- Top of page
- Materials and methods
- Supporting Information
The Brassicaceae form a large plant family with over 3000 species. Six Brassica species are cultivated worldwide: three diploids, Brassica rapa (AA, 2n = 20), B. nigra (BB, 2n = 16) and B. oleracea (CC, 2n = 18); and three amphidiploids, B. juncea (AABB, 2n = 36), B. napus (AACC, 2n = 38) and B. carinata (BBCC, 2n = 34). B. oleracea and B. rapa are among the most important vegetable crops worldwide.
The best characterized phytonutrient compounds in the Brassicaceae are the glucosinolates (Fahey et al., 2001; Wittstock & Halkier, 2002). These secondary metabolites are predominantly found in plants of the order Brassicales/Capparales and are normally produced as part of the natural defense against generalist herbivores (Blau et al., 1978; Kliebenstein et al., 2002). As such they generally also have an anti-nutritional value for animals (Mithen, 2001) and great efforts have been made to breed for ‘double zero’ rapeseed, which is low in erucic acid and the anti-nutritional 3-hydroxy-3-butenyl glucosinolate (Rosa et al., 1997). Remarkably, though, several glucosinolate hydrolysis products, such as isothiocyanates, which are formed upon tissue damage, have an anti-cancer effect when provided in the human diet (Zhang et al., 1994; Fahey et al., 1998; Mithen et al., 2000; Talalay & Fahey, 2001).
The isothiocyanate sulphoraphane, the myrosinase converted form of 4-(methylsulfinyl) butyl glucosinolate (glucoraphanin) present in broccoli, was reported to induce detoxifying enzymes in mammalian cells and acts as a long-lasting antioxidant that detoxifies carcinogens in the body (Zhang et al., 1994; Talalay & Fahey, 2001). Likewise, phenylethyl isothiocyanate, the myrosinase converted form of the main glucosinolate in watercress, inhibits the action of lung carcinogens (Neo et al., 2005).
The Brassica species accumulate aliphatic glucosinolates derived from the amino acid methionine, indole glucosinolates derived from tryptophane and aromatic glucosinolates derived from phenylalanine. In B. napus, approx. 30 glucosinolates, and in B. oleracea 12 different glucosinolates, have been identified (Mithen et al., 2000; Fahey et al., 2001; Branca et al., 2002). In 58 accessions of B. rapa representing all different morphotypes (Chinese cabbage, pak choi, turnips, oil-types and Japanese leafy types like mizuna and komatsuna), eight different glucosinolates were detected (He et al., 2000, unpublished), while in 113 different accessions of turnip greens (Brassica rapa subsp. rapa L.), 16 different glucosinolates were detected (Padilla et al., 2007). Despite the wide variation in glucosinolate profiles among genotypes, individual plants mostly contain a limited number of major glucosinolate forms. In all B. rapa genotypes studied, the aliphatic (4-C) 3-butenyl and the aliphatic (5-C) 4-pentenyl glucosinolates were the predominant glucosinolates.
In the species A. thaliana, more than 34 different glucosinolates have been identified (Kliebenstein et al., 2001a; Reichelt et al., 2002). The glucosinolate biosynthetic pathway consists of three stages in Arabidopsis and Brassica ssp., namely an amino-acid chain elongation stage for methionine- and phenylalanine-derived glucosinolates, a glucosinolate skeleton formation and a side-chain modification stage (Halkier & Du, 1997; Mithen, 2001; Halkier & Gershenzon, 2006). Upon tissue disruption, glucosinolates come into contact with the plant enzyme myrosinase, resulting in the formation of many, often volatile, glucosinolate degradation products with diverse biological activities.
To date, the structural genes responsible for most biosynthetic steps have been identified and the core pathway of glucosinolate biosynthesis has been clarified in Arabidopsis (Wittstock & Halkier, 2000; Bak & Feyereisen, 2001; Grubb et al., 2004; Mikkelsen et al., 2004; Piotrowski et al., 2004; Halkier & Gershenzon, 2006). Genetic studies in Arabidopsis and Brassica spp. proved an important role for AOP (2-oxoglutarate-dependent dioxygenase) and MAM (methyl-thioalkylmalate synthase) in the regulation of aliphatic glucosinolate accumulation. The MAM gene family is involved in glucosinolate side-chain elongation, and the AOP family is involved in side-chain modification (Kliebenstein et al., 2001a,b; Kroymann et al., 2001, 2003; Field et al., 2004, 2006; Heidel et al., 2006).
In B. oleracea, glucosinolate-Elong (MAM) and glucosinolate-Alk (AOP), responsible for glucosinolate side-chain elongation and modification in Arabidopsis, were cloned based on Arabidopsis homology (Li & Quiros, 2002; 2003) and the organization of AOP genes was further compared with Arabidopsis sequences (Gao et al., 2004). Additionally, BoGSL-Alk expressed in Arabidopsis resulted in changed aliphatic glucosinolate profiles in leaves and seeds (Li & Quiros, 2003).
Besides these and other structural genes, several transcription factors were identified as regulators of indole glucosinolate or aliphatic glucosinolate accumulation, such as IQD1, ART1/Myb34, AtDof1.1(OBP2), HAG1/Myb28, HAG3/Myb29, HAG2/Myb76 and HIG1/Myb51 (Celenza et al., 2005; Hirai et al., 2005, 2007; Levy et al., 2005; Skirycz et al., 2006; Gigolashvili et al., 2007a,b,c; Sonderby et al., 2007; Wentzell et al., 2007).
Genetic factors strongly influence glucosinolate accumulation, although environmental and developmental factors may modify the expression of genes. Quantitative trait locus (QTL) analysis is a powerful method to study the genetics underpinning quantitative variation in glucosinolate profiles, and to estimate the number of variable loci affecting a trait (Koornneef et al., 2004). In Arabidopsis, one QTL analysis was performed for glucosinolate accumulation in both seeds and leaves using Ler × Cvi recombinant inbred lines. Total aliphatic glucosinolate accumulation was controlled by six QTL, total indolic glucosinolate accumulation was controlled by another six QTL and benzylic glucosinolate accumulation was controlled by three QTL (Kliebenstein et al., 2001c). Kliebenstein et al. (2001c) showed that two QTL controlling total aliphatic glucosinolates that map to AOP and GS-elong (MAM) loci interact epistatically, which was also observed in expression QTL analysis by Keurentjes et al. (2006) and Wentzell et al. (2007).
In the Ler × Col RI population there was much less variation for total aliphatic glucosinolate accumulation and only two chromosome 5 QTL were identified, with the major QTL explaining 60% of the variation and colocalizing with MAM (Kliebenstein et al., 2002). QTL for glucosinolate accumulation were also identified in a new Arabidopsis inbred line population (Da(1)-12 × Ei-2 (Pfalz et al., 2007). Total aliphatic glucosinolates in this population were controlled by three QTL, one co-localizing with AOP, one near the top of chromosome 5, which was also identified previously, and a new QTL on chromosome 1. Genetic analysis of a series of Arabidopsis ecotypes revealed the MAM locus at the top of chromosome 5, explaining a large proportion of the observed variation in aliphatic glucosinolate composition (Kroymann et al., 2003).
In Arabidopsis lyrata interpopulation crosses, one QTL was identified at the MAM locus for aliphatic glucosinolate ratios plus an additional QTL for total indole glucosinolates and the ratio of aliphatic to indole glucosinolates (Heidel et al., 2006). Multiple QTL have been described both for seed and leaf glucosinolates in both B. napus (Howell et al., 2003) and B. juncea (Mahmood et al., 2003; Ramchiary et al., 2007).
In B. rapa, there has been no report on QTL controlling glucosinolate accumulation until now. The objectives of this study are to map QTL for glucosinolate composition and accumulation in B. rapa leaves in two new double haploid (DH) populations derived from wide crosses.
- Top of page
- Materials and methods
- Supporting Information
Populations of DH lines have been used extensively for QTL mapping because they are immortal and completely homozygous. They have obvious advantages above F2 populations, which are represented by single, often heterozygous, plants. However, recombination events between parental genomes are lower in DH lines compared with recombinant inbred lines (RILs). The latter are, however, hard to create and to maintain because of the self-incompatibility of most B. rapa accessions. We analyzed two DH populations which both shared yellow sarson R500 (YS-143) as male parent and had as female parents very different east Asian B. rapa types, providing ample variation for molecular and phenotypic traits, in contrast to many earlier DH populations that were made between either Chinese cabbage types or between oil types (summer oil , yellow sarson and winter oil types) (Wang et al., 2004; Zhang et al., 2005; Kim et al., 2006;Suwabe et al., 2006).
In order to compare our linkage map with public B. rapa maps, SSR primers were used to assign linkage groups to international standard A groups (http://www.brassica.info). One important criterion used during the selection of SSR loci was the requirement that each primer only has one map position in reference linkage maps. This avoids the ambiguity that specific SSR markers map at more, or different, positions in different B. rapa maps. The positions of SSR loci in the DH maps presented here were consistent with those in other published Brassica maps except for Na12A01, which mapped to A1 in DH38 and not to A4 (N4) as listed in the MBGP website (http://www.brassica.info). However Piquemal et al. (2005) showed that Na12A01 maps to at least three loci in the B. napus linkage map, one of them on N01, which corresponds to A1(Piquemal et al., 2005). Marker coverage in our DH maps was good, with some clustering of markers on several linkage groups. The orientation of linkage groups A3 and A10 is different in published B. rapa genetic maps; one orientation in the F2 linkage map of Kim et al. (2006) and the opposite orientation in the DH linkage map published by Choi et al. (2007). The orientation as published by Choi et al. (2007) is now the international standard, and is used in this paper. This orientation is similar to that of the B. napus maps used for comparative mapping published by Parkin et al. (2005). We need to state here that in our earlier publications, we also used DH-30 and DH-38 to map QTL for phytate, phosphate and developmental traits, and used the orientation as published by Kim et al. (2006), and thus A3 and A10 have different orientations compared with this paper (Lou et al., 2007; Zhao et al., 2008). Sixteen genetic loci controlled individual or total aliphatic glucosinolate accumulation and their ratios; four QTL are detected in both populations and four QTL are identified in both seasons, while others are population/season-specific. Not all QTL were identified in both experiments, which partly illustrates environmental effects on glucosinolate composition. Additional causes may be different developmental stages as a result of different growth rates in autumn and spring, and the fact that population size was relatively small with no replications within the experiments. Ali-QTL3.2 at the bottom of A3 is a major QTL detected in both populations and both seasons for the individual aliphatic glucosinolates and their ratios, which co-localizes with Ind-QTL3.2 for total indole glucosinolates detected in DH-30 spring 2005. This is the only clear co-localization for different classes of glucosinolates. Ali-QTL3.2 also affects the ratio between 4-C and 5-C aliphatic glucosinolates (ratio 1: (NAP + PRO)/TAli), explaining 28–50% of the variation in both populations and seasons, while AliQTL3.1 explains between 6 and 12% for ratio 1, also in both seasons and populations. AliQTL3.2 explained between 12 and 25% of the variation for ratio 2 between PRO and PRO + NAP in both populations and seasons, while AliQTL3.1 affects these ratios only in DH38, explaining 28% in 2005. These data suggest a role for enzymes like 2-oxoglutarate-dependent dioxygenases (AOP) (Kliebenstein et al., 2001b), involved in side-chain modification.
The genome of B. rapa has triplicate counterparts to corresponding homeologous segments of Arabidopsis; as a result, many genes are present in duplicate or triplicate in the B. rapa genome, and multiple copies of genes may contribute to phenotypic diversity (Parkin et al., 2005; Schranz et al., 2006). Based on these data, we can predict the putative location of homologues of Arabidopsis genes involved in glucosinolate biosynthesis, such as MAM and AOP, but also transcription factors like Myb28 and Myb29 which regulate aliphatic glucosinolate biosynthesis (Hirai et al., 2007) (Fig. 4). AOP is predicted to have two homeologous loci, one on A3 and another on A9 in the B. rapa genome, which was confirmed by the mapping of SSR-AOPa (an SSR physically linked to an AOP locus on BAC KBrB002P01) and SCAR-AOPc in the DH populations on A3 and A9, respectively (Figs 1, 4). For MAM (segment Q) three paralogues are predicted on A2, A6 and A9, and for Myb29 (segment R) three paralogues on A2, A3 and A10, while Myb28 (segment X) paralogues are predicted on chromosomes 2 and 6. The QTL analysis reported in this paper is too imprecise to associate specific QTL with candidate gene loci. However, we can use the comparisons between our QTL and the map positions of predicted glucosinolate biosynthesis genes in reference maps based on SSR markers and map distance to suggest candidate genes. To obtain proof of the hypothesis, the candidate genes need to be mapped on our DH maps and allelic variation needs to be investigated.
Figure 4. Graphical presentation of aliphatic glucosinolate quantitative trait loci (QTL) positions in double haploid (DH) populations.
Download figure to PowerPoint
In a recent study, Myb28, Myb76 and Myb29 were identified as positive regulators of aliphatic glucosinolate accumulation, and it was suggested that Myb29 had a preference for short-chain aliphatic glucosinolates as identified in this study (Hirai et al., 2007; Gigolashvili et al., 2007c). Two other studies in a population of 403 Arabisopsis Bay-0 × Sha RILs also investigated the roles of the R2R3 Myb transcription factors MYB28, MYB29 and MYB76, and also concluded that these genes evolved both overlapping and specific regulatory capacities. By genetic co-localization, MYB28 was identified as candidate gene for both aliphatic glucosinolate QTL and expression QTL of genes involved in aliphatic glucosinolate biosynthesis. This suggests that these QTL may be explained by variation in the expression of this Myb gene (Sonderby et al., 2007; Wentzell et al., 2007). In B. rapa we do not see evidence for co-localization of MYB28 or MYB29 with the major aliphatic QTL AliQTL3.2. However, synthetic relations may suggest the orthologous gene of MYB28 as a putative candidate gene for AliQTL6.1, explaining around 15% of the variation in NAP and Tali in DH 38 (both seasons), and an orthologue of MYB29 as a putative candidate gene for AliQTL10.2, explaining 6% for ratio 1 (elongation) in DH38 (autumn 2004). The mapped AOP locus on A3 maps in the vicinity of Ali-QTL3.1 (ratio 1; elongation, ratio 2; modification, NAP, PRO) may thus account for these QTL. Another candidate for both AliQTL9.1 and Ali-QTL6.1 that needs to be tested by genetic mapping is MAM, since the Arabidopsis orthologue maps at a synthetic position. Genetic mapping of the Brassica paralogues is under way to confirm co-localization of candidate genes with QTL intervals.
In conclusion, many QTL for aliphatic glucosinolates are identified in two B. rapa DH populations, with the major Ali-QTL3.2 on A3 co-localizing with a QTL for indole glucosinolates. From 16 genetic loci controlling individual or total aliphatic glucosinolate accumulation and their ratios, four QTL are detected in both populations. Another four QTL are identified in both seasons while others are population/season-specific. Studies that combine gene expression profiles with metabolite profiles are presently being conducted in DH populations and these studies, together with the mapping of candidate genes, may provide more information on genes underlying the QTL. As the genetic basis of glucosinolate biosynthesis becomes clear, it will be feasible to breed for Brassica vegetables with different types and concentrations of glucosinolates by marker-assisted selection or by transgenic approaches.