Multiple Gibberellin Receptors Contribute to Phenotypic Stability under Changing Environments

The three tomato GID1 GA receptors exhibit redundancy under optimal growth conditions, but under changing environments, their overlapping functions contribute to phenotypic stability and robust growth.

Abstract

The pleiotropic and complex gibberellin (GA) response relies on targeted proteolysis of DELLA proteins mediated by a GA-activated GIBBERELLIN-INSENSITIVE DWARF1 (GID1) receptor. The tomato (Solanum lycopersicum) genome encodes for a single DELLA protein, PROCERA (PRO), and three receptors, SlGID1a (GID1a), GID1b1, and GID1b2, that may guide specific GA responses. In this work, clustered regularly interspaced short palindromic repeats (CRISPR) /CRISPR associated protein 9–derived gid1 mutants were generated and their effect on GA responses was studied. The gid1 triple mutant was extremely dwarf and fully insensitive to GA. Under optimal growth conditions, the three receptors function redundantly and the single gid1 mutants exhibited very mild phenotypic changes. Among the three receptors, GID1a had the strongest effects on germination and growth. Yeast two-hybrid assays suggested that GID1a has the highest affinity to PRO. Analysis of lines with a single active receptor demonstrated a unique role for GID1a in protracted response to GA that was saturated only at high doses. When the gid1 mutants were grown in the field under ambient changing environments, they showed phenotypic instability, the high redundancy was lost, and gid1a exhibited dwarfism that was strongly exacerbated by the loss of another GID1b receptor gene. These results suggest that multiple GA receptors contribute to phenotypic stability under environmental extremes.

INTRODUCTION

The plant hormone gibberellin (GA) regulates various developmental processes, including seed germination, cell and shoot elongation, leaf expansion, transition to flowering, flower growth, and fruit development (Davière and Achard, 2013). The nuclear DELLA proteins inhibit GA responses (Locascio et al., 2013). GA binding to the soluble GIBBERELLIN-INSENSITIVE DWARF1 (GID1) receptor increases the affinity of the latter to DELLA, leading to the formation of a GID1-GA-DELLA complex. This facilitates the interaction of DELLA with an Skp, Cullin, F-box (SCF) E3 ubiquitin ligase complex, via GID2/SLEEPY1 (SLY1) F-box proteins. The SCF^SLY1 complex then polyubiquitinates DELLA, targeting it for degradation by the 26S proteasome (Sasaki et al., 2003; Dill et al., 2004; Griffiths et al., 2006; Harberd et al., 2009; Hauvermale et al., 2012). DELLA degradation releases DELLA-interacting transcription factors, leading to transcriptional reprogramming and activation of GA responses.

GID1s can be divided into several evolutionarily conserved groups, with monocots expressing one group and eudicots expressing two: A type and B type (Yoshida et al., 2018). GID1 is a soluble protein showing sequence similarity to carboxylesterase (CXE) enzymes that hydrolyze short-chain fatty-acid esters; however, it lacks CXE activity (Ueguchi-Tanaka et al., 2007; Yoshida et al., 2018). GID1 was first identified in rice (Oryza sativa); the gid1-1 rice mutant is dwarf and insensitive to GA (Ueguchi-Tanaka et al., 2005). Although rice has a single GID1 gene, Arabidopsis (Arabidopsis thaliana) has three homologs named GID1a, GID1b, and GID1c (Griffiths et al., 2006; Nakajima et al., 2006). GID1a and GID1c belong to the A type group and GID1b to B type (Yoshida et al., 2018). Because the triple gid1a gid1b gid1c mutant in Arabidopsis is extremely dwarf and completely insensitive to GA, it was concluded that GID1 proteins are the only GA receptors in this species (Griffiths et al., 2006; Iuchi et al., 2007; Willige et al., 2007). GID1 contains two main motifs: the CXE domain that functions as a GA binding “pocket” and the N-terminal extension (N-Ex) domain. When GA binds to GID1, the N-Ex closes the GA binding pocket and then binds DELLA (Murase et al., 2008). Yamamoto et al. (2010) showed that Arabidopsis GID1b interacts with DELLA in a GA-independent manner. They also showed that although rice GID1 requires GA for interaction with DELLA, a mutation at the loop region (P99S) enables GA-independent interaction and responses.

The role of GAs in tomato (Solanum lycopersicum) plant development has been studied for many years. GA biosynthetic mutants, such as gib-1, gib-2, and gib-3 (GA-deficient mutants), as well as the GA signaling mutant procera (pro), were identified and characterized (Bensen and Zeevaart, 1990; Koornneef et al., 1990; Martí et al., 2007; Carrera et al., 2012; Livne et al., 2015). PROCERA (PRO) codes for the single tomato DELLA protein and loss-of-function pro alleles were identified (Van Tuinen et al., 1999; Livne et al., 2015) and found to cause significant stem elongation, late flowering, parthenocarpic fruit development, and increased transpiration (Livne et al., 2015; Nir et al., 2017). Tomato plants overexpressing the gain-of-function, stable DELLA mutant gene pro∆17, were dwarf and exhibited reduced transpiration (Nir et al., 2017). A recent study identified three putative SlGID1 (GID1) GA-receptor genes in tomato, but mutants were not found yet (Gazara et al., 2018; Shinozaki et al., 2018; Yoshida et al., 2018).

Arabidopsis has three GID1s, but rice and barley only one, raising the question of the necessity for multiple receptors. Mutant analyses in Arabidopsis revealed high redundancy, with differences in the contribution of each GID1 to GA responses (Griffiths et al., 2006). Although gid1a, gid1b, and gid1c single mutants and the gid1b gid1c double mutant exhibited normal development, gid1a gid1c and gid1a gid1b double mutants displayed clear developmental defects, including reduced inflorescence-stem length and lower fertility. This suggests that in Arabidopsis, GID1a has the most significant contribution to GA responses, which has been attributed to its high expression level (Griffiths et al., 2006). Althouh it has been speculated that different Arabidopsis GID1s bind specific DELLAs, analyses in yeast (Saccharomyces cerevisiae) suggested that all GID1s can interact with all five DELLA proteins (Griffiths et al., 2006; Nakajima et al., 2006). By contrast, Gallego-Giraldo et al. (2014) showed that during Arabidopsis fruit development, each GID1 variant binds specific DELLAs, dictating a unique role. The fact that Arabidopsis has three GID1s and five DELLAs makes it difficult to dissect the specific activities of each receptor. Tomato has three putative receptors (Shinozaki et al., 2018) but only one DELLA. Thus, tomato presents an ideal system for studying the specific and overlapping roles of the different SlGID1s (GID1s), and the importance of multiple GA receptors in the overall GA activity and plant development.

Here, we assessed the role of the tomato GID1 receptors in GA responses throughout tomato plant life cycle. The results show high redundancy in the activities of the three receptors in GA sensing and signaling under optimal controlled growth conditions. However, when grown in the field under changing environments, the redundancy was lost and all mutants showed phenotypic variability with increased dwarfism, suggesting that redundancy in GA sensing contributes to phenotypic stability under extreme environments.

RESULTS

The Three Tomato GID1s Are Functional GA Receptors

The tomato genome contains three putative GID1 genes (Shinozaki et al., 2018); one belongs to the type A group, and two to type B. We named them GID1a (Solyc01g098390), GID1b1 (Solyc09g074270), and GID1b2 (Solyc06g008870; Supplemental Figure 1; Supplemental Data Set). To test whether the putative GID1s function as GA receptors, we heterologously expressed them in the semi-dwarf Arabidopsis gid1a gid1c mutant (Griffiths et al., 2006). All three GID1s restored normal growth in the transgenic mutant (Figure 1A), suggesting that all are functional GA receptors. Following GA binding, GID1 interacts with DELLA to initiate GA responses (Hauvermale et al., 2012). We then tested the interaction of the tomato GID1s with the tomato DELLA protein PRO, in a yeast two-hybrid assay. In the presence of GA₃, all three GID1s interacted with PRO (Figure 1B).

Figure 1.

The Tomato GID1.

(A) Complementation of the Arabidopsis gid1a gid1c (gid1ac) mutant with the tomatoGID1 genes. The three GID1 genes were heterologously expressed under the regulation of the 35S promoter in the Arabidopsis gid1a gid1c mutant (in Col-0 background). Bar = 3 cm.

(B) The interaction between PRO and the three GID1s in the presence of 10⁻⁵M GA₃ in yeast two-hybrid assays. proΔ17 was used as a negative control.

(C) Two-month-old representative M82, stable DELLA (proΔ17)-overexpressing M82, GA-deficient mutant gib-2 and gid1 triple mutant (gid1^TRI). Bar = 5 cm.

(D) Sixteen-month-old gid1^TRI plant. Bar = 2 cm.

(E) The response of M82 and gid1^TRI to repeated application of GA₃ (100µM). Bar = 3 cm.

We next generated clustered regularly interspaced short palindromic repeats (CRISPR) /CRISPR associated protein 9 (Cas9)-derived gid1 mutants. The mutations were analyzed by PCR and sequenced (Supplemental Figures 2A and 2B), and two independent mutant lines were identified for GID1a, four for GID1b1, and three for GID1b2. Homozygous mutants were obtained for each line, and the Cas9 construct was segregated out by backcrossing to M82. One homozygous line for each gene was selected for further study: gid1a-1-4 had a 1-bp insertion, causing a frame-shift and premature stop-codon; gid1b1-12 had a 672-bp deletion; and gid1b2-4-2 had a 132-bp deletion (Supplemental Figure 2A). We then generated double and triple (gid1^TRI) mutants by crosses. Homozygous gid1^TRI seeds did not germinate even after scarification. For germination, the seed coat had to be removed, and the embryos rescued and placed on Murashige and Skoog (MS) medium. gid1^TRI plants exhibited very slow growth, extreme dwarfism, and very small and dark-green leaves (Figures 1C and 1D). Flowers of gid1^TRI were very small, did not open, their style did not elongate, and their carpels slowly degenerated (Figure 2E). As a result, gid1^TRI plants were sterile. Some old gid^TRI plants (more than 12 months old) produced extremely small parthenocarpic fruits. Although the lifespan of M82 plants is approximately five months, the first generated gid1^TRI plants were more than 16 months old (Figure 1D).

Figure 2.

Phenotypic Characterization of the gid1 Single and Double Mutants.

(A) Germination of M82 and gid1a gid1b2 seeds. Values (percentage of germinated seeds) are means of three replicates (plates) each containing 30 seeds ±se.

(B) Epicotyl length of control M82 and single and double gid1 mutants. Data are graphically presented as whisker and box plots. Statistical significance was tested with Student's t test, (n = 12, P < 0.05). Each set of letters above the columns represents significant differences.

(C) Representative seven-week-old single and double gid1 mutant plants. Bar = 3 cm.

(D) Representative single and double gid1 leaves (leaf number 4 from the apex down). Bar = 5 cm.

(E) Representative single and double gid1 flowers. Bar = 2 mm.

(F) and (G) Steady state level of GA20ox1 (F) and GA3ox1 (G) expression in single, double, and triple (gid1^TRI) gid1 mutants. Values (gene to ACTIN) are means of four biological replicates ±se. Each set of letters above the columns represents significant differences (Student's t test, P < 0.05).

In comparison with plants overexpressing the DELLA gain-of-function mutant gene pro∆17 (Nir et al., 2017) and the GA-deficient mutant gib-2 (Koornneef et al., 1990), gid1^TRI dwarfism was much stronger (Figure 1C). To test if gid1^TRI is fully insensitive to GA, we treated M82 and gid1^TRI repeatedly with 100 µM GA₃. The GA treatment had no effect on gid1^TRI development (Figure 1E). As expected, the steady state level of GA20ox1, whose expression is suppressed by GA as a result of the negative feedback regulation (Livne et al., 2015), was six times higher in gid1^TRI than in M82 (Supplemental Figure 3). Although GA application to M82 strongly suppressed GA20ox1 expression, it had no effect in gid1^TRI, further demonstrating the insensitivity of gid1^TRI to GA and suggesting that gid1^TRI is null in all three receptor genes.

High Redundancy in the Regulation of Plant Growth by GID1s

To study specific and overlapping activities of the different GID1s, we examined GA-regulated developmental responses in the single and the double mutants. All of the single mutants seeds and the double mutant gid1a gid1b1 and gid1b1 gid1b2 seeds germinated normally, similar to wild type M82 seeds (Supplemental Figure 4A). However, gid1a gid1b2 seeds showed reduced and delayed germination (Figure 2A). We then examined epicotyl elongation and leaf development in the different mutants. Whereas gid1b1 and gid1b2 plants did not show obvious phenotypic changes, gid1a stems were slightly shorter (Figures 2B and 2C). Epicotyl length of gid1b1 gid1b2 plants was similar to M82, but their leaves were smaller with normal leaflets (Figures 2B to 2D). gid1a gid1b1 plants were semi-dwarf and had long leaves but dark-green leaflets (Figures 2B to 2D; Supplemental Figure 4B). gid1a gid1b2 had the strongest phenotype; they were semi-dwarf with smaller and darker, green leaves (Figures 2B to 2D; Supplemental Figure 4B). Taken together, GID1a had the strongest effect on germination, stem elongation, and leaflet growth and color.

We previously showed that increased GA activity in pro delayed flowering (Livne et al., 2015). Flowering time (measured as leaves to first inflorescence) was slightly delayed in gid1a gid1b2 only (Supplemental Figure 4C). These suggest that both increased and decreased GA activity delay flowering time in tomato. Flower size was affected only in gid1b1, with an additive effect to gid1b2 in the double mutant gid1b1 gid1b2 (Figure 2E; Supplemental Figure 4D). No developmental defect in the flowers of the strongest double mutant gid1a gid1b2 were identified, suggesting that GID1b1 is the most prominent GID1 acting in GA-regulated flower organ growth. We also examined root development of all single and double mutants grown hydroponically. Primary-root length was reduced in gid1a, with an additive effect to gid1b2 in the double mutant gid1b1 gid1b2 (Supplemental Figures 4E and 4F).

Finally, the effect of the single, double, and triple mutants on the steady state expression of GA20ox1 and GA3ox1 was examined. The expression of GA20ox1 and GA3ox1 was upregulated in gid1^TRI only (Figures 2F and 2G), suggesting a high redundancy in the regulation of the feedback response by the three receptors.

GA-Dependent and Independent Interactions between GID1s and PRO

To examine whether the relative contributions of the different GID1s is determined by their expression levels, we analyzed their expression in various tissues in M82 plants. In imbibed seeds, roots, and flowers, GID1b1 exhibited the highest expression, whereas in elongating stems and young leaves, GID1a and GID1b1 showed similar expression levels (Figure 3A). GID1b2 exhibited the lowest expression in all tested tissues. The relatively low expression of GID1a in stems, seeds, and roots cannot explain its prominent role in stem and root elongation and germination. We therefore examined whether the different GID1s display different affinities to PRO. To this end, we tested the interaction between the three GID1s and PRO in yeast in the presence of different GA₃ concentrations. GID1a interacted with PRO in the absence of GA; addition of the hormone had no effect on the intensity of the interaction (Figure 3B). GID1b2 and GID1b1 interacted with PRO only when the GA₃ concentration exceeded 10⁻⁸ M and 10⁻⁷ M, respectively. These results suggest that GID1a has the highest and GID1b1 the lowest affinity for PRO.

Figure 3.

Molecular Analyses of GID1s.

(A) RT-qPCR analysis of GID1a, GID1b1, and GID1b2 expression in various tissues. Values (normalized to ACTIN) are means of four biological replicates ±se.

(B) The interaction between PRO and the three GID1s and the effect of GA₃ concentration (10⁻⁸–10⁻⁶ M) on this interaction in yeast two-hybrid assay. proΔ17 was used as a negative control.

(C) Seedlings of the three double mutant lines (each with one active receptor as indicated) were treated repeatedly, for 5 weeks, with 10 mg/l Pac, and then stem elongation was measured for 2 weeks. Values (cm) are means of 18 biological replicates ±se. Each set of letters above the columns represents significant differences (Student's t test, P < 0.05).

(D) Transgenic Arabidopsis gid1a gid1c mutant expressing the tomato GID1a, GID1b1, or GID1b2 and Col seedlings were treated repeatedly with 1 mg/l Pac for 3 weeks, and then rosette diameter was measured. Values (cm) are means of 10 biological replicates ±se. Each set of letters above the columns represents significant differences (Tukey–Kramer HSD, P < 0.05).

(E) GID1a-PRO interaction with SlSLY1 and the effect of GA₃ in yeast three-hybrid assay. SlSLY1 fused to GAL4 BD and PRO were expressed in yeast together with GID1a fused to GAL4 AD. The addition of Met to the growth medium (+Met) suppressed PRO expression. Cells were grown with or without 10⁻⁵ M GA₃. β-Galactosidase activity (the interaction of all three proteins) was visualized by X-Gal staining.

To determine whether the GA-independent interaction between GID1a and PRO occurs in planta and induces spontaneous GA responses, we followed stem elongation in double mutants treated with the GA biosynthesis inhibitor paclobutrazol (Pac). We reasoned that in each double mutant combination a single GID1 is active: for example, in gid1b1 gid1b2 only GID1a is active. Double mutant seedlings were treated repeatedly, three times a week, for 5 weeks, with 10 mg/l Pac before stem elongation was monitored for 2 weeks. All three double mutants showed strong suppression of stem elongation, and the inhibition of gid1b1 gid1b2 (active GID1a) growth was similar to that of gid1a gid1b2 (active GID1b1; Figure 3C). Because it was shown previously that GA-induced stem elongation in tomato is PRO dependent (Livne et al., 2015), these results suggest that GID1a, similar to GID1b1 and GID1b2, depends on GA to promote PRO degradation and stem elongation.

We also tested if GID1a promotes growth independently of GA when expressed in Arabidopsis. Arabidopsis Col and gid1a gid1c mutants expressing the tomato GID1a, GID1b1, or GID1b2 (Figure 1A) were treated repeatedly with 1mg/l Pac for 3 weeks, and then rosette diameter was measured. Pac had a similar inhibiting effect on the growth of all lines (Figure 3D). These results suggest that either the GA-independent interaction between GID1a and PRO does not occur in planta or that GID1a and PRO interact in planta but the GID1a-PRO complex cannot bind the F-box SLY1 without GA. Similar to Arabidopsis and rice, tomato has a single SLY1 (encoded by SlSLY1). We tested the interaction between GID1a, PRO, and SlSLY1 in yeast, in the presence or absence of GA₃ (10⁻⁵ M). A yeast three-hybrid assay showed that SlSLY1 interacts with GID1a-PRO independently of GA (Figure 3E), suggesting that GA-independent interaction between GID1a, PRO, and SLY occurs in yeast but probably not in planta.

The Unique Role of GID1a in the Regulation of Plant Growth in Response to High GA Levels

Because each double mutant has a single active receptor, we used them to study the contributions of each GID1 to GA-induced stem elongation. Seedlings of the double mutants were treated for 10 d with Pac and then with increasing concentrations of GA₃. All three double mutants responded similarly to GA₃ doses up to 1 µM (Figure 4A). However, although in gid1b1 gid1b2 (active GID1a) and M82 the elongation response increased further with higher GA₃ concentrations up to 100 µM, gid1a gid1b2 (active GID1b1) and gid1a gid1b1 (active GID1b2) exhibited a very mild elongation response to GA₃ concentrations above 1 µM (Figures 4A and 4B). In addition to stem elongation, other phenotypes were affected by the high GA₃ doses in gid1b1 gid1b2 but not in the other two double mutants, including leaf color and form (simpler leaves with smoother margins in M82 and gid1b1 gid1b2; Figure 4B). These results imply that GID1a is the GA receptor that most contributes to the response of tomato stem to high GA doses. It should be noted, however, that the high GA₃ doses used in this experiment are not physiologically relevant. To further explore this phenomenon, we examined shoot elongation responses in the single mutant plants to a single treatment with high GA₃ concentrations (100 µM). Although gid1b1 and gid1b2 exhibited a strong elongation response, similar to M82, GA₃-induced elongation was strongly suppressed in gid1a (Figure 4C; Supplemental Figure 5A); similar results were obtained with GA₄ (Supplemental Figure 5B). Following repeat treatments (every 3 d with 100µM GA₃), GA-induced elongation was strongly suppressed in gid1a, but not in gid1b1 or gid1b2 (Supplemental Figure 5C). Similar responses were observed with another gid1a allele (Supplemental Figure 5D).

Figure 4.

GID1a Is Responsible for Stem Elongation in Response to High GA Doses.

(A) Epicotyl elongation in response to GA₃ treatment. Seedlings of the double mutants with two leaves were treated for 10 d with Pac, followed by single treatment with different concentrations of GA₃ (0.5 µM–100 µM). Then 7 d later, stem elongation was measured. The active GID1 is indicated in parenthesis for each double mutant. Values are means of eight replicates ±se.

(B) Representative plants of M82 and the three double gid1 mutants treated for 10 d with Pac, followed by single treatment with 100 µM GA₃. Then 7 d later, pictures of the treated plants were taken. Bar = 4 cm.

(C) The effect of GA₃ treatment (100 µM) on stem elongation in the different single mutants. Representative seedlings of M82 and the three single gid1 mutants are presented. Bar = 2 cm.

(D) RT-qPCR analysis of GID1a, GID1b1, and GID1b2 expression in response to GA treatment. M82 seedlings were treated for 4 d with Pac and then with different concentrations of GA₃ (0.1 to 100 µM); and 3 h later, leaves were taken for the analysis of GID1a, GID1b1, and GID1b2 expression. Values (normalized to ACTIN) are means of four biological replicates ±se. The highest value for each gene was set to 1.

In Arabidopsis, GID1 expression is repressed by GA (Middleton et al., 2012), but the severity of the repression differs among the three GID1 genes (Griffiths et al., 2006). We tested if the three tomato GID1s differ in their feedback response to GA treatment. M82 seedlings were treated for 4 d with Pac and then with different concentrations of GA₃ (0.1 to 100 µM), and 3 h later, leaves were taken for the analysis of GID1a, GID1b1, and GID1b2 expression. GID1b1 and GID1b2 expression was reduced by 1 µM GA₃, and the inhibition effect increased with higher GA concentrations (Figure 4D). By contrast, GID1a expression was repressed only by 100 µM GA₃, and the level of inhibition was significantly lower: although GID1b1 and GID1b2 were inhibited by approximately 75%, GID1a was inhibited only by approximately 30% following the application of 100 µM GA₃. We also examined the expression of the active GID1 genes in each of the respective double mutants, following treatments with 100 µM GA₃, and found similar results (Supplemental Figure 5E). A similar feedback response was also found in elongating stems (Supplemental Figure 5F). These results suggest that the strong inhibition of GID1b1 and GID1b2 by high GA doses may reduce their contribution to GA-induced stem elongation.

gid1s Exhibit Phenotypic Instability under Semi-Controlled, Nonoptimal Growth Conditions

GA activity promotes transpiration due to increased stomatal conductance (Nir et al., 2017). To examine the contribution of each receptor to whole-plant transpiration, M82 and the double gid1 mutants were grown in a partially controlled greenhouse (natural light and day-length and temperature ranging from 22°C to 32°C) in pots on an array of load cells (lysimeters), which simultaneously followed plant weight change and provided information on biomass gain and transpiration (Nir et al., 2017). Whole-plant daily transpiration and weight gain in M82 were significantly higher than in all double mutants (Figure 5A; Supplemental Figure 6). However, the mutant lines exhibited relatively high phenotypic variability, with few gid1a gid1b2 plants showing much stronger dwarfism than when grown under optimal growth conditions (Figure 5B). We evaluated the degree of variability in plant weight, using the parameter of coefficient of variation (CV- sd/mean; Fisher et al., 2017). The variability in M82 plant weight was much lower than that of all double mutants (Figure 5C). Among the three receptors, GID1a had the strongest effect on phenotypic stability, and among the three double mutants, gid1b1 gid1b2 (with an active GID1a) exhibited the lowest variability. These results show that the stable phenotype of the gid1 mutants observed under optimal growth conditions is partially lost under less optimal growth conditions.

Figure 5.

gid1 Mutants Exhibit Phenotypic Instability Under Semi-Controlled, Nonoptimal Growth Conditions.

(A) Plants were placed on lysimeters in semi-controlled greenhouse. Pot (pot + soil + plant) weight was measured every 3 min. Values (daily transpiration in g) are means of thirteen plants ±se. Each set of letters above the columns represents significant differences (Student's t test, P < 0.05).

(B) Relative weight gain in M82 and gid1a gid1b2 mutants grown as in A during 6 d. Total weight, including plant, pot, and soil, was taken each pre-dawn, immediately after irrigation to saturation and drainage. The weight difference (delta) from previous pre-dawn measurement is the accumulated plant weight during a single day. Data are graphically presented as whisker and box plots.

(C) Coefficient of variation (CV - sd/mean) of plant weight taken at d 6 (see [B]).

The Loss of Redundancy under Changing Environments

We next tested the development of all single and double mutant gid1 lines under an ambient changing environment. When grown in the soil in a greenhouse (Supplemental Figure 7A), under ambient light and temperature conditions (temperature ranging from 20°C to 40°C), the different lines exhibited increased dwarfism compared with their counterparts grown under controlled optimal growth condition. More specifically, gid1a and gid1a gid1b2 exhibited approximately 50% and 65% reduction in stem length (compared with wild type M82) under ambient growth conditions and only 15% and 20% stem length reduction under optimal growth conditions, respectively (Figures 6A and 6B versus Figure 2B; Supplemental Figure 8A). Moreover, all mutant lines exhibited a high degree of phenotypic variability (Figures 6B and 6C; Supplemental Figure 7B), which was significantly higher than that of M82. Phenotypic variability (based on epicotyl length) of gid1a was 3-fold and that of gid1a gid1b2 was 4-fold higher than that of M82 (Figure 6C). PCR and sequencing were performed to confirm the genotype of individual plants showing strong and mild dwarfism (Supplemental Figure 9). Because gid1a gid1b2 exhibited the strongest dwarfism and the highest phenotypic instability (Figures 6B and 6C), it was grown next to M82 under optimal controlled growth conditions to compare their phenotypic stability. As expected gid1a gid1b2 was only semi-dwarf under these conditions (Supplemental Figure 8A) and exhibited low and similar phenotypic variability as compared with M82 plants (Figure 6D). We collected seeds from field-grown semi-dwarf and dwarf gid1a gid1b2 fruits (after selfing) and grew the plants under controlled optimal conditions. All plants were semi-dwarf and exhibited low phenotypic variability, similar to M82 plants (Supplemental Figures 8B and 8C).

Figure 6.

The High Redundancy Between GID1s Is Lost Under Changing Environments.

All gid1 single and double mutants were grown in the soil in a greenhouse under ambient light and temperature conditions.

(A) Epicotyl length. Data are graphically presented as whisker and box plots. Statistical significance was tested with Student's t test, (n = 12, P < 0.05). Each set of letters above the columns represents significant differences.

(B) Representative M82 and gid1a gid1b2 semi-dwarf (SD) and dwarf (D) plants grown under ambient conditions. Bar = 3 cm.

(C) Coefficient of variation (% CV) of epicotyl length of plants grown under ambient conditions. Values are means of four replicates (each the mean length of three plants) grown in randomized block design ± se. Percentages are presented above columns.

(D) CV of epicotyl length of M82 and gid1a gid1b2 plants grown under optimal growth conditions. Values are means of nine plants ± se. Percentages are presented above columns.

Under optimal growth conditions, the expression of the GA-biosynthesis genes GA20ox1 and GA3ox1 was affected only in the triple gid1^TRI and not in the single and double mutants (Figures 2F and 2G). To test if the high redundancy in the regulation of the feedback response is lost under a changing environment, GA20ox1 and GA3ox1 expression was then analyzed in field-grown M82 and semi-dwarf and dwarf gid1a gid1b2 plants. GA20ox1 and GA3ox1 expression was increased in gid1a gid1b2, and the levels of their expression were in line with the severity of dwarfism (Figures 7A and 7B). Previously, we showed that PRO promotes the expression of the abscisic acid-regulated gene RESPONSIVE TO ABSCISIC ACID18 (SlRAB18; Nir et al., 2017). We tested the effect of gid1a gid1b2 on the expression of SlRAB18 under the different growth conditions. Although the expression of SlRAB18 was similar in gid1a gid1b2 and M82 under controlled optimal growth conditions (Figure 7C), under ambient conditions it was much higher in the dwarf gid1a gid1b2 (Figure 7D), suggesting that under these conditions GA activity is reduced and PRO accumulates. Taken together, these results suggest that the dwarfism of gid1a gid1b2 in the field was caused by loss of redundancy in GID1 activity and the suppression of GA responses.

Figure 7.

Loss of Redundancy in GA Sensing Under Ambient Growth Conditions.

(A) and (B) RT-qPCR analysis of SlGA20ox1 (A) and SlGA3ox1 (B) expression in M82 and semi-dwarf (SD) and dwarf (D) gid1a gid1b2 grown under ambient growth conditions. Values (normalized to ACTIN) are means of four biological replicates ± se. Letters above the columns represent significant differences between respective treatments (Student’s t test, P < 0.05).

(C) and (D) RT-qPCR analysis of SlRAB18 and GID1b1 expression in M82 and gid1a gid1b2 plants (in fully expanded young leaves) grown under optimal growth conditions (C) or ambient conditions in the field (D). Values (normalized to ACTIN) are means of four biological replicates ± se. Letters above the columns represent significant differences between respective treatments (Student's t test, P < 0.05). Expression values for all analyzed genes in M82 were set to 1.

We then tested if the loss of redundancy and dwarfism of gid1a gid1b2 and gid1a gid1b1 was caused by reduced expression of the remaining active receptor. The expressions of GID1b1 in dwarf gid1a gid1b2 and of GID1b2 in dwarf gid1a gid1b1 were similar to their expression in M82 under controlled optimal growth conditions and in the field (Figures 7C and 7D; Supplemental Figures 8D to 8F). Moreover, we did not find correlations between dwarfism to changes in the expression levels of the remaining active receptor in the double mutants in the field (Supplemental Figures 8E and 8F); whereas GID1b1 expression was lower, that of GID1b2 was higher. Thus, reduced expression of the remaining active receptor is probably not the cause for the loss of redundancy under ambient growth conditions.

DISCUSSION

Most eudicot plants express multiple GA GID1 receptors (Yoshida et al., 2018), whereas some monocots express only one, raising questions regarding the possibility of GA responses driven by specific receptors and of evolutionary benefits of multiple receptors. Because tomato has three GID1s but only one DELLA, we used this system to study specific and overlapping roles of the different receptors, and the importance of multiple GA receptors to the overall GA activity and plant development under different growth conditions.

The three tomato GID1s were expressed in all examined tissues and exhibited overlapping activity. However, their expression levels did not correlate with their relative contribution to GA-regulated processes. Only in developing flowers was GID1b1 expression highest and its effect on flower organ growth the strongest. In all other tested developmental and physiological responses, GID1a played the dominant regulatory role. GID1a, however, did not show the highest expression, and in most tissues, it exhibited similar or lower levels of expression than the less active receptor, GID1b1. Yeast two-hybrid assays suggested that GID1a has the highest affinity to the DELLA protein PRO. It also showed that the affinity of GID1b2 to PRO was higher than that of GID1b1. Thus, the affinity of the three tomato GA receptors to DELLA may determine their relative contributions to most GA-regulated developmental processes.

GA application or complete loss of DELLA activity has a dramatic effect on stem elongation in tomato, suggesting that, in tomato, GA activity is not saturated under normal growth conditions (Livne et al., 2015). Our results suggest that only GID1a drives the strong response to high GA doses in tomato. The growth-promoting effects of activated GID1b1 and GID1b2 were rather weak, and the corresponding double mutants (gid1a gid1b1 and gid1a gid1b2) exhibited a very mild elongation response to exogenous GA treatments. Thus, in this specific response, the three receptors did not show overlapping activity. The unique role of GID1a may be the result of differences in the expression levels of the different receptors following GA application. GID1b1 and GID1b2 expression was strongly suppressed by GA, due to feedback inhibition (Middleton et al., 2012), whereas the expression of GID1a was much less affected. This, together with the higher affinity of GID1a to PRO, may underlie the strong response to the hormone and subsequent effect on stem elongation, as well as on other GA-related developmental responses, such as leaf color and form. Whether this unique activity of GID1a has a true biological role is not yet clear. GA levels increase in response to specific environmental conditions to induce rapid and strong stem elongation (e.g., shade-avoidance response; Yang and Li, 2017). GID1a may be the only receptor mediating these types of responses.

The type A GID1 (GID1a) exhibited GA-independent interaction with PRO in yeast. Similar results were shown recently by Shinozaki et al. (2018). In Arabidopsis, type B GID1 (GID1b) exhibits GA-independent interaction with DELLA (Griffiths et al., 2006), suggesting that this GID1 unique characteristic evolved independently in different species (Yoshida et al., 2018). Yamamoto et al. (2010) speculated that GA-independent interactions between specific GID1s and DELLAs are enabled by a unique conformation of GID1 in which the N-Ex domain is partially closed even in the absence of GA. Our yeast three-hybrid assay showed that the GID1a-PRO complex interacts with the F-box protein SlSLY without GA, suggesting that GID1a can induce PRO degradation and GA responses in the absence of GA. However, continuous Pac treatment (GA-limited conditions) had a similar dwarfing effect on all gid1 double mutants (each containing a different active GID1), suggesting that similar to GID1b1 and GID1b2, GID1a depends on GA for its activity. It is possible that the high affinity between GID1a and PRO, together with the high expression levels, allowed for spontaneous interaction in yeast even though they do not interact in planta without the hormone.

The three tomato GA receptors, GID1a, GID1b1, and GID1b2, exhibited extensive overlapping activity in the regulation of germination, growth, transition to flowering, flower development, and gene expression. Redundancy, caused by gene duplication, is very common in flowering plants (Veitia, 2005), rendering them resistant to mutation (Abley et al., 2016). Under optimal, controlled growth conditions, mutation in a single GID1 had almost no effect on most tested GA-regulated developmental processes. However, when grown in the field and exposed to ambient changing environment, the gid1 mutants showed phenotypic instability and loss of redundancy. Among the three receptors, the loss of GID1a had the strongest effect. This loss of redundancy in gid1a gid1b2 plants was evident by the strong dwarfism, the activation of the feedback response (upregulation of GA20ox1 and GA3ox1 expression), and the upregulation of the DELLA-induced gene SlRAB18. All these phenotypes were not found in gid1a gid1b2 under controlled, optimal growth conditions. Changes in the expression of the remaining active GID1s in the gid1 single and double mutants, under ambient conditions, are probably not the cause for the loss of redundancy and instability, because they did not correlate with plant phenotype.

The mechanism buffering phenotypes against environmental influences is not fully clear; several factors have been proposed, including redundancy (Abley et al., 2016). Our results support this hypothesis; under optimal growth conditions, the activity of a single receptor was sufficient to maintain rather normal growth, but under a changing environment all three receptors were needed for stable and normal development. GA biosynthesis is affected by environmental cues, such as light, temperature, water availability, and salinity (Yamaguchi, 2008; Colebrook et al., 2014; Wang et al., 2019). Thus, under ambient conditions, extreme changes in the environment may lead to strong fluctuations in GA levels. It is possible that the overlapping activity of the three tomato GID1 receptors is needed to buffer these changes in GA level and that this buffering effect allows stable and normal growth under a changing environment.

METHODS

Plant Materials, Growth Conditions, and Hormone Treatments

Tomato Solanum lycopersicum cv M82 (sp⁻/sp⁻) plants were used throughout this study. The CRISPR-Cas9 gid1 mutants were generated in the M82 background. The gib-2 mutant originally in ’Money maker’ was backcrossed three times to M82. All plants were grown in a growth room set to a photoperiod of 12/12-h night/days, light intensity (cool-white bulbs) of ∼250 μmol m⁻² s⁻¹, and 25°C. In other experiments, plants were grown in a greenhouse under natural day-length conditions, light intensity of 700 to 1000 µmol m⁻² s⁻¹ and 18-29°C. For growth under ambient conditions, plants were grown in the soil in a greenhouse under natural day-length conditions, with light intensity of ∼500-1200 µmol m⁻² s⁻¹ and 20-40°C. For root analysis, seedlings were grown hydroponically in Hoagland nutrient solution (pH 6.5), in a growth room, under the above-described conditions. Arabidopsis (Arabidopsis thaliana) gid1a gid1c mutant in the Columbia (Col-0) background and wild type Col-0 plants were used in this study. The plants were grown in a growth room, under a controlled temperature (22°C) and long (16-h light/8-h dark) day conditions.

GA₃, GA₄ (Sigma-Aldrich), and Pac were applied to plants by spraying.

Seed Germination Assays

After surface sterilization of the tomato seeds, ∼100 seeds were sown on 0.5× MS plates. Seeds were germinated in the dark at 23°C, for 3 d and then transferred to the light∼100 μmol m⁻² s⁻¹ at the same temperature. To assess seed germination, radicle emergence was scored each day after sowing.

Molecular Cloning and Arabidopsis Transformation

Solyc01g098390 (GID1a), Solyc09g074270 (GID1b1), and Solyc06g008870 (GID1b2) coding sequences in pENTR were inserted into the Gateway-compatible pGWB6 vector (Bensmihen et al., 2004). The pGWB6 vector drives expression of the recombined gene under control of the Cauliflower mosaic virus (CaMV) 35S promoter. The pGWB6 constructs were transferred into Agrobacterium tumefaciens GV3101 by electroporation and used to transform gid1ac Arabidopsis (Columbia) mutants by the floral dip method. T1 transgenic seeds were selected based on their resistance to hygromycin.

CRISPR/Cas9 Mutagenesis, Tomato Transformation, and Selection of Mutant Alleles

Two single-guide RNAs (sgRNAs; Supplemental Table 1) were designed for each gene, using the CRISPR-P tool (http://cbi.hzau.edu.cn/crispr). Vectors were assembled using the Golden Gate cloning system, as described by Weber et al. (2011). Final binary vectors, pAGM4723, were introduced into A. tumefaciens strain GV3101 by electroporation. The constructs were transferred into M82 cotyledons using transformation and regeneration methods described by McCormick (1991). Kanamycin-resistant T0 plants were grown, and independent transgenic lines were selected and self-pollinated to generate homozygous transgenic lines. The genomic DNA of each plant was extracted and genotyped by PCR for the presence of the Cas9 construct. The CRISPR/Cas9-positive lines were further genotyped for mutations using a forward primer to the upstream sequence of the sgRNA1 target and a reverse primer to the downstream of the sgRNA2 target sequence. The target genes in all mutant lines were sequenced. Several homozygous and heterozygous lines were identified, and at least two independent mutant lines for each SlGID1 gene were selected for further analysis. The Cas9 construct was segregated out by crosses with M82.

RNA Extraction and cDNA Synthesis

Total RNA was isolated from various tissues: seeds, roots, young leaves, elongating stems, flower bud, and all flower organs. Frozen tissues were ground and resuspended in guanidine HCl, and then phenol/chloroform was added. Samples were mixed by vortexing for 30 s, and after 30 min at 4°C the samples were centrifuged at 4°C for 45 min. Ethanol (100% v/v) and 1M acetic acid were then added, and the samples were mixed and stored overnight at −80°C. NaAc (3M) was added, and samples were washed with cold 70% (v/v) ethanol. cDNA was then synthesized using SuperScript II reverse transcriptase (18064014; Invitrogen) and 3 μg of total RNA, according to the manufacturer’s instructions. The RNeasy Micro Kit (QIAGEN) was used, according to manufacturer’s instructions, to extract RNA from roots, seeds, and flower organs.

RT-Quantitative PCR Analysis

RT-qPCR analysis was performed using an Absolute Blue qPCR SYBR Green ROX Mix (AB-4162/B) kit (Thermo Fisher Scientific). Reactions were performed using a Rotor-Gene 6000 cycler (Corbett Research). A standard curve was obtained using dilutions of the cDNA sample. The expression was quantified using Corbett Research Rotor-Gene software. Three independent technical repeats were performed for each sample. Relative expression was calculated by dividing the expression level of the examined gene by that of ACTIN. Gene to ACTIN ratios were then averaged. All primer sequences are presented in Supplemental Table 2.

Yeast Two- and Three-Hybrid Assays

GID1a, GID1b1, and GID1b2 coding regions were fused to GAL4 DNA binding domain (BD) in pBD-GAL4 (Clontech) by PCR amplification, with primers bearing EcoRI and SalI restriction sites. Following restriction digests, the SlGID1 products were ligated into pBD-GAL4 at the EcoRI and SalI sites. The coding sequence of PRO and pro∆17 were fused to the transcriptional activation domain (AD) in pACT (Clontech) at the BamHI and XhoI sites, which enables expression of proteins containing a GAL4 AD, by PCR amplification of fragments flanked with EcoRI and SalI restriction sites. All enzymes used were acquired from New England Biolabs. Plasmids were then transformed into Escherichia coli (DH5α) by heat shock, and the protein coding regions were sequence-verified. Each pBD GAL-GID1 was individually transformed into yeast (Saccharomyces cerevisiae) strain Y190 that contained the pACT vector sub-cloned with either PRO or pro∆17. Yeast transformants were selected for the presence of plasmids by growth on synthetic dextrose (SD) agar plates lacking Leu and Trp (LT), and examined for PRO interactions by using 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) staining to monitor β-galactosidase reporter gene expression levels. Individual clones were spotted onto glass, round SD–LT plates containing GA₃ (0 to 10⁻⁵ μM). After incubation at 28°C, for 2 d, colonies were chloroform lysed and stained to estimate β-galactosidase accumulation. The plate assay was repeated three times.

For yeast three-hybrid assay, PRO and SlSLY1 coding regions were cloned into pBridge (Clontech) at PflmI and NotI (PRO) and EcoRI and SalI (SlSLY1) sites. The cDNAs in pBridge allowed the expression of two proteins: SlSLY1 fused to GAL4 BD, and PRO. PRO expression was driven by a Met-suppressed promoter (suppressed by 2 mM Met). The coding sequence of GID1a was cloned into pACT (Clontech) at EcoRI and NcoI sites to express GID1a fused to the GAL4 AD. The pBridge (containing PRO and SlSLY1) and pACT (containing GID1a) were transformed into yeast strain Y190. Yeast transformants were selected for the presence of the plasmids on SD agar plates lacking LT, and examined for GID1a-PRO-SlSLY1 interaction using X-Gal staining to monitor β-galactosidase reporter gene expression levels. Individual clones were spotted onto SD (-LT) plates containing combinations of GA₃ (10⁻⁵ μM) and/or Met. In the presence of Met, PRO is not expressed. After incubation at 28°C, for 2 d, colonies were chloroform lysed and stained to visualize β-galactosidase accumulation.

Seed Scarification and Embryo Rescue

Seeds were placed on MS plates (without sugar). After 24-48 h, the seed coats were cut with a sharp knife under a binocular microscope, the outer layer of the seed was peeled off, and then a small scar was made at the root tip side of the seed coat. The scarified seeds were placed on MS plates. The full embryo rescue was achieved by peeling off the outer seed layer and then gently making a horizontal cut on the seed coat, peeling away the entire seed coat and exposing the embryo. The naked embryos were placed on MS plates for 2 to 3 d and then planted in the soil.

Soil Plant Analysis Development

The intensity of the green color (greenness) of the leaves was measured using a soil plant analysis development‐502 chlorophyll meter (Minolta Camera Co.).

Whole-Plant Transpiration Measurements

Whole-plant transpiration rates were determined using an array of lysimeters placed in the greenhouse (Plantarry 3.0 system; Plant-DiTech) in the “iCORE Center for Functional Phenotyping” (http://departments.agri.huji.ac.il/plantscience/icore.phpon), as described in detail by Halperin et al. (2017). Briefly, plants in 4L pots were grown under semi-controlled temperature conditions (20–32°C d and 18–24°C night), natural day-length, and light intensity of ∼1000 µmol m⁻² s⁻¹. Each pot was placed on a temperature-compensated load cell with digital output (Vishay Tedea-Huntleigh) and sealed to prevent evaporation from the surface of the growth medium. The weight output of the load cells was monitored every 3 min. Daily plant transpiration (weight loss between predawn and sunset) was calculated from the weight difference between the two data points.

Statistical Analysis

All data were analyzed with JMP Pro 14 software. One-way analysis of variance (compared with means), each pair Student’s t, or all pairs of Tukey’s Honestly Significant Difference test were used. Different letters represent differences at a significance level of P < 0.05.

Accession Numbers

Sequence data from this article can be found in the Sol Genomics Network (https://solgenomics.net/) under the following accession numbers: SlACTIN, Solyc11g005330; SlGA20ox1, Solyc03g006880.2.1; SlGA3ox1, Solyc03g119910; GID1a, Solyc01g098390; GID1b1, Solyc09g074270; GID1b2, Solyc06g008870; PRO, Solyc11g011260; SlRAB18, Solyc02g084850; SlSLY1, Solyc04g078390.

Supplemental Data

• Supplemental Figure 1. Molecular phylogenetic analysis of GID1s.

• Supplemental Figure 2. Sequence of gid1 mutant alleles.

• Supplemental Figure 3. gid1^TRI is resistant to GA.

• Supplemental Figure 4. Characterization of the gid1 mutants.

• Supplemental Figure 5. The response of the different gid1 mutants to GA treatments.

• Supplemental Figure 6. Relative weight gain in M82 and gid1 double mutants.

• Supplemental Figure 7. gid1 mutants showed increased dwarfism and phenotypic instability under changing environmental conditions.

• Supplemental Figure 8. GID1 expression and gid1 growth under controlled and ambient growth conditions.

• Supplemental Figure 9. Genotyping of gid1 mutants grown in the field under ambient conditions.

• Supplemental Table 1. RNA-guides used in this study.

• Supplemental Table 2. Primers used in this study.

• Supplemental Data Set. Text file of the alignment used for the phylogenetic analysis shown in Supplemental Figure 1.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• SLY1 Gramene: AT4G24210

• SLY1 Araport: AT4G24210

• gid1a Gramene: AT3G05120

• gid1a Araport: AT3G05120

• GID1c Gramene: AT5G27320

• GID1c Araport: AT5G27320

• GID1B Gramene: AT3G63010

• GID1B Araport: AT3G63010

• TRI Gramene: AT2G29330

• TRI Araport: AT2G29330

• seed AmiGo: PO:0009010

• GA20ox1 Gramene: Solyc03g006880

• GA20ox1 Araport: Solyc03g006880

• GA3ox1 Gramene: Solyc06g066820

• GA3ox1 Araport: Solyc06g066820

• PROCERA Gramene: Solyc11g011260

• PROCERA Araport: Solyc11g011260