Understanding the cellular mechanisms of plant tolerance to mercury (Hg) is important for developing phytoremediation strategies of Hg-contaminated soils. The early responses of alfalfa (Medicago sativa) seedlings to Hg were studied using transcriptomics analysis.
A Medicago truncatula microarray was hybridized with high-quality root RNA from M. sativa treated with 3 μM Hg for 3, 6 and 24 h. The transcriptional pattern data were complementary to the measurements of root growth inhibition, lipid peroxidation, hydrogen peroxide (H2O2) accumulation and NADPH-oxidase activity as stress indexes.
Of 559 differentially expressed genes (DEGs), 91% were up-regulated. The majority of DEGs were shared between the 3 and 6 h (60%) time points, including the ‘stress’, ‘secondary metabolism’ and ‘hormone metabolism’ functional categories. Genes from ethylene metabolism and signalling were highly represented, suggesting that this phytohormone may be relevant for metal perception and homeostasis.
Ethylene-insensitive alfalfa seedlings preincubated with the ethylene signalling inhibitor 1-methylcyclopronene and Arabidopsis thaliana ein2-5 mutants confirmed that ethylene participates in the early perception of Hg stress. It modulates root growth inhibition, NADPH-oxidase activity and Hg-induced apoplastic H2O2 accumulation. Therefore, ethylene signalling attenuation could be useful in future phytotechnological applications to ameliorate stress symptoms in Hg-polluted plants.
Mercury (Hg) is a highly toxic heavy metal that occurs naturally in some environments, such as the mining district of Almadén, Spain. Several industrial activities also release large amounts of Hg that cause severe environmental problems and pose serious risks to human health because of its biomagnification in the trophic chain (Li et al., 2009). Hg2+ is the predominant chemical form of Hg bioavailable to plants in most soils (Heaton et al., 2005), which cause different toxic effects in exposed plants related to metabolic alterations that can lead to severe cell damage (Chen & Yang, 2012). Typical toxic effects include stunted plant growth (Cho & Park, 2000; Rellán-Álvarez et al., 2006), inhibition of photosynthetic activity (Patra et al., 2004) and decreased nutrient uptake (Patra & Sharma, 2000). One of the earliest responses of plants exposed to Hg is the occurrence of oxidative stress (Ortega-Villasante et al., 2005) caused by the accumulation of reactive oxygen species (ROS), which ultimately leads to cell death in a matter of minutes as observed in Hg-challenged Medicago sativa root epidermal cells (Ortega-Villasante et al., 2007). Induction of the oxidative burst coupled with the accumulation of Hg results in the alteration of the activity of several antioxidant enzymes, such as ascorbate peroxidase and superoxide dismutase, which are relevant components of the ROS scavenging system (Rellán-Álvarez et al., 2006; Zhou et al., 2008). Concurrently, the metabolism of glutathione (GSH), a key antioxidant in metal homeostasis (Seth et al., 2012), is readily altered by Hg stress, either by the synthesis of phytochelatins (Carrasco-Gil et al., 2011) or by the accumulation of oxidized GSH (GSSG, Ortega-Villasante et al., 2007). In addition, glutathione reductase (GR), which mediates the conversion of GSSG to GSH (Sobrino-Plata et al., 2009), is extremely sensitive to Hg and most likely contributes to oxidative stress.
During the past few years, the objective of intensive research has been to characterize the direct effects of heavy metals in order to understand the mechanisms underlying metal perception, defence responses and tolerance (Hernández et al., 2012). Acclimation to metal stress requires the coordination of complex physiological and biochemical processes, because it is one of the most relevant aspects that contribute to the occurrence of global changes in gene expression (DalCorso et al., 2010). The transcriptional profile of plants under heavy metal stress has recently been studied in short-term experiments. These experiments are primarily performed with model plants treated with cadmium (Cd) in which transcription factors related to phytohormone signalling or belonging to the WRKY, bZIP, and MYB families were up-regulated (Weber et al., 2006; Van De Mortel et al., 2008). In addition, advanced transcriptomics using a massive sequencing approach revealed that Solanum nigrum seedlings, a Cd accumulator, overexpressed a group of genes related to metal transport and antioxidant processes after only 24 h of Cd treatment (Xu et al., 2012). Therefore, relevant information has recently been gained regarding redox homeostasis, metal chelators, metal transporters, and cis-regulatory elements involved in Cd tolerance in plants (Thapa et al., 2012).
Several genes were differentially overexpressed in Arabidopsis thaliana plants exposed to long-term Hg treatment using a suppression subtractive hybridization technique, among them jasmonate-induced genes, which were primarily related to environmental stress (Heidenreich et al., 2001). Similarly, pea plants exposed to Hg for only 6 h also showed an increase in the genes involved in stress responses, such as antioxidative reactions (Sävenstrand & Strid, 2004). Short-term treatment of alfalfa seedlings with Hg for 24 h also caused the overexpression of several antioxidant-related genes involved in GSH metabolism, such as glutathione synthetase and homoglutathione synthetase (Ortega-Villasante et al., 2007). Recently, a short-term transcriptomics analysis of Medicago truncatula seedlings exposed to 10 μM Hg was performed using massive tag sequencing (Zhou et al., 2013). In this analysis, the majority of differentially expressed genes (DEGs), such as the mitochondrial phosphate transporter, glutathione S-transferases, flavonoid glycosyltransferase and prenyltransferase/squalene oxidases, corresponded to energy conversion and secondary metabolism. By contrast, Cd and Hg exert different phytotoxic effects on plants grown under similar environmental conditions (Sobrino-Plata et al., 2009), suggesting that it is difficult to propose molecular mechanistic models of plant heavy metal tolerance based on partial information because the responses are highly varied depending on the metal type and growth conditions (Thapa et al., 2012).
To better understand the early plant responses to Hg toxicity, a comprehensive study using physiological, cellular and genomic approaches was performed in a microscale experimental setting (Ortega-Villasante et al., 2007). The early transcriptional response to Hg was studied in M. sativa plants treated for 3, 6 and 24 h using oligonucleotide microarrays. A relatively large number of genes were differentially expressed, and among these was a group of ethylene-responsive genes. Involvement of the ethylene signalling pathway in Hg perception was further studied using a functional approach with alfalfa plants pretreated with the ethylene signalling inhibitor 1-methylcyclopropene (1-MCP; pharmacological assay) and an Arabidopsis thaliana ein2-5 mutant (genetic test).
Materials and Methods
Sterilized alfalfa seeds (Medicago sativa var. Aragón) were germinated on plates and then transferred to a microscale hydroponic system (see Supporting Information Methods S1, Fig. S1). The seedlings were treated with different concentrations of Hg (3, 10 or 30 μM HgCl2; Merck, Whitehouse Station, NJ, USA) and/or 10 μM 1-MCP (AgroFresh Inc., Philadelphia, PA, USA) as described in each experiment, and were carefully collected to minimize mechanical stress at different time intervals (3, 6 and 24 h after treatment; Methods S1). Metal doses were selected considering previously known thresholds of toxicity (Ortega-Villasante et al., 2007). Before freezing, the plants were rinsed with Na2EDTA and deionized water, and the seedling length was measured.
Wildtype (Col-0), mutant (ein2-5 and AtrbohD/F) and transgenic over8expressor 35S::AtrbohD Arabidopsis thaliana seeds were sterilized and sown on vertical plates containing Murashige–Skoog (MS) sucrose agar media (Duchefa-Biochemie, Haarlem, the Netherlands). Growth inhibition assays were performed in 5-d-old plantlets that were transferred to a new plate supplemented with 0.30 μM HgCl2, and root growth was followed for 24, 48 and 72 h (Methods S1; Fig. S2). Root length was measured using ImageJ software in photographs taken of the plates.
A hydroponic growing system (Araponics; Tocquin et al., 2003) was also used with Arabidopsis plantlets (Col-0, ein2-5) first grown for 5 wk in control nutrient solution and then treated with 1.5 μM HgCl2 for 24 and 48 h (Methods S1; Fig. S3). The roots were rinsed with Na2EDTA and deionized water, collected and frozen at −80°C.
Extracellular H2O2 and lipid peroxidation
Extracellular H2O2 release was measured in apical root segments of untreated alfalfa seedlings (1 cm long) equilibrated in MS medium titrated with 2 mM MES at pH 6.0 in the dark for 1 h (Ortega-Villasante et al., 2007). Root segments were placed individually in a 96-well microtitre plate containing MS-MES with different concentrations of Hg and/or 10 μM 1-MCP. Root H2O2 production was also measured in 10-d-old intact Col-0, ein2-5, AtrbohD/F and 35::AtrbohD Arabidopsis seedlings placed individually and carefully (avoiding shoot interference) as described earlier in MS-MES medium with 0 (control) and 0.2 μM HgCl2. At least 12 individual replicates per treatment were analysed. Amplex Red (Molecular Probes, Eugene, OR, USA; 5 μM final concentration) was added immediately before fluorescence recording at λexc = 542 nm and λem = 590 for 6 h every 5 min (Synergy HT Biotek, Winooski, VT, USA). Fluorescence percentage was calculated relative to time 0, and representative data of five independent experiments are shown. Lipid peroxidation was measured as the malondialdehyde concentration, as described by Ortega-Villasante et al. (2005).
Mercury tissue concentration
Whole alfalfa seedlings were collected, washed in deionized water, rinsed briefly in 10 mM Na2EDTA at pH 6.0, and dried for 72 h at room temperature. After acid digestion, Hg was analysed using an AMA-254 analyser (LECO, St Joseph, MI, USA) equipped with a golden amalgam (Sobrino-Plata et al., 2009).
RNA extraction and quantification
Total RNA from the M. sativa roots was isolated with TRI Reagent (Ambion) as described by Portillo et al. (2006). The RNA was cleaned using in-column DNAse treatment with the RNeasy Mini Kit (Qiagen). The integrity of the extracted RNA was determined with an Agilent 2100 Bioanalyzer equipped with a RNA 6000 Nano LabChip Kit (Agilent Technologies, Santa Clara, CA, USA). The algorithm RNA integrity number (RIN) was calculated as the RNA quality of three independent biological replicates (Schroeder et al., 2006).
DNA microarray hybridization
The transcriptional profile analysis was performed by hybridization of M. sativa RNA to the M. truncatula Mt16koli-plus 70 mer microarray (Küster et al., 2007). Eight independent biological experiments were performed per metal dose (control and 3 μM HgCl2) and time of exposure (3, 6 and 24 h); in total 48 independent RNA samples were prepared (Methods S1). One RNA ‘independent hybridization replicate’ was prepared from two equivalent RNA samples pooled in equimolar ratios, and four hybridization replicates were performed following a two-colour array design (control vs Hg-treated in a flourophore-swap manner) per treatment time (i.e. 12 chips were analysed in total). Hybridization data are available at Array Express (http://www.ebi.ac.uk/arrayexpress/; accession no. E-MEXP-3876). To control the false discovery rate (FDR), P-values were corrected using the method of Benjamini & Hochberg (1995). Gene expression differences were significant at an FDR < 0.01 and only fold-changes (FCs) greater than two (up-regulated) or less than two (down-regulated) were selected. Hierarchical clusters were calculated using the Multi Experiment Viewer software by TIGR (Saeed et al., 2003) with the Pearson uncentred metric distance and a complete linkage. Functional categories of the DEGs were obtained from MapMan (Thimm et al., 2004). Enriched categories with a large number of DEGs with respect to the Medicago database background were selected and analysed with the χ2 test and classified by significance level (P < 0.05).
Quantitative RT-PCR was performed with the total RNA from alfalfa and Arabidopsis roots of five independent biological replicates to synthesize the complementary DNA strand (cDNA). Oligonucleotide primers were designed based on M. truncatula sequences and homologous sequences of A. thaliana using the Primer Express software (Applied Biosystems, Foster City, CA, USA; Table S1). Gene expression quantification was performed using the relative method (Livak & Schmittgen, 2001). The glyceraldehyde 3-phosphate dehydrogenase gene (GAPDH) was used as the reference because it showed steady expression over the time course of the experiment.
NADPH-oxidase enzymatic activity
The in-gel NADPH-oxidase enzymatic activity was determined after the separation of protein extracts by nondenaturing polyacrylamide electrophoresis. The protein concentration of the extracts was determined using the Bio-Rad Protein Assay reagent (BioRad) with BSA as the standard. Adequate loading (10 μg) was corrected using sodium dodecyl sulphate polyacrylamide denaturing gel electrophoresis (SDS-PAGE) and Coomassie Blue staining (Laemmli, 1970). NADPH-oxidase staining solution was prepared in 50 mM Tris–HCl buffer (pH 7.4) supplemented with 0.5 mg ml−1 nitrotetrazolium blue chloride, 0.2 mM NADPH, 1 mM CaCl2 and 0.2 mM MgCl2 (Sagi & Fluhr, 2001). Randomly chosen sample aliquots were preincubated with 100 μM diphenyleneiodonium chloride (DPI) as negative controls to identify genuine NADPH-oxidase bands. The optical density (analysed with ImageLab™ software; Bio-Rad) of two major nonsaturated NADPH-oxidase activity bands clearly inhibited by DPI was measured, and the FC intensity calculated relative to the control (0 μM HgCl2). Two independent experiments with two biological replicates were performed.
Analyses of variance between treatments were done using ANOVA and Tukey's post-hoc test as indicated in the figure legends. Gene expression statistical analysis of qRT-PCR was accomplished using the BootsRatio web tool (Clèries et al., 2012).
Physiological parameters of stress induced by Hg
Root growth inhibition is a very sensitive parameter of metal toxicity in seedlings (Ortega-Villasante et al., 2005). Root growth was inhibited concomitantly with the Hg dose (0, 3, 10 and 30 μM) and time of exposure (3, 6 and 24 h; Fig. 1a,b). Root growth inhibition was much less significant in seedlings treated with 3 μM Hg than in those treated with 10 and 30 μM Hg at all the measured time intervals (two times lower) with minor differences between the last doses (Fig. 1b). Mercury rapidly accumulated in the plants after 3 h of treatment, followed by a saturation trend for 10 and 30 μM Hg. A minimum accumulation of Hg of c. 0.5 μmol Hg g−1 DW was required to produce significant growth inhibition. This degree of Hg accumulation was quickly reached after 3 h in the plantlets exposed to 10 and 30 μM Hg, which accumulated 30 and 45%, respectively, of the total Hg found after 24 h (Fig. 1c). Concurrent with growth inhibition, there was a sharp increase in extracellular H2O2 produced in vivo in the roots exposed to 3 and 30 μM Hg (Fig. 1d), which increased during the 6 h treatment. H2O2 release did not increase with Hg concentration from 3 to 30 μM Hg, indicating that apoplastic H2O2 generation was saturated at the lowest dose (Fig. 1d). A good index of oxidative stress damage is the peroxidation of membrane lipids quantified as the accumulation of malondialdehyde (MDA). Lipid peroxidation augmented significantly (P <0.05) for both the highest doses of Hg and prolonged Hg exposure (Table 1).
Table 1. Lipid peroxidation in the roots and shoots of Medicago sativa exposed to different doses of HgCl2 (0, 3, 10 and 30 μM) for 3, 6 and 24 h
Lipid peroxidation (nmol MDA g−1 FW)
The values represent the means of three independent experiments ± SD. The values with different lowercase letters are significantly different from the control conditions with P <0.05 according to the Tukey test. MDA, malondialdehyde.
12.23 ± 1.36 a
9.23 ± 1.52 a
10.19 ± 2.64 a
12.89 ± 1.55 a
15.16 ± 2.51 ab
13.26 ± 2.71 ab
21.00 ± 1.77 ab
24.72 ± 2.08 bc
18.36 ± 2.14 bc
27.50 ± 2.03 bc
29.25 ± 3.90 c
21.57 ± 2.36 c
24.80 ± 2.07 a
21.21 ± 1.72 a
17.69 ± 1.03 a
23.38 ± 3.71 a
23.26 ± 0.99 a
18.71 ± 3.09 a
23.04 ± 1.63 a
21.31 ± 2.40 a
15.79 ± 1.44 a
22.30 ± 1.21 a
24.02 ± 1.46 a
20.00 ± 1.87 a
Critical parameters to evaluate for holistic approaches of the transcriptional analyses of plants subjected to stress are the quality and integrity of the RNA obtained (Portillo et al., 2006). Electrophoretic fluorescence profiles were obtained for the RNA extracted from the alfalfa roots treated with 0, 3, 10 or 30 μM Hg after 3, 6 and 24 h, and the RIN was calculated (Schroeder et al., 2006). The partial degradation of RNA occurred in the seedlings treated with doses > 10 μM Hg, especially in those treated with 30 μM Hg (Fig. 2). This caused the RIN to decrease, with values < 9 representative of the degraded RNA. After 24 h of exposure to 30 μM Hg, the RNA was dramatically altered, and insufficient RNA was extracted for further analysis (data not shown). Hg may influence the total RNA population, leading to analytical artefacts for transcriptomics analysis, and only unaltered RNA extracted from the plants exposed to 3 μM Hg for 3, 6 and 24 h was used to characterize the early transcriptional response of plants under Hg stress; plants that were readily responding in terms of growth inhibition and H2O2 release.
Transcriptomics profiling of early responses to Hg
The whole-genome transcriptional profile was obtained from M. sativa seedlings exposed to 3 μM Hg after 3, 6 and 24 h of treatment as compared with nontreated plants using two-colour 70-mer microarrays from M. truncatula (Küster et al., 2007). Of the 16 470 M. truncatula unigenes represented in the tested DNA chip, a total of 559 genes (3.4% of the genes in the chip) were differentially expressed at all time points (510 up-regulated and 49 repressed) with FDR cutoff < 0.01 and FC criteria of ± 2. For simplicity, the terms ‘induced/up-regulated/overexpressed’ and ‘repress/down-regulated’ are used throughout the text to define transcript abundances higher or lower than the control, respectively.
A hierarchical cluster analysis (HCL) was performed to characterize the expression patterns of the DEGs after Hg exposure at the three sampling time points (Saeed et al., 2006). A large proportion of DEGs after 3 and 6 h of treatment with 3 μM Hg followed a similar expression pattern, which clustered together with a good bootstrap value after 1000 interactions. By contrast, the expression profile of 24 h-exposed seedlings clustered separately, indicating obvious differences in the global gene expression patterns at 24 h (Fig. 3a). A vast majority of DEGs at 3 and 6 h were induced (401 and 423 genes, respectively), with a minor number of repressed genes (11 and 36 genes, respectively, representing 2.7 and 7.8% of the total DEGs; Fig. 3b). The number of common up-regulated genes after 3 and 6 h of treatment accounted for > 60% of the total induced DEGs. The proportion of repressed genes at 24 h was increased relative to the up-regulated genes (20%; Fig. 3b). The Venn diagram also shows that the number of common repressed or induced genes at 3 and 6 h is significantly > 24 h, whereas 6 h had the greatest number of exclusive genes. A comprehensive list of DEGs at all time points is given in Table S2.
The DEGs were also grouped into four clusters based on their expression profiles along the time of exposure by K-means analysis (KMC; Fig. 3c). Cluster 1 included 159 early overexpressed genes (higher expression at 3 h), which decreased in expression after 24 h. Cluster 2 grouped 49 DEGs that were repressed at all sampling points. Cluster 3 represented 93 genes that had the highest expression after 6 h of treatment. Cluster 4 showed a similar pattern to that of cluster 1 but with more steady expression up to 24 h of exposure.
The expression pattern observed by microarray hybridization was validated by qRT-PCR using the set of primers described in Table S1. A total of 28 independent genes were tested at the different treatment intervals with 3 μM Hg, which were selected among those with similar expression profiles and high FC values. These genes covered several functional categories after MapMan analysis. qRT-PCR revealed similar transcriptional patterns to those in the microarray hybridization experiments for the induced and repressed genes except for particular time points for the three genes marked in yellow boxes in Table 2. Therefore, the data obtained from the microarray analysis were consistently validated by qRT-PCR.
Table 2. Transcriptional profile (log2 expression ratio values (M)) of selected genes using quantitative reverse transcription polymerase chain reaction (qRT-PCR) (PC) and microarray hybridization data (Arr) of Medicago sativa seedlings treated with 3 μM HgCl2 for 3, 6 and 24 h
The qRT-PCR values are relative to the control samples normalized to endogenous glyceraldehyde 3-phosphate dehydrogenase gene (GAPDH) expression and were obtained from three independent biological replicates. Differential expression between the qRT-PCR and array results is shown in the yellow boxes. Over-expressed and down-regulated genes are shown respectively in red and green boxes. *Differentially expressed genes in the array (false discovery rate (FDR)< 0.01) and the statistical differences on gene expression in qRT-PCR experiments (P < 0.05) using the BootsRatio web tool (http://regstattools.net/br) .
TC77005 (sp|P2165) Tonoplast intrinsic protein root-specific RB7-5A; Nicotiana tabacum
We grouped the DEGs into functional categories using MapMan (Thimm et al., 2004). The significant differences for the expression profiles followed the Benjamini & Hochberg correction (see Tables S3, S4). The majority of DEGs (80%) were allocated to 16 categories, and the distribution varied between times of treatment. ‘Secondary metabolism’, ‘stress’ and ‘protein’ were the DEG categories with a higher collective expression pattern (asterisks in Fig. 4), particularly after 3 h of treatment. The functional families ‘secondary metabolism’, ‘miscellaneous’, ‘stress’ and ‘protein’ were highly represented in the 3 and 6 h Hg-treated alfalfa, and the number of genes in these categories decreased at 24 h (Fig. 4a–c). Notably, ‘stress’ and ‘secondary metabolism’ included a high proportion of induced genes related to biotic and abiotic stresses, such as those encoding several pathogenesis-related proteins, chitinases, endoglucanases and heat shock-related proteins, and genes involved in the production of phenylpropanoids, lignins, lignans and flavonoids (Fig. S4). Four NADPH-oxidase/Rboh homolog genes were transiently overexpressed after 3 h of treatment – expression that decayed with time of exposure (Table S2). The ‘protein’ category included proteasome and ubiquitin E3 ligase SCF box-related genes that were up-regulated after 3 and 6 h of treatment.
A second analysis of the relative DEG-enriched functional categories based on the χ2 test also revealed interesting results (Fig. 4d–f). ‘Metabolism’ was highly represented in the seedlings exclusively treated for 3 and 6 h. In this group, we identified genes encoding alternative oxidases and alternative NADH-dehydrogenases that function in the mitochondrial electron transport chain. Sulphur metabolism-related genes were only overrepresented in the 6 h Hg-treated seedlings, which included up-regulated ATP sulphurylase, adenosine 5′-phosphosulphate reductase and sulphite reductase. A completely differentiated pattern was observed in the ‘transport’ category at 3, 6 and 24 h of Hg treatment. At 3 h, there were no overrepresented DEGs, and after 6 h, there was up-regulation of various genes encoding ABC transporters and repression of a nitrate transporter NRT2, whereas there was only the repression of two genes of the NRT2 family after 24 h. Other overrepresented up-regulated DEGs grouped in the ‘miscellaneous’ category at 3, 6 and 24 h, including a relevant number of glutathione S-transferases.
Detailed analysis of the genes involved in ‘hormone metabolism’ was performed, because phytohormones may contribute to heavy metal stress perception and may be involved in root growth inhibition. All DEGs were distributed primarily into three hormone-related groups: jasmonate, auxin and ethylene. In particular, the ‘ethylene-related group’ contained the highest number of DEGs with high FCs during the earliest hours of treatment. Among these ethylene-related DEGs, we found up-regulated genes involved in ethylene biosynthesis, such as 1-aminocyclopropane carboxylic acid (ACC) synthase (ACCS) and ACC oxidase (ACCO), transcription factors AP2 and ERF1, and downstream genes such as ER66 and ethylene-induced esterase (Fig. S5).
Ethylene as a key regulator in the early response to Hg
Functional experiments were performed to confirm the participation of ethylene in the early response to Hg, a phytohormone that plays a central role in the responses of plants to different types of abiotic stresses (Mittler, 2006). This followed a pharmacological approach using the ethylene receptor inhibitor 1-MCP in alfalfa seedlings and a genetic approach using the ethylene-insensitive A. thaliana mutant ein2-5. Root growth inhibition, a reliable toxicity index, was significantly lower in alfalfa seedlings pretreated with 10 μM 1-MCP (no inhibition after 6 h of exposure to 3 μM Hg; Fig. 5a) or in ein2-5 (50% diminution of Col-0 after 2 and 3 d of treatment with 0.3 μM Hg, Fig. 5b). No differences in Hg concentration were detected between alfalfa seedlings incubated or not with 1-MCP (data not shown).
To confirm the participation of ethylene signalling in Hg-induced stress responses, expression of the ethylene-related genes ERF1, AP2 and ACCO was investigated in 1-MCP-pretreated alfalfa and ein2-5 by qRT-PCR (Fig. 6), using primers designed after a detailed phylogenetic analysis of M. truncatula and Arabidopsis gene homologues (Fig. S6). The accumulation of ERF1 transcripts decreased in 1-MCP-pretreated alfalfa exposed to 3 μM Hg for 3 and 6 h (c. 50% decrease of the control; Fig. 6a). However, the 1-MCP did not interfere with ACCO and AP2 expression (Fig. 6b,c). Similar results were observed in Arabidopsis, because ERF1 up-regulation induced by 1.5 μM Hg in Col-0 plants was completely blocked in the ein2-5 mutants (Fig. 6d), whereas the expression responses were lower for AP2 and ACCO (Fig. 6e,f).
The production of extracellular H2O2 is an index of early oxidative stress induced by Hg and depends on plasma membrane NADPH-oxidases (Ortega-Villasante et al., 2007). We tested the role of ethylene signalling in apoplastic H2O2 evolution in 1-MCP-pretreated alfalfa and ein2-5 exposed to Hg (Fig. 7a), because it is known that ethylene mediates the production of ROS under different stresses (Mersmann et al., 2010). In alfalfa, the release of H2O2 increased during the first 90 min when treated with 3 and 30 μM Hg. By contrast, extracellular H2O2 was significantly lower in plants pretreated with 1-MCP and exposed to 3 μM Hg than in plants treated only with 3 μM Hg. 1-MCP had a much more moderate effect in the 30 μM Hg-treated seedlings, most likely because of the extremely toxic effects at this very high dose. In-gel NADPH-oxidase activity increased in the seedlings exposed to 3 μM Hg for 6 h, but this enzymatic activity was similar to the control when 1-MCP was added to the culture medium (Fig. 7c). Involvement of NADPH-oxidases in apoplastic H2O2 release triggered by Hg was confirmed by studying the double AtrbohD/F mutant (Torres et al., 2002) and an Arabidopsis transgenic line overexpressing AtrbohD (harbouring a 35S::AtrbohD transgene; Torres et al., 2005). Production of H2O2 was reduced in the ein2-5 ethylene-insensitive mutant under Hg stress (Fig. 7c), and decreased drastically in the AtrbohD/F double mutant (Fig. 7c). Interestingly, transgenic seedlings overexpressing AtrbohD treated with Hg suffered an abrupt increase in apoplastic H2O2 generation (Fig. 7c).
Clear signs of oxidative stress after Hg exposure were similar to that described for different plant species (Cho & Park, 2000; Cargnelutti et al., 2006; Rellán-Álvarez et al., 2006; Zhou et al., 2008). Our results showed higher lipid peroxidation and apoplastic H2O2 production in Hg-treated alfalfa seedlings grown in a microscale hydroponic system, which was indicative of an oxidative burst, in agreement with previous results (Ortega-Villasante et al., 2007). Growth inhibition was also a sensitive marker of Hg toxicity that paralleled metal accumulation and was significant at the lowest doses (3 μM for 6 h) and highly significant at doses above 10 μM for 3 h (Fig. 1). Damage to alfalfa seedlings treated with the highest doses of Hg and long exposure time was also reflected in the RIN parameter of RNA stability. Although RNA quality is routinely validated prior transcriptomics studies, it is strongly recommended this preliminary assessment in experiments in plant material subjected to highly toxic substances (such as Hg), which may lead to artefacts. Therefore, we confined our transcriptional study to plants exposed to the lowest dose of Hg (3 μM).
Transcriptomics analyses of plants exposed to metals remain scarce, although several analyses have been performed for short-term Cd-treated Arabidopsis (Herbette et al., 2006; Weber et al., 2006) and Thlaspi (Van De Mortel et al., 2008), As-treated Arabidopsis (Abercrombie et al., 2008) and rice (Norton et al., 2008), Cu-treated Arabidopsis (Weber et al., 2006; Sudo et al., 2008) and Hg-treated M. truncatula (Zhou et al., 2013). The novelty of our study is the detailed characterization of the early response of M. sativa treated with low doses of Hg. We observed that 459 genes were differentially expressed after 6 h of treatment with 3 μM Hg, even when the seedlings did not show severe stress symptoms. This suggests that transcriptional changes occurred well before extensive cell damage was detected. Notably, the number of DEGs decreased sharply to 62 genes after 24 h of exposure, indicating a timescale-specific response. This time-dependent response is typical of other transcriptomics studies, because time of exposure is an important parameter that must be well established (Herbette et al., 2006).
The distribution of DEGs into functional categories by MapMan helped to identify cellular processes that may participate in the early response to Hg. Among the qRT-PCR-validated genes, a group of oxidative stress-related genes already characterized in Cd- and Hg-treated plants, including those encoding cytochrome P450 enzymes, glutathione S-transferases and quinone oxide reductases (Weber et al., 2006; Zhou et al., 2013), was identified. In addition, other up-regulated genes were observed that corresponded to ABC-type transporter proteins, which may participate in metal detoxification similar to that described in the presence of Zn (Moons, 2003), Cd (Bovet et al., 2005), As (Norton et al., 2008) or Hg (Zhou et al., 2013).
Several induced genes classified in the ‘secondary metabolism’ group encoded enzymes that control the synthesis of flavonoids, phenylpropanoids and monolignols (i.e. chalcone synthases), which are known antioxidants under stress conditions or components of cell walls. These phenol metabolites accumulate in plants exposed to Cd and are hypothesized to enhance cell wall stiffness, which may contribute to the inhibition of root elongation observed under conditions of heavy metal toxicity (Van De Mortel et al., 2008). The early transcriptional activity of the phenyl propanoid pathway occurred in parallel to the oxidative burst detected under Hg stress caused by apoplastic H2O2 accumulation, which is also required for cell wall polymer cross-linking via peroxidase activity (Zhou et al., 2007). Another interesting group of DEGs was the ‘protein’ category, which included regulatory proteins of the proteasome machinery, such as the E3 SCF ubiquitin ligases. The up-regulation of these genes after 3 and 6 h exposure to Hg may indicate an increase in protein ubiquitination, which is known to occur in plants under Cd stress (Djebali et al., 2008) and may exert early post-translational regulation of proteins.
Notably in the ‘stress’ category, there were a significant number of induced genes related to ‘biotic stress’ and ‘heat shock’. Among the biotic stress-related genes, different pathogen-related proteins, such as xyloglucan endoglucanase inhibitor protein and class IV chitinase, were correlated with expression changes in several WRKY transcription factors (Dong et al., 2003), and these proteins may regulate the plant stress response (Chen et al., 2012a). There was a strong induction of genes coding a group of small heat shock proteins (17.7, 17.9 and 18.2 kDa), HSP70, HSP83 and the RHSF2 transcription factor in the alfalfa seedlings exposed to Hg (see Table 2). Therefore, metals promote the expression of these genes in a manner similar to other environmental stresses, such as heat, whenever protein stability may be compromised (Swindell et al., 2007; Agudelo-Romero et al., 2008). Interestingly, there is parallelism between the transcriptional responses to biotic-stressed plants and heavy metal-stressed plants, in which several molecules, including transcription factors, kinases, phytohormones and ROS, are key players (Cruz-Ortega & Ownby, 1993; Pitta-Alvarez et al., 2000; Fujita et al., 2006).
Among the repressed genes validated by qRT-PCR, a relevant group encodes aquaporins of the TIP (TobRB7-5A like, Table 2) and PIP (ZmPIP2-7 like; Table S2) large families. It is well known that Hg blocks various classes of aquaporins and affects water balance in Hg-stressed plants (Javot & Maurel, 2002). Similarly, Norton et al. (2008) found that several aquaporin genes were down-regulated in rice seedlings exposed to As (V). The inhibition of these proteins and the consequent closing of the water channels disturb the water balance of the plants, which causes root dehydration and cell shrinkage and deformation, symptoms that appeared in Hg-treated alfalfa plants (Ortega-Villasante et al., 2005; Carrasco-Gil et al., 2013).
Notably, the functional groups ‘metabolism’, ‘sulphur assimilation’, ‘transport’ and ‘hormone metabolism’ (Fig. 4) were clearly overrepresented with respect to the Medicago microarray database background. Among others, the alternative-oxidase (AOX) and external NADH-dehydrogenase genes of the mitochondrial electron transport machinery were up-regulated after 3 and 6 h of treatment with 3 μM Hg. Overexpression of AOX was detected in maize mitochondria under Cd-stress conditions (Nocito et al., 2008) and most likely contributes to energy dissipation at the mitochondrial electron transport chain to control aberrant redox reactions that may cause the accumulation of ROS (Heyno et al., 2008). Another protein group that was highly represented in the DEGs corresponded to ‘sulphur assimilation’, which was only significantly overexpressed after 6 h of treatment with Hg. This was consistent with the results obtained by Zhou et al. (2013). Genes that encode enzymes such as ATP sulphurylase, APS reductase and sulphite reductase, which control the first steps of sulphate assimilation, are also overexpressed under Cd stress conditions (Herbette et al., 2006). It is hypothesized that S-depletion caused by PC synthesis from GSH is responsible for activation of the S-assimilatory pathway (Ernst et al., 2008). The χ2 analysis revealed the down-regulation of the NRT2 nitrate transporter genes under Hg stress conditions with increased Hg exposure times (6 and 24 h). It is known that inhibition of nitrate transport to the aerial portion of A. thaliana plants is important for Cd tolerance through down-regulation of the NRT1 and NRT2 genes (Chen et al., 2012b). Accumulation of nitrate in the roots may be part of the metabolic changes that occur in response to different endogenous and environmental factors in Lotus japonicus (Criscuolo et al., 2012), similar to the pattern detected under toxic metal stress conditions.
Special attention was focused on the ‘hormone metabolism’ category because signal processes involved in perception and homeostatic adjustment to Hg stress may be mediated through hormone signalling (Rodríguez-Serrano et al., 2009; Hernández et al., 2012). In our study, the most abundant hormone-related genes included members of the jasmonic acid (JA), auxin and ethylene signalling and metabolic pathways. The expression of genes participating in JA synthesis was noticeably higher in Hg-treated alfalfa seedlings and included the 12-oxo-phytodienoic acid reductase (OPRs) gene family, which consists of enzymes that catalyse the β-oxidation reactions involved in JA production. Incidentally, OPR up-regulation may also contribute to the early-response mechanisms of plants exposed to several biotic stresses, such as wounding, drought or heavy metals (Agrawal et al., 2004). Similarly, JA accumulation occurred after 7 h exposure to Cd and Cu in A. thaliana (Maksymiec et al., 2005). Auxin glucosyl transferases that are involved in auxin conjugation and homeostasis were also induced as described previously (Douglas-Grubb et al., 2004). In rice, auxins were involved in the early responses of the plants to Cd, most likely in a process mediated by H2O2 (Zhao et al., 2012).
Special attention was directed towards ethylene, as several classes of ethylene-related genes were overexpressed under Hg stress. Ethylene is synthesized in plants subjected to various types of biotic and abiotic stresses and is responsible for the induction of cell senescence (Wang et al., 2002). In particular, plants accumulating heavy metals, such as Cd or Cu, produce ethylene at a higher rate than controls (Arteca & Arteca, 2007; Rodríguez-Serrano et al., 2009). Overexpression of ERF1 and AP2 was accompanied by the significant inhibition of alfalfa root growth, which is a common negative effect of Cd, Cu and Al accumulation in plants (Lequeux et al., 2010; Sun et al., 2010). Therefore, it is feasible that root growth inhibition observed in Hg-treated plants is part of the ethylene-mediated responses (Swarup et al., 2007).
The role of ethylene in the early responses of plants to Hg was further studied in functional experiments using alfalfa preincubated with the ethylene inhibitor 1-MCP and the Arabidopsis ethylene-insensitive ein2-5 mutant. As hypothesized, the roots of alfalfa pretreated with 1-MCP and ein2-5 grew longer when exposed to moderate concentrations of Hg (Fig. 5). No differences in Hg concentration were detected between seedlings grown in the presence or absence of 1-MCP, which is possibly explained by the preferential apoplastic movement (i.e. nonmetabolic uptake) of Hg in roots (Carrasco-Gil et al., 2013). Similar behaviour was observed in Al-polluted Arabidopsis, because the ethylene synthesis inhibitors Co2+ and aminoethoxy vinylglycine (AVG) and the ethylene signalling blocker Ag+ abolished the Al3+-induced root growth inhibition (Sun et al., 2010). Therefore, ethylene may be involved in the root growth inhibition that occurred during the early responses to Hg.
ERF homologous genes were overexpressed in Arabidopsis seedlings treated with Cd for 2, 6 and 30 h (Herbette et al., 2006), which is similar to the pattern described here for Hg. However, ERF1 expression was abolished when ethylene signalling was blocked with 1-MCP in alfalfa and ein2-5 seedlings (Fig. 6). Differences between the plant species were observed for AP2 expression. M. sativa AP2 was up-regulated under Hg stress conditions but was not blocked by 1-MCP, whereas AP2 Hg-mediated expression was blocked in Arabidopsis ein2-5 plants. This differential AP2 expression pattern may be the result of the specific responses of each plant species, related to the different experimental settings used, or may be dependent on genetic differences. Similarly, the Hg-induced ACCO expression was also reduced in the ein2-5 plants but only after 24 h Hg treatment. A phylogenetic comparison between the ERF1, ACCO and AP2 gene families revealed that M. sativa and Arabidopsis ERF1 orthologues were much more similar than AP2 and ACCO (Fig. S6). ERF1 appears to be the primary transcription factor of the ethylene response cascade, and several experiments have shown that overexpression of ERF1 on an ein2-5 genetic background was sufficient for the expression of ethylene-related genes (Glazebrook, 2005). Therefore, the induction of ethylene-responsive genes by Hg was generally reduced in both Arabidopsis ein2-5 and 1-MCP-pretreated alfalfa plants.
The loss of ethylene perception in 1-MCP-pretreated alfalfa and ein2-5 mutants led to a significant decrease in the in vivo extracellular H2O2 production, compared with the H2O2 released by Hg-treated alfalfa and Col-0 roots (Fig. 7a,c). Concurrently, a similar stress response regulation was detected at the enzymatic level, and the observed increase in NADPH-oxidase activity detected with 3 μM Hg treatment was abolished in 1-MCP-treated alfalfa roots after 6 h of treatment (Fig. 7b). Involvement of NADPH-oxidases in apoplastic H2O2 production under Hg stress was confirmed by studying AtrbohD/F double mutants and 35S::AtrbohD overexpressor transgenic plants, which suffered a remarkable depletion or enhancement of apoplastic H2O2 generation, respectively (Fig. 7c). NADPH-oxidases are the primary producers of extracellular H2O2 under metal stress conditions (Romero-Puertas et al., 2004; Ortega-Villasante et al., 2007; Rodríguez-Serrano et al., 2009). Four homologue genes in M. sativa were early and transiently overexpressed (3–6 h) under Hg stress (Table S5), suggesting that a transcriptional regulation may also influence NADPH-oxidase activity. Notably, Mersmann et al. (2010) demonstrated that ROS generation triggered by flagellin FLS22 was blocked in etr1 and ein2 ethylene-insensitive mutants, which was related to the ability of ethylene to promote the release of extracellular H2O2 via the activation of NADPH-oxidases. It is possible that a primary response to Hg stress, the induction of an oxidative burst (Ortega-Villasante et al., 2007) and H2O2 release (Figs 1d, 7), is connected to ethylene perception. Heyno et al. (2008) determined that NADPH-oxidases are not the only source of H2O2 in Cd-treated plants because mitochondria may also be involved in H2O2 generation. Ortega-Villasante et al. (2007) observed that DPI, an NADPH-oxidase inhibitor, did not block extracellular H2O2 release in Cd-exposed alfalfa roots, whereas it significantly decreased Hg-induced H2O2 production. Therefore, each metal may exert different toxic effects, and Hg may be particularly effective in causing ROS production at the plasma membrane, although alternative sources of ROS cannot be discarded.
Other studies with the ein2-5 mutants confirmed the primary role of the EIN2 protein in sensing different abiotic and biotic stresses and in mediating different hormone response pathways (Thomma et al., 1999; Wang et al., 2007; Cao et al., 2009; Lei et al., 2011). Therefore, it is possible that Hg-induced stress follows a similar general signalling process, which intervenes in ROS induction. It is suggested that an accumulation of ROS, such as apoplastic H2O2, activates signalling processes in cells that influence the expression of stress genes (Xiang & Oliver, 1998; Foyer & Noctor, 2003), such as proteins that participate in the phenylpropanoid pathway (Yastreb et al., 2012) and ethylene (Desikan et al., 2005), auxin (Zhao et al., 2012) and jasmonate metabolism and signalling (Maksymiec & Krupa, 2006). Recent evidence suggests that components of the ethylene signalling pathway (e.g. ERF1) are also upstream of jasmonate-mediated responses (Cheng et al., 2013), implying a complex interplay between different stress-related phytohormones. Therefore, it will be necessary in the future to determine the relationship between these signalling molecules to understand the mechanisms underlying heavy metal perception and homeostatic adjustments in plants, which will require a multifactorial approach.
This work was supported by the Spanish Ministry of Economy and Competitiveness (AGL2010-015150 PROBIOMET), Fundación Ramón Areces, Comunidades Castilla-La Mancha (POII10-0087-6458 FITOALMA2). M.B.M-P. was supported by an FPU PhD grant from the Spanish Ministry of Education. Microarrays were purchased from Dr H. Küster, (Institute for Plant Genetics, Leibniz Universität, Hannover, Germany). We are indebted to Prof. J. M. Alonso (Department of Genetics, North Carolina State University, NC, USA) for the gift of Arabidopsis ein2-5 mutants and to Prof. M. A. Torres (Centro de Biotecnología y Genética de Plantas, Universidad Politécnica Madrid, Spain) for the donation of Arabidopsis AtrbohD/F mutants and 35S::AtrbohD overexpressor trangenics. We are grateful to T. Malefyt (AgroFresh Inc., Philadelphia, USA) for donating the 1-methylcyclopropene. We thank M. L. Flores-Cáceres and J. Sobrino-Plata (UAM) for invaluable technical assistance.