Comparative Genome-Wide Transcriptional Analysis of Al-Responsive Genes Reveals Novel Al Tolerance Mechanisms in Rice

Rice (Oryza sativa) is the most aluminum (Al)-tolerant crop among small-grain cereals, but the mechanism underlying its high Al resistance is still not well understood. To understand the mechanisms underlying high Al-tolerance, we performed a comparative genome-wide transcriptional analysis by comparing expression profiling between the Al-tolerance cultivar (Koshihikari) and an Al-sensitive mutant star1 (SENSITIVE TO AL RHIZOTOXICITY 1) in both the root tips and the basal roots. Exposure to 20 µM AlCl3 for 6 h resulted in up-regulation (higher than 3-fold) of 213 and 2015 genes including 185 common genes in the root tips of wild-type and the mutant, respectively. On the other hand, in the basal root, genes up-regulated by Al were 126 and 2419 including 76 common genes in the wild-type and the mutant, respectively. These results indicate that Al-response genes are not only restricted to the root tips, but also in the basal root region. Analysis with genes up- or down-regulated only in the wild-type reveals that there are other mechanisms for Al-tolerance except for a known transcription factor ART1-regulated one in rice. These mechanisms are related to nitrogen assimilation, secondary metabolite synthesis, cell-wall synthesis and ethylene synthesis. Although the exact roles of these putative tolerance genes remain to be examined, our data provide a platform for further work on Al-tolerance in rice.


Introduction
Aluminum (Al) toxicity is a major factor limiting crop production on acid soils, which comprise approximately 40% of the world's arable soils and up to 70% of potentially arable land [1]. At soil pH below 5.0, toxic forms of Al (mainly Al 3+ ) are solubilized into the soil solution, which inhibit root growth and function, consequently reducing crop yields [2,3]. However, there is a great variation for the ability to withstand Al-toxicity between plant species and cultivars within a species. To survive on acidic soils, some plant species or cultivars have evolved mechanisms to tolerate high levels of toxic Al. Many mechanisms for both Altolerance and -toxicity have been proposed [3].
Rice (Oryza sativa) is the most Al-tolerant crop among smallgrain cereals [4]. A number of quantitative trait loci (QTLs) for Altolerance have been identified in rice by using different populations [5], but responsible QTL genes have not been isolated. Recently, through genome-wide association analysis and QTL mapping, 48 loci associated with Al 3+ tolerance have been identified [6] in rice. On the other hand, mutant approaches have revealed an ART1-regualted Al-tolerance mechanism in rice [7]. ART1 (AL 3+ RESISTANCE TRANSCRIPTION FACTOR 1) is a Cys2His2-type zinc-finger transcription factor [8]. ART1 is constitutively expressed in the roots and its expression is not induced by Al 3+ treatment. ART1 regulates the expression of at least 31 genes with a cis-element [GGN(T/g/a/C)V(C/A/g)S(C/ G)] (ART1-binding affinity of nucleotides with small characters is weaker than those with large characters) [9]. Among them, only six genes have been functionally characterized. OsSTAR1 and OsSTAR2 (SENSITIVE TO ALUMINUM RHIZOTOXICITY 1 & 2) encode a ATP-binding domain and a transmembrane domain, respectively, of a bacterial-type ATP binding cassette (ABC) transporter, which transports UDP-glucose [10]. The complex is implicated in cell wall modification [10]. OsFRDL4 (FERRIC REDUCTASE DEFECTIVE3-LIKE 4) encodes a citrate transporter, which secretes citrate from the roots to chelate Al in the rhizosphere [11]. On the other hand, OsNrat1 (NRAMP ALUMI-NUM TRANSPORTER 1) encodes an Al transporter localized at the plasma membrane, which transports Al into the cells [12], while OsALS1 (ALUMINUM SENSITIVE 1) encodes a tonoplastlocalized transporter for Al, which sequestrates Al into the vacuoles [13]. Recently, up-regulation of a Mg transporter, OsMGT1 (MAGNESIUM TRANSPORTER 1), is reported to be required for conferring Al-tolerance in rice [14]. All of these genes are specifically induced by Al and knockout of either gene results in decreased Al-tolerance, indicating their important roles in Al-tolerance. However, the mechanisms underlying high Altolerance in rice are not fully understood at the molecular level.
In the present study, we performed a genome-wide transcriptional analysis of Al-responsive genes in rice. By comparing transcriptional profiling between a wild-type rice and an Alsensitive rice mutant star1, we found that rice possesses novel mechanisms of Al-tolerance in addition to ART1-regulated mechanism in rice.

Plant Materials and Growth Conditions
Seeds of wild-type rice (Oryza sativa cv. Koshihikari) and an Alsensitive mutant, star1 [10], were germinated for 2 days at 30uC. The seedlings were then transferred to a plastic net floating on a 0.5 mM CaCl 2 solution in a 1.5 L plastic box. At day 4, the seedlings were exposed to a 0.5 mM CaCl 2 solution (pH 4.5) containing 0 or 20 mM AlCl 3 . Root length was measured with a ruler before and after 6 h treatments. Ten seedlings were used for each treatment.

RNA Isolation, Microarray and Data Analysis
Root tips (0-1 cm) and basal region (1-2 cm) of the roots (20-30 plants per sample) were excised from the seedlings of both wildtype rice and star1 mutant which had been exposed to 20 mM AlCl 3 for 6 h and immediately frozen in liquid nitrogen. Total RNA was isolated using the RNeasy Plant Mini Kit (Qiagen, Germany). The RNA quality was assessed on agarose gels and with the Nanodrop ND-1000 (Thermo Fisher Scientific, USA). Microarray analysis was performed according to Agilent Oligo DNA Microarray Hybridization protocols using the Agilent 44 K Rice Oligo DNA Microarray RAP-DB (Agilent Technologies, USA; G2519F#15241) [15] with three biological replicates (Agilent Technologies, USA; G2519F#15241) [15]. The hybridized slides were scanned using a DNA microarray scanner (Agilent Technologies, USA). Signal intensities were extracted by Feature Extraction software (Agilent Technologies, USA). For statistical analysis, we excluded genes we excluded genes with low signal intensities less than 500 (sum of +Al and -Al signal intensity) in all treatments of the wild-type and star1 mutant. This is based on expression level of known Al-tolerance genes (10)(11)(12)(13)(14). The average value (arithmetic mean) of fold change (the ratios of Cy3 and Cy5) and standard deviation (SD) of each probe were calculated using three biological replicates. Since the expression of known Al-tolerance genes is usually up-regulated by higher than three folds (10)(11)(12)(13)(14), we extracted genes up-regulated or down-regulated by Al more/less than three-fold in the wild-type and star1 mutant. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE40964 (http://www.ncbi.nlm.nih.gov/ geo/query/acc.cgi?acc=GSE40964).

Quantitative Real-time PCR
To validate microarray data, 12 genes were randomly selected for quantitative real-time PCR (qRT-PCR) (Table S1). Total RNA was prepared from the root tips and basal root regions of wild-type and star1 using RNeasy Plant Mini Kit (Qiagen, Germany) and reversely transcribed using SuperSript TM II Reverse Transcriptase (Invitrogen, USA) and Oligo(dT) primers. The qRT-PCR was performed on an Eppendorf MasterCycler ep realplex real-time PCR (Eppendorf, Germany) using the specific primers described in Table S1.
One-fifth dilutions of the cDNAs were used as a template for the qRT-PCR in a total volume of 20 mL as follows; 10 mL SYBR Premix Ex Taq TM Perfect Real Time (TaKaRa Biol Inc., Japan), 0.4 mL ROX Reference Dye, 0.8 mL primer mix (50:50 mix of forward and reverse primers at 10 pmol mL 21 each), 6.8 mL distilled water and 2 mL template. The reaction conditions were: 30 s at 95uC followed by 40 cycles of 30 s at 95uC, 20 s at 60uC and 35 s at 72uC. The rice Histone H3 was used as an internal control. Relative expression levels were calculated by the  comparative Ct method. Three independent biological replicates were made for each gene.

Results and Discussion
Tolerance and toxicity of Al stress are a complicated phenomenon, involving many genes and a number of signaling pathways [19]. However, microarray technique has provided a useful tool for investigation of genome-wide changes in transcripts. So far, microarray analysis for Al response has been reported in Arabidopsis [20][21][22], maize [23,24], Medicago truncatula [25,26], and wheat [27]. Since the mechanisms for Al-tolerance differ with plant species, in the present study, we performed a microarray analysis with rice, a well-known Al-tolerant species, to understand genes involved in high Al-tolerance at genome-wide scale.
Al-toxicity is characterized by inhibition of root elongation, which occurs within a few hours after exposure to Al [3]. Therefore, to exclude genes associated with Al-toxicity, we sampled the roots exposed to Al solution for 6 h for microarray analysis. Furthermore, to extract genes related to Al-tolerance, we compared the transcriptional profiling between the wild-type rice and an Al-sensitive mutant, star1 [10]. Moreover, we selected a concentration of 20 mM for Al treatment. At this concentration, the root elongation of the wild-type rice was hardly inhibited, whereas that of the mutant was inhibited by 75% ( Figure 1), which make possible to extract genes possibly associated with Al-tolerance.

Verification of Microarray Results by Quantitative Realtime PCR
To validate the reliability of the microarray data, we randomly selected 12 genes from root tips and basal root regions for the quantitative real-time PCR (qRT-PCR) analysis. There was a good correlation (r = 0.84) between the microarray data and the  Table 1. Functional classification of Al-responsive genes in the roots of the wild-type rice and star1 mutant.

Up-regulated a Down-regulated b
Root tips Basal roots Root tips Basal roots Genes which expression was changed higher than 3-fold (fluorescence signal more than 500) in the root tips and the basal roots were categorized. b Genes which expression was changed lower than 3-fold (fluorescence signal more than 500) in the root tips and the basal roots were categorized. doi:10.1371/journal.pone.0048197.t001 qRT-PCR results ( Figure 2). These results indicated that the microarray data could reflect the transcriptional changes caused by Al stress.

Overview of Al-induced Transcriptional Profiling
Agilent 44 K Rice Oligo DNA Microarray RAP-DB covers almost genes in rice genome [15]. In the root tips (0-1 cm) of wildtype rice, exposure to 20 mM AlCl 3 for 6 h resulted in upregulation of 213 genes and down-regulation of 21 genes ( Figure 3A and 3B). By contrast, much more genes (2015 and 1521) were up-and down-regulated by the same treatment in the star1 mutant ( Figure 3A and 3B). In the basal root region (1-2 cm), 126 and 112 genes, respectively, was up-and down-regulated in the wild-type rice, whereas the numbers of up-and downregulated genes were 2419 and 1663, respectively, in the star1 mutant ( Figure 3A and 3B).
Functional category analysis showed that 27.0-42.8% of the upand down-regulated genes are assigned to unknown function (Table 1). Genes related to 'Metabolism' and 'Abiotic or biotic stress response' were mostly affected by Al stress in both the wildtype rice and star1 roots (Table 1).
Since the root elongation was hardly inhibited in the wild-type rice, but severely inhibited in the star1 mutant, three different groups for Al-responsive genes could be divided by comparing expression profiling between wild-type rice and star1 mutant. Group 1 includes genes which are up-or down-regulated by Al only in the wild-type rice. These genes are probably involved in Al-tolerance. Twenty eight up-regulated and three down-regulated genes in the root tips, 50 up-regulated and 43 down-regulated genes in the basal root region, belong to this group ( Figure 3A and 3B). Group 2 includes genes, which are up-or down-regulated by Al in both the wild-type rice and star1 mutant. These genes are probably involved in Al-tolerance or -toxicity. There are 185 upregulated and 18 down-regulated genes in this group in the root tip, 76 up-regulated and 69 down-regulated genes in the basal root region ( Figure 3A and 3B, Table S2-S5). Genes in Group 3 are those up-or down-regulated only in the mutant. These genes are related to Al-toxicity and included 1830 up-regulated genes and 1503 down-regulated genes in the root tip, 2343 up-regulated and 1594 down-regulated genes in the basal root region ( Figure 3A and 3B). Most genes in this group are also response to general stresses and found in microarray data of other plant species such as Arabidopsis [20], maize [23,24] M. truncatula [25,26] and wheat [27]. For example, the genes encoding a NADPH oxidase, peroxidase, oxalate oxidase, which are reactive oxygen species (Table S2). Most types of abiotic stresses disrupt the metabolic balance of cells, resulting in enhanced production of ROS [28]. The accumulation of ROS such as 1 O 2 , O 2 2 , H 2 O 2 and HO N , during abiotic stresses was considered to be a by-product of stress metabolism as well as an overall unwelcome by-product of aerobic metabolism [29]. These findings indicate that these genes are involved in arrest of plant root elongation in response to general stress.

Spatial Profiling of Al-responsive Genes
Root tip has been considered as the target of Al-toxicity [3] based on root elongation inhibition, however, surprisingly, similar numbers of genes were up-and down-regulated by Al in the root tips and mature regions of both wild-type rice and star1 mutant ( Figure 4A and 4B). This result raises a question on whether the root tip is only the target of Al-toxicity. Among genes affected, 49 up-regulated and 7 down-regulated genes were the same between root tip and basal root region in the wild-type rice ( Figure 4A and 4B), but most Al-responsive genes were different between the root tip and basal root region. This was the same in the star1 mutant; 1385 up-regulated and 522 down-regulated genes were the same between the root tip and basal root region, whereas other genes showed root region-dependent ( Figure 4A and 4B). These results suggest that basal root region is also a target of Al-toxicity in addition to the root tip. In M. truncatula, Al-induced gene expression is also found not to be restricted to the root tip [25]. In fact, some genes identified from rice were expressed in both the root tips and basal root region. For example, OsFRDL4 was expressed in both the root tip and the mature root zones [11]. The expression of OsSTAR1 and OsSTAR2 was also induced in both regions [10]. These findings suggest that the basal root region is also involved in Al-tolerance and -toxicity.
Transcriptional Profiling of ART1-regulated Genes in the Wild-type and the star1 Mutant Roots ART1-regulated Al-tolerance has been identified as a major mechanism responsible for high Al-tolerance in rice [7,8]. We compared expression profiling of ART1-reguated downstream genes between wild-type rice and star1 mutant. Among 31 downstream genes, 11 genes showed higher fold changes in the expression in the mutant than in the wild-type rice (Table 2), whereas 13 genes showed higher fold changes in the wild-type rice than in the mutant ( Table 2). Seven genes showed similar fold changes in the expression between wild-type rice and mutant  Table 2. Expression changes of ART1-regulated genes in the roots of wild-type rice and star1 mutant.     (Table 2). Six genes were only up-regulated in the wild-type rice, including genes encoding Expansin (Os04g0583500), Mg 2+ transporter/OsMGT1 (Os01g0869200), OsNramp4/OsNrat1 (Os02g0131800), LrgB-like protein family protein (Os10g0578800), Allyl alcohol dehydrogenase (Os12g0227400) and uncharacterized plant-specific domain 01627 containing protein (Os11g0490100, Table 2). Among them, OsNrat1 (Al 3+ transporter) and OsMGT1 (Mg 2+ transporter) have been demonstrated to be involved in Al-tolerance [12,14]. Although several ART1-regulated genes were also up-regulated in the mutant, the Al-tolerance was severally decreased, indicating that not a single gene, but multiple genes are required to function together for high Al-tolerance in rice.

Novel Al-tolerance Mechanism in Rice
Among 28 genes only up-regulated in the root tips of the wildtype ( Table 2, Table 3), 6 genes are ART1-regulated, indicating that there are other mechanisms for Al-tolerance except ART1regulated pathway in rice. Seven genes out of 22 genes belong to unknown function group (Table 3), while other genes are related to transporter (nitrate transporter, iron-regulated transporter), metabolism (nitrate reductase), oxidative stress-responsive genes (germin-like protein), polysaccharide/cell wall metabolism (cell wall invertase, beta-1,3-glucanase precursor) and so on.
Genes encoding nitrate transporter1 (OsNRT1; Os05g0410900) and nitrate reductase (OsNR; Os08g0468100) were up-regulated by 6.9-and 3.0-fold in the root tips of wild-type rice (Table 3). OsNRT1 is a low-affinity transporter for nitrate uptake [30], while OsNR is responsible for the reduction of nitrate to nitrite [31]. Rice takes up nitrogen mainly in the form of ammonium, therefore, it is unlikely that up-regulation of OsNRT1 and OsNR is for enhancing nitrogen uptake. One possibility is that the upregulation is associated with nitric oxide (NO) production. Nitric oxide is produced from nitrite and a key signal molecule involved in many physiological processes in plants [32]. In fact, addition of exogenous NO enhanced Al-tolerance in rice roots by decreasing the contents of pectin and hemicellulose, increasing the degree of methylation of pectin, and decreasing Al accumulation in root cell walls [32], supporting that up-regulation of OsNRT1 and OsNR is required for Al-tolerance in rice.
Genes related with secondary metabolism were also upregulated by Al. Chalcone synthase (CHS, EC 2.3.1.74) is a key enzyme of the flavonoid/isoflavonoid biosynthesis pathway. A gene encoding this enzyme was up-regulated by 4.4-fold (Table 3). CHS is quite commonly induced in different plant species under different forms of stress like UV, wounding, herbivory and microbial pathogens, resulting in the production of compounds that have e.g. antimicrobial activity (phytoalexins), insecticidal activity, and antioxidant activity or quench UV light directly or indirectly [35]. CHS expression causes accumulation of flavonoid and isoflavonoid. On the other hand, 4-Coumarate:CoA ligase has a pivotal role in the biosynthesis of plant secondary compounds at the divergence point from general phenylpropanoid metabolism to several major branch pathways [36]. Al is known to induce peroxidation and ROS formation in rice roots [37,38]. Increased secondary metabolites such as flavonoids may increase antioxidative capacity, subsequently alleviating Al-toxicity. In line with this aspect, a gene encoding germin-like protein (OsGLP; Os03g0804500) was also up-regulated (Table 3). Germin-like proteins (GLPs) constitute a diverse family of ubiquitous plant glycoproteins [39]. Many GLPs have manganese-containing superoxide dismutase (SOD) activity [40,41]. The SOD activities catalyze the dismutation of superoxide into oxygen and hydrogen peroxide. In this study, OsSOD was not up-regulated in rice roots after short exposure to Al stress (Table 3, Table 4, Table S2, S3), indicating OsSOD might not function in rice root after short exposure to Al stress. Thus, OsGLP might function as SOD. Furthermore, the H 2 O 2 produced by OsGLPs is detoxicated by peroxiredoxin (PrxR) and thioredoxin (Trx) because they are only up-regulated antioxidant genes in rice root tips (Table S2). These results suggested that OsGLP, OsPrxR and OsTrx function as major ROS-scavenging enzymes in the rice roots after short exposure to Al stress. Among genes up-regulated by Al only in the basal region of wild-type rice, some are related to polysaccharide/cell wall metabolism, including genes encoding glycoside hydrolase (GH; Os02g0802200) and glycosyl transferases (GTs; Os04g0506800, Os04g0477500) ( Table 4). Glycoside hydrolases (GHs) catalyze the hydrolysis of the glycosidic linkage to release smaller sugars [42]. Glycosyl transferases (GTs) catalyze the transfer of sugar moieties from activated donor molecules to specific acceptor molecules, thereby forming glycosidic bonds [43]. Al causes the thickening and rigidification of cell walls [44]. Increased expression of OsGH and OsGTs may contribute to the cell wall synthesis, hence alleviating the Al-induced inhibition of longitudinal cell expansion.
Gene encoding 1-aminocyclopropane-1-carboxylate oxidase (OsACC; Os05g0149400) was down-regulated in the basal root (Table 4). OsACC is related to biosynthesis of ethylene. Ethylene production is associated with inhibition of root elongation in Lotus japonicus and M. truncatula [45]. Down-regulation of OsACC may prevent further inhibition of root growth caused by Al. The association between other genes and Al-tolerance remain to be examined in future.
As a conclusion, our comparative genome-wide transcriptional analysis reveals that there are other mechanisms for Al-tolerance except for ART1-regulated one in rice including those related to nitrogen assimilation, secondary metabolite synthesis, cell-wall synthesis and ethylene synthesis. Although the exact roles of these putative tolerance genes remain to be examined, our data provide a platform for further work on Al-tolerance in rice.

Supporting Information
Table S1 Primer sequences used for quantitative realtime PCR.