Knockdown of the partner protein OsNAR2.1 for high-affinity nitrate transport represses lateral root formation in a nitrate-dependent manner

The morphological plasticity of root systems is critical for plant survival, and understanding the mechanisms underlying root adaptation to nitrogen (N) fluctuation is critical for sustainable agriculture; however, the molecular mechanism of N-dependent root growth in rice remains unclear. This study aimed to identify the role of the complementary high-affinity NO3− transport protein OsNAR2.1 in NO3−-regulated rice root growth. Comparisons with wild-type (WT) plants showed that knockdown of OsNAR2.1 inhibited lateral root (LR) formation under low NO3− concentrations, but not under low NH4+ concentrations. 15N-labelling NO3− supplies (provided at concentrations of 0–10 mM) demonstrated that (i) defects in LR formation in mutants subjected to low external NO3− concentrations resulted from impaired NO3− uptake, and (ii) the mutants had significantly fewer LRs than the WT plants when root N contents were similar between genotypes. LR formation in osnar2.1 mutants was less sensitive to localised NO3− supply than LR formation in WT plants, suggesting that OsNAR2.1 may be involved in a NO3−-signalling pathway that controls LR formation. Knockdown of OsNAR2.1 inhibited LR formation by decreasing auxin transport from shoots to roots. Thus, OsNAR2.1 probably functions in both NO3− uptake and NO3−-signalling.

Recent studies have highlighted the role of AtNRT2.1 in root architecture response to low NO 3 − availability, especially in LR initiation 16,19 . Experiments using atnar2.1-1 identified the role of AtNAR2.1 in LR responses to low NO 3 − supply 20 ; further work showed that the essential role of AtNAR2.1 for the localization of AtNRT2.1 to the plasma membrane 20,21 . miR393/AFB3 is a NO 3 − -responsive module that controls LR density in responses to external and internal N concentrations in Arabidopsis 12,13 . Interestingly, AGL7-clade MADS-box gene AtAGL21 plays a crucial role in both LR formation and elongation 22 . Thus, NO 3 − -dependent root development is apparently under the control of complex mechanisms, although its signalling components have remained largely unidentified.
Our understanding of NO 3 − -regulated LR development is limited in monocot plants, including rice, the model cereal plant (Oryza sativa L.) 8,23 . Elevated NO 3 − responsiveness in rice plants is strongly related to increased LR initiation regardless of whether the roots system was split or whole 8,9 . Recent work showed that the transcriptional levels of four ANR1-like genes (OsMADS25, OsMADS27, OsMADS57 and OsMADS61) were markedly modulated by NO 3 − availability 24 . Furthermore, in miR444a-overexpressing rice lines expression of the target ANR1-like genes was down-regulated and LR elongation was less responsive to localised NO 3 − 23 . These studies indicated that ANR1-like genes might have a similar role in regulating root developmental response to NO 3 − in rice, despite the evolutionary distance between Arabidopsis and rice.
We previously reported that OsNAR2.1, a rice NAR2-like gene, has no known transport activity, but is required to complement high-affinity NO 3 − transport. Yeast two-hybridisation showed that OsNAR2.1 interacted with OsNRT2.1/2.2/2.3a, and knockdown of OsNAR2.1 suppressed expression of the three high-affinity-transportsystem NO 3 − transporters, unlike OsNRT1.1 [25][26][27][28] . Knockdown of OsNAR2.1 impaired NO 3 − uptake; mutants had only ~65% of the N concentration measured in wild-type (WT) plants when the NO 3 − supply was limited. No such difference was found when a low concentration of NH 4 + was provided as the sole inorganic N source 27 . In this study, we confirmed inhibition of LR formation in the osnar2.1 mutants (compared to WT plants) under low NO 3 − concentrations. 15 N-labelling of NO 3 − supplies (provided at concentrations of 0-10 mM) and our localised-NO 3 − treatments showed that defective LR formation in OsNAR2.1 knockdown plants may be due (i) an impairment of NO 3 − uptake, and (ii) a NO 3 − -signalling function.

Results
Knockdown of OsNAR2.1 inhibited root growth when N was supplied as NO 3

−
. Total root lengths were shorter in osnar2.1 mutants than in WT plants at a NH 4 + :NO 3 − concentration ratio of 25:75. The difference in root length between the genotypes was greatest when the ionic ratio was 0:100 ( Supplementary Fig. S1 online); mutant line total root lengths were ca. 30% of those in WT plants at this concentration ratio. However, total root lengths were similar between genotypes when NH 4 + -N was supplied (Fig. 1a,b). Reductions in total root lengths in the mutants were largely attributable to reduced total LR lengths (Fig. 1c). The total LR lengths of seminal and adventitious roots in the RNAi lines were ca. 64 and 53%, respectively, of those in WT plants when NO 3 − -N was supplied. The LR density in the RNAi lines was reduced to a greater extent than the mean LR length (Fig. 1c, Supplementary Fig. S2). The LR densities of the seminal and adventitious roots in the two osnar2.1 mutants were reduced by ca. 29% and 40%, respectively, in comparison with those in the WT plants (Fig. 1d). The lengths of the seminal and all adventitious roots in the osnar2.1 mutants were ca. 15% shorter than those of WT plants when NO 3 − -N was supplied (Fig. 1b). However, the numbers and mean lengths of adventitious roots were not different from those of WT plants ( Supplementary Fig. S3 online).

Knockdown of OsNAR2.1 repressed LR initiation when N was supplied as NO 3
− . Two-week-old rice seedlings had seminal roots that were longer than the adventitious roots. Our preliminary experiment showed that the response of seminal root LRs to N treatment was similar to that of adventitious root LRs. We therefore chose seminal roots to represent root responses in our study of the effects of OsNAR2.1 on the rice root system. Compared to WT plants, the number of LR primordia was reduced markedly in the RNAi lines when NO 3 − -N was supplied (Fig. 2a). When NH 4 + -N was supplied, we observed no difference in LR development between the WT and the mutants (Fig. 2a,c). When NO 3 − -N was supplied, our microscopic observations detected significant differences between genotypes in (i) the numbers of unemerged and emerged LR primordia and (ii) the numbers of LRs at distances 0-8 cm behind the seminal root tips (Fig. 2b). A marked decrease in the number of LR primordia (number of primordia from the first division of the pericycle cells to emergence) resulted in a 23% reduction in the LR numbers in 6-8 cm-lengths of roots in the mutants (compared to roots in the WT plants) after incubation for 7 d with a NO 3 − -N supply.
N accumulation in the osnar2.1 mutants was inhibited across a wide range of NO 3 − concentrations. We determined whether defective LR formation in the osnar2.1 mutants under NO 3 − supply was attributable to the inhibition of N uptake, by quantifying the level of N limitation experienced by rice seedlings through measurements of the cumulative uptake of 15 N-NO 3 − during the entire 7 d period following transfer to media containing a range of labelled NO 3 − concentrations (Fig. 3a,b). Compared to WT plants, the mutants had reduced N accumulations in the shoots and the roots; reductions ranged between 61% and 18% at external NO 3 − concentrations of 0.05-10 mM.
Lateral root formation responses to reduced root N accumulation. We investigated LR formation in rice plants after transfer from medium containing 10 mM NO 3 − to media containing a range of NO 3 − concentrations (0-10 mM). After 7 d, the elongation of seedling seminal and adventitious roots had responded little to the treatments (Fig. 1, Supplementary Fig. S3), but subsequent LR formation was strongly affected by external NO 3 − concentrations (Fig. 4a). Lateral root numbers on the seminal roots of WT plants increased with increasing external NO 3 − concentration (0.05-2 mM), but decreased with continued increases in external NO 3 − concentrations above 2 mM (Fig. 4a), in agreement with our previous findings 29 Lateral root development is modulated by internal plant N status 11,16,18,31 . A plot of the number of new LRs formed during the experimental period against the internal root N content in WT plants (Fig. 4b) showed clearly that LR numbers increased with internal root N content. N accumulation increased when the concentration in the root was less than ca. 780 μ mol g −1 (external NO 3 − ≤ 2 mM); however, LR formation was inhibited when the internal N content in the root was greater than ca. 780 μ mol g −1 (external NO 3 − ≥ 2 mM). A similar trend was detected when we plotted the number of newly formed LRs in the osnar2.1 knockdown mutants against a range of internal root N contents. These results suggest that the inhibition of LR initiation in osnar2.1 mutants at a low NO 3 − concentrations is due to impaired NO 3 − uptake. However, defects in LR formation of osnar2.1 mutants were not completely explainable by reduced NO 3 − uptake because when internal N concentrations in the roots of WT plants and mutants were similar, the mutants still had fewer visible LRs than the WT plants.

LR formation was less sensitive to localised NO 3
− supply in the osnar2.1 mutants than in WT plants. Our data to this point suggested that OsNAR2.1 may be involved in a signalling function; we therefore examined the responses of seminal root LRs to localised NO 3 − supply in WT plants and osnar2.1 mutants. LR densities on seminal roots provided with a localised NO 3 − supply were 33% lower in osnar2.1 mutants than in WT plants (Fig. 5a,b). However, LR densities were similar on seminal roots of mutants and WT plants when we Knockdown of OsNAR2.1 inhibited auxin transport from the shoot to the root when N was supplied as NO 3

−
. Several plant hormones control LR formation; auxin plays a pivotal role 32,33 . To determine whether auxin affected LR formation in osnar2.1 mutants, we analysed endogenous IAA concentrations in the first leaf and in the root. IAA concentrations in the first leaf of the two osnar2.1 mutants supplied with NO 3 − were elevated by ca. 62% in comparison with WT seedlings; conversely, IAA concentrations in the roots of two osnar2.1 mutants were ca. 50% lower than those in WT plants (Fig. 6a). When NH 4 + -N was supplied, auxin distributions were similar between genotypes (Fig. 6b). Thus, knockdown of OsNAR2.1 probably inhibited auxin polar transport from shoots to roots when NO 3 − -N was supplied. We therefore conducted [ 3 H]IAA transport assays. When NO 3 − -N was supplied to the two mutant seedlings, [ 3 H]IAA transport from their shoots to their roots was reduced significantly, and the [ 3 H]IAA activity in the roots was reduced in consequence. These data confirm that auxin polar transport from shoots to roots was inhibited in osnar2.1 mutants supplied with NO 3 − as an N source.

Exogenous application of NAA recovered LR initiation in osnar2.1 mutants. Application of NAA
(1 nM) counteracted the effects of OsNAR2.1 knockdown on LR primordium numbers and LR density when N was supplied as NO 3 − (Fig. 7a,b). LR initiation in osnar2.1 mutants supplied with NO 3 − and treated with NAA was similar to LR initiation in WT plants supplied with NO 3 − as the N source. Furthermore, exogenous NAA application enhanced LR initiation in WT rice seedlings supplied with NO 3 − . These findings were concordant with changes in DR5::GFP expression levels in seminal root tips (Fig. 7c). Thus, reduced IAA concentrations in the two osnar2.1 mutants played a key role in the inhibition of LR initiation.

Knockdown of OsNAR2.1 reduced expression levels of PINs in the roots. Most auxin transport
occurs via the polar transport stream, which is facilitated by proteins of the PIN family 34 . Our qRT-PCR analysis showed that the expression levels of PIN1c, PIN2, PIN9, and PIN10a-b in the roots subjected to NO 3 − treatments were significantly lower in the two osnar2.1 mutants than in the WT rice seedlings (Fig. 8a). When NH 4 + -N was supplied, PIN expression levels were similar between the two osnar2.1 mutants and the WT (Fig. 8b).

Discussion
Mounting evidence shows that LR development is highly plastic and responsive to N treatments 1,4,11,17,20,29,30,35 . The striking effects of localised-NO 3 − and high NO 3 − media on LR growth have been investigated extensively in the model plant Arabidopsis 5,10,11,15 ; however, the mechanism of N-dependent LR growth in rice remains unclear. We observed inhibition of LR formation in the osnar2.1 mutants (compared to WT plants) subjected to low NO 3 − concentrations, but not in those subjected to low NH 4 + concentrations. Knockdown of OsNAR2.1 reportedly impairs NO 3 − uptake at a NO 3 − concentration of 0.2 mM; the concentration of N in the mutants was only ca. 65% of that in the WT plants, but no differences were observed among genotypes at an NH 4 + concentration of 0.2 mM 27 . Our study is the first to determine whether the inhibition of LR initiation in osnar2.1 mutants at low NO 3 − concentrations is due to impaired NO 3 − uptake. When we plotted LR number under varying external NO 3 − concentrations against root N content, we found that trends in the relationship were similar between mutants and WT plants: LR numbers increased with root N content up to a concentration of 700-780 μ mol g −1 ; above this concentration, LR numbers decreased (Fig. 4b). Under external NO 3 − ( 15 N-NO 3 − ) concentrations of 5 and 10 mM, impaired NO 3 − uptake in the osnar2.1 mutants resulted in root N accumulations of 700-780 μ mol g −1 . Therefore, similar LR formations in WT plants and the two mutants were not unexpected at high NO 3 − concentrations (Fig. 4a). Reduced root N accumulation in osnar2.1 mutants subjected to low external NO 3 − concentrations inhibited LR formation (in comparison with WT rice plants). Overall, these results suggest that the knockdown of OsNAR2.1 probably reduced LR formation under low external NO 3 − concentrations via impairment of NO 3 − uptake. Interestingly, inhibition of LR growth by OsNAR2.1 knockdown cannot be fully explained by reduced N uptake in the mutants. When internal root N contents in the roots of the WT and mutant plants were similar across the entire range of N contents (i.e. under external 15 N-NO 3 − concentrations ranging from 0.05 to 10 mM), the osnar2.1 mutant plants always had significantly fewer LRs than the WT plants; e.g. when internal root N contents reached 520 μ mol g −1 , LR numbers in the two mutants were 33% fewer than those in the WT plants. Furthermore, the LR formation response to localised NO 3 − supply was weaker in the osnar2.1 mutants than in WT plants (Fig. 5). Thus, the NO 3 − -responsive OsNAR2.1 gene may be involved in a signalling pathway controlling LR formation in environments supplied with NO 3 − -N. The expression levels of OsNRT2.1, OsNRT2.2, and OsNRT2.3a in the roots of two osnar2.1 mutants are reportedly much reduced in comparison with the abundant expression levels in WT plants 27 . Recent evidence suggests that in the model plant Arabidopsis, AtNRT1.1 (CHL1) is a NO 3 − sensor that activates the ANR1-mediated NO 3 − -signalling pathway to regulate LR proliferation 5,11,[35][36][37] . Furthermore, AtNRT2.1 may have a direct stimulatory role in LR initiation under low NO 3 − concentrations 19 . Interestingly, AtNAR2.1 may also participate in the signalling pathway that integrates nutritional cues for LR proliferation by targeting AtNRT2.1 at the plasma membrane 20,21 . Therefore, it is reasonable to postulate that the inhibition of LR initiation in the osnar2.1 mutants  osnar2.1 knockdown lines (r1 and r2). Rice seedlings were grown for 1 wk in hydroponic medium containing 10 mM NO 3 − and then transferred for a further week to media containing 0-10 mM 15  is also related to the interaction of proteins affected by the knockdown of OsNAR2.1. The transcript levels of OsNAR2.1 and the three NO 3 − transporters are rapidly induced by NO 3 − supply, with peak levels occurring 1-2 h after the initiation of treatments 25,27,38,39 . Furthermore, OsNAR2.1, OsNRT2.1, and OsNRT2.2 are expressed abundantly throughout the primary and lateral roots [25][26][27] , and OsNRT2.3a is expressed abundantly in the root stelar cells, particularly in the xylem parenchyma tissue 40 . More experiments are required to elucidate the participation of OsNAR2.1 regulatory networks in rice LR formation under low NO 3 − concentrations. Plants adjust their growth and development in response to changing environmental conditions through the perception and integration of external signals into the signalling pathways of plant hormones, such as auxin 12,13,33,[41][42][43][44][45] . Auxin plays dominant roles in the specification of the founder cells that initiate LR formation and in the later stages of LR development 32,33 . Diverse environmental and endogenous signals may be integrated to mediate changes in auxin distribution via effects on polar transport 46,47 . Lateral root growth is not stimulated by localised NO 3 − supply in the auxin-insensitive mutant axr4, which suggests an overlap between the auxin and NO 3 − signalling pathways 6 . These two signalling pathways are further linked through an effect on auxin transport via AtNRT1.1 37 .

. Changes in lateral root (LR) numbers in response to varying external and internal nitrate concentrations in wild-type (WT) and
In our study, (i) elevated IAA concentrations in mutant leaves, (ii) reduced IAA concentrations in mutant roots, and (iii) repressed mutant [ 3 H]IAA transport from the shoot to root (in comparison with WT plants) suggested that auxin polar transport was inhibited by the knockdown of OsNAR2.1. These effects were correlated with a decrease in the transcript levels of five PIN genes in mutant roots (relative to WT plants). Exogenous NAA application restored LR initiation and DR5::GFP expression levels in seminal root tips of osnar2.1 mutants to levels similar to those in WT plants, further demonstrating that a reduced auxin concentration contributed to repressed LR initiation in the osnar2.1 mutants. OsNAR2.1 expression is dominant in roots and minimal in shoots 25,38 . Furthermore, in comparison with normal nutrient supply conditions, IAA transport is reduced in rice when N supply is limiting 29,30 . Thus, the repression of [ 3 H]IAA transport from the shoot to the root in the osnar2.1 mutants (relative to WT plants) under low external NO 3 − concentrations can probably be attributed to N starvation system signals that inhibited auxin polar transport from the shoot to the root.
In conclusion, OsNAR2.1 had a key function in coordinating LR formation at low external NO 3 − concentrations. Its effects on LR formation most likely operated through a combination of roles in both NO 3 − uptake and NO 3 − signalling.

Methods
Plant material and growth conditions. The Nipponbare rice ecotype was used as the WT. The osnar2.1 mutant lines (T2 generation) obtained from RNA interference were reported in previous work 27 . Plants were grown in a greenhouse under natural light at day/night temperatures of 30 °C/18 °C. Seven-day-old seedlings of uniform size and vigour were transplanted into holes in a lid placed over the tops of 7-L pots (four holes per lid and three seedlings per hole). Nutrient solutions varying from one-quarter to half strength were applied for 2 d, and then full-strength nutrient solution was applied for a further week. Pots receiving NH 4 + or NO 3 − were filled Measurement of root system architecture. The rice root system comprises seminal and adventitious roots, each bearing LRs 48 . We used previously-described LR developmental stage categories 3 ; stages I-XII were grouped here into the "unemerged primordia" category. The primordia of LR were classified into unemerged and emerged primordia. An emerged LR primordium > 0.5 mm long (visible to the unaided eye) was classified as a LR, and the primordium was considered to have been activated 9 .
To visualise the development of LRs, we stained the seminal roots with methylene blue 49 . These roots were fixed in formalin/acetic acid/alcohol (FAA) solution (10:5:85 v/v/v) at 4 °C for at least 24 h. After fixation, LR primordia were rinsed for 10 min in water and then stained with a 0.01% w/v methylene blue solution. After the roots had been stained, counting the numbers of LR primordia and LRs was straightforward. The scaleplate in the stereomicroscope we used (Olympus Optical Co. Ltd, Tokyo, Japan) simplified determinations of the lengths of emerged primordia and LRs. The lengths of seminal and adventitious roots were measured using a ruler, and LR density was calculated by dividing the LR number by root length. Total root length and LR length were measured using the WinRhizo scanner-based image analysis system (Regent Instruments, Montreal, QC, Canada).
Lateral root responses to localised NO 3 − supply. Rice seeds were germinated in trays over 2 d and then transferred into plant culture dishes (23 × 23 cm) containing 0.6% Phytagel ™ media. Culture dishes were divided The full chemical composition in the procedures using Phytagel medium was similar to that used in the hydroponic medium. To exclude the potential effects of calcium (Ca 2+ ) on the treatments, the solutions in the three segments were supplemented with Ca 2+ (as CaCl 2 ) up to concentrations matching those in the NO 3 − treatments.
Measurement of 15 N concentration. 15 N concentration was assayed in plants grown hydroponically 27 . After grinding in liquid N 2 , one aliquot of powder was dried to a constant weight at 70 °C. Approximately 6 mg powder from each sample were analysed using an Isotope Ratio Mass Spectrometer system (Thermo Fisher Scientific). To analyse the change in LR growth in response to the nutrient solution containing varying NO 3 − concentrations, the same experiment without NO 3 − labelling was performed at the same time. After harvest, LR number in the seminal root was analysed through recording of the LR number before and after the transfer of rice plants to nutrient solutions containing various NO 3 − concentrations. Determination of IAA. We measured the concentrations of IAA in the first leaf and in the root 9 . Fresh weights of the samples were measured, after which specimens were immediately frozen in liquid N 2 . We performed sample measurement of free IAA by high-performance liquid chromatography. A standard IAA sample was obtained from Sigma-Aldrich (St. Louis, MO, USA). The pDR5::GFP construct was transformed into WT plants and osnar2.1 mutants using Agrobacterium tumefaciens (strain EHA105) to determine the patterns of IAA distribution in rice plants. To construct the pDR5::GFP vector, we amplified a 720-bp cDNA fragment of green fluorescent protein (GFP) from the cloning vector pSAT6A-EGFP-N1 using the following primer set: FP, GGATCCATGGTGAGCAA GGGCGAGGAGCT; RP, GAGCTCTCACTTGTACAGCTCGTCCATG. This fragment was inserted into the vector of pDR5::GUS at the BamHI and SacI sites. The pDR5::GUS construct was kindly provided by the Ping Wu laboratory at Zhejiang University, Hangzhou, China. We analysed the fluorescence of GFP in the cells using 543-nm helium-neon and 488-nm argon lasers using a confocal laser scanning microscope (LSM410; Carl Zeiss, Oberkochen, Germany). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis. Total RNA was isolated from the roots of rice seedlings. RNA extraction, reverse transcription, and the qRT-PCR procedures followed a previously-reported procedure 50 . The primer sets targeting the PIN genes are listed in Supplementary  Table S1. Data analysis. Data from experiments were pooled for calculation of means and standard errors (SE) and analysed by one-way ANOVA followed by the LSD test at P ≤ 0.05 to determine the statistical significance of the differences between individual treatments. All statistical evaluations were conducted using the SPSS (version 11.0) statistical software (SPSS Inc., Chicago, IL).