Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Cadmium (Cd) is a major pollutant of several industrialized or developing countries and its effects on human health are particularly deleterious (Ishihara et al. 2001). Industrial activities and use of sewage sludge account form the principal sources of Cd, and lead to a continuous increase in its concentration in crop fields and water sources. Cd is considered as a non-essential metal, even though it has been reported to act as a cofactor of carbonic anhydrase in Thalassiosira weissflogii and other marine diatoms (Lane et al. 2005). Like other non-essential metals, it is assimilated by plant roots via non-specific transport activity (Clemens 2006). Once in the xylem, it spreads over the entire organism where it affects numerous biological functions probably by replacing metal cofactors, by interacting with protein thiols and by inducing the production of reactive oxygen species. In fact, the mechanistic basis of its toxicity is still poorly understood. In order to cope with Cd, plants have evolved various mechanisms leading to its chelation, exclusion and/or stabilization (Clemens 2006; Cobbett and Goldsbrough 2002; Verbruggen et al. 2009). One of the main well-known protection mechanism in plants involves its chelation by phytochelatins, which are cysteine-rich peptides of general formula (γGlu-Cys)n-X [with n = 2–11, and X = Gly (PC) or Ser, Ala or Glu (isoPC)], followed by their transport into the vacuoles (Cobbett 2000; Rauser 1990; Zenk 1996). Understanding the molecular activities underlying these processes could allow one to take advantage of them for phytoremediation of contaminated soils or biofortification for instance to avoid Cd in the edible parts of plants (Pilon-Smits 2005; Palmgren 2008).

Transcriptome-, proteome- and metabolome-wide studies have become increasingly used, and a wide variety of techniques allows for quantitative and/or qualitative analyses of samples at the plant, tissue or organelle level (Agrawal et al. 2011; Baginsky et al. 2010; Fukushima et al. 2009). They provide accurate snapshots of their molecular content at a given time and growth conditions, and comparisons of these datasets are able to give an invaluable dimension of dynamic to this profiling. A number of transcriptomic analysis using microarrays have been performed in the last few years to study the impact of Cd on plants (Herbette et al. 2006; Kovalchuk et al. 2005; van de Mortel et al. 2008; Weber et al. 2006; Yamaguchi et al. 2010; Zhao et al. 2009, 2010). Highly informative in terms of gene expression regulation over a time and/or concentration course, they allow the identification of early processes to be switched on and off by the plant during exposure to the metal. Recently, the monitoring of transcript levels of Solanum torvum plants subjected to low Cd stress was achieved by high-throughput RNA sequencing technologies (RNA seq) (Yamaguchi et al. 2010). This technique, yet still more expensive than microarray-based analyses, allows one to qualitatively and quantitatively explore transcriptome of species for which no microarray chip is commercially available, or species which their genome has not been annotated or even sequenced.

Proteomic analyses usually involves a pre-fractionation step of the sample, followed by a trypsic digestion of the proteins (Agrawal et al. 2011). Peptides are then analyzed by peptide mass fingerprinting (PMF) or tandem MS (MS/MS), often after decomplexification by chromatography (nanoLC-MS). Protein fractionation prior to MS analyses is, in most cases, based on 1D or, more often, 2D-polyacrylamide gel electrophoresis (PAGE), and this applies to most of the studies analyzing the plant response to Cd at the leaf, root and cell culture levels (Aina et al. 2007; Aloui et al. 2009; Alvarez et al. 2009; Durand et al. 2010; Gianazza et al. 2007; Hajduch et al. 2001; Hradilova et al. 2010; Kieffer et al. 2008, 2009a, b; Lee et al. 2010; Rodriguez-Celma et al. 2010; Roth et al. 2006; Sarry et al. 2006; Semane et al. 2010). These comparative 2D-PAGE studies rely on the densitometry intensities of the protein spots on the gel, measured using dedicated analysis software after classical (colloidal Coomassie blue or silver staining) or fluorescent (DIGE) staining. Protein identification of the different spots is then performed by MS. In addition to these techniques, procedures allowing direct quantification in the mass spectrometer are being developed, and here again several methods have been successfully used. They consist in the simultaneous analysis, during a single MS run, of “heavy” and “light” samples, the “heavy” term referring to as stable isotope (15N or 13C for instance) labelling of the sample prior to protein extraction, during cell culture or plant growth. Other methods, such as iTRAQ, require post-extraction labeling, but allow for multiplexing of up to 8 samples. Whichever method being used, peptides from the different samples conserve identical properties, allowing them to enter the mass spectrometer at the same time when a nanoLC-based separation is performed. Then, simultaneous MS or MS/MS analysis of these peptides permits the discrimination of their origin, and thereby relative quantification. Both quantification methods, e.g. 15N- (Lanquar et al. 2007) and iTRAQ- (Alvarez et al. 2009; Schneider et al. 2009) labeling have been used to evaluate Cd-induced changes in plants, as opposed to label-free methods such as spectral counting or, Accurate Mass and Time (AMT) tag approaches, the latter being particularly promising for quantitative proteomic works.

While transcriptomic and proteomic methodologies have been, to some extent, particularly used to study effects of Cd on plants, metabolomic-based work is much less well represented in the literature. A range of analytical platforms can be used to analyze the maximum number of metabolites in a given sample. These include Gas Chromatography–MS (GC–MS), comprehensive GC × GC–MS, Liquid Chromatography–MS (LC–MS) and variants including Ultra Performance Liquid Chromatography (UPLC), Capillary Electrophoresis–MS (CE–MS), Direct Infusion Mass Spectrometry (DIMS), Fourier Transform InfraRed Spectroscopy (FT–IR) and NMR spectroscopy (Fiehn et al. 2000; Goodacre et al. 2003; Hirai et al. 2004; Sato et al. 2004; Tolstikov et al. 2003; Ward et al. 2003; Wolfender et al. 2003). Both global (without any a priori) analyses and more targeted studies have been undertaken. For example, metabolite analyses performed using NMR revealed modifications of organic and amino acids contents after a Cd exposure (Bailey et al. 2003; Hediji et al. 2010) while a global approach performed by LC-MS/MS pointed out that the main variation at the metabolite level came from the presence of several types of chelating peptides (phytochelatins, PC) in Cd-stressed Arabidopsis thaliana cells (Sarry et al. 2006). Targeted metabolite studies revealed new insights into the regulation of the biosynthesis of glutathione (GSH) and PC biosynthesis in A. thaliana cells (Ducruix et al. 2006) whereas MS analyses of the phloem sap of Brassica napus suggested a role of the thiol-peptides (GSH and PCs) in the long-distance transport of Cd (Mendoza-Cozatl et al. 2008). With the aim of understanding targets of Cd in plants, as well as general and specific defence mechanisms it induces as a response to its toxicity, researchers have taken advantage of these “omics” techniques to draw the complex signaling and metabolic network resulting from Cd contamination. The picture, however, is still incomplete, and some key aspects of the plant defence mechanisms remain unknown. We present here a summary of the different results, from “ome”-wide methodology, that participate in the establishment of the current state of the art of the plant−Cd interaction. Other chapters intend to suggest new additional directions to be considered for further improvement of our knowledge. We particularly focus on the interest and promises of combining and cross-analyzing these data in a system biology-oriented framework to decipher actions and reactions of the plant−Cd conflict.

2 Global Response of Plants to Cadmium Exposure

2.1 Overview of a Cadmium Exposure in Plants as Evaluated by Transcriptomic, Proteomic and Metabolomic Approaches

2.1.1 Primary Metabolism

Even though its manifestation and amplitude importantly differs between plant species and stress severity, the harmful effect of Cd on the photosynthetic machinery is constantly reported as a major cause of its toxicity on plants (DalCorso et al. 2008; Sanità di Toppi and Gabbrielli 1999). Observed at both macroscopic (morphological) (Semane et al. 2010) and molecular (protein) levels, Cd exposure induces a strong disorganization of the light collecting antennas (Fagioni et al. 2009), electron transfer chain (Durand et al. 2010; Kieffer et al. 2009b), Calvin cycle (Durand et al. 2010) and membranes (Semane et al. 2010), as established by proteomic approaches. Kieffer and co-workers (2008) reported for instance a down regulation of several enzymes (chloroplast ferredoxin-NADP+ oxidoreductase, Cyt B6-f complex iron-sulfur subunit, oxygen-evolving enhancer protein 1 and 2) on poplar, and Hajduch et al. (2001) observed changes in RuBisCO levels as well as other enzymes of the Calvin cycle on rice leaves, later confirmed on poplar (Durand et al. 2010) or spinach leaves (Fagioni and Zolla 2009). Interestingly these changes at the protein levels are correlated with a global down regulation of the expression of genes involved in photosynthesis, meaning that not only is this reduction of protein accumulation the result of an enhanced degradation, but also the consequence of an active repression of their synthesis. Microarray analysis performed by Herbette et al. (2006) highlighted a transcriptional repression of genes coding for subunits of the PSI (psI-G, psI-K, psI-N, psaE, psaD, etc.), PSII (psbQ, psbP, psbW), enzymes of the chlorophyll biosynthetic pathway (Mg chelatases, 4–aminobutyrate transaminases, hydroxymethylbilane synthase, etc.), and enzymes of the Calvin cycle (Transketolase, Rubisco, Ribose-5-phosphate isomerase, fructose-bisphosphate aldolase, etc.) (schematized in Fig. 1). Overall, understanding the effect of Cd on photosynthesis is of prime importance to decipher molecular mechanisms leading to Cd resistance as the ability of plants to control the damage that is done to this Achilles’ heel has been proposed to be a major contributor to their sensitivity/tolerance (Bah et al. 2010). Indeed, destabilization of the photosynthetic apparatus by Cd is thought to lead to an important number of subsequent modifications of the plant physiology, including carbohydrate metabolism re-organization or oxidative stress.

Fig. 1
figure 1_6

Some cellular functions and molecular mechanisms highlighted by “omics” investigations under cadmium stress. Information concerning Cd-induced modifications of main cellular functions or metabolic pathways presented here (photosynthesis; carbohydrate metabolism, sulphur assimilation, amino acids (AAs), Glutathione (GSH) and phytochelatins (PC) syntheses; pathogenesis-related proteins and other stress proteins; proteolysis and oxidoreduction control) are further detailed in the text

Full size image

Plant response to Cd is, in contrast, less clear at the carbohydrate metabolism level, yet important modifications of its constituents are invariably mentioned. On the one hand, the shutdown of the photosynthetic machinery calls for an up-regulation of other sources of reducing power and ATP, such as glycolysis or TCA cycle; but on the other hand, supplying these pathways with necessary carbohydrates is becoming precarious as the main source of carbon has dried up. In poplar leaves, Kieffer et al. (2008) showed that an α–amylase was up-regulated, as well as a fructose kinase 2, involved in starch degradation and the phosphorylation of fructose prior to its entry in the glycolysis pathway. Similarly, genes encoding α– and β–amylases were shown to be up-regulated at the transcriptional level (Herbette et al. 2006). In parallel, numerous enzymes of this pathway (hexose kinase, fructose-bisphosphate aldolase, triose phosphate isomerase, cytosolic glyceraldehyde 3-phosphate dehydrogenase, phosphoglycerate mutase, 2-phosphoglycerate hydrolase, pyruvate kinase) were shown to be up-regulated in different proteomic works performed on Arabidopsis (Roth et al. 2006; Sarry et al. 2006), rice (Aina et al. 2007; Lee et al. 2010), poplar (Durand et al. 2010; Kieffer et al. 2009a, b), tomato (Rodriguez-Celma et al. 2010) or flax (Hradilova et al. 2010) (Fig. 1). Accumulation of these enzymes was also correlated with an up-regulation of several enzymes of the TCA cycle (pyruvate dehydrogenase complex enzymes 1, 2 and 3, citrate synthase, aconitase, NADP-specific isocitrate dehydrogenase, succinate dehydrogenase or malate dehydrogenase) (Kieffer et al. 2009b; Sarry et al. 2006; Semane et al. 2010; Visioli et al. 2010). However, even though some enzymes of this pathway (glucose-6-phosphate dehydrogenase, transketolase) seem to be up-regulated at the protein level (Kieffer et al. 2009b; Sarry et al. 2006), transcriptomic analysis revealed a repression of the expression for most of the enzymes, including 2-glucose-6-phosphate dehydrogenases and the transketolase, but also the 3-hydroxyisobutyrate dehydrogenase and several phosphogluconate dehydrogenases and gluconolactonases (Herbette et al. 2006). These contradictory results may reflect differences in time of exposure and concentration of Cd employed, as these pathways are, though dependent upon carbon source availability, required for preserving suitable levels of reducing molecules and carbon skeletons that are necessary for the synthesis of amino acids and molecules involved in Cd2+ chelation such as GSH and PC.

It is noticeable that enzymes of the sulfur assimilation pathway, all the way down to cysteine synthesis and GSH production, are actively up-regulated in order to sustain the production of PC (Figs. 1 and 2). Herbette et al. (2006) reported a transcriptional activation of 6 sulfate transporters, out of 14, in response to Cd, including the high-affinity Sultr1:1 at as early as 2 h of a 50-μM Cd exposure (Fig. 2). In addition to an increased sulfur uptake from the medium, intracellular stocks are also remobilized and Cd exposure leads to an overexpression of the vacuolar Sultr4:1 transporter, as well as others involved in long-distance transport (Sultr2:1, Sultr3:5). Genes coding for enzymes involved in the subsequent reduction of sulfur in sulfide (ATP sulfurylase, APS reductase) are also activated early. Interestingly, an ATP sulfurylase was also found up-regulated in the Cd2+-hypertolerant facultative metallophyte A. halleri at 125 μM Cd2+ after 2 h of exposure and was present in the five genes that constitute the A. halleri core response (Weber et al. 2006). In Thlaspi caerulescens, another Zn/Cd hyperaccumulator, an ATP sulfurylase was also up-regulated under Zn-deficiency and high Cd exposure (van de Mortel et al. 2008). This gene was also strongly up-regulated in Solanum torvum roots stressed with a mild Cd2+ treatment (0.1 μM) for 3 days (Yamaguchi et al. 2010).

Fig. 2
figure 2_6

Simplified and schematic representation of transcriptional changes of genes involved in sulphur and glutathione metabolisms in roots. The metabolites are represented in gray boxes; arrows represent enzymatic reactions; when determined, changes in gene expression are indicated schematically to the right of each gene

Full size image

During Cd stress, the up-regulation of enzymes involved in the sulfur assimilation pathway, which ultimately leads to cysteine and enzymes involved in its amino acid precursor biosynthesis, was also confirmed by proteomic works with the accumulation of ATP sulfurylase, serine hydroxymethyltransferase (both mitochondrial and cytosolic forms), components of the glycine decarboxylase complex and O-acetylserine (thiol)lyase (Fagioni and Zolla 2009; Kieffer et al. 2009a; Roth et al. 2006; Sarry et al. 2006). In addition to these enzymes, others belonging to the amino acid metabolism and in particular to the synthesis of the other aminoacid precursors of glutathione, were also shown to be upregulated. This includes a glutamine synthetase, a glutamate synthase, an alanine aminotransferase, an aspartate aminotransferase, or the alanine:glyoxylate aminotransferase 2, which produces glycine from alanine (Aina et al. 2007; Kieffer et al. 2008, 2009a, b; Roth et al. 2006; Sarry et al. 2006; Semane et al. 2010).

Interestingly, Herbette et al. (2006) observed that Cd treatment induced an early and sustained down-regulation of many genes involved in the biosynthesis pathway of glucosinolates (Fig. 2), sulfur-rich compounds which play an important role in defense against herbivores and pathogens (Grubb and Abel 2006; Halkier and Gershenzon 2006). The down-regulation by Cd of most of the genes involved in this biosynthesis pathway observed by Herbette et al. (2006) suggests that, to support the sulfate demand for GSH and PC biosynthesis, the synthesis of glucosinolate is limited. In agreement with this hypothesis, the down-regulation of two genes encoding APS kinases and two genes encoding sulfotransferases were also observed, whereas three genes encoding 3’(2’),5’–biphosphate nucleotidase were up-regulated (Fig. 2).

2.1.2 Defence Mechanisms

Particularly highlighted in transcriptomic and proteomic studies, the deleterious effects of Cd on the plant redox status account for an important part of its toxicity. As opposed to iron, which mediates ROS production via the Haber-Weiss/Fenton reactions, Cd induces a redox imbalance by disorganizing the photosynthetic apparatus (Herbette et al. 2006) and by interfering with antioxidant enzymes and molecules (Clemens 2006; Sanità di Toppi and Gabbrielli 1999). A repression of several ROS-coping proteins, including, for instance, enzymes of the ascorbate biosynthesis pathway, can be observed (Kieffer et al. 2008), as well as the degradation of a Cu/Zn superoxide dismutase (Alvarez et al. 2009; Kieffer et al. 2008, 2009a). In contrast, a Fe-superoxide dismutase is up-regulated (Herbette et al. 2006; Kieffer et al. 2009a; Semane et al. 2010), together with different catalases and peroxidases (Alvarez et al. 2009; Herbette et al. 2006; Kieffer et al. 2008; Lee et al. 2010; Sarry et al. 2006), monodehydroascorbate reductase (Sarry et al. 2006; Alvarez et al. 2009; Herbette et al. 2006) and glutathione reductase (Herbette et al. 2006; Lee et al. 2010), transcriptional and protein levels being remarkably well correlated. In relation to this, the overexpression of numerous enzymes involved in the reduction of various oxidized molecules is particularly important. Peroxiredoxins (Ahsan et al. 2007; Durand et al. 2010; Sarry et al. 2006), phospholipid hydroperoxide glutathione peroxidase (Sarry et al. 2006), aldehyde dehydrogenases (Kieffer et al. 2009a; Sarry et al. 2006), formate dehydrogenase (Durand et al. 2010; Hradilova et al. 2010; Sarry et al. 2006), glyoxalase I (Ahsan et al. 2007), 2–nitropropane dioxygenase (Alvarez et al. 2009; Roth et al. 2006), NADPH oxidoreductase (Sarry et al. 2006), several quinone or other oxidoreductases (Kieffer et al. 2008, 2009a; Sarry et al. 2006; Semane et al. 2010; Visioli et al. 2010) were shown to be up-regulated under Cd treatment. In Brassica juncea, a methionine sulfoxide reductase, probably involved in the reduction of oxidized sulfur-containing residues of proteins (Davies 2005), was also reported to be induced by Cd (Alvarez et al. 2009). Interestingly, metabolite analyses performed by GC-MS or 1 H-NMR (Hediji et al. 2010; Sun et al. 2010) revealed a significant increase of antioxidant molecules including α-tocopherols, confirming the important role of antioxidant defences in the mechanisms of plant resistance to Cd (Collin et al. 2008).

This deleterious effect of Cd on proteins is further exemplified by the overexpression of a number of chaperones and HSPs, such as HSP70, DnaK subfamily proteins and HSP60 at the protein level (Ahsan et al. 2007; Hradilova et al. 2010; Kieffer et al. 2009b; Rodriguez-Celma et al. 2010; Sarry et al. 2006), and a disulfide isomerase (Kieffer et al. 2009b; Sarry et al. 2006), involved in protein correct folding. Up-regulation of HSPs, in particular small HSPs, was also highlighted at the transcription level (Herbette et al. 2006; Weber et al. 2006) (Fig. 1). However, and despite the overproduction of this important set of repairing enzymes, Cd-induced damages on proteins might be irreversible, as it is often the case when carbonylation occurs, for instance upon oxidative stress (Moller et al. 2007). In such a case, degradation of affected proteins is required, which importance was highlighted by in vitro assays (Polge et al. 2009). Polge et al. (2009) also showed that, in response to Cd, the 20S proteasome proteolytic pathway (i.e., the 20S proteasome and the peptidases acting sequentially downstream from the proteasome: tripeptidyl-peptidase II, thimet oligopeptidase and leucine aminopeptidase) is up-regulated at both RNA and protein activity levels in Arabidopsis leaves and may play a major role in degrading oxidized proteins generated by the Cd stress (Fig. 1). Experiments performed on yeast also demonstrated that mutants of the proteasome pathway are more sensitive to Cd than wt (Vido et al. 2001). Subunits of the 20S and 26S proteasomes, ubiquitin (Ub) family proteins, Ub-activating enzyme E1, Ub-conjugating enzyme E2, Ub-protein ligase E3, subtilase (serine-type endopeptidase), aspartyl protease family protein and cytosol aminopeptidase were shown to be up-regulated under Cd stress (Ahsan et al. 2007; Aina et al. 2007; Alvarez et al. 2009; Kieffer et al. 2008; Lee et al. 2010; Rodriguez-Celma et al. 2010; Roth et al. 2006; Sarry et al. 2006; Semane et al. 2010). In addition to proteins, other molecules might be affected by Cd and therefore require sequestration and degradation. The accumulation of several glutathione S-transferases (GST), mainly belonging to the τ- and ϕ-families, involved in the detoxification of toxic compounds by conjugation with GSH (Edwards et al. 2000; Marrs 1996), is often mentioned in the literature. Proteomic works performed in Arabidopsis revealed that AtGSTF2, 6, 7, 8, 9, 10 and AtGSTU8, 19 and 28 (Roth et al. 2006; Semane et al. 2010) were overexpressed by Cd. OsGSTU3, 4 and 12 were shown to be Cd-responsive in rice (Lee et al. 2010), and induction of various GSTs was also reported from studies conducted on soybean cells (Sobkowiak and Deckert 2006) and poplar (Kieffer et al. 2008, 2009b). Again, a good correlation exists between proteomic and transcriptomic data as several transcriptomic analyses also report up-regulated GST (Herbette et al. 2006; Kovalchuk et al. 2005; van de Mortel et al. 2008; Zhao et al. 2009) (Fig. 1).

Such detoxification processes require an important production of glutathione. While the sulfur assimilation/cysteine synthesis pathway is globally up-regulated upon Cd stress, enzymes involved in GSH synthesis (GSH1 and GSH2), in contrast, remain stable at both transcriptional and protein levels (Herbette et al. 2006; May et al. 1998; Weber et al. 2006). This strongly suggests that amino acid availability, rather than enzyme activity, is limiting in the production of GSH, used for both conjugation of toxic compounds and production of PC. While the different isoforms of PC (defined by their C-terminal amino acid) were thought to be species specific (Zenk 1996), metabolic profiling performed on Arabidopsis cells actually revealed that all the known isoforms were produced in this same organism, and even specified different times and levels of production for each of them (Ducruix et al. 2006, 2008; Sarry et al. 2006). Further metabolomic approaches highlighted that, as suggested by transcriptomic and proteomic works, amino acid availability, rather than enzyme activity, was critical for GSH and PC synthesis. Indeed, severe Cd stress led to the accumulation of the GSH immediate precursor: the dipeptide γ-GluCys, immediately consumed when glycine was added to the medium (Ducruix et al. 2006). This research elegantly illustrates the meaningfulness of combining different “omics” approaches (e.g., proteomics and metabolomics), which in this case pictures strengths and weaknesses of the PC-based defence mechanism in plants.

Interestingly, transcriptomic studies have shown that genes involved in cell wall metabolism were modulated in response to Cd in roots and leaves (Herbette et al. 2006; van de Mortel et al. 2008). Genes involved in lignin biosynthesis such as cinnamoyl-coA reductases, cinamyl-alcohol dehydrogenases, 4–Coumarate: CoA ligase 2 were up-regulated (Herbette et al. 2006; van de Mortel et al. 2008) whereas several genes coding for expensins and pectinesterases involved in cell expansion were repressed (Herbette et al. 2006), suggesting that components to strengthen the cell wall were modified probably to protect the plants from Cd stress. The modification of the cell wall at the root level could be a specific detoxification mechanism that limits the entry of toxic metals in Arabidopsis, similar to what has been observed in Phragmites australis (Ederli et al. 2004). Proteomic approaches combined with genetic and biochemical analysis performed on candidate proteins that accumulates in Arabidopsis cells treated with Cd (Sarry et al. 2006) have highlighted potential new detoxification mechanisms that plants use to face Cd toxicity (Dutilleul et al. 2008). This was the case of the Arabidopsis homologue to the human selenium binding protein, SBP1. SBP1 overexpression was shown to protect Arabidopsis from the toxic effects of Cd and binding studies indicated direct interactions between Cd and the protein (Dutilleul et al. 2008). Expression analysis performed on Arabidopsis SBP1::LUCIFERASE lines and by immunodetection assays demonstrated that SBP1 expression was not specifically induced in response to Cd stress but enhanced in response to stresses provoking a cellular sulfur demand (Hugouvieux et al. 2009). In line with these results, sulfur starvation increased SBP1 expression and SBP1 overexpression enhanced tolerance to stresses that require GSH for detoxification (Hugouvieux et al. 2009).

2.2 Transcriptomic Analysis Allow for Large Scale Comparisons Between Species and/or Treatment

Transcriptomics has been used to analyze the similarities and differences in the metal response among different plant species. In particular, comparative transcriptomic analysis on Arabidopsis thaliana and in the Zn2+ and Cd2+-hypertolerant facultative metallophyte Arabidopsis halleri were performed in order to identify response elements shared by these two species and highlight specific aspects of A. halleri that might lead to a better understanding of the metal-tolerance mechanisms (for more details see chapters of Marmiroli and colleagues). In pioneering works, Becher et al. (2004) and Weber et al. (2004) have shown that metal-homeostasis genes are constitutively expressed at much higher levels in A. halleri than in A. thaliana and independently from micronutrient status. It therefore appears that the effect of Cd on the A. halleri root transcriptome is much smaller than on that of A. thaliana (Weber et al. 2004). Indeed, after 2 h of treatment, only five genes were more than twofold up-regulated in A. halleri, versus 112 genes in A. thaliana. Four genes were up-regulated in the two species, constituting the “Cd2+ core response”. A reduced Cd2+ uptake capacities by roots and its higher faculty to sequestrate Cd2+ could contribute, at least in part, to Cd2+ tolerance of A. halleri (Weber et al. 2006). More recently, Courbot et al. (2007) have shown that a QTL involved with Zn and Cd tolerance co-localizes in A. halleri with the plasma membrane P1B-ATPase, HMA4, and it was demonstrated that the metal tolerance was due to both a triplication and altered cis regulation of the HMA4 gene (Hanikenne et al. 2008) (see also chapter on HMAs by Leonhardt et al.). Weber et al. (2006) have also shown that a short-term Cd exposure could provoke an apparent Zn2+ deficiency as judged by a high induction of the AtZIP9 uptake system and known as a marker of Zn deficiency. Another toxic effect of Cd could be an increased protein denaturation as illustrated by a high induction of several HSPs, and also mentioned by several proteomic data and other transcriptomic works. A transcriptomic study between Arabidopsis thaliana and the Zn/Cd-hyperaccumulator Thlaspi caerulescens exposed to different concentrations of Cd and Zn was performed by Aarts and colleagues (van de Mortel et al. 2008). Among the results they obtained, and as mentioned for A. halleri, a number of genes induced by Cd exposure were constitutively highly expressed in T. caerulescens. They also showed that Cd appears to induce Fe deficiency in Arabidopsis; an effect on lignin biosynthesis in both species as mentioned above and an effect on sulfate assimilation only in T. caerulescens. The discrepancy observed with other transcriptomic works could be explain by the difference in exposure times, since Herbette et al. (2006) mentioned that gene expression in response to Cd was more time-regulated than dose-regulated in Arabidopsis. Comparative transcriptomic analyses between plants exposed to different stresses were also established by other groups. Comparative microarray studies of seedlings exposed to Cd, Al, Cu and NaCl (applied at a similar degree of severity as estimated by root growth inhibition) were performed in order to identify the specific and generic responses they trigger (Zhao et al. 2009, 2010). Among the commonly up-regulated genes, they identified genes encoding ROS-scavenging enzymes such as GST and peroxidases, several Ca-binding proteins including calmodulin-like proteins, several enzymes involved in the tryptophane synthesis, the alanine amino-transferase 1 (ALAAT1), the trehalose–6–phosphate phosphatase and the arginine decarboxylase, these latter two enzymes being involved in the biosynthesis of trehalose and polyamine synthesis, respectively, which are two stress-responsive metabolites. Interestingly, when the trehalose synthesis pathway was analyzed using the Kappa-view tool (Tokimatsu et al. 2005), an activation of the trehalose pathway was predicted, correlating with an increase of root trehalose content that was experimentally determined in all treatments (Zhao et al. 2010). In the specific group of genes that were up-regulated by Cd, Zhao et al. (2009, 2010) identified a GDP-mannose pyrophosphorylase (CYT1) involved in the ascorbate biosynthesis and critical for ROS-related stress resistance, heat shock proteins and pathogenesis-related (PR) proteins. Interestingly, several proteomic studies also revealed that some of the mechanisms involved in the plant response to a Cd stress are common, with inducible defence mechanisms activated upon pathogen attack. These include in particular the production of different PR proteins normally associated with the oxidative burst during biotic stress responses. This common feature may be due to the fact that signaling pathways are probably shared, as application of some hormones (e.g. ethylene, jasmonate and salycilic acid) are known to mimic the effect of a pathogen attack (Sels et al. 2008). Among the PR proteins induced by Cd are a major latex-protein, a PR-10 family member (Roth et al. 2006), also induced in response to salt stress and bacteria (Jiang et al. 2007; Jones et al. 2006), a traumatin-like protein, a PR-5 family member (Kieffer et al. 2008; Semane et al. 2010), class I chitinases, members of the PR-3 family and several β-1,3-glucanases belonging to the PR-2 family (Kieffer et al. 2008) (Fig. 1).

3 Future Directions

3.1 Subcellular-Level Analysis of the Cd Response

The major advantage of proteome-wide analyses of whole plants or tissues, which is a large coverage of most of the major proteins for a “no a priori” study, also constitutes a considerable drawback as peptides from minor proteins often get hidden during MS runs by other, more represented, peptides. This is particularly exemplified in the case of the vacuole, which overall protein content is suggested to represent less than 2%, in mass, of the total cellular proteome. While extremely important in the processes leading to Cd sequestration and tolerance, its protein constitution was, until recently, rather unknown and the vacuolar actors of the Cd tolerance poorly identified. The lack of exhaustiveness resulting from an under representation of a particular organelle proteome can be more or less easily circumvented by deciding to focus on this single subcellular compartment, and therefore doing a sub-fractionation of the plant material prior to protein extraction (Agrawal et al. 2011). First proteomic studies of the vacuole of Arabidopsis appeared between 2004 and 2007, and they led to the establishment of a list of about 1100 non-redundant proteins, including more than 120 transporters (Carter et al. 2004; Jaquinod et al. 2007a, b; Shimaoka et al. 2004; Szponarski et al. 2004). Others have been completed, performed on barley (Endler et al. 2006) or cauliflower (Schmidt et al. 2007), which opened the way to differential proteomic analyses of this organelle under Cd stress. In this matter, Schneider et al. (2009) have used iTRAQ-labeled tonoplast samples isolated from barley subjected to 0, 20 or 200 μM Cd. Out of 56 transporters identified, 6 were found to be differentially expressed. Surprisingly, only one – an ABC-type MRP-like transporter homologous to AtMRP3 – was upregulated at 20 and 200 μM Cd, while others were only affected at 20 μM. It is interesting to notice that AtMRP3 was previously shown to be upregulated by Cd (Bovet et al. 2003), and to partly rescue the Cd-sensitive phenotype of DTY168 yeasts (Tommasini et al. 1998). A CAX1a-like transporters was found to be over-expressed at 20 μM (× 1.4), confirming that this kind of transporter might play an important role in Cd2+ detoxification, and a member of the natural resistance-associated macrophage protein (NRAMP) family (× 2.3), which homologous proteins AtNRAMP3 and 4 are known actors of the plant metal homeostasis (Lanquar et al. 2005, 2010). In addition, an isoform of the vacuolar H+-pyrophosphatase was overexpressed (× 2) as well as a γ-TIP-like aquaporin (× 1.9). Such organelle-targeted differential studies were also conducted on the thylakoid membrane (Fagioni et al. 2009), where a disruption of the PSI was observed, concomitant with the diminution of chlorophyll a and b contents. In contrast, the PSII was less affected and no change was observed on the cytochrome b6/f and ATP-synthase complex organization. Changes at the plasma membrane level were also studied by 15N labeling of cell cultures (Lanquar et al. 2007), and evidenced the overexpression by 2.5-fold of the ABC transporter AtPDR8, shown to be an extrusion pump conferring Cd resistance (Kim et al. 2007). An ammonium transporter (AtAMT1.1) was also shown to be fivefold upregulated, and seems to confirm the importance of side mechanisms, such as nutrient uptake to sustain amino acid production, as probable key players in the tolerance mechanisms. While very promising, these results did not lead to the clear identification of major, vacuolar actors of Cd tolerance, such as the ABCC1 and 2 ABC-type transporters recently characterized for their PC−arsenic transport activity in the vacuole (Song et al. 2010). The evaluation of their implication in the transport of PC−Cd complexes towards the lumen of the vacuole needs to be assessed, and may, after decades of research, finally identify the transporter responsible for this translocation. However it is particularly noticeable that, in the quest of this Holy Grail that is (or used to be?) the plant vacuolar PC−Cd transporter, authors have emphasized on the dynamic of membrane proteins. The yeast homologous SpHMT1 or ScYCF1, on the other hand, are not described to be up-regulated upon Cd stress, their activity being probably not limiting in the processes of vacuolar sequestration. Catalytic activities subsequent to Cd internalization in the vacuole and allowing stable and safe storage of this toxic are in contrast rather unknown, as well as a plausible recycling of the PCs, as suggested by some authors. Preliminary results of experiments aiming to identify soluble proteins differentially expressed under Cd stress revealed an over-expression of some of them, weakly expressed in control conditions, which are currently in the process of being characterized (Villiers F and Bourguignon J, unpublished data). Altogether, these results confirm an organelle-targeted proteomic approach as very valuable, but also point out the need for researchers to expend their digging area, to soluble proteins for instance, or to activity regulation other than over-expression, such as post-translational modifications (PTM).

3.2 Post-translational Modifications: Phosphoproteomic Studies

PTM studies are an important challenge for the future and proteomic analysis of most PTM, except for protein phosphorylation, is still in the initial phases in plants (Ytterberg and Jensen 2010). Phosphoproteomic analysis of Arabidopsis upon various types of stimuli has tremendously increased in the literature over the last few years (Benschop et al. 2007; Kersten et al. 2009; Li et al. 2009; Niittylae et al. 2007), and has allowed the identification of numerous proteins as important actors of the processes being considered (Schulze 2010). These techniques have also been successfully applied to isolated organelles such as vacuoles (Endler et al. 2009; Whiteman et al. 2008a, b) or plasma membrane (Hem et al. 2007; Nuhse et al. 2007; Whiteman et al. 2008a), highlighting hundreds of phosphorylation sites. In parallel, other works have pointed out the phosphorylation status of different proteins such as transporters (Kobayashi et al. 2007) as essential in the regulation of aluminum tolerance in Arabidopsis. This should soon lead people interested in understanding the regulation of Cd tolerance to perform phosphorylation profiling of whole protein extract, as well as more tissue- or organelle-specific samples, from Cd-challenged plants.

3.3 Systems Biology and Integrated Analysis

As mentioned above, post-genomic biology has allowed advances in our understanding of plant biological systems by mining genomic data, making use of high-throughput methods (transcriptomics, proteomics, metabolomics, screening, etc.), and resources for mutants functional studies. This new era of integrative biology enables scientists to determine how the interconnected networks of genes, proteins and metabolites (substrates, other ligands, etc.) work together in complex cellular processes which can now be followed in the cell owing to revolutionary imaging technologies (Mano et al. 2009). However, the functioning of a plant as a system is not the result of a simple network, but rather a combination of multiple, intertwined, dynamic, and linear or nonlinear interactions between its components (DNA, RNA, proteins, metabolites, organelles, cell types, organs, etc.). It is now well accepted that the comprehensive knowledge of how a plant system functions in fluctuating environmental conditions cannot be achieved by the characterization of its elements one by one, or a single class of elements in isolation of the others (Ruffel et al. 2010). One of the most challenging issues for biology in this twenty-first century is deciphering the diversity and complexity of biological processes in plant science as well as in other fields of fundamental biology. Addressing this priority is a multistep procedure as biological processes are first studied at a molecular scale, while seeking to elucidate the fine organization of the proteins involved, in order to monitor their dynamic interactions with other partners and to understand their reactivity. Then, the molecular information is integrated on a larger scale (organelle, cell and organism) and related to biological functions. A nice example of this approach is illustrated by the work of Curien and colleagues (2009). Combining expertise on molecular description of proteins, their interactions and their dynamics together with those on integrative biology, including well-thought modeling, should ultimately lead to a paradigm shift in how we understand complex biological processes, as a step towards systems biology and its technological counterpart, synthetic biology (Joyard and McCormick 2010).

3.4 Data Mining and Integration: The Bioinformatics Challenge

Because of their size, diversity and complexity, all the data generated by such multi-omics projects (DNA chips, qPCR, MS data as well as imaging and biochemical data) should be further integrated in order to search for the key information they contain. Several bottlenecks are already identified. Properly integrating data into dedicated knowledge bases (KBs) as well as designing user-oriented tools to efficiently query and visualize their content are two necessary steps towards higher level analysis, like network reconstruction or dynamic modeling of physiological states. In such a framework, standards are also necessary to ensure data quality and integrity in terms of capture, curation, verification and storage. For instance, depositing data in public repositories is necessary but not sufficient and the setting up of guidelines for reporting (namely “minimum information”) and standards is an important step to ensure that submitted omics data are sufficient for non-ambiguous interpretation and querying by other scientists (Chervitz et al. 2011). Within this context, considerable efforts have been carried out to make these datasets publicly available through databases and web interfaces (for a recent list of plant database resources, see Mano et al. 2009). As such, these webtools often represent a good starting point for planning experiments or generating hypotheses in silico or exploring your own dataset by comparative approach (for review see Brady and Provart (2009)). A supplementary challenge lies in techniques of data mining for high-throughput data which require a high level of expertise and represent a constantly evolving methodological research field for functional genomics and biological networks studies. Among these resources, transcriptome datasets represent the most comprehensive of all of the large-scale data types and have led to the development of many tools for querying these both in a directed manner and correlatively. Hence, several co-expression gene databases and webtools dedicated to Arabidopsis have been developed for functional inference purposes, mainly by calculating standard Pearson correlation coefficients between gene pairs (for review, see Usadel et al. 2009). In addition, on the basis of a more sophisticated methodological framework (Bansal et al. 2007), a compendium of public micro-arrays has been shown to be valuable either to explore the whole transcriptional network (Ma et al. 2007) or to focus on a smaller subpart such as cell wall biosynthesis (Persson et al. 2005), glucosinolate metabolic pathway (Hirai et al. 2007) or amino acid metabolism in response to abiotic stress (Less and Galili 2008). To go further, some bioinformatic programs and omics datasets can be smartly combined into a strategy to elucidate the function of a gene of interest, including modes of regulation and synthesis, and potential interacting partners, and last but not least processes that are regulated by its gene products. With the growing number of programs, webtools and various datasets available, examples of combinatory applications of bioinformatic tools that lead to the generation of testable hypotheses (sometimes with subsequent confirmation) have been recently proposed with a focus on a cross-talk between nitrate response and hormone signals for the control of the plant development (Nero et al. 2009). Regarding the field of plant response to Cd, the number of published integrated analysis remains very limited. Owing to the low amount of “omics” data produced under these experimental perturbations, a set of genes known to be deregulated in Cd stress condition was used to select co-expression graph through a large compendium of microarrays datasets encompassing various experimental conditions including biotic and abiotic stresses. The resulting transcriptional modules are assessed by GO term enrichment and promoter analysis to identify shared combinatorial control (Boyer et al. 2009). Our preliminary results highlight a potential dialog between the sulfur/nitrate assimilation pathways, the biosynthesis of glutathione amino-acid precursors and the glucosinolate biosynthesis pathway. Further studies are under investigation for confirmation and targeted experiments (Boyer et al., in preparation). This may lead to new findings of gene regulation under stress condition in plant.

4 Conclusion

Finally, if data generation is no longer the limiting factor in advancing biological research, their integration, analysis, and interpretation have become key bottlenecks and challenges that biologists conducting genomic research face daily. While advanced data analysis tools for exploiting genomic data are rapidly emerging, the narrow specialization of most current software tools forces biologists to employ many tools to analyze the data in a single biological study. Despite the improved integration in software tools, the sheer number and diversity of analytical and visualization tools available might be bewildering to biologists who wish to apply them correctly. In this regard, the setup of a sustained and open dialog between computer scientists, statisticians and biologists actually represents one of the key steps for successful collaboration in plant systems biology (Shasha 2003). By integrating selected genomic data, visualization and analysis tools into a single Web-accessible software platform, the Virtual Plant project is promising for bridging the gap between biologists and bioinformaticians by providing an intuitive way to integrate and mine diverse data sources (Katari et al. 2010).