OsSHI1 Regulates Plant Architecture Through Modulating the Transcriptional Activity of IPA1 in Rice

OsSHI1 represses the DNA binding activity of IPA1 to regulate plant architecture in rice.

Abstract

Tillering and panicle branching are important determinants of plant architecture and yield potential in rice (Oryza sativa). IDEAL PLANT ARCHITECTURE1 (IPA1) encodesSQUAMOSA PROMOTER BINDING PROTEIN-LIKE14, which acts as a key transcription factor regulating tiller outgrowth and panicle branching by directly activating the expression of O. sativa TEOSINTE BRANCHED1 (OsTB1) and O. sativa DENSE AND ERECT PANICLE1 (OsDEP1), thereby influencing grain yield in rice. Here, we report the identification of a rice mutant named shi1 that is characterized by dramatically reduced tiller number, enhanced culm strength, and increased panicle branch number. Map-based cloning revealed that O. sativa SHORT INTERNODES1 (OsSHI1) encodes a plant-specific transcription factor of the SHI family with a characteristic family-specific IGGH domain and a conserved zinc-finger DNA binding domain. Consistent with the mutant phenotype, OsSHI1 is predominantly expressed in axillary buds and young panicle, and its encoded protein is exclusively targeted to the nucleus. We show that OsSHI1 physically interacts with IPA1 both in vitro and in vivo. Moreover, OsSHI1 could bind directly to the promoter regions of both OsTB1 and OsDEP1 through a previously unrecognized cis-element (T/GCTCTAC motif). OsSHI1 repressed the transcriptional activation activity of IPA1 by affecting its DNA binding activity toward the promoters of both OsTB1 and OsDEP1, resulting in increased tiller number and diminished panicle size. Taken together, our results demonstrate that OsSHI1 regulates plant architecture through modulating the transcriptional activity of IPA1 and provide insight into the establishment of plant architecture in rice.

INTRODUCTION

As a major staple crop worldwide, the yield of rice is multiplicatively determined by three major agronomic traits: panicle number, grain number per panicle, and grain weight. The numbers of panicles and grains per panicle are mainly determined by the ability of the plant to produce tillers, the primary and secondary branches of the panicle. These traits are also important determinants of overall plant architecture in rice (Oryza sativa; Wang and Li, 2008; Xing and Zhang, 2010; Wang et al., 2018a).

The formation of a tiller can be divided into two consecutive steps: tiller bud formation and outgrowth, both of which are regulated by elaborate crosstalk among hormonal, developmental, and environmental factors (Domagalska and Leyser, 2011). Recent molecular and genetic studies have revealed much insight into the control of bud formation in both dicots and monocots. Rice MONOCULM1 (MOC1) encodes a transcription factor of the GRAS family orthologous to Lateral suppressor (LS) of tomato (Solanum lycopersicum) and LATERAL SUPPRESSOR (LAS) of Arabidopsis (Arabidopsis thaliana; Schumacher et al., 1999; Greb et al., 2003; Li et al., 2003). The rice moc1 mutant has no tillers due to the defect in axillary meristem formation, which is similar to the phenotype of the tomato ls mutant, implying that LS/LAS/MOC1 plays a conserved role in maintaining the potential for axillary meristem initiation in both monocots and dicots. Further studies demonstrated that Tillering and Dwarf 1 (also named TE) acts as a component of the APC/C^TAD1/TE E3 ligase to modulate tiller bud formation by facilitating the degradation of MOC1 (Lin et al., 2012; Xu et al., 2012).

Studies of various branching mutants, such as the dwarf mutants in rice (Arite et al., 2007, 2009; Gao et al., 2009; Lin et al., 2009; Jiang et al., 2013; Zhou et al., 2013), more axillary growth mutants in Arabidopsis (Sorefan et al., 2003; Booker et al., 2004, 2005; Stirnberg et al., 2007), ramosus mutants in pea (Pisum sativum; Sorefan et al., 2003; Johnson et al., 2006) and decreased apical dominance mutants in petunia (Petunia hybrida; Snowden et al., 2005) revealed an essential role of the plant hormone strigolactones (SLs) in bud outgrowth. Deficiencies in both SL production and signaling lead to excessive outgrowth of the axillary buds (Smith and Li, 2014). Recent studies have demonstrated that SLs repress tiller outgrowth through promoting the degradation of a central repressor protein, D53 (Jiang et al., 2013; Zhou et al., 2013).

SQUAMOSA PROMOTER BINDING PROTEIN-box genes (SBP-box genes) encode plant-specific transcription factors that share a highly conserved DNA binding domain, the SBP domain. At least 18 putative SQUAMOSA PROMOTER BINDING PROTEIN-LIKE genes exist in the rice genome, and most of them are regulatory targets of microRNA (miRNA) OsmiR156 (Xie et al., 2006). Several members of the SPL family, OsSPL7, OsSPL13, OsSPL14, OsSPL16, and OsSPL17, have been shown to regulate vegetative and inflorescence architecture in rice (Jiao et al., 2010; Miura et al., 2010;Wang et al., 2012, 2015a, 2015b; Liu et al., 2016; Si et al., 2016; Wang and Wang, 2017). Among these factors, OsSPL14,also known as IDEAL PLANT ARCHITECTURE1 (IPA1) or WEALTHY FARMER’S PANICLE, is the best studied so far (we use “IPA1” hereafter). It is predominantly expressed in the shoot apex at both the vegetative and reproductive stages. A C-to-T Single Nucleotide Polymorphism that escapes OsmiR156 targeting or increasing IPA1 expression via epigenetic regulation confers an ideal plant architecture to rice, including reduced tiller number, stronger culm, enlarged panicle and, ultimately, enhanced grain yield (Jiao et al., 2010; Miura et al., 2010). IPA1 binds directly to the promoter regions of several important regulators of rice plant architecture, including O. sativa TEOSINTE BRANCHED1 (OsTB1), O. sativa DENSE AND ERECT PANICLE1 (OsDEP1), LONELY GUY, SLENDER RICE1, and PIN-FORMED (Lu et al., 2013), as well as WRKY45 to promote both yield and immunity in rice (Wang et al., 2018b). OsTB1, a transcription factor of the TEOSINTE BRANCHED1/CYCLOIDEA/PCF family, is also referred to as FINE CULM1 and was initially identified as a counterpart of maize (Zea mays; ZmTB1), which is involved in inhibiting lateral branching in maize (Doebley et al., 1995; Lukens and Doebley, 2001; Dong et al., 2017). Later studies confirmed that OsTB1 also acts to suppress axillary buds outgrowth in rice (Takeda et al., 2003; Minakuchi et al., 2010). OsDEP1 encodes the γ-subunit of the heterotrimeric G protein complex. Gain-of-function mutation of OsDEP1 results in increased primary and secondary branches and number of grains per panicle and consequently, increased grain yield (Huang et al., 2009). In addition, recent studies implicated OsDEP1 in regulating nitrogen-use efficiency and grain size determinacy in rice (Sun et al., 2014, 2018; Liu et al., 2018).

The Arabidopsis SHI family contains 10 members referred to as SHORT INTERNODES (SHI), STYLISH1 and 2 (STY1, STY2), SHI-RELATED SEQUENCE3 to SHI-RELATED SEQUENCE8, and LATERAL ROOT PRIMORDIUM1. This gene family encodes plant-specific transcription factors characterized by a conserved zinc finger domain of Cys/His consensus sequence C₃HC₃H and a unique IGGH domain, which was named after the four highly conserved residues within this region (Fridborg et al., 2001). Genetic studies of shi-related mutants implied that SHI family members play indispensable roles for gynoecium and leaf development and photomorphogenesis in Arabidopsis, probably by regulating auxin homeostasis or expression of HY5, BBX21, and BBX22 (Smith and Fedoroff, 1995; Fridborg et al., 1999, 2001; Kuusk et al., 2002, 2006; Sohlberg et al., 2006; Eklund et al., 2010; Baylis et al., 2013; Yuan et al., 2018). In addition, SHORT AWN2 and SIX-ROWED SPIKE2, encoding two SHI family transcription factors, regulate awn elongation, pistil morphology, and inflorescence patterning in barley (Hordeum vulgare; Yuo et al., 2012; Youssef et al., 2017). However, the precise roles and significance of this gene family in regulating rice development and plant architecture establishment are still not well characterized.

In this study, we characterize a rice mutant named shi1, which has significantly reduced tiller number, enhanced culm strength, and increased panicle branch numbers. Molecular cloning revealed that the mutant defects are caused by deletion of the SHI family gene OsSHI1. We show that OsSHI1 regulates tillering and panicle branching through physical interacting with IPA1 and modulating its transcriptional activity on downstream target genes.

RESULTS

Characterization of the shi1 Mutant Phenotype

In a screen for regulators of plant architecture in rice, we identified the shi1 mutant from a ⁶⁰Co-γ irradiation-induced mutant population of the indica cultivar 9311. Compared with the wild type, shi1 exhibited dramatically reduced tiller number from the 4th-leaf stage to the mature stage (Figures 1A to 1H). Histological analysis revealed that axillary bud initiation was largely normal; however, the outgrowth of axillary buds was obviously delayed in the shi1 mutant (Supplemental Figure 1). Notably, shi1 had a more compact plant architecture with significantly reduced tiller number at the reproductive developmental stage, compared with the wild-type plant (Figures 1H and 1I). Panicles of shi1 were also more compact and erect with slightly increased primary branch number and substantially increased secondary branch and spikelet numbers (Figures 1J to 1N). However, due to the trade-off between spikelet number and grain size and various defects in floral organ development, the grain size, 1,000-grain weight, and the seed setting rate of shi1, were significantly reduced (Supplemental Figures 2A to 2C). Leaves of shi1 were shorter but wider, especially for the flag leaves (Supplemental Figures 2D to 2F), and more dark-green with increased chlorophyll contents (Supplemental Figure 2G). More strikingly, the culm diameters of shi1 were greatly increased due to the increased parenchyma tissue layers and vascular bundles (Supplemental Figures 2H to 2M). These observations suggest that OsSHI1 plays a pleiotropic role in regulating plant architecture establishment in rice.

Figure 1.

Phenotypic Characterization of the shi1 Mutant.

(A) to (G) Tillering phenotypes of wild type (9311) and shi1 at 2 weeks after germination (WAG) (A), 3 WAG (B) and (C), 4 WAG (D), 5 WAG (E), 6 WAG (F), and 7 WAG (G). (C) is the enlarged image of the dotted box in (B). White arrows indicate the tillers. Bars = 2 cm (A) and (B), 1 cm (C), 5 cm (D) to (G). WT, wild type.

(H) Plant architectures of wild type and shi1 at the grain-filling stage. Bar = 20 cm. WT, wild type.

(I) Statistical analysis of tiller numbers of wild type and shi1. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 15, **P < 0.01). WT, wild type.

(J) Panicle morphologies of wild type and shi1 at the mature stage. Bar = 5 cm. WT, wild type.

(K) Panicle branch architectures of wild type and shi1 with grains removed. Bar = 5 cm. WT, wild type.

(L) to (N) Statistical analysis of the primary branch numbers (L), secondary branch numbers (M), and spikelet numbers per panicle (N) of wild type and shi1. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 10, *P < 0.05, **P < 0.01). WT, wild type.

Map-Based Cloning of OsSHI1

Genetic analysis of two F₂ populations derived from the reciprocal crosses between shi1 and wild type (9,311) indicated that the shi1 phenotype is controlled by a single recessive nuclear locus, given that the numbers of wild-type and mutant individuals approximately fit the expected 3:1 ratio (Supplemental Table 1).

To identify the causal gene, an F₂ population was generated from a cross between shi1 and 02428 (O. sativa ssp japonica). Linkage analysis revealed that the OsSHI1 locus is associated with the simple sequence repeat markers RM107 and RM189 on the long arm of chromosome 9. Subsequent fine-mapping using 7,547 progeny from the F₂ population delimited the OsSHI1 locus to a 50-kb region with four predicted Open Reading Frames (ORFs; Figure 2A). Sequencing and PCR analysis revealed that a ∼18-kb genomic region covering ORF2 is deleted in the shi1 mutant (Figures 2A and 2B), but no DNA sequence alterations were found in the promoter and coding regions of ORF1, ORF3, and ORF4. Further, RT-PCR analysis and immunoblot analysis using anti-OsSHI1–specific polyclonal antibodies confirmed no expression of ORF2 in the shi1 mutant (Figures 2C and 2D). These results suggest that ORF2 likely corresponds to OsSHI1.

Figure 2.

Map-Based Cloning of OsSHI1.

(A) OsSHI1 was narrowed down to an ∼50-kb region of chromosome 9 containing four ORFs. A genomic region of ∼18 kb is deleted in shi1. The markers, BACs and numbers of recombinants are indicated. WT, wild type. Rec, recombinants. BACs, Bacterial Artifical Chromosomes.

(B) PCR amplifications by primer pairs located at the flanking boundaries of the deleted region. No PCR product could be amplified from wild type due to the large size of genomic region using the primer pair F48 and R56. WT, wild type.

(C) RT-PCR analysis of the four ORFs. Expression of ORF2 was not detected in shi1. Actin2 was used as an endogenous control. WT, wild type.

(D) Protein levels of OsSHI1 in the panicle tissues of wild type and shi1 detected by immunoblot using anti-OsSHI1–specific polyclonal antibodies. HSP antibody was used as the loading control. Molecular masses of proteins (kDa) are shown on the left. WT, wild type.

(E) Plant phenotypes of wild-type, shi1, and two independent complementation lines before the heading stage. Bar = 10 cm. WT, wild type.

(F) Tiller numbers of the complemented transgenic lines compared with the wild-type and mutant levels. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 5, **P < 0.01). WT, wild type.

To verify whether ORF2 is indeed responsible for the shi1 phenotype, an ∼5-kb genomic fragment of ORF2 (full-length ORF2 including a 3-kb promoter, two exons, one intron, and a 400-bp downstream region) was transformed into shi1. As expected, positive transgenic plants displayed normal plant development with recovered tiller numbers (Figures 2E and 2F). Moreover, ORF2 knockout transgenic plants generated by clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR associated protein 9 (Cas9) genome-editing approach (in Kitaake background, O. sativa ssp japonica) showed reduced tiller number, but increased panicle branch number, compared with the wild-type plants (Kitaake; Supplemental Figure 3). On the contrary, the ORF2 overexpression lines had reduced plant height (Supplemental Figures 4A and 4B) and ectopic tillers usually formed at the upper internodes (Supplemental Figures 4C and 4D). At the reproductive developmental stage, the primary and especially secondary branch numbers of ORF2 overexpression lines were remarkably decreased (Supplemental Figures 4E to 4G). Taken together, these molecular and genetic lines of evidence confirmed that ORF2 indeed represents OsSHI1.

Expression Pattern of OsSHI1

Sequence analysis revealed that OsSHI1 encodes a transcription factor homologous to the Arabidopsis SHI family with the intrinsic C₃HC₃H zinc finger domain and the SHI family-specific IGGH domain (Figure 3A; Supplemental Figure 5). Consistent with its function as a transcription factor, transient expression analysis in rice protoplasts showed that the OsSHI1-GFP fusion protein was exclusively localized to the nucleus (Figure 3B). The transactivation activity assay indicated that OsSHI1 exhibited weak transcriptional activation activity in yeast cells (Supplemental Figure 6) and was capable of forming homodimer via its C terminus (Figure 3C). RT-quantitative (q)PCR analysis revealed that OsSHI1 was expressed in various tissues, with higher transcript abundance being detected in root, young panicle, and axillary buds (Figure 3D). Histochemical staining analysis of the pOsSHI1:GUS (β‑glucuronidase) transgenic plants showed strong GUS staining in root, young panicle, and axillary buds, but not in the culm, leaf blade, or leaf sheath, further confirming that the OsSHI1 promoter is active in these tissues (Supplemental Figure 7). Moreover, immunoblot analysis using anti-OsSHI1–specific antibodies (Supplemental Figure 8) validated that OsSHI1 protein was mainly accumulated in young panicle and axillary buds, consistent with its role in regulating tiller and panicle development (Figure 3E).

Figure 3.

Expression Analysis of OsSHI1 and Functional Characterization of OsSHI1 Protein.

(A) Schematic representation of the OsSHI1 protein. The conserved zinc finger domain and IGGH domain are indicated.

(B) Subcellular localization of the OsSHI1-GFP fusion protein in rice protoplast. OsD53-mCherry was used as the nuclear marker. Bar = 20 μm.

(C) OsSHI1 is capable of forming homodimer through the C terminus in yeast cells. Transformed yeast cells were spotted on the control medium DDO (SD/-Trp/-Leu) and selective medium QDO (SD/-Trp/-Leu/-His/-Ade). The empty pGADT7 was used as the negative control. OsSHI1-N” and “OsSHI1-C indicate the N- and C-terminal regions of OsSHI1 including the zinc finger domain and IGGH domain, respectively.

(D) RT-qPCR analysis of the expression pattern of OsSHI1 in various tissues. R, root; C, culm; LB, leaf blade; LS, leaf sheath; YP, young panicle; MP, mature panicle; AB, axillary buds. Values are presented as mean ± sd (n = 3).

(E) Immunoblot analysis showing the accumulation of OsSHI1 protein in various tissues. HSP antibody was used as the loading control. Molecular masses of proteins (kDa) are shown on the left. R, root; C, culm; LB: leaf blade; LS, leaf sheath; YP, young panicle; MP, mature panicle; AB, axillary buds.

OsSHI1 Physically Interacts with IPA1

To elucidate the regulatory mechanism of OsSHI1 on rice tiller and panicle development, a yeast two-hybrid (Y2H) screening was performed to identify the interacting partners of OsSHI1 (Supplemental Table 2). Intriguingly, two positive clones containing the coding region of IPA1 were isolated. Considering that IPA1 plays a critical role in the establishment of rice plant architecture, we pursued their interaction and physiological significance further. Dissection of their interactive domains showed that both the N- and C-terminal regions of OsSHI1 could interact with the C terminus, but not the SBP domain of IPA1 (Figures 4A and 4B). The interaction between IPA1 and OsSHI1 was further confirmed by in vitro pull-down assay (Figure 4C; Supplemental Figure 9), in vivo bimolecular fluorescence complementation (BiFC) assay in leaf epidermal cells of Nicotiana benthamiana (Figure 4D), and coimmunoprecipitation (CoIP) assay in axillary buds of wild-type seedlings (Figure 4E).

Figure 4.

OsSHI1 Physically Interacts with IPA1.

(A) Schematic representation of the various truncated visions of OsSHI1 and IPA1 proteins. The conserved zinc finger, IGGH, and SBP domains are indicated.

(B) Both the N- and C-terminal regions of OsSHI1 interact with the C terminus of IPA1 in yeast cells. Transformed cells were spotted on the control medium (DDO, SD/-Leu/-Trp) and selective medium (QDO, SD/-Leu/-Trp/-His/-Ade). The empty pGADT7 was used as the negative control.

(C) In vitro pull-down assay confirms that OsSHI1-GST, but not GST itself, could precipitate IPA1 as detected by anti-MBP antibody. The symbols “–” and “+” indicate the absence and presence of the corresponding proteins.

(D) BiFC assay verifies the interaction between OsSHI1 and IPA1 in the nuclei of epidermal cells of N. benthamiana. OsSHI1 and IPA1 were fused with the N- and C terminus of YFP, respectively. eYNE and eYCE were used as the negative controls. OsSPL16, a homologous protein of IPA1, was also used as a negative control to demonstrate the specific interaction between OsSHI1 and IPA1. DIC, differential interference contrast; Merged, merged images of YFP channel and DIC. Bar = 30 μm.

(E) In vivo CoIP assay shows that OsSHI1 interacts with IPA1 in the axillary buds of wild-type seedlings. Total protein extracts were immunoprecipitated by the anti-IPA1–specific polyclonal antibodies and analyzed by immunoblot probed with the anti-IPA1 and anti-OsSHI1 polyclonal antibodies. Immunoglobulin G was used as the negative control.

OsSHI1 Directly and Negatively Regulates the Expression of OsTB1 and OsDEP1

Previous reports revealed that IPA1 directly activates the expression of OsTB1 and OsDEP1, two key regulators for tiller and panicle development in rice (Jiao et al., 2010; Lu et al., 2013). The SBP-box of IPA1 functions as the conserved DNA binding domain and its C terminus confers transcriptional activation activity (Lu et al., 2013). As the shi1 mutants exhibited reduced tillers and increased panicle branch number, we speculated that OsSHI1 may act together with IPA1 to coregulate the expression of OsTB1 and OsDEP1. In support of this notion, RT-qPCR analysis revealed that indeed, the expression levels of both OsTB1 and OsDEP1 were significantly increased in the shi1 background (Figures 5A and 5B). Sequence analysis identified two and one T/GCTCTAC motifs in the promoter regions of OsTB1 and OsDEP1, respectively. Notably, these elements are quite similar to the binding motif (ACTCTAC) of the Arabidopsis AtSTY1 homologous protein (Eklund et al., 2010). Intriguingly, these T/GCTCTAC motifs are located near the IPA1 recognition sites (GTAC) in the promoter regions of both OsTB1 (59 bp) and OsDEP1 (113 bp; Supplemental Figures 10 and 11), hinting that OsSHI1 and IPA1 may coordinately regulate the expression of OsTB1 and OsDEP1 to affect plant architecture.

Figure 5.

OsSHI1 Binds Directly to the Promoter Regions of OsTB1 and OsDEP1.

(A) and (B) RT-qPCR analysis of OsTB1 (A) and OsDEP1 (B) expression levels in wild-type and shi1 axillary buds or young panicle tissues, respectively. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, *P < 0.05, **P < 0.01). WT, wild type.

(C) and (D) Y1H assays to dissect the binding regions of OsSHI1 in the promoter regions of OsTB1 (C) and OsDEP1 (D). Series of promoter fragments of OsTB1 and OsDEP1 were fused to the upstream region of the LacZ reporter gene and tested for OsSHI1 binding.The empty pB42AD was used as the negative control.

(E) to (G) EMSAs to test OsSHI1 binding to the two TCTCTAC (E) and (F) and one GCTCTAC (G) motifs in the OsTB1 and OsDEP1 promoters, respectively. The T/GCTCTAC motif was mutated into T/GAAAAAC to test for sequence specificity. The triangles indicate increased amounts of competing probes. GST or MBP proteins were used as the negative controls. The symbols “–” and “+” indicate the absence and presence of the corresponding proteins or probes.

(H) and (I) ChIP-qPCR analyses of the P3 promoter regions of OsTB1 (H) and OsDEP1 (I) in ChIP samples precipitated by anti-OsSHI1–specific polyclonal antibodies. The fold enrichment was calculated as IP/Input. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, **P < 0.01).

(J) ChIP-reChIP analysis of IPA1 and OsSHI1 co-occupy common target promoters. The chromatin of wild-type plants was first immunoprecipitated by anti-OsSHI1–specific polyclonal antibodies and then by anti-IPA1–specific polyclonal antibodies, and the precipitated DNA was quantified by qPCR analysis. The fold enrichment was calculated as IP/Input and normalized to that of the UBIQUITIN promoter region as an internal control. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, **P < 0.01).

To test this, we first performed yeast one-hybrid (Y1H) assay to test for direct binding of OsSHI1 to the OsTB1 and OsDEP1 promoters. As shown in Figures 5C and 5D, OsSHI1 bound directly to the F4 (−1∼−336) and F3 (−1274∼−1578) promoter regions of OsTB1 and OsDEP1, respectively, where the three OsSHI1 recognition cis-elements reside. We further used electrophoretic mobility shift assay (EMSA) to verify the binding specificity of OsSHI1 to these motifs. Full-length recombinant OsSHI1 proteins (fused with glutathione s-transferase [GST] or maltose-binding protein [MBP] tags) were expressed in Escherichia coli BL21 (DE3) (TransGen Biotech) and affinity-purified (Supplemental Figure 9). We found that the GST- or MBP-OsSHI1 fusion proteins could bind DNA probes containing the T/GCTCTAC motifs. Moreover, nonlabeled competing probes could effectively reduce the binding ability of OsSHI1 in a dosage-dependent manner and mutation of the core sequence (T/GCTCTAC mutated to T/GAAAAAC) abolished the binding (Figures 5E to 5G). Furthermore, chromatin immunoprecipitation assay (ChIP) using anti-OsSHI1 specific polyclonal antibodies verified that OsSHI1 could be specifically recruited to the P3 promoter regions of OsTB1 and OsDEP1, adjacent to the IPA1 recognition sites (Figures 5H and 5I). Moreover, ChIP-reChIP analysis with chromatin immunoprecipitated sequentially by OsSHI1- and IPA1-specific polyclonal antibodies showed that OsSHI1 and IPA1 co-occupy common target promoters (OsTB1 and OsDEP1) in vivo (Figure 5J; Supplemental Figure 12). Further domain dissection analysis revealed that the N-terminal region of OsSHI1 (containing the conserved C₃HC₃H zinc finger domain) confers the DNA binding ability (Supplemental Figure 13).

OsSHI1 Represses the DNA Binding Activity of IPA1 to the Promoters of OsTB1 and OsDEP1

Previous studies demonstrated that IPA1 acts as a transcriptional activator, promoting the accumulation of OsTB1 and OsDEP1 transcripts (Lu et al., 2013). The upregulation of OsTB1 and OsDEP1 in the shi1 background indicates that OsSHI1 and IPA1 act antagonistically in regulating the expression of OsTB1 and OsDEP1. We thus performed transient dual-luciferase (LUC) assay in rice protoplasts to evaluate the transcriptional regulatory relationship between OsSHI1 and IPA1. As shown in Figures 6A to 6C, IPA1 alone greatly enhanced the expression of the luciferase (LUC) reporter gene driven by the OsTB1 and OsDEP1 promoters, while OsSHI1 alone had no significant effect. However, when coexpressed with OsSHI1, the transcriptional activation activity of IPA1 was significantly attenuated. Moreover, mutations of the T/GCTCTAC motifs did not compromise the effect of OsSHI1-mediated repression of the transcriptional activation of OsTB1 conferred by IPA1 (Figure 6D). To test whether OsSHI1 affects the DNA binding affinity of IPA1, in vitro EMSAs were performed. As previously reported, IPA1 could bind directly to the GTAC motifs in the OsTB1 and OsDEP1 promoter regions, and no shifted bands were observed for OsSHI1 to the GTAC motifs (Figures 6E and 6F). However, the presence of increasing amounts of OsSHI1 protein in the reactions significantly reduced the binding ability of IPA1 to the target probes, independent of OsSHI1 binding (Figures 6E and 6F; Supplemental Figure 14), indicating that OsSHI1 could interfere with the DNA binding ability of IPA1. Moreover, immunoblot analysis using anti-IPA1–specific polyclonal antibodies (Supplemental Figure 12) revealed that no differences of IPA1 protein abundance were observed in either the young seedling or young panicle tissues of wild-type and shi1 (Figures 6G and 6H). However, in vivo ChIP-qPCR assay performed with DNA precipitated by anti-IPA1 antibodies revealed that the P3 and P4 promoter regions of OsTB1 and OsDEP1 were significantly more enriched in the shi1 mutant (Figures 6I and 6J), which is consistent with the upregulation of expression levels of OsTB1 and OsDEP1 in shi1 mentioned above.

Figure 6.

OsSHI1 Represses the DNA Binding Activity of IPA1.

(A) Schematic representation of the effector and reporter constructs. Full-length coding regions of OsSHI1 and IPA1 under control of the double 35S promoter were used as the effectors. The Firefly luciferase gene LUC driven by the OsTB1 or OsmTB1 and OsDEP1 promoters and the Renilla luciferase gene Ren driven by the 35S promoter were used as the reporter and internal control, respectively. d35S, double 35S promoter.

(B) to (D) OsSHI1 represses the transcriptional activation activities of IPA1 on OsTB1 (B), OsDEP1 (C), and OsmTB1 (D) promoters in rice protoplasts. Relative LUC activity was calculated by LUC/Ren and normalized to that of vector control which was set as 1. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 3, **P < 0.01).

(E) and (F) EMSAs show that OsSHI1-IPA1 interaction attenuates the DNA binding activity of IPA1 to the GTAC motifs in the promoter regions of OsTB1 (E) and OsDEP1 (F). The triangles indicate increased amounts of OsSHI1 proteins. MBP proteins were used as negative controls. The symbols “–” and “+” indicate the absence and presence of the corresponding proteins.

(G) and (H) Immunoblot analyses of IPA1 protein accumulation in young seedling (G) and young panicle (H) tissues of wild type and shi1. HSP antibody was used as the loading control. The molecular masses of proteins (kDa) are shown on the left. WT, wild type.

(I) and (J) ChIP assays of the P3 and P4 promoter regions of OsTB1 (I) and OsDEP1 (J) in shi1 compared with the wild type. DNAs precipitated from axillary buds or young panicle tissues of wild type and shi1 by anti-IPA1–specific polyclonal antibodies were subjected into ChIP-qPCR analysis. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, **P < 0.01).

To further investigate the biological significance of the OsSHI1 and IPA1 interaction, we generated 35S:IPA1-Flag transgenic plants, which displayed significantly repressed tiller development as expected (Figures 7A and 7B). We further overexpressed OsSHI1 under the control of the ACTIN1 promoter in the 35S:IPA1-Flag transgenic background (Figure 7C). Immunoblot analysis showed that OsSHI1 accumulation did not obviously affect the stability or abundance of IPA1 protein (Figure 7C). In vivo ChIP-qPCR performed with DNA precipitated by anti-Flag antibody from the 35S:IPA1-Flag transgenic plants revealed that the P3 promoter region of OsTB1 (where the two IPA1 recognition sites reside) was predominantly enriched (Figure 7D). However, the enrichment of the P3 promoter region was significantly reduced in the Actin1:OsSHI1/35S:IPA1-Flag transgenic plants, when compared with the scenario of 35S:IPA1-Flag transgenic plants (Figure 7D). Consistent with this, the Actin1:OsSHI1/35S:IPA1-Flag transgenic plants displayed obviously reduced plant height and somewhat recovered tiller development, as well as significantly reduced expression levels of OsTB1, compared with the 35S:IPA1-Flag parental plants (Figure 7E and 7F). Taken together, these results support the conclusion that OsSHI1 negatively regulates the transcriptional activation activity of IPA1 on OsTB1 and OsDEP1 by repressing its DNA binding activity.

Figure 7.

OsSHI1 Acts Antagonistically with IPA1 to Regulate Tillering in Rice.

(A) Plant morphologies of Kitaake, 35S:IPA1-Flag and Actin1:OsSHI1/35S:IPA1-Flag transgenic lines at the heading stage. Bar = 10 cm.

(B) Determination of IPA1-Flag protein accumulation in the 35S:IPA1-Flag transgenic young seedlings. IPA1-Flag protein was detected using anti-Flag antibody. HSP antibody was used as the loading control. The molecular masses of proteins (kDa) are shown on the left.

(C) Immunoblot analyses showing the accumulation of OsSHI1 and IPA1-Flag proteins in the Actin1:OsSHI1/35S:IPA1-Flag transgenic seedlings. OsSHI1 and IPA1-Flag proteins were detected with anti-OsSHI1–specific polyclonal antibodies and anti-Flag antibody, respectively. HSP antibody was used as the loading control. The molecular masses of proteins (kDa) are shown on the left.

(D) OsSHI1 reduces the enrichment of the promoter region of OsTB1 immunoprecipitated by IPA1. DNAs precipitated from 35S:IPA1-Flag and Actin1:OsSHI1/35S:IPA1-Flag transgenic seedlings by anti-Flag antibody were subjected into ChIP-qPCR analysis. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, *P < 0.05).

(E) Overexpression of OsSHI1 in the 35S:IPA1-Flag transgenic background results in increased tiller number. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 5 independent plants, *P < 0.05, **P < 0.01).

(F) RT-qPCR analysis showing OsTB1 expression levels in Kitaake, 35S:IPA1-Flag, and Actin1:OsSHI1/35S:IPA1-Flag transgenic lines. Values are presented as mean ± sd, and the statistically significant differences were determined by Student’s t test (n = 4, **P < 0.01).

IPA1 Acts Downstream of OsSHI1 to Regulate Plant Architecture in Rice

To determine the genetic relationship of OsSHI1 and IPA1, a series of shi1, ipa1, tb1, dep1, ipa1 shi1, tb1 shi1, and dep1 shi1 mutants were generated in the same genetic background (Kitaake) using CRISPR/Cas9-mediated genome-editing approach (Supplemental Figure 15). As expected, shi1 mutant exhibited reduced tillering and increased panicle branching (Supplemental Figure 3). In contrast with the phenotype of shi1 mutant, the tiller number of ipa1 mutant was significantly increased, accompanied with diminished panicle size. Tiller development was greatly intensified in the tb1 mutant. Panicle branching was compromised in both the tb1 and dep1 mutants, similar to ipa1 mutant (Figures 8A and 8B). Further phenotypic observations of the ipa1 shi1, tb1 shi1, and dep1 shi1 double mutants revealed their similarity to the ipa1, tb1, and dep1 single mutant, respectively (characterized by enhanced tillering and compromised panicle branching; Figures 8A and 8B). These observations support the conclusion that OsSHI1 functions upstream of IPA1, OsTB1 and OsDEP1 to regulate plant architecture in rice.

Figure 8.

OsSHI1 Acts Upstream of IPA1 to Regulate Plant Architecture in Rice.

(A) Plant morphologies of Kitaake, shi1, ipa1, tb1, dep1, ipa1 shi1, tb1 shi1, and dep1 shi1 mutants at the heading stage. Bar = 10 cm.

(B) Panicle architectures of Kitaake, shi1, ipa1, tb1, dep1, ipa1 shi1, tb1 shi1, and dep1 shi1 mutants. Grains were removed to show the primary and secondary branch patterns of the panicles. Bar = 2 cm.

DISCUSSION

OsSHI1 Acts Antagonistically with IPA1 in Regulating Plant Architecture in Rice

Tiller number and panicle size are critical determinants of plant architecture and crop yield. To meet the constantly increasing demand for food productivity, a new breeding approach of the New Plant Type or IPA strategy has been proposed (Khush, 2001; Wang and Li, 2008). The IPA traits characterized by fewer sterile tillers, larger panicles, and stronger culms are closely correlated with the accumulation of IPA1 protein, which further activates the expression of OsTB1 and OsDEP1 to regulate plant architecture in rice (Jiao et al., 2010; Miura et al., 2010; Lu et al., 2013). Notably, several aspects of the shi1 mutant phenotype (remarkably reduced tiller number, enlarged panicles, and enhanced culm strength; Figures 1H to 1N; Supplemental Figures 2H to 2M) are reminiscent of the effects of IPA1 overexpression. Furthermore, OsSHI1 overexpression lines exhibit significantly reduced plant height, ectopically formed tillers, and diminished panicle size (Supplemental Figure 4), similar to the previously reported IPA1 transgenic RNAi plants (Wang et al., 2015a). In addition, OsSHI1 is predominantly expressed in axillary buds and young panicle (Figures 3D and 3E; Supplemental Figure 7) and its expression pattern partially overlaps with that of IPA1 (Jiao et al., 2010; Miura et al., 2010; Lu et al., 2013). These observations hint that OsSHI1 and IPA1 may antagonistically regulate plant architecture in rice. This notion is further supported by the observation that the expression levels of OsTB1 and OsDEP1, two positively regulated targets of IPA1, are significantly upregulated in shi1 (Figures 5A and 5B). Moreover, the Actin1:OsSHI1/35S:IPA1-Flag double-overexpression plants exhibit significant dwarfism and somewhat increased tiller number, in comparison with the 35S:IPA1-Flag parental plants (Figures 7A and 7E). Further, in contrast with the repressed tiller development and enlarged panicle architecture of shi1, the tiller and panicle branch numbers of ipa1 shi1, tb1 shi1, and dep1 shi1 double mutants are significantly increased or reduced, similar to the scenario of ipa1, tb1, or dep1 single mutants (Figure 8). These results together suggest that OsSHI1 acts antagonistically and upstream of IPA1 to regulate plant architecture in rice.

OsSHI1 Represses the Transcriptional Activity of IPA1 by Interfering with Its DNA Binding Activity

Recent studies have identified three IPA1-interacting proteins, IPA1 INTERACTING PROTEIN1 (OsIPI1), ovarian tumor domain-containing ubiquitin aldehyde-binding protein 1 (OsOTUB1), and OsD53, and demonstrated their roles in regulating plant architecture in rice (Jiang et al., 2013; Zhou et al., 2013; Song et al., 2017; Wang et al., 2017a, 2017b). OsIPI1 acts as a RING-type E3 ligase to promote the degradation of IPA1 in panicles by adding K48-linked poly-ubiquitin chains while stabilizing IPA1 in shoot apexes by mediating K63-linked poly-ubiquitin modification (Wang et al., 2017a). OsOTUB1 is a deubiquitinating enzyme with both K48- and K63-linked poly-ubiquitin cleavage activities, and OsOTUB1-IPA1 interaction limits the K63-linked ubiquitination of IPA1, which in turn promotes K48-linked ubiquitination -dependent proteasomal degradation of IPA1 (Wang et al., 2017b). Thus, both OsIPI1 and OsOTUB1 are enzymes involved in the posttranscriptional poly-ubiquitin modification of IPA1. Recent studies also showed that OsD53 represses the transcriptional activation activity of IPA1 by interacting with the TOPLESS transcriptional corepressors, which in turn recruit histone deacetylases complexes to modulate local chromatin status, thereby influencing downstream target gene expression. The degradation of OsD53 by the 26S proteasome system releases IPA1 to proceed with the activation of downstream target genes (like OsTB1), allowing tiller development (Jiang et al., 2013; Zhou et al., 2013; Song et al., 2017). Notably, OsIPI1, OsOTUB1, and OsD53 all interact with IPA1 through the conserved SBP domain and they do not affect the DNA binding ability of IPA1 (Song et al., 2017; Wang et al., 2017a, 2017b).

In this study, we used various assays to show that OsSHI1 is an interacting partner of IPA1 (Figure 4; Supplemental Table 2). In contrast with the previous identified IPA1 interacting partners, we showed that OsSHI1 interacts with IPA1 through the C-terminal region but not the SBP domain (Figure 4B). In addition, we showed that the levels of IPA1 transcript and IPA1 protein are not affected in the shi1 mutant (Figures 6G and 6H; Supplemental Figure 16), suggesting that OsSHI1 regulates IPA1 through a previously uncharacterized mechanism. In support of this notion, our transient analysis revealed that OsSHI1 represses the transcriptional activation activity of IPA1 in protoplasts (Figures 6B and 6C). Further, both in vitro EMSAs and in vivo ChIP assays revealed that the OsSHI1-IPA1 interaction reduces the binding affinity of IPA1 to the promoter regions of both OsTB1 and OsDEP1 (Figures 6E, 6F, 6I, and 6J, and 7D), but no obvious effect of IPA1 on the DNA binding ability of OsSHI1 was observed (Supplemental Figure 17). Compromised DNA binding affinity of IPA1 conferred by OsSHI1 was partially independent of its binding to its own cis-element (Figure 6D; Supplemental Figure 14).

Based on our findings, we propose a model to illustrate how OsSHI1 and IPA1 act cooperatively to regulate plant architecture in rice (Supplemental Figure 18). In wild-type plants, OsSHI1/IPA1 heterodimer formation reduces the DNA binding ability of IPA1 to modulate the expression of downstream target genes and plant architecture. However, in shi1, the absence of OsSHI1 enhances the binding of IPA1 to the promoter regions of OsTB1 and OsDEP1 to upregulate their expression levels, consequently altering plant architecture. Taken together, our results suggest that OsSHI1 regulates IPA1 activity through affecting its transcriptional activity. However, other possibilities exist, such as that the formation of a OsSHI1-IPA1 complex may alter the protein/DNA conformation of target genes thus reducing the access of IPA1 or block the transcriptional activation activity conferred by the C-terminal region of IPA1 or interfere with the interaction of IPA1 with other proteins (such as chromatin remodeling factors). Given the demonstration that plant architecture could be improved through fine-tuning the tissue-specific expression or protein accumulation pattern of IPA1 (Wang and Wang, 2017), our findings may provide new alternative approaches to modify the activity of IPA1 and thus bear important implications for genetic improvement of rice plant architecture in future breeding.

METHODS

Plant Materials and Growth Conditions

The shi1 mutant was initially isolated from a mutant library of 9,311 (Oryza sativa ssp indica) mutagenized by ⁶⁰Co-γ irradiation. Two F₂ populations derived from the reciprocal crosses between shi1 and wild type (9311) were utilized for genetic analysis of segregation. For map-based cloning, an F₂ population was generated from the cross between shi1 and cv 02,428 (O. sativa ssp japonica). Plants were grown in the paddy fields at the Chinese Academy of Agricultural Sciences and Nanjing Agricultural University under natural conditions with conventional management.

5-ethynyl-2´-deoxyuridine (EdU) Staining Observation

Wild-type and shi1 young seedlings were incubated in the Murashige and Skoog solution with the 5-ethynyl-2´-deoxyuridine substrate overnight. Shoot bases with axillary buds were carefully dissected and fixed in the formaldehyde - acetic acid - alcohol (FAA) fixative solution (50% (v/v) ethanol, 5% (v/v) acetic acid, and 10% (v/v) formaldehyde) at 4°C overnight. Samples were rehydrated through the 70% (v/v), 50% (v/v), 30% (v/v), 15% (v/v), and 0% (v/v) ethanol series (each for 15 mins) and incubated in 1% (v/v) Triton X-100 for 2 h. Then the samples were incubated in a staining solution (invitrogen) (1× Click-iT reaction buffer, 100 mM of CuSO₄, 10 mM of Alexa Fluor azide, and 1× Reaction buffer additive) for 3 h in darkness. The samples were subsequently washed three times in water and dehydrated through the 15% (v/v), 30% (v/v), 50% (v/v), 70% (v/v), 85% (v/v), and 95% (v/v) ethanol series (each for 20 min) followed by incubating with 100% (v/v) ethanol for 2 h. After sufficient dehydration, samples were hyalinized through the 2:1, 1:1, 1:2, and 0:1 ethanol:methyl salicylate series (each for 1 h). Fluorescence signals were observed using a confocal laser scanning microscope (LSM 700; Zeiss).

Paraffin Section Analysis

Shoot bases with axillary buds were carefully dissected and fixed in the formaldehyde - acetic acid - alcohol (FAA) solution at 4°C for 24 to 72 h. Samples were dehydrated through the 70% (v/v), 80% (v/v), 90% (v/v) and 100% (v/v) ethanol series (each for 60 mins). After sufficient dehydration, samples were hyalinized through the 1:2, 1:1, 2:1, and 1:0 xylene:ethanol series (each for 45 min). Samples were then incubated in the 1:1 ethanol: Paraplast Plus solution (Sigma-Aldrich) at 42°C for 24 h and repeated once, after which samples were embedded in Paraplast Plus at 60°C for 4 d. Then the samples were sectioned into 8-µm–thick sections using a model no. RM2245 Rotary Microtome (Leica). After the removal of Paraplast Plus by series of xylene and ethanol solutions, sections were stained with toluidine blue before pictures taken using a model no. ICC50 HD Microscope (Leica). The images were processed using the software suite ACDSee (https://www.acdsee.com/en/index).

Map-Based Cloning of OsSHI1

A total of 137 genome-wide primer pairs that exhibit polymorphisms between 9311 and 02428 were identified from our primer library. The OsSHI1 locus was initially mapped to an interval between the simple sequence repeat markers RM107 and RM189 on the long arm of chromosome 9 using 180 F₂ mutant plants. Subsequent fine-mapping based on 7,547 progeny delimited the mutant locus to a ∼50-kb genomic region with additional newly developed molecular markers (Supplemental Table 3). cDNAs of the four ORFs in the fine-mapped region were amplified from both wild type and the shi1 mutant (primer pairs listed in Supplemental Table 3), and the PCR products were confirmed by sequencing.

RT-PCR and RT-qPCR Analyses

Total RNA was extracted from various tissues using the ZR Plant RNA MiniPrep Kit (Zymo Research) following the manufacturer’s recommendations. The first-strand cDNA was synthesized based on the QuantiTect Reverse Transcription Kit (Qiagen). RT-PCR with 28 cycles was performed to amplify the four ORFs in the fine-mapped region and 24 cycles for ACTIN2 (which was used as the endogenous control). Real time-qPCR was performed on an ABI7500 Real-Time PCR System using SYBR Premix Ex Taq (Takara) with rice Ubiquitin as the endogenous control. Relative changes in gene expression levels were quantitated based on three biological replicates via the 2^−△△Ct method (Livak and Schmittgen, 2001). All primer pairs used for RT-PCR and Real time-qPCR are listed in Supplemental Tables 3 and 4.

Vector Construction and Plant Transformation

A ∼5-kb genomic DNA fragment (consisting of a 2.9-kb upstream region, the entire OsSHI1 coding region including two exons, one intron, and a 400-bp downstream region) was amplified with the primer pair OsSHI1-complementation (Supplemental Table 5) and inserted into the HindIII/BamHI restriction sites of the pCUbi1390 binary vector to generate the proOsSHI1:OsSHI1 construct, which was introduced into the calli of shi1 via Agrobacterium-mediated transformation using a method described in Hiei et al. (1994).

To knock out the OsSHI1, IPA1, OsTB1, and OsDEP1 genes, 20-bp gene-specific spacer sequences were cloned into the sgRNA-Cas9 vector (Miao et al., 2013) and subsequently introduced into the calli of Kitaake (or OsSHI1-CRISPR-#1), a japonica variety suitable for transformation, via Agrobacterium-mediated transformation. Positive transgenic individuals were identified by sequencing or immunoblot analyses.

To verify the expression pattern of OsSHI1, an ∼2-kb promoter fragment upstream of the ATG start codon was amplified using the primer pair OsSHI1-GUS (Supplemental Table 5) and the PCR product was fused into the EcoRI/NcoI restriction sites of the binary vector pCAMBIA1305. The generated pOsSHI1:GUS construct was introduced into the calli of Kitaake via Agrobacterium-mediated transformation to generate the pOsSHI1:GUS reporter lines.

To generate the IPA1-overexpressing plants, full-length cDNA of IPA1 was amplified using the primer pair IPA1-Flag (Supplemental Table 5) and fused into the XbaI restriction site of the 1300-221-3×Flag binary vector. The generated 35S: IPA1-Flag construct was introduced into the calli of Kitaake via Agrobacterium-mediated transformation.

To determine the effect of OsSHI1 overexpression, full-length cDNA of OsSHI1 was amplified using the primer pair OsSHI1-overespression (Supplemental Table 5). The PCR product was inserted into the SmaI restriction site of the pCAMBIA2300 binary vector to generate the Actin1:OsSHI1-overexpression construct, which was subsequently introduced into the calli of Kitaake or 35S:IPA1-Flag transgenic plants via Agrobacterium-mediated transformation.

Subcellular Localization

For subcellular localization of OsSHI1 protein, the 945-bp coding region of OsSHI1 was inserted into the XbaI restriction site upstream of GFP in the transient expression vector pAN580 driven by the double CaMV35S promoter to generate the OsSHI1-GFP construct (primer pair OsSHI1-GFP in Supplemental Table 5). The OsSHI1-GFP plasmid was introduced into the rice protoplasts according to protocols described in Zhang et al. (2011). Fluorescence of GFP was observed using a confocal laser scanning microscope (LSM 700; Zeiss).

Transactivation Activity Assay

The full-length OsSHI1 coding region was cloned into the pGBKT7 (Clontech) vector at the EcoRI/PstI restriction sites (primer pair OsSHI1-BD in Supplemental Table 5) to generate the OsSHI1-BD construct which was then transformed into the Saccharomyces cerevisiae strain AH109, together with empty pGADT7. Transactivation activity assay was performed on selective solid medium followed with quantitative β-gal assay of the LacZ reporter gene using chlorophenolred-beta-D-galactopyranoside (CPRG) as the substrate. Early heading date1 (Cho et al., 2016) and the empty pGBKT7 vector were used as the positive and negative control, respectively. All procedures were performed according to the manufacturer’s recommendations (Clontech).

Y2H

The coding region of OsSHI1 was fused to the GAL4 binding domain of the “bait” pGBKT7 vector (Clontech; primer pairs listed in Supplemental Table 5). A cDNA library prepared from rice young inflorescences was used to perform the Y2H screening and positive clones were identified by sequencing. Full-length and various truncated versions of IPA1 were inserted into the EcoRI/XhoI restriction sites of the “prey” pGADT7 vector (Clontech; primer pairs listed in Supplemental Table 5). Series of combined bait and prey constructs were cotransformed into the yeast strain AH109 (Clontech). After growing on SD-Trp/-Leu plates for 3 d at 30°C, interactions between baits and preys were examined on the selective medium SD-Leu/-Trp/-His/-Ade. Yeast strain containing OsSHI1-BD in combination with the empty pGADT7 vector was used as the negative control. Detailed procedures were performed according to the manufacturer’s recommendations (Clontech).

GUS Staining Assay

GUS histochemical staining was performed according to the method described in Duan et al. (2016). Briefly, various tissues of the pOsSHI1:GUS transgenic plants were detached and incubated in the GUS staining buffer overnight. Tissues were then transferred into alcohol solution to extract the chlorophyll before pictures taken using a model no. ICC50 HD microscope (Leica). The images were processed using the software suite ACDSee.

In Vitro Pull-Down Assay

Full-length coding sequences of OsSHI1 and IPA1 were cloned into the expression vectors pGEX4T-1 and pMAL-c2x, respectively (primer pairs listed in Supplemental Table 5), to generate GST or MBP tag fusion proteins. Expression of GST, GST-OsSHI1, and MBP-IPA1 in Escherichia coli BL21 (DE3) cells (TransGen) was induced by 0.4-mM isopropyl-β-d-thiogalactoside at 18°C for 16 h. Fusion proteins were purified using the GST Bind Resin (Novagen) or Amylose Resin (New England BioLabs) according to the manufacturer’s protocols, and protein concentrations were determined by the BSA quantitative assay.

For pull-down assay, roughly equal amounts of purified GST and GST-OsSHI1 proteins were incubated with 30 μL of GST Bind Resin in 1 mL of PBS solution for 30 min with gentle rotation, after which ∼2 µg of IPA1-MBP fusion protein was added. After a further incubation for 60 min, the resin was washed five times with PBS, diluted in 50 μL 1× Protein Loading Buffer, and denatured at 100°C for 10 min before separation on 10% SDS-PAGE gel. Proteins were then transferred to the nitrocellulose (NC) membrane, detected with horseradish peroxidase conjugated anti-GST (PM013-7; MBL International) or anti-MBP (E8032L; New England BioLabs) antibodies at 1:5000 dilutions, and visualized with an enhanced chemiluminescence reagent (GE Healthcare).

BiFC

Full-length coding region of OsSHI1 was amplified (primer pair listed in Supplemental Table 5) and cloned into the EcoRI/SalI restriction sites of the pSPYNE173 (for split YFP N-terminal fragment expression) expression vector to generate the OsSHI1-eYNE construct. Full-length coding regions of IPA1 and OsSPL16 were amplified (primer pair listed in Supplemental Table 5) and inserted into the EcoRI/SalI restriction sites of the pSPYCE (for split YFP C-terminal fragment expression) expression vector to generate the IPA1-eYCE or OsSPL16-eYCE constructs. For transient expression, Agrobacterium tumefaciens strain EHA105 carrying the combined OsSHI1-eYNE and IPA1- or OsSPL16-eYCE constructs was coinfiltrated with the p19 strain into leaves of 5-week–old Nicotiana benthamiana. The yellow fluorescent protein (YFP) fluorescent signals were monitored 48 to 72 h after infiltration using a laser confocal scanning microscope (model no. LSM 700; Zeiss).

Preparation and Determination of OsSHI1- and IPA1-Specific Antibodies

The synthetic peptides of OsSHI1 (Os09g0531600, SRDPTKRPRARPSATTP) and IPA1 (Os08g0509600; RIDPGSGSTFDQTSNTMD) were injected into rabbits to generate the corresponding polyclonal antibodies at ABclonal Technology. The specificities of anti-OsSHI1 and anti-IPA1 polyclonal antibodies were determined by immunoblot analysis using wild-type young panicle tissues. Samples were ground into fine powder in liquid N, suspended in 2× volumes of Protein Extraction Buffer (50 mM of Tris–HCl at pH 8.0, 150 mM of NaCl, 10 mM of MgCl₂, 1 mM of EDTA, 10% glycerol, and 1 ± Protease inhibitor cocktail) and incubated at 4°C for 30 min with rotation. After centrifuging at 12,000g at 4°C for 10 min, the supernatant was boiled in 1× Protein Loading Buffer before separation on 10% SDS-PAGE gel. Proteins were then transferred to the NC membrane, detected with anti-OsSHI1 or anti-IPA1 antibodies at 1:1,000 dilutions and visualized with enhanced chemiluminescence reagent (GE Healthcare). HSP antibody (AbM51099-31-PU; Beijing Protein Innovation) was used as the endogenous control.

In Vivo CoIP Assay

Axillary buds of wild-type seedlings were carefully detached and ground into fine powder in liquid N. Samples were suspended in 4× volumes of Protein Extraction Buffer (50 mM of Tris–HCl at pH 8.0, 150 mM of NaCl, 10 mM of MgCl₂, 1 mM of EDTA, 10% (v/v) glycerol, and 1× Protease inhibitor cocktail) and incubated at 4°C for 30 min with rotation. After centrifuging at 12,000g for 10 min at 4°C, the supernatant was mixed with 30-μL Protein A beads (Millipore) and incubated with rotation at 4°C for 1 h. Beads were pelleted and the cleaned supernatant were incubated with 10 μL of anti-IPA1–specific polyclonal antibodies for at least 2 h at 4°C with gentle rotation. Thirty-microliter Protein A beads were added to precipitate the protein complex with rotation at 4°C for 1 h. Beads were pelleted and washed four times with prechilled Protein Extraction Buffer. The bound proteins were eluted from the beads with 1× Protein Loading Buffer by boiling at 100°C for 10 min. Protein samples were then separated by 10% SDS-PAGE gel, transferred to NC membrane, and immunoblotted with anti-IPA1 and anti-OsSHI1–specific polyclonal antibodies.

Y1H

Y1H analysis was performed according to the method described in Lin et al. (2007). Briefly, the full-length coding region of OsSHI1 was cloned into the pB42AD vector at the EcoRI restriction site to generate the AD-OsSHI1 construct (primer pair OsSHI1-42AD in Supplemental Table 5). The full ∼2 kb and various truncated versions of promoter regions of OsTB1 or OsDEP1 were amplified (primer pairs listed in Supplemental Table 5) and ligated into the XhoI restriction site of the pLacZi reporter vector. Constructs were then cotransformed into the yeast strain EGY48. Transformants were grown on SD-Trp/-Ura plates for 3 d at 30°C and then transferred onto 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside plates for blue color development. Yeast strains containing the empty pB42AD in combination with the LacZ reporter constructs were used as negative controls.

Transient Expression Assay in Protoplasts and LUC Activity Determination

Approximately 1-kb and 2-kb promoter regions of OsTB1 and OsDEP1 were amplified using the primer pairs OsTB1-LUC or OsmTB1-LUC and OsDEP1-LUC (primer pairs listed in Supplemental Table 5) and cloned into the upstream of LUC reporter gene to generate the OsTB1-LUC or OsmTB1-LUC and OsDEP1-LUC reporter constructs. The luciferase gene from Renilla reniformis (Ren) under the control of CaMV35S promoter was used as the internal control. Full-length cDNAs of OsSHI1 and IPA1 were amplified (primer pairs listed in Supplemental Table 5) and inserted into the BamHI/PstI restriction sites of the pAN580 vector to generate the d35S:OsSHI1 and d35S:IPA1 effector constructs, respectively. The combined reporter and effector plasmids were cotransformed into the rice protoplasts according to a protocol described in Zhang et al. (2011). The LUC activity was quantified with a Dual-Luciferase Assay Kit (Promega) following the manufacturer’s recommendations and the relative LUC activity is calculated as the ratio of LUC/Ren.

ChIP Assay

To assess the enrichment of OsSHI1 or IPA1 at the promoter regions of OsTB1 and OsDEP1 in vivo, ChIP assays were performed using anti-OsSHI1 or anti-IPA1–specific polyclonal antibodies. Approximately 4 g of wild-type axillary buds or young panicle tissues (1 to 2 g for transgenic plants) were ground into fine powder in liquid N, resuspended in 20 mL Extraction buffer 1 (0.4 M of Suc, 10 mM of Tris-HCl at pH 8.0, 10 mM of MgCl₂, 5 mM of β-mercaptoethanol [β-ME], 0.1 mM of phenylmethylsulfonyl fluoride [PMSF], 1× proteinase inhibitor, and 1% (v/v) formaldehyde), and thoroughly mixed to release the nuclei. After incubation under vacuum conditions for 30 min, 0.125 M of Gly was added and incubated for another 5 min to stop the crosslink reaction. The solution was filtered by two layers of Miracloth (Millipore) and centrifuged at 3000g at 4°C for 20 min to remove the supernatant. The pellet was resuspended in 1-mL Extraction buffer 2 (0.25 M of Suc, 10 mM of Tris-HCl at pH 8.0, 10 mM of MgCl₂, 1% (v/v) Triton X-100, 5 mM of β-ME, 0.1 mM of PMSF, and 1× proteinase inhibitor) and centrifuged at 12,000g at 4°C for 10 min. The pellet was resuspended in 300 μL of Extraction buffer 3 (1.7 M of Suc, 10 mM of Tris-HCl at pH 8.0, 2 mM of MgCl₂, 0.15% (v/v) Triton X-100, 5 mM of β-ME, 0.1 of mM PMSF, and 1× proteinase inhibitor), laid on top of another clean 300 μL of Extraction buffer 3 and centrifuged at 15,000g at 4°C for 1 h. The chromatin pellet was resuspended in 200 μL of Nuclei lysis buffer (50 mM of Tris-HCl at pH 8.0, 10 mM of EDTA, 1% (w/v)SDS, and 1× proteinase inhibitor) and sonicated to 200∼500 bp with 3-s burst/7s interval frequency at 2-W power for 33 min. After centrifugation at 12,000g at 4°C for 10 min, the chromatin supernatant was diluted with Dilution buffer (0.01 (w/v)% SDS, 1.1% (v/v) Triton X-100, 1.2 mM of EDTA, 16.7 mM of Tris-HCl at pH 8.0, and 167 mM of NaCl) and 1 mL of the diluted chromatin sample was precleared with 20-μL Protein A beads (Millipore) for 1 h at 4°C with rotation. Then 10 μL of anti-OsSHI1 or anti-IPA1–specific polyclonal antibodies together with 10 μL of BSA (10 mg/mL) were added to the precleared sample and incubated overnight with gentle rotation. One microliter of salmon sperm DNA (10 mg/mL) was added as the blocking reagent and the immune complexes were collected by 30-μL Protein A beads for 1 h at 4°C with rotation. The Protein A beads were washed stepwise with a low salt wash buffer (0.1% (w/v) SDS, 1% (v/v) Triton X-100, 2 mM of EDTA, 20 mM of Tris-HCl at pH 8.0, and 150 of mM NaCl), a high salt wash buffer (0.1%(w/v) 1% (v/v) Triton X-100, 2 mM of EDTA, 20 mM of Tris-HCl at pH 8.0, and 500 mM of NaCl), the LiCl wash buffer (0.25 M of LiCl, 1% (v/v) Nonidet P-40, 1% (w/v) deoxycholic acid sodium, 1 mM of EDTA, and 20 mM of Tris-HCl at pH 8.0), and TE buffer (10 mM of Tris-HCl at pH 8.0 and 1 mM of EDTA) each two times. The immune complexes were eluted from the Protein A beads by incubating with 250 μL of Elution buffer (0.1 M of NaHCO₃ and 1% (w/v)SDS) at 65°C for 20 min with agitation. The supernatant was transferred to another tube to repeat elution and the two eluates were combined. Twenty microliters of NaCl (5 M), 10 μL of RNase A (10 mg/mL), and 2.5 μL of protease K (10 mg/mL) were added to the eluted solution (for the input sample, two volume amounts were added) and incubated at 65°C for at least 6 h with agitation. The immunoprecipitated DNA was extracted with isopropyl alcohol precipitation. The recovered DNA was used as the template for ChIP-qPCR and the enrichment was calculated as the ratio of immunoprecipitation (IP) to Input. The ChIP-reChIP assay was performed according to protocol reported in Furlan-Magaril et al. (2009). All primer pairs are listed in Supplemental Table 6.

EMSA

A quantity of 3ʹ-digoxigenin-labeled probes containing the putative OsSHI1 binding sites were synthesized by Invitrogen (primer pairs listed in Supplemental Table 7). EMSAs were performed with the DIG Gel Shift Kit (Roche, 03353591910) following the manufacturer’s recommendations. Briefly, equal amounts of complementary oligonucleotides were incubated at 95°C for 10 min, cooled down slowly to 15°C (0.1°C/1 s), and diluted to 50 fmol/μL final concentrations. The DNA binding reaction was performed with 100-fmol probe, 2-µg poly (dI-dC), and 100-ng purified MBP or MBP-OsSHI1 proteins and incubated at room temperature for 30 min. Then the samples were immediately applied to the pre-run native polyacrylamide gel containing 6.5% acrylamide in 0.5× Tris-Borate-EDTA buffer. After electro-blotting onto a nylon membrane (Millipore), the oligonucleotides were crosslinked using UV-light. The membrane was incubated in a blocking solution for 30 min, followed by incubating in a DIG antibody solution for another 30 min. After intensive washing with a washing buffer, CSPD working solution was applied to the membrane to visualize the signal.

Accession Numbers

Sequence data from this article can be found in the EMBL/GenBank data libraries under the following accession numbers: OsSHI1, Os09g0531600; IPA1, Os08g0509600; OsSPL16, Os08g0531600; OsTB1, Os03g0706500; OsDEP1, Os09g0441900; UBIQUITIN, Os03g0234200; and ACTIN2, Os03g0718100.

Supplemental Data

• Supplemental Figure 1. Normal initiation of axillary buds in shi1.

• Supplemental Figure 2. Grain, leaf, and culm morphologies of wild type and shi1.

• Supplemental Figure 3. Phenotypes of ORF2 (OsSHI1) knockout plants.

• Supplemental Figure 4. Phenotypes of ORF2 (OsSHI1) overexpression plants.

• Supplemental Figure 5. Characterization and sequence analysis of OsSHI1.

• Supplemental Figure 6. OsSHI1 exhibits weak transcriptional activation activity in yeast cells.

• Supplemental Figure 7. Histochemical staining of the pOsSHI1:GUS transgenic plants.

• Supplemental Figure 8. Specificity tests of the anti-OsSHI1 polyclonal antibodies used for ChIP and immunoblot analyses.

• Supplemental Figure 9. Protein induction and purification used for immunoblot and EMSAs.

• Supplemental Figure 10. Schematic depictions of the cis-elements in the promoter regions of OsTB1 and OsDEP1.

• Supplemental Figure 11. Sequence analysis of the promoter regions of OsTB1 and OsDEP1.

• Supplemental Figure 12. Specificity tests of the anti-IPA1 polyclonal antibodies used for ChIP and immunoblot analyses.

• Supplemental Figure 13. The zinc finger domain is indispensable for the DNA binding ability of OsSHI1.

• Supplemental Figure 14. OsSHI1 represses the DNA binding ability of IPA1 independent of OsSHI1 binding.

• Supplemental Figure 15. Generation and identification of mutants generated by CRISPR/Cas9 genome-editing approach.

• Supplemental Figure 16. OsSHI1 does not affect the expression level of IPA1.

• Supplemental Figure 17. IPA1 does not interfere with the DNA binding ability of OsSHI1.

• Supplemental Figure 18. A proposed working model for OsSHI1 in the regulation of plant architecture in rice.

• Supplemental Table 1. Phenotypic segregation identified by test for goodness-of-fit.

• Supplemental Table 2. Putative interacting partners of OsSHI1 identified by Y2H screening.

• Supplemental Table 3. Primers used for map-based cloning.

• Supplemental Table 4. Primers for detection of the RNA levels of selected genes.

• Supplemental Table 5. Primers used for vector constructions.

• Supplemental Table 6. Primers used for ChIP-qPCR analyses.

• Supplemental Table 7. DIG-labeled oligonucleotides used for EMSAs.

Dive Curated Terms

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

• TPL Gramene: AT1G15750

• TPL Araport: AT1G15750

• HY5 Gramene: AT5G11260

• HY5 Araport: AT5G11260

• LAS Gramene: AT1G55580

• LAS Araport: AT1G55580

• STY1 Gramene: AT3G51060

• STY1 Araport: AT3G51060

• STY2 Gramene: AT4G36260

• STY2 Araport: AT4G36260

• WRKY45 Gramene: AT3G01970

• WRKY45 Araport: AT3G01970

• BBX21 Gramene: At1g75540

• BBX21 Araport: At1g75540

• BBX22 Gramene: At1g78600

• BBX22 Araport: At1g78600