Genome-wide analysis in response to nitrogen and carbon identifies regulators for root AtNRT2 transporters

A systems biology approach reveals three transcription factors involved in the regulation of NRT2 nitrate transporters in response to combinations of nitrogen and carbon treatments.

Abstract

In Arabidopsis (Arabidopsis thaliana), the High-Affinity Transport System (HATS) for root nitrate (${\text{NO}}_3^{-}$) uptake depends mainly on four NRT2 ${\text{NO}}_3^{-}$ transporters, namely NRT2.1, NRT2.2, NRT2.4, and NRT2.5. The HATS is the target of many regulations to coordinate nitrogen (N) acquisition with the N status of the plant and with carbon (C) assimilation through photosynthesis. At the molecular level, C and N signaling pathways control gene expression of the NRT2 transporters. Although several regulators of these transporters have been identified in response to either N or C signals, the response of NRT2 gene expression to the interaction of these signals has never been specifically investigated, and the underlying molecular mechanisms remain largely unknown. To address this question we used an original systems biology approach to model a regulatory gene network targeting NRT2.1, NRT2.2, NRT2.4, and NRT2.5 in response to N/C signals. Our systems analysis of the data identified three transcription factors, TGA3, MYC1, and bHLH093. Functional analysis of mutants combined with yeast one-hybrid experiments confirmed that all three transcription factors are regulators of NRT2.4 or NRT2.5 in response to N or C signals. These results reveal a role for TGA3, MYC1, and bHLH093 in controlling the expression of root NRT2 transporter genes.

Introduction

Plants must integrate internal and external signals to adapt to fluctuating environmental conditions. This is particularly the case concerning mineral nutrition, because most nutrients display dramatic changes in external availability, whereas their internal concentrations must be kept within a limited range to be compatible with physiological processes. Accordingly, root nutrient uptake systems are finely tuned by regulatory mechanisms activated by local signaling of external nutrient availability and systemic signaling of the nutrient status of the whole plant (Schachtman and Shin, 2007). Furthermore, acquisition of the various nutrients has to be coordinated to remain consistent with the global chemical composition of plant tissues and with the fact that most nutrients contribute to the synthesis of biomolecules with a relatively strict elemental stoichiometry (e.g. carbon [C], nitrogen [N] and S for amino acids). Therefore, the signaling pathways that are specific for the different nutrients must interact to ensure this coordination. Although coordinated regulation of uptake systems for different nutrients have been clearly demonstrated at the physiological level, the underlying molecular mechanisms remain largely obscure (Schachtman and Shin, 2007). The cross-talks between N and C signaling mechanisms are certainly those that have been most often investigated (Coruzzi and Zhou, 2001; Nunes-Nesi et al., 2010; Ruffel et al., 2014), first because N and C are the two mineral nutrients plants require in largest quantities, and also because they connect two key functions of plants as autotrophic organisms, that is, photosynthesis and assimilation of inorganic N. Moreover, the importance of N/C signaling interaction is dramatically illustrated by the fact that most N-responsive genes in Arabidopsis (Arabidopsis thaliana), are actually regulated by C/N interaction (Gutierrez et al., 2007).

The N nutrition of most herbaceous plants relies on the uptake of nitrate (${\text{NO}}_3^{-}$), which is ensured in root cells by two classes of transport systems. The High-Affinity Transport System (HATS) is predominant in the low range of ${\text{NO}}_3^{-}$ concentrations (up to ∼ca 1 mM), whereas the Low-Affinity Transport System (LATS) makes an increasing contribution to total ${\text{NO}}_3^{-}$ uptake with increasing external ${\text{NO}}_3^{-}$ concentration (Crawford and Glass, 1998). In all species investigated to date, genes encoding the various transporter proteins involved in either HATS or LATS have mostly been identified in the NRT2 and NPF (formerly NRT1/PTR) families, respectively (Nacry et al., 2013; O’Brien et al., 2016). The respective roles of HATS and LATS in the total ${\text{NO}}_3^{-}$ acquisition by the plant are still a matter of debate. However, field studies suggest that even in agricultural conditions, the HATS has a major contribution over the whole developmental cycle (Malagoli et al., 2004; Garnett et al., 2013). Both the structure and regulation of the HATS have been extensively studied in Arabidopsis. In this species, almost all the HATS activity depends on four NRT2 transporters, namely NRT2.1, NRT2.2, NRT2.4, and NRT2.5 (Filleur et al., 2001; Kiba et al., 2012; Lezhneva et al., 2014), which all require an interaction with the NAR2.1 protein to be active in ${\text{NO}}_3^{-}$ transport (Kotur et al., 2012). Under most conditions, NRT2.1 is the main contributor to the HATS (Cerezo et al., 2001; Filleur et al., 2001). However, NRT2.4 and NRT2.5 display a very high-affinity for ${\text{NO}}_3^{-}$ and are important for taking up this nutrient when present at very low concentration (<50 µM) in the soil solution (Kiba et al., 2012; Lezhneva et al., 2014). Furthermore, unlike NRT2.1 and NRT2.4, NRT2.5 does not require the presence of ${\text{NO}}_3^{-}$ to be expressed, and is therefore considered crucial for ensuring the initial uptake of ${\text{NO}}_3^{-}$ as soon as it appears in the external medium (Kotur and Glass, 2015).

Most interestingly, the HATS is the target of almost all regulations governing root ${\text{NO}}_3^{-}$ acquisition in Arabidopsis (Nacry et al., 2013), and this is associated with control of NRT2.1, NRT2.2, NRT2.4, and NRT2.5 expression at the mRNA level. In particular, previous reports have shown that NRT2.1 is induced both by N starvation (Lejay et al., 1999; Cerezo et al., 2001; Gansel et al., 2001), and by light and sugars, indicating coordination with photosynthesis (Lejay et al., 1999; Lejay et al., 2003). This makes NRT2.1 a very relevant model gene for investigating the interaction between N and C signaling networks in roots. This also holds true for NRT2.4 (Lejay et al., 2008; Kiba et al., 2012), but not for NRT2.5, which until now has only been reported to be up-regulated by N starvation (Lezhneva et al., 2014). For these reasons, and also due to its high functional importance as the main component of the HATS, NRT2.1 has been extensively investigated to unravel its regulatory mechanisms. Accordingly, a quite significant number of genes were found to encode regulators of NRT2.1 expression, such as LBD37-39 (Rubin et al., 2009), TGA1 and TGA4 (Alvarez et al., 2014), NLP6 and NLP7 (Marchive et al., 2013; Guan et al., 2017), NRG2 (Xu et al., 2016), BT1-2 (Araus et al., 2016), NRT1.1 (Munos et al., 2004), CIPK8 (Hu et al., 2009), HNI9/IWS1 (Widiez et al., 2011), and HY5 (Chen et al., 2016). Most of these genes contribute to the regulation of NRT2.1 expression in response to changes in N provision. The only exception is HY5, which encodes a transcription factor reported to ensure long-distance signaling of the stimulation of NRT2.1 expression in roots by illumination of the shoot. Strikingly, none of the above regulators seem to be involved in the cross-talk between N and C signaling pathways. Even more surprising, the response of NRT2.1 expression itself (as well as those of the other NRT2s) to the interaction of N and C signals was not specifically investigated. As a consequence, the molecular mechanisms responsible for the coordinated regulation of the ${\text{NO}}_3^{-}$ HATS by N and C status of the plant are unknown.

Our study aimed at filling this gap. Therefore, using NRT2.1 as a marker gene to identify relevant combinations of N/C treatments, we developed a systems biology approach based on genome-wide transcriptome analysis in roots of Arabidopsis plants to model a regulatory gene network targeting NRT2.1, NRT2.2, NRT2.4, and NRT2.5 in response to N/C signals. This highlighted the potential role of three putative transcription factors, TGA3, MYC1, and bHLH093 in controlling the expression of these transporter genes. Functional analysis of loss-of-function mutants confirmed that all three transcription factors are regulators of NRT2.4 or NRT2.5 in response to N or C signals. Furthermore, yeast one-hybrid (Y1H) experiments confirmed that at least TGA3 and MYC1 are able to bind NRT2.4 and NRT2.5 promoters.

Results

Regulation of root ${\text{NO}}_3^{-}$ transporters by interaction between N and light provision

We wished to determine whether induction of NRT2.1 by N starvation is dependent on light, and conversely, if NRT2.1 induction by light is dependent on the availability of ${\text{NO}}_3^{-}$ (Figure 1, A and B). In order to reveal possible interactions between C and N signaling pathways for the regulation of NRT2.1, we performed two different sets of experiments. In the first set of experiments, plants were starved for N for up to 72 h either in the dark or at three different light intensities, 50 μmol m⁻²s⁻¹ (low light [LL]), 250 μmol m⁻²s⁻¹ (intermediate light [IL]), and 800 μmol m⁻²s⁻¹ (high light [HL]; Figure 1A). In the second set of experiments, plants were treated with 10 mM ${\text{NO}}_3^{-}$, 1 mM ${\text{NO}}_3^{-}$ or no N and transferred from the dark to HL conditions for 8 h (Figure 1B).

Figure 1.

Interaction between N and Light/C provision modulates NRT2.1 mRNA accumulation in roots. (A) Different light regimes modulate NRT2.1 regulation in roots of plants experiencing high ${\text{NO}}_3^{-}$ provision (10 mM) to N deprivation (−N). The light regimes encompass dark, LL intensity (50 μmol m⁻² s⁻¹), IL intensity (250 μmol m⁻² s⁻¹), and HL intensity (800 μmol m⁻² s⁻¹). Plants were supplied with ${\text{NO}}_3^{-}$ 10 mM one week before the experiment and acclimated for 24 h in the different light regimes before applying the N deprivation for 24, 48, or 72 h. (B) Different N provisions modulate NRT2.1 regulation in roots of plants experiencing a dark to light transition. The N provisions encompass 10 mM ${\text{NO}}_3^{-}$, 1 mM ${\text{NO}}_3^{-}$ (for 72 h) and N deprivation for 48 h (−N). Plants were kept in the dark (i.e. 40 h) before transition to HL intensity (800 μmol m⁻² s⁻¹; HL) and roots were collected at time 0 (Dark) and 1, 2, 4, and 8 h after the light transition. (C) Regulation of NRT2.1 by photosynthesis activity. Plants were grown in regular ${\text{NO}}_3^{-}$ regime (1 mM) and IL intensity until they were transferred for 4 h in a CO₂-deprived atmosphere (0 ppm) or in high CO₂-supplied atmosphere (600 ppm), either in the dark or in the light. Means ± sd (n = 4) with different letters are significantly different within each block of data determined by one-way ANOVA followed by a post hoc Tukey test (P < 0.05). In these three experimental conditions, roots were collected to assess NRT2.1 mRNA accumulation by RT-qPCR (relative accumulation to Clathrin housekeeping gene). Expression pattern of NRT2.1 across the 35 conditions tested (16 in A, 15 in B, and 4 in C) drove the choice of 18 conditions to investigate gene reprogramming associated to the regulation of ${\text{NO}}_3^{-}$ transport. These 18 conditions are indicated with arrows and numbers on the x-axis of the 3 NRT2.1 bar graphs (Each arrow corresponds to one condition with two independent biological repeats constituted of a pool of ∼10 plants each).

In LL and IL conditions, NRT2.1 expression was, as expected, induced when plants were starved for N even if both the kinetic and the level of induction were different depending on light intensity (Figure 1A). When plants were kept in the dark, NRT2.1 expression was not induced by N starvation but it remained very low both on 10 mM ${\text{NO}}_3^{-}$ and on N-free solution. More surprisingly, the induction of NRT2.1 expression by N starvation was also almost completely abolished when plants were treated in HL conditions. However, under HL NRT2.1 mRNA levels were always high, even under repressive conditions such as 10 mM ${\text{NO}}_3^{-}$. This unexpected result is specific of NRT2.1 since NRT2.2, NRT2.4, and NRT2.5, known to be also induced by N starvation in roots (Li et al., 2007; Kiba et al., 2012; Lezhneva et al., 2014), are still regulated by N starvation in HL (Supplemental Figure S1A). However, just like NRT2.1, NRT2.2, NRT2.4, and NRT2.5 were not regulated by N starvation in the absence of light. These data confirm the need of light for the regulation by N starvation of root ${\text{NO}}_3^{-}$ transporters. It also suggests that the mechanisms involved in NRT2.1 regulation by N starvation are somewhat different from the mechanisms involved in the regulation of NRT2.2, NRT2.4, and NRT2.5.

The second set of experiments confirmed the strong interaction between C/N signals as it revealed that the level of N nutrition affects the regulation of NRT2.1 expression by light (Figure 1B). Indeed, when plants were N starved for 48 h, NRT2.1 expression was much less induced by light as compared to plants grown on 10 or 1 mM ${\text{NO}}_3^{-}$ (Figure 1B). Among other root ${\text{NO}}_3^{-}$ transporters, only NRT2.2 and NRT2.4 were induced by light and their level of induction seemed to be also dependent on N nutrition (Supplemental Figure S1B). However, in contrast to NRT2.1, the level of expression of both NRT2.2 and NRT2.4 was high when plants were starved for N and low when plants were grown on 1 or 10 mM ${\text{NO}}_3^{-}$ (Supplemental Figure S1B). For NRT2.4, these results confirm that this transporter is more sensitive to high N repression than NRT2.1 (Kiba et al., 2012). The same result was obtained for NRT2.5, whose expression was barely detectable on either 10 mM or 1 mM ${\text{NO}}_3^{-}$ (Supplemental Figure S1B). However, concerning regulation by light, even when NRT2.5 expression was high in N starved plants, light did not induce but rather seemed to repress NRT2.5 mRNA accumulation after 8 h in the light (Supplemental Figure S1B).

In a previous study, we showed that expression of NRT2.1 and NRT2.4 is induced by light only in the presence of CO₂ in the atmosphere, suggesting that light regulation of these genes corresponds to a control exerted by photosynthesis (Lejay et al., 2008). As in the rest of our study we used micro-array experiments to look for genes involved in the regulation of root ${\text{NO}}_3^{-}$ transporters by photosynthesis, it was important for us to be able to discriminate between genes regulated by light itself or by photosynthesis. To do so, we performed a third set of experiments where plants were transferred from dark to light for 4 h in an atmosphere containing 0 or 600 ppm CO₂. The results confirmed (1) that both NRT2.1 and NRT2.4 are only induced by light in the presence of CO₂ and (2) that NRT2.5 is not induced by light or photosynthesis as suggested by our previous experiment (Figure 1C; Supplemental Figure S1C).

Gene network for the regulation of root ${\text{NO}}_3^{-}$ transporters by light and N starvation

The experiments performed above allowed us to reveal interesting interactions between C and N regulation of root NRT2 ${\text{NO}}_3^{-}$ transporters. We took advantage of this experimental design to develop a systems biology approach aiming at inferring a gene regulatory network underlying the interactions between N and C signals in the regulation of root high-affinity ${\text{NO}}_3^{-}$ transporters. Due to the central position of NRT2.1 as regulatory target affecting N acquisition and the high and complex regulation of its level of expression in response to N and C, we used it as a focus gene around which to find associated gene networks.

We performed Affymetrix microarrays on selected combinations of light and N treatments, which were found discriminant for regulation of NRT2.1. Altogether, we chose 18 treatments labeled with numbered arrows in Figure 1. These 18 treatments correspond to 3 sets of conditions representative of (1) the light-dependent induction of NRT2.1 expression in response to N starvation, (2) the light induction of NRT2.1 on 10 mM ${\text{NO}}_3^{-}$ and (3) the specific regulation of NRT2.1 by photosynthesis and not by light itself. For each treatment, two independent biological replicates were generated and used for Affymetrix ATH1 microarray hybridization.

Regulation of NRT2.1, NRT2.2, NRT2.4, and NRT2.5 gene expression in response to N starvation and light/photosynthesis was similar on microarrays as compared to the results obtained by reverse transcription-quantitative PCR (RT-qPCR) in Figure 1 and Supplemental Figure S1 (Supplemental Figure S2). These results also confirmed that these four ${\text{NO}}_3^{-}$ transporters are the main NRT2s expressed in roots. NRT2.3, NRT2.6, and NRT2.7 showed very low expression levels on the microarrays under our experimental conditions. It is also noteworthy that NRT2.1 was the most highly expressed member of the family among the 7 NRT2s (5–50-fold higher expression as compared to NRT2.2, NRT2.4, and NRT2.5; Supplemental Figure S2).

To find gene regulatory networks that could integrate N and C signaling and thus control NRT2.1 expression, we defined five different subsets of conditions addressing the regulation by N on one side and by C on the other side, as described in Figure 2A. Genes defined as regulated by N-deprivation like NRT2.1 are differentially regulated by N provision in Conditions 1–4 in Experiment 1, where NRT2.4 is also found regulated and in Conditions 7–14 in Experiment 2, where NRT2.2 and NRT2.5 were also found regulated. To select the most robust genes regulated by N provision only the intersection between the two groups was isolated. In addition to NRT2.1, the intersection defines a set of 33 genes including the 2 transcription factors TGA3 (At1g22070) and MYC1 (At4g00480). On another hand, genes considered as regulated by C provision like NRT2.1 are differentially regulated by light intensity in Conditions 1, 3, 5, and 6 in Experiment 1, by light time exposure in Conditions 9, 11, and 13 in Experiment 2 and by photosynthesis in Conditions 15–18 in Experiment 3. Similarly, to narrow down the specificity of gene regulation by C factor, only common genes to at least two experiments were isolated. This core set corresponds to 142 genes including NRT2.1 but also two others transcription factors bHLH093 (At5g65640) and ARR14 (At2g01760; Figure 2A).

Figure 2.

Gene expression multi-analysis driven by NRT2.1 expression pattern combined to an integrative analysis identified a candidate gene regulatory network connected to the ${\text{NO}}_3^{-}$ transport system. (A) Venn diagrams identifying common genes regulated by N provision in L condition and dark to light transition (34 genes) or regulated by light/C (142 genes). The union of these gene lists defines a population of 174 genes, including 4 transcription factors. (B) The core set of 174 genes differentially expressed structured into a Gene Regulatory Network using the Gene Networks analysis tool in VirtualPlant software (http://virtualplant.bio.nyu.edu/cgi-bin/vpweb/; Katari et al., 2010). The network includes 124 nodes (genes) and 260 edges connecting genes. The nodes have been organized according to their connection to the three transcription factors MYC1, TGA3, and bHLH093 and are detailed in the Network Legend. ARR14 was excluded from the network due to its lack of connectivity to other nodes according to the edges selected to generate the network.

Next, we only focused on the 174 genes that showed a response to N starvation (34 genes) and/or C provision (142 genes); NRT2.1 being the common gene between the two responsive gene lists together with a Kinesin3 gene (At5g54670-ATK3) coding for a microtubule motor protein. The possible connection of the four transcription factors with NRT2.1 and the other genes was determined by a Gene Networks analysis performed on the VirtualPlant platform (Katari et al., 2010). The generated network contains 124 gene nodes. These genes are connected to each other by 260 edges, representing regulatory relationships such as predicted transcription factor-target gene interactions (Figure 2B). Regulatory interactions were proposed based on detection of at least one predicted binding site for a given transcription factor within the promoter region of the target gene as done previously (Gutierrez et al., 2008). According to the parameters used, 50 genes out of the 174 are not connected to any other genes in the network (See Material and Methods for details about the parameters). Among these 50 genes, the transcription factor ARR14 was excluded due, for instance, to a low level of correlation between this gene and NRT2.1 expression pattern. However, TGA3, MYC1, and bHLH093 have predicted regulatory interactions with NRT2.1 plus 40 other genes of the network (indicated in blue in Figure 2B). The network predicts also that only one or only two of these transcription factors putatively regulate the 79 remaining genes (Supplemental Table S6; one gene being connected to the network by predicted protein–protein interaction with two TGA3-targets). Nevertheless, almost all sub-networks are interconnected through protein–protein interaction prediction, suggesting possible coordination within the network at a whole.

Regulation of MYC1, TGA3, and bHLH093 in response to C and N

The gene regulatory network we obtained revealed three main transcription factors: MYC1 and TGA3 which were co-regulated with NRT2.1 in response to N starvation and bHLH093 which was co-regulated with NRT2.1 in response to light/photosynthesis. In order to validate their regulation, we measured gene expression by RT-qPCR across all the conditions performed in Experiments 1 and 2 (Figure 3A). The results confirmed that expression of TGA3 and MYC1 genes is induced 2- to 3-fold after transferring the plants to a N-free solution, especially under LL or HL conditions. Furthermore, similar to NRT2.1, the regulation of TGA3 and MYC1 expression by N requires the presence of light (Figure 3A; Supplemental Figure S3). The results also confirmed that bHLH093 gene expression is only induced by light (between 3- and 4-fold after 8 h of HL), independent of N nutrition. This is supported by the fact that bHLH093 is not regulated by N starvation (Figure 3A; Supplemental Figure S3). On the contrary, MYC1 and TGA3 genes are not only regulated by N starvation, but their expression is also induced by light, especially in plants starved for N (Figure 3A; Supplemental Figure S3). Like for NRT2.1, putative cis-binding elements for TGA3, MYC1, and bHLH093 were also found in the promoters of NRT2.2, NRT2.4, and NRT2.5 (Figure 3B). Furthermore, DNA affinity purification sequencing experiments performed by O’Malley et al. (2016) showed that TGA3 binds in silico to the promoter of NRT2.1, NRT2.2, and NRT2.4 (Figure 3C). Unfortunately, no data are available for MYC1 and bHLH093 in this work. Altogether, these results support the hypothesis that the transcription factors we identified are involved in the regulation of several root NRT2s.

Figure 3.

TGA3, MYC1, and bHLH093 are candidate transcription factors for the control of the expression of NRT2 gene family. (A) Gene expression analysis of the 3 candidate transcription factors in the extended set of N/C combinations; Expression patterns have been determined by RT-qPCR (relative accumulation to Clathrin housekeeping gene). Means ± sd (n = 4) with different letters are significantly different within each block of data determined by one-way ANOVA followed by a post hoc Tukey test (P < 0.05). (B) NRT2.2, NRT2.4, and NRT2.5 as well as NRT2.1 display putative cis-binding elements for the three transcription factors in their promoter region. The gene network has been done using the Gene Networks analysis tool in VirtualPlant software (http://virtualplant.bio.nyu.edu/cgi-bin/vpweb/; Katari et al., 2010); only Regulated Edges box and One Binding Site option has been selected in this case. (C) TGA3 binds in silico with the promoter of NRT2.1, NRT2.2, and NRT2.4. The analysis was conducted using the Plant Cistrome Database (http://neomorph.salk.edu/PlantCistromeDB; O'Malley et al., 2016).

To our knowledge, the transcription factors TGA3, MYC1, and bHLH093 have not been isolated in previous transcriptomic approach as candidates for regulation of root ${\text{NO}}_3^{-}$ transporters. In order to understand why they have not been found before we looked at the expression pattern of the known regulatory elements for NRT2.1 in our experimental set up. The known regulators for NRT2.1 were not co-regulated with NRT2.1 expression in our conditions (Figure 4). This was also the case for HY5, a transcription factor recently identified as involved in the regulation of NRT2.1 by light/photosynthesis (Chen et al., 2016). In our hands, this transcription factor was only induced by light independently of the presence of CO₂ and therefore not by photosynthesis like NRT2.1 (Figure 4). As most of the previous transcriptomic experiments were performed to study the signaling pathways involved in short-term induction by ${\text{NO}}_3^{-}$, we also looked at the regulation of TGA3, MYC1, and bHLH093 in those conditions (Supplemental Figure S4). We chose the transcriptomic experiments performed by Wang et al. (2004). In this study wild-type (WT) plants and the null mutant for ${\text{NO}}_3^{-}$ reductase (NR) were treated with 5 mM KNO₃ for 2 h and compared to control plants treated with 5 mM KCl for 2 h. The data sets allowed the authors to determine the genes that respond specifically to ${\text{NO}}_3^{-}$ in both WT and NR-null plants. The results show that, as expected, NRT2.1, NRT2.2, and NRT2.4 are induced by ${\text{NO}}_3^{-}$ while NRT2.5 seems to be repressed (Supplemental Figure S4A). In the same time, most of the known regulators for NRT2.1 are also induced by ${\text{NO}}_3^{-}$ except NLP7 and TCP20, two transcription factors which have not been isolated using transcriptomic approaches (Supplemental Figure S4B). On the contrary, in the same conditions, our three transcription factors, TGA3, MYC1, and bHLH093 were not regulated by ${\text{NO}}_3^{-}$ supply neither in WT nor NR-null plants. All these results reinforced the originality of our experimental set up and explain why we found new candidates that have never been isolated in previous transcriptomic experiments.

Figure 4.

Most of the genes previously determined as NRT2s regulators do not display expression patterns similar to the patterns of the three candidate transcription factors in the set of N and Light/C combinations. Graphs display the expression pattern of the 20 genes extracted from the whole transcriptomic dataset. Data ± sd (n = 2) are organized according to the multi-analysis (i.e. Supplemental Figure S1–S5; Figure 2). LBD37, LBD38, and LBD39 repress the expression of genes involved in ${\text{NO}}_3^{-}$ uptake (NRT2.1 and NRT2.5; Rubin et al., 2009). TGA1, TGA4, NLP6, NLP7, NRG2, NRT1.1, CIPK8, and CIPK23 are required for the ${\text{NO}}_3^{-}$-dependent induction of NRT2.1 (Munos et al., 2004; Castaings et al., 2009; Ho et al., 2009; Hu et al., 2009; Konishi and Yanagisawa, 2013; Marchive et al., 2013; Alvarez et al., 2014; Xu et al., 2016). TCP20 and HNI9/IWS1 are involved into NRT2.1 regulation controlled by systemic signaling (Widiez et al., 2011; Guan et al., 2014). BT2 represses expression of NRT2.1 and NRT2.4 genes under low ${\text{NO}}_3^{-}$ conditions (Araus et al., 2016). CBL7 regulates NRT2.4 and NRT2.5 expression under N-starvation conditions (Ma et al., 2015). HY5 has been recently identified as a regulator of NRT2.1 by mediating light promotion of ${\text{NO}}_3^{-}$ uptake (Chen et al., 2016). HRS1, HHO1, HHO2, and HHO3 are repressors of NRT2.4 and NRT2.5 expression under high N conditions (Kiba et al., 2018; Safi et al., 2018)

Role of MYC1, TGA3, and bHLH093 in the regulation of NRT2 root ${\text{NO}}_3^{-}$ transporters

To determine if MYC1, TGA3, and bHLH093 are involved in regulation of NRT2 root ${\text{NO}}_3^{-}$ transporters we used two independent insertion mutants for each of the transcription factors: tga3.2, tga3.3 for TGA3, myc1.2, myc1.3 for MYC1, and bHLH093.1, bHLH093.5 for bHLH093. As both TGA3 and MYC1 were regulated by N starvation, we also produced a double mutant, tga3.2/myc1.2, to test a potential additive effect of those transcription factors on the regulation of NRT2s. In addition, to reinforce our conclusions concerning the role of bHLH093, we also produced an overexpressing line by transforming the bhlh093.1 mutant with a 35S::bHLH093 construct. The measurement of MYC1, TGA3, and bHLH093 expression level confirmed an almost complete absence of their transcripts in their respective mutants and a strong overexpression of bHLH093 in the overexpressing line (Supplemental Figure S5, A and B).

As expected for a role of TGA3 and MYC1 in the regulation of NRT2s by N starvation, the induction of both NRT2.4 and NRT2.5 is overall reduced in tga3 and myc1 mutants compared to WT plants, especially after 72 h of N starvation (Figure 5A). This lower induction in response to N starvation is stronger in the double mutant tga3.2/myc1.2 and is observed in that case consistently after 24, 48, and 72 h of N starvation for NRT2.4 and NRT2.5 and after 48 h for NRT2.2. This result suggests that TGA3 and MYC1 are not redundant and that both factors may function as transcriptional activators under low N conditions. This result is supported by the fact that neither the level of expression nor the regulation of MYC1 in tga3 mutants and of TGA3 in myc1 mutants is affected compared to WT plants (Supplemental Figure S5A). Interestingly, the role of MYC1 and TGA3 seems to depend on the amount of light since the effect on NRT2.4 and NRT2.5 induction by N starvation is preferentially observed in HL and not in LL conditions (Figure 5A; Supplemental Figure S6). However, these two transcription factors are specific to N regulation since none of them has a negative impact on the induction of NRT2s expression by light in the double mutant tga3.2/myc1.2 (Supplemental Figure S7). Surprisingly, MYC1 and TGA3 never affect the regulation of NRT2.1 by N starvation (Figure 5A; Supplemental Figure S6). In agreement with a role of MYC1 and TGA3 in the regulation of NRT2.4 and NRT2.5, Y1H experiments show that both transcription factors are able to bind to the promoter of these two transporters (Figure 5B). Finally, the lower induction of NRT2.4 and NRT2.5 in response to N starvation in the double mutant tga3.2/myc1.2 has functional consequences. Indeed, the level of root ${\text{NO}}_3^{-}$ influx decreased in the double mutant tga3.2/myc1.2 compared to WT plants (Figure 5C).

Figure 5.

TGA3 and MYC1 are required for NRT2.4 and NRT2.5 full induction during N-deprivation. (A) Characterization of the knock-out mutants for TGA3 (tga3.2 and tga3.3), MYC1 (myc1.2 and myc1.3), and the TGA3/MYC1 double mutants (tga3.2 myc1.2). The plants were supplied with 10 mM ${\text{NO}}_3^{-}$ one week before the experiment and acclimated for 24 h in HL conditions (800 μmol m⁻² s⁻¹) before applying the N deprivation for 24, 48, or 72 h. Roots were collected to assess NRT2.1, NRT2.2, NRT2.4, and NRT2.5 mRNA accumulation by RT-qPCR (relative accumulation to Clathrin housekeeping gene). Values are means of three biological replicates ± sd. (B) Characterization of TGA3 and MYC1 interaction with NRT2.4 and NRT2.5 promoters in a Y1H assay. Yeast cells were grown on SD-H-U-T minimal media without histidine (H), uracil (U), tryptophan (T), and containing 3-amino-1,2,4-triazole at 0, 15, 30, and 50 mM. Interaction between the transcription factors and the promoters results in HIS3 reporter activation in contrast to the empty vector that does not interact. (C) Root ${\text{NO}}_3^{-}$ influx measured at the external concentration of 5 µM ¹⁵${\text{NO}}_3^{-}$. Plants were treated in the same conditions as for NRT2s mRNA level measurements. Values are means of 12 replicates ± sd. Differences between WT (Col-0) and the KO mutants are significant at *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

Out of the three NRT2s, which are clearly induced by light, NRT2.4 and to a lower extent NRT2.1, have a significantly lower induction after 4 and 8 h of light in the bHLH093 mutants as compared to WT plants (Figure 6, A and B). Conversely, the induction by light of both NRT2.1 and NRT2.4 is higher in the 35S::bHLH093 plants (Figure 6B). Interestingly, this phenotype seems to depend on the amount of ${\text{NO}}_3^{-}$ in the nutritive solution since the effect of bHLH093 is preferentially seen when plants are starved for N or on 1 mM ${\text{NO}}_3^{-}$ and is almost absent when plants are grown on 10 mM ${\text{NO}}_3^{-}$ (Figure 6, A and B). For NRT2.5, despite the fact that its induction by light seems to be only transient, bHLH093 is also involved in its regulation, at least on 1 mM ${\text{NO}}_3^{-}$ (Figure 6, A and B). However, contrary to TGA3 and MYC1, the effect of bHLH093 mutation on NRT2s induction by light has no real impact on the level of root ${\text{NO}}_3^{-}$ influx compared to WT plants, no matter how much ${\text{NO}}_3^{-}$ is in the nutritive solution (Figure 6C).

Figure 6.

bHLH093 is required for NRT2.4 full induction by light. (A) Characterization of the knock-out mutants for bHLH093 (bHLH093-1 and bHLH093-5) on 0 N or 10 mM ${\text{NO}}_3^{-}$. The plants were either N starved for 48 h (light gray bars) or supplied with 10 mM ${\text{NO}}_3^{-}$ one week ahead of the experiment (black bars) and were kept in the dark 40 h before transition to HL intensity (800 μmol m⁻² s⁻¹) for 1, 2, 4, and 8 h. Roots were collected to assess NRT2.1, NRT2.2, and NRT2.4 mRNA accumulation by RT-qPCR (relative accumulation to Clathrin housekeeping gene). Values are means of three biological replicates ± sd. (B) Characterization of the knock-out (bHLH093-1 and bHLH093-5) and the over-expressor (35S::bHLH093) mutants for bHLH093 on 1mM ${\text{NO}}_3^{-}$. The plants were grown on 1 mM ${\text{NO}}_3^{-}$ and were kept in the dark 40 h before transition to IL intensity (250 μmol m⁻² s⁻¹) for 1, 2, 4, and 8 h. Roots were collected to assess NRT2.1, NRT2.2, and NRT2.4 mRNA accumulation by RT-qPCR (relative accumulation to Clathrin housekeeping gene). Values are means of three biological replicates ± sd. (C) Root ${\text{NO}}_3^{-}$ influx measured at the external concentration of 5 µM ¹⁵${\text{NO}}_3^{-}$. Plants were treated in the same conditions as for NRT2s mRNA level measurements and were transferred to the light for 2, 4, and 8 h. Values are means of 12 replicates ± sd. Differences between WT (Col-0) and the mutants are significant at *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

Discussion

Interaction between N and light provision affect regulation of NRT2.1 expression

As part of its central physiological role, the root ${\text{NO}}_3^{-}$ HATS is a main target of the C/N regulatory networks ensuring the necessary integration of both, N acquisition by roots and C acquisition by shoots. The HATS regulation by N starvation has been well characterized in previous studies, especially through the study of NRT2.1 expression. The classical response of NRT2.1 expression to N starvation corresponds to an initial increase due to the relief of the repression exerted by high N status followed by a decrease due to the loss of the induction by ${\text{NO}}_3^{-}$ (Lejay et al., 1999). Split-root experiments have demonstrated that the regulation by the N status relies on systemic signaling pathways (Gansel et al., 2001), and underlying molecular mechanisms have recently been unraveled (Ohkubo et al., 2017). On the other hand, NRT2.1 expression is also dramatically induced by light and sugars through an Oxidative Pentose Phosphate Pathway (OPPP)-dependent signaling mechanism (Lejay et al., 1999; Lejay et al., 2003; Lejay et al., 2008; de Jong et al., 2014). Over the past decade, the importance of signal interaction for the regulation of gene expression has become more and more obvious and especially for C/N regulation (Palenchar et al., 2004; Gutierrez et al., 2007; Krouk et al., 2009). However, the details of how this interaction affects regulation of NRT2.1 expression in response to combined N/C treatments were unknown. Our results clearly show that the interplay of N and C signaling mechanisms has a major role as light conditions can totally suppress the regulation of NRT2.1 expression by the N status, and vice versa (Figure 1, A and B). Similar to the case for inorganic N assimilation, it seems that low sugars inhibit NRT2.1 expression, overriding signals from N metabolism (Stitt et al., 2002; Nunes-Nesi et al., 2010). Surprisingly, the regulation of NRT2.1 by N starvation is not only abolished when plants are treated in the dark. It happens also under HL conditions (Figures 1A and 5A). However, in that case, the level of NRT2.1 expression is high, even on normally repressive conditions like 10 mM ${\text{NO}}_3^{-}$ (Figures 1A and 5A), while in the dark the level of NRT2.1 stays low, independently of the level of N (Figure 1A). Comparison between Figures 1A and 5A also seems to indicate that the kinetics of the decrease in NRT2.1 expression, due to the loss of induction by ${\text{NO}}_3^{-}$, can vary, occurring fast in Figure 5A and likely later on in Figure 1A. Indeed, this variability in the time course of NRT2.1 response is something classically observed. For example, in Lejay et al. (1999), NRT2.1 expression increases at 24 and 48 h in response to N starvation, and starts to decrease at 72 h, while in some other experiments such as the one performed by Jung et al. (2018), the overall process occurred much faster, with the transient de-repression of NRT2.1 observed after 6 and 12 h of N starvation and the following decrease of NRT2.1 expression observed as soon as after 24 h of N starvation. On the contrary in other experiments such as the one presented in Figure 1A in LL, the decrease in the expression of NRT2.1 is not yet observed after 72 h of N starvation. Such variability in the kinetics of this response is not surprising as it results from the balance between the two above-mentioned opposite regulatory mechanisms, which are both likely to vary depending on the precise experimental conditions of each experiment (internal N status, tissue ${\text{NO}}_3^{-}$ stores). One model to explain these results is that enhancement of growth due to combination of HL and high ${\text{NO}}_3^{-}$ supply results in (1) a sustained high N demand for growth, relieving the feedback repression normally associated with high ${\text{NO}}_3^{-}$ supply and (2) a variation in ${\text{NO}}_3^{-}$ consumption depending on the growth rate leading to variation in the kinetics of the decrease of NRT2.1 expression due to the loss of the induction by ${\text{NO}}_3^{-}$. This model is supported by metabolomics analysis performed on Arabidopsis under diverse C and N nutrient conditions (Sato and Yanagisawa, 2014). Taken together, these results clearly support the idea that the control of NRT2.1 expression involves a complex network of interactions between signals emanating from N and C metabolisms. However, this level of complexity seems to be rather specific for NRT2.1. In contrast to NRT2.1, expression of NRT2.2, NRT2.4, and NRT2.5 is always repressed on 10 mM ${\text{NO}}_3^{-}$, independent of light levels (Figure 5A; Supplemental Figure S1A). It should be noted that in the N starvation experiments plants are transferred on a media with no N. This leads to the variation of two factors, the N status of the plants, which decreases when plants are starved for N, and the presence of ${\text{NO}}_3^{-}$ in the nutritive solution, which is suppressed by the transfer to N-free solution. Concerning the regulation of NRT2.2, NRT2.4, and NRT2.5 it is not known which one of these two factors is predominant since their expression was only measured in N starvation experiments (Kiba et al., 2012; Lezhneva et al., 2014; Kotur and Glass, 2015). It is thus possible that NRT2.2, NRT2.4, and NRT2.5 are only regulated locally by ${\text{NO}}_3^{-}$ and not by systemic signals of N demand. This idea is supported by the work of (Ma et al., 2015) showing that the regulation of NRT2.4 and NRT2.5 by N starvation depends on CBL7, which is specifically induced by ${\text{NO}}_3^{-}$ deficiency. Moreover, NIGT/HRS1s have been shown to act as transcriptional repressor of NRT2.4 and NRT2.5 upon ${\text{NO}}_3^{-}$ treatment (Kiba et al., 2018; Safi et al., 2018). Local regulation by ${\text{NO}}_3^{-}$ would explain why these transporters, unlike NRT2.1, are always repressed when plants are on 10 mM ${\text{NO}}_3^{-}$, regardless of the light conditions (Supplemental Figure S1A).

Identification of three candidates for regulation of NRT2 genes using a systems biology approach

Over the past few years, transcriptomic approach and systems biology have been powerful tools to identify regulatory elements involved in N signaling (For review Medici and Krouk, 2014; Vidal et al., 2015). For root NRT2s genes and HATS activity in Arabidopsis, it enabled the identification of CIPK23 and CIPK8 in response to ${\text{NO}}_3^{-}$, LBDs transcription factors in response to high N and BT2, a negative regulator of NRT2.1 and NRT2.4 under low N conditions (Figure 7; Ho et al., 2009; Hu et al., 2009; Rubin et al., 2009; Araus et al., 2016). For C and N signaling, previous microarray studies in response to transient treatments with ${\text{NO}}_3^{-}$, sucrose or ${\text{NO}}_3^{-}$ plus sucrose have been used to reveal, at the level of the genome, the existence of interaction between C and N signaling (Wang et al., 2003; Price et al., 2004; Scheible et al., 2004; Wang et al., 2004; Gutierrez et al., 2007; Huang et al., 2016). In Arabidopsis, 300 genes have been found differentially expressed by combined C:N treatments compared to C or N treatments (Palenchar et al., 2004). However, because of the number of genes affected by C and/or N regulation and the complex interactions between the signaling pathways, none of these studies have led so far to the identification and the validation of new regulatory elements. The unexpected regulations of root NRT2s and especially of NRT2.1 in our experimental set-up offer an interesting opportunity to find genes more specifically involved in the regulation of root ${\text{NO}}_3^{-}$ transporters by C and/or N, and to build a gene network model integrating regulators responding to N and/or C signals. As compared to previous transcriptomic approaches on N and C signaling in plants, we were able to narrow down the number of candidate genes by (1) using NRT2.1 as a specific target and (2) integrating the data from several Affymetrix microarrays to find gene networks co-regulated with the expression of NRT2.1 in response to different combinations of light and N treatments.

Figure 7.

Schematic representation of the known regulatory elements for the regulation of root high-affinity ${\text{NO}}_3^{-}$ transporters in response to external ${\text{NO}}_3^{-}$, the N status of the plant and light/photosynthesis. Purple circles represent the transcription factors identified in previous studies while red circles represent the transcription factors identified in our study. Light blue squares represent ${\text{NO}}_3^{-}$ transporter. Light orange squares represent CIPK kinases and CBL identified in previous studies. Yellow and green round squares represent the treatments and the pink diamond the chromatin factor HNI9 identified in previous studies.

Therefore, the gene regulatory network includes only three transcription factors, bHLH093, MYC1, and TGA3 (Figure 2B). bHLH093 was found co-regulated with NRT2.1 in response to light through photosynthesis because, like NRT2.1, it is not induced by light in the absence of CO₂ (Supplemental Figure S3). MYC1 and TGA3 were found co-regulated with NRT2.1 in response to N starvation. The analysis of their level of expression across all the experiments revealed that TGA3 and MYC1 are induced by N starvation but especially in LL and HL conditions, while bHLH093 seems overall induced by light no matter what the level of N (Figure 3A). Furthermore, MYC1 is also clearly induced by light (Figure 3A; Supplemental Figure S3). Taken together these results support the validity of our approach to find regulatory elements affected by C and/or N signaling and which are thus candidates for the regulation by C/N interaction. Interestingly, none of these three transcription factors was found involved in the regulation of root ${\text{NO}}_3^{-}$ transporters by previous studies. One explanation for this relates to the fact that the expression of bHLH093, MYC1, and TGA3 is not responsive to the induction by ${\text{NO}}_3^{-}$, which was by far the major environmental change investigated by previous studies (Supplemental Figure S4A). Conversely, none of the regulatory genes identified in previous studies was found with our approach. Indeed, most of them are not affected by N starvation and/or by light (Figure 4). The only exception is HY5, which encodes a recently identified mobile transcription factor involved in the regulation of NRT2.1 by sugar signals (Chen et al., 2016) and that is not found co-regulated with NRT2.1 in our analysis. This is explained by the fact that, unlike NRT2.1, we found HY5 induced by light even in the absence of CO₂ in our dataset (Figure 4). It indicates that expression of HY5 does not depend of the production of sugars through photosynthesis and is directly regulated by light. The role of HY5 in light signaling and not in C signaling is supported by previous studies showing that HY5 works downstream of phytochrome signaling (Quail, 2002; Li et al., 2010). Taken together, these results suggest that NRT2.1 would be induced by both a light component dependent on HY5 and a C component dependent on the OPPP (Lejay et al., 2008; de Jong et al., 2014; Chen et al., 2016). Accordingly, both Lejay et al. (2008) and Chen et al. (2016) found that induction of NRT2.1 by light is higher as compared to the addition of sucrose in the dark. Furthermore, there is still an induction of NRT2.1 expression by increasing supply of sucrose in the mutant hy5 (Chen et al., 2016).

bHLH093, MYC1, and TGA3, three transcription factors involved in the regulation of NRT2.4 and NRT2.5 gene expression

The use of mutants validated our approach and showed that bHLH093 has mainly a role in the induction by light of NRT2.4, while MYC1 and TGA3 affect induction by N starvation of both NRT2.4 and NRT2.5 and in a more modest way NRT2.2 (Figures 5A and 6). Furthermore, Y1H experiments support the fact that MYC1 and TGA3 are direct regulators of NRT2.4 and NRT2.5 as already suggested for TGA3 with NRT2.4 by the results obtained by O’Malley et al. (2016; Figures 3 and 5B). Conversely, chromatin immunoprecipitation experiments, using plants expressing bHLH093 fused to GFP, failed to reveal a robust interaction with the promoter of NRT2.4. It suggests that bHLH093 is an indirect regulator and that it is rather involved in the signaling pathway governing the regulation of NRT2.4 and in a lesser extent NRT2.1 and NRT2.5 by photosynthesis. The fact that bHLH093 mutation mainly affected the regulation of NRT2.4 and that NRT2.4 and NRT2.5 kinetics of regulation by light are different could also explain why it has no functional impact on HATS activity (Figure 6). Indeed, the activity of one of the transporter could compensate for the activity of the other one across the different timepoints. Furthermore, the induction of NRT2.1 expression by light could also participate to maintain HATS activity at the same level in the mutants compared to WT plants (Figure 6). By comparison, MYC1 and TGA3 mutations in the double mutant tga3.2/myc1.2 affect both the induction of NRT2.4 and NRT2.5 by N starvation across all the time points (Figure 5A). And in that case, NRT2.1 expression is low in the plants starved for N (Figure 5A). It could explain why the mutation of TGA3 and MYC1 has a functional impact on HATS activity in those conditions (Figure 5C).

As represented in Figure 7, most of the regulatory elements identified to date concern the primary ${\text{NO}}_3^{-}$ response, with only three elements involved in the repression by high N or high ${\text{NO}}_3^{-}$ and one in the induction by light. Along with CBL7, MYC1, and TGA3 seem thus to be part of an independent signaling pathway involved in the induction of root ${\text{NO}}_3^{-}$ transporters in response to low N, while bHLH093 is, to our knowledge, the first element involved in a regulatory mechanism linked to photosynthesis (Ma et al., 2015). As discussed above, the role of these transcription factors in the regulatory mechanisms involved in C/N interactions is also supported by our results. Indeed, the role of bHLH093 in the regulation by light seems to be dependent of the level of N and the role of MYC1 and TGA3 seems to be stronger in HL conditions for the regulation of NRT2.4 and NRT2.5 (Figures 5A and 6; Supplemental Figure S6). Altogether, these results show that NRT2 genes are subjected to multiple regulations involving a large number of regulators. This could explain why the impact of MYC1 and TGA3 mutations on NRT2.4 and NRT2.5 expression is only 20–30% of WT. Indeed, with the exception of the key NRT1.1 transceptor, limited differences (<50%) between WT and regulatory mutants concerning changes in NRT2s expression are not the exception but the general case for most of the regulators presented in Figure 7 (see for instance: Castaings et al., 2009; Hu et al., 2009; Ho et al., 2009; Ma et al., 2015; Xu et al., 2016). It shows that most regulators contribute to a limited part of the overall regulation, which seems to be a characteristic for NRT2s.

However, surprisingly, none of the three transcription factors we found affect strongly the regulation of NRT2.1, which we used as a target gene in our systems biology approach. This result could indicate that the regulatory mechanisms differ between the four NRT2 genes involved in the HATS. Indeed, NRT2.1 is regulated by at least 4 different mechanisms (local induction by ${\text{NO}}_3^{-}$ and repression by high ${\text{NO}}_3^{-}$, systemic repression by N metabolites and induction by C), while NRT2.4 and NRT2.5 are only regulated by one and/or two mechanisms (C and/or N starvation). Furthermore, our experimental setup revealed obvious complex interactions between N and C signaling for NRT2.1, which do not exist for NRT2.2, NRT2.4, and NRT2.5. As discussed above, if NRT2.2, NRT2.4, and NRT2.5 are only repressed by ${\text{NO}}_3^{-}$ and not by N metabolites, MYC1 and TGA3 could be involved in a ${\text{NO}}_3^{-}$-specific signaling pathway upregulating the very high-affinity transporters (NRT2.4 and NRT2.5) when the external ${\text{NO}}_3^{-}$ concentration becomes too low to be efficiently taken up by NRT2.1. However, despite the fact that TGA3 is part of a family of seven genes in Arabidopsis, in which two of them, TGA1 and TGA4, have already been involved in the induction of NRT2.1 and NRT2.2 in response to ${\text{NO}}_3^{-}$, there is no direct evidence to support a role of both TGA3 and MYC1 in a ${\text{NO}}_3^{-}$ signaling pathway (Bruex et al., 2012; Alvarez et al., 2014;).

Since NRT2.1 and NRT2.4 are both regulated by C through OPPP, it was even more surprising to find that the absence of bHLH093 affects mainly the induction by light of NRT2.4 compared to NRT2.1 (Figure 6; Lejay et al., 2008). However, the role of bHLH093 seems to be dependent on the level of N since it plays a significant role in the regulation of NRT2.4 only under low N conditions, whereas the induction of NRT2.1 by light is mostly seen in this experiment under high N conditions (10 mM ${\text{NO}}_3^{-}$). It could explain why, in those conditions, bHLH093 mutation does not affect the regulation of NRT2.1 by light, while in the second experiment, where plants were grown on a moderate level of ${\text{NO}}_3^{-}$ (1 mM), NRT2.1 is well induced by light and the ${\text{NO}}_3^{-}$ concentration could be low enough to reveal the impact of bHLH093 on NRT2.1 regulation (Figure 6B). To our knowledge, the role of bHLH093 in the roots and in response to light has never been characterized before. The only information concerns a role in flowering promotion under non-inductive short-day conditions through the gibberellin pathway (Sharma et al., 2016).

Materials and methods

Plant material

Arabidopsis (A. thaliana) genotypes used in this study were the WT Col-0 ecotype and mutants obtained from the Salk Institute: tga3.2 (Salk_081158), tga3.3 (Salk_088114), myc1.2 (Salk_057388), myc1.3 (Salk_006354), bHLH093.1 (Salk_121082), and bHLH093.5 (Salk_104582).

In all experiments plants were grown hydroponically under non-sterile conditions as described by Lejay et al. (1999). Briefly, the seeds were germinated directly on top of modified Eppendorf tubes filled with pre-wetted sand. The tubes were then positioned on floating rafts and transferred to tap water in a growth chamber under the following environmental conditions: light/dark cycle of 8/16 h, light intensity of 250 µmol·m⁻²·s⁻¹, temperature of 22°C/20°C, and RH of 70%. After 1 week, the tap water was replaced with a complete nutrient solution. The experiments were performed on plants grown on 1 mM ${\text{NO}}_3^{-}$ as N source. The other nutrients were added as described by Lejay et al. (1999). The plants were allowed to grow for three additional weeks before the experiments. Nutrient solutions were renewed weekly and on the day before the experiments.

Treatments

Two different sets of experiments were performed to (1) study the impact of light on the regulation of ${\text{NO}}_3^{-}$ transporter genes in the roots by N starvation, and (2) study the impact of the N status of the plants on the regulation of these genes by light.

In the first set of experiments 4 weeks old plants were transferred to a solution containing 10 mM ${\text{NO}}_3^{-}$. After 1 week the plants were transferred in the morning either to continuous dark or to light/dark cycle at three different light intensities (50, 250, and 800 μmoles.h⁻¹.m⁻²) and N starved for 24, 48, and 72 h, by replacing ${\text{NO}}_3^{-}$ with CaCl₂ 2.5 mM and K₂SO₄ 2.5 mM.

In the second set of experiments 4 weeks old plants were transferred to a solution containing 10 mM ${\text{NO}}_3^{-}$. They were then pre-treated during 3 d on nutrient solution containing different levels of N: (1) no N, (2) 1 mM ${\text{NO}}_3^{-}$, or (3) 10 mM ${\text{NO}}_3^{-}$. After 32 h in the dark the plants were transferred to light for 1, 2, 4, and 8 h under 3 different light intensities (50, 250, and 800 μmoles.h⁻¹.m⁻²).

The dependence of the expression of ${\text{NO}}_3^{-}$ transporter genes on photosynthesis was investigated by modifying the CO₂ concentration in the atmosphere. After a pre-treatment of 40 h in the dark, plants grown on 1 mM ${\text{NO}}_3^{-}$ were placed for 4 h in the light (∼150 µmol·m⁻²·s⁻¹) or in the dark in a 240-L airtight plexiglass chamber connected to a computerized device for controlling temperature, humidity, and CO₂ concentration in the atmosphere (Atelliance Instruments; see Delhon et al. (1996) for details). The CO₂ concentration in the atmosphere was held constant during the treatments at 0 or 600 μL L⁻¹. All experiments were repeated 2 or 3 times.

RNA extraction and gene expression analysis

Root samples were frozen in liquid N₂ in 2-mL tubes containing one steel bead (2.5 mm diameter). Tissues were disrupted for 1 min at 30 s⁻¹ in a Retsch mixer mill MM301 homogenizer (Retsch, Haan, Germany). Total RNA was extracted from tissues using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Subsequently 4 µg of RNA were treated with DNase (DNase I; SIGMA-ALDRICH, St. Louis, MO, USA) following the manufacturer’s instructions. RT was achieved in the presence of Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI, USA) after annealing with an anchored oligo(dT)₁₈ primer as described by Wirth et al. (2007). The quality of the cDNA was verified by PCR using specific primers spanning an intron in the gene APTR (At1g27450) forward 5′- CGCTTCTTCTCGACACTGAG-3′; reverse 5′-CAGGTAGCTTCTTGGGCTTC-3′.

Gene expression was determined by RT-qPCR (LightCycler; Roche Diagnostics, Basel, Switzerland) with the kit LightCycler FastStart DNA Master SYBR Green I (Roche Diagnostics, Basel, Switzerland) according to the manufacturer’s instructions with 1 µL of cDNA in a total volume of 10 µL. The amplifications were performed as described previously by Wirth et al. (2007). All the results presented were standardized using the housekeeping gene Clathrin (At4g24550). Gene-specific primer sequences were: NRT2.1 forward, 5′-AACAAGGGCTAACGTGGATG-3′; NRT2.1 reverse, 5′-CTGCTTCTCCTGCTCATTCC-3′; NRT2.2 forward, 5′-GCAGCAGATTGGCATGCATTT-3′; NRT2.2 reverse, 5′-AAGCATTGTTGGTTGCGTTCC-3′; NRT2.4 forward, 5′- GAACAAGGGCTGACATGGAT -3′; NRT2.4 reverse, 5′- GCTTCTCGGTCTCTGTCCAC -3′; NRT2.5 forward, 5′-TGTGGACCCTCTTCCAAAAA-3′; NRT2.5 reverse, 5′- TTTGGGGATGAGTCGTTGTGG-3′; MYC1 forward, 5′-AACCTTAACGACTCTGTG-3′; MYC1 reverse, 5′-CCGCAACTATGTAGTCTCTG-3′; TGA3 forward, 5′-CTCTCAGAAAGTGTTGGC-3′; TGA3 reverse, 5′-CATATACGAGGAGATGAGTG-3′; bHLH093 forward, 5′-AGCTTGAAGGCCAACC-3′; bHLH093 reverse, 5′-GCTCTTTCATGTAATCTATGGCA-3′; Clathrin forward, 5′-AGCATACACTGCGTGCAAAG-3′; Clathrin reverse, 5′-TCGCCTGTGTCACATATCTC-3′.

${\text{NO}}_3^{-}$ influx studies

Root ${\text{NO}}_3^{-}$ influx was assayed as described by (Delhon et al., 1995). Briefly, the plants were sequentially transferred to 0.1 mM CaSO₄ for 1 min, to a complete nutrient solution, pH 5.8, containing 0.005 mM ¹⁵${\text{NO}}_3^{-}$ (99 atom % excess¹⁵N) for 5 min, and finally to 0.1 mM CaSO₄ for 1 min. Roots were then separated from shoots, and the organs dried at 70°C for 48 h. After determination of their dry weight, the samples were analyzed for total N and atom % ¹⁵N using a continuous-flow isotope ratio mass spectrometer coupled with a C/N elemental analyzer (model Euroflash Eurovector, Pavia, Italy) as described in (Clarkson, 1986).

Acquisition of genome-wide expression and statistical analysis

Genome-wide expression was determined using Affymetrix ATH1 GeneChip expression microarrays according to manufacturer’s instructions. To do so, biotinylated cRNA was synthesized from 200 ng of total RNA from Arabidopsis roots. Affymetrix data were normalized in R (http://www.r-project.org/) using MAS5.

Then, normalized data were subjected to different statistical analyses, all centered on NRT2.1 expression pattern but including various sets of microarray data among the whole data set. As a first approach to build a gene network involved in the regulation of root ${\text{NO}}_3^{-}$ transporters, we examined genes displaying expression pattern correlated to NRT2.1 expression pattern across the entire dataset. An R² coefficient cut-off 0.8 or  −0.8 led to the identification of 79 AGIs displaying an expression pattern correlated to NRT2.1, including 77 genes positively correlated with NRT2.1 (Supplemental Table S1). Among these 79 genes, none of them displays a function related to gene regulation but rather related to metabolic activity and more precisely to carboxylic acid metabolic process as, for example, the Glutamate synthase 2 gene (Supplemental Figure S8A). Moreover, a hierarchical clustering of the treatments according to the expression pattern of these genes clearly revealed that their response is largely driven only by the light/C factor, putting aside any possible regulation by N provision (Supplemental Figure S8B). Therefore, a global analysis of the entire data set was not relevant to identify regulators of ${\text{NO}}_3^{-}$ transport integrating C and N availability and an analysis of gene expression in different subsets of treatments was deemed more powerful. The list of genes regulated by N-deprivation specifically under LL regime was determined by a t test analysis (P-value < 0.05) between Conditions 3 and 4. All genes also found regulated between Conditions 1 and 2 based on the same analysis are removed from this list (Figure 1; Supplemental Table S1). Genes regulated by N-deprivation during light induction are determined by a two-way ANOVA using N as one factor (presence = Conditions 7, 9, 11, and 13/absence = Conditions 8, 10, 12, and 14) and Light as the second factor (no Light = Conditions 7, 8/1 h-light = Conditions 9, 10/2 h-light = conditions 11, 12/4 h-light = Conditions 13, 14). Genes of interest are regulated by the interaction of the two factors (P-value < 0.05) and display a similar regulation by N from dark to 2 h-light as observed for NRT2.1 (Figure 1; Supplemental Table S2). Genes regulated by light intensity under high N-provision and by light time exposure under high N-provision are both determined by a linear modeling of gene expression across light intensity (Conditions 1, 3, 5, and 6) or time exposure (Conditions 7, 9, 11, and 13) using an R²  0.9 (P-value is ˂0.003; Figure 1; Supplemental Tables S3 and S4). Finally, genes regulated by photosynthesis activity are determined by a two-way ANOVA using CO₂ level as one factor (0 ppm = Conditions 15, 17/600 ppm = Conditions 16 and 18) and Light as the second factor (Dark = conditions 15, 16/Light = Conditions 17, 18). To narrow down the list of NRT2.1-like genes, only those passing post hoc Tukey tests comparing Conditions 18 to all 3 others (P-value < 0.05) and displaying a ratio >2 or <0.5 are selected (Figure 1; Supplemental Table S5).

Visualization of gene connectivity by clustering and gene network analysis

Heat map hierarchical cluster of gene expression and samples was generated with the MeV software using Pearson correlation as distance metric and Average as linkage method (www.tm4.org; Saeed et al., 2003). The Gene Network was generated with the VirtualPlant version 1.3 software (http://virtualplant.bio.nyu.edu/cgi-bin/vpweb/; Katari et al., 2010). The connectivity of the nodes is based on five categories corresponding to literature data, post-transcriptional regulation, protein–protein interactions, transcriptional regulation, and regulated edges meaning transcription factor–target relationship based at least on one binding site in the promoter of the target gene. Two nodes are linked by an edge if they fall in any of these categories combined to an expression pattern correlated at an R² > 0.7 or <−0.7. Visualization of the gene regulatory network has been performed with Cytoscape (http://www.cytoscape.org/; Shannon et al., 2003). Node properties have been modified to reveal connectivity with the three transcription factors and highlight NRT2.1 position within the network.

Y1H assays

For the generations of the plasmids for promoter analysis by Y1H, particular promoter fragments of NRT2.4 (1,968 bp), NRT2.5 (1,692 bp) were first amplified by PCR with overlapping ends as described by Gibson et al. (2009). For the bait, the pMW2 and pMW3 vectors were used (Deplancke et al., 2006). pMW vectors were amplified by PCR with overlapping ends as a single sequence (pMW2) or as two independent sections (pMW2). Final vectors were made as described by Gibson et al., 2009. The Y1H prey vectors for TGA3 and MYC1 transcriptions factors were a kind gift from Franziska Turck (Castrillo et al., 2011). All the fragments generated for all constructs were validated by DNA sequencing.

The Y1H assay was performed according to protocol described by Grefen (2014) with minor modifications. Briefly, the vectors pMW2-NRT2.4, pMW3-NRT2.4, pMW2-NRT2.5, and pMW3-NRT2.5 were first linearized with restriction enzymes. For pMW2 vectors BamH1 (NEB) was used and for pMW3 vectors Xho1 (NEB). The resulting linearized constructs were subsequently co-integrated into the yeast (Saccharomyces cerevisiae) strain: YM4271 as described by Grefen (2014). The transformed yeast strains were tested for autoactivation and the selected colonies with the higher sensitivity to 3-AT were then transformed with the construct pDEST-AD-TGA3 or pDEST-AD-MYC1 or pDEST-AD (Empty vector). Empty vector was included as a negative control. Resulting yeast was dropped on selection media (SD –His–Ura–Trp) supplemented with increasing concentrations of 3-AT (0, 15, 30, 50, 80, and 100 mM). Yeast growth was verified after 48 h.

Accession numbers

NRT2.1: AT1G08090

NRT2.2: AT1G08100

NRT2.4: AT5G60770

NRT2.5: AT1G12940

MYC1: AT4G00480

TGA3: AT1G22070

BHLH093: AT5G65640

Supplemental Data

The following materials are available in the online version of this article.

Supplemental Figure S1 Interaction between N and Light/C provision modulates mRNA accumulation in roots of most of the NRT2 family members.

Supplemental Figure S2 Expression pattern of NRT2 family genes in the set of N and C/Light combinations as determined by Arabidopsis Affymetrix ATH1 microarray hybridization.

Supplemental Figure S3 Expression pattern of TGA3, MYC1 and bHLH093 transcription factors in the set of N and C/Light combination as determined by Arabidopsis Affymetrix ATH1 microarrays hybridization.

Supplemental Figure S4  TGA3, MYC1 and bHLH093 display expression patterns different than most of the known regulators of NRT2 genes in response to ${\text{NO}}_3^{-}$.

Supplemental Figure S5 Expression pattern of TGA3, MYC1, and bHLH093.

Supplemental Figure S6 TGA3 and MYC1 are not required for NRT2.1, NRT2.4, and NRT2.5 full induction during N-deprivation in LL conditions

Supplemental Figure S7 MYC1 and TGA3 are not required for NRT2s induction by light.

Supplemental Figure S8 Biomaps and hierarchical clustering of the 79 genes most correlated with NRT2.1 expression across all experiments.

Supplemental Table S1 List of 430 probes regulated by N deprivation under LL regime only

Supplemental Table S2 List of 573 probes regulated by the interaction between N and light.

Supplemental Table S3 List of 128 probes linearly regulated by light intensity

Supplemental Table S4 List of 985 probes linearly regulated during light induction

Supplemental Table S5 List of 509 probes regulated by the interaction between light and CO₂

Supplemental Table S6 List of 80 probes coregulated based on NRT2.1 expression and Pearson correlation

 

S.R. performed the transcriptomic experiments, the analysis, and the generation of the gene regulatory network. S.R. and V.C. obtained and performed the experiments to characterize the mutants along with J.P.T., I.F., P.T., and L.L. J.A.O. with C.I. performed Y1H experiments. T.M., C.F., and R.G. performed bioinformatics and statistical analysis for the gene regulatory network and the interaction of the transcription factors with the NRT2 promoters. L.L., S.R., A.G., and R.G. designed the experiments. L.L., S.R., and A.G. wrote the manuscript.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Laurence Lejay (laurence.lejay@inrae.fr)