Evidence for functional interaction between brassinosteroids and cadmium response in Arabidopsis thaliana

Abstract

Plant hormones, in addition to regulating growth and development, are involved in biotic and abiotic stress responses. To investigate whether a hormone signalling pathway plays a role in the plant response to the heavy metal cadmium (Cd), gene expression data in response to eight hormone treatments were retrieved from the Genevestigator Arabidopsis thaliana database and compared with published microarray analysis performed on plants challenged with Cd. Across more than 3000 Cd-regulated genes, statistical approaches and cluster analyses highlighted that gene expression in response to Cd and brassinosteroids (BR) showed a significant similarity. Of note, over 75% of the genes showing consistent (e.g. opposite) regulation upon BR and Brz (BR biosynthesis inhibitor) exposure exhibited a BR-like response upon Cd exposure. This phenomenon was confirmed by qPCR analysis of the expression level of 10 BR-regulated genes in roots of Cd-treated wild-type (WT) plants. Although no change in BR content was observed in response to Cd in our experimental conditions, adding epibrassinolide (eBL, a synthetic brassinosteroid) to WT plants significantly enhanced Cd-induced root growth inhibition, highlighting a synergistic response between eBL and the metal. This effect was specific to this hormone treatment. On the other hand, dwarf1 seedlings, showing a reduced BR level, exhibited decreased root growth inhibition in response to Cd compared with WT, reversed by the addition of eBL. Similar results were obtained on Brz-treated WT plants. These results argue in favour of an interaction between Cd and BR signalling that modulates plant sensitivity, and opens new perspectives to understand the plant response to Cd.

Cadmium, brassinosteroids, gene expression, transcriptional interaction

Introduction

Plants are sessile organisms that have developed complex signalling networks to respond and adapt to adverse conditions such as biotic and abiotic stresses, including drought, cold or salt stress. Plants produce a wide variety of hormones that, in addition to regulating growth and development, are involved in biotic and abiotic stress responses (Bari and Jones, 2009; Gill and Tuteja, 2010).

Soil pollution by metals is a major problem in many industrialized and Third World countries, and is responsible for a number of human maladies including cancer, bone fragility, and kidney dysfunction (Ishihara et al., 2001). Widely present in cultured soils due to applications of contaminated sewage sludge, it accumulates in plants from where it is disseminated along the whole food chain. In this context, understanding the mechanisms for plant protection and toxic sequestration can be considered as one of the most important challenges in the coming decades in order to reduce the amount and effects of metal pollutants. The impact of heavy metals on plant physiology has, therefore, been extensively studied and the effect of cadmium (Cd) or zinc (Zn) as model heavy metals has led to the identification of some detoxification pathways in plants as well as in yeast. Several molecular components involved in plant Cd uptake, accumulation, and tolerance, have now been identified (di Toppi and Gabbrielli, 1999; Clemens, 2006a). In particular, low specificity ion (Fe²⁺, Ca²⁺, and Zn²⁺) transporters have been suggested to enable Cd to enter the plant cells (Thomine et al., 2000; Connolly et al., 2002; Perfus-Barbeoch et al., 2002). Once inside the cell, it interferes with many biological processes and mainly leads to reduced growth and leaf chlorosis. Nutrient uptake is disturbed and, in leaves, Cd triggers the degradation of the photosynthesis apparatus (Fagioni et al., 2009) and disturbs water status by inducing stomatal closure. Cd is also suggested to target several enzymes, especially those which require metallic ions as co-factors, particularly Zn binding proteins. Cd does not mediate the direct production of reactive oxygen species (ROS) via the Fenton and Haber–Weiss reaction, but the increase of ROS it triggers via the deregulation of redox control mechanisms also explains part of its toxicity.

Cd detoxification is mainly mediated through chelators such as phytochelatins (PCs), metallothioneins, and organic acids (Cobbett and Goldsbrough, 2002; Clemens, 2006a). In particular, PC chelation is one of the best characterized Cd detoxification process (Howden et al., 1995; Ebbs et al., 2002; Clemens, 2006b). Phytochelatins are thiol-rich peptides that are synthesized in the presence of heavy metals from glutathione (GSH) and related thiols by PC synthases, and facilitate the sequestration of heavy metals into vacuoles (Steffens et al., 1986; Vogeli-Lange and Wagner, 1990; Salt and Rauser, 1995; Clemens et al., 1999; Ha et al., 1999; Vatamaniuk et al., 1999, 2000; Romanyuk et al., 2006).

Up to now very few data concern hormone signalling networks involved in the plant response to heavy metals. Large-scale transcriptomic, proteomic, and metabolomic analyses were undertaken in Arabidopsis challenged with Cd (Herbette et al., 2006; Sarry et al., 2006). They revealed that transcription of many genes involved in the sulphur assimilation pathway and GSH metabolism was enhanced in response to the metal, providing an adequate supply of GSH for PC production. Many other genes involved in oxidative stress and calcium signalling were also found to be regulated and a cross-talk between ROS, nitric oxide, and calcium was proposed to regulate the cellular response of pea plants to Cd (Rodriguez-Serrano et al., 2009). Nitric oxide was shown to be a key regulator of Cd-induced programmed cell death in Arabidopsis suspension cultures (De Michele et al., 2009). However, despite the increasing knowledge of the cellular Cd response, several key points remain unclear. Among them, the hormone signalling pathway occurring during this stress is still poorly characterized and some authors suggest that the specific response of the plant against toxic metal ions would probably implicate at least one hormone (Clemens, 2006a; Weber et al., 2006). Rodriguez-Serrano et al. (2009) reported an over-production of ethylene and methyl jasmonate that, together with ROS, were proposed to regulate the induction of pathogenesis-related protein in the plant in order to protect proteins from Cd toxicity-related damages. Recently, these two hormones were shown to promote resistance in Arabidopsis to the metalloid selenite (Tamaoki et al., 2008).

The goal of the present work was to gain a better understanding of the hormone signalling network involved in response to Cd. Microarray experiments evaluating gene expression changes in Arabidopsis roots and shoots under Cd stress were published (Herbette et al., 2006). The behaviour of 3022 Cd-regulated genes was carefully analysed in response to eight hormone treatments available on the Genevestigator database. These analyses pointed out a significant similarity between Cd- and BR-induced responses at the gene expression level. The impact of modifying BR levels on the plant response to Cd was investigated. A substantially increased sensitivity to Cd was observed on plants treated with BR. This relationship between Cd sensitivity and BR level was further confirmed using BR-deficient mutants. This work provides novel and interesting data on the molecular elements responsive to Cd and able to influence the Arabidopsis response to this metal.

Materials and methods

Plant material and culture conditions

Experiments were conducted on Arabidopsis thaliana (var. Col-0) plants. T-DNA insertion lines in DWARF1 (N506932), DWARF4 (N520761), and BRI1 (N503371) were obtained from the Nottingham Arabidopsis Stock Center (NASC). Selection of T-DNA lines was performed based on phenotype and genotype. Gene knock-out was confirmed by RT-PCR for the dwarf1 lines (DWF1-1, 5′ ATGTCGGATCTTCAGACACC 3′; DWF1-2, 5′ AGCCTGATCTCAGC AGCTACA 3′; and DWF1-3, 5′ GCCTCGGCATAAGCAGTTTCG 3′). For in vitro experiments, seeds were sterilized in 15% bleach, 0.01% Triton X-100, and sown on half-strength MS medium [0.22% (w/v) Murashige and Skoog basal medium (Sigma #M0404), 0.5% (w/v) sucrose, 0.05% (w/v) MES (pH 5.7), and 0.8% (w/v) agar type A]. After 4 d at 4 °C for stratification, plates were placed in a growth chamber, vertically, under 16 h of day (120 μE m⁻² s⁻¹, 56% humidity, 21 °C) and 8 h of night (56% humidity, 20 °C). After 4 d or 7 d, depending on the experiment, plants were transferred to half-strength MS medium with or without Cd (CdNO₃), eBL (epibrassinolide, Sigma #E1641), and Brz (brassinazole, TCI-Europe # B2829) at the indicated concentrations. After 3–6 more days of growth, root elongation was measured as an indicator of Cd toxicity and plants were eventually collected for the measurement of GSH/PC or Cd content and total RNA extraction.

Bioinformatic resources

Data collection

Published microarray experiments data from Herbette et al. (2006) were inserted in a home-made local database, developed with the Relational DataBase Management System mySQL (v4.0.15) obtained from the EasyPHP package (v1.7). The database was created using the PhpMyAdmin interface (v2.5.3) and scripts for data parsing and insertion were written in PHP (v4.3.3). The web interface, allowing data management and consultation, was set up on a local server (Apache v1.3.27, EasyPHP) and contains HTML, JavaScript, and PHP scripts. A total of 3022 genes presenting a statistical and homogeneous change of their expression upon Cd treatment across all times and concentrations (after Bonferoni correction; Herbette et al., 2006) were retrieved. Each gene was assigned a single value representing its averaged response to Cd stress regardless of any time- or concentration-dependent fluctuations. For these genes, available microarray experiment results from the Genevestigator database (https://www.genevestigator.com;Zimmermann et al., 2004), were downloaded and also included in the database. Microarray data from these different experiments were then normalized using Genesis v1.7.5 (Graz University of Technology, Austria).

Distance calculation

Expression values were retrieved after Genesis processing and implemented in a PHP script as variables. For each gene, a distance d was calculated as the absolute value of the difference between experiment results (e.g. Cd- and hormone-treated plants, for each considered hormone), divided by the average absolute induction factor across both experiments raised to the second power (cf. equation 1 where Cd_i and Hormone_i are induction values of the gene i after Cd and hormone exposure respectively).

(1)

A global distance D_Cd–Hormone for each Cd–Hormone pair was then calculated as the average distance of the 3022 genes composing this dataset (Equation 2).

(2)

Significance calculation

To assess the statistical significance of D_Cd–Hormone, a Monte-Carlo analysis by random sampling (Manly, 1997) was performed by 1000 bootstraps of shuffling cadmium data, each followed by D_Cd-Hormone calculation (then noted D_{Cd–Hormone*}). The 1000 resulting D_{Cd–Hormone*} determined a distribution for possible values to be obtained by random sampling, which probability density function was computed using the non-parametric Gaussian kernel density estimation method. The definite integral of this function was calculated using the trapezoidal rule, and a probability determined for the real D_Cd–Hormone value to be greater than or equal to any D_{Cd–Hormone*}, which was used as the P-value. A stringent threshold of P-value <0.0001 was applied to consider a D_Cd–Hormone value as significant.

Total RNA extraction, cDNA synthesis, and quantitative real-time PCR (qPCR) analyses

Total RNA were extracted from control and Cd-treated seedlings from about 100 mg fresh tissues using the RNeasy plant mini kit (Qiagen). One column DNAse-treatment was performed. One μg of total RNA was used for cDNA synthesis (Reverse-it first strand synthesis kit, ABgene) in a total volume of 20 μl. At the end of the reaction, 180 μl of water was added. Specific primer sequences designed for each gene are listed below: At3g50660 (Forward ACAGCAAAACAACGGAGCG, Reverse TCTGAACCAGCACATAGCCT); At5g38970 (Forward CCACTCGGTCTTGAGG ACG, Reverse AGCATGAGTTTTGTGACTCC); At3g19820 (Forward CACTC AAGGTGAAGCTATCAGG, Reverse TAGGACACAGCCAGGTGCGTAG); At1g77760 (Forward CTGAG CTGGCAAATTCCGAAGC, Reverse ATCTCTGCG TGACCAGGTGTTG); At5g64000 (Forward AGAGGGAGCTCCAACCTG ATAAAC, Reverse TCCGCAGATCTCCAGTTTC CTC); At5g52640 (Forward AAGCTCGATGGACAGCCTGAAC, Reverse TCCCAAGTTGTTCACCAAA TCTGC); At2g 45210 (Forward AGCAAAGTTTCAGACGCAGGTC, Reverse GAACCGGGTCGGCTTTCTTATG); At5g15230 (Forward AT GTGAAGTGG AGCCAGAAACG, Reverse ATTCCGATGGGCATTGGGTACG); At2g47550 (Forward AGTGCTGCCATCA CCAATGAGC, Reverse AGTGCTGCCATCA CCAATGAGC); At5g47370, Forward AACGTCGAGGAAGAAGCTCAGG; Reverse AACGTCGAGGAAGAAGCTCAGG). The real-time PCR reactions were carried out on 5 μL of diluted cDNA mixture, in triplicates, on a Rotor-Gene 3000 instrument (Corbett Research) using SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich). Quantification of gene expression was performed using the comparative C_T method with the Rotor-Gene 3000 Software. The actin2/7 gene (At5g09810) was used to normalize qPCR data.

Glutathione and phytochelatins measurement

GSH and PC levels in roots and leaves of Arabidopsis were determined by HPLC of monobromobimane-labelled compounds as described by Sauge-Merle et al. (2003) using 50 mg of plant material. GSH and PC were quantified as nmol of the thiol equivalent.

Cadmium content quantification

Leaves and roots were washed in distilled water for 30 min and mineralized in 4 ml of 75% (v/v) HNO₃ (HNO₃ 65%, Merck, ref. 1.00441) and 25% (v/v) HCl (HCl 30%, Merck, ref. 1.00318) for 3 h at 85 °C. Residues were rehydrated in 5 ml 1% (v/v) HNO₃ and Cd concentration was measured by Inductively Coupled Plasma Mass Spectrometer (HP4500 ChemStation ICP-MS, Yokogawa Analytical Systems Inc, Tokyo, Japan). Set-ups were performed to quantify ¹¹²Cd, ¹¹⁴Cd (3 replicates of 1 s integration time).

BR purification and quantification

For sterols and brassinosteroids profiling, Arabidopsis plants were grown in hydroponic conditions and treated with 50 μM Cd as described by Herbette et al. (2006). Ten to 12 g of roots were collected and lyophilized, and sterols / BR quantification was performed as published (Fujioka et al., 2002).

Results

Transcriptome-wide analysis approach identified brassinosteroids and cadmium effects as very similar at the gene expression level

Published microarray analysis performed on Arabidopsis thaliana challenged with Cd (Herbette et al., 2006) were used to investigate whether a hormone signalling pathway was affected in response to the metal. To evaluate such a possibility, the expression level of 3022 Cd-responsive genes was analysed upon modification of eight hormone treatments retrieved from the Genevestigator database (Zimmermann et al., 2004) and compared as described in the Materials and methods. For each hormone treatment, a dyadic distance value d_Cd–Hormone was calculated for every single gene, which evaluates the dis/similarity of its response between the two experiments (Cd versus Hormone). The global relationship between Cd-mediated and hormone-mediated genes response was then calculated as a D_Cd–Hormone value, corresponding to the average of all 3022 d_Cd–Hormone. These D_Cd–Hormone values were calculated for all the different hormone treatments (Fig. 1). D_Cd–eBL (1.91) was found to be significantly smaller than any of the other D_Cd–Hormone values (ranging from 2.83 to 3.63), highlighting the response to eBL as the closest to the Cd one (Fig. 1). To assess the statistical meaningfulness of our results, a Monte-Carlo approach was used (Manly, 1997) to compute a P-value associated with each distance value. As described in the Materials and methods, random sampling from the original Cd dataset was performed 1000 times to create 1000 new, shuffled datasets. This allowed 1000 new D* values for each comparison of Cd–Hormone to be determined. Distribution of these D* values is shown as bar charts (see Supplementary Fig. S1 at JXB online), together with the function of this distribution (see Supplementary Fig. S1, blue curve, at JXB online) and the location of the original D_Cd–Hormone (see Supplementary Fig. S1, red line, at JXB online). This approach defined three hormone treatments (eBL, ABA, and IAA2) as significantly related to the Cd treatment as the original D_Cd–Hormone value was significantly lower than any of the D_{Cd–Hormone*} values obtained after suffling (P-value ≤0.0001). D_Cd–eBL being the smallest distance (1.91 versus 2.83 and 3.09 for, respectively, D_Cd–ABA and D_Cd–IAA-2), the focus was on the involvement of BR signalling in the Cd response.

Fig. 1.

Analysis of the similarity between Cd- and hormone-triggered gene expression. The expression level of 3022 Cd-regulated genes was retrieved from microarray analysis published by Herbette et al. (2006) and analysed in response to various hormone treatments available in the Genevestigator database. The similarity of each Cd–Hormone pair was evaluated by calculating a D value that takes into account the similarity of the response of every single gene after Cd treatment compared with each hormone treatment. For IAA treatment, three experiments available on the Genevestigator are reported. Among all the hormone treatments, eBL appeared to trigger the closest response to the Cd-induced one. eBL, Epibrassinolide (brassinosteroid); ABA, abscisic acid; GA3, gibberellin A3 (gibberellin); IAA, indole-3- acetic acid (auxin); MJ, methyl jasmonate; Zeatin (cytokinin). The asterisk (*) denotes the significantly smallest distance (P <0.01) evaluated by Student’s t test.

Expression profiles of the 3022 Cd-regulated genes in response to eBL and to two forms of brassinazole (Brz91 and Brz220, two triazole-type BR biosynthesis inhibitors (Asami et al., 2001)) are shown in Fig. 2. Genes are listed with their expression levels in response to the four treatments in Supplementary Table S1 at JXB online. The hierarchical clustering (Fig. 2A) shows the similarity of the global gene response of plants exposed to Cd and eBL, while the opposite pattern was observed in response to Brz91 or Brz220. Among the 1511 genes exhibiting a coherent regulation by eBL, Brz220 and Brz91 (e.g. opposite regulation by eBL versus Brz91 and Brz220) more than 75% showed comparable Cd- and eBL-induced regulations (Fig. 2B, C). Indeed, 547 genes were up-regulated by both Cd and eBL and down-regulated upon Brz treatment (Fig. 2B; see Supplementary Table S1b at JXB online), while 597 genes were down-regulated by Cd and by eBL but up-regulated by Brz treatment (Fig. 2C; see Supplementary Table S1c at JXB online). Many fewer genes were found to be Cd-down-regulated and eBL-up-regulated (181) or Cd-up-regulated and eBL-down-regulated (186) (Fig. 2A; data not shown). As expected, Monte-Carlo analyses confirmed the absence of significant similarity between Cd- and Brz-triggered responses (P-value=0.9 and 0.972 for Brz220 and Brz91, respectively, Fig. 2D). All together these results suggested a possible cross-talk between Cd and BR.

Fig. 2.

Expression profiles of 3022 Cd-regulated genes in response to eBL and Brz treatments. (A) Hierarchical clustering showing the similarity of the global gene response of plants exposed to Cd compared with plants exposed to eBL. Gene regulation after Brz91 or Brz220 exposure globally shows the opposite pattern. Gene expression data for eBL and Brz treatments were retrieved from the Genevestigator database. Expression is represented as log₂ ratio values after experiment normalization. (B) Cd-induced genes presenting an up-regulation after eBL exposure and a down-regulation after Brz treatments. (C) Cd-repressed genes presenting a down-regulation upon eBL treatment and an up-regulation after Brz treatments. (D) Monte-Carlo-based analysis showing the statistical similarity of Cd versus eBL responses and the dissimilarity of Cd versus the two BRZ responses. A distance D (red line) was determined to characterize the similarity of the responses and 1000 randomizations of Cd values allowed to calculate 1000 new distances D* (bar chart and blue curve), determining a distribution of possible distances to be obtained by chance. Distance values D and distribution of D* values were computed, and the P-value characterizing the similarity was calculated.

Similarity between Cd and BR response is confirmed by qPCR analysis

There was a need to validate the Cd and BR transcriptional similarity at the single gene level. Published marker genes of BR content were selected in the literature (Goda et al., 2002; Mussig et al., 2002) and their expressions were carefully analysed in the roots of Arabidopsis seedlings challenged with Cd for different times of exposure (Fig. 3). Our analysis included genes, the expression of which was affected in the BR deficiency mutants SAL2 (At5g64000), DWARF1 (At3g19820), and GAS4 (At5g15230) (Mussig et al., 2004) and genes consistently affected by BR deficiency and BR treatment in the wild type (WT) such as DWARF4 (At3g50660), NIA1 (At1g77760), BR6OX (At5g38970), and HAT2, (At5g47370). HSP83, induced by eBL in WT (At5g52640) (Mussig et al., 2002) and three genes specifically regulated by eBL versus auxin (Goda et al, 2002), GA3.9 (At2g47550), DWARF4 (already mentioned above), and SAUR36 (At2g45210), were also included. Eight of these genes were found to have similar behaviour in response to Cd and BR in our global analysis (see Supplementary Table S1a and b at JXB online). DWARF4, BR6OX, HAT2, and SAUR36 did not respond to Cd in the microarray analysis (Herbette et al., 2006). As shown in Fig. 3, all the genes described as being down-regulated in BR-deficient mutants and up-regulated by eBL in the WT showed enhanced expression in response to the metal (Fig. 3, four upper graphs). Conversely, all the genes characterized to be over-expressed in BR- deficient mutants and repressed by eBL in the WT were down-regulated in the presence of Cd (Fig. 3, six lower graphs). This expected behaviour was observed in at least two of the four time points tested for all genes except for SAUR36. SAUR36 regulation was observed only after 8 h Cd treatment (Fig. 3). Each gene expression was regulated at both 50 μM and 200 μM Cd, with a stronger regulation observed at the highest Cd dose. These results confirmed our global analysis that Cd treatment induces the same regulation as BR. Moreover, the four genes that did not show any regulation in response to Cd in the microarray analysis were in fact down-regulated in our analysis, as observed in the presence of BR. It is important to note that, among them, DWARF4 and BR6OX belong to the BR biosynthesis pathway and have been described to be specifically negatively regulated by BR in order to maintain adequate hormone levels (Goda et al., 2002; Mussig et al., 2002; Tanaka et al., 2005; Kim et al., 2006).

Fig. 3.

Expression profiles of genes regulated by BR in response to Cd stress. Ten genes whose expression level appeared to be BR-dependent based on publications (Goda et al., 2002; Mussig et al., 2002) and this work (see Supplementary Table SI at JXB online) (HSP83, HAT2, GH3.9, SAL2, NIA1, GAS4, SAUR36, DWARF1, DWARF4, and BR6OX) were randomly selected and their relative steady-state transcript levels were determined by qPCR in roots of Arabidopsis thaliana seedlings exposed to Cd stress. A minimum of 100 7-d-old seedlings per treatment and time point, were transferred to 0, 50, and 200 μM Cd-containing plates. Expression levels were analysed at 4, 8, 24, and 48 h. The actin2/7 gene (At5g09810) was used to normalize qPCR data. For HSP83, at 24 h and 48 h, relative transcript accumulation in the presence of 200 μM Cd was 280% and 350%, respectively, compared to control conditions with no Cd. For GA3.9, relative transcript accumulation reached 300%, at 24 h and 48 h at 50 μM Cd.

Altogether, transcriptome-wide analyses and single-gene-level studies show a close similarity between Cd-induced and BR-triggered transcription regulation, suggesting that, in plants, the Cd response includes an activation of the BR-dependent signalling pathway.

Similarity between Cd- and BR- regulated gene is not a general response to stress

To find out whether this similarity with eBL was specific to Cd, or due to a global stress state of the plant upon Cd exposure, microarray data from the ATGenExpress repository (www.arabidopsis.org/info/expression/ATGenExpress.jsp) corresponding to drought stress treatment (Submission number ME00338; D’Angelo and colleagues) and Pseudomonas aeruginosa treatment (Submission number ME00353, Dong and colleagues) performed on Arabidopsis were obtained. It is now well established that the plant response to drought involves ABA signalling and that the response to pathogen attack includes the MeJA and SA pathways (Gosti et al., 1995; Vijayan et al, 1998; Berger, 2002; Makandar et al, 2010; Yoshida et al., 2010; An and Mou, 2011). Responsive genes from these datasets (3000 and 2573, respectively) were retrieved and expression values of these same genes were downloaded from Genevestigator in response to eBL, ABA, MeJa, and SA. Distance calculations and significance evaluations were performed as previously described. As expected, drought stress was significantly similar to ABA treatment (p-val < 0.0001) while Pseudomonas exposure was not (P-value=0.171), and Pseudomonas exposure was similar to both MeJa and SA treatments, while drought stress was not (P-value=0.955 and 0.649) (Fig. 4). The results also showed the absence of similarity between eBL treatment and drought stress (P-value=0.497), as well as between eBL treatment and Pseudomonas stress (P-value=0.029) (Fig. 4). Overall, these results clearly demonstrate that the global comparative approach used to highlight the signalling pathways involved in Cd response was pertinent and further strengthen the hypothesis that Cd could trigger BR signalling activation.

Fig. 4.

Relevance of the used Monte-Carlo-based significance calculation of distance values was evaluated using two stimuli: drought stress and Pseudomonas exposure. Regulated genes (3000 and 2573, respectively) were retrieved from AtGenExpress, and corresponding expression values upon four hormone treatments downloaded from Genevestigator: epibrassinolide (eBL), abciscic acid (ABA), methyljasmonate (MeJa), and salicylic acid (SA). Distance values D (red line) and distribution of D* values (bar chart and blue curve) were computed, and the p-val characterizing the similarity calculated. As expected, results indicate that Pseudomonas exposure is significantly similar to both MeJa and SA treatment (P-value <0.001), but not to eBL and ABA treatments (P-value of 0.029 and 0.171, respectively). Drought-induced gene response is not significantly similar to eBL (P-value=497), MeJa (P-value=0.995), nor SA (P-value=0.649), but shows high similarity with ABA-triggered gene response (P-value <0.001).

Increasing the BR level in Arabidopsis WT seedlings enhances response to Cd

BR profiling experiments were performed in control- and Cd treated-Arabidopsis roots as this tissue contains higher BR level compared with shoot in Arabidopsis (Kim et al., 2006) and, moreover, accumulates up to 10 times more Cd than shoots (Herbette et al., 2006; Hugouvieux et al., 2009). For this experiment, Arabidopsis seedlings were grown hydroponically as described in Herbette et al. (2006) to provide enough root tissue for hormone measurements, and were challenged with 50 μM Cd for 24 h. In these conditions, it appeared that BR levels were very low in both control and Cd-treated roots and no significant difference was noticed (see Supplementary Table S2 at JXB online). These results suggested that Cd activation of BR signalling could be independent of a BR rise.

However, to evaluate whether the observed Cd-BR cross-talk influences the plant response to Cd, Arabidopsis seedlings treated with eBL were challenged with Cd. When submitted to eBL alone, WT plants exhibited typical curved and largely expanded leaves (Fig. 5A) with longer petiole as already described (Arteca and Arteca, 2001). Root growth was 10–30% reduced in the presence of eBL (P ≤0.05) compared with control conditions (see Supplementary Fig. S2A at JXB online). Cd treatment in the absence of hormone induced a slight leaf yellowing (Fig. 5A) that is probably due to the expected degradation of the photosynthesis apparatus (Pietrini et al., 2003) and root growth inhibition ranging from 15% (10 μM Cd) to 60% (50 μM Cd) (Fig. 5B, first set of columns). When both eBL and Cd were added, a marked tanning of the leaves was observed at 50 μM Cd for plants treated with 1 μM eBL and at 25 μM Cd for plants treated with 10 μM eBL (Fig. 5A). In parallel, Cd-induced root growth inhibition was significantly enhanced in the presence of eBL (Fig. 5B). When exposed to 10, 25, and 50 μM Cd, the inhibition increased from 15, 34, and 60% in the absence of eBL to 43, 66, and 82% in the presence of 10 nM eBL, respectively. This inhibition appeared to be eBL concentration-dependent up to 10 nM and seemed to reach a plateau for higher concentrations (Fig. 5B). As ABA and one of the IAA treatments were also shown to be significantly related to Cd treatment (see Supplementary Fig. S1 at JXB online), it was decided to evaluate the impact of these two hormones on the Arabidopsis Cd response (Fig. 5C). As observed for eBL, IAA or ABA alone, at 100 nM and 10 μM respectively, significantly reduced (P ≤0.05) the root growth compared with control plates with no hormone (see Supplementary Fig. S2B at JXB online). However, in combination with IAA or ABA, Cd-induced root growth inhibition was similar to Cd alone (P >0.05) as it reached 50–57% in the presence of the hormone versus 59% with no hormone (Fig. 5C). The synergic and specific effect between eBL and Cd on root growth provides additional evidence that the Cd and the BR hormone signalling pathways share common elements. In addition, such interaction enhances the plant sensitivity to the metal.

Fig. 5.

Effect of epibrassinolide (eBL) on Arabidopsis thaliana WT response to Cd. After growing on standard half-strength MS medium for 4 d, WT plants were transferred on Cd–(0–50 μM) containing plates with or without addition of eBL (A, B, C), auxin (IAA, C) and abscisic acid (ABA, C). After 6 d, the shoot phenotype (A) of WT plants was observed and root elongation (B, C) was measured. Data show a representative experiment performed on a minimum of 40 seedlings. At concentration lower than 1 μM eBL, shoot phenotype was similar to Cd alone in (A). Asterisks in (B) and (C) denote root length in Cd-treated plants statistically different (P <0.05) than mock-treated plants. Graphs presenting the absolute root elongation values and significant difference are shown in Supplementary Fig. S2 at JXB online. Significant differences were evaluated by Student’s t test.

To assess whether the observed effect was due to an actual difference in the response to the toxic, and not to lower Cd assimilation, the Cd content of roots and shoots of Arabidopsis seedlings exposed to Cd upon eBL and Brz treatment was measured (see Supplementary Fig. S3 at JXB online). No difference (in mg Cd g⁻¹ FW) was observed between the different samples, highlighting Cd assimilation independent from the endogenous BR level, strongly suggesting an increased sensitivity of the BR-treated plants.

Reducing BR level enhances cadmium tolerance

To test whether lowering the BR level could trigger the opposite effect, experiments were first performed with Arabidopsis seedlings mutated in the DWARF1 gene that acts at different levels of the BR biosynthesis pathway (Klahre et al., 1998; Choe et al., 1999). Plants deficient in DWARF1 present a strong depletion in most essential brassinosteroids (Klahre et al., 1998). A T-DNA line (N506932) exhibiting an insertion in the first exon of DWARF1 was obtained from the Salk Institute (Fig. 6A). Homozygous seedlings selected by PCR (data not shown) showed no expression of DWARF1 or a severe dwarf phenotype, confirming the reduced level in BR content in the T-DNA insertion line (Fig. 6A, B). In our growth conditions, dwarf1 root length was about 30% reduced compared with the WT in 10-d-old seedlings (Fig. 6B, C; see Supplementary Fig. S4 at JXB online). In the presence of Cd, dwarf1 seedlings exhibited no statistical root growth inhibition upon 10 μM Cd exposure (11% inhibition in the WT) and only 20% inhibition by 25 μM Cd versus 50% in the WT (Fig. 6B, C). Importantly, at 50 and 100 μM Cd, growth inhibition was still reduced in dwarf1 and absolute root length was even significantly longer (P <0.05) compared with the WT (Fig. 6B, C; see Supplementary Fig. S4 at JXB online) indicating that lowering the BR level substantially enhances tolerance to high concentrations of Cd. Similar results were also observed based on fresh weight measurements (see Supplementary Fig. S5 at JXB online). In addition, WT plants treated with BRZ showed reduced sensitivity to Cd (see Supplementary Fig. S6 at JXB online). Lowering the BR level in plants did not change Cd accumulation in roots and translocation to the shoots (see Supplementary Fig. S3 at JXB online). A check was made that dwarf1 reduced root growth inhibition was not a general response to stress by challenging the WT and dwarf1 seedlings to salt stress (Fig. 6D; see Supplementary Fig. S4 at JXB online). In contrast to the data obtained in the presence of Cd, it was found that dwarf1 was hypersensitive to 100 mM NaCl compared with WT plants, as evaluated by root elongation measurement (Fig. 6D), consistent with previous reports indicating that BR enhances salt stress tolerance in Arabidopsis (Kagale et al., 2007). The addition of eBL to dwarf1 seedlings increased root length in control conditions (see Supplementary Fig. S4 at JXB online) and restored WT sensitivity to Cd (Fig. 6E). Similar results were observed in shoots of dwarf1 mutant where photosystem degradation appeared to be stronger in plants exposed to both 50 μM Cd and 10 μM eBL than in those submitted to Cd only (Fig. 6F). These data indicate that the reduced hormone level in dwarf1 seedlings was responsible for the reduced sensitivity to Cd and correlated well with the results described above with eBL in the WT. The same kind of experiments were also performed with T-DNA insertion lines in DWARF4, which is involved in BR biosynthesis downstream of DWARF1 (Choe et al., 1998); and BRI1, which encodes a receptor kinase involved in BR perception (Li and Chory, 1997) (Fig. 7; see Supplementary Fig. S7 at JXB online). Results were comparable with those obtained with dwarf1 as Cd-induced root growth inhibition was reduced in dwarf4 and bri1 seedlings compared with WT, further demonstrating the involvement of BR signalling in the Cd response.

Fig. 6.

Enhanced tolerance to Cd of the dwarf1 seedlings is abolished by eBL addition. (A) Location of the T-DNA in the At3g19820 (DWARF1) gene in the Salk line N506932 and semi-quantitative RT-PCR confirming that the T-DNA insertion leads to no expression of the entire DWARF1 mRNA. Phenotypic analysis pointed to a typical BR-depleted phenotype. (B) Whole-plant phenotype of WT and dwarf1 seedlings in response to Cd stress is shown. Asterisks denote root length in dwarf1 significantly (P <0.05) different from the WT. (C) Root length of typical 10-d-old WT and dwarf1 plants measured after 6 d of exposure to Cd ranging from 0 to 100 μM. (D) Experiments were performed in parallel in the presence of salt to test the ability of dwarf1 seedlings root growth to be more reduced than the WT in response to stress. (E) Evaluation of dwarf1 sensitivity to Cd in the presence of eBL. Experiments were conducted as previously described, with Cd concentrations ranging from 10 μM to 25 μM and on plates supplemented with 0 or 1 to 1000 nM eBL. Data show the mean ±sem of a typical experiment performed on 45 seedlings. Asterisks in (B) indicate statistical difference from WT root length (P <0.05). In (C), (D), and (E) the asterisks denote the root length of Cd- or NaCl-treated plants statistically different from that of mock-treated plants. (F) Cd effect on leaves pigmentation of dwarf1 plants for different hormone concentrations. In (C), (D), and (E), data are expressed as % of the control for clarity purposes and graphs presenting the absolute root elongation values are shown in Supplementary Fig. S4 at JXB online. Statistical differences were evaluated by Student’s t test.

Fig. 7.

Evaluation of the Cd resistance of several BR-defective mutants. After growing on standard half-strength MS medium for 4 d, WT, dwarf1 (N506932), dwarf4 (N520761), and bri1 (N503371) seedlings were transferred on to 0 or 25 μM Cd. Root elongation was measured after 6 d of exposure. Absolute values are presented in Supplementary Fig. S7 at JXB online. Asterisks denote the root length of Cd-treated plants different (P <0.05) from that of mock-treated plants evaluated by Student’s t test.

Discussion

The aim of this work was to determine whether a hormone signalling pathway was predominantly involved in the plant response to Cd stress. Our results strongly suggest the existence of a transcriptional Cd-BR cross-talk and link for the first time the BR signalling pathway with the Cd-induced response in Arabidopsis. These experiments also demonstrate that a modulation of the BR content in Arabidopsis seedlings affects their response to Cd.

Cd and BR signalling pathways share common elements

It was hypothesized that, if a hormone signalling pathway was predominantly involved in the plant response to Cd, a correlation should be observed at the gene expression level between the two treatments. A bioinformatics method was developed that allowed us to compare the behaviour of 3022 Cd-regulated genes (Herbette et al., 2006) in response to eight different hormone treatments (Zimmermann et al., 2004) and show that gene expression changes in response to BR was the most similar to Cd-induced changes. To strengthen this analysis, it was confirmed that this similarity was statistically significant by showing that none of the 1000 data randomizations (Monte-Carlo method; Manly, 1997) could lead to such a close homology. Using the same approach, the two stress-related treatments—drought and a biotic stress—showed significant similarity with the hormones known to be involved in their response, for example, ABA and SA/MeJa, respectively (Gosti et al., 1995; Vijayan et al., 1998; Berger, 2002; Makandar et al., 2010; Yoshida et al., 2010; An and Mou, 2011), while they did not show significant similarity with eBL-triggered gene regulation. These results demonstrated the validity of the approach used here to highlight the signalling pathways involved in Cd response, and showed that the similarity between Cd and eBL treatment was not a general stress response but was specific to the metal.

Brassinosteroids are a class of sterol-derived hormones, often compared with the mammalian steroids. They are known to regulate cell elongation and division, vascular differentiation or senescence and, therefore, are essential for plant growth and development (Clouse and Sasse, 1998; Asami et al., 2005). During the last few decades, both BR biosynthetic and signalling pathways have been well characterized (Asami et al., 2005; Kim and Wang, 2010). BR level in planta is tightly controlled and feedback mechanisms have been described that regulate BR homeostasis (Goda et al., 2002; Mussig et al., 2002; Tanaka et al., 2005; Kim et al., 2006). In particular, the DWARF4 gene encodes a 22-hydroxylase that was proposed to be a rate-limiting step during hormone synthesis (Kim et al., 2006). When the BR level increases, DWARF4 expression is repressed, and conversely (Kim et al., 2006). Among the different genes analysed by qPCR analysis DWARF4 was down-regulated early in response to Cd, and this repression was sustained for up to 48 h for the highest metal concentration. This Cd-induced down-regulation of gene expression was observed for BR6OX as well, also known to mediate a negative feedback of the BR biosynthetic pathway upon BR increase (Goda et al., 2002; Mussig et al., 2002; Tanaka et al., 2005; Kim et al., 2006). These feedback mechanisms have been shown to require BR signalling elements (Mathur et al., 1998; He et al., 2005). Beside this feedback regulation, it was also shown that eight other BR reporter genes analysed by qPCR showed a BR like response in response to Cd treatment.

Both together, the transcriptome-wide analysis and the targeted study of the reporter genes, show that, in Arabidopsis, gene expression in response to Cd mimics a BR increase and that Cd exposure most probably triggers an activation of the BR signalling pathway. Importantly, however, BR profiling experiments did not show any change of BR level in response to the metal. However, subsequent experiments with seedlings exhibiting an altered level of BR challenged with different Cd concentrations confirmed a relationship between the BR signalling pathway and the response to Cd. This validates the existence of an interaction, in vivo, between Cd and BR. A very plausible hypothesis, therefore, is a possible BR-independent activation of the BR-signalling pathway by Cd, which is supported by numerous reports of a steroid-like activity of Cd in mammalian cells, recently reviewed by Byrne and co-workers (Byrne et al., 2009). They indeed refer to Cd as a metallohormone and describe numerous evidences of a direct activation of steroid receptor by Cd. In Arabidopsis, gene expression regulation in response to BR notably involves two transcription factors, BES1 and BZR1 (Kim and Wang, 2010). Both factors show specific and common target genes (Sun et al., 2010; Yu et al., 2011). BZR1 is the major transcription factor responsible for BR-regulated gene expression, including DWARF4 repression and other BR biosynthetic genes regulation (Sun et al., 2010). Three of the genes tested by qPCR in this study, At5g52640 (HSP83), At2g47550 (GA3.9), and At2g45210 (SAUR36), were recently characterized as BES1 target genes (Yu et al., 2011). These results suggest that Cd activation of the BR signalling pathway might happen before BES1 and BZR1 activation. Discovering the underlying mechanisms leading to the activation of BR signalling in plants in response to Cd should be the purpose of future investigations.

Cd sensitivity is modulated by BR level and activation of BR signalling

The previously mentioned activation of the BR signalling pathway may, in turn, affect plant response to Cd. Our data clearly demonstrate that providing eBL significantly enhances inhibition of root growth by Cd, while plants with a lower BR level appeared much less sensitive to the metal. While this proves a functional interaction between BR and Cd response, BRs are, however, known to tightly regulate root growth and any modification of the level of endogenous BR in the WT results in lowering the root elongation even in the absence of Cd. Our experiment performed with two other hormones (IAA and ABA), used at concentrations leading to a comparable root growth inhibition in the absence of Cd, showed that lowering the root length in Cd-free conditions did not affect the rate of Cd-induced root growth inhibition: 50–59% of inhibition by 25 μM Cd in mock-, IAA-, and ABA-treated plants versus over 80% of inhibition by 25 μM Cd when plants were treated with eBL. It also demonstrates that enhancement of the Cd-induced root growth inhibition by adding BR is specific to this hormone. IAA and ABA were chosen, among others in this assay because, in addition to BR, the results of our transcriptome-wide analysis highlighted a putative cross-talk between them and Cd responses (see Supplementary Fig. S1 at JXB online). Recent data (Yu et al., 2011) showed that genes involved in IAA, GA, and ABA signalling were overrepresented in BR responsive BES1 target genes which could explain why Supplementary Fig. S1 shows a significant cross-talk between Cd and the two hormones, ABA and IAA. Interestingly, neither IAA nor ABA (Fig. 5; see Supplementary Fig. S2 at JXB online) or GA (data not shown) had any effect on Cd -induced root growth inhibition. However, it cannot be excluded that GA, IAA, and ABA impact on other aspects, not studied in this work, of the plant response to Cd stress.

There was also a control for the smaller root growth rate of dwarf1 in Cd-free conditions and it was demonstrated that this could not lead to an apparent tolerance by just preventing the root growth from being further inhibited. This eventuality was first abolished by confirming an enhancement of the root growth inhibition of dwarf1 compared with the WT plants in a condition of expected hypersensitivity (100 mM NaCl) (Kagale et al., 2007). It was shown that the growth inhibition rate was higher in dwarf1 plants than in WT. In addition to this, it is important to note that Cd concentrations (50 μM and 100 μM) were also tested, leading to an absolute root length that was smaller in WT plants than in dwarf1 seedlings. This definitely confirmed that Arabidopsis thaliana dwarf1 plants are less sensitive to Cd than WT.

It is interesting to note that in our global transcriptomic analysis in Supplementary Fig. S1, no significant cross-talk was observed between Cd and SA or Cd and ethylene treatments, consistent with the fact that neither the Arabidopsis SA-deficient mutant nahG nor the ethylene insensitive mutant ein2 showed compromised response to Cd (Weber et al., 2006) as compared with the BR-deficient and the BR-insensitive mutants (dwarf1, dwarf4, and bri1, respectively; this work).

BR have been shown to enhance tolerance to biotic stresses such as bacterial or viral challenges as well as abiotic ones, such as cold, drought or salt stresses (Kagale et al., 2007; Xia et al., 2009; Divi and Krishna, 2009; Krishna, 2003), notably by enhancing HSP synthesis (Kagale et al., 2007) and H₂O₂ production (Xia et al., 2009). Brassinosteroids were also shown to protect against Cd toxicity in Brassica juncea, tomato cultivars, and Raphanus sativus and to reduce nickel toxicity in Triticum aestivum (Hayat et al., 2007; Hayat et al., 2010; Hasan et al., 2011;,Yusuf et al., 2011). They were even suggested as a biotechnological target for enhancing crop yield and stress tolerance (Divi and Krishna, 2009). While we were expecting to increase the resistance against Cd upon BR treatment in Arabidopsis thaliana, a marked specific hypersensitivity of plants when subjected to exogenous eBL was actually observed. It is important to note that, in our experimental conditions, dwarf1 seedlings that are BR deprived, are hypersensitive to NaCl, confirming the previously described enhanced tolerance to salt stress by eBL in Arabidopsis (Kagale et al., 2007), and demonstrating that this Cd-specific effect of eBL is new and strong enough to counteract the global protective role of BR against other stresses. The discrepancy observed between our work in Arabidopsis and the work described previously by others workers (Hayat et al., 2007, 2010; Hasan et al., 2011; Yusuf et al., 2011) could first come from the plant material itself: different species were tested at different ages (young seedlings in our work versus older plants in their work). Probably more important are the differences in the experimental procedures: eBL and Cd were provided at the same time in the in vitro medium of young seedlings, whereas they first provided the toxic metal in soil, or during germination, and then sprayed eBL on the leaves more than one week post-metal treatment. The global response that was followed was also quite different: while the focus was on root growth after 3 d of treatment, they mainly looked at enzyme activity in leaves and fresh weight after a longer period of incubation. They therefore focused on the effect of BR on the physiological response to Cd whereas our study revealed cross-talk between early signalling events common to BR and Cd.

Hypothesis regarding how modulation of BR signalling regulates plant response to Cd

How BR signalling affects plant sensitivity to Cd still requires further investigation. One hypothesis could be that activation of the BR signalling pathway decreases the expression of tolerance mechanism and/or increases the expression of sensitivity mechanisms. Changes in plant Cd content were investigated (see Supplementary Fig. S3 at JXB online; data not shown) as ion efflux across roots is well known to be a major tolerance mechanism in heavy-metal-resistant species (Verkleij et al.,2001), but it was observed that similar amounts of Cd were accumulated in the roots and shoots of Brz-, eBL-, and mock-treated plants, as well as dwarf1 versus WT seedlings. GSH and PC production was also investigated (see Supplementary Fig. S8 at JXB online) in Brz-treated plants compared with a mock treatment, as this biosynthetic pathway is one of the best characterized mechanisms that plants use to counter Cd toxicity (Cobbett et al., 1998; Cobbett, 2000). However, GSH and PC quantification by HPLC did not provide any significant evidence supporting the hypothesis that a BR-dependent modulation of their synthesis could explain the observed phenotypes (see Supplementary Fig. S8 at JXB online). Although modulation of BR content in plants drastically changes Cd response, the existence of a specific, essential but unknown mechanism allowing plants to modulate Cd toxicity, which would be controlled by the BR level, is rather unlikely. By contrast, it is proposed that the observed phenotypes could result from several, more or less important, side mechanisms which, taken together, may greatly affect the Cd response. It is indeed noticeable that BR level controls the expression of several genes encoding proteins that could modulate Cd sensitivity (see Supplementary Table S1 at JXB online). The characterization of the actual involvement of each of them is beyond the scope of this article but these results provide a rich set of data for future investigation of Cd, and, more generally, heavy metal, response and tolerance in plants. For instance, several plasma membranes localized P-type H⁺-ATPases are down-regulated upon BR treatment (AHA5, AHA11, ACA7). As Cd is known to substitute for many cations in biological processes, these pumps could represent one way of Cd elimination from the cytosol, repressed by BR treatment. This importance of decreasing cytosolic Cd is supported by the characterization of metal transporters (such as the Nramp family, AtMHX) whose activities, from the vacuole to the cytosol, represent a sensitivity factor with Cd stress (Thomine et al., 2000). In line with these observations, quantitative trait loci for metal tolerance in different plant species have been associated with P-type ATPases (Papoyan and Kochian, 2004; Courbot et al., 2007). AtIREG3 (IRON REGULATED3), a transporter belonging to the same family as AtIREG2, which has been shown to enhance metal tolerance (Schaaf et al., 2006) is also down-regulated by high eBL levels and up-regulated by Brz treatment. Interestingly, two metallothioneins (At3g15353, At3g09390) are expressed in a BR-dependent manner and could also explain the phenotype (Cobbett and Goldsbrough, 2002). So far, few proteins or protein families have been identified for their ability to trigger Cd/heavy metal sensitivity, and no particular candidate was observed that an up-regulation by eBL could lead to an enhanced Cd effect. This possibility, however, should not be excluded and further analysis will probably allow new components of the Cd tolerance/sensitivity in Arabidopsis to be identified.

Conclusion

Two phenomena have been described here that link BR signalling and Cd response: Cd exposure is very likely to activate the BR signalling pathway and artificial alteration of the BR content/BR signalling in planta leads to a modulation of the plant sensitivity to Cd. These findings are novel and first reported in plants, and they can provide new directions for better understanding the plant response to Cd and, eventually, engineering phytoremediation capable biotechnology. A working model is proposed (Fig. 8) linking these two aspects: upon Cd exposure, the BR signalling pathway is activated. As shown in this work, activation of the BR signalling pathway is able to influence the plant response to Cd. This Cd-induced activation of the BR-signalling pathway would therefore result in the modulation of the expression of genes able to affect the Cd response. Molecular actors of this BR-mediated hypersensitivity still have to be identified, which would allow one to investigate their role in Cd tolerance of WT plants, thus opening interesting opportunities for future research.

Fig. 8.

Hypothesis regarding how Cd response interacts with BR signalling and biosynthetic pathways. Following Cd exposure, the BR signalling pathway would be activated through a BR independent pathway (Cd-mediated direct activation) (A). The resulting cell response includes therefore a BR-triggered-like gene expression pattern (B) which leads to the up-regulation and down-regulation of many genes. It is proposed that, among them, some repressed genes could represent minor or major tolerance factors during Cd exposure (C), and/or that activated genes could enhance Cd toxicity (D). These effects would constitute a part of the Cd response, other mechanisms remaining BR-independent such as phytochelatins production or Cd root/shoot partition (E). Transcriptional repression of BR biosynthesis is observed in response to BR signalling activation (F).

We thank D Grando (CEA Grenoble, France) for her help in bioinformatics tools design, S Cuiné (CEA Cadarache, France) for his help in HPLC analysis, V Pautre (CEA Grenoble, France) for her help in real-time PCR analysis, and R Lombart-Latune and Dr C Dutilleul (CEA Grenoble, France) for their technical assistance. We are also grateful to Dr S Takatsuto (Joetsu University of Education, Japan) for supplying deuterium-labelled internal standards. We also want to thank Dr M Blasquez (University of Valencia, Spain) for discussion and advice, Dr J Kwak (University of Maryland, USA) for suggestions and availability, and Dr C Zubieta (ESRF, Grenoble, France) for careful reading of the manuscript.

Author notes

*

Present address: Plant Signal Transduction laboratory (J. Kwak), University of Maryland, 0219 Bioscience Research Building (#413), College Park, MD 20742, USA.