Introduction

Cadmium is considered a non-essential metal element for plants and is one of the most toxic heavy metals. It causes a wide range of deleterious effects on plants, including oxidative stress, protein inactivation, nutrient uptake disturbance and even plant death (Mohamed et al. 2012). Plants have evolved several collective mechanisms to prevent excess Cd effects, such as cell wall binding, extracellular exudates, redistribution by definite transporters, compartmentation in vacuole, chelation with phytochelatins, restoration of impaired proteins and antioxidant defense (Hall 2002; Mendoza-Cózatl et al. 2005; Weber et al. 2006; Ahsan et al. 2009; Lin and Aarts 2012; Gao et al. 2016; Singh et al. 2016). By limiting heavy metal transport activity, plants can diminish toxic metal inflow and enhance metal exclusion from the cytosol (Wysocki and Tamás 2010; Clemens et al. 2013). A number of heavy metal transporters involved in the acquisition, distribution, and homeostasis of Cd in plants have been characterized, comprising ATP-binding cassette transporters (ABC), heavy metal ATPases (HMA), metal transporter proteins (MTPs), low-affinity cation transporters (LCT), and natural resistance-associated macrophage proteins (Nramp) (Verrier et al. 2008; Thomine et al. 2000; Uraguchi et al. 2014; Slamet-Loedin et al. 2015; Yuan et al. 2012). OsNramp5, the uptake transporter for the essential element manganese (Mn), was found to enable rice to accumulate large amounts of Cd (Sasaki et al. 2012; Yoneyama et al. 2015). AtNramp3 and AtNramp4 play key roles in the export of vacuolar Cd (Thomine et al. 2003; Lanquar et al. 2010; Pottier et al. 2015), and AtNramp6 functions as an intracellular metal transporter, which can be modified to affect the distribution and availability of cadmium within the cell (Cailliatte et al. 2009). OsHMA2, an HMA family transporter, is involved in the translocation of Cd into shoots, while OsHMA3 helps sequester Cd in vacuoles (Ueno et al. 2010; Nocito et al. 2011; Satoh-Nagasawa et al. 2012; Takahashi et al. 2012). OsLCT1, which is localized at node I, plays an essential role in regulating Cd transport into rice grains (Uraguchi et al. 2011, 2014). The ABC transporter AtPDR8 is a Cd extrusion pump conferring heavy metal resistance in Arabidopsis (Kim et al. 2007), while the two closely related phytochelatin transporters, AtABCC1 and AtABCC2, mediate tolerance to Cd and mercury (Hg) (Park et al. 2012). Recently, another ABC-type transporter, AtABCC3, was recognized as a key factor for phytochelatin-mediated Cd tolerance in Arabidopsis (Brunetti et al. 2015).

Cd often induce the generation of toxic reactive oxygen species (ROS) such as hydrogen peroxide (H2O2), hydroxyl radical (OH•), and superoxide anion (O2 •−), which are able to oxidize biological macromolecules such as proteins, lipids, and nucleic acids, and consequently promote enzyme inactivation, lipid peroxidation, and membrane impairment (Daud et al. 2016). Plants have evolved antioxidant defense mechanisms, including enzymatic antioxidants such as peroxidase (POD), catalase (CAT), superoxide dismutase (SOD), ascorbate peroxidase (APX), glutathione reductase (GR), and non-enzymatic antioxidants including reduced glutathione (GSH), ascorbic acid (AsA), and phenolics to circumvent the damaging effects of ROS (Ahsan et al. 2009; Mittler 2002). A few studies of heavy metal-induced antioxidant stress defense processes in legumes have been conducted using common vetch (Vicia sativa L.) (Romman 2016). Growth and antioxidative metabolism characteristics have been examined previously in V. sativa under Cd stress (Zhang et al. 2009). GSH is a biothiol tripeptide that plays a fundamental role as an antioxidant in mitigating the redox imbalance caused by heavy metal accumulation and as precursor of phytochelatins which play crucial role in Cd chelation. Several studies have indicated the importance of GSH metabolism in plant tolerance to heavy metals (Hernandez et al. 2015; Ahsan et al. 2008; Yang et al. 2007a). GSH also serves as a substrate for glutathione S-transferases (GSTs), which catalyze the conjugation of GSH with potentially dangerous xenobiotics and the reduction of toxic organic hydroperoxides (Edwards et al. 2000). Glutathione peroxidase (GPX) reduces H2O2 to H2O using GSH as an electron donor, and glutaredoxins (GRXs) are small redox proteins that use GSH as a cofactor. The induction of GSTs, GRXs and GPX by heavy metals has been described previously in soybean, poplar and rice (Villiers et al. 2011; Song et al. 2013; Ma et al. 2012). Plant GSH synthesis depends on plant sulfur assimilation, beginning with the uptake and distribution of sulfate (SO4 2−). Sulfate is activated by conversion to adenosine 5′-phosphosulfate (APS), which is reduced to sulfite (SO3 2−) by adenylyl-sulfate reductase (APR). SO3 2− is reduced to sulfide (S2−), which is utilized to produce Cys and GSH (Hernandez et al. 2015).

Retention of Cd in cell wall is the first defense against Cd stress. Lignin is a branched phenylpropanoid polymer located in the cell wall, generally formed from three distinct phenylpropanoid alcohols including ρ-coumaryl, coniferyl, and sinapyl alcohol. More than 8 enzymes are involved in the biosynthesis of these precursors, including phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H), 4-coumarate:CoA ligase (4CL), hydroxycinnamoyl-CoA shikimate/quinate hydroxyl cinnamoyl transferase (HCT), ρ-coumarate 3-hydroxylase (C3H), cinnamoyl-CoA reductase (CCR), ferulate 5-hydroxylase (F5H), caffeoyl-CoA O-methyltransferase (CCoAOMT), caffeic acid O-methyltransferase (COMT), cinnamyl alcohol dehydrogenase (CAD) (Liu et al. 2015a, b). The polymerization of these precursors of lignin is catalyzed by POD in the presence of H2O2 and by laccases (LAC) in the presence of O2. Increased lignin deposition occurs in a wide range of plant species in response to Cd (Kováčik and Klejdus 2008; Elobeid et al. 2012; Yang et al. 2007b). It has been reported that lignification of epidermis and exodermis cell walls limited the entry of cadmium into the xylem (Ederli et al. 2004), and that increased xylem lignification might prevent the translocation of Cd from the roots to the aerial parts (Ahsan et al. 2012).

However, until now plant Cd tolerance mechanisms have been poorly understood. Identifying the Cd stress response genes involved in Cd tolerance in V. sativa via modern biotechnological techniques such as transcriptomic analysis might be a useful method to explore the mechanisms of plant Cd tolerance and a useful step toward developing future transgenic lines resistant to Cd stress.

Quite a few studies have used transcriptomic analysis to investigate the transcriptional and post-transcriptional regulatory mechanisms involved in plant responses to Cd stress (He et al. 2015; Dal Corso et al. 2010; Yue et al. 2016; Xu et al. 2015; Dubey et al. 2014; Cao et al. 2014; Liu et al. 2015a, b). Genes related to the Cd response have been identified using various transcriptomic analysis methods, including chip analysis in Arabidopsis roots under Cd stress, a comparative transcriptomic study of Cd-treated roots of A. thaliana and A. halleri (a Cd-hypertolerant metallophyte), a time-course investigation of gene regulation under Cd toxicity in rice shoots and roots, and a study of early transcriptomic responses to Cd in rice roots (Lin et al. 2013; Herbette et al. 2006; Ogawa et al. 2009; Weber et al. 2006; Zhao et al. 2009). These transcriptomic analyses have given rise to a broad understanding of Cd-induced gene expression in model plants. However, few studies have been conducted on the response to Cd stress in non-model plants, and high-throughput sequencing approaches can lead to improved prospects for the identification of distinct tolerance mechanisms in non-model plants such as V. sativa.

The leguminous plant Vicia sativa has been cultivated largely in areas at high altitude with extremely cold climates because of its high nutritional value and tolerance to cold and drought. It can be used as food for humans as well as fodder for livestock. Kim et al. (2015) conducted transcriptomic analysis of two V. sativa subspecies, mining molecular markers to enhance genomic resources for vetch improvement. Recently, we identified two varieties of V. sativa with distinct tolerances to Cd (Rui et al. 2016). In that study, we found elevated levels of ROS and a greater extent of lipid peroxidation in the roots of Cd-sensitive variety (ZM) than in the Cd-tolerant variety (L3), and observed that hydrogen peroxide (H2O2) and superoxide anions (O2 •−) mainly accumulated in the cell walls and extracellular spaces in response to Cd treatment. In addition, we elucidated the physiological relationships of Cd tolerance with apoplastic H2O2 accumulation, guaiacol peroxidase (GPOD) and LAC activities and lignin deposition in roots. In the present study, we aimed to identify the molecular mechanisms and defensive genes responsible for avoiding Cd toxicity in the same two varieties of V. sativa.

Materials and methods

Plant materials, growth conditions, and treatments

According to our previous study (Rui et al. 2016), seeds of two Vicia sativa varieties (L3, ZM) were surface-sterilized with 3% H2O2 and germinated in plastic dishes at 20–25 °C filled with vermiculite. After 3 days, seedlings were shifted to vessels comprising Hoagland nutrient solution. The growth parameters were set as follows: 25/20 °C day/night temperatures, 60–80% relative humidity, and 12/12-h day/night photoperiod. 7-day-old seedlings were treated with 0, 5 and 50 μM Cd (added as CdCl2·2.5H2O) for 24 h. The experiment was arranged in a completely randomized design with three replicate vessels having 10 seedlings correspondingly. The pH of the nutrient solution was adjusted to 5.5 with NaOH or HCl and nutrient solution was resupplied every 3 days. Subsequent to 24-h Cd treatments, roots, stems and leaves samples were harvested and immediately frozen in liquid nitrogen and stored at −80 °C for RNA extraction and subsequent de-novo library construction, RNA sequencing (RNA-Seq) and quantitative reverse transcription–polymerase chain reaction (qRT-PCR) analysis. All the experiments were performed at least twice, each with three biological replicates.

RNA isolation, RNA-Seq library preparation, and sequencing

Total RNA for RNA-Seq was extracted from roots, stems, and leaves of samples as described above using a plant RNA kit (Omega Bio-Tek, Norcross, GA, USA), following the manufacturer’s protocol. Purified RNA was quantified using an ND-8000 spectrophotometer (Nanodrop Technologies, Inc., Wilmington, DE, USA), a 2100-Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), or agarose gel electrophoresis. RNA samples with no smearing observed on agarose gels, a 260/280 ratio above 2.0, and an RNA integrity number greater than 8.0 were selected for use in subsequent processes. For de novo transcriptome analysis, equal amounts of RNA from roots, stems, and leaves were mixed to construct the cDNA library. For comparative RNA-Seq analysis, we mixed three replicate samples of roots into one total RNA sample for each treatment, which was then sent to Majorbio Bio-pharm Technology Corporation (http://www.majorbio.com/) for sequencing. The cDNA libraries were constructed using the TruSeq™ RNA Sample Preparation Kit (Illumina, Inc.) according to the manufacturer’s instructions. Briefly, the protocol consists of the following steps: Poly(A)-containing mRNA was purified from 5 μg of total RNA using oligo(dT) magnetic beads and fragmented into ~200-bp pieces using divalent cations at 94 °C for 5 min. The cleaved RNA fragments were copied into first-strand cDNA using SuperScript II reverse transcriptase (Life Technologies, Inc. Carlsbad, CA, USA) and random primers. After second-strand cDNA synthesis, fragments were end-repaired, A-tailed, and indexed adapters were ligated. The products were purified and amplified via PCR to create the final cDNA library. Target bands were harvested by 2% agarose gel electrophoresis and quantified using a TBS380 mini-fluorometer (Picogreen). The tagged cDNA libraries were pooled at equal ratios and used for 101-bp paired-end sequencing on the Illumina HiSeq2000 platform, according to the manufacturer’s instructions.

Data processing, assembly, and annotation

The original image data were processed by base calling, generating sequenced reads, namely, raw reads in the FASTQ format. Next, the data were subjected to four-step quality control: (1) base mistake ratio counting: <1%; (2) A/T/G/C percentage and distribution analysis, removal of AT/GC separated sequences, and transforming raw reads into clean reads; (3) removal of reads with more than 10% N (uncertain base type); and (4) removal of the adaptor and reads less than 20 bp in final length. Trinity software (v.2013-02-25) (Grabherr et al. 2011) was used for assembly, during which clean reads of different isoforms derived from one gene were assembled into distinct transcripts of the same subcomponent, which can be regarded as a gene. The longest transcript of each subcomponent was defined as a “unigene” for functional annotation. All assembled unigenes from the three tissues were searched against the Nr (NCBI non-redundant protein sequences) database to identify putative mRNA functions using BLASTX (Version 2.2.25) with an E-value cut-off of 10−5. Additionally, Gene Ontology (GO) terms were extracted from the best hits obtained from BLASTX against Nr using the Blast2GO program (Conesa and Gotz 2008). The BLAST algorithm was also used to align unique sequences by applying String software (Szklarczyk et al. 2011), the Pfam database (Punta et al. 2012), and the KEGG Ortholog (KO) database (Kanehisa et al. 2008) (with an E-value cut-off of 10−5) to predict possible functional classifications and molecular pathways. BLAST analysis were performed in a local protein sequence database downloading all protein sequences (proteomes) of various organisms used in NR, Swissprot, STRING, Pfam, and KEGG database.

Gene expression analysis

After mapping the results from all samples to the reference genome, RSEM (RNA-Seq by Expectation-Maximization) (http://deweylab.biostat.wisc.edu/rsem/) was used to calculate the reads per kilobase of transcript per million mapped reads (RPKM) value of each gene or transcript, which represents the relative expression level of that gene or transcript in the samples. EdgeR (http://www.bioconductor.org/packages/2.12/bioc/html/edgeR) was adopted to identify differentially expressed genes among groups of samples using the following criteria: FDR (false discovery rate) <0.05 and log2 (fold change) ≥1. A Venn diagram of differentially expressed genes was generated using the GeneVenn application (Pirooznia et al. 2007).

Validation of gene expression

To validate the RNA-Seq results, the expression of selected up- or down-regulated genes were confirmed by quantitative reverse transcription–polymerase chain reaction (qRT-PCR). After 7-days of growth, V. sativa seedlings were treated with or without 5 and 50 μM CdCl2 solutions for 24 h, and the roots samples were harvested for RNA extraction. After RNA extraction, genomic DNA was removed with the RNase free DNase I Set (Omega, USA) following the manufacturer’s guidelines. RNA was reverse-transcribed using a Super Reverse transcription kit (BioTeke, China). Gene-specific PCR products were amplified with SYBR pre-mix EX Taq (TaKaRa Bio, Japan), and the quantitative real-time PCR was performed on Mastercycler ep realplex (Eppendorf) with the primer sets shown in Supplemental Table 1. The PCR protocol was as follows: initial denaturation at 95 °C for 30 s, followed by 95 °C for 5 s, and 60 °C for 34 s in a 40-cycle reaction, then followed by dissociation stage. 2 –ΔΔCt method calculation was used for data analysis and ACTIN2 was used as an internal standard.

Table 1 Summary of RNA-seq of V. sativa transcriptome
Full size table

Results

De novo assembly of a V. sativa reference transcriptome

To compare the transcriptomes of V. sativa varieties (L3 and ZM) treated with different CdCl2 concentrations, a de novo assembled V. sativa transcriptome was generated. In total, 61,335,826 raw reads were generated from the mixed samples (Table 1). After filtration, 58,789,686 clean reads were obtained. More than 95.21% of bases in raw reads had a Q value ≥20 (an error probability of 0.0312%) and more than 97.15% of bases in clean reads had a Q value ≥20 (an error probability of 0.0275%). The GC contents were 43.33% and 43.25% for raw reads and clean reads, respectively. Clean reads data were used for de novo assembly. Trinity software generated 63,100 total transcripts (Table 2) with an average length of 921.55 bp and a N50 of 1532 bp, and a total of 49,062 sequences were identified as unigenes. Among these unigenes, 9398 (32.9%) were more than 500 bp in length, while 3854 (13.5%) of unigenes were longer than 1 kb (Supplemental Fig. 1).

Table 2 Summary of de novo assembly of V. sativa transcriptome
Full size table
Fig. 1
figure 1

Species distribution of matched proteins in NR database in V. sativa transcriptome

Full size image

Functional annotation

All assembled transcripts and unigenes were searched against the Nr, Swissprot, String, Pfam and KEGG databases using BLAST (Version 2.2.25, E-value <1E−5). Among unigenes, 57.30% showed homology with Medicago truncatula, 27.85% showed homology with Cicer arietinum, 4.32% showed homology with Glycine max, and 1.40% showed homology with Pisum sativum in the Nr database (Fig. 1). The numbers of transcripts with significant similarity to sequences in the Pfam, Swissprot, KEGG, String, and Nr databases were 24,093, 25,930, 13,247, 19,411, and 40,402, respectively. The numbers of unigenes with significant similarity to sequences in the Pfam, Swissprot, KEGG, String, and Nr databases were 17,227, 18,662, 9543, 13,854, and 29,354, respectively.

During GO classification, all unigenes were classified based on three functional categories: molecular function, biological process, and cellular component. In biological process analysis, unigenes were clustered into 22 classifications, with the most abundant being metabolic processes,cellular processes, and single-organism processes, containing 11,943, 10,345, and 8575 unigenes, respectively. In the cellular component category, unigenes were clustered into 17 classifications, with the largest being the cell, cell parts, and organelle classes, which contained 7823, 7823, and 5582 unigenes, respectively. In terms of molecular function, unigenes were clustered into 17 classifications, the largest of which were binding and catalytic activity, containing 9509 and 9636 unigenes.

A total of 10,218 unigenes were clustered into 25 Eukaryotic Orthologous Group (KOG) categories (Fig. 2). Among these categories, the largest was general function prediction only, accounting for 1829 unigenes; the second largest category was signal transduction mechanisms, including 1141 unigenes; and the third was post-translational modification/protein turnover/chaperones, comprising 1070 unigenes.

Fig. 2
figure 2

Histogram presentation of KOG classification of unigenes in V. sativa transcriptome

Full size image

A total of 13,247 unigenes had BLAST hits in KO database with E-values <1E−5, and 7237 were assigned to 330 metabolism pathways. Among these pathways, the six with the most genes were ribosome, plant hormone signal transduction, starch and sucrose metabolism, protein processing in the endoplasmic reticulum, plant-pathogen interaction, and phenylpropanoid biosynthesis.

Differential gene expression between L3 and ZM under Cd stress

To identify genes in V. sativa that are involved in Cd stress responses and those that contribute to the differing Cd tolerances between the two varieties, L3 and ZM, 7-day-old seedlings of both varieties were treated with 5 or 50 μM solutions of Cd2+ or solution without Cd (control) for 24 h, after which root samples were harvested from each treatment and labeled as L3(0); without Cd, L3(5); with 5 μM Cd and L3(50); with 50 μM Cd) and ZM(0), ZM(5), and ZM(50) accordingly. Libraries were constructed from the six RNA samples separately and Illumina Hiseq sequencing was performed. More than 27 million raw reads were obtained, and more than 24 million clean reads remained for each sample after processing (Supplemental Table 2). Using the de novo assembled sequence as a reference, clean reads were mapped, with results between 23 and 34 million reads (Supplemental Table 3).

Based on the results above, gene expression was calculated using the RPKM method, which takes into account the influences of both sequencing depth and gene length on read count. On the basis of the criteria applied [FDR < 0.05 and log2 (fold change) ≥1], the two cultivars grown under low- and high-Cd conditions were compared with the control. Under low-Cd treatment, few genes had different expression levels from the control in both varieties (i.e. 69 in L3 and 28 in ZM). Under high-Cd treatment, 1036 genes showed differential expression in L3, whereas in ZM, only 335 genes had altered expression levels. As shown in the Venn diagram (Fig. 3), in L3, 53 genes were differentially expressed under both low- and high-Cd stress, 16 genes were solely expressed under low-Cd stress, and 983 genes specifically changed expression under high-Cd stress. In ZM, 21 genes were differentially expressed under both low- and high-Cd stress, 7 genes were specifically expressed under low-Cd stress, and 314 genes changed expression solely under high-Cd stress. As commonalities between the L3 and ZM varieties, there were 13 and 214 genes differentially expressed under low- and high-Cd treatments, respectively.

Fig. 3
figure 3

The Venn diagram of the number of differentially expressed unigenes in each comparison group

Full size image

To identify possible biological processes or pathways that were altered under the two levels of Cd stress, in relation to the two varieties of V. sativa, Gene Ontology (GO) and KEGG pathway enrichments were performed using an FDR adjusted p-value ≤0.05 as the criterion. Our analysis revealed that the common differentially expressed genes were highly enriched in cell wall organization or biogenesis, phenylpropanoid biosynthesis, drug metabolism-cytochrome P450, metabolism of xenobiotics by cytochrome P450, cutin, suberine, and wax biosynthesis, protein processing in the endoplasmic reticulum, glutathione metabolism, starch and sucrose metabolism, the tricarboxylic acid cycle, and glycerophospholipid metabolism (Supplemental File 1).

Cell wall-related genes

Cutin, suberine, and wax biosynthesis genes were considerably overrepresented in L3(50) vs. L3(0) as compared to ZM(50) vs. ZM(0). Five genes were significantly up-regulated in L3(50) vs. L3(0); in contrast, only one gene (FAR) showed increased expression in ZM(50) vs. ZM(0) (Supplemental Table 4). Lignin biosynthesis pathway genes were overrepresented in both varieties [L3(50) vs. L3(0) and ZM(50) vs. ZM(0)]; one COMT, two CCR, and three CAD genes showed similar up-regulation patterns. However, one 4CL gene displayed decreased expression in the comparison of L3(50) vs. L3(0) and one CCoAOMT gene down-regulated in the comparison of ZM(50) vs. ZM(0). Among POD genes, 13 were up-regulated and 1 gene was down-regulated in L3(50) vs. L3(0), whereas 10 and 3 POD genes were up- and down-regulated, respectively, in ZM(50) vs. ZM(0). Among these POD genes, 9 were up-regulated in both varieties (Supplemental Table 5).

Common stress response genes

Under Cd stress, both varieties differentially expressed many heat shock protein (HSP) genes and other chaperone/co-chaperone protein-encoding genes (Supplemental Table 6). In total, 19 and 16 HSP genes were up-regulated in L3(50) vs. L3(0) and ZM(50) vs. ZM(0), respectively, and 15 genes were shared among them. Eleven of these (c21159_g1; c24577_g1; c45652_g1; c28022_g1; c39617_g1; c8867_g1; c32996_g1; c3732_g1; c25243_g1; c27249_g1; c9550_g1) are among the 13 genes that are differentially expressed in the presence of Cd regardless of dose or variety. For chaperone/co-chaperone genes, 5 were up-regulated and 2 down-regulated in L3(50) vs. L3(0), while 3 were significantly up-regulated in ZM(50) vs. ZM(0). We also identified that two germin-like protein-encoding genes were commonly up-regulated in both groups. In contrast, one pathogenesis-related protein (PR) gene was up-regulated in L3(50) vs. L3 (0), but not ZM(50) vs. ZM(0).

GSH pathway genes

We identified 11 GST genes that were significantly up-regulated in the assessment of L3(50) vs. L3(0), while only one GST gene showed increased background expression in ZM. Two GPX and 6 GRX genes were significantly down-regulated in the background of L3, while no GPX and 4 GRX genes were down-regulated in the ZM background. For the APR gene, 4 and 3 genes were significantly up-regulated in the background of L3 and ZM, respectively (Supplemental Table 7).

Metal transporter genes

We found significant differences in metal transporter gene expression regulation between L3(50) vs. L3(0) and ZM(50) vs. ZM(0) (Supplemental Tables 8, 9 and 10). More than 10 transporter genes were differentially regulated under Cd exposure, and most of these gene expression changes were variety-specific. There were 9 ABC transporter family genes which up-regulated. Among them, one PDR family ABC transporter gene (c28232_g2) was up-regulated in ZM but not in L3, whereas 8 ABC family genes (c20094_g1, c21832_g1, c2540_g1, c28624_g2, c27925_g1, c13627_g1, c28427_g1, c25102_g1) were up-regulated in L3 background, but not in ZM. Moreover, four MATE (Multidrug and Toxic Compound Extrusion) efflux family protein genes, three auxin transporter genes and five phosphate transporter genes were up- or down- regulated in L3 but not in ZM. As for sugar transporter genes, ZIP (ZRT/IRT-like protein) family transporter genes and lipid-transfer protein genes, the numbers of changed gene were much greater in L3 than in ZM. However, three sulfate transporter genes and two Nrmap3 genes were only up-regulated in ZM but not in L3. In general, the change in expression of transporter family genes was much higher in the L3 than in ZM under Cd stress.

Transcription factors

Transcription factors (TFs) play a major role in the network of Cd-responsive genes. We identified several Cd-inducible TFs belonging to different families. Significant differences in TF gene expression regulation between L3(50) vs. L3(0) and ZM(50) vs. ZM(0) were observed (Supplemental Tables 11 and 12). Two GRAS family TF genes were down-regulated in L3(50) vs. L3(0) and one was down-regulated in ZM(50) vs. ZM(0). Six bHLH TF genes were down-regulated in L3(50) vs. L3(0), whereas only one down-regulated bHLH TF gene was identified in ZM(50) vs. ZM(0). Four and two myeloblastosis protein (MYB) family TF genes were up-regulated in L3(50) vs. L3(0) and ZM(50) vs. ZM(0), respectively, while four MYB TFs were repressed in L3(50) vs. L3(0). Besides, four up-regulated ERF (ethylene-responsive) genes, two up-regulated zinc finger protein (ZFP) TF genes, one up-regulated NAC TF-like gene, one down-regulated RF2a-like TF gene, two basic leucine zipper (bZIP) TF genes, and two up-regulated and one down-regulated WKRY TF genes were identified in L3(50) vs. L3(0), while none was found in ZM(50) vs. ZM(0).

qPCR validation

To test the reliability of RNA-Seq results, we selected several transporter gene family members for real-time quantitative polymerase chain reaction (RT-PCR) validation. As shown in Fig. 4, the differential expression changes were largely enhanced in the L3 background, especially in the case of two ABC transporter genes, ABCC3 and ABCC9, under 5 μM Cd stress. Under 50 μM Cd, the differential expression was significant and much higher (Fig. 4).

Fig. 4
figure 4

Validation of RNA-Seq data by real-time qRT-PCR under Cd-treated condition in two V.sativa variesties L3 (Cd-tolerant) and ZM (Cd-sensitive). The relative mRNA levels of several genes under 5 μm (a) and 50 μm (b) CdCI2 stress for 24 h. Steady-state mRNA levels in roots were normalized against actin2 gene and were expressed relative to those of control plants, which were given a value of 1 (not shown). Bar values are means ± SD of three biological replicates, each corresponding to different RNA extractions from roots of three plants grown independently

Full size image

Discussion

Plants have evolved different strategies for improving their heavy metal tolerance (Gallego et al. 2012). In our previous work (Rui et al. 2016), we observed that the V. sativa variety L3 had a higher Cd tolerance than the variety ZM. In this study, using high-throughput sequencing technologies, we aimed to provide a deeper exploration of the molecular basis of responses to Cd stress, revealing Cd tolerance genes in two varieties with contrasting Cd tolerance traits at the global transcriptome level. On one hand, the different expression regulation patterns under different levels of Cd stress suggested that the genes related to cell wall organization or biogenesis, phenylpropanoid biosynthesis, GSH metabolism and common stress response play essential roles in response to Cd in V. sativa plants, while the difference between Cd tolerances in the L3 and ZM varieties is correlated with metal transporter genes and transcription factors.

In our previous study (Rui et al. 2016), we noted that lignin accumulation in the roots was enhanced under Cd stress, and this accumulation was more significant in the sensitive variety, ZM. In RNA-Seq data, the lignin biosynthesis pathway was found to be significantly overrepresented in KEGG pathway analysis. One COMT, two CCR, and three CAD genes were significantly up-regulated in both varieties. However, aside from one 4CL gene and one LAC gene that displayed decreased expression only in L3(50) vs. L3(0) and one CCoAOMT gene that displayed decreased expression only in ZM(50) vs. ZM(0), there were no differences in lignin biosynthesis pathway gene expression between the two varieties. As a number of lignin biosynthesis pathway genes have previously been reported as inducible by Cd stress in Arabidopsis and copper (Cu) toxicity in rice (Liu et al. 2015a), we suggest that this pathway is a conserved adaptation mechanism that plants have gained through evolution. In addition, GPOD (both ionically bound cell wall GPOD and soluble GPOD) and LAC activity in the roots of two V. sativa varieties increased rapidly under 10 μM Cd stress (Rui et al. 2016), which is analogous with our study on Cu in rice roots (Liu et al. 2015a). Phenols can participate in lignin synthesis, acting as the electron donor for POD (Sgherri et al. 2004), and LAC can also oxidize phenolic substrates, a process which requires O2 as a secondary substrate. So we suggest that lignin accumulation is a common response by plants to heavy metal stress. Moreover, one KEGG pathway (cutin, suberine, and wax biosynthesis) showed significant enrichment only in L3(50) vs. L3(0). Metabolites including cutin, suberine, and wax on the cuticle can decrease water loss during dehydration stress, and thus enrichment of the cutin, suberine, and wax pathway may play a role in Cd-tolerance differences found between the two varieties.

There are a greater number of GSH metabolism genes up- or down-regulated in L3(50) vs. L3(0) than in ZM(50) vs. ZM(0), in particular, there are 11 GST genes up-regulated in L3(50) vs. L3(0) but only one GST gene up-regulated in ZM(50) vs. ZM(0), which suggested that the GSH metabolism had an important role in the difference of Cd tolerance in two V. sativa varieties.

Molecular chaperones (e.g. HSPs) are protein-repairing bodies, and are active in cells under both normal and adverse conditions. HSPs, including HSP70s, HSP90s, HSP60s, and small heat-shock proteins (sHSPs), are stress-responsive proteins, protecting plants from damage under stress (Xu et al. 2015). In the present experiment, up-regulation of a large number of HSPs during Cd stress indicates that HSPs in particular play important roles in protecting plants against stress by re-establishing normal protein conformation and thus cellular homeostasis (Dubey et al. 2014). There are more stress-responsive genes in L3(50) vs. L3(0) than in ZM(50) vs. ZM(0), which may contribute to the different levels of cadmium tolerance of these two varieties.

Heavy metal accumulation can be mediated by a group of metal transporters in plants. Plant metal transporters can be classified into different families based on their sequence specificity, subcellular location, and the metals they carry (Hall 2002). Several metal transporter family members are implicated in heavy metal detoxification in plants (Yoneyama et al. 2015), including Nramp, MATE, HMA, ZIP, and ABC. Cd is biologically non-essential and often accumulate in plants through the same transporters used for essential divalent metals, such as Ca, Fe, Mn and Zn due to their structural similarity (Verbruggen et al. 2009). The repression of these transporters under Cd toxicity could reduce plant Cd uptake and translocation and increase Cd tolerance, however, in the other hand might cause secondary effects on essential metal uptake and distribution in planta which will also affect plant Cd tolerance. OsNramp5 is a major transporter of Mn and Cd in rice root. The OsNramp5 knockout line lost the ability to take up Mn and Cd and resulted in a significant reduction in rice growth and grain yield (Sasaki et al. 2012). It has been reported that Nramp3 and Nramp4 are responsible for transporting Cd2+ out of the vacuole in Arabidopsis. The nramp3nramp4 mutant impaired in release of vacuolar metal stores showed hypersensitive to cadmium stress (Molins et al. 2013). In V. sativa, up-regulation of two Nramp3 under Cd stress in ZM but not in L3 might play some important roles in different Cd tolerance of these two varieties. As a large family of membrane proteins, ABC transporters are involved in the transport of a broad range of substances and in plant responses to various environmental stresses. Both OsPDR9 and AtPDR8 encode PDR-type ABC transporters, which are induced by Cd in rice and Arabidopsis as an efflux pump of Cd2+ or Cd conjugates (Kim et al. 2007; Moons 2003). In V. sativa, one PDR family ABC transporter gene was up-regulated in ZM but not in L3. In contrast, there were 8 ABC family genes that showed a response to Cd in L3 background, but not in ZM. This is also confirmed by the qRT-PCR results (Fig. 4). The sensitivity expression response of ABC transporter genes in the L3 background is an adaptation mechanism to detoxify Cd from cytosol. We speculate these difference in the ABC transporters in response to Cd stress is at least one main contribution in the difference in Cd-tolerance of L3 and ZM. MATE proteins bind to a variety of potentially toxic compounds and function as proton dependent efflux transporters to remove toxic compounds from the cell (Dubey et al. 2014). A MATE-related efflux protein from Arabidopsis (AtDTX1) was reported to serve as an efflux transporter for plant-derived alkaloids, antibiotics, and other toxic compounds involving in detoxification of heavy metal Cd (Li et al. 2002). Four MATE efflux protein genes were down-regulated in Cd-tolerant L3 but not in Cd-sensitive ZM, if these differences were responsible for their different Cd tolerances, it needs further research. The uptake of Cd may occur via ZIP family transporters. One root iron transporter protein gene was down-regulated in L3 but not in ZM and one ZIP family protein was up-regulated in both varieties, which indicated that V. sativa might regulate root Cd or iron/zinc uptake through ZIP transporter expression regulation, which might influence plant Cd tolerance by control of Cd absorption and essential metal utilization. Besides, there are three auxin transporter genes up- or down-regulated and five phosphate transporter genes up-regulated in L3 but not in ZM, which indicated that auxin regulation and phosphate transport might play important role in response to Cd stress in L3. Cd affected sulfur assimilation pathway in both of the two V. sativa varieties. Three sulphate transporter genes were up-regulated in ZM but not in L3, which might mean that ZM absorbed more sulphate than L3 in response to Cd stress. Corresponding to this, three APR genes were up-regulated in L3 and four APR genes were up-regulated in ZM (Supplemental Table 7). Sulfur assimilation leads to the activation of pathway responsible for GSH biosynthesis (Gill and Tuteja 2011). GSH acts as an important antioxidant in mitigating Cd-induced oxidative stress and plays an important role in phytochelatins synthesis (Hernandez et al. 2015). These results indicated that sulphur assimilation pathway and synthesis of GSH were crucial steps for alleviating the phytotoxicity of Cd in two V. sativa varieties. In our study, we observed differential expression of various transporter genes between L3 and ZM under Cd stress. These results indicated that transporters might play an essential role in the differential tolerance to Cd stress among the two V. sativa varieties.

Transcription factors are involved in many processes related to plant stress signaling. TFs, such as the bZIP, WRKY, NAC, ERF, and MYB play a vital part in directing the expression of specific stress-related genes in response to Cd stress (Dal Corso et al. 2010). Genes including MYB, WRKY, NAC, and AP2 were up-regulated in rice roots treated with 10 μM Cd for 3 h. It has also been shown in Arabidopsis thaliana roots that some genes encoding C2H2 ZFP, AP2 domain-containing protein, HSF, and MYB TFs were specifically induced by Cd2+ but not by Cu2+ (Weber et al. 2006). Wheat ZAT7 and ZAT12 in the C2H2 ZFP family were shown to improve ROS tolerance in Arabidopsis (Ali-Benali et al. 2013). In Arabidopsis, RAP2.6, an AP2/ERF transcription factor, contributes to abscisic acid (ABA), osmotic, and salt toxicity responses (Zhu et al. 2010). SNAC2, isolated from Oryza sativa, is a stress-inducible NAC gene that can elevate stress tolerance in rice (Hu et al. 2008). In addition, wheat overexpressing the TaNAC6 gene was shown to exhibit greater resistance to stress (Xue et al. 2011). MYB genes can also respond to one or more stress treatments, such as elevated salinity, exogenous ABA, and polyethylene glycol (PEG) (Zhang et al. 2012). A number of WRKY genes are also stimulated by lead (Pb) heavy metal stress in Arabidopsis (Liu et al. 2009). For example, WRKY38 from barley has been implicated in responses to drought and cold stress (Marè et al. 2004). TcWRKY53 is also induced markedly by NaCl, drought, cold, and salicylic acid stresses in Cd-treated Thlaspi caerulescens (Wei et al. 2009). Four up-regulated WRKY genes comprising OsWRKY71, OsWRKY28, OsWRKY22, and OsWRKY42 were also shown to be involved in plant stress resistance responses (Liu et al. 2007; Abbruscato et al. 2012; Chujo et al. 2013; Han et al. 2014). In rice, OsWRKY42 also inhibits OsMT1d and induces ROS production and senescence of leaves. One up-regulated gene, OsWRKY72, was found to be involved in the ABA signal and auxin transport pathway (Song et al. 2010). In our experiment, we observed differential expression of various TF genes (WRKY, GRAS, MYB, bHLH, ZFP, ERF, and NAC transcription factors) in two V. sativa varieties under Cd stress. These results suggested that TFs play significant role in Cd stress response of V. sativa, especially in Cd-tolerant variety L3 which had greater number of up- or down-regulated TF genes in response to Cd stress. Further analysis of these TF genes may lead to decipher the network related to the whole pathway that may differ in Cd stresses. Currently, we are testing their potential role in heavy metal tolerance.

In conclusion, we demonstrated that Cd caused lignification in the roots of V. sativa at the transcriptome level, which is consistent with our previous physiological data (Rui et al. 2016). More interestingly, our gene expression analysis results suggest that the differential expression of transcription factors and transporter genes plays a major role in controlling the contrasting Cd stress sensitivities of the L3 and ZM varieties of V. sativa.