Transcriptomic profiling of major carbon and amino acid metabolism in the roots of Arabidopsis thaliana treated with various rhizotoxic ions

Authors


H. KOYAMA, Laboratory of Plant Cell Technology, Faculty of Applied Biological Sciences, Gifu University, 1-1, Yanagido, Gifu 501-1193, Japan. Email: koyama@gifu-u.ac.jp

Abstract

Alteration of metabolic processes is a common adaptive response of plants to various stress conditions and is likely to be under complex regulatory control. To understand the metabolic responses to rhizotoxic treatments in Arabidopsis thaliana, transcriptome profiles of major carbon and amino acid metabolic pathways were compared among aluminum (Al), copper (Cu) and cadmium (Cd) ion and NaCl treatments with a similar level of severity. All stress treatments induced genes encoding enzymes for synthesizing trehalose and polyamine, as well as tryptophan-synthesizing enzymes previously identified as critical for resistance to various stresses. Genes encoding enzymes critical for ascorbic acid and spermine synthesis had higher specificity to Cd and NaCl among the genes upregulated by each stress. Major isoforms of malic enzymes and glutamate decarboxylases were more specifically upregulated by the Al treatment than were other genes; these enzymes belong to cellular pH-regulating pathways, namely the biochemical pH stat pathway and the γ-amino butyric acid shunt. Characterization of the grouped genes with higher Cu specificity indicated that amino acid degradation and sugar starvation-like symptoms were enhanced by Cu treatment. Pathway analysis in the trehalose synthesis pathway accounted for the activation of the pathway and for the accumulation of trehalose by all stressors. These metabolic alterations might form part of the tolerance mechanisms of Arabidopsis roots against rhizotoxic ions.

Introduction

Many physiological and biochemical studies have revealed that alteration of metabolic processes by rhizotoxic treatments forms part of the resistance mechanisms to particular rhizotoxins, for example, alteration of organic acid metabolism by aluminum (Al) treatment (Ma et al. 2001) and enhancement of proline synthesis with NaCl treatment (Lutts et al. 2002). Similar altered metabolism in the γ-aminobutyric acid (GABA) shunt is proposed to form part of the resistance mechanisms to acidosis-related stress factors, such as hypoxia and anoxia (Bouche and Fromm 2004). Identification of alterations to metabolism at the molecular level would be useful in understanding plant responses to rhizotoxic ions. However, it is difficult to determine plant responses to rhizotoxic ions because plant metabolism is controlled by many enzymes and is regulated in a complex manner.

The application of omics-based technology is one reasonable approach to understand such complex responses (see the review by Rochfort 2005). Genome-wide transcriptome analysis using an oligo-DNA array has been developed in model plants, and the technology is now becoming available for application in a variety of crop plant species. When applying the technology to clarify the response to each stressor, it is important to consider “common” and “specific” responses to each stress. For example, enzymes relating to reactive oxygen species (ROS) scavenging are responsive to both biotic (e.g. pathogen infection) and abiotic (e.g. high light [Kimura et al. 2003] and heavy metals [Zhao et al. 2009]) stressors. In contrast, some responses are thought to be more specific to particular stressors, such as the induction of heat-shock proteins with heat treatment (Schlesinger 1990). This situation is similar to the response of plants to rhizotoxic ions. For example, Al induces genes encoding organic acid transporters such as AtALMT1 (Arabidopsis thaliana Al-activated malate transporter; Kobayashi et al. 2007), which shows high specificity to Al. In contrast, Al induces genes encoding proteins for protecting cells from ROS that are commonly induced by a variety of other rhizotoxins (Hasegawa et al. 2000; Heidenreich et al. 2001). Identification of these “specific” and “common” genes is important for understanding the complex nature of metabolic alteration by transcriptome.

Recently, we showed that a comparison of transcriptome among different rhizotoxins is useful to distinguish specific and general responsive genes among highly upregulated genes in Arabidopsis thaliana, that is, genes with a fold change (FC) in the upper 2.5% of all active genes (Zhao et al. 2009). Our analysis identified specific gene groups that involve many known genes regulating resistance to particular rhizotoxins, such as AtALMT1 (Hoekenga et al. 2006) in the Al-specific group, DREB (dehydration responsive element binding protein) transcription factors (see the review by Shinozaki and Yamaguchi-Shinozaki 2007) in the NaCl-specific group and metal-thionein (Guo et al. 2008) in the copper (Cu) specific group, together with unidentified genes reflecting characteristics of each stressor, such as HSP in cadmium (Cd) treatment (Zhao et al. 2009). A similar approach would be useful to understand the complex nature of plant metabolic responses to rhizotoxic ions.

In contrast, one approach that can be used to understand the complex responses of altered metabolism to a particular stress is the application of bioinformatics techniques, such as pathway analysis tools for visualizing transcriptomic changes in metabolic processes (e.g. MAPMAN [http://gabi.rzpd.de/projects/MapMan/] and KaPPA-View [The Kazusa Plant Metabolic Pathway Viewer; http://kpv.kazusa.or.jp/kappa-view/]. A combination of comparative microarray and pathway analyses would be useful for characterizing metabolic alterations during rhizotoxic treatments and would contribute to our understanding of plant responses to rhizotoxic ions.

In the present study, we used a comparative transcriptomic approach to characterize the adaptation and activation of major carbohydrate and amino acid metabolism in Arabidopsis roots under rhizotoxic ion treatments. This approach identified distinct gene sets that were induced by each stressor. These groups contained several genes encoding enzymes contributing resistance to all stressors or to a particular stressor (e.g. trehalose synthase in the gene group was induced by all stressors). Using publicly available gene ontological information, these gene sets were characterized to identify the major metabolic responses in Arabidopsis roots to rhizotoxic ions.

Materials and methods

Plant material and growing conditions

Arabidopsis thaliana accession Columbia (Col-4; N933) was obtained from the Nottingham Arabidopsis Stock Center (Nottingham, UK). Plants were grown hydroponically in modified MGRL solution (Fujiwara et al. 1992) according to Kobayashi et al. (2007) (35 μmol L−1 sodium phosphate, 30 μmol L−1 MgSO4, 40 μmol L−1 Ca(NO3)2, 60 μmol L−1 KNO3, 1.34 μmol L−1 Na2-ethylenediaminetetraacetic acid [EDTA], 172 nmol L−1 FeSO4, 206 nmol L−1 MnSO4, 600 nmol L−1 H3BO3, 20 nmol L−1 ZnSO4, 0.48 nmol L−1 (NH4)6Mo7O24, 2.6 nμmol L−1 CoCl2 and 20 nmol L−1 CuSO4) with extra CaCl2 to give a final concentration of 200 μmol L−1 Ca (pH 5.6). The culture solution was renewed every 2 days and the seedlings were grown for 10 days in a controlled environment (30 μmol E m−2 s−1; 12 h day/12 h night cycle; 23–25 °C).

Rhizotoxin treatments

Hydroponically grown Arabidopsis seedlings were transferred to modified MGRL solution containing either 25 μmol L−1 AlCl3 (pH 4.9), 15 μmol L−1 CdCl2, 1.6 μmol L−1 CuSO4 or 50 μmol L−1 NaCl. These concentrations induced 90% inhibition of root growth after 1-week of culture and triggered the expression of responsive genes (Zhao et al. 2009). After 24-h incubation in these solutions under continuous illumination (30 μmol E m−2 s−1) at 23–25 °C, the roots of the seedlings were harvested and used for trehalose assay and RNA isolation. Root samples were collected in 1.5 mL microcentrifuge tubes and stored at −80 °C until use. Seedlings grown in a non-toxic solution (pH 5.0, no rhizotoxins) were used as a control. The treatment conditions were identical to previously reported genome-wide microarray analyses (Zhao et al. 2009).

RNA isolation

Total RNA was isolated following the method of Suzuki et al. (2004). In brief, root tissue was ground to a fine powder in liquid nitrogen and then mixed with 5–10 volumes of extraction buffer (100 mmol L−1 Tris–HCl [pH 9.5], 10 mmol L−1 EDTA [pH 8.0], 2% [w/v] lithium dodecyl sulfate, 0.6 mol L−1 NaCl and 0.4 mol L−1 trisodium citrate, 5% [v/v]2-mercaptoethanol]). The RNA was then purified with chloroform : isoamylalcohol (CIA; 24:1)-phenol-lithium chloride extraction followed by isopropanol precipitation. For the microarray analysis, three independent root samples derived from approximately 200 seedlings were used for RNA isolation.

Transcriptomic profiling of the major carbohydrate and related pathways in the roots

Genes involved in the major pathways related to carbohydrate and amino acid metabolism in Arabidopsis were identified by pathway analysis with KaPPA-View, which is a web-based tool that can be used for the analysis of transcriptome and metabolome in plant metabolic pathways. A total of 534 genes included in the selected metabolic maps (Table 1) of KaPPA-View (33 maps, 540 genes), and which exist on the Agilent Arabidopsis ver. 2 oligo-DNA chip, were used for the following analyses. The microarray dataset was identical to that used in a previous comparative transcriptomic analysis (Zhao et al. 2009), which consisted of three biologically independent replications of a competitive array (Al, Cd, Cu and NaCl treatments; see the above rhizotoxic treatments). The raw data are available from the ArrayExpress database under the accession number E-MEXP-1907 (transcription profiling of Al, Cu, Cd and NaCl stresses; http://www.ebi.ac.uk/arrayexpress). The original datasets were obtained using the Arabidopsis ver. 2 oligo microarrays followed by standard experimental protocols for the Agilent microarray system as described previously (Zhao et al. 2009). Data points based on less than three measurements and spots with low fluorescence intensity (i.e. <500) were excluded from the analyses, leaving 295 genes for use. Upregulated genes were defined as those in which the treatment/control (FC) value was >1.2 (< 0.05, t-test from FC = 1 or FC >1.2 with three replicates). Downregulated genes were defined as those in which FC <1/1.2 (< 0.05, t-test from FC = 1 or FC <1/1.2 with three replicates). Cluster analyses were carried out using the cluster software (Eisen et al. 1998) (available at http://rana.lbl.gov/eisen/) to identify genes with greater specificity to a particular stressor using “relative fold change” ([RFC] the mean FC with other ion treatments divided by the mean FC with a particular stressor) as described in a previous study (Zhao et al. 2009). The RFC was calculated for genes upregulated and downregulated in response to each stressor, but which were not identified as commonly responsive genes to all stressors using a Venn diagram approach (Fig. 1). The RFCs were analyzed using the cluster software and were grouped with hierarchical clustering using the average linkage clustering method with normalization of the mean RFC for each treatment. The output data were visualized using a web available software TREEVIEW (http://rana.lbl.gov.eisen/). Genes were manually subgrouped by the formed cluster and its color and specificity was assessed by comparison of the subgroups using a Sheffe test (< 0.05).

Table 1.   List of the map ID and designation of the metabolic pathways at KaPPA-View 2
Map ID number (designation)
  1. A comparative microarray approach was carried out with genes belonging to the maps in the list. dTDP, deoxythymidine diphosphate; GDP, guanosine diphosphate; TCA, tricarboxylic acid; UDP, uridine diphosphate.

CO2 fixation and central carbohydrate metabolism
 Ath21100 (CO2 fixation and central carbohydrate metabolism)
 Ath00118 (Glycolate pathway) 31 genes
 Ath00120 (Phosphoenolpyruvate and pyruvate metabolism) 73 genes
 Ath00150 (TCA cycle) 42 genes
 Ath00152 (Glyoxylate cycle) 19 genes
 Ath00090 (Glycerol metabolism) 6 genes
Mono-, di- and oligosaccharide metabolism
 Ath21200 (Mono-, di- and oligosaccharide metabolism)
 Ath00020 (Hexose phosphate pool) 27 genes
 Ath00022 (Sucrose metabolism) 43 genes
 Ath00099 (Trehalose metabolism) 20 genes
 Ath00134 (UDP-sugar metabolism) 20 genes
 Ath00145 (GDP-sugar and ascorbate metabolism) 6 genes
 Ath00025 (dTDP-sugar biosynthesis) 10 genes
 Ath00443 (Inositol phosphate metabolism) 8 genes
Miscellaneous carbohydrate metabolism
 Ath21900 (Miscellaneous carbohydrate metabolism)
 Ath00036 (Pyridoxal 5-phosphate metabolism) 8 genes
 Ath12000 (Amino acid, nucleic acid and nitrogen-containing derivative metabolism)
Aspartate and related amino acid metabolism
 Ath22100 (Aspartate and related amino acid metabolism)
 Ath00011 (Aspartate and asparagine metabolism) 13 genes
 Ath00010 (Lysine, threonine and methionine biosynthesis) 17 genes
 Ath00079 (Lysine degradation) 13 genes
 Ath00081 (Methionine metabolism) 5 genes
 Ath00085 (Threonine and methylglyoxal metabolism) 2 genes
Glutamate and related amino acid metabolism
 Ath22200 (Glutamate and related amino acid metabolism)
 Ath00006 (Glutamate and Glutamine metabolism) 26 genes
 Ath00013 (Arginine and proline metabolism) 25 genes
 Ath00082 (Proline and 4-hydroxyproline metabolism) 1 gene
Leucine, valine, isoleucine and alanine metabolism
 Ath22300 (Leucine, valine, isoleucine and alanine metabolism)
 Ath00009 (Leucine, valine, isoleucine and alanine biosynthesis) 31 genes
 Ath00077 (Leucine, valine and isoleucine degradation) 30 genes
Aromatic amino acid metabolism
 Ath22400 (Aromatic amino acid metabolism)
 Ath00017 (Aromatic amino acid biosynthesis) 57 genes
 Ath00086 (Tryptophan metabolism) 7 genes
 Ath00087 (Tyrosine metabolism) 11 genes
Serine, Glycine and cystein metabolism
 Ath22500 (Serine, glycine and cysteine metabolism)
 Ath00008 (Serine and glycine metabolism) 22 genes
 Ath00141 (Sulfur and cysteine metabolism) 23 genes
 Ath00076 (Glycine degradation) 2 genes
 Ath00014 (Homocysteine and cysteine interconversion) 3 genes
 Ath00072 (L-Cysteine degradation) 1 gene
 Ath00031 (Glutathione biosynthesis) 2 genes
Histidine and nucleic acid metabolism
 Ath22600 (Histidine and nucleic acid metabolism)
 Ath00154 (Histidine metabolism) 10 genes
Figure 1.

 Identification of the genes responsive to various rhizotoxic stressors in the roots of Arabidopsis thaliana and that are involved in major carbon and amino acid metabolism. Venn diagrams showing the classification of (A) upregulated (FC [treatment/control] was >1.2 and < 0.05 in t-test from FC = 1, or >1.2 in all three replications) and (B) downregulated genes (FC [treatment/control] was <1/1.2 and < 0.05 in t-test from FC = 1, or <1/1.2 in all three replications) in each rhizotoxic treatment. The roots were treated for 24 h with rhizotoxic solutions containing AlCl3 (25 μmol L−1), NaCl (50 mmol L−1), CdCl2 (15 μmol L−1) or CuSO4 (1.6 μmol L−1); these solutions resulted in similar levels of growth inhibition in the roots (I90).

Semi-quantitative reverse transcription polymerase chain reaction

Semi-quantitative reverse transcription polymerase chain reaction (RT-PCR) was carried out according to the method described by Kihara et al. (2003) using specific PCR primers under real-time conditions, where amplicons were logarithmically amplified. The sequence of the specific primers was as follows (the fragment size amplified and the number of PCR cycles carried out with each primer pair are in parentheses): At5g51460 (TPPA) forward: 5′-CATATCCAAAACTACGGCTAACACA-3′, reverse: 5′-GAGACTAATATACCATAACCGTGGT-3′ (220 bp, 25 cycles); At1g23870 (TPS9) forward: 5′-CTCGCAAATGAGCCTGTAGTCGTC-3′, reverse: 5′-CAAACGCCACTTGCGTGTGCAATGA-3′ (388 bp, 25 cycles); At2g18700 (TPS11) forward: 5′-TGTCAAGAGAGGCCAGCACATAG-3′, reverse: 5′-AAGCATCTTGATAACACTTGGGG-3′ (286 bp, 23 cycles); At1g35910 forward: 5′-TCCCGGACTTTACTATGCAGGTAGC-3′, reverse: 5′-TACTCACTCAATATCGATGTCACGT-3′ (279 bp, 23 cycles); UBQ1 forward: 5′-TCGTAAGTACAATCAGGATAAGATG-3′, reverse: 5′-CACTGAAACAAGAAAAACAAACCCT-3′ (215 bp, 20 cycles).

Quantification of trehalose

Sample preparation followed the method of Leyman et al. (2004) with minor modifications to make it suitable for Arabidopsis root samples. Frozen root samples (50 mg aliquots) in microcentrifuge tubes were ground to a powder in liquid N2 with a plastic pestle. Ice-cold water (300 μL) was added to the microcentrifuge tube and then immediately floated on boiling water for 20 min. Recovery of trehalose by this treatment exceeded 97%. The trehalose was enzymatically quantified using the method of Lee and Goldberg (1998). In brief, the trehalose in the sample was converted to glucose using porcine kidney trehalase (Sigma, Tokyo, Japan). Under these conditions, trehalose was converted to glucose (>97% conversion for the trehalose standard) and thus the glucose concentration after the trehalase reaction was an indicator of the trehalose content in the sample. The glucose was quantified using a Glucose-CII-Test kit (Wako, Osaka, Japan), either with or without trehalase treatments. Three independent root samples were used for each trehalose assay.

Results

Identification of upregulated and downregulated genes by a comparative microarray approach

Arabidopsis roots were incubated in solutions containing rhizotoxic minerals (Al, Cd, Cu) and salt (NaCl) with similar levels of severity (i.e. levels that gave approximately 90% growth inhibition) (Zhao et al. 2009). A total of 534 genes belonging to the major carbohydrate and amino acid metabolisms, which were defined in the pathway maps of KaPPA-View, were filtered by the intensity of the fluorescence (<500) of the spots in a microarray, and then 295 genes were used in a comparative microarray approach. Upregulating genes (mean of FC > 1.2 with P < 0.05 in t-test from 1, or FC > 1.2 in three replications) and downregulating genes (mean of FC < 1/1.2 [= 0.833] with P < 0.05 in t-test from 1, or FC < 1/1.2 in three replications) were grouped using a Venn diagram (Fig. 1). Fifteen and three genes were identified as being upregulated or downregulated, respectively, by all stressors (Fig. 1A,B). In addition, 48, 71, 40 and 39 genes were identified as being upregulated and 30, 120, 78 and 22 genes were identified as being downregulated in response to Al, Cu, Cd and NaCl, respectively (Fig. 1A,B). The gene groups are listed in Supplementary file S1. Because some of genes in these gene groups responded to multiple ions (e.g. Al and Cu), the specificity of the response of the genes to a particular stressor was examined later (Figs 2,3).

Figure 2.

 Dendrograms from the hierarchical cluster analysis of upregulated genes by treatment with (A) Al ions, (B) NaCl, (C) Cd ions and (D) Cu ions. The relative fold change was analyzed. Genes upregulated by all stressors are not included. The fold change ratios of the genes are indicated by the different colors. The magnified gene groups have a higher specificity to each stressor than the other subgroups of genes (see Supplementary file S3).

Characteristics of the genes upregulated or downregulated by all stressors

The upregulated group contained five genes (33%) involved in “aromatic amino acid biosynthesis”, comprising approximately 10% of the total genes that were tested (Table 2). The upregulated genes included two genes belonging to the synthetic pathways of two stress-responsive metabolites, namely trehalose and polyamine. One of the putative trehalose-6-phosphate phosphatases that catalyzes a direct step in trehalose synthesis and the gene encoding arginine decarboxylase 2 (ADC2), a rate-limiting enzyme for polyamine synthesis, were identified among the commonly upregulated genes (Supplementary file S1). The A-methyl tryptophan resistant 1 (AMT1) gene, encoding a rate-limiting enzyme for tryptophan synthesis, and two other genes encoding tryptophan-synthesizing enzymes (tryptophan synthase α subunit 1 [TSA1] and phosphoribosylanthranilate transferase [PAT1] annotated as tryptophan biosynthesizing enzyme) were identified in the commonly upregulated gene group, whereas those were identified as critical genes for resistance to various stress. Genes encoding two of enzymes in Uridine diphosphate (UDP) sugar metabolism and one of isoforms of alcohol dehydrogenase were belong to the generally downregulated genes group. (Supplementary file S1). One of the major enzymes catalyzing keto acid to amino acid, namely alanine amino-transferase 1 (ALAAT1), belonged to the generally upregulated genes group.

Table 2.   Classification of the genes responsive to Al, Cd and Cu ions or NaCl in primary carbohydrate metabolism and amino acid metabolism
Metabolic pathwayNo. genes
Genes on the map (s)Gene category
Map IDDesignationAllAnalyzed (intensity >500)Responsive to All ionsGenes have higher specificity to each ion
AlNaClCdCu
UD§UDUDUDUD
(Total)
534295153103226943327
  1. Metabolic pathway was designated by the KaPPA-View database. Upregulated. §Downregulated. Genes were grouped by Venn diagram (Fig. 1) and cluster analysis (Figs 2,3). ** and * indicate that the value in each group is significantly larger or smaller than the value of the analyzed gene group (χ2-test, < 0.05).

Ath00006Glutamate and glutamine metabolism2613002122**0123
Ath00008Serine and glycine metabolism22120010201000
Ath00009Leucine, valine, isoleucine and alanine biosynthesis31161010211123
Ath00010Lysine, threonine and methionine biosynthesis1790000000002
Ath00011Aspartate and asparagine metabolism1390000210113
Ath00013Arginine and proline metabolism25162000211002
Ath00014Homocysteine and cysteine interconversion310*0*0*0*00*0*0*01
Ath00017Aromatic amino acid biosynthesis55295**000200130
Ath00020Hexose phosphate pool27170000100001
Ath00022Sucrose metabolism42151000001040
Ath00025dTDP-sugar biosynthesis1080010000000
Ath00031Glutathione biosynthesis200000000000
Ath00036Pyridoxal 5-phosphate metabolism8400*00*000010
Ath00072L-Cysteine degradation110*0*0*0*00*0*0*00
Ath00076Glycine degradation210*0*0*0*00*0*0*00
Ath00077Leucine, valine and isoleucine degradation3016000011007**1
Ath00079Lysine degradation12600*00000020
Ath00081Methionine metabolism510*0*0*0*00*00*00
Ath00082Proline and 4-hydroxyproline metabolism100000000000
Ath00085Threonine and methylglyoxal metabolism210*00*0*10*0*0*01
Ath00086Tryptophan metabolism6300*00*000010
Ath00087Tyrosine metabolism10400*00*200110
Ath00090Glycerol metabolism6400*00*100100
Ath00099Trehalose metabolism2071000100030
Ath00118Glycolate pathway31151011111021
Ath00120Phosphoenolpyruvate and pyruvate metabolism7345114**0211058
Ath00134UDP-sugar metabolism201512**02**002011
Ath00141Sulfur and cysteine metabolism23121010011011
Ath00145GDP-sugar and ascorbate metabolism6400*00001000
Ath00150TCA cycle41291000600010
Ath00152Glyoxylate cycle19131000300010
Ath00154Histidine metabolism1070000000001
Ath00443Inositol phosphate metabolism860000000002

Characteristics of the stressor-specific gene groups among the responsive genes

Some of the upregulated and downregulated genes were differentially responsive to specific stressors. These genes reflected specific characteristics of toxicity and resistant mechanisms of each stressor. Using RFC, both upregulated and downregulated genes, and those that were not grouped as commonly responsive genes (Fig. 1), were grouped by cluster analysis. The cluster analysis indicated that the gene groups possessing larger RFCs were more specifically responsive to individual stressors than the other gene groups. These gene groups are shown in Figures 2 and 3 and are described more fully in the following sections.

Figure 3.

 Dendrograms from the hierarchical cluster analysis of downregulated genes by treatment with (A) Al ions, (B) NaCl, (C) Cd ions and (D) Cu ions. The relative fold change was analyzed. Genes downregulated by all stressors are not included. The fold change ratios of the genes are indicated by the different colors. The magnified gene groups have a higher specificity to each stressor than the other subgroups of genes (see Supplementary file S3).

Specifically responsive genes in the Al-responsive gene group

A total of 10 genes and three genes were identified as more specific than others among the upregulated and downregulated genes, respectively, to Al (Figs 2A,3A). Approximately half of the specific genes upregulated by Al were involved in phosphoenolpyruvate metabolism. Genes encoding two isoforms of NADP-ME (malic enzyme 1 and 2) and a putative pyruvate carboxykinase (PCK1), which were related to the production of pyruvate from malate and oxalate, were placed in this group. A cytosolic isoform of glutamate synthase (GLN1;4) and glutamate decarboxylase 1 (GAD1), whose substrate is glutamate, were also found in this gene group. Some of these genes (i.e. ME1, ME2 and GAD1) have been identified as being regulated by STOP1 (Cys2-Hys2 zinc-finger protein regulating multiple genes for the protection of proton and aluminum rhizotoxicity) (Sawaki et al. 2009). A gene for an UDP-glucose epimerase (i.e. defective use of root [DUR]), which catalyses a direct step in indole acetic acid (IAA) degradation, was identified in the downregulated Al-specific gene group.

Specifically responsive genes in the NaCl-responsive gene group

A total of 22 and six specific NaCl-responsive genes were upregulated and downregulated, respectively (Figs 2B,3B). The upregulated gene group contained five genes belonging to the map for the TCA cycle (designated by KaPPA-View and not limited to the mitochondrial TCA cycle), including mitochondrial citrate synthase (AtCS) and NAD-dependent isocitrate dehydrogenase (IDH2). Genes encoding cytosolic aconitase and mitochondrial malate dehydrogenase were also identified in this gene group. In addition, genes encoding one of the trehalose-synthesizing enzymes (Arabidopsis thaliana trehalose-6-phosphate phosphatase [ATTPA]) and spermine synthase (a paralog of spermidine synthase [ATSPDS3]) belonged to this gene group. The gene cytosolic glutamine synthase (GLN1;4) and a homologue of spermine synthase (ACL5) were among the downregulated NaCl-responsive genes.

Specifically responsive genes in the Cd-responsive gene group

A comparative cluster analysis identified nine genes as being more specifically upregulated by the Cd treatment than the other Cd-responsive genes (Fig. 2C). One gene encoding an enzyme belonging to the GDP-sugar and ascorbate metabolic pathway, namely cytokinesis defective 1 (CYT1), is involved in vitamin C synthesis and a dysfunctional mutation causes phenotypes sensitive to various stress factors (e.g. UV-B, ozone and NH4+ toxocity) as a result of vitamin C deficiency. A gene encoding ALF1 (aberrant lateral root formation 1) relating to IAA response, which belongs to the aromatic amino acid pathway, was one of four downregulated genes (Supplementary file S1) that was more specifically responsive to Cd than to the other stressors.

Specifically responsive genes in the Cu-responsive gene group

A total of 33 and 27 upregulated and downregulated genes, respectively, were more specifically responsive to Cu than to the other stressors (Figs 2D,3D). Six genes encoding enzymes in the leucine, valine and isoleucine degradation pathways were upregulated, whereas another six genes encoding enzymes belonging to the phosphoenolpyruvate and pyruvate metabolic pathways were downregulated (Fig. 3D). Three homologues of trehalose phosphate synthase and genes encoding tryptophan synthase subunits (α and β) were identified in the specific gene cluster as being upregulated. Two genes inducible by dark treatment in leaves (e.g. dark inducible [DIN]) as a result of sugar starvation were also identified in this gene group.

Effect of rhizotoxic treatments on trehalose concentration in the roots

Genes encoding a trehalose-synthesizing enzyme, trehalose-6-phosphate phosphatase, were identified in the gene group upregulated by all stressors and by certain individual stressors (NaCl and Cu) (Supplementary file S1). First, the FC of all isoforms was analyzed with KaPPA-View to view overall changes in the trehalose metabolic pathway. As shown in Fig. 4A, the trehalose synthetic pathway was slightly increased by all treatments, as measured by the mean FC of all isoforms, which was shown by the color of the lines in the pathway map. In contrast, trehalose degradation was also upregulated in the NaCl (slightly), Cd and Cu treatments, but remained stable in the Al treatment. The first step in trehalose synthesis, which is catalyzed by trehalose-6-phosphate synthase, was downregulated in the NaCl treatment, whereas it was stable (Al) or upregulated in the Cu and Cd treatments (Fig. 4A; Supplementary file S2). Second, the same analysis was carried out after eliminating genes with a lower expression level (i.e. intensity of spots on the DNA chip was <500) and that were not used in other experiments in the present study. In this case, upregulation of the synthetic pathway was more enhanced in all stress treatments (Fig. 4B). Both the Cd and Cu treatments upregulated the genes involved in the first process of synthesis, whereas Al and NaCl downregulated these genes. The expression level of the upregulated genes was confirmed by semi-quantitative RT-PCR (Fig. 5). Under these conditions, the trehalose contents of the roots were significantly increased in all stress treatments 24 h after treatment (Fig. 6).

Figure 4.

 Changes in transcript levels of the genes in the trehalose synthesis pathway in the roots of Arabidopsis thaliana 24 h after exposure to rhizotoxic solutions containing either AlCl3 (25 μmol L−1), NaCl (50 mmol L−1), CdCl2 (15 μmol L−1) or CuSO4 (1.6 μmol L−1), which resulted in similar levels of growth inhibition in the roots (I90). Transcriptome data (no filtering by intensity of microarray spot: (A), intensity of the spot >500; (B) of the trehalose metabolism pathway were analyzed by Kappa-view2 (map ID; Ath00099 from KaPPA-View 2 (see raw data in Supplementary file S2). The boxes correspond to isoforms and the color indicates the fold change (FC) (treatment/control). The color of the arrows indicates the mean FC among the detected genes, which was automatically calculated with KaPPA-View 2. UDP, Uridine diphosphate.

Figure 5.

 Semi-quantitative reverse transcription polymerase chain reaction (RT-PCR) image of the major genes in the trehalose metabolic pathway. Transcripts of genes of higher intensity (>500) were amplified by semi-quantitative RT-PCR using specific primers. Amplicons were visualized with SYBR Green I and the image was captured using Typhoon 9410. The expression patterns judged by semi-quantitative RT-PCR were similar to those examined by microarray analysis. Cont, control.

Figure 6.

 Trehalose contents in the roots of Arabidopsis thaliana after exposure to rhizotoxic treatments. The roots were exposed for 24 h to rhizotoxic solutions containing either AlCl3 (25 μmol L−1), NaCl (50 mmol L−1), CdCl2 (15 μmol L−1) or CuSO4 (1.6 μmol L−1), which resulted in similar levels of growth inhibition in the roots (I90). Asterisks indicate a significant difference from the control (t-test; < 0.05, = 3). Cont, control.

Discussion

Using a comparative transcriptomic approach for major carbon and amino acid metabolism, we could identify the gene groups that had higher specificity to a particular stressor, and that were related to resistance and toxicity mechanisms for each rhizotoxic ion (Figs 2,3). We suggest that a simple comparative approach using different transcriptome data would be useful to understand complex biological events that occur in both global (genome-wide) and local (e.g. specific biochemical pathways, biological processes) biological systems.

Using RFC, we determined that some genes had a higher specificity to a particular treatment (Figs 2,3; supplementary file S3). A direct role for most of these genes in the response to a particular stressor has not yet been identified, but the molecular and physiological functions of several genes belonging to each gene group have been identified. These “relatively” specific genes may reflect the toxicity of the stressors or resistance mechanisms in response to the stressors. For example, all stressors induced expression of TSA1 and AMT1, which are involved in tryptophan synthesis and are also related to ethylene–auxin signaling (Ouyang et al. 2000; Voll et al. 2004). Thus, these genes were responsive to a variety of stressors (e.g. Ascencio-Ibanez et al. 2008; Bray 2002). In addition, these enzymes are involved in defense mechanisms against both biotic and abiotic stressors (e.g. innate immunity of Arabidopsis; Clay et al. 2009). Enhancement of the trehalose synthetic pathways also reflects activation of general defense mechanisms by all rhizotoxic stressors (Fig. 4) because trehalose can protect tissues from various stressors, including high salt (Garg et al. 2002) and ROS (Benaroudj et al. 2001). This “common response” to ROS might also account for previous quantitative trait locus (QTL) studies that have identified pleiotropic loci controlling ROS, Cd, Cu and Al tolerance in Arabidopsis thaliana (Ikka et al. 2008; Tazib et al. 2009).

The function and physiological roles of the induced genes are suggestive of different toxicity mechanisms for rhizotoxic ions. For example, the genes with the highest specificity to Cu (Fig. 2D) included a large number of genes identified as dark-inducible genes (i.e. DIN genes), which are induced by dark as a result of a shortage of sugar. This was concomitant with activation of pyruvate-derived amino acid degradation (i.e. leucine, valine and isoleucine) and downregulation of several isoforms of pyruvate decarboxylases converting pyruvate to acetyl coenzyme A (Fig. 3D; Supplementary file S1). These findings suggest that Cu induces a greater sugar deficiency than other rhizotoxic ions. In contrast, several genes encoding enzymes of the TCA bypass were expressed in the Al-ion treatment. Genes encoding one of the major isoforms of glutamate decarboxylase (GAD1) and two isoforms of malic enzymes (e.g. ME1 and ME2) were identified in the specifically upregulated gene group (Fig. 2A; Supplementary file S1). These enzymes are involved in cellular pH-regulating pathways (GABA shunt [Bown and Shelp 1997] and the biochemical pH stat pathway [Sakano 1998]) and are induced by acidosis under hypoxia and anoxia (Crawford et al. 1994). This indicates that Al may induce acidosis in Arabidopsis roots as in the roots of wheat (Lindberg and Strid 1997). Furthermore, CYT1, which is critical for ROS-related stress resistance via vitamin C synthesis (Colville and Smirnoff 2008), was identified in the specific gene group upregulated by Cd, and ATSPDS3, which encodes spermine synthase (the precursor of polyamine) and which is critical for high salt resistance (Yamaguchi et al. 2007), was identified among the specific NaCl-responsive upregulated genes (Fig. 2A; Supplementary file S1). Alterations that account for tolerant mechanisms might be regulated by a gene-expression network related to a particular stressor.

Because one of the homologs encoding the enzyme directly involved in trehalose synthesis was upregulated by all stressors (Supplementary file S1), we estimated the change in trehalose synthesis during rhizotoxin treatment using KaPPA-View (Tokimatsu et al. 2005) to integrate the transcriptome and metabolite(s) data on the pathway maps (Fig. 4). This approach revealed that the rhizotoxic ion treatments triggered trehalose accumulation by modification of the trehalose synthesis/degradation balance. Further analysis using additional “metabolite” data is necessary to evaluate this hypothesis, but our data suggest that trehalose accumulation is involved in the defense response of the roots to various rhizotoxic ions. In contrast, some of the genes specifically induced by each stressor would be regulated by the same signal transduction pathway. For example, ME1, ME2 and GAD1 showed higher specificity to Al (Fig. 2A), but are repressed in the Arabidopsis mutant carrying the dysfunctional missense mutation in the Cys2–His2 domain of STOP1 zinc-finger transcription factor (Iuchi et al. 2007; Sawaki et al. 2009). In contrast, we only focused on commonly upregulated or downregulated genes by all stressors and the genes relatively specific to particular stressor than to other stressors (Figs 1–3). This indicates that other genes would be cumulatively responsive to some stressor combinations. This would reflect cross talk of signal transduction pathways that overlap multiple stressors (e.g. overlap of the NaCl response to those of ABA and cold stresses; Shinozaki and Yamaguchi-Shinozaki 2007). Bioinformatic approaches, such as the prediction of cis-elements in the promoter region and coexpression gene network analysis, might improve our understanding of the regulatory mechanism(s) of the metabolic processes under rhizotoxic stress. Further research is needed to evaluate this possibility.

In conclusion, a simple comparative microarray approach is useful to determine specific responses to particular rhizotoxins and common responses to a variety of rhizotoxins at a genome-wide level (Zhao et al. 2009), and this approach is also useful for grouping genes involved in local events (i.e. a limited number of genes involved in selected metabolic pathways). Similar approaches have been used to characterize common and specific factors in the signal transduction pathways of salt, cold and abscisic acid stress (e.g. Kant et al. 2007). These findings indicate that the procedure is applicable for identifying specific gene responses to a variety of stressors. Recent advances in transcriptome and biological information in other plants (e.g. metabolic pathway information in rice and other species; Jaiswal et al. 2006) may allow us to conduct comparative genomics of stress-responsive genes among a diverse range of plant species.

Acknowledgments

This work was supported by the Japan Society for the Promotion of Science and the Ministry of Economy for HK, Trade and Industry, Japan for HK and DS. We thank the Nottingham Arabidopsis Stock Center for providing Arabidopsis seeds.

Ancillary