Microarray Analysis of the Nitrate Response in Arabidopsis Roots and Shoots Reveals over 1,000 Rapidly Responding Genes and New Linkages to Glucose, Trehalose-6-Phosphate, Iron, and Sulfate Metabolism

Comparison of Root and Shoot Responses

The data from the array give signal (RNA) levels and induction ratios, which were averaged for roots and for shoots using the biological replicates. An overall view of the expression data is presented as a scatter plot (Fig. 1). Signal intensities from nitrate-treated samples (y axis) were plotted over control samples (x axis) for roots (Fig. 1A) and shoots (Fig. 1B). Guide lines are provided in the plots showing when signal ratios are 1, 2, and 0.5 (–2). The data show that after 20 min of nitrate treatment, many genes are both induced and repressed greater than 2-fold in both roots and shoots. In addition, the scatter is much greater for roots than in shoots, indicating that the RNA response to nitrate is much larger in roots under these conditions. Care should be taken in interpreting the data points for signal intensities less than 100 (boxed off in the plots), which, as described below, is an approximate limit of reliable detection. If a signal is below 100, it is often not meaningful, and the resulting ratios can be artificially high.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Scatter plots of ATH1 array data. Plotted are signal intensities for roots (A) and shoots (B). Values are averaged from two biological replicates for nitrate-treated samples and chloride-treated (control) samples. Guide lines are given showing signal values of 100 and showing signal ratios of 1 (no change), 2 (induced), and –2 (repressed).

To get an accurate count of the number of genes that are induced or repressed by the nitrate treatment, an assessment of the sensitivity of detection is needed. If signal values fall below a reliable detection limit, the values are not useful for determining signal ratios. To determine which genes show a reliable level of RNA, the Affymetrix Microarray Suite software (MAS version 5.0) was used to analyze the signals for each probe set. The results of this analysis are summarized in Table I. First, those genes that showed reliable signal levels in both biological replicates were identified. When a transcript was reliably detected, it was given a detection call of “P” (present), and when it was not detected, it was given a detection call of “A” (absent). For roots, 13,846 genes of a total of 22,626 had P call values in both root samples, and 13,221 genes had P call values for both shoot samples (Table I). Thus, a little more than one-half of the genes represented on the ATH1 chip gave reliable signals. Next, changes in RNA levels in response to nitrate (i.e. induction or repression) were assessed using Wilcoxon's signed rank tests as described (Affymetrix Statistical Algorithms Reference Guide). When the increase in signal level in response to nitrate is significant, a call of “I” (increase) is given. When a decrease in signal is significant, a call of “D” (decrease) is given. When this test was performed, the following results were obtained. In roots, 555 genes had signals that significantly increased, and 621 had signals that decreased. In shoots, 171 genes showed increased signals, and 12 showed a decrease. If a threshold value of 2 was used for the signal ratio, the number of genes with increased signals was 251 in roots and 76 in shoots. The number of genes with decreased signals was less: 78 with decreased signals in roots and 2 with decreased signals in shoots. Thus, there is a 6-fold greater number of genes responding to nitrate in roots using the Wilcoxon's test (1,176 responsive genes in roots versus 183 in shoots) and a 4-fold greater response in roots using a cutoff value of 2 (Fig. 2 showing 329 in roots versus 78 in shoots). These results provide quantitative evidence on a genomic scale showing that roots have a more dramatic response to nitrate than shoots after 20 min.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Numbers of induced/repressed genes in roots and shoots. Histogram pair shows the number of genes that were given a call of present (reliable signal intensity) for RNA samples from both biological replicates and had average ratios 2.0 and above and –2.0 and below. Note that the signal ratios were determined by the Affymetrix MAS software by comparing each probe pair on the induced array with the corresponding probe pair on the control array and, thus, are not simply the ratio of the average signal intensities shown in Figure 1. Bars above the 25 value correspond to values that were greater than 25 and were grouped together.

Analysis of Nitrate Transporter Genes

Compared with the 25 nitrate-responsive genes identified previously for 250 μm nitrate at 20 min (Wang et al., 2000), the ATH1 data provide more responsive genes by more than an order of magnitude. To analyze the genes from the ATH1 array in more detail, data for specific groups of genes were extracted from the array tables and studied. First, data for the nitrate transporter gene families NRT1 and NRT2 gene families were examined (Fig. 3). Of 11 NRT genes, only four showed significant induction in roots (NRT1.1, 2.1, 2.2, and 2.4; note that for NRT2.6, the signal calls were “absent,” indicating that expression levels were too low to be reliable). Of these four genes, NRT2.1 and NRT2.2 show the most dramatic induction ratios. In fact, NRT2.2 had the third highest induction ratio of any gene in roots; however, the signal from the chloride sample for this gene was not reliable (“absent”) so that the value of the ratio is not reliable. If one considers only genes that had significant (“present”) signals for both the induced and control samples, NRT2.1 was the 10th most induced gene in the complete array behind those encoding G6PDH, NiR, phosphoglycerate mutase (PGM), and six unknown genes. The strong induction of NRT2.1 is noteworthy because this gene is subject to repression by ammonium or high nitrogen. In our previous experiments where plants were grown on 20 mm ammonium, induction of NRT2.1 was not observed at 250 μm nitrate for 20 min or 2 h (Wang et al., 2000). It took 5 to 10 mm nitrate for 2 h to induce NRT2.1. The fact that NRT2.1 is induced so strongly in the ATH1 experiments described here indicates that the ammonium repression of NRT2.1 is very weak in these plants, presumably due to the fact that the ammonium has been reduced to undetectable levels by the time the plants were treated with nitrate. In shoots, only two NRT genes showed response to nitrate (NRT2.1 and NRT2.5), but the response by NRT2.5 was very small, and the ratio for NRT2.1 is not reliable because the control signal was “absent.” Thus, shoot NRT genes show very little response to 250 μm nitrate after 20 min.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

NRT family of genes. Shows average signal ratios and signal intensities for the NRT nitrate transporter gene families. Note that the signal ratios were determined by the Affymetrix MAS software by comparing each probe pair on the induced array with the corresponding probe pair on the control array and, thus, are not simply the ratio of the average signal intensities. NC, No change as determined by the Affymetrix MAS software. A value is given for the ratio only if both RNA samples had a call of I or D. An asterisk by a signal value indicates an “Absent” call.

Another interesting finding shown in Figure 3 is the distribution of RNA signals for each family. Based on signal levels, NRT1.1 and NRT2.1 account for 78% of the total NRT signal in nitrate-treated roots, indicating that these two genes produce the vast majority of nitrate transporter mRNA in nitrate-treated roots of these 10-d-old seedlings. In control roots, NRT1.1 and NRT1.2 account for 64% of the NRT mRNA. For shoots, the data show that the mRNA from the NRT1 family of genes makes up over 90% of the total NRT mRNA in treated and control plants. These results indicate that, at least at the mRNA level, NRT1.1 and NRT2.1 dominate in roots that are freshly treated with nitrate, and NRT1 genes predominate in shoots of plants regardless of the nitrate treatment. It should be noted that for these experiments plants were grown with Suc in the media, which raises the level of NRT1.1 mRNA in roots (Lejay et al., 1999).

It is interesting to compare the results from the ATH1 analysis with those reported previously for NRT2 genes in Arabidopsis using quantitative PCR (Orsel et al., 2002). In the study by Orsel et al. (2002), plants were grown on 1 mm KNO₃ with short days and without Suc, and then analyzed at 29 d and after flowering. Orsel et al. (2002) reported that NRT2.1 mRNA predominates in roots with NRT2.4, 2.5, 2.6, and 2.3 showing detectable levels in that order. Our results (obtained for 10-d-old seedlings grown with 24 h of light with Suc) show that NRT2.1 mRNA levels also dominate in nitrate-treated roots followed by NRT2.4, 2.5, and 2.2. In shoots, Orsel et al. (2002) reported that NRT2.4 had the highest mRNA levels with detectable signals from NRT2.5, whereas our nitrate-treated plants showed that NRT2.5, 2.7, and 2.1 gave equivalent signals, and NRT2.4 gave no detectable signal. This comparison shows that NRT2.1 encodes the majority of NRT2 mRNA in roots in the presence of nitrate, whereas shoots show different patterns of NRT2 gene expression depending on the age and growth conditions.

Analysis of Nitrate Assimilation Genes

The next group of genes that was examined encodes nitrate assimilatory enzymes involved in reduction of nitrate to nitrite and nitrite to ammonium (Fig. 4). The two NR genes, NIA1 and NIA2, were both induced in roots and shoots as expected. In roots, induction of NIA1 was much higher (15-fold) than for NIA2 (2-fold). This differential induction of NIA1 and NIA2 has been described previously (Cheng et al., 1991). NiR was also dramatically induced in both organs as previously described (Wang et al., 2000). In fact, NiR was the third most induced gene in shoots and the eighth highest in roots. UPM1, which encodes an enzyme that makes a cofactor for NiR, was also highly induced in both organs. In our previous work, two FNR and one Fd genes, involved in supplying reductant to NiR, were found to be induced (Wang et al., 2000). In the ATH1 array, the same result was obtained. Two of six FNR genes and one of six Fd (or Fd-like) genes were induced by nitrate in both shoots and roots. The induction of the two FNR genes was substantial in roots. Another group of genes involved in producing reductant for nitrite reduction encodes enzymes in the pentose phosphate pathway, which generates NADPH. As described previously, genes encoding two enzymes in the oxidative arm of the pathway, G6PDH and 6PGDH, and genes encoding enzymes in the nonoxidative arm, transketolase and transaldolase, were induced (Wang et al., 2000). The ATH1 array gave the same result (Fig. 4). The induction was strongest in the root except for one of the G6PDH genes. This gene (At1g24280) was the second most induced gene in shoots and the fifth highest in roots.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Nitrate/nitrite assimilatory genes. Shows average signal ratios and signal intensities for NR, NiR, and genes involved in nitrite reduction, including those in the pentose phosphate pathway. An asterisk by a signal value indicates an “Absent” call.

Last, genes involved in ammonium assimilation were examined (Fig. 5). We found that one GS gene (GLN2), one GOGAT gene (NADH dependent, GLT1), and ASN2 are induced in roots after 20 min. In shoots, no GS gene, only one GOGAT gene (GLT1), and two ASN genes (ASN1 and ASN2) were induced, although the response was low. These results are similar to what we reported previously with the glass slide arrays where one GS, one Asn synthetase (ASN2), and two NADH-GOGAT genes were found to be induced either at 20 min or 2 h (Wang et al., 2000).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Ammonium assimilatory genes. Shows average signal ratios and signal intensities for GS, GOGAT, and Asn synthetase (AS). An asterisk by a signal value indicates an “Absent” call.

Analysis of Glycolysis-Related Genes

The most interesting and novel finding in our data was the induction of genes that are involved or potentially involved in glycolysis. First, two genes encoding the glycolytic enzymes PGM and Glc-6-phosphate isomerase (G6PI) were induced. Glycolytic enzymes and their corresponding genes are not known to respond to nitrate. Both PGM and G6PI are involved in isomerization reactions that involve little change in free energy and, thus, are not rate-limiting steps in glycolysis. It is interesting, however, that G6PI serves both glycolysis and the pentose phosphate pathways. G6PI interconverts Glc-6-P and Fru-6-P, both intermediates in glycolysis. Fru-6-P is also a product of the pentose phosphate pathway, and Glc-6-P is the initial substrate. Glycolysis and pentose phosphate oxidation occur in the same compartments (cytosol and plastids) and are integrated (Dennis and Blakeley, 2000). It is possible that G6PI induction helps recycle carbon back to the pentose phosphate pathway and, thus, aids in the generation of NADPH by pentose phosphate oxidation. Another product of the pentose phosphate pathway is glyceraldehyde-3-P. This is also a glycolytic intermediate, which when oxidized to glycerate-3-P, becomes the substrate for PGM. Perhaps PGM induction is needed to help metabolize glyceraldehyde-3-P generated by the pentose phosphate pathway. The induction of G6PI is modest, but induction of one PGM gene is very strong, being in the top 20 induced genes in roots and the top three induced genes in shoots. There are six potential PGM genes identified in the ATH1 array, and three of these genes are induced by nitrate.

The next six genes involve the metabolism of trehalose-6-P. Trehalose is a disaccharide of Glc and is synthesized in two steps catalyzed by trehalose-6-P synthase (TPS) and trehalose-6-P phosphatase (TPP). In yeast (Saccharomyces cerevisiae), trehalose-6-P regulates glycolysis. In plants, its role is not known, but trehalose-6-P is implicated in controlling photosynthetic capacity (Paul and Pellney, 2003), and mutations in one of 11 possible TPS genes in Arabidopsis leads to an embryo lethal phenotype (for review, see Eastmond et al., 2003). Genes encoding enzymes that are implicated in both steps of trehalose synthesis are induced in roots but not shoots (Fig. 6). The induction of one of the TPP genes is quite substantial. There are a total of 10 potential TPP genes and 11 potential TPS genes in the ATH1 array. This result suggests that nitrate may be altering trehalose-6-P levels, which may, in turn, affect glycolysis if it behaves in plants as it does in yeast. Given the strong connection between nitrate reduction and pentose oxidation, it is also possible that trehalose-6-P is a regulator of the pentose phosphate pathway.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Other metabolic genes. Shows average signal ratios and signal intensities for additional and novel metabolic and transporter genes. An asterisk by a signal value indicates an “Absent” call.

Analysis of Other Metabolic and Regulatory Genes

As has been reported before, nitrate signals a shift from starch biosynthesis to organic acid production (for review, see Stitt, 1999; Foyer et al., 2003). Three enzymes involved in this process, PPC kinase, PPC, and NAD-dependent malate dehydrogenase (NMDH), all show gene induction in roots, and the latter two show induction in shoots (Fig. 6). There are four PPC genes on the ATH1 array, and only one shows induction. The NMDH induction is interesting because it has been found in the nitrate assimilation gene cluster in Chlamydomonas reinhardtii (Quesada et al., 2000). Together with the malate-oxalacetate translocator (also induced in roots and shoots, Fig. 6), NMDH is thought to export reducing power from the chloroplast to the cytosol to support nitrate reduction.

As observed previously (Wang et al., 2000), the nicotianamine synthase gene is induced by nitrate. Induction of two nicotianamine synthase genes is observed with the ATH1 array in roots. Nicotianamine is involved in iron (Fe) acquisition, transport, and homeostasis in plants (van Wiren et al., 1999; Pich et al., 2001). Fe is required for many of the enzymes in nitrate assimilation including NR, NiR, and Fd. Perhaps nitrate is inducing the synthesis of nicotianamine to facilitate the transport of Fe to support synthesis of nitrate assimilatory enzymes.

Another set of genes induced by nitrate is implicated in sulfur metabolism. Two putative sulfate transporter genes and a 5-adenylylphosphosulfate reductase gene are induced in roots; the latter is also induced in shoots. 5-Adenylylphosphosulfate reductase is part of a novel pathway in plants for the assimilation of sulfate (Leustek et al., 2000). A linkage between sulfur and nitrogen metabolism has been known for many years, with a deprivation of one leading to a disruption of the metabolism of the other (Reuveny et al., 1980; Prosser et al., 2001). It has also been shown in barley (Hordeum vulgare) that high levels of nitrate or ammonium can induce a high-affinity sulfate transporter gene and sulfate uptake in nitrogen-starved plants, suggesting that a nitrogen metabolite may affect sulfate transporter gene expression (Vidmar et al., 1999). The data in Figure 6 suggest that nitrate itself can induce sulfate uptake and assimilation genes and, thereby, may increase sulfate assimilation.

Two genes involved in molybdopterin biosynthesis are induced in roots and shoots (Fig. 6). Molybdopterin serves as a cofactor for NR, and many of the genes involved in the synthesis of this molecule have been identified (Mendel and Hansch, 2001).

Lastly, none of the 35 aquaporin water channel genes on the ATH1 array are induced by nitrate in either roots or shoots. However, one is significantly repressed in roots (NIP2.1, Fig. 6). In fact, it is the third most repressed gene in the list of expressed genes in the ATH1 list. In tomato roots, several aquaporin genes were found to be induced by nitrate (Wang et al., 2001); however, this induction was observed 6 to 48 h after treatment so that these results are not comparable with the ATH1 data, which were taken 20 min after treatment.

In addition to the metabolic genes described above, many potential regulatory genes were found to respond to nitrate after 20 min. If the analysis is restricted to genes that show reliable signals and response ratios equal to or greater than 2-fold, 65 genes in roots and 12 genes in shoots were identified as possible regulatory genes. As an example, the two most highly induced genes in roots are a putative protein phosphatase (At4g32950, 8.9-fold increase) and a putative protein kinase (At1g49160, 8.6-fold increase). A putative Leu zipper transcription factor (At5g65210, similar to the tobacco [Nicotiana tabacum] transcription factor TGA1a; Katagiri et al., 1989) that is induced by nitrate in tomato roots (Wang et al., 2001) is also induced 4-fold in roots and 1.8-fold in shoots in the ATH1 data. A summary of the results showing how many genes in each class for roots and shoots is shown (Fig. 7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Potential regulatory genes. Pie charts showing the number of genes identified on the ATH1 array in one of several categories (transcription factors/DNA-binding proteins, protein kinase/phosphatases, response regulators, RING Zn finger proteins, and other) for (A) roots and (B) shoots showing 2.0-fold or greater induction or repression.

Verification of the ATH1 Array Data

We used several criteria to verify the array data. First, two biological replicates were performed, and only genes that showed reliable signals and signal ratios in both replicates were considered. Second, we performed real-time quantitative PCR on a select group of genes and compared the results with those obtained from the array (Fig. 8). The induction ratios from the PCR data generally exceeded those from the array; however, if the overall difference is used to normalize the data, then the correlation between the PCR and array data are very good. Those genes that were the most highly induced (e.g. NiR, NRT2.1, and TPP-At1g78090) gave the highest induction ratios in both methods, whereas those that were the least induced (e.g. NRT1.1 in shoots) gave the lowest ratios in both methods.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Comparison of real-time PCR and ATH1 array data. Histograms shows ATH1 signal ratios and quantitative real-time PCR ratios for select genes. The array data are averages of two biological replicates, and the error bar shows the spread between the two data points. The PCR data were obtained from two independent reactions for each biological replicate (total of four reactions) and averaged. Error bars = se. Ubiquitin 10 was used as the standard to normalize each sample. A, Induced genes involved in nitrate uptake and nitrate or nitrite reduction. B, Induced genes involved in trehalose-6-P synthesis.

Plant Growth Media

The growth medium contained the following components (final concentration): 10 mm potassium phosphate (pH 6.5), 2.5 mm (NH₄)₂succinate, 2 mm MgSO₄, 1 mm CaCl₂, 0.1 mm FeNa₂EDTA, 0.5% (w/v) Suc, and micro-nutrients (50 μm H₃BO₃, 12 μm MnSO₄, 1 μm ZnCl₂, 1 μm CuSO₄, and 0.2 μm Na₂MoO₄). Component solutions were autoclaved (FeNa₂EDTA was filter sterilized) separately and added to sterile deionized water before use.

Hydroponics Growth Conditions

Arabidopsis ecotype Columbia was grown at 25°C to 27°C under constant illumination with agitation at 120 rpm for 10 d. Plants were grown in a 50-mL glass beaker containing a segment (20 mm in height) of a 50-mL Falcon centrifuge tube with triangular openings at bottom to allow for liquid flow. On top of the tube was a disc of polypropylene mesh (250-micron pore size, 25 mm in diameter). Beakers initially contained 25 mL of sterile medium (described above) and approximately 50 surface-sterilized seeds evenly distributed on the mesh disc. On d 5, 5 mL of medium was removed from the beaker to allow shoots to stay above the medium liquid. The beaker top was covered with three layers of aluminum foil to keep the culture sterile.

Real-Time Quantitative PCR

Real-time PCR was performed using a LightCycler System from Roche Diagnostics (Indianapolis). Template cDNA samples were prepared using the SuperScript First Strand Synthesis System Kit (Invitrogen, San Diego) for reverse transcriptase-PCR with 3 μg of total RNA. The cDNA synthesis reaction mixture was diluted 25 times before being used for PCR. Primers for the PCR reactions were designed to have a melting temperature of about 63°C to 69°C and to give a PCR product between 175 and 250 bp. The genes tested for expression levels by quantitative PCR were as follows: NRT1.1, forward primer (5′-AAAGCTGCCACACACTGAAC) and reverse primer (5′-ATTGTGCGACTGATAATGTCGT); NIA1, forward primer (5′-GGTGCTGGTGTTTCTGGTCACT) and reverse primer (5′-GGGTCTGGTCGGGTGTTC); NIA2, forward primer (5′-CGAGACACACAACAGCAACGC) and reverse primer (5′-GCAACGGGACAGGGGTGA); NiR, forward primer (5′-CCGGTAGCCAGTTCTGCG) and reverse primer (5′-CCTATTCGTCCCCCGACGT); NRT2.1, forward primer (5′-GGCTAACGTGGATGGGAGTG) and reverse primer (5′-CGGCGTAGCAGCAGAGC); TPP-At1g78090, forward primer (5′-CCATCTTGGGTTTGAAAACCATC) and reverse primer (5′-GGCTGCTTCAAGAAGTTTCA); TPP-At5g65140, forward primer (5′-TAACATGTTCGAACGAATCATC) and reverse primer (5′-TTAGCCATTTTTTTCACTGTTCTT); and TPS-At1g23870, forward primer (5′-GAGTTTCAGAGTTCTGTCACTGTC) and reverse primer (5′-TTCGGCATTAGGGGTTTTGATGAT). The internal control gene was ubiquitin 10: forward primer, 5′-GTCCTCAGGCTCCGTGGTG; and reverse primer, 5′-TGCCATCCTCCAACTGCTTTC.

The LightCycler-FastStart DNA Master SYBR Green I Kit (Roche) was used for the PCR reactions. Each PCR reaction contains 2 μL of cDNA and 0.5 μm of each of the primers. The initial denaturing time was 10 min., followed by 35 PCR cycles consisting of 94°C for 0 s, 63°C for 5 s, and 72°C for 10 s. A melting curve was run after the PCR cycles followed by a cooling step. Quantification was performed with LightCycler Relative Quantification Software 1.0.

Target Preparation/Processing for GeneChip Analysis

Isolated total RNA samples were processed as recommended by Affymetrix (Affymetrix GeneChip Expression Analysis Technical Manual, Affymetrix, Inc.). In brief, total RNA was initially isolated using an RNeasy Plant Mini Kit (Qiagen). Total RNAs were quantified with a portion of the recovered total RNA adjusted to a final concentration of 1 μg μL^–1. All starting total RNA samples were quality assessed before beginning target preparation/processing steps by running out a small amount of each sample (typically 25–250 ng well^–1) onto a RNA Lab-On-A-Chip (Caliper Technologies Corp., Mountain View, CA) that was evaluated on an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA). Single-stranded, then double-stranded cDNA was synthesized from the poly(A⁺) mRNA present in the isolated total RNA (10 μg of total RNA starting material each sample reaction) using the SuperScript Double-Stranded cDNA Synthesis Kit (Invitrogen Corp., Carlsbad, CA) and poly (T)-nucleotide primers that contained a sequence recognized by T₇ RNA polymerase. A portion of the resulting double-stranded cDNA was used as a template to generate biotintagged cRNA from an in vitro transcription reaction (IVT), using the Bio-Array High-Yield RNA Transcript Labeling Kit (T₇; Enzo Diagnostics, Inc., Farmingdale, NY). Fifteen micrograms of the resulting biotin-tagged cRNA was fragmented to strands of 35 to 200 bases in length following prescribed protocols (Affymetrix GeneChip Expression Analysis Technical Manual). Subsequently, 10 μg of this fragmented target cRNA was hybridized at 45°C with rotation for 16 h (Affymetrix GeneChip Hybridization Oven 320) to probe sets present on an Affymetrix ATH1 array. The GeneChip arrays were washed and then stained (streptavidin-phycoerythrin) on an Affymetrix Fluidics Station 400, followed by scanning on a Hewlett-Packard GeneArray scanner (Hewlett-Packard, Palo Alto, CA). The results were quantified and analyzed using MicroArray Suite 5.0 software (Affymetrix, Inc.).