Research ArticleChannels

A Plant Homolog of Animal Glutamate Receptors Is an Ion Channel Gated by Multiple Hydrophobic Amino Acids

See allHide authors and affiliations

Science Signaling  11 Jun 2013:
Vol. 6, Issue 279, pp. ra47
DOI: 10.1126/scisignal.2003762

Abstract

Ionotropic glutamate receptors (iGluRs) are ligand-gated cation channels that mediate neurotransmission in animal nervous systems. Homologous proteins in plants have been implicated in root development, ion transport, and several metabolic and signaling pathways. AtGLR3.4, a plant iGluR homolog from Arabidopsis thaliana, has ion channel activity and is gated by asparagine, serine, and glycine. Using heterologous expression in Xenopus oocytes, we found that another Arabidopsis iGluR homolog, AtGLR1.4, functioned as a ligand-gated, nonselective, Ca2+-permeable cation channel that responded to an even broader range of amino acids, none of which are agonists of animal iGluRs. Seven of the 20 standard amino acids—mainly hydrophobic ones—acted as agonists, with methionine being most effective and most potent. Nine amino acids were antagonists, and four, including glutamate and glycine, had no effect on channel activity. We constructed a model of this previously uncharacterized ligand specificity and used knockout mutants to show that AtGLR1.4 accounts for methionine-induced membrane depolarization in Arabidopsis leaves.

Introduction

Ionotropic glutamate receptors (iGluRs) are glutamate-gated nonselective cation channels that mediate fast excitatory neurotransmission in the central nervous system of vertebrates. In 1998, iGluR homologs were discovered in an organism lacking a nervous system: the model plant Arabidopsis thaliana (1). These proteins, which were termed AtGLRs, occur in a remarkable diversity, comprising 20 subunits in three subgroups (2). Similar proteins have also been identified in other plants, such as rice and radish (3, 4). In planta studies provided evidence that GLRs might be involved in the response to light (1, 5), control of the allocation of Ca2+ in various pools (6), sensing of the carbon-to-nitrogen status (7), regulation of plant hormone biosynthesis and water balance (7, 8), coordination of mitotic activity during root development (3, 9, 10), pollen tube growth (11), resistance against fungal infection (4), and the response to aluminum toxicity (12).

The similarity to animal iGluRs prompted the assumption that plant GLRs function as glutamate-activated cation channels. Indeed, glutamate, glycine, and d-serine—co-agonists of animal N-methyl-d-aspartate (NMDA)–type iGluRs—induced depolarization, Ca2+ influx, and cation currents in various cell types of plants (11, 1317). In transgenic plants, it was shown that depolarization and Ca2+ influx induced by glutamate and other amino acids in root cells depend on the subunit AtGLR3.3 (18, 19) and that AtGLR1.2 is involved in the generation of cytosolic Ca2+ oscillations in Arabidopsis pollen tubes (11); however, whether these proteins form the channel or function as channel regulators is unknown. AtGLR3.4 and AtGLR3.7 expressed in Xenopus oocytes appeared to be constitutively active cation channels, yet activation by agonists had never been observed (15, 20). A patch-clamp study in transfected human embryonic kidney (HEK) 293 cells showed that AtGLR3.4 is a Ca2+-permeable cation channel activated by asparagine, serine, and glycine but not by glutamate (21). Using a domain transplantation approach, we had previously demonstrated that AtGLR1.1 and AtGLR1.4 have functional nonselective cation channel domains when expressed in Xenopus oocytes (22). Here, we report the cloning of the full-length AtGLR1.4 from Arabidopsis C24 root tissue and analysis of its functional properties in the Xenopus oocyte heterologous expression system.

Results

AtGLR1.4—An amino acid–gated ion channel

Two in planta studies showed that amino acids other than glycine and glutamate caused GLR-dependent depolarization and Ca2+ fluxes in Arabidopsis (18, 19). Furthermore, all iGluR ligand-binding domains (LBDs) resemble bacterial periplasmic amino acid–binding proteins (PBPs)—a family of proteins that bind and transport various amino acids (23). Therefore, we tested all 20 standard amino acids for their ability to induce AtGLR1.4-mediated currents in Xenopus oocytes injected with the AtGLR1.4–encoding transcripts. Surprisingly, several different amino acids activated AtGLR1.4, albeit to different extents (Fig. 1 and Table 1). The most effective agonist was methionine, followed by tryptophan (43 ± 3% of methionine-induced currents), phenylalanine, leucine, tyrosine, asparagine, and threonine. Of these seven amino acids, methionine was not only the most effective but also the most potent agonist (Fig. 1, inset, and Table 1), with an EC50 of 7.3 ± 0.6 μM. Asparagine was the least potent, with an EC50 of 264 ± 7 μM. Norleucine, a structural analog of methionine with the sulfur atom replaced by a methylene group, was much less effective (17 ± 1% of methionine-induced currents) and less potent than methionine (EC50 = 26.4 ± 3.5 μM), indicating that the thioether group of methionine plays an important role in binding to the receptor. Methionine-induced currents were insensitive to extracellular pH changes between pH 5.7 and pH 9.2 (fig. S1). The relatively high efficacy of tryptophan raised the possibility that its derivatives, such as serotonin, melatonin (24), and the phytohormone indole-3-acetic acid, might also act as AtGLR1.4 agonists. None of these compounds, however, activated AtGLR1.4, nor did dopamine and acetylcholine, which are also present in plants (25) (fig. S2). Combined with the finding that many amino acids are effective agonists, this suggests that the primary functional groups of amino acids, that is, the α-amino and α-carboxyl groups, might be essential for binding to AtGLR1.4.

Fig. 1 Agonists of AtGLR1.4.

Relative efficacies of amino acids at AtGLR1.4. Current amplitudes induced by 1 mM of the indicated amino acid (3 mM for asparagine because of its low potency) were corrected for background currents evoked by the same amino acid in uninjected oocytes (typically between 0 and 5 nA) and normalized to the response induced by 1 mM methionine (88 ± 5 nA, ranging from 54 to 196 nA). Inset: Concentration-response curves for the seven effective amino acids. Data are presented as means ± SEM from three to eight oocytes. See Table 1 for EC50.

Table 1 Efficacies and potencies of amino acids at AtGLR1.4.

Agonist efficacies were calculated from current amplitudes induced by 1 mM of the respective amino acid (3 mM for asparagine), corrected for background currents evoked by the same amino acid in uninjected oocytes, and normalized to the response induced by 1 mM methionine. Potencies are given as EC50 values determined from concentration-response curves. All data are presented as means ± SEM from n oocytes. n/a, not available.

View this table:

Antagonists of AtGLR1.4

The ability of a number of amino acids to act as agonists at AtGLR1.4 with various efficacies and potencies led us to the hypothesis that some of the ineffective amino acids might act as competitive antagonists. We analyzed potential inhibition by co-applying these amino acids with methionine. Indeed, many of the amino acids that were inactive as agonists inhibited methionine-induced currents to different extents (Fig. 2, A and B). Strikingly, glutamate, glycine, and aspartate did not. Increasing the methionine concentration reduced the inhibition by other amino acids, indicating that they indeed competed with methionine at the same binding site (Fig. 2, A and B). We investigated this behavior in more detail for arginine, which was the most effective amino acid antagonist. Analysis of concentration-response curves for the agonist methionine in the presence of varying concentrations of the antagonist arginine confirmed that arginine acted competitively, with a dissociation constant Kb of 3.5 ± 0.3 μM (Fig. 2, C to E).

Fig. 2 Antagonists of AtGLR1.4.

(A) Inhibition of methionine-induced currents by 1 mM of amino acids or competitive iGluR antagonists. Black columns, inhibition of currents induced by 100 μM methionine; light gray columns, inhibition of currents induced by 1 mM methionine. Methionine-induced currents had amplitudes ranging from 49 to 91 nA. Data are presented as means ± SEM from three to six oocytes. (B) Representative current traces showing the inhibition of methionine-induced currents of AtGLR1.4 by antagonists. Black traces, currents induced by 100 μM methionine; gray traces, currents induced by 1 mM methionine. To facilitate comparison of antagonist efficacies, we normalized all responses to the amplitudes before antagonist application. Vertical scale bars next to each trace represent an absolute current amplitude of 20 nA. The presence of methionine (M) and the amino acid (single-letter abbreviation) is indicated above each set of traces. (C) Concentration-response curves of AtGLR1.4 for the agonist methionine in the presence of the indicated concentrations of the antagonist arginine. Solid curves have individual slopes; dashed curves are constrained to a common slope. (D) Effect of increasing arginine concentrations on logEC50 values for methionine. The curve shows the result of nonlinear regression to determine Kb. (E) Clark plot displaying the effect of arginine on logEC50 values for methionine. The straight line represents ideal competitive antagonism calculated from the Kb determined in (D). (F) Concentration-inhibition curves for the noncompetitive antagonists DNQX, philanthotoxin-7,4, and MK-801. All data in (C) to (F) are presented as means ± SEM from three to five oocytes.

To test whether competitive antagonists of animal iGluRs also inhibited this plant GLR, we investigated the effect of two such antagonists, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 6,7-dinitroquinoxaline-2,3-dione (DNQX) (26), on AtGLR1.4-mediated currents. Addition of either CNQX or DNQX inhibited methionine-induced currents of AtGLR1.4, with DNQX being more effective (Fig. 2, A and B). The IC50 value (half-maximal inhibitory concentration) for DNQX was 869 ± 66 μM (Fig. 2F). Surprisingly, the extent of inhibition was the same when the methionine concentration was 100 μM or 1 mM (Fig. 2, A and B), indicating that CNQX and DNQX were not functioning as competitive antagonists and thus act differently on AtGLR1.4 than on animal iGluRs.

The ion pores of Ca2+-permeable AMPA and kainate receptors are blocked by extracellular polyamine toxins, such as philanthotoxin (27, 28). For AtGLR1.4, we observed a relatively weak, reversible block: 100 μM philanthotoxin-7,4 blocked 49 ± 0.3% of methionine-induced currents (Fig. 2F). At Ca2+-permeable AMPA and kainate receptors, a complete block is achieved with lower concentrations. The NMDA receptor open channel blocker MK-801 (29) also inhibited methionine-induced currents through AtGLR1.4, but 100 μM was required to reversibly block 35 ± 4% of methionine-induced currents (Fig. 2F). In contrast, NMDA receptor–mediated currents are completely and irreversibly blocked by low micromolar concentrations of MK-801. The potencies of philanthotoxin-7,4 and MK-801 at AtGLR1.4 were similar, with IC50 values of 96 ± 9 μM and 120 ± 16 μM, respectively (Fig. 2F).

The ligand-binding domain of AtGLR1.4

The unusual, unprecedented agonist specificity we found for AtGLR1.4 suggested that this subunit actually should not be termed a glutamate receptor, because it acts as an amino acid sensor with broad but distinct specificity. Glutamate, glycine, and aspartate, which all can activate certain animal iGluRs (30), were among the completely ineffective amino acids. In general, amino acids with bulky, hydrophobic side chains were the most effective agonists at AtGLR1.4. To gain insight into the structural basis of this specificity, we calculated a homology model of the AtGLR1.4 LBD on the basis of the crystal structure of the LBD of the rat NMDA receptor subunit GluN1 (31), which of the animal iGluR LBDs with known structure has the highest sequence identity with the AtGLR1.4 LBD (fig. S3). This initial model suggested that the α-amino and α-carboxyl groups of methionine might interact with residues D499, T501, and Q659, and the methionine side chain might interact with residues F658, F702, and V703 (Fig. 3A).

Fig. 3 Effects of point mutations in the AtGLR1.4 LBD.

(A) Initial homology model of the LBD of AtGLR1.4 in complex with the agonist methionine based on the GluN1 LBD crystal structure [Protein Data Bank (PDB) ID 1PBQ] (31). The D1 domain is highlighted in blue, and the D2 domain in red. The agonist and LBD residues within a radius of 4 Å are shown in stick representation. In the detailed view, the six residues predicted to interact with the ligand are labeled in different colors. (B) Methionine (300 μM)–induced steady-state current amplitudes of AtGLR1.4 wild type (WT), EGFP-tagged AtGLR1.4 (WT-EGFP), and EGFP-tagged AtGLR1.4 LBD point mutants expressed in oocytes. (C) Concentration-response curves for methionine of EGFP-WT and EGFP-tagged LBD point mutants expressed in oocytes. (D) Confocal cross sections of two oocytes demonstrating plasma membrane localization of the nonfunctional D499A-EGFP mutant. As controls, oocytes expressing nontagged AtGLR1.4 (WT, top) or WT-EGFP (bottom) were used. (E) Relative efficacies of the indicated agonist amino acids at EGFP-tagged LBD point mutants expressed in oocytes. Current amplitudes induced by saturating concentrations of the respective amino acid were corrected for background currents evoked by the same amino acid in uninjected oocytes (typically between 1 and 5 nA) and normalized to the response induced by 1 mM methionine. NL, norleucine. (F) Refined homology model of the AtGLR1.4 LBD in complex with the agonist methionine resulting from molecular dynamics simulation using the initial model shown in (A) as the starting point and incorporating the results of the mutagenesis study. The presumed sulfur-aromatic interaction between the sulfur atom of methionine and F658 of AtGLR1.4 is highlighted as a dotted line in magenta. Data in (B), (C), and (E) are presented as means ± SEM from 2 to 11 oocytes, with only the concentration-response curve in (B) for the F702A mutant from only two oocytes. Color coding of the mutated residues is the same in all panels.

To check the validity of our model, we individually mutated to alanines these six residues that were predicted to shape the ligand-binding pocket. Analysis of the point mutants, which were expressed as fusion proteins with enhanced green fluorescent protein (EGFP) in the Xenopus oocyte system, revealed that D499 and T501 critically influence agonist binding. The receptor with the T501A mutation was functional, but steady-state current amplitudes were decreased and methionine potency was reduced more than 40-fold compared to that observed with the wild-type channel coupled to EGFP (Fig. 3, B and C). The D499A mutant was not functional (Fig. 3B), although we confirmed that it was produced and targeted to the plasma membrane of the oocytes (Fig. 3D). All other mutations reduced the amplitude of methionine-induced currents to different extents (Fig. 3B), but only slightly affected methionine potency (Fig. 3C), suggesting that D499 and T501 are the main determinants of agonist affinity. To assess the influence of the presumed binding site residues on ligand specificity, we analyzed the relative efficacies of tryptophan, phenylalanine, leucine, and norleucine (current compared to methionine-induced current) in activating the functional point mutants (Fig. 3E). Q659, F702, and V703 appear not to be critical for agonist binding, because their mutation to alanines resulted in a receptor with a specificity pattern similar to that of the wild-type channel (Fig. 3E). By contrast, mutation of F658 to alanine changed the specificity pattern by increasing the relative efficacies of leucine and norleucine and decreasing those of tryptophan and phenylalanine (Fig. 3E). This suggests that the aromatic ring of F658 interacts with the electron-rich side chains of methionine, tryptophan, and phenylalanine and thus stabilizes their binding.

Incorporating the information that D499, T501, and F658 probably interact with the ligand, we refined our initial homology model by a molecular dynamics simulation. During the first 5 ns of this simulation, the ligand’s side chain reoriented, resulting in a conformation of the protein in complex with methionine that was stable for the rest of the simulation (over 55 ns). The final model shows the typical structure of iGluR LBDs (23): Two globular domains, D1 and D2, are connected by a hinge-like region consisting of two short sequence stretches with the agonist bound in a cleft between the two globular domains (Fig. 3F). According to our model, the α-amino and α-carboxyl groups of methionine bind to charged and polar residues, as well as to backbone amino and carbonyl groups in D1 (residues D499, T501, and R506 in Fig. 3F). These interactions are similar to those observed in rat iGluRs (23). Of the three predicted interacting residues, one (R506 in AtGLR1.4) is completely conserved among all Arabidopsis and rat iGluRs, and another one (T501) is conserved in the majority of subunits (fig. S4). The third residue (D499), however, shows a clear distinction between Arabidopsis and rat iGluRs: The former all have an aspartate, and most of the latter have a proline or serine, but never an aspartate, at that position. Our model supports a previous prediction that this exchange stabilizes ligand binding in AtGLRs compared to rat iGluRs by an additional interaction with the side chain of aspartate (14).

Nonselective cation channel properties of AtGLR1.4

Animal iGluRs are cation channels that hardly discriminate between monovalent cations, but the permeabilities to divalent cations, such as Ca2+ and Ba2+, differ depending on the particular subunits forming the channel. The ion pore region of plant GLRs differs considerably from that of animal iGluRs, making prediction of ion selectivity impossible. To estimate monovalent cation permeabilities, we recorded current-voltage (I/V) relationships of methionine-induced currents in extracellular solutions containing 1.8 mM Ca2+ and 115 mM of Na+, K+, Rb+, Cs+, or NH4+ (Fig. 4A). All I/V relationships were virtually identical, with no shift of reversal potentials, demonstrating that AtGLR1.4 was equally permeable to all the tested monovalent cations and thus as unselective as animal iGluRs. Furthermore, when expressed in Xenopus oocytes, AtGLR1.4 was strongly inwardly rectifying: Outward currents at positive membrane potentials were almost completely blocked (Fig. 4A). Similar properties are observed for Ca2+-permeable AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionate) and kainate receptors, which display a voltage-dependent block by intracellular polyamines causing inward rectification (32). However, animal iGluRs show virtually linear I/V curves at negative membrane potentials, whereas the slope of the AtGLR1.4 I/V curve constantly increased at more negative potentials. We had also observed such an I/V curve for the ion pore of AtGLR1.1, which we had characterized by putting the ion pore domain into an animal iGluR, creating a chimeric receptor (22).

Fig. 4 Electrophysiological properties of AtGLR1.4.

(A) Monovalent cation permeabilities of AtGLR1.4. Current-voltage relationships were recorded in extracellular solutions containing 115 mM of the indicated salt, 1.8 mM CaCl2, and 10 mM Hepes (pH 7.2). NaGluc, sodium gluconate. The curves are normalized to the amplitude at −150 mV and presented as means from the indicated number of oocytes. (B) Ca2+ permeability of AtGLR1.4. Current-voltage relationships were recorded in extracellular solutions containing CaCl2 at the indicated concentrations, 10 mM KCl, 10 mM Hepes (pH 7.2), and mannitol for osmotic balance. Current amplitudes at −100 mV in these low-salt solutions ranged from 4 to 23 nA. The curves are normalized to the amplitude at −120 mV and presented as means (solid lines) ± SEM (shading) from three oocytes.

Because GLRs contribute to Ca2+ influx (4, 6, 11, 13, 15, 18, 19, 22), we tested AtGLR1.4 for Ca2+ permeability. Methionine-induced current amplitudes in a Hepes-buffered 80 mM CaCl2 solution containing no monovalent cations and Ca2+ as the only permeable cation were 63 ± 6% (n = 13) of the amplitudes in normal frog Ringer’s solution (NFR), demonstrating that AtGLR1.4 is Ca2+-permeable. In a solution containing Ca2+ and 10 mM K+, the reversal potential shifted by ~+20 mV each time when the extracellular Ca2+ concentration was increased by a factor of 10 from 0.2 to 2 mM and from 2 to 20 mM (Fig. 4B). Thus, when expressed in the oocyte system, AtGLR1.4 is permeable to Ca2+ in a physiological concentration range in the presence of a physiological concentration of K+. Although some animal iGluR variants have been reported to be Cl permeable (33), AtGLR1.4 was not permeable to Cl, as indicated by identical reversal potentials in Cl-free and Cl-containing extracellular solutions (Fig. 4A).

Methionine-induced depolarization mediated by AtGLR1.4 in Arabidopsis seedlings

After characterizing AtGLR1.4 in a heterologous system, we investigated the relevance of these results in plants. Although we demonstrated that GFP- or yellow fluorescent protein (YFP)–tagged AtGLR1.4 was localized to the plasma membrane in Xenopus oocytes (Fig. 3) and in HEK293 cells (fig. S5), we needed to determine whether AtGLR1.4 was targeted to the plasma membrane or other membranes, such as the tonoplast (the vacuolar membrane), in plant cells to assess its physiological function. We determined the subcellular localization of AtGLR1.4 by transiently overexpressing a GFP-tagged form in mesophyll protoplasts of Arabidopsis leaves. GFP-tagged AtGLR1.4 was detected in the plasma membrane 24 to 48 hours after transformation and colocalized with the red fluorescent protein (RFP)–tagged plasma membrane marker StREM1.3 (34) (Fig. 5A and fig. S6A), but was not detectable in the membrane of vacuoles released by lysis of protoplasts (fig. S6B). Because YFP-tagged AtGLR3.4 overexpressed in Arabidopsis protoplasts is targeted to the same compartments as the nontagged wild-type protein in nontransfected plants (35) and the GFP tag did not alter the properties nor expression of AtGLR1.4 in our experiments in the oocyte system, our results with the protoplasts indicate that AtGLR1.4 is located in the plasma membrane in wild-type Arabidopsis plants.

Fig. 5 Subcellular localization and function of AtGLR1.4 in planta.

(A) Representative confocal cross section of an Arabidopsis leaf mesophyll protoplast transiently expressing EGFP-tagged AtGLR1.4 and the RFP-tagged plasma membrane marker StREM1.3. From left to right: Fluorescence of AtGLR1.4-EGFP, fluorescence of StREM1.3-RFP, overlay of AtGLR1.4-EGFP and StREM1.3-RFP fluorescence, bright-field image. (B) Maximal depolarization determined from membrane potential recordings from intact cotyledons of the AtGLR1.4 knockout (KO) mutants (light purple and light turquoise traces) and WT Arabidopsis [ecotype Col-0 (dark purple traces) and a segregating WT derived from heterozygous glr1.4-2 mutant plants (dark turquoise traces)] in response to 500 μM methionine (left) or glutamate (right). The WTs display two distinct methionine response populations, which are shown with separate means ± SEM. For the WTs, the columns represent the means of the combined data of both populations. (C) Representative membrane potential recordings from the experiment in (B). For Col-0 (WT-1): representative methionine-activated trace of the low-responding population; for the segregating WT (WT-2): a trace of the high-responding population. (D) AtGLR1.4 gene structure and position of T-DNA insertions in two independent homozygous AtGLR1.4 knockout lines. Primers used for mutant line verification are indicated. (E) Polymerase chain reaction (PCR)–mediated verification of identity and homozygosity of AtGLR1.4 knockout mutant lines. With gene-specific primers P1 and P2 or P3 and P4, a fragment of the AtGLR1.4 WT gene was amplified, whereas insertion of the T-DNA prevented amplification in knockout mutants. With primer Pt1 or Pt2 binding to the left border of the T-DNA insert, a fragment was only amplified for the knockout mutant. (F) Absence of the AtGLR1.4 transcript in the knockout mutant lines verified by reverse transcription PCR (RT-PCR). To amplify an AtGLR1.4 fragment, we used primers P5 and P6 or P3 and P4 for glr1.4-1 and glr1.4-2, respectively. The constitutively expressed actin-2 partially amplified with specific primers served as a reference.

This plasma membrane localization suggested that AtGLR1.4 may play a role in sensing extracellular amino acids, and its biophysical properties determined in the oocyte system suggested that its activation by amino acids in planta should result in changes in membrane potential. Therefore, we performed membrane potential recordings on Arabidopsis Col-0 seedlings. Application of 500 μM methionine caused a transient depolarization of 56 ± 8 mV, and 500 μM glutamate induced a larger depolarization of 117 ± 5 mV (Fig. 5, B and C). To assess whether the methionine-induced depolarization involved AtGLR1.4 activity, we recorded the membrane potential from two independent AtGLR1.4 knockout lines (Fig. 5, B to F) and from a segregating wild type derived from one of these knockout lines. The segregated wild type showed responses identical to those observed in the Col-0 seedlings (57 ± 8 mV for methionine; 107 ± 5 mV for glutamate) (Fig. 5, B and C). In both knockout lines, however, methionine-induced depolarizations were strongly reduced to 18 ± 3 mV and 22 ± 3 mV, whereas glutamate-induced depolarizations did not differ significantly from those of wild-type plants (106 ± 4 mV and 104 ± 5 mV) (Fig. 5, B and C).

Mesophyll cells of wild-type plants exhibited two distinct responses to methionine. About 60% of the tested cells from wild-type plants displayed low responses to methionine that were only slightly higher than those of cells from AtGLR1.4-knockout plants (30 ± 2 mV and 35 ± 4 mV; Fig. 5B), whereas the remaining 40% responded much more strongly, reaching depolarizations that were almost as high as those induced by glutamate (101 ± 3 mV and 93 ± 7 mV; Fig. 5B). These two populations may result from differences in the abundance of AtGLR1.4, with the low-responding cells having little or no AtGLR1.4. This finding matches data by Roy et al. who detected strongly variable expression of AtGLR1.4 in different cells of the same plant, as well as different individual plants (20). Despite the variability, we showed that AtGLR1.4 accounted for the major fraction of methionine sensitivity in Arabidopsis seedlings.

Discussion

Our studies in the Xenopus oocyte expression system revealed that AtGLR1.4 is a nonselective cation channel that reacts to various amino acids, with some acting as agonists and others acting as competitive antagonists. The residues in the LBD that, according to our modeling study, presumably bind the α-amino and α-carboxyl groups of the amino acid ligands are highly conserved among plant iGluR homologs (fig. S4), suggesting that all AtGLR subunits bind amino acids just like the animal iGluRs. This assumption is supported by the finding that AtGLR3.4 is activated by asparagine, serine, and glycine (21).

Specificity for particular amino acids is difficult to predict because the area of the LBD that binds the side chain of the amino acid ligand differs considerably between Arabidopsis and rat iGluRs and thus cannot be modeled easily. This may explain the erroneous prediction of glycine as the ligand of most AtGLRs, including AtGLR1.4, in a previous modeling study that provided no experimental data to be incorporated (14). Our AtGLR1.4 model, supported by the mutant study, indicates that the side chain of methionine binds mainly to a phenylalanine residue (F658) within a large hydrophobic pocket in D2. Such sulfur-aromatic interactions have already been shown to occur between amino acid residues within proteins as well as between proteins and ligands (36, 37). In general, van der Waals interactions with hydrophobic residues seem to play an important role in binding of the bulky, mostly hydrophobic, amino acid side chains of ligands to the AtGLR1.4 LBD, as it is also observed in the Escherichia coli leucine/isoleucine/valine-binding protein (38). In animal iGluRs, on the other hand, the ligand’s side chain forms hydrogen bonds with main-chain atoms and polar residues (23).

Other AtGLRs might have different agonist profiles, as suggested by the sequence diversity within the area of D2 that binds the side chain of the agonist as well as by experimental studies. A number of amino acids, including alanine, asparagine, cysteine, glutamate, glycine, and serine, externally applied to Arabidopsis roots triggered ion fluxes in root apex cells dependent on the presence of AtGLR3.3 and partly AtGLR3.4 (18, 19). Whereas these investigations only relied on data from wild-type and GLR knockout plants, not allowing to rule out indirect effects of GLRs on other proteins, a recent study in HEK293 cells confirmed that AtGLR3.4 is indeed an ion channel activated by asparagine, serine, and glycine (21). Combined with our results, this supports the assumption that the ion fluxes activated in root cells by various external amino acids are directly mediated by GLRs. To explain the broad ligand specificity of AtGLR1.4 and particularly the gradual differences in efficacy of agonists and antagonists, more detailed structural information is necessary, as the generally low sequence identity between animal and plant iGluRs precludes precise modeling outside the highly conserved regions in D1.

Although it has been shown that glutamate, which is sensed by the root tip of Arabidopsis, modifies root growth and branching (9, 10), essentially nothing is known about methionine—the most effective amino acid agonist of AtGLR1.4—as a morphogenetic amino acid. The methionine-induced depolarizations that we observed in leaf cells of Arabidopsis seedlings raise the possibility that, in plants, methionine not only serves as a nitrogen source but also as a signaling molecule. Only 20% of the total methionine pool in plants serves as a building block for protein biosynthesis (39). The major part of the methionine pool is converted to S-adenosylmethionine, which not only serves as the methyl donor in many cellular transmethylation reactions but also constitutes the precursor of ethylene, a well-known stress hormone in plants (40). Studies suggest that GLRs provide the molecular pathway for calcium signals during biotic stress (41). Furthermore, proteomic studies identified cobalamin-independent methionine synthase (AtMS1) as secreted into the extracellular space during bacterial infections, providing an explanation how methionine can be produced extracellularly under conditions of biotic stress (42). Future studies will investigate the possible involvement of methionine and AtGLR1.4 in innate immunity-related responses.

Alternatively to being part of a signaling cascade that involves endogenously produced agonists that are either secreted or generated extracellularly by the plant, AtGLR1.4 may detect methionine and other amino acids that occur in the environment. Such sensing of external chemical signals as part of an evolutionarily conserved adaptation strategy to changing environmental conditions might represent a functional connection to bacterial PBPs (fig. S7). Evolutionary connections between iGluRs and PBPs had been suggested on the basis of sequence similarities (43). PBPs, however, occur in the periplasmic space, which is between the inner and outer membranes of bacteria, and scavenge or sense nutrients in the environment by coupling to transporters or chemotaxis receptors located in the inner membrane. Some PBPs show broad specificity for ligands, such as various amino acids and peptides, whereas the previously known iGluRs are highly ligand-specific ion channels within the synaptic membrane that mediate intercellular communication. Thus, an iGluR with broad specificity that senses external chemicals to signal changing environmental conditions would represent a functional intermediate between these two rather distant protein families. Another large family of proteins, called the ionotropic receptors (IRs), that are closely related to iGluRs and that sense external chemicals have been discovered in Drosophila (44). Like the plant GLRs, these insect IRs cannot be classified into the well-known families of iGluRs. They are present in sensory neurons and seem to serve as odorant receptors. As such, most of them lack the amino acid–binding residues common to synaptic iGluRs and plant GLRs.

Our characterization of AtGLR1.4 as a broad-range amino acid sensor with ion channel function opens new doors to the understanding of amino acid sensing in plants. Phylogenetic analysis of LBDs shows that, as previously described (45), the AtGLRs do not group with any of the animal iGluR subfamilies and split into three distinct clades (fig. S7). On the basis of the high degree of sequence identity within these clades and the complete conservation of the general amino acid–binding residues in all subunits, we hypothesize that the AtGLRs all serve as amino acid sensors, but likely with different ligand specificities, as observed for the animal iGluRs, which separate into distinct clades corresponding to their pharmacological subfamilies. iGluRs appear to have evolved in quite different directions to serve as sensors of tissue-intrinsic signals in cellular communication, as well as of direct external signals within the environment. It remains to be determined which of these two roles plant GLRs fulfill—or if they are involved in both.

Materials and Methods

Cloning and mutagenesis

RNA from A. thaliana C24 roots was prepared by a guanidinium thiocyanate procedure as previously described (22). First-strand complementary DNA (cDNA) was synthesized from this RNA with SuperScript II reverse transcriptase (Invitrogen) using oligo(dT) or random hexamer primers. The AtGLR1.4 cDNA was then amplified with Phusion DNA polymerase (Finnzymes) and cloned into the oocyte expression vector pSGEM (46). To append an EGFP or enhanced YFP (EYFP) tag to the C terminus of AtGLR1.4 for localization studies, we cloned the AtGLR1.4 coding sequence in-frame with its stop codon deleted into pEGFP-N1 and pEYFP-N1 (Clontech), respectively. To facilitate expression of EGFP-labeled AtGLR1.4 in oocytes, we subcloned the coding sequence of the AtGLR1.4-EGFP fusion protein from pEGFP-N1 into pSGEM. Point mutants were constructed by overlap extension PCR with mutagenic oligonucleotide primers (47). Complementary RNA (cRNA) was synthesized from 1 μg of linearized template DNA with the mMESSAGE mMACHINE in vitro transcription kit (Ambion).

Electrophysiological studies in Xenopus laevis oocytes

To obtain oocytes, we surgically removed parts of the ovaries from Xenopus laevis (Nasco) anesthetized with 3-aminobenzoic acid ethyl ester (1.5 g/liter; Sigma). To remove the follicular cell layer, the ovary clippings were digested with collagenase type I (784 U/ml; 4 mg/ml) (Worthington) in Ca2+-free Barth’s solution [88 mM NaCl, 1.1 mM KCl, 2.4 mM NaHCO3, 0.8 mM MgSO4, 15 mM Hepes-NaOH (pH 7.6)] with gentle agitation for 1.5 to 2 hours at 20°C. Collagenase digestion was stopped by extensive washing with Barth’s solution [88 mM NaCl, 1.1 mM KCl, 2.4 mM NaHCO3, 0.3 mM Ca(NO3)2, 0.4 mM CaCl2, 0.8 mM MgSO4, 15 mM Hepes-NaOH (pH 7.6)]. Defolliculated oocytes of stages V and VI were selected and maintained at 17°C in Barth’s solution supplemented with gentamicin (100 μg/ml), streptomycin (40 μg/ml), and penicillin (63 μg/ml). Oocytes were microinjected with 20 ng of cRNA with a nanoliter injector (WPI). Five to 8 days after cRNA injection, current responses were recorded under voltage clamp at −100 mV with a Turbo Tec-10CD amplifier (npi electronic) controlled by Pulse software (HEKA). Currents were filtered with a 20-Hz low-pass filter and then digitized with a sampling rate of 50 Hz. Recording electrodes pulled from borosilicate glass (Hilgenberg) with a PIP5 vertical pipette puller (HEKA) were filled with 3 M KCl and had resistances of 0.5 to 5 megohms (voltage electrode) or 0.5 to 1.5 megohms (current electrode). Recordings were performed in a 50-μl chamber under constant superfusion with extracellular solution at a flow rate of 3 to 5 ml/min. Unless otherwise stated, the extracellular solution was NFR [115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl2, 10 mM Hepes-NaOH (pH 7.2)]. Current-voltage relationships were recorded with 2-s voltage ramps from −150 to +50 mV at a sampling rate of 1 kHz in the presence of agonist, corrected for background conductances determined likewise in the absence of agonist, normalized to the amplitude at −150 mV, and then averaged. Agonist potencies were determined by measuring steady-state current responses induced by the application of increasing agonist concentrations to the same oocyte. The obtained responses were normalized to the maximal response induced by a saturating agonist concentration. The normalized data from each oocyte were fitted separately to the Hill equation using Prism (GraphPad). Finally, the resulting EC50 values for each agonist were averaged. To investigate antagonist efficacies, we first activated the receptor by a 40-s application of the agonist alone to determine the equilibrium response. Then, we co-applied the antagonist with the agonist for 20 s, followed by another 30- to 40-s application of the agonist alone to check the reversibility of inhibition. To determine the affinities of competitive antagonists, we recorded the concentration-response curves as described above in the presence of varying concentrations of antagonist. Antagonist binding constants (Kb) were then calculated from the logEC50 values by nonlinear regression (48).

Verification of AtGLR1.4 knockout lines

A transferred DNA (T-DNA) insertion line (N629955, generated at the Salk Institute) for AtGLR1.4 (locus At3g07520) was obtained from the Nottingham Arabidopsis Stock Centre (NASC) (49). Genomic DNA extraction and PCR amplification were performed with the Extract-N-Amp Plant PCR Kit (Sigma). Homozygosity was checked by PCR with gene-specific primers P1 (5′-ACACAGCTGTTTGATTGATTGACTTG-3′) and P2 (5′-TTGCAGGAGGTCCATGGCTGTTTA-3′) as well as a primer binding in the T-DNA insert (Pt1, 5′-TGGTTCACGTAGTGGGCCATCG-3′). Expression of the AtGLR1.4 gene was analyzed by RT-PCR (50) with primers P5 (5′-GATCTTGTGTACACACTGTAT-3′) and P6 (5′-CATATAAAGATCTGTTATCAGTA-3′). Actin-2 was amplified in parallel as a reference, with primers 5′-TCCAAGCTGTTCTCTCCTTG-3′ and 5′-GAGGGCTGGAACAAGACTTC-3′. A second AtGLR1.4 T-DNA insertion line was obtained from GABI-KAT (51) (ID 349A04). The homozygosity of this line was checked by PCR with gene-specific primers P3 (5′-AGTTCCAGTGATTTCTTCTTTCCA-3′) and P4 (5′-CTGACAAACCTGGTAACTTTGCAA-3′) as well as a primer binding in the T-DNA insert (Pt2, 5′-ATAATAACGCTGCGGACATCTACATTTT-3′). Expression of the AtGLR1.4 gene was analyzed by RT-PCR (50) with primers P3 and P4. Actin-2 was amplified in parallel as a reference, with primers 5′-GGTGATGGTGTGTCT-3′ and 5′-ACTGAGCACAATGTTAC-3′.

Electrophysiological studies in seedlings

A. thaliana plants were grown on 1% agar plates containing Murashige and Skoog medium (Duchefa) supplemented with 1% sucrose, with a 9-hour photoperiod at a photon flux density of 150 μmol m−2 s−1. Before electrophysiological recordings, 14-day-old seedlings were glued to the bottom of a 600-μl recording chamber and left to recover for at least 12 hours in extracellular solution [5 mM KCl, 1 mM CaCl2, 5 mM MES (pH 6.0) with bis-tris-propane]. The free-running membrane potential was then recorded from leaf cells impaled with a microelectrode using a NanoControl NC30 micromanipulator (Kleindiek Nanotechnik) and digitized with a sampling rate of 1 kHz. Recording electrodes pulled from borosilicate glass (Hilgenberg) with a P2000 horizontal laser puller (Sutter Instruments) were filled with 300 mM KCl, had resistances of 50 to 100 megohms, and were connected to HS-2-x0.01 headstages of an Axoclamp-2B microelectrode amplifier (Axon Instruments) through an Ag/AgCl half cell.

Confocal microscopy

EGFP-tagged receptors expressed in oocytes were visualized at room temperature with a TCS SP2 AOBS confocal microscope with a 20× HC PL APO 0.7 numerical aperture (NA) immersion objective (Leica). EGFP was excited at 488 nm and detected between 500 and 600 nm. To express EYFP-tagged receptors in HEK293 cells, we seeded 100,000 to 200,000 cells on polyornithine-coated coverslips in 35-mm cell culture dishes. One to 3 days after seeding, cells were transiently transfected with the cDNA of AtGLR1.4 fused to the N terminus of EGFP cloned into the expression vector pEGFP-N1 (Clontech). Transfection was performed with 2 μg of DNA using a modified calcium phosphate method as previously described (52). Four days after transfection, cells were fixed with formaldehyde and embedded in Fluoromount-G (SouthernBiotech). EYFP-tagged receptors were then visualized at room temperature with a TCS SP2 AOBS confocal microscope with a 63× HCX PL APO CS 1.4 NA oil immersion objective (Leica). EYFP was excited at 514 nm and detected between 525 and 600 nm. EGFP-tagged receptors expressed in leaf protoplasts isolated from 4-week-old Arabidopsis (53) were visualized at room temperature on an LSM 5 PASCAL confocal microscope (Zeiss) 24 to 48 hours after transformation, with excitation at 488 nm and detection between 520 and 560 nm. RFP was excited at 543 nm and detected with a 560-nm long-pass filter. For colocalization studies, scanning was performed in multitrack mode.

Molecular modeling

The AtGLR1.4 LBD homology model was calculated on the basis of the GluN1 LBD (PDB ID 1PBQ) (31) using the SWISS-MODEL server (54) and was energy-optimized using the GROMOS96 43B1 force field (55). Methionine was docked into a position analogous to that of the ligand in the GluN1 LBD structure, followed by energy optimization of the model. Molecular dynamics simulations were performed with YASARA version 9.2.27 with the YASARA2 highly accurate force field in default settings. The simulation cell was set at least 5 Å around the atoms of the protein and filled with water at a density of 1 g/cm3. The time step was set to 1 fs, and intermolecular forces were calculated every two simulation substeps.

Sequence alignment and phylogenetic analysis

For comparative analysis, the amino acid sequences of various iGluR subunits were retrieved from GenBank (table S1) and aligned using the ClustalW algorithm in MegAlign (DNASTAR) with modified parameters (gap penalty = 15.00, gap length penalty = 3.00, slow-accurate pairwise alignment, identity matrix). The resulting alignment was visually inspected, manually adjusted, and assembled with MacClade 4.08 (56). An unrooted phylogenetic analysis of the 329 amino acids of the LBD (S1 and S2 domains combined) was performed with MrBayes 3.1 (57), calculating 105 generations.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/6/279/ra47/DC1

Fig. S1. pH dependence of methionine-induced AtGLR1.4-mediated currents.

Fig. S2. Representative current responses of AtGLR1.4 upon application of signaling molecules that occur in plants.

Fig. S3. Sequence alignment of the LBDs of AtGLR1.4 and GluN1.

Fig. S4. Sequence alignment of the most conserved region of the LBDs of Arabidopsis iGluR homologs and rat iGluRs.

Fig. S5. Localization of AtGLR1.4 in HEK293 cells.

Fig. S6. Localization of AtGLR1.4 in mesophyll protoplasts isolated from Arabidopsis leaves.

Fig. S7. Phylogenetic tree of the LBDs of iGluRs and related proteins.

Table S1. GenBank accession numbers of iGluRs and related proteins used for phylogenetic analysis.

Reference (58)

References and Notes

Acknowledgments: We thank E. Weiler and M. Piotrowski for providing Arabidopsis plants; B. Peters, S. Nolte, and A. Mnich for technical assistance; I. Fuchs and J. Blachutzik for providing the StREM-RFP construct; and P. Walch-Liu, B. G. Forde, and M. Tester for support in the isolation of AtGLR knockout lines. Funding: Part of this work was supported by the Deutsche Forschungsgemeinschaft (FOR964 to D.B.) and the National Natural Science Foundation of China (grant nos. 30771288 and 30821003 to L.-H.L.). Author contributions: D.T. designed the study; performed cloning, mutagenesis, and all oocyte and HEK cell experiments; analyzed the data; and wrote the paper. U.A. designed and performed the in planta experiments and analyzed the corresponding data. L.-H.L. generated knockout plants. T.H. performed phylogenetic analyses. G.S. was involved in study design and performed molecular modeling. D.B. designed the in planta experiments, analyzed the corresponding data, and contributed to the discussion of in planta data. M.H. was involved in study design. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The sequence of AtGLR1.4 has been deposited with GenBank under accession no. KC589118.
View Abstract

Stay Connected to Science Signaling

Navigate This Article