• Open Access

Targeted silencing of BjMYB28 transcription factor gene directs development of low glucosinolate lines in oilseed Brassica juncea

Authors


Correspondence (fax +91 11 26741658; email ncbisht@nipgr.ac.in)

Summary

Brassica juncea (Indian mustard), a globally important oilseed crop, contains relatively high amount of seed glucosinolates ranging from 80 to 120 μmol/g dry weight (DW). One of the major breeding objectives in oilseed Brassicas is to improve the seed-meal quality through the development of low-seed-glucosinolate lines (<30 μmol/g DW), as high amounts of certain seed glucosinolates are known to be anti-nutritional and reduce the meal palatability. Here, we report the development of transgenic B. juncea lines having seed glucosinolates as low as 11.26 μmol/g DW, through RNAi-based targeted suppression of BjMYB28, a R2R3-MYB transcription factor family gene involved in aliphatic glucosinolate biosynthesis. Targeted silencing of BjMYB28 homologs provided significant reduction in the anti-nutritional aliphatic glucosinolates fractions, without altering the desirable nonaliphatic glucosinolate pool, both in leaves and seeds of transgenic plants. Molecular characterization of single-copy, low glucosinolate homozygous lines confirmed significant down-regulation of BjMYB28 homologs vis-à-vis enhanced accumulation of BjMYB28-specific siRNA pool. Consequently, these low glucosinolate lines also showed significant suppression of genes involved in aliphatic glucosinolate biosynthesis. The low glucosinolate trait was stable in subsequent generations of the transgenic lines with no visible off-target effects on plant growth and development. Various seed quality parameters including fatty acid composition, oil content, protein content and seed weight of the low glucosinolate lines also remained unaltered, when tested under containment conditions in the field. Our results indicate that targeted silencing of a key glucosinolate transcriptional regulator MYB28 has huge potential for reducing the glucosinolates content and improving the seed-meal quality of oilseed Brassica crops.

Introduction

Glucosinolates are a group of amino acid-derived plant secondary metabolites distinctive to the order Capparales. This order includes the family Brassicaceae, containing agriculturally important vegetables, condiments and oilseed crops and the model species Arabidopsis thaliana. Glucosinolates are broadly classified into three major groups, namely aliphatic, indole and aromatic glucosinolates, based on their precursor amino acid. The biosynthesis of both aliphatic and indole glucosinolates can be divided into three steps: (i) side-chain elongation of precursor amino acid, (ii) core-glucosinolate structure formation and (iii) modification of side group (Halkier and Gershenzon, 2006; Wittstock and Halkier, 2002). Together with side-chain elongation of the R-group, side-chain modifications generate a plethora of diverse glucosinolate compounds, with more than 200 structures identified till date (Clarke, 2010).

In recent years, glucosinolate research has gained increasing importance as these metabolites have both useful and harmful effects. Besides imparting distinct flavour to the cruciferous vegetables, certain glucosinolates and their degradation products have been reported to act as anti-carcinogenic and chemo-protective agents in mammalian systems (Fahey et al., 1997; Hayes et al., 2008). Both aromatic and indole glucosinolates are known to play important roles in plant defence against pathogens and herbivores (Clay et al., 2009; Hopkins et al., 2009). However, on the negative side, some of the aliphatic glucosinolates and their degradation products in seed meal are associated with anti-nutritional properties (Cartea and Velasco, 2008; Fahey et al., 2001; Mawson et al., 1993). For example, 2-hydroxy-3-butenyl glucosinolate (progoitrin) degrades to goitrogenic products while other glucosinolate such as 3-butenyl and 4-pentenyl produce isothiocyanates, which ultimately reduce the meal palatability (Campos de Quiros and Mithen, 1996) and food intake of cattle, swine and poultry.

Over the last few decades, one of the major breeding objectives in oilseed Brassica crops (B. rapa, B. napus, and B. juncea) has been the improvement of its seed-meal quality through the development of lines having total glucosinolate <30 μmol/g DW in seed-meal (Potts et al., 1999). Canola quality lines (having low glucosinolate and low erucic acid in seed) have been successfully developed in B. napus and B. rapa. The development of these varieties with major improvements in agronomic, oil and meal quality has greatly influenced the rapid increase of canola consumption over the last two decades mainly in the North American continent and the European Union. In contrast, the B. juncea gene pools contain significantly higher amounts of seed aliphatic glucosinolates, ranging from 80 to 120 μmol/g DW (Pradhan et al., 1993; Sodhi et al., 2002). The presence of such a high quantity of seed aliphatic glucosinolates limits the value of this crop in the international market both as edible oil and animal feed. However, B. juncea has several potential advantages over B. napus (canola rapeseed) as a global oilseed crop of semi-arid regions, viz., more vigorous seedling growth, less pod shattering problem, greater tolerance to heat and drought and enhanced resistance to the blackleg fungus, Leptosphaeria maculans (Burton et al., 1999; Potts et al., 1999).

Till date, no productive and agronomically viable low glucosinolate line has been reported in B. juncea (Indian mustard). Glucosinolate trait in B. juncea follows a complex inheritance pattern being controlled by multiple quantitative trait loci, QTL (Bisht et al., 2009; Lionneton et al., 2004; Mahmood et al., 2003; Ramchiary et al., 2007; Sodhi et al., 2002). Even though a canola quality line is available in exotic gene pool of B. juncea (Burton et al., 1999; Potts et al., 1999), the introgression of low glucosinolate alleles into well adapted Indian B. juncea cultivars through conventional breeding has been largely impeding. Some of the inherent problems associated with breeding low glucosinolate QTLs were the existence of epistatic QTL, and context-dependent interactions of the loci (Bisht et al., 2009; Ramchiary et al., 2007). In addition, a significant amount of linkage drag from donor genome was also reported because of which negative linkage between QTL alleles of low glucosinolate and yield-related QTL in few linkage groups was observed (Ramchiary et al., 2007). Thus, recombination-based transfer inevitably requires fine mapping of each QTL and screening a large number of progeny in every back-cross generation, which makes the approach time-consuming, labour-intensive and more importantly germplasm dependent.

Alternatively, transgenic technologies could be exploited to engineer low glucosinolate lines in oilseed Brassica crops. Our understanding of genes involved in the complex glucosinolate biosynthesis pathway has been possible from accumulating studies in the closest model species A. thaliana (Grubb and Abel, 2006; Halkier and Gershenzon, 2006; Sønderby et al., 2010b). Briefly, till date, >20 biosynthesis pathway genes and three transcriptional regulator genes have been reported to control the aliphatic glucosinolate biosynthesis in A. thaliana (Figure S1, Gigolashvili et al., 2007, 2009; Hirai et al., 2007; Sønderby et al., 2007, 2010a). However, because of the inherent polyploidy associated with the Brassica crops, multiplicity of glucosinolate biosynthesis and regulatory genes are quite expected. Recently, multiple homologs of glucosinolate candidate genes were observed in B. rapa genome (Wang et al., 2011; Zang et al., 2009). It is likely that allotetraploid B. juncea and B. napus might have even greater inventory of glucosinolate candidate genes, possibly resulting in more complex regulatory and metabolic networks in these oilseed crops. In view of the above, metabolic engineering of low glucosinolate content in B. juncea is a major challenge for plant biotechnology and requires a robust genetic engineering strategy.

Here, we describe the development of low glucosinolate B. juncea transgenic lines, having seed glucosinolate content <30 μmol/g DW, through RNAi-based targeted suppression of a key glucosinolate transcriptional regulator, BjMYB28. The development of low glucosinolate B. juncea lines will significantly improve its seed-meal quality and raise its international trade potential both as oilseed and feed crop.

Results and discussion

BjMYB28-3 was differentially expressed in glucosinolate contrasting lines of B. juncea

In order to study the differential regulation of glucosinolate biosynthesis in B. juncea cultivars, we selected a well adapted high glucosinolate Indian line (Varuna) and the canola quality exotic line (Heera) and measured their leaf and seed glucosinolate contents and profiles. In general, the Indian line showed a very high amount of total glucosinolates both in leaves and seeds compared to the canola quality line (Table S1). More than 95% of the total glucosinolate in B. juncea was found to be aliphatic glucosinolates. The two B. juncea lines were highly contrasting for aliphatic glucosinolate pools both in leaves and seeds; however, the nonaliphatic glucosinolate pools were found to be almost similar. For example, the total aliphatic seed glucosinolate content of B. juncea Indian line was found to be 105.12 ± 5.8 μmol/g DW, where as the canola quality line showed a total aliphatic seed glucosinolate content of 12.46 ± 0.9 μmol/g DW (Table S1).

Recent reports in Arabidopsis showed that levels of aliphatic glucosinolate are transcriptionally regulated by at least three members, namely AtMYB28, AtMYB29 and AtMYB76 of subgroup-12 of R2R3-MYB transcription factor family genes (Gigolashvili et al., 2007, 2009; Hirai et al., 2007; Sønderby et al., 2007, 2010a). These studies concluded that AtMYB28 is the principal regulator of aliphatic glucosinolate biosynthesis and affects the production of both short- and long-chain aliphatic glucosinolates, whereas AtMYB29 and AtMYB76 (encoded by tandemly duplicated genes in A. thaliana) are additional independent control elements affecting biosynthesis of only short-chain aliphatic glucosinolates.

We therefore investigated the inventory of these key regulators for aliphatic glucosinolates in allopolyploid B. juncea. Four full-length homologs of the Arabidopsis AtMYB28 (At5 g61420) were isolated from B. juncea high glucosinolate line Varuna, namely BjMYB28-1, BjMYB28-2, BjMYB28-3 and BjMYB28-4 (Augustine et al., communicated; accession nos. JQ666166, JQ666167, JQ666168, JQ666169). In contrast, only two full-length BjMYB29 homologs viz., BjMYB29-1 and BjMYB29-2 (accession nos., JX316031, JX316032) were isolated from B. juncea (Figure S2). We could not identify any MYB76-specific orthologs from B. juncea genome, which was also found to be absent in B. rapa genome (Wang et al., 2011; Zang et al., 2009).

The expression profile of BjMYB28 and BjMYB29 homologs across different tissue types was determined in two B. juncea lines contrasting for glucosinolate contents. Differential expression of BjMYB28 homologs was observed across the developing stages of glucosinolate contrasting lines, wherein the fold difference was found to be highest in the developing silique (Figure 1a–d). Interestingly, expression of one of the homolog, namely BjMYB28-3, was observed exclusively in the high glucosinolate line, whereas the other three BjMYB28 homologs showed expression in both the B. juncea lines, even though the levels of transcripts were variable. The BjMYB28-3 transcript was undetectable in the low glucosinolate canola quality line, despite using primers from different regions of the same gene (Figures 1c and S3). The expression of two BjMYB29 homologs was found to be significantly lower compared to that of BjMYB28 homologs across the developing stages of B. juncea (Figure S4). Our observation was in accordance with that of Arabidopsis, wherein AtMYB29 showed lower expression level compared to the AtMYB28 across all the developmental stages (Gigolashvili et al., 2009).

Figure 1.

Expression analysis of BjMYB28 homologs between Brassica juncea lines contrasting for total glucosinolate content. qRT-PCR analysis of (a) BjMYB28-1, (b) BjMYB28-2, (c) BjMYB28-3 and (d) BjMYB28-4 homologs. Seedling stage of B. juncea line, Heera was used as a reference calibrator (set as 1). The qRT-PCR experiments were conducted with three independent sets of plants and values represent mean ± standard error. Significant difference in expression profile of BjMYB28 homologs of high glucosinolate line Varuna in comparison with the low glucosinolate line Heera at different stages are indicated by asterisks (*P < 0.05, **P < 0.01 in Fishers LSD test determined by ANOVA).

The differential expression of BjMYB28-3 across glucosinolate contrasting lines of B. juncea could be an important cue suggesting its involvement towards controlling glucosinolate variation. In order to investigate the differential regulation of BjMYB28-3 expression, we performed Southern blot analysis on glucosinolate contrasting B. juncea cultivars using probe from 5′ upstream region specific to BjMYB28-3 homolog. The low glucosinolate cultivar of B. juncea did not show any hybridization signal (Figure S5). Very recently, genomic deletions underlying the two QTLs for seed glucosinolate contents in B. napus were identified using associative transcriptomics approach (Harper et al., 2012). The deleted regions of B. napus were found to contain orthologs of AtMYB28. Our data also suggest that there could be genomic deletion of BjMYB28-3 homolog in the low glucosinolate cultivar, Heera. The Polish spring rape variety ‘Bronowski’ is regarded as the sole donor source of low glucosinolate trait to the present day ‘canola’ quality cultivars of Brassica species (Bisht et al., 2009; Potts et al., 1999). With the advent of genomics tools, genetic dissection of glucosinolate QTLs across Brassica species can be undertaken to identify the historically important ‘Bronowski gene(s)’ controlling the low glucosinolate trait in Brassica crops.

Additionally, mapping and association of pathway genes with seed-glucosinolate QTLs in B. juncea and B. napus also suggested that the MYB28 orthologs contribute prominently towards controlling variability in glucosinolate contents across these complex allopolyploid genomes (Bisht et al., 2009; Feng et al., 2012; Ramchiary et al., 2007). For example, in B. juncea, two of six QTLs identified for seed-glucosinolate trait harbour BjMYB28 homologs (Bisht et al., 2009; Ramchiary et al., 2007). Further, one of the A-genome-specific QTLs identified in B. juncea was orthologous to that identified in B. napus genome (Feng et al., 2012). Thus, in accordance with our data, it seems that MYB28 orthologs could be important genetic determinants controlling the variation of seed-glucosinolate content across Brassica species.

Development of low glucosinolate transgenic lines in B. juncea

Our initial investigations led us to a tentative conclusion that BjMYB28-3 could be a potential candidate for the development of low glucosinolate lines in B. juncea. An intron spliced hairpin RNAi (ihpRNAi) construct (Figure 2a) was therefore developed, wherein a 342 bp region downstream to the conserved R2R3 repeats of BjMYB28-3 homolog was used, hereafter referred to as BjMYB28(RNAi) construct. The BjMYB28-3 target sequence used to develop the said construct showed 77.0–87.4% sequence identity to other three BjMYB28 homologs and shared lower levels of sequence identity with the two BjMYB29 homologs (56.4–61.5%) (Figure S6). Because MYB28 is a member of large MYB transcription factor family genes, engineering with a strong, constitutive promoter could provide off-shoot targets and alter other biological processes which may be deleterious to the plant. Therefore, to ensure controlled and specific silencing across the plant developmental stages and cell-types, the said cassette was driven by the native promoter of BjMYB28-3 gene, the homolog showing differential expression profile between the contrasting glucosinolate lines of B. juncea (Figure 1c). The transformation constructs were introduced into the well-adapted high seed glucosinolate Indian line of B. juncea (cv. Varuna; total seed glucosinolate content ca. 107.52 μmol/g DW) through Agrobacterium-mediated genetic transformation. A total of 36 and 15 primary transgenic (T0) lines were generated using the BjMYB28(RNAi) and vector control constructs, respectively. Southern blot analysis of the T0 transgenic events showed stable integration of T-DNA cassette in variable copies (Figure 2b). The transgenic lines were maintained by selfing in each generation under the containment field conditions and were analyzed for glucosinolate phenotype.

Figure 2.

Development and characterization of BjMYB28(RNAi) transgenic lines. (a) Map of the T-DNA construct of BjMYB28(RNAi) transformation vector. A 342 bp region of the BjMYB28-3 gene was cloned in sense and antisense orientations, with gene-specific second intron (409 bp) as spacer, under the control of native promoter. (b) Southern blot analysis of BjMYB28(RNAi) transgenic lines. A 390 bp fragment of bar gene was used as the probe. Asterisks (*) represent transgenic lines with single-copy integration of T-DNA.

Transgenic lines developed were initially subjected to near infrared reflectance spectroscopy (NIRS) to estimate total glucosinolate contents in T1 seeds. The total seed glucosinolate content in the BjMYB28(RNAi) transgenic lines ranged from 15.21 to 124.76 μmol/g DW, suggesting variable degree of silencing efficiency in these lines (Table S2). We classified the transgenic lines into four different classes based on their seed glucosinolate content (Figure 3a) viz., class I (<30 μmol/g DW; the internationally acceptable limit), class II (31–60 μmol/g DW), class III (61–90 μmol/g DW) and class IV (>90 μmol/g DW). Ten (30.6%) transgenic lines generated using the BjMYB28(RNAi) construct showed seed glucosinolate content <30 μmol/g DW (hereafter, referred as low glucosinolate transgenic lines). Further, a significant reduction in total seed glucosinolate content (down to 86% of the wild-type level) suggested a very high silencing efficiency of the BjMYB28(RNAi) construct. All transgenic lines developed using the vector control construct showed seed glucosinolate content similar to that of transformation host, Varuna (107.52 ± 5.90 μmol/g DW). Transgene-induced RNAi has been shown to be an effective mechanism for silencing un-desirous genes/traits in plants; however, a great deal of variation in the silencing efficacy and silencing-induced phenotype among transgenic events has been reported in crop plants (Ali et al., 2010; Mansoor et al., 2006).

Figure 3.

Average seed-glucosinolate estimation of BjMYB28(RNAi) transgenic lines. (a) Frequency (%) of T0 plants within different classes based on total seed glucosinolates content in T1 seeds. The T0 transgenic plants obtained using BjMYB28(RNAi) and vector control constructs were categorized under four different classes viz., <30, 31–60, 61–90, and >90 μmol/g DW. (b) Comparison of average total glucosinolate observed in T1 and T2 seeds of single-copy transgenic events having total seed glucosinolate <60 μmol/g DW. T2 seeds of five independent T1 plants were analyzed for the total glucosinolate content using NIRS, in replicates.

The BjMYB28(RNAi) lines showed reduced accumulation of aliphatic glucosinolates without compromising the nonaliphatic glucosinolate pool

In order to determine the effect of BjMYB28(RNAi) construct on suppression of total seed glucosinolates, we analyzed representative transgenic lines for individual glucosinolate components. HPLC analysis was carried out to estimate the glucosinolate profile in seeds of nine representative single-copy transgenic lines, which showed >50% reduction in the total glucosinolate content (i.e. class I and class II) compared to the wild-type plants in initial NIRS analysis. HPLC estimation of component glucosinolates in T1 seeds of these nine transgenic lines showed significant reduction in total aliphatic glucosinolate levels (Table 1). Gluconapin (3-butenyl), the predominant aliphatic glucosinolate which is also the precursor of progoitrin (a major anti-nutritional glucosinolate) was reduced to as low as 9.51 ± 1.06 μmol/g DW in the transgenic lines (wild-type level: 100.03 ± 7.77 μmol/g DW). Similarly, sinigrin (2-propenyl), the second major seed glucosinolate (wild-type levels: 14.66 ± 1.56 μmol/g DW) was also reduced to 1.16 ± .07 μmol/g DW in the transgenic lines. In addition to these major anti-nutritional glucosinolates, other aliphatic glucosinolate fractions were also reduced significantly (down to ca. 84%) in the BjMYB28(RNAi) lines compared to that of wild-type (Varuna) and vector control lines. However, the transgenic lines showed only marginal changes in the total indole glucosinolates content (Table 1), thereby reflecting highly specific suppression of seed aliphatic glucosinolates by the BjMYB28(RNAi) construct.

Table 1. Average glucosinolate contents and profile (in μmol/g DW) in T1 seeds of BjMYB28(RNAi) transgenic lines. The glucosinolate profiles of single-copy transgenic lines with total glucosinolates <60 μmol/g DW were estimated using HPLC. Value represent mean ± SD of three independent biological replicates
Plant codeAliphatic glucosinolatesNonaliphatic glucosinolatesTotal glucosinolates
Sinigrin (2-propenyl)Gluconapin (3-butenyl)Glucobrassica-napin (4-pentenyl)
  1. a

    Significantly different (P < 0.01) in comparison with wild-type control (Varuna) plants.

Control14.66 ± 1.56100.03 ± 7.771.39 ± 0.121.95 ± 0.26118.05 ± 6.67
RNAi-21.58 ± 0.41a15.97 ± 0.70a0.47 ± 0.12a1.72 ± 0.4319.83 ± 0.97a
RNAi-36.90 ± 2.90a47.56 ± 3.68a0.93 ± 0.26a1.81 ± 0.4457.20 ± 3.55a
RNAi-51.98 ± 0.44a16.40 ± 1.01a0.42 ± 0.12a1.80 ± 0.3420.74 ± 1.06a
RNAi-92.70 ± 1.14a17.78 ± 1.06a0.55 ± 0.26a1.71 ± 0.3622.90 ± 2.50a
RNAi-122.68 ± 0.50a22.94 ± 2.40a0.60 ± 0.18a1.98 ± 0.4028.21 ± 2.41a
RNAi-151.16 ± 0.07a9.51 ± 1.06a0.23 ± 0.14a1.46 ± 0.3212.58 ± 1.20a
RNAi-171.83 ± 0.66a15.54 ± 0.89a0.42 ± 0.22a1.67 ± 0.4219.48 ± 0.42a
RNAi-185.47 ± 2.40a43.43 ± 5.32a0.83 ± 0.25a1.78 ± 0.4551.67 ± 4.42a
RNAi-192.17 ± 0.41a21.90 ± 4.83a0.63 ± 0.32a1.96 ± 0.4626.67 ± 5.05a

In recent years, both glucosinolate content and profiles in oilseed B. napus has been engineered by silencing Brassica homologs of aliphatic glucosinolate pathway genes including MAM3 and GSL-ALK (Liu et al., 2011, 2012). RNAi-based silencing of MAM3 (Methylthio-alkyl-malate synthase), a biosynthetic pathway gene involved in the chain elongation step of aliphatic glucosinolate biosynthesis, was used to reduce the aliphatic glucosinolate contents in B. napus seeds (Liu et al., 2011). Although total seed glucosinolate contents in MAM3(RNAi) transgenic plants were found to be reduced in B. napus canola (low glucosinolate) line; silencing of MAM3 in high glucosinolate background of B. napus could elicit only marginal decrease in total glucosinolate contents. The MAM3(RNAi) transgenic lines showed induced production of 2-propenyl glucosinolate while reducing the level of both C4 (2-hydroxy-3-butenyl and 3-butenyl) and C5 (5-methylsulfinylpentyl and 4-pentenyl) glucosinolates. In contrast, our results clearly demonstrated that the targeted silencing of a key transcriptional regulator, BjMYB28 in a high glucosinolate background of oilseed B. juncea could provide significant and stable suppression of total seed glucosinolate content (including C3, C4 and C5 glucosinolates) in the transgenic lines, without affecting the overall profile of aliphatic glucosinolate components.

One of the greatest applicability of transgene-induced RNAi is towards multigene families and polyploids, as it is not straightforward to create loss-of-function phenotype (knock-out) of multiple genes and their homologs by conventional breeding approaches, particularly, if members of the gene family are tightly linked (Lawrence and Pikaard, 2003). Thus, for agronomical traits like glucosinolates which are quantitative, sufficient degree of suppression of a target gene (as well as its multiple homologs) may be required for stable propagation of low glucosinolate trait in subsequent generations.

The low glucosinolate trait was stable in subsequent generations

To check the stability of BjMYB28 silencing, nine BjMYB28(RNAi) single-copy transgenic lines showing >50% reduction in total seed glucosinolate were analyzed in subsequent generations under containment field conditions. The T2 seeds derived from independent Basta resistant T1 progeny of each transgenic event showed glucosinolate levels comparable to or lower than that of T1 seeds (Figure 3b; Table S3), thereby indicating stable integration and performance of the BjMYB28(RNAi) silencing cassette in the following generation. The homozygous T3 seeds showed even lower levels of total seed glucosinolate contents compared to the T1 seeds (Table 1). For example, the transgenic line, RNAi-9 with seed glucosinolate content of 22.90 ± 2.50 μmol/g DW in T1 seeds had average seed glucosinolate content of 13.78 ± 0.62 μmol/g DW in homozygous T3 seeds (Table 2). This in all possibility could be attributed to the dosage-mediated silencing in the homozygous state. The single-copy homozygous T3 lines showed total aliphatic glucosinolate content ranging from 9.64 to 18.19 μmol/g DW. The nonaliphatic glucosinolate profile in the low glucosinolate lines remained almost similar to that of wild-type and vector control lines.

Table 2. Average glucosinolate contents and profile (in μmol/g DW) in T3 homozygous seeds of BjMYB28(RNAi) transgenic lines. Only single-copy low glucosinolate lines (<30 μmol/g DW) were analyzed in advance generations. Value represent mean ± SD of three independent biological replicates
Transgenic eventSinigrin (2-propenyl)Gluconapin (3-butenyl)Glucobrassica-napin (4-pentenyl)Total seed aliphatic glucosinolatesNon aliphatic glucosinolatesTotal seed glucosinolates
  1. a

    Significantly different (P < 0.01) in comparison with wild-type control (Varuna) plants.

Control13.45 ± 1.6190.88 ± 6.112.39 ± 0.31106.69 ± 7.571.74 ± 0.45108.43 ± 8.00
RNAi-21.88 ± 0.14a13.09 ± 1.01a0.59 ± 0.13a15.57 ± 1.23a1.98 ± 0.1017.55 ± 1.22a
RNAi-52.18 ± 0.28a14.59 ± 2.04a0.68 ± 0.09a17.46 ± 2.26a1.81 ± 0.1119.27 ± 2.15a
RNAi-91.48 ± 0.09a10.19 ± 0.21a0.39 ± 0.05a12.06 ± 0.33a1.72 ± 0.3213.78 ± 0.62a
RNAi-151.27 ± 0.28a8.01 ± 1.57a0.36 ± 0.12a9.64 ± 1.84a1.62 ± 0.5511.26 ± 2.04a
RNAi-171.55 ± 0.06a11.98 ± 0.95a0.52 ± 0.04a14.06 ± 0.49a1.77 ± 0.6015.84 ± 0.78a
RNAi-192.81 ± 0.41a14.96 ± 1.63a0.41 ± 0.08a18.19 ± 0.99a1.78 ± 0.3019.97 ± 2.26a

Glucosinolates are known to be synthesized mainly in the vegetative organs such as leaves and silique walls, and then transported actively to seeds through phloem transporters (Nour-Eldin et al., 2012), although synthesis in immature seeds has also been proposed (Chen et al., 2001; Toroser et al., 1995). In order to investigate the effect of BjMYB28(RNAi) suppression in vegetative tissues, we estimated the glucosinolate content and profile in leaves of homozygous T2 transgenic lines having seed glucosinolate level <30 μmol/g DW. The six transgenic lines showed total leaf glucosinolates ranging from 5.92 to 11.43 μmol/g DW. A drastic reduction in total aliphatic glucosinolates to 80–90% in the young leaves of transgenic lines was observed (Figure 4a). The indole glucosinolates pool in leaves was found to be almost unaltered compared to the control lines (Table S4). Thus, use of native BjMYB28 promoter can ensure sufficient suppression of glucosinolate biosynthesis in leaves as well as other vegetative tissues, which in turn reduces the total seed glucosinolate accumulation in a significant way.

Figure 4.

Average leaf glucosinolate analysis and molecular characterization of low glucosinolate BjMYB28(RNAi) lines. Single-copy homozygous T2 lines with total seed glucosinolate <30 μmol/g DW were analyzed. (a) Average leaf aliphatic glucosinolate content in low glucosinolate transgenic lines. The mean ± SD of aliphatic glucosinolate content of three independent biological measurements were determined using HPLC. (b) BjMYB28-specific siRNA accumulation in low glucosinolate lines. Upper panel shows Northern blot analysis of small RNA using BjMYB28-3 target region. Equal loading of RNA samples is shown in the lower panel. (c) Reduction in steady-state levels of four BjMYB28 transcripts using qRT-PCR analysis. Wild-type (Varuna) and a vector line were used as control for qRT-PCR (set as 1) as well as siRNA accumulation experiments. Asterisks (*) indicate significant difference at P < 0.01 (Fishers LSD test determined by ANOVA).

Low glucosinolate transgenic lines showed increased siRNA accumulations and reduced levels of BjMYB28 transcripts

One of the major aims of our study was to develop the low glucosinolate lines in B. juncea which could meet international standard as canola quality oilseed (Potts et al., 1999). We therefore carried out detailed molecular characterization on six of the nine BjMYB28(RNAi) single-copy transgenic lines showing strong reduction in total seed glucosinolate content, that is, <30 μmol/g DW.

To check whether the reduced glucosinolate pool in low glucosinolate transgenic lines was due to BjMYB28-specific siRNA accumulation, small RNA Northern blot was performed in seedling stage of T2 homozygous plants. BjMYB28-specific siRNAs were detected in the low glucosinolate transgenic lines (Figure 4b). The decrease in glucosinolate accumulation was found to be associated with increased levels of siRNA accumulation in the homozygous transgenic lines. The siRNA accumulation was not observed either in the wild-type or in the vector control plants. Thus, our data clearly demonstrated RNAi-based suppression of BjMYB28 transcripts in low glucosinolate transgenic lines.

We further determined the target specificity of BjMYB28(RNAi) construct towards regulating the BjMYB28 homologs at post-transcriptional level. Steady-state transcript accumulation, using qRT-PCR analysis, was measured in seedlings, the stage at which all four BjMYB28 homologs showed optimal expression (Figure 1). There was a significant 4–10-fold reduction in BjMYB28-3 transcript, in the transgenic seedlings compared to the control plants (Figure 4c). Expression of other three BjMYB28 homologs was also affected in a limited manner across the low glucosinolate transgenic lines, with BjMYB28-1 being the second-most down-regulated transcript. The differential suppression of BjMYB28 homologs in transgenic lines could be possibly attributed to the presence of variable sequence identity among these homologs at the RNAi target region (Figure S6). A significant reduction in BjMYB28-3 transcript was also observed in the siliques of low glucosinolate transgenic lines (Figure S7). The steady-state level of BjMYB29 transcripts however remained almost unaltered. Thus, use of the native BjMYB28-3 promoter ensured controlled and specific knock-down of BjMYB28 homologs vis-à-vis reduced accumulation of aliphatic glucosinolates across B. juncea developmental stages.

Suppression of BjMYB28 homologs resulted in significant down-regulation of aliphatic glucosinolate biosynthetic genes in transgenic lines

If BjMYB28 is a major regulator of the aliphatic glucosinolate biosynthesis pathway, the expression of the genes involved in aliphatic glucosinolate biosynthesis pathway in low glucosinolate lines should be consequently down-regulated. To investigate this, steady-state mRNA levels of the downstream genes were assessed in the seedling stage of single-copy, homozygous T2 low glucosinolate lines. Our analysis revealed significant down-regulation of all the selected genes of aliphatic glucosinolate biosynthetic pathway in B. juncea, including those involved in side-chain elongation (GSL-ELONG), core structure biosynthesis (CYP79F1, CYP83A1) and secondary modifications of aliphatic glucosinolate (GSL-ALK, GSL-OH) (Figure 5a). When expression levels of indole glucosinolate biosynthetic pathway genes like CYP79B2, CYP83B1 and SOT16 were measured in the low glucosinolate transgenic lines, no significant change in the transcripts levels was observed for CYP83B1 and SOT16 (Figure 5b). The low glucosinolate lines, RNAi-2, RNAi-9 and RNAi-15 showed reduction in the transcript level of CYP79B2; however, the indole glucosinolate pools in these lines remained almost unaltered. This could be possibly because of the redundancy of CYP79B2 gene homologs in polyploid B. juncea. Thus, our results clearly indicated that BjMYB28 specifically regulate the aliphatic glucosinolate biosynthetic genes in B. juncea but not the indole glucosinolate biosynthetic genes, which was in complete accordance with the glucosinolate profiles observed in the transgenic lines.

Figure 5.

Gene expression analysis of glucosinolate candidate genes in the seedling stage of single-copy homozygous T2 low glucosinolate BjMYB28(RNAi) lines. Expression profiles of genes involved in (a) aliphatic glucosinolate biosynthesis pathway and (b) indole glucosinolate biosynthesis pathway. qRT-PCR experiments were conducted thrice with at least two technical replicates each. Wild-type (Varuna) and a vector line were used as control for qRT-PCR analysis (set as 1). Asterisks (*) indicate significant difference at P < 0.01 (Fishers LSD test determined by ANOVA).

Low glucosinolate transgenic lines had unaltered levels of seed quality and yield parameters

MYB28 belongs to a gigantic R2R3 transcription factor multigene family, the members of which are known to regulate a large array of biological processes and functions in plants (Dubos et al., 2010). For biotechnological applications, alteration of these transcription regulators needs to be performed in a very controlled and precise manner so as to have no or minimum detrimental effects to the plants. Recently, ectopic over-expression of aliphatic glucosinolate regulators in Arabidopsis showed retardation of plant growth, impaired gravitropic response and fertility in transgenic events (Gigolashvili et al., 2007, 2009). Thus, for an important oilseed crop like B. juncea, the effect of transgene-mediated RNAi suppression of BjMYB28 in low glucosinolate lines needs to be tested stringently for various growth and agronomical parameters. The low glucosinolate transgenic lines developed in this study showed proper seed germination, growth phenotype, male- and female-fertility and seed setting in subsequent generations, thereby reflecting no visible off-target effects on plant growth and development.

B. juncea is an important oilseed crop of the Indian subcontinent, yielding 37%–40% seed oil. In recent years, demand for Brassica oil for the production of biofuels has also increased to a greater extent (Hill et al., 2006). Hence, both seed quality and yield parameters of the low glucosinolate transgenic lines developed in this study also need to be considered. The six low glucosinolate transgenic lines homozygous for the T-DNA cassette were grown for two successive generations under the containment field conditions and the open-pollinated T2 and T3 seeds were analyzed for various oil-quality parameters using NIRS. The results showed no significant difference in the total oil content among the wild-type control plants and the six low glucosinolate transgenic lines developed (Figure 6a). After expelling oil, the oilcake is left out which is rich in proteins and therefore serves as an animal feed. The total protein content of the low glucosinolate transgenic lines and the control plants were also comparable, ranging from 25 to 27% (Figure 6b). Fatty acid compositions of the transgenic lines were also analyzed. The levels of three major fatty acids, namely oleic acid (C18:1), linolenic acid (C18:3) and erucic acid (C22:1) were found to be unchanged in the low glucosinolate transgenic lines compared to the control plants (Figure 6c). One of the major yield parameters namely 100 seed weight of the transgenic lines was also found to be unaltered compared to Varuna, the transformation host and the national check cultivar of B. juncea (Figure 6d). Detailed assessment of various seed quality and yield traits of the low glucosinolate B. juncea transgenic lines needs to be performed at multilocation open field trials in future.

Figure 6.

Characterization of low glucosinolate transgenic B. juncea lines for various seed quality parameters. Six single-copy, low glucosinolate BjMYB28(RNAi) lines were analyzed for (a) Total oil content; (b) Protein content; (c) Fatty acid profiles (oleic, linolenic and erucic acid); and (d) Hundred (100) seed weight. The NIRS data of T3 homozygous seeds obtained from at least 12 independent T2 progeny are represented along with SD. Wild-type (Varuna) and vector control lines were used as control. Asterisks (*) indicate significant difference, P < 0.05 (one-way ANOVA).

Future prospects of low glucosinolate transgenic B. juncea

One of the major concern of development of low glucosinolate lines in Brassica crops is its susceptibility to pest and diseases as glucosinolates forms integral part of plants immune system. There are various reports from Brassica species suggesting mixed response of pests and pathogens towards glucosinolate contents and profile (Bodnaryk, 1997; Giamoustaris and Mithen, 1997; Williams, 1989). The glucosinolate–herbivore interactions are known to be very complex, wherein specific glucosinolate fraction may act as attractant or repellent for specialist or generalist herbivores (Hopkins et al., 2009). In general, indole glucosinolates are known to be critical determinants for pest and disease resistance (Kim and Jander, 2007), whereas aliphatic glucosinolates play major role as deterrents for herbivore attack as well as for the overall fitness of the plant (Lankau and Kliebenstein, 2009).

In this study, we found that indole glucosinolate content of the low glucosinolate B. juncea transgenic lines remained almost unaltered compared to that of the wild-type control (Table S4). When these lines were grown under the containment field conditions for three successive growing seasons, we did not observe any altered susceptibility to pests and diseases compared to the B. juncea control (cv. Varuna). However, detailed assessment of the vulnerability of low glucosinolate B. juncea lines developed in the current study to various pests and pathogens will be undertaken in future.

Consumer acceptability is another major concern related to genetically modified crops. The hpRNAi cassette used to develop the low glucosinolate B. juncea lines in the current study contains only cis-DNA sequences (including the promoter, target exon sequence and the intronic spacer of the native BjMYB28-3 homolog) derived from the host B. juncea. Besides, the bar gene cassette (conferring resistance to herbicide glufosinate) was cloned within the loxP sites, which in turn will facilitate excision of the selectable marker from the homozygous low glucosinolate stock in the subsequent generations by using the bacterial Cre-lox system of recombination (Arumugam et al., 2007). Thus taking into account all these facts, we presume that the low glucosinolate B. juncea transgenic lines will broaden the scope and acceptability of this oilseed crop in international market.

Conclusion

Quantitative traits like glucosinolates are known to be controlled by highly complex gene expression networks comprising multiple biosynthetic pathway genes and their transcriptional regulators (Hirai et al., 2007). Further, in polyploid crops like B. juncea and B. napus, because of the existence of structural and expression divergence among multiple homologs, the manipulation of such quantitative trait is quite challenging either through conventional breeding or transgenic approaches.

In this study, we identified BjMYB28 as major transcription factor gene controlling the aliphatic glucosinolates accumulation in B. juncea. We also showed that using transgene-based RNAi-suppression strategy against BjMYB28, a significant suppression of aliphatic glucosinolate contents could be achieved, without compromising the nonaliphatic glucosinolates pool. We successfully developed stable low glucosinolate transgenic lines in B. juncea having total seed glucosinolate levels as low as 11.26 μmol/g DW (ca. 89% reduction compared to the wild-type level). The low glucosinolate transgenic lines performed well for growth and various oil-quality parameters under the containment field conditions. The work provides a significant advancement over the previously adopted breeding approaches which are encumbered by linkage-drags, and necessitate introgression of multiple loci to achieve the low glucosinolate trait. The study will contribute immensely towards developing low aliphatic glucosinolate lines across related Brassica species thus raising their oil- and seed-meal quality in the global market.

Experimental procedures

Plant materials and growth condition

Brassica juncea, high glucosinolate Indian line (Varuna) and canola quality east-European line (Heera) were grown in a growth chamber (Conviron) at 10 h light/14 h dark cycle, with temperature of 22 °C/15 °C and 70% relative humidity, respectively. Different developmental stages were collected, immediately frozen in liquid nitrogen and stored at −80 °C.

Generation of BjMYB28 hairpin RNAi transformation construct

A modified binary vector (pPZP200lox) containing the ‘lox’ tandem repeats was used (Arumugam et al., 2007). A PCR amplified 35Sde-bar-ocspA fragment conferring resistance to the herbicide phosphinothricin was cloned between the lox repeats at EcoRV site, and used as vector control construct. A 1054 bp fragment of BjMYB28-3 promoter was cloned directionally within the SmaI/SpeI sites of the above mentioned vector to create the vector pPZP200lox:35Sde-bar-ocspA::ptBjMYB28. A 409 bp fragment of the second intron of BjMYB28-3 gene was amplified using specific primers and cloned into the SmaI and SacI site of vector pRT100 (Topfer et al., 1987) to create the vector pRT100:int2. To this vector, a 342 bp fragment from third exon (encompassing 721–1062 bp sequence from ATG start codon) of BjMYB28-3 was amplified and cloned in both sense and antisense orientations. The Exon(s)-int2-Exon(as)-35SpA cassette thus created was excised and finally cloned directionally at XhoI/PstI sites of binary vector pPZP200lox:35Sde-bar-ocspA::ptBjMYB28 to construct the final binary vector BjMYB28(RNAi). The transformation vectors were mobilized into A. tumefaciens strain GV3101 using freeze thaw method (Nishiguchi et al., 1981). All DNA manipulations were performed using standard protocols. Primers used in the current study are provided in Table S5.

B. juncea genetic transformation

Genetic transformation of B. juncea was performed using the protocol described earlier (Jagannath et al., 2003) with minor modifications. Briefly, seeds were cleaned with detergent, followed by treating with 70% ethanol. Surface sterilization was carried out with 0.05% HgCl2 for 10 min. Seeds were then inoculated on to basal MS media and grown in culture room set at 10 h light (day)/14 h dark (night) cycle at a constant temperature of 23 ± 1 °C. Hypocotyls of 5 day old seedlings were cut into 0.5–1.0 cm pieces, and used as explants. Explants were cultured in MS liquid media supplemented with 1 mg/ml each of NAA and BAP (N1B1) for 24 h. Agrobacterium strain harbouring desired construct was grown in YEB medium supplemented with proper antibiotics (Rif10Gent20Spec50). Bacterial cells at OD600 of 0.5–0.6 were harvested and re-suspended in N1B1 medium to a final OD600 of 0.3. Explants were incubated with the bacterial suspension for 30 min and cocultivated at 23 °C, 110 r.p.m. for 16–24 h. After cocultivation, the explants were washed with liquid N1B1 supplemented with Augmentin (200 mg/L) to remove excess Agrobacterium cells. The explants were plated onto shoot induction media containing N1B1, Augmentin (200 mg/L), AgNO3 (20 μm) and 10 mg/L of Basta (Agrevo, active ingredient phosphinothricin). Regenerated shoots obtained within 25–35 days were transferred onto rooting media supplemented with 2 mg/L of IBA along with Augmentin (200 mg/L) and 10 mg/L Basta. The well-rooted transformants were transferred directly on to soil during the growing season. Transgenic plants were grown in a containment net-house in the field according to the guidelines of Department of Biotechnology, Government of India.

Southern blotting of transgenic plants

Genomic DNA from (approximately 1 month old) B. juncea transgenic lines as well as the wild-type Varuna (−ve control) was isolated using CTAB method. About 10 μg of genomic DNA was digested with EcoRV. After proper treatments (depurination, denaturation and neutralization), the DNA fragments were transferred onto the nylon membrane (Amersham Hybond XL, GE Healthcare, Buckinghamshire, UK) and UV cross-linked. Prehybridization and hybridization steps were carried out at 42 °C in Denhardts buffer for 16–18 h. Probe (ca. 390 bp fragment of bar gene) was labelled with dCTP [32Pα] by incubating at 37 °C for 1 h using Amersham Megaprime DNA Labelling Systems (GE Healthcare). After hybridization and washing, the blot was kept in a cassette, exposed to X-ray film in dark for 48–72 h and developed in an automatic film processor (Hyperprocessor, GE Healthcare).

Gene expression analysis using qRT-PCR

Total RNA from plant samples was isolated using Spectrum Total RNA Isolation Kit (Sigma-Aldrich, St. Louis, MO). Approximately, two microgram of total RNA was reverse transcribed using high capacity first strand cDNA synthesis kit (Applied Biosystems, Foster City, CA) according to manufacturer's instructions. The relative expression of glucosinolate candidate genes was analyzed by real-time qRT-PCR in ABI 7900HT Fast Real-time PCR machine (ABI) using SYBR green protocol. BjACTIN2 gene was used as endogenous control (Chandna et al., 2012). Data were analyzed in three independent sets of plants with three technical replicates each. Statistical analyses were conducted using one-way ANOVA following Fishers LSD test of significance.

Specific primers for BjMYB28 homologs were designed based on the nucleotide sequence alignment; preferably from the 3′ region of the gene. Nucleotide sequence of genes involved in both aliphatic and indole glucosinolate biosynthetic pathways was adapted from the recently published A-genome sequences of B. rapa (Wang et al., 2011; Zang et al., 2009). Various primers used for qRT-PCR analysis are tabulated in Table S5.

Detection of small RNA by Northern blotting

Total RNA was extracted with TRIZol reagent (Invitrogen, Life Technologies Corp., Carlsbad, CA) and RNA integrity was checked in Agilent-2100 Bioanalyzer and approximately, 20 μg of total RNA was used for blotting. RNA molecules were fractionated on 17% polyacrylamide gel containing 7 m urea. Miniprotean cell unit (Bio-Rad, Hercules, CA) was used to resolve the fragments in the gel. The transfer of RNA to nylon membrane was set up overnight at cold room with 0.5X TBE. After transfer, the membrane was soaked in 2X SSC and UV cross-linked and proceeded for hybridization. Prehybridization was carried out for 6 h followed by hybridization for 12–16 h at 50 °C. Probes specific to the BjMYB28-3 region selected for RNAi design was radiolabelled with dCTP [32Pα] using Amersham Megaprime DNA Labelling Systems (GE Healthcare) according to the manufacturer's instructions. Probes to marker (New England Biolabs, Ipswich, MA) were labelled with dATP [32Pγ] using polynucleotide kinase. The probes were added to the blot simultaneously. Blots were exposed to phosphor screen (GE Healthcare) for 12 h and developed in multi-image scanner (Typhoon 9210, Amersham Biosciences, Buckinghamshire, UK).

Estimation of total glucosinolates, protein, oil content and fatty acid composition

Total glucosinolates, oil, protein contents and fatty acid composition of field grown seeds (T2 and T3 homozygous seeds) were determined using near infrared reflectance spectroscopy (NIRS-5000, FOSS, Denmark) in duplicates as per manufacturer's instructions. Seeds of wild-type Varuna and the vector control plants were used as control. The 100 seed weight of open-pollinated T3 seeds of at least 10 independent T2 progeny was calculated, in duplicates. Statistical analyses were conducted using one-way ANOVA following Fishers LSD test of significance.

Glucosinolate analysis using HPLC

The transgenic events were analyzed for component seed/leaf glucosinolate profiles using HPLC as per the protocols described earlier (Kraling et al., 1990) with minor modifications. Samples were analyzed in Prominence UFLC 20A (Shimadzu, Japan) machine with reverse phase C18 column (Shimadzu LC column-XR-ODS; 100 mm × 3.0 mm with 5 μm internal diameter). Benzyl glucosinolate, glucotropaeolin (Applichem, Darmstadt, Germany) was used as the internal standard. A gradient of water (solvent A) and acetonitrile (solvent B) was used with a flow rate of 1 ml/min at an oven temperature of 35 °C. Elution was achieved with a gradient of 1–19% of solvent B over a period of 10 min. Glucosinolate components were detected at 229 nm and peaks were identified with reference to the retention time of already published chromatograms (Brown et al., 2003). Concentrations of individual glucosinolates were calculated in micromoles per gram dry weight (μmol/g DW) relative to the area of the internal standard peak applying their relative response factors. At least three independent biological replicates were analyzed and statistical test was conducted using one-way ANOVA following Fishers LSD test of significance.

Data deposition

The sequences reported in this paper are deposited in the National Center for Biotechnology Information GeneBank database with accession nos. JQ666166 (BjMYB28-1), JQ666167 (BjMYB28-2), JQ666168 (BjMYB28-3), JQ666169 (BjMYB28-4), JX316031 (BjMYB29-1) and JX316032 (BjMYB29-2).

Acknowledgements

The work was supported by Department of Biotechnology, India (project schemes: BT/PR271/AGR/36/687/2011 and Rapid Grant for Young Investigators) and the core-grant provided by NIPGR, India to NCB. RA was funded with Junior Research Fellowship from Council of Scientific and Industrial Research, India. We are grateful to Central Instrumentation Facility at NIPGR. Critical suggestions from Prof. Deepak Pental, Prof. Akshay K. Pradhan, Prof. Roger Beachy and Dr. Swarup K. Parida are highly acknowledged. Two anonymous reviewers are also acknowledged.

Ancillary