Strigolactone promotes cytokinin degradation through transcriptional activation of CYTOKININ OXIDASE/DEHYDROGENASE 9 in rice

Significance

Strigolactone plays a vital role in plant growth and development, but its response genes remain to be identified. In this study, we found that cytokinin content is markedly increased in the strigolactone signaling mutant d53, and that OsCKX9, which encodes a cytokinin oxidase to catalyze the degradation of cytokinin, functions as a primary strigolactone-responsive gene to regulate rice tillering, plant height, and panicle size, likely via a secondary response gene, OsRR5, which encodes a cytokinin-inducible rice type-A response regulator, demonstrating that strigolactone regulates rice shoot architecture through enhanced cytokinin catabolism by modulating OsCKX9 expression.

Abstract

Strigolactones (SLs), a group of terpenoid lactones derived from carotenoids, are plant hormones that control numerous aspects of plant development. Although the framework of SL signaling that the repressor DWARF 53 (D53) could be SL-dependently degraded via the SL receptor D14 and F-box protein D3 has been established, the downstream response genes to SLs remain to be elucidated. Here we show that the cytokinin (CK) content is dramatically increased in shoot bases of the rice SL signaling mutant d53. By examining transcript levels of all the CK metabolism-related genes after treatment with SL analog GR24, we identified CYTOKININ OXIDASE/DEHYDROGENASE 9 (OsCKX9) as a primary response gene significantly up-regulated within 1 h of treatment in the wild type but not in d53. We also found that OsCKX9 functions as a cytosolic and nuclear dual-localized CK catabolic enzyme, and that the overexpression of OsCKX9 suppresses the browning of d53 calli. Both the CRISPR/Cas9-generated OsCKX9 mutants and OsCKX9-overexpressing transgenic plants showed significant increases in tiller number and decreases in plant height and panicle size, suggesting that the homeostasis of OsCKX9 plays a critical role in regulating rice shoot architecture. Moreover, we identified the CK-inducible rice type-A response regulator OsRR5 as the secondary SL-responsive gene, whose expression is significantly repressed after 4 h of GR24 treatment in the wild type but not in osckx9. These findings reveal a comprehensive plant hormone cross-talk in which SL can induce the expression of OsCKX9 to down-regulate CK content, which in turn triggers the response of downstream genes.

Strigolatones (SLs), a group of carotenoid-derived plant hormones, are pivotal regulators of plant growth and development (1). Intensive research has revealed a conserved SL perception mechanism in plants. In rice (Oryza sativa L.), α/β hydrolase receptor DWARF 14 (D14) functions as both an SL receptor and an enzyme that binds and cleaves SL to trigger its conformational change (1). D14 interacts with F-box protein DWARF 3 (D3) and nuclear-localized repressor DWARF 53 (D53) in a SL-dependent manner, causing the ubiquitination and subsequent degradation of D53 within minutes (2, 3). D53 has been reported to act as a key component in the assembly of a repressor-corepressor-nucleosome complex via recruitment of the transcriptional corepressor TOPLESS (TPL) and TPL-related proteins (4). In the d53 mutant, the dominant form of D53 is resistant to SL-induced degradation, and the SL signaling pathway is constitutively inhibited. Recently in rice, Ideal Plant Architecture 1 (IPA1) was identified as a D53-targeted downstream transcription factor that mediates SL-regulated tiller development and SL-induced D53 expression (5). However, besides the feedback regulation of D53, the SL-responsive genes in rice are largely unknown but have long been speculated (1), especially the primary response genes and their downstream secondary response genes.

Cytokinins (CKs) constitute a class of plant hormones that play important roles in axillary bud initiation and outgrowth, as well as in various aspects of plant growth (6). CK and SL have shown antagonistic or combined/synergistic effects on several development processes. In pea (Pisum sativum), axillary bud outgrowth is induced by CK but inhibited by SL (7), and a similar phenomenon is seen in rice mesocotyl elongation (8). Repression of lateral root development by SL is abolished in CK-signaling mutants in Arabidopsis thaliana (9). It is speculated that SL and CK may share specific targets, or that one may regulate the key components in the metabolism or signaling pathway of the other. In pea, BRANCHED 1 (BRC1) is activated by SL and repressed by CK (7, 10); however, in rice, FINE CULM 1 (FC1), an orthologous gene of BRC1, is repressed by CK but insensitive to SL (11). The transcriptional regulation of SL and the cross-talk between SL and CK are not fully conserved between monocotyledonous and dicotyledonous plants, which remains to be elucidated.

In Arabidopsis, the type-A Arabidopsis response regulator (ARR) genes are rapidly induced by CK (12) and are considered negative-feedback regulators of CK signaling (13–17). Different type-A ARRs have been extensively studied in Arabidopsis for their critical roles in diverse developmental processes (14, 15, 18–21). Comparatively, 13 type-A rice response regulators (OsRRs) have been annotated in the rice genome, but knowledge of their function is limited (22). Inhibition of CK signaling by OsRR1 and OsRR2 are known to be involved in rice crown root development (23, 24). Overexpression of OsRR3, OsRR5, and OsRR6 can each inhibit the shoot regeneration from rice calli (25, 26). OsRR6-overexpressing transgenic plants exhibit dwarfism with small roots and inflorescences (26). These data suggest that OsRR1, OsRR2, OsRR3, OsRR5, and OsRR6 are functionally negative regulators of CK signaling in rice.

In this study, we identify one primary SL-responsive gene, OsCKX9, which is rapidly induced by rac-GR24 in a D53-dependent manner and is likely responsible for the increased CK levels in d53. OsCKX9 encodes a nuclear/cytosolic dual-localized CK catabolic enzyme. Both disruption and overexpression of OsCKX9 could significantly increase tiller number, with reduced plant height and panicle size. Moreover, we show that rac-GR24 represses the expression of OsRR5, a CK-inducible rice type-A response regulator, and this repression is released in OsCKX9 loss-of-function mutants, indicating that the CK signaling pathway is regulated by SL via OsCKX9. Collectively, these findings demonstrate that SLs directly active CK catabolism to regulate shoot architecture in rice.

Results

Shoot Bases of d53 Have Elevated CK Levels.

The local CK biosynthesis in tiller nodes is known to play an important role in bud elongation in rice (27, 28). To examine whether CK levels are altered in SL-related dwarf (d) mutants, we harvested the shoot bases of 20-d-old seedlings of the wild type (WT) and d53, and measured the endogenous concentrations of trans-zeatin (tZ) and N⁶-(Δ²-isopentenyl)adenine (2iP), two major forms of CK in rice (29), as well as their riboside derivatives tZR and iPR. The results showed significantly increased levels of tZ and 2iP in shoot bases of d53, ∼2.0-fold and 1.3-fold higher, respectively, than those in WT (Table 1). For their riboside derivatives, iPR levels showed no significant difference between d53 and WT, whereas the level of tZR in d53 was ∼1.4-fold higher than that in WT (Table 1). These results imply that the repression of the SL signaling pathway could result in an elevated CK content in rice shoot bases.

Table 1.

CK content in the shoot bases of the WT and d53

	CK content, pg/mg fresh weight
	tZ	tZR	2iP	iPR
WT	0.342 ± 0.038	0.651 ± 0.045	0.053 ± 0.006	0.258 ± 0.033
d53	0.681 ± 0.080**	0.902 ± 0.090**	0.067 ± 0.005**	0.202 ± 0.045ns

Values are mean ± SD, n = 6. **P < 0.01, Student’s t-test; ns, no significant difference.

OsCKX9 Acts as a Primary SL-Responsive Gene.

Previous research has shown that the CK levels are mainly determined by the balance between their biosynthesis and catabolism, which involves three biosynthetic enzymes—adenosine phosphate isopentenyltransferase (IPT), cytochrome P450 monooxygenase 735A (CYP735A), and the CK-activating enzyme LONELY GUY (LOG)—and one catabolic enzyme, cytokinin oxidase/dehydrogenase (CKX) (30). Since CK levels are increased in the shoot bases of d53, we asked whether SL could directly regulate the expression of CK biosynthesis or catabolism genes by quantifying the expression levels of all 34 CK metabolism genes. We found that OsCKX9, a CK catabolism gene, was significantly induced by rac-GR24 within 1 h (Fig. 1A and SI Appendix, Fig. S1), indicating that OsCKX9 might be a primary SL-responsive gene.

Fig. 1.

Expression levels of CK metabolism genes on treatment with rac-GR24, 6-BA, tZ, or 2iP. (A) Fold change in CK metabolism gene expression levels after 1 h treatment with 5 μM rac-GR24 compared with mock. **P < 0.01, Student’s t test. (B) Expressions of eight OsCKXs on treatment with 5 μM 6-BA. Results are presented relative to mock. Values are mean ± SD, n = 3. **P < 0.01 compared with mock, Student’s t test; ns, no significant difference. (C) Expression levels of eight OsCKXs on treatment with 5 μM tZ or 2iP. Results are presented relative to mock. Values are mean ± SD, n = 3. **P < 0.01 compared with mock, Student’s t test; ns, no significant difference.

D53 is a primary SL-responsive gene that is significantly down-regulated in rice SL-related d mutants, including SL biosynthesis mutants d10, d17, and d27 and SL signaling mutants d3, d14, and d53 (3). Similarly, we found that OsCKX9 is significantly down-regulated in all these d mutants compared with WT (cv. Nipponbare) (Fig. 2A). We found that the OsCKX9 protein level was remarkably decreased in d53 (Fig. 2B), in agreement with the decreased transcript level (Fig. 2A). We further examined the time course response of OsCKX9 transcripts to rac-GR24 in shoot bases of d27 and d53 and found that OsCKX9 is gradually up-regulated within 12 h in d27 (Fig. 2C) but shows no response in d53 (Fig. 2D). The foregoing results indicate that OsCKX9 is a primary SL-responsive gene, and that the response of OsCKX9 to SL requires the intact function of D53.

Fig. 2.

OsCKX9 is a primary strigolactone-responsive gene. (A) Expressions of OsCKX9 in the WT and strigolactone-related dwarf mutants. Results are presented relative to WT. Values are mean ± SD, n = 3. **P < 0.01, Student’s t test. (B) OsCKX9 protein levels in WT and d53. α-OsCKX9 and α-HSP90, anti-OsCKX9 and anti-HSP90 antibodies. (C and D) Expression levels of OsCKX9 on treatment with 5 μM rac-GR24 in d27 (C) and d53 (D). Results are presented relative to mock at 0 h. Values are mean ± SD, n = 3. **P < 0.01, Student’s t test; ns, no significant difference. (E) GUS staining with or without treatment with 5 μM rac-GR24 in OsCKX9_pro:GUS rice lines. (Left) Root-shoot junction of 20-d-old plants. (Right) Bottom view below the roots of heading-stage plants. (Scale bars: 1 mm in Left, 1 cm in Right.)

We further generated transgenic lines expressing OsCKX9pro:GUS and treated the transgenic lines at the seedling and heading stages with rac-GR24 for 12 h. Compared with the mock treatment, we observed enhancement in GUS activity in both rac-GR24–treated young seedlings and heading-stage plants (Fig. 2E), suggesting that OsCKX9 expression is subject to SL regulation at different growth stages.

The rapid up-regulation of different CKXs by CK within 2 h was observed in Arabidopsis, maize, and rice, forming a negative feedback loop of the CK responses (31–33). However, this raises the question of whether OsCKX9 expression is regulated by both SL and CK. Therefore, we treated rice seedlings with 6-benzylaminopurine (6-BA, a synthetic CK analog), tZ, or 2iP and quantified the expression levels of all 11 OsCKX genes in rice by qPCR. We found that 7 of the 8 detectable OsCKX genes (OsCKX1–OsCKX5, OsCKX8, and OsCKX11) were significantly induced by 6-BA, tZ, or 2iP, but the OsCKX9 gene was not (Fig. 1 C and D). These results suggest that in CK catabolism, OsCKX9 responds to SL rather than to CK.

OsCKX9 Encodes a Functional CKX Enzyme.

To examine the CK degradation activity of OsCKX9, the GST-OsCKX9 fusion protein was expressed in Escherichia coli cells and purified for in vitro assays (SI Appendix, Fig. S2). The CKX activity assays were carried out using tZ and 2iP as substrates, with 2,3-dimetoxy-5-methyl-1,4-benzoquinone as the electron acceptor at pH 5.0 and 2,6-dichlorophenol indophenol as the electron acceptor at pH 7.5 using a modified endpoint method (34). The GST-OsCKX9 fusion protein exhibited relatively high activity for tZ at pH 5.0 (23.3 µkat g⁻¹), moderate activity for tZ at pH 7.5 (3.3 µkat g⁻¹), moderate activity for 2iP at pH 5.0 (3.7 µkat g⁻¹), and relatively low activity for 2iP at pH 7.5 (0.7 µkat g⁻¹) (Fig. 3A).

Fig. 3.

OsCKX9 encodes a functional CKX. (A) In vitro CKX activity assay of recombinant GST-OsCKX9. Values are mean ± SD (n = 3). (B) Expression levels of OsCKX9 in WT, Ubi_pro:OsCKX9, and Act_pro:OsCKX9-GFP rice seedlings at the mature stage. Results are presented relative to WT. Values are mean ± SD, n = 3. **P < 0.01, Student’s t test. (C) Gross (Upper) and panicle (Lower) phenotypes of WT, Ubi_pro:OsCKX9, and Act_pro:OsCKX9-GFP at the mature stage. (D) Subcellular localization of OsCKX9-eGFP fusion protein in rice protoplasts. SV40NLS-mCherry is used to label the nucleus. (E) Phenotypes of calli derived from the WT, Ubi_pro:OsCKX9/WT, WT cultured with 6-BA, Ubi_pro:OsCKX9/WT cultured with 6-BA, d53, and Ubi_pro:OsCKX9/d53. (Scale bars: 10 cm in C, Upper; 3 cm in C, Lower; and 10 µm in D.)

In Arabidopsis, overexpression of CKXs leads to CK-deficient phenotypes, including reduced shoot development, dwarfism, late-flowering, enhanced root growth, and reduced fertility (35). In rice, ren1-D, a dominant mutant of OsCKX4, exhibits decreases in plant height, tiller number, and primary and secondary branch number per panicle and an increase in crown root number (36), while the effects of other OsCKX genes remain elusive. Therefore, we generated two types of transgenic rice lines, one with Ubi_pro:OsCKX9 expressing OsCKX9 driven by the maize ubiquitin 1 promoter and the other with Act_pro:OsCKX9-GFP expressing an OsCKX9-GFP fusion gene driven by the rice actin 1 promoter. The OsCKX9 transcript levels were up-regulated by 106-fold in Ubi_pro:OsCKX9 and 56-fold in Act_pro:OsCKX9-GFP plants (Fig. 3B). Compared with WT (cv. Nipponbare), Act_pro:OsCKX9-GFP plants formed shorter culms, more tillers, and fewer primary and secondary branches per panicle and had a lower setting rate, similar to the morphological alterations caused by the CK deficiency in the CKXs-overexpressing plants in Arabidopsis (35). More severe defects were also observed in Ubi_pro:OsCKX9 plants (Fig. 3C and SI Appendix, Table S1), possibly resulting from a higher expression level of OsCKX9 in Ubi_pro:OsCKX9. Taken together, these results indicate that OsCKX9 encodes a functional CKX enzyme in rice.

To identify the subcellular localization of the OsCKX9 protein, we analyzed the Act_pro:OsCKX9-GFP fluorescence pattern in roots and found that the OsCKX9-GFP proteins are expressed throughout the cytosol and accumulate mainly in nuclei (SI Appendix, Fig. S3). To confirm this, we transiently expressed a carboxyl-terminal eGFP fusion of OsCKX9 in rice protoplasts and found OsCKX9-eGFP protein accumulation predominantly in the cytosol and nuclei (Fig. 3D). These findings demonstrate that OsCKX9 is a cytosolic/nuclear dual-localized protein.

In plant tissue culture systems, a moderate concentration of CK is essential for shoot regeneration (37), and 6-BA treatment can result in the browning of calli (Fig. 3E). Overexpression of OsCKX9 largely prevented the callus browning in WT treated with 6-BA (Fig. 3E), possibly because OsCKX9 degrades the endogenous CK in the callus and thus decreases the total amount of CK, including 6-BA. Inconsistent with the finding of elevated CK levels in d53, the calli derived from d53 showed apparent browning (Fig. 3E). To test whether the browning of the d53 calli is indeed caused by an increase in CK content, we generated transgenic d53 calli expressing Ubi_pro:OsCKX9 and found that the browning phenotype of d53 calli could be largely rescued by overexpressing OsCKX9 (Fig. 3E). Consistent with this, the expression levels of OsRR1 and OsRR2 were significantly increased in d53 calli, which can be suppressed by overexpression of OsCKX9 (SI Appendix, Fig. S4). These findings suggest that OsCKX9 is a functional CKX enzyme, and that its overexpression could rescue the browning of d53 calli caused by an elevated CK level.

OsCKX9 Regulates Rice Plant Architecture.

To further understand the biological function of OsCKX9, we investigated the spatial expression patterns of OsCKX9 in various rice organs and found that OsCKX9 is widely expressed in all the tissues at the heading stage, with the highest level in shoot bases (Fig. 4A). We then generated a loss-of-function mutant osckx9 using CRISPR/Cas9 technology in the Nipponbare background. Sequence analysis revealed a 1-bp insertion at the first exon of OsCKX9 in osckx9, which results in a premature stop codon (Fig. 4B). Indeed, absence of the OsCKX9 protein was confirmed by the immunoblotting analysis in osckx9 (Fig. 4C). Compared with WT, osckx9 displayed altered architecture in shoot and panicle (Fig. 4 D and G). The tiller number was significantly increased (Fig. 4E), while the plant height was slightly but significantly reduced in osckx9 plants (Fig. 4F). For panicle morphology, osckx9 showed reduced panicle length, primary and secondary branches per panicle, and grain number per panicle (Fig. 4 H–K). These results demonstrate that OsCKX9 regulates rice plant architecture at both the vegetative and reproductive stages.

Fig. 4.

Characterization of the osckx9 mutant. (A) OsCKX9 transcript levels in various organs at the heading stage. Results are presented relative to the expression level in roots. Values are mean ± SD; n = 3. (B) The mutation site in the osckx9 coding region and its amino acid changes. (C) OsCKX9 protein levels in the WT and osckx9. α-OsCKX9 and α-HSP90, anti-OsCKX9 and anti-HSP90 antibodies. Molecular mass markers are shown on the left. (D) Phenotypes of WT and osckx9 at the mature stage. (E and F) Comparisons of tiller number (E) and plant height (F) in D. (G) Panicles of WT and osckx9 at the mature stage. (H–K) Comparisons of panicle length (H), primary branches per panicle (I), secondary branches per panicle (J), and grain number per panicle (K) in G. Values are mean ± SD, n = 18 (**P < 0.01; Student’s t test). Bars = 10 cm in D and 5 cm in G.

OsCKX9 Works in the SL Signaling Pathway.

In rice, SL deficiency leads to increased shoot branching and dwarfism. The osckx9 plant exhibits similar phenotypes to d27 but has less severe branching and dwarf stature than d27 (Fig. 5 A–C). To further test whether OsCKX9 functions in the SL signaling pathway, we generated and characterized the osckx9 d27 double mutant. As shown in Fig. 5 A–E and SI Appendix, Fig. S5, compared with d27, the osckx9 d27 double mutant has similar tiller number, plant height, panicle length and branches, and grain number per panicle; however, compared with osckx9 and WT, the double mutant is markedly increased in tiller number but decreased in other features, indicating that the OsCKX9 functions in the SL signaling pathway as a downstream primary SL-responsive gene to regulate rice shoot architecture.

Fig. 5.

Phenotypic comparison of osckx9 and the strigolactone pathway mutants. (A) Phenotypes of the WT, osckx9, d27, and osckx9 d27 at the heading stage. (B and C) Comparison of tiller number (B) and plant height (C) in A. (D) Panicles of WT, osckx9, d27, and osckx9 d27 at the mature stage. (E) Comparison of grain number per panicle in D. Values are means ± SD, n = 18 (**P < 0.01; Student’s t test). (F) Expressions of OsCKX9 upon 5 µM rac-GR24 treatment in ipa1-10. Results are presented relative to mock at 0 h. Values are means ± SD, n = 3 (**P < 0.01; ns, no significant difference; Student’s t test). (G) Responses of rice seedlings to 10 μM rac-GR24 treatment. Bars = 10 cm in A and G and 2 cm in D.

Based on the recent discoveries that IPA1 functions as a direct downstream transcription factor of D53 and that its transcriptional induction by SL depends on IPA1 in rice (5), we asked whether the OsCKX9 induction by SL also requires IPA1 by examining the responses of OsCKX9 transcripts to rac-GR24 in ipa1-10, a loss-of-function mutant of IPA1 (5). As shown in Fig. 5F, on rac-GR24 treatment, the OsCKX9 transcripts were significantly increased in ipa1-10 plants, indicating that the SL-induced activation of OsCKX9 is not mediated by IPA1. We also found that rac-GR24 could repress the tiller number of osckx9 (Fig. 5G) but was unable to inhibit the bud outgrowth of ipa1-10 (5), indicating that IPA1 is a crucial regulator in SL-induced rice tillering suppression. Taking together, these findings suggest that OsCKX9 works downstream in the SL signaling pathway but independently of IPA1.

Reduced CK Content by SLs Leads to Decreased OsRR5 Expression.

The expression levels of most type-A OsRRs are increased in leaves and roots after 6-BA treatment for 2 h, five of which— OsRR1, OsRR2, OsRR3, OsRR5, and OsRR6— have been reported to act as negative regulators in CK signaling to regulate various traits in rice (23–26). As SL could up-regulate the expression of OsCKX9 in shoot bases (Figs. 1B and 2C), we wondered whether and which type-A OsRRs are in the downstream of CK catabolism regulated by SL. We first examined the expression levels of these five type-A OsRRs in the shoot bases of rice seedlings treated with 6-BA and found that OsRR1, OsRR2, OsRR5, and OsRR6 were strongly induced by CK (SI Appendix, Fig. S6), but transcripts for OsRR3 were undetectable. We then compared the expression levels of these four type-A OsRRs in the shoot bases of SL-related d mutants with the WT and found that OsRR1, OsRR2, and OsRR6 showed no significant differences among these materials at transcript levels, but the expression of OsRR5 was significantly higher in all the SL-related d mutants than in WT (Fig. 6A).

Fig. 6.

OsRR5 is a secondary strigolactone-responsive gene. (A) Expression levels of four OsRRs in the WT and strigolactone mutants. Results are presented relative to WT. Values are mean ± SD, n = 3. **P < 0.01, Student’s t test. (B and C) Expression levels of OsRR5 on treatment with 5 μM rac-GR24 in WT (B) and osckx9 (C). Results are presented relative to mock at 0 h. Values are mean ± SD, n = 3. **P < 0.01, Student’s t test; ns, no significant difference. (D) Proposed model of the OsCKX9-mediated strigolactone signaling pathway. In the shoot bases of rice, perception of strigolactone (SL) leads to degradation of D53, which in turn releases the repression of OsCKX9 to degrade cytokinin (CK) and then triggers the reduction of OsRR5.

We tested whether the transcripts of OsRRs respond to SL using D53 as a positive control (SI Appendix, Fig. S7A). Consistent with the results in SL-related d mutants, OsRR5 expression was significantly repressed in shoot bases of WT after a 4-h treatment with rac-GR24 (Fig. 6B), while OsRR1, OsRR2, and OsRR6 are all insensitive to rac-GR24 within 12 h (SI Appendix, Fig. S7 B–D). Furthermore, we found that the rac-GR24 repressed expression of OsRR5 was released in shoot bases of the osckx9 mutants (Fig. 6C), while this treatment resulted in pronounced induction of D53 in shoot bases of the osckx9 mutants (SI Appendix, Fig. S7E). These findings suggest that OsRR5 is one of the downstream components of the CK catabolism regulated by SL-induced OsCKX9.

Discussion

Interactions between plant hormones are considered essential for their intact functions. SL and CK have been shown to function antagonistically in pea bud elongation (7, 10) and rice mesocotyl elongation in darkness (8); however, compared with the well-studied interactions between auxin and CK and between auxin and SL, relatively little is known about the interaction between CK and SL, especially in monocotyledonous plants. Here we report a comprehensive hormonal cross-talk between SL and CK in the shoot bases of rice. In response to rac-GR24, D53, the repressor of the SL signaling pathway, is degraded in minutes, leading to rapid transcriptional activation of the CK catabolism gene OsCKX9 within 1 h and subsequent down-regulation of the CK-responsive gene OsRR5 within 4 h, probably due to the OsCKX9-dependent degradation of CK (Fig. 6D). Consistent with these results, OsCKX9 is largely down-regulated in all six SL-related d mutants tested in rice, resulting in an increased CK content. These findings demonstrate that SL could promote CK degradation through transcriptional activation of OsCKX9 in rice. In Arabidopsis, AtCKX1- and AtCKX3-overexpressing plants exhibit multiple developmental alterations, including retarded shoot development and increased branch number, possibly resulting from decreased auxin content due to reduced auxin-producing tissues (35). However, in rice, overexpression and knockout of OsCKX9 did not show opposite phenotype as expected, but both resulted in increased tiller number, reduced plant height, and decreased panicle size, suggesting that the phenotypes cannot be explained simply by the total CK content, instead pointing to a fine-tuned spatial and temporal-specific control of CK content in rice development.

Compared with other plant hormones that could trigger rapid responses of thousands of genes at the transcriptional level, surprisingly, significantly fewer genes are transcriptionally regulated after SL treatment over a short time frame (38, 39). One possibility is that significant responses to SL occur only in a very specific temporospatial manner. In rice, D53 is the only reported gene that is rapidly up-regulated after SL treatment (2, 3). Here we demonstrate that OsCKX9 is another primary SL-responsive gene downstream of D53 in rice, the expression of which is significantly induced within 1 h after SL treatment (Fig. 1B). Consistent with the finding of elevated CK levels in d53 (Table 1), it is likely that SL may constitutively regulate CK content by affecting OsCKX9 homeostasis. Consistent with this, significantly elevated CK content at node 2 from the top of D10-RNAi plants has been reported (40). In addition, OsCKX9 appears to be a special member of the OsCKX family, because its expression is insensitive to CK (Fig. 1 C and D) but rapidly induced by SL (Fig. 1B). It should be mentioned that the interactions of auxin with CK and auxin with SL have been established and proven to be critical for plant development, such as that the SL can inhibit auxin transport capacity and auxin inhibit CK biosynthesis (41). Therefore, along with the rapid induced activation of OsCKX9, SL may also affect CK content through hormonal cross-talk with auxin.

The transcriptional regulation of plant hormone action has been described as primary and secondary responses, with the former indicating a rapid protein synthesis-independent induction and the latter referring to an alteration depending on the products of the primary response genes (42, 43). OsRR5 is significantly repressed by SL treatment after 4 h, and, more importantly, this repression depends on the intact function of the primary SL-responsive gene OsCKX9 (Fig. 6B). Therefore, OsRR5 functions as a secondary SL-responsive gene, which suggests that a secondary response program can occur as within 4 h, and that in such time, SL treatment could induce the transcriptional and translational responses of OsCKX9, the product of which could degrade CK and inhibit the OsRR5 expression.

Different type-A OsRRs have overlapping/differential expression patterns in various organs (44), implying their diverse functions in specific organs. In shoot bases of rice, the expression of OsRR5 was significantly up-regulated in all six SL-related d mutants (Fig. 6A) and inhibited by SL treatment (Fig. 6B), but the expression levels of OsRR1, OsRR2, and OsRR6 were not (Fig. 6A and SI Appendix, Fig. S7 B–D), indicating that OsRR5 specifically participates in SL-controlled CK responses in shoot bases. Corresponding to this, in green seedlings, the expression level of OsRR5 is higher than that of the other type-A OsRR genes (44). CK is involved in various aspects of development, and osckx9 plants display obvious alterations in plant height, tiller number, and panicle morphology (Fig. 4 D–K). It is quite possible that different type-A OsRRs may work in different organs downstream of SL-induced CK catabolism. A comprehensive dissection of the downstream genes will shed light on the regulation of different traits by SL-induced CK catabolism.

Identification of the transcription factors downstream of D53 is critical for understanding the SL signaling pathway. IPA1 has been reported as a direct downstream transcription factor regulated by D53, and the transcription activation activity of IPA1 can be suppressed by D53 (5), similar to AUXIN/INDOLE-3-ACETIC ACID INDUCIBLE proteins in the auxin-signaling pathway (45) and JASMONATE-ZIM DOMAIN proteins in the jasmonate-signaling pathway (46, 47). Although rac-GR24 can repress the tiller number of osckx9, it is unable to inhibit the bud outgrowth of ipa1-10 (5), suggesting that IPA1 is a crucial regulator of SL-induced rice tillering suppression. A number of transcription factors have been revealed to be regulated by these hormonal repressors for their diverse functions (48, 49), but whether other transcription factors are regulated by D53 remains to be elucidated. Our finding that SL-induced activation of OsCKX9 is dependent on D53 (Fig. 2D) but independent of IPA1 (Fig. 5F) demonstrates that other transcription factors besides IPA1 may function downstream of D53 in the SL signaling pathway, and these factors need to be identified for a better understanding of the SL signaling pathway.

Materials and Methods

Detailede information on plant growth, CK measurement, gene expression analysis, vector construction and plant transformation, GUS staining, antibody preparation, CKX activity assays, chemical treatment, and subcellular localization assays are provided in SI Appendix, Materials and Methods. The primers used in this study are listed in SI Appendix, Table S2.