To analyze the OsNPF7.3 promoter, a 2136 bp fragment upstream of OsNPF7.3 open reading frame (ORF) was amplified by PCR using the primers shown in Supplementary Table 1, and inserted in front of the β-glucuronidase (GUS) coding region in pCAMBIA1301¹ with EcoR I and Nco I to make the pOsNPF7.3-GUS plasmid. To construct OsNPF7.3-overexpressing plants, a 1782-bp OsNPF7.3 ORF was inserted downstream of the 35S promoter in pCAM1301 with Bgl II and Afl II, producing the p35S-OsNPF7.3 plasmid. To generate the OsNPF7.3-RNAi construct, two 314-bp fragments of the OsNPF7.3 cDNA were amplified using the primers listed in Supplementary Table 1 and transferred downstream of the Ubi-1 promoter in vector pTCK303 (Wang et al., 2004) with BamHI/KpnI and SpeI/SacI, respectively, to generate the OsNPF7.3-RNAi vector. To produce OsNPF7.3-GFP, the OsNPF7.3 ORF was amplified by PCR using the primers listed in Supplementary Table 1. The 1779-bp amplified OsNPF7.3 ORF (lacking the stop codon) was cloned in front of the green fluorescent protein (GFP) coding region in the pCAM1302 vector with Bgl II and Spe I and the final pOsNPF7.3-GFP construct was created. All of the plasmid constructs were transferred into Agrobacterium strain EHA105, and then transformed into japonica rice variety Zhonghua 11 (ZH11) by Agrobacterium-mediated transformation method (Hiei et al., 1997). T2 or T3 transgenic lines were selected using real time-PCR. The corresponding primers are indicated in Supplementary Table 1.

To understand the subcellular localization of OsNPF7.3, p35S-OsNPF7.3-GFP, and p35S-GFP were introduced into rice protoplasts. These were prepared and transformed according to previously protocols (Zhang et al., 2011). Protoplasts were isolated from leaf sheaths of rice seedlings after sowing for 7–15 days. OsNPF7.3-GFP under control of the 35S promoter was introduced into tobacco epidermal cells using transferred into tobacco by Agrobacterium-mediated transformation. The green fluorescence of GFP and coexpressed membrane marker AtTPK-mkate (Voelker et al., 2006) were observed using a confocal laser scanning microscope (Leica SP8 AOBS, Wetzlar, Germany).

Hydroponic experiments were conducted with filled sand under glasshouse using the basic rice culture solution (Yoshida et al., 1976). To analyze OsNPF7.3 expression in the presence of different N sources, promoter-GUS transgenic seedlings and ZH11 were grown in the basic nutrient solution for 3 weeks using 1 mM NH₄NO₃ as the N source. They were then transferred to basic nutrient solution without N where they remained for 3 days. The N-starved seedlings were transferred to basic nutrient solutions supplemented with one of the following as the sole N source: 0.5 mM NaNO₃; 2.0 mM NaNO₃; 5.0 mM NaNO₃; 0.25 mM (NH₄)₂SO₄; 1.0 mM (NH₄)₂SO₄; 2.5 mM (NH₄)₂SO₄; 0.25 mM NH₄NO₃; 1.0 mM NH₄NO₃; 2.5 mM NH₄NO₃; 2% peptone; 2 mM Gln; or 2 mM Glu. Samples were harvested at different times for histochemical GUS staining and RNA extraction. OsNPF7.3 transgenic plants on sand were grown in basic nutrient solution supplemented with 1.0 mM NH₄NO₃. The nutrient solution was renewed every 3 days. The greenhouse condition is 32°C for sodium lamp 400w 14 h in the daytime and 25°C for dark 10 h in the evening. The paddy rice grew at season June to October at rice experimental base of Huazhong Agricultural University under fertilizer (N/P/K = 19%/7%/14%) 180 kg/hm², and the nitrogen level of soil in rice paddy is 90 mmol/kg. Generally, the number of rice plants was 30 for each experiment and the planting density was 19.98 × 19.98 cm. For the field yield trials, the number of rice plants was 100 for each OE line and ZH11.

Histochemical GUS staining was performed according to Jefferson (1987). The stained materials were observed using light or fluorescence microscopy. To make sections, the stained tissues of root, stem, leaf, and panicle were rinsed and fixed in FAA (formalin-acetic acid-70% ethanol [1:1:18]) at 4°C for 24 h, gradually dehydrated with ethanol for 15 min each time, and washed two times with 100% ethanol for 30 min each time. Finally, embedded in Spurr resin, and then sectioned. The sections were observed using Zeiss Axio Imager M2 (Carl Zeiss AG, Oberkochen, Germany).

Total free amino acid concentration was measured by the ninhydrin method (Fang et al., 2013). Single free amino acid concentration was measured by HPLC method using amino acid analyzer L-8800 HITACHI. The samples were prepared as follows: 1 g rice tissue was mixed, then was soaked in 80% ethanol 10 ml at 80°C water bath for 20 min, and the water bath was repeated two times. The collected extracts were placed at 80°C dry bath to remove ethanol, and 1 ml 0.5 M NaOH was used to dissolve the substance, then centrifuged with 14000 rpm for 15 min. The supernatant was filtered with filter membrane 2 μm, and the filtrates 0.8 ml was tested with the amino acid analyzer. Total nitrogen content and total protein content were determined using the semi-micro Kjeldahl method by using a nitrogen analyzer (Smart Chem 200, Westco, Italy). Total soluble protein were pyrolyzed then measured by the Coomassie brilliant blue G250 method (Sedmak and Grossberg, 1977). Chlorophyll content was measured using the technique of Dhindsa et al. (1981). Photosynthesis was measured by portable photosynthetic apparatus LI-6400 (LI-COR Company, America). Nitrogen utilization efficiency was determined from the formula [grain yield (g)/grain nitrogen content (g) + straw nitrogen content (g)] × 100%.

Total RNA was extracted using TRIzol reagent (Invitrogen, Beijing, China). First-strand cDNA were synthesized using oligo (dT) primers and MLV reverse transcriptase (TaKaRa Bio, Beijing, China). Quantitative real-time PCR was performed using SYBR Green mix (TaKaRa Bio, Beijing, China) and monitored with the 7500 RT qPCR system (Applied Biosystems, Foster City, CA, United States). The primers used for qPCR are listed in Supplementary Table 1.

For all treatments, Duncan’s test was run using one-way ANOVA. Significance was accepted at P < 0.05 and P < 0.01.

First, we attempted to determine its intracellular localization of peptide transporter OsNPF7.3. The transient expression of 35S: OsNPF7.3-GFP in rice protoplast produced green fluorescence at the vacuolar membrane (Figures 1D–F), but for the 35S-GFP control it occurred at the plasma membrane and nucleus (Figures 1A–C). To further determine the subcellular localization of OsNPF7.3, OsNPF7.3-GFP fusion protein (under the control of the CMV 35S promoter) was transiently coexpressed with the vacuolar membrane marker AtTPK-mkate in rice protoplasts. The results showed that the OsNPF7.3-GFP fusion protein signal was localized at the vacuolar membrane and overlap with the vacuolar membrane marker signal (Figures 1G–I). In addition, it showed OsNPF7.3-GFP localized to the membrane system using Agrobacterium carried OsNPF7.3-GFP injection to tobacco leaf (Figures 1J–L). It was also observed that OsNPF7.3 has 12 transmembranes structure (TM). Overall, these data indicate that OsNPF7.3 is vacuolar membrane-localized peptide transporter.

OsNPF7.3 is expressed in most of organs, as shown in transgenic rice plants with the promoter of OsNPF7.3 driving GUS expression (Figure 2). In roots, GUS staining was observed in root tip (Figure 2A), and lateral roots (Figure 2B). In the seedling shoots, GUS staining is noted in the outgrowth buds (Figure 2C), leaf blade (Figure 2D), stems (Figures 2E,F). Furthermore, we found that OsNPF7.3 is expressed in young panicles (Figures 2G,H), but it is not expressed in anther of panicle (Figure 2H) and filled grains (Figure 2I). In addition, the expression of OsNPF7.3 is higher in lateral root and stem by real time-PCR in further experiment (Figure 2J).

We produced promoter-GUS transgenic rice plants to determine the transcriptional response of OsNPF7.3 to various nitrogen treatments. Promoter-GUS plants were transferred to various nitrogen source solutions to monitor if the GUS staining is induced. GUS staining was noted deeply in the root tip and lateral roots at 24 h with the nitrate treatment, but at 8 h with the ammonium and ammonium nitrate treatments (Figure 3A). Similarly, the expression of OsNPF7.3 was significantly increased with the nitrate between 0 and 24 h, but decreased in ammonium and ammonium nitrate treatments at 24 h by real time-PCR (Supplementary Figures 1A,B). To systematically characterize the regulation of OsNPF7.3 expression by other nitrogen sources, GUS staining was performed using the organic nitrogen treatments. Intense GUS staining (Figure 3B) and increased expression level of OsNPF7.3 (Supplementary Figures 1C,D) at 2, 8, and 24 h was observed where organic nitrogen (peptone, Gln, or Glu) was the sole nitrogen source.

To determine the tissue-specific expression pattern of OsNPF7.3, cross- or longitudinal sections of the GUS-stained organs were made (Figure 4). Sectioning confirmed that higher expression of OsNPF7.3 was observed in the cortex parenchyma and the vascular bundle of the root (Figures 4A–C), the vascular bundle of the leaf (Figure 4D), the glume epidermis (Figure 4E), and the vascular bundle of the stem (Figures 4F,G). It was revealed that OsNPF7.3 expressed in the cortex parenchyma cells in the root by transverse sectioning (Figures 4B,C), and the parenchyma and sieve cells of the vascular bundle in the elongation zone (Figure 4C). The expression level of OsNPF7.3 was also observed in the parenchyma and sieve cells of the vascular bundles in the leaf and stem but only in the epidermal cells of the glume (Figures 4D–G). When the roots were treated with peptone, sectioning also showed that GUS staining obviously intensified in the epidermal vascular bundle (Figure 4I) from epidermis (Figure 4H). Overall, these observations show that OsNPF7.3 is expressed higher in the parenchyma cells adjacent to the veins and sieve cells in response to organic nitrogen application, and this may suggest that OsNPF7.3 contributes to nitrogen allocation.

To determine the function of OsNPF7.3 in rice plants, we generated OsNPF7.3-RNAi transgenic rice plants (RNAi) under the control of the rice Ubi-1 promoter. Sixty-five independent OsNPF7.3-RNAi T0 lines were obtained, and we chose five homozygous T2 lines for transcript level measurement, the result showed that the OsNPF7.3 transcript levels of Ri-D1 to Ri-D5 are clearly reduced in leaves (Supplementary Figure 2), but OsNPF7.3 expression is not decreased in the leaves of the null segrigant, which are separated from OsNPF7.3-RNAi (Ri-S1 to Ri-S5) lines (Supplementary Figure 2). We then selected the lines Ri-D1, Ri-D2, Ri-D3 for subsequent tests and labeled them Ri1, Ri2, and Ri3, respectively. To study the effects of increased OsNPF7.3 expression, OsNPF7.3 over-expressing (OE) rice was constructed under the control of the 35S promoter. Forty-eight independent OE T0 lines were obtained, and we chose five homozygous T2 lines for transcript level determination. The result showed that the transcript levels is quietly high in leaves of OE-D1, OE-D3, and OE-D5 (Supplementary Figure 2), and they were used for subsequent tests and labeled them OE1, OE2, and OE3, respectively.

Since nitrogen regulates OsNPF7.3 expression, we tested the responses of OsNPF7.3 transgenic plants raised on sand with 1.0 mM NH₄NO₃. The results indicated that the tiller numbers of OE lines were higher compared with that WT, but lower of these RNAi lines (Figures 5A,B). The root dry weight per plant and leaf dry weight per plant of three OE seedlings were apparently higher than those of WT (Figures 5C,D). Inversely, down-expression of OsNPF7.3 decreased those characteristics (Figures 5A–D). To investigate why OsNPF7.3 affects rice growth, we measured the free amino acid content (Figure 5E) and total nitrogen content in the roots, leaf sheaths and the leaves of RNAi and OE lines. It showed that the free amino acid content per organ and total nitrogen content per organ of RNAi plants were lower than those of WT (Figure 5F). Nevertheless, free amino acid content and total nitrogen content of OE plants were relatively higher in all organs (Figures 5E,F). Furthermore, we found that sole free amino acid concentration (Asp, Gly, Ala, Cys, Val, Leu, Tyr, Phe, His, and Arg) was higher in the leaf sheath of RNAi lines, resulting in accumulation of total free amino acid concentration in leaf sheath (Supplementary Table 1). However, total free amino acid concentration was not changed in root and leaf of the RNAi lines, but increased in root and leaf of OE lines (Supplementary Table 1). Therefore, OsNPF7.3 could affect nitrogen allocation from the roots to the leaves for rice growth.

To investigate the role of OsNPF7.3 in N transport and recycling during leaf senescence and grain filling, some indices of N recycling were measured. From the data, we found that the soluble protein content gradually decreased in the leaves of the OE lines during the reproductive stage (Figure 6A). This factor declined faster in the OsNPF7.3 OE lines than WT and RNAi lines (Figure 6A). More importantly, the soluble protein content in the panicles was obviously higher in OsNPF7.3 OE lines than the WT at all reproductive stages (Figure 6B). Chloroplast degradation accompanied the decrease in leaf protein content, so we measured the chlorophyll content of different transgenic plants. The results showed that the chlorophyll content accumulated in the RNAi lines (Figure 6C), and that the total grain protein per panicle increased in the OsNPF7.3 OE lines but decreased in the RNAi lines (Figure 6D). Based on leaf phenotype observations, leaf senescence in the OsNPF7.3 OE lines was accelerated but slowed down in the RNAi lines in the later reproductive stages (Figure 6E). Grain numbers increased in the OsNPF7.3 OE lines but decreased in the RNAi lines (Figure 6F). Moreover, the OsNPF7.3 OE lines accelerate soluble protein degradation under different nitrogen treatments during the later reproductive stages (Supplementary Figure 3). Nitrogen concentrates in the leaves and is transported to the panicles for seed growth. Therefore, leaf soluble protein content decreased with increasing OsNPF7.3 expression and the translocation of nitrogen from leaf to panicle was accelerated. Consequently, more proteins were stored in the panicles, seed growth was enhanced, and grain yield increased.

The effects of altered OsNPF7.3 expression on agronomic traits associated with grain yield were evaluated using field-grown plants. The agronomic traits of three OsNPF7.3 OE lines (OE1–OE3) were superior to those of the WT (Figure 7A). The three RNAi (Ri1–Ri3) lines presented with relatively lower tiller numbers, dwarfism, and short panicles (Figure 7A). The tiller numbers per plant (Figure 7B) and the panicle numbers per plant (Figure 7C) of the OE lines were greater than those of the WT and clearly greater in three of the lines. Nevertheless, the tiller and panicle numbers of the RNAi lines were distinctly lower (Figures 7B,C) than those of the WT. The filled grain numbers per panicle of the OsNPF7.3 OE lines enhanced but apparently decreased in the RNAi line (Figure 7D). In addition, OsNPF7.3 OE lines can clearly increase panicle length and secondary branch numbers per panicle but not affected primary branch number per panicle and 1000-grain weight (Supplementary Figure 4). The expression of N recycling related gene OsGS1.1 increased in OE lines but decreased in RNAi lines (Figure 7E). Besides, the photosynthetic rate Pn (Figure 7F) and the nitrogen utilization efficiency (Figure 7G) of OsNPF7.3 OE lines increased evidently in three lines relative to the WT but obviously decreased in RNAi lines compared to the WT. The grain yield traits were also measured under fertilizer regimes. Filled grain number per plant was significantly greater in the OE lines compared with the WT (Table 1). Furthermore, the higher grain yields of the three OE lines (over WT 5.5, 6.7, and 6.1%) were verified in a field test in 2016 (Table 1). Therefore, OsNPF7.3 participates in rice grain yield maintenance by increasing tiller and filled grain numbers.