Sucrose nonfermenting-1-related protein kinase 1 regulates sheath-to-panicle transport of nonstructural carbohydrates during rice grain filling

Abstract

The remobilization of nonstructural carbohydrates (NSCs) reserved in rice (Oryza sativa) sheaths is essential for grain filling. This assimilate distribution between plant tissues and organs is determined by sucrose non-fermenting-1-related protein kinase 1 (SnRK1). However, the SnRK1-mediated mechanism regulating the sheath-to-panicle transport of NSCs in rice remains unknown. In this study, leaf cutting treatment was used to accelerate NSC transport in the rice sheaths. Accelerated NSC transport was accompanied by increased levels of OsSnRK1a mRNA expression, SnRK1a protein expression, catalytic subunit phosphorylation of SnRK1, and SnRK1 activity, indicating that SnRK1 activity plays an important role in sheath NSC transport. We also discovered that trehalose-6-phosphate, a signal of sucrose availability, slightly reduced SnRK1 activity in vitro. Since SnRK1 activity is mostly regulated by OsSnRK1a transcription in response to low sucrose content, we constructed an snrk1a mutant to verify the function of SnRK1 in NSC transport. NSCs accumulated in the sheaths of snrk1a mutant plants and resulted in a low seed setting rate and grain weight, verifying that SnRK1 activity is essential for NSC remobilization. Using phosphoproteomics and parallel reaction monitoring, we identified 20 SnRK1-dependent phosphosites that are involved in NSC transport. In addition, the SnRK1-mediated phosphorylation of the phosphosites directly affected starch degradation, sucrose metabolism, phloem transport, sugar transport across the tonoplast, and glycolysis in rice sheaths to promote NSC transport. Therefore, our findings reveal the importance, function, and possible regulatory mechanism of SnRK1 in the sheath-to-panicle transport of NSCs in rice.

SnRK1, the signaling hub that controls NSC sheath-to-panicle transport via phosphorylation, is regulated by OsSnRK1a expression and T6P level in response to sucrose availability in rice leaf sheath.

Introduction

Grain filling, which determines the grain weight and yield in rice (Oryza sativa), depends on the assimilate supply from two sources—photosynthesis of functional leaves after anthesis and nonstructural carbohydrate (NSC) accumulation in the stems and sheaths. The NSCs reserved in the stems and sheaths contribute to 1/6–1/3 of the grain weight (Okamura et al., 2018), and are mainly composed of starch and sucrose (Perez et al., 1971). Starch is the predominant component of NSCs (Slewinski, 2012). Sheaths experience a sink-to-source transition during grain filling. Starch and sugar accumulate before flowering and then export during the grain filling stage (Togari and Sato, 1954; Perez et al., 1971; Yoshida, 1972). There is also a sink-to-source transition in source leaves in the diurnal cycle. A considerable proportion of photo assimilate synthesized by source leaves is exported to support the growth and development of heterotrophic sink organs, whereas excess photo assimilate is stored in the form of starch in leaves during the day. The starch is then degraded at night to provide carbon and energy to support maintenance and growth (Martins et al., 2013). The accumulation and degradation trends of starch vary in the course of transition between the sheaths and source leaves. The organelles where the starch is reserved also vary. The starch is reserved in the chloroplasts of parenchyma cells in source leaves (Stitt and Zeeman, 2012), and starch has been reported to be stored in leucoplasts of parenchyma cells in the sheath (Sato, 1955). Starch reserved in sheaths is degraded to glucose and glucose-1-phosphate (G1P) by α-amylase, β-amylase, and α-1,4-glucan phosphorylase (Stitt and Zeeman, 2012). Sucrose is re-synthesized by sucrose phosphate synthase (SPS) and sucrose phosphorylase (Schaffer and Petreikov, 1997). The re-synthesized sucrose is loaded into the Sieve Element/ Companion Cell complex through apoplastic or symplastic pathways and transported to the grains via mass flow (Li et al., 2017). The main sucrose transporters (SUTs) in the rice sheaths are OsSUT1 and OsSUT4 (Chen and Wang, 2008). Previous studies revealed that insufficient leaf photosynthesis accelerated sheath NSC transport after plants were exposed to abiotic stress (Slewinski, 2012). In recent years, various omics techniques have assisted in identifying potential molecular mechanisms that facilitate plant NSC transport (Wang et al., 2017, 2020a, 2021). However, the mechanism that mediates the sheath-to-panicle transport of NSCs in rice remains unknown.

Sucrose nonfermenting-1-related protein kinase 1 (SnRK1) is an evolutionarily conserved protein kinase that shares homology with SNF1 in yeasts and AMP-activated protein kinase (AMPK) in mammals (Emanuelle et al., 2016). SnRK1 plays a crucial role in the source–sink interactions and nutrient redistribution between tissues and organs (Lawlor and Paul, 2014; Smeekens, 2015; Yu et al., 2015). Additionally, SnRK1 promotes catabolism and suppresses anabolism through transcriptional and posttranslational regulation in response to starvation via sugar sensing and signaling (Baena-González et al., 2007; Polge and Thomas, 2007). In rice, the three genes encoding SnRK1 are classified into two subfamilies, namely SnRK1a (OSK1) and SnRK1b (OSK24 and OSK35). SnRK1 is involved in the starch and sucrose metabolism of plants (Sugden et al., 1999a, 1999b; Lu et al., 2007). The transcription factor MYBS1 is transcriptionally activated by SnRK1 to upregulate amylase expression in rice embryos during sugar starvation (Lu et al., 2007). Furthermore, SnRK1 directly phosphorylates and inactivates SPS, a key enzyme for sucrose synthesis (Sugden et al., 1999b). The starch content in rice sheaths was also found to be associated with SnRK1 gene expression (Kanegae et al., 2005). In response to sugar availability, SnRK1 activity is inhibited by trehalose-6-phosphate (T6P), the metabolic precursor of trehalose (Paul et al., 2008). T6P inhibits the SnRK1 activity in the seedlings and young leaves of Arabidopsis (Arabidopsis thaliana), but not in the mature leaves (Zhang et al., 2009). Moreover, trehalose phosphate synthase and trehalose phosphorylase, which are involved in T6P metabolism, reportedly play an important role in sheath NSC redistribution (Wang et al., 2017). Therefore, we speculate that SnRK1 may be involved in sheath-to-panicle NSC remobilization and regulated by T6P in rice.

To verify our hypothesis, we performed leaf cutting (LC) treatment (the three upper leaves were cut in half) at anthesis to accelerate NSC remobilization to demonstrate the involvement of SnRK1. An snrk1a mutant was constructed to investigate the function of SnRK1 in NSC transport. In addition, quantitative phosphoproteomic and proteomic analyses were adopted to identify the SnRK1-regulated phosphosites involved in NSC transport. The identified phosphosites were validated and quantified using samples from control (CK) and leaf cutting (LC) plants by parallel reaction monitoring (PRM).

Results

Sheath is an important source for grain filling

Dry matter, NSC content, and starch content of -1/-2/-3/-4 sheaths and -1/-2/-3/-4 internodes were measured during the grain filling stage to clarify the importance of sheaths in grain filling (Figure 1, A–F). In general, the values and ranges of dry matter, NSC, and starch content of sheaths were larger than those in internodes. The dry matter, NSC content, and starch content of -1/-2/-3/-4 sheaths first increased and then decreased, exhibiting an inverted “V” shape after anthesis (the apex at 10 d after anthesis [DAA]). The results suggest that all sheaths are temporary storage organs for gain filling, and experience the sink-to-source transition. From 10 to 30 DAA, the dry matter of -1/-2/-3/-4 sheaths decreased by 0.13, 0.29, 0.28, and 0.22 g, respectively. Among the four sheathes, the -2 sheath had the most dramatic changes (Figure 1, A–C). Similar to the sheaths, the -3 and -4 internodes experienced sink-to-source transition, with an inverted “V” shape. NSC content and starch of -3 and -4 internodes increased at 0–10 DAA and decreased at 10–20 DAA. There was an increase in NSC at 20–30 DAA. The decrease in dry matter of -3 and -4 internodes from 10 to 30 DAA was 0.21 g and 0.17 g, respectively, which were lower than those in -2/-3/-4 sheaths. NSC and starch content of -1 and -2 internodes did not exhibit the inverted “V” trend. The dry matter of -1 and -2 internodes increased at 5 DAA and remained stable at 5–30 DAA (Figure 1, D–F). The results suggest that the sheath is a more important source organ than the internode during grain filling in Nipponbare. Among the sheaths, the -2 sheath is the most representative sheath.

Figure 1.

The physiological function of rice sheath during grain filling. A–C, the dry weight (A), NSC content (B) and starch content (C) of -1/-2/-3/-4 sheaths during grain filling in Nipponbare. D–F, the dry weight (D), NSC content (E), and starch content (F) of -1/-2/-3/-4 internodes during grain filling in Nipponbare. Each value in (A and D) represents the mean ± sd of four replicates. Each value in (B, C, E, and F) represents the mean ± sd of three replicates. -1/-2/-3/-4 sheaths represent the first, second, third and fourth sheaths from the top, respectively. -1/-2/-3/-4 internodes represent the first, second, third and fourth internodes from the top, respectively. G–J, a mesophyll cell of the top of -2 sheath at 0 DAA (G), 5 DAA (H), 10 DAA (I), and 15 DAA (J). K, a chloroplast of the top part at 10 DAA. L–O, a mesophyll cell of the bottom of -2 sheath at 0 DAA (L), 5 DAA (M), 10 DAA (N), and 15 DAA (O). P, a chloroplast of the bottom part at 5 DAA. Abbreviation: Cl, chloroplast; CW, cell wall; SG, starch granules; Gr, grana.

We also observed the starch granules of -2 sheath at 0, 5, 10, and 15 DAA using a transmission electron microscope (TEM; Figure 1, G–P). Starch content at the bottom of sheaths has been reported to be higher than at the top (Ishimaru et al., 2004). The top of -2 sheath is green, while the bottom is light green (almost white, Figure 1). At the top part, there were plenty of chloroplasts with obvious thylakoid grana and stroma lamellae. The starch granules were located in the chloroplasts, and were small and flattened (Figure 1, G–K). There were one to two starch granules in each chloroplast at 0 and 5 DAA, and the starch granules were small. At 10 DAA, there were two to three starch granules in each chloroplast. The starch granules at 10 DAA were larger than those at 0 DAA and 5 DAA. At 15 DAA, almost no starch granules were observed in the chloroplasts. Besides, the chloroplast gradually became smaller (Figure 1, G–J). At the bottom part, the starch granules were larger and rounder than those at the top part (Figure 1, L–P). The plastids where the starch granules were located also had thylakoid grana and stroma lamellae (Figure 1P), suggesting that the starch granules are also stored in chloroplasts. The starch granules were so large that the grana and stoma lamellae of the bottom part were squeezed to the edge of the chloroplasts, or surrounded starch granules. The grana and stoma lamellae of the bottom part were thinner than those of the top part. Compared with 0 DAA, the size of starch granules at 5 DAA was larger. The size of starch granules at 15 DAA was smaller than that at 10 DAA and was similar to 0 DAA (Figure 1, L–O). The change in starch granules at the top and the bottom is consistent with the change in starch content (Figure 1, C and G–P). The results suggest that -2 sheath is an important and representative source organ for grain filling. Therefore, the -2 sheath was selected for our subsequent studies.

Effects of LC treatment on sheath NSC transport and grain filling

The effects of LC treatment on sheath NSC transport were examined in LC and CK plants (Figure 2A). Results showed that the grain weights and grain filling rates between LC and CK plants were similar (Figure 2, B and C). The grain filling rates in both groups peaked at approximately 10 DAA (Figure 2C). Furthermore, the 1,000-grain weight and seed setting rate decreased by 0.9% and 0.3% after LC treatment, respectively, but the differences were not significant (Table 1). In CK plants, the starch content in sheaths increased after flowering, and began to decrease after 10 DAA. In contrast, the starch content sharply decreased after LC treatment and was significantly lower in LC than in CK plants during grain filling. These corroborate the results of starch staining, in which smaller areas were stained in the sheaths of LC plants (Figure 2, D and E). The sucrose contents in the sheaths of both CK and LC plants decreased after anthesis. However, the abrupt decrease after LC treatment resulted in lower sucrose content in LC plants at the early filling stage (Figure 2F). Additionally, the relative expression of OsSUT1 increased after LC; this high expression was maintained during the grain filling stage, which explains the decreased sucrose content observed in LC plants (Figure 2, F and G). The changes in OsSUT4 expression were significantly different from those of OsSUT1 expression after LC treatment. Specifically, OsSUT1 exhibited lower expression than OsSUT4 at the early filling stage, but had higher expression at the late filling stage (Figure 2H).

Figure 2.

Effects of LC treatment on grain filling and NSC remobilization in the sheaths. A, Schematic diagram of the LC treatment (three upper leaves were cut in half) at anthesis. B and C, The grain weight (B) and grain filling rate (C) of control (CK) and LC plants during grain filling. Each value in (B) represents the mean ± sd of six replicates. The grain filling rate was calculated using Richards’ equation (Richards, 1959). D, Starch staining of the sheaths from CK and LC plants during grain filling. E–H, The starch content (E), sucrose content (F), and relative expression levels of OsSUT1 (G) and OsSUT4 (H) in the sheaths of CK and LC plants during grain filling. Each value in (E–H) represents the mean ± sd of three replicates. Asterisks indicate significant differences between CK and LC plants at each time point by Student’s t test (B, E, F, G, and H): *P < 0.05, **P < 0.01.

Table 1.

Grain weight and seed setting rate of control (CK) and LC treatment plants

Treatment	1,000-Grain Weight (g)	Seed Setting Rate (%)
CK	22.85	88.22
LC	22.60	87.95

CK, no treatment; LC, three upper leaves were cut in half. Each value represents the mean ± sd of six replicates. The significance analysis of 1,000-grain weight and seed setting rate between CK and LC plants was performed by Student’s t test. The P-value of 1,000-grain weight and seed setting rate were 0.073 and 0.357, respectively.

In CK plants, the activity of α-amylase was maintained at ∼0.3 mg min⁻¹ g⁻¹ FW at 0–6 DAA and stably increased at 8 DAA (∼0.35 mg min⁻¹ g⁻¹ FW). In LC plants, α-amylase activity rapidly increased from 0.39 to 0.45 mg min⁻¹ g⁻¹ FW after LC treatment and remained at this high level (Supplemental Figure S1A). In both groups, β-amylase showed the same dynamic pattern as α-amylase during grain filling (Supplemental Figure S1B). In contrast, SPS activity first increased, then decreased, and reached its peak at 15 and 20 DAA in CK and LC plants, respectively. The increase in SPS activity was faster after LC treatment, resulting in higher SPS activity in LC plants at the early filling stage. However, there was no significant difference in the SPS activities between LC and CK plants at the late filling stage (Supplemental Figure S1C). Measurement of the photosynthetic parameters of the flag leaves at 6 DAA revealed that the net photosynthetic rate, stomatal conductance, and intercellular CO₂ concentration in LC plants significantly increased by 9.1%, 13.6%, and 6.2%, respectively. However, there was no significant difference in transpiration rate between LC and CK plants (Table 2).

Table 2.

Net photosynthetic rate (Pn), stomatal conductance (Gs), intercellular CO₂ concentration (Ci), and transpiration rate (Tr) in the flag leaves of control (CK) and LC treatment plants at 6 DAA

Treatment	Pn (umol·m−2s−1)	Gs (mmol·m−2s−1)	Ci (umol·mol−1)	Tr (mmol·m−2s−1)
CK	16.81	0.22	218.75	7.29
LC	18.34**	0.25*	232.25**	7.55

CK, no treatment; LC, three upper leaves were cut in half. Each value represents the mean ±sd of four replicates. Asterisks indicate significant differences between CK and LC plants by Student’s t test: *P < 0.05, **P < 0.01.

Effects of LC treatment on SnRK1 activity and T6P content

In CK plants, the SnRK1 activity in the sheaths was maintained at ∼0.2 nmol min⁻¹ mg⁻¹ protein at 0–10 DAA, increased at 10 DAA, peaked at 20 DAA, and decreased thereafter. In LC plants, the SnRK1 activity increased after LC treatment and was maintained at ∼0.4 nmol min⁻¹ mg⁻¹ protein at 2–25 DAA, then increased again at 30 DAA (Figure 3A). The T6P content was similar to the sucrose content during grain filling (Figures 2, F and 3, B). In CK plants, the T6P content gradually decreased at the early filling stage and slightly changed at 10–30 DAA. In LC plants, the T6P content rapidly dropped to 0.2 nmol g⁻¹ FW and then remained stable (Figure 3B). Correlation analysis revealed a significant correlation between the T6P and sucrose contents in CK and LC plants, indicating that T6P content is directly affected by the changes in sucrose availability in the rice sheaths. The Pearson correlation coefficients for CK and LC plants were 0.8570 and 0.9478, respectively (Figure 3C). T6P content and SnRK1 activity exhibited opposite trends. However, T6P content was not negatively significantly correlated with SnRK1 activity. The correlation coefficients for CK and LC plants were −0.5685 and −0.5307, respectively (Figure 3, A and B; Supplemental Figure S2). To examine whether T6P inhibits SnRK1 in rice sheaths, the SnRK1 activity was assayed using desalted extracts of different tissues at 10 DAA treated with or without 1-mM T6P (Figure 3D). Results showed that the SnRK1 activity in developing grains was inhibited by 64% after T6P treatment, while no significant change was observed in the flag leaves. Notably, the SnRK1 activity in the sheaths was inhibited by 18.5% and 17.0% in CK and LC plants, respectively.

Figure 3.

Effects of LC treatment on SnRK1 activity in the sheaths during grain filling. A, SnRK1 activity in control (CK) and LC plants during grain filling. B, T6P contents in the sheaths of CK and LC plants during grain filling. C, The T6P content at each time point was plotted against the sucrose content. The Pearson correlation coefficients (r) for CK and LC plants are shown. Asterisks indicate significant correlations between T6P content and sucrose content using the data in Figures 2, F and 3, B: **P < 0.01, ***P < 0.001. D, SnRK1 activity in different tissues with 1 mM T6P at 10 DAA. E–G, Relative expression levels of OsSnRK1a (E), OSK24 (F), and OSK35 (G) in CK and LC plants during grain filling. H, Western blot analysis of SnRK1a protein abundance and phosphorylation level of the catalytic subunit in CK and LC plants at 0, 4, 10, and 15 DAA, with tubulin as internal control. Each value in (A) and (D) represents the mean ± sd of four replicates. Each value in (B) and (E–G) represents the mean ± sd of three replicates. Asterisks indicate significant differences between CK and LC plants at each time point by Student’s t test (A, B, E, F, and G): *P < 0.05, **P < 0.01. Asterisks indicate significant differences between −T6P and +T6P by Student’s t test (D): *P < 0.05, **P < 0.01.

Effects of LC treatment on gene expression, protein abundance, and phosphorylation level of SnRK1

The observed inhibition (18%) of SnRK1 by 1-mM T6P in vitro was not sufficient to explain the changes in SnRK1 activity after LC treatment (2.08-fold increase; Figure 3, A and D). Therefore, the expression of the gene encoding SnRK1 was determined to verify the transcriptional regulation of SnRK1 in rice sheaths during grain filling (Figure 3, E–G). Results showed that the relative expression of OsSnRK1a in the sheaths of CK plants was low at the early filling stage and elevated at 10 DAA. In contrast, the relative OsSnRK1a expression in the sheaths of LC plants sharply increased after LC treatment and remained high during grain filling. No significant difference was observed in the relative expression of OSK24 and OSK35, except for that of OSK24 at 15 DAA, between CK and LC plants. These results suggest that the changes in SnRK1 activity were closely related to OsSnRK1a expression (Figure 3, A, E, F, and G). Hence, we hypothesize that OsSnRK1a is primarily responsible for modulating the SnRK1 activity in rice sheaths during grain filling. To verify this, specific SnRK1a and p-AMPKα (T172) antibodies were used to detect the SnRK1a protein and phosphorylation levels in CK and LC plants at 0, 4, 10, and 15 DAA (Figure 3H). The results revealed that the SnRK1a protein abundance and phosphorylation level of the SnRK1 catalytic subunit were elevated after LC treatment. These changes were consistent with the changes in the OsSnRK1a expression in LC plants, which verifies the function of OsSnRK1a in regulating SnRK1 activity.

NSC transport in snrk1a mutant plants

Since OsSnRK1a was found to be responsible for the changes in SnRK1 activity (Figure 3, E–H), we then constructed a snrk1a mutant using the clustered regularly interspaced short palindromic repeats/CRISPR-associated 9 (CRISPR/Cas9) system to verify the function of SnRK1 in NSC transport. Similar to SNF1 and AMPK, SnRK1 is a heterotrimeric holoenzyme comprising a catalytic subunit α and two regulatory subunits (β and γ). The α-subunit of SnRK1 comprises an N-terminal kinase catalytic domain, a ubiquitin-associated domain, and a C-terminal regulatory domain (RD; Polge and Thomas, 2007). The RD is largely involved in binding β- and γ-subunits to facilitate heterotrimer formation, which is essential for membrane location and catalytic regulation (Sanz 2008; Emanuelle et al., 2016). The 9–11th exon encodes the RD of SnRK1a in rice. Therefore, the sgRNA targeted the region of the ninth exon of OsSnRK1a. The snrk1a mutant had 16-bp deletion and 6-bp insertion sequences in the coding sequence of OsSnRK1a, which caused translation into a truncated protein with 373 amino acid (aa) (Figure 4A). Correspondingly, it lost the conserved α-subunit RD, which has an interval with 385–502 aa (Supplemental Figure S3). This mutation reduced the SnRK1 activity of the sheaths at the grain filling stage significantly (Figure 4B). In addition, the phenotype of the mutant was mild during grain filling, and only poor grain filling appeared (Figure 4C), which facilitates the exploration of the role of SnRK1 in sheath NSC transport.

Figure 4.

Analysis of NSC remobilization in WT and snrk1a mutant plants during grain filling. A, The mutation site in the OsSnRK1a coding region. The triangle indicates the small guide RNA target. B, SnRK1 activity in the sheaths of snrk1a mutant plants at 0 and 4 DAA. Each value represents the mean ± sd of four replicates. C, Photo of the snrk1a mutant at 15 DAA. D, Starch staining of the sheaths at 0, 4, 10, 15, 20, 25, and 30 DAA. E and F, The starch (E) and sucrose (F) contents in the sheaths during grain filling. Each value represents the mean ± sd of three replicates. Asterisks indicate significant differences between WT and snrk1a plants at each time point by Student’s t test (B, E, and F): *P < 0.05, **P < 0.01.

Compared to the wild-type (WT), the snrk1a mutant exhibited a lower seed setting rate and 1,000-grain weight (Table 3). However, there was no difference in the panicle number and number of grains per panicle between the WT and snrk1a mutant plants (Table 3). The starch staining also showed that the sheaths of snrk1a mutant plants contained more starch than WT plants during grain filling (Figure 4D). The starch content in the WT first increased and then decreased at 10 DAA, while that of the snrk1a mutant remained stable and ranged from 190.2 FW to 200.3 mg g⁻¹ FW (Figure 4E). The WT and snrk1a mutant plants showed similar changes in sucrose contents during monitoring. However, the sucrose content was higher in the snrk1a mutant than that in the WT at 2–10 DAA but was lower in the snrk1a mutant than in the WT at 20 and 30 DAA (Figure 4F).

Table 3.

Panicle length, number of grains per panicle, grain weight, and seed setting rate of WT and snrk1a mutant plants

Materials	Panicle Length (cm)	Grains per Panicle	1,000-Grain Weight (g)	Seed Setting Rate（%）
WT	16.22	93.92	22.85	85.72
snrk1a	15.78	96.42	21.50**	15.43**

Each value represents the mean ± sd of six replicates. Asterisks indicate significant differences between WT and snrk1a by Student’s t test: **P < 0.01.

Phosphoproteomic and proteomic analyses of WT and snrk1a mutant plants

Quantitative phosphoproteomic and proteomic analyses were performed to explore the mechanisms involved in the SnRK1-mediated regulation of NSC transport in the rice sheaths (Figure 5A). The sheaths from WT and snrk1a mutant plants at 0, 4, 10, and 15 DAA were selected as samples for proteomics and phosphoproteomics. The four turning points included the initial, increased, peak, and decreased starch contents in the sheaths of WT plants after anthesis (Figure 4D). The samples were lysed and digested with trypsin. One part of the samples was used for proteomics, and the remaining part was used for phosphoproteomics after affinity enrichment. After normalization (based on protein quantification) to remove the influence of protein expression on the phosphorylation signal, the data were used for further bioinformatics analysis. A total of 6,519 proteins and 6,592 phosphosites were identified (Supplemental Tables S1 and S2). For subsequent analysis, “C” and “S” were used to represent the WT and snrk1a mutant plants, respectively. Principal component analysis of all proteins and phosphosites verified the reproducibility of the results (Figure 5B).

Figure 5.

Proteomic and phosphoproteomic profiling of the sheaths from WT and snrk1a mutant plants during grain filling. A, Experimental workflow of the analyses performed. B, Principal component analysis of all identified proteins and phosphosites per analysis. C, Number of DEPs and DESs in the sheaths at 0, 4, 10, and 15 DAA. The red and blue bars represent the upregulated and downregulated DEPs/DESs, respectively. D, Venn diagram of the DESs among the four comparisons. E, Cluster analysis of the 599 DESs identified. The color scale represents the relative phosphorylation level, from low (green) to high (red). F, The phosphorylation motifs identified using Motif-X algorithm within the putative SnRK1 targets and/or SnRK1-dependent phosphosites. G, Predicted kinase activity of MAPKs, CPKs, and TOR. The color scale represents the kinase activity score, from low (blue) to high (red).

The proteins or phosphosites with ratios >1.5 or <1/1.5 and P-values <0.05 were considered as differentially expressed between snrk1a mutant and WT plants. The summarized information of the differentially expressed proteins (DEPs) and phosphosites (DESs) at 0, 4, 10, and 15 DAA are shown in Supplemental Tables S3 and S4, and Figure 4C. In the snrk1a mutant, there were 156, 120, 154, and 120 upregulated and 148, 126, 187, and 227 downregulated proteins at 0, 4, 10, and 15 DAA, respectively. Furthermore, there were 56, 42, 81, and 64 upregulated and 165, 155, 145, and 160 downregulated phosphosites at 0, 4, 10, and 15 DAA, respectively (Figure 5C). The number of downregulated phosphosites was significantly higher than that of upregulated phosphosites at all four time points, indicating that as a kinase, SnRK1 plays a key role in the phosphorylation of proteins in the rice sheaths. In addition, a total of 599 nonredundant phosphosites were identified among the four comparisons (Figure 5D;Supplemental Table S4) and discovered to be regulated by an SnRK1-dependent pathway in the sheaths during grain filling. Cluster analysis of these phosphosites revealed that the expression patterns were clearly different between snrk1a mutant and WT plants (Figure 5E).

Six significantly enriched phosphorylation motifs, namely DxxxxxSP, SP, RxxS, SDxE, PxTP, and TP, were extracted from the 599 SnRK1-dependent phosphosites using the Motif-X algorithm (Figure 5F;Supplemental Table S5). Recent studies reported that the SP and RxxS motif are two classic phosphorylation targets of SnRK1 (Cho et al., 2016; Nukarinen et al., 2016). RxxS is also a motif commonly targeted by mitogen-activated protein kinase kinases (Wijk et al., 2014) that appears in target of rapamycin (TOR)-dependent phosphosites (Van Leene et al., 2019). Furthermore, the SDxE motif is targeted by calcium-dependent protein kinases (CPKs; Zhang et al., 2014). In addition to p-serine motifs, the two p-threonine motifs, PxTP and TP, were found to be targets of mitogen-activated protein kinases (MAPKs; Rayapuram et al., 2018). Prediction results from the Group-based Prediction System (GPS) version 5.0 software and Gene Set Enrichment Analysis (GSEA) showed that the kinase activities of MAPKs, CPKs, and TOR were different between WT and snrk1a mutant plants during grain filling (Figure 5G), while no significant differences were observed in the activities of CPK3, CPK5, CPK13, CPK18, CPK19, CPK23, and CPK25 between WT and snrk1a mutant plants. However, the snrk1a mutant plants showed lower activities at 4 DAA and higher activities at 15 DAA for CPK4, CPK6, CPK7, CPK9, CPK10, CPK12, CPK15, CPK16, CPK17, CPK20, CPK21, CPK22, CPK24, CPK26, CPK28, and CPK29, while TOR activity was only detected at 0 DAA. In contrast, the MPK3, MPK4, MPK5, and MPK6 activities were higher in WT than in snrk1a mutant plants during grain filling.

Identification of phosphosites involved in NSC transport

A three-step analytical method integrating co-expression clustering and functional enrichment analyses was developed to identify the SnRK1-dependent phosphosites involved in NSC transport (Figure 6A). First, we performed weighted gene correlation network analysis (WGCNA) for the 599 SnRK1-dependent phosphosites (Langfelder and Horvath, 2008), which grouped 582 phosphosites into seven co-expression modules (phosphoproteomics modules [PPMs]), while 17 phosphosites were not clustered into any module (Supplemental Figure S4; Supplemental Table S6). Second, we analyzed the relationships between the modules and sheath starch content to screen those with high correlation values (|r| > 0.5, P < 0.05; Figure 6B). The results showed that PPM1 with 77 phosphosites was highly correlated with starch content (r = 0.73, P = 1 × 10⁻⁵), while PPM3 with 125 phosphosites (r = −0.63, P = 0.001) and PPM4 with 60 phosphosites were negatively associated with starch content (r = −0.51, P = 0.01). The remaining modules were not associated with starch content. Third, we performed functional enrichment analysis of PPM1, PPM3, and PPM4 to identify the phosphosites involved in NSC transport (Figure 6C;Supplemental Table S7). For PPM1, two enriched pathways were discovered to be involved in NSC transport, specifically “linear malto-oligosaccharide phosphorylase activity” and “active transmembrane transporter activity.” PPM3 was enriched for “carbohydrate catabolic process” and “starch and sucrose metabolism,” while PPM4 was enriched for “carbohydrate catabolic process,” “carbon metabolism,” “water transport,” and “glycolysis.” A total of 30 SnRK1-dependent phosphosites were found to be associated with these pathways, suggesting their involvement in the remobilization of NSCs in the rice sheaths (Supplemental Table S8).

Figure 6.

Phosphoproteomic analysis of SnRK1-dependent phosphosites involved in the sheath transport of NSCs. A, Overview of the analysis performed. B, Module–trait (starch content) relationships and their corresponding P-values. The eight-colored part (left) shows the seven PPMs and one unclustered module. The color scale (right) represents the module–trait (starch content) correlations, from low (green) to high (red). C, Functional enrichment analysis of the PPMs using the GO and KEGG databases.

Validation of SnRK1-dependent phosphosites

PRM, the latest application of targeted mass spectrometry, has been widely used to quantify the abundance and/or posttranslational modification level of target proteins because of its specificity and sensitivity (Bourmaud et al., 2016). To verify the regulatory function of SnRK1, the phosphorylation levels of phosphosites identified in the sheaths of CK and LC plants at 6 DAA were quantified by PRM. Results showed that the starch content in the sheaths increased in CK plants and decreased in LC plants (Figure 2E). Additionally, the SnRK1 activity was 2.08-fold higher in LC than in CK plants (Figure 3A). Of the 30 phosphosites, 20 were successfully quantified (Figure 7; Supplemental Table S9), while the remaining 10 were not, which may be due to the different samples or detection methods. Among the 20 SnRK1-dependent phosphosites, 18 had higher phosphorylation levels in LC than in CK plants. In contrast, two phosphosites, namely S142-SPS4 and S292-TMT2, exhibited lower phosphorylation levels in LC than in CK plants.

Figure 7.

PRM assay of 20 phosphosites involved in the sheath transport of NSCs at 6 DAA. PRM assay results showing the associated pathways of the 20 phosphosites identified in the sheaths of control (CK) and LC treatment plants at 6 DAA. Each value represents the mean of three replicates. For more information, please refer to Supplemental Table S9.

Discussion

Sheaths and culms (internodes) are both carbon sources for grain filling in rice (Perez et al., 1971). Xu et al. (2021) observed more obvious dry matter dynamics in sheaths in WYJ3, WYJ7, and NJ8. Pan et al. (2016) observed that the sheaths transport more NSC to the panicle than the internodes in LYP9 and YD6. Such inconsistency could be due to variation across cultivars. We observed that the sheath, as a source organ, is more important than the internode during grain filling in Nipponbare (Figure 1, A–F). Although -1/-2/-3/-4 sheaths and -3/-4 internodes all function as sources for grain filling, the -2/-3/-4 sheaths have higher dry matter than -3/-4 internodes at 10–30 DAA. In addition, the dry matter of -1 and -2 internodes was stable after a dramatic increase. Xu et al. (2021) found that the three upper internodes are sink organs during the early filling stage in WYJ3, WYJ7, and NJ8. The accumulated assimilate is used for elongation. In wheat (Triticum aestivum), the -1 internode still elongates after heading and the assimilate is mainly used for elongation and structural compound synthesis (Scofield et al., 2009). The elongation of -1/-2 internodes and grain filling occurred simultaneously in Nipponbare. The reserved NSC of -1/-2 internodes was used for their elongation and development, not for grain filling.

Sheaths and leaves are both green organs. The accumulation and degradation of starch vary between the sheaths and leaves. Under TEM observation, the top part of -2 sheath had complete chloroplasts with plenty of grana and stroma lamellas, similar to the leaves. The starch granules were located in chloroplasts (Figure 1, G–K); however, there were only few grana and stroma lamellas in the plastids that stored starch granules at the bottom of -2 sheath. The grana and stroma lamellas were squeezed to the edges or among the starch granules by big starch granules (Figure 1, L–P). These phenomena also exist in the chloroplasts of K-deficient maize (Zea mays) and cotton (Gossypium hirsutum) leaves, in which the grana and stroma were squeezed to the edge by large starch granules (Hall et al., 1972; Zhao et al., 2001). Therefore, we suggest that the starch granules are also reserved in the chloroplasts at both the top and bottom of sheaths. Our observation is consistent with the results of Guo et al. (2011). They found that the starch granules are located at chloroplasts in the middle part of the -1 sheath. Sato (1955) found that the starch granules of sheaths were in leucoplasts, using an optical microscope, and the accumulation and degradation of starch were related to the number of leucoplasts. A probable explanation for the inconsistency between our results and the results of Sato (1955) is due to the different methods of observation. The squeezed grana and stroma lamellas might be too small to be observed using an optical microscope. Starch in chloroplasts of source leaves fluctuates with the day/night cycle, a process regulated by T6P (Martins et al., 2013; Peixoto et al., 2021). Starch and sucrose in rice sheaths remained stable over 24 h (Supplemental Figure S5). Leaves are autotrophic organs where sucrose and starch accumulated through photosynthesis are exported at night to sustain growth and development. Unlike leaves, sheaths are temporary storage organs during grain filling. The accumulation and output of sucrose and starch in sheaths may be mainly regulated by the source–sink relationship, which results in no drastic changes in sucrose and starch levels in a short period (24 h). Besides, the starch pools accumulating in leaf sheaths are huge as compared to starch pools established by day/night cycles in leaves. Therefore, some subtle changes in sucrose and starch in the day/night cycle may be masked in sheaths. T6P in sheaths was also stable in 24 h, while T6P exhibited a diurnal cycle in Arabidopsis leaves (Supplemental Figure S5; Martins et al., 2013). Differential dynamics in T6P of sheaths and leaves may result from the different fluctuations in sucrose levels in one day/night cycle. Considering the starch content of the filling stage, we believe that the accumulation and degradation of starch in sheaths is a long-period process, not a daily change. Overall, the sheath is an important source for grain filling. Determination of the signals that mediate the sheath-to-panicle transport of NSCs is essential for improving NSC transport.

SnRK1 is involved in sheath NSC transport

In plants, SnRK1 lies at the hub of metabolic homeostasis and is essential for normal plant development and stress response (Emanuelle et al., 2016). SnRK1 can regulate nutrient distribution between tissues and organs for forming new organs and improving stress tolerance (Schwachtje et al., 2006; Lawlor and Paul, 2014; Oszvald et al., 2018). Most published data on plant SnRK1 activity are from studies on seeds and seedlings (Martinez-Barajas et al., 2011; Nunes et al., 2013; Coello and Martínez-Barajas, 2014a). There have been a few detailed and comprehensive analyses of SnRK1 in temporary storage organs, such as sheaths. The NSCs reserved in rice sheaths are remobilized to supply grain filling after anthesis (Okamura et al., 2018). Given the importance of NSC transport in rice grain formation and the function of SnRK1 in nutrient allocation, elucidating the role of SnRK1 in NSC transport is a key step for promoting NSC remobilization, thereby improving rice yield. In this study, the SnRK1 activity and NSC transport in CK and LC plants were evaluated to determine the function of SnRK1 in NSC transport. Our results showed that starch degradation occurred around the peak of grain filling (10 DAA) and was accompanied by an increase in SnRK1 activity in CK plants (Figures 2, B–F and 3, A). In LC plants, NSC transport was advanced to anthesis and simultaneously activated SnRK1 activity, suggesting that the accelerated NSC remobilization and higher photosynthetic rate compensated for the loss of yield due to less leaf area after LC treatment (Figures 2, A–F and 3, A; Tables 1 and 2). These data imply that SnRK1 plays a critical role in the transition of NSC accumulation to remobilization in the rice sheaths in response to insufficient leaf photosynthesis.

To verify the function of SnRK1 in NSC remobilization, we compared the starch and sucrose contents between WT and snrk1a mutant plants during grain filling (Figure 4, D–F). We discovered that SnRK1 activity in the sheaths was significantly lower in the snrk1a mutant than in the WT (Figure 4B). Notably, the starch stored in the sheaths was not remobilized to supply grain development in snrk1a mutant plants (Figure 4, D and E). Sucrose transport was also slightly inhibited in the snrk1a mutant (Figure 4F). Furthermore, insufficient NSC transport partly resulted in low seed setting rate and 1,000-grain weight (Table 3), confirming that SnRK1 activity is essential for starch degradation and NSC transport in the sheaths. In rice, SnRK1 acts upstream of MYBS1 and amylase and mobilizes the starch stored in the endosperm for seed germination and seedling growth. A previous study revealed that seed germination and seedling growth were retarded in the snrk1a mutant and SnRK1A RNA interference transgenic line (Lu et al., 2007). The snf1a snf1b double knockout mutant of Physcomitrium patens failed to mobilize the reserved starch under normal day–night cycles and could only survive under continuous light (Thelander et al., 2004). In Arabidopsis, the double kin10 kin11 mutant exhibited impaired starch remobilization and seedling growth (Baena-González et al., 2007). In addition, the SnRK1 complex was found to regulate the transport of carbon resources to the roots after herbivore attack (Schwachtje et al., 2006). Overall, these findings support our hypothesis that SnRK1 can reallocate the assimilate supply between organs in response to stress or nutrient starvation.

SnRK1 activity is regulated by the transcriptional level and T6P content in response to sucrose availability

T6P reportedly acts as a specific signal for sucrose availability (Yadav et al., 2014). Previous studies revealed that SnRK1 activity is inhibited by T6P in vivo and in vitro to regulate the metabolic network (Zhang et al., 2009; Nunes et al., 2013). T6P inhibits the catalytic activity of SnRK1 from developing tissues such as developing wheat grains, such inhibition was not reproduced for SnRK1 from mature leaves (Zhang et al., 2009; Martinez-Barajas et al., 2011). The T6P-mediated inhibition of SnRK1 was found to be dependent on an unknown protein factor, which was counteracted after purification by antibodies (Zhang et al., 2009). This inhibitory effect has only been observed in rice seedlings, developing seeds, and mature leaves (Zhang et al., 2009; Martinez-Barajas et al., 2011; Nunes et al., 2013), but not in the leaf sheath, a transient storage organ. In this study, the T6P and sucrose contents were similar in CK and LC plants (Figures 2F and 3B). There was also a high correlation between the T6P and sucrose contents in CK and LC plants (Figure 3C). These results demonstrate that T6P serves as a signal of sucrose availability in the rice sheath. T6P content and SnRK1 activity exhibited opposite trends in the sheaths of CK and LC plants during grain filling (Figure 3, A and B; Supplemental Figure S2). The in vitro assay showed that the SnRK1 activity was strongly inhibited by T6P in developing rice grains at 10 DAA, but not in the leaves (Figure 3D), which is consistent with the results of previous studies (Zhang et al., 2009; Martinez-Barajas et al., 2011). The SnRK1 activity in the sheaths at 10 DAA was also slightly inhibited (18%) by 1-mM T6P in vitro (Figure 3D), suggesting that the unknown protein factor had low expression in the rice sheaths during grain filling. A recent study reported that T6P inhibits SnRK1 partly through GEMINIVIRUS REP-INTERACTING KINASES GRIK1 and GRIK2. T6P directly binds to SnRK1 and diminishes the interaction between GRIKs and SnRK1 and thereby SnRK1 phosphorylation and activity. The inhibition of T6P on SnRK1 activity was reduced in extracts of the grik1 or grik2 mutant (Zhai et al., 2018). The difference in the inhibition of T6P among sheaths, grains, and leaves may be attributed to the variation in the abundance of GRIK homologs in different tissues. Taken together, these results suggest that T6P inhibits SnRK1 depending on sucrose availability in the rice sheaths during grain filling.

However, the level of T6P-inhibited SnRK1 activity was too low to explain the elevated SnRK1 activity after LC treatment. Hence, the levels of SnRK1 gene expression, SnRK1 protein, and phosphorylation of the catalytic subunit were examined. The expression of OsSnRK1a, instead of OSK24 and OSK35, rapidly increased after LC treatment, resulting in the increased SnRK1a abundance and phosphorylation of the catalytic subunit. In CK plants, these increased levels were only observed at 15 DAA (Figure 3, E–H). The phosphorylation of the catalytic subunit is essential for the catalytic activity of SnRK1 (Sugden et al., 1999a). Thus, we speculated that the activation of SnRK1 after LC treatment was caused by the increased OsSnRK1a expression in the rice sheaths. Lu et al. (2007) discovered that the OsSnRK1a expression was stable in rice embryos during sugar starvation, while the SnRK1a protein expression and SnRK1 activity increased, indicating that these two are regulated at the posttranscriptional level. These results suggest that the mechanism regulating the SnKR1 abundance and activity in rice may be organ-specific. Glucose-6-phosphate and G1P has also been reported to inhibit SnRK1 activity. G1P inhibits SnRK1 in combination with T6P synergistically (Nunes et al., 2013). ADPG and UDPG are also inhibitors of SnRK1, which reduced the activity of SnRK1 from beans (Phaseolus vulgaris) cotyledon and embryo extracts (Coello and Martínez-Barajas, 2014b). These metabolites may also be involved in regulating SnRK1 activity in rice sheaths.

Sucrose starvation and increased sucrose demand were discovered as the signals that transcriptionally activate SnRK1. The changes in sucrose content were found to be directly proportional to the changes in WPK4 expression, an SnRK gene family member, in wheat seedlings (Ikeda et al., 1999), while sucrose depletion stimulated the VfSnRK1 promoter activity in Arabidopsis protoplasts (Radchuk et al., 2009). In this study, the sucrose content was maintained at ∼20 mg g⁻¹ FW after LC treatment and was accompanied by an increase in OsSnRK1a expression. In CK plants, the sucrose content continued to decrease after anthesis, which dropped to <20 mg g⁻¹ FW until 10 DAA, but was accompanied by increased OsSnRK1a expression (Figures 2F and 3E). Based on these results, we hypothesize that the level of OsSnRK1a expression increased in response to the low sucrose signal (sucrose content <20 mg g⁻¹ FW) and induced SnRK1 activation.

SnRK1 regulates the phosphorylation of proteins involved in NSC transport

SnRK1 modulates the transcriptional network through the phosphorylation of various transcription factors in response to sugar starvation and energy stress (Baena-González et al., 2007; Mair et al., 2015). SnRK1 is also involved in the metabolic adaptation and response to biotic and abiotic stresses via direct phosphorylation of key metabolic enzymes, such as SPSs, nitrate reductase, fructose-2,6-bisphosphatase, and trehalose phosphate synthase (Sugden et al., 1999b; Harthill et al., 2006; Cho et al., 2016). In this study, quantitative phosphoproteomics was performed to investigate the function of SnRK1 in sheath NSC transport. We successfully screened 30 SnRK1-dependent phosphosites that are involved in NSC transport (Figure 6, A–C; Supplemental Table S8), of which 20 were validated and quantified in CK and LC plants (Figure 7; Supplemental Table S9).

Starch is broken down into maltose, glucose, or G1P in the chloroplast, which can be transported to the cytoplasm (Stitt and Zeeman, 2012). There are two pathways for starch degradation in plants—hydrolytic and phosphorolytic (Stitt and Heldt, 1981a). SnRK1 functions in the hydrolytic pathway via phosphorylation. SnRK1 phosphorylates the transcription factor MYBS1 to promote amylase transcription in response to sugar starvation (Lu et al., 2007). However, we did not detect MYBS1 using phosphoproteomics. On the other hand, the expression levels of OsRAM3D and OsBAM3, which encode α-amylase and β-amylase, respectively, were significantly higher in LC plants at 6 DAA (Supplemental Figure S6). The activities of α-amylase and β-amylase were also higher in LC than in CK plants (Supplemental Figure S1, A and B), implying that SnRK1 may regulate starch degradation via this pathway. In the phosphorolytic pathway, α-1,4 glucan phosphorylase (PHS) and plastidic phosphoglucomutase (PGM) were found to be involved in catalyzing reversible reactions (Stitt and Heldt, 1981b). In this study, LC treatment resulted in higher S411-PHS and S322-PGM phosphorylation levels (Figure 7), indicating that SnRK1 may also regulate the phosphorolytic pathway via phosphorylation, thereby promoting starch degradation in the rice sheaths.

The glucose and G1P produced by starch degradation are synthesized into sucrose in the cytoplasm (MacNeill et al., 2017). Previous studies have noted that SPS is a key enzyme for sucrose re-synthesis in rice sheaths during grain filling (Yang et al., 2001). In rice, SPSs are encoded by a small gene family composed of five genes (Castleden et al., 2004). SPS1 (OsSPS1), SPS4 (OsSPS8), and SPS5 (OsSPS11) were identified to be differentially phosphorylated in both LC and CK plants. The phosphorylation of S740-SPS1, S170-SPS4, S703-SPS4, S719-SPS4, and S716-SPS5 was induced, while S142-SPS4 phosphorylation was inhibited by LC treatment (Figure 7). SnRK1 plays a role in sucrose homeostasis and SPS activity regulation (Peixoto et al., 2021; Sugden et al., 1999b). Overexpressing the homologous or heterologous SnRK1 gene increases sucrose content in both sink and source organs in plants (McKibbin et al., 2006; Wang et al., 2012; Jiang et al., 2013; Peixoto et al., 2021). The SnRK1 loss-of-function mutants exhibited decreased sucrose contents of rosettes in Arabidopsis (Peixoto et al., 2021). SnRK1 inactivates SPS by phosphorylating S158 in vitro, a conserved phosphorylation site in SPS (Sugden et al., 1999b). SPS activity in leaves was decreased in transgenic tobacco (Nicotiana tabacum) plants overexpressing heterologous IbSnRK1 gene. However, SPS activity did not change in potato (Solanum tuberosum) tubers and tomato (Solanum lycopersicum) leaves of transgenic lines overexpressing SnRK1 gene (McKibbin et al., 2006; Wang et al., 2012). In our work, SnRK1 activity and the phosphorylation level of S170-SPS4 (homologous site to S158) increased in LC plants, and SPS activity also increased (Figures 3A and 7; Supplemental Figure S1C). The results are inconsistent with the inhibitory effect of SnRK1 on SPS activity in vitro. We speculate that, besides SnRK1 phosphorylating S150, there are other regulatory mechanisms for SPS activity, which results in higher SPS activity in LC plant. The other phosphorylation sites we identified such as S740-SPS1 and S703-SPS4 may also play important roles in SPS activity. Cytosolic PGM is essential for glucose-6-phosphate partitioning and sucrose synthesis in the leaves (Mühlbach and Schnarrenberger, 1978; Fettke et al., 2009). Here, the S124-PGM phosphorylation was upregulated in LC plants (Figure 7), suggesting that SnRK1 may phosphorylate cytosolic PGM to maintain sucrose synthesis. Cytosolic invertase (CINV) catalyzes the irreversible breakdown of sucrose in the cytoplasm (Vargas et al., 2008), and CINV activity is also regulated via phosphorylation. The S547 of AtCINV1 was found to be phosphorylated by CPK3 and CPK21, which in turn increased the 14-3-3 protein affinity and activity (Gao et al., 2014). The S18-CINV1 and S37-INV phosphorylation levels have not yet been reported (Figure 7). We hypothesize that SnRK1 potentially regulates the phosphorylation of these two sites to regulate sucrose breakdown.

The proton motive force is generated by H⁺-ATPase to load sucrose into the phloem against a concentration gradient (Lalonde et al., 2004). Similarly, the plasma membrane H⁺-ATPase is regulated via phosphorylation (Robertson et al., 2004), and regulatory sites are mainly clustered in the terminal tail with 100 aa residues (Haruta et al., 2015). Although several DESs at the C-terminal tail of plasma membrane H⁺-ATPase were identified, these sites did not cluster into the PPMs associated with starch content (Supplemental Table S7). Interestingly, S545-plasma membrane ATPase (H⁺-ATPase), a site not located at the C-terminal tail, had a correlation with starch content, and its phosphorylation level was significantly higher in LC than in CK plants (Figures 6, B and 7). The Münch pressure-flow hypothesis (Münch, 1930) states that the movement of the phloem sap is driven by mass flow and water transport. Aquaporins are believed to support sucrose transport by facilitating water movement into the sieve elements (Schulz, 2015). Hence, the phosphosites at the C-terminal tail are essential for the gatekeeping and subcellular localization of plasma membrane aquaporins (Maurel et al., 2015). In Arabidopsis, confocal microscopy revealed that S283-AtPIP2-1 phosphorylation is necessary for guiding the AtPIP2;1 to the plasma membrane (Prak et al., 2008). In the present study, the amino acid sequence alignment of AtPIP2-1, AtPIP2-6, OsPIP2-1, and OsPIP2-6 revealed a consensus phosphosite between S283-AtPIP2-1, S288-PIP2-1, and S278-PIP2-6 (Supplemental Figure S7). Furthermore, the phosphorylation of S288-PIP2-1 and S278-PIP2-6 was higher in LC than in CK plants at 6 DAA, indicating that SnRK1 may regulate the subcellular localization of aquaporins in the rice sheaths during grain filling.

The sucrose stored in the vacuole is first exported to the cytoplasm by OsSUT2, a tonoplast-localized H⁺-Suc symporter, and subsequently loaded into the phloem (Eom et al., 2011). The proton gradient across the vacuolar membrane was established by vacuolar inorganic diphosphatase or H⁺-pyrophosphatase (H⁺-PPase; Lin et al., 2012). The LC treatment resulted in the increased phosphorylation level of S71-H⁺-PPase and reduced sucrose level (Figures 7 and 2F). In addition, the phosphorylation of S292-TMT2 was found to be lower in LC than in CK plants. Cold stress can reportedly induce the phosphorylation of S385-TMT2 and S376-TMT2 and enhance glucose and fructose accumulation in the cytoplasm (Schulze et al., 2012). Therefore, SnRK1-mediated phosphorylation potentially promotes the exportation of vacuole-stored sucrose to the cytoplasm and inhibits monosaccharide import into the vacuoles of the rice sheaths during grain filling.

Through enzymatic reactions, glycolysis produces ATP that can be used for sucrose transport. Seven glycolytic enzymes are known to be regulated via serine/threonine phosphorylation (O’Leary and Plaxton, 2020). Among them, fructose-2,6-bisphosphatase, nicotinamide adenine dinucleotide (NAD)-dependent glyceraldehyde-3-P dehydrogenase, nonphosphorylating NADP-dependent glyceraldehyde-3-P dehydrogenase, and cytosolic pyruvate kinase are phosphorylated by SnRK1 (Tang et al., 2003; Plaxton and Podestá, 2006; Piattoni et al., 2011; Zhang and Liu, 2017). Here, we found that the phosphorylation levels of S81-phosphoglycerate mutase, S60-ATP-dependent 6-phosphofructokinase (ATP-PFK), S71-ATP-PFK, and S32-fructose-bisphosphate aldolase 3 were upregulated after LC treatment at 6 DAA (Figure 7). These results suggest that SnRK1 may activate glycolysis to supply the ATP required for sugar transport during grain filling.

Conclusion

In summary, our findings demonstrate that SnRK1 acts as a signaling hub for the sheath-to-panicle transport of NSCs in rice (Figure 8). In specific, when the assimilates supplied by the leaves are sufficient for grain development, starch accumulation occurs in the rice sheaths. However, in case of reduced photosynthetic leaf area or when the peak of grain filling is reached, the assimilate supply will not be sufficient for grain development. Hence, the sucrose stored in the sheaths is rapidly transported to the panicles, resulting in decreased sucrose content in the sheaths. The low sucrose content induces an increase in OsSnRK1a expression, which in turn promotes SnRK1 activity. Additionally, the T6P level decreases in response to the low sucrose content, consequently reducing the T6P-mediated inhibition of SnRK1. These processes result in elevated SnRK1 activity and subsequently promote the sheath-to-panicle transport of NSCs through the regulation of starch degradation, sugar transport across the tonoplast, sucrose metabolism, phloem transport, and glycolysis via phosphorylation. Therefore, our study confirms the function of high SnRK1 activity for the remobilization of NSCs in the rice sheaths. Future strategies involving chemical regulation and biotechnology may be utilized to further enhance the SnRK1 activity in the rice sheaths during grain filling, thereby improving rice yield.

Figure 8.

A model for the SnRK1-mediated regulation of NSC sheath-to-panicle remobilization in response to sucrose availability. NSC transport in sheaths is regulated by sucrose availability and SnRK1 activity to ensure sufficient assimilation for grain filling. If the source leaves are sufficient for grain development, starch accumulation occurs in the rice sheaths. In case of reduced photosynthetic leaf area or when the peak of grain filling is reached, the leaves will not be sufficient for grain development. Hence, the sucrose stored in the sheaths is transported to the panicles, resulting in a low sucrose level in the sheaths. The low sucrose content induces an increase in OsSnRK1a expression and reduces the T6P mediated inhibition of SnRK1. Elevated SnRK1 activity subsequently promotes the sheath to panicle transport of NSCs through the regulation of starch degradation, sugar transport across the tonoplast, sucrose metabolism, phloem transport, and glycolysis via phosphorylation. The orange arrows represent sucrose flow (left). The red and blue represent upregulation and downregulation, respectively (right).

Materials and methods

Plant materials and growth conditions

The WT and snrk1a mutant plants of japonica rice (Oryza sativa L.) cultivar Nipponbare were used in this study. The CRISPR–Cas9 (Mao et al., 2013) system was used to construct the OsSnRK1a gene editing vector. The sgRNA was inserted into PSH-U3 vector at the Kpn I site and driven by the OsU3 promoter. Afterward, the transformed vector was digested using Hind III restriction enzyme to obtain a combined “OsU3-sgRNA” fragment. Subsequently, fragment was inserted into pCAMBIA1300-Cas9 vector at the Hind III site for the CRISPR construct of OsSnRK1A. Genetic transformation was performed to introduce the plasmid into WT Nipponbare.

The experiment was conducted at the Danyang research base of Nanjing Agricultural University, Jiangsu Province, China (31°54′N, 119°28′E) during the rice-growing season. The plant materials were cultivated in plastic pots. Each plot is filled with 15 kg of soil taken from the field. Each pot contained four holes, with two seedlings per hole. The seedlings were transplanted at 15 d after sowing. Basal and panicle fertilizers were applied twice throughout the growing season, with a dosage of 1.0 g N, 0.5 g P₂O₅, and 0.8 g K₂O per pot. The temperature and light conditions after transplanting were shown in Supplemental Figure S8.

Treatment and sampling

For the LC treatment, three upper leaves of the WT plants were cut in half at anthesis. No treatment was used as a CK (Figure 1A). The CK and LC groups both consisted of 50 pots. Four hundred plants with a similar growth pattern and flowered on the same day were selected, tagged, and subjected to LC. Twenty tagged plants were sampled at 0, 2, 4, 6, 8, 10, 15, 20, 25, and 30 DAA every dusk (17:00 h), after which the whole -2 leaf sheaths and panicles were removed. The -2 sheaths were frozen in liquid nitrogen for 1 min before storage at −80°C for indoor experiments. To weigh the grain dry weights, the panicles were dried at 80°C to a constant weight after deactivation at 105°C for 30 min. The grain-filling processes were fitted with Richards’ growth equation (Richards, 1959). In addition, frozen samples of flag leaves and grains of CK plants were collected at 10 DAA for the measurement of SnRK1 activity. The sampling for the comparison between WT and snrk1a mutant plants was the same as that of the LC experiment.

For the dry matter and NSC content and starch content determinations in different internodes and sheaths, 200 plants with similar growth pattern that flowered on the same day were selected and tagged. Twenty tagged plants were harvested at 0, 5, 10, 15, 20, and 30 DAA. The first, second, third, and fourth sheaths and internodes from the top (-1/-2/-3/-4 sheaths and -1/-2/-3/-4 internodes) were removed and dried at 80°C to a constant weight after deactivation at 105°C for 30 min. The samples were ground to a high-throughput tissue grinder and sieved through a 100-mesh sieve for determination.

For sheath starch, sucrose and T6P content determination in a day/night cycle, 100 plants with a similar growth pattern and flowered on the same day were selected and tagged. The midnight of 5 DAA was the first sampling time, and plants were harvested at 3-h intervals until the midnight of 6 DAA. Ten plants were harvested each time. The -2 sheaths were removed and frozen in liquid nitrogen for 1 min before storage at −80°C.

Photosynthesis measurement and starch staining

The photosynthetic parameters were measured using Li-6400 Portable Photosynthesis System (Li-Cor Inc., Lincoln, NE, USA). To visualize the starch in the sheaths, the harvested rice sheaths were cleared in 80% (v/v) ethanol and stained with Lugol’s iodine.

Measurement of sucrose, starch, and NSC contents

The procedure for the measurement of sucrose and starch was performed according to the method of Chen et al. (2019). Sucrose solution was extracted with 80% ethanol (v/v) and analyzed by UltiMate 3000 UHPLC (Thermo Fisher Scientific, Waltham, MA, USA) coupled with an ELSD 6000 (Alltech, Deerfield, IL, USA) and using a Shodex Asahipak (NH2P-50 4E column). After sugars were extracted, the residue was used to extract starch with 9.2 mol L⁻¹ HClO₄. Starch concentrations were determined using the anthrone method. The NSC content was measured according to the method of Li et al. (2017).

Microscopy analysis of starch grains

The top and bottom of -2 sheaths at 0, 5, 10, and 15 DAA were sampled and fixed in 0.1-M phosphate-buffered saline (pH = 7.2) with 2.5% (v/v) glutaric dialdehyde at 4°C for 24 h. Then, the samples were postfixed in 1% OsO₄ solution for 12 h. After being dehydrated with an acetone series, the samples were embedded in epoxy resin, penetrated for 12 h at 37°C, and then polymerized at 60°C for 36 h. Ultra-thin sections (50 nm) were cut using an ultramicrotome (Pharmacia LKB, Sweden) and observed under an H7650 TEM (Hitachi, Tokyo, Japan).

Measurement of enzymatic activity

The activities of α-amylase, β-amylase, and SPS were measured using biochemical kits from Suzhou Comin Biotechnology Co., Ltd. (Suzhou, China; http://www.cominbio.com/). The samples were measured following the manufacturer’s instructions.

Determination of relative gene expression

Total RNA of the rice sheaths was isolated using RNAprep Pure Plant Kit (Tiangen Biotech, Beijing, China) and then reverse-transcribed with oligo(dT) primers using PrimeScript RT Reagent Kit (Takara, Kyoto, Japan). Relative gene expression was performed by qPCR using ABI 7300 sequencer (Thermo Fisher Scientific) and SYBR Green (Takara), according to the manufacturer’s protocol. Three biological replicates (each composed of a pool of four to five randomly sampled sheaths) were conducted from each time point and treatment with three technical replicates. The primers used are listed in Supplemental Table S10.

Assay for SnRK1 activity

The SnRK1 activity was according to the method previously described by Zhang et al. (2009). Frozen tissues were ground in liquid nitrogen using a high-throughput tissue grinder. About 150 mg of ground tissue was used to the extracted total protein with 700-μL precooled buffer (pH 8) containing 100-mM Tricine-NaOH, 0.5-mM Ethylenediaminetetraacetic acid, 0.5-mM EGTA, 25-mM NaF, 5-mM DTT, 1-mM benzamidine, 2-mM tetrasodium pyrophosphate, 2% polyvinyl pyrrolidone (PVP) (w/v), 1-mM PMSF, and 1% P9599 protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO, USA) and PhosStop (Roche, Basel, Switzerland). After centrifugation at 13,000g for 20 min at 4°C, and the 300-μL supernatant was spin desalted and concentrated using Amicon Ultra Filters (0.5 mL, 10 K; Merck Millipore, Billerica, MA, USA). The concentrated solution was added with 1% P9599 and 2.5-mM okadaic acid to a concentration of before storage in a −80°C freezer. The SnRK1 activity was measured using a 35-μL concentrated solution placed in 96-well plates. The reaction solution contained 70-mM HEPES-NaOH (pH 7.5), 1-mM okadaic acid, 8.75-mM DTT, 8.75-mM MgCl₂, P9599, 200-mM okadaic acid, 8.75-mM DTT, 8.75-mM MgCl₂, P9599, 200-μM [γ³²P]ATP (Perkin-Elmer, Waltham, MA, USA), and 200-μM AMARA peptide (Sangon Biotech, Shanghai, China). The assay was initially performed using a 5-μL extract for 6 min at 30°C, followed by transferring a 25-μL extract to a 4-cm² P81 phosphocellulose paper. About 1% phosphoric acid immersed in acetone was used to wash this paper. After drying, the paper was transferred to a tube with 4-mL scintillation fluid (Perkin-Elmer, Waltham, MA, USA). This assay was also performed using 1-mM T6P to examine the T6P-mediated inhibition of SnRK1 in different tissues.

Determination of T6P content

The T6P extraction method was according to the method described by Delatte et al. (2009). T6P was determined using ACQUITY UPLC I-Class PLUS System (Waters, Milford, MA, USA) coupled with Xevo TQ-S Micro MS/MS (Waters) and using a ACQUITY UPLC BEH HILIC column (3 × 100 mm, 1.7 μm; Waters). The liquid chromatography tandem-mass spectrometry (LC–MS/MS) conditions were modified based on the method by Sastre Toraño et al. (2012). The standards for T6P, sucrose-6-phosphate (S6P), and lactose-1-phosphate (L1P) were obtained from Sigma-Aldrich. The spectrum of the solution with 400 ppm L1P, S6P, and T6P showed that L1P and S6P did not co-elute with T6P (Supplemental Figure S9).

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and western blotting

Total protein was isolated using T-PER Tissue Protein Extraction Reagent (Thermo Pierce, Rockford, IL, USA). Enhanced BCA Protein Assay Kit (Beyotime, Shanghai, China) was used to determine the protein content. For western blotting, the specific antibody used for SnRK1a was generated according to the procedures described by Lu et al. (2007). The phosphorylation level of SnRK1 was evaluated using p-AMPKα (T172) antibody (CST), with the goat anti-rabbit IgG (Thermo Pierce) as the secondary antibody. Tubulin was used as the internal control. Western blotting was performed using ECL luminescence reagent (Absin, Shanghai, China), following the manufacturer’s protocol.

Preparation of samples for phosphoproteomic and proteomic analyses

Total proteins were extracted from 500 mg of ground tissue using 4 mL lysis buffer (8-M carbamide, 10-mM DTT, 1% protease inhibitor cocktail, and 1% Triton-100) in an ultrasonic processor on ice. Cold 20% TCA was used to precipitated the protein after centrifugation. After another round of centrifugation, the residue was washed with pre-cooled acetone and then dissolved in 8-M carbamide. Protein content was determined using a BCA Protein Assay Kit (Merck, Kenilworth, NJ, USA). The protein extract was alkylated with 11-mM iodoacetamide after reduction with 5-mM DTT. The protein was incubated with trypsin twice for digestion.

The phosphopeptides were enriched using immobilized metal affinity chromatography (IMAC). The peptide solution was incubated with the IMAC microsphere suspension, and then the IMAC microspheres were collected by centrifugation, which contained enriched phosphopeptides. To remove the nonspecifically adsorbed peptides, the IMAC microspheres were washed several times. After incubation with elution buffer, the solution was shaken for 10 min to isolate phosphopeptides and microspheres. After centrifugation, the supernatant containing phosphopeptides was collected and lyophilized for LC–MS/MS analysis.

LC–MS/MS analysis

LC–MS/MS was performed using timsTOF Pro MS (Bruker Daltonics, Bremen, Germany) connected to NanoElute 1000 UPLC system (Bruker, Billerica, MA, USA). The liquid chromatography solvents were water (solvent A) and acetonitrile (solvent B) with 0.1% formic acid. The elution gradient was as follows: 6%–24% solvent B for 70 min, 24%–35% for 14 min, 35%–80% for 3 min, and holding at 80% for 3 min. The flow rate was 450 nL min⁻¹. After being subjected to an ion source, the peptides and fragments were detected using timsTOF Pro MS via parallel accumulation-serial fragmentation data collection mode, with the scanning range set at 80–1,700 m/z for the mass spectrometry scan.

Database search and data analysis

Raw data were processed using the MaxQuant software package version 1.6.6.0 and searched against Oryza sativa subsp. japonica in the UniProtKB database. Trypsin/P was designated as the lyase, allowing the two missing cleavages. The mass tolerance of the first and main searches for precursor ions was twenty ppm, while the mass tolerance for fragment ions was set to 0.02 Da. The fixed and variable modifications were Carbamidomethyl on cysteine and phosphorylation (serine, threonine, and tyrosine), respectively. For protein and peptide spectrum match identification, the false-discovery rate was adjusted to <1%, while the minimum score for phosphopeptides was adjusted to >40.

The identified protein domains were functionally annotated using InterProScan and InterPro domain database. The UniProt-GOA database was used to derived the gene ontology (GO) annotation proteome. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database descriptions of proteins was annotated using KEGG online service tools KAAS, while the annotation results to the KEGG pathways were mapped using KEGG mapper. The subcellular localization was predicted using Wolfpsort.

Additionally, the model of sequences was analyzed using the Motif-X algorithm (Chou and Schwartz, 2011), which constituted with amino acids at specific positions of the modified-13-mers (6 aa upstream and downstream of the site). The background database parameters were all database protein sequences, with the number of occurrences >20. The GPS version 5.0 software (Wang et al., 2020b) was used to predict kinase–substrate regulation, and the corresponding proteins in the kinase family were obtained from the iEKPD version 2.0 database (http://iekpd.biocuckoo.org/). GSEA method was used to predict kinase activities (Subramanian et al., 2005). WGCNA R package was used to analyzed the co-expression clusters (Langfelder and Horvath, 2008). OmicShare tools, a free online platform for data analysis, used to perform GO and KEGG enrichment analyses.

Quantification of protein abundance and phosphorylation level of selected phosphosites by PRM

Protein extraction, trypsin digestion, and affinity enrichment were conducted using the same methods for phosphoproteomic analyses. Three biological replicates (each composed of a pool of four to five randomly sampled sheaths) were conducted from CK and LC plants. PRM was performed using Q Exactive Plus (Thermo Fisher Scientific) coupled to the EASY-nLC 1000 UPLC system (Thermo Fisher Scientific). The liquid chromatography solvents were water (solvent A) and acetonitrile (solvent B) with 0.1% formic acid. The gradient was as follows: 3%–17% solvent B for 40 min, 17%–28% for 12 min, 28%–80% for 6 min, and holding at 80% for 4 min. The flow rate was 500 nL min⁻¹. After being subjected to an NSI source, the peptides were detected at a resolution of 70,000 in the Orbitrap. The full scan was performed using the range from 410 to 1,255 m/z. Then, a DIA setting of 27 was used to select peptides for MS/MS. The fragments were detected at a resolution of 35,500 in the Orbitrap. The AGC was set at 3E6. The maximum IT for full MS and MS/MS was at 50 and 145 ms, respectively. The AGC for MS/MS was 1E5, with isolation window set as 1.6 m/z. The resulting MS data was processed using Skylin (version 3.6).

Data analysis

SPSS version 19.0 software (SPSS Inc., Chicago, IL, USA) was used for statistical analysis included one-way analysis of variance and Student’s t test. Pearson correlation coefficient (r) analysis was used to determine the correlation coefficient. GraphPad Prism version 8 (GraphPad Software, Inc., San Diego, CA, USA) was used for data visualization.

Data Submission

The proteomic and phosphoproteomic raw data have been submitted into the ProteomeXchange Consortium via the PRIDE partner repository with identifier PXD024670 for proteomics and PXD024675 for phosphoproteomics data (Ternent et al., 2014).

Accession numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: OsSnRK1a (Os05g0530500), OSK24 (Os08g0484600), OSK35 (Os03g0289100), OsSUT1 (Os03g0170900), OsSUT4 (Os02g0827200), OsRAM3D (Os08g0473900), OsRAM2A (Os06g0713800) and OsBAM3 (Os03g0141200), and Actin (Os03g0718100).

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Enzyme activities of α-amylase, β-amylase, and SPS in the sheaths of CK and LC treatment plants during grain filling.

Supplemental Figure S2. T6P content at each time point was plotted against SnRK1 activity.

Supplemental Figure S3. The SnRK1a protein sequence of the snrk1a mutant.

Supplemental Figure S4. Hierarchical cluster tree of DESs identified in the WT and snrk1a mutant plants at 0, 4, 10, and 15 DAA.

Supplemental Figure S5. Dynamics of metabolite levels of the sheath within 1 day.

Supplemental Figure S6. Relative expression levels of OsRAM3D, OsRAM2A, and OsBAM3 in the sheaths of CK and LC treatment plants at 6 DAA.

Supplemental Figure S7. Sequence alignment of the C-terminal ends in the aquaporins of Arabidopsis and rice.

Supplemental Figure S8. The daily mean temperature and daily mean photosynthetically active radiation after transplanting.

Supplemental Figure S9. Spectrum of the solution containing 400 ppm L1P, S6P, and T6P at m/z 421.1 using the LC/MS method described in the study.

Supplemental Table S1. List of identified proteins of the sheaths from WT and snrk1a mutant plants, as determined by proteomics.

Supplemental Table S2. List of identified phosphosites of the sheaths from WT and snrk1a mutant plants, as determined by phosphoproteomics.

Supplemental Table S3. List of upregulated and downregulated proteins from snrk1a mutant versus WT at 0, 4, 10, and 15 DAA.

Supplemental Table S4. List of upregulated and downregulated phosphosites from snrk1a mutant versus WT at 0, 4, 10, and 15 DAA.

Supplemental Table S5. Motif-X analysis results of significantly deregulated phosphosites from snrk1a mutant versus WT.

Supplemental Table S6. List of phosphosites of seven PPMs.

Supplemental Table S7. Enrichment analysis of PPMs based on current pathway databases (KEGG and GO).

Supplemental Table S8. List of the 30 SnRK1-dependent phosphosites involved in NSC transport.

Supplemental Table S9. PRM analysis of 20 phosphosites in sheaths involved in NSC transport under CK and LC at 6 DAA.

Supplemental Table S10. Primers used in this study for qPCR.

Funding

This work was supported by the National Natural Science Foundation of China (Grants No 31901454), the Natural Science Foundation of Jiangsu Province for Distinguished Young Scientists (Grants No BK20200079) and the Collaborative Innovation Center for Modern Crop Production co-sponsored by Province and Ministry (CIC-MCP).

Conflict of in terest statement. The authors declare that they have no conflict of interest.

 

L.C., C.D., Y.D., and Y.H. conceived the research plan and designed the experiments. X.X., Y.X., and J.B. assisted with sample collection. Y.H., J.L., Y.L., and Y.Y. performed the experiments. F.X. assisted with SnRK1 activity determination. Y.H. and L.C. analyzed data and wrote the manuscript. L.C. edited the manuscript. All the authors proofread the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Lin Chen (linchen@njau.edu.cn).