Regulation of Root Nitrate Uptake at the NRT2.1 Protein Level in Arabidopsis thaliana*

In Arabidopsis the NRT2.1 gene encodes a main component of the root high-affinity nitrate uptake system (HATS). Its regulation has been thoroughly studied showing a strong correlation between NRT2.1 expression and HATS activity. Despite its central role in plant nutrition, nothing is known concerning localization and regulation of NRT2.1 at the protein level. By combining a green fluorescent protein fusion strategy and an immunological approach, we show that NRT2.1 is mainly localized in the plasma membrane of root cortical and epidermal cells, and that several forms of the protein seems to co-exist in cell membranes (the monomer and at least one higher molecular weight complex). The monomer is the most abundant form of NRT2.1, and seems to be the one involved in \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{NO}_{3}^{-}\) \end{document} transport. It strictly requires the NAR2.1 protein to be expressed and addressed at the plasma membrane. No rapid changes in NRT2.1 abundance were observed in response to light, sucrose, or nitrogen treatments that strongly affect both NRT2.1 mRNA level and HATS activity. This suggests the occurrence of post-translational regulatory mechanisms. One such mechanism could correspond to the cleavage of NRT2.1 C terminus, which results in the presence of both intact and truncated proteins in the plasma membrane.

The NRT2.1 gene of Arabidopsis thaliana is part of a small multigene family comprising 7 members, which with the exception of NRT2.7, are predominantly expressed in the roots (1,2). NRT2 genes are found in a large variety of organisms (fungi, certain yeasts, green algae, and plants) and belong to the nitrate nitrite porter family of transporter genes (3).
It is generally assumed that NRT2 genes encode high-affinity nitrate (NO 3 Ϫ ) or nitrite transporters (4 -11), and that in higher plants, they play a key role in the root high-affinity transport system (HATS), 3 which ensures uptake of NO 3 Ϫ from the soil solution (3,(12)(13)(14). In all plant species investigated to date, the NO 3 Ϫ HATS displays a saturable activity, with a V max generally reached for NO 3 Ϫ concentrations comprised between 0.2 and 0.5 mM (3,12,14). Although the functional characterization of almost all higher plant NRT2 transporters remains to be done, it is now well documented that NRT2.1 is a major component of the HATS in A. thaliana as shown by the fact that (i) of the seven NRT2 members, only NRT2.1 transcript abundance showed significant correlation (r 2 ϭ 0.74) with HATS activity (2,11) and (ii) several mutants disrupted for the NRT2.1 gene (15,16) or for both NRT2.1-NRT2.2 genes (17,18) have lost up to 75% of the high-affinity NO 3 Ϫ uptake activity. As a consequence, growth of these mutants is severely impaired at low NO 3 Ϫ concentration (15,19,20), but not at high NO 3 Ϫ concentration when low-affinity transporters, most probably those of the NRT1 family (3,12,14), are active.
Despite its firmly established role in root NO 3 Ϫ uptake, several aspects of NRT2.1 function remain enigmatic. First, unlike the Aspergillus nidulans or Chlorella sorokiniana NRT2 transporters (CRNA or ChNRT2.1, respectively) (21,22), but similarly to NRT2 proteins of Chlamydomonas reinhardtii, NRT2.1 does not seem to be able to mediate NO 3 Ϫ transport on its own. In Xenopus oocytes, CrNRT2 and AtNRT2.1 (as well as its Hordeum vulgare homologue, HvNRT2.1) need to be co-expressed with a NAR2 protein, to yield NO 3 Ϫ transport (19,23,24). This has suggested that the actual transport system corresponds in fact to a dual component (NRT2/NAR2) transporter (19,24). Indeed, a crucial role of the NRT2.1 putative partner, NAR2.1 (also called NRT3.1), is confirmed by the observation that mutants disrupted in the NAR2.1 gene (At5g50200) display an even stronger defect in HATS activity than the NRT2.1 mutants in Arabidopsis (19,25). A second surprising aspect of NRT2.1 function is that it seems to be involved in the control of lateral root initiation, in a way that is independent from its transport activity (16,26). Because NO 3 Ϫ is not only a major nitrogen source for nutrition of the plants, but also acts as a signal to modulate plant metabolism and development (27,28), this gave rise to the hypothesis that NRT2.1 may also be a NO 3 Ϫ sensor, or a signal transducer (16).
The regulation of NRT2.1 expression has been thoroughly investigated at the mRNA level. NRT2.1 transcript accumulation predominantly occurs in epidermis and cortex of the mature root regions (29), and is affected by a wide range of environmental changes. Expression of NRT2.1 is induced by low NO 3 Ϫ concentration (2,4,11,30), feedback repressed by NH 4 ϩ and amino acids (11, 29 -31), and stimulated by light and sugars (30,32). These mechanisms are postulated to modulate root NO 3 Ϫ uptake as a function of both nitrogen and carbon status of the plant. More recently, NRT2.1 has been shown to be down-regulated by NO 3 Ϫ itself, through a mechanism independent of the feedback repression exerted by nitrogen metabolites, but specifically triggered by the dual-affinity NRT1.1 NO 3 Ϫ transporter (33,34). The NO 3 Ϫ HATS is subjected to the same controls, and in all cases, a strong correlation was found between the changes in NRT2.1 transcript level and those in NO 3 Ϫ HATS activity, suggesting that the transcriptional regulation of NRT2.1 expression plays a major role in governing root high-affinity NO 3 Ϫ uptake. However, a yet limited number of reports suggest that posttranscriptional regulation of NRT2 transporters may also participate to the modulation of root NO 3 Ϫ uptake in response to environmental changes. In Nicotiana plumbaginifolia ectopic overexpression of the NpNRT2.1 gene did not prevent the inhibition of root NO 3 Ϫ uptake by exogenous NH 4 ϩ supply (35). Likewise, in barley NH 4 ϩ accumulation in roots of plants treated with the glutamine synthetase inhibitor methionine sulfoximine decreased root NO 3 Ϫ uptake but did not change HvNRT2.1 transcript level (36). Further evidence for posttranslational control of NRT2 transporters is provided by the recent report on YNT1 in the yeast Hansenula polymorpha, showing that this protein undergoes trafficking to the vacuole and proteolysis in response to glutamine supply (37).
To gain further insights on NRT2.1 at the protein level in Arabidopsis we initiated the biochemical characterization of its localization and regulation, by combining a GFP fusion strategy and an immunological investigation. Our results provide strong evidence that NRT2.1 is predominantly localized in the plasma membrane of root cortical and epidermal cells of the mature regions of the roots, and show that NAR2.1 is essential for NRT2.1 expression. They also reveal that several forms of the protein seem to co-exist in the plasma membrane. Furthermore, treatments, which rapidly modulate both NRT2.1 transcript level and NO 3 Ϫ HATS activity, are shown to yield only very slow responses of the NRT2.1 protein, suggesting the occurrence of important post-translational regulatory mechanisms. One such putative mechanism is unraveled, which could be associated with partial proteolysis of NRT2.1 resulting in the cleavage of the C terminus of the protein.

EXPERIMENTAL PROCEDURES
Production of the P NRT2.1 -NRT2.1-GFP Construct-Cloning of P NRT2.1 -NRT2.1 (3114 bp, including the 1335-bp 5Ј untranslated region and promoting sequence upstream the ATG) and fusion with GFP coding sequence at the 3Ј end of NRT2.1 was performed using Gateway TM Technology, according to the manufacturer's instructions (Gateway cloning manual, Invitrogen). The primers NRT2.1 GATE forward (CACCCACGTCA-GCGAGATTGATCG) and NRT2.1 GATE reverse (AACATT-GTTGGGTGTGTTCTCAGGC) were used to amplify the P NRT2.1 -NRT2.1 complete DNA sequence from the bac clone T6D22 (ARBC, Columbus, OH). After gel purification of the PCR product with the Nucleo Spin Extract Kit (Machery Nagel), P NRT2.1 -NRT2.1 was cloned into the pENTR TM / D-Topo vector (Invitrogen) to create an entry clone. After transformation of One Shot TOP10 thermocompetent Escherichia coli (Invitrogen) the vector was sequenced. LR (Invitrogen) reaction was performed to transfer P NRT2.1 -NRT2.1 from entry clone to the destination binary vector pGWB4 (no promoter, C-sGFP) obtained from Tsuyoshi Nakagawa (Research Institute of Molecular Genetics, Shimane University, Matsue, Japan). Following the LR reaction thermocompetent DH5␣ E. coli were transformed and positive clones were selected with hygromycin. Prior transformation of Agrobacterium, part of the expression construct was sequenced to verify the translational fusion of P NRT2.1 -NRT2.1 with the GFP tag. In addition we used plants, called P43NRT2.1, transformed with NRT2.1 fused to GFP in N-terminal, under the control of the 35 S promoter as described in Chopin et al. (38).
Plant Transformation-Binary vectors containing the GFP fusion construct were introduced into Agrobacterium tumefaciens strain GC3101. A. thaliana nrt2.1-1 mutant plants, ecotype Wassilewskija (18), were transformed by dipping the flowers in the presence of Silwet L77 (39). The transformants were selected on a medium containing 30 mg/liter of hygromycin. For further analyses, T1 segregation ratios were analyzed to select transformants with one T-DNA insertion and to isolate T3-homozygous plants.
Growth Conditions-For all experiments, except those devoted to confocal imaging, plants were grown hydroponically using the experimental set-up described previously (30). Briefly, seeds were sown directly on the surface of wet sand in modified 1.5-ml microcentrifuge tubes, with the bottom replaced by a metal screen. The tubes supporting the seeds were placed on polystyrene floating rafts, on the surface of a 10-liter tank filled with tap water. The culture was then performed in a controlled growth chamber with 8 h/16 h day/night cycle at 24/20°C. Light intensity during the light period was at 250 mol m Ϫ2 s Ϫ1 . One week after sowing, the tap water was replaced by nutrient solution until the age of 6 -7 weeks depending on the size of the plants. The basal nutrient solutions supplied to the plants are those described by Gansel et al. (31) and contained either 0.3 mM NO 3 Ϫ , 1 mM NO 3 Ϫ , 5 mM NO 3 Ϫ , or 10 mM NH 4 NO 3 as nitrogen source. The nutrient solution was replaced every week during this period and the day before the experiment.
For confocal imaging, plants were grown in sterile conditions in vertical agar plates (12 ϫ 12 cm) on the same basal medium as used for hydroponic cultures plus 2.5 mM MES and 1.2% (w/v) agar type A (Sigma, product A4550). It contained 1 mM NO 3 Ϫ as nitrogen source and either 3% sucrose (w/v) or no sugar. The pH was adjusted to 5.8 with KOH. After sowing, the plates were transferred in a growth chamber with 16 h/8 h day/ night cycle at 21/18°C and 70% relative humidity. Light intensity during the light period was at 125 mol m Ϫ2 s Ϫ1 . Observations were performed after 14 days of growth.
In Vivo Protein Cross-linking-According to Rohila et al. (40) in vivo protein cross-linking was performed by adding formaldehyde to the nutrient solution of hydroponically grown plants to get a final concentration of 1% (v/v). Plants were treated during 45 min before harvesting the roots for plasma membrane (PM) extraction.
Confocal Microscopy-GFP images were acquired with a Zeiss LSM 510 axiovert 200M inverted microscope. GFP (excitation/emission maxima ϳ488/507 nm) was excited with the 488 nm line of an Argon laser and detected via a 505-530 nm bandpass filter (green). Autofluorescence was detected via a 650-nm long pass filter (red). Dichroic mirrors used were HFT 488 and NFT 545.
To visualize the different plant cell membranes, several stains were used. Images of the plasma membrane and the tonoplast were obtained after 5 min and at least 16 h of incubation with 8.2 M of the endocytic tracer FM4-64 (Invitrogen, product F34653), respectively. The endoplasmic reticulum was visualized after 30 min staining with 5 M endoplasmic reticulum (ER) Tracker (Blue-White DPX, Invitrogen). FM4-64 was visualized with the microscope used for GFP imaging. FM4-64 (excitation/emission maxima ϳ515/640 nm) was excited with the 543-nm line and detected via LP 650 and LP 585, respectively. Dichroic mirrors used were HFT 488/543 and NFT 545. ER Tracker (excitation/emission maxima ϳ347-640 nm) images were obtained with the Zeiss LSM 510 Meta Axioplan 2 microscope. ER Tracker was excited with the diode 405-nm laser for blue dye and detected via a 420 -480 BP filter (blue). Dichroic mirrors used were HFT 405/488/543 and NFT 490.
NRT2.1 Immunodetection and Membrane Purification-Root total proteins were extracted as described by Santoni et al. (41). Microsomes were prepared as described by Giannini et al. (42) and plasma membrane vesicles were purified from microsomes by aqueous two-phase partitioning, as described by Santoni et al. (43).
For ELISA, serial 2-fold dilutions in a carbonate buffer (30 mM Na 2 CO 3 , 60 mM NaHCO 3 , pH 9.5) of 500 ng of PM proteins were loaded in duplicate on Maxisorb immunoplates (Nunc, Denmark) and left overnight at 4°C. The immunodetection was performed according to the manufacturer's instructions. The primary anti-NRT2.1 20 antibody (1:2500 dilution) and a secondary peroxidase-coupled anti-rabbit antibody were successively applied for 2 h at 37°C. A linear regression between the absorbance signal due to oxidized 2,2Ј-azinobis-3-ethylbenzothiazoline-6-sulfonic acid diammonium salt, as read with a multiplate reader (Victor, PerkinElmer Life Sciences), and the amount of proteins was obtained for each sample and used for relative comparison between samples.
RNA Extraction and Reverse Transcription-RNA extraction was performed on roots as described previously (44) using guanidine hydrochloride and lithium chloride. Subsequently 40 g of RNA were treated with DNase (RNase Free DNase Kit, Qiagen) and purified (RNeasy MinElute TM Cleanup Kit, Qiagen) following the manufacturer's instructions. The absence of genomic DNA was verified by PCR using specific primers spanning an intron in the gene APTR (At1g27450).
Reverse transcription was performed with 4 g of purified RNA and oligo(dT) 18 primers. The mixture was heated for 5 min at 72°C and progressively (Ϫ1°C per 10 s) cooled down to allow hybridization of the primers. The reaction was carried out in a volume of 20 l in the presence of 200 units of Moloney murine leukemia virus reverse transcriptase (Promega) at 42°C during 90 min. The quality of the cDNA was verified by PCR using the primers for the gene APTR.
Quantitative PCR-Real-time amplification was performed in a LightCycler (Roche Diagnostics) with the kit SYBR Green (LightCycler FastStart DNA Master SYBR Green1, Roche Diagnostics) according to the manufacturer's instructions with 1 l of cDNA in a total volume of 10 l. The following conditions of amplifications were applied: 10 min at 95°C; 45 cycles of 5 s at 95°C; 7 s at 65°C; and 8 s at 72°C. A melting curve was then performed to verify the specificity of the amplification. Successive dilutions of one sample were used as a standard curve. Amplification efficiency was around 1. All the results presented were standardized using the housekeeping gene Clathrin (At4g24550). The primers used were: NRT2.1 forward, AACA-AGGGCTAACGTGGATG; NRT2.1 reverse, CTGCTTCTCC-TGCTCATTCC; Clath forward, AGCATACACTGCGTGCA-AAG; Clath reverse, TCGCCTGTGTCACATATCTC.
Root NO 3 Ϫ Influx Measurements-Root 15 NO 3 Ϫ influx was assayed as described by Dehlon et al. (45). Briefly, the plants were sequentially transferred to 0.1 mM CaSO 4 for 1 min, to complete nutrient solution, pH 5.8, containing 0.2 mM 15 NO 3 Ϫ (99 atom % excess 15 N) for 5 min, and finally to 0.1 mM CaSO 4 for 1 min. Roots were then separated from shoots, and the organs dried at 70°C for 48 h. After determination of their dry weight, the samples were analyzed for total nitrogen and atom % 15 N using a continuous flow isotope ratio mass spectrometer coupled with a C/N elemental analyzer (model ANCA-MS; PDZ Europa, Crewe, UK) as described in Clarkson et al. (46). Each influx value is the mean of 6 to 12 replicates.

RESULTS
Tissular and Subcellular Localization of NRT2.1-To investigate the localization of NRT2.1 in the roots, NRT2.1-GFP transgenic lines were generated by expressing a GFP-tagged NRT2.1 protein (C-terminal translational fusion) in the nrt2.1-1 knock-out mutant of NRT2.1. The GFP coding sequence was fused in-frame to the 3Ј end of the NRT2.1 gene under the control of its own promoter. Two transgenic lines (GFP10 and GFP12) displayed both a correct regulation of the expression of the transgene and a functional complementation of the mutant phenotype (supplemental materials Fig. S1). In these lines, the expression of P NRT2.1 -NRT2.1-GFP was induced by both light and sucrose in the roots, as it was the case for NRT2.1 in WT plants. The NO 3 Ϫ HATS activity, determined by root 15 NO 3 Ϫ influx assays (0.2 mM external 15 NO 3 Ϫ concentration), was also stimulated by the light/sucrose treatments in both GFP10 and GFP12 plants, and restored at higher values than those measured in the nrt2.1-1 mutant.
Cell-type specific expression of NRT2.1-GFP was therefore studied by confocal microscopy in both GFP10 and GFP12 plants grown in vitro. In roots of GFP10 plants, GFP fluorescence was mainly localized in the cortical cells along the primary and secondary roots (Fig. 1, A and C), except in the apical part of these roots where GFP was not present (Fig. 1B). No fluorescence was detected in the stele of the roots (Fig. 1C), and in the leaves (results not shown). These results were confirmed using the GFP12 line (data not shown). In plants grown hydroponically, NRT2.1-GFP was also found in mature regions of the roots, where in addition to the cortex, it was also expressed in the epidermal cells, and in root hairs (Fig. 1D).
The subcellular localization of the NRT2.1-GFP protein was investigated in roots of plants grown in vitro, using specific markers for cellular membranes. The PM and the tonoplast were visualized after a short and a long incubation with the red fluorescent dye FM4-64, respectively (47). After a short incubation time (5 min), FM4-64 fluorescence co-localized with that recorded from NRT2.1-GFP, yielding a yellow staining of the PM in the overlaid image ( Fig. 1, E-G). After a prolonged incubation time (16 h), FM4-64 fluorescence had entirely moved to the tonoplast and became clearly distinguishable from the GFP fluorescence in the overlaid image ( Fig. 1, H-J). Finally, to test the presence of NRT2.1-GFP in the ER, plants were treated with the vital ER Tracker blue-white DPX. Images showed blue staining of the ER around the nucleus, which co-localized with NRT2.1-GFP fluorescence in the overlaid image ( Fig.  1, K-M). Despite the fact that the spatial resolution of these images may not be sufficient to allocate NRT2.1 to one specific membrane, they do show that NRT2.1 is not in the tonoplast, and suggest that this protein is localized in both the PM and ER.
Immunochemical Characterization of NRT2.1 in Root Cell Membranes-To further investigate NRT2.1 at the protein level, two specific polyclonal antibodies, called anti-NRT2.1 19 and anti-NRT2.1 20, were raised in rabbit against specific peptidic sequences within, respectively, an internal loop and the C terminus of the protein ( Fig. 2A). The affinity-purified anti-NRT2.1 antibodies were tested on Western blots with total microsomal membranes purified from roots of hydroponically grown WT plants or nrt2.1-1 and nrt2.1-2 knock-out mutants (16,18). As shown in Fig. 2, B and C, both anti-NRT2.1 antibodies revealed several bands at ϳ45, ϳ75, and ϳ100 -120 kDa. An additional band at ϳ60 kDa was found with the anti-NRT2.1 19 antibody, but not with the anti-NRT2.1 20 antibody. These bands were all specific for NRT2.1 because they were absent in the microsomes from the two nrt2.1 knock-out mutants. For all the bands, the anti-NRT2.1 20 antibody always gave the strongest signal and was thus preferentially used in most of the experiments. In PM-enriched fractions, only the bands at ϳ45 and ϳ100 -120 kDa were visible (Fig. 2D). With both antibodies the strong band observed at ϳ45 kDa was by far the most abundant form of NRT2.1 in both microsomes and PM because it could be readily observed after a few minutes of exposure to film. By contrast, several hours were required to get a visible signal for the other ones. It is likely that the band at ϳ45 kDa corresponds to the monomeric isolated form of NRT2.1. The apparent size of ϳ45 kDa differs from its theoretical molecular mass of 57 kDa. However, this may be due to the hydrophobic nature of the protein, as already seen with previously characterized sugar (48), ammonium (49), and nitrate transporters (50, 51), which all display in Western blots an apparent lower molecular mass than the theoretical one. The bands at ϳ75 and ϳ100 -120 kDa may reveal higher molecular mass complexes incorporating NRT2.1. For simplicity, the highest molecular mass band is referred to as the ϳ120-kDa complex in the following text, although its apparent size was found to vary between 100 and 150 kDa, depending on the blots.
To make sure that the ϳ45-kDa band, corresponding to the isolated form of NRT2.1, is actually present in the native PM, and is not a product resulting from the dissociation of the higher molecular mass complexes, we performed in vivo crosslinking to stabilize protein complexes before PM extraction. Cross-linking was induced by adding 1% formaldehyde to the nutrient solution 45 min prior to harvesting for PM extraction. This markedly improved the detection of the band at ϳ120 kDa (Fig. 2E). However, this did not lead to the disappearance or a decrease in the intensity of the band at ϳ45 kDa, which remained strongly predominant in PM from plants treated with formaldehyde. As a control for the efficiency of protein-protein cross-linking, the same blot was probed with an anti-PIP2 antibody (Fig. 2F). PIP2 is a plasma membrane aquaporin that assembles in membranes as tetramers (52). Without cross-linking, only a monomer at 26 -29 kDa and a dimer at 52 kDa of PIP2 were immunodetected on SDS-PAGE gels (43). By contrast a trimer at 75 kDa and a tetramer at 100 kDa were clearly immunodetected upon in vivo cross-linking. This last result validates the efficiency of formaldehyde-induced cross-linking.
Role of NAR2.1 in the Expression and Localization of NRT2.1-It has been recently suggested that the NO 3 Ϫ HATS in Arabidopsis may actually be a dual component transport system, involving both NRT2.1 and NAR2.1 proteins (19,24,25). Thus, a possibility would be that the high molecular mass bands at ϳ75 and ϳ120 kDa correspond to NRT2.1/NAR2.1 complexes. To investigate this hypothesis, we used the nar2.1-1 knock-out mutant for NAR2.1 (19) (also named nrt3.1-2 in Ref. 25) to perform Western blots on microsomes in comparison with the wild-type Wassilewskija. Before extraction of microsomes, plants were treated with 1% formaldehyde during 45 min to induce in vivo protein cross-linking. The data obtained showed that the band at ϳ120 kDa is only slightly affected in the mutant nar2.1-1 compared with the WT, whereas the band at ϳ75 kDa, and surprisingly the major one at ϳ45 kDa, disappeared completely in the mutant microsomes (Fig. 3, A and B). The absence of the major form of NRT2.1 in the nar2.1-1 mutant was also found using PM fractions (data not shown), and was not due to a defect in NRT2.1 gene expression because in our conditions, the NAR2.1 mutation did not affect NRT2.1 mRNA accumulation in the roots (Fig. 3C). Furthermore, the analysis of transgenic lines expressing a P 35 S-GFP-NRT2.1 transgene either in a wild-type, or in a nar2.1-1 background confirmed this result because despite ectopic expression of the transgene, NAR2.1 mutation resulted in an almost complete loss of GFP fluorescence associated with the PM (Figs. 3, D-L).
Regulation of NRT2.1 Abundance by Light, Sugars, and Nitrogen-To determine whether the abundance of NRT2.1 is regulated like NRT2.1 expression and NO 3 Ϫ HATS activity, ELISA and Western blots were performed with PM or microsomes extracted from WT plants during a light/dark cycle and after various sugar or nitrogen treatments.
To study the regulation of NRT2.1 in response to light and sugars, PM were isolated from plants harvested at the end of a normal light period (5 p.m.), or after 4 h into the night (9 p.m.) with or without 1% sucrose supply (Fig. 4, A-C). These treatments strongly affect both NRT2.1 mRNA accumulation and NO 3 Ϫ HATS activity because both decreases more than 50% after 4 h of darkness, but remain high when sucrose is supplied after the light/dark transition (30, 32) (Fig. 4C). Surprisingly, Western blots did not reveal any significant changes in the abundance of the main form of NRT2.1 (at ϳ45 kDa) between the end of the light period and after 4 h in the dark with or without sucrose (Fig. 4A). Quantitative analysis by ELISA confirmed this result, showing that the total amount of NRT2.1 in the PM was not decreased after 4 h of darkness, and not increased by sugar supply in the dark (Fig. 4B). Furthermore, NRT2.1 abundance did not show any change at the beginning of the light period, when plants were illuminated for 4 h following the usual night period (Fig. 4, D and E). Sucrose supply also failed to increase the NRT2.1 level even during an extended 4-h dark period after the 16 h of normal night (Fig. 4, D and E).
Concerning the band at ϳ120 kDa, it was detected in almost all PM extracts from plants in light or darkness, with roughly the same intensity (Fig. 4, A and D). These results, obtained with both anti-NRT2.1 antibodies, are representative of several independent experiments, leading to the conclusion that the abundance of NRT2.1 (at ϳ45 kDa) is neither regulated during the diurnal cycle, nor by short-term sucrose supply. Nevertheless, the amount of NRT2.1 in root PM did show a slight (but statistically significant) increase in response to light and sucrose supply following a 40-h prolonged period of darkness (supplemental materials Fig. S2). This increase remained, however, much lower than that observed at the mRNA level.
To further investigate the changes in NRT2.1 abundance in response to long-term treatments down-regulating both NRT2.1 expression and NO 3 Ϫ HATS activity, we shifted to the study of the repressive effect of high nitrogen supply to the plants, because we found this more physiologically relevant than several days of extended darkness. Therefore, we compared hydroponically grown plants provided with 1 mM KNO 3 as a nitrogen source, or transferred either to nonrepressive, moderately repressive, or highly repressive media (0.3 mM KNO 3 , 5 mM KNO 3 , or 10 mM NH 4 NO 3 , respectively). These treatments are known to strongly affect NRT2.1 mRNA accumulation within a few hours (11,30,33). A significant difference was observed in the abundance of both the main form of NRT2.1 at ϳ45 kDa and the protein complex at ϳ120 kDa in the root PM between plants supplied with 0.3 mM KNO 3 or 10 mM NH 4 NO 3 for 7 days (Fig. 5, C and D). Both bands were clearly more intense under 0.3 mM KNO 3 conditions, suggest-  ing regulation of the two forms of NRT2.1 by the nitrogen status of the plant. This result was confirmed by ELISA, indicating a 70% decrease in total NRT2.1 signal at 10 mM NH 4 NO 3 , compared with 0.3 mM KNO 3 (Fig. 5E). Time course experiments, however, indicated that a significant decay of NRT2.1 in response to high nitrogen required several days of treatment. Indeed, transfer of the plants from 1 mM NO 3 Ϫ to 10 mM NH 4 NO 3 for 8 h or 3 days resulted in a 20 and 60% decrease in the total amount of NRT2.1, respectively (Fig.  5A). This repressive effect of high nitrogen supply was apparently not specifically due to the presence of NH 4 ϩ in the nutrient solution, because a moderate decrease of NRT2.1 abundance was also observed in response to the supply of 5 mM NO 3 Ϫ as the sole nitrogen source (Fig. 5B). Taken together, these data indicate that, unlike NRT2.1 expression, NRT2.1 abundance is only slowly modified by light, sugar, or nitrogen treatments. This suggests that the NRT2.1 protein is relatively stable in the PM and remains abundant for hours or days after transcription of the gene has been down-regulated. This hypothesis is supported by the fact that the pro-tein synthesis inhibitor cycloheximide has only a slow effect on NRT2.1 abundance in the PM, whereas it quickly reduced both NRT2.1 mRNA level and 15 NO 3 Ϫ influx (supplemental materials Fig. S3).
Cleavage of the C Terminus of NRT2.1-To validate by another independent approach the conclusion that NRT2.1 abundance on PM is not regulated during the day/night cycle, we used GFP10 transgenic plants to determine how the NRT2.1-GFP protein responds to light and sugar. GFP10 plants were grown for 14 days on modified MS medium containing 1 mM KNO 3 with or without 3% sucrose. Plants were pre-treated for 24 h in the light or dark before confocal imaging. Under light conditions, GFP fluorescence was predominantly located in the PM and ER as described above, independently on the presence of sucrose (Fig. 6, A and B). Plants pre-treated for 24 h in the dark with or without sucrose also showed strong GFP fluorescence associated with PM/ER (Fig. 6, C and D). However, GFP fluorescence was also surprisingly strongly observed inside the cortical cells, filling a compartment, that most likely is the vac-  uole (Fig. 6, C and D). Time course studies indicated that when plants were transferred in the dark after 24 h of continuous light, accumulation of GFP in the vacuole was observed after at least 9 h of darkness (Fig. 6, E-G), and increased until 24 h of darkness (Fig. 6H). In the reverse experiment, where plants were transferred to the light after 24 h of darkness, GFP fluorescence, which was observed both inside the vacuole and in the PM/ER of cortical cells at the beginning of the experiment (Fig.  6I), quickly disappeared from the inside of the cells after only 1 h 30 min of illumination (Fig. 6J), and stayed only visible in the PM/ER. The same results were found with GFP12 plants (results not shown).
To explain this observation Western blots were performed with soluble proteins and PM fractions extracted from WT and GFP10 plants using both anti-NRT2.1 and anti-GFP antibodies. As expected for a membrane protein, no signal was detected with anti-NRT2.1 antibody on the Western blot performed with soluble proteins (data not shown). However, using the same extract, a strong band was detected with the anti-GFP antibody, specifically in the GFP10 plants, at ϳ27-28 kDa (Fig.  7A), which fits exactly the molecular mass of free GFP. In agreement with the confocal microscopy data, the GFP band was only recorded at night, and quickly disappeared after only 4 h of light (Fig. 7A). This unexpected result suggests that the NRT2.1-GFP protein is cleaved, and that the GFP fluorescence seen inside the vacuole in the dark is due to free GFP generated by this cleavage. This conclusion is fully supported by Western blots performed with PM fractions of GFP10 plants using anti-NRT2.1 antibody. As shown in Fig. 7B, three specific bands were detected in the PM of GFP10 plants: a weak band at ϳ120 kDa corresponding to the high molecular mass complex observed in the WT, a strong band at ϳ65 kDa, specific for GFP10 plants, which approximately corresponds to the expected size of the NRT2.1-GFP fusion protein, and surprisingly a strong band at ϳ45 kDa, i.e. at the size of the native NRT2.1 protein in the WT. This last result was totally unexpected because no native NRT2.1 protein should be found in the PM from GFP10 plants as this line was obtained after transformation of the nrt2.1-1 knock-out mutant (see Fig. 2, B and C). The simplest interpretation of these data is that the band found at ϳ45 kDa in GFP10 PM is the portion of the NRT2.1-GFP protein that remains after GFP has been cleaved off. Interestingly, free GFP was not found in transgenic plants expressing a GFP-NRT2.1 protein with GFP fused at the N terminus of NRT2.1 (Fig. 7C), and accordingly, no GFP fluorescence was recorded inside the root cortical vacuoles when these plants were transferred for 24 h in the dark (Fig. 7, D and E). Furthermore, no GFP is observed in the root vacuoles of transgenic plants expressing GFP alone under the control of the 35 S promoter (supplemental materials Fig. S4).

Tissular and Subcellular Localization of NRT2.1-Despite
extensive studies describing NRT2.1 as a major component of the NO 3 Ϫ HATS in A. thaliana (4,17,18,20,30,32), the cellular and subcellular localization of this protein remained unknown. To address this point we developed a GFP strategy using the nrt2.1-1 knock-out mutant complemented with a transgene expressing NRT2.1 fused at the C terminus with a GFP tag. To avoid the loss of specific tissular localization, NRT2.1-GFP was expressed under the control of the native NRT2.1 promoter (29). Two transgenic lines were used, GFP10 and GFP12, that were functionally complemented by the NRT2.1-GFP construct (supplemental materials Fig. S1). These results strengthen the fact that our imaging of NRT2.1-GFP accurately reflects the native cellular distribution of NRT2.1. As expected for a NO 3 Ϫ transporter involved in uptake from the soil into the roots, GFP localization is found in the epidermal and cortical cells of mature roots, where the bulk of nutrient uptake into the symplasm occurs. These results are fully consistent with the localization of NRT2.1 gene expression using ␤-glucuronidase (GUS) or luciferase (LUC) reporter genes (26,29), and indicate no discrepancy between gene and protein expression territories.
At the subcellular level, NRT2.1 is predominantly localized in the PM. This was confirmed by both NRT2.1-GFP and GFP-NRT2.1 imaging (Figs. 1, 3, and 6) and immunological assays (Western blots and ELISA) performed with purified PM fractions (Figs. 2, 4, 5, and 7). Furthermore, membrane fractionation on sucrose gradients indicated that the monomeric form of NRT2.1 (at ϳ45 kDa) is specifically present on PM (38). This localization corresponds to what was found for the NRT2.1 homologue in the yeast H. polymorpha (37), and is in agreement with the hypothesis that NRT2.1 is involved in root NO 3 Ϫ influx from the external medium (15,17,18).
The Unexpected Structural Complexity of NRT2.1 in Root Cell Membranes-Western blots reveal that at least two forms of NRT2.1, detected with both NRT2.1 antibodies, seem actually present in the PM: a major form detected at ϳ45 kDa, and a much less abundant form with a higher mass at ϳ120 kDa (Fig. 2). It suggests that, in PM, NRT2.1 may be part of a protein complex of two or more associated proteins, but is mainly present in its isolated monomeric form at ϳ45 kDa. Furthermore, a third band at ϳ75 kDa was detected when Western blots were performed with microsomal fractions, suggesting that there is another complex involving NRT2.1, not predominantly localized on PM. This third form of NRT2.1 could correspond to the GFP fluorescence detected in the network of the ER in NRT2.1-GFP transgenic lines (Fig. 1). The most obvious candidate protein possibly involved in complex formation with NRT2.1 is NAR2.1. Indeed, recent studies have shown that in Arabidospsis, like in C. reinhardtii and barley, NRT2.1 needs to interact with the NAR2.1 protein to be functional (19,(23)(24)(25)53). The physiological characterization of a NAR2.1 knock-out mutant (nar2.1-1), provided in planta evidence that the absence of NAR2.1 reduces NO 3 Ϫ -inducible high-affinity NO 3 Ϫ influx by more than 90%, even though NRT2.1 is normally expressed (19,25). Furthermore Orsel et al. (19) showed, using a mating based split-ubiquitin system, that the interaction between NRT2.1 and NAR2.1 occurs at the protein level. These data clearly suggest that NAR2.1 is involved in a protein-protein interaction with NRT2.1, and thus that the higher molecular mass bands found in our Western blots might be constituted of both proteins. Surprisingly, immunological studies of NRT2.1 expression in the nar2.1-1 mutant revealed that disruption of NAR2.1 does not dramatically affect the band at ϳ120 kDa, but leads to the complete disappearance of the band at ϳ45 kDa, corresponding to the major form of NRT2.1 (Fig. 3). Accordingly, GFP fluorescence recorded from GFP-NRT2.1 protein is considerably decreased in the nar2.1-1 mutant background as compared with the wild-type background (Fig. 3). Thus, it is concluded from these data that NAR2.1 is not part of the high molecular mass complex at ϳ120 kDa, but is strictly required for expression of the main form of NRT2.1, at ϳ45 kDa.
As already mentioned by Orsel et al. (19), one hypothesis for the function of NAR2.1 is that this protein is involved in targeting NRT2.1 to the PM, thus playing a similar role as PHF1 and AXR4, which were recently shown to be responsible for the trafficking of PHT1 and AUX1 transporters from the ER to the PM (54,55). The complete absence of the main form of NRT2.1 (ϳ45 kDa) from the PM of nar2.1-1 roots (Fig. 3) is consistent with this hypothesis, especially because it cannot be attributed to the lack of NRT2.1 gene expression in the nar2.1-1 mutant (Fig. 3C). However, there are two intriguing aspects of the nar2.1-1 phenotype, which call for a more complex explanation. First, unlike PHT1 and AUX1, which are retained in the ER in the phf1 and axr4 mutants, respectively (54,55), NRT2.1 is absent in the root microsomal fractions of the nar2.1-1 mutant (Fig. 3), and GFP-NRT2.1 was not found to accumulate in the ER of root cells in the transgenic lines in nar2.1-1 background (Fig. 3, D-L). This suggests that, if involved in NRT2.1 trafficking, NAR2.1 also plays a role in either stimulating NRT2.1 synthesis or preventing NRT2.1 degradation. In either case, this would correspond to a mechanism for controlling expression of membrane proteins that has not been described yet in plants. Second, the high molecular mass complex at ϳ120 kDa was still present in the nar2.1-1 microsomal fraction (Fig.  3), indicating that this form of NRT2.1 does not depend on NAR2.1 to be expressed in root cell membranes.
Because their composition remains unknown, it is difficult to speculate about the physiological role of the high molecular mass complexes revealed by anti-NRT2.1 antibodies. However, these results raised an important question about the relationship between the presence of NRT2.1 in two different forms in root PM, and the fact that this protein has two distinct functions, in high-affinity root NO 3 Ϫ uptake on the one hand (17,18), and in regulation of lateral roots initiation on the other hand (16,26). The phenotype of the nar2.1-1 mutant (19,25) allows providing a partial answer to this. As the suppression of high-affinity NO 3 Ϫ influx in this mutant corresponds to the disappearance of the band at ϳ45 kDa and not to the band at ϳ120 kDa (Fig. 3), we hypothesize that the active form of NRT2.1 for NO 3 Ϫ transport across the PM most probably corresponds to the isolated monomeric form at ϳ45 kDa, which is also the most abundant form of NRT2.1. This also indicates that the presence of the high molecular mass form of NRT2.1 at ϳ120 kDa in root cell membranes is not sufficient per se to mediate high-affinity NO 3 Ϫ uptake by the roots. Regulation of the NRT2.1 Protein-The strong correlation observed between the changes in root accumulation of NRT2.1 transcript and those of root NO 3 Ϫ influx in response to light, sugar, and high nitrogen supply has suggested that the regulation of the NO 3 Ϫ HATS may mainly be accounted for by a transcriptional control of NRT2.1 expression (11, 17, 29 -31). However, so far, no studies at the protein level have been performed to confirm that NRT2.1 abundance in root PM follows similar changes as NRT2.1 mRNA or NO 3 Ϫ HATS activity. Very surprisingly, our data provide ample evidence that this is not the case. Both immunological assays in wild-type plants and confocal microscopy analysis of NRT2.1-GFP in transgenic plants showed that neither the day/night cycle nor sugar treatments had a marked influence on the abundance of NRT2.1 in root Regulation of the NRT2.1 Nitrate Transporter Protein AUGUST 10, 2007 • VOLUME 282 • NUMBER 32 PM (Figs. 4 and 6, and supplemental materials Fig. S2), whereas these conditions strongly affect the NRT2.1 mRNA level and NO 3 Ϫ influx (30,32). Significant decrease in the amount of NRT2.1 is only observed in plants transferred for at least 3 days on high nitrogen (10 mM NH 4 NO 3 or 5 mM NO 3 Ϫ ) compared with plants grown on low nitrogen (0.3 or 1 mM NO 3 Ϫ ) (Fig. 5). However, little change in the amount of NRT2.1 is observed after a short treatment (8 h) on high nitrogen, whereas this time scale is known to be largely sufficient to down-regulate both NRT2.1 expression and NO 3 Ϫ HATS activity (11,17). Even after 1 week on 10 mM NH 4 NO 3 , the 3-fold reduction of the NRT2.1 protein level is surprisingly limited compared with the decreases in NRT2.1 mRNA level and root NO 3 Ϫ influx observed in the same conditions (20-and 10-fold, respectively). 4 This slow response of the protein to strongly repressive treatments can be explained by its relatively high stability, as suggested by the results from cycloheximide treatment. Interestingly, a relatively long half-life of the transporters involved in NO 3 Ϫ HATS has already been suggested from physiological studies with the inhibitor of functional protein synthesis fluorophenylalanine (56).
The fact that the amount of NRT2.1 in the PM does not change in response to environmental conditions that affect the NRT2.1 mRNA level and NO 3 Ϫ influx is, to our knowledge, pretty unique compared with what is known for the regulation of other ion or water transporter proteins. Several examples concerning the iron transporter IRT1 (57), the high-affinity NO 3 Ϫ transporter from Tuber borchii (58), the sulfate transporter HVST1 from barley (59), the phosphate transporter PT1 from Medicago and tomato (60,61), and aquaporines (62) all showed a correlation between the level of mRNA and the level of the protein in response to iron starvation, nitrogen availability, sulfate re-supply, phosphate starvation, and salinity, respectively. However, recent large-scale experiments showed that transcript levels undergo marked and rapid changes during diurnal cycles and after transfer to darkness, whereas changes in enzyme activities are smaller and delayed (63). This strongly suggests that in many cases additional regulation at the level of protein is required. Evidence for the occurrence of post-translational control of NRT2 proteins have been provided by the observations that NO 3 Ϫ HATS activity is repressed by high nitrogen supply in transgenic lines of N. plumbaginifolia, constitutively overexpressing NRT2.1 and by an increase in internal NH 4 ϩ concentration in roots of barley treated with MSO, whereas transcript levels remains high (35,36). A possible posttranslational regulatory mechanism that can be considered for controlling NRT2.1 activity relates to the NRT2.1/NAR2.1 interaction. The requirement for co-expression of both proteins to yield NO 3 Ϫ HATS activity in Xenopus oocytes has highlighted the hypothesis of a two-component transporter (19,24). Accordingly, association/dissociation of the NRT2.1/NAR2.1 complex may be an efficient way to modulate transport activity. Without excluding this hypothesis, we found no evidence for the presence of a NRT2.1/NAR2.1 complex in PM (Fig. 3). Alternatively, the presence of a number of conserved protein kinase C recognition motifs in the N-and C-terminal domains of NRT2.1 (3) may suggest that phosphorylation events are involved in regulating NRT2.1 activity in response to environmental cues, as it was shown to be the case for NRT1.1 (51).
Evidence for Partial Proteolysis of NRT2.1 at the C Terminus of the Protein-The GFP filling of the vacuole observed in the GFP10 transgenic lines after transfer of the plants to the dark (Fig. 6) initially suggested a mechanism for NRT2.1 degradation similar to that recently described for the YNT1 NO 3 Ϫ transporter of H. polymorpha (37). In this particular yeast species, YNT1 is ubiquitinylated in response to glutamine supply, and is transferred to the vacuole where it is rapidly degraded by a specific proteinase A. In our case we showed that the fluorescence that slowly appears in the vacuole in darkness is due to the free GFP generated by the cleavage of the NRT2.1-GFP protein (Fig. 7A) and not to NRT2.1 trafficking for degradation in the vacuole. This mechanism does not remove the NRT2.1 part of the tagged protein from the PM (Fig. 7B), suggesting that this partial proteolysis of NRT2.1-GFP occurred in the PM, and not after a transfer to the vacuole. Furthermore, GFP filling of the vacuole in the dark does not correspond to a decrease in the fluorescence observed in PM (Fig. 6). This makes a profound difference with what happens in H. polymorpha, where degradation of YNT1 in the vacuole was actually associated with the removal of this transporter from the PM (37). The fact that both the fluorescence in the vacuole and the free GFP in soluble proteins are not found after long dark treatment when GFP is fused in N terminus of NRT2.1 (Fig. 7), demonstrates that the free GFP generation is specific of the C terminus NRT2.1-GFP fusion protein. Furthermore, the NRT2.1 part of NRT2.1-GFP that remains in the PM is still recognized by the anti-NRT2.1 antibody (Fig. 7B), which targets an epitope located only 18 amino acids upstream of the C terminus of NRT2.1. Taken together, both data from GFP imaging and immunological studies with anti-NRT2.1 and anti-GFP antibodies are all consistent with the hypothesis that NRT2.1 is subject to partial proteolysis at its C terminus. With the native NRT2.1 protein, this partial degradation mechanism is thus expected to generate on the one hand, the same truncated NRT2.1 protein as with NRT2.1-GFP, and on the other hand, a short peptide corresponding to the C terminus of NRT2.1. Both native and truncated NRT2.1 proteins would be indistinguishable in our Western blots, due to their very similar size.
An important question is to determine whether the partial proteolysis of NRT2.1 constitutes a post-translational mechanism for regulating its activity. The fact that cleavage of NRT2.1-GFP is only apparent at night, when the NO 3 Ϫ HATS activity is down-regulated, may suggest that it is associated with NRT2.1 inactivation. However, several observations do not fit with this hypothesis. First, GFP localization in the vacuole resulting from cleavage of NRT2.1-GFP is not suppressed by sucrose supply (Fig. 6), whereas this treatment is very effective in preventing down-regulation of the NO 3 Ϫ HATS (30,32). Second, it is not certain that partial proteolysis of NRT2.1 only occurs at night. Indeed, Tamura et al. (64) showed that GFP located in the vacuole can only be visualized in the dark because light, and especially blue light, induces a conformational change in the protein, followed by a rapid degradation by vacuolar papain-type cysteine proteinases under acidic pH. Thus, it cannot be ruled out that in our case, NRT2.1-GFP is always cleaved, independently of light or darkness, but that cleaved GFP can only be observed in the vacuole in the dark because of its rapid degradation in the light. This rapid degradation of GFP under light conditions was confirmed by transferring GFP10 plants to the light after 24 h of darkness. Only 1.5 h after transfer to the light, no more fluorescence could be observed in the vacuole (Fig. 7). Independently of the putative role of the NRT2.1 C terminus cleavage, it is noteworthy that the GFP part of the NRT2.1-GFP protein generated by this cleavage moves to the vacuole, and does not remain confined to the cytoplasm and nucleus, as usually observed with free GFP (65), even in the dark (see supplemental materials Fig. S4). If still fused to GFP, the C-terminal peptide of NRT2.1 may be responsible for addressing the tag protein to the vacuole, suggesting trafficking of the NRT2.1 C-terminal part to this compartment in wild-type plants.
In conclusion, our data provide new insights on several important aspects of NRT2.1 function and regulation. First, the complexity of NRT2.1 function (NO 3 Ϫ transport and root development) seems to be associated with a structural complexity of this protein, with at least two, and possibly three, different forms of NRT2.1 present in roots PM. The confirmation of the existence and the investigation of the respective roles of these different NRT2.1 forms will certainly be a major issue for the future. Second, the monomeric form of the NRT2.1 protein is absent in the nar2.1-1 mutant suggesting that NAR2.1 is essential for expression of the NRT2.1 protein in the PM. Third, the lack of correlation between regulation of NRT2.1 expression and NO 3 Ϫ HATS on the one hand, and regulation of the NRT2.1 protein level on the other hand, suggests that post-translational mechanisms are crucial for the fast modulation of NO 3 Ϫ uptake in response to environmental changes, as it is also the case for the control of the first step of NO 3 Ϫ reduction catalyzed by nitrate reductase (66,67). One such mechanism could correspond to a partial proteolysis of NRT2.1. All these mechanisms seem to be original and possibly specific to plants as they differ compared with what was found for the NO 3 Ϫ transporter YNT1 in the yeast H. polymorpha (37).