Genetic improvement of phosphate-limited photosynthesis for high yield in rice

Significance

Phosphate availability is an important limiting factor realizing the potential of photosynthesis in the field, and understanding genetic approaches of how phosphate limitation can be relieved is essential for further yield improvement. We found that OsPHO1;2 functions as one such regulator counteracting phosphate limitation through phosphate allocation toward leaves. Loss function of OsPHO1;2 caused Pi deficiency, disrupted photosynthetic electron transport, and CO₂assimilation. Overexpression of OsPHO1;2 postponed the occurrence of Pi limitation and substantially improved the performance of photosynthesis, thereby contributing to grain yield. We also found a correlation between OsPHO1;2 expression, photosynthesis, and grain yield among natural rice varieties. Comparable to foliar phosphate application, genetic strategy is proven to be effective in modulating leaf Pi for efficient photosynthetic production.

Abstract

Low phosphate (Pi) availability decreases photosynthesis, with phosphate limitation of photosynthesis occurring particularly during grain filling of cereal crops; however, effective genetic solutions remain to be established. We previously discovered that rice phosphate transporter OsPHO1;2 controls seed (sink) development through Pi reallocation during grain filling. Here, we find that OsPHO1;2 regulates Pi homeostasis and thus photosynthesis in leaves (source). Loss-of-function of OsPHO1;2 decreased Pi levels in leaves, leading to decreased photosynthetic electron transport activity, CO₂ assimilation rate, and early occurrence of phosphate-limited photosynthesis. Interestingly, ectopic expression of OsPHO1;2 greatly increased Pi availability, and thereby, increased photosynthetic rate in leaves during grain filling, contributing to increased yield. This was supported by the effect of foliar Pi application. Moreover, analysis of core rice germplasm resources revealed that higher OsPHO1;2 expression was associated with enhanced photosynthesis and yield potential compared to those with lower expression. These findings reveal that phosphate-limitation of photosynthesis can be relieved via a genetic approach, and the OsPHO1;2 gene can be employed to reinforce crop breeding strategies for achieving higher photosynthetic efficiency.

Phosphorus (P) is an indispensable element for plant growth and function. Phosphate (Pi), the common inorganic form of phosphorus, plays vital roles in major plant metabolic processes, particularly photosynthesis (1). Pi is an essential component of adenosine triphosphate (ATP), nicotinamide adenine dinucleotide phosphate (NADPH), sugar phosphates, and phospholipids, all of which are crucial in photosynthetic processes (2). Pi deficiency has a strong negative impact on plant growth, development, and grain filling. As a result, large volumes of Pi fertilizer are commonly applied to crop plants to ensure high production (3, 4). Previous research has uncovered several mechanisms evolved by plants to overcome Pi deficiency, such as enhancing Pi uptake from the soil, improving Pi use efficiency (PUE), and recycling Pi (5, 6). Identification of additional Pi-efficiency mechanisms or regulators could allow improvement of photosynthesis and ultimately increase yield (6, 7), as current crop yield is far from sufficient to meet the world’s population facing reduced arable lands, threats of pests and pathogens, as well as weather extremes (8).

The process of photosynthesis involves the light reactions followed by the Calvin–Benson–Bassham (CBB) cycle. In the light reactions, ATP is generated using adenosine diphosphate (ADP) and Pi. In the CBB cycle, ATP and NADPH from the light reactions are consumed to generate triose phosphate (TP), which is then used in carbohydrate synthesis (9). Leaf photosynthetic activity is limited by three major components: Rubisco activity, which is reflected by the maximum carboxylation rate (V_cmax); photosynthetic electron transport (J_max), which determines the ability to regenerate the Rubisco substrate RuBP; and TP utilization (TPU), which drives sucrose synthesis from precursors in the CBB cycle (10). V_cmax and J_max are considered major biological limitations of photosynthesis, while TPU has received less attention from researchers (11–13). Improved understanding of Pi limitations (including TPU limitation) may contribute to yield improvement, because sugar production in source tissues and consumption by sink tissues directly affect crop yield (14–16). Sucrose synthesis involves TP export from the chloroplast to the cytosol in exchange for Pi import via the TP/Pi translocator (TPT). In Arabidopsis thaliana TPT-knockout mutants, the CO₂ assimilation rate is not sensitive to CO₂ levels after a threshold is reached, causing Pi-limited photosynthesis to occur above that point (17, 18). A recent study suggested that a sufficient capacity of chloroplast TP isomerase (cpTPI) is necessary to prevent the occurrence of Pi-limited photosynthesis and that cpTPI content affects photosynthetic capacity at elevated CO₂ levels (19). Indeed, Pi deficiency caused phosphate limitation counteracting efficient photosynthesis in crops. However, further mechanisms by which Pi limitation in photosynthesis can be relieved or avoided genetically, particularly in cereals, remains to be discovered.

Pi starvation, a consequence of Pi deficiency, places considerable constraints on photosynthetic activity (16). Typically, roots absorb soluble Pi from the soil, and Pi is then transferred to aerial tissues. Experimental Pi-deficiency treatments (limiting the P supplementation in the soil) may lead to Pi starvation of the entire plant. Numerous Pi transporters are responsible for Pi uptake and translocation between tissues, including members of the PHT and PHO families identified in cereal plants (20). In rice, mutants for PHT1 family members (e.g., Ospht1;4 and Ospht1;8) show decreased Pi content throughout the plant, including in the roots, leaves, and seeds (21, 22), similar to the Pi-deficiency treatments. However, these make it difficult to precisely evaluate photosynthetic performances that are specifically associated with Pi deficiency in the leaves. In contrast, mutants for the Pi transporter OsPHO1;2, an important grain-filling regulator in cereals, show increased Pi concentration in the seeds but reduced Pi concentration in the flag leaves (23). Thus, OsPHO1;2 likely has important functions in Pi allocation and homeostasis in leaves and may be used to more accurately study Pi-related photosynthetic regulation.

Photosynthetic activity during the grain-filling phase in cereal plants is by far the most important contributor to yield (24). In the present study, we found a genetic modulator OsPHO1;2 for phosphate limitation of photosynthesis in rice. Loss of OsPHO1;2 function decreased the photosynthetic rate, and caused earlier photosynthetic Pi limitation in flag leaves at critical grain-filling stage. More importantly, OsPHO1;2 overexpression or Pi supplementation by foliar spraying of Pi fertilizers improved the photosynthetic rate and prolonged the efficient period of flag leaf photosynthesis, contributing considerably to increased grain yield. This study demonstrates the impact of leaf-targeted Pi deficiency on crop photosynthesis, and establishes a viable path for enhancing grain yield by modulating Pi-related photosynthetic performance.

Results

OsPHO1;2 Expression Correlates with Leaf Pi Content and Photosynthetic Rate.

Our previous study of OsPHO1;2 demonstrated that it has a role in Pi reallocation in developing seeds; besides, Pi content was significantly decreased in the leaves of Ospho1;2 mutants but increased in OsPHO1;2-OE leaves (23). We here sought to explore whether OsPHO1;2-mediated changes in Pi concentration were associated with changes in leaf photosynthetic performance. Pi accumulation and photosynthetic parameters were therefore measured in the leaves of several genotypes which were generated from our previous study: a pair of near-isogenic lines (NILs) (NIL-OsPHO1;2 and NIL-Ospho1;2); an OsPHO1;2 knockout mutant (Ospho1;2-ko1); OsPHO1;2 overexpression (OE) (driven by CaMV 35S promotor) lines; and the wild-type (WT) (23). Time-course data were recorded from field-grown plants during the grain-filling stage.

The Pi concentration and net photosynthetic CO₂ assimilation rate (A_net) were significantly lower in all the measured leaves of NIL-Ospho1;2 and Ospho1;2-ko1 mutant plants than that of NIL-OsPHO1;2 and WT plants, respectively, at the early grain-filling stage [5 d after flowering (DAF)] (SI Appendix, Fig. S1 A–D). Flag leaves showed the highest Pi concentration and photosynthetic rate, therefore, flag leaf was chosen for most leaf-Pi and photosynthetic performance analysis. We then performed Pi treatments under external Pi sufficient and deficient supplies in the independent field pools (Materials and Methods). As expected, the leaf Pi level and photosynthetic rate of all the measured leaves were lower in Pi-deficient condition than those in Pi-sufficient condition. Importantly, we confirmed that flag leaf Pi content and A_net were significantly decreased in NIL-Ospho1;2 and Ospho1;2-ko1 mutants than those in NIL-OsPHO1;2 and WT plants, at both Pi sufficient and deficient conditions during grain filling (SI Appendix, Fig. S1 E–H). We also detected increased flag leaf Pi concentration and A_net in all three OsPHO1;2-OE lines during grain filling (SI Appendix, Fig. S1 I and J), and therefore one of the OsPHO1;2 overexpression lines OsPHO1;2-OE #1 (hereafter called OsPHO1;2-OE) was randomly selected for following study.

To clarify the relationship between Pi accumulation and leaf photosynthesis, flag leaf Pi levels and photosynthetic rates were compared throughout the grain-filling stage. First, NIL-Ospho1;2 and Ospho1;2-ko1 mutants had distinctly lower Pi concentration and A_net in the flag leaves throughout the grain-filling stage compared to the corresponding controls (Fig. 1 A, B, D, and E). Second, Pi concentration and A_net were higher in OsPHO1;2-OE than in WT plants, particularly in the middle and late grain-filling stages (starting from 10 DAF) (Fig. 1 C and F). These suggested that increased levels of OsPHO1;2 aided Pi availability in flag leaves, relieving the Pi deficiency that is common in the middle and late grain-filling stages. Notably, during the late grain-filling stage, both Pi content and A_net displayed a downtrend due to leaf senescence (25); the higher Pi level and A_net in OsPHO1;2-OE compared to the WT may have also resulted from a slower senescence rate (Fig. 1 C and F), which was further supported by higher relative chlorophyll content in OE leaves (SI Appendix, Fig. S2 A–C). Overall, there was a strong positive correlation between Pi concentration and photosynthesis in each of the plant lines measured, and this was further reflected by scatter plots (Fig. 1G).

Fig. 1.

OsPHO1;2 allocates Pi to the leaves and facilitates photosynthesis. (A–C), Measurements of Pi concentration in flag leaves at the grain-filling stage from 0 DAF to 30 DAF in NILs (A), Ospho1;2-ko1 mutant (B), and OsPHO1;2 overexpression line (C). Data represent six plants (n = 6). (D–F) Measurements of net carbon assimilation rate (A_net) in flag leaves at 0 to 20 DAF in NILs (D), Ospho1;2-ko1 (E), and OsPHO1;2-OE (F). n = 6 plants in (D) and n = 12 plants in (E and F), respectively. (G), The dot plot of correlation between flag leaf Pi concentration and A_net among WT, Ospho1;2-ko1, OsPHO1;2-OE, and NILs. The trend line and R² value were indicated in the graph. (H–J), Measurement of Pi concentration in isolated chloroplasts of NILs (H), WT/Ospho1;2-ko1 (I), and WT/OsPHO1;2-OE (J) flag leaves at the early and middle grain filling stage (n = 24). (K) ³²P short-term stem-fed assay indicates the distribution ratio of newly absorbed Pi, in different tissues of the above-last-node part of rice plants. Four tissues were analyzed; error bars represent means ± SD of four plant repeats (n = 4). (L) Relative expression of OsPHO1;2 in WT and spdt-1 mutant flag leaves detected by qPCR. (M) Relative expression of SPDT in WT and Ospho1;2-ko1 mutant flag leaves detected by qPCR. The rice OsActin gene was used as internal control. Values are means ± SD (n = 3 biological repeats). P-values were indicated according to two-tailed Student’s t-tests.

Ospho1;2 Specific Pi Deficiency Affects Pi Availability in Chloroplasts.

Leaf photosynthesis takes place in chloroplasts. To further illustrate whether the reduced photosynthetic activity in Ospho1;2 mutants was related to changes of Pi content in chloroplasts, we performed Pi concentration measurement in isolated chloroplasts during grain filling. We found that Pi concentration in the chloroplasts of both NIL-Ospho1;2 and Ospho1;2-ko1 mutants were significantly reduced in flag leaves compared to their wild types (Fig. 1 H and I), suggesting that the decreased photosynthetic activity in Ospho1;2 mutant leaves likely resulted from reduced Pi availability in chloroplasts. Importantly, the chloroplast Pi concentration in OsPHO1;2-OE plants was increased particularly in middle grain filling stage (Fig. 1J), consistent with the flag leaf Pi level changes.

We next confirmed that mutants for the other two PHO1 family members in rice, Ospho1;1 and Ospho1;3, showed no differences in A_net or Pi concentration in the flag leaves (SI Appendix, Fig. S2 D and E). This suggested that OsPHO1;2 was a specific PHO1-type transporter that not only coordinated Pi reallocation and grain filling, but also positively regulated leaf photosynthesis in rice. During the vegetative growth stage in the field, there were no obvious morphological differences between WT and Ospho1;2 plants (SI Appendix, Fig. S3 A and C), although both Pi concentration and A_net were significantly decreased in the second and third leaves of Ospho1;2 mutants (SI Appendix, Fig. S3 E–H), together with significantly reduced Pi content in the chloroplasts of Ospho1;2 mutants (SI Appendix, Fig. S3 I and J). This may have been because that root-to-shoot Pi translocation was perturbed during the vegetative stage without functional OsPHO1;2 (23, 26), but not sufficient to affect the overall morphology.

OsPHO1;2 Preferentially Allocates Pi to Source Leaves to Promote Photosynthesis.

To further assess the relationship between Pi level and photosynthesis, we also assayed several low phytic acid (lpa) mutants (27, 28), which showed comparable seed Pi accumulation to Ospho1;2 mutants (SI Appendix, Fig. S4A). Z9B-lpa and MH86-lpa were caused by OsSULTR3;3 mutation that disturbed phytic acid metabolism (28), so that both lpa and Ospho1;2 mutants showed greatly accumulated Pi but decreased phytic acid in the seeds (23, 27, 28). We also observed reduced OsPHO1;2 expression in lpa grains and flag leaves, together with reduced grain weight and yield in lpa mutants (SI Appendix, Fig. S4 B, I, J, and K). Significantly, decreased Pi concentration and A_net in the flag leaves of lpa mutants were found during grain filling (SI Appendix, Fig. S4 C–H). Although Ospho1;2 and lpa mutants caused decreased leaf Pi concentration via different mechanisms, the effect of OsPHO1;2-mediated variation in Pi level on leaf photosynthesis was further supported.

In rice, the SULTR-like P distribution transporter (SPDT, also known as OsSULTR3;4) reportedly functions in the nodes as a key regulator of P allocation to the grains. Different from OsSULTR3;3, SPDT/SULTR3;4 function relies on its strong inward-facing Pi transport activity, and its mutation disturbed P allocation to grains thus decreased phytic acid content (28, 29). Similarly, OsPHO1;2 is highly expressed in the nodes and is known to be involved in P allocation (23, 30). To compare the functions of these two Pi transporters in Pi homeostasis and leaf photosynthesis, spdt knockout mutants were generated using the CRISPR-Cas9 system; all independent mutant alleles were verified with sequencing and gene expression analysis (SI Appendix, Fig. S5 A and B). Spdt mutants showed similar morphology to WT plants (SI Appendix, Fig. S5 C and D), although they accumulated higher levels of Pi in the flag leaves, particularly at the late grain-filling stage. A_net was also consistently higher in the leaves of spdt mutants, likely due to the increased Pi (SI Appendix, Fig. S5 E and F), however, Pi contents in the grains of spdt mutants were significantly decreased (SI Appendix, Fig. S5G). These results were distinct from those observed in the Ospho1;2 mutants (Fig. 1 A–D), suggesting that the two genes have distinct effects on leaf photosynthesis.

A ³²P isotope pulse labeling assay was next conducted to explore Pi distribution patterns in Ospho1;2 and spdt mutants. During the grain-filling stage, there was significantly less ³²P absorption in the flag leaves and stems of Ospho1;2-ko1 mutants compared to the WT; the opposite effect was observed in spdt mutants. On the contrary, ³²P levels were increased in the nodes and panicles of Ospho1;2-ko1 mutants compared to the WT, while an opposite was again found in spdt mutants. These results further supported the hypothesis that the grain-filling defects in Ospho1;2 was associated with high Pi concentration in grains according to our previous study (23). Together, these demonstrated contrasting results in Pi distribution between the two mutants (Fig. 1K). Our data suggested that OsPHO1;2 preferentially allocated Pi to the leaves rather than the panicles, thus supporting leaf photosynthesis; this function was distinct from that of SPDT. Gene expression analysis showed that OsPHO1;2 was up-regulated in spdt mutants and that SPDT was up-regulated in Ospho1;2-ko1 mutants (Fig. 1 L and M), which suggested the two transporters together maintain Pi homeostasis.

OsPHO1;2 Alleviates Pi-Limitation of Photosynthesis during Grain Filling.

Pi limitation is one of the key factors that affect leaf photosynthetic rates (16, 19). Typical Pi-limited photosynthesis in the field is characterized by limited CO₂ assimilation rates under high-light and high-CO₂ concentrations, as indicated by a downward turn of the A_net/intercellular CO₂ concentration (A-C_i) response curve (31). To test whether OsPHO1;2 was associated with this process, especially during the grain-filling stage (when efficient photosynthesis is most necessary), we measured A-C_i response curves and calculated the related photosynthetic parameters. Pi-limited photosynthesis was observed at 10 DAF in both NIL-Ospho1;2 plants and Ospho1;2-ko1 mutants compared with NIL-OsPHO1;2 and WT plants, respectively; in NIL-Ospho1;2, A_net deviated more clearly downward compared to the control (Fig. 2 A and B). At 10 DAF, A_net was significantly lower in Ospho1;2-ko1 mutants than in the WT or OsPHO1;2-OE plants. However, C_i levels were similar between those three lines, indicating that C_i was not the causative factor (SI Appendix, Table S1). In addition, V_cmax, J_max, and TPU were all reduced in Ospho1;2-ko1 mutants and in NIL-Ospho1;2 plants compared with WT and NIL-OsPHO1;2 plants, respectively (SI Appendix, Table S1).

Fig. 2.

OsPHO1;2 functions to relieve the Pi-limitation of photosynthesis during grain filling. (A and B), Measurements of the response of carbon assimilation rate (A_net) to intracellular CO₂ concentration (C_i) (A-C_i curves) at 30 °C and 1,200 µmol photons, in NILs (A) and WT/Ospho1;2-ko1 (B) at 10 DAF. (C and D), Measurements of A-C_i curves at 30°C and 1,200 µmol photons in OsPHO1;2-OE line at 10 DAF (C) and 14 DAF (D). Flag leaves were measured. The shaded part between dotted lines represents the means ± SD of four plants (n = 4).

Based on the A-C_i response curves of WT and OsPHO1;2-OE plants, it did not appear that either line experienced Pi limitation at 10 DAF, whereas mild Pi limitation was observed in the WT at 14 DAF (Fig. 2 C and D). At 14 DAF, A_net, V_cmax, J_max, and TPU were correspondingly higher in OsPHO1;2-OE compared to WT plants, but there were no differences in C_i (SI Appendix, Table S1), suggesting ectopic overexpression of OsPHO1;2 delayed the occurrence of Pi-limited photosynthesis. Neither WT nor OsPHO1;2-OE plants showed signs of Pi limitation at the earlier grain-filling stage (3 to 7 DAF) (SI Appendix, Fig. S6 A–D). These data demonstrated that OsPHO1;2 enhanced Pi availability in leaves and thus moderated Pi limitation of photosynthesis during grain filling, and loss of OsPHO1;2 causes earlier occurrence of Pi limitation.

OsPHO1;2 Function Is Necessary for Proper Photosynthetic Protein Expression and Phosphorylation Levels.

To investigate the effects of alterations in OsPHO1;2 expression and Pi availability, we conducted an RNA-sequencing analysis of the flag leaves at 5 and 15 DAF (early and middle grain filling, respectively) in WT, Ospho1;2-ko1, and OsPHO1;2-OE plants. The resulting data were analyzed to identify differentially expressed genes (DEGs) between WT and Ospho1;2-ko1 mutant at the two time points. We identified 69 (at 5 DAF) and 2,265 DEGs (at 15 DAF) in Ospho1;2-ko1 mutant compared to the WT respectively (SI Appendix, Fig. S7 A and B and Dataset S1). Gene ontology (GO) term enrichment analysis was then conducted among the DEGs to identify enriched processes or functions at each time point. At the early grain-filling stage, DEGs were enriched in molecular function terms such as “phosphotransferase”, “kinase”, and “transcription factor activity” (SI Appendix, Fig. S7C); at the middle grain-filling stage, DEGs were enriched in biological process terms such as “phosphate ion transport” and “responses to Pi starvation” (SI Appendix, Fig. S7D). Many known Pi starvation-related (PSR) genes (32, 33) were up-regulated in Ospho1;2-ko1 mutant (Fig. 3A and SI Appendix, Fig. S7 D and E), indicating that these plants experienced Pi deficiency in the leaves. However, levels of only a few photosynthetic genes, including AtpΓ, PsbD, Lhcb, PsaL, and Lhca, were altered in Ospho1;2-ko1 mutant (SI Appendix, Fig. S7F). We also assessed expression levels of senescence-associated genes (SAGs) because phosphorus metabolism and transport are closely associated with leaf senescence (25, 34). Four positive regulators of senescence were up-regulated in Ospho1;2-ko1 mutant (WRKY53, OsH36, Osl30, and MYC2), whereas the negative senescence regulator OsRCCR1 was down-regulated (SI Appendix, Fig. S7F). These results indicated that OsPHO1;2-mediated photosynthetic regulation might occur less at the transcriptional level.

Fig. 3.

Loss of OsPHO1;2 causes up-regulated PSR genes, declined photosynthetic protein expression, and phosphorylation levels. (A) A heatmap of 97 differentially expressed P-starvation related (PSR) genes in flag leaves relative to WT during grain filling. The relative fold change is displayed on a log2 scale: red indicates up-regulated and blue indicates down-regulated, respectively. (B) Changes of photosynthetic proteins in flag leaves at grain filling stage detected by western blot, proteins of AtpB, PsaA, PsbA, Lhca1, Lhcb1, Rubisco activase (RCA), ndhH were detected. ko1 represents Ospho1;2-ko1, and OE represents OsPHO1;2-OE. Antibody of Actin was used as an internal control. The experiment was repeated twice with similar results. (C) The heatmap of differentially expressed photosynthetic, glycolysis-related, and photorespiration-related proteins between WT and Ospho1;2. Flag leaf proteomics was performed at early grain filling stages. (D) The heatmap of differential phosphorylation of photosynthetic and glycolysis-related proteins between WT and Ospho1;2 flag leaves. The relative fold change is displayed on a log2 scale: red indicates up-regulated and blue indicates down-regulated, respectively.

At the protein level, we measured the expression of several key proteins involved in light reactions and CO₂ assimilation. Immunoblot assays revealed that the levels of light reaction proteins PsaA, Lhca1, Lhcb1, and AtpB were down-regulated in Ospho1;2-ko1 mutant (Fig. 3B). Proteomic analysis of samples in the early grain-filling stage was conducted and further revealed decreased levels of PSII, PSI, and ATP synthase related subunits and of proteins involved in the CBB cycle. Furthermore, protein levels of several dephosphorylating enzymes that function in glycolysis and photorespiration were increased in Ospho1;2 mutant (Fig. 3C and Dataset S2). These results indicated that Pi deficiency in Ospho1;2-ko1 mutant exerted a strong negative effect on photosynthetic protein accumulation.

Protein phosphorylation is important for the function of many critical cellular processes, including photosynthetic reactions (35, 36). We therefore asked whether Pi deficiency in Ospho1;2 mutants would decrease the phosphorylation level of photosynthetic proteins, which could be also responsible for reduced photosynthetic rates. To answer this question, we conducted a phosphorylated proteomic analysis in WT and Ospho1;2 flag leaves during the grain-filling stage. Differentially phosphorylated proteins were identified between WT and Ospho1;2 mutant, and we analyzed the enrichment of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways among those proteins. As expected, the differentially phosphorylated proteins were significantly enriched in the “photosynthesis” and “carbohydrate metabolism” pathways (SI Appendix, Fig. S8). Importantly, many photosynthetic proteins had drastically decreased phosphorylation levels in Ospho1;2 mutant in the early grain-filling stage. These included PsbC, PsbS1, PsaD, PsaF, AtpA, AtpB, RbcS, RCA, FBA, and GAPDH (Fig. 3D and Dataset S3), belonging to light reactions, CBB cycle, or the regulation of these processes and most of which are positively related to photosynthesis (35, 37–39). Taken together, these findings suggested that Pi was required to optimize photosynthetic protein expression and phosphorylation levels, to maintain high rates of photosynthesis and carbon assimilation, and that the loss of OsPHO1;2 markedly disturbed these processes.

Changes in the Metabolic Profile of Ospho1;2 Mutants.

Pi is involved in many steps of photosynthesis and related metabolic processes. To determine how the loss of OsPHO1;2 would affect these processes, we performed metabolite profiling in the leaves of Ospho1;2-ko1 mutant and WT plants. Most metabolites that participate in the CBB cycle were markedly decreased at both 5 and 15 DAF in Ospho1;2-ko1 mutant; there were corresponding reductions in protein expression and phosphorylation levels (Figs. 3 C and D and 4 A and B and Dataset S4). Levels of the glycolytic metabolite phosphoenolpyruvate (PEP) declined in Ospho1;2-ko1 mutant, whereas there was significant accumulation of the subsequent dephosphorylation product, pyruvate (Pyr), consistent with the increased enzyme expression and phosphorylation levels of participants in the relevant reactions (Figs. 3 C and D and 4 C and D). Isocitric acid (ICIT) and succinic acid (Succ), which participate in the tricarboxylic acid (TCA) cycle, were also accumulated at higher levels in Ospho1;2-ko1 mutant than in the WT at 5 DAF; other participants, including Alpha-ketoglutarate (α-KG) and malate (Mal), were found accumulated to higher levels at 15 DAF (Fig. 4 C and D and Dataset S4). Levels of the early photorespiration product 2-phosphoglycolate (2-PG) were decreased in Ospho1;2-ko1 mutant at 5 DAF, but levels of the dephosphorylated product glycolate (Glyco) were significantly increased (Fig. 4E). Metabolites from later photorespiratory steps, such as glycine (Gly), serine (Ser), and glyceric acid (Glyce), were decreased in Ospho1;2-ko1 plant at 5 DAF (Fig. 4E and Dataset S4). At 15 DAF however, Ser and Glyce were present at higher levels in Ospho1;2-ko1 mutant, while other photorespiratory metabolites were similar between WT and the mutant (Fig. 4F and Dataset S4). Metabolite profiling was also conducted in NILs (NIL-OsPHO1;2 and NIL-Ospho1;2) plants and similar results were found (SI Appendix, Fig. S9 A–F and Dataset S4).

Fig. 4.

Metabolic profiles reveal decreased CBB metabolites, ATP, and NADPH levels in response to Ospho1;2 mutation. (A–F) Metabolite analysis of the Calvin–Benson–Bassham cycle (CBB, A and B), tricarboxylic acid cycle (TCA cycle, C and D), and photorespiration process (E and F) among WT and Ospho1;2-ko1 flag leaves at grain filling stages, including early grain filling (5 DAF, A, C, and E) and middle grain filling (15 DAF, B, D, and F). The relative levels of metabolites were shown as bar graphs, with gray and green columns indicate WT and Ospho1;2-ko1, respectively. (G–J), Relative ATP (G), ADP (H), NADPH (I), and NADP⁺ (J) contents in the flag leaves of WT and Ospho1;2-ko1 during grain filling. (K and L), Measurement of relative sucrose contents (K) and starch content (L) in WT and Ospho1;2-ko1 flag leaves at grain filling stage. For box-and-whisker plots, the central line, box, and whiskers indicate the median, IQR, and 1.5 times the IQR, respectively. Data represent five biologically independent samples (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001, according to two-tailed Student’s t-tests.

Photosynthetic carbon assimilation is driven by ATP and NADPH produced from the light reactions. It is thus crucial for ATP and NADPH production to be well balanced with the demands of the CBB cycle (40). Metabolite analysis indicated that the relative levels of ATP, ADP, NADPH, and NADP⁺ were decreased in Ospho1;2-ko1 mutant compared to the WT, and in NIL-Ospho1;2 compared to NIL-OsPHO1;2 plant at both 5 DAF and 15 DAF (Fig. 4 G–J and SI Appendix, Fig. S10 A–H). This suggested that the availability of energy molecules and reducing power, which is essential for photosynthetic carbon assimilation, were correlated with OsPHO1;2 function. Consequently, reduced content of assimilated sucrose was found in Ospho1;2 mutants flag leaves during grain filling (Fig. 4K and SI Appendix, Fig. S10I). Moreover, it is worth noting that more starch accumulation was observed in Ospho1;2 mutants than that in WT, likely due to limitation of Triose-P export to the cytosol in exchange of reduced Pi import to the chloroplasts (Fig. 4L and SI Appendix, Fig. S10J). Overall, these demonstrated that knocking out OsPHO1;2 resulted in substantial metabolic perturbation of photosynthetic CO₂ assimilation and respiratory reactions.

We also detected the metabolites in OsPHO1;2-OE flag leaves during grain filling. The relatively decreased content of RuBP coupled with the elevated contents of many downstream metabolites participating in the CBB cycle in OsPHO1;2-OE plant suggested the augmented carboxylation efficiency particularly in middle grain filling stages (SI Appendix, Fig. S11 A and B), which was consistent with the enhanced leaf Pi availability and photosynthetic rate (Fig. 1). Importantly, we also observed that the relative amounts of ATP, ADP, NADPH, and NADP⁺ were increased in OsPHO1;2-OE plant (SI Appendix, Fig. S11 G–J), further supported that overexpression of OsPHO1;2 facilitates photosynthetic metabolism and thereby leaf photosynthesis.

Decreased Photosynthetic Electron Transport and Increased Photorespiration in Ospho1;2 Mutants.

Photosynthetic electron transport drives NADPH production and is coupled with ATP synthesis. Measurements of chlorophyll fluorescence and Photosystem I absorption (P700) (41) were performed in plants at 5 DAF to evaluate the activities of linear and cyclic photosynthetic electron transport. The linear electron transport rate in Photosystem II (ETRII) was significantly decreased in Ospho1;2-ko1 compared to WT plants, whereas there were no notable differences between OsPHO1;2-OE and WT plants (Fig. 5A). The decrease in P700 absorption after switching-off of far-red light was much slower in Ospho1;2-ko1 mutants than in WT plants (Fig. 5 B and C), indicating that the cyclic electron transport rate was lower in Ospho1;2-ko1 mutant. Similar results were observed in NIL-Ospho1;2 compared to NIL-OsPHO1;2 plants (SI Appendix, Fig. S12 A and B). In contrast, the cyclic electron transport rate was higher in OsPHO1;2-OE compared with WT plants (Fig. 5 B and C).

Fig. 5.

Loss of OsPHO1;2 results in decreased photosynthetic electron transport activities and trans-thylakoid proton motive force. (A) Detection of the electron transport rate at photosystem II (ETRII) in WT, Ospho1;2-ko1, OsPHO1;2-OE (n = 5 plants). (B and C), Measurement of representative P700⁺ reduction curves in Photosystem I (B) and analysis of the initial rate of P700⁺ reduction (C) in WT, Ospho1;2-ko1, OsPHO1;2-OE (n = 6 plants). (D–F) Analysis of the representative light-off response curves (D), ΔpH components (slow rise phase amplitude indicates estimated ΔpH components of pmf, E), and decline amplitude (indicates estimated proton motive force, pmf, F) (n = 4 plants), the pmf and ΔpH components labeling in (D) is based on the blue curve of WT. Measurements were performed among WT, Ospho1;2-ko1, and OsPHO1;2-OE flag leaves at 5 DAF. P-values were indicated according to two-tailed Student’s t-tests.

The operation of both linear and cyclic electron transport is coupled with generation of the trans-thylakoid membrane proton gradient, which drives ATP synthesis. We therefore also measured the electrochromic shift (ECS) of the leaves, and the dynamics of the P515 signal were used to assess the trans-thylakoid proton motive force (pmf). The ΔpH as well as pmf were considerably lower in Ospho1;2-ko1 mutants and in NIL-Ospho1;2 plants compared to WT and NIL-OsPHO1;2 plants, suggesting a decreased thylakoid lumen acidification (42) (Fig. 5 D–F and SI Appendix, Fig. S12 C–E). These results were consistent with the lower rates of electron transport, reduced levels of NADPH, and obstructed ATP synthesis in Ospho1;2 mutants.

Because several photorespiratory metabolites were differentially accumulated in Ospho1;2 mutants (Fig. 4 E and F and SI Appendix, Fig. S9 C and F), we also assessed the photorespiration rates of WT, Ospho1;2, and OsPHO1;2-OE plants via low-oxygen suppression experiments. A_net was measured under ambient and low O₂ (21% and 2%, respectively), the latter of which suppresses photorespiration. Photorespiration rate (photosynthetic O₂ inhibition) was then evaluated based on the differences in values between the two states (43). We next calculated the photorespiratory O₂ inhibition ratio (photosynthetic O₂ inhibition/A_net at 2% O₂) considering the photorespiratory Rubisco oxidation proportion was different in Ospho1;2 mutant. The result indicated more abundant photorespiration in the mutants compared to the WT (SI Appendix, Table S2). Although photorespiration and photosynthesis are generally considered to compete for resources, previous studies have suggested that photorespiration may contribute to maintenance of normal physiological functions under Pi-deficient conditions (44, 45). Therefore, the increased photorespiration rates in Ospho1;2 mutants may have partially compensated for the Pi deficiency.

High OsPHO1;2 Expression Was Associated with High Levels of Photosynthesis in Different Rice Varieties.

After detailing the relationship between OsPHO1;2 function and leaf photosynthetic rates, we sought to identify correlations between leaf Pi content and photosynthesis in a broader range of rice cultivars. Twenty rice varieties were selected from the Mini-Core Collection of 217 worldwide accessions (46, 47) (SI Appendix, Table S3), 10 each with high or low OsPHO1;2 expression. During early grain filling (5 DAF), lines with high OsPHO1;2 expression tended to have higher Pi levels, photosynthetic rates in flag leaves, and higher grain yield; the opposite effects were seen in varieties with low expression of OsPHO1;2 (Fig. 6 A–C). Among these 20 varieties, flag leaf Pi concentration had strong positive correlations with A_net and with grain yield, and A_net also showed positive correlation with grain yield (Fig. 6 D–F). These results were also observed in the middle grain-filling stage (15 DAF) (Fig. 6 A–C and SI Appendix, Fig. S13 A–C). Our survey of multiple rice accessions thus suggested that varieties with high OsPHO1;2 expression were associated with higher photosynthetic rate and better yield potential than those with lower OsPHO1;2 expression.

Fig. 6.

High OsPHO1;2 expression and foliage Pi application improve leaf photosynthesis and yield potential in different rice varieties. (A–C), Measurements of flag leaf Pi concentration (A), net photosynthetic rates (A_net, B) at 5 DAF and 15 DAF, and grain yield (C) of 20 varieties from the Mini-Core Collection (with high or low OsPHO1;2 expression). The “High” or “Low” group each contains 10 varieties, n = 10 plants for each variety in (A and B) and n = 8 plants for each variety in (C). (D) Dot plot of the correlation between A_net and Pi contents among the 20 selected varieties at 5 DAF. (E) Dot plot of the correlation between grain yield and leaf Pi concentration among the 20 varieties. (F) Dot plot of the correlation between grain yield and A_net (5 DAF). The trend lines and R² values were indicated in (D–F). (G), Measurements of flag leaf Pi concentration among cultivars NIP, WYJ, and WS3 under different foliar Pi supplements at 5 DAF (n = 12 plants). (H) A_net of cultivars NIP, WYJ, and WS3 detected under different foliar Pi supplements (n = 20 plants). (I–K) The plant morphologies of NIP (I), WYJ (J), and WS3 (K) at mature stage. (Scale bars, 20 cm.) (L) The grain yield per plant of NIP, WYJ, and WS3 under different foliar Pi supplements (n = 36 plants). (M) The grain yield per plot of NIP, WYJ, and WS3 under different foliar Pi supplements (n = 5 plots). Each plot includes 48 plants from NIP, WYJ, and WS3. Different lower-case letters indicate significant differences (P < 0.05) assessed by one-way ANOVA followed by Tukey’s HSD tests.

Foliar Pi Supplementation Greatly Enhanced Rice Photosynthesis and Yield.

Due to the positive correlation between leaf Pi concentration and photosynthetic rate/grain yield, we hypothesized that an additional Pi supply to the leaves, especially at grain filling stage, could increase photosynthesis and grain yield. We therefore performed field experiments in which leaves were sprayed with Pi fertilizers. Appropriate concentrations of Pi fertilizer were first determined by spraying the cultivar “Nipponbare” (NIP) during the heading period; treatments consisted of 1% (w/v) calcium superphosphate (CaS), 2% (w/v) CaS, 1 mM potassium dihydrogen phosphate (KH₂PO₄), or 5 mM KH₂PO₄. Lower concentrations of Pi fertilizer were found to promote photosynthesis more effectively than the higher concentrations (SI Appendix, Fig. S14 A and B). We then tested the effects of Pi spraying on NIP and the local high-yield cultivars WYJ (a japonica variety) and WS3 (an indica variety) in Shanghai, China. During the early and middle grain-filling stages, flag leaf Pi contents and photosynthetic rates were significantly increased in Pi-treated plants of all varieties compared with the control treatment (Fig. 6 G and H and SI Appendix, Fig. S15 A–F). At the mature stage, growth morphologies were similar between treatment groups, while the Pi-supplemented plants showed heavier panicles (Fig. 6 I–K). Tiller numbers were similar between control and treated plants (SI Appendix, Fig. S15 G–I), whereas the 1,000-grain weight was markedly increased in plants treated with either CaS or KH₂PO₄ (SI Appendix, Fig. S15 J–L). Importantly, grain yield per plant and grain yield per plot were dramatically increased (by up to 13 to 21% and 11 to 15%, respectively) in the treatment groups of all three varieties (Fig. 6 L and M). Further, we performed another Pi fertilizer spraying field trail in Hainan, China, which confirmed the positive effects by either CaS or KH₂PO₄ treatment (SI Appendix, Fig. S16 A–H). Additionally, we also tested the effects of Pi spraying on NIL-OsPHO1;2 and NIL-Ospho1;2. Pi content and A_net were both increased in treated NIL-OsPHO1;2 and NIL-Ospho1;2 plants. Although spraying of CaS or KH₂PO₄ to the leaves of NIL-Ospho1;2 restored the Pi concentration and A_net to a level similar to that of the WT plants without Pi spraying, the grain weight and yield did not show corresponding improvements in NIL-Ospho1;2 (SI Appendix, Fig. S17 A–I). This should be because the high grain Pi accumulation in NIL-Ospho1;2 suppresses grain filling (23). Overall, the results demonstrated that supplying additional Pi to rice leaves during grain filling stage enhanced leaf photosynthetic rate and ultimately increased yield, which was inseparable from effective operation of OsPHO1;2.

Discussion

Optimization of photosynthesis is considered a key route for genetic improvement of yield in major cereal crops. To date, there have been few improvements to photosynthesis in modern crops, and the genetic yield potential remains to be further increased (48). Although many approaches have been used to attempt improvements in crop photosynthetic efficiency and yield (8, 48, 49), as well as the management of photosynthetic machinery upon stress conditions (50), mechanisms associated with Pi levels have been understudied. Internal P concentrations are closely associated with photosynthetic PUE (7, 51), and improvements in Pi-mediated photosynthesis could therefore be an effective approach to increase yield.

Here, we suggest that OsPHO1;2 functions opposite to SPDT, another key P distributor in leaves and grains (29). This was reflected by their distinct expression pattern in nodes, that is, SPDT highly expressed at diffuse vascular bundles (DVBs, connecting from node to upper node/panicle) of xylem (29), while OsPHO1;2 only expressed at enlarged vascular bundles (EVBs, connecting from lower node to leaf) of xylem (23, 30). Supportively, we observed that SPDT was up-regulated in Ospho1;2 mutant and OsPHO1;2 was up-regulated in spdt mutant (Fig. 1 L and M). Based on their opposite Pi allocation pattern, we inferred that in the Ospho1;2 mutant, deficient in OsPHO1;2 caused compromised allocation of Pi to leaves, while SPDT and other Pi transporters could be induced to transport more Pi to seeds to participate in grain filling, phytic acid synthesis, and other necessary processes. In contrast, in spdt mutants, deficient in SPDT disturbed P allocation into grains, so that the upregulation of OsPHO1;2 should be responsible to allocate more Pi into leaves, leading to increased leaf Pi concentration. Taken together, relative gene expression levels, Pi accumulation, and leaf photosynthetic capacities in Ospho1;2 and spdt mutants (Fig. 1 and SI Appendix, Fig. S5) further indicated that the combined functions of OsPHO1;2 and SPDT facilitated Pi homeostasis in rice plants during grain filling. Therefore, SPDT and OsPHO1;2 play, at least partially, compensatory roles in maintaining Pi homeostasis in whole plant, while the underlying mechanism between two transporters needs further investigation in the future. Given the strong efflux transport activity of OsPHO1;2 (23), our current knowledge suggests that on the one hand, OsPHO1;2 plays key roles in Pi reallocation and grain filling in developing grains (23); on the other hand, OsPHO1;2 also exerts functions in Pi allocation through nodes exporting to leaves to modulate Pi-limited photosynthesis. In this way, OsPHO1;2 coordinates source (leaf photosynthesis) and sink (grain filling) activities, thereby substantially improving grain yield in rice.

Photosynthetic CO₂ assimilation can be constrained by RuBP carboxylation, RuBP regeneration, and TPU (16). Here, leaf Pi deficiency in Ospho1;2 mutants suppressed not only RuBP carboxylation and regeneration but also TPU (Fig. 2 and SI Appendix, Table. S1). Ospho1;2 mutants showed decreased photosynthetic electron transport flow and a lower pmf across the thylakoid membrane in the light reactions, which indicated that less NADPH and ATP were synthesized for subsequent RuBP carboxylation and regeneration in the CBB cycle (52). These changes were associated with reduced photosynthetic protein expression and phosphorylation levels, and decreased accumulation of photosynthetic metabolites in Ospho1;2 mutants during grain filling. Furthermore, Ospho1;2 mutants exhibited a higher proportion of photorespiration activity than WT plants. These could be resulted from the inhibition of photosynthesis, and may also have been associated with P recycling (53), TP export, and other Pi-requiring processes. TPs, which are products of the CBB cycle, require translocation out of the chloroplast to the cytosol to be converted into carbohydrates (16); Pi is needed for TP export because the transporter (TPT) uses a counterexchange method (54). Thus, when leaves encounter Pi deficiency, TP-Pi translocation will be limited. If insufficient Pi is imported into the chloroplast, V_cmax, J_max, photophosphorylation, and phosphorylation of proteins associated with photosynthesis may all be repressed (55). Significantly, decreased chloroplast Pi concentration was observed in Ospho1;2 mutants during grain filling (Fig. 1), which meets well with the above assumptions. TPU is an important component of the photosynthetic regulatory network (16, 56). In WT plants, TPU limitation can temporarily occur under high-CO₂ and high-light conditions (56). However, TPU limitation appeared to be constant in Ospho1;2 mutants due to cytoplasmic Pi deficiency. We observed corresponding upregulation of PSR genes and decreases in photosynthetic protein expression and phosphorylation levels in Ospho1;2 mutants, which repressed light reactions and the CBB cycle (Fig. 3). This was likely responsible for the earlier occurrence of Pi limitation in Ospho1;2 mutants compared to the WT, and indicated that long-term Pi-limited photosynthesis was taking place. Overall, our current study uncovered that OsPHO1;2 functions in moderating phosphate-limitation of photosynthesis in leaves, providing genetic evidence for Pi-control of photosynthetic operation.

Based on this phenomenon of long-term Pi limitation, we here propose a working model for OsPHO1;2-mediated regulation of photosynthesis (Fig. 7). During grain filling, OsPHO1;2 is responsible for Pi allocation to the flag leaves through the nodes while SPDT mainly allocated P to the grains. Adequate Pi levels in the flag leaves facilitate activity of the photosynthetic light reactions and carbon assimilation. Loss of OsPHO1;2 function causes severe Pi deficiency in the leaves particularly in chloroplasts, which markedly represses photosynthetic protein expression and phosphorylation, balance of CBB cycle components, ATP and NADPH synthesis, and TP-Pi translocation, all of which induces accumulated effects of long-term Pi limitation of photosynthesis. In contrast, OsPHO1;2 overexpression enhances the Pi efflux activity thereby more allocation to leaves that increases the Pi availability in chloroplasts and thus delays the induction of Pi-limited photosynthesis and contributes to high photosynthetic rates over a long duration, which could promote accumulation of photosynthates and increase grain yield. This model was supported by data from OsPHO1;2 mutation and overexpression plants as well as various Mini-Core Collection cultivars. In recent years, a range of engineering strategies have been employed to increase photosynthetic efficiency and crop yield, including nuclear-encoded synthesis of the D1 subunit of photosystem II (57), increasing the relaxation rate of nonphotochemical quenching (NPQ) (58), and co-overexpressing Rubisco and Rubisco activase (59). Each of these approaches has helped to increase the efficiency of one aspect of photosynthesis. In the present study, we found that OsPHO1;2 was directly involved in multiple aspects of leaf photosynthesis, giving this gene more regulatory power.

Fig. 7.

Proposed model of OsPHO1;2 in regulating leaf photosynthesis in rice. During grain filling, OsPHO1;2 is responsible for Pi allocation to leaves through the node, and proper Pi accumulation level in flag leaves constantly facilitate photosynthetic light reactions and carbon assimilation. While loss-of-function of OsPHO1;2 causes severe Pi-deficiency in leaves, which markedly represses the phosphorylation of photosynthetic proteins, CBB cycle components, ATP and NADPH synthesis, as well as the TP-Pi translocation efficiency. Additively, mutation of OsPHO1;2 causes earlier occurrence of Pi-limitation of photosynthesis, whereas overexpression of OsPHO1;2 delays it, thus greatly contributing to longer duration of high photosynthetic rate and higher grain yield. The purple arrowheads indicate the transport function of OsPHO1;2, and the brown arrowheads indicate the transport function of SPDT in the model.

We hypothesized that the effects of Pi deficiency on the photosynthetic machinery could be alleviated by resupplying Pi to the leaf tissue (52), which was supported by the studies in barley and wheat that foliar phosphate application could improve grain yield (60, 61). We therefore tested foliar supplementation with Pi fertilizer at grain filling stage in rice, which turned out to have greatly improved leaf photosynthetic rates and rice yield (Fig. 6). Foliar spraying is expected to replace rhizosphere fertilization as a field management strategy in the future due to the weak Pi absorption capacity and PUE associated with rhizosphere fertilization (62). The reserves of rock phosphate that serve as Pi fertilizer sources are expected to be exhausted in the next 60 to 100 y (63), making it more valuable to adopt fertilization methods with high Pi absorption capacity and PUE. Furthermore, excessive Pi fertilizer use in agriculture leads to P accumulation in the soil, which is then released via leaching and water erosion, causing eutrophication in the environment (4). We should therefore aim to resolve Pi-deficiency-induced photosynthetic inefficiencies in crop plants with more sustainable and effective strategies. Our findings illustrate the positive effects of OsPHO1;2-mediated Pi homeostasis on photosynthesis, suggesting an effective genetic method of increasing crop yield and PUE. In summary, this study provides an efficient approach by which PUE can be improved through rice breeding, contributing to both food security and environmentally sustainable Pi use.

Materials and Methods

Plant Materials, qPCR, and Western Blot Analysis.

The details of plant materials and growth conditions are shown in SI Appendix. Total RNAs and proteins were extracted from rice flag leaf tissues during grain filling. The detailed experiment protocols are described in SI Appendix.

Measurement of Pi Concentration in Leaf Tissues and Chloroplasts.

The rice leaf tissues were sampled and used for Pi extraction according to our previous study (23). Chloroplasts were carefully isolated from mature leaves of different rice materials according to previous report with modifications. The detailed methods of Pi concentration measurement are listed in SI Appendix.

Gas-Exchange and Chlorophyll Fluorescence Measurements.

The steady-state photosynthetic rate (A_net) was measured on rice leaves at the vegetative and filling stages. The Li-COR 6400 instrument equipped with a 6 cm² cuvette and a 6400-02B red–blue light source (Li-Cor, Lincoln, NE) was used to measure the light-saturated (1,200 µmol m⁻² s⁻¹) net CO₂ assimilation rates (A_net) at 30 °C among all the rice lines. A_net was measured with the following additional settings: ambient relative humidity, 400 ppm reference CO₂, and a flow rate of 500 μmol s⁻¹. All steady-state photosynthesis measurements were taken on sunny days from 9 am to 11:30 am in the field.

For the A-C_i measurements, photorespiration rates (Rp) estimation, chlorophyll fluorescence, P700, and P515, the details were described in SI Appendix.

Proteomics, Phosphoproteomic, and Metabolomics analysis.

Flag leaf samples of rice plants were collected at 5 and 15 DAF during grain filling stage. The detailed methods of multiomics analysis are shown in SI Appendix.

Foliage P Fertilizers Spraying Treatment.

For Pi-supplement treatments, 1-mo-old seedlings were transplanted into normal field (with normal soil) in Shanghai or Hainan, China, followed by standard water and fertilizer management during the whole growth period. During flowering stage, commercial P fertilizers calcium superphosphate (CaS) and potassium dihydrogen phosphate (KH₂PO₄) were used for leaf spraying. The P fertilizer spraying treatments were conducted once at early flowering stage, and Pi content and A_net were analyzed during grain filling. The detailed treatments are shown in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

The RNA-seq datasets have been deposited in the NCBI Gene Expression Omnibus: GSE218345 (64). Description of data statistics as well as entire data availability are included in SI Appendix. All other data are included in the manuscript and/or supporting information.