Analysis of different strategies adapted by two cassava cultivars in response to drought stress: ensuring survival or continuing growth

Plant materials

Two cassava cultivars, Argentina 7 (Arg7) and South China 124 (SC124), were used in the study. Cassava plants were propagated clonally from cuttings of parental stems with at least two nodes and 8cm in length. Plants were grown in plastic pots (25cm height×35cm diameter) in potting mix, which was made by mixing two parts latosolic red soil with one part sand. Two plants were grown in each pot: one Arg7 plant and one SC124 plant. The pots were placed in a non-air-conditioned glass house on the campus of the Chinese Academy of Tropical Agricultural Sciences (Haikou, China). The glass house was covered with a glass roof with a wall that was half open and half sealed with glass to allow air flow. The plants were grown from June to December 2013 during which time the local temperature ranged from 19 to 33 °C and the temperature in the glass house ranged from 20 to 38 °C. No supplementary light was used other than sunlight received in the glass house. The plants were watered with 2 l of tap water per pot two to three times every week and fertilized with a compound fertilizer (N:P:K=1:1:1).

Drought treatment

When the plants were 80 d old after planting the cuttings, they were treated by withholding water. Some pots of plants were rewatered after certain days of the drought treatment as indicated in the experimental results to determine their recovery rates after the stress and for analysis of root proteomes in response to rewatering. The control plants kept receiving water as normal.

Plant tissue collection

For the proteome analysis and linamarin measurement, leaves and roots were harvested at the specified days after the initiation of drought treatment (withholding water). Roots were also harvested from rewatered plants following the drought treatment. From each plant, three leaves (leaves 4, 5, and 6 counting from the top of a plant) were collected. These three leaves were fully or nearly fully expanded. In this way, we excluded older senescing leaves (mainly in the Arg7 plants), which might have very different proteomes and other compositions from other leaves. For the root samples, all roots from a plant were collected and the roots were washed quickly in tap water to remove attached soil. Three biological replicates were generated for each treatment. Each replicate included leaf or root materials from at least five plants to reduce variations from different pots. The leaves and roots were frozen immediately in liquid nitrogen and stored in a –70 °C freezer for further analysis.

Protein extraction

Plant tissues were ground into powder in liquid nitrogen and the powder was mixed with extraction buffer (7M urea, 2M thiourea, 4% CHAPS, 40mM Tris/HCl, pH 8.5) containing 1mM PMSF and 2mM EDTA. After 5min, 10mM dithiothreitol (DTT; final concentration) was added to the samples. The suspension was sonicated at 200W for 15min and then centrifuged at 4 °C at 30 000g for 15min. The supernatant was removed and mixed well with 5 vols of chilled acetone containing 10% (v/v) trichloroacetic acid and incubated at –20 °C overnight. After centrifugation at 4 °C and 30 000g, the supernatant was discarded and the precipitate was washed with chilled acetone three times. The pellet was air dried and dissolved in the protein suspension buffer (7M urea, 2M thiourea, 4% NP-40, 20mM Tris/HCl, pH 8.0–8.5). The suspension was sonicated at 200W for 15min and centrifuged at 4 °C and 30 000g for 15min. The supernatant was transferred to another tube. To reduce disulfide bonds in the proteins, 10mM DTT (final concentration) was added to the protein solution and incubated at 56 °C for 1h. Subsequently, 55mM iodoacetamide (final concentration) was added to the solution, which was incubated for 1h to block reduced thiols in cysteines. The supernatant was mixed well with 5 vols of chilled acetone for 2h at –20 °C to precipitate proteins. After centrifugation at 4 °C and 30 000g, the supernatant was discarded and the pellet was air dried for 5min, dissolved in 500 μl of 0.5M TEAB (Applied Biosystems, Milan, Italy), and sonicated at 200W for 15min. Finally, samples were centrifuged at 4 °C and 30 000g for 15min. The supernatant was transferred to a new tube, quantified by the Bradford assay using BSA as a standard, and kept at –70 °C for further analysis.

iTRAQ labelling of the protein samples

Protein (100 μg) was taken out of each sample and digested with Trypsin Gold (Promega, Madison, WI, USA) at a ratio of protein:trypsin=30:1 at 37 °C for 16h. After trypsin digestion, the peptides were dried by vacuum centrifugation. The peptides were reconstituted in 25 μl of 0.5M TEAB and 60 μl of organic solvent and processed according to the manufacturer’s protocol for 8-plex iTRAQ reagent (Applied Biosystems). Briefly, one unit of iTRAQ reagent was thawed and reconstituted in 24 μl of isopropanol. The samples were labelled with different 8-plex iTRAQ isobaric tags and incubated at room temperature for 2h. The labelled peptide mixtures were then pooled and dried by vacuum centrifugation. The samples were labelled with the iTRAQ tags as follows (sample/isobaric tag): for the leaf samples, Arg7 control/113, SC124 control/116, Arg7 drought-treated/118, SC124 drought-treated/121; for the root samples, Arg7 control/113, SC124 control/114, Arg7 drought-treated/116, SC124 drought-treated/118, Arg7 rewatered/119, SC124 rewatered/121.

The four leaf samples labelled with the four different isobaric tags were pooled together; similarly the six root samples tagged with the six isobaric tags were combined. SCX chromatography was performed with an LC-20AB HPLC pump system (Shimadzu, Kyoto, Japan). The iTRAQ-labelled peptide mixtures were reconstituted with 4ml of buffer A (25mM NaH₂PO₄ in 25% acetonitrile, pH 2.7) and 4ml of the solution was loaded onto a 4.6×250mm Ultremex SCX column containing 5 μm particles (Phenomenex). The peptides were eluted at a flow rate of 1ml min^–1 with a gradient of buffer A for 10min, 5–60% buffer B (25mM NaH₂PO₄, 1M KCl in 25% acetonitrile, pH 2.7) for 27min, and 60–100% buffer B for 1min. The system was then maintained at 100% buffer B for 1min before equilibrating with buffer A for 10min prior to the next injection. Elution was monitored by measuring the absorbance at 214nm, and fractions were collected every 1min. The eluted peptides were pooled into 20 fractions, each of which contained up to 20 μg (for the leaf samples) or 30 μg (for the root samples), and desalted with a Strata X C18 column (Phenomenex) and vacuum dried.

Liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) analysis based on Q Exactive

Each fraction was resuspended in buffer A (2% acetonitrile, 0.1% formic acid) and centrifuged at 20 000g for 10min. The final concentration of peptide was approximately 0.5 μg μl^–1. Ten microliters of supernatant was loaded onto an LC-20AD nanoHPLC (Shimadzu) by the autosampler onto a 2cm C18 trap column. The peptides were then eluted onto a 10cm analytical C18 column (inner diameter 75 μm) packed in house. The samples were loaded at 8 μl min^–1 for 4min, and the 44min gradient was run at 300 nl min^–1 starting from 2 to 35% B (98% acetonitrile, 0.1% formic acid), followed by a 2min linear gradient to 80%, maintained at 80% B for 4min, and finally returned to 5% in 1min. The peptides were subjected to nano-ESI followed by MS/MS in an Q Exactive (Thermo Fisher Scientific, San Jose, CA, USA) coupled online to high-performance LC (HPLC). Intact peptides were detected in the Orbitrap at a resolution of 70 000 full width at half maximum (FWHM). Peptides were selected for MS/MS using the high-energy collision dissociation operating mode with a normalized collision energy setting of 27.0, and ion fragments were detected in the Orbitrap at a resolution of 17 500 FWHM. A data-dependent procedure that alternated between one MS scan followed by 15 MS/MS scans was applied for the 15 most abundant precursor ions above a threshold ion count of 20 000 in the MS survey scan with a dynamic exclusion duration of 15 s. The electrospray voltage applied was 1.6kV. Automatic gain control was used to optimize the spectra generated by the Orbitrap. The automatic gain control target for full MS was 3e⁶ and 1e⁵ for MS2. For MS scans, the scan range was 350–2000 m/z. For MS2 scans, the scan range was 100–1800 m/z.

Mass spectrometer data analysis

Raw data files acquired from the Orbitrap were converted into MGF files using Proteome Discoverer 1.2 (Thermo Scientific) (5600ms converter) and the MGF files were searched. Protein identification was performed using the Mascot search engine (Matrix Science, London, UK; version 2.3.02) against the cassava genome database containing 33 113 annotated cassava proteins (http://www.phytozome.net/cassava.php).

For protein identification, a mass tolerance of 20 ppm was permitted for intact peptide masses and 0.05Da for fragmented ions, with allowance for one missed cleavage in the trypsin digests. Gln→pyro-Glu (N-term Q), Oxidation (M), and Deamidated (NQ) were set as the potential variable modifications, and Carbamidomethyl (C), iTRAQ 8plex (N-term), and iTRAQ 8plex (K) were set as fixed modifications. The charge states of peptides were set to +2 and +3. An automatic decoy database search was performed in Mascot (with a peptide false-discovery rate of <2%) by choosing the decoy checkbox in which a random sequence from the database was generated and tested for raw spectra as well as the real database. To reduce the probability of false peptide identification, only peptides with significance scores (≥20) at the 99% confidence interval by a Mascot probability analysis greater than ‘identity’ were counted as identified. For protein quantitation, it was required that a protein contained at least two identified unique peptides. The quantitative protein ratios were weighted and normalized by the median ratio from Mascot. For a comparison of differences in protein abundance, Student’s t-test was performed using log₂-transformed ratios, followed by Benjamini–Hochberg correction.

Functional annotation of proteins

Functional annotation of proteins was conducted using the Blast2GO program against the non-redundant (nr) NCBI protein database. The KEGG database (http://www.genome.jp/kegg/) and the COG database (http://www.ncbi.nlm.nih.gov/COG/) were used to classify and group these identified proteins.

Linamarin measurement

Leaf numbers 4–6 of each plant (counting from the top) and whole roots were harvested at the indicated days after drought treatment. A half gram of tissue was ground in liquid nitrogen. The power was mixed with pre-warmed 80% methanol and transferred to a 50ml tube and boiled in a water bath for 30min. The mixture was centrifuged at 10 000g for 15min at 4 °C. The supernatant was transferred to a clean tube. One millilitre of the supernatant was dried in a 37 °C oven to remove the methanol and resuspended in 1ml of water. Analytical HPLC was carried out using a Waters e2695 (Waters Corp., Milford, USA) coupled to a 3300 Evaporative Light Scattering Detector (ELSD) (W. R. Grace and Company, Columbia, MD, USA) fitted with an Atlantis C18 column. The measurement was carried out according to previous reports (Siritunga and Sayre, 2003; Jorgensen et al., 2005).

Measurement of Fv/Fm

Ten pots of Arg7 and SC124 plants (one Arg7 and one SC124 plant per pot) were used to measure Fv/Fm, which measures chlorophyll fluorescence. Three leaves of each plant (nos 4–6 counting from the top of the plants) were measured at around 9:00 p.m. after 2h of dark treatment using a MINI-PAM Photosynthesis Yield Analyzer (Heinz Walz GmbH, Effeltrich, Germany) according to the manufacturer’s instruction and a previous report (Calatayud et al., 2000).

Measurement of stomatal aperture

Stomata of the fifth leaf of each plant (counting from the top) were collected at around 9:00 a.m. for measuring stomatal aperture. The sample processing and measurement were carried out using an FV-1000 Confocal Laser Scanning Microscope (Olympus) according to a previously reported method (Calatayud et al., 2000).

Morphological and physiological responses of two cassava cultivars to drought stress

Two cassava cultivars, Arg7 and SC124, were used for comparison in this study. One Arg7 plant and one SC124 plant were planted in the same pot so that their responses to drought stress could be compared under essentially the same soil water level. Arg7 grew slightly bigger and taller than SC124 under normally irrigated conditions (Fig. 1A).

When the plants were 80 d old (counting from the date of planting the stem cuttings), watering was stopped to initiate drought stress treatment. Onset of drought stress became apparent within 10 d after withholding water, but the symptoms were different between the two cultivars. The Arg7 plants started to show yellowing of older leaves but their young leaves remained turgid and green. In contrast, leaves of the SC124 plants displayed a wilting symptom but remained green. Figure 1B and C shows the plants at 14 and 18 d after withholding water, respectively. At d 14, many older leaves of Arg7 were yellowish but shoot growth continued, although at a reduced rate; however, the SC124 plants remained droopy and green and their growth had completely stopped (Fig. 1B and Supplementary Fig. S1 at JXB online). At d 18, several older leaves of Arg7 had dropped off and the plants also started to show wilting, whereas the SC124 plants remained green but droopy (Fig. 1C). Their growth quickly resumed following rewatering after 15 d of withholding water (Supplementary Fig. S1). When the plants were rewatered after 18 d of withholding water, all plants of both cultivars could still be recovered. After the plants were kept in non-watering conditions for an extended period, SC124 plants also dropped some leaves (Fig. 1D). After the plants had been unwatered for 35 d, no Arg7 plants were able to recover, whereas 67% of SC124 plants were able to revive after irrigation was resumed.

The stomatal apertures of the plants were measured during drought treatment. Six days after withholding water, the average stomatal aperture in the stressed SC124 leaves had dropped to approximately 1.3 µm compared with approximately 4.7 µm in normally irrigated plants, whereas the aperture of the stressed Arg7 leaves was around 3.5 µm (Supplementary Fig. S2 at JXB online). At 12 d after withholding water, the stomata of SC124 leaves were almost completely closed, whereas the stomatal aperture of the Arg7 plants dropped to below 3 µm. Stomatal closure is expected to hinder photosynthesis due to lack of CO₂. From the ratio of Fv/Fm, which measures chlorophyll fluorescence to determine effects of the stress on photosystem II, it appeared that 5 d after withholding water, the SC124 plants started to show a more significant drop in the Fv/Fm ratio than Arg7 plants (Fig. 2). The reduction of Fv/Fm was progressive during the 18 d of measurement following water withholding. After that, the plants were rewatered and their Fv/Fm ratios started to return to a relatively normal level in a couple days.

Proteome analysis of leaves and roots of the two cultivars in response to drought stress

To gain a global view of molecular responses to drought stress, we obtained the proteomes of leaves and roots from stressed and normally irrigated (control) plants of these two cultivars. As in the morphological and physiological analysis described above, one Arg7 plant and one SC124 plant were grown in each pot. The drought treatment was also the same as above. Leaves and roots were collected separately at d 14 after the beginning of withholding water, when the plants showed moderate drought stress judging from the morphological and physiological traits. Other pots of plants were rewatered and the roots were collected from the rewatered plants 24h later. For controls, leaves and roots were collected from plants that were the same age but were under the regular watering scheme. All tissue samples were collected around the same time (between 11:00 a.m. and noon). Each sample contained leaves or roots pooled from at least five plants per pot, and three biological replicates were generated for each treatment. Only three leaves (nos 4, 5, and 6 counting from the top of a plant) were collected from each plant to exclude older senescing leaves, which might have very different proteomes from other leaves. Following protein extraction and trypsin digestion, peptides from different tissue samples were tagged with different iTRAQ isobaric tags and identified and quantified using a Q Exactive™ Hybrid Quadrupole-Orbitrap Mass Spectrometer.

The peptides were searched against proteins derived from the cassava genome database (Prochnik et al., 2012). In total, 5445 unique proteins were identified and quantified, among which 3765 proteins were from the leaf samples and 4382 from the root samples. Supplementary Tables S1 and S2 (at JXB online) contain lists of all proteins identified from the iTRAQ analysis (including those identified from one replicate), their abundance differences and the P values in various comparisons between the samples, functional annotations, and the sequences of the identified peptides. Supplementary Table S1 lists those from the leaf samples and Supplementary Table S2 from the root samples. The ID numbers of these proteins are based on the cassava genome database (Prochnik et al., 2012). The differences in protein abundance (fold change) are based on comparisons between drought-stressed and normally watered plants of the same cultivar and between the two different cultivars. Among the proteins that showed a significant change (P<0.05) in abundance, we selected those whose levels differed by at least 1.67-fold as differentially expressed proteins for further analysis and discussion in this report. Supplementary Tables S3 and S4 (at JXB online) contain the lists of all leaf and root proteins, respectively, that were found differentially expressed in at least one of the comparisons. In these tables, a ratio of >1.67 and <0.6 indicates an increase and decrease by over 1.67-fold, respectively, in a comparison between a treatment sample and the control sample. It should be pointed out that, although we applied a 1.67-fold difference as the threshold for differentially expressed proteins for further analysis in this report, those proteins whose abundance change was less than 1.67 fold with a P value of <0.05 could also be truly differentially expressed.

Comparisons of leaf proteomes

Among the leaf proteins showing a >1.67-fold difference, 337 were found to be upregulated and 224 downregulated in the drought-stressed leaves compared with the control leaves for at least one of the two cultivars (Fig. 3A and Supplementary Table S3). In total, 102 proteins showed higher levels (Fig. 3A, red arrows) and 68 proteins showed lower levels (Fig. 3, green arrows) under drought conditions in both cultivars. The number of drought-responsive proteins was higher in SC124 (439) than in Arg7 (292), indicating that the difference between the stressed leaves and normal leaves was more pronounced in SC124 than in Arg7, which is consistent with the more pronounced differences in the morphological and physiological traits in the former under drought stress.

Among the proteins showing a reduced level in the stressed leaves, the most common functional group was the proteins involved in primary metabolism. The drought-responsive proteins involved in photosynthesis were all found to be downregulated (except PsbP-1; see below) in the drought-stressed leaves compared with the control plants (Table 1 and Supplementary Table S3). These included the proteins/enzymes involved in the photosynthetic light reaction and electron transfer (such as chlorophyll A/B-binding protein and a few photosystem I and II subunits), carbon fixation (Rubisco activase), and the Calvin cycle (fructose-1,6-bisphosphatase and fructose-bisphosphate aldolase). These proteins were reduced by around 2-fold in the stressed SC124 leaves. They were also reduced in the stressed Arg7 leaves compared with their control plants but to a lesser extent (Table 1), further indicating that photosynthesis was affected more in SC124 than in Arg7 by the drought stress. However, PsbP-1, a component of photosystem II that is required for assembly/stability of the oxygen-evolving complex, was upregulated in the stressed leaves of both cultivars. The levels of many other primary metabolic enzymes such as a phosphoglycerate mutase involved in glycolysis and an aspartate aminotransferase and asparagine synthetase 1 involved in amino acid biosynthesis were also reduced in the drought-stressed leaves of both cultivars.

Table 2 lists the top 10 most strongly up- or downregulated proteins in the stressed leaves in the two cultivars. The proteins that were strongly up- or downregulated proteins in one cultivar were often strongly up- or downregulated, respectively, in the other cultivar, but with some exceptions (Table 2 and Supplementary Table S3). For instance, a linamarase, the cyanogenic glycoside breakdown enzyme, was the most strongly (20-fold) reduced protein in the stressed SC124 leaves but was not significantly changed in the Arg7 leaves in response to the stress.

The most common types of proteins that showed higher levels in the stressed leaves than in the controls included those involved in secondary metabolism (such as in the flavonoid and phytoalexin biosynthetic pathways), the antioxidant system (such as superoxide dismutase, thioredoxin, and glutathione S-transferase), several proteases, and in modifications of cell-wall components (Supplementary Table S3). Many of these proteins are known to be associated with abiotic and biotic stress responses, and activation of the antioxidant system is also considered a key factor for drought tolerance in cassava (Turyagyenda et al., 2013). The levels of cyanohydrin UDP-glucosyltransferase (UGT85K4) were significantly increased in the stressed leaves of SC124 but were not significantly changed in Arg7 (Supplementary Table S3). UGT85K4 and methylbutanal oxime monooxygenase (cytochrome P450 protein CYP71E) are the key enzymes in biosynthesis of linamarin.

Although some antioxidant enzymes were found to be upregulated, others were downregulated in the stressed leaves, suggesting a dynamic feature of the cellular redox status. Similarly, some heat-shock proteins were upregulated but others were downregulated under the drought stress. Other proteins upregulated by the drought stress include two myo-inositol 1-phosphate synthases that are involved in inositol phosphate metabolism, 1-aminocyclopropane-1-carboxylate oxidase involved in ethylene formation, and a few proteins involved in vesicle trafficking (such as ENTH/VHS family protein, two dynamin proteins, and a clathrin adaptor complex protein). In addition, a number of putative regulatory proteins were identified as upregulated proteins in the stressed leaves, which included several protein kinases, two 14-3-3 proteins, two leucine-rich repeat-containing proteins, three RNA-binding proteins, a winged-helix DNA-binding transcription factor, and a general transcription factor. Two histone deacetylases were upregulated in the stressed SC124 leaves but showed no significant change in the Arg7 leaves.

Unexpectedly, a nitrate reductase and a sucrose synthase were among the most strongly upregulated proteins in the stressed leaves, although the plants were expected to have reduced levels of nitrate uptake and glucose formation from photosynthesis. In addition, their levels were increased even more in the SC124 leaves than in the Arg7 leaves under drought stress (Table 2). A nitrate reductase and a sucrose synthase were also upregulated in the stressed roots of SC124 (see below).

Comparisons of the root proteomes

The total number of proteins identified and quantified from the root samples (4382) was slightly higher than that from the leaf samples (3765). However, only 129 root proteins were found to show at least a 1.67-fold change (with a P value of <0.05) between the drought-stressed roots and normal roots (Fig. 3B and Supplementary Table S4), which is much lower than the number of the differentially expressed proteins in the leaves (561) with the same threshold level of change. Among the drought-responsive proteins in the roots, 80 were upregulated and 49 were downregulated. Although many more proteins showed drought-responsiveness in the SC124 leaves than in the Arg7 leaves, the number of drought-responsive proteins in the SC124 roots (51) was lower than in the Arg7 roots (103). Seventeen proteins were upregulated and eight downregulated in both cultivars. Table 3 contains a list of the 10 most strongly up- or downregulated proteins in the roots under drought conditions in at least one of the cultivars.

Most common functional categories of the drought-responsive proteins in the roots included those involved in secondary metabolism, pathogenesis-related (PR) proteins (such as a few PR2 family glucanases), proteases, and cytoskeleton (such as a few tubulins). Among the most strongly upregulated proteins by drought in the stressed roots of both cultivars were those involved in biosynthesis of linamarin including P450 71E7 (CYP71E7) (Table 3) and UGT85K4 (Supplementary Table S4). The latter was also increased in the stressed leaves.

Other proteins that were upregulated in the stressed roots included an inorganic phosphate transporter, a RNA-binding protein, aquaporin, two SAM-dependent methyltransferases, several proteases, thioredoxin, heat-shock proteins, a starch synthase, and a CLAVATA1-like receptor kinase (Supplementary Table S4 and Table 3). The aquaporin was induced more in Arg7 (>3-fold) than in SC124 (1.85-fold) under drought. The downregulated proteins under the stress conditions included a serine-threonine protein kinase, several cell-wall modification enzymes, and two germin-like proteins. A linamarase was reduced in the stressed Arg7 roots but unchanged in the SC124 roots. The comparison of the proteomes between the drought-stressed roots and the roots from the plants 24h after rewatering showed a much smaller number of proteins that responded to rewatering (Fig. 3C and Supplementary Table S4). In Arg7, 14 proteins were found to be downregulated, but no protein was found to be significantly upregulated. In SC124, 10 proteins were upregulated and 10 downregulated. A linamarase was induced by 2.2-fold by rewatering in the SC124 roots but was unchanged in the Arg7 roots. A transducin/WD40 repeat protein, a RNA-binding (RRM/RBD/RNP motifs) protein, and a transcription factor were reduced by rewatering in the SC124 roots.

Measurement of linamarin levels

The proteome analysis showed that the enzymes involved in linamarin biosynthesis were present at higher levels in the stressed SC124 leaves and in the stressed roots of both cultivars. In the stressed SC124 leaves, a linamarase was sharply reduced but was little changed in Arg7. These results suggested that drought enhanced linamarin accumulation, particularly in the SC124 leaves.

We determined the levels of linamarin in normally irrigated and drought-stressed leaves and roots of both cultivars (Fig. 4). It was found that the linamarin levels in the control leaves were similar between the two cultivars. In the stressed leaves of Arg7, the linamarin level was slightly lower than in the control leaves, but in the stressed SC124 leaves, the linamarin level was higher than in the control leaves. In roots, SC124 had a higher level of linamarin than Arg7 under normally irrigated conditions. The drought stress led to a higher level of linamarin in the Arg7 roots, whereas in the stressed SC124 roots, the level was lower than that in the control roots.