGA:GFP-RGA and their corresponding Δ17 version seedlings were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N) in the absence or presence of 100 mM glucose (100C). b) Hue values recorded for seedlings photographed in a) were individually analyzed with Image J. For statistical analyses C/N stress measurements were normalized to the 0.3N values. No statistically significant differences found (P < 0.01) using Welch's test. Welch's t-statistic and effect size measurement eta-squared (η2) are indicated. (c) Protein extracts from pGAI:gaiΔ17-3xYPET seedlings were incubated for the indicated times with extracts from Ler seedlings harboring or not a 35S:ATL31 construct and grown under C/N stress (200C:0.3N). An anti-GFPHRP antibody was used for immunoblotting. Analog results were obtained 3 times. Lower band coincides with gaiΔ17-1xYPET size. d) Relative intensity of gaiΔ17 in 3 independent experiments was quantified by Image J. The initial protein levels were defined as 1. Coomassie blue staining served as loading control. Asterisks indicate statistically significant differences (*P < 0.05; P < 0.01) for Ler versus ATL31ox using multiple Student’s t-test.

ATL31 reduces sensitivity to high C/N even in the absence of GA biosynthesis. a) Growth of Col-0 and Ler seedlings with and without a 35S:ATL31 construct (ATL31ox) after 5 d. MS-N plates were supplemented with 0.3 mM nitrogen (0.3N) in the absence or presence of 100 mM glucose (100C). b) The same analyses of panel a) were conducted with the ga1-3 mutant background and the indicated combinations of della loss of function alleles (rgl2-1, rga-t2, gai-t6).  c) Hue values for seedlings photographed in a) and b) were individually analyzed with Image J. Image J was used for image scoring. For statistical analyses C/N stress measurements were normalized to the 0.3N values. Asterisks indicate statistically significant differences (P < 0.001) using Welch's test. d)  ga1-3 rgl2-1 gai-t6 seedlings with and without a 35S:ATL31 construct (ATL31ox) were grown for 5 d under C/N stress conditions (100C:0.3N). Proteins were then extracted and subjected to immunoblotting with an anti-RGA antibody. (e)  GA3ox1 mRNA levels of Col-0, atl31 and ATL31ox under C/N stress. RT-qPCR analysis of RNA from Col-0 (C), atl31 (K) and ATL31ox (O) grown for 2 or 3 d under the indicated conditions. The values were normalized to UBC21 expression. GA3ox1 and UBC21 specific primers are listed in Table S1. Averages and SE of 3 replicates are shown.

Model for the role of the ATL31-mediated DELLA degradation mechanism on plant growth and C/N stress responses. a) Germination percentage (%) of FH Col-0 versus 35S:ATL31 (ATL31ox) FH seeds at 72 h after imbibition (hai) and germination kinetics of seeds AR seeds for 4 wk. b) Hypocotyl lengths of seedlings grown for 7 d with or without lighting in the absence or presence of PAC. c) Proposed model. Under balanced C/N conditions, rising in N levels promotes DELLA degradation by the GA pathway (Camut et al. 2021) and ATL31 expression is reduced (Aoyama et al. 2014). C/N stress represses the GA pathway (Fig. 2) and enhances ATL31 mRNA and protein levels (Aoyama et al. 2014; Yasuda et al. 2017). ATL31 interacts in turn with specific DELLAs in the plasma membrane, targeting them for degradation.

Full text

Introduction

Plant growth and development must be coordinated with nutrient availability. Carbon (C) and nitrogen (N) are the 2 most abundant elements in plant cells. N is the main limiting element in most agricultural soils and sugar metabolism is significantly affected by N levels. Complex interactions between C and N signaling have been observed (Coruzzi and Bush 2001; Coruzzi and Zhou 2001; Palenchar et al. 2004; Zheng 2009; Osuna et al. 2015). Photosynthetic aerial tissues provide reduced C, and photosynthetic rates are subjected to daily and seasonal fluctuations (Bauerle et al. 2012; Koyama and Takemoto 2014). N is taken up by roots from 2 major sources, nitrate (NO_3_^−^) and ammonium (NH_4_^+^). N uptake is conditioned by environmental cues acting on root architecture and physiology as well as by its availability in the soil, which in turn, will affect global gene expression and modulate root and shoot development (Medici and Krouk 2014; Wang et al. 2018; Fredes et al. 2019; Jia and von Wirén 2020). C and N also affect germinating embryos undergoing the phase transition from heterotrophic to autotrophic growth, crucial for seedling establishment and plant survival. Such transition is repressed by high sugar levels but can be counteracted by increasing N availability (Martin et al. 2002; Cho et al. 2010). Thus, elevated C/N ratios are unfavorable for photosynthesis, storage lipid breakdown, cotyledon and seedling growth and promote anthocyanin accumulation (Martin et al. 2002). Only a few genes involved in C/N stress responses have been described via genetic analyses (Osuna et al. 2015). One of them, the ATL31 RING-H2-type ubiquitin ligase, plays a key role in promoting growth and counteracting the response to unbalanced C/N ratios during seedling establishment (Sato et al. 2009, 2011).

The central role of gibberellins (GA) and DELLA proteins in the control of growth and stress responses has been widely documented (Claeys et al. 2014; Davière and Achard 2016). The genome of the model plant Arabidopsis thaliana harbors 5 DELLA genes (RGA, GAI, RGL1-3) with both specific and overlapping functions (Claeys et al. 2014; Hedden and Sponsel 2015; Davière and Achard 2016; Nelson and Steber 2016; Vera-Sirera et al. 2016). These proteins are transcriptional regulators that act as signaling hubs connecting environmental cues with endogenous pathways through interactions with many transcription factors (Minguet et al. 2014; Vera-Sirera et al. 2016; Blanco-Touriñán et al. 2020a). DELLAs were identified as master negative regulators of GA signaling that, among other processes, arrest growth to promote survival under adverse conditions (Achard et al. 2006, 2008a, 2008b). They have also been involved in the regulation of responses to nutrients, including phosphate (Jiang et al. 2007), sucrose (Li et al. 2014), iron (Wild et al. 2016), and nitrate (Camut et al. 2021). When nutrient environment is favorable for growth, DELLAs are targeted for GA-dependent degradation, which involves the GA receptor GID1 (Ueguchi-Tanaka et al. 2005) and a SCF-type E3 ubiquitin ligase complex where the F-box protein GID2/SLY1 provides the substrate specificity (McGinnis et al. 2003; Sasaki et al. 2003; Lechner et al. 2006). GA perception leads to DELLA ubiquitination and degradation by the 26S proteasome in the nucleus (Dill et al. 2004; Vera-Sirera et al. 2016; Blanco-Touriñán et al. 2020a). Two GA-independent DELLA degradation pathways, mediated by COP1 (CONSTITUTIVELY PHOTMORPHOGENIC 1) and FKF1 (FLAVIN-BINDING KELCH REPEAT F-BOX 1), have been unveiled (Yan et al. 2020; Blanco-Touriñán et al. 2020b). COP1 is a RING-type E3 ubiquitin ligase that interacts with RGA and GAI in nuclear bodies when seedlings are placed in shaded or warm environments, thus leading to their polyubiquitination and degradation by the 26S proteasome to promote hypocotyl growth (Blanco-Touriñán et al. 2020b). FKF1 is the F-box protein of a SCF-type E3 ubiquitin ligase whose interaction with GAI and RGA in the nucleus triggers their degradation to promote flowering under long photoperiods (Yan et al. 2020).

Here we show that the stress phenotypes produced during seedling establishment under high C/N ratios can be alleviated by exogenous GAs, even in the absence of the membrane-associated ATL E3-ligases known to counteract stress. By using combinations of della loss of function mutants and a GA biosynthetic mutant we identified that RGA and GAI are the Arabidopsis DELLAs involved in this response. Whereas C/N stress promotes DELLA stability and reduces GA levels, it also triggers a GA-independent pathway for DELLA degradation relying on their physical interaction with the ATL31 protein at the plasma membrane. This pathway has a substantial impact on plant growth adaptation to nutrient stress.

Results

Exogenous gibberellins alleviate C/N stress in wild-type and hypersensitive mutant plants

Arabidopsis seedlings show growth reduction and anthocyanin accumulation when exposed to high C/N ratios. It is known that ATL31 and ATL6 E3-ligases loss of function mutants are hypersensitive to such stress compared with wild-type (WT) plants (Sato et al. 2009; Fig. 1a). Given that DELLAs have been associated with growth arrest (Achard et al. 2006) and anthocyanin biosynthesis (Li et al. 2014), we tested whether promoting DELLA degradation by adding exogenous GA had an impact on the response to unbalanced C/N. We used 1 μM GA since it was the lowest concentration required for the GA-deficient mutant ga1-3 (Koornneef and Van Der Veen 1980) to reach 100% germination (Figure S1a). GA addition significantly alleviated the C/N stress phenotypes in WT and mutant plants, which reverted to almost those observed under control growth conditions (0.3N; Fig. 1a). Although these results were quantitatively confirmed by visual scoring (Figure S1b) as well as by measuring anthocyanin content (Figure S1c), both methods previously used to grade C/N stress (Martin et al. 2002; Sato et al. 2009), we implemented an additional scoring method that also captures variability in the severity of the phenotype. Briefly, all photographed seedlings were individually analyzed with Image J, so that channels a* and b* of the CIELAB color space (Kendal et al. 2013) were combined to estimate the mean hue value of each cotyledon pair (Fig. 1b). Validation of this approach is shown in Figure S2 and fully explained in the Methods section. Under control conditions (0.3N), the atl mutants showed a general trend toward higher (greener) hue values than WT plants (Fig. 1c). To eliminate the effect of basal differences between genotypes, the hue values obtained under C/N stress (200C:0.3N) were normalized to their averaged hue values under 0.3N. Cotyledons scored lower hue values when seedlings were grown under C/N stress and those for atl seedlings were clearly lower (redder) than in the WT (Fig. 1c). GA addition counteracted C/N stress, producing values comparable to the control condition and eliminating the variance attributable to atl mutants (Fig. 1c). Whereas GA and abscisic acid (ABA) have antagonistic effects on growth and a reduction in GA/ABA ratios enhance DELLA stability (Piskurewicz et al. 2008; Lorrai et al. 2018), growth of WT and atl mutants in the presence of paclobutrazol (PAC; inhibitor of GA biosynthesis) or ABA did not phenocopy the C/N response (Figure S3). Therefore, our results indicate that exogenous GAs alleviate C/N stress, probably by reducing DELLA levels, and suggest that the C/N response requires additional signaling events besides DELLA stabilization.

GAI and RGA DELLA proteins mediate C/N stress responses. a) The indicated genotypes were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N), 200 mM glucose (200C), or/and 1 µM GA as indicated. b) Cotyledon color intensity scale (hue values) is based on the angle originating from the a (green-red) and b* (blue-yellow) axes in polar coordinates and is used in c), e, and other figures. c) Hue values recorded for seedlings from the plates shown in a) were individually analyzed. Image J was used for image scoring (see Methods and Figure S2). For statistical analyses, high C/N ratio measurements were normalized to 0.3N. Different letters denote statistically significant differences (P < 0.01) using Welch's 1-way ANOVA with the Games-Howell test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated. d) The indicated genotypes were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N), 100 (100C), or 200 (200C) mM glucose or/and 0.05 or 1 µM GA. Mutant alleles used in all panels were ga1-3, rgl2-1, rga-t2, gai-t6, and rgl1-1. e) Hue values recorded for seedlings. For statistical analyses, high C/N measurements were normalized to the 0.3N plate. Different letters denote statistically significant differences (P < 0.01) using Welch's 1-way ANOVA with the Games-Howell test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated.*

The GAI and RGA DELLA proteins mediate the C/N response

To determine whether the C/N stress response is mediated by specific DELLAs, we tested different combinations of Arabidopsis della mutants in the ga1-3 background, in which DELLA proteins over accumulate. As shown in Fig. 1d, the WT (Ler) seedlings are less sensitive to the C/N stress than the ga1-3 mutants containing single (rgl2-1) or double (rgl2-1 rga-t2 or rgl2-1 gai-t6) della loss of function alleles. This suggests that the hypersensitivity is caused by enhanced levels of the remaining 4 or 3 DELLAs, respectively. Indeed, the hypersensitivity is lost in the higher order mutants ga1-3/rgl2-1/rga-t2/gai-t6 and ga1-3 rgl2-1 rga-t2 gai-t6 rgl1-1 (Fig. 1d). Although full germination of a ga1-3 single mutant required 1 μM GAs (Figure S1a), we found that C/N stress alleviation is highly sensitive to GAs, as the response was saturated at 0.005 μM GA (Figure S4, a and b). By measuring cotyledon color (Fig. 1e) and anthocyanin content (Figure S4c) we confirmed that the simultaneous loss of RGL2, RGA and GAI function is sufficient to reduce sensitivity to levels obtained with GA treatments. The similar sensitivity shown by ga1-3 rgl2-1 rga-t2 gai-t6 and ga1-3 rgl2-1 rga-t2 gai-t6 rgl1-1 (Fig. 1e), and the higher sensitivity of ga1-3 rgl2-13 rga-28 rgl1-2 relative to WT (Col-0; Figure S5), points to GAI, RGA and RGL2 as the relevant DELLAs for the C/N stress response. However, RGL2 might not have a direct role given that its loss of function is a requisite for ga1-3 seeds to germinate without GA addition. We ruled out the involvement of RGL2 given that a gai-t6 rga-24 double mutant, unlike its WT (Ler), did not increase its sensitivity to C/N stress in the presence of PAC (Figure S6a). In addition, the gai-td1 and rga-28 single mutants showed WT sensitivity to C/N stress in the presence of PAC, unlike the gai-td1 rga-29 double mutant (Figure S6b). These results confirm that the relevant role in C/N stress is exerted by GAI and RGA but not by the RGL DELLAs and, that either GAI or RGA function is sufficient for the response.

C/N stress promote DELLA accumulation

Our results with the della mutants and modulation of GA levels suggest that RGA and GAI stability conditions the C/N stress response. To test this hypothesis we used protein extracts from Arabidopsis pGAI:GAI-GFP (Fleck and Harberd 2002) and pRGA:GFP-RGA (Silverstone et al. 2001) seedlings grown under different C/N ratios to quantify GAI-GFP and GFP-RGA protein levels. As shown in Fig. 2a, GAI and RGA levels consistently increased under high C/N ratios compared with the 0C:0.3N condition; on the other hand, their levels decreased under reduced C/N ratios. Moreover, DELLAs could not be detected after addition of exogenous GAs even under high C/N ratios (Figure S7a). The kinetics of their response to high C/N ratios indicated that GAI and RGA accumulated in a gradual and sustained manner for at least 48 h ensuing stress (Figure S7b). RT-qPCR experiments did not support that protein upregulation produced by C/N stress was due to increased mRNA levels (Figure S7c). Instead, quantitation of GA3ox1 mRNA levels (GA biosynthesis gene) and analyses of a GA biosensor line (HACR-GAI; Khakhar et al. 2018) suggested that the observed increase in GAI-GFP and GFP-RGA stability is likely due to a reduction in GA levels promoted by C/N stress (Fig. 2, b to d). To further investigate this effect, we obtained protein extracts from seedlings carrying GFP-tagged DELLAs where GAI and RGA were readily detected (labeled as 0 in Fig. 2e). After incubation for 5 and 30 min in the presence of protein extracts from WT (Ler) plants grown under control conditions, GAI and RGA abundance decreased rapidly to reach minimal and nearly undetectable levels, respectively. Such reduction in protein abundance was less pronounced when protein extracts from C/N stressed WT plants were used (Fig. 2e). Taken together, our results support that the increase in GAI and RGA stability under C/N stress is due to reduced GA levels and GA-mediated protein degradation, not to an increase in transcription rates.

C/N stress promotes DELLA accumulation. a) pGAI:GAI-GFP and pRGA:GFP-RGA seedlings were grown for 5 d under nitrogen deficiency (0.3N), C/N stress conditions (100C:0.3N or 200C:0.3N) or nutritionally balanced conditions (200C:3N or 200C:30N). Proteins were then extracted and subjected to immunoblotting with an anti-GFP antibody. b) RT-qPCR analysis using RNA from Arabidopsis WT seedlings (Ler and Col-0) grown for 2 or 3 d under the indicated conditions. Values were normalized to UBC21 expression. GA3ox1 and UBC21 specific primers are listed in Table S1. Averages and SE of 2 replicates are shown. c) HACR-GAI seeds germinated for 3 d in control plates (0.3N) were transferred to control or C/N stress conditions (100C:0.3N or 200C:0.3N). After 24 and 48 h, seedlings were sprayed with 200 mM luciferin and luciferase activity was quantified using a NightOWL imaging system. Different letters denote statistically significant differences (P < 0.05) using 1-way ANOVA with the Tukey HSD test for multiple comparisons. F-statistic and effect size measurement eta-squared (η2) are indicated in Supplementary Data Set S1. d) Confocal microscopy images of Venus fluorescence alone and merged with bright field as observed in seedlings from b) after 48 h. Bars = 30 μm. e) Protein extracts from pGAI:GAI-3xYPET and pRGA:GFP-RGA seedlings grown in MS were incubated for the indicated times with protein extracts from WT (Ler) plants grown under control (0.3N) or C/N stress (200C:0.3N) conditions. Extracts from 5-d-old seedlings were used in all cases.

ATL31 interacts with DELLA proteins at the plasma membrane and promote their degradation by the proteasome

ATL31 is a well characterized E3-ligase, with roles in C/N stress response (Sato et al. 2009, 2011). C/N stress increases ATL31 mRNA and protein levels in adult plants (Aoyama et al. 2014; Yasuda et al. 2017). To assess the possibility that ATL31 might be targeting GAI and RGA under C/N stress during seedling establishment, we first generated transcriptional and translational fusions between ATL31 and the luciferase reporter gene to analyze its expression kinetics. We observed that ATL31 transcription is enhanced between 54 and 72 h upon seed imbibition. However, ATL31 protein levels increase earlier (between 24 and 48 hai) and show enhanced accumulation at later time points in the presence of high C/N ratios (Figure S8). We then tested protein interactions in yeast 2-hybrid experiments (Y2H) using BD-GAI and BD-RGA proteins as baits and, as prey, an ATL31 version compatible with the Y2H system (AD-ATL31C143SΔTM; Sato et al. 2011; see methods). Unlike yeast cells harboring control plasmids (AD-0 and AD-GFP), those cells carrying BD-GAI or BD-RGA and the AD-ATL31C143SΔTM constructs were able to grow on selective medium lacking histidine, even in the presence of 1 mM 3-AT (Fig. 3a). To validate this protein interaction in planta, we carried out co-immunoprecipitation (Co-IP) experiments in Nicotiana benthamiana (hereafter Nicotiana) using hemagglutinin (HA) and GFP translational fusions for GAI/RGA and ATL31. As shown in Fig. 3b, HA-GAI and RGA-HA proteins co-immunoprecipitated in extracts of leaves co-expressing GFP-ATL31 (α-HA IP panels, lanes 2 and 4). These findings support that ATL31 can interact with GAI and RGA in both a Y2H system and in planta.

ATL31 interacts with DELLAs at the plasma membrane. a) Yeast strains containing the GAI or RGA CDS fused to the GAL4-BD (BD-; bait) were mated to strains containing the AD-Ø, AD-GFP or AD-ATL31C143SΔTM constructs. Diploid cells were grown on diploid (-L-W) and screening (-L-W-H) plates with or without 3-AT (mM). b) Co-IP assays with GFP-ATL31 and HA-DELLAs co-expressed in Nicotiana leaves. Soluble proteins before (input) and after (IP: α-GFP) immunoprecipitation with an anti-GFP antibody were immunoblotted with anti-GFPHRP (α-GFP) or an anti-HAHRP (α-HA) antibodies. The HA-GAI and RGA-HA proteins were detected by the anti-HAHRP antibody in protein extracts from leaves agroinfiltrated with HA-GAI/RGA-HA constructs (α-HA input panels, lanes 1 to 4). Likewise, the anti-GFPHRP antibody successfully detected the GFP-ATL31 fusion protein (α-GFP input panel, lanes 2 and 4). The GFP epitope fused to ATL31 was efficiently immunoprecipitated by an anti-GFP antibody (α-GFP IP panels, lanes 2 and 4). c) cCFP translational fusions to GRF6 (positive control), GAI and RGA were co-infiltrated with an ATL31C143S-nYFP construct in Nicotiana cells as indicated. Images were acquired using a confocal laser microscope. Bars = 100 μm. d) A GAI translational fusion with cCFP was co-infiltrated with a ATL31ΔTM-nYFP construct in Nicotiana cells as indicated. Images were acquired using a confocal laser microscope. Bars = 100 μm. e) FLIM analysis of GAI-GFP, ATL31-mCherry and ATL31C143S-mCherry in Nicotiana leaf cells. Fusion proteins were expressed from an estradiol-inducible promoter and imaged 3 to 4 d after infiltration. Micrographs display pixel-wise fitted average fluorescence lifetime values of GFP in representative individual cells expressing GAI-GFP with or without free mCherry, ATL31-mCherry and ATL31C143S-mCherry. Twenty-six to 35 different cells were analyzed per genotype. Different letters denote statistically significant differences (P < 0.05) using 1-way ANOVA with the Tukey HSD test for multiple comparisons.

Since ATL31 is a plasma membrane protein and DELLAs are known to act in the nucleus, we used bimolecular fluorescence complementation (BiFC) to locate the ATL31-GAI/RGA interaction. As controls we used GRF6, a protein known that interacts with ATL31 at the plasma membrane (Sato et al. 2011; Yasuda et al. 2014), and a plasma membrane marker based on the aquaporin PIP2A fused to mCherry (Nelson et al. 2007). Merged GFP and mCherry images show that interactions between ATL31 and the GAI and RGA proteins occur at the plasma membrane (Fig. 3c). Moreover, this emplacement relies on the ATL31 transmembrane domain (TM), as its deletion leads to nuclear localization of the GFP signal (Fig. 3d). To confirm the BiFC results, we carried out FRET-FLIM experiments known to be less prone to false positives and to retain quantitative information about the kinetics and dynamics of the interaction (Strotmann and Stahl 2022; Maika et al. 2023). We selected GAI-GFP as donor, because of its stronger tendency to show cytoplasmic location compared with RGA (Figure S9), and ATL31-mCherry to serve as an acceptor. Additionally, we included a catalytically inactive ATL31 version (ATL31C143S; Sato et al. 2009) in the analysis to test if its E3-ligase activity might interfere with the interaction being tested. As shown in Fig. 3e, GFP lifetime decreased significantly in the presence of the ATL31:mCherry, an effect accentuated when the catalytically inactive version was used. In contrast, no significant reduction in GFP lifetime was detectable by using free mCherry as an acceptor. These results strongly support that membrane-linked ATL31 interacts with cytoplasmic DELLAs. They also suggest that such interaction promotes DELLA degradation.

To visualize the dual nuclear-cytosolic localization in Arabidopsis we incubated ProGAI:GAI-GFP seedlings in 1 M mannitol solution, which promotes plasmolysis and separation of the cell membrane from the cell wall. Confocal imaging demonstrated that the subcellular distribution of the GAI protein was similar to that observed after transient expression in Nicotiana cells, with stronger fluorescence in the nucleus but also present in the cell periphery (Fig. 4a). Biochemical support for these findings was obtained by immunoblotting subcellular protein fractions from C/N stressed and nonstressed pGAI:GAI-GFP and pRGA:GFP-RGA seedlings. As shown in Fig. 4b, the GAI-GFP and GFP-RGA proteins were both detected in the nuclear and cytosolic fractions. To further confirm the DELLA-ATL31 interaction in Arabidopsis, we analyzed lines overexpressing the ATL31 or its catalytically inactive version (ATL31C143S) fused to the GFP epitope. Compared with WT plants, overexpression of ATL31 resulted in enhanced growth and tolerance to C/N stress as well as lower RGA levels. In contrast, ATL31C143S lines showed similar growth, stress sensitivity and RGA levels (Ler; Fig. 4c; Figure S10). We carried out Co-IP experiments with these lines. As shown in Fig. 4c, the RGA protein co-immunoprecipitated from extracts of leaves overexpressing the ATL31C143S-GFP protein. However, we did not detect co-immunoprecipitation of DELLAs with the active WT-ATL31 protein. ATL31 is known to ubiquitinate target proteins (Sato et al. 2009, 2011; Liu et al. 2022). We show that only WT-ATL31 promotes DELLA ubiquitination, despite the observation that the active and inactive versions of ATL31 can interact with DELLAs (Fig. 4d; Figure S11). These findings, along with our results from FRET-FLIM experiments (Fig. 3e), support that ATL31-DELLA interaction in Arabidopsis and Nicotiana triggers the ubiquitination and degradation of DELLA proteins. We also tested whether the interaction promotes proteasome-dependent DELLA degradation by using transient expression in Nicotiana. As shown in Fig. 4e, the abundance of the GAI and RGA proteins was significantly reduced when co-expressed with ATL31, an effect reverted by the proteasomal inhibitor MG132.

ATL31 promotes cytosolic GAI and RGA degradation by the proteasome. a) Plasmolysis experiment for GAI-GFP. The indicated genotypes were grown for 4 d in MS/2 and then were treated with 5 µM PAC. After 48 h, the seedlings were immersed in a 1 M mannitol solution for 80 min. Images were acquired using a confocal laser microscope. Bars = 10 μm. b) Western blots of subcellular fractions isolated from pGAI:GAI-GFP and pRGA:GFP-RGA seedlings grown for 5 d in the indicated media. The GAI-GFP and GFP-RGA proteins were detected by anti-GFP antibodies. Histone H3 (anti-H3) was used as marker for the nuclear fraction, and UGPase (anti-UGPase) was used as marker for the cytosolic fraction. c) Ler, ATL31-GFP and ATL31C143S-GFP overexpressing seedlings were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen and 200 mM glucose. Proteins were then extracted and subjected to immunoblotting with an anti-RGA antibody. In parallel, Co-IP assay with anti-GFP was performed. Immunoprecipitation with an anti-GFP antibody was immunoblotted with anti-GFPHRP (α-GFP) or an anti-RGA (α-RGA) antibodies. d) Ubiquitination experiment for GAI-GFP. GAI-GFP, ATL31-mCherry and ATL31C143S-mCherry were expressed from an estradiol-inducible promoter in Nicotiana leaves and collected 3 d after infiltration. 50 µM MG132 was infiltrated 24 h before collection to prevent degradation. A Co-IP assay with anti-GFP was performed and immunoblotted with anti-GFPHRP (α-GFP) or an anti-UBQ (α-UBQ). e) Degradation assay for ATL31 target proteins. GFP-tagged GAI and RGA proteins were expressed alone or in combination with a 35S:ATL31 construct (ATL31ox) in Nicotiana leaves. Protein extracts were incubated in the presence or absence of 50 µM MG132 for 2 h and immunoblotted with an anti-GFPHRP antibody.

ATL31 reduces sensitivity to C/N stress even in the absence of GA biosynthesis

We next try to determine whether the ATL31-mediated control of DELLA stability was independent of the GA pathway. Firstly, we evaluated the C/N stress sensitivity of Arabidopsis lines carrying GA-resistant DELLA genes (Δ17; Peng et al. 1997; Dill et al. 2001; Barro-Trastoy et al. 2022) and compared them with their corresponding WT versions (Fig. 5a). All lines tested performed similarly under stress conditions (Fig. 5b), supporting that the presence of DELLA alleles nondegradable by GAs was not altering stress sensitivity. Next, we investigated whether ATL31 might target GA-resistant DELLA alleles, as it was able to interact with DELLAs lacking the N-terminal residues required for GA-mediated degradation (Fig. 3a). To test this hypothesis, we obtained protein extracts from pGAI:GAI△17*-3xYPET* seedlings and measured the kinetics of the GAI△17-3xYPET protein stability, after incubation with protein extracts from Arabidopsis WT (Ler) and 35S:ATL31 seedlings grown under C/N stress conditions. While the WT extracts caused a gradual reduction in GAI△17 abundance (it was still detected after 180 min incubation), those from 35S:ATL31 caused a quick reduction (GAI△17 was not detectable after 45 min incubation; Fig. 5, c and d).

*GA-resistant DELLAs can be degraded by ATL31 and do not enhance sensitivity to C/N stress. a) pGAI:GAI-3xYPET, pRGA:GFP-RGA and their corresponding Δ17 version seedlings were grown for 5 d in MS-N plates supplemented with 0.3 mM nitrogen (0.3N) in the absence or presence of 100 mM glucose (100C). b) Hue values recorded for seedlings photographed in a) were individually analyzed with Image J. For statistical analyses C/N stress measurements were normalized to the 0.3N values. No statistically significant differences found (P < 0.01) using Welch's test. Welch's t-statistic and effect size measurement eta-squared (η2) are indicated. (c) Protein extracts from pGAI:gaiΔ17-3xYPET seedlings were incubated for the indicated times with extracts from Ler seedlings harboring or not a 35S:ATL31 construct and grown under C/N stress (200C:0.3N). An anti-GFPHRP antibody was used for immunoblotting. Analog results were obtained 3 times. Lower band coincides with gaiΔ17-1xYPET size. d) Relative intensity of gaiΔ17 in 3 independent experiments was quantified by Image J. The initial protein levels were defined as 1. Coomassie blue staining served as loading control. Asterisks indicate statistically significant differences (*P < 0.05; *P < 0.01) for Ler versus ATL31ox using multiple Student’s t-test.

To establish a genetic relationship between ATL31, DELLAs and the C/N stress, we introduced a 35S:ATL31 construct into different backgrounds. The sensitivity to C/N stress was reduced when ATL31 was overexpressed (ATL31ox) in Ler (Fig. 6a; Figure S12) and Col-0 (Sato et al. 2009; Fig. 6a; Figure S12). In line with our previous findings (see Fig. 1, d and e), the sensitivity to C/N stress was higher in a GA-deficient background (ga1-3) containing specific combinations of della mutations (rgl2-1, rgl2-1 rga-t2, or rgl2-1 gai-t6). Overexpression of ATL31 reduced C/N stress sensitivity in these backgrounds, pointing to a GA-independent role in promoting DELLA degradation (Fig. 6, b and c; Figure S12). Indeed, RGA protein levels drastically decreased in the ga1-3 rgl2-1 gai-t6 mutant when ATL31 was overexpressed (Fig. 6d), an observation that cannot be attributed to alterations in GA3ox1 or DELLA mRNA levels due to changes in ATL31 function (Fig. 6e; Figure S13).

ATL31 reduces sensitivity to high C/N even in the absence of GA biosynthesis. a) Growth of Col-0 and Ler seedlings with and without a 35S:ATL31 construct (ATL31ox) after 5 d. MS-N plates were supplemented with 0.3 mM nitrogen (0.3N) in the absence or presence of 100 mM glucose (100C). b) The same analyses of panel a) were conducted with the ga1-3 mutant background and the indicated combinations of della loss of function alleles (rgl2-1, rga-t2, gai-t6).  c) Hue values for seedlings photographed in a) and b) were individually analyzed with Image J. Image J was used for image scoring. For statistical analyses C/N stress measurements were normalized to the 0.3N values. Asterisks indicate statistically significant differences (P < 0.001) using Welch's test. d)  ga1-3 rgl2-1 gai-t6 seedlings with and without a 35S:ATL31 construct (ATL31ox) were grown for 5 d under C/N stress conditions (100C:0.3N). Proteins were then extracted and subjected to immunoblotting with an anti-RGA antibody. (e)  GA3ox1 mRNA levels of Col-0, atl31 and ATL31ox under C/N stress. RT-qPCR analysis of RNA from Col-0 (C), atl31 (K) and ATL31ox (O) grown for 2 or 3 d under the indicated conditions. The values were normalized to UBC21 expression. GA3ox1 and UBC21 specific primers are listed in Table S1. Averages and SE of 3 replicates are shown.

Early stages of plant development regulated by DELLAs are not affected by ATL31 in the absence of C/N stress

We evaluated the possibility that ATL31 might be involved in DELLA degradation in the absence of C/N stress, interfering with early stages of plant growth regulated by these proteins (Cao et al. 2005; de Lucas et al. 2008; Feng et al. 2008; Blanco-Touriñán et al. 2020b). For this purpose, we carried out seed germination and hypocotyl elongation experiments using WT and ATL31ox seeds under standard growth conditions. Freshly harvested (FH) Arabidopsis seeds are dormant and unable to germinate unless dormancy is relieved by storage in dry conditions, the so-called after-ripening (AR; Carrera-Castaño et al. 2020). As shown in Fig. 7a, overexpression of ATL31 did not alter the dormancy of FH seeds, nor the germination of AR seeds. Likewise, no significant differences were found in the hypocotyls of WT and ATL31ox seedlings grown in darkness, under the light or in the presence of PAC (Fig. 7b). These results indicate that, under nutritionally balanced conditions, ATL31 function does not have any noticeable impact on seed dormancy, seed germination, and seedling establishment.

Model for the role of the ATL31-mediated DELLA degradation mechanism on plant growth and C/N stress responses. a) Germination percentage (%) of FH Col-0 versus 35S:ATL31 (ATL31ox) FH seeds at 72 h after imbibition (hai) and germination kinetics of seeds AR seeds for 4 wk. b) Hypocotyl lengths of seedlings grown for 7 d with or without lighting in the absence or presence of PAC. c) Proposed model. Under balanced C/N conditions, rising in N levels promotes DELLA degradation by the GA pathway (Camut et al. 2021) and ATL31 expression is reduced (Aoyama et al. 2014). C/N stress represses the GA pathway (Fig. 2) and enhances ATL31 mRNA and protein levels (Aoyama et al. 2014; Yasuda et al. 2017). ATL31 interacts in turn with specific DELLAs in the plasma membrane, targeting them for degradation.

Discussion

Our results support a mechanism for DELLA degradation independent of the canonical GA-GID1 pathway, previously associated with responses to low N availability. Low nitrate enhances DELLA accumulation, while concentrations above 1 mM increase both AtGA3ox1 transcripts and GA levels, which in turn decrease DELLA stability via GA-mediated degradation (Camut et al. 2021). Likewise, ATL31 expression is reduced upon increased N availability (from 0.3 to 30 mM; Aoyama et al. 2014), suggesting that this E3 ligase is not involved in DELLA degradation under balanced C/N contexts. Whereas high DELLA levels are observed under low N availability (0.3 mM; Camut et al. 2021), our results show that they can be further raised at high C/N ratios (100C:0.3N or 200C:0.3N; Fig. 2a). We also found that C/N stress further reduces GA3ox1 gene expression and GA levels relative to low N conditions (Fig. 2, b to d), suggesting that enhanced DELLA stability is due to reduced GA-mediated degradation. C/N stress also increases ATL31 mRNA and protein levels in adult plants (Aoyama et al. 2014; Yasuda et al. 2017). We have shown that GA3ox1 repression precedes the accumulation of ATL31 in seedlings (Fig. 2b; Figure S8), raising the question of whether DELLAs might be involved in promoting ATL31 expression. In any case, these kinetics are congruent with our findings that ATL31 controls the stability of DELLAs and reduces the hypersensitivity of a GA biosynthetic mutant to C/N stress (Fig. 6). These results reinforce the notion of a GA-independent mechanism, which is further supported by the fact that the C/N sensitivity of WT plants is not enhanced by expression of GA-resistant DELLAs, as they are also degraded by ATL31 (Fig. 5). Two findings support a complementary role for GAs in preventing excessive DELLA accumulation based on: (1) Under C/N stress, lower hue values (higher sensitivity) were observed when RGA and/or GAI WT alleles were present in a ga1-3 background, relative to its WT (Ler; Fig. 1c) or to atl mutants (Fig. 1a). (2) while ATL31ox significantly increases hue values under C/N stress in a ga1-3 rgl2-1 background, these are still lower than in WT lines overexpressing ATL31 (Ler or Col-0; Fig. 6c). Besides C/N stress and N availability (Camut et al. 2021), DELLAs are involved in other nutritional stresses that promote their accumulation by reducing GA-mediated degradation (Jiang et al. 2007; Wild et al. 2016). It follows (from our data) that ATL31 is probably involved in these responses.

Whereas DELLA stabilization is known to increase stress tolerance at the expense of growth, we show here that ATL31-mediated GAI and RGA degradation can enhance C/N stress tolerance and growth. This response cannot be phenocopied by reducing GA-mediated DELLA degradation via blocking GA biosynthesis with PAC or altering GA/ABA ratios (Figure S3), which suggests that other mechanisms must be involved. In this regard, it has been shown that ABA synthesis is not a predominant factor in the C/N stress response, although ABA signaling components (ie, ABI1) are involved independently of ABA biosynthesis (Lu et al. 2015). At the same time, it is unknown whether specific GRF proteins that accumulate under C/N stress (Sato et al. 2011) affect DELLA functions. Enhanced ATL31 activity limits senescence in adult plants under unbalanced C/N conditions (Aoyama et al. 2014) and improves pathogen resistance (Maekawa et al. 2014). Despite the involvement of DELLAs in these processes (Wild et al. 2012; Chen et al. 2014; De Vleesschauwer et al. 2016; Li et al. 2019), a role for their degradation remains to be studied.

Besides their nuclear activity, DELLA levels have nontranscriptional effects on processes occurring outside the nucleus. For instance, nuclear DELLA levels were found to have a negative impact on the amount of the prefoldin complex available in the cytoplasm that regulates cell expansion (Locascio et al. 2013). Similarly, DELLAs regulate bidirectional protein trafficking between the vacuole and the cell surface (Salanenka et al. 2018). Here, we have found that cytosolic RGA and GAI interact with the membrane-bound ATL31 protein (Figs. 3 and 4). The association between DELLAs and a membrane protein reported here adds an extra layer of complexity to GA control by affecting the amount of DELLAs available for nuclear import. Interestingly, GAs are known to increase cytosolic Ca^+2^ (Okada et al. 2017), a second messenger required to regulate the ATL31-mediated C/N-nutrient response (Yasuda et al. 2017) and CALCINEURIN B-LIKE 8 (CBL8) mediates Ca^+2^-dependent recruitment of CIPK14 at the plasma membrane to enhance phosphorylation and activation of ATL31 (Yasuda et al. 2017). DELLA stability is controlled by phosphorylation when targeted by the GA pathway (Qin et al. 2014), opening the possibility that this modification is also important for ATL31-mediated degradation.

The finding that enhanced expression of ATL31 does not affect early stages of plant development (Fig. 7, a and b) suggests that the effect of ATL31 on DELLA levels is negligible under balanced C/N ratios. Similarly, it has been shown that the germinative response does not depend on ATL31 expression levels, whether in the presence or absence of ABA (Sato et al. 2009; Pavicic et al. 2019). In Fig. 7c, we propose a model for the roles of ATL31-mediated DELLA degradation. Upon stress perception, GA-mediated degradation decreases, leading to DELLA accumulation, reduced growth, and improved stress tolerance. By focusing on C/N stress, a situation typically associated with low N levels, we have found that GAI and RGA accumulation is partially counteracted by active ATL31-mediated degradation, most likely to balance growth and stress tolerance. Failure to control DELLA levels under these conditions might lead to excessive accumulation, negatively affecting survival metabolism and/or growth resumption upon stress withdrawal (Fig. 7c). This suggests that manipulation of ATL31 levels can be used to fine tune the DELLA-mediated stress response, with presumably lower impact on plant development than manipulating the GA pathway. The level of conservation of the ATL31 pathway in crop species will condition transferability of this mechanism to improve productivity.

Methods

Plant materials and treatments

Unless otherwise specified all seeds were in the Ler background. The *ga1-3 rgl2-*1 (Lee et al. 2002), ga1-3 rgl2-1 rga-t2, ga1-3 rgl2-1 gai-t6, ga1-3 rgl2-1 rga-t2 gai-t6, and ga1-3 rgl2-1 rga-t2 gai-t6 rgl1-1 seeds (Yu et al. 2004) were obtained from Dr. S. Prat (CRAG, Spain). The gai-t6 rga-24 seeds were obtained from Dr D. Alabadí (IBMCP, Spain) and the pRGA:GFP-RGA (Silverstone et al. 2001), pGAI:GAI-GFP (Fleck and Harberd 2002), pGAI:GAI-3xYPet (Gómez et al. 2023), and pGAI:gaiΔ17-3xYPet (Barro-Trastoy et al. 2022) seeds were obtained from Dr. MA Pérez-Amador (IBMCP, Spain). The ga1-3 rgl2-13 rga-28 rgl1-2 seeds were in the Col-0 background (Tyler et al. 2004) and kindly provided by Dr Tai-ping Sun (Duke University, United States). The gai-td1, rga-28 and gai-td1 rga-29 were in the Col-0 background (Conti et al. 2014; Plackett et al. 2014) and kindly provided by Dr Stephen G. Thomas. The atl31-1 (GABI_746D08) and atl6-1 (SALK_083652) seeds were in the Col-0 background and obtained from NASC. The GA biosensor line in the Col-0 background containing the Venus and Luciferase reporter genes under the control of a GAI-derived degron (hormone activated Cas9-based repressors; HACRs; Khakhar et al. 2018) was kindly provided by Dr Nemhauser (University of Washington, United States). Surface-sterilized seeds were sown in Murashige and Skoog Modified Basal Salt Mixture (MS-N) medium supplemented with different concentrations of nitrogen and/or carbon sources. Nitrogen added to plants was an equimolar mix of NH_4_NO_3_ and KNO_3_ (0.3 mM = 0.1 mM NH_4_ + 0.2 mM NO_3_) and D-glucose was used as a carbon source as described in Sato et al. (2009). After stratification (3 days in the dark at 4 °C), plates were sealed with Micropore tape and transferred to 16 h light (cool-white fluorescent; 110 μmol m^−2^ s^−1^) at 22 °C/8 h dark at 20 °C and 60% relative humidity. For C/N stress 2 different ratios were used, as described in Sato et al. (2009), by using either 100 or 200 mM glucose in the presence of 0.3, 3, or 30 mM total nitrogen. GA_4+7_ was used for GA treatments (Duchefa).

Carbon/nitrogen assay and visual scoring

Experiments were essentially as in Sato et al. (2009). MS-N medium without nitrogen or sugars was used for these assays. The following amounts of nitrogen or/and glucose were added when indicated: 0.3 mM N (0.1 mM KNO_3_, 0.1 mM NH_4_NO_3_, and 10 mM KCl), 100 mM or 200 mM glucose. In figures, the concentrations of GAs are given in μM units. Approximately 50 to 100 seeds of each genotype were surface-sterilized and sown on the appropriate media. After cold stratification for 3 days, seeds were transferred to 16 h light at 22 °C/8 h dark at 20 °C and 60% relative humidity. Visual scoring was calculated by recording percentages of green and red cotyledons in 5-d-old seedlings. At least 70 seedlings were visually scored for each genotype and growth condition.

Quantification of chlorophylls and anthocyanins

As a preliminary approach and to test the image analysis method shown in Figure S2, pigment extractions were carried out as in Lichtenthaler (1987) and Yin et al (2012). We processed 5 d-old seedlings by freezing in liquid nitrogen and macerating them while in a frozen state. Then, 2 samples of 50 mg of tissue were separated for each pigment extraction with 1 mL acetone 80% for chlorophylls and carotenoids and 1 mL MeOH + 1% HCl for anthocyanins, respectively. After 5 min shaking and 1 h at 4 °C in the dark, extracts were clarified by centrifugation (17,000 × g at room temperature for 5 min) and placed in 96 well microplate. Finally, the full visible light spectrum absorbance (400 to 700 nm) was obtained with a SPECTROstar Nano spectrophotometer. Pigment quantities were obtained following the equations provided by the references. The remaining anthocyanin quantifications were performed using a modified version of Yin et al (2012). Briefly, 750 μL of acidic methanol (1% HCl, w/v) were added to 5 d-old Arabidopsis seedlings (between 40 and 65 per genotype and growth condition). After shacking for 5 min, anthocyanins were extracted by incubating at 4 °C for 1 h in the dark. Extracts were clarified by centrifugation (17,000× g at room temperature for 5 min) and anthocyanins were quantified from A530 and A657 values by using the empirical formula Q = (A530–0.25*×A657)×A*^−1^, where Q is a corrected absorption value linearly correlated with the amount of anthocyanins and A is the mean cotyledonary area of the plant material used for extraction.

Cotyledon color measurements by digital imaging

Five-days-old seedlings were photographed (Sony ILCE-6000) on a matt black baseboard with an integrated lighting system KAISER RB 5000 DL equipped with 2 daylight fluorescent lamps (36 W, 5,400 °K). In order to minimize color fluctuations between plates and experiments, identical lighting conditions and camera settings were applied. Next, ImageJ was used to convert pictures into the CIELAB color space (strictly CIE 1976 Lab*) as in Kendal et al (2013). In this space, the coordinates used to specify colors are device- and set-up independent. This system expresses color as 3 values: (i) lightness dimension (L*); (ii) position in the chromatic a* axis (from green [−a*] to red [+a*]); (iii) position in the chromatic b* axis (from blue [−b*] to yellow [+b*]). Cotyledon areas were selected for each seedling, saved in a ROI Manager tool, and mean gray values were measured for channels a* and b*. Both channels were then combined by calculating the hue angle [hue = arctan(b*/a*); range: 0 to 360 degrees], which allows association of each cotyledon pair to a specific position in the color spectrum wheel. Lightness was excluded from these measurements to avoid skewing due to reflections or shadows. An average of 50 seedlings per genotype and growth condition were analyzed.

Statistical analysis

For each plant assay with color measurements, statistical analysis was performed considering genotype as the fixed factor and hue values as the response variable, with data stratified by growth medium. For statistical analyses, high C/N ratio measurements were normalized to 0.3N. Statistical significance (Supplementary Data Set S1) was analyzed using 1-way ANOVA followed by a Tukey's HSD post hoc test. When, according to Levene's test, data did not meet the assumption of homoscedasticity, 1-way ANOVA followed by a Games-Howell post hoc test, or t-test, depending on the number of groups, both with Welch's correction, were applied. In addition, as a way to determine the relevance of the studied mutations in plant responses to treatments, the phenotypic divergence between the lines was quantified. Each test was accompanied with eta-squared (η^2^) effect size measurements eta-squared (η^2^), thus reflecting variance proportion between samples in the dependent variable which can be associated with genotype belonging. All statistical analyses applied to digital color measurements were conducted using the R environment (v. 4.1.1; R Core Team, 2021, https://www.R-project.org/), the rstatix (v0.7.0; Kassambara, 2021, https://CRAN.R-project.org/package=rstatix), onewaytests (v2.6.0; Dag et al. 2018), and effectsize (v0.6.0.1; Ben-Shachar et al. 2020) packages.

Germination scoring and hypocotyl length measurements

Freshly harvested seeds and seeds after 4 wk of dry storage were used for germination assays. For each genotype, at least 150 seeds from 6 different mother plants were scored for germination at the indicated time points as described in Carrera-Castaño et al. (2024). To measure the hypocotyl length, seeds were surface-sterilized, placed on MS/2 plates (with or without 0.1 μM PAC), sealed with porous tape (Micropore 3 M) and stratified at 4 °C in the dark for 3 to 5 d. Plates were incubated at 22 °C under 16/8-h light/dark conditions or in the dark for 7 d, as indicated. At least 25 seedlings per genotype and condition were scanned. Hypocotyl length was measured with Image J software (https://imagej.net/ij/).

Generation of constructs for plant and yeasts assays

The ATL31 coding sequence (CDS) was amplified by PCR with primers LO2170 and LO2171 or LO2352 and cloned into a donor vector (pDONR207) with BP clonases (Invitrogen). Also, a point mutation to prevent degradation was generated in the RING domain of ATL31 CDS (ATL31C143S; Sato et al. 2009). We used primers LO2170 and LO2353 in combination with LO2354 and LO2352 to clone the mutated version into pDONR207. To generate 35S:ATL31 and 35S:ATL31C143S Arabidopsis plants, the ATL31 and ATL31C143S CDS were transferred from the donor plasmid to the pEarlyGate201 and/or pGWB605 plasmid (Earley et al. 2006; Nakamura et al. 2010) by a LR reaction (Invitrogen) and introduced into different Arabidopsis genotypes by floral dipping (Clough and Bent 1998). For ATL31 transcriptional and translational fusions to luciferase (LUC), we used the primer LO2417 in combination with LO2419 or LO2418, respectively. The PCR-amplified DNA fragments were cloned into the pBGWL7 plasmid (VIB-Ugent, Belgium) to generate the proATL31:LUC and proATL31:ATL31-LUC constructs. These constructs were used to generate Arabidopsis plants as explained above. For Co-IP assays, the ATL31 CDS was transferred from the donor plasmid to pMDC43 (Curtis and Grossniklaus 2003) to obtain a translational fusion with the GFP CDS. The DELLA CDSs were cloned into pGWB14 (Nakagawa et al. 2007) or the pEarlyGate 201 plasmid (Earley et al. 2006) to obtain translational fusions with the HA epitope as follows: (i) The GAI CDS was transferred from a pDEST22-GAI construct isolated from a yeast library (Castrillo et al. 2011) to a pDONR207 before recombination with the pEarlyGate201 plasmid. The RGA CDS was obtained from Dr Salomé Prat (pYL:RGA-YFP; Nieto et al. 2015) and recombined into pGWB14.

For BiCF experiments, the ATL31C143S CDS, as well as those of GAI and RGA (CDS obtained from Dr Federico Valverde), were transferred to plasmids PXCGW or pNXGW plasmids (Kim et al. 2009; Rombolá-Caldentey et al. 2014) for overexpression as translational fusions to cCFP or nYFP, respectively. The binary plasmid containing the plasma membrane aquaporin PIP2A fused to mCherry (pm-rb CD3-1008) has been previously described (Nelson et al. 2007). ATL31-ΔTM was generated by site-directed mutagenesis by using primers LO2170 and LO2385 in combination with LO2352 and LO2384 and cloned into pDONR207. This ATL31-ΔTM CDS was transferred from its donor to the PXCGW plasmid by a LR reaction for overexpression as a translational fusion to nYFP.

For FRET-FLIM and ubiquitination assays, the GAI CDS (amplified by PCR with primers LO2346 and LO2363) was transferred to pABindGFP (Bleckmann et al. 2010) from its pDONR207:CDS, to obtain a translational fusion to the complete GFP CDS under the control of an estradiol-inducible promoter. ATL31 and ATL31C143S ORFs were transferred from its donor to pABindmCherry (Bleckmann et al. 2010) to obtain translational fusions to mCherry under the control of estradiol-inducible promoters.

For degradation assays in Nicotiana, the GAI CDS was transferred to pMDC43 (Curtis and Grossniklaus 2003) from its pDONR207:CDS, to obtain translational fusions to the complete GFP CDS. To overexpress the RGA-YFP fusion protein we used the pYL:RGA:YFP construct (Nieto et al. 2015). ATL31 ORF was transferred from its donor to pEarlyGate201 to obtain a translational fusion to the HA sequence.

Control constructs (AD-0 and AD-GFP) and constructs carrying GAL4BD-GAI or RGA (BD-GAI or BD-RGA) used in yeast 2 hybrid (Y2H) assays have been previously described (Rombolá-Caldentey et al. 2014). The ATL31C143SΔTM CDS lacked the transmembrane domain and contained a point mutation abolishing the ligase activity, both required to use this protein in Y2H assays (Sato et al. 2011). It was amplified by nested PCR, first using primers LO2304 and LO2171 and then primers LO1 and LO604. The ATL31C143S:pDONR207 construct, previously generated for BiFC experiments, was used as template. The amplified product was cloned into pDONR207 by a BP reaction and transferred into the pGADT7 plasmid (GAL4AD fusion) by a LR reaction.

The Platinum SuperFi II DNA Polymerase was used for PCR amplifications and all constructs were fully sequenced.

Yeast transformation and 2-hybrid assays

For a complete description of the methods and yeast strains used here, see Sánchez-Montesino and Oñate-Sánchez (2017, 2018). Yeast colonies were photographed (Lumix DMC-FZ300) on a matt black baseboard.

RNA isolation and RT-qPCR

Total RNA isolation from seedlings (15 mg) has been previously described (Oñate-Sánchez and Vicente-Carbajosa 2008; Oñate-Sánchez and Verdonk 2021). First-strand cDNA synthesis and RT-qPCR were carried out as previously described (Rombolá-Caldentey et al. 2014). To compare data from different cDNA samples, expression values for all selected genes were normalized to the corresponding expression values for UBC21 (LO729 and LO730; Dekkers et al. 2012). Primers used are listed in Table S1

Transient expression analyses in Nicotiana benthamiana

Agroinfiltration of 4-wk-old Nicotiana benthamiana leaves was carried out as previously described (Rombolá-Caldentey et al. 2014). A pBIN61-35S:P19 plasmid was always co-infiltrated to avoid gene silencing (Voinnet et al. 2003). A plasma membrane marker based on aquaporin PIP2A fused to mCherry (Nelson et al. 2007) was co-infiltrated when indicated. For BiFC experiments, Agrobacterium cultures were used at OD_600_ = 0.3, and for degradation assays, 0.2 or 0.5 were used for target proteins or ATL31, respectively. On the morning of the fourth day after agroinfiltration, leaves were harvested or used for microscopy. BiFC images were taken with a Leica CLSM TCS-SP8 confocal microscope. Transient transformation of N. benthamiana for FRET-FLIM experiments was performed as described in Blümke et al. (2021). For GAI-GFP ubiquitination analysis, 4-wk-old N. benthamiana leaves were agroinfiltrated and collected after 72 h. Twenty-four hours prior to collection, 50 µM MG132 and 20 µM estradiol were co-infiltrated.

Confocal microscopy and plasmolysis

Images were acquired using a Leica CLSM TCS-SP8 or a Zeiss LSM 880 camera with Airyscan confocal microscope applying a 20× water immersion objective. For detection and localization of the fluorophores, GFP was excited with a 488 nm argon ion laser and detected between 500 and 520 nm, while mCherry was excited at 561 nm and detected between 609 and 629 nm, and Venus was excited at 515 nm and detected between 517 and 550 nm.

For plasmolysis experiment, 4-d-old plants were treated with 5 µM PAC. After 48 h, the seedlings were immersed in a 1 M mannitol solution for 80 min prior to confocal visualization.

Fluorescence lifetime imaging

Fluorescence lifetime was measured at a Zeiss LSM 780 confocal microscope (40× water immersion objective, Zeiss C-PlanApo, NA 1.2). For TCSPC a PicoQuant Hydra Harp 400 (PicoQuant, Berlin, Germany) was used. Photon counting was performed with picosecond resolution. GFP was excited with a 485 nm (LDH-D-C-485, 32 MHz, PicoQuant, Berlin, Germany) pulsed polarized laser. mCherry was excited with a 561 nm laser with 1% laser power. Laser power at the objective lens was adjusted to 1 µW for the 485 nm laser. Light, emitted from the sample, was separated by a polarizing beam splitter before photons were selected with a band-pass filter. For GFP a 520/30 band-pass filter and for mCherry a 607/70 band-pass filter was used. When GFP served as a donor, a LP610 beam splitter was used. Photons were detected in both donor and acceptor channel simultaneously with Tau-SPADs (PicoQuant, Berlin, Germany). Images were acquired at zoom 8 resolution of 256 × 256 pixel with a pixel size of 0.1 µm and a pixel dwell time of 12.54 µs and laser repetition rate of 32 MHz. Photons were collected over 40 to 60 frames. To avoid pileup effects, cells containing high donor concentrations were avoided. Before image acquisition, the system was calibrated for each donor. For this, the objective was adjusted to reach a maximal count rate. FCS curves of Rhodamine110 dye and water were acquired to monitor the system function. Internal response function for the 485 nm laser was determined by measuring the fluorescence decay of quenched erythrosine in saturated KI using the same hardware settings as for the FRET pair.

Fitting of the fluorescence decays

The fluorescence decays of selected ROIs in the FLIM image were analyzed with the SymPhoTime FLIM analysis software (SymPhoTime 64, version 2.4; PicoQuant, Berlin, Germany). TCSPC bins of channel 1 (parallel light) were binned by 8 resulting in a bin width of 8 ps. Cell peripheries were selected by hand using the ROI tool. Chloroplasts and pixels above the pileup limit (10% of the laser repetition rate) were manually removed. Decays from donor only samples were fitted with the FLIM analysis tool (fitting model: n-exponential reconvolution). Judged by fitting residuals and χ^2^ test, 1 lifetime (model parameter n = 1) was needed to fit donor only decays containing GFP.

FRET samples (containing GFP and mCherry) as well as donor only samples (only GFP) were fitted using the grouped FLIM image analysis tool. Parameter “τ_1_” was fixed as the average donor only lifetime measured beforehand in the FLIM analysis.

The second lifetime parameter “τ_2_” is corresponding to the FRET fraction of the sample and was fitted within limits corresponding to 10% and 80% of the average lifetime acquired in the donor only samples. The intensity-weighted lifetime was considered as the sample's apparent lifetime. All measurements were performed in at least 2 independent experiments.

In vivo imaging of bioluminescence

In vivo imaging and quantification of luciferase activity was carried out using a cooled CCD camera (NightOWL II LB 983 NC-100; Berthold Technologies) and the provided software. HACR-GAI seeds were germinated for 3 d under control conditions (0.3N). After that, seedlings were placed in control (0.3N) or C/N stress conditions (100C:0.3N or 200C:0.3N). After 24 and 48 h, seedlings were sprayed with 200 mM of luciferin and imaged after 20 min.

Four T3 lines with ATL31-LUC fusions (2 transcriptional and 2 translational) were used to assess the regulation of the gene expression and protein stability by the C/N treatment. Three replicates of ca. 30 seedlings per line each, were grown in both low N (0.3 mM N) and C/N media (100C:0.3N). Pictures were taken with the CCD camera NightOwl at 24, 48, 54, and 72 h after imbibition and analyzed with ImageJ. The mean luciferase signal for each plant was quantified.

Immunoblot analysis and antibodies

For DELLA accumulation analysis, 30 to 50 mg of 5-d-old seedlings were ground in liquid nitrogen and homogenized in 100 µL of Co-IP extraction buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 5 mM β-mercaptoethanol and EDTA-free protease inhibitor cocktail; cOmplete-Roche). Cell debris was removed by centrifugation and total protein concentration was determined using the Bradford Protein Assay Kit (Bio-Rad). Samples with the same amount of protein were resolved on 8 or 10% SDS-PAGE gels and transferred to polyvinylidene fluoride membranes. Immunoblots were probed with anti-GFP^HRP^ antibody (Miltenyi Biotec; #130-091-833) at 1:4,000 dilution, anti-HA^HRP^ (Roche; #12013819001) at 1:1,000 dilution, anti-H3 (Abcam; #ab1791) at 1:2,500 dilution, anti-UGPase (Agrisera; #AS14 2813) at 1:10,000 dilution, anti-RGA (Agrisera; #AS11 1630) at 1:1,000 dilution, anti-UBQ11 (Agrisera; #AS08 307A) at 1:10,000 dilution or Goat anti-Rabbit IgG^HRP^ (Agrisera; #AS09 602) at dilution 1:25,000. Coomassie blue staining was used for loading control.

Co-immunoprecipitation assays

They were carried out as described in Sánchez-Montesino et al. (2019), with the following modifications. For immunoprecipitations, each protein extract was incubated with 25 µL of GFP-Trap Agarose (Proteintech; #gta-20) beads previously equilibrated as indicated in the manufacturer’s application notes. Beads were washed 4 times with Co-IP extraction buffer, eluted with 80 µL of loading buffer (SDS-sample buffer 2×) by boiling at 95 °C for 5 min, and then loaded onto SDS-PAGE gels.

Degradation assays

The Nicotiana in vitro degradation assays were performed as previously described (Igawa et al. 2009). For each combination tested, extraction buffer C was used to homogenize 7 agroinfiltrated leaf discs followed by incubation at room temperature for 2 h with 50 µM MG132 or DMSO. SDS-sample buffer (2×) was added to stop the reaction before immunoblotting. Degradation assays for each protein were replicated 2 times. For Arabidopsis in vitro degradation assays, Igawa et al. (2009) procedure was used but with modifications based on Wang et al. (2009). Briefly, 100 to 150 mg of 5-d-old WT or 35S:ATL31 seedlings and 250 mg of DELLA-reporter seedlings were harvested and homogenized in extraction buffer C. DELLA reporter extracts were clarified by 20 min centrifugation (17,000*×g*) at 4 °C and then mixed with miracloth-filtered WT or 35S:ATL31 extracts. These combinations were incubated at room temperature and reactions were stopped with (2×) SDS-sample buffer. After 5 min incubation at 95 °C, samples were centrifugated before immunoblotting. Degradation assay for gai-Δ17 was replicated 3 times and band intensity quantified by Image J. In both cases, Coomassie blue staining was used for loading control.

Subcellular fractionations

Subcellular fractionations were performed as previously described (LaMontagne et al. 2016) with modifications based on Bowler et al. (2004). Briefly, 1 to 2 g of 5-d-old pRGA:GFP-RGA and pGAI:GAI-GFP seedlings were harvested and homogenized in liquid nitrogen. We followed Bowler et al. (2004) for nuclear fraction isolation. To isolate cytosol fraction, the supernatant was subjected to 2 steps of centrifugation at 8,000*×g* at 4 °C. SDS-sample buffer (2×) was added to all fractions before immunoblotting. Coomassie blue staining was used for loading control.

Accession numbers

ATL6, At3g05200; ATL31, At5g27420; GA1, At4g02780; GA3ox1, At1g15550; GAI, At1g14920; GRF6, At5g10450; RGA, At2g01570; RGL1, At1g66350; RGL2, At3g03450; UBC21, At5g25760.

Supplementary Material

koag013_Supplementary_Data