Karrikin Signaling Acts Parallel to and Additively with Strigolactone Signaling to Regulate Rice Mesocotyl Elongation in Darkness

Abstract

Seedling emergence in monocots depends mainly on mesocotyl elongation, requiring coordination between developmental signals and environmental stimuli. Strigolactones (SLs) and karrikins are butenolide compounds that regulate various developmental processes; both are able to negatively regulate rice (Oryza sativa) mesocotyl elongation in the dark. Here, we report that a karrikin signaling complex, DWARF14-LIKE (D14L)-DWARF3 (D3)-O. sativa SUPPRESSOR OF MAX2 1 (OsSMAX1) mediates the regulation of rice mesocotyl elongation in the dark. We demonstrate that D14L recognizes the karrikin signal and recruits the SCF^D3 ubiquitin ligase for the ubiquitination and degradation of OsSMAX1, mirroring the SL-induced and D14- and D3-dependent ubiquitination and degradation of D53. Overexpression of OsSMAX1 promoted mesocotyl elongation in the dark, whereas knockout of OsSMAX1 suppressed the elongated-mesocotyl phenotypes of d14l and d3. OsSMAX1 localizes to the nucleus and interacts with TOPLESS-RELATED PROTEINs, regulating downstream gene expression. Moreover, we showed that the GR24 enantiomers GR24^5DS and GR24^ent-5DS specifically inhibit mesocotyl elongation and regulate downstream gene expression in a D14- and D14L-dependent manner, respectively. Our work revealed that karrikin and SL signaling play parallel and additive roles in modulating downstream gene expression and negatively regulating mesocotyl elongation in the dark.

INTRODUCTION

Seed germination and seedling establishment require the coordination of developmental programs in response to various environmental signals (Gommers and Monte, 2018). Epigeal seedlings of dicotyledonous plants growing in soil or in the dark have elongated hypocotyls that allow the seedlings to emerge from the soil to reach the light (de Wit et al., 2016). By contrast, monocot seedling emergence depends on mesocotyl elongation, an important agronomic trait of grain crop species (de Wit et al., 2016). Mesocotyl elongation reflects the coordination of cell elongation and cell division processes that are regulated precisely by phytohormones that integrate developmental signals and environmental stimuli (Sawers et al., 2002). The phytohormones that promote rice (Oryza sativa) mesocotyl elongation include auxin (Epel et al., 1992), brassinosteroids (BRs; Yamamuro et al., 2000), cytokinins (Hu et al., 2014), ethylene (Xiong et al., 2017), and gibberellins (Takahashi, 1972), whereas abscisic acid (Takahashi, 1972), jasmonic acid (JA; Riemann et al., 2003; Xiong et al., 2017), and strigolactones (SLs; Hu et al., 2010) inhibit mesocotyl elongation. The interaction of SLs with cytokinins (Hu et al., 2014) and BRs in the regulation of dark-induced rice mesocotyl elongation (Sun et al., 2018) suggests that mesocotyl development is an excellent model for investigating the crosstalk between SLs and other phytohormones.

SLs are butenolide compounds that are involved in various developmental processes and environmental responses, including parasitic weed germination stimulation, arbuscular mycorrhizal (AM) fungi hyphal branching, shoot branching inhibition, root system architecture formation, secondary growth of the cambium, leaf senescence, and drought tolerance (Al-Babili and Bouwmeester, 2015). Deficiency in SL biosynthesis or signaling results in tiller bud outgrowth in rice (Ishikawa et al., 2005), increased tiller angles (Sang et al., 2014), enhanced sensitivity to senescence (Yamada et al., 2014), and mesocotyl elongation in the dark (Hu et al., 2010). SLs are derived from β-carotene, and the activities of the following catalytic enzymes lead to carlactone production: carotene isomerase DWARF27 (D27; Lin et al., 2009; Alder et al., 2012), carotenoid cleavage dioxygenase HTD1/D17 (CCD7; Zou et al., 2006), and CCD8 (Arite et al., 2007). Carlactone is further converted by the MORE AXILLARY GROWTH1 (MAX1; Lazar and Goodman, 2006; Abe et al., 2014; Cardoso et al., 2014; de Saint Germain et al., 2016; Yoneyama et al., 2018; Zhang et al., 2018; Wakabayashi et al., 2019) and the LATERAL BRANCHING OXIDOREDUCTASE (LBO) enzymes (Brewer et al., 2016). The SL signaling cascade is initiated when the receptor DWARF14 (D14), an α/β hydrolase fold protein (Arite et al., 2009; de Saint Germain et al., 2016; Yao et al., 2016; Shabek et al., 2018; Seto et al., 2019), covalently binds and hydrolyzes SL (de Saint Germain et al., 2016; Yao et al., 2016), enhancing the interactions of D14 with the F-box protein DWARF3 (D3; Yan et al., 2007; Hamiaux et al., 2012; Jiang et al., 2013; Zhou et al., 2013) and the Clp ATPase protein DWARF53 (D53; Jiang et al., 2013; Zhou et al., 2013). In the presence of SL, the D53 protein is ubiquitinated by SCF^D3 for degradation via the 26S proteasome pathway, relieving the TOPLESS-RELATED PROTEIN (TPR) transcriptional corepressors that mediate repression of D53 downstream targets (Jiang et al., 2013; Zhou et al., 2013).

SL signaling was revealed to be more complex when MAX2 was discovered to participate in both SL and karrikin pathways. Karrikins are butenolide chemical regulators present in the smoke of burned plant material, and their molecular structure is similar to that of the D-ring of SLs (Scaffidi et al., 2013). Karrikins can stimulate seed germination (Flematti et al., 2004) and are involved in regulating root skewing (Swarbreck et al., 2019) and root hair development (Villaécija-Aguilar et al., 2019) in Arabidopsis. Genetic screening of Arabidopsis karrikin-insensitive (kai) mutants revealed that loss of function of MAX2/KAI1 (Nelson et al., 2011) led to an elongated-hypocotyl phenotype in the light. HTL/KAI2 was initially identified by screening light-responsive mutants (Sun and Ni, 2011; Waters et al., 2012) and later was found to be insensitive to karrikin (Nelson et al., 2011; Waters et al., 2012). Compared with the wild type, both max2 and kai2 are hypersensitive to drought stress (Bu et al., 2014; Ha et al., 2014; Li et al., 2017), indicating that karrikin plays roles in drought adaptation. The phenotypes of Atd14 and kai2 single-mutant seedlings somewhat resemble the phenotype of max2 seedlings (Waters et al., 2012; Scaffidi et al., 2013), and the Atd14 kai2 double mutant phenotypically mimics max2 (Waters et al., 2012). Neither the hypocotyl-elongation phenotype nor the shoot-branching phenotype of Atd14 kai2 and max2 respond to karrikin or rac-GR24 (Waters et al., 2012; Scaffidi et al., 2013). Loss of function of SMAX1 and SMXL2 can restore most aspects of the max2 seedling phenotype, but not shoot-branching defects of max2 (Stanga et al., 2013, 2016; Soundappan et al., 2015), while the max2 shoot-branching phenotype is rescued by the smxl6 smxl7 smxl 8 triple mutation (Soundappan et al., 2015; Wang et al., 2015; Liang et al., 2016). Recently, SMAX1 and SMXL2 were shown to suppress the root-skewing and root-hair phenotypes of max2 (Swarbreck et al., 2019; Villaécija-Aguilar et al., 2019). The karrikin signaling and SL signaling pathways act in parallel to regulate MAX2 activity in a ligand-dependent manner (Soundappan et al., 2015; Wang et al., 2015). The commonly used synthetic SL analog rac-GR24 is a mixture of two enantiomers, GR24^5DS and GR24^ent-5DS (Scaffidi et al., 2014), and the effects of GR24^5DS and GR24^ent-5DS on the inhibition of hypocotyl elongation and the regulation of gene expression seem to be mediated by AtD14 and KAI2, respectively (Scaffidi et al., 2014). Collectively, these genetic data suggest that the SL and karrikin signaling pathways act in a similar manner: they recruit different members of the same receptor family (α/β fold hydrolase) that recognize different ligands to trigger the SCF^MAX2 complex to target different members of the same protein family for degradation (Morffy et al., 2016).

Under phosphorus-limiting conditions, increased SL biosynthesis and exudation from roots promote AM symbiosis with plants (Akiyama et al., 2005). In rice, D3, but not D14, is required for root colonization by AM fungi (Yoshida et al., 2012). The karrikin signaling component DWARF14-LIKE (D14L) is required for colonization of rice roots by AM fungi (Gutjahr et al., 2015b). The D14L-mediated pathway may act in parallel with the CERK1-dependent pathway (Gutjahr et al., 2015b; Chiu et al., 2018). Moreover, the mesocotyls of the d3 mutant are longer than those of other SL pathway mutants, including d10 and d14 (Hu et al., 2010), and loss of function of D14L increases mesocotyl length in the dark (Gutjahr et al., 2015b; Kameoka and Kyozuka, 2015). These findings suggest the existence of a D14L-D3–dependent (but D14-independent) karrikin signal cascade in rice.

Here, we report that mesocotyl elongation in the dark is regulated by the D14L-D3-O. sativa SUPPRESSOR OF MAX2 1 (OsSMAX1) module in rice. OsSMAX1 interacted with TPR transcriptional corepressors in an Ethylene-responsive element binding factor-associated amphiphilic repression (EAR) motif-dependent manner and regulated the expression of downstream target genes. D14L and D3 were required for the karrikin signal-induced degradation of OsSMAX1, which is necessary for inhibition of mesocotyl elongation in the dark, but not the regulation of shoot branching. We further revealed that D14L and D14 are required for the recognition of the stereospecific enantiomers of GR24 and for the recruitment of SCF^D3 for ubiquitination and degradation of substrate proteins, respectively. Our work demonstrates that the parallel and additive actions of SL and karrikin signaling in the regulation of mesocotyl elongation in the dark largely depend on their convergence in the regulation of the expression of common downstream genes.

RESULTS

D14L Acts Parallel to and Additively with D14 to Regulate Rice Mesocotyl Elongation in the Dark

Both the karrikin and SL signaling pathways are involved in the regulation of mesocotyl elongation in the dark, which requires the function of D14L and D14, respectively (Hu et al., 2010; Gutjahr et al., 2015b; Kameoka and Kyozuka, 2015). A deficiency in SL biosynthesis or signaling leads to the accumulation of D53 and results in tiller bud outgrowth (Jiang et al., 2013; Zhou et al., 2013). As a gain-of-function SL-insensitive mutant, the shoot-branching phenotype of d53 is similar to that of the loss-of-function SL biosynthesis mutants d27, d17, and d10 and the loss-of-function SL signaling mutants d14 and d3 (Jiang et al., 2013; Zhou et al., 2013). Compared with the wild type, both the SL-deficient mutants (d27, d17, and d10) and the SL-insensitive mutants (d14 and d3) have longer mesocotyls (Hu et al., 2010). To investigate the role of D53 in the inhibition of mesocotyl elongation in the dark, we measured the mesocotyl length of d53, which was as long as that of d14 (Figures 1A and 1B). ACT:D53m-GFP transgenic seedlings, which constitutively express d53 and GFP fusion proteins, exhibited a shoot-branching phenotype similar to that of d53 (Supplemental Figures 1A and 1B; Jiang et al., 2013). The mesocotyl length of the ACT:D53m-GFP transgenic seedlings was also similar to that of d14 (Supplemental Figures 1C and 1D). These results suggested that D53 accumulation in these seedlings promotes mesocotyl elongation in the dark and confirmed that D14- and D3-dependent D53 degradation is involved in the inhibition of rice mesocotyl elongation in the dark. However, the d3 mesocotyl length was longer than that of other SL mutants, including d14 and d53, strongly suggesting that dark-induced mesocotyl elongation in rice is also regulated by an SL-independent pathway. D14L, D14L2, D14L3, and D14L4 are homologs of D14 (Supplemental Figure 2; Supplemental Data Set 1; Arite et al., 2009; Walker et al., 2019). Moreover, D14L is homologous to HTL/KAI2 and is involved in regulating rice mesocotyl elongation in the dark (Gutjahr et al., 2015b; Kameoka and Kyozuka, 2015). D14L2 and D14L3 are similar to DLK2, which can be induced by karrikin treatment (Nelson et al., 2011), and enhance hypocotyl elongation (Végh et al., 2017).

Figure 1.

D14L Acts in Parallel and Additively with D14 in Mesocotyl Inhibition in the Dark.

(A) Mesocotyls of seedlings grown for 7 d in darkness; the arrowheads indicate the boundaries of the mesocotyl. Bar = 5 mm. WT, wild type.

(B) Lengths of mesocotyls of dark-grown seedlings. Data are presented as means ± sds, and the numbers above the columns indicate the sample sizes. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5). WT, wild type.

(C) and (D) Relative expression levels of the mesocotyl-related genes OsTCP5  (C) and GY1  (D) detected by RT-qPCR in seedlings of mutants compared to the wild type (WT).

(E) Venn diagram of genes for which expression was upregulated in the mutants compared to the wild type.

(F) Venn diagram of genes for which expression was downregulated in the mutants compared to the wild type.

(G) Pathways identified by enrichment analysis of genes with upregulated expression in all indicated mutants compared with the wild type.

(H) to (J) Relative expression of D14L  (H), D14L2  (I), and D14L3  (J) in seedlings of mutants compared to the wild type (WT). In (C), (D), and (H–J), the expression of each indicated gene is relative to that of ACTIN as the internal reference; data shown are from one of three replicate experiments. Data are presented as means ± ses (n = 3). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Tukey's test (P < 0.05; Supplemental Data Set 5).

To investigate how karrikin regulates rice mesocotyl development in the dark, we generated D14L knockout plants using clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated 9 (Cas9)–mediated genome editing as previously described (Song et al., 2017). The d14l-1 mutant contained a 1-bp deletion at site 493 of D14L that shifted the reading frame, the d14l-2 mutant contained a 3-bp in-frame deletion at site 493 to 495, and the d14l-3 mutant had a G-to-T substitution at site 495 and a 7-bp deletion at site 496 to 502, causing a reading frame shift in D14L (Supplemental Figure 3). The d14l-1 line was used for further analysis and is referred to as d14l henceforth (Supplemental Figure 3). Under light conditions, mesocotyl elongation was not observed in d14, d14l, d3, d14 d14l, or wild-type seedlings (Supplemental Figure 4). In the dark, the mesocotyl length of d14l was similar to that of d14, which was longer than that of the wild type but shorter than that of d3 (Figures 1A and 1B). The d14l mesocotyl phenotype in this study is consistent with the phenotypes reported previously (Gutjahr et al., 2015b; Kameoka and Kyozuka, 2015). However, unlike the D14L RNAi d14 seedlings in a previous study (Kameoka and Kyozuka, 2015), the mesocotyl length of the d14 d14l double mutant was as long as that of d3 (Figures 1A and 1B). In addition, the mesocotyl of the d53 d14l double mutant was longer than that of d14l and similar to that of d3 (Supplemental Figures 1E and 1F). These results suggested that the D14L-mediated pathway acts additively with the D14-mediated pathway in the inhibition of rice mesocotyl elongation in the dark. SL negatively regulates the expression of OsTCP5 during mesocotyl development (Hu et al., 2014). Deficiency in either D14 or D14L resulted in the downregulation of OsTCP5 (Figure 1C). GY1 is involved in JA biosynthesis and negatively regulates mesocotyl elongation (Xiong et al., 2017). Deficiency in either D14 or D14L resulted in significant downregulation of GY1 expression (Figure 1D). The downregulation of OsTCP5 and GY1 expression in d14 d14l and d3 was similar to that in d14 and d14l (Figures 1C and 1D), indicating that the D14- and D14L-mediated pathways potentially regulate common target genes to modulate mesocotyl elongation in the dark.

Next, we compared the gene expression profiles of wild-type, d14, d14l, d3, and d14 d14l seedlings grown in the dark. More than two-thirds of the differentially expressed genes (DEGs) between the wild type and d3 overlapped with the DEGs between the wild type and d14 d14l. Interestingly, more than one-third of the DEGs between the wild type and d14 overlapped with the DEGs between the wild type and d14l (Figures 1E and 1F). Gene Ontology analysis of these DEGs revealed a significant enrichment of genes associated with the responses to abiotic stress and light stimuli, which are likely subject to regulation by both the D14- and D14L-mediated pathways (Figure 1G). Loss of function of either D14 or D14L resulted in a decrease in the expression of D14L (Figure 1H). The expression of D14L2 has been suggested to be dependent on D14L (Gutjahr et al., 2015b). Indeed, disrupting either D14 or D14L function led to a dramatic decrease in D14L2 expression (Figure 1I). Moreover, the expression of D14L3 also depended on the function of both D14 and D14L (Figure 1J). Taken together, these results suggested that, in terms of the regulation of rice mesocotyl elongation in the dark, the D14L- and D14-mediated pathways act additively and in parallel, and both depend on the function of D3.

D14L Interacts with OsSMAX1 and D3

The D53-like/SMAXL protein family has nine members in rice (Jiang et al., 2013). In the presence of SL, D14 recruits D3 to degrade D53 (Jiang et al., 2013; Zhou et al., 2013). We hypothesized that D14L may be similar to D14 with respect to the recruitment of D3 for the ubiquitination of a D53-like/SMAXL protein in the regulation of mesocotyl elongation in the dark. First, we tested the interaction of individual D53-like/SMAXL proteins with D14, D14L, D14L2, and D14L3 proteins (Supplemental Figure 2) in a yeast two-hybrid assay. Among the rice D53-like/SMAXL proteins, D53L, which is encoded by LOC_Os12g01360, was most similar to D53 (Supplemental Figure 5; Supplemental Data Set 2) and could interact with D14 in the presence of rac-GR24 (Supplemental Figure 6). The protein encoded by LOC_Os08g15230 was most similar to Arabidopsis (Arabidopsis thaliana) SMAX1 and could interact with D14L (Figure 2A; Jiang et al., 2013; Stanga et al., 2013), but not with D14, D14L2, or D14L3 (Supplemental Figure 7). We therefore refer to LOC_Os08g15230 as OsSMAX1 hereafter. Pull-down assays confirmed that OsSMAX1 interacts with D14L, but not D14 (Figure 2B). Moreover, HA-OsSMAX1 was coprecipitated by GFP-D14L, but not by GFP-D14 when tagged proteins were transiently expressed in rice protoplasts, further confirming the interaction between OsSMAX1 and D14L (Figure 2C).

Figure 2.

D14L Forms a Complex with OsSMAX1 and D3.

(A) D14L interacts with OsSMAX1 in a yeast two-hybrid assay.

(B) In vitro pull-down assay with amylose resin. His-MBP-OsSMAX1 was detected by anti-His antibodies. His-Nus-D14L and His-Nus-D14 were detected by anti-Nus antibodies. MBP was detected by anti-MBP antibodies.

(C) Co-IP of transiently expressed proteins in rice protoplasts by GFP-Trapcoupled agarose beads. HA-OsSMAX1 was detected by anti-HA antibodies. GFP-D14L and GFP-D14 were detected by anti-GFP antibodies.

(D) In vitro pull-down assay by amylose resin. His-MBP-OsSMAX1 (191 to 444) was detected by anti-His antibodies. His-Nus-D14L and His-Nus-D14 were detected by anti-Nus antibodies.

(E) Pull-down assay using glutathione magnetic agarose beads. GST-D3-His was detected by anti-His antibodies. His-Nus-D14L were detected by anti-His antibodies. GST was detected by anti-GST antibodies.

(F) Co-IP of transiently expressed proteins in rice protoplasts using GFP-Trapcoupled agarose beads. HA-D3 was detected by anti-HA antibodies. GFP-D14L and GFP were detected by anti-GFP antibodies. The proteins indicated by red arrows were used as bait, and the proteins indicated by black arrows were used as prey.

We found that D14L interacts with OsSMAX1, but not with other D53-like/SMAXL proteins in the yeast two-hybrid assay (Supplemental Figure 6). To identify the OsSMAX1 domain responsible for its interaction with D14L, various OsSMAX1 fragments were tested via yeast two-hybrid assays and pull-down assays. The putative interaction domain was narrowed to OsSMAX1 (191 to 444; Figure 2D; Supplemental Figure 8A), which contained a region similar to the D14-interacting fragment of D53 (181 to 404; Supplemental Figure 8B; Zhou et al., 2013). D14 displays hydrolase activity on rac-GR24 (Zhao et al., 2013a), enhancing the interaction between D14 and D53 (Jiang et al., 2013). However, rac-GR24 showed little effect on the interaction between D14L and OsSMAX1 (Supplemental Figure 7). We mutated the conserved sites of the hydrolase catalytic triad in D14L (Ser-96, Asp-218, and His-247) and found that the mutated D14L proteins still interacted with OsSMAX1 (Supplemental Figure 9). These findings are consistent with previous findings that rice D14L has little hydrolase activity on GR24 (Zhao et al., 2013a).

To detect the interaction between D14L and D3, we performed a pull-down assay. Nus-D14L was pulled down by glutathione S-transferase (GST)-D3, but not by GST, suggesting that D14L can interact with D3 (Figure 2E). Both HA-D3 and GFP-D14L were transiently expressed in rice protoplasts. HA-D3 could be coimmunoprecipitated by GFP-D14L, but not by GFP (Figure 2F). These results suggested that D14L may form a complex with D3 and OsSMAX1, as D14 forms a complex with D3 and D53.

Accumulation of OsSMAX1 Determines Rice Mesocotyl Elongation in the Dark

SL-induced D53 degradation requires the function of D14 and D3 (Jiang et al., 2013; Zhou et al., 2013). Mutants deficient in SL biosynthesis or signaling accumulate D53 and show outgrowth of tiller buds (Jiang et al., 2013; Zhou et al., 2013). To investigate whether the protein stability of OsSMAX1 requires the function of D14L and D3 in vivo, we used OsSMAX1 antibodies to measure OsSMAX1 protein levels in the wild-type, d14, d14l, d3, and d14 d14l seedlings (Supplemental Figure 10). OsSMAX1 protein accumulated in d14l, d3, and d14 d14l, but not in the wild type or d14 (Figure 3A). These results indicated that the regulation of OsSMAX1 protein stability depends on the function of D14L and D3, but not on that of D14. By contrast, we found that the accumulation of D53 in d14l was similar to that in the wild type (Figure 3B), suggesting that D14L has little effect on the abundance of D53.

Figure 3.

Accumulation of OsSMAX1 Leads to Rice Mesocotyl Elongation in the Dark.

(A) OsSMAX1 protein levels in the wild-type (WT) and mutant seedlings detected by immunoblotting with anti-OsSMAX1 polyclonal antibodies.

(B) D53 protein levels in wild-type (WT) and mutant seedlings detected by immunoblotting with anti-D53 polyclonal antibodies.

(C) Mutation sites of OsSMAX1m.

(D) Yeast two-hybrid analysis showing that D14L interacts with both OsSMAX1and OsSMAX1m equally well.

(E) In vitro pull-down assay using amylose resin. His-MBP-OsSMAX1 and His-MBP-OsSMAX1m were detected by anti-His antibodies. His-Nus-D14L and His-Nus-D14 were detected by anti-Nus antibodies. The protein indicated by the red arrow was used as bait, and the proteins indicated by black arrows were used as prey.

(F) 35S:REN-2A-OsSMAX1-FF and 35S:REN-2A-OsSMAX1m-FF transiently expressed in rice protoplasts of the wild type (WT) and mutants. The FF:REN is the average ratio of the bioluminescence of firefly luciferase to that of Renilla luciferase. The lowercase letters indicate samples expressing 35S:REN-2A-OsSMAX1-FF, and the uppercase letters indicate samples expressing 35S:REN-2A-OsSMAX1m-FF.

(G) Seedlings of wild-type (WT) and transgenic Ubi:OsSMAX1-GFP-3XFlag and Ubi:OsSMAX1m-GFP-3XFlag seedlings grown in the dark for 7 d. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrow. Bar = 2 cm.

(H) Length of the mesocotyl of wild-type (WT) and transgenic Ubi:OsSMAX1-GFP-3XFlag and Ubi:OsSMAX1m-GFP-3XFlag seedlings.

(I) OsSMAX1 protein levels in seedlings of the wild-type (WT) and transgenic Ubi:OsSMAX1-GFP-3XFlag and Ubi:OsSMAX1m-GFP-3XFlag seedlings as revealed by immunoblotting with anti-Flag monoclonal antibodies. In (F), data are presented as means ± ses (n = 3). In (H), the numbers above the columns indicate the sample sizes. Data are presented as means ± sds. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5).

In d53 plant, the lack of the Gly-Lys-Thr (GKT) motif prevents d53 from being subjected to SL-induced degradation. When the GKT motif-mutated D53 (D53m) was overexpressed in the wild type, the shoot-branching phenotypes were similar to those of d53 (Supplemental Figures 1A and 1B; Jiang et al., 2013). Since the GKT motif region of OsSMAX1 is highly similar to that of D53, we generated a mutated OsSMAX1 (OsSMAX1m) by an in-frame deletion of the GKT motif at 744 to 747 (Gly-Lys-Thr-Ala; Figure 3C), which is similar to that in D53 at 813 to 817 (Gly-Lys-Thr-Gly-Ile). Since d53 interacts with D14 as well as D53, yeast two-hybrid assays (Figure 3D) and pull-down assays (Figure 3E) were conducted to verify the interaction between OsSMAX1m and D14L. The results showed that OsSMAX1m interacts with D14L as well as OsSMAX1.

To further evaluate the stability of D53/SMAXL family proteins, we performed a StrigoQuant-like reporter transient expression assay in rice protoplasts (Supplemental Figure 11A; Samodelov et al., 2016). To verify the viability of this assay, we first tested the effects of rac-GR24 on the stability of D53 and D53m reporters that were transiently expressed in the wild-type, d3, d14, d14l, and d14 d14l protoplasts (Supplemental Figure 11B). The ratio of firefly luciferase activity to Renilla luciferase activity (FF:REN) of the D53 reporter in d14, d3, and d14 d14l treated with 1 μM rac-GR24 was higher than that in the wild type and d14l. By contrast, there were no obvious differences in the FF:REN of the D53m reporter in response to rac-GR24 treatment among the wild type, d3, d14, d14l, and d14 d14l (Supplemental Figure 11B). To evaluate the stability of OsSMAX1 and OsSMAX1m, StrigoQuant-like reporters of OsSMAX1 and OsSMAX1m were transiently expressed in wild-type, d14, d14l, d3, and d14 d14l protoplasts. The stability of OsSMAX1, indicated by the FF:REN, was higher in d14l, d3, and d14 d14l than in the wild type, while the stability of OsSMAX1 in d14 was similar to that in the wild type (Figure 3F). However, the stability of OsSMAX1m did not obviously differ in d14, d14l, d3, or d14 d14l compared to that in the wild type (Figure 3F). These results suggested that the stability of OsSMAX1 is regulated by D14L and D3, but not by D14, and that the mutation of the GKT motif of OsSMAX1 enables OsSMAX1m to resist D14L- and D3-dependent degradation.

To test whether the elongation of the mesocotyls of d14l, d3, and d14 d14l resulted from the accumulation of OsSMAX1 in these mutants, we constructed vectors to express OsSMAX1 and OsSMAX1m driven by the maize UBIQUITIN promoter (Ubi:OsSMAX1-GFP-3XFlag and Ubi:OsSMAX1m-GFP-3XFlag, respectively) and transformed them into rice cv Nipponbare (Figures 3G to 3I). OsSMAX1 and OsSMAX1m overexpression promoted mesocotyl elongation in the dark (Figures 3G and 3H). The length of the mesocotyls in the transgenic seedlings was consistent with the protein levels of the OsSMAX1 or OsSMAX1m fusion proteins (Figure 3I). We further generated constructs overexpressing OsSMAX1 and OsSMAX1m driven by the rice ACTIN promoter (ACT:OsSMAX1-Flag and ACT:OsSMAX1m-Flag, respectively), and transformed them into Nipponbare. In addition, we generated an OsSMAX1 promoter-driven OsSMAX1m expression vector (OsSMAX1:OsSMAX1m-GFP-3XFlag) and subsequently obtained transgenic seedlings (Supplemental Figure 12A). The length of the mesocotyl of the ACT:OsSMAX1-Flag seedlings was similar to that of wild-type seedlings. The length of the mesocotyl of the ACT:OsSMAX1m-Flag and OsSMAX1:OsSMAX1m-GFP-3XFlag transgenic seedlings was significantly greater than that of wild-type seedlings (Supplemental Figure 12B), which is consistent with the abundance of OsSMAX1m-Flag fusion protein in the ACT:OsSMAX1m-Flag transgenic seedlings and the OsSMAX1m-GFP-Flag fusion protein in the OsSMAX1:OsSMAX1m-GFP-3XFlag transgenic seedlings (Supplemental Figure 12C). These results indicated that OsSMAX1m is more likely resistant to ligand-induced degradation than OsSMAX1 and that the accumulation of OsSMAX1 determines the rice mesocotyl-elongation phenotype in the dark.

Karrikin-Like Signals (KLs) Induce Ubiquitination and Degradation of OsSMAX1

To determine the role of karrikins in the inhibition of rice mesocotyl elongation in the dark, we added either KAR₁ or KAR₂ to the medium used for seedling growth in the dark for 7 d (Figures 4A to 4C). Treatment with either 20 μM KAR₁ or KAR₂ inhibited the elongation of the mesocotyls of the wild-type and d14 seedlings but did not inhibit the elongation of the mesocotyls of either the d14l or the d3 seedlings (Figures 4A and 4B). Treatment with karrikins consistently resulted in decreased abundance of OsSMAX1 in the wild type and d14 but had little effect on the abundance of OsSMAX1 in d14l and d3 (Figure 4C). These results indicated that karrikins could induce the degradation of OsSMAX1, which is dependent on the function of D14L and D3. To confirm whether karrikins induce the degradation of OsSMAX1, which depends on the function of the GKT motif, we treated calli derived from Ubi:OsSMAX1-GFP-Flag and Ubi:OsSMAX1m-GFP-Flag transgenic seedlings with 10 μM KAR₁ and found that KAR₁ had no obvious effects on the induction of OsSMAX1 degradation within 2 h (Figure 4D).

Figure 4.

KL Signal-Induced Degradation of OsSMAX1 Depends on the Function of D14L and D3.

(A) Dark-grown 7-d-old wild-type (WT), d14, d14l, and d3 seedlings under karrikin treatment. KAR₁ and KAR₂ (20 µM) were added to the media. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrows. Bars = 2 cm.

(B) Length of the mesocotyl of the indicated seedlings grown in the dark for 7 d under karrikin treatment. Data are presented as means ± sds, and the numbers above the columns indicate the sample sizes. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5).

(C) OsSMAX1 protein levels in the dark-grown 7-d-old wild-type (WT), d14, d14l, and d3 seedlings under karrikin treatment. Anti-OsSMAX1 and anti-ACTIN antibodies were used for immunoblotting.

(D) OsSMAX1 and OsSMAX1m protein levels in calli of Ubi:OsSMAX1-GFP-Flag and Ubi:OsSMAX1m-GFP-Flag transgenic plants after chemical treatment (10 µM KAR₁, GR24^5DS, and GR24^ent-5DS) at the indicated time points. The anti-Flag and anti-ACTIN antibodies were used for immunoblotting.

(E) Ubiquitination assay of OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag in response to 10 μM GR24^ent-5DS. Polyubiquitinated proteins detected by monoclonal anti-ubiquitin antibodies. OsSMAX1-GFP-Flag, OsSMAX1m-GFP-Flag, and GFP proteins detected by anti-GFP antibodies. IP, immunoprecipitation; Ub, ubiqitin.

It has been suggested that different GR24 stereoisomers have different effects and that the nonnatural enantiomer GR24^ent-5DS can mimic the inhibitory effects of karrikin on hypocotyl elongation in Arabidopsis (Scaffidi et al., 2014). We therefore measured the abundance of the OsSMAX1-GFP-Flag fusion protein in response to GR24 stereoisomer treatment. OsSMAX1-GFP-Flag degraded upon treatment with 10 μM GR24^ent-5DS, but not with 10 μM GR24^5DS (Figure 4D). However, the abundance of OsSMAX1m was not affected by treatment with either GR24^ent-5DS or GR24^5DS (Figure 4D). Moreover, the results showed that GR24^ent-5DS treatment could induce the ubiquitination of OsSMAX1, but not of OsSMAX1m (Figure 4E). Taken together, these results indicated that different GR24 stereoisomers have distinct effects on the induction of OsSMAX1 degradation and that the mutation of the GKT motif of OsSMAX1 is resistant to GR24^ent-5DS-induced ubiquitination and degradation.

To further determine whether karrikins could enhance the interaction between D14L and OsSMAX1 in a manner similar to that of SLs with respect to the interaction between D14 and D53, we performed a yeast two-hybrid assay with different small molecules. The results showed that the addition of 20 μM rac-GR24 or GR24^5DS could enhance the interaction between D14 and D53, but neither rac-GR24 and GR24^5DS nor KAR₁ and GR24^ent-5DS could enhance the interaction between D14L and OsSMAX1 (Supplemental Figure 13). The pull-down assay also revealed that addition of 20 μM rac-GR24, KAR₁, GR24^5DS, or GR24^ent-5DS has no obvious effect on the interaction between D14L and OsSMAX1. Even when the concentration of these chemicals was increased to 50 μM, KAR₁ and GR24^ent-5DS still had no obvious effect on the interaction between D14L and OsSMAX1 (Supplemental Figure 14). These results indicated that there may be different recognition mechanisms between KLs and D14L and between SLs and D14.

OsSMAX1 Acts Downstream of D14L and D3 in Regulating Mesocotyl Elongation

To further dissect the role of OsSMAX1 in the regulation of mesocotyl elongation in the dark, we generated OsSMAX1 knockout plants by CRISPR/Cas9-mediated genome editing (Supplemental Figure 15) and found that loss of function of OsSMAX1 reduced rice mesocotyl elongation in the dark compared to that of the wild type (Supplemental Figure 15). Ossmax1-1, which has a 4-bp deletion at site 70 to 73, was used for further analysis and is henceforth referred to as Ossmax1. The expression levels of the OsTCP5, GY1, and OsGSK2 genes are consistent with the mesocotyl phenotypes of Ossmax1 and OsSMAX1:OsSMAX1m-GFP-3XFlag, referred to OsSMAX1m-overexpression(OE) hereafter (Supplemental Figures 16A to 16C). These results indicated that OsSMAX1 may regulate mesocotyl elongation in the dark by regulating the expression of these genes during mesocotyl development. Moreover, the expression of the D14L, D14L2, and D14L3 genes increased in Ossmax1 but decreased in OsSMAX1m-OE (Supplemental Figures 16D to 16F). Taken together, these results indicated that the D14L-D3-OsSMAX1 complex might act additively in conjunction with the D14-D3-D53 complex to control the expression of downstream genes, such as the D14L, D14L2, and D14L3.

In Arabidopsis, loss of function of SMAX1 and SMXL2 suppresses the hypocotyl-elongation phenotype of max2 (Stanga et al., 2013, 2016; Soundappan et al., 2015). To test whether OsSMAX1 acts downstream of D14L and D3 in the regulation of mesocotyl elongation, we obtained Ossmax1 d14l and Ossmax1 d3 double mutant plants. The length of the mesocotyl of both the Ossmax1 d14l and Ossmax1 d3 double mutants in the dark was similar to that of Ossmax1 (Figures 5A and 5B), indicating that OsSMAX1 acts downstream of D14L and D3. Moreover, we measured the protein levels of OsSMAX1 in the wild type, d3, d14l, Ossmax1, Ossmax1 d3, and Ossmax1 d14l and found that the loss of function of OsSMAX1 resulted in no accumulation of OsSMAX1 in d3 and d14l (Figure 5C). The transcript levels of both OsTCP5 and GY1 in these mutants compared with the wild type were consistent with the abundance of OsSMAX1 and with the mesocotyl-elongation phenotypes in these mutants (Figure 5D). These results strongly suggested that OsSMAX1 acts downstream of D14L and D3 to inhibit mesocotyl elongation in the dark.

Figure 5.

Loss of Function of OsSMAX1 Suppresses the Elongated-Mesocotyl Phenotype of d14l and d3.

(A) Dark-grown 7-d-old seedlings of the wild type (WT), d14l, d3, Ossmax1, Ossmax1 d14l, and Ossmax1 d3. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrows. Bar = 2 cm.

(B) Length of the mesocotyl of dark-grown 7-d-old seedlings in (A).

(C) Abundance of OsSMAX1 proteins in dark-grown 7-d-old seedlings in (A).

(D) Relative expression levels of OsTCP5 and GY1 in dark-grown 7-d-old seedlings of the wild type (WT), d14l, d3, Ossmax1, Ossmax1 d14l, and Ossmax1 d3.

(E) Dark-grown 7-d-old wild-type (WT), d10, d14, Ossmax1, Ossmax1 d10, and Ossmax1 d14 seedlings. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrows. Bar = 2 cm.

(F) Length of the mesocotyl of dark-grown 7-d-old seedlings in (E).

(G) Relative expression levels of OsTCP5 and GY1 in dark-grown 7-d-old seedlings of the wild type (WT), d10, Ossmax1, and Ossmax1 d10.

(H) Relative expression levels of OsTCP5 and GY1 in dark-grown 7-d-old seedlings of the wild type (WT), d14, Ossmax1, and Ossmax1 d14. In (B) and (F), data are presented as means ± sds, and the numbers above the columns indicate the sample sizes. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5). In (D), (G), and (H), the expression of each indicated gene is relative to that of ACTIN as the internal reference and is replicated three times. Data are presented as means ± ses (n = 3). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Tukey's test (P < 0.05; Supplemental Data Set 5).

D3 can be recruited by D14 to degrade D53 (Jiang et al., 2013; Zhou et al., 2013) and can be recruited by D14L to degrade OsSMAX1 (Figures 2 and 4D). Since the mesocotyl phenotype of d14 d14l is similar to that of d3, it is reasonable to expect that loss of function of OsSMAX1 may rescue the elongated-mesocotyl phenotype of d14l and that the mesocotyl length of Ossmax1 d3 would be similar to that of d14. Surprisingly, the loss of function of OsSMAX1 fully rescued the elongated-mesocotyl phenotype of d3 instead of partially rescuing the mesocotyl phenotype of d3 in the dark (Figures 5A to 5D). Thus, we generated Ossmax1 d10 and Ossmax1 d14 double mutants. The length of the mesocotyls of both Ossmax1 d10 and Ossmax1 d14 was similar to that of Ossmax1 (Figures 5E and 5F). Furthermore, the transcript levels of both OsTCP5 and GY1 in Ossmax1 d10 and Ossmax1 d14 were similar to those in Ossmax1 (Figures 5G and 5H), and the expression levels of OsTCP5 and GY1 were consistent with the observed mesocotyl-elongation phenotypes of these mutants. Together, these findings indicated that OsSMAX1 might act downstream of D14-D3-D53 signaling in the regulation of rice mesocotyl elongation in the dark.

OsSMAX1 transcript levels were downregulated in d14l, d3, d14 d14l, and OsSMAX1m-OE compared to those of the wild type but were upregulated in Ossmax1 and Ossmax1 d3 (Supplemental Figure 16G), indicating that OsSMAX1 expression was negatively regulated by the accumulation of the OsSMAX1 protein. Surprisingly, OsSMAX1 transcript levels were also downregulated in d14 and D53m-OE but upregulated in Ossmax1 d14 (Supplemental Figure 16G). D53 transcription is subject to feedback regulation from SL signaling (Jiang et al., 2013). Interestingly, we found that D53 expression levels decreased in d14l and OsSMAX1m-OE, similar to that in d14 and D53m-OE, but increased in Ossmax1 (Supplemental Figure 16G). Collectively, these results indicated that the D14-D3-D53 and D14L-D3-OsSMAX1 signaling complexes act interdependently to maintain the expression levels of D53 and OsSMAX1.

OsSMAX1 Is Required for SL-Mediated Regulation of Mesocotyl Elongation, but Not for SL-Mediated Inhibition of Shoot Branching

To determine whether D14L-D3-OsSMAX1 regulates shoot branching like D14-D3-D53 does, we measured the tiller number of the wild-type, d3, d14, d14l, and d14 d14l plants (Figures 6A and 6B). No significant differences in tiller number were detected between the wild type and d14l, and the tiller number of d14 d14l was similar to that of d14 and d3 (Figures 6A and 6B). Moreover, the tiller number of the Ubi:OsSMAX1-GFP-3XFlag and Ubi:OsSMAX1m-GFP-3XFlag transgenic plants did not significantly differ from that of wild-type plants (Supplemental Figures 17A and 17B). Consistent with these results, the tiller numbers of the ACT:OsSMAX1-Flag, ACT:OsSMAX1m-Flag, and OsSMAX1:OsSMAX1m-GFP-3XFlag transgenic plants were also similar to that of the wild type (Supplemental Figures 17C and 17D). Although the loss of function of OsSMAX1 inhibits mesocotyl elongation in the dark, compared with that of wild-type plants, the tiller number of OsSMAX1 knockout plants differed little (Supplemental Figures 17E and 17F). Although the loss of function of OsSMAX1 suppresses the elongated-mesocotyl phenotype of d3 (Figures 5A and 5B), the tiller number of Ossmax1 d3 was similar to that of d3 (Figures 6A and 6B), which indicated that OsSMAX1 has little effect on rice shoot branching. Furthermore, the tiller numbers of Ossmax1 d14 and Ossmax1 d10 were similar to those in d14 and d10 (Figures 6A and 6B), respectively. These results suggested that the D14L-D3-OsSMAX1 complex, unlike the D14-D3-D53 complex, is not required to regulate shoot branching in rice.

Figure 6.

D14L-D3-OsSMAX1–Mediated Signaling Pathway Does Not Regulate Shoot Branching.

(A) Morphology of wild-type (WT), d10, d14, d3, d53, d14l, Ossmax1, Ossmax1 d10, Ossmax1 d14, Ossmax1 d3, d53 d14l, and d14 d14l plants at the heading stage.

(B) Tiller number of the indicated plants at the heading stage.

(C) Height of the indicated plants at the heading stage. Bars = 5 cm. Data are presented as means ± sds (n = 10). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5).

We observed that the plant height was reduced both in the OsSMAX1 and OsSMAX1m overexpression plants (Supplemental Figure 17G) and in the Ossmax1 plants (Figure 6C). Similarly, the plant height of Ossmax1 d3, Ossmax1 d14, and Ossmax1 d10 was also reduced compared to that of d3, d14, and d10, respectively (Figure 6C). These results suggested that the D14L-D3-OsSMAX1 complex may be involved in regulating plant height in rice.

OsSMAX1 Interacts with TPRs to Regulate Downstream Gene Expression

D53/SMAXL family proteins are localized in the nucleus and are able to recruit TPR transcriptional corepressors through their EAR motifs to regulate the expression of downstream genes (Causier et al., 2012; Jiang et al., 2013; Liang et al., 2016; Ma et al., 2017). Similar to D53, OsSMAX1 has an EAR motif (Supplemental Figure 18A), indicating that OsSMAX1 might interact with TPRs to regulate downstream gene expression, as observed for D53 (Jiang et al., 2013; Ma et al., 2017). Interactions between OsSMAX1 and rice TPRs were detected via two-hybrid assays (Figure 7A), and the interaction between OsSMAX1 and the N terminus of TPR2 was further verified via pull-down assays (Figure 7B). The mutated EAR motif of OsSMAX1 disrupted the interaction between OsSMAX1 and OsTPR2 (Supplemental Figure 18B). In addition, OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag fusion proteins localized mainly to the nucleus (Figure 7C). These observations indicated that OsSMAX1 may recruit TPR transcriptional corepressors to regulate the expression of downstream genes.

Figure 7.

OsSMAX1 Interacts with TPRs and Regulates Downstream Gene Expression.

(A) Yeast two-hybrid assay showing that OsSMAX1 interacts with the N terminus (1 to 300) of TPRs.

(B) In vitro pull-down assay by amylose resin. His-MBP-OsSMAX1 was detected by anti-His antibodies. TPR2(1-600)-His was detected by anti-His antibodies. MBP was detected by anti-MBP antibodies. The proteins used as bait are indicated by red arrows, and the proteins used as prey are indicated by black arrows.

(C) Subcellular localization of OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag fusion proteins. The images show the roots of Ubi:OsSMAX1-GFP-Flag and Ubi:OsSMAX1m-GFP-Flag transgenic plants. Bars = 100 μm. DIC, differential interference contrast; PI, propidium iodide.

(D) Venn diagram of differentially regulated genes in Ossmax1 and OsSMAX1:OsSMAX1m-GFP-Flag transgenic plants compared to wild-type (WT) plants.

(E) Heatmap of the expression fold change of upregulated genes in Ossmax1 mutant plants and downregulated genes in OsSMAX1:OsSMAX1m-GFP-Flag (OsSMAX1m-OE) transgenic plants compared to wild-type (WT) plants.

(F) to (H) Relative expression levels of the indicated OsSMAX1 repressed genes, LOC_Os06g49750 (F), LOC_Os02g40240 (G), and LOC_Os05g11414(H), in different mutant and transgenic plants.

(I) to (K) Relative expression levels of the indicated OsSMAX1-activated genes, LOC_Os04g15840 (I), LOC_Os06g4032355 (J), and LOC_Os03g64330(K), in different mutant and transgenic plants compared to wild-type (WT) plants. In (F) to (K), the expression of each indicated gene is relative to that of ACTIN as the internal reference. Data are presented as means ± ses (n = 3). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Tukey's test (P < 0.05; Supplemental Data Set 5).

To identify the downstream genes regulated by OsSMAX1, we compared the gene expression profiles of the wild type, Ossmax1, and OsSMAX1m-OE and identified 84 genes potentially regulated by OsSMAX1. Among these genes, the expression of 18 genes was upregulated in Ossmax1 and downregulated in OsSMAX1m-OE, and the expression of 66 genes was downregulated in Ossmax1 and upregulated in OsSMAX1m-OE (Figure 7D). The expression profiles of these OsSMAX1 negatively regulated genes and positively regulated genes are shown, respectively, in a heatmap (Figure 7E; Supplemental Figure 19). We selected several genes for RT-qPCR analysis and confirmed their expression levels. The expression of LOC_Os06g49750, LOC_Os02g40240, and LOC_Os05g11414 was downregulated in OsSMAX1m-OE and upregulated in Ossmax1 (Figures 7F to 7H). LOC_Os06g49750 encodes the rice homolog of the karrikin-inducible marker gene KARRIKIN UPREGULATED F-BOX1 (KUF1) in Arabidopsis (Figure 7F; Nelson et al., 2011). LOC_Os02g40240 (Figure 7G) is a light-inducible gene that encodes a plasma membrane receptor-like kinase of LEAF PANICLE2 (LP2; Thilmony et al., 2009). LOC_Os05g11414 encodes the MADS-box family transcription factor OsMADS58, which regulates the expression of photosynthesis-related genes (Figure 7I; Chen et al., 2015). By contrast, the expression of LOC_Os04g15840, LOC_Os06g32355, and LOC_Os03g64330, which encode members of the Expansin, Thionin, and Aquaporin family proteins, respectively, was upregulated in OsSMAX1m-OE and downregulated in Ossmax1 (Figures 7I and 7J). These results provided further evidence that OsSMAX1 influences rice mesocotyl elongation in the dark through the regulation of the expression of downstream genes.

KL and SL Signaling Complexes Perceive Different GR24 Stereoisomers to Inhibit Rice Mesocotyl Elongation in the Dark

To determine the role of karrikin and SL in the inhibition of mesocotyl elongation in the dark, we added chemicals to the media used for seedling growth and observed the seedlings under darkness for 7 d (Supplemental Figure 20). We found that 1 µM rac-GR24 or GR24^ent-5DS is sufficient to inhibit the elongation of the mesocotyl of d10, but not of the wild type, and that 10 µM rac-GR24 or GR24^ent-5DS is able to significantly inhibit the elongation of the mesocotyls of the wild type and d10. Treatment with 20 µM rac-GR24 caused more obvious inhibitory effects than did treatment with 10 µM rac-GR24 for the wild type and d10, while treatment with 20 µM GR24^ent-5DS caused a more obvious inhibitory effect than did 10 µM GR24^ent-5DS in the wild type, but not in d10 (Supplemental Figure 20). Treatment with 20 μM rac-GR24 dramatically inhibited the elongation of the mesocotyls of the wild-type and d17 seedlings, slightly inhibited the elongation of the mesocotyls of d14 and d14l, and had little effect on the elongation of the mesocotyls of the d3 and d14 d14l mutants. However, 20 μM KAR₁ inhibited the elongation of the mesocotyls of the wild-type, d17, and d14 seedlings but had little effect on those of the d14l, d3, and d14 d14l mutants (Figures 8A and 8B). rac-GR24 treatment induced D53 expression in a D14-dependent manner, while KAR₁ treatment induced OsSMAX1 expression in a D14L-dependent manner (Figures 8C and 8D). The expression of OsKUF1 responded more specifically to KAR₁ than to rac-GR24 in the wild type and responded to KAR₁ only in d17 and d14 (Figure 8E). Expression of the OsSMAX1-downregulated gene LP2 was induced in response to both KAR₁ and rac-GR24 in the wild type and d17 and induced in response to KAR₁ in d14 (Figure 8F). Expression of the OsSMAX1-upregulated genes LOC_Os04g15840 and LOC_Os06g32355 was not induced in response to KAR₁ or rac-GR24 in the wild type (Figures 8G and 8H). Deficiency of the SL or karrikin pathway led to increased expression of these two genes, such that their expression was inhibited in response to both KAR₁ and rac-GR24 in the d17 mutants (Figures 8G and 8H). The expression of LOC_Os04g15840 and LOC_Os06g32355 in d14, d14l, and d14 d14l was responsive to both KAR₁ and rac-GR24. However, rac-GR24 did not inhibit their expression in d14l as well as KAR₁ did in d14 (Figures 8G and 8H), which was consistent with the inhibition effect of rac-GR24 and KAR₁ on the mesocotyl elongation in d14l and d14 (Figure 8B). Given that rac-GR24 is a racemic mixture of two enantiomers (GR24^5DS and GR24^ent-5DS) and that GR24^ent-5DS triggers the ubiquitination and degradation of OsSMAX1, we expected rac-GR24 to be capable of activating both D14- and D14L-mediated downstream responses in the wild type and activating the karrikin-induced response in d14 and the SL-induced response in d14l, respectively. Surprisingly, seedlings treated with 20 μM rac-GR24 showed a slight inhibition in the elongation of the mesocotyls in the d14 and d14l seedlings (Figures 8A to 8C; Supplemental Figure 20). Treatment of seedlings with 40 μM rac-GR24 revealed a significant inhibitory effect on the elongation of the mesocotyls of the wild type and d14 but no inhibitory effect on the elongation of the mesocotyl of d14l (Supplemental Figure 20). However, when the wild-type, d17, d14, d14l, d3, and d14 d14l seedlings were grown in media only with 20 μM GR24^5DS or GR24^ent-5DS, GR24^5DS inhibited the elongation of the mesocotyls of the wild-type, d17, and d14l seedlings, but not of the d14 seedlings (Figures 9A and 9B). By contrast, GR24^ent-5DS inhibited the elongation of the mesocotyls of the wild-type, d17, and d14 seedlings, but not of the d14l seedlings (Figures 9A and 9B). Neither GR24^5DS nor GR24^ent-5DS had a significant inhibitory effect on the length of the mesocotyl of d3 or d14 d14l (Figures 9A and 9B). Consistent with its ability to inhibit mesocotyl elongation, GR24^ent-5DS treatment reduced the accumulation of OsSMAX1 in the wild-type, d17, and d14 seedlings but had little effect on the abundance of OsSMAX1 in the d14l and d14 d14l seedlings (Figures 9C and 9D). Taken together, these results showed that D14L responds specifically to GR24^ent-5DS in the regulation of mesocotyl development.

Figure 8.

Mesocotyl Elongation in the Dark in Response to Both SL and Karrikin Signals.

(A) Dark-grown 7-d-old wild-type (WT), d17, d14, d14l, d14 d14l, and d3 seedlings with or without treatment with the indicated chemical at 20 µM. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrows. Bars = 2 cm.

(B) Length of the mesocotyls of seedlings shown in (A). WT, wild type.

(C) Relative expression levels of D53 in response to chemical treatment. WT, wild type.

(D) Relative expression levels of OsSMAX1 in response to chemical treatment. WT, wild type.

(E) Relative expression levels of the OsSMAX1-repressed gene KUF1 (LOC_Os06g49750) in response to chemical treatment. WT, wild type.

(F) Relative expression levels of the OsSMAX1-repressed gene LP2 (LOC_Os02g40240) in response to chemical treatment. WT, wild type.

(G) Relative expression levels of the OsSMAX1-enhanced gene LOC_Os04g15840 in response to chemical treatment. WT, wild type.

(H) Relative expression levels of the OsSMAX1-enhanced gene LOC_Os06g32355 in response to chemical treatment. In (B), data are presented as means ± sds, and the numbers above the columns indicate the sample sizes. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5). In (C) to (H), the expression of each indicated gene is relative to that of ACTIN as the internal reference. The expression values are scaled to the expression levels in the mock-treated wild type (WT). Data are presented as means ± ses (n = 3). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Tukey's test (P < 0.05; Supplemental Data Set 5).

Figure 9.

SL and Karrikin Signaling Complexes Selectively Perceive GR24 Stereoisomers in the Inhibition of Mesocotyl Elongation in the Dark.

(A) Dark-grown 7-d-old wild-type (WT), d17, d14, d14l, d14 d14l, and d3 seedlings with or without treatment of the indicated chemical at 20 µM. The center seedling is shown enlarged in the top right corner, and the mesocotyl is indicated by the arrows. Bars = 2 cm.

(B) Length of the mesocotyls of seedlings shown in (A). Data are presented as means ± sds, and the numbers above the columns indicate the sample sizes. Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Duncan's test (P < 0.05; Supplemental Data Set 5). WT, wild type.

(C) OsSMAX1 protein levels in the dark-grown 7-d-old wild-type (WT), d14, and d14l seedlings treated with GR24 stereoisomers.

(D) OsSMAX1 protein levels in the dark-grown 7-d-old (WT), d17, and d14 d14l seedlings treated with GR24 stereoisomers. The anti-OsSMAX1 and anti-ACTIN antibodies were used for immunoblotting in (C) and (D).

(E) Relative expression levels of D53 in response to GR24 stereoisomer treatment. WT, wild type.

(F) Relative expression levels of OsSMAX1 in response to GR24 stereoisomer treatment. WT, wild type.

(G) Relative expression levels of LP2 (LOC_Os02g40240) in response to GR24 stereoisomer treatment. WT, wild type.

(H) Relative expression levels of KUF1 (LOC_Os06g49750) in response to GR24 stereoisomer treatment. In (E) to (H), the expression of each indicated gene is relative to that of ACTIN as the internal reference. The expression values are scaled to the expression level in the mock-treated wild type (WT). All data are presented as means ± ses (n = 3). Statistically significant differences were determined by one-way ANOVA; the different letters indicate significant differences between samples according to Tukey's test (P < 0.05; Supplemental Data Set 5).

GR24^5DS induced D53 expression in the wild-type, d17, and d14l seedlings, but not in d14, d14 d14l, and d3 seedlings (Figure 9E). By contrast, GR24^ent-5DS induced OsSMAX1 expression in the d17 and d14 seedlings, but not in the wild-type, d14l, d14 d14l, and d3 seedlings (Figure 9F). RT-qPCR analysis showed that LP2 is responsive to GR24^ent-5DS or GR24^5DS in a D14- or D14L-dependent manner, respectively (Figure 9G). The OsKUF1 seems responsive specifically to GR24^ent-5DS in a D14L- and D3-dependent manner. However, OsKUF1 is also responsive GR24^5DS in d17, but not in d14l (Figure 9H). In addition, both LOC_Os04g15840 and LOC_Os06g32355 are responsive to GR24^5DS or GR24^ent-5DS in a D14- or D14L-dependent manner, respectively (Supplemental Figure 21). Our results showed that the OsSMAX1-regulated genes either respond specifically to GR24^ent-5DS or respond to both GR24^5DS and GR24^ent-5DS. The response to GR24^5DS depends on the function of D14 and D3, while the response to GR24^ent-5DS depends on the function of D14L and D3. These results demonstrated that the nonnatural enantiomer GR24^ent-5DS may mimic the inhibitory effect of karrikin on mesocotyl elongation in the dark, as it did with the inhibition of Arabidopsis hypocotyl elongation under light conditions. These findings suggested that D14-D3-D53 and D14L-D3-OsSMAX1 specifically recognize different GR24 stereoisomers and regulate specific and common downstream target genes during rice mesocotyl elongation in the dark.

DISCUSSION

Comparative approaches to the study of karrikin signaling and SL signaling have provided crucial information about both pathways, with progress in one helping to inform the other (Waters et al., 2017). KAR₂ and rac-GR24 inhibit hypocotyl elongation in the light in Arabidopsis (Nelson et al., 2011; Waters et al., 2012). In addition, in Arabidopsis, a previous promoter-swapping analysis of KAI2 and D14 revealed that their distinct roles in plant growth are likely due to the formation of distinct signaling complexes (Waters et al., 2015b). It has been reported that SMAX1 and SMXL6/SMXL7/SMXL8 proteins can suppress distinct or overlapping phenotypes of max2 (Soundappan et al., 2015). AtD14, MAX2, and SMXL6/SMXL7/SMXL8 are responsible for SL responses in the regulation of shoot branching (Soundappan et al., 2015; Wang et al., 2015). Genetic analysis revealed that SMAX1 and SMXL2 act downstream of MAX2 and KAI2 in karrikin signaling (Stanga et al., 2013, 2016; Villaécija-Aguilar et al., 2019). Both SMAX1 and SMXL2 can be degraded in response to karrikin signaling in a KAI2- and MAX2-dependent manner in Arabidopsis (Khosla et al., 2020; Wang et al., 2020). The specificity of the karrikin or SL signaling pathway is determined by the perception of ligands by the karrikin or SL receptor (KAI2 or AtD14, respectively), each of which can recruit SCF^MAX2, leading to a decrease in the stability of different subgroups of D53/SMAXL proteins (Soundappan et al., 2015; Wang et al., 2015, 2020; Liang et al., 2016; Khosla et al., 2020). Previous work in rice revealed that D14 and D14L play distinct roles in shoot branching and AM symbiosis, both of which require the function of D3 (Arite et al., 2009; Yoshida et al., 2012; Jiang et al., 2013; Zhou et al., 2013; Gutjahr et al., 2015b). D53 has been revealed to act downstream of D14 and D3 and regulate the shoot branching of rice (Jiang et al., 2013; Zhou et al., 2013). By contrast, OsSMAX1 acts downstream of D14L and D3 to regulate root colonization by AM fungi and the elongation of the mesocotyl in rice (Choi et al., 2020). In this study, we show that OsSMAX1 acts downstream of D14L and D3 and parallel to D14-D3-D53 in regulation the elongation of the mesocotyl. We further demonstrate that D14 and D14L specifically recognize different GR24 stereoisomers, but both are required for the function of D3 to regulate the accumulation of specific members of the D53/SMAXL family protein in the regulation of downstream events. In addition, we determined that, similar to AtKUF1 in Arabidopsis, OsKUF1 in rice is regulated specifically by KL signals (Figures 8C and 9E). These results indicate that the karrikin signaling pathway is relatively conserved in dicotyledonous and monocotyledonous plants.

Here, we proposed a working model for the interaction of karrikin signaling in rice (Figure 10). In the absence of a KL ligand, the activity of unidentified OsSMAX1-interacting transcription factors is repressed by the OsSMAX1-TPR complex. In the presence of a KL signal, D14L perceives the ligand and recruits the SCF^D3 complex to polyubiquitinate OsSMAX1, which is subsequently degraded by the 26S proteasome. The degradation of OsSMAX1 leads to the release of OsSMAX1-interacting transcription factors from the TPR transcriptional corepressor and alters the expression of its downstream target genes. Our results suggested that the specificity of karrikin and SL signaling could be determined by the degradation of distinct D53/SMAXL proteins in response to the recognition of specific ligands by their receptors. Thus, SL and KL signals act independently to trigger the expression of specific downstream genes that distinguish D14L-dependent processes (AM symbiosis) and D14-dependent processes (shoot branching) mediated by transcription factors that interact specifically with D53 and OsSMAX1, respectively. Disruption of either the SL or karrikin pathway alters the expression of common downstream target genes, such as D14L2 and D14L3. In some cases, the disruption of both the SL and karrikin pathways resulted in additive changes in the expression levels of downstream genes, such as LOC_Os01g41310 and LOC_Os03g52080 (Supplemental Data Set 3). The GR24 stereoisomers GR24^5DS and GR24^ent-5DS could mimic SL-specific and KL-specific signals and act through D14 and D14L to degrade D53 and OsSMAX1, respectively, leading to the release of activity of common transcription factors from repression mediated by TPR transcriptional corepressors, which resulted in changes in the expression of common target genes and inhibited rice mesocotyl elongation in the dark.

Figure 10.

KL Signaling Pathway Mirrors the SL Signaling Pathway.

A working model of karrikin signaling mediated by the D14L-D3-OsSMAX1 complex. Karrikin signaling mirrors the SL signaling complex in rice. In the absence of ligands, both OsSMAX1 and D53 are able to interact with TPR transcriptional corepressors and repress the expression of downstream genes. In the presence of ligands, D14L and D14 perceive specific ligands (such as GR24^ent-5DS and GR24^5DS) and recruit the SCF^D3 complex to ubiquitinate OsSMAX1 and D53 for degradation by the 26S proteasome. In turn, this releases OsSMAX1- and D53-mediated repression of the activity of their interacting transcription factors, thus regulating the expression of downstream target genes. SL signals specifically regulate shoot branching, and KL signals might specifically regulate root colonization by AM fungi. The specificity of the output of SL signals and KL signals is likely determined by transcription factors that interact specifically with D53 or OsSMAX1. It is possible that some common transcription factors both interact with D53 and are responsible for the expression of a subset of common genes, which could respond to both KL signaling and SL signaling. During skotomorphogenesis, KL and SL signals act through the D14L-D3-OsSMAX1 complex and D14-D3-D53, respectively, and act in parallel and/or additively to trigger the expression of their specific or commonly regulated downstream genes, which leads to the inhibition of rice mesocotyl elongation.

Our work on OsSMAX1 and the work on SMXL2 revealed that downstream genes may be subject to regulation by both the SL signaling pathway and the karrikin pathway (Wang et al., 2020). These findings may help to explain why it is difficult to investigate the expression of downstream gene in response to rac-GR24 treatment. We showed that the karrikin pathway is not involved in the regulation of the branching of rice shoots (Figure 6). By contrast, the SL-mediated regulation of mesocotyl elongation requires the function of OsSMAX1 (Figure 5). The transcript levels of D53 and OsSMAX1 were revealed to be controlled by negative feedback regulation (Supplemental Figure 16G). In addition, the expression of D53 and OsSMAX1 could respond to both SL and KL signals in a D14- and D14L-dependent manner (Figures 9E and 9F). These results indicated a possible mechanism of crosstalk between the SL pathway and KL pathway. Notably, the expression of a MAX1 homolog in rice (Os01g0700900), which is referred to as both LOC_Os01g50520 and LOC01g50530 (Cardoso et al., 2014; Zhang et al., 2014), was downregulated in the d14, d14l, d14 d14l, and d3 mutants as well as in D53m-OE and OsSMAX1m-OE transgenic seedlings but was upregulated in the Ossmax1, Ossmax1 d14, and Ossmax1 d3 seedlings (Figure 7E). Therefore, OsSMAX1 negatively regulates SL biosynthesis, suggesting an alternative mechanism of the crosstalk between the SL pathway and KL pathway (Choi et al., 2020).

The stereospecificity of SL analogs is crucial for their recognition and for the induction of downstream activities (Scaffidi et al., 2014). Although the protein structures of both D14 and D14L have been resolved, it remains unclear how their stereospecific recognition of substrates is achieved. In SL signaling, ligand binding or hydrolysis-induced conformational changes of receptor is essential for the interactions between receptor and its downstream signaling partners (Yao et al., 2016; Shabek et al., 2018; Burger et al., 2019). Previous reports have indicated that D14/AtD14 require an intact catalytic triad for signal perception and interaction with D3/MAX2 (Hamiaux et al., 2012; Jiang et al., 2013; Kagiyama et al., 2013; Zhao et al., 2013a; Waters et al., 2015b; de Saint Germain et al., 2016; Yao et al., 2016; Shabek et al., 2018; Xu et al., 2018). However, when the catalytic triad of D14L was mutated, there was no significant change in the interaction of D14L with OsSMAX1 (Supplemental Figure 9). In Arabidopsis, both AtD14 and KAI2 hydrolyze rac-GR24 (Zhao et al., 2013a; Toh et al., 2014). However, rice D14, but not D14L, hydrolyzes rac-GR24 (Zhao et al., 2013a). In addition, AtD14 is degraded in response to SL in a MAX2-dependent manner (Chevalier et al., 2014), whereas KAI2 is degrade in response to karrikin in a MAX2-independent manner (Waters et al., 2015a). Surprisingly, recent work indicated that the turnover of SMAX1 could be mediated by KAI2 and MAX2 and may be subjected to a MAX2-independent regulation (Khosla et al., 2020). These results suggest that the recognition of ligands might differ between D14 and D14L. It is possible that KAR₁ or GR24^ent-5DS might not be the direct ligands perceived by D14L, which is stimulated by an unidentified endogenous KL signal. In this scenario, KAR₁ or GR24^ent-5DS would exhibit activity in the in planta assay, but not in the in vitro assays.

Given that GR24^ent-5DS is a nonnatural compound that can mimic karrikin activity and that carlactone requires MAX1-dependent activity for repressing shoot branching and limited activity for inhibiting hypocotyl elongation (Scaffidi et al., 2013), it has been proposed that endogenous KL molecules from a carlactone-independent pathway may be responsible for the regulation of seedling development (Conn and Nelson, 2016; Burger et al., 2019). KAI2 and AtD14 homologs are present throughout seed plants (Bythell-Douglas et al., 2017), and SMXL proteins can be traced back to moss (Bennett and Leyser, 2014). The evolution of distinct ligand recognition by KAI2 and AtD14 homologs indicated that, compared with the AtD14 signaling pathway, the KAI2-mediated signaling pathway is more ancient (Bythell-Douglas et al., 2017). KAR₁ and GR24^ent-5DS were found to be recognized by different groups of KAI2-like proteins in moss (Burger et al., 2019). Diversification of KAI2 proteins in Brassica tournefortii conferred differential responses to karrikins (Sun et al., 2020). SL perception through KAI2 homologs occurs in parasitic plants of the Orobanchaceae family, which seems to have evolved relatively more recently (Conn et al., 2015; Toh et al., 2015; Tsuchiya et al., 2015; Xu et al., 2018). When a KAI2 homolog from Selaginella moellendorffii was expressed in Arabidopsis kai2, it could complement the seedling and leaf development phenotypes of kai2, but the plants were not responsive to SL or karrikin treatment (Waters et al., 2015b). SMXL2 can be degraded in response to both karrikin signaling and SL signaling (Wang et al., 2020). When ShHTL receptors from Striga were expressed in Arabidopsis, the seeds could germinate without the requirement of gibberellin in a MAX2- and SMAX1-dependent manner (Bunsick et al., 2020). These results support that coevolution of receptor–target pairs was involved in the establishment of distinct signaling pathways for SL and karrikin, which have been selected in adaptive evolution of land plants (Waters et al., 2017).

Identifying the endogenous KL signal that is recognized by D14L, resolving the structure of the D14-D3-D53 complex, identifying the transcriptional factors that interact with D53/SMXL proteins, and characterizing the D14L-D3-OsSMAX1 complex of different species will provide insight into the molecular mechanism underlying how these signals are perceived and passed to downstream signaling components and into how stereospecific recognition of substrates is achieved and evolves. It has been suggested that D14L acts in parallel with the OsCERK1-mediated common symbiosis signaling pathway (CSSP) in the regulation of early colonization events of AM fungus–plant symbiosis (Gutjahr et al., 2015b; Chiu et al., 2018). However, the genes involved in CSSP downstream of OsCERK1 were found to be upregulated in Ossmax1 (Choi et al., 2020). It seems that the karrikin signaling pathway plays an essential role in coordinating with CSSP and SL signaling during AM symbiosis. To reveal how karrikin signaling mediates the communication between plants and AM fungi requires intensive further investigation. The signaling paradigm of karrikin strongly suggests that some unidentified chemicals play roles in AM symbiosis (Choi et al., 2020). The recent discovery of zaxinone synthase (Wang et al., 2019) and β-cyclocitral (Dickinson et al., 2019) and their roles in rice root development and stress adaptation indicated that apocarotenoid-derived small molecules might be candidates for the KL molecule that regulates mesocotyl development and/or symbiosis between plant and AM fungi.

In rice, coleoptile and mesocotyl elongation protects the shoot apical meristem during seedling emergence from the soil. The mesocotyl immediately stops expanding upon exposure to light, while the coleoptile still grows even when mesocotyl elongation ceases (Takahashi, 1972). The relative length of the mesocotyl and coleoptile are affected by developmental and environmental factors (Hu et al., 2010), which could be used as a model system to investigate the molecular mechanism underlying crosstalk between developmental and environmental factors in monocots. Elongation of the mesocotyl and coleoptile has been observed in the light-signaling-deficient mutants phyA and cpm1 (Takano et al., 2001; Biswas et al., 2003; Haga et al., 2005) and JA-deficient mutants hebiba and cpm2 (Riemann et al., 2003, 2013). SL and karrikin negatively regulate rice mesocotyl elongation in the dark. In Arabidopsis, SL inhibition of hypocotyl elongation depends on both cryptochrome and phytochrome signaling (Jia et al., 2014). Notably, KAI2 is required for normal seedling photomorphogenesis but does not affect seedling morphology in darkness, suggesting that the regulation of seedling development mediated by karrikin signaling may act downstream of light signaling.

The cell length of the lower parts of the mesocotyls of d10 and d14 is similar to that of the wild type, while the cells of the lower parts of the mesocotyl of d3 are shorter than those of the wild type. However, there are more cells in the mesocotyls of d10 and d14 than in those of the wild type, and there are more cells in the mesocotyl of d3 than in d10 and d14 (Hu et al., 2010). These results indicated that the SL-mediated inhibition of mesocotyl elongation occurs via negative regulation of cell division. SL has been suggested to be involved in crosstalk with cytokinin and BRs to regulate cell division in the mesocotyl (Sun et al., 2018). Since the mesocotyl length in d14 d14l is similar to that in d3, disruption of both the SL and karrikin pathways resulted in downregulation of the expression of OsTCP5 and GY1, and loss of OsSMAX1 function reversed the downregulated expression of OsTCP5 and GY1 as well as the elongated-mesocotyl phenotypes of d10, d14, d14l, and d3 (Figure 5). It is possible that karrikin may control mesocotyl development by negatively controlling the action of cytokinin as well as JA to regulate cell division and cell elongation.

It has been reported that JA biosynthesis in seedlings is induced by red light (Haga and Iino, 2004) and that exogenous application of JA can suppress the elongation of the mesocotyls of dark-grown d mutants (Hu et al., 2010). The facts that karrikin signaling regulates the expression of light-responsive genes (such as LP2) and that JA synthesis genes can be induced by red light (Haga and Iino, 2004) in rice seedlings may help explain why the elongation of the mesocotyls of the SL- and KL-deficient mutants was not observed under light conditions. Previous work suggested that the elongation of the mesocotyls of hebiba was enhanced under darkness as well as under red light because of deficiency in JA biosynthesis (Riemann et al., 2003). However, the large fragment deletion found in hebiba included not only a JA biosynthesis gene but also D14L (Riemann et al., 2013; Gutjahr et al., 2015b). The defective colonization by AM fungi in hebiba is determined by the function of D14L and not by the deficiency in JA biosynthesis (Gutjahr et al., 2015a, 2015b). Thus, further investigation of the interaction between karrikin signaling and light signaling as well as the role of JA in mesocotyl development may help elucidate the molecular mechanism underlying crosstalk between karrikin signaling and light signaling as well as other phytohormone signaling pathways in plants.

Because of limited water resources and labor costs, there has been a trend toward rice cultivation by direct seeding instead of transplanting in rice-growing areas in recent years (Feng et al., 2017). Breeding elite rice varieties suitable for direct seeding requires the improvement of various early developmental traits, including fast and uniform seed germination, high seedling vigor, early tillering capability, strong root growth, and lodging resistance. Recent identification of natural variation in GY1 (Xiong et al., 2017) and OsGSK2 (Sun et al., 2018) highlights the practical application of beneficial alleles of phytohormone-related genes that regulate mesocotyl growth. Our work has indicated that karrikin signaling and SL signaling may act upstream of cytokinin-, JA-, and BR-mediated pathways in the regulation of belowground development of rice seedlings (Supplemental Figures 16A to 16C). Improving our understanding of the molecular mechanism through which karrikin and SL regulate mesocotyl development and mining the natural variation in karrikin signaling and SL signaling components as well as OsSMAX1-regulated downstream genes will contribute to the development of new elite rice varieties suitable for direct seeding.

METHODS

Plant Materials and Growth Conditions

Plants were grown in paddy fields at the experimental stations of the China National Rice Research Institute at Fuyang, Zhejiang Province, in the summer or at Lingshui, Hainan Province in the winter. Oryza sativa cv Nipponbare, mutants (d10, d17, d14, d3, and d53), and the pACT:D53m-GFP transgenic plants in the Nipponbare genetic background were the same as previously described by Jiang et al. (2013). d14l and Ossmax1 mutants in the Nipponbare genetic background were generated via CRISPR/Cas9 genome editing (Song et al., 2017). The mutation sites of the d14l and Ossmax1 mutants are shown in Supplemental Figures 3 and 15, respectively. d14 d14l, d53 d14l, Ossmax1 d14, Ossmax1 d14l, Ossmax1 d3, and Ossmax1 d10 double mutants were generated by crossing. Plant height and tiller number were recorded at the heading stage.

Mesocotyl Length Measurements

For phenotype determination, unshelled rice seeds were sterilized with sodium hypochlorite for 1 h and then germinated for 20 h at 30°C. Uniform seeds were selected, placed on solid medium consisting of 0.9% (w/v) agar (Sigma-Aldrich), and then grown in the dark at 30°C for 7 d. For seedlings grown in the light, the growth conditions consisted of 30°C with a 16-h-light/8-h-dark cycle via fluorescent lamps with a light intensity of 75 μE m⁻² s⁻¹. For chemical treatment, seedlings were grown on 0.9% (w/v) agar plates with the indicated concentrations of chemicals (KAR₁, rac-GR24, GR24^5DS, and GR24^ent-5DS). Except where specified, at least 20 seedlings were used for mesocotyl length measurements that were performed manually or via ImageJ.

Plasmid Construction

For yeast two-hybrid analyses, the coding DNA sequences (CDSs) of OsSMAX1, D14, and D53 were amplified and cloned into both pGBKT7 and pGADT7 between EcoRI and BamHI. To construct the mutated OsSMAX1-binding domain with the GKT motif deletion (Figure 3C) or EAR mutation (Supplemental Figure 18A), two fragments of each indicated OsSMAX1 mutation were individually amplified and then cloned together into pGBKT7 via an In-Fusion HD cloning kit (Clontech). The CDSs of D53L, DL1, DL2, DL3, DL4, DL5, and DL6 were subsequently amplified and cloned into pGBKT7. The different truncated CDSs of OsSMAX1 and D53 were amplified using the indicated primers and then cloned into pGBKT7, and the coding regions of D14L, D14L2, and D14L3 were amplified and cloned into pGADT7. To construct different site-specific mutants at potential catalytic triad sites of D14L, two fragments of a given mutated D14L were individually amplified and then cloned together into pGADT7 via an In-Fusion HD cloning kit. To construct TPRs-activation domain (AD), TPR1(1-300)-AD, TPR2(1-300)-AD, and TPR3(1-300)-AD, the DNA sequences encoding the full-length CDS of TPR2 and the 1 to 300 N-terminal amino acid residues of each TPR were amplified and cloned into pGADT7 via an In-Fusion HD cloning kit.

To construct vectors for expressing recombinant proteins in Escherichia coli, the full-length, truncated, or mutated CDSs of the genes of interest were amplified using the primers listed in Supplemental Table 1 and then cloned into pDONR221 via BP reactions (Invitrogen) to generate entry vectors, followed by LR reactions (Invitrogen) with pET-55-DEST, pET-57-DEST, pET-60-DEST, and pDEST-His-Maltose binding protein (MBP) destination vectors. TPR2(1-600) was cloned into a pET-55-DEST destination vector, D14 and D14L were cloned into a pET-57-DEST destination vector, D3 was cloned into a pET-60-DEST destination vector, and D53 and OsSMAX1 were cloned into a pDEST-His-MBP destination vector. E. coli strain BL21 was used for expression vectors and for purification of tagged proteins. For coimmunoprecipitation (Co-IP) transient expression vectors, pBeacon-EGFP and pBeacon-HA were used to transiently express proteins with the respective recombinant tags in rice protoplasts (Wang et al., 2015). The genes were cloned into pDONR221 by BP reactions to create entry plasmids, followed by LR reactions with appropriate transient expression vectors.

To generate a PUC57-StrigoQuant-like backbone vector containing the cauliflower mosaic virus-35S promoter, Renilla luciferase-2A-firefly luciferase coding sequences, and the nopaline synthase terminator, the corresponding elements were synthesized by GENEWIZ. Multiple cloning sites were located between the 2A and firefly luciferase sequences. For the proteins of interest, the full-length CDSs of the corresponding genes were amplified and cloned into PUC57-StrigoQuant using an In-Fusion HD cloning kit through the SpeI and HpaI double digestion sites. All primers used for vector construction are listed in Supplemental Table 1.

Plant Transformation

The oligos corresponding to the guide RNA sequences of D14L and OsSMAX1 were synthesized and cloned into the CRISPR/Cas9 genome editing vector as previously described by Song et al. (2017). A pCAMBIA1300-pUbi-GFP-3XFlag binary vector was derived from pCAMBIA1300 and was generated by inserting the maize Ubi promoter, nopaline synthase terminator, and GFP-3×Flag CDS using an In-Fusion HD cloning kit between KpnI and HindIII. The BamHI and PmlI sites were introduced between Ubi promoter and GFP-3×Flag CDS. To construct pUbi:OsSMAX1-GFP-3XFlag and pUbi:OsSMAX1m-GFP-3XFlag overexpression vectors, the OsSMAX1 and OsSMAX1m coding regions were amplified and then cloned into pCAMBIA1300-pUbi-GFP-3XFlag between BamHI and PmlI using an In-Fusion HD cloning kit. For pOsSMAX1:OsSMAX1m-GFP-Flag, the OsSMAX1 promoter was amplified and used to replace the pUbi promoter between KpnI and BamHI. For pAct:OsSMAX1-Flag and pAct:OsSMAX1m-Flag, the tag-fused OsSMAX1 and OsSMAX1m coding regions were amplified and digested by ApaI and SpeI, after which they were ligated into an AHLG vector (Actin promoter: HA-linker-GFP-Nos terminator expression cassette in pCAMBIA 1300). All overexpression plasmids were confirmed by sequencing (Sangon Biotech). The binary vectors for D14L and OsSMAX1 knockout or for overexpression of OsSMAX1 and OsSMAX1m were transformed into Nipponbare by the Agrobacterium-mediated transgenic method (Hiei et al., 1994). All primers used for vector construction are listed in Supplemental Table 1.

Yeast Two-Hybrid Assays

To detect protein interactions, plasmids were transformed into Y2H-Gold Yeast (Clontech) cells using a lithium acetate and polyethylene glycol–mediated protocol. After incubation for 2 d at 28°C in synthetic defined (SD)/-Leu-Trp medium (Clontech), the yeast clones were transferred to SD/-Leu-Trp-His-Ade medium (Clontech) with or without chemical treatment. To assay the effects of rac-GR24, GR24^5DS, GR24^ent-5DS, and KAR₁ on the interaction of OsSMAX1 and D14L, the yeasts cotransformed with OsSMAX1-binding domain and D14L-AD were diluted to different concentrations and grown for 60 h on SD/-Leu-Trp-His-Ade medium with the indicated chemicals at 20 μM. rac-GR24 and KAR₁ were obtained from Chiralix (Nijmegen), and the GR24 stereoisomers GR24^5DS and GR24^ent-5DS were prepared as described by Wang et al. (2020).

Protein Expression and Purification

E. coli strain BL21 (CWBIO) containing the fusion protein expression plasmids was grown in Luria-Bertani broth with the corresponding antibiotics to an OD₆₀₀ of 0.6 to 0.8 and then continuously cultured at 28°C overnight together with 0.4 mM isopropyl-β-d-thiogalactopyranoside. The collected cells were lysed using a JN-02C low-temperature high-pressure biomixer (JNBIO) and then centrifuged at 8000g for 15 min. The supernatants were subsequently collected for recombinant protein purification. For GST-tagged recombinant protein purification, the supernatants and glutathione magnetic agarose beads (BeaverBeads GSH; Beaver) were incubated together for 3 h. After washing three times with 1× PBS, the recombinant protein eluent was removed from the beads by competitive binding with 20 mM reduced glutathione. For MBP-tagged recombinant protein purification, the supernatants were incubated with amylose resin (BioLabs) for 3 h and then washed three times with wash buffer (20 mM Tris-HCl, pH 7.4, 200 mM NaCl, 1 mM EDTA, and 10 mM 2-mercaptoethanol), after which the recombinant protein eluent from amylose resin was obtained by competitive binding with 10 mM maltose. The His-MBP-OsSMAX1 protein was concentrated using Amicon Ultra-4 centrifugal filters (Ultracel-100K; Millipore). For His-tagged protein purification, the supernatants were incubated with Ni Sepharose 6 Fast Flow (GE Healthcare) for 3 h. After washing three times with PBS, the recombinant protein eluent from Sepharose was washed with a gradient of different concentrations (10, 20, 50, and 250 mM) of imidazole in elution buffer consisting of 50 mM NaH₂PO₄, pH 7.8, 300 mM NaCl, and 0.2% (v/v) Tween 20. All steps were performed at 4°C or on ice.

Pull-Down Assays

To detect protein interactions, His-tagged proteins and GST-tagged proteins (3 μg) or GST (6 μg) were incubated together at 4°C in a 700-μL incubation buffer (1× protease inhibitor cocktail [Roche], MG132 [Millipore], and 0.5% [v/v] Triton X-100 [Sigma-Aldrich] in PBS). After incubation for 1 h, glutathione magnetic agarose beads (BeaverBeads GSH; 50 μL) were added, followed by incubation at 4°C for 2 h. The beads were then washed three times with wash buffer (0.1% Triton X-100 in PBS) and then eluted with 30 μL of SDS-PAGE sample buffer for SDS-PAGE and immunoblotting; Nus protein (6 μg) was used as a negative control. To detect the interaction between His-MBP–tagged proteins and His-Nus–tagged proteins, 2 μg of His-MBP–tagged recombinant proteins was bound to amylose resin, after which it and 4 μg of His-Nus–tagged proteins were incubated together at 4°C in 700 μL of binding buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10 μM EDTA, 1× protease inhibitor cocktail, 50 µM MG132, and 1% [v/v] Triton X-100) for 2 h. The amylose resin was washed three times with wash buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% [v/v] Triton X-100); Nus (6 μg) and MBP (6 μg) proteins were used as negative controls. The proteins were detected using the following monoclonal antibodies: mouse anti-His (Transgen Biotech, cat. no. HT501-02) at a 1:5000 dilution; mouse anti-GST (Transgen Biotech, cat. no. HT601-02) at a 1:10,000 dilution; anti-NusA (Abnova, cat. no. MAB0049-M01) at a 1:10,000 dilution; and mouse anti-MBP (Transgen Biotech, cat. no. HT701-02) at a 1:10,000 dilution. To assay the effects of rac-GR24, GR24^5DS, GR24^ent-5DS and KAR₁ on the interaction of OsSMAX1 and D14L, the indicated concentrations (20 or 50 μM) of the chemicals were added to the incubation buffer.

Co-IP Assays

Protoplasts generated from the young stems of 3-week-old rice seedlings grown under light were transformed with transient expression plasmids as previously described by Bart et al. (2006). After incubation of the protoplasts at 28°C for 12 h, protein extraction buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10 μM EDTA, 1× protease inhibitor cocktail, 50 μM MG132, and 1% [v/v] Triton X-100) was added, after which the mixture was vortexed (IKA Vortex). The lysate was centrifuged at 14,000g for 15 min at 4°C, after which the supernatant was collected for Co-IP experiments. In accordance with the manufacturer's instructions, 25 μL of GFP-Trap coupled to agarose (ChromoTek) was added to 1 mL of extracted protein and incubated for 3 h at 4°C. After incubation, the agarose was washed three times with wash buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% [v/v] Triton X-100), followed by elution with 30 μL of SDS-PAGE sample buffer for SDS-PAGE electrophoresis and immunoblot analysis. GFP protein was used as a negative control in each set of experiments. The proteins were detected by mouse anti-GFP monoclonal antibodies (Roche, cat. no. 11814460001) at a 1:5000 dilution or by anti-HA antibodies (Roche, cat.no. 11867423001) at a 1:5000 dilution.

Dual Luciferase Reporter Assays in Protoplasts

To determine the stability of D53, the respective PUC57-StrigoQuant-like reporter vectors of D53 and D53m were transformed into rice protoplasts from the wild-type, d3, d14, d14l, and d14 d14l plants as previously described by Bart et al. (2006). After incubation at 28°C for 11 h, 1 μM rac-GR24 was added, followed by incubation for another 1 h. The Renilla and firefly luciferase activities were then measured using a Dual Luciferase Reporter Assay System (Promega). To test the stability of OsSMAX1, the respective PUC57-StrigoQuant-like reporter vectors of OsSMAX1 and OsSMAX1m were transformed into rice protoplasts from the wild type, d3, d14, d14l and d14 d14l as described. After incubation at 28°C for 12 h, the Renilla and firefly luciferase activities were measured using a Dual Luciferase Reporter Assay System. FF:REN was used as a metric of protein stability.

Antibody Preparation

To detect endogenous OsSMAX1, the cDNA fragment encoding OsSMAX1 (571 to 951 amino acids) was cloned into pDONR221 via BP reactions (Invitrogen) and then cloned into a pDEST-His-MBP destination vector by LR reactions (Invitrogen) to express the MBP-OsSMAX1 (571- to 951–amino acid) fusion protein. The fusion protein was subsequently expressed in E. coli BL21 and then purified with amylose resin. The MBP-OsSMAX1 (571 to 951 amino acids) purified protein was used as an antigen to produce polyclonal anti-OsSMAX1 antibodies in rabbits. The specificity of the anti-OsSMAX1 antibodies was confirmed by immunoblot analysis of OsSMAX1 knockout plants (Figure 5C) and OsSMAX1m-OE transgenic plants (Supplemental Figure 10).

Endogenous and Transgenic Protein Detection

The aerial parts of 7-d-old light-grown rice seedlings were ground in liquid nitrogen and incubated in extraction buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 μM EDTA, 1% [w/v] SDS, and 1× protease inhibitor cocktail) for 20 min at 4°C. After centrifugation for 10 min at 14,000g, SDS-PAGE sample buffer was added to the supernatant for SDS-PAGE electrophoresis and immunoblot analysis. Endogenous OsSMAX1 protein was detected by anti-OsSMAX1 antibodies at a 1:4000 dilution. For endogenous D53, the shoot bases of 3-week-old plants were used, after which the detection was performed using anti-D53 polyclonal antibodies (Jiang et al., 2013). To determine the effects of the chemicals on endogenous OsSMAX1 degradation, the aerial parts of seedlings treated with 20 μM KAR₁, KAR₂, GR24^5DS, or GR24^ent-5DS and grown in the light for 7 d were collected for immunoblotting. rac-GR24 and KAR₁ were obtained from Chiralix, and GR24^5DS and GR24^ent-5DS were prepared as described (Wang et al., 2020). KAR₂ was obtained from Daqin. Except where specified, the abundance of OsSMAX1 or OsSMAX1m proteins from light-grown 7-d-old seedlings of different OsSMAX1 and OsSMAX1m transgenic seedlings was measure by anti-Flag antibodies (DDDDK monoclonal antibody; MBL, cat. no. M185-3L) at a 1:10,000 dilution. To determine the stability of OsSMAX1 in response to rac-GR24, GR24^5DS, GR24^ent-5DS, and KAR₁ treatment, calli of OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag transgenic plants were cultured on solid medium for 6 d at 28°C and then transferred to liquid media containing a specific chemical compound at a specific concentration (Jiang et al., 2013). After treatment with 10 μM rac-GR24, GR24^5DS, GR24^ent-5DS, and KAR₁, the calli were collected at specified time points. The OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag proteins collected from the calli were detected by anti-Flag antibody. In the ubiquitin assay, the OsSMAX1-GFP-Flag and OsSMAX1m-GFP-Flag proteins were detected by anti-GFP monoclonal antibody (Roche) at a 1:5000 dilution and monoclonal anti-ubiquitin antibody (Zhao et al., 2013b). The rice ACTIN protein was detected by anti-ACTIN antibodies (Abmart, cat. no. M20009) at a 1:4000 dilution, as an internal control.

Subcellular Localization

To observe the subcellular localization of OsSMAX1 in plants, the primary roots of 5-d-old seedlings of Ubi:OsSMAX1-GFP-Flag and Ubi:OsSMAX1m-GFP-Flag transgenic plants were used for analysis. The GFP signal was detected using a confocal microscope (SP8; Leica) at an excitation wavelength of 488 nm. Propidium iodide was used for cell wall staining, and the excitation wavelength was 540 nm.

RNA Sequencing and Data Analysis

Dark-grown 7-d-old seedlings were collected for total RNA extraction. Eight seedlings were pooled as one sample for each indicated genotype. Three biological replicates were used for RNA extraction and RNA sequencing. Total RNA was extracted using TRI-Reagent (MRC) according to the manufacturer's protocol. RNA-seq was performed on an Illumina NovaSeq platform. The reads were aligned to the reference genome MSU version_7.0 and the gene model annotation file (http://rice.plantbiology.msu.edu/) using TopHat2. Gene expression profiling and the DEGs identification were performed with cufflinks. The genes whose expression was upregulated in the mutants and transgenic plants of all three biological replicates are provided in Supplemental Data Sets 3 and 4. A Venn diagram and heatmap were constructed by R using the VennDiagram package (https://cran.r-project.org/web/packages/VennDiagram) and the pheatmappackage (https://cran.r-project.org/web/packages/pheatmap), respectively. Gene ontology and Kyoto Encyclopedia of Genes and Genomes analyses were performed by the online tool KOBAS (http://kobas.cbi.pku.edu.cn).

RT-qPCR Analysis

All the aerial parts of dark-grown 7-d-old seedlings were sampled for total RNA extraction. The total RNA was extracted using TRI-Reagent according to the manufacturer's instructions. The total RNA (0.5 to 0.7 μg) was used for first-strand cDNA synthesis using ReverTra Ace qPCR Master Mix in conjunction with gDNA Remover (Toyobo). RT-qPCR analysis was performed using gene-specific primers (Supplemental Table 2) on a CFX Connect Real-Time System (Bio-Rad) according to the manufacturer's instructions. The 10-μL reaction volume consisted of 1 μL of diluted cDNA, primers at 0.3 μM, and 5 μL of ChamQ Universal SYBR qPCR Master Mix (Vazyme). The rice ACTIN gene was used as an internal control.

Accession Numbers

The RNA-seq information has been deposited in the BioProject ID PRJNA553596. Sequence data from this article can be found in the GenBank/EMBL libraries under the following accession numbers: D10 (LOC_Os01g54270); D17 (LOC_Os04g46470); D14 (LOC_Os03g10620); D14L (LOC_Os03g32270); D14L2 (LOC_Os05g51240); D14L3 (LOC_Os01g41240); D3 (LOC_Os06g06050); D53 (LOC_Os11g01330); D53L (LOC_Os12g01360); OsSMAX1 (LOC_Os08g15230); DL1 (LOC_Os02g26600); DL2 (LOC_Os02g33460); DL3 (LOC_Os02g54720); DL4 (LOC_Os04g23220); DL5 (LOC_Os04g33980); DL6 (LOC_Os11g05820); TPR1 (LOC_Os01g15020); TPR2 (LOC_Os08g06480); TPR3 (LOC_Os03g14980); OsTCP5 (LOC_Os02g51280); OsGSK2 (LOC_Os05g11730); GY1 (LOC_Os01g67430); LP2 (LOC_Os02g40240).

Supplemental Data

• Supplemental Figure 1. Accumulation of D53 promotes mesocotyl elongation in the dark.

• Supplemental Figure 2. D14 subfamily proteins in rice.

• Supplemental Figure 3.  D14L knockout plants generated by CRISPR/CAS9.

• Supplemental Figure 4. Seedlings of different mutants in the light and in the dark.

• Supplemental Figure 5. Phylogenetic tree of D53/SMAX1 family proteins.

• Supplemental Figure 6. Interaction of D14 homologs with D53 family proteins in yeast two-hybrid assay.

• Supplemental Figure 7. Interaction of OsSMAX1 with a D14 homolog in yeast two-hybrid assay.

• Supplemental Figure 8. Yeast two-hybrid analysis of OsSMAX1 domain interaction with D14L.

• Supplemental Figure 9. Mutation of the catalytic triad of D14L did not affect its interaction with OsSMAX1.

• Supplemental Figure 10. Validation of the anti-OsSMAX1 antibody.

• Supplemental Figure 11. StrigoQuant-like reporters transiently expressed in protoplasts.

• Supplemental Figure 12. OsSMAX1 accumulation promotes mesocotyl elongation in the dark.

• Supplemental Figure 13. Interaction between D14L and OsSMAX1 is not enhanced by 20 µM Karrikin treatment.

• Supplemental Figure 14. Interaction between D14L and OsSMAX1 is not enhanced by 50 µM Karrikin treatment.

• Supplemental Figure 15.  OsSMAX1 knockout plants generated by CRISPR/CAS9.

• Supplemental Figure 15. Relative expression levels of OsSMAX1-regulated genes.

• Supplemental Figure 17. OsSMAX1 is not involved in regulation of shoot branching.

• Supplemental Figure 18. The EAR motif of OsSMAX1 is essential for the interaction between OsSMAX1 and TPR2.

• Supplemental Figure 19. Heatmap of OsSMAX1 positively regulated genes in wild-type, mutant, and transgenic plants.

• Supplemental Figure 20. Mesocotyl elongation in response to different concentration of chemicals in the dark.

• Supplemental Figure 21. Relative expression levels of OsSMAX1-regulated genes in response to GR24 stereoisomers.

• Supplemental Table 1. Primers used for vector construction.

• Supplemental Table 2. Primers used for RT-qPCR.

• Supplemental Data Set 1. Text file of the alignment used for the phylogenetic analysis of D14 proteins shown in Supplemental Figure 2.

• Supplemental Data Set 2. Text file of the alignment used for the phylogenetic analysis of D53/SMAX1 proteins shown in Supplemental Figure 5.

• Supplemental Data Set 3. Downregulated genes in mutants and overexpression plants compared to the wild type.

• Supplemental Data Set 4. Upregulated genes in mutants and overexpression plants compared to the wild type.

• Supplemental Data Set 5. Statistical Analyses.

DIVE Curated Terms

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

• CCD8 Gramene: AT4G32810

• CCD8 Araport: AT4G32810

 

• mesocotyl AmiGo: PO:0020037

 

• MAX2 Gramene: At2g42620

• MAX2 Araport: At2g42620

 

• SMAX1 Gramene: At5g57710

• SMAX1 Araport: At5g57710

 

• SMXL2 Gramene: At4g30350

• SMXL2 Araport: At4g30350

Acknowledgments

We thank Shengben Li (Nanjing Agricultural University) and Zhixi Tian (Institute of Genetics and Developmental Biology, Chinese academy of Sciences) for critical reading. We thank Qi Xie (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) for providing the antibody of ubiquitin. We thank High-Performance Computing Centers at Agricultural Genomics Institute at Shenzhen, Chinese Academy of Agricultural Sciences, for bioinformatics support and also thank Jiangsu Collaborative Innovation Center for Modern Crop Production for support. This work was supported by founding from National Key Research and Development Program of China (grant 2016YFD0101801); National Natural Science Foundation of China (31501384 and 31201004); and the Science, Technology and Innovation Commission of Shenzhen Municipality (grants JCYJ20170303154319837, JCYJ20170412155447658, and KQJSCX2018323140312935).

AUTHOR CONTRIBUTIONS

J.L. and G.X. conceived the project. J.Z., K.H., L.J.Z., G.X., and J.L. designed the experiments. J.Z., K.H., L.J.Z., L.W., S.K., J.D., L.Y.Z., L.X.Z., Z.T., X.M., J.H., Y.Z., and Q.W. performed experiments. J.Z., K.H., L.J.Z., M.Q., D.Z., B.W., C.P., Q.W., Q.Q., Y.W., G.X., and J.L. analyzed data. J.Z., G.X., and J.L. wrote the article.

Author notes

1

Address correspondence to [email protected].

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: [email protected].

Articles can be viewed without a subscription.

¹These authors contributed equally to this work.

www.plantcell.org/cgi/doi/10.1105/tpc.20.00123