Arabidopsis root architecture is highly responsive to changes in the nitrogen supply. External NO3− stimulates lateral root growth via a signalling pathway involving the ANR1 MADS box transcription factor, while the presence of exogenous l-glutamate (Glu) at the primary root tip slows primary root growth and stimulates root branching. We have found that NO3−, in conjunction with Glu, has a hitherto unrecognized role in regulating the growth of primary roots. Nitrate was able to stimulate primary root growth, both directly and by antagonising the inhibitory effect of Glu. Each response depended on direct contact between the primary root tip and the NO3−, and was not elicited by an alternative N source (NH4+). The chl1-5 mutant, which is defective in the NRT1.1 (CHL1) NO3− transporter, was insensitive to NO3− antagonism of Glu signalling, while an anr1 mutant retained its sensitivity. Sensitivity to NO3− was restored in a chl1-5 mutant constitutively expressing NRT1.1. However, expression in chl1-5 of a transport-competent but non-phosphorylatable form of NRT1.1 not only failed to restore NO3− sensitivity but also had a dominant-negative effect on Glu sensitivity. Our results indicate the existence of a NO3− signalling pathway at the primary root tip that can antagonise the root’s response to Glu, and they further suggest that NRT1.1 has a direct NO3− sensing role in this pathway. We discuss how the observed signalling interactions between NO3− and Glu could provide a mechanism for modulating root architecture in response to changes in the relative abundance of organic and inorganic N.
Root system architecture is one of the most important traits affecting the efficiency with which a plant explores the soil and captures limiting supplies of water and nutrients (Lynch, 1995). A significant feature of root architecture is its plasticity, arising from complex interactions with biotic and abiotic factors (Forde and Lorenzo, 2001; Malamy, 2005; Osmont et al., 2007). One of the best known examples of the plasticity of root development is the localised stimulation of root growth that occurs in many plant species when their roots encounter a localised source of NO3− (Forde and Lorenzo, 2001; Robinson, 1994). Studies with Arabidopsis have shown that stimulation of lateral root growth in NO3−-rich patches depends on perception by the lateral root tip of a NO3− signal leading to increased meristematic activity (Zhang and Forde, 1998; Zhang et al., 1999) and have identified the ANR1 MADS box transcription factor as a component of the NO3− signalling pathway (Zhang and Forde, 1998). More recently, the NRT1.1 (CHL1) NO3− transporter has been shown to have a direct or indirect role in NO3− sensing in the same pathway (Remans et al., 2006b). Other evidence of a regulatory function for the NRT1.1 protein has come from studies indicating its involvement in modulating the growth of nascent primary and lateral roots (Guo et al., 2001), in NO3− repression of the NRT2.1 nitrate transporter gene (Muňos et al., 2004) and in NO3− stimulation of germination (Alboresi et al., 2005).
Here we show that NO3− acts as an external signal to modulate primary root growth and root architecture in Arabidopsis, both directly and by acting as an antagonist of the response of the root to Glu. Using a mutant defective in the NRT1.1 nitrate transporter, and transgenic derivatives of this mutant, we also provide the clearest evidence to date that NRT1.1 has a direct role in NO3− sensing.
Nitrate antagonises the effect of l-glutamate on root growth and branching
We have reported that Arabidopsis roots are extremely sensitive to the presence of exogenous Glu, which inhibits primary root growth and stimulates lateral root growth behind the primary root tip (Walch-Liu et al., 2006). Subsequently, we observed that this response did not occur if the seedlings were grown on 0.5× MS medium (Murashige and Skoog, 1962) rather than the dilute B5 medium used previously (data not shown). To investigate whether the high concentrations of inorganic N present in 0.5× MS could be affecting Glu sensitivity, we examined the effect of treating Arabidopsis seedlings (C24) with 50 μm Glu in the presence and absence of 5 mm NO3−. As shown in Figure 1(a), in the presence of NO3− the effects of Glu on both primary root growth and root branching were completely suppressed.
The concentration dependence of the interaction between NO3− and Glu was investigated using two different accessions, C24 and Col-0. In these and subsequent experiments we used the inhibition of primary root growth as the most convenient measure of the sensitivity of the root system to exogenous Glu. Figure 1(b) shows that for C24 (which is the most Glu-sensitive accession known), 0.5 mm NO3− almost fully alleviated the strong inhibitory effect of 50 μm Glu, while 5 mm NO3− only partially alleviated the effect of 100 μm Glu. Thus an excess of NO3− appears to be necessary to overcome the Glu effect, although the response is not strictly dependent on the ratio of NO3− to Glu. In Col-0, where higher concentrations of Glu are required to affect root growth, 5 mm NO3− was necessary and sufficient to fully alleviate the Glu effect, even at 1 mm Glu (Figure 1c).
Signalling interactions between nitrate and l-glutamate occur at the primary root tip
We previously showed that the effect of Glu on primary root growth depended on direct exposure of the root tip to exogenous Glu (Walch-Liu et al., 2006). We were interested in whether NO3− similarly needs to be in direct contact with the root tip to alleviate Glu inhibition. The alternative possibility would be that the response is a whole-organism one, requiring the uptake and accumulation of NO3− (or its assimilates) in the plant. To distinguish between these options we used a segmented agar plate technique (Walch-Liu et al., 2006; Zhang and Forde, 1998), which allowed NO3− (or NH4+) treatments to be applied separately to the primary root tip and the mature part of the primary root (see Figure S1).
As shown in Figure 1(d), it was only when the primary root tip was in contact with the NO3− that the inhibitory effect of Glu was alleviated. Furthermore, having NO3− only in the bottom segment was as effective as having NO3− in both segments. Repeating the experiment with 5 mm NH4+ showed that this alternative N source had no influence on Glu sensitivity, regardless of which part of the root system received the treatment (Figure 1e). These results indicate that the effect is due to NO3− itself, and not a product of NO3− assimilation, and that the root tip must be in direct contact with the NO3−.
To confirm the specific role of the NO3− ion in alleviating the inhibitory effect of Glu, we used a double mutant (chl3-5/nia1-2) that lacks detectable nitrate reductase (NR) activity and is therefore unable to assimilate NO3− (Wang et al., 2004). As this NR-null mutant was derived from a cross involving two different genetic backgrounds, we compared its responses with those of both the relevant parents (Ler and Columbia). Figure 1(f) shows that the parental lines are equally sensitive to growth inhibition by 0.5 mm and 1 mm Glu, but Columbia is much more responsive to the antagonistic effect of NO3−. The NR-null mutant resembles Columbia in its responsiveness to NO3− (Figure 1f), confirming that NO3− reduction is not required for NO3− to antagonise the effect of Glu on primary root growth. Note that the Columbia accession used here (and in Figures 2 and 3 below) is genetically distinct from the Col-0 accession used in Figure 1(c) and previously (Walch-Liu et al. 2006) (see Experimental procedures) and is much more sensitive to Glu than Col-0.
Nitrate does not inhibit uptake of l-glutamate at the root tip
Rather than interacting with Glu signalling per se, it was possible that NO3− was alleviating the effect of Glu on primary root growth by blocking its uptake into Glu-sensing cells at the root tip. We therefore adapted the segmented agar plate technique to measure the uptake of [15N]Glu specifically at the primary root tip over a 24 h period in the presence and absence of 5 mm KNO3. The data (see Figure S2) show that NO3− had no significant effect on Glu uptake in either C24 or Col-0. However, Glu uptake was inhibited when 2.5 mm Gln was used instead of NO3−, possibly due to competitive inhibition as the two amino acids are reported to share common transporters (Hirner et al., 2006). Since excess Gln does not antagonise the Glu response (Walch-Liu et al., 2006), whereas excess NO3− does, we conclude that Glu uptake is probably not required for Glu sensing and that NO3− may interact with Glu signalling directly rather than indirectly through effects on Glu transport. The finding that the cellular uptake of Glu is unlikely to be required for its effect on root growth is consistent with previous work that led to the suggestion that cells at the root tip are able to sense localised changes in the extracellular Glu concentration (Walch-Liu et al., 2006).
Nitrate also stimulates primary root growth directly
In some segmented agar plate experiments we noted that NO3− appeared to stimulate primary root growth independently of the Glu treatment, something which became more noticeable over longer growth periods (data not shown). Because of the possibility that this phenomenon is mechanistically related to the growth stimulation caused by NO3− antagonism of the Glu effect, we investigated it further.
We first screened a small collection of Arabidopsis accessions to establish the generality of the effect and the extent of natural variation amongst different Arabidopsis genotypes. Seedlings of six different accessions were pre-germinated on NO3−-free medium and transferred to segmented agar plates where NO3− could be applied to different parts of the primary root. Figure 2(a) shows that for five of the six accessions, primary root growth was stimulated when NO3− was present only in the bottom segment, relative to the most directly comparable control treatment where the NO3− was present only in the top segment. Although in Bch-1, Col-0, Ler and RLD1 the growth stimulation was modest (17–22%), the stimulatory effect was also evident when the whole root was exposed to NO3− (Figure 2a), and the effects were observed consistently in replicate experiments. Two accessions differed from the others in their response. Primary root growth in C24 was not significantly affected, while in No-0 it was highly responsive, showing a 90% increase in both treatments where the primary root tip was in direct contact with the NO3−.
For two of the accessions (Col-0 and No-0) we followed the time course of changes in primary root growth rates in the period after transfer to the localised NO3− treatments (Figure 2b,c). In both of the treatments where NO3− was excluded from the bottom segment, the growth rate in No-0 remained more or less constant over the treatment period, whereas in both treatments where the primary root tip was in direct contact with the NO3−, the growth rate accelerated continuously from the first to the sixth day after transfer (Figure 2c). Surprisingly, the difference between No-0 and Col-0 was not in their growth dynamics when NO3− was present at the tip (which were almost identical), but rather in their behaviour when NO3− was excluded from the tip. Figure 2(b) shows that the rate of primary root growth in Col-0 accelerated continuously in both the presence and absence of NO3− (but slightly faster in the presence of NO3− than in its absence). Thus an accelerating pattern of primary root growth is constitutive in Col-0 but is NO3−-inducible in No-0.
The dose–response curve in Figure 2(d) shows that significant stimulation of primary root growth in No-0 occurred with a localised application of just 50 μm NO3−. Even stronger responses were observed with increasing NO3− concentrations, reaching a >2.5-fold stimulation at 5 mm NO3−. Localised treatments using NH4+ as an alternative N source (at 0.2 or 2 mm) had no effect on primary root growth (Figure 2e), indicating that the effect is specific to the NO3− ion.
Role of the NRT1.1 and ANR1 genes in nitrate antagonism of the l-glutamate response
Previous studies have indicated a role for the NRT1.1 and ANR1 genes in a NO3− signalling pathway that modulates the rate of lateral root growth (Remans et al., 2006b; Zhang and Forde, 1998). As both genes are also expressed in the primary root tip (Remans et al., 2006b) we adopted a genetic approach to investigate their possible role in the pathway by which NO3− antagonises the Glu response.
Figure 3(a) shows that an anr1 mutant was as sensitive to NO3− suppression of the Glu effect as the wild type. However, as seen in Figure 3(b), an NRT1.1 deletion mutant (chl1-5) was completely insensitive to the presence of excess NO3−. Primary root growth in chl1-5 was as strongly inhibited by Glu in the presence of excess NO3− as in its absence (Figure 3b). As the deletion in chl1-5 affects two additional genes (Muňos et al., 2004), we examined the phenotype of the 35S-CHL1 line, which is derived from chl1-5 but constitutively expresses NRT1.1 under the control of the cauliflower mosaic virus 35S promoter (Huang et al., 1996). 35S-CHL1 has been shown to be rescued for both the low-affinity and high-affinity components of NO3− transport that are affected in chl1-5 (Huang et al., 1996; Liu et al., 1999). As shown in Figure 3(b), NO3− antagonism of the Glu effect has been fully restored in 35S-CHL1, confirming that the NRT1.1 gene itself is responsible for conferring NO3− sensitivity. Thus NRT1.1, but not ANR1, appears to be involved in the signalling pathway by which NO3− antagonises Glu-induced changes in root architecture. The enhanced sensitivity of the 35S-CHL1 line to NO3− suppression of the Glu effect compared to the Columbia parent, which is evident from Figure 3(b), seems likely to be a dosage effect due to the high level of NRT1.1 expression driven by the 35S promoter.
Another chl1-5-derived line (35S-CHL1T101A) carries a construct that differs from 35S-CHL1 only in encoding a Thr to Ala substitution at residue 101 (Liu and Tsay, 2003). The T101A mutation specifically inactivates the high-affinity component of NRT1.1, by preventing phosphorylation at Thr101 (Liu and Tsay, 2003). Consequently, the 35S-CHL1T101A line is rescued for the low-affinity component but not the high-affinity component of the NO3− transport activity of NRT1.1. Figure 3(b) shows that unlike the 35S-CHL1 line, the 35S-CHL1T101A line is strongly inhibited by Glu even in the presence of 5 mm NO3−. Even in the absence of NO3− we found that the 35S-CHL1T101A line (but not the 35S-CHL1 line) was significantly more sensitive to Glu than chl1-5, an effect that was reproducible at different Glu concentrations and in several independent experiments (Figure S3).
Nitrate regulates primary root growth by two distinct pathways
The generally held view has been that the NO3− supply modifies root architecture predominantly through its effects on lateral root length and numbers, and that primary root growth is largely unresponsive to NO3− (Bloom et al., 2002; Drew and Saker, 1975; Granato and Raper, 1989; Linkohr et al., 2002; Zhang and Forde, 1998). Here we have shown that, in Arabidopsis, NO3− can have a very strong positive effect on primary root growth and that it achieves this in two ways. The first is by antagonising the inhibitory effect of exogenous Glu. This amino acid has previously been shown to be sensed at the primary root tip, resulting in inhibition of meristematic activity and an increase in root branching (Walch-Liu et al., 2006). We found that if an excess of NO3− was present, these effects could be partially or even completely suppressed, depending on the accession and the ratio of NO3− to Glu. The second way in which NO3− stimulates primary root growth is independent of the presence of exogenous Glu. Accessions differed markedly in their sensitivity to this direct effect of NO3−, but primary root growth in the most sensitive line (No-0) was significantly stimulated by an external NO3− concentration of only 50 μm, and a ∼2.5-fold stimulation was achieved by NO3− concentrations above 0.5 mm. The magnitude of these responses is comparable to what was seen when a localised NO3− treatment was applied to lateral roots, although the minimum concentration needed to stimulate lateral root growth was 100 μm (Zhang and Forde, 1998; Zhang et al., 1999).
For both types of primary root response to NO3− we provide evidence that it is the external presence of the NO3− ion itself that is being perceived and that NO3− sensing is taking place at the primary root tip. These conclusions are based on the demonstration that in both cases the primary root tip must be in direct contact with NO3−, that NH4− will not substitute and, in the case of NO3− antagonism of the Glu effect, that a NR-null mutant retains its responsiveness to NO3−. Based on similar evidence, it was previously concluded that stimulation of lateral root growth by NO3− is also a response to the external presence of the NO3− ion, independently of its role as a nutrient (Zhang and Forde, 1998; Zhang et al., 1999).
We considered the possibility that direct stimulation of primary root growth by NO3− might be a manifestation of the same phenomenon as NO3− antagonism of the inhibitory effect of Glu on primary root growth. In this model, primary root growth would be negatively regulated by the endogenous Glu pool, and NO3− would stimulate primary root growth by alleviating this effect. The exceptionally slow growth of primary roots of the Glu-hypersensitive accession, C24 (see Figure 2a), could potentially be explained by a model in which endogenous Glu pools were negatively regulating root growth. Since C24 is also very sensitive to the antagonising effect of NO3−, this model would predict that primary root growth in this accession would also be highly responsive to NO3− in the absence of exogenous Glu. However, C24 was the one accession in which NO3− failed to stimulate primary root growth in the absence of exogenous Glu. Although it is possible that the sensitivity of the root to exogenous Glu is an unreliable indicator of its sensitivity to endogenous Glu pools, we conclude that there is at present no evidence to suggest that the two types of primary root response to NO3− share a common mechanism.
It may be that the signalling pathway by which NO3− stimulates primary root growth directly is the same as the one by which a localised NO3− supply stimulates lateral root growth (Remans et al., 2006b; Zhang and Forde, 1998). However, we were unable to confirm this by demonstrating a role for the NRT1.1 and ANR1 genes because the relevant mutants were in a background (Columbia) whose primary root growth is insufficiently responsive to this NO3− effect.
The evidence presented here indicates that NRT1.1 is a component of at least the signalling pathway that leads to antagonism of the negative effect of Glu on primary root growth. We found that NO3− did not alleviate this effect in the chl1-5 NRT1.1-deficient mutant, but that the NO3− response was fully restored in a chl1-5 line constitutively expressing the NRT1.1 cDNA. However, a major barrier to deciding whether a transporter is directly or indirectly involved in nutrient sensing is eliminating the possibility that it is simply facilitating the uptake of the nutrient for detection by intracellular sensors. Thus although a previous study established that lateral root growth in NRT1.1-deficient mutants was unresponsive to NO3− under conditions where there was no general defect in NO3− uptake (Remans et al., 2006b), a NO3− sensing role for NRT1.1 could not be confirmed because it remained possible that the transporter was necessary for NO3− uptake into specific NO3−-sensing cells in the root tip. The most convincing way to show that a transporter is also a sensor is to be able to uncouple the two functions by mutation (Donaton et al., 2003; Smith et al., 2003). In the case of NRT1.1, we have evidence suggesting that a Thr to Ala substitution at residue 101 may have achieved this effect.
The T101A mutation in NRT1.1 has been shown to inactivate the high-affinity component of this dual-affinity NO3− transporter, while leaving the low-affinity component intact (Liu and Tsay, 2003). We found that in marked contrast to the 35S-CHL1 line, constitutive expression of the mutant form of NRT1.1 in the 35S-CHL1T101A line failed to confer any increase in NO3− sensitivity, even at 5 mm NO3−. To explain this result in terms of the NO3− transport function of NRT1.1 we would have to propose that while constitutive expression of the wild-type (dual-affinity) form of the protein was able to more than fully restore NO3− uptake in the chl1-5 mutant, constitutive expression of the mutant (low-affinity) form made no significant contribution to NO3− uptake in the low-affinity range. This would contradict previous studies showing that the T101A mutant protein expressed in oocytes could catalyse rates of NO3− uptake (from 5 mm NO3−) that were two-thirds those of the wild-type, and experiments showing that the rate of NO3− uptake (again at 5 mm NO3−) by the same 35S-CHL1T101A line as used here was only 35% less than the 35S-CHL1 line (Liu and Tsay, 2003). The presence of alternative N sources such as NH4+ or amino acids at high concentrations (typically 10 mm) can cause feedback inhibition of the high-affinity NO3− transport system, but their effect on uptake in the low-affinity range is much less pronounced (Forde and Clarkson, 1999). l-Glutamate has not been reported to have any effects on NO3− uptake beyond those attributable to its influence on the plant’s N status (Nazoa et al., 2003). Thus there is no reason to expect that the low Glu concentrations used here (0.5–1 mm) would have biased the results, e.g. by differentially inhibiting NO3− uptake in the 35S-CHL1 and 35S-CHL1T101A lines.
The difficulty in explaining the phenotype of the 35S-CHL1T101A line solely in terms of the transport function of NRT1.1 leads us to propose that NRT1.1 has a direct rather than indirect role in NO3− sensing, and that this sensing function is inactivated by the T101A mutation. Since the Thr to Ala substitution blocks phosphorylation at residue 101 (Liu and Tsay, 2003), this would imply that only the phosphorylated protein has NO3−-sensing activity. This would be analogous to the yeast Mep2p NH4+ transceptor, where mutagenesis of a putative phosphorylation site abolished its sensing function while leaving its transport activity intact (Smith et al., 2003).
A regulatory role for NRT1.1 would also help to explain the observation that the 35S-CHL1T101A line was consistently more sensitive to Glu than chl1-5 (or the 35S-CHL1 line), even in the absence of NO3−. This dominant-negative effect of overexpressing the mutant form of the protein suggests that NRT1.1 may be interacting directly with component(s) of the Glu signalling pathway. Thus constitutive expression of NRT1.1 in its unphosphorylated form (but not in its phosphorylatable form) may lead, through protein–protein interactions at the plasma membrane, to conformational changes that make the Glu-sensing system more sensitive to its ligand.
It has been found that phosphorylation/dephosphorylation at Thr101 is nitrogen regulated, being most highly phosphorylated under N-limiting conditions and then rapidly dephosphorylated when N is resupplied (Liu and Tsay, 2003). If, as our results suggest, phosphorylation at Thr101 is required to activate the sensing function of NRT1.1, then reversible phosphorylation/dephosphorylation of the protein would provide a useful mechanism for regulating its NO3− sensitivity according to changes in N availability. Phosphorylation-dependent sensitization/desensitization of receptor proteins is a commonly observed phenomenon in animal systems (e.g. Ribas et al., 2007).
The observation that the NRT2.1 gene remains inducible by NO3− in an NRT1.1-defective mutant (Muňos et al., 2004) indicates that there are NO3− sensors additional to NRT1.1 in the Arabidopsis root. Possible candidates include other NRT1-related proteins with a confirmed NO3− transport function (Tsay et al., 2007), or the NRT2.1 NO3− transporter itself, which is reported to have a sensing or signalling role in the regulation of lateral root initiation (Little et al., 2005; Remans et al., 2006a).
How does nitrate antagonise l-glutamate signalling?
If uptake of Glu by Glu-sensing cells at the root tip were required for perception of the Glu signal, then one straightforward explanation for nitrate’s ability to antagonise the Glu effect would be by blocking Glu uptake. However, we found that excess NO3− had no inhibitory effect on Glu uptake at the root tip. The evidence suggests that exogenous Glu is most likely sensed at the cell surface (Walch-Liu et al., 2006), and although there is a plant family of GLR genes that encode homologues of mammalian ionotropic Glu receptors (Chiu et al., 2002), their involvement in this process has not yet been established. Once the receptor(s) and other components of the Glu signalling pathway at the root tip have been identified, we can begin to address the question of how the NO3− signal, sensed by the NRT1.1 protein, is able so effectively to antagonise the Glu signal. A model for how NO3− and Glu signalling pathways may interact at the primary root tip is presented in Figure 4(a).
As previously discussed (Walch-Liu et al., 2006), Glu is one of the most abundant amino acids in soil and the concentrations that affect root growth in Arabidopsis are in the range likely to be encountered in organic N-rich soil patches. Likewise, the NO3− concentrations that were able to alleviate Glu inhibition (0.5–5 mm) and directly stimulate primary root growth (0.05–5 mm) are commonly found in many soil types (Farley and Fitter, 1999; Glass and Siddiqi, 1995). While NO3− is the major form of N available in fertile aerobic soils, the dissolved organic N pool can be the main N source available in nutrient-poor soils (Christou et al., 2005). Thus the antagonistic effects of NO3− and Glu on primary root growth suggest a mechanism that would allow root architecture to be modulated according to differences in soil fertility, or in response to spatial or temporal variations in the relative abundance of organic and inorganic N. Figure 4(b) illustrates how differences in soil N composition and in a plant’s intrinsic sensitivity to the different N signals could combine to influence the development of the architecture of the root system.
Arabidopsis thaliana L. (Heynh.) accessions were sourced from Lehle Seeds (http://www.arabidopsis.com/), unless stated otherwise. The NR-null mutant (chl3-5/nia1-2), which was a gift from Nigel Crawford (University of California at San Diego, USA), lacks detectable NR activity (Wang et al., 2004). The parental line for nia1-2 was Ler and for chl3-5 it was Columbia (Redei) (Wilkinson and Crawford, 1991). The 35S-CHL1 line (Huang et al., 1996) and the 35S-CHL1T101A line (35S-T101A-14; Liu and Tsay, 2003) were gifts from Yi-Fang Tsay (Academica Sinica, Taiwan). The Columbia accession used as the parental line for the anr1, chl1-5 and NR-null mutants, was from the European Arabidopsis Stock Centre (catalogue no. N60000). This accession is distinct from the Col-0 accession which we have been using routinely and which was originally obtained from Lehle Seeds (Walch-Liu et al., 2006).
The dilute B5 growth medium (dB5) was prepared as described (Walch-Liu et al., 2006). Seeds were surface-sterilised, sown on dB5 medium containing 1% Phytagel® and 0.5 mm Gln as sole N source and stratified in the dark for 2 days at 4°C. Routinely, seedlings were grown on 9-cm Petri dishes, vertically orientated, at 19–22°C with a 16-h/8-h photoperiod at ∼120 μmol m−2 sec−1. For the treatments, a homogeneous subset of 4- or 5-day-old seedlings was transferred to treatment plates. Unless stated otherwise, treatment plates contained 0.5% sucrose and there was no N source other than those used for the treatments. Where K glutamate or KNO3 treatments were applied, control treatments contained appropriate concentrations of K2SO4 or KCl, respectively.
Segmented agar plate experiments
When N treatments were to be applied independently to different parts of the primary root, dB5 plates (10 × 10 cm) were prepared in which the top and bottom halves of the agar were separated by an air gap (Walch-Liu et al., 2006; Zhang and Forde, 1998). The appropriate N source (as KNO3, K glutamate or NH4SO4), was applied as a concentrated solution and allowed to soak in overnight. To initiate the treatment, 5-day-old seedlings were transferred to the segmented plates and positioned so that only the tip of the primary root was in contact with the bottom segment (see Figure S1).
Imaging and root measurements
After transfer to treatment plates, the position of the primary root tip was marked on the base of the plate. Roots were imaged using a flat-bed digital scanner and root growth analysed using Optimas Image Analysis software (Version 6.1, Media Cybernetics Inc., http://www.mediacy.com/). All experiments were performed at least twice with similar results.
We are very grateful to Yi-Fang Tsay (Academica Sinica, Taiwan) and Nigel Crawford (University of California, San Diego) for generous gifts of seed and to Darren Sleep (Centre for Ecology and Hydrology, Lancaster) for 15N analysis. This work was supported in part by grants from the UK Biotechnology and Biological Sciences Research Council no. BB/C005120/1 and by European Commission Research Training Network grant no. HPRN-CT-2002-00247.