Upstream Open Reading Frame and Phosphate-Regulated Expression of Rice OsNLA1 Controls Phosphate Transport and Reproduction^,

An upstream open reading frame participates in phosphate-responsive gene expression to control phosphate acquisition, phosphate translocation, and reproductive success in rice.

Abstract

Rice (Oryza sativa) OsNLA1 has been proposed to play a crucial role in regulating phosphate (Pi) acquisition in roots, similar to that of Arabidopsis (Arabidopsis thaliana) AtNLA. However, unlike AtNLA, OsNLA1 is not a target of miR827, a Pi starvation-induced microRNA. It is, therefore, of interest to know whether the expression of OsNLA1 depends on Pi supply and how it is regulated. In this study, we provide evidence that OsNLA1 controls Pi acquisition by directing the degradation of several OsPHT1 Pi transporters (i.e. OsPT1/2/4/7/8/12). We further show that OsNLA1 has an additional function in reproduction and uncover the mechanism of its expression regulation. Analysis of mRNA levels, promoter-GUS activity, and protoplast transient expression showed that the expression of OsNLA1.1, the most abundant transcript variant, is up-regulated in response to increasing Pi supply. The OsNLA1 promoter region was found to contain an upstream open reading frame that is required for Pi-responsive expression regulation. OsNLA1 promoter activity was observed in roots, ligules, leaves, sheaths, pollen grains, and surrounding the vascular tissues of anthers, suggesting that OsNLA1 is important throughout the development of rice. Disruption of OsNLA1 resulted in increased Pi uptake from roots as well as impaired pollen development and reduced grain production. In summary, our study reveals that Pi-induced OsNLA1 expression regulated by a unique mechanism functions in Pi acquisition, Pi translocation, and reproductive success.

Phosphorus (P) is an essential element for maintaining life. It serves various basic cellular functions in bioenergetics as a component in signal transduction cascades and as a structural constituent in nucleic acids and phospholipids. Whereas P is abundant, an adequate supply for plant productivity is usually limited because of the low mobility of phosphate (Pi) in the soil, a major form of P acquired by plant roots. Pi is acquired via Pi transporters at plasma membranes, whose function and expression are tightly regulated to coordinate internal cellular activities with external Pi supply. Under Pi starvation, the transcription levels of high-affinity Pi transporters belonging to the PHOSPHATE TRANSPORTER1 (PHT1) family at the root periphery are induced in order to extract Pi from the root-soil interface (Rausch and Bucher, 2002; Smith et al., 2003). Besides initial uptake of Pi from the rhizosphere, the PHT1 Pi transporters also mediate subsequent Pi allocation inside plants (Shin et al., 2004; Misson et al., 2005; Chiou and Lin, 2011), indicating the importance of these transporters in maintaining whole-plant Pi homeostasis via modulating Pi acquisition at the roots and Pi distribution among tissues/organs.

In rice (Oryza sativa), there are 13 OsPHT1 family members (OsPT1–OsPT13; Paszkowski et al., 2002). Eight of these showed high expression (fragments per kilobase of transcript per million mapped reads > 25) after Pi starvation, among which OsPT3 and OsPT10 exhibited the greatest induction (Secco et al., 2013). The activity and function of many of the OsPTs have been characterized. For example, OsPT1 was reported to be a high-affinity Pi transporter and OsPT1 loss of function reduced shoot Pi concentrations (Sun et al., 2012). OsPT2 is a low-affinity Pi transporter that facilitates the transport of Pi from Pi-starved roots to shoots, and OsPT6 and OsPT8 are high-affinity Pi transporters responsible for Pi uptake and translocation (Ai et al., 2009; Jia et al., 2011). OsPT4 loss of function reduced Pi uptake and also affected grain yield and embryo development (Zhang et al., 2015). OsPT9 and OsPT10, two other high-affinity Pi transporters, also play roles in Pi uptake (Wang et al., 2014). With overexpression of OsPT2, OsPT8, OsPT9, or OsPT10, transgenic rice showed reduced biomass as a result of Pi overaccumulation (Ai et al., 2009; Jia et al., 2011; Wang et al., 2014), highlighting the need for a proper control of the expression and activity of OsPHT1 Pi transporters.

It has been shown that the activity of PHT1 is regulated at the transcriptional and posttranslational levels. Under Pi deficiency, PHT1 is transcriptionally activated by PHOSPHATE STARVATION RESPONSE (PHR) transcription factors (Rubio et al., 2001; Nilsson et al., 2007; Zhou et al., 2008; Bustos et al., 2010; Liu et al., 2010). When Pi is adequate, PHR transcription factors are sequestered by SPX (SYG1/Pho81/XPR1) to prevent transcriptional activation of PHT1s (Lv et al., 2014; Puga et al., 2014). The exit of newly synthesized nonphosphorylated PHT1 proteins from the endoplasmic reticulum is facilitated by PHOSPHATE TRANSPORTER TRAFFIC FACILITATOR1 (PHF1) and trafficking to the plasma membrane in a COPII-dependent manner (González et al., 2005; Bayle et al., 2011; Chen et al., 2011, 2015; Kant et al., 2011; Wu et al., 2013). In rice, the interaction between OsPHT1 and OsPHF1 is weakened by CK2α3β3 holoenzyme-dependent phosphorylation of OsPHT1 to prevent its trafficking to plasma membranes (Chen et al., 2015). In addition, the protein abundance of AtPHT1 is modulated by ubiquitin-mediated degradation pathways, an important regulatory mechanism to avoid Pi toxicity, especially under Pi-replete conditions (Huang et al., 2013; Lin et al., 2013; Park et al., 2014).

AtPHO2 encodes a ubiquitin E2 conjugase (Aung et al., 2006). In Arabidopsis (Arabidopsis thaliana), AtPHO2 mediates the degradation of AtPHT1 proteins and PHOSPHATE1 (AtPHO1) at the endomembrane system (Chiou et al., 2006; Liu et al., 2012; Huang et al., 2013). NITROGEN LIMITATION ADAPTATION (NLA) containing an N-terminal SPX domain and a C-terminal RING domain with putative ubiquitin E3 ligase activity interacts with AtPHT1 proteins at plasma membranes, mediates their ubiquitination, and triggers endocytosis followed by vacuolar degradation (Lin et al., 2013). Loss of function of AtPHO2 or AtNLA in Arabidopsis results in Pi toxicity (Hamburger et al., 2002; Kant et al., 2011; Liu et al., 2012; Huang et al., 2013; Lin et al., 2013). Although it is not well understood, the toxicity phenotype of Arabidopsis nla mutants is known to be augmented under low-nitrate conditions (Kant et al., 2011), suggesting cross talk between N and P signaling pathways. Of note, the posttranscriptional cleavage of AtPHO2 and AtNLA mRNA in Arabidopsis is mediated by Pi deprivation-induced microRNA399 (miR399) and microRNA827 (miR827), respectively (Aung et al., 2006; Chiou et al., 2006; Hsieh et al., 2009), indicative of cooperation between different microRNAs and coordinated regulation of Pi transport at the posttranscriptional and posttranslational levels (Lin et al., 2013).

We previously showed that a rice NLA homolog (OsNLA1) is able to degrade the Arabidopsis AtPHT1 proteins (Lin et al., 2013), suggesting the conservation of NLA function in different plant species. Recently, a similar role of OsNLA1 in mediating the degradation of rice OsPHT1 proteins (OsPT2 and OsPT8) was reported (Yue et al., 2017). Like Arabidopsis, osnla1 mutants also accumulate an increasing level of Pi (Yang et al., 2017; Yue et al., 2017; Zhong et al., 2017). However, several questions remain. Because of inconsistent results in previous reports, it is still unclear whether the expression of OsNLA1 is regulated by Pi supply (Yang et al., 2017; Yue et al., 2017). It is also unclear whether the Pi toxicity phenotype of osnla1 depends on nitrate and whether OsNLA1 plays other roles in addition to Pi acquisition in the roots (Yue et al., 2017; Zhong et al., 2017). How OsNLA1 expression is regulated is also unknown. This is particularly interesting because, unlike Arabidopsis, there is no canonical miR827 target sequence in OsNLA1 transcript despite up-regulation of rice miR827 by Pi starvation (Lin et al., 2010, 2018).

Here, we showed that OsNLA1 indeed has a similar function to AtNLA in regulating Pi uptake in roots via posttranslational regulation of several rice OsPHT1 proteins and that this regulation is nitrogen dependent. Moreover, OsNLA1 is involved in pollen development and grain production, a function that has not been observed in Arabidopsis. Importantly, OsNLA1 expression is also induced by Pi supply, but in contrast to AtNLA, its regulation is not controlled by miR827. Instead, an upstream open reading frame (uORF) identified in the 5′ untranslated region (5′ UTR) of OsNLA1 plays such a regulatory role. The conservation of this uORF in cultivated rice species implies its importance in controlling Pi acquisition during domestication and modern rice cultivation.

RESULTS

OsNLA1.1 Is Up-Regulated by Pi Supply in Both Roots and Shoots

According to the OrygenesDB database (http://orygenesdb.cirad.fr/), OsNLA1 possesses three transcript variants as a result of alternative splicing at the 5′ UTR. OsNLA1.1 has the longest 5′ UTR, containing a uORF encoding 30 amino acids (Fig. 1A; Supplemental Fig. S1, A and B). OsNLA1.2 and OsNLA1.3 contain shorter 5′ UTRs than OsNLA1.1. To characterize the expression of different variants of OsNLA1, 1-week-old wild-type rice (‘Nipponbare’) seedlings were transferred to hydroponic solution supplemented with different Pi levels (0, 25, 500, or 1,000 µm) for a further 4 weeks. The Pi treatment was effective, since Pi starvation-induced expression of OsPT6 and OsIPS1 could be confirmed (Supplemental Fig. S2A). We first performed reverse transcription (RT)-PCR using the same set of primer pairs to amplify all three splicing variants. The amplicon size of OsNLA1.1, OsNLA1.2, and OsNLA1.3 was 460, 248, and 180 bp, respectively. Among these, OsNLA1.1 was the most abundant, and its expression was induced by high Pi in both shoots and roots (Fig. 1B). The expression level of OsNLA1.2 was very low and that of OsNLA1.3 was undetectable. We next focused on OsNLA1.1 and examined its expression by reverse transcription quantitative PCR (RT-qPCR) using OsNLA1.1-specific primers. The expression of OsNLA1.1 in both shoots and roots was increased when the Pi supply was elevated (Fig. 1C). Regarding the response to nitrate, OsNLA1.1 expression was down-regulated by increasing the supply of nitrate (from 0.45 to 7 mm) under Pi-sufficient conditions (500 µm) in both shoots and roots (Fig. 1D). Changes in OsNLA1.1 expression in response to external Pi supply suggest that OsNLA1.1 has a role in adaptation to Pi availability.

Figure 1.

OsNLA1.1 is up-regulated by high Pi but low nitrate supply. A, Gene structure of three OsNLA1 transcript variants. The black boxes represent the ORF comprising exons separated by introns, and the white boxes represent the UTR. The slash box represents the uORF. Black arrowheads indicate primers used for RT-qPCR analysis, and gray arrowheads indicate primers used for RT-PCR analysis. B, RT-PCR analysis of three OsNLA1 transcript variants in the shoot and root of plants grown under two Pi regimes. C, RT-qPCR analysis of OsNLA1.1 mRNA in the shoot and root of plants grown in Kimura B medium containing 0.45 mm nitrate with four Pi regimes. Different lowercase letters indicate significant differences among treatments at P < 0.05 after ANOVA and lsd test within root or shoot samples. D, RT-qPCR analysis of OsNLA1.1 mRNA in the shoot and root of plants grown under Pi-sufficient conditions (500 μm) with two nitrate regimes. Gene expression was analyzed in 1-week-old seedlings grown under different treatments for an additional 4 weeks. E, RT-qPCR analysis of OsNLA1.1 mRNA in the anther and spikelet collected from wild-type plants grown under high- or low-Pi conditions. Gene expression was normalized with Cyclophilin2, and se values refer to three to four biological replicates. Asterisks in D indicate statistically significant differences from the low-nitrate condition: **, P < 0.05 and *, P < 0.1.

OsNLA1 Expression Is Not Regulated by miR827

AtNLA expression is subject to posttranscriptional repression by miR827 in response to Pi deficiency in Arabidopsis (Hsieh et al., 2009). However, sequence analysis did not identify an miR827 target sequence in the OsNLA1 transcript (Supplemental Fig. S1A; Lin et al., 2010, 2018); instead, rice miR827 targets the OsPHT5 family (SPX-MFS) encoding vacuolar Pi transporters containing an N-terminal SPX domain and a C-terminal MSF domain (Lin et al., 2010; Liu et al., 2016). To examine whether miR827 regulates the expression of OsNLA1, we analyzed the expression of OsNLA1 in OsmiR827-overexpressing lines (Lin et al., 2010). The expression of OsmiR827 was highly up-regulated in the root of two overexpression lines compared with that of the wild type and an empty vector control. However, OsNLA1 expression was not repressed but instead exhibited slight up-regulation in these two miR827-overexpressing lines (Supplemental Fig. S1C). These results confirm that in rice, OsNLA1 expression is not regulated by OsmiR827.

Regulation of OsNLA1 Expression Is Controlled by Its Promoter Sequence and uORF

Because the expression of OsNLA1 is altered by external Pi supply (Fig. 1C) but it is not regulated by OsmiR827, we postulated that its upstream sequence, including its promoter and/or 5′ UTR, is responsible for such regulation. In order to investigate this possibility, different lengths of the upstream sequence of OsNLA1 were fused with a luciferase reporter gene and transiently expressed in rice protoplasts incubated in medium with (1.5 mm) or without Pi to mimic high- or low-Pi conditions. OsNLA1 promoter activity was very low and close to the background control when the upstream sequence between −1,500 and −1,000 bp was omitted (Fig. 2A), suggesting that this region is required for basal expression levels. Notably, when the included upstream sequence was extended to 2,472 bp comprising the 2,000-bp promoter region and the 5′ UTR (472 bp), in which a uORF is present, Pi-stimulated expression was observed in addition to enhancement of the overall activity (Fig. 2A). Because of the potential role of the uORF in regulating gene expression, we disrupted the uORF by mutating its start codon from ATG to AAG. Intriguingly, the Pi-stimulated expression was lost when the uORF was no longer present. When the promoter region between −2,472 and −2,000 bp was deleted, the expression was diminished and Pi-stimulated expression was not observed either. We then asked if the −2,472 to −2,000 bp upstream sequence alone or combined with the 5′ UTR is sufficient for Pi-induced expression but observed no promoter activity regardless of the Pi condition (Fig. 2A). Collectively, these results suggest that two distant regions, the uORF and the promoter region between −2,472 and −2,000 bp, are required to cooperatively regulate the Pi-induced expression of OsNLA1; however, these two regions alone are insufficient because the sequence between them is also needed for the expression of OsNLA1.

Figure 2.

Regulation of OsNLA1 expression by promoter sequence and uORF. A, The promoter activity driven by different lengths of OsNLA1 upstream sequence was analyzed in rice protoplasts incubated with (1.5 mm; +Pi) or without (−Pi) Pi. The activity is presented as relative luciferase activity (luciferase activity/GUS activity/protein content). se values refer to four to 24 biological replicates. Different lowercase letters represent significant differences among the constructs with Pi treatment at P < 0.05 after ANOVA and lsd test. Asterisks indicate statistically significant differences (**, P < 0.05 and ***, P < 0.01) between two Pi conditions in the same construct. B, Luciferase activity driven by the 2,472-bp OsNLA1 upstream sequence containing a uORF with the original (ATG) or a mutated (AAG) start codon, as detected in rice protoplasts incubated with different concentrations of Pi. se values refer to six to 11 biological replicates. Asterisks indicate statistically significant differences from the original uORF (ATG): *, P < 0.1.

After analysis with different software programs (Okumura et al., 2007), we identified several cis-elements in the promoter region (Supplemental Fig. S3A), but their importance requires future investigation. To further examine the Pi-dependent regulation, protoplasts transformed with the 2,472-bp promoter region containing either the original uORF (ATG) or the mutated uORF (AAG) were incubated under different Pi concentrations. Pi-stimulated OsNLA1 promoter activity was observed only when the original uORF was present and the activity increased prominently when Pi supply was above 1.5 mm Pi (Fig. 2B), suggesting uORF-mediated translational regulation of Pi-dependent expression of OsNLA1.

To characterize how prevalent this uORF identified from cv Nipponbare is in other rice species, the 5′ UTR sequences of OsNLA1 homologs from different rice species, including cultivated rice species Oryza sativa japonica, Oryza sativa indica, Oryza barthii, and Oryza glaberrima and wild rice species Oryza brachyantha, Oryza punctata, Oryza glumaepatula, Oryza nivara, Oryza rufipogon, Oryza meridionalis, and Oryza longistaminata, were collected from the Ensembl Plants and Phytozome plant genome databases for analysis. The alignment result showed that the 5′ UTR sequences could be distinguished into several groups, and four cultivated rice species were grouped within the same cluster (Supplemental Fig. S3B). Notably, the uORF identified from the cv Nipponbare OsNLA1 5′ UTR (designated as uORF2 in Supplemental Fig. S3B) was also present in three other cultivated rice species sharing 100% sequence identity, but it was not found in the wild rice species examined (Supplemental Fig. S3, B and C). It is therefore postulated that this uORF2 evolved after domestication of rice.

OsNLA1 Regulates Protein Abundance of Rice OsPHT1 Transporters

We previously showed that OsNLA1 plays a similar role to AtNLA in regulating the protein abundance of Arabidopsis Pi transporters, AtPHT1;1 and AtPHT1;4 (Lin et al., 2013). To find out if OsNLA1 can also regulate rice Pi transporters, the protein abundance of six rice OsPHT1 transporters, OsPT1, OsPT2, OsPT4, OsPT7, OsPT8, and OsPT12, was evaluated while coexpressed with OsNLA1 in tobacco (Nicotiana benthamiana) leaves. These Pi transporters were chosen because of their high expression in the root under Pi-sufficient conditions (Secco et al., 2013) or their predominant expression in anthers (Supplemental Fig. S4B). The protein amount of all tested rice Pi transporters was reduced markedly when coexpressed with OsNLA1 (Fig. 3, A and B), suggesting that these OsPTs are the potential targets degraded by OsNLA1. Previous results showed that the ubiquitin E3 ligase activity of AtNLA is essential for degrading AtPHT1 proteins (Lin et al., 2013). To examine if this is also the case for OsNLA1, we mutated the conserved Cys (residue 265) in the RING domain of OsNLA1 to Ala (OsNLA1^C265A) and found that the reduction of OsPT1 and OsPT4 proteins was impaired (Fig. 3B), indicating that the ubiquitin ligase activity of OsNLA1 is required to modulate the protein abundance of these OsPTs.

Figure 3.

OsNLA1-mediated regulation of protein abundance of rice OsPHT1 transporters. A, Protein abundance of OsPT2, OsPT7, OsPT8, and OsPT12 detected in tobacco leaves coexpressed with or without OsNLA1. GFP or ST-mRFP was used as a control of the infiltration event. OsNLA1 expression was detected by RT-PCR, and NbActin was used as an internal control. B, OsPT1 and OsPT4 protein abundance detected in tobacco leaves coexpressed with OsNLA1 or OsNLA1^C265A. C, BiFC analysis of the interaction between OsNLA1^C265A and OsPT1, OsPT4, OsPT7, or OsPT12. Reconstituted fluorescence signals were detected in the plasma membrane of tobacco leaf cells labeled by FM4-64 when OsNLA1^C265A-nYFP was coexpressed with OsPT1-, OsPT4-, OsPT7-, or OsPT12-cYFP but not with IRT1-cYFP. Bars = 20 µm.

To further examine whether OsPTs are the direct substrate of OsNLA1, the physical interaction between OsNLA1 and OsPTs was inspected by bimolecular fluorescence complementation (BiFC) in tobacco leaves. The interaction between OsNLA1 and OsPT2 or OsPT8 was reported recently (Yue et al., 2017). We therefore examined OsPT1, OsPT4, OsPT7, and OsPT12. Reconstituted yellow fluorescent protein (YFP) signals from coexpression of OsNLA1^C265A and OsPT1, OsPT4, OsPT7, or OsPT12 were observed in the membrane colocalized with FM4-64, indicating that the interaction takes place at the plasma membranes. This interaction is specific because coexpression of OsNLA1^C265A with an iron transporter (IRON REGULATED TRANSPORTER1 [IRT1]) did not show any YFP signal (Fig. 3C). Taken together, our results show that OsNLA1 functions similarly to AtNLA in directing the ubiquitin-mediated degradation of OsPTs at plasma membranes.

Misdegradation of OsPTs Results in Pi Overaccumulation in osnla1 Mutants in a Nitrate-Dependent Manner

To reveal the role of OsNLA1 in controlling Pi transport in rice, two rice mutant lines with Tos17 retrotransposon insertions residing within the OsNLA1 third exon (osnla1-2; NF0010) and second intron (osnla1-4; ND7049) were examined (Fig. 4A). To coordinate the nomenclature with the previous publications, NF5039 and NF0010 (Yue et al., 2017) are referred to as osnla1-1 and osnla1-2, respectively. NG3578 (Zhong et al., 2017) was renamed osnla1-3, and ND7049 first used in our study was designated as osnla1-4. Based on previous reports, osnla1-1 and osnla1-3 were characterized as knockdown mutants (Yue et al., 2017; Zhong et al., 2017). Our RT-PCR analysis of the full-length transcript showed that osnla1-2 is a knockout mutant and osnla1-4 is a knockdown mutant with ∼40% of transcript remaining (Fig. 4B).

Figure 4.

Importance of OsNLA1 in controlling Pi uptake and distribution in rice. A, The structure of OsNLA1 drawn to scale with the positions of respective insertions of Tos17 retrotransposon in nla1-4 and nla1-2 mutants. Black boxes represent exons separated by introns. Arrows indicate the locations of primers used in B. B, Analysis of full-length OsNLA1 mRNA in the root of the wild type (WT) and two mutants by RT-PCR. Samples were run on the same gel except for the nla1-4 sample, which was loaded separately from the others. The relative levels of OsNLA1 normalized with Actin in the wild type and nla1-4 are shown below the images. C, Root and shoot Pi contents of plants grown under Pi-sufficient conditions (500 μm) with two nitrate regimes (0.45 or 7 mm). FW, Fresh weight. D, Phenotypes of plants grown under Pi-sufficient and low-nitrate conditions. Arrowheads indicate wilted leaves. E, ³³Pi uptake activity of seedlings grown under two different conditions as in C. F, The Pi contents of different tissues of plants grown in the full-nutrient supplied field for 3 months. FLB, Flag leaf blade; FLS, flag leaf sheath; inN, internode; LB, leaf blade; LS, leaf sheath. se values in C, E, and F refer to three to four biological replicates. Asterisks indicate statistically significant differences from the wild type: ***, P < 0.01; **, P < 0.05; and *, P < 0.1.

The germinated seedlings were grown under sufficient Pi (500 µm) medium with two different concentrations of nitrate (0.45 and 7 mm), and the fourth leaf and roots were harvested for Pi analysis after 4 weeks. Under high-nitrate conditions, only osnla1-2 mutants showed higher Pi content than the wild-type plants (Fig. 4C). Under low-nitrate conditions, the accumulation of Pi in the mutants became aggravated. The Pi contents in osnla1-2 and osnla1-4 leaves were 3- and 2-fold that of the wild-type control, respectively. osnla1-2 mutants also showed high root Pi content under these conditions. Moreover, both mutants exhibited early senescence of mature leaves, and osnla1-2 mutants exhibited much reduced plant height compared with the wild type (Fig. 4D). This growth impairment was not observed under Pi-limited conditions, implying the importance of OsNLA1.1 in mediating plant fitness under sufficient Pi supply. Next, we examined Pi transport activity by incubation with radioactively labeled ³³Pi. Consistent with Pi accumulation, only osnla1-2 mutant plants displayed enhanced ³³Pi uptake activity under high-nitrate conditions. However, under low-nitrate conditions, ³³Pi uptake activity in both mutants was elevated (Fig. 4E). These results indicate that OsNLA1 negatively regulates Pi uptake and OsNLA1 loss of function results in high Pi accumulation and affects plant fitness under high Pi supply in a nitrate-dependent manner. Consistent with its role as a ubiquitin E3 ligase, none of the mRNA levels for OsPT1, OsPT2, OsPT4, and OsPT8 were altered in osnla1 mutants, suggesting that OsNLA1 is involved in modulating the protein abundance of these Pi transporters rather than their transcript levels (Supplemental Fig. S4A).

OsNLA1 Promoter Activity Is Active in Different Tissues and Developmental Stages

To further reveal the biological relevance of OsNLA1, we mapped the spatial expression pattern of OsNLA1 by analyzing promoter-GUS transgenic lines. Based on the GUS staining, the OsNLA1 promoter activity was strong in the vascular tissues of roots, coleoptiles, and shoots of 1-week-old seedlings grown under one-half-strength Murashige and Skoog medium (Fig. 5, A–F). Strong GUS staining responding to high-Pi conditions was clearly observed in the whole root (Fig. 5, G and H), including epidermis, cortex, and vascular tissues of 5-week-old plants (Fig. 5K). Moreover, all root types, including crown roots, large lateral roots, and fine lateral roots, showed strong activity under high-Pi conditions (Fig. 5, I and J). Intriguingly, during the flower developmental stage, GUS activity was observed in specific regions, such as the bracts of rachises and ligules of flag leaves (Fig. 5, M and N), anthers (Fig. 5, L and P), stigmas (Fig. 5P), and flag leaf sheaths (Fig. 5O). Cross sections revealed strong signals inside microspores and vascular tissue of anthers (Fig. 5Q) and parenchyma surrounding xylem vessels and phloem in flag leaf sheaths (Fig. 5R). To verify the expression of OsNLA1 in reproductive organs, we examined its mRNA level in spikelets and anthers collected at 1 d before anthesis from wild-type plants grown under high- or low-Pi conditions (see “Materials and Methods”). The RT-qPCR results showed that OsNLA1.1 was highly expressed in anthers and spikelets; however, no differences were detected between the samples collected from different Pi conditions (Fig. 1E). No change in OsNLA1.1 expression was correlated with the steady Pi level in anthers and spikelets, in spite of the effectiveness of Pi treatment in vegetative tissues (e.g. flag leaf sheaths; Supplemental Fig. S2C). We postulate that remobilization of Pi to reproductive organs in adult plants may diminish the response to Pi starvation. Taken together, the broad expression pattern of OsNLA1 suggests that it not only regulates Pi uptake in the root, as previously reported, but is also involved in modulating Pi translocation or redistribution in the shoot and reproductive organs.

Figure 5.

OsNLA1 promoter activity is active in different tissues and developmental stages. GUS staining results of pOsNLA1 (2,472 bp)-GUS transgenic plants are shown. A, Root. B, Coleoptile. C, Shoot. D to F, Corresponding cross sections of A to C. G, Root grown under 0 μm Pi conditions. H to J, Roots grown under 500 μm Pi conditions. K, Cross section of crown root and large lateral root. L, Panicle. M, Rachis. N, Flag leaf blade. O, Flag leaf sheath. P, Spikelet. Q, Cross section of anther. R, Cross section of flag leaf sheath. The staining parts in M and N are bract and ligule, respectively. Samples in A to F, G to K, and L to R were collected from 1-week-old seedlings, 5-week-old plants, and plants at flowering stage, respectively. CR, Crown root; FLR, fine lateral root; LLR, large lateral root. Bars = 200 µm (A–C), 2 mmi(L–O), 1 mm (G, H, and P), and 50 µm (D–F, I– K, Q, and R).

OsNLA1 Is Involved in Pollen Development and Grain Production

Since OsNLA1 promoter activity was active in leaves and reproductive organs, the relevance of OsNLA1 for Pi distribution or panicle development was further evaluated. Wild-type plants and osnla1 mutants were grown with full nutrient supply, and various tissues were sampled for Pi analysis (Fig. 4F). At the reproductive growth stage, nodes were numbered from top to bottom in roman numerals. The Pi content in the flag leaves and leaf I (blades and sheaths) of both osnla1 mutants was higher than that of wild-type plants. The Pi content in the ligules of two osnla1 mutants, which showed high OsNLA1 promoter activity, was also higher than that of wild-type plants. Except for node I, the Pi contents in the tissues examined were increased in at least one mutant (Fig. 4F).

To further investigate the role of OsNLA1 during reproductive growth, rice plants were grown in pots containing soil collected from a nutrient-limited field. Each plant was supplied with a defined fertilizer regime until the reproductive stage (see “Materials and Methods”). Pollen development, tiller number, panicle number, and grain yield were assessed. Compared with wild-type plants, osnla1-2 and osnla1-4 mutants were stunted, had reduced shoot dry weight, and produced fewer tillers (Fig. 6, A–C), although panicle numbers were not different from the wild type (Fig. 6D). The seed-setting rate in the two osnla1 mutants was also significantly reduced, from 76% observed in the wild type to 48% and 62% observed in osnla1-2 and osnla1-4, respectively (Fig. 6E). When the first five panicles were evaluated in isolation, grain number and grain and panicle dry weight were found to be reduced in the two osnla1 mutants compared with the wild type (Fig. 6, F–H). Because of the expression of OsNLA1 in pollen grains and the reduction in grain number in osnla1 mutants, we examined pollen viability by the iodine staining method (Baker and Baker, 1979). The degree of pollen staining indicates the pollen starch content, which indicates fertility (Fig. 7, A–C). Compared with wild-type pollen with 88% fertility (black, full stain), osnla1-2 and osnla1-4 mutants had reduced pollen fertility of 55% and 83%, respectively (Fig. 7D). Also, osnla1-2 and osnla1-4 mutants had greater proportions of partially fertile pollen (brown, partial stain), specifically 30% and 10%, respectively, compared with 5% in wild-type plants. Moreover, the amount of sterile pollen (yellow or transparent) in the osnla1-2 mutant (15%) was twice that in the wild type (7.3%; Fig. 7D). Interestingly, the Pi content in the anthers of both osnla1 mutants was higher than that of wild-type plants (Fig. 7E), hinting at an association between Pi content and pollen viability. To determine which Pi transporter might be the potential target of OsNLA1 in anthers, we queried the rice eFP Browser and found abundant expression of OsPT7 and OsPT12 in anthers (Li et al., 2007; Supplemental Fig. S4B). Our results of protein-protein interaction and degradation analysis have validated OsPT7 and OsPT12 as potential targets of OsNLA1 (Fig. 3, A and C). It is possible that the role of OsNLA1 during pollen development and subsequent grain production is through regulating Pi allocation.

Figure 6.

OsNLA1 is required for shoot growth and grain production. Characterization of morphological features of wild-type (WT), nla1-2, and nla1-4 plants after seed setting is shown. se values refer to 10 biological replicates. Asterisks indicate statistically significant differences from the wild type: **, P < 0.05 and *, P < 0.1. A, Plant height. B, Shoot dry weight per plant. C, Tiller number per plant. D, Panicle number per plant. E, Seed-setting rate. Grain number (F), grain dry weight (G), and panicle dry weight (H) per first five panicle.

Figure 7.

OsNLA1 is required for pollen viability. A to C, Pollen viability assay determined by starch staining in the wild type (WT; A), nla1-2 (B), and nla1-4 (C). Arrow indicates a pollen grain with a high level of starch (full stain); dashed arrow indicates a pollen grain with a low level of starch (partial stain); arrowhead indicates a sterile pollen grain. Bars = 100 µm. D, The percentage of each category quantitated as a bar chart. E, Pi contents of anthers. se values refer to four biological replicates. FW, Fresh weight. Asterisks in D and E indicate statistically significant differences from the wild type: ***, P < 0.01 and *, P < 0.1.

DISCUSSION

uORF Is Involved in Regulating OsNLA1 Expression in Response to Pi Supply

Previous studies reported contradictory results regarding the Pi-dependent regulation of OsNLA1 expression (Yang et al., 2017; Yue et al., 2017). Here, our analyses of mRNA level (Fig. 1C), promoter-GUS activity (Fig. 5, G and H), and protoplast transient expression (Fig. 2) showed that the expression of OsNLA1.1 is indeed up-regulated by external Pi supply. This discrepancy may result from the differences in plant age and duration of experimental treatment. Under our growth conditions, significant changes in OsNLA1.1 mRNA level were detected after 4-week treatment but not after 1- to 2-week treatment of 1-week-old seedlings (Fig. 1C; Supplemental Fig. S2B), which is likely because of the high amount of P stored in the seeds. The Pi-dependent expression of OsNLA1 is particularly interesting because, unlike AtNLA, OsNLA1 is not regulated by miR827, which is Pi starvation induced (Supplemental Fig. S1C; Lin et al., 2010; Yue et al., 2017). We found that the upstream sequences of OsNLA1, including the uORF and promoter region from −2,472 to −2,000 bp, mediate this Pi-dependent regulation cooperatively; however, additional sequences are required for the basal expression of OsNLA1 (Fig. 2). It has been reported that regulatory sequences positioned up to several kilobases and even more than 1 Mb away from core promoter can influence transcription rates (Hernandez-Garcia and Finer, 2014). The interaction between these two regions may occur by folding and association with chromatin.

uORF-mediated translational control can occur through different mechanisms, such as the interference of the main ORF translation through ribosome subunit dissociation, nonsense-mediated mRNA decay, or translation reinitiation of the main ORF after uORF translation (Barbosa et al., 2013). Recent studies have reported the importance of the uORF in regulating the expression of NIP5;1 (a boric acid channel) and BOR1 (a borate exporter) in response to the change in external boron concentrations (Tanaka et al., 2016; Aibara et al., 2018). Although we do not know which mechanism mediates the uORF-regulated expression of OsNLA1, the conservation of this uORF (uORF2 in Supplemental Fig. S3) among cultivated rice varieties implies its functional role in regulating Pi uptake of cultivated rice bred in a Pi-rich field. We speculate that during domestication, response to high Pi supply regulated by OsNLA1 might have been important for the growth of cultivated rice species. Future investigation into the OsNLA1 expression levels under different Pi supply in cultivated and wild rice species will provide further insights.

OsNLA1 Mediates Pi Homeostasis in a Nitrate-Dependent Manner

Enhancement of Pi accumulation under low-nitrate conditions was observed in Arabidopsis nla mutants (Kant et al., 2011). However, inconsistent results regarding Pi accumulation were reported for rice nla1 mutants in response to low nitrate (Yue et al., 2017; Zhong et al., 2017). Our study here showed that like AtNLA, OsNLA1 maintains Pi homeostasis in a nitrate-dependent manner. The OsNLA1.1 transcript was up-regulated by low nitrate (Fig. 1), which augmented Pi accumulation in osnla1 mutants (Fig. 4). Moreover, a low nitrate-dependent early-senescence phenotype was both previously observed in Arabidopsis nla mutants and observed here in osnla1 mutants (Peng et al., 2007; Kant et al., 2011; Fig. 4). Recently, in Arabidopsis, NLA/PHO2 was found to function in posttranslational regulation of ORE1, a key transcription factor of leaf senescence, during nitrogen limitation (Park et al., 2018). In contrast to transcriptional up-regulation on OsNLA1, low-nitrate conditions suppressed the translation of AtNLA (Liu et al., 2017), suggesting that nitrate may regulate the expression of AtNLA homologs in different plant species by distinct mechanisms. It was found that nitrogen limitation directly represses the Pi-starvation response in rice, which is comparable to low nitrate-induced OsNLA1 expression (Takehisa and Sato, 2019). In Arabidopsis, the HRS1 transcription factor has been proposed to be the molecular gate integrating P and N signals (Medici et al., 2015). It would be interesting to examine whether rice OsHRS1 homologs regulate the expression of OsNLA1.

OsNLA1 Modulates Pi Transport by Posttranslational Regulation of Rice OsPHT1 Transporters

In the tobacco transient expression system, we showed that six OsPHT1 transporters are the targets of OsNLA1. In addition to OsPT2 and OsPT8, which were previously reported (Yue et al., 2017), we identified an additional four targets, OsPT1, OsPT4, OsPT7, and OsPT12 (Fig. 3). Although the ubiquitination activity of OsNLA1 has not been proven, the prerequisites for the ubiquitin ligase activity of OsNLA1 to degrade OsPTs and their interaction suggest that OsNLA1 possesses a similar function to AtNLA in facilitating the degradation of PHT1s through the protein ubiquitination machinery.

Since all the tested OsPHT1 transporters can be degraded by OsNLA1, we postulate that OsNLA1 regulates the degradation of all rice OsPHT1 members through recognition of the conserved sequence or structure. OsNLA1 promoter activity can be observed in various tissues and at different developmental stages (Fig. 5); thus, its native OsPHT1 target might be identified by temporal and spatial coexpression patterns. Investigation of the expression patterns of OsNLA1 and OsPTs in the rice eFP Browser (Li et al., 2007; Supplemental Fig. S4B) indicated that OsPT7 and OsPT12 may be the targets of OsNLA1 in anthers and OsPT1 and OsPT4 may be the major targets of OsNLA1 in shoots, roots, embryos, endosperm, and seeds. Yue et al. (2017) showed that Pi overaccumulation in osnla1 mutants might be a result of misdegradation of OsPT2 and OsPT8, since small interfering RNA-mediated repression of OsPT2/PT8 restored leaf Pi concentration to a wild-type level in osnla1 mutants. However, the off-target effects of small interfering RNA on other OsPTs were not investigated. Measurement of the protein abundance of endogenous OsPTs and their ubiquitination status in the wild type and osnla1 mutants by OsPT-specific antibodies will elucidate the native target of OsNLA1 in rice.

The Role of OsNLA1 in Pi Mobilization from Shoots to Reproductive Tissues

Strong OsNLA1 promoter activity was observed in the vascular tissue of flag leaf sheaths and ligules (Fig. 5). OsNLA1 loss of function resulted in Pi accumulation in ligules and young leaves during the reproductive stage (Fig. 4F). Given the function of OsNLA1 in degrading OsPHT1 Pi transporters, OsNLA1-mediated suppression of Pi transporters in these specific tissues might control Pi reallocation from shoots to reproductive organs. Recently, Yamaji et al. (2017) identified the expression of a plasma membrane-localized Pi transporter, SPDT, in the node where it transports Pi from xylem to phloem. SPDT loss of function caused decreased P in the grains but increased P in the leaves. Based on the proposed model, Pi is taken up to the leaves via xylem flow, reallocated to the phloem, and transferred to the node for further transport to the reproductive tissues. We hypothesized that during Pi reallocation from the xylem to the phloem in the leaves, down-regulated expression of OsPHT1 Pi transporters by OsNLA1 in xylem parenchyma cells could prevent Pi from retrieval back into the xylem and thereby promote xylem-to-phloem transport. Indeed, several rice OsPHT1 transporters, such as OsPT1, OsPT2, OsPT6, and OsPT8, have been reported to show promoter activity in the xylem parenchyma of shoots (Ai et al., 2009; Jia et al., 2011; Sun et al., 2012). Thus, they are the potential candidates regulated by OsNLA1 in these cells. Coexpression of OsPT4 and OsNLA1 in flag leaves further implied that OsNLA1 might regulate the protein abundance of OsPT4, which was shown to play a role in remobilizing Pi from flag leaves to panicles (Ye et al., 2015).

A rice leaf is composed of three parts: a leaf sheath, a leaf blade, and a laminar joint that contains a pair of auricles and the ligules. It is interesting that ligules showed high expression of OsNLA1 and at least threefold increased Pi accumulation in the osnla1-2 mutant compared with the wild type (Fig. 4). The ligule is a thin, white, tongue-like organ that is considered to be the degenerated tip of the leaf sheath (Hoshikawa, 1989). Although the function of the ligule is still not clear, its protective role through the exclusion of dust or harmful spores and the synthesis of a secretory product has been proposed (Chaffey, 2000). Given that the ligule is the tip of the leaf sheath, we postulate that Pi driven by transpiration through xylem flow might be accumulated to a high level in the ligule. During the reproductive stage, suppression of OsPHT1 Pi transporters by OsNLA1 in the ligule might facilitate the redirection of Pi to the shoot phloem and panicle. The role of the ligule in Pi mobilization is an attractive target for future inspection.

In addition to vegetative tissues, we found that OsNLA1 had strong promoter activity in pollen grains and vascular tissues of anthers (Fig. 5Q), which was not observed in Arabidopsis (Peng et al., 2007). Concomitant with increased Pi content, the pollen viability of both osnla1 mutants was significantly reduced compared with the wild type (Fig. 7), which led to impaired grain yield (Fig. 6). It was shown that Pi is transported to female organs and also stored in pollen to sustain their growth and development (Lau and Stephenson, 1994). Several transgenic rice plants or mutants with disturbances in the expression of genes involved in the Pi signaling pathway, such as OsPHR2-overexpressing lines and ospho2 mutants, also showed reduced pollen viability (Zhou et al., 2008; Cao et al., 2014). These results suggest a close association between Pi homeostasis and pollen viability.

In summary, our results show that uORF-regulated Pi-induced OsNLA1 controls Pi uptake, translocation, and reproductive growth, likely through posttranslational regulation of rice OsPHT1 transporters in different tissues and cell types. The conservation of the uORF in cultivated rice species further implies the importance of this region in regulating Pi homeostasis during domestication and modern rice growth.

MATERIALS AND METHODS

Plant Material and Growth Conditions

The rice (Oryza sativa ssp. japonica) mutant genotypes and transgenic lines were generated in the cv Nipponbare background. Homozygous seeds of rice Tos17 insertional mutant lines NF0010 (nla1-2) and ND7049 (nla1-4) disrupted in OsNLA1 were obtained from the Rice Genome Resource Center of the National Institute of Agrobiological Sciences, Japan (Miyao et al., 2003). Seeds were sterilized and germinated in one-half-strength Murashige and Skoog agar medium at 30°C for 1 week and then transferred to modified one-half-strength Kimura B hydroponic medium containing an additional 0.5 mm MES for a further 4 weeks (Ma et al., 2001). Plants were grown at 30°C under a 12-h photoperiod with light intensity at 100 to 150 μmol m⁻² s⁻¹, and the nutrient solution was replenished weekly. To measure Pi level during the reproductive stage shown in Figure 4F, plants were grown in the soil collected from fields that had been continuously fertilized. Every plant received an additional 5 g of No. 5 fertilizer twice during the 3-month growth period. The No. 5 fertilizer was purchased from the Taiwan Fertilizer Company, and the composition of the fertilizer was 16 (N):8 (P):12 (K):5 (S):10.5 (CaO):50 (organic matter). To investigate yield production shown in Figure 6, plants were grown inside containers with 8 kg of soil collected from fields that had not been fertilized for years. Every plant received an additional 10 g of Ca(NO₃)₂, 11 g of superphosphate (N:P₂O₅:K₂O = 0:18:0), and 5 g of KCl twice during the 3-month growth period. To investigate OsNLA1.1 expression in the anther and spikelet in response to Pi, the same condition was used with two Pi regimes, 0 and 11 g of superphosphate as low and high Pi, respectively. Samples were harvested at the 1 d before anthesis stage. All plants grown with soil were in flooding conditions.

Measurement of Pi Content, Pi Uptake Activity, and Ion Content

To measure Pi content, tissues were homogenized with 1% (v/v) glacial acetic acid. After centrifugation, the Pi content of the supernatant was determined by colorimetric assay based on the formation of phosphomolybdate (Chiou et al., 2006). To monitor Pi uptake activity, 5-week-old plants were incubated with modified Kimura B hydroponic medium (500 μm Pi) containing [³³P]orthophosphate (Pi) for 4 h. Roots and shoots were harvested at different time points and digested with 30% (v/v) H₂O₂ and perchloric acid (5:2, v/v). ³³P radioactivity was measured with a Beckman liquid scintillation counter.

RNA Isolation and RT-qPCR

Total RNA was isolated using Trizol reagent (Invitrogen). Total RNA was treated with DNaseI (Invitrogen), and cDNA was synthesized from 1 µg of total RNA using Moloney murine leukemia virus reverse transcriptase (Invitrogen) and oligo(dT) primers. Sequences of primers used for RT-qPCR are listed in Supplemental Table S1. qPCR was performed by using the Power SYBR Green PCR Master Mix kit on the 7500 Real-Time PCR system. Gene expression was normalized to the constitutively expressed Cyclophilin2 gene and displayed as a function of Cyclophilin2 expression.

Generation of Constructs for Promoter-GUS Lines, Transient Tobacco Expression, BiFC Assays, and Transient Promoter Activity Assays

To generate promoter-GUS transgenic lines, a 2,472-bp fragment of the promoter region of OsNLA1 including 472-bp 5′ UTR was cloned into the PCR8/GW/TOPO vector (Invitrogen) for subsequent introduction into the Gateway pHGWFS7.0 vector containing an enhanced GFP-GUS fusion reporter gene (Karimi et al., 2002). For transient tobacco (Nicotiana benthamiana) expression, the open reading frames of OsPT1, OsPT2, OsPT4, OsPT7, OsPT8, OsPT12, OsNLA1, and OsNLA1^C265A were cloned into the PCR8/GW/TOPO vector and then subcloned into the Gateway pMDC32, pK7FWG2.0, or pH7FWG2.0 vector. For BiFC assays, the open reading frame of OsNLA1^C265A and OsPT1/PT4/PT7/PT12 was cloned into Gateway pUBC-nYFP and pUBC-cYFP vectors, respectively. For transient promoter activity, different fragments of OsNLA1 promoter were cloned into the PCR8/GW/TOPO vector (Invitrogen) and subcloned into the Gateway vector pGW2-Luc (Lu et al., 2007) to fuse with luciferase. The pUbiP:GUS vector (Lu et al., 1998) was used as a control for normalization. Gateway LR Clonase II Enzyme Mix (Invitrogen) was used for cloning into the destination vector. Phosphorylated primers with the desired change of nucleotide sequence were used to generate point mutations. Sequences of primers used for cloning are listed in Supplemental Table S2.

Rice Protoplast Preparation, Transformation, and Transient Promoter Activity Assays

The sterilized seeds were germinated under modified Kimura B medium (no Pi) with 0.8% (w/v) agar. The seedlings were grown at 30°C under light for 4 d and moved to the dark for a further 3 d. A bundle of rice plants including shoots and roots were cut into approximately 0.5-mm strips. The strips were incubated in 20 mL of enzyme solution (2% [w/v] Cellulase RS [Yakult], 1% [w/v] macerozyme R10 [Yakult], 10 mm MES, pH 5.6, 0.6 m mannitol, 10 mm CaCl₂, and 0.1% [w/v] BSA) for 3 h in the dark with gentle shaking and filtered through 50-µm nylon mesh to release protoplasts. The protoplasts were collected by centrifugation at 250g, washed once with W5 solution (154 mm NaCl, 125 mm CaCl₂, 5 mm KCl, 2 mm MES, and 5 mm Glc), and resuspended with MMg solution (0.6 m mannitol, 15 mm MgCl₂, and 4 mm MES, pH 5.7). For transient promoter activity assays, 5 µg of plasmid DNA containing the OsNLA1 promoter and 3 µg of plasmid DNA of pUbiP:GUS vector was mixed with 100 µL of protoplasts. Then, 110 µL of freshly prepared polyethylene glycol solution (40% [v/v] PEG4000 and 0.6 m mannitol) was added and mixed gently, and the mixture was incubated at room temperature for 20 min. After incubation, 440 µL of W5 solution was added, and protoplasts were collected by 250g centrifugation. The supernatant was removed, and the protoplasts were resuspended gently in 1 mL of W5 solution with or without Pi (1.5 mm KH₂PO₄) and incubated in the dark for 16 h. Then, the protoplasts were collected, and total protein was extracted by CCLR solution (100 mm potassium phosphate, pH 7.8, 1 mm EDTA, 10% [v/v] glycerol, 1% [v/v] Triton X-100, and 7 mm 2-mercaptoenthanol). The activity of luciferase and GUS was analyzed by using protoplast total protein extraction. The concentration of total protein was measured by Pierce 660-nm protein assay solution. OsNLA1 promoter activity is presented as relative luciferase activity (luciferase activity/GUS activity/protein content) normalized by the cotransformed GUS activity driven by the ubiquitin promoter (pUbiP:GUS) and total protein amount.

Protein Isolation and Immunoblot Analysis

Total protein in tobacco leaves was isolated using extraction buffer (60 mm Tris, 2.5% [v/v] glycerol, 0.2 mm EDTA, and 2% [v/v] SDS). Fifty micrograms of total protein of each sample was loaded for immunoblot analysis. GFP and ST-mRFP (a Golgi marker) proteins, used as the control of the infiltration event, were detected by rabbit polyclonal GFP-HRP antibody (1:5,000; Proteintech) or rabbit anti-GFP antibody (1:500; Abcam) and rabbit polyclonal RFP antibody (0.5 µg mL⁻¹; Abcam), respectively. The polyclonal antibodies against the peptides of OsPT1 (amino acid residues 510–526), OsPT2 (amino acid residues 509–526), OsPT4 (amino acid residues 520–538), OsPT7 (amino acid residues 270–287), and OsPT8 (amino acid residues 524–540) were raised individually. The antibodies were affinity purified and applied at 1 µg mL⁻¹.

Pollen Viability Assay

Anthers were collected from five to six rice spikelets at 1 d before anthesis and submerged in 2% (v/v) I₂-KI solution for starch staining. Pollen grains were placed on a glass slide by using forceps to squeeze the anther. Stained pollen grains were observed and counted using a light microscope. Pollen viability was categorized into three groups: fertile with full stain (black), partial stain (brown), and sterile (yellow or transparent). An average of 750 pollen grains per plant were counted.

Fluorescence Microscopy

Fluorescence signals were observed with a Zeiss LSM 780 plus ELYRA S.1 with objective Plan-Apochromat 320/0.8 mm and LD C-Apochomat 340/1.1 W. Excitation/emission wavelengths were 514 nm/520 to 550 nm for YFP and 561 nm/575 to 630 nm for FM4-64.

Statistical Analysis

The data were statistically analyzed by Student’s t test, one-way ANOVA, and Fisher’s lsd for multiple comparisons. The significance level decided by P value was labeled and described in each figure.

Accession Numbers

Accession numbers are as follows: OsNLA1, Os07g47590; OsPT1, Os03g05620; OsPT2, Os03g05640; OsPT4, Os04g10750; OsPT7, Os03g04360; OsPT8, Os10g30790; and OsPT12, Os03g05610.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. OsNLA1 expression is not regulated by OsmiR827 in rice.

• Supplemental Figure S2. Gene expression responses to different supplies of Pi.

• Supplemental Figure S3. Sequence analysis of OsNLA1 promoter and 5′ UTR in different Oryza species.

• Supplemental Figure S4. Potential OsPTs regulated by OsNLA1 in rice.

• Supplemental Table S1. RT-PCR and RT-qPCR primers for gene expression analysis.

• Supplemental Table S2. Sequences of primers used for cloning.