Sugar starvation activates the OsSnRK1a‐OsbHLH111/OsSGI1‐OsTPP7 module to mediate growth inhibition of rice

Summary

Sugar deficiency is the persistent challenge for plants during development. Trehalose‐6‐phosphate (T6P) is recognized as a key regulator in balancing plant sugar homeostasis. However, the underlying mechanisms by which sugar starvation limits plant development are unclear. Here, a basic helix–loop–helix (bHLH) transcription factor (OsbHLH111) was named starvation‐associated growth inhibitor 1 (OsSGI1) and the focus is on the sugar shortage of rice. The transcript and protein levels of OsSGI1 were markedly increased during sugar starvation. The knockout mutants sgi1‐1/2/3 exhibited increased grain size and promoted seed germination and vegetative growth, which were opposite to those of overexpression lines. The direct binding of OsSGI1 to sucrose non‐fermenting‐1 (SNF1)‐related protein kinase 1a (OsSnRK1a) was enhanced during sugar shortage. Subsequently, OsSnRK1a‐dependent phosphorylation of OsSGI1 enhanced the direct binding to the E‐box of trehalose 6‐phosphate phosphatase 7 (OsTPP7) promoter, thus rose the transcription inhibition on OsTPP7, then elevated trehalose 6‐phosphate (Tre6P) content but decreased sucrose content. Meanwhile, OsSnRK1a degraded phosphorylated‐OsSGI1 by proteasome pathway to prevent the cumulative toxicity of OsSGI1. Overall, we established the OsSGI1‐OsTPP7‐Tre6P loop with OsSnRK1a as center and OsSGI1 as forward, which is activated by sugar starvation to regulate sugar homeostasis and thus inhibits rice growth.

Introduction

The higher plants tightly control sugar levels to maintain energy homeostasis and ensure survival in changing environments throughout the lifecycle. The “blood sugar” of plants is sucrose (Suc) (Eom et al., 2012). Suc produced by seeds during germination and early seedling stages, and by photosynthesis of leaves (source organ) during the vegetative stage and reproductive stage, is the energy source of the embryonic axis, sink organs and filling grain (Wingler, 2018). In addition, Suc acts as a regulator of gene transcription. For instance, the transcript level of basic leucine zipper 63 (AtbZIP63) is inhibited by Suc in Arabidopsis thaliana (Sanagi et al., 2018). ADP‐glucose pyrophosphorylase (AGPase), the rate‐limiting enzyme in starch biosynthesis, is activated by Suc while not by glucose or fructose in potatoes and Arabidopsis (Harn et al., 2000). In A. thaliana, Suc induces the expression of AtWRKY20 and promotes its transcription initiation on AGPase gene ApL3 (Nagata et al., 2012). The transcript level of AtbZIP11 is induced by Suc, but its protein level is repressed by Suc (Ma et al., 2011). In Oryza sativa, the expression of OsNAC23 is up‐regulated by Suc (Li et al., 2022).

The core of energy homeostasis, which integrates information on nutrient status and environmental conditions, is Sucrose non‐fermenting‐1 (SNF1)‐related protein kinase 1 (SnRK1) (Hardie, 2007). Under energy shortage stresses, including sugar starvation, flood, and darkness, activated SnRK1 promotes survival and stress tolerance by regulating massive genes transcription and metabolic enzymes activities, which in turn controls metabolism and ultimately inhibits growth (Baena‐González et al., 2007; Cho et al., 2012). In A. thaliana, the SnRK1.1 dominant negative mutant is more sensitive to submergence than the wild type (Cho et al., 2016). In O. sativa, SnRK1a RNAi mutant exhibits slower rate of germination and growth of root and shoot, due to decreased expression of MYBS1 and α‐amylase 3 (αAmy3) (Lu et al., 2007). OsCIPK15 promotes seeds’ germination and seedlings growth in flood and semi‐submerged fields, as it regulates SnRK1a to produce more sugar and energy to promote flood tolerance of rice (Lee et al., 2009). There are one catalytic subunit α and two regulatory subunits β and γ in the heterotrimeric complex SnRK1 (Polge and Thomas, 2007). As a kinase, SnRK1 controls many metabolism processes by binding and phosphorylating transcription factors (TFs). As reported, the phosphorylation of TF AtbZIP63 by SnRK1 during starvation leads to changes in its dimerization preference, target gene transcription and primary metabolism, and ultimately affects starvation response and leaf shape of A. thaliana (Mair et al., 2015; Sanagi et al., 2018). SnRK1 cooperates with the seed maturation essential regulator B3‐domain TF FUSCA3 (FUS3) in developmental phase transitions and organogenesis, and SnRK1‐dependent phosphorylation delays the degradation of FUS3 (Tsai and Gazzarrini, 2012). In rice, OsSnRK1 phosphorylates and degrades OsNAC23, the TF increases rice yields (Li et al., 2022).

Uridine diphosphate‐glucose (UDPG) and glucose‐6‐phosphate (Glc6P) generate trehalose 6‐phosphate (Tre6P) under catalysis of Tre6P synthase (TPS), and generates trehalose by dephosphorylation via Tre6P phosphatase (TPP) (Avonce et al., 2006). In plants, the increased Tre6P inhibits SnRK1 activity under sufficient sugar supply and leads to expression block of stress survival response genes while expression activation of growth‐related genes (Nunes et al., 2013; Zhang et al., 2009b). Therefore, Tre6P integrates sugar levels, SnRK1 activity and SnRK1‐controlled genes (Delatte et al., 2011). Tre6P also acts as signal and specific regulator of Suc in plants, similar to glucagon and insulin (Figueroa and Lunn, 2016). Tre6P ensures the level of Suc within normal range: high level Suc causes Tre6P increased, and then Tre6P leads to Suc level decrease by negative feedback regulation, Tre6P finally returns to normal as Suc level decline (Figueroa and Lunn, 2016). Therefore, there are close associations among Tre6P, Suc, and SnRK1. Although Tre6P inhibits Arabidopsis growth with high concentration (Schluepmann et al., 2004), it indeed act as a vital signalling molecule during plants growth, including seed germination, vegetative growth, and reproduction, similar to phytohormones (Figueroa and Lunn, 2016). The Arabidopsis TPS1 mutant tps1 fails to germinate at the torpedo stage due to lack of Tre6P, but rescued by exogenous expression of Escherichia. coli TPS (Eastmond et al., 2002; Schluepmann et al., 2003). In O. sativa, OsTPP7 improves germination under anaerobic stress by increasing the turnover of Tre6P (Kretzschmar et al., 2015). OsNAC23 inhibits OsTPP1 transcription and increases Tre6P content of rice (Li et al., 2022). Hitherto, the regulatory mechanism for rice growth limit under sugar deficiency is elusive.

The transcription activation and inhibition of specific genes regulated by TFs are vital for plants responses to endogenous and exogenous stimuli and survival (Feller et al., 2011). The bHLH TF family contains 167 members in Arabidopsis and 177 members in rice (Carretero‐Paulet et al., 2010; Li et al., 2006). The typical bHLH domain is comprised of a basic region for DNA binding and an HLH region for protein homodimerization and heterodimerization (Feller et al., 2011). The bHLH TFs regulate plant development and stress responses. In O. sativa, OsbHLH035 mutants exhibit delayed germination, particularly under salt stress (Chen et al., 2018). Overexpression of OsbHLH073 results in a semi‐dwarfism phenotype (Lee et al., 2020). OsPGL1 increases rice grain size and weight (Heang and Sassa, 2012). A recent report suggests that the Hd1 binding protein (OsbHLH111) binds with the Partner of HBP1 (POH1) to initiate the transcription of Heading date 1 (Hd1), thus negatively controlling the flowering time of rice (Yin et al., 2023). However, there are no studies about the role of bHLH TFs in Suc shortage of rice. Here, the function mechanism of OsbHLH111 in Suc shortage was explored, and OsbHLH111 was named starvation‐associated growth inhibitor 1 (OsSGI1). By generating mutants and overexpression lines, we unravel the function of OsSGI1 in inhibiting rice growth and subsequently proposed the OsSnRK1a‐OsSGI1‐OsTPP7 module to explain the growth inhibition mechanism under sugar starvation.

Results

Expression of OsbHLH111/OsSGI1 was activated by sugar starvation

Endosperm is the main energy source during seed germination. We performed endosperm excision experiments to simulate energy deficiency in order to search new energy regulator. Data showed that seed germination process was significantly inhibited after endosperm excision, such as radicle and germ growth (Figure 1a). Surprisingly, the hypothetical bHLH TF gene, OsbHLH111/OsSGI1 (Os04g0489600), was markedly activated by energy deficiency in the radicle (Figure 1b). A further investigation indicated the transcripts of OsSGI1 in seedling also surged under carbon starvation treatment by darkness (Figure 1c), and the addition of Suc significantly inhibited the increased trend of OsSGI1, which induced by subsequent carbon starvation (Figure 1c). We inferred that OsSGI1 plays an important role in the energy deficiency caused by sugar starvation.

Figure 1.

The expression profile of OsSGI1. (a) The growth of radicle and germ under sugar starvation. ‐S: Sugar starvation. (b) and (c) The transcript level of OsSGI1 in radicle (b) and seedling leaves (c) under sugar starvation treatment by qRT‐PCR. (d) The transcript level of OsSGI1 in the endosperm by qRT‐PCR. The time after the 2‐days of seed‐soaking period was marked as 0 h. (e) The transcripts of OsSGI1 in the 1‐, 3‐, 5‐, 7‐, and 10‐days roots by qRT‐PCR. (f) The expression profile of OsSGI1 in different tissues and developmental stages by qRT‐PCR. β‐actin was used as the internal gene. Each dot represents a repetition. Error bars indicate mean ± SE. Different letters indicate significant differences at P < 0.01 (capital letter) or P < 0.05 (small letter) level by Tukey's test. ***P < 0.001, Student's t‐test. (g) The GUS signals of OsSGI1pro::GUS plants.

The spatial and temporal expression profile of OsSGI1 was explored in various tissues from gemination stage to reproductive stage. The quantitative real‐time PCR (qRT‐PCR) results showed that the transcripts of OsSGI1 in the germinated‐endosperm were remarkably elevated at 2 days after germination (DAG) and subsequently maintained at a high expression level (Figure 1d). In young seedlings before 10 DAG, OsSGI1 transcripts in the root increased with growth (Figure 1e). The expression profile from 2‐week to reproductive stage exhibited that the OsSGI1 transcripts were higher in roots compared to other tissues (Figure 1f). Intriguingly, the expression level of OsSGI1 in roots was continuously enhanced with rice maturity (Figure 1f). We also generated OsSGI1pro::GUS lines to confirm the expression profile of OsSGI1. The GUS signals were detected in various tissues and enhanced in the root (Figure 1g). These results suggested that OsSGI1 was activated with seed germination and more highly expressed in the roots.

OsSGI1 inhibited rice growth and grain size

To study the biological roles of OsSGI1, CRISPR‐Cas9 technology was used for OsSGI1 mutants and we ultimately obtained three independent knock‐out mutants, thereafter named sgi1‐1/2/3, all of which caused premature translation termination due to a mutation at the first exon (Figures 2a and S1). Meanwhile, we generated stable transgenic plants expressing ubi::OsSGI1 construction in wild‐type (Oryza sativa ssp. japonica, cv. Nipponbare [Nip]) background. The qRT‐PCR results confirmed that OsSGI1 was highly expressed at these over‐expression (OE) lines (Figure 2b). At germination stage, compared with Nip, sgi1‐1/2/3 mutants had increased germination rate and germination index, whereas those of OE‐1/2/3 were decreased (Figure 2c–e). The seedlings of sgi1‐1/2/3 were conspicuously stronger than Nip, whereas OE‐1/2/3 exhibited a dwarf growth phenotype and weak roots (Figure 2f). We counted agronomic traits at the tillering stage and the results showed that the root length, root number, fresh weight, and plant height of sgi1‐1/2/3 mutants were significantly increased, whereas those of OE‐1/2/3 lines were reduced (Figure 2g–j). Moreover, sgi1‐1/2/3 exhibited markedly rise in grain length, grain width, and 1000‐grain weight than Nip (Figure 2k–n), while OE‐1/2/3 had shorter grain length and lighter 1000‐grain weight (Figure 2k–n). Accordingly, OsSGI1 was an important growth inhibitor in response to sugar starvation stress for rice.

Figure 2.

OsSGI1 Inhibited rice growth. (a) Mutation sites of knockout lines sgi1‐1/2/3. (b) The transcript level of OsSGI1 in OE‐1, OE‐2 and OE‐3 by qRT‐PCR. β‐actin was used as the internal gene. Each dot represents a repetition. Error bars indicate M ± SE. ***P < 0.001, Student's t‐test. (c)–(e) The germination status (c), gemination rate at the 3‐day (d), and germination index (e) of Nip, sgi1‐1/2/3, and OE‐1/2/3. The time after the 2‐days of seed‐soaking period was marked as 0 h. DAG: Days after germination. The bar was 1 cm. (f) The seedling phenotypes of rice lines. The colour images showed 30‐day‐old seedling, and the black‐and‐white photos showed 21‐day‐old seedling roots. The bar was 1 cm. (g)–(j) The statistics analyses of root length (g), root number (h), fresh weight (i), and plant height (j) of 21‐day‐old Nip, sgi1‐1/2/3, and OE‐1/2/3. (k) The seed images of Nip, sgi1‐1/2/3, and OE‐1/2/3. The bar was 1 cm. (l)–(n) The statistics analysis of grain length (l), grain width (m), and 1000‐grain weight (n) of Nip, sgi1‐1/2/3, and OE‐1/2/3. Each dot represents a repetition. Error bars indicate mean ± SE. Different letters indicate significant differences at P < 0.01 (capital letter) or P < 0.05 (small letter) level by Tukey's test.

Sugar starvation enhanced the direct binding of OsSGI1 and OsSnRK1a

The yeast two‐hybrid (Y2H) assay was employed to investigate the function mechanism of OsSGI1. OsSGI1 remained inactive in self‐activation assay (Figure S2). Then, we screened the yeast library of rice seedling with OsSGI1 as a bait (Figure S2) and obtained 10 putative interacting proteins (Figure S2). Considering that OsSGI1 may be involved in energy regulation processes, we focused our work on OsSnRK1a, which has been reported as a key energy sensing hub (Hardie, 2007). OsSnRK1a was also inactive in self‐activation assay (Figure S2), and the Y2H point‐to‐point assay proved the binding of OsSnRK1a and OsSGI1 (Figure 3a). Different from the nucleus‐cytoplasm localization of GFP protein, OsSGI1‐GFP localized to the nucleus in Arabidopsis protoplasts and N. benthamiana, and OsSnRK1a‐GFP showed nucleus‐cytoplasm localization (Figures 3b and S3). Bimolecular fluorescence complementation (BiFC) assays confirmed that OsSGI1 was associated with OsSnRK1a in the nucleus (Figure 3c). The in vitro GST pull‐down assay suggested that His‐OsSGI1 directly binds to GST‐OsSnRK1a, in contrast to GST+ His‐OsSGI1 control (Figure 3d). Given that OsSnRK1a was also up‐regulated under sugar starvation (Figure S4), the interaction between OsSGI1 and OsSnRK1a under sugar starvation was explored. We prepared polyclonal antibodies and carried out coimmunoprecipitation (Co‐IP) experiment in rice seedlings. Results not only confirmed that OsSGI1 bound with OsSnRK1a in vivo but also suggested this binding was enhanced in the induction of Suc starvation treatment (Figures 3e and S5). Above results proved that sugar starvation enhanced the direct binding of OsSGI1 to OsSnRK1a.

Figure 3.

The binding of OsSGI1 and OsSnRK1a was enhanced by sugar starvation. (a). Y2H experiment proved the binding between OsSGI1 and OsSnRK1a. OsSGI1‐pGBKT7 + OsSnRK1a‐pGADT7 was the experimental group, pGBKT7‐P53 + pGADT7‐largeT was the positive control, pGBKT7‐Laminc + pGADT7‐largeT was the negative control. (b) The subcellular localization of OsSGI1 and OsSnRK1a in Arabidopsis protoplasts. Protoplasts transfected with pCMBIA1300‐GFP was the control. The bar was 10 μm. (c) The interaction between OsSGI1 and OsSnRK1a in tobacco by BiFC. nYFP+ cYFP, nYFP+ cYFP‐SGI1, cYFP+ nYFP‐OsSnRK1a were the control groups. (d) GST pull‐down and WB assays proved the directly interaction between His‐OsSGI1 and GST‐OsSnRK1a. (e) Co‐IP and WB assays proved the increased interaction between OsSGI1 and OsSnRK1a under Suc starvation treatment for 6 h. 2‐week‐old seedling protein was extracted. Actin was used as the internal control. (f) The expression of OsSGI1 and OsSnRK1a in the seedlings under sugar starvation by WB. 2‐week‐old seedling was treated with Suc starvation for 6 h and then protein was extracted. Actin was used as the internal control. (g) The contents of disaccharide including Suc, fructose, glucose, and trehalose in the roots of 2‐week‐old Nip, sgi1‐1, and OE‐1. (h) The contents of Tre6P in 7‐day‐old Nip, sgi1‐1, and OE‐1. (i) The SnRK1 activities of 21‐day‐old Nip, sgi1‐1, and OE‐1. Each dot represents a repetition. Error bars indicate mean ± SE. Different letters indicate significant differences at P < 0.01 (capital letter) or P < 0.05 (small letter) level by Tukey's test.

The protein levels of both OsSnRK1a and OsSGI1 were induced by sugar starvation, in line with their transcriptional activation (Figure 3f). We hypothesized that the sugar starvation activates SnRK1a and transmits signal to the OsSGI1 for tuning sugar metabolism. Therefore, we measured the OsSGI1‐mediated changes in the endogenous sugar‐related substances content. By contrasted with Nip, the contents of disaccharide, including Suc, fructose, glucose, and trehalose, were all decreased in OE‐1 but increased in sgi1‐1 (Figure 3g). Conversely, Tre6P content rose in OE‐1 but fell in sgi1‐1 (Figure 3h). Unexpectedly, the OsSnRK1 activity was increased both in OE‐1 and sgi1‐1 (Figure 3i). These results hinted that OsSGI1 might mediate sugar metabolism by regulating Tre6P accumulation in response to sugar deficiency.

OsSnRK1a phosphorylated and degraded OsSGI1 under sugar starvation

A series of experiments were performed to study the regulation of kinase OsSnRK1a on OsSGI1. In vitro, His‐OsSGI1 protein was phosphorylated by GST‐OsSnRK1a which showed weaker auto‐phosphorylation activity (Figure 4a). The His‐OsSGI1 protein was incubated with GST‐OsSnRK1a protein for phosphorylation reaction, and LC–MS/MS data showed that OsSGI1 was phosphorylated at serine‐167 only in the presence of exogenetic ATP (Figure S6). Thus, we proposed that Serine‐167 was the OsSnRK1a‐mediated phosphorylation site of OsSGI1. Then, the OsSGI1 and OsSnRK1a proteins were expressed in tobacco and Arabidopsis protoplasts to study their relationship. The protein level of OsSGI1 was significantly decreased in the presence of OsSnRK1a under sugar starvation (Figure S7). Phos‐tag assay was used for further analysis, and the phosphorylated‐OsSGI1 was proved by lambda phosphatase (λPP) treatment (Figure 4b). Surprisingly, the phosphorylated OsSGI1 protein level was more significantly inhibited upon co‐expression of OsSnRK1a, as compared to the non‐phosphorylated OsSGI1 (Figure 4b). Following, the protein biosynthesis inhibitor CHX and proteasome inhibitor MG132 were employed to analyse the degradation mechanism of OsSGI1. OsSGI1‐Myc was decreased after CHX treatment and this degradation was inhibited by MG132 (Figure 4c). Notably, the protein degradation was accelerated under sugar starvation (Figure 4c). We thus hypothesized that sugar starvation activated OsSnRK1a to promote OsSGI1 degradation. Compared with the GFP control (Figure 4d), OsSGI1‐Myc showed faster degradation in the presence of OsSnRK1a‐GFP under CHX treatment (Figure 4e). Expectedly, under sugar starvation, phosphorylated OsSGI1 accumulated significantly after undergoing CHX + MG132 treatment (Figure 4f). Accordingly, OsSnRK1a phosphorylated OsSGI1 and promoted its degradation through proteasome pathway under sugar starvation.

Figure 4.

The regulation of OsSnRK1a on OsSGI1. (a). The phosphorylation between GST‐OsSnRK1a and His‐OsSGI1 by autoradiography (Typhoon 9410). The control was MBP‐His+GST protein. CBB: coomassie brilliant blue staining. (b) The phosphorylation of OsSGI1‐Myc by OsSnRK1a‐GFP and in Arabidopsis protoplasts by phos‐tag PAGE gel. λPP: lambda protein phosphatase, pCMBIA1300‐GFP + OsSGI1‐pCMBIA1300‐Myc was the control. (c) The protein level of OsSGI1‐Myc in Arabidopsis protoplasts under sugar starvation, CHX (80 μM) or MG132 (25 μM) treatments. (d) and (e) The protein level of OsSGI1‐Myc in GFP (d) or OsSnRK1a‐GFP (e)‐expressed Arabidopsis protoplasts under CHX (80 μM) or MG132 (25 μM) treatments. (f) The level of phosphorylated‐OsSGI1 in OsSnRK1a‐GFP expressed Arabidopsis protoplasts under sugar starvation or CHX (80 μM) + MG132 (25 μM) treatments by phos‐tag gel.

OsSnRK1 enhanced the direct transcription repression of OsSGI1 on OsTPP7

RNA‐seq were employed to analyse the effects of OsSGI1 on global genes expression in leaf and root. DEGs analysis showed that OsSGI1 overexpression changed numerous genes transcription. There were 2818 up‐regulated and 8570 down‐regulated genes in leaves (OE‐1 vs Nip, set 1), and 1107 up‐regulated and 2258 down‐regulated genes in roots (OE‐1 vs Nip, set 2) (Figure 5a,b). The KEGG pathway rich results indicated that set 1 DEGs were mainly involved in carbon, starch, and sucrose metabolism (Figure 5c). Differently, the set 2 DEGs were mainly involved in the MAPK signalling, phenylpropanoid biosynthesis, and phenylalanine metabolism (Figure 5d). There were 1580 overlapped genes between DEGs of set1 and DEGs of set2 (Figure 5e). The KEGG pathway rich results indicated that these 1580 genes were mainly involved in the photosynthesis, the MAPK signalling pathway, and plant‐pathogen interaction (Figure 5f). Considering the changes of Tre6P level in OsSGI1 mutants and OE lines, the expression of OsTPPs was focused. Interestingly, compared to Nip, OsTPP family genes, including OsTPP1, OsTPP2, OsTPP7, and OsTPP9, were all down‐regulated both in OE‐1 leaves and roots, OsTPP3 was reduced in the root of OE‐1, OsTPP6 declined in the leaves of OE‐1, OsTPP4 was up‐regulated both in in OE‐1 leaves and roots, and OsTPP5, OsTPP8, and OsSnRK1a were unchanged in leaves and roots of OE‐1 (Figure 5g).

Figure 5.

RNA‐seq analysis of genes regulated by OsSGI1. (a) and (b) Volcano plot of DEGs in leaves (a) and roots (b) of Nip VS OE‐1. (c) and (d) The KEGG pathway rich map of leaves (c) and roots (d) DEGs of Nip VS OE‐1. (e) The overlapped genes number between DEGS in leaves and DEGs in roots. (f) The KEGG pathway rich map of overlapped DEGs in (e). (g) The heatmaps of DEGs clustering in Nip VS OE‐1.

The changes of TPPs transcripts prompted us to study about the regulation of TF OsSGI1 on TPPs transcription. The dual‐luciferase reporter assay data showed that OsSGI1 protein directly repressed the promoter activity of OsTPP7 under sugar starvation (Figure 6a). The bHLH TFs has been reported binds to the conservative motifs, such as E‐box (CACATG) and G‐box (CACGTG) cis‐elements (Li et al., 2006). Yeast one‐hybrid (Y1H) assays indicated that OsSGI1 could bind to the G‐box and E‐box, compared with the positive control P53‐AbAi+pGADT7‐Rec53 and blank load controls pGbox‐AbAi/pEbox‐AbAi + pGADT7 (Figure 6b). There were three G‐box or E‐box elements in the 3 kb promoter region of OsTPP7. Following, EMSA suggested that the binding between OsSGI1 and P2/P3 probes containing E‐box was stronger than P1 probe with G‐box (Figure S8), the specific binding of P2 and P3 probes to OsSGI1 was proved by the addition of cold and mutant probes (Figure 6c,d). Accordingly, OsSGI1 directly repressed the transcription of OsTPP7. The expression profile analysis showed that OsTPP7 was distributed in all tissues, and peaked in 3‐week‐old roots and tillering stage leaves (Figure S9). Furthermore, we analysed the phenotypes of OsTPP7 knockout mutants. Two homozygous lines tpp7‐1 and tpp7‐2 exhibited reduced growth compared with the wild type ZH11 (Figures 6e and S10). We measured several growth parameters and trehalose metabolism intermediates in these mutants. The root length, fresh weight, plant height, and trehalose content of tpp7‐1/2 were significantly lower than those of ZH11, whereas the Tre6P content was markedly higher (Figure 6f–j).

Figure 6.

OsSGI1 directly repressed the transcription of OsTPP7. (a) The effect of OsSGI1 on the promoter activity of OsTPP7 by dual‐luciferase reporter assay under sugar starvation. The control: pCMBIA1300‐Myc + pGreenII 0800‐pTPP7‐Luc; experiment group: OsSGI1‐pCMBIA1300‐Myc + pGreenII 0800‐pTPP7‐Luc. Each dot represents a repetition. Error bars indicate mean ± SE. Different letters indicate significant differences at P < 0.05 level by Tukey's test. (b) The direct binding of OsSGI1 to G‐box and E‐box of OsTPP7 promoter by Y1H. AD: pGADT7. AD‐Rec53 + p53‐AbAi was the positive control. AD+pGbox/Ebox‐AbAi were the empty vector controls. (c) and (d) The direct binding of OsSGI1 protein with P2 (c) and P3 (d) probes by EMSA. Cold probe with 20× and 50×. (e) The 6‐week‐old seedling phenotypes of tpp7‐1/2. The bar was 1 cm. (f)–(h) The statistics analyses of root length (f), plant height (g), and fresh weight (h) of 6‐week‐old ZH11 and tpp7‐1/2. (i) The contents of Tre6P in 10‐day‐old ZH11 and tpp7‐1/2. (j) The content of trehalose in the leaves of 6‐week‐old ZH11 and tpp7‐1/2. (k) The effect of OsSnRK1a kinase on the direct binding of OsSGI1 with P2 and P3 probes by EMSA. (l) The transcript levels of OsTPP7 in different mutants under sugar starvation (dark) by qRT‐PCR. β‐actin was used as the internal gene. Each dot represents a repetition. Error bars indicate mean ± SE. Different letters indicate significant differences at P < 0.05 level by Tukey's test.

Considering the relationship between OsSGI1 and OsSnRK1a, the effect of OsSnRK1a on OsTPP7 was studied. Interestingly, after phosphorylated by OsSnRK1a, the binding of OsSGI1 to P2 probe was enhanced, while unchanged of P3 probe (Figure 6k). In addition, the transcription changes of genes in mutants provided new evidence for the regulatory relationship among OsSnRK1a, OsSGI1, and OsTPP7. The qRT‐PCR data indicated that the transcript level of OsSGI1 was unchanged in the OsSnRK1a knockout mutant snrk1a (Figures S11 and S12), compared with that of Nip (Figure S12). Similarly, the expression of OsSnRK1a was unchanged in the sgi1‐1 and OE‐1, compared with that of Nip (Figure S12). Accordingly, OsSnRK1a and OsSGI1 were unrelated on transcription. However, the transcript level of OsTPP7 was up‐regulated by sugar starvation in Nip, which exhibited more obviously increase trend in the knockout mutants, including sgi1‐1 and snrk1a in CK and sugar starvation groups, while was markedly down‐regulated in the OE‐1 line (Figure 6l). Above results suggested that OsSnRK1a enhanced the binding of OsSGI1 to the E‐box of OsTPP7 promoter, thereby promoting the repression of OsTPP7 transcription.

Discussion

The function of bHLH TFs in rice has been widely studied. For instance, OsPIL15 inhibits grain size (Ji et al., 2019). OsBLR1 increases leaf angle and grain size (Wang et al., 2020). During life cycle, plants constantly adjust status to survive in sugar deficiency stress. However, relevant studies, especially the role of bHLHs, are not in‐depth. In the present research, OsSGI1, a TF localized in the nucleus (Figure 3), inhibited rice germination, seedling growth, and grain size in response to sugar starvation (Figures 1 and 2). In sugar deficiency stress, the energy hub OsSnRK1a directly bound and phosphorylated OsSGI1 (Figures 3 and 4), then enhanced the binding of phosphorylated‐OsSGI1 to the promoter of Tre6P repressor OsTPP7, thus promoted the transcription inhibition on OsTPP7 and finally increased Tre6P content (Figures 3 and 6). OsSnRK1a also degraded the phosphorylated‐OsSGI1 protein by proteasome pathway (Figure 4), which prevented growth inhibitor OsSGI1 from unremitting working and even death.

OsSGI1 inhibits rice growth in response to sugar deficiency stress

The sharply increased transcripts under sugar starvation but decreased under Suc supplement (Figure 1), as well as the highly transcripts from 0 to 10 days when sugar continuously consumed (Figure 1), all suggested that OsSGI1 was fired under sugar starvation. Different to MYB family member OsMYBS1, a TF which is up‐regulated under glucose starvation (Lu et al., 2007), and NAC family member OsNAC23, a TF which is induced by glucose and sucrose but repressed by sugar starvation (Li et al., 2022), OsSGI1 is first reported as sugar starvation induced bHLH TF.

Well known are the important roles of bHLHs in germination, seedling growth, and grain size. SlPRE2 (bHLH TF) overexpression in tomato promotes morphogenesis during seedling development (Zhu et al., 2017). Overexpression of rice ILI1 (HLH protein) promotes cell elongation and inhibits the dwarf phenotype caused by IBH1‐overexpression (bHLH protein) in Arabidopsis (Zhang et al., 2009a). The OsBUL1 (HLH protein) knockout mutant osbul1 has smaller grains, while overexpression lines exhibit bigger grains (Jang et al., 2017). Here, the slow‐growing OE‐1/2/3 but fast‐growing sgi1‐1/2/3 mutants (Figure 2) suggested that OsSGI1 inhibited germination, seedling growth, and grain size. Interestingly, the increased grain length and width of sgi1‐1/2/3 (Figure 2) were in line with those of hbp1, the knockout mutants of OsHBP1 (OsbHLH111) (Yin et al., 2023). Together with the previously reported agronomic traits of hbp1 at heading, flowering, and grain filling stages, and OE lines phenotypes in rice flowering (Yin et al., 2023), these two separate studies not only provide a comprehensive characterization of the role of OsSGI1 throughout the rice life cycle but also suggest that OsSGI1 has complex functions both in growth and stress. Notably, insufficient sugar led to rice growth inhibition (Figure 1), and the phenotypes of OsSGI1 mutants were also consistent with its increased expression under sugar starvation and Suc changes of mutants (Figures 2 and 3). Thus, our research provided experimental evidence to the function studies of bHLHs and, importantly, proved that OsSGI1 was rice growth inhibitor in response to sugar starvation.

OsSGI1 represses OsTPP7 transcription and increases Tre6P content

The content of Tre6P was increased in OE‐1 but decreased in sgi1‐1 (Figure 3). And TF OsSGI1 was more likely to inhibit genes transcription than up‐regulate transcription from RNA‐seq data (Figure 5). The transcript level of OsTPP7 was down‐regulated in OE‐1 but up‐regulated in sgi1‐1 (Figure 6). Interestingly, we proved that OsSGI1 directly repressed the transcription of OsTPP7 by binding to the E‐box (Figure 6). This binding is consistent with the previously studies about the E‐box element binding ability of bHLH TF, such as OsbHLH002, which directly initiates OsTPP1 transcription by binding to the E‐box element (Li et al., 2006; Zhang et al., 2017). In addition to Hd1 (Yin et al., 2023), we demonstrated that OsbHLH111 also regulates the expression of OsTPP7. Surprisingly, OsTPP7 was up‐regulated under sugar starvation (Figure 6), imply that there are OsTPP7 transcription activators in addition to OsSGI1, which needs further study. The role of TPP in the catalysis of Tre6P to trehalose is well‐known (Avonce et al., 2006), and OsTPP family transcripts partially decreased in OE‐1 (Figure 5). According to previous reports, OsTPP7 has no effect on Tre6P content in O. sativa, and OsTPP7 enhances the turnover of Tre6P to trehalose (Kretzschmar et al., 2015). However, this discrepancy could be explained by the different experimental conditions and tissue sources. The former study measured Tre6P content in embryos with coleoptiles that germinated underwater in the dark, whereas we measured Tre6P content in seedlings grown under normal conditions (Figure 6). Underwater germination is a stressful situation that requires survival as the primary goal. Therefore, it is possible that other OsTPPs compensate for the loss of OsTPP7 and catalyse the conversion of Tre6P to trehalose, which confers stress tolerance. Therefore, the content changes of Tre6P in OE‐1 and sgi1‐1 were reasonable (Figure 3). Different to OsNAC23 increases Tre6P content of rice by directly repressing the transcription of OsTPP1 (Li et al., 2022), the present study developed the new idea about the roles of OsSGI1 and OsTPP7 in controlling Tre6P content.

Tre6P is the availability signal and negative feedback regulator of Suc (Schluepmann et al., 2003). The E. coli TPS1 overexpression Arabidopsis plants 35s:ostA sense a high Suc signal which represented by high Tre6P, and then negatively feedback to reduce Suc content. Therefore, 35s:ostA exhibits growth inhibition with increased Tre6P and decreased Suc (Schluepmann et al., 2003). Here, the altered Tre6P content is linked with the phenotypes of OsSGI1 and OsTPP7 mutants. OE‐1/2/3 with slower growth phenotypes exhibited increased Tre6P and reduced Suc (Figures 2 and 3). The fast‐growing sgi1‐1 exhibited decreased Tre6P, and thus increased Suc by negatively feedback (Figures 2 and 3). The inhibition growth of tpp7‐1/2 was consistent with its increased Tre6P content (Figure 6). OsSGI1 inhibited the transcription of OsTPP7 (Figure 6), and both OE‐1/2/3 and tpp7‐1/2 exhibited growth inhibition (Figures 2 and 6), which also proved their regulatory relationship. Therefore, the links among OsSGI1‐dependent transcription repression of OsTPP7, the Tre6P content changes, and the phenotypes of OsSGI1 mutants are powerful.

The push and pull regulation of OsSnRK1a on OsSGI1 ensure energy homeostasis even survival of rice

In the present study, OsSnRK1a was proved as binding partner and phosphorylation kinase of OsSGI1, and the binding was increased under sugar starvation (Figure 3). SnRK1a acts as binding protein and kinase for many TFs. For instance, SnRK1a binds and phosphorylates AtbZIP63 (Mair et al., 2015). FUS3, a crucial regulator of seed maturation in Arabidopsis, acts as binding partner and phosphorylation substrate of SnRK1a (Chan et al., 2017). The Arabidopsis ATAF1, a member of NAC TF family, interacts with SnRK1a (Kleinow et al., 2009). In rice, SnRK1a phosphorylates TF OsNAC23 (Li et al., 2022). Our study creatively reported about the kinase‐substrate relationship between SnRK1a and bHLH TF. The serine‐167 of OsSGI1 was preliminarily identified as phosphorylation site by OsSnRK1a (Figure S6). As reported, the SAMS peptide is commonly used for AMPK kinase (homology of SnRK in mammalian) activity measurement. It is known that the common sequence of SnRK1‐mediated phosphorylation site of HMG‐CoA reductases, nitrate reductases, and sucrose phosphate synthases is the phosphorylated serine or threonine (0) is preceded by a residue (M/L/V/F/I) at position −5 and a residue (R/K/H/X) at position −3, and followed by a residue (L/F/I/M/V) at position +4 (Halford and Hardie, 1998). Although serine‐167 of OsSGI1 (MKAPLSFASS) does not fully match this motif, the specific phosphorylation site of OsSGI1 needs to be confirmed by site‐directed mutagenesis in future experiments. OsSnRK1a and OsSGI1 were similar in many aspects. First, OsSnRK1a and OsSGI1 transcripts exhibited similar increase trends under sugar starvation (Figures 1 and S4), which were consistent with the increased expression of OsSnRK1a under sugar starvation (Lu et al., 2007). Second, the overexpression line OsSnRK1a‐OX has reduced seed weight, plant height, and root growth (Filipe et al., 2018), which are similar to OE‐1/2/3 (Figure 2). However, snrk1a mutant exhibits slower growth on germination, root, and shoot (Lu et al., 2007), which are opposite to those of sgi1‐1/2/3 (Figure 2). Anyway, above discussion suggested the intense relationship between OsSnRK1a and OsSGI1.

More importantly, OsSnRK1a precisely controlled OsSGI1. The phosphorylation mediated by OsSnRK1a enhanced the direct repression of OsSGI1 on OsTPP7 transcription (Figure 6), and qRT‐PCR also proved that OsTPP7 transcript level was inhibited by OsSnRK1a and OsSGI1 (Figure 6), which led to an increase in Tre6P and a decrease in Suc (Figure 3). The decreased Suc inevitably activates SnRK1a activity (Baena‐González et al., 2007; Cho et al., 2012), and Tre6P inhibits SnRK1a activity only under Suc sufficient status (Zhang et al., 2009b; Nunes et al., 2013). Therefore, SnRK1a activity elevated both in OE‐1 and sgi1‐1 (Figure 3). Different from the phosphorylation of OsNAC23 mediated by OsSnRK1a, which inhibits its binding to OsTPP1 promoter (Li et al., 2022), similar to the OsSnRK1a‐dependent phosphorylation of TF OsMYBS1, which is required for the transcription initiation of αAmy3 under sugar starvation (Lu et al., 2007), the phosphorylation of OsSGI1 by OsSnRK1a enhanced its binding to OsTPP7 promoter (Figure 6). Surprisingly, OsSnRK1a also degraded phosphorylated OsSGI1 (Figure 4), which is consistent with the decreased protein stability of OsNAC23 by OsSnRK1a (Lu et al., 2007). This degradation is related to the growth inhibitory role of OsSGI1. Considering that OsSGI1 reduced thousands of genes transcription and decreased Suc content while elevated Tre6P content (Figures 3 and 5), the continuous working of OsSGI1 will result in rice death inevitably. Therefore, the degradation controlled by OsSnRK1a ensured OsSGI1 in the controllable state. The push and pull regulation of OsSnRK1a on OsSGI1 proved that OsSnRK1a act as energy hub indeed.

The present study sheds light on the mechanism of OsSGI1 in rice survival under sugar starvation (Figure 7): 1. The expression of OsSGI1 and its binding to OsSnRK1a are both increased in response to sugar starvation. 2. OsSnRK1a‐mediated phosphorylation of OsSGI1 enhances its binding to the E‐box of OsTPP7 promoter, thus increases the transcriptional repression on OsTPP7, results in increased Tre6P content while decreased Suc content, and finally growth inhibition. 3. OsSnRK1a degrades phosphorylated‐OsSGI1 protein through proteasome pathway to prevent the accumulation toxicity of OsSGI1 protein.

Figure 7.

The function model of OsSGI1 in rice growth inhibition under sugar starvation stress.

Methods

Plasmids construction

The full‐length coding sequence (CDS) of OsSGI1 was driven by Ubiquitin(ubi) promoter in pTCK303‐OsSGI1 vector for OE‐1/2/3 lines. The OsSGI1‐pET32a and OsSnRK1a‐pGEX‐4T‐1 plasmids with CDS were used for protein expression. The pGreenII 0800‐pOsTPP7::LUC (3050 bp upstream towards ATG of OsTPP7) plasmid for transcription regulation. The OsSGI1‐pGBKT7 and OsSnRK1a‐pGADT7 plasmids for yeast two‐hybrid (Y2H). The OsSGI1‐pCMBIA1300‐GFP, OsSGI1‐pCMBIA1300‐Myc, and OsSnRK1a‐pCMBIA1300‐GFP plasmids for protein expression in tobacco and Arabidopsis protoplasts. The nYFP‐OsSnRK1a and cYFP‐OsSGI1 plasmids for BiFC. The primers sequences are listed in Table S1.

Transgenic rice construction and cultivation

Nipponbare (Nip) was the background for sgi1‐1/2/3, OE‐1/2/3, and snrk1a. The sgi1‐1/2/3 were generated by CRISPR/Cas9 technology as description previously (Ma et al., 2015), and PAM sequence was “GCGATGAGGCAGCAGCACTACGG”. The OsSGI1‐pTCK303 plasmid was used for OE‐1/2/3 lines construction. For OsSGI1pro::GUS lines, the promoter sequence of OsSGI1 (from 2932 bp to 226 bp upstream towards ATG) was ligated to the pCAMBIA1301 LB‐RB binary vector containing GUS reporter. The knockout mutants snrk1a were generated by CRISPR/Cas9 method, and the two PAM sequences were “CAATCCGGTAACCGCCAAGAGGG” and “ATTGGCAAAACCCTAGGGATTGG”. The homozygous seeds of knockout mutants tpp7‐1 and tpp7‐2 were purchased from the company (Biogle, Changzhou, China), which were constructed by the CRISPR/Cas9 method using ZH11 as background. The PAM sequence of tpp7‐1 was “GCTCACCGTCTCGCTCATGAAGG”, and the PAM sequence of tpp7‐2 was “TGTGCAGGCGAGGCACCCGTCGG”. The sterilized seeds were germinated in water after soaking at 30 °C, and grown in half‐strength Yoshida's culture solution for 7 days (Yoshida et al., 2013). Then, 7‐day‐old seedlings were grown in Yoshida's culture solution in a climate chamber with 28 °C light/25 °C dark (14 h/10 h) photoperiod, 60% humidity and ~200 μmol/m²/s light intensity, or were planted in a paddy field.

Gene expression analysis

The TRIzol reagent and FastKing RT Kit (Tiangen, Beijing, China) were used for total RNA extraction and cDNA, respectively. The qRT–PCR was performed in real‐time PCR detection system (CFX96, Bio‐Rad). The internal reference gene was β‐actin. The 2^−▵▵Ct algorithm was used for data analysis referred to previous description (Wang et al., 2020). Three biological replicates and three technical replicates were performed.

Sugar starvation treatment

Protoplasts:the W5 solution contains 154 mM NaCl, 125 mM CaCl₂, 5 mM KCl, 5 mM glucose, and 2 mM MES‐KOH (pH 5.7). W5 solution without 5 mM glucose was used as sugar starvation solution. Arabidopsis protoplasts incubated with W5 or glucose‐without W5 were considered as control and sugar starvation treatment group, respectively.

Radicle: The intact embryos of seeds were isolated as description previously (Yu et al., 1996) and then germinated on dishes with water. Finally, the roots were collected at 0, 2, 3, and 4 DAG. 15 seeds in each group.

Seedlings: The 2‐week‐old seedlings were cultured in darkness for 0, 6, 12, 24 and 36 h. Then, seedlings were treated with or without 10 mM Suc for another 6, 12, and 24 h. The leaves and roots were collected at different time points. At least 5 seedlings at each time point.

Histochemical GUS staining

Tissues, including germinated seeds, 7‐day root and leaf, 14‐day root, 21‐day root, 21‐day shoot, and 21‐day leaf, were stained, decolorized, and photographed as decribed above (Zhao et al., 2021).

Phenotypes identification

The time after 2‐days seed‐soaking was marked as 0 h. The germination index and germination rates (at 72 h) were calculated, and there were 100 seeds per group. The following indexes were recorded for 3‐week‐old seedlings of OsSGI1 transgenetic lines and 6‐week‐old seedlings of tpp7‐1/2: the longest root was taken as “root length”, the visible root number was taken as “root number”, the height of the above‐ground part was regarded as “plant height”, seedlings were weighted with electronic balance for “fresh weight”, grain width and length were measured with Vernier callipers, and 1000 seeds were weighted with electronic balance for “1000‐grain weight”. Three biological replicates and three technical replicates were performed in T2 and T3 generations lines for sgi1‐1/2/3 and OE‐1/2/3.

Subcellular localization

The pCMBIA1300‐GFP, OsSGI1‐pCMBIA1300‐GFP, and OsSnRK1a‐pCMBIA1300‐GFP vectors were used for subcellular localization in Nicotiana benthamiana tobacco as described above (Wang et al., 2020). The pCMBIA1300‐GFP, OsSGI1‐GFP/OsSnRK1a‐pCMBIA1300‐GFP, and NLS‐mCherry plasmids were co‐transformed into Arabidopsis protoplasts by PEG method for 16 h as previously description (Tong et al., 2017). The LSM‐710 laser confocal microscope was used for imaging.

Protein interaction identification

Yeast two‐hybrid assay (Y2H): The self‐activation activity of OsSGI1‐pGBKT7 and OsSnRK1a‐pGBKT7 was tested as previously description (Wang et al., 2020). The OsSGI1‐pGBKT7 contained AH109 cells mated with O. sativa seedling Y2H library (constructed by Shanghai biogene biotechnology company) and performed as previously description (Chang et al., 2022) for Y2H library screening. Point‐to‐point: the OsSGI1‐pGBKT7 and OsSnRK1a‐pGADT7 plasmids were co‐transformed into AH109 competent cells, and then cells were coated on SD‐Trp/‐Leu (SD‐TL), SD‐Trp/‐His/‐Leu (SD‐THL), and SD/‐Trp/‐His/‐Leu/‐Ade (SD‐THLA) plates and then lacZ assay. The pGBKT7‐P53 + pGADT7‐largeT and pGBKT7‐Laminc+pGADT7‐largeT were positive and negative controls, respectively.

GST pull‐down: OsSGI1‐pET32a and OsSnRK1a‐pGEx‐4T1 plasmids were used for protein expression and purification as previous description (Wang et al., 2020). Glutathione beads containing 8 μg GST‐OsSnRK1a or 1.5 μg GST purified proteins were incubated with 1 μg His‐OsSGI1 in pull‐down buffer at 4 °C for 4 h. Then, beads were washed, and proteins were eluted in sodium dodecyl sulphate (SDS) loading buffer. Finally, boiled proteins were examined by western blot (WB) using GST and His antibodies (Abmart, Shanghai, China).

BiFC: The GV3101 chemically competent cells that transformed with OsSnRK1a‐ nYFP+OsSGI1‐cYFP, OsSnRK1a‐cYFP+OsSGI1‐nYFP, nYFP+OsSGI1‐cYFP, OsSnRK1a‐nYFP+cYFP, or nYFP+cYFP were injected into N. benthamiana leaves for 2 day. Subsequently, fluorescence signals were observed by LSM‐710 laser confocal microscope.

Co‐immunoprecipitation (Co‐IP): The purified His‐OsSGI1 and GST‐OsSnRK1a proteins were used to prepare polyclonal antibodies by HuaAn Biotechnology Company (HuaAn, Hangzhou, China). The proteins of 2‐week‐old Nip that grew in MS or sucrose‐without MS solution (Solarbio, M8521) were extracted. After taking the input, the residual solution was rotated with anti‐OsSnRK1a (1 : 100) at 4 °C for 8 h and then incubated with Pierce™ Protein A/G Magnetic beads (ThermoFisher, 88 802) for 2 h to purify target proteins. Finally, samples were examined by WB.

Measurement of endogenous substances contents

After growth in sterile tissue culture bottles with 1/2 MS medium for 7 days of OsSGI1 transgenetic lines and 10 days of OsTPP7 mutants, seedlings without seeds were collected. Tre6P was extracted with 30% acetonitrile and measured by high‐performance liquid chromatography–tandem mass spectrometry (HPLC‐MS/MS) (Agilent 1260 high performance liquid chromatograph in tandem with 6420A mass spectrometer). The Tre6P standard was from Sigma (Sigma, St Louis). The experiments were performed by ProNestBio Company (Wuhan, China). The 2‐week‐old rice roots of OsSGI1 transgenetic lines and leaves of 6‐week‐old OsTPP7 mutants were collected. Sucrose hydrolyzed to glucose and fructose under acidic conditions. Fructose was determined by the resorcinol method, glucose was measured by the glucose oxidase method, and trehalose was measured by anthrone colorimetry. Each group contains 15 seedlings, and three biological replicates were performed.

SnRK1 enzymatic activity

The 21‐day‐old roots were collected for double‐antibodies one‐step sandwich ELISA kit. The samples, standards, and horse radish peroxidase (HRP)‐labelled detection antibody were added to the SnRK1 antibody pre‐coated microwells and incubated. The colour was developed with TMB substrate, and absorbance was measured with a microplate reader at 450 nm. Every group contained 9 seedlings. Three biological replicates were performed.

Phosphorylation level analysis

In vitro: The His‐OsSGI1 (2 μg) and GST‐OsSnRK1a (5 μg) proteins were incubated in kinase reaction buffer (20 mM pH 7.5 Tris–HCl, 20 mM MgCl₂, 50 μM ATP, 1 μCi [γ‐³²P] ATP, and 1 mM DTT) at 30 °C for 30 min. Then, the proteins were separated in 10% SDS‐PAGE gel and visualizes by coomassie bright blue stain and autoradiography (Typhoon 9410). The GST‐OsSnRK1a, His‐OsSGI1, GST+ His‐OsSGI1, and MBP‐His + GST‐OsSnRK1a were control groups.

In vivo: Arabidopsis protoplasts transfected with OsSnRK1a‐pCMBIA1300‐GFP + OsSGI1‐pCMBIA1300‐Myc plasmids were incubated with normal W5 or sugar starvation W5 solutions at 22 °C in the dark for 16 h. Taking normal W5 group as an example, the above protoplasts were treated with phos‐tag protein buffer (50 mM Tris–HCl [pH 7.5], 150 mM NaCl, 10 mM MgCl₂, 1% [v/v] NP‐40, 2 mM NaF, 1 × Cocktail, and 5 mM DTT) for total protein extraction. Then, proteins were treated with or without lambda protein phosphatase (λPP, P0753S, NewEngland Biolabs). These proteins were separated by 8% phos‐tag‐PAGE gel (containing 50 μM phos‐tag AAL solution [Boppard, China] and 150 μM MnCl₂) and 12% SDS‐PAGE gel, respectively. Then, phos‐tag gel was washed in buffer (47.88 mM Tris, 38.64 mM glycine, 20% [v/v] methanol, and 10 mM EDTA [pH 8.0]). Finally, proteins were transferred to polyvinylidene fluoride membranes. The pCMBIA1300‐GFP + OsSGI1‐pCMBIA1300‐Myc was the control.

The phosphorylation site identification: To analyse the phosphorylation sites of OsSGI1 by OsSnRK1a in vitro, His‐OsSGI1 protein (4 μg) was incubated with GST‐OsSnRK1a protein (5 μg) in the kinase reaction buffer (20 mM pH 7.5 Tris–HCl, 20 mM MgCl₂, and 1 mM DTT) with or without 100 μM ATP at 30 °C for 30 min. The reaction was terminated by addition of 5 × loading buffer and these two protein samples were separated in SDS‐PAGE followed by cutting His‐OsSGI1 protein. Then, two samples were digested with trypsin and subjected to a Thermo Q‐Exactive high‐resolution mass spectrometer (Thermo Fisher Scientific) for LC–MS/MS analysis. Results were analysed with the Scaffold PTM software.

Protein degradation assay

Target proteins were expressed in Arabidopsis protoplasts as described above. Then, protoplasts were resuspended by W5 or sugar starvation W5 solution at 22 °C in the dark for 16 h, and then 80 μM Cycloheximide (CHX, protein biosynthesis inhibitor) was added at different times. Protoplasts under sugar starvation and CHX treatments were treated with another 25 μM MG132 (proteasome inhibitor) for another 4 h in the dark. The proteins were extracted for WB or phos‐tag‐PAGE gel.

Bioinformatics analysis of RNA‐sequencing (RNA‐seq) data

The roots and leaves of 3‐week‐old Nip and OE‐1 were collected. The BGISEQ‐500 platform (BGI Genomics, Shenzhen, China) was used for sequencing of generated cDNA libraries. The clean reads were aligned to Nip genome sequence (IRGSP‐1.0) in RAP‐DB. The differentially expressed genes (DEGs) (fold change ≥2 and false discovery rate [FDR] ≤ 0.001) between Nip and OE‐1 were screened by NOISeq method. The heatmaps was drew by TBtools with reference to mean of fragments per kilo base per million mapped reads (FPKM) values.

Transcription regulation on OsTPP7

Dual‐luciferase reporter assay: The pGreenII 0800‐pOsTPP7:LUC was co‐transfected with pCMBIA1300‐Myc or OsSGI1‐pCMBIA1300‐Myc into Arabidopsis protoplasts with W5 or sugar starvation W5 solution at 22 °C in the dark for 16 h. Firefly LUC and REN activities were measured with dual‐luciferase reporter assay kit (Promega, Madison) and GloMax 20/20 luminometer.

Y1H: pGbox‐AbAi and pEbox‐AbAi plasmids were constructed. In the case of pEbox‐AbAi. Y1H Gold competent cells with pEbox‐AbAi or P53‐AbAi were spotted on different concentration Aureobasidin A (AbA) SD‐Ura plates, and the lowest AbA concentration was 100 and 200 ng/mL, respectively. The Y1H Gold competent cells with pGADT7‐OsSGI1+ pEbox‐AbAi were spotted on normal and 100 ng/mL AbA‐containing SD‐Leu plates. pGADT7‐Rec53 + p53‐AbAi and pGADT7+ pGbox‐AbAi were the positive and blank controls.

Electrophoretic mobility shift assay (EMSA): The oligonucleotide sequences of OsTPP7 promoter elements were synthesized and labelled with biotin at 5′‐end. The unlabeled sequence was the cold probe, and mutation probe sequence contained “AAAAAA”. EMSA was performed with the LightShift^® Chemiluminescent EMSA Kit (ThermoFisher, #20148). Briefly, purified His‐OsSGI1 protein was incubated with biotin‐probe, cold (20‐fold and 50‐fold) probe, or mutant probe at 25 °C for 30 min. SnRK1a‐phosphorylation on DNA binding ability of OsSGI1: GST‐OsSnRK1a and His‐OsSGI1 proteins were incubated in kinase reaction buffer containing 50 μM ATP or ATP‐free at 30 °C for 30 min, and then reaction mixture was incubated with probes. Finally, images were photographed by the chemiluminescent imaging system.

Data significant test

Significant differences were analysed by Student's t‐test or Tukey's test.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Authous contributions

Conceptualization: WZ. Data curation: KW and ML. Formal analysis: KW. Funding acquisition: WZ. Investigation: KW, ML, and YC. Methodology: SA. Writing—original draft: KW and WZ. Writing—review and editing: WZ. All authors have read and agreed to the published version of the manuscript.

Supporting information

Figure S1 The mutation sites of sgi1‐1/2/3 by sequencing.

Figure S2 The Y2H library screening results using OsSGI1‐pGBKT7 as bait.

Figure S3 The subcellular localization of OsSGI1 and OsSnRK1a in N. benthamiana.

Figure S4 The transcript level of OsSnRK1a in radicle and seedling leaves under sugar starvation.

Figure S5 Antibodies detection by WB with Nip seedling proteins.

Figure S6 The OsSnRK1a‐mediated phosphorylation site analysis of OsSGI1 in vitro.

Figure S7 Co‐expression of OsSnRK1a inhibited the protein accumulation of OsSGI1.

Figure S8 The binding ability of P1/P2/P3 probe to His‐OsSGI1 protein by EMSA.

Figure S9 The expression profile of OsTPP7 by qRT‐PCR.

Figure S10 The mutation sites of tpp7‐1/2 by sequencing.

Figure S11 The mutant sites of OsSnRK1a knockout mutant snrk1a by sequencing.

Figure S12 The transcript levels of OsSGI1 and OsSnRK1a in mutants under sugar starvation (dark) by qRT‐PCR.

Table S1 Primer sequences used in the present study.

Data S1 The DEGs of leaves between Nip and OsSGI1‐OE1.

Data S2 The DEGs of roots between Nip and OsSGI1‐OE1.

Data S3 The overlapped DEGs.

Data S4 The other supplementary files.

Data availability statement

The raw RNA‐seq data in this paper have been deposited at NCBI under Bioproject with the accession number PRJNA908276.