The Japonica rice (Oryza sativa L.) variety of Nipponbare was selected for this study. To determine the presence of Arabidopsis NLA homolog in rice, its full-length protein sequence was BLAST searched in Rice Genome Annotation Database server¹. The whole length of NLA protein sequences were used for phylogenetic analysis. The phylogenetic tree was computed with MEGA5 software² using maximum likelihood method with 1000 bootstrap replicates.

The T-DNA line of Tos17 insertion (NG3578) was obtained from rice the Tos17 insertion mutant database³. The genotyping of the T-DNA mutant plants was performed by using OsNLA1-specific primers NG3578LP (5′-GGCGACAGAGAAAATGCTTC-3′) and NG3578RP (5′-CTTGGCAAAATGGCATACCT-3′), along with Tos17-specific primer Tos17-tail6 (AGGTTGCAAGTTAGTTAAGA-3′). The construct for OsNLA1 over-expression was created by cloning the full-length OsNLA1 cDNA coding region into expression vector pCAMBIA 1300 driven by a maize ubiquitin promoter.

To generate the OsNLA1 –RNAi construct, a 430 bp OsNLA1 cDNA fragment was amplified by OsNLA1GLF (5′-CACCGAACAACGGTGCTATGGAGC-3′) and OsNLA1GLR (5′-TCACATGCCCAAGAATGCCCT-3′) then subcloned into the pENTR/D-TOPO cloning vector (Invitrogen, Carlsbad, CA, United States). Subsequently, the fragment was transferred inverted repeatedly into pANDA binary vector (miki 2004) downstream of a maize ubiquitin promoter.

All the transgenic lines were generated through Agrobacterium-mediated transformation. Japonica rice variety Nipponbare was used for all transgenic lines as well as Tos17 insertion line.

To grow plants in hydroponics, rice seeds of different genotypes (WT, T-DNA or Osnla1-1, two RNAi lines or Osnla1-2 and Osnla1-3, and three over-expression lines or OsNLA1-OE1, OsNLA1-OE2, and OsNLA1-OE3) were germinated between moistened filter papers in Petri dishes for 7 days in a climatic chamber (day/night rhythm: 12/12 h, 28/23°C, light intensity: 550 μmol m^-2 s^-1). Seven days after germination, the seedlings were transferred to five different hydroponic systems for different nutrient treatments: (i) control (all nutrients were supplied at optimum levels as previously described by Ranathunge et al., 2003), (ii) low nitrate with 30 μM KNO₃, (iii) low phosphate (LP) with 5 μM KH₂PO₄, (iv) high phosphate (HP) with 300 μM KH₂PO₄, and (v) low nitrate and high phosphate together. In treatments, all other nutrients were provided in an optimum concentration. The plants were grown in the same environmental conditions as those used for seed germination. Nutrient solutions were renewed every week. Four weeks old plants were used for genotyping, Pi and total P measurement of shoots and roots as well as nitrate and phosphate permeability measurements of roots.

In soil pot experiments, control plants were grown with full nutrients that included Sunshine LA4 (Sun Grow Horticulture, Agawam, MA, United States), peat moss and vermiculite (Perlite Canada Inc., Lachine, QC, Canada) with a ratio of 1:1:4, and 1 g of 14-14-14 Nutricote slow releasing fertilizer (Plant Products, Ltd Co, Brampton, ON, Canada). For P and N treatment experiments, the seedlings were planted in nutrient-free soil (peat-moss and vermiculite with a ratio of 1:4). It is known that positively charged nutrient ions can be bound to the soil particles, whereas, negatively charged ions can be lost by leaching (Marschner, 1995). Such bound and leached nutrient ions are not available for plants. Hence, soil-grown plants were provided with a greater concentration of nutrients than in hydroponics. The nutrient solution that had been used for hydroponics was modified with different KNO₃ levels. To study how different nitrate levels affect/alter Pi uptake, a set of plants were grown in soil with low (3 mM) and optimum/sufficient (10 mM) KNO₃ levels. For P treatment, plants were grown either with 0.1 mM (low) or 1 mM (optimum/sufficient) or 6.5 mM (high) KH₂PO₄ levels. Four weeks old plants were used for genotyping, phenotypic assays and other physiological and metabolic analyses.

In order to investigate whether or not ammonium affects Pi uptake in rice, wild-type, knockdown lines and over-expression lines were grown both in hydroponics and soil-pots with different ammonium levels but with an optimum/constant Pi level. In hydroponics, a concentration of 90 μM (NH₄)₂SO₄ was used as low, whereas, 300 μM (NH₄)₂SO₄ was used as the optimum. In soil-pot experiments, a relatively higher ammonium levels were used than in hydroponics due to binding of ions to negatively charged soil particles. In soil, the applied (NH₄)₂SO₄ concentrations were 1 and 10 mM; the latter concentration being the optimum while the former concentration was sub-optimal. Four weeks old plants were harvested and used for Pi analysis.

To analyze the expression level of OsNLA1 gene in different tissues of wild-type plants (leaf, root, stem, sheath, and inflorescence with open and closed spikelets), 2-month-old plants were harvested at noon and tissue samples were frozen immediately in liquid nitrogen. Total RNA was extracted from each tissue using TriZol regent (Invitrogen, Carlsbad, CA, United States) and RNEasy RNA extraction kit (Qiagen Inc., Toronto, ON, Canada). To eliminate genomic DNA from RNA, the samples were treated with RQ1 RNAse-free DNAse enzyme (Promega Cooperation, Madison, WI, United States). The cDNA was synthesized from total RNA with qScript cDNA SuperMix from Quanta Biosciences (Gaithersburg, MD, United States) and used for qRT-PCR analysis. The PCR primers were designed using the Applied Biosystems (Forster City, CA, United States) software Primer Express 2.0 (Supplementary Table S1). The qRT-PCR was carried out and analyzed with a 7300 real-time PCR system (Applied Biosystems, Foster City, CA, United States) according to the manufacturer’s instruction. For the selection of endogenous control (reference gene), OsUBQ1, OsGAPDH, and OsActin8 were tested. OsUBQ1 showed a uniform and most stable expression in all plant tissues as well as under different treatments. Hence, the constitutively expressed OsUBQ1 was used as endogenous control. Three technical and three biological replicates were used for each measurement, and relative quantity was calculated using the 2^-ΔΔT method (Livak and Schmittgen, 2001).

Quantitative RT-PCR was used to determine the expression level of OsNLA1 in shoots of 4-week-old mutant and wild-type plants. Shoots from each plant were collected at noon and were frozen immediately in liquid nitrogen. Frozen samples were then ground into powder and used for RNA extraction. Preparation of cDNA and analysis of relative expression levels using qRT-PCR were performed as described earlier in the previous section. OsNLA1 primers and endogenous control of OsUBQ1 specific primers were used for this analysis. Three biological replicates were used for each assay.

All plants (over-expression lines, RNAi lines and Tos17 insertion line along with wild-type) grown in hydroponics and soil with different nitrogen and phosphorous treatments (see plant growth section), were used for Pi analysis. In some experiments, total P in shoots and roots was also measured. After analyzing phenotype, plants were harvested and whole root and shoot tissues were homogenized with liquid nitrogen separately. Pi concentration was measured as previously described by Chiou et al. (2006), which was slightly modified from the method of Ames (1966). For analysis, approximately 0.1 g tissue powder was used for each sample. One milliliter of extraction buffer (1mM EDTA, 1 mM β–mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 10 mM Tris and 100 mM NaCl) was added to the homogenized sample. 100 μl of supernatant was added to 900 μl of 1% glacial acetic acid and incubated at 42°C for 30 min. The samples were centrifuged for 5 min at 13,000 g. 300 μl of supernatant was mixed with 700 μl assay solution (0.35% NH₄MoO₄, 1.4% ascorbic acid, and 0.86N H₂SO₄) and incubated at 42°C for 30 min. The OD value of samples were measured at A820 using a spectrophotometer (MULTISKAN GO, Thermo Fisher Scientific, Vantaa, Finland). At least 3–5 biological replicates were used for each assay.

The nitrate and Pi permeability of roots were measured by a root pressure probe. For this method, excised end segments of roots lacking laterals were used as previously described by Miyamoto et al. (2001). These measurements were performed for plants grown in both optimum (300 μM) and low (30 μM) nitrate levels for 3–4 weeks. Excised root segments were tightly but carefully connected to a root pressure probe without damaging them with the aid of cylindrical seals prepared from liquid silicone material (Xantopren; Bayer, Leverkusen, Germany). Once roots attained steady-state root pressures (usually it took 1–2 h), the osmotic experiments were performed. To measure nitrate and phosphate permeability of roots, the original nutrient solution was rapidly exchanged with a medium containing either 25 mM KNO₃ (∼50 mOsmol/kg) or 25 mM KH₂PO₄ (∼50 mOsmol/kg), dissolved in the nutrient solution. During the osmotic experiment, the external root medium was rapidly stirred to mix the solution and to minimize the effects of the external, unstirred layers. The changes in root pressure in response to changes in the osmotic pressure of the medium were biphasic. There was a rapid water phase followed by a slower solute phase (see Miyamoto et al., 2001 for osmotic curves). From the solute phase, the permeability coefficients of KNO₃ and KH₂PO₄ (P_sr in m s^-1) were calculated according to Steudle et al. (1987):

where k_sr is the rate constant of permeation of solutes. They were calculated from the half-time of solute exchange (t^s_1/2) of each osmolyte used here. V_x is the volume of mature xylem, which was 1 to 2% of the total root volume (Miyamoto et al., 2001).

Cutting experiments were conducted to validate the readings after each experiment. When the xylem of the root remained open, there was a quick drop in root pressure to zero and a drastic decrease in the t^w_1/2. If this did not occur, the results were discarded. All transport measurements were conducted at 27°C. Ten roots were used for each genotype with different nitrate levels.

The full-length coding sequence of OsNLA1 (Os07g0673200) was amplified using primers OsNLA-F (GGGGACAAGTTTGTACAAAAAAGCAGGCTTTATGAAGTTTGCCAAGAAGTACG), and OsNLA-R (GGGACCACTTTGTACAAGAAAGCTGGGTTTCACATGCCCAAGAATGC) and cloned into pDONR^TM 221 vector using a BP Clonase^TM enzyme mix (Invitrogen^®). The coding region was then shuttled into compatible destination vector, pB7WG2D, using the LR Clonase^TM enzyme mix by following the instructions given in the manufacturer protocol (Invitrogen^®, Invitrogen, Carlsbad, CA, United States). Atnla mutant was transformed with the destination vector using agrobacterium-mediated transformation by floral-dip method (Clough and Bent, 1998). Putative positive transformants were obtained by spraying with BASTA (0.1%). Integration and expression of the transgene OsNLA1 in Arabidopsis nla background was confirmed using PCR analysis in transgenic lines. Seeds from homozygous transgenic lines identified (from T2 generation) were used in atnla complementation assay. cDNA samples from five independent putative transgenic lines in nla background were subjected to PCR analysis to confirm the expression of OsNLA1 transgene. Primers set (qRTOsNLA1-F and qRTOsNLA1-R) was used to detect the amplification of OsNLA in Arabidopsis in different lines.

Wild-type (Col-0), atnla mutant and atnla-complementation lines transformed with OsNLA1, were grown under nitrogen sufficient (10 mM nitrate) and nitrogen limitation (3 mM nitrate) conditions as previously described by Peng et al. (2007). Rosette leaves from plants were harvested after 27 days of sowing, and Pi contents/concentrations were analyzed as described earlier by Chiou et al., 2006. Three to four replicates were used for each genotype.

The full-length coding regions of the OsNLA1 and OsPHO2 (Phosphate over accumulator 2) were amplified using the primer pairs, yOsNLA1-F and yOsNLA1-R, and yOsPHO2-F and yOsPHO2-R, and cloned into pGEM-T easy vector. The coding regions of the OsNLA1 and OsPHO2 were then restricted from pGEM-T easy vector using Nde1 and BamH1, and ECOR1 and Xho1 restriction enzymes, and cloned into pGADT7-AD and pGBKT7-BD vectors, respectively. pGADT7-empty and pGADT7-OsNLA1constructs were transformed individually into Y187 strain. pGBKT7-empty and pGBKT7-OsPHO2 constructs were transformed into Y2HGold strain. Yeast mating between Y187 and Y2HGold strains was performed following the instructions given in Matchmaker Gold Yeast Two-Hybrid System User Manual (Clontech, United States). The hybrids were identified as positive clones based on their ability to grow DDO (SD/-Leu/-Trp) plates. Interactions were analyzed by growing the hybrids on DDO/X/A (SD/-Leu/-Trp/X-a-gal/Aureobasidin A) and QDO/X/A (SD/-Leu/-Trp/-His/-Ade/X-a-gal/Aureobasidin A) plates.

Data presented in the figures are means ± SD. Analysis of variance (ANOVA), and the LSD test were employed to compare the means of different genotypes with different nitrate and Pi levels at a confidence level of 95%.

Using the protein sequence of AtNLA gene (At1g02860), one homologous OsNLA1 in rice (Os07g0673200 or LOC_Os07g47590 given by Rice Annotation Project) was discovered by TBLASTN search⁴. The analysis of amino acid sequence of OsNLA1 showed that it has 59.7% sequence identity to AtNLA. OsNLA1 consisted of an N-terminal SPX domain and a C-terminal RING domain, which was similar to AtNLA (Figure 1A). Phylogenetic analysis with other related SPX proteins in Arabidopsis and rice showed that these genes together form a clade (Figure 1B).

The spatial expression of OsNLA1 was determined using quantitative RT-PCR analysis in different plant tissues (leaf, root, stem, sheath, and inflorescence with open and closed spikelets) of rice (Japonica rice variety of Nipponbare) (Figure 2). The tissue of open spikelets was set as the reference or 1 for comparison with other tissues. Although OsNLA1 transcripts were detectable in all the tissue types examined, the sheath and leaf tissues had the greatest transcript levels, reaching approximately 18- and 10-fold higher than in the reference tissue, respectively, and 14- and 8-fold higher than in the roots, respectively (Figure 2). The lowest expression of OsNLA1 was found to be in the open spikelets, which was approximately half of the inflorescence with closed spikelets (Figure 2).

To determine the role of OsNLA1 in rice (in Pi homeostasis), both loss-of-function and gain-of-function approaches were employed. In the context of the loss-of-function approach, a homozygous Tos17 insertion line (NG3578) with T-DNA insertion in the first intron was included in the study (Figure 3A). qRT-PCR analysis of the T-DNA line (referred to as Osnla1-1, henceforth) showed that the transcripts of OsNLA1 were still detectable (∼24% of the wild-type) suggesting that Osnla1-1 is not a null mutant, and thus represent a knockdown mutation (Figure 3B). In addition to this, RNAi lines expressing an OsNLA1 hair-pin construct were also generated. Amongst these, two independent RNAi lines, Osnla1-2 and Osnla1-3, which demonstrated the highest decrease in the transcript levels of the endogenous OsNLA1 compared to the wild-type, were selected for further analysis (Figure 3B). The TOS17 line (Osnla1-1) and RNAi lines (Osnla1-2 and Osnla1-3) lines have the same background. In the case of the gain-of-function approach, transgenic lines overexpressing OsNLA1 under the UBQ promoter were generated. Three independent over-expression lines, OsNLA-OE1, OsNLA-OE2, OsNLA-OE3 with maximum expression levels of the OsNLA1 transgene were selected for further analysis (Figure 3C).

Alteration of OsNLA1 expression did not affect the shoot biomass content of rice plants under optimal nutrient levels (nitrate and Pi; Supplementary Figure S1). Even under low Pi (0.1 mM), all the transgenic lines and the wild-type grew similarly (Figure 3D). However, when grown under high phosphorus (HP) condition, the knockdown lines, Osnla1-1, Osnla1-2, and Osnla1-3, showed reduced plant size with lower tiller numbers and leaf necrosis (starting from the tip of the leaf blade), which were apparent Pi toxicity symptoms (Figures 3E,F). Biochemical analysis of the leaf tissues showed that knockdown lines accumulated significantly higher levels of Pi compared to wild-type, especially under HP level (Figure 3G). On average, under high Pi level, the knockdown lines had 1.4- to 3.8-fold greater Pi content compared to the wild-type. In contrast, the over-expression lines accumulated significantly lower levels of Pi containing 50–60% of the Pi content of wild-type (Figure 3G). Interestingly, even under the optimal Pi level, the transgenic lines exhibited the same trend for Pi accumulation but at a much lesser level. However, no differences were found for Pi accumulation when transgenic lines were grown under low Pi level (Figure 3G).

In general, irrespective of the genotype, when grown under sufficient and low nitrate levels, shoots had markedly greater Pi content than roots (Figure 4A). In all genotypes, low nitrate treatment significantly increased Pi content of both leaf and root tissues. The comparison of genotypes showed that T-DNA and RNAi lines had significantly greater Pi contents in shoots and roots than wild-type at both growth conditions but it was more pronounced under low nitrate levels (Figure 4A). In contrast, the over-expression lines had slightly lower Pi contents than wild-type in both shoots and roots (Figure 4A). The calculated percent Pi increase under low nitrate revealed that there was a marked increment in both shoots and roots but this was more pronounced in the latter (Figure 4B). This data clearly showed that the knockdown plants had the greatest percent Pi increase over wild-type and over-expression lines (Figure 4B). The Pi increase was the same for wild-type and over-expression lines. When compared with the wild-type, the percent Pi increase of knockdown lines under low nitrate was 9–35% in shoots and 22–73% in roots (Figure 4C). This percent Pi change was slightly negative for over-expression lines. On average, it was 0.5–8% and 1–7% lower than wild-type for shoots and roots, respectively. In summary, the knockdown lines had the greatest Pi accumulation under low nitrate treatment, whereas over-expression lines had the lowest Pi accumulation.

In all genotypes, a low nitrate treatment led to a markedly increased total P content in both leaf and root tissues, which was the same as for Pi (Supplementary Figure S2A). The comparison of genotypes showed that OsNLA1 knockdown lines had greater total P contents in shoots and roots than wild-type and over-expression lines grown at a low nitrate level (Supplementary Figure S2A). The total P content of over-expression lines and wild-type was the same, except for roots grown under low nitrate. The percent P increase under low nitrate showed that there was a marked increment in both shoots and roots but the knockdown lines had the greatest values, whereas, over-expression lines had the lowest values compared to wild-type (Supplementary Figure S2B). On average, the knockdown lines had 15–62% total P increase in shoots and 15–52% in roots than that of wild-type (Supplementary Figure S2C). This percent P change was slightly negative for over-expression lines. On average, it was 1–4% and 5–8% lower than wild-type for shoots and roots, respectively.

The resulted negative plant phenotypic effects, such as shorter plants and less tiller numbers of OsNLA1 knockdown plants were apparently due to the accumulation of high P/Pi levels under low nitrate or high phosphate levels (Figures 3, 4). These observations led us to test the response of transgenic lines grown in both low nitrate and high phosphate together. As expected, when grown in low nitrate and high phosphate levels together, plants showed a more severe leaf chlorotic and necrotic phenotype compared to the plants grown in optimum nitrate and high phosphate together (Figures 5A vs. 5B). All knockdown lines had stronger phenotypes with leaf necrosis at the tip of the leaf blade than the wild-type grown in low nitrate and high phosphate together (Figure 5A). The phenotype of the over-expression lines and the wild-type was the same. These chlorotic/necrotic patches in knockdown lines were what one would expect for Pi toxicity symptoms due to hyper-accumulation of Pi in the shoot. Consistent with this hypothesis, in both growth conditions, the measured Pi content of shoots showed that the knockdown lines had markedly greater levels than wild-type, and over-expression lines (Figure 5C). The latter had the lowest Pi contents grown under either low or optimum nitrate levels with high Pi (Figure 5C). In all genotypes, low nitrate treatment significantly increased Pi acquisition in rice plants. In summary, OsNLA1 knockdown lines showed the strongest Pi toxicity symptoms and the greatest Pi accumulation under high Pi/low nitrate levels. In contrast, all three over-expression lines had the mildest leaf necrotic symptoms and the least Pi accumulation under high Pi/low nitrate levels.

To test whether there is a relationship between ammonium and Pi accumulation in rice plants, wild-type and Osnla1-1 knockdown line were grown in both hydroponics and soil, providing them with optimum Pi and different ammonium levels (optimum and low). The results showed that there was no clear correlation between different ammonium levels and Pi accumulation in the shoot of wild-type and Osnla1-1 knockdown line (Supplementary Figure S3). Lower ammonium treatments failed to accumulate higher Pi levels in the shoot of both genotypes either in soil or hydroponics, which was in contrast compared to plants grown at a low nitrate level (see Figure 4A). Osnla1 had significantly greater Pi content in the shoot than wild-type in all ammonium treatments.

The Pi and nitrate permeability (Ps_r) was measured for adventitious roots grown in either optimum/control (300 μM), or low (30 μM) nitrate levels. In general, regardless of the genotype, roots grown in low nitrate had higher Pi and nitrate permeabilities than roots from the optimum nitrate treatment (Figures 6A,B). When grown and measured under low nitrate level, the Pi permeability of roots of the knockdown lines had markedly greater values than wild-type and the over-expression lines (Figure 6A). Under low nitrate level, on average, the knockdown lines had 34–66% greater permeability values than wild-type plants (P < 0.05; Figure 6A). In contrast, the Ps_r of roots for Pi was the same for wild-type and over-expression lines under both nitrate levels. Further, there was no significant difference in nitrate permeability of roots between the genotypes (Figure 6B). Plants grown in low nitrate had higher nitrate permeability, likely due to weak root barriers, which were induced by sub-optimal N that resulted in higher permeability. In summary, when plants were provided with low nitrate, it enhanced Pi uptake of roots but it was more pronounced in the knockdown lines than in the wild-type and over-expression lines.

Given the high sequence similarity between rice and Arabidopsis NLA proteins, we next examined if rice NLA could complement the nla mutation in Arabidopsis. Five independent transgenic lines, expressing rice NLA under control of the 35S CaMV promoter in the Arabidopsis nla background, were grown under nitrogen (N) sufficient (10 mM KNO₃) and limitation (3 mM KNO₃) conditions (Figures 7A,B). Under N sufficient conditions at 27 DAS, all the transgenic lines including nla and wild-type grew normally with no drastic senescence symptoms (Figure 7A). However, under N limitation conditions, the nla mutant displayed the premature senescence phenotype compared to wild-type (Figure 7A). Interestingly, all five transgenic lines expressing rice NLA did not show premature senescence and grew phenotypically similar to wild-type under N limitation condition. These observations were further confirmed through the analysis of Pi content. When grown under optimum N level, on average, the Pi content of the nla mutant was 1.7-fold greater than the wild-type, whereas, complemented lines had 1.4-fold of the wild-type (Figure 7C). In contrast, the over-accumulation of Pi was more pronounced under low N conditions in nla mutant reaching 4.3-fold compared to wild-type. However, the complemented lines had significantly lower content of Pi than the nla mutant with 1.6- to 2.1-times greater than the wild-type (Figure 7C). These results suggest that OSNLA1 partially complemented the nla mutant phenotype.

A yeast two-hybrid (Y2H) assay was carried out to determine whether there is an interaction between OsNLA1 and OsPHO2 as earlier found for Arabidopsis (Park et al., 2014). In this assay, one protein is translationally fused with the DNA binding domain (GAL4-BD) and the other with the activation domain (GAL4-AD) of the GAL4 transcription factor. The result showed that there is an interaction between OsNLA1 and OsPHO2. The yeast hybrid cells carrying the pAD-empty and pBD-OsPHO2, and pAD-OsNLA1 and pBD-empty constructs were unable to grow on DDO/X/A and QDO/X/A plates (Figure 8). However, the hybrids harboring the pAD-OsNLA1 and pBD-OsPHO2 constructs were able to grow on DDO/X/A and QDO/X/A plates (Figure 8), suggesting that OsNLA1 and OsPHO2 interact with each other and this interaction may play an important role in phosphate acquisition/homeostasis of rice plants.