Mutation of OsLPR3 Enhances Tolerance to Phosphate Starvation in Rice

Low Phosphate Root (LPR) encodes a protein localized to the endoplasmic reticulum (ER) and cell wall. This gene plays a key role in responding to phosphate (Pi) deprivation, especially in remodeling the root system architecture (RSA). An identification and expression analysis of the OsLPR family in rice (Oryza sativa) has been previously reported, and OsLPR5, functioning in Pi uptake and translocation, is required for the normal growth and development of rice. However, the role of OsLPR3, one of the five members of this family in rice, in response to Pi deficiency and/or in the regulation of plant growth and development is unknown. Therefore, in this study, the roles of OsLPR3 in these processes were investigated, and some functions were found to differ between OsLPR3 and OsLPR5. OsLPR3 was found to be induced in the leaf blades, leaf sheaths, and roots under Pi deprivation. OsLPR3 overexpression strongly inhibited the growth and development of the rice but did not affect the Pi homeostasis of the plant. However, oslpr3 mutants improved RSA and Pi utilization, and they exhibited a higher tolerance to low Pi stress in rice. The agronomic traits of the oslpr3 mutants, such as 1000-grain weight and seed length, were stimulated under Pi-sufficient conditions, indicating that OsLPR3 plays roles different from those of OsLPR5 during plant growth and development, as well as in the maintenance of the Pi status of rice.


Introduction
Phosphorus (P) is an essential macronutrient for plant growth and development. In addition to being a key constituent of molecules, such as nucleic acids, ATP, and membrane phospholipids, P plays crucial roles in signal transduction, energy transfer, photosynthesis, and respiration [1,2]. Inorganic orthophosphate (Pi), the major form of P absorbed and assimilated by plants, is the least accessible macronutrient in many natural and agricultural ecosystems. Its inaccessibility is mainly due to its poor mobility in soils and/or its propensity to form sparingly soluble salts with oxides or hydroxides of aluminum and iron in acidic soils and with calcium and magnesium in alkaline soils [2][3][4]. Its low availability in soil often limits plant growth and productivity in soils across climatic zones [5].
Plants have evolved several strategies that adapt their physiology to Pi concentration variations. Molecular responses to low Pi availability have been divided into two partially independent signaling pathways known as the local response and the systemic response to Pi starvation [6,7]. Systemic responses depend on the internal P status of the plant and concentration, and it promoted the growth and development of rice during Pi deficiency. The agronomic traits of oslpr3 mutants, such as 1000-grain weight and seed length, were stimulated under Pi-sufficient conditions, indicating the key roles of OsLPR3 in the growth and development, as well as in the maintenance of Pi homeostasis, of rice.

OsLPR3 Was Responsive to Pi Deprivation
To investigate the expression patterns of OsLPR3 at various nutrient deficiencies, Nipponbare seedlings were initially grown in an IRRI solution for 10 days (10-d-old seedlings). The seedlings were then transferred to a complete nutrient solution (CK) or to a nutrient solution lacking either nitrogen (−N), phosphate (−P), potassium (−K), magnesium (−Mg), or iron (−Fe), and they were grown for 7 d. The relative transcript levels of OsLPR3 in the leaf blades, leaf sheaths, and root samples of the treated seedlings were assessed via a quantitative polymerase chain reaction (qPCR). In CK, the relative transcript levels of OsLPR3 were high in the roots and low in the leaf blades and leaf sheaths. Moreover, most nutrient deficiencies, such as −N, −K, −Mg, and −Fe, did not exert a significant impact on the transcript levels of OsLPR3 compared with CK ( Figure 1). The relative transcript levels of OsLPR3 were induced by Pi deficiency in all three plant tissues. In the leaf blades, the relative transcript levels of OsLPR3 were 150-fold higher in −P than in CK ( Figure 1A); however, in the roots, the relative transcript levels of OsLPR3 were 13-fold higher in −P than in CK ( Figure 1C). Thus, after the induction by −P, the relative transcript levels of OsLPR3 were only three times more abundant in the roots than in the leaf blades. The relative expression of OsLPR3 remained much lower in the leaf sheaths than in the roots and leaf blades after the Pi deficiency induction. Values are means ± SE (n = 3). Different letters above the bars indicate significant differences in the relative OsLPR3 transcript levels (p < 0.05, oneway ANOVA).

OsLPR3 was Localized to the ER
The ER is the entry point for membrane and secretory proteins into the endomembrane system in all eukaryotic cells [31]. It is the site at which these proteins are folded in Values are means ± SE (n = 3). Different letters above the bars indicate significant differences in the relative OsLPR3 transcript levels (p < 0.05, one-way ANOVA).

OsLPR3 Was Localized to the ER
The ER is the entry point for membrane and secretory proteins into the endomembrane system in all eukaryotic cells [31]. It is the site at which these proteins are folded in order to acquire biological functions at their final destination [32]. The protein localization program Softberry (http://linux1.softberry.com/, accessed on 3 September 2022) suggested the putative ER localization of the OsLPR3 protein. For the subcellular localization of the putative ER-targeted proteins, a red fluorescent protein (mCherry) is a potent in vivo marker applicable in diverse plant species [33]. To provide experimental evidence for the ER localization of OsLPR3, a reporter gene encoding a green fluorescent protein (EGFP) driven by a CaMV35S promoter was fused upstream and in frame with the OsLPR3 open-reading frame (35S::eGFP::OsLPR3). This construct was transiently co-expressed with 35S::mCherry::HDEL, which exhibits characteristic red fluorescence in the ER [34] in the leaf epidermal cells of Nicotiana benthamiana. Confocal microscopy was used to demonstrate the co-localization of EGFP::OsLPR3 and mCherry::HDEL to the ER ( Figure 2). The co-localization of the two proteins was particularly apparent in the cytoplasmic strands extending across the vacuole (white arrows). The ER localization of OsLPR3 was consistent with the localization of AtLPR1 in Arabidopsis [24].

OsLPR3 was Involved in Vegetative Growth and RSA of Rice
The functions of OsLPR3 were explored using transgenic plants in which OsLPR3 was overexpressed (Ox) and knocked out. The Ox lines with the strongest upregulated expressions of OsLPR3 (designated Ox5, Ox10, Ox29, and Ox31 (Figure S1A)) were selected for further analyses. Southern blot analyses demonstrated that the Ox transgenic lines all arose from independent transformation events ( Figure S1B). The knockout mutants of OsLPR3 (designated oaslpr3-1, oslpr3-2, and oslpr3-3) were confirmed after sequencing the lesion site ( Figure S1C).
The effects of OsLPR3-Ox and the knockout mutants on the vegetative growth of the 10-d-old rice seedlings grown hydroponically for an additional 21 d under +P and −P conditions were determined. The plant height of the Ox plants was shorter than that of the WT and oslpr3 lines under both +P and −P conditions. The plant height of the oslpr3 lines did not differ from that of WT under +P conditions ( Figure 3C). However, under −P conditions, the plant height of the oslpr3 plants was higher than that of WT ( Figure 3D). The root length of both the Ox and knockout lines did not differ from that of WT under +P conditions ( Figure 3E). However, both the Ox and knockout lines had similarly long roots under −P conditions compared with those of WT ( Figure 3F). The effects of the OsLPR3-Ox and knockout lines on the biomass and root-shoot ratio were also determined. The shoot and root biomasses of the OsLPR3-Ox plants were repressed under both +P and −P conditions, but no significant differences were observed between their root-shoot ratio and that of WT ( Figure S2). The shoot biomass, root biomass, and root-shoot ratio of the oslpr3 lines were not significantly different from those of WT under +P conditions ( Figure  S2A,C,E). However, under −P conditions, both the shoot biomass and root biomass were increased in the oslpr3 plants, and they were ~54% and ~24% higher in the transgenic plants than in the WT plants, respectively. In addition, the root-shoot ratio was ~22% lower in the oslpr3 mutants than in WT ( Figure S2B,D,F).

OsLPR3 Was Involved in Vegetative Growth and RSA of Rice
The functions of OsLPR3 were explored using transgenic plants in which OsLPR3 was overexpressed (Ox) and knocked out. The Ox lines with the strongest upregulated expressions of OsLPR3 (designated Ox5, Ox10, Ox29, and Ox31 (Figure S1A)) were selected for further analyses. Southern blot analyses demonstrated that the Ox transgenic lines all arose from independent transformation events ( Figure S1B). The knockout mutants of OsLPR3 (designated oaslpr3-1, oslpr3-2, and oslpr3-3) were confirmed after sequencing the lesion site ( Figure S1C).
The effects of OsLPR3-Ox and the knockout mutants on the vegetative growth of the 10-d-old rice seedlings grown hydroponically for an additional 21 d under +P and −P conditions were determined. The plant height of the Ox plants was shorter than that of the WT and oslpr3 lines under both +P and −P conditions. The plant height of the oslpr3 lines did not differ from that of WT under +P conditions ( Figure 3C). However, under −P conditions, the plant height of the oslpr3 plants was higher than that of WT ( Figure 3D). The root length of both the Ox and knockout lines did not differ from that of WT under +P conditions ( Figure 3E). However, both the Ox and knockout lines had similarly long roots under −P conditions compared with those of WT ( Figure 3F). The effects of the OsLPR3-Ox and knockout lines on the biomass and root-shoot ratio were also determined. The shoot and root biomasses of the OsLPR3-Ox plants were repressed under both +P and −P conditions, but no significant differences were observed between their root-shoot ratio and that of WT ( Figure S2). The shoot biomass, root biomass, and root-shoot ratio of the oslpr3 lines were not significantly different from those of WT under +P conditions ( Figure S2A,C,E). However, under −P conditions, both the shoot biomass and root biomass were increased in the oslpr3 plants, and they were~54% and~24% higher in the transgenic plants than in the WT plants, respectively. In addition, the root-shoot ratio was~22% lower in the oslpr3 mutants than in WT ( Figure  Furthermore, the OsLPR3-Ox lines seemed to have a spindly RSA, whereas the oslpr3 lines had an RSA thicker than that of WT. These observations were examined further by determining the number of adventitious roots in each mutant line. Compared with WT, the number of adventitious roots in the OsLPR3-Ox lines was lower under both +P and −P conditions, whereas the number of adventitious roots in the oslpr3 lines was not significantly different under +P conditions, but it was higher under −P conditions ( Figure S3). Moreover, the lateral root densities in the WT and transgenic lines were not significantly different under either Pi-sufficient or Pi-deficient conditions ( Figure S4). Furthermore, the OsLPR3-Ox lines seemed to have a spindly RSA, whereas the oslpr3 lines had an RSA thicker than that of WT. These observations were examined further by determining the number of adventitious roots in each mutant line. Compared with WT, the number of adventitious roots in the OsLPR3-Ox lines was lower under both +P and −P conditions, whereas the number of adventitious roots in the oslpr3 lines was not significantly different under +P conditions, but it was higher under −P conditions ( Figure S3). Moreover, the lateral root densities in the WT and transgenic lines were not significantly different under either Pi-sufficient or Pi-deficient conditions ( Figure S4).

Alteration of OsLPR3 Expression Affected the Agronomic Traits of Rice during the Reproductive Stage
The growth inhibition of the OsLPR3-Ox lines observed at the seedling stage ( Figure 3) was also clear in the 20-week-old mature plants (Figure 4). Both plant height and yield were significantly reduced by OsLPR3 overexpression (Figure 4A,B). The plant height was also lower in the 20-week-old oslpr3 plants, although the yield was not significantly affected ( Figure 4C). We measured the total tiller number and effective tiller number of the WT and OsLPR3 transgenic lines. The alteration of OsLPR3 expression did not affect the total tiller number and effective tiller number of the rice. However, the knockout of OsLPR3 decreased the seed setting rate of the rice ( Figure S5). OsLPR3 overexpression decreased the 1000-grain weight by 10%, whereas the knockout of OsLPR3 increased the 1000-grain weight by 5-11% ( Figure 4D). Furthermore, the seed width was reduced in the OsLPR3-Ox lines but was unchanged in the oslpr3 knockout lines compared with that of WT ( Figure 5A). In contrast, the seed length was significantly greater in the oslpr3 mutants but was unchanged in the OsLPR3-Ox lines compared with that of WT ( Figure 5B). These results suggest that OsLPR3 affects the agronomic traits of rice by regulating seed size. were significantly reduced by OsLPR3 overexpression (Figure 4A,B). The plant height w also lower in the 20-week-old oslpr3 plants, although the yield was not significantly fected ( Figure 4C). We measured the total tiller number and effective tiller number of WT and OsLPR3 transgenic lines. The alteration of OsLPR3 expression did not affect total tiller number and effective tiller number of the rice. However, the knockout OsLPR3 decreased the seed setting rate of the rice ( Figure S5). OsLPR3 overexpress decreased the 1000-grain weight by 10%, whereas the knockout of OsLPR3 increased 1000-grain weight by 5-11% ( Figure 4D). Furthermore, the seed width was reduced in OsLPR3-Ox lines but was unchanged in the oslpr3 knockout lines compared with that WT ( Figure 5A). In contrast, the seed length was significantly greater in the oslpr3 muta but was unchanged in the OsLPR3-Ox lines compared with that of WT ( Figure 5B). Th results suggest that OsLPR3 affects the agronomic traits of rice by regulating seed size  Values are means ± SE (n = 10). Different letters above the bars indicate significant differences (p < 0.05, one-way ANOVA).

Alteration of OsLPR3 Expression Affected the Pi Status
The effects of the OsLPR3-Ox and knockout mutants on Pi uptake and utilization were determined in the 10-d-old seedlings of the WT and transgenic lines that were grown hydroponically under +P conditions (200 µM Pi) and −P (5 µM Pi) conditions for an additional 21 d. The Pi concentration in the OsLPR3-Ox lines was not significantly different from that in WT in the roots, leaf blades, and leaf sheaths under either +P or −P conditions ( Figure 6). However, the Pi concentration of the OsLPR3 knockout mutant was significantly higher in all tissues than in WT under +P conditions, including in the roots, leaf blades, and leaf sheaths ( Figure 6A), and the Pi concentration of the OsLPR3 knockout mutant was significantly higher in the leaf blades and leaf sheaths than in WT under −P conditions ( Figure 6B). Whereafter, we measured the total P concentration in the OsLPR3 transgene lines and WT. The total P concentrations in the roots, leaf blades, and leaf sheaths of the oslpr3 mutant were higher in the oslpr3 knockout lines than in WT ( Figure 7A), but they were unchanged under −P conditions ( Figure 7B). Furthermore, radioisotope labeling was used to evaluate the effect of OsLPR3 on the phosphate absorption rate. The uptake of 32 Pi was significantly higher in the oslpr3 knockout lines than in the WT and OsLPR3-Ox lines under +P conditions but not under −P conditions ( Figure 7C,D). The above results indicate that the knockout of OsLPR3 improved the phosphate utilization capacity of the rice.

Alteration of OsLPR3 Expression Affected the Relative Expressions of OsPTs and Pi-Starvation-Induced Genes
To determine the function of OsLPR3 in Pi homeostasis, the expressions of the relative genes that maintained Pi homeostasis were explored in the OsLPR3 knockout lines and WT. The expression levels of the Pht1 family members were assessed in the roots of the WT and oslpr3 lines under +P conditions. The transcript levels of OsPT2, OsPT4, OsPT8, and OsPT10 were significantly higher in the oslpr3 mutants than in WT, whereas the transcript levels of OsPT1, OsPT6, and OsPT9 were not significantly different between the oslpr3 mutants and WT ( Figure S6). Further, the relative expression levels of Pi-starvation-inducible genes, including OsIPS1, OsPAP10a, and OsSQD2, were determined in the roots of the WT and oslpr3 lines under +P and −P conditions. The relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were not significantly different from those in WT under +P conditions. Under −P conditions, the relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were severely repressed compared with those in WT. The relative expressions of OsIPS1, OsPAP10a, and OsSQD2 were highly induced in response to Pi deficiency in WT, which is consistent with a previous study [35]. Notably, the relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were only a little higher under −P conditions than under +P conditions, but the relative expression levels of these genes were far higher in WT under −P conditions than under +P conditions (Figure 8).

Alteration of OsLPR3 Expression Affected the Pi Status
The effects of the OsLPR3-Ox and knockout mutants on Pi uptake and utilization were determined in the 10-d-old seedlings of the WT and transgenic lines that were grown sheaths of the oslpr3 mutant were higher in the oslpr3 knockout lines than in WT ( Figure  7A), but they were unchanged under −P conditions ( Figure 7B). Furthermore, radioisotope labeling was used to evaluate the effect of OsLPR3 on the phosphate absorption rate. The uptake of 32 Pi was significantly higher in the oslpr3 knockout lines than in the WT and OsLPR3-Ox lines under +P conditions but not under −P conditions ( Figure 7C,D). The above results indicate that the knockout of OsLPR3 improved the phosphate utilization capacity of the rice. Values are means ± SE (n = 4). Different letters above the bars indicate significant differences (P < 0.05, one-way ANOVA).

Alteration of OsLPR3 Expression Affected the Relative Expressions of OsPTs and Pi-Starvation-Induced Genes
To determine the function of OsLPR3 in Pi homeostasis, the expressions of the relative genes that maintained Pi homeostasis were explored in the OsLPR3 knockout lines and WT. The expression levels of the Pht1 family members were assessed in the roots of the WT and oslpr3 lines under +P conditions. The transcript levels of OsPT2, OsPT4, OsPT8, and OsPT10 were significantly higher in the oslpr3 mutants than in WT, whereas the transcript levels of OsPT1, OsPT6, and OsPT9 were not significantly different between the oslpr3 mutants and WT ( Figure S6). Further, the relative expression levels of Pi-starvation-inducible genes, including OsIPS1, OsPAP10a, and OsSQD2, were determined in the roots of the WT and oslpr3 lines under +P and −P conditions. The relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were not significantly different from those in WT under +P conditions. Under −P conditions, the relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were severely repressed compared with those in WT. The relative expressions of OsIPS1, OsPAP10a, and OsSQD2 were highly induced in response to Pi deficiency in WT, which is consistent with a previous study [35]. Notably, the relative expression levels of OsIPS1, OsPAP10a, and OsSQD2 in the oslpr3 mutants were only a little higher under −P conditions than under +P conditions, but the relative expression levels of these genes were far higher in WT under −P conditions than under +P conditions ( Figure 8).

Discussion
As a model plant for monocotyledon species, rice has a larger genome, a longer growth cycle, and a higher biomass than A. thaliana, and the mechanism of the growth regulation of rice is more complex and precise than that of A. thaliana. In Arabidopsis, a low Pi availability relieves the negative regulation of AtLPR1 by AtPRD2 in the ER, and it induces AtLPR1 transport to the plasma membrane. The AtLPR1 activity in the plasma membrane triggers Fe and callose deposition in the apoplast, which causes impaired movement of the SHORT ROOT (AtSHR) protein, interferes with symplastic communication, and is responsible for root meristem differentiation [18,19,24]. In a previous study, OsLPR3 (Loc_Os01g03630), a rice homolog of AtLPR1, was highly induced at the mRNA level by 21 d of Pi starvation, and it rapidly returned to basal transcript levels after 1 h of Pi resupply to the roots [36]. In this study, the strong induction of the relative expression of OsLPR3 mRNA to low Pi in the leaf blades, leaf sheaths, and roots ( Figure 1) is consistent with an earlier study [30]. However, in previous studies, the relative expression of AtLPR1 was induced by low Pi at the protein level rather than at the mRNA level [13,19]. AtLPR2 has been found to be abundant in roots and show an increased expression under NH 4 + conditions compared to under NO 3 − conditions [27]. This suggests a different function of LPR in dicotyledon Arabidopsis and monocotyledon rice.
Proteins require signal sequences in order to be secreted to the exterior of the cell. Guided by a signal peptide, proteins enter the cytosol through the ER and are finally secreted outside the cell [37]. AtLPR1 mainly functions as a ferroxidase in the cell wall. To be transported to the exterior of the cell, a protein requires a signal peptide to first deliver it to the ER for translocation through the endomembrane system to the plasma membrane. OsLPR5 and AtLPR1/2 have a putative ER-specific signal peptide, and they are localized to the ER and cell wall [24,27,38], and LPR1/2 are required to oxidize Fe(II) and maintain Fe(III)-citrate stability and mobility during xylem translocation against photoreduction [39]. However, OsLPR3 did not appear to have an ER-specific signal peptide sequence and was different from OsLPR5 and AtLPR1/2 ( Figure S7). The purified pCold-OsLPR3 fusion protein was heterologously expressed, and the pCold protein and pGS-OsLPR5 were used as negative and positive controls, respectively. The ferroxidase activity of the OsLPR3 fusion protein was comparably lower than that of the pCold negative control and significantly lower than that of the OsLPR5 fusion protein ( Figure S8A-C). Furthermore, OsLPR3 overexpression showed~20% increased ferroxidase activity under +P conditions but showed~35% decreased ferroxidase activity under −P conditions ( Figure S8D,E). However, in a previous study, OsLPR5 overexpression showed more than a 430% increase in the ferroxidase activity under both +P and −P conditions [38]. Hence, OsLPR3 might not function as a ferroxidase and, thus, differs from OsLPR5.
Different members of a gene family may be involved in functional conservation or redundancy [40][41][42]. In a previous study, the expressions of OsLPR3 and OsLPR5 were significantly induced and repressed, respectively, in ospho2 mutants [30]. In OsPDR2 RNAi lines, the expressions of OsLPR3 and OsLPR5 were decreased and increased, respectively, in −P split roots compared with +P split roots [43]. These results revealed that OsLPR3 and OsLPR5 might function differently in mediating Pi deficiency response networks under −Pi conditions. Subsequent research showed that the OsLPR5 mutant increased Pi uptake but repressed root growth and affected the translocation of Pi from the root to the shoot under −P conditions at the vegetative stage. Furthermore, the knockout of OsLPR5 impacted the plant height, seed setting rate, 1000-grain weight, and grain yield per plant at the reproductive stage [38]. However, in this study, the OsLPR3 mutation increased the 32 P uptake and total P concentration of rice under +P conditions but had no significant effect on the 32 P uptake and total P of rice under −P conditions (Figure 7), and the mutation of OsLPR3 increased the Pi concentration of rice under both +P and −P conditions at the vegetative stage ( Figure 6). In addition, the mutation of OsLPR3 increased the 1000-grain weight and seed length of the rice at the reproductive stage ( Figures 4D and 5B). The alteration of OsLPR3 expression did not affect the total tiller number and effective tiller number of the rice. However, the knockout of OsLPR3 decreased the seed setting rate of the rice ( Figure S5), and this could be the reason why the oslpr3 mutant had a higher 1000-seed weight compared to the wild type but a similar yield. Furthermore, we conducted a correlation analysis between the expression results and the seed production of the OsLPR3 transgene lines. The correlation coefficient was −0.5105, indicating a strong negative correlation. These results highlight the different roles of the LPR family members OsLPR3 and OsLPR5 in the growth, development, and P utilization of rice, indicating their different functions in response to Pi deficiency. This can provide some ideas for breeding nutrient-efficient and high-yield varieties.
A low Pi availability restricts plant growth and development. To adapt to low or limited Pi availability, plants usually grow longer roots and increase their root-shoot ratio by increasing their root growth more than their shoot growth [44,45]. Here, the number of rice tillers, an important indicator of Pi nutrient status, was positively correlated with tolerance to low Pi stress [46][47][48]. RSA is sensitive to the Pi status of a growth medium [49]. Improved root growth and RSA are thus important traits to increase P-acquisition efficiency [50,51]. In this study, the phenotypes of the oslpr3 knockout mutants were not significantly different from those of WT under +P conditions. However, under −P conditions, the growth of the oslpr3 knockout mutants was significantly promoted. The WT plants did not grow young tillers due to a low Pi, whereas the oslpr3 mutants grew young tillers under −P conditions ( Figure 3B). In addition, the plant height, root length, shoot biomass, root biomass, and number of adventitious roots of the oslpr3 mutants were increased compared with those of WT under −P conditions. Furthermore, the lower root-shoot ratio of the oslpr3 knockout mutants under −P conditions indicated that they were less sensitive to a low Pi than the WT plants ( Figure S2).
Furthermore, the oslpr3 mutation increased the total P concentration of the rice under +P conditions, and the 32 P-isotope-labeling assay further confirmed that the Pi uptake rate of the oslpr3 mutants was higher than that of WT, which might be because the OsLPR3 knockout stimulated the expressions of OsPT2, OsPT4, OsPT8, and OsPT10 in the rice ( Figure S6). Under −P conditions, the total P concentration and Pi uptake rate of the oslpr3 mutants were not significantly different from those of WT. However, the oslpr3 mutants showed a high tolerance to low Pi stress. The increased soluble Pi content of the oslpr3 mutant lines might have led to the high tolerance to low Pi stress in the oslpr3 mutants.
Several genes that play pivotal roles in maintaining Pi homeostasis have been identified in rice [35]. OsIPS1 has been found to be rapidly induced in the Pi-deprived roots of rice [52], OsPAP10a encodes an acid phosphatase [53] and OsSQD2 is involved in sulfolipid biosynthesis activated by Pi starvation in rice [53,54]. The expression levels of the Pi-starvation-inducible genes in the oslpr3 mutants were analyzed, and the expressions of all three genes were induced in WT under −P conditions compared with +P conditions. However, the expression levels of OsIPS1, OsPAP10a, and OsSQD2 were only slightly higher but not significantly different in the oslpr3 mutants under −P conditions compared with +P conditions, implying that the oslpr3 mutants were less sensitive to low Pi, and the mutation of OsLPR3 increased the tolerance to low Pi stress in the rice. The mutations of LPR1/2 result in Fe and callose deposition in the apoplast, subsequently impacting the movement of SHR and interfering with symplastic communication, which is responsible for root meristem differentiation [18,19]. There is a possibility that the alteration of OsLPR3 expression may influence the expression of its homologous genes, thus affecting Fe and callose deposition in the apoplast. The relative expressions of OsPTs and Pi-starvation-induced genes may also be affected by this process.

Plant Materials and Growth Conditions
Rice (Oryza sativa L. ssp. japonica) plants, namely, WT (Nipponbare), OsLPR3 overexpression lines (Ox5, Ox10, Ox29, and Ox31), and oslpr3 mutant lines (oslpr3-1, oslpr3-2, and oslpr3-3), with a Nipponbare background, were grown hydroponically in a temperaturecontrolled growth room (14 h light (30 • C)/10 h dark (22 • C) photoperiod, with a relative humidity of ∼70%). Seed germination and seedling growth conditions were as those previously described [30,55] 4 , and the pH was adjusted to 5.3. The nutrient medium was replaced every 3 d. The soil for the pot experiments was obtained from an experimental farm at Nanjing Agricultural University. Each pot was filled with 15 kg of air-dried soil supplemented with 40 mg Pi kg −1 soil. Standard cultural practices recommended for rice were followed as previously described [56].

qRT-PCR
The total RNA was isolated from the rice samples using the TRIzol reagent (Invitrogen), and it was treated with RNase-free DNase. First-strand cDNA was reverse-transcribed from~1 µg total RNA using the oligo (dT) 18 primer (Superscript II TM Reverse Transcriptase, Invitrogen). OsActin1 (LOC_Os03g50885) was used as an internal control for the qPCR. Each qPCR assay was performed in triplicate using an SYBR green-based master mix (Vazyme) on a StepOnePlus™ real-time PCR system (Applied Biosystems). The relative expression levels of the genes were computed by using the 2 −∆CT method. The gene-specific primers used in this study are listed in Table S1.

Transient Expression of OsLPR3 in N. benthamiana Leaves for Subcellular Localization
Agrobacterium-mediated transformation was used for the transient co-expression of 35S::eGFP::OsLPR3 and 35S::mCherry::HDEL in the epidermal leaf cells of N. benthamiana as previously described [57]. The N. benthamiana leaves were collected 2-3 d after infiltration. Using a diode laser for excitation, EGFP and mCherry fluorescence was visualized at 488 nm and 561 nm, respectively, under a confocal laser scanning microscope (Leica SP8).

Construction of OsLPR3 Overexpression and Mutation Vectors and Generation of Transgenic Plants
For overexpression, the coding sequence (1.605 kb) of OsLPR3 was amplified from the cDNA isolated from the WT Nipponbare plants using OsLPR3-specific primers. The PCR product was then inserted into pCAMBIA1305, which was digested with KpnI and BamHI using a ClonExpress II One Step Cloning Kit (C112-01, Vazyme). For the CRISPR/Cas9mediated mutation in OsLPR3, two gene-specific spacers residing in the exons were selected from the rice-gene-specific spacers library provided by Miao et al., 2013 [58]. The intermediate vector pOs-sgRNA was digested by BsaI (#R0535L, NEB), and then it was ligated to the spacers by using T4 DNA ligase (C301-01, Vazyme). The recombinated vector was then introduced to the final expression vector pH-Ubi-cas9-7 (GATEWAY recombination system, Invitrogen, USA). The constructs were transformed into Agrobacterium tumefaciens EHA105 and then into the mature embryos of WT Nipponbare, as described by Upadhyaya et al., 2000 [59].

Southern Blot Analyses
Genomic DNA (~100 µg) was extracted from the leaves of the WT (Nipponbare) and independent overexpression lines. It was then digested with EcoRI and BamHI overnight at 37 • C, separated on a 0.8% (w/v) agarose gel, and transferred to a Hybond-N + nylon membrane (Amersham). Hybridization was performed with a digoxigenin-labeled hygromycin-resistant gene as the probe at 65 • C overnight. The blots were washed under a stringent condition at 65 • C and analyzed using a phosphorimager (Typhoon-8600) [53].

Quantification of Pi and Total P Concentrations
Pi and total P concentrations were quantified as described by Zhou et al. (2008).

32 Pi Uptake Assay
The seedlings (10-d-old) of the WT, OsLPR3-Ox, and oslpr3 lines were grown hydroponically under +P or −P conditions for 7 d. Subsequently, these seedlings were grown for 24 h in a +P and/or −P uptake solution (200 mL) indicated by 8 µCi of 32 Pi (KH 2 PO 4 , Perkin-Elmer). After the uptake, apoplastic 32 Pi was removed by incubating the roots of the seedlings in an ice-cold desorption solution (2 mM MES (pH 5.5), 0.5 mM CaCl 2 , 0.1 mM NaH 2 PO 4 ) for 10 min. The seedlings were blotted dry, their roots and shoots were harvested separately, and their fresh weights were determined. The plant parts were digested in a mixture containing HClO 4 and 30% (v/v) H 2 O 2 at 28 • C for 8-12 h. A scintillation cocktail (3 mL) was added to the digested tissue, and 32 Pi activity was determined using a liquid scintillation counter (Tri-Carb 2100, Packard). The uptake of 32 Pi was calculated by summing the 32 Pi activity in the root plus shoot and dividing it by the fresh weight of the root for each plant separately.