Expression Profiling of Genes Associated with Cyanogenesis in Three Cassava Cultivars Containing Varying Levels of Toxic Cyanogens ()
1. Introduction
Cassava (Manihot esculenta) is the fourth most important staple food crop in the world after rice, wheat and corn [1]. Cassava roots are consumed daily by more than 500 million people and are the sixth highest source of energy in the world [2]. In agronomical terms, cassava is characterized for its high potential yield, ability to grow in poor soils and drought tolerance [3]. These qualities have made cassava roots an important food crop in the tropics, particularly for subsistence farmers in Africa, who are prompt to suffer droughts and famine.
Cassava has the ability to release hydrogen cyanide (HCN) due to the presence of cyanogenic glycosides linamarin (>90% total cyanogen) and lotaustralin (<10% total cyanogen) in its tissue [4,5]. Cyanogenic glucosides generally function as protective agents against herbivores but more recently has been reported to act as translocable form of reduced nitrogen [4,6,7]. Cyanogenic glucosides are present in all cassava tissue with the exception of seeds, with the leaves having the highest cyanogenic glycoside levels (5 g linamarin/kg fresh weight) and the roots having approximately 20-fold lower linamarin levels [5]. Cyanogenesis in cassava is a complex and variable phenomenon which is influenced by the involvement of different genes and proteins in different tissues, the environment [8] and the developmental stage of the plant [9]. This has lead to the development of a spectrum of cultivars with different levels of cyanogenic glycosides in the roots. Considering the multiple factor involved in cassava cyanogenesis an explanation for the cultivar-dependent differences in root linamarin content is a controversial aspect in cassava biology [4]. In general, based on cyanogenic glucoside levels all cassava cultivars are categorized into 3 groups: low cyanide (≤50 mg linamarin/kg fresh weight), intermediate cyanide (50 - 100 mg linamarin/kg fresh weight) and high cyanide (≥100 mg linamarin/kg fresh weight [10].
Cyanogenic glucoside pathway consists of the initial steps in the synthesis of linamarin which then can branch into two pathways depending if the plant is disrupted or if the plant is intact (Figure 1). In the synthesis of cyanogenic glucoside, valine is converted to linamarin (or lotaustralin whose precursor is isoleucine) by the enzymes CYP79D1/D2, CYP71E1 and UDP-glucosyl transferase [11]. Linamarin is stored in the vacuole and does not come in contact with the cell wall bound enzymes, linamarase and hydroxynitrile lyase (HNL), involved in the breakdown process unless the integrity of the cell is compromised [10] usually by a herbivore. Then linamarase can deglucosylate linamarin into acetone cyanohydrin which in turn is decomposed to produce acetone and HCN by HNL [5,12,13]. The generation of HCN from acetone cyanohydrins may also occur at pH > 5.0 and temperatures > 35˚C. Overall, in intact plants linamarin is primarily synthesized in the leaves and transported to the roots. Due to the cell wall localization of linamarin metabolizing enzymes, an apoplastic mode of linamarin transportation has been suggested. Linustatin, the glycosylated form of linamarin, is thought to be the translocable form, since its presence in other
Figure 1. Cyanogenic glycosides metabolic pathways. Three main sets of reaction are illustrated: 1) the synthesis; 2) the metabolic steps under plant disruption and; 3) the metabolic steps in intact plants. All the reactions have not been studied in detail with most of them being inferred from research in Hevea brasilensis and Sorgum bicolor.
cyanogenic plants has been demonstrated [14-16].
Further support for leaves being the primary site of linamarin synthesis in cassava is derived from the silencing of CYP79D1/D2 genes selectively in leaves of transgenic plants. This resulted in the significant reduction in both leaf and root linamarin content, while a root specific silencing of CYP79D1/D2 led to no change in linamarin accumulation in any tissue [7,17]. Siritunga et al. [17] has shown that linamarin is a mobile source of reduced nitrogen (NH3) to cassava root. Thus, the linamarin is metabolized in differently in intact plants. Selmar [16] by extrapolating the results and evidence found in Hevea brasilensis, suggested that once in the roots, linustatin can be deglucosilated in two different manners by “sequential” and “simultaneous” cleavage. In the sequential cleavage the linustatin is converted back to linamarin and possibly stored, but in the simultaneous reaction the linustatin is converted in acetone-cyanohydrin, which can produce free HCN at pH > 5.0 and temperatures > 35˚C. Therefore the plant needs to re-assimilate the HCN through detoxification reactions to avoid autotoxicity within its cells. Nartey [18] was the first to demonstrate the presence of such a detoxification pathway in cassava when 49% of the 14C-radiolabeled cyanide fed to leaves accumulated as asparagine in two week old cassava seedlings, thus proposing a cyanogen re-assimilation pathway in cassava. Subsequent studies have shown that re-assimilation of cyanide to asparagine occurs through an intermediate, β-cyanoalanine [18-20], in reactions catalyzed by β-cyanoalanine synthase and β- cyanoalanine hydrase [21].
The contribution of each gene involved in the cyanogenic glucoside synthesis, breakdown and re-assimilation pathway to the total cyanogenic glucoside content is unknown. The previous transcriptional studies have focused on a particular gene under a specific condition [6,7,8,17] and never has there been an integral study comparing the activity of the key genes of the cyanogenic glocuside patherway together (Table 1). Here we present the complete tissue-specific expression analysis of key genes in the cyanogenic glucoside pathway in three agronomically important cassava cultivars of varying cyanogen levels (Mcol2215-low level, 60444-mid level, Mtai16-high level) in order to better understand the contribution of each gene to the different cyanogenic glucoside content of cultivars.
2. Methods
2.1. Plant Material and Growth Conditions
Three cassava cultivars with varying levels of cyanogens were used: Mcol2215 (sweet cultivar), 60444 (bittersweet cultivar) and Mtai16 (bitter cultivar). Eight month old roots of Mcol2215, 60444 and Mtai16 have 142, 182 and 569 ppm total cyanogens per gram dry weight, respectively (Teresa Sanchez, CIAT, personal communication). All plants were grown under in vitro conditions in 4E liquid media [17] using 14 h/day photoperiod (5 µmol photons m−2·s−1) at 28˚C.
2.2. RNA Isolation and cDNA Synthesis
Leaves and roots from 4 month old plants were collected from in vitro plants, frozen immediately in liquid nitrogen and stored at −80˚C until use. RNA isolations were performed using RNeasy® Mini Kit followed by the elimination of DNA contaminants by RNAse-Free DNAse treatment according to manufacturers’ recommendations (Qiagen Inc, Valencia, CA, USA). After analysis of quality in a 0.8% agarose gel and quantification, the RNA was protected with 1 unit of RNasin® Ribonuclease inhibitor (Promega Inc, Madison, WI, USA). The cDNA was synthesized from 1 mg of total RNA using Omniscript RT® kit (Qiagen Inc, Valencia, CA, USA) with 1 mM of Oligo (dT)12-18 and 12.5 ng/mL of random primers. The mixture was incubated for 2 hours at 37˚C.
2.3. Primers Design
The genes studied were CYP79D1/D2, linamarase, HNL, b-CAS and 18s rRNA (reference gene). Due to the 85% identity between CYP79D1 and CYP79D2 primers were
Table 1. Summary of previous studies on the expression of key genes involved in cyanogenic glycoside metabolism in cassava.
designed for consensus regions on both genes [11]. Each primer (Table 2) was designed using Primer QuestSM software from IDT (Integrated DNA Technologies Inc, Coralville, IA, USA) and was checked for their compliance to the recommendations of the ABI guidelines for Real-Time PCR Primers and amplicons (Real-Time PCR handbook, University of Illinois, Urbana-Champagne, IL, USA). Subsequently, the primers generated and the original GenBank sequences were analyzed using BLAST program to verify the absence of homologies between the primers and other reported cassava sequences.
2.4. Conventional PCR Conditions
Conventional PCR was performed with 10 mL from a 1:8 cDNA dilution. For the genes CYP79D1/D2, linamarase and β-CAS the reactions contains 1× of PCR Buffer, 2.5 mM of MgCl2, 0.4 mM from each gene-specific primer, 0.2 mM of dNTPs and 1 unit of Taq polymerase in a final volume of 20 mL. For amplification of HNL and 18s rRNA the reaction conditions were similar, except for the concentration of the gene-specific primers were 0.2 mM. The PCR conditions used were an initial denaturation at 96˚C for 5 minutes followed by 35 cycles of 96˚C for 1 minute, specific annealing temperature for each primer (CYP79D: 60˚C, linamarase: 60˚C; HNL: 58˚C; b-CAS: 56˚C and 18s rRNA: 55˚C) for 1 minute and an extension at 72˚C for 1 minutes. Final extension at 72˚C for 5 minutes was also performed for all reactions. All the reactions were run in an Eppendorf Mastercycler® thermocycler (Eppendorf, Hamburg, Germany).
2.5. Quantitative Real Time PCR Conditions
All real time PCR amplifications were performed with 2 mL from a 1:8 cDNA dilution and in a final volume of 20 mL using DyNAmo SYBR® Green qPCR kit (Finnzymes Keilaranta, Finland) and 1× of ROX® (Invitrogen Inc, Carlsband, CA, USA) as internal reference. Varied genespecific primer concentrations were used (0.05 mM of each primer for CYP79D and b-CAS amplification, 0.1 mM of each primer for linamarase and HNL amplification and 0.15 mM of each primer for 18s rRNA amplification). Three technical repetitions (of each sample) were performed. All reactions were conducted in an ABI7300 real-time PCR thermocycler (Applied Biosystems, Foster City, CA, USA) which automatically calculated the dissociation curve. The amplification consisted of an initial denaturation at 94˚C for 5 minutes followed by 30 cycles of 94˚C for 20 seconds (for CYP79D and linamarase) or for 15 seconds (for HNL, b-CAS and 18s rRNA), annealing at 58˚C for 1.25 minutes (for CYP79D1/D2) or 62˚C for 1 minute (for linamarase, HNL and b-CAS) or 53˚C for 35 seconds (for 18s rRNA).
2.6. Data Acquisition
The quality of amplification from each real-time PCR reaction was assessed through the dissociation curves generated by the SDS Software (version 1.3.1) incorporated to the ABI7300 real-time PCR thermocycler. The Ct data were collected using the manual threshold method which posts the threshold in logarithmic phase.
Table 2. Genes studied and their respective primer sequences utilized.
Amplifications with primer-dimmer or non-amplifications were eliminated from the analysis or replaced with cycle number corresponding to the Limit of Detection + 1 cycle (LOD + 1 cycle) in order to correct for off-scale measurements. The LOD is the highest CT which is obtained for a truly positive sample, thus the LOD + 1 corresponds to 50% of the amount that you are able to detect (Mikael Kubista, personal communication).
The following criteria were followed for acquisition of data: 1) if the three technical repetitions amplified clearly, no data was removed or replaced. The Ct for the biological sample was the average from these three technical repetitions; 2) if one technical repetition amplified with some problem it was eliminated from the data. The Ct for the biological sample was the average from the two remaining technical repetitions; 3) if two technical repetitions were not reliable, these were eliminated from the data and the Ct for the remaining replicate was reported for the corresponding biological sample; 4) if none of the three technical repetitions amplified then it was considered that the sample has very low transcriptional activity and the data is replaced with the LOD + 1 cycle. In this experiment the LOD was calculated for the genes CYP79D1/D2 and linamarase using dilutions of amplicons purified.
2.7. Data Pre-Treatment and Statistical Analysis
Two types of univariable comparison for the transcriptional activity for each gene were developed: 1) Comparison between tissues (leaves vs roots) in each cultivar; 2) Comparisons among the three cultivars for each tissue. The Ct data of all the genes were formatted in different matrices depending on the comparisons being conducted. The matrices were n × 5 type where the 5 refers to the number of genes studied (CYP79D1/D2, linamarase, HNL, b-CAS and 18s) and n refers to the number of samples being compared, depending on the corresponding biological replicates (3) and technical replicates (3). Thus, in the comparison between tissues matrices 18 × 5 were obtained, whereas on the comparison among cultivars (in each tissue separated) matrices 27 × 5 were used. The matrices were introduced into GenEx Light® software following the next data pre-processing which are correction for primer efficiency (calculated from standard curves), technical replicates average, correction for reference gene (18s), production of relative quantities for the average of Ct followed by the transformation of data to fold changes using Logarithmic in base 2 (all data preprocessing were conducted using equations described in GenEx® manual). The GenEx-processed data from each matrix were used for statistical analysis according to the corresponding experimental design. In tissue comparison (leaves vs Roots) a t-paired test was conducted while for the comparison between cultivars an ANOVA for completely randomized design and contrast was conducted. The statistical significance used was 10%. The confidence intervals for each mean and for the mean difference were made using 1 − α = 90%.
The overall comparison among the genes as well as the cluster analysis were developed using a multivariable approach. It was a matrix of 57 × 5 that included the Ct data from the three cultivars (including the three biological replicates), the two tissues and an artificial sample with its respective technical replicates (3). The artificial sample is a mixture of the target amplicons purified, which are at the same concentration. The matrix was introduced into GenEx Light® software and followed the same data pre-processing described above, but the relative quantities were obtained with respect to the Ct averages from the artificial sample. These GenEx processed data was used for chart bar visualization of all the genes in the three cultivars in both tissues. For Gene expression clusters according to each cultivar, the data processed from this matrix was separated according to the cultivar. The data from leaves and roots were analyzed together as one vector and thus was not distinguished. For cultivar clusters according to the transcriptional activity of all the genes was used the full data released after GenEx pre-processing. In both cases the clusters were generated with Infostat® for non-standardized data using Euclidean distance and unweighted centroid method.
3. Results
3.1. Comparison of Quantitative Gene Expression within Tissues
In all three cultivars analyzed the expressions of CYP79D1/D2, linamarase and HNL genes were higher in the leaves than the roots (Figure 2; Table 3). In the cases of linamarase and HNL the lowest leaves/roots expression ratio was observed for the low cyanide cultivar Mcol2215, 38.86 and 178.52, respectively. Though melting curves for amplicons were detected for CYP79D1/D2 activity in cassava roots from the three cultivars, its detection was intermittent. The leaves/roots expression ratio of CYP79D1/D2 genes showed a pattern with lowest ratio being in the low cyanogen cultivar and the highest ratio being in the high cyanogen cultivar (Mcol2215 = 4.44, 60444 = 16.44 and Mthai16 = 354.78) (Table 3). The expression of β-CAS gene also showed a pattern with the low cyanogen cultivar Mcol2215 having a higher root expression compared to expression in the leaf (roots/leaves ratio = 4.89). This ratio reduced in the intermediate cyanogen cultivar 60444 to 2.86 with the
Figure 2. Comparison between leaves and roots of the transcriptional activity of the genes CYP79D1/D2, linamarase, HNL and β-CAS in three cassava cultivars. The genes CYP79D1/D2, linamarase and HNL have highest transcriptional activity in leaves in relation to roots while b-CAS gene has more transcriptional activity in roots or at least the same level in both tissues. The error bar is the upper value of 90% confidence interval. A different letter above the bars indicates significant differences (p ≤ 0.10).
Table 3. Ratios of gene expression between leaves and roots in the three cassava cultivars.
high cyanogen cultivar Mthai16 having a leaves/roots ratio of 1.12. Out of the four gene expressions analyzed in this study only β-CAS gene was shown to be having an equal or higher expression level in the roots compared to leaves (Figure 2; Table 3).
3.2. Comparison of Quantitative Gene Expression within Cassava Cultivars
In leaves the average transcriptional activity of all the genes with exception of HNL has a pattern directly related to the cyanogens levels reported for each cultivar (Figure 3; Table 4). In each case, the expression levels in Mtai16 leaves are statistical different (1 − α = 10%)
from Mcol2215, as well as from the cultivar 60444 in CYP79D1/D2, HNL and β-CAS genes. However, no significant differences were found in any case between Mcol2215 and 60444 or between Mtai 16 and 60444 for linamarase gene. When comparing the leaf expression ratio for each gene between any two cultivars, 60444: Mcol215 ratios were the lowest with CYP79D1/D2, linamarase, HNL and β-CAS being 2.82, 1.75, 1.34 and 1.67 respectively (Table 4). In roots the average transcriptional activity of any of the genes did not show any pattern based on the cyanogen content of the cultivars. Though no pattern was observed, statistically significant differences in the gene expression were observed for
Table 4. Ratios of expression among cassava cultivars with different cyanide levels.
Figure 3. Expression pattern of the key genes involved in cyanogenic glycoside metabolism in leaves and roots of three cassava cultivars with different cyanide level. Only illustrated are the gene expression pattern where significant differences were found. The cultivars used were Mcol2215 (142 ppm of HCN), 60444 (182 ppm of HCN) and Mtai16 (569 ppm of HCN). The error bar is the upper value of 90% confidence interval. A different letter above the bars indicates significant differences (p ≤ 0.10).
linamarase and HNL (Figure 3). The expression of linamarase and HNL were significantly different between Mcol2215 and TMS6044. Gene expression of CYP79D1/ D2 and β-CAS in the roots showed no differences between cultivars. Overall, greater differences in gene expression between cultivars were observed in the leaves and not the roots (Table 4).
3.3. Overall Gene Expression Comparison
Figure 4 illustrates a general and comprehensive comparison of the expression levels of the four genes in the leaves and roots of three cultivars having varying levels of cyanogens. With the exception of β-CAS all genes are expressed higher in the leaves compared to the roots. The expression level of β-CAS remained constant between leaves and roots in the three cultivars analyzed. Overall HNL had the highest expression in each of the cultivars irrespective of the tissue while the lowest levels were observed for the expression of CYP79D1/D2 genes.