Positive and negative regulation of seed germination by the Arabidopsis GA hormone receptors, GID1a, b, and c

Abstract Epistasis analysis of gid1 single and double mutants revealed that GID1c is a key positive regulator of seed germination, whereas the GID1b receptor can negatively regulate germination in dormant seeds and in the dark. The GID1 GA receptors were expected to positively regulate germination because the plant hormone gibberellin (GA) is required for seed germination in Arabidopsis thaliana. The three GA hormone receptors, GID1a, GID1b, and GID1c, positively regulate GA responses via GA/GID1‐stimulated destruction of DELLA (Asp‐Glu‐Leu‐Leu‐Ala) repressors of GA responses. The fact that the gid1abc triple mutant but not gid1 double mutants fail to germinate indicates that all three GA receptors can positively regulate non‐dormant seed germination in the light. It was known that the gid1abc triple mutant fails to lose dormancy through the dormancy breaking treatments of cold stratification (moist chilling of seeds) and dry after‐ripening (a period of dry storage). Previous work suggested that there may be some specialization of GID1 gene function during germination because GID1b mRNA expression was more highly induced by after‐ripening, whereas GID1a and GID1c mRNA levels were more highly induced by cold stratification. In light‐germinated dormant seeds, the gid1b mutation can partly rescue the germination efficiency of gid1a but not of gid1c seeds. Thus, GID1b can function as an upstream negative regulator GID1c, a positive regulator of dormant seed germination. Further experiments showed that GID1b can negatively regulate dark germination. Wild‐type Arabidopsis seeds do not germinate well in the dark. The gid1b and gid1ab double mutants germinated much more efficiently than wild type, gid1c, or gid1ac mutants in the dark. The observation that the gid1ab double mutant also shows increased dark germination suggests that GID1b, and to some extent GID1a, can act as upstream negative regulators of GID1c. Since the gid1abc triple mutant failed to germinate in the dark, it appears that GID1c is a key downstream positive regulator of dark germination. This genetic analysis indicates that the three GID1 receptors have partially specialized functions in GA signaling.

The fact that the gid1abc triple mutant but not gid1 double mutants fail to germinate indicates that all three GA receptors can positively regulate non-dormant seed germination in the light. It was known that the gid1abc triple mutant fails to lose dormancy through the dormancy breaking treatments of cold stratification (moist chilling of seeds) and dry after-ripening (a period of dry storage). Previous work suggested that there may be some specialization of GID1 gene function during germination because GID1b mRNA expression was more highly induced by after-ripening, whereas GID1a and GID1c mRNA levels were more highly induced by cold stratification. In light-germinated dormant seeds, the gid1b mutation can partly rescue the germination efficiency of gid1a but not of gid1c seeds. Thus, GID1b can function as an upstream negative regulator GID1c, a positive regulator of dormant seed germination. Further experiments showed that GID1b can negatively regulate dark germination. Wild-type Arabidopsis seeds do not germinate well in the dark. The gid1b and gid1ab double mutants germinated much more efficiently than wild type, gid1c, or gid1ac mutants in the dark. The observation that the gid1ab double mutant also shows increased dark germination suggests that GID1b, and to some extent GID1a, can act as upstream negative regulators of GID1c. Since the gid1abc triple mutant failed to germinate in the dark, it appears that GID1c is a key downstream positive regulator of dark germination. This genetic analysis indicates that the three GID1 receptors have partially specialized functions in GA signaling.

| INTRODUCTION
Plant embryos can survive for extended periods of time in a dry seed, in a state resembling suspended animation (reviewed in Bewley, Bradford, Hilhorst, & Nonogaki, 2013). This allows plants to be spread by wind or animal activity, and allows humans to store crop varieties for extended periods as metabolically quiescent dry seeds.
The seeds of many temperate plant species are dormant at maturity, meaning that they cannot germinate unless they experience dormancy-breaking conditions. In Arabidopsis, seed dormancy can be released by a period of dry storage called after-ripening or by cold stratification, taking up or imbibing water in the cold. Seed germination is a process beginning with water uptake and ending with the final germination event when the embryonic root penetrates the surrounding structures of the seed coat. Seed dormancy allows time for seed dispersal, prevents seeds from germinating out of season since a spring annual must experience an extended period of wintry cold stratification to germinate, and can help species to survive natural disasters as seeds stored in the soil.
It is well established that seed dormancy and germination are regulated by the opposing action of two plant hormones, abscisic acid (ABA) and gibberellin (GA) (reviewed by Finkelstein, Reeves, Ariizumi, & Steber, 2008). ABA is needed to establish dormancy during seed maturation, maintains dormancy in mature seeds, and can inhibit germination when externally applied to seeds. GA stimulates seed germination and is absolutely required for germination in some plant species like Arabidopsis and tomato. GA also stimulates stem elongation, the transition to flowering, and fertility. Arabidopsis seeds require light to germinate efficiently. Red light breaks Arabidopsis seed dormancy while far-red light promotes dormancy. GA and ABA also participate in light regulation of germination. Red light stimulates GA biosynthesis and inhibits ABA biosynthesis, whereas far-red light promotes GA turnover and stimulates ABA biosynthesis (reviewed by Yamaguchi, 2008). In Arabidopsis, the ga1 biosynthesis mutant completely fails to germinate unless GA hormone is exogenously applied or the seed coat is cut. Moreover, GA-insensitive signaling mutants like sly1 (sleepy1) cause increased seed dormancy, whereas GA hypersensitive mutants like spy (spindly) cause decreased seed dormancy.
GA stimulates seed germination and other developmental events by lifting DELLA protein repression of GA responses (reviewed by Hauvermale, Ariizumi, & Steber, 2012;Thomas, Blázquez, & Alabadi, 2016;Urbanova & Leubner-Metzger, 2016). DELLA proteins are named for a conserved amino acid sequence (Asp-Glu-Leu-Leu-Ala), and are a family of nuclear-localized transcriptional regulators that negatively regulate GA signaling and responses. The partially specialized roles of the five DELLA family members in Arabidopsis have been defined by characterizing how loss of function mutations in DELLA genes suppress ga1-3 phenotypes (Cheng et al., 2004;Dill, Jung, & Sun, 2001;King, Moritz, & Harberd, 2001;Lee et al., 2002;Tyler et al., 2004;Yu et al., 2004;Wild et al., 2012). For example, loss of the DELLA gene RGL2 rescued the germination of the GA biosynthesis mutant ga1-3 in the light, but complete rescue of ga1-3 dark germination required mutations in DELLA genes RGL2, RGA, and GAI. Thus, RGL2 is considered to be the main DELLA negatively regulating germination. RGA and GAI are the main DELLAs negatively regulating cell expansion, whereas RGA, RGL1, and RGL2 negatively regulate the transition to flowering. DELLA RGL3 interacts with jasmonic acid signaling to regulate plant defense responses.
GA lifts DELLA repression of seed germination and other GA responses by directing DELLA destruction via the ubiquitin-proteasome pathway (Ariizumi & Steber, 2007;McGinnis et al., 2003;Nelson & Steber, 2016;Tyler et al., 2004). The GA hormone signal is perceived by the GIBBERELLIN-INSENSITIVE DWARF1 (GID1) GA receptors (Nakajima et al., 2006;Ueguchi-Tanaka et al., 2005). In the absence of GA, DELLA proteins repress seed germination and other GA responses. GA binding to the GID1 receptor results in a conformational change in the receptor, enabling it to bind DELLA protein.
Formation of the GID1-GA-DELLA complex, causes the SLY1 F-box protein to bind the DELLA protein (Ariizumi, Lawrence, & Steber, 2011;Wang & Deng, 2011). SLY1 is the F-box subunit of an SCF E3 ubiquitin ligase that catalyzes DELLA polyubiquitination, causing DELLA to be degraded by the 26S proteasome. DELLA destruction allows seeds to germinate.
Since GID1 and SLY1 are both positive regulators of GA signaling, loss of function should result in increased seed dormancy/reduced germination. The sly1 mutants are GA-insensitive dwarves that exhibit increased seed dormancy (Ariizumi & Steber, 2007). The sly1-2 allele requires one to two years to after-ripen, whereas the stronger sly1-t2 fails to after-ripen. Rescue of sly1-2 germination by after-ripening and by overexpression of the GID1 receptor genes on the CaMV 35S promoter is associated not with decreased DELLA protein accumulation, but with increased formation of the GID1-DELLA protein complex (Ariizumi et al., 2013;Fukazawa, Ito, Kamiya, Yamaguchi, & Takahashi, 2015). It appears that GID1-binding to DELLA blocks DELLA interaction with downstream regulators of GA signaling.
The GID1a, GID1b, and GID1c genes appear to function redundantly as positive regulators of seed germination in the light (Griffiths et al., 2006;Iuchi et al., 2007;Voegele, Linkies, Mueller, & Leubner-Metzger, 2011;Willige et al., 2007). The fact that the gid1a gid1b gid1c triple mutant is unable to germinate unless the seed coat is cut, while the gid1 single mutants can germinate suggests that the three genes act redundantly to stimulate germination. Moreover, a positive role in seed germination is suggested by the fact that overexpression of each Arabidopsis GID1 gene increases sly1 mutant germination and increases the GA-sensitivity of ga1-3 during germination (Ariizumi et al., 2013;Hauvermale, Ariizumi, & Steber, 2014;Hauvermale, Tuttle, Takebayashi, Seo, & Steber, 2015). Interestingly, overexpression of the GID1b gene more strongly stimulated germination than overexpression of the GID1a or GID1c genes. This suggested that there might be some specialization in GID1 gene function in seed germination.
In order to examine whether there may be more specialization in GID1 gene function during seed germination, this study closely examined the relative effects of gid1 single and double mutants on initial Arabidopsis seed dormancy, dormancy loss, and germination.
We reasoned that conditions likely to inhibit germination would accentuate the germination phenotype of gid1 single and double mutants allowing us to better examine their relative effects on seed dormancy and germination. Voegele et al. (2011) provided a careful examination of the germination phenotypes of after-ripened gid1 mutants germinated in the light. Using the same gid1 alleles, this study examined the effect of gid1 single and double mutants on initial dormancy, after-ripening, dark germination, and GA sensitivity.
Examination of mutant phenotypes told a new story about GID1 regulation of germination. GID1b appeared to behave like a negative regulator of germination in dormant seeds germinated in the light and in dark-germinated seeds. GID1c is a strong positive regulator of germination, whereas the role of GID1a appears to depend upon dormancy and lighting conditions.

| Plant materials and growth conditions
The gid1a-1, gid1b-1, and gid1c-2 single mutants, double mutants, and segregating triple mutant (gid1b/gid1b gid1c/gid1c gid1a/+) in the Arabidopsis thaliana ecotype Columbia (Col-0) background were a gift from Claus Schwechheimer (Willige et al., 2007). Each of these three alleles contains a T-DNA insertion in the major second exon of the GID1 gene, predicted to result in a loss-of-function. All genotypes were confirmed by PCR using Taq (NEB) and the primers reported by Willige et al., 2007. Genomic DNA was prepared according to McKinney et al., 1995, andPCR was performed using the PCR program: 3 min at 95°C, 42 cycles of 1 min at 95°C, 50 s at the annealing temperature, 1 min at 72°C, then 10 min at 72°C.
To make sure that seeds used were exposed to the same environment during development and were of similar age, all gid1 mutants and the Columbia (Col-0) wild type (WT) were grown side-by-side in a Conviron growth chamber under a 16 hr day under fluorescent lamps at 200 μmol m −2 sec −1 at 22°C. The accession numbers for GID1a, GID1b, and GID1c are AT3G05120, AT3G63010, and AT5G27320, respectively.
In order to obtain high initial seed dormancy, seeds were harvested from plants near physiological maturity, when 40%-60% of the siliques were yellow but some were still green (as in Nelson, Ariizumi, & Steber, 2017). When such plants were hand threshed, only brown mature seeds, not green seeds, were obtained. Experiments were performed using two independent sets of seeds, and the second set of seeds had more initial dormancy than the first.
For dry after-ripening (AR), seeds were stored in open tubes at room temperature (~22°C) and 15%-20% humidity for 2 days (d), 1 week (wk), 2 weeks, or 4 weeks. Light germination experiments were plated immediately upon reaching the desired after-ripening time point. The dark germination plating experiments were performed using seed stored at −20°C for less than 2 months to maintain dormancy.

| Determining seed size
Seed size was determined using two independent sets of gid1 single and double mutant, and of Col WT growth side-by-side in two greenhouse experiments. For each set, seeds of each genotype were pooled. We stochastically selected three subsamples of 500 seeds from each set of the pooled samples. Seeds were counted under a dissection microscope (Leica), and then the mass of each subsample determined using a quantitative balance (Denver Instruments).

| Effects of gid1 single and double mutants on light germination
The effect of gid1 single and double mutants on dormancy and after-ripening were examined by comparing light germination to Col GE AND STEBER | 3 WT at 2 days, 1, 2, and 4 weeks of dry after-ripening. The gid1a and gid1c, but not gid1b, single mutants showed higher initial dormancy than WT (Figure 1a). Of the double mutants, gid1ac had the strongest seed dormancy, showing a complete failure to germinate through 2 weeks after-ripening ( Figure 1b). All mutants showed increasing germination with after-ripening (Figure 1a-h). The gid1ac double mutant reached 41% germination by 4 weeks after-ripening.
To determine if gid1ac seeds were viable, seeds coats were cut at 2 weeks-AR resulting in 86% germination within 5 days of imbibition. Since embryos are sometimes injured when the seed coat is cut, this suggests that the gid1ac seeds are dormant but viable. This is consistent with previous work showing that the gid1ac double mutant seeds are viable based on tetrazolium staining (Voegele et al., 2011). These results suggest that GID1a and GID1c are the main GA receptors serving as positive regulators of germination in the light. Previous work showed that the gid1a gid1b gid1c triple mutant fails to germinate and does not after-ripen (Willige et al., 2007). However, no single or double gid1 mutant is sufficient to prevent after-ripening, suggesting that the three GID1 genes act additively to positively regulate seed germination and dormancy loss.
Another story about the relative roles of GID1a, GID1b, and The gid1bc double mutant germination (52.3%) was similar to gid1c (56.8%), but lower than gid1b germination (88.5%) at 4 days incubation ( Figure 2c). Thus, gid1c is epistatic to gid1b. In contrast, the gid1ab double mutant had a germination phenotype that was intermediate (72.1%) between gid1a (33.9%) and gid1b (88.5%) at 4 days incubation ( Figure 2b). Thus, the gid1b mutation appears to partly rescue gid1a germination, suggesting that GID1b acts as a negative regulator of germination in dormant seeds. The observation that gid1b and gid1a have additive effects suggests that they act in parallel. The fact that gid1b can partly rescue gid1a but not gid1c germination suggests that GID1b functions upstream of GID1c as a negative regulator of seed germination. The gid1ab germination appears depend on stimulation of germination by GID1c.  incubation for 14 days at 22°C. While similar trends were observed at 2 and 4 weeks of after-ripening, the 4 weeks after-ripening data are described in more detail because it is easier to observe significant differences when the germination rates are higher (Supporting Information Figure S2AB; Supporting Information Table S1; Supporting Information Table S2). The effects of gid1 mutants on dark germination varied with light intensity during plating. Generally, higher fluorescent light intensity during plating stimulated dark germination in WT and in all gid1 mutants, and green light resulted in the least efficient seed germination (Supporting Information Figure S3).

|
Many of the effects of the gid1 single and double mutants were observed regardless of lighting during plating, although some effects were clearer under one or more condition. Of the single mutations, only gid1c showed a significant decrease in dark germination compared to WT at 4 weeks AR and only after plating under bright lighting ( Figure 4; Supporting Information Figure S2). Surprisingly, gid1b consistently germinated more efficiently than Col WT in the dark.
The gid1a mutation did not show a significant decrease in germination compared to Col WT, and actually germinated more efficiently An epistasis analysis of dark germination was made by comparing double mutants to single mutant parents (Figure 4). The experiment was repeated using a second batch of seeds that had higher initial dormancy (Supporting Information Figure S4). The gid1 mutants showed similar effects and epistasis relationships in the two experiments. The gid1ac double mutant failed to germinate in the dark and, therefore, germinated less efficiently than either gid1a or gid1c mutants (Figure 4; Supporting Information Figure S4). This additive effect suggests that GID1c and GID1a act in parallel as positive regulators of dark germination. Interestingly, the gid1ab double mutant germinated more efficiently than either the gid1a or gid1b single mutants when plated under green or dim lighting conditions, suggesting that GID1a can also act as a negative regulator of dark germination. Because the second set of seeds had more initial dormancy, it was more apparent that the gid1a mutation mildly stimulated germination, and that the gid1ab double mutant germinated more efficiently than the gid1b single mutant. Thus, GID1a is a weak negative regulator of dark germination in the presence of GID1c. Thus, GID1c is a strong positive regulator of seed germination that appears to be negatively regulated by GID1b and GID1a. The role of GID1a appears to be context dependent, acting as a positive regulator when GID1b is present and acting as a negative regulator when GID1c is present.
These results strongly suggest that the Arabidopsis GA receptors have specialized roles in dark seed germination. The gid1bc double mutant germinated more efficiently than gid1c but less efficiently than gid1b. This suggests that GID1c and GID1b act in parallel to positively and negatively regulate dark germination, respectively. If GID1b and GID1c do act in parallel, then we might expect the gid1abc triple to germinate in the dark in a manner similar to gid1ab.
Previous work showed that the gid1abc triple mutant fails to germinate in the light unless germination is rescued by cutting the seed coat (Willige et al., 2007). However, the effect of the gid1abc triple mutation on dark germination is unknown. Since the gid1abc triple homozygous mutant is infertile, seeds of the gid1 b/b c/c a/+ line that segregates for the gid1abc triple homozygous mutant were plated to compare dark and light germination phenotypes ( Figure 5).

| The effect of gid1 single and double mutants on seed size
While performing germination experiments, we observed some morphological differences in the gid1 mutant seeds themselves. For example, the gid1ab seeds appeared to be larger than Col WT, F I G U R E 4 Epistasis analysis of the gid1 dark germination phenotype. Compared in order are dark germination of: (a) gid1a and gid1c to gid1ac, (b) gid1a and gid1b to gid1ab, and (c) gid1a and gid1c to gid1ac. Seeds were plated at 4 weeks AR. The x-axis labels "Bright", "Dim", "Dimmer" indicate the fluorescent light intensity or quality during plating prior to incubation for 14 days in the dark at 22°C. Shown is the mean ±SE (n = 3) whereas the gid1ac double mutant seeds appeared somewhat shriveled (Figure 6a,b). To examine whether there was a quantitative difference in seed size, we measured the dry weight of 500 seeds of Col WT, the gid1 single, and gid1 double mutants in the two sets of seeds harvested at different times (Figure 6c,d). The gid1ab double mutant seeds were significantly larger and the gid1ac double mutant, significantly smaller than WT seeds in both experiments (p < 0.001).
Thus, GID1 genes appear to regulate Arabidopsis seed size, suggesting that they play a role during seed development as well as during seed germination.

| DISCUSSION
GA hormone stimulates germination in many plant species, and it is known that GA hormone biosynthesis is required for germination in Arabidopsis and tomato (reviewed in Finkelstein et al., 2008).  (Figure 7).
The gid1ab and gid1ac double mutants also reproducibly impacted seed size and morphology. This is interesting given that previous work suggested a role for GID1 in control of ovule development (Ferreira, et al., 2008). The gid1ab double mutant produced larger seeds than WT ( Figure 6). Larger seed size is sometimes associated with higher germination efficiency, but it is more often associated with better seedling emergence and vigor, presumably due to the presence of more stored reserves (Chaisurisri, Edwards, & El-Kassaby, 1992;Edwards & Hartwig, 1971;Gómez, 2004;Leishman, Wright, Moles, & Westoby, 2000;Wu & Du, 2007). Increased seed size alone is not sufficient to explain the increased dark germination, since the gid1a and gid1b single mutants did not consistently increase seed size but did increase in dark germination compared to WT (Figure 4; Figure 6). It is possible, however, that the larger The gid1ac embryos appeared to have developed to a normal shape and size when the seed coat was cut to stimulate germination.
Future work will need to examine if gid1ac causes altered endosperm development.
Previous studies of loss-of-function alleles indicated that GID1a, GID1b, and GID1c act additively as positive regulators of seed germination in the light, but suggested that one GID1 gene or another may play a more prominent role (Iuchi et al., 2007;Voegele et al., 2011;Willige et al., 2007). This study examined the same gid1 mutant alleles used in both the Willige et al., 2007 andVoegele et al., 2011 studies. In the Voegele et al., 2011 study, the after-ripened gid1c single mutant had a slightly lower, and gid1b a slightly higher, germination rate than Col WT. But those differences were not significant. Similarly, we found that none of the gid1 single mutants had a statistically significant effect on germination when after-ripened for 4 weeks (Figure 1). In seeds showing mild dormancy (2 day afterripened), however, the gid1b mutation was able to improve gid1a germination efficiency, but not gid1c germination in the light (Figure 2). Thus, loss of GID1b as a negative regulator of dormant seed germination can stimulate germination when GID1c is present, suggesting that GID1b may sometimes function as a negative regulator of the downstream positive regulator GID1c (Figure 7). The notion that GID1b can negatively regulate germination is supported by the fact that the gid1b and the gid1ab double mutants showed a strong increase in dark seed germination compared to Col WT (Figure 4; Supporting Information Figure S4). The gid1a mutation only showed increased dark germination compared to WT under green light or in more dormant seeds. This suggests that GID1b, and to a lesser extent GID1a, can act as negative regulators of dark seed germination. GID1c appears to be a key positive regulator of seed germination. This is interesting given that of the gid1 single mutants, gid1c had the strongest decrease in GA dose-response (Figure 3).
These results were the reverse of what we initially expected based on overexpression experiments (Ariizumi et al., 2013;Hauvermale et al., 2014Hauvermale et al., , 2015. Other lines of evidence raised the possibility that the Arabidopsis GA receptors have evolved specialized functions. Based on predicted amino acid sequence alignment, Arabidopsis GID1a more closely resembles GID1c and the sole copy of GID1 in rice (Oryza sativa, OsGID1) than GID1b (Nakajima et al., 2006;Yamamoto et al., 2010).
GID1a has 85% amino acid identity with GID1c, but only 66% amino acid identity with GID1b (Nelson & Steber, 2016). This difference in amino acid similarity is associated with functional differences in DELLA and GA-binding activities (Nakajima et al., 2006;Yamamoto et al., 2010). Rice GID1 and GID1a/c-type receptors absolutely require GA in order to bind DELLA protein, whereas GID1b-type receptors have a limited ability to bind DELLA without GA that is and GID1c. The three GA receptors also differ in their affinity for GA hormone. GA associated with GDI1b more quickly than GID1a and GID1c, while GA dissociated from GID1a more slowly than from GID1c and GID1b. It is possible that the stronger rescue of sly1-2 germination by GID1b-OE is due to either to a lower GA requirement or to higher DELLA protein affinity. The GID1 transcripts also show differences in mRNA expression during germination (Hauvermale et al., 2015). GID1a and c transcripts are more strongly induced by cold stratification than GID1b, whereas GID1b transcript showed higher relative expression levels in the dark. These expression differences initially prompted us to more closely examine the relative function of GID1 genes during germination. By analogy to SA signaling, different GID1 proteins may allow germination to respond to different environmental signals, possibly due to differences in GA hormone levels and/or receptor affinities