Involvement of abscisic acid-responsive element-binding factors in cassava (Manihot esculenta) dehydration stress response

Cassava (Manihot esculenta) is a major staple food, animal feed and energy crop in the tropics and subtropics. It is one of the most drought-tolerant crops, however, the mechanisms of cassava drought tolerance remain unclear. Abscisic acid (ABA)-responsive element (ABRE)-binding factors (ABFs) are transcription factors that regulate expression of target genes involved in plant tolerance to drought, high salinity, and osmotic stress by binding ABRE cis-elements in the promoter regions of these genes. However, there is little information about ABF genes in cassava. A comprehensive analysis of Manihot esculenta ABFs (MeABFs) described the phylogeny, genome location, cis-acting elements, expression profiles, and regulatory relationship between these factors and Manihot esculenta betaine aldehyde dehydrogenase genes (MeBADHs). Here we conducted genome-wide searches and subsequent molecular cloning to identify seven MeABFs that are distributed unevenly across six chromosomes in cassava. These MeABFs can be clustered into three groups according to their phylogenetic relationships to their Arabidopsis (Arabidopsis thaliana) counterparts. Analysis of the 5′-upstream region of MeABFs revealed putative cis-acting elements related to hormone signaling, stress, light, and circadian clock. MeABF expression profiles displayed clear differences among leaf, stem, root, and tuberous root tissues under non-stress and drought, osmotic, or salt stress conditions. Drought stress in cassava leaves and roots, osmotic stress in tuberous roots, and salt stress in stems induced expression of the highest number of MeABFs showing significantly elevated expression. The glycine betaine (GB) content of cassava leaves also was elevated after drought, osmotic, or salt stress treatments. BADH1 is involved in GB synthesis. We show that MeBADH1 promoter sequences contained ABREs and that MeBADH1 expression correlated with MeABF expression profiles in cassava leaves after the three stress treatments. Taken together, these results suggest that in response to various dehydration stresses, MeABFs in cassava may activate transcriptional expression of MeBADH1 by binding the MeBADH1 promoter that in turn promotes GB biosynthesis and accumulation via an increase in MeBADH1 gene expression levels and MeBADH1 enzymatic activity. These responses protect cells against dehydration stresses by preserving an osmotic balance that enhances cassava tolerance to dehydration stresses.

in the cassava cultivar South China 5 (SC5) by BLAST searches of the Phyzotome database and NCBI database. The resulting genes, termed MeABF1, MeABF2, MeABF3, MeABF4, MeABF5, MeABF6, and MeABF7, were then cloned and sequenced, and their sequences were aligned with the counterparts of the cassava cultivar AM560-2.
To examine phylogenetic relationships among ABFs, a phylogenetic tree was constructed using the amino acid sequences of ABFs from Arabidopsis and cassava (Supplementary Fig. S1 and Table S1). The phylogenetic tree revealed that the seven MeABFs could be classified into three groups (A, B, and C). Group A MeABFs underwent a significant expansion, as two MeABFs (MeABF2 and MeABF3) existed in cassava, whereas one corresponding Arabidopsis ortholog (Arabidopsis thaliana ABF2, AtABF2) were present. The size of Group B MeABFs decreased as evidenced by a single MeABF1 and three AtABFs (AtABF1, AtABF3, and AtABF4). Surprisingly, Group C included four MeABFs (MeABF4, MeABF5, MeABF6, and MeABF7) that were exclusive to cassava. Furthermore, MeABF2 and MeABF3, and MeABF5 and MeABF6 showed high sequence similarity with each of the other among seven MeABFs.
MeABF chromosomal distribution and cis-acting element analysis. To determine the chromosomal distribution of the identified MeABFs, we subjected the genomic sequences of the seven MeABFs to BLAST analysis against the cassava genome database. The location image shows that the seven MeABFs could be successfully mapped to six of the 18 chromosomes present in the cassava genome (Fig. 1). MeABF2 and MeABF4 were distributed on chromosome 18, and MeABF1, MeABF3, MeABF5, MeABF6, and MeABF7 were present on chromosomes 10,5,8,11, and 2, respectively.
To further explore transcriptional regulation of MeABFs, the promoter sequence cis-elements of these genes were predicted using PlantCare software (Supplementary Table S2). Previous studies showed that promoters for most MeABFs contained one or more cis-elements of genes that act in hormone signaling pathways involving ABA (ABRE), jasmonic acid (CGTCA-motif and TGACG-motif), gibberellic acid (GARE-motif, P-box, and TATC-box), salicylic acid (TCA-element), and auxin (TGA-element). In this study, we found that the upstream regions of most of the identified MeABFs contained one or more cis-elements related to stress responses. These elements include Box-W1 (fungal elicitor responsive element), HSE (heat stress response element), LTR (low-temperature response element), MBS (MYB binding site involved in drought-inducibility), TC-rich repeats (cis-acting element involved in defense and stress responsiveness), W box (WRKY binding site involved in abiotic stress responsiveness), and box S (elicitor-responsive element involved in the wounding and pathogen responsiveness). Interestingly, the promoter regions of most MeABFs contain one element involved in circadian clock and additional elements involved in light responsiveness. The cis-regulatory elements present in the 5′ upstream regions of the MeABFs identified here suggested that they play an important mediating role in plant growth and development, as well as in responses to various stresses.

MeABF expression levels in cassava leaves under dehydration stresses. All of the measured
MeABF genes of cassava leaf tissue showed differential expression profiles after drought, osmotic, or salt stress treatments (Fig. 2), except for MeABF5 and MeABF6, whose expression levels were too low to detect. This phenomenon also occurred in stems, roots and tuberous roots (Figs 3-5).
The transcript levels of the MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 of cassava leaves were significantly elevated or remained at a similar level as the control at all points in time in response to drought stress (Fig. 2, Supplementary Tables S3 and S4), with significant up-regulation by 33.33%, 83.33%, 100.00%, 33.33%, and 50.00%, respectively (Supplementary Table S5).
Compared to the control, the expression of MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 was significantly increased or unchanged in contrast to the control at any point in time in cassava leaves treated by osmotic stress, with the exception of MeABF2 (2 h and 8 h) and MeABF4 (2 h), which were suppressed at one or two points in time (Fig. 2, Supplementary Tables S3 and S4). The observed significant increases relative to control were 50.00%, 33.33%, 50.00%, 33.33%, and 50.00% for MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7, respectively (Supplementary Table S5).
The expression levels of MeABF1, MeABF2, and MeABF3 in cassava leaves were significantly increased only at 12 h, whereas the levels for MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 were significantly reduced or sustained at a similar level as the control at most time points by salt stress (Fig. 2   Expression profiles of MeABFs in cassava leaves under three stress conditions. Relative expression values for the target genes were calculated according to the 2 −ΔΔCt method. The heatmap was generated based on the log 2 of relative expression values using MeV software. Red, green and white indicate up-regulated, down-regulated and unchanged expression, respectively. Asterisk on the right corner of the value indicates a statistically significant difference at P -value < 0.05 and the absolute value of log 2 relative expression values >1. D, drought stress; O, osmotic stress; S, salt stress. In cassava leaves, drought stress significantly elevated the expression of the highest number of MeABFs (60.00%), followed by osmotic stress (43.33%) and then salt stress (10.00%, Supplementary Table S5).

MeABF expression levels in cassava stems under dehydration stresses. In cassava stems, MeABF1
and MeABF3 expression levels were significantly up-regulated or sustained at a level similar to that of the control throughout the drought treatment period, except at 15 d when MeABF3 levels were significantly reduced. MeABF2, MeABF4, and MeABF7 expression was significantly decreased or unchanged at any point in time under drought stress (Fig. 3, Supplementary Tables S6 and S7). In addition, MeABF1 and MeABF3 expression was significantly upregulated by 83.33% and 33.33%, respectively, relative to control (Supplementary Table S8).
After osmotic stress treatment, MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 expression in cassava stems was significantly increased or unchanged relative to the control at all points in time, with the exception of MeABF7 levels at 2 h, which were significantly reduced (Fig. 3 MeABF expression levels in cassava roots under dehydration stresses. In cassava roots, transcript levels of MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 were significantly increased or sustained at a level similar to that of the control by drought stress, with the exception of MeABF2 and MeABF4, which were significantly reduced at only one point in time (9 d and 1 d, respectively; Fig. 4 and Supplementary Tables S9 and S10). The expression of these five was significantly induced by 66.67%, 66.67%, 33.33%, 50.00%, and 66.67%, respectively, relative to the control (Supplementary Table S11).
MeABF3 and MeABF7 expression in cassava roots was significantly up-regulated or unchanged in contrast to the control in response to osmotic stress. The expression levels of MeABF1, MeABF2, and MeABF4 were significantly decreased or sustained at a level similar to that of the control at any point in time under osmotic stress   Table S11).
Under salt stress, MeABF3 expression was significantly induced (66.67%) or remained at a level similar to that of the control in cassava roots, except at 24 h, when the level was significantly suppressed. The expression levels of MeABF1, MeABF2, MeABF4, and MeABF7 in cassava roots were significantly reduced or unchanged at any point in time by salt stress (Fig. 4, Supplementary Tables S9-11).
In cassava roots, drought stress significantly elevated the expression of the highest number of MeABFs (56.67%), followed by osmotic stress (16.67%) and then salt stress (13.33%, Supplementary Table S11).

MeABF expression levels in tuberous roots under dehydration stresses. The transcript levels of
MeABF1, MeABF2, MeABF3, and MeABF7 in cassava tuberous roots were significantly increased or sustained at a level similar to that of the control at all points in time by drought stress. MeABF4 expression levels in tuberous roots were significantly reduced or unchanged after drought stress treatment (Fig. 5, Supplementary Tables S12 and S13). The induction rate for MeABF1, MeABF2, MeABF3, and MeABF7 was 100.00%, 66.67%, 83.33% and 16.67%, respectively (Supplementary Table S14).
MeABF1, MeABF2, MeABF3, MeABF4, and MeABF7 expression in cassava tuberous roots was significantly up-regulated or unchanged relative to the control at any point in time by osmotic stress treatment, with the exception of MeABF2 (1 h), MeABF3 (1 h), and MeABF7 (1 h and 4 h), which were significantly down-regulated at one or two points in time (Fig. 5, Supplementary Tables S12 and S13). The rate of significant induction of expression was 100.00%, 16.67%, 66.67%, 50.00%, and 66.67%, respectively (Supplementary Table S14).
Osmotic stress significantly elevated the expression of the highest number of MeABFs (60.00%) in cassava tuberous roots, followed by drought stress (53.33%) and then salt stress (0.00%; Supplementary Table S14).

Measurement of GB content in cassava.
We also measured GB content in the fresh leaves of cassava that had been treated by drought, osmotic, or salt stress (Fig. 6, Supplementary Tables S15 and S16). The GB content was significantly altered by the three stresses and similar patterns manifested as an initial increase, a decrease, and then an increase and a decrease. Furthermore, GB levels increased and then decreased during the final stages under drought stress conditions. GB exhibited a maximum accumulation at 4 h after osmotic and salt stress treatments and at 15 d after drought stress treatment. Overall, GB content was elevated under drought, osmotic, and salt stress conditions in cassava leaves (Supplementary Table S17).

Identification of MeBADHs, cis-acting elements, and expression analysis. Two members of the
MeBADH family, termed MeBADH1 and MeBADH2, were identified in cassava using the same methods as those used to identify MeABFs. Cis-acting element analysis in the promoter sequence of these two genes showed that the upstream region of MeBADH1 contained three types of cis-elements that act in hormone signaling pathways involving ABA (ABRE), gibberellin (GARE-motif), and auxin (TGA-element), whereas the MeBADH2 promoter contained only the ERE cis-element (ethylene-responsive element) (Supplementary Table S18). The MeBADH1 and MeBADH2 promoter regions both contained two or more kinds of cis-elements related to stress responses, and seven or more kinds of cis-elements involved in light responsiveness. The MeBADH1 promoter also contained a circadian cis-element. MeBADH1 and MeBADH2 showed differential expression patterns in cassava leaf tissue after drought, osmotic, and salt stress treatments ( Fig. 7; Supplementary Tables S19 and S20). MeBADH1 expression was significantly induced by 50%, 16.67%, and 0% under drought, osmotic, and salt stresses, respectively. Meanwhile, induction of MeBADH2 expression was minimal in response to all three stresses (Supplementary  Table S21).

Effect of dehydration stresses on MeABF transcription levels in cassava.
Abiotic stresses, such as drought or salinity, cause intensive losses to agricultural production worldwide 25 . Plants have evolved complex molecular, cellular, and physiological mechanisms to respond to these stresses in order to survive adverse conditions 21 . These mechanisms include signaling pathways, such as those activated by ABA, and other stress-responsive gene families 21 . ABA acts as a growth regulator in response or tolerance to abiotic stressors such as drought, salinity, cold, and heat [26][27][28][29] . The ABF family of bZIP transcription factors functions in ABA signaling pathways and plays an important role in plant responses to stresses 18,21 . Here we found that in various cassava organs, MeABFs exhibited differential expression patterns after drought, osmotic, and salt stress treatments (Figs 2-5), which was consistent with trends we observed in cassava in our previous study that identified genes involved in ethylene signaling in response to dehydration stresses 8 . Our results of this study showed that drought stress in cassava leaves and roots, osmotic stress in cassava tuberous roots, and salt stress in cassava stems induced expression of the highest number of MeABFs in terms of significantly elevated expression (Supplementary Tables S5, S8, S11 and S14). Although many of the MeABFs overlap with AtABF homologs in Arabidopsis in that their expression is induced by similar abiotic stresses and their target genes overlap, differences in their temporal and spatial expression patterns indicate that each has a unique function 16,19,[30][31][32][33] .
The main negative effect of drought on plants is a shortage of water available to tissues. To reduce the amount of water lost to transpiration under drought stress, in cassava plants, leaf stomata partially close and leaf area is reduced via ABA signaling pathway, which restricts the formation of new leaves, results in leaves having a smaller size and that droop, and promotes leaf loss [34][35][36] . Leaf formation is regarded as an important indicator used to assess the drought tolerance of various cassava varieties 37 .
NaCl can negatively affect the energetic, hydric, and nutritional equilibria of plants 38 . Plants grown under salt stress are first affected by water stress and then by sodium and chloride ion-mediated toxicity and nutritive stresses 39 . Salinity can reduce the biomass, leaf area, and photosynthetic rate of cassava plants 40 . Therefore, NaCl not only causes the dehydration of cassava plants, similar to drought stress, but also disturbs other physiological processes and metabolic pathways related to ion homeostasis.
Unlike NaCl, which readily enters the cells to cause toxicity, PEG, a nonabsorbable, non-metabolizable and non-toxic osmotic agent, is often used because of its high molecular weight, which precludes its entry into plant cells via the cell wall 41,42 . Our results indicated that PEG affect plants through a different mechanism than that involved drought and salt stresses; this mechanism awaits further elucidation.

Other environmental factors that affect MeABF transcription in cassava.
Under dehydration stress, circadian clock and light conditions also appear to regulate ABA-mediated gene expression, likely conferring versatile tolerance and repressing growth under stress conditions 20 . Analysis of cis-acting elements in MeABF promoter sequences showed that MeABF transcription can be regulated by circadian clock, light, and hormones, except for by biotic and abiotic stresses (Supplementary Table S2).  www.nature.com/scientificreports www.nature.com/scientificreports/ Circadian clock is an endogenous and cell-autonomous biological timekeeper that generates roughly 24 h rhythms and provides an adaptive advantage by synchronizing physiological and metabolic processes of the plant to the external environment 43 . Plants are usually exposed to drought stress during the day, and water deficit crises would be at maximum levels near the end of the day 44 . Consequently, ABA-dependent drought responses are usually gated primarily around dusk 45 . The rhythmic expression of drought-responsive genes confers rhythmic modulation of drought responses throughout the day, as seen in Arabidopsis and poplar 46,47 . The ABA-inducible MYB96 transcription factor activates TIMING OF CAB EXPRESSION 1 (TOC1) expression by binding directly to its gene promoter 48 . This regulation may be direct, as PSEUDO-RESPONSE REGULATOR 7 directly binds to the promoter region of ABA DEFICIENT 1, which encodes a zeaxanthin epoxidase involved in ABA biosynthesis 49 . The pervasive TOC1 also interacts with ABA INSENSITIVE 3 (ABI3), which has roles in ABA signaling and drought tolerance 50 . Bidirectional interactions between circadian oscillator and output pathways have also been observed in ABA-related physiological processes 51 .
ELONGATED HYPOCOTYL 5 (HY5), a constitutively-nuclear bZIP protein, was the first transcription factor that was found to promote photomorphogenesis and has been extensively studied 52,53 . HY5 physically interacts with B-box 21 (BBX21), BBX22, BBX24 and BBX25, and the bZIP domain of HY5 and the B-box domains of these BBX proteins mediate their interactions [54][55][56][57][58][59] . HY5 also promotes BBX22 expression by directly binding to its promoter 54 , whereas BBX24 and BBX25 repress BBX22 expression by interfering with HY5 transcriptional activity 58 . BBX21 negatively regulates ABI5 expression by interfering with HY5 binding to the ABI5 promoter 60 . In addition, ABI5 can directly activate its own expression, whereas BBX21 negatively regulates this activity through interactions with ABI5. These results indicate that BBX21 coordinates with HY5 and ABI5 on the ABI5 promoter and that these transcriptional regulators work in concert to integrate light and ABA signaling in Arabidopsis 60 . The influence of ABA on phototropin expression is negligible at the mRNA level, but prominent at the protein level, and ABA appears to enhance plant sensitivity to light and promote the chloroplast avoidance response 61 . An influence of ABA on light-mediated chloroplast positioning has also been implicated in succulent plants exposed to water-deficit stress 62 .
Collectively, these observations may support the view that the circadian clock and light mediate the ability of a plant to adapt to daily changes in water status by controlling endogenous MeABF levels and subsequently MeABF gene expression. Interestingly, the cis-acting element ABRE is present in MeABF promoter sequences (Supplementary Table S2). Therefore, MeABFs could likely bind their own promoter to activate their expression as was demonstrated in studies on ABI5 60 . Furthermore, expression of two genes encoding 9-cis-epoxycarotenoid dioxygenase, a key enzyme in ABA biosynthesis, was up-regulated in response to water deficits 63 and might be responsive to ABA itself 64 . These findings indicated that a positive feedback loop controls ABA biosynthesis and ABA signaling during water deficit stress. A more recent study found that ABFs played a role in the negative feedback regulation of ABA signaling by mediating rapid ABA-mediated induction of group A PP2C gene expression 65 .

Effect of dehydration stresses on GB content in cassava leaves.
To cope with harsh habitats such as high salt, drought, heat, and cold, plants have evolved various types of tolerance mechanisms, among which the accumulation of compatible solutes plays a key role in balancing the intracellular osmotic potential of plants 24 . Under conditions of water deficit or salinity stress, plant ABA levels increase dramatically, restricting water loss by stimulating stomatal closure and protecting cellular machinery against dehydration damage by promoting the accumulation of osmo-compatible solutes 44 . GB is regarded as one of the most effective compatible solutes 66 . GB can protect cells from stresses by maintaining an osmotic balance with the surrounding environment and by stabilizing the quaternary structures of complex proteins, such as antioxidant enzymes and the oxygen-evolving PSII complex 67 .
Taken together, we found that cassava leaves had increased GB concentrations in response to drought, osmotic, and salt stresses (Fig. 6, Supplementary Table S17). Furthermore, osmotic stress had the largest impact on the levels of GB content followed by drought stress and then salt stress (Supplementary Table S17). These results suggest that cassava plants may induce expression of BADH genes via the ABA signaling pathway to increase the GB content and adapt to harsh habitats.

MeABf regulation of MeBADH transcription.
In higher plants, BADH catalyzes the key step of GB biosynthesis 23 . In Ammopiptanthus nanus, endogenous expression of AnBADH is strongly induced by exposure to high salt, drought, ABA, heat, or cold 24 . Heterologous expression of AnBADH in Arabidopsis enhanced its tolerance to high salt and drought stresses, suggesting that BADH could play a critical role in plant abiotic tolerance through ABA signaling pathways 24 . An incremental increase in BADH activity induced by ABA treatment of maize also promoted increases in GB content under drought stress 68 . Activated SnRK2s phosphorylate and activate BADH, as do scavenging reactive oxygen species and many other proteins related to abiotic tolerance [69][70][71] . Therefore, we speculate that ABFs may activate BADH production in plants after ABFs are phosphorylated and activated by SnRK2s via the ABA signaling pathway.
To examine this possibility, in this study we assessed transcriptional expression of MeBADH1 and MeBADH2 in cassava ( Fig. 7; Supplementary Tables S19 and S20). MeBADH1 expression was induced in cassava leaves in response to drought and osmotic stresses, whereas MeBADH2 levels were not affected. A similar pattern was observed in Arabidopsis, wherein one BADH family member was targeted to leucoplasts and involved in tolerance to high salt and drought, while another family member was targeted to peroxisomes and did not confer abiotic tolerance [72][73][74] . MeBADH1 promoter sequences contained an ABRE, a cis-acting regulatory element that interacts with ABFs (Supplementary Table S18). Furthermore, in cassava leaves exposed to drought, osmotic, or salt stress, the MeBADH1 expression pattern was concomitant with expression profiles of MeABFs ( www.nature.com/scientificreports www.nature.com/scientificreports/ of MeABFs in cassava plants may activate MeBADH1 transcription by binding to the MeBADH1 promoter in response to ABA signaling, and promote GB biosynthesis and accumulation by mediating increases in MeBADH1 gene expression and MeBADH1 enzymatic activity. These responses could protect cassava plant cells from dehydration stresses by preserving the osmotic balance, thereby improving tolerance of cassava plants to dehydration stresses (Fig. 8).

conclusions
In summary, increasing evidence supports that ABF is an ABA-dependent transcription factor that regulates expression of downstream ABA-responsive genes in response to a variety of abiotic stresses in plants. In this study, we identified seven MeABFs that were distributed unevenly across six chromosomes, and clustered into three groups according to their phylogenetic relationships to Arabidopsis counterparts. Analysis of the 5′-upstream region of MeABFs revealed putative cis-acting elements related to hormone signaling, stress, and light responses, as well as circadian clock. Expression profiles of MeABF genes displayed clear differences among leaf, stem, root, and tuberous root tissues under normal and stress conditions, including drought, osmotic, or salt stress. Drought stress in cassava leaves and roots, osmotic stress in tuberous roots, and salt stress in stems induced expression of the highest number of MeABFs that showed significantly elevated expression. The GB content increased under Figure 8. Proposed mechanism by which ABF promotes plant tolerance to dehydration stress 65 . In response to dehydration stress, ABA is ligated to PYR/PYL/RCAR proteins that interact with PP2Cs and inhibit their activity, in turn activating SnRK2, which phosphorylates ABF. ABF gene and protein expression is promoted by ABF themselves during ABA signaling. ABF promotes rapid BADH protein synthesis and triggers generation and accumulation of GB, which maintains osmotic balance to increase cassava tolerance to dehydration stresses. www.nature.com/scientificreports www.nature.com/scientificreports/ drought, osmotic, or salt stress conditions in cassava leaves. BADH1 is involved in GB synthesis and the promoter sequence of MeBADH1 contains an ABRE cis-acting element that would interact with ABFs. Furthermore, MeBADH1 expression profiles were consistent with those for MeABFs in cassava leaves after the three stress treatments. These findings suggest that MeABFs activate MeBADH1 transcription by binding to the MeBADH1 promoter to induce MeBADH1 production that in turn promotes GB biosynthesis that can preserve the osmotic balance of cassava cells to provide tolerance to dehydration stresses. After the experimental treatments, cassava plants were removed carefully from the plastic pots to allow collection of leaf, stem, root, and tuberous root tissues. All samples were frozen by immersion in liquid nitrogen immediately after collection and stored at −80 °C until analyzed. The cDNA sequences of likely MeABF and MeBADH genes in the cassava cultivar SC5 were cloned, sequenced, and then aligned with their counterparts from the cassava cultivar AM560-2. If these gene sequences showed more than 95% homology to those of the AM560-2, then identification was considered positive. Identified MeABF and MeBADH genes were reannotated and named.

Isolation and identification of
In silico analysis of MeABFs and MeBADHs. Phylogenetic and molecular evolutionary analyses of amino acid sequences of ABFs from cassava and Arabidopsis were performed using the MEGA 6.0 software 75 by the neighbor-joining method 76 with 1,000 bootstrap replicates.
To investigate the cis-elements in the promoter sequences of MeABFs and MeBADHs, the 1.5 kb genomic DNA sequences located the upstream of initiation codon (ATG) representing the core promoter regions of these genes were retrieved from the database of Phyzotome (http://phytozome.jgi.doe.gov/pz/portal.html). These sequences were converted to FASTA formatted sequences, and saved to pure text files. PlantCare software (http://bioinformatics.psb.ugent.be/webtools/plantcare/html) was used to identify the putative cis-acting regulatory elements located within these sequnces. Firstly, click the 'Search for CARE' button, and a new window will appear. Then, click the 'Select file' button, and submit the FASTA formatted sequences. Simultaneously, fill in the email address and the sequence name or ID. Finally, click the 'Search' button, and the predicted results will be obtained. expression analysis of MeABFs and MeBADHs. Total RNA was extracted from the various tissues of cassava plants using the cetyltrimethyl ammonium bromide method 8 . Contaminating DNA in total RNA samples was eliminated by treatment with DNase I (RNase-free; Thermo Scientific, USA). First-strand cDNA was synthesized from 1.0 µg of total RNA using the first-strand cDNA synthesis kit according to the manufacturer's instructions (Thermo Scientific, USA).
Gene-specific primer pairs (Supplementary Table S22) for quantitative real-time PCR (qRT-PCR) were designed using Primer Express Software v3.0 (Thermo Scientific, USA). The PCR mix for qRT-PCR contained 1.0 µl of diluted cDNA, 10 µl of 2× SYBR Green PCR Master Mix (Thermo Scientific, USA), and 200 nM of each gene-specific primer in a final volume of 20 µl. To determine the expression profiles of MeABFs and MeBADHs in cassava from stress-treated and control plants, qRT-PCR was performed using a Stratagene Mx3005P thermal cycler (Agilent Technologies, USA). Three independent biological replicates were performed for each time point of three stress treatments, with three technical replicates per qRT-PCR. The cassava Actin gene (Supplementary Tables S1 and S22) was used as reference gene for the normalization of RNA steady-state level, and the 2 −ΔΔCt method was used to calculate the relative expression values for the target genes 8 .
The log 2 relative expression values were then visualized as heatmaps using the Multiple Experiment Viewer (MeV) software 8 . The absolute value of log 2 relative expression values > 1.0 and P -value < 0.05 (determined by two-tailed Student's t -test) was used as the threshold to assess the significant change in gene expression.
Quantitation analysis of GB. GB content was measured using the commercially available plant GB assay kit (Cat. No. ml036338; Shanghai Enzyme-linked Biotechnology Co., Ltd., China). The GB content of fresh leaves from stress-treated and control cassava plants was measured using an enzyme-linked immunosorbent assay according to the manufacturer's instructions. Statistical analysis was carried out using Student's t -test and assigned P -value < 0.05, which was considered statistically significant.