Molecular basis of nitrate uptake by the plant nitrate transporter NRT1.1

Summary

The NRT1/PTR family of proton-coupled transporters are responsible for nitrogen assimilation in eukaryotes and bacteria through the uptake of peptides. However, in the majority of plant species members of this family have evolved to transport nitrate as well as additional secondary metabolites and hormones. In response to falling nitrate levels, NRT1.1 is phosphorylated on an intracellular threonine that switches the transporter from a low to high affinity state. Here we present both the apo and nitrate bound crystal structures of Arabidopsis thaliana NRT1.1, which together with in vitro binding and transport data identify a key role for His356 in nitrate binding. Our data support a model whereby phosphorylation increases structural flexibility and in turn the rate of transport. Comparison with peptide transporters further reveals how the NRT1/PTR family has evolved to recognize diverse nitrogenous ligands, whilst maintaining elements of a conserved coupling mechanism within this superfamily of nutrient transporters.

Nitrogen is a key element in biology, required for the synthesis of amino and nucleic acids and is a fundamental nutrient for cellular metabolism. The PTR or peptide transporter family, also known as the POT, or proton dependent oligopeptide transporter family, plays an important role in nitrogen assimilation in bacteria, fungi and mammals through the uptake of short peptides from the environment ^1,2. In contrast, in the majority of plant species nitrogen is largely obtained through the uptake of nitrate (NO_3⁻) from the soil ³. This is achieved through the actions of specific nitrate transporters in the plasma membrane of root cells ⁴ and is controlled through the NRT1 and NRT2 gene families that encode for low (K_M mM) and high (K_M μM) affinity transporters respectively ^5,6. Intriguingly the NRT1 family of nitrate transporters phylogenetically belongs to the PTR family, and may have evolved from an ancestral peptide transport protein ⁷.

In plants the NRT1/PTR family, recently renamed the NPF (NRT1/PTR Family) ⁸, has functionally diverged with individual members recognizing peptides, glucosinolate defense compounds, plant hormones or nitrate ^9-12. The NRT1/PTR family belongs to the Major Facilitator Superfamily (MFS) of secondary active transporters ^13,14 that use the proton electrochemical gradient to drive substrate uptake into the cell ^15-17. MFS transporters operate via an alternating access mechanism wherein a centrally located binding site is re-orientated to either side of the membrane to uptake and release substrate ^18-20. Recent crystal structures of bacterial members of the NRT1/PTR family have identified pairs of salt bridges that orchestrate the orientation of the transporter and revealed proton binding residues that are conserved across the family from pro- to eukaryotes ^21,22. An important question therefore is how the binding site in this transporter family has evolved to recognize and transport diverse molecules while retaining elements of the same coupling mechanism.

The Arabidopsis thaliana nitrate transporter, AtNRT1.1 (AtNPF6.3), is the founding member ⁷ of the NRT1/PTR family of transporters and shares significant sequence identity to mammalian and bacterial PTR peptide transporters (Extended Data Fig. 1). Interestingly AtNRT1.1 has two K_M values for nitrate ²³; in conditions of high nitrate availability (> 1mM) AtNRT1.1 behaves as a low-affinity transporter (K_M ~ 4 mM). However, when nitrate levels fall below 1mM, AtNRT1.1 is switched into a high-affinity mode (K_M ~ 40 μM). This switch occurs following phosphorylation of an intracellular threonine, Thr101, by the kinase CIPK23 ²⁴. This regulatory mechanism allows for the rapid adaption to changing nitrate levels before the dedicated high affinity transporter family is expressed ²⁵. These observations suggest a complex interplay between transporter activity and post-translational modification at the molecular level that currently has no structural or biochemical basis for explanation.

Structural basis of nitrate recognition in AtNRT1.1

To study the binding of nitrate to AtNRT1.1 we developed a microscale thermophoresis assay ^26,27 (Extended Data Fig. 2a) and determined the K_D for nitrate to be 1 ± 0.15 mM (Fig. 1a & Extended Data Fig. 2). A wide range of nitrate concentrations were tested (1 μM – 50 mM) and we did not observe a high affinity state. We further tested a range of different ligands and show that AtNRT1.1 specifically recognizes both nitrate and chlorate, a herbicide for which the AtNRT1.1 protein was originally named (Chlorate Resistance 1) ⁷ but does not recognize nitrite, alanine, sulphate, phosphate or the di-peptide Ala-Ala (Extended Data Fig. 2c). A key question surrounding the function of AtNRT1.1 is the switch between high and low affinity states. Using the phosphomimetic variant Thr101Asp, we investigated the effect on nitrate binding. The K_D for nitrate binding to the Thr101Asp was the same as WT (K_D 1 ± 0.12 mM) (Fig. 1a), suggesting that phosphorylation is unlikely to directly alter the nitrate-binding site. To follow up this finding and identify the nitrate-binding site we crystallised AtNRT1.1 in the presence and absence of nitrate. The crystal structures were determined to 3.7 Å (Extended Data Table 1).

Figure 1. Structural and biophysical characterisation of AtNRT1.1.

a. Binding isotherms for nitrate to both wild type (black squares) and T101D (red triangles) reveals no significant change in the K_D. Data shown is representative of three independent experiments and the error is calculated as standard deviation. b. Cartoon representation of the crystal structure of AtNRT1.1 viewed from the extracellular side of the plasma membrane. Transmembrane helices (TM) 1-12 have been coloured from blue at the amino terminus to red at the carboxy terminus. T101 is shown as magenta spheres. c. AtNRT1.1 viewed in the plane of the membrane represented as both cartoon and electrostatic surface, positively charged and aromatic residues at the end of the lateral helix are shown in sticks and may help to anchor this helix in the interfacial region of the plasma membrane. Residues 270-325 of the intracellular domain could not be modeled due to insufficient electron density.

The apo structure of AtNRT1.1 contains 12 transmembrane (TM) spanning alpha helices, consisting of N-(TM1-TM6) and C-terminal (TM7-TM12) bundles forming the canonical MFS fold (Fig. 1b). Separating these two bundles is a previously uncharacterized intracellular domain consisting of 84 amino acids. This domain is predominantly alpha helical and extends outwards from the transporter domain at an approximate 90° angle, which we have termed the lateral helix (Fig. 1c). At the distal end of this helix are three conserved positively charged residues that may help to stabilize this domain on the intracellular side of the membrane. It is unlikely this domain is a crystallographic artifact as it does not make any crystal contacts and is sufficiently mobile that we cannot model residues 270-325. AtNRT1.1 crystallizes as a dimer in the asymmetric unit cell, despite being monomeric in solution (Extended Data Fig. 3). The dimer interface forms between TM3 and TM6 of monomer A packing against the equivalent helices in monomer B and stabilized by hydrophobic interactions (Extended Data Fig. 3). The buried surface area between the two monomers is ~ 2136 Å2 and may be physiologically significant as both monomers are in the same orientation. Both monomer A and B adopt the same inward open conformation (with an r.m.s.d. between monomer A and B of ~ 0.69 Å for 472 Cα atoms) with a clearly defined cavity extending outward from the middle of the protein towards the intracellular side of the membrane. Surprisingly the cavity is much larger than expected given the size of nitrate, being ~ 21 × 18 × 17 Å in diameter (Fig. 2a). The extracellular side of the binding site is tightly sealed through the packing together of TMs1-2 against TMs7-8. These helices form the extracellular gate in MFS transporters, which control access to the binding site from the outside of the cell ^19,28,29. Interestingly there exist no salt bridge interactions between these helices, unlike the prokaryotic NRT1/PTR homologues ^14,21,22,30, instead the extracellular gate is stabilized through extensive hydrophobic interactions (Extended Data Fig. 4).

Figure 2. Crystal structure of nitrate bound AtNRT1.1.

a. A section through the protein volume showing the binding site in the plane of the membrane, with the mFo-DFc difference electron map contoured at 3 sigma (green) identifying the position of nitrate. b. Zoomed in view of the binding site. Shown in sticks are the conserved PTR/NRT1 residues and His356 that can be seen coordinating nitrate within the binding site, supported by Thr360. Dashed lines indicate a potential interaction network between His356, Tyr388 and Glu476 in the C-terminal bundle. c. Binding isotherms showing the effect of mutating positively charged residues in the binding site on the affinity of AtNRT1.1 for nitrate. Data shown is representative of three independent experiments and the error is calculated as standard deviation. d. Similarity between the binding sites of the peptide transporter PepT_So and AtNRT1.1. Shown in sticks are the residues important for peptide transport. The relative position of nitrate from the AtNRT1.1 structure is superimposed (green mesh).

The nitrate bound structure was unchanged from the apo state with an r.m.s.d. of ~ 0.44 Ų for 472 C_α atoms. However, we observed a significant (> 3 sigma) mFo-DFc difference electron density peak within the central cavity, which we observed in crystals grown in the presence of nitrate. We therefore modeled nitrate into this density (Fig. 2 a & b & Extended Data Fig. 5). The nitrate molecule sits approximately 2.8 Å away from His356 on TM7; given the pH of the crystallization condition was 4.5 His356 is likely to be protonated and forming an electrostatic interaction with nitrate. An additional interaction via a hydrogen bond occurs with Thr360 on TM7, which sits approximately 2.9 Å away at the apex of the binding site. The binding of nitrate through an electrostatic interaction is similar but not identical to the recent crystal structure of NarU, a member of the biochemically distinct Nitrate Nitrite Porter (NNP) family, where two conserved arginine residues act to coordinate nitrate through hydrogen bonds (Extended Data Fig. 6) ³¹. The other possible sites of interaction for nitrate therefore could be either Arg45 (TM1) and/or Lys164 (TM4). We mutated all three of these positively charged residues to alanine, however mutation of only His356 resulted in complete loss of nitrate binding, indicating an essential role of His356 in AtNRT1.1 (Fig. 2c).

Substrate specificity in the NRT1/PTR family

As discussed previously, members of the NRT1/PTR family can transport peptides, nitrate and in the case of plant members, hormones and metabolites. A key question we wished to address was how the binding site of AtNRT1.1 differed to that of the peptide transporters. Comparison with the binding site from the bacterial peptide transporter from Shewanella oniendensis, PepT_So, reveals that the two binding sites are strikingly similar (Fig. 2d). The previously identified ExxERF motif on TM1, which plays an important role in coupling proton binding to peptide transport ²² is present in the same position, as are Lys164 (TM4) and Glu476 (TM10). As predicted for their bacterial counterparts ^21,22 Lys164 and Glu476 are likely to interact via a salt bridge in the outward facing conformation and form an important link between the N and C terminal bundles during transport (Extended Data Fig. 7).

Our comparison suggests the mechanism required for coupling proton movement to structural changes during transport is conserved regardless of substrate specificity. It is known the stoichiometry of AtNRT1.1 and mammalian peptide transporters is at least 2:1 proton:nitrate, as transport is electrogenic ^7,32. It is therefore conceivable the ExxERF motif couples one proton, leaving His356 on TM7 to bind another proton and nitrate. In peptide transporters the equivalent region on TM7 has been shown to couple proton binding to peptide recognition ²² through an acidic residue, Asp316 in PepT_So. One major difference between the binding sites is the absence of an obvious salt bridge connecting the extracellular gate helices TM1 and TM7. Indeed in PepT_So the arginine of this salt bridge occupies the same position in the protein as the nitrate (Figure 2c & d). Another significant difference is the replacement of the two conserved aromatic residues on TM1 (Phe28, Tyr29), which are responsible for peptide specificity ²¹ with residues containing smaller side chains helping to create space for nitrate to bind. Given that di-peptides are coordinated horizontally between the N- and C-terminal bundles ^22,30, it is conceivable that the increased width of the cavity in AtNRT1.1 can no longer coordinate the amino and carboxy termini simultaneously. Coupled with the absence of the salt bridge and tyrosines, which help position the peptide and coordinate movement of TM1 and TM7, these seemingly esoteric alterations could explain why AtNRT1.1 cannot recognize or transport peptides. 7

Functional role of phosphorylation at Thr101

Our structures show that Thr101 is situated at the bottom of TM3 and points towards a hydrophobic pocket constructed from residues in TM1 and TM4 (Fig. 3a). Given the position of Thr101 it is highly likely that phosphorylation would cause localized structural distortion in the packing of TM3 with TM1 and TM4 on the intracellular side of the protein. Consistent with this hypothesis we discovered that the Thr101Asp variant of AtNRT1.1 was significantly less stable compared to WT protein (9 °C lower melting temperature) (Fig. 3b). To investigate the effect of the Thr101Asp substitution on nitrate uptake we reconstituted AtNRT1.1 into liposomes and monitored the relative uptake of nitrate as a function of proton movement using a pH sensitive fluorophore (Extended Data Fig. 8). As AtNRT1.1 is a proton coupled nitrate transporter this assay provides a reliable readout for nitrate transport. We discovered that compared to the WT protein, the Thr101Asp variant showed increased uptake of ~ 2.8 fold, whereas the His356Ala variant showed no detectable transport, consistent with an essential role in nitrate recognition (Fig. 3c).

Figure 3. Proposed mechanism for the affect of phosphorylation at Thr101 on nitrate transport.

a. Thr101 (represented as magenta spheres) in AtNRT1.1 is situated in a hydrophobic cavity formed at the intracellular ends of TMs 2 and 4. b. Thermal stability as assessed by the loss of alpha helical secondary structure using circular dichroism shows a reduction in the Thr101Asp variant when compared to wild type protein. c. The Thr101Asp variant of AtNRT1.1 shows a significant increase in transport when compared to wild type protein in a reconstituted assay system, where as the His356Ala mutant shows no activity. Data represents the mean value from eight experiments and error bars are standard deviation.

Given the high sequence conservation of the N-terminal bundle within the NRT1/PTR family (Extended Data Fig. 1 & 9a), we mutated the equivalent threonine in PepT_So (Thr87) to investigate any general effects disrupting the packing between these helices may have on this family. Interestingly the Thr87Asp variant also resulted in an increase in the rate of peptide transport, whereas other variants Thr87Ala/Ser/Lys showed WT like rates (Extended Data Fig. 9b-d) showing this region is sensitive to the addition of a negative charge. Comparable to what we observed for AtNRT1.1, the Thr87Asp variant was also less stable than WT protein by 4 °C (Extended Data Fig. 9e). However a Thr87Arg variant was inactive suggesting gross structural distortion in this region cannot be tolerated. This shows that disrupting the helix packing between TMs 1, 3 and 4 in the N-terminal bundle of the NRT1/PTR family can dramatically affect transport rates. Similar studies have shown that altering the structural flexibility within the MFS fold can affect the rate-limiting step of transport ^33,34. We propose a similar mechanism may occur in AtNRT1.1 to switch the protein between a high and low Km state. Phosphorylation causes a localized disruption of the N terminal bundle helices that leads to an increased flexibility of the protein. This in turn leads to an increase in the transport rate, that would result in the lower Km observed in the previous study ²⁴.

Model for proton coupled nitrate transport via AtNRT1.1

Taken together a working model for proton coupled nitrate transport emerges from our study (Fig. 4). The protonation of His356 is clearly a key component of the transport mechanism. Our model predicts that protonation must occur prior to nitrate binding and may be conditional upon the presence of nitrate to stabilize the charged state of His356. Following closure of the extracellular gate and adoption of the occluded state the release of both proton and nitrate from His356 must be coupled to the opening of the intracellular gate. A possible mechanism in achieving this is that adoption of the occluded state brings His356 into close proximity to Glu476 causing the disruption of the intracellular gate salt bridge (Extended Data Fig. 7). The interaction between His356 and Glu476 may be facilitated by Tyr388 (TM8), which can be seen making a hydrogen bond to Glu476 in the crystal structures. The binding of additional protons potentially to the ExxERF motif might also affect the strength of the salt bridge between Lys164 and Glu476 facilitating opening of the intracellular gate and release of bound nitrate and protons into the interior of the cell. Our data suggest that in AtNRT1.1 the effect of phosphorylation is to cause a localized disruption in the N-terminal bundle that in turn may affect the position of the ExxERF motif with respect to Lys164. The increase in transport rate that occurs is possibly the result of altering the efficiency in the formation and breakage of the Lys164-Glu476 salt bridge and release of nitrate from the binding site. Perhaps the most surprising finding from our study however is that only minor changes to the previously characterized peptide transporter binding site are required to accommodate nitrate and that these are predominantly located in the C-terminal bundle. The ability of the PTR family to recognize multiple ligands may therefore stem from the separation of a fundamentally conserved proton coupling mechanism that resides largely in the N-terminal bundle and orchestrates gross conformational changes between TMs4-5 and 10-11, whilst leaving the C-terminal bundle to evolve to recognize different nitrogenous ligands.

Figure 4. Alternating access model for nitrate-proton symport.

In the outward facing state (a) the extracellular gate, constructed from TMs1-2 (dark blue) and TMs7-8 (yellow) is open and the intracellular gate, constructed from TM4-5 (cyan) packing against TM10-11 (orange) is closed and stabilized by a salt bridge between Lys164 (TM4) and Glu476 (TM10) - similar to the bacterial homologues. Following protonation of His356, nitrate is able to bind triggering closure of the extracellular gate (b). Additional sites of protonation exist possibly within the ExxERF motif (TM1). Upon transition to the occluded state (c) the intracellular gate salt bridge will break as a result of Glu476 moving to form a new salt bridge with His356. This will result in the release of nitrate and protons into the interior of the cell (d). Phosphorylation of Thr101 (purple sphere) by CIPK23 results in increased flexibility within the N-terminal bundle, increasing the overall transport rate (e).

Methods

Cloning, Expression and Purification of AtNRT1.1

The gene encoding AtNRT1.1 was amplified from an Arabidopsis image clone and cloned into a modified form of a C-terminal GFP^His fusion yeast expression vector ³⁵. The yeast vector had an additional selection marker the LEU2 gene under the control of a truncated promoter to increase copy number within Saccharomyces cerevisiae ³⁶. Wild type and mutant AtNRT1.1 were expressed by growing an overnight in minus leucine with 2 % glucose, this culture was diluted 10 fold in minus leucine with 2 % lactate, after 8-10 hours expression was induced thorough the addition of 2 % galactose from a 25 % stock. The yeast was harvested after 20 hours and membranes were prepared. AtNRT1.1 and mutants were purified to homogeneity using standard IMAC protocols in n-dodecyl- β-D-maltopyranoside (DDM) detergent as described previously ³⁷. For MST binding experiments the C-terminal GFP^His was not removed. The protein was concentrated to 10 mg.ml⁻¹ for crystallization or 0.5 mg.ml⁻¹ for binding assays and stored at −80°C.

Crystallization and Structure Determination

Crystals of AtNRT1.1 were initially obtained in the MemGold crystallization screen ³⁸ and optimized to 22 % PEG 400, 0.05 M sodium citrate pH 4.5, 0.07 M sodium chloride and 1.5 % PEG 600 using the hanging drop vapor diffusion technique at 4 °C. For cryoprotection the crystals were transferred to a solution containing 33 % PEG 400, before being flash cooled in liquid nitrogen. The crystals always showed anisotropic diffraction, with the best crystals diffracting X-rays to Bragg spacing’s of 3.50 Å in the strongest direction. Initial phases were calculated using single wavelength anomalous dispersion (SAD) using a single gold cyanide derivatised crystal. Data were collected on beamlines IO2, IO3, IO4 and IO4-1 at the Diamond Light Source Ltd., UK. Data were processed by the XIA2 ^39,40 pipeline to XDS ⁴¹ and scaled using AIMLESS ⁴². Diffraction data were collected to 3.5 Å (nitrate bound) and 3.6 Å (apo) resolutions and this data was used for calculating maps during model building however the final deposited model was refined against the 3.7 Å data shown in Extended Data Table 1, as the I/σI and completeness of the data dropped significantly after this resolution cut off. The space group was determined to be P2₁2₁2₁ with two molecules in the asymmetric unit forming a back to back dimer (Supp. Fig. 7). Two gold sites were initially located using SHELXC/D ⁴³ with their positions further refined and initial phases calculated using SHARP with solvent flattening in SOLOMON ⁴⁴. The resulting experimental maps were of sufficient quality to see all 12 trans membrane helices from each of the two molecules (Supp. Fig. 9). Further improvement in map quality was obtained following cross crystal averaging in DMMulti ^45,46 with the native dataset. The lateral helix appeared in the maps following model building and refinement.

Model building and refinement

The model was built into experimental maps calculated from SHARP and DMMulti, using O ⁴⁷. The partial models were further cycled back into SOLOMON to improve the initial solvent envelope used for the solvent flipping procedure. The amino acid side chains were then built using an homology model of AtNRT1.1 built from the template of PepT_St (4APS) and GkPOT (4IKV) using modeller v9.12 ⁴⁸. Refinement of the model was carried out in BUSTER ⁴⁹ with inclusion of experimental phase information. Model validation was carried out using the Molprobity server ⁵⁰. Images were prepared using PyMol ⁵¹.

Microscale thermophoresis binding assay

Binding was calculated for AtNRT1.1 and the control POT family transporter, PepTSo ¹⁴, using microscale thermophoresis ²⁶. A range of concentrations of the required ligand (range from 0.1 μM to 50 mM) was incubated with 0.8 μM of purified GFP tagged protein 5 minutes in assay buffer (20 mM Bis-Tris pH 6.5, 150 mM NaCl, 0.03 % DDM). The sample was loaded into the NanoTemper glass capillaries and microthermophoresis carried out using 10 % LED power and 80 % MST. K_Ds were calculated using the mass action equation via the NanoTemper software from duplicate reads of triplicate experiments. The instrument used was a NanoTemper monolith NT.115.

PepT_So transport assays using a proton driven proteoliposome system

PepT_So was purified as previously described ¹⁴ and reconstituted by the dilution method. The efficiency of reconstitution was analysed by densitometry of a fixed amount of proteoliposomes using a serial dilution of purified protein as a standard. For proton driven uptake assays, artificially imposed potassium ion diffusion potentials were generated as previously described ⁵². Proteoliposomes were thawed and centrifuged at 90,000 g for 30 min at 4 °C and re-suspended in 20 mM Potassium Phosphate, pH 6.50, 100 mM Potassium Acetate, 2mM Magnesium Sulfate, followed by 11 cycles of extrusion through a 400 nm polycarbonate filter to obtain small unilamellar vesicles of relatively homogenous size ⁵³. Proteoliposomes were subsequently diluted 1:50 (v/v) to final protein concentration of 0.25 μM into external buffer containing 20 mM Na Phosphate, pH 6.5, 2 mM magnesium sulphate with 10 μM valinomycin and ³H-labeled di-Ala peptide (40 μM). Uptake of ³H substrate was assayed at 25 °C. Diluting aliquots into 1.5 ml of ice-cold 0.1 M lithium chloride stopped the reaction. Proteoliposomes were collected on 0.22 μm nitrocellulose filters and washed under vacuum with 0.1 M Lithium Chloride prior to scintillation counting. The ³H signal was converted to molar concentrations of peptide using standard curves for each substrate.

Reconstitution of AtNRT1.1 into liposomes

AtNRT1.1 was reconstituted into liposomes using the biobead method. Polar soy lipids (Avanti polar lipids) were washed twice in pentane and resuspended at 5 mg/ml in lipid buffer (50 mM potassium phosphate pH 7.5). The lipids were extruded through a 0.4 μm filter and titrated with Triton X-100 until the absorbance at 540 nm was ~90% of maximal. Purified AtNRT1.1 (0.2 mg) was added to the lipids at a final lipid:protein ratio of 60:1 and incubated for 1 hour on ice or for the no protein liposome control the same volume of gel filtration buffer used to purify AtNRT1.1 was added. After this time the required amount of biobeads (~0.1 mg) were added and incubated at 4 °C with rotary mixing for 3 hours. The biobead addition was repeated a further three times with the final incubation being overnight. The biobeads were removed and proteoliposomes were harvested by centrifugation at 120,000 g for 30 minutes. Proteoliposomes were resuspended in lipid buffer at a final lipid concentration of 30 μg/μl and dialysed overnight in lipid buffer to remove any traces of residual detergent. After dialysis the proteoliposomes were harvested and resuspended as before and subjected to three rounds of freeze thawing in liquid nitrogen before storage at −80 °C. Protein amount was quantified by SDS-PAGE and densitometry.

Nitrate transport assays using pyranine dye

AtNRT1.1 proteoliposomes were thawed and harvested at by centrifugation at 120,000g for 25 minutes at 4 °C. The liposomes were resupended in inside buffer (5 mM Hepes pH 6.8, 120 mM KCl, 2 mM MgSO₄) and 1 mM pyranine and loaded by two rounds of freeze thawing in liquid nitrogen followed by extrusion through first a 0.4 μm filter then a 0.2 μm filter. Excess pyranine was removed by washing once in inside buffer and then by application to a G25 gel filtration (GE Healthcare) column equilibrated in inside buffer. After this step the liposomes were split into 8 μg aliquots (enough for one assay per tube) and harvested by centrifugation at 120,000 g for 25 minutes at 15 °C.

The prepared liposomes were resuspended in 99 ul external buffer (5 mM MES pH 6.0, 1 mM KCl, 2 mM MgSO₄, 109 mM NaCl) and immediately the fluorescence was read at 460/510 and 415/510 in a SpectraMax M3 plate reader using 96 well black clear bottom plates (Molecular Devices). Each assay was measured individually. Either 10 mM sodium nitrate or 10 mM sodium chloride (no nitrate control) from 1 M stocks were added and again fluorescence read immediately after the addition and mixing. Data were collected for 30 seconds in 10-second increments. The difference between nitrate and no nitrate after 20 seconds was calculated (averaged from 8 experiments) and used to compare the variants of AtNRT1.1 (data shown in figure 3B).

Supplementary Material

Extended data figure 1. Alignment of AtNRT1.1 with PTR family members.

Amino acid alignment of A. thaliana NRT1.1 (Uniprot: Q05085) with PepT_So (Q8EKT7), PepT_St (Q5M4H8), human PepT1 (B2CQT6) and PepT_Gk (Q5KYD1) using MAFFT (Katoh & Toh, 2008) with manual adjustment in JalView (Clamp et al, 2004). AtNRT1.1 shares 23 % identity to human PepT1 and 22 % identity to PepT_So at the primary structure level. Identical residues shared across the NRT1/PTR family are highlighted in red with key functionally conserved residues highlighted by blue triangles. Nitrate binding site residues His356, Tyr388 are highlighted as magenta stars. Thr101 is conserved in all mammalian PTR family members and a small subset of bacterial homologues including PepT_So. AtNRT1.1 topology. The central cavity is shown as an open triangle, representing the inward open state observed in the crystal structure. The 12 TM helices identified from the crystal structures are colored from blue to red and the observed intracellular domain between TM6-7 is indicated in grey.

Extended data figure 2. Microscale thermophoresis binding assay as a technique to calculate K_D for ligand binding to NRT1/PTR transporters.

The accuracy of the microscale thermophoresis binding assay was examined through comparison with data obtained from transport assays. PepT_So was functionally reconstituted into liposomes as described and an IC₅₀ value calculated for Ala-Ala was determined and found to be ~ 50 μM (a, left). Using the NanoTemper monolith NT.115 instrument the K_D for Ala-Ala was calculated to be approximately 35 μM (a, right) similar to the IC₅₀ calculated from proton-driven transport assays. In this assay the orientation of the PepT_So molecules are highly likely to adopt multiple conformations, as observed for other members of the MFS ⁵⁴, and so the K_D we have determined is a global parameter for this protein and should be the same for either conformation of the transporter.

The binding of nitrate to AtNRT1.1 was determined at pH 7.0, 6.5 and 6.0 the K_D calculated at each of these pH values did not dramatically differ from one another (b). However as the pH was reduced the fluorescent signal produced by GFP was partially quenched, therefore all subsequent analysis was performed at a pH of 6.5. Nitrite, alanine, di-alanine, phosphate and sulphate show no observable interaction with AtNRT1.1 using the MST binding assay under the same conditions as for nitrate. Chlorate was shown to bind AtNRT1.1 with a K_D of ~ 5 mM (c). Data shown is representative of three independent experiments and mean and standard deviations are shown.

Extended data figure 3. Purification, secondary structure analysis and crystallization of AtNRT1.1.

Gel filtration trace of the purified AtNRT1.1 after TEV cleavage of the GFP^His (a) the protein elutes as a single peak corresponding to a size of approximately 65 KDa following subtraction of the DDM micelle (~50 KDa) which would correspond to a monomer in solution. The protein was purified to homogeneity as judged by coomassie stained SDS PAGE (b). Analysis of secondary structure content of AtNRT1.1 using circular dichroism spectroscopy shows the protein is predominately alpha helical with 22 % disordered regions and no significant beta strands (c). Typical crystal obtained with scale bar at 100 μm (d). AtNRT1.1 crystallizes as a dimer in the asymmetric unit. The buried surface area between the two monomers is 2135 Ų as calculated using PDBe PISA (http://www.ebi.ac.uk/msd-srv/prot_int/) and may be physiologically significant as both monomers are in the same orientation. Structural comparisons between bacterial members of the PTR family in different conformations of the transport cycle suggest the N-terminal bundle is less flexible compared to the C-terminal helices ²¹, which may explain why we observe AtNRT1.1 packing through these helices in the crystal. The oligomeric state of the PTR family in the membrane is currently unclear, with a recent bacterial member conjectured to form a tetramer ³⁰. In this case however, the dimer interface was found to be via the C- rather than the N-terminal bundle.

Extended data figure 4. Extracellular gate of AtNRT1.1 is stabilized through hydrophobic interactions as opposed to the salt bridges observed in the bacterial and mammalian members of this family.

AtNRT1.1 is shown in the plane of the membrane with a slice through the protein volume highlighting the central nitrate binding site and the exit tunnel. The extracellular gate formed from TMs 1 and 2 packing against TM 7 and 8 is highlighted and the interface between these helices is illustrated using electrostatic surface representations.

Extended data figure 5. Representative electron density maps used during model building.

Stereo view of final 2mFo-DFc electron density from refinement, contoured at 1 σ around chain A (PDB: 4CL5). View shown is equivalent to figure 2B in main text (PDB: 4CL4). Nitrate binding site: The position of modeled nitrate was resolved much better in the electron density maps in chain A compared to chain B, possibly due to higher occupancy in this molecule and the analysis in the paper was therefore conducted on chain A. In chain B the nitrate was seen to occupy a slightly different position within the binding site, displaced down ~ 1.7 Å compared to the position in molecule A but still sitting close to and interacting with His356. Given the resolution is 3.7 Å; it is difficult to ascertain the significance of this observation with respect to the coordination of nitrate. The low K_D value determined from the binding experiments could be consistent with nitrate being able to adopt slightly different positions within the binding site with respect to His356.

Extended data figure 6. Structural comparison between nitrate binding sites in AtNRT1.1 and the bacterial NNP family members NarK and NarU.

Although AtNRT1.1 shares no sequence identity to the Nitrate-Nitrite Porter (NNP) family, they do both belong to the Major Facilitator Superfamily, and share the same canonical architecture of 12 trans membrane helices ⁴. Shown in panel a is the superposition of both NarU (from E. coli) and NarK (from E. coli), whose crystal structures were recently published in complex with nitrate and nitrite respectively with AtNRT1.1 ^31,55.

Consistent with previously determined crystal structures for ligand bound MFS family members the ligand-binding site is located in the center of the molecule, constructed from side chains contributed by the N- and C-terminal bundles. In AtNRT1.1 the nitrate in chain A is located much further into the cavity, ~ 9 Å, than observed in either NarU or NarK and makes far fewer interactions with side chain functional groups. However, as discussed in the main text the binding sites do exhibit similar physiochemical features. As illustrated in the close up of the binding sites (b) and depicted in the cartoon schematic below (c), the nitrate in NarU is held tightly by two electrostatic interactions to Arg87 and Arg303 and by two further hydrogen bonds to Asn173 and Tyr261 (yellow sticks). These interactions are further sandwiched between two layers of hydrophobic residues, which will likely increased the relative strength of the electrostatic interactions by lowering the dielectric constant in the immediate vicinity of the nitrate. By comparison and as discussed in the main text, in AtNRT1.1 the nitrate is also held by an electrostatic interaction with a protonated His356 with contributions in chain A from Thr360 (purple sticks). The immediate vicinity of the nitrate in the present conformation of AtNRT1.1 is also hydrophobic and highlighted (purple sticks & black oval). However, as shown in the cartoon schematic, the number and strength of these interactions are substantially lower compared to the NNP family of transporters, potentially explaining the much lower K_D we observe for AtNRT1.1 (1 mM) compared to NarU (33 μM).

Extended data figure 7. Superposition of AtNRT1.1 onto the outward facing crystal structure of the fucose permease, FucP.

The coordinates for the AtNRT1.1 structure was divided into the N-and C-terminal six helix bundles, coloured blue and orange respectively, before being superimposed onto the structure of FucP (PDB: 3O7Q) using the CCP4 Superpose program. The r.m.s.d values are indicated. Helices 2, 7 and 11 were removed for clarity. Superposition of both N and C bundles of AtNRT1.1 onto the proton coupled fucose transporter FucP places Lys164 in the same position as a conserved Glu135 in FucP, which upon protonation is predicted to trigger the switch to the inward facing state ²⁹. Similarly, in superimposing the respective C terminal domains, Glu476 of AtNRT1.1, which this study suggests is involved in structural aspects of nitrate binding through an electrostatic interaction with His356 (see Fig. 2B and Extended Data Fig. 5), is close to Tyr365 in FucP which is the proposed interaction partner of Glu135. A speculative function of Lys164 may therefore be to form a salt bridge with Glu476 in the outward open conformation (Fig 4. panel A main text) facilitating the closer packing of helices H4 and H10 and sealing the binding site to the interior of the cell in the outward facing state. Supporting this interaction in the POT family are mutational data from both PepT_St and GkPOT, which show these residues are important for proton driven peptide transport in the POT family ^21,22.

Extended data figure 8. Proton transport is increased in the Thr101Asp variant of AtNRT1.1 when compared to wild type protein.

A reconstituted proteoliposome system was used to study the effect of the Thr101Asp variant on nitrate uptake. In this assay the membrane impermeant pH indicator, pyranine (trisodium 8-hydroxypyrene-1,3,6-trisulfonate) was trapped within the AtNRT1.1 reconstituted liposomes (a). In the presence of a pH gradient, AtNRT1.1 couples the import of nitrate with protons into the lumen of the liposome. The trapped pyranine becomes protonated (b) whereupon the excitation maximum shifts from 460 to 415 nm, the emission maximum remaining constant at 510 nm. Therefore proton coupled nitrate transport can be measured as a function of the proton dependent quenching of pyranine.

Protein was reconstituted into liposomes as detailed in the materials and methods and assays were performed on the same amount of protein for both the WT, Thr101Asp and His356Ala variants (c). Representative data obtained from the nitrate transport assays showing three independent experiments for both protein free liposomes and those containing the variants were studied (d). Nitrate uptake was driven using a pH gradient (pH 6.8 inside, 6.0 outside) as the addition of valinomycin caused an anomalous proton leak in AtNRT1.1 containing liposomes. The difference between the addition of nitrate (blue) verses a no nitrate control (black) is shown, data were normalised to the values obtained before the addition of nitrate. Due to the use of a plate reader to measure the experiments, after the addition of nitrate there is a slight delay (~10 seconds) before the first measurement is taken (time zero) which is why at time zero a drop in fluorescence is observed. The presence of WT AtNRT1.1 caused a larger change in fluorescence (indicative of acidification) in the presence of nitrate, which was not seen with the inactive His356Ala variant. The Thr101Asp variant however caused a substantial increase in acidification indicative of more protons and therefore nitrate being transported by this variant.

Extended data figure 9. Effects of mutating Thr87 in PepT_So on protein stability, peptide binding and proton coupled peptide uptake in a reconstituted proteoliposome system.

The N terminal domains of AtNRT1.1 (also shown in Fig. 3a) and the bacterial peptide transporters (a) are very similar. The location of Thr101 and the equivalent threonine in PepT_So, Thr87 are highlighted as magenta spheres, both threonines are located in a similar hydrophobic cavity formed at the intracellular ends of TMs 2 and 4. Thr87Asp variant of PepT_So showed a four-fold increase transport rate (b) where as the Thr87Arg variant was inactive. Mutation to alanine, serine, or lysine had no significant impact on the rate of transport (c) Error bars represent S.D. from three independent experiments. Comparison of substrate specificity between WT and The87Asp variant showed no significant change indicating the binding site has not been adversely affected (d) Error bars represent S.D. from three independent experiments. The stability as assessed using a high performance circular dichroism spectrometer (Jasco J-1500 instrument) of the Thr87Asp (Tm 52 °C) and Arg (Tm 46 °C) mutants is reduced when compared to WT protein (56 °C) (e). Loss of alpha helical secondary structure was monitored continuously at 220 nm while the temperature was increased from 20 °C to 80 °C over 1 degree increments. 10 μg of protein were monitored in a 100 μl cuvette. (f) Binding isotherms showing there is no change in affinity for peptide upon mutating Thr87 to Asp.