A 2-Oxoglutarate-Dependent Dioxygenase Mediates the...A 2-Oxoglutarate-Dependent Dioxygenase Mediates the Biosynthesis of Glucoraphasatin in Radish1[OPEN] Tomohiro Kakizaki*, Hiroyasu

A 2-Oxoglutarate-Dependent Dioxygenase Mediates theBiosynthesis of Glucoraphasatin in Radish1[OPEN]

Tomohiro Kakizaki*, Hiroyasu Kitashiba, Zhongwei Zou2, Feng Li3, Nobuko Fukino, Takayoshi Ohara,Takeshi Nishio, and Masahiko Ishida4

Division of Vegetable Breeding, Institute of Vegetable and Floriculture Science, NARO, Ano, Tsu, Mie514-2392, Japan (T.K., N.F., T.O., M.I.); and Graduate School of Agricultural Science, Tohoku University,Aoba-ku, Sendai, Miyagi 980-0845, Japan (H.K., Z.Z., F.L., T.N.)

ORCID IDs: 0000-0002-3636-6616 (T.K.); 0000-0002-8955-1153 (Z.Z.).

Glucosinolates (GSLs) are secondary metabolites whose degradation products confer intrinsic flavors and aromas toBrassicaceae vegetables. Several structures of GSLs are known in the Brassicaceae, and the biosynthetic pathway andregulatory networks have been elucidated in Arabidopsis (Arabidopsis thaliana). GSLs are precursors of chemical defensesubstances against herbivorous pests. Specific GSLs can act as feeding blockers or stimulants, depending on the pestspecies. Natural selection has led to diversity in the GSL composition even within individual species. However, in radish(Raphanus sativus), glucoraphasatin (4-methylthio-3-butenyl glucosinolate) accounts for more than 90% of the total GSLs, andlittle compositional variation is observed. Because glucoraphasatin is not contained in other members of the Brassicaceae, likeArabidopsis and cabbage (Brassica oleracea), the biosynthetic pathways for glucoraphasatin remain unclear. In this report, weidentified and characterized a gene encoding GLUCORAPHASATIN SYNTHASE 1 (GRS1) by genetic mapping using amutant that genetically lacks glucoraphasatin. Transgenic Arabidopsis, which overexpressed GRS1 cDNA, accumulatedglucoraphasatin in the leaves. GRS1 encodes a 2-oxoglutarate-dependent dioxygenase, and it is abundantly expressed inthe leaf. To further investigate the biosynthesis and transportation of GSLs in radish, we grafted a grs1 plant onto a wild-typeplant. The grafting experiment revealed a leaf-to-root long-distance glucoraphasatin transport system in radish and showedthat the composition of GSLs differed among the organs. Based on these observations, we propose a characteristicbiosynthesis pathway for glucoraphasatin in radish. Our results should be useful in metabolite engineering for breeding ofhigh-value vegetables.

Glucosinolates (GSLs), the sulfur-containing sec-ondarymetabolites, are precursors of chemical defensecompounds of the Brassicaceae. When plants areattacked by herbivores and/or pathogens, GSLs arerapidly hydrolyzed by endogenous thioglucosidases(also known as myrosinases) to yield isothiocyanates

(Rask et al., 2000; Wittstock and Halkier, 2002; Halkierand Gershenzon, 2006). Isothiocyanates, such as sul-foraphane (4-methylsulfinylbutane isothiocyanate),are known as anticarcinogenic compounds that inducephase 2 detoxification enzymes (Zhang et al., 1994).Specific isothiocyanates can act as repellents depend-ing on the feeding species. In response to natural se-lection by pests, qualitative variation of GSLs mayhave arisen among Arabidopsis (Arabidopsis thaliana)accessions (Giamoustaris and Mithen, 1995).

Plants contain more than 200 types of GSLs, which,on the basis of their precursors, are classified into thefollowing three groups: aliphatic, aromatic, and indolicGSLs. The main precursors of aliphatic, aromatic, andindolic GSLs are Met, Phe, and Trp, respectively (Faheyet al., 2001; Halkier and Gershenzon, 2006). The struc-ture and composition of GSLs depend on plant species,variety, developmental stage, and tissue (Mithen et al.,2000; Fahey et al., 2001). In vegetables of the Brassica-ceae, GSLs not only act as antipest compounds but alsoconfer specific flavors and tastes.

Radish (Raphanus sativus; 2n = 2x = 18) is an impor-tant root vegetable of the Brassicaceae family thatgenerally contains glucoraphasatin (4-methylthio-3-butenyl glucosinolate), one of the aliphatic GSLsderived from Met, in the root. Glucoraphasatin isthe predominant GSL in the root and accounts for

1 This work was supported by the Ministry of Agriculture, For-estry, and Fisheries of Japan (Genomics-Based Technology for Agri-cultural Improvement, grant no. HOR-1006).

2 Present address: Department of Plant Science, University of Man-itoba, Winnipeg, R3T 2N2, Canada.

3 Present address: Division of Radiation Breeding, Institute ofCrop Sciences, NARO, Hitachiomiya, Ibaraki 319-2293, Japan.

4 Present address: Institute of Vegetable and Floriculture Science,NARO, Tsukuba, 305-8666, Japan.

* Address correspondence to [email protected] author responsible for distribution of materials integral to the

findings presented in this article in accordance with the policy de-scribed in the Instructions for Authors (www.plantphysiol.org) is:Tomohiro Kakizaki ([email protected]).

T.K. and M.I. designed the study and performed most of the ex-periments; H.K., Z.Z., and F.L. constructed the BAC library and per-formed the screening; N.F. and T.O. performed the HPLC; T.K. andT.N. wrote the article.

[OPEN] Articles can be viewed without a subscription.www.plantphysiol.org/cgi/doi/10.1104/pp.16.01814

Plant Physiology�, March 2017, Vol. 173, pp. 1583–1593, www.plantphysiol.org � 2017 American Society of Plant Biologists. All Rights Reserved. 1583

Dow

nloaded from https://academ

ic.oup.com/plphys/article/173/3/1583/6115984 by guest on 07 June 2021

http://orcid.org/0000-0002-3636-6616http://orcid.org/0000-0002-8955-1153http://crossmark.crossref.org/dialog/?doi=10.1104/pp.16.01814&domain=pdf&date_stamp=2017-02-28mailto:[email protected]://www.plantphysiol.orgmailto:[email protected]://www.plantphysiol.org/cgi/doi/10.1104/pp.16.01814

more than 90% of the total GSLs present in Japanesecultivars (Ishida et al., 2012). Recently, the genomesequence and transcriptome profiles of radish havefacilitated studies on the GSL biosynthetic pathway byprofiling of GSL synthesis-associated genes of Ara-bidopsis (Wang et al., 2013; Kitashiba et al., 2014;Mitsui et al., 2015; Wu et al., 2015). However, becauseglucoraphasatin is characteristic of radish and notfound in other Brassica spp., the pathway for its bio-synthesis remains unclear.

In a comprehensive screening of radish landracesfrom Japan using their GSL composition, a mutantlacking glucoraphasatin was identified in a previousstudy (Ishida et al., 2015). The mutant contains glu-coerucin (4-methylthiobutyl glucosinolate) in roots andshoots instead of glucoraphasatin. Glucoerucin is the

major GSL in cabbage (Brassica oleracea) and some ac-cessions of Arabidopsis but is almost absent in wild-type radish. The structural difference between glucor-aphasatin and glucoerucin is the presence of a doublebond between third and fourth carbon chain (Visentinet al., 1992; Barillari et al., 2005; Montaut et al., 2010).Genetic analysis revealed that the absence of glucor-aphasatin is controlled by a single recessive gene lo-cated at the end of the Raphanus linkage group R1 andsuggested that the mutant has a lesion in a gene en-coding an enzyme catalyzing the synthesis of glucor-aphasatin from glucoerucin (Ishida et al., 2015). Wenamed the mutant glucoraphasatin synthase1 (grs1) andattempted to elucidate the function of GRS1. Here, wereport the map-based cloning of GRS1 and its functionin the GSL biosynthetic pathway in radish.

Table I. GSL content in radish plant

Aliphatic and indolic GSLs were measured in leaves of 3-week-old soil-grown wild-type (cv. Taibyosoubutori), grs1-1, grs1-2, and F1 plants.Numbers are averages 6 SD (n = 10). Values given are mmol g21 dry weight. Data within a column followed by the same letter are not significantlydifferent (P , 0.05). n.d., Not detected.

Plant Glucoraphanin Glucoraphenin Glucoerucin Glucoraphasatin Glucobrassicin4OH-

Glucobrassicin

4OMe-

GlucobrassicinTotal

Wild type 1.4 6 0.1 a 2.9 6 0.5 0.3 6 0.3 a 42.7 6 10.3 0.9 6 0.4 a n.d. n.d. 48.1 6 11.2 a

grs1-1 3.9 6 0.9 b n.d. 6.6 6 1.3 b n.d. 1.3 6 0.6 a,b n.d. 0.1 6 0.1 11.9 6 2.0 b

grs1-2 9.7 6 1.9 c n.d. 21.7 6 3.2 c n.d. 1.8 6 0.9 b 0.1 6 0.1 0.1 6 0.1 33.4 6 2.7 c

grs1-2 3 grs1-1 4.4 6 1.9 b n.d. 9.1 6 2.7 b n.d. 1.0 6 0.5 a n.d. n.d. 14.6 6 3.1 b

Figure 1. Map-based cloning of GRS1. Themutation in GRS1 was mapped between in-sertion/deletion markers s8D07 and ssG02 onRaphanus linkage group R1. Black bars, Ho-mozygous region for the grs1-1 allele; whitebars, heterozygous region. The GSL content inthe recombinants was measured by HPLC, andthe plants were classified into glucoerucin-rich(E) and glucoraphasatin-rich (R) types. Thenumber of recombinants with the same geno-type is indicated in parentheses. Gray bars in-dicate bacterial artificial chromosome clonesisolated using the linked markers. The pre-dicted ORFs in this region are shown by blackarrows. cM, Centimorgan.

1584 Plant Physiol. Vol. 173, 2017

Kakizaki et al.

Dow



RESULTS

Map-Based Cloning of GRS1

To identify GRS1, we employed two mutant plantsdeficient in glucoraphasatin synthesis, grs1-1 and grs1-2.Both the mutants are completely lacking in the accu-mulation of glucoraphasatin in leaves (Table I). Thesemutants contain a high concentration of glucoerucin,which is hardly detected in the wild-type plants. More-over, glucoraphenin (4-methylsufinyl-3-butenyl gluco-sinolate), an S-oxygenated product of glucoraphasatin,also was not detected in the mutants. To determine theallelism of the two mutants, F1 progeny (grs1-23 grs1-1)were obtained by cross-pollination. The GSL profile ofthe F1 plant was similar to that of the respective parents,indicating that these mutants have lesions in the samegene that is involved in glucoraphasatin biosynthesis(Table I).In our previous genetic mapping experiment, GRS1

was mapped onto a 4.2-centimorgan interval end ofRaphanus linkage group R1 using genome-wide single-nucleotide polymorphism markers (Ishida et al., 2015).To fine-mapGRS1, we first screened a large segregatingpopulation (;5,000 F2 and F3 plants) derived from across between a grs1-1 mutant and the wild-typeHAGHN. The GSL analysis of the segregating popu-lation revealed that three group had recombinationbetween two insertion/deletion markers, s8D07 andssG02, putting GRS1 in a 23-kb genomic region (Fig. 1).By screening a BAC library, constructed fromwild-typecv. Miyashige, BAC contigs were built using theneighboring genetic markers (Fig. 1). The nucleotidesequence between s8D07 and ssG02 DNAmarkers wasdetermined by shotgun sequencing of the P64M08 BACclone. Although seven open reading frames (ORFs)were identified in this region by the AUGUSTUS geneprediction program (Stanke and Morgenstern, 2005),only ORF3 showed differences in the gene expressionlevels between the leaf and root of grs1-1 and the wildtype. The expression level of ORF3 was reduced dras-tically in the grs1-1 mutant (Supplemental Fig. S1).These findings strongly suggest that ORF3 is a candi-date GRS1.

Prediction of the Amino Acid Sequence andPhylogenetic Tree of 2-Oxoglutarate andFe(II)-Dependent Dioxygenases

Wenext determined the nucleotide sequence ofORF3in the grs1-1 and grs1-2 mutants. The 8.6- and 1.2-kbinsertions were identified in the first exon of grs1-1 andthe third exon of grs1-2, respectively (Fig. 2A). Thenucleotide sequence of the 8.6-kb insertion in the grs1-1allele contained high similarity to a Ty1-copia long ter-minal repeat retrotransposon. The 1.2-kb insertion ofgrs1-2 showed no similarity to long terminal repeatretrotransposons. Both the insertions led to an in-framestop codon just downstream of the insertion site (Fig.2B). Real-time PCR analysis revealed that the expression

of ORF3 in grs1-1 and grs1-2 was approximately 1:1,000and 1:10 of that in the wild type (HAGHN and cv. Tai-byosoubutori), respectively (Fig. 2C). Therefore, the lesionin the accumulation of glucoraphasatin in the two mu-tants was caused by the production of a truncated ORF3protein and/or the suppression of ORF3 expression.

A BLAST search against the GenBank ConservedDomain Database version 3.14 (http://www.ncbi.nlm.nih.gov/cdd/) indicated that the 372-amino acid proteinencoded by ORF3 is a member of the 2-oxoglutarateand Fe(II)-dependent dioxygenase (2OGD) superfamily.2OGDs consist of a nonheme dioxygenase in morphinesynthesis N-terminal (DIOX_N) in the N-terminal regionand a 2OG-Fe(II) oxygenase superfamily (2OG-FeII_Oxy)motif in the C-terminal region (De Carolis and DeLuca, 1994). In plants, 2OGD superfamily members are

Figure 2. Gene structure and expression ofORF3. A, Polymorphisms inORF3. The introns (horizontal lines), exons (black boxes), and 59 and 39untranslated regions (white boxes) of theORF3 are shown. Arrowheadsshow the large insertions. Bar = 100 bp. B, The insertion sequence in-dicated by lowercase letters. Asterisks show the in-frame stop codon. C,Quantitative analysis ofORF3mRNA expression in wild-type (HAGHNand cv. Taibyosoubutori [TIB]) and grs1 mutant plants. Each bar repre-sents the mean 6 SD (n = 3). The asterisk above the bar indicates sig-nificant differences (P , 0.01) between HAGHN and the respectivemutant, as determined by Student’s t test.

Plant Physiol. Vol. 173, 2017 1585

Biosynthesis Pathway of Glucosinolate in Radish

Dow



http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.ncbi.nlm.nih.gov/cdd/http://www.ncbi.nlm.nih.gov/cdd/

involved in various oxygenation/hydroxylation reac-tions. Arabidopsis and rice (Oryza sativa) contain 130 and114 2OGD proteins, respectively, classified into threeclasses, namely DOXA, DOXB, and DOXC, based onsimilarity of the deduced amino acid sequences (Kawaiet al., 2014). The domain organization of the predictedORF3 protein suggested that it belongs to theDOXC class.2OGDs of the DOXC class are classified into 57 phyloge-netic clades and are involved in the biosynthesis of variousmetabolites. To assess the biological function of ORF3, aphylogenetic tree was constructed with 11 functionallycharacterized 2OGDs in the DOXC class (Fig. 3). Phylo-genetic analysis revealed that ORF3 belongs to theDOXC31 clade, which includes AtGSL-OH of Arabi-dopsis, CrD4H of Catharanthus roseus, and ZmBX6 of Zeamays. AtGSL-OH is involved in GSL biosynthesis andcatalyzes the conversion of 3-butenyl glucosinolate to2-hydroxy-3-butenyl glucosinolate (Hansen et al., 2008).Homology between the deduced amino acid sequences ofORF3 and AtGSL-OHwas 50.1%. In the GSL biosyntheticpathway, another clade of DOXC class 2OGDs, includingAOP1, AOP2, and AOP3 (for Arabidopsis 2-oxoglutarate-dependent dioxygenases), is involved (Kliebenstein et al.,2001). However, these AOPs were classified into theDOXC20 clade. These data clearly showed that ORF3 is amember of the 2OGDs and is classified into the same cladeas the enzyme that modifies the aliphatic GSL side chain.However, it is difficult to infer a distinct function from thephylogenetic analysis.

Analysis of Transgenic Arabidopsis Overexpressing ORF3

Further functional characterization of ORF3 wasachieved by transgenic experiments inArabidopsis. It isknown that the efficiency of transgenesis in radish is

extremely low compared with other plants of theBrassicaceae and that accessions that show a high re-generation rate are limited. Therefore, an overexpressionconstruct of ORF3 was introduced into Arabidopsis. Asthe phenotype of grs1 mutants indicated that the sub-strate of GRS1 is glucoerucin, we surveyed and selecteda glucoerucin-rich accession, Ts-1, from the RIKENArabidopsis core collection and generated transgenicplants overexpressing ORF3 in the Ts-1 background. T3plants of three independent overexpression and emptyvector lines were obtained. In HPLC for desulfo-GSLs inleaf, an additional peak (peak 4) that was not detectedin the vector controls was obtained in all the over-expression lines (Fig. 4A). We confirmed that the reten-tion time of the additional peak was identical to that ofdesulfo-glucoraphasatin by comparing the GSL profileswith those of wild-type radish plants. In liquidchromatography-mass spectroscopy (LC-MS) analysis,peak 4 had mass-to-charge ratio (m/z) values of 178, 340,and 362, which corresponded to the previously reportedm/z values of desulfo-glucoraphasatin for [M+H2Glc],[M+H], and [M+Na], respectively (Fig. 4B; Kusznierewiczet al., 2013). Similarly, the mass spectra of peak 3 wereidentical to that of desulfo-glucoerucin (data not shown).The concentration of glucoraphasatin was 2.3 to 5.8mmol g21 dryweight in the overexpression lines; however,it was not detected in the empty vector controls(Supplemental Table S1). Contrary to the accumulation ofdesulfo-glucoraphasatin, the deslufo-glucoerucin (peak 3)concentrationwas decreased in the overexpression lines to40% of that in the vector control (Supplemental Table S1).In the overexpression lines, desulfo-glucoraphenin wasdetected byHPLC (Supplemental Table S1); the samewasnot detected in the vector control. Both glucoraphasatinand glucoraphenin contain a double bound in their sidechains between the third and fourth carbons. These results

Figure 3. Phylogenetic tree of 2OGDproteins in the DOXC class. Thededuced amino acid sequenceswere aligned with the MUSCLEprogram. The phylogenetic tree wasconstructed by the maximum like-lihood method using the MEGA6program (Tamura et al., 2013). Thebranches indicate bootstrap valuescalculated by the 1,000-permutationtest. The classificationwas accordingto Kawai et al. (2014). ArabidopsisALKBH2 (At2g22260), which wasclassified into the DOXA5 class, wasused as an outgroup. The accessionnumbers and Arabidopsis GenomeInitiative codes are indicated in pa-rentheses.


Kakizaki et al.

Dow



http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1

indicated that GRS1 is a gene of ORF3 and is responsiblefor the desaturation of the side chain in aliphatic GSLs, inparticular glucoerucin (Fig. 5).

In Vitro Assay of GRS1

To determine if GRS1 encodes a glucoraphasatinsynthase, we investigated its ability to desaturate

the side chain of glucoerucin. To this end, the full-lengthcDNA of GRS1 was cloned into a pColdIII vector andthe recombinant protein was expressed in Escherichiacoli. The recombinant GRS1 protein was solubilizedby sonication, and the crude supernatant was used forthe in vitro assay. After incubation of the mixture contain-ing recombinant GRS1, glucoerucin, Fe2+, 2-oxogulutarate,and ascorbate were analyzed by HPLC. However, noenzymatic activity was detected.

Figure 4. ORF3 mediates the biosyn-thesis of glucoraphasatin in transgenicArabidopsis. A, Chromatograms of desulfo-GSLs recorded at 229nm for the transgenicTs-1 containing the p35S::ORF3 construct,empty vector, andwild-type radish. Peak 1,Sinigrin (internal control); peak 2, glucor-aphanin; peak 3, glucoerucin; peak 4,glucoraphasatin. The leaves of 39-d-old T3transformants were subsequently used inthe HPLC analysis. B, LC-MS and tandemmass spectrometry (MS/MS) spectra show-ing peak 4 obtained in transgenic Arabi-dopsis overexpressingORF3.



Dow



Change in GSL Composition in Grafted Radish Plants

An analysis of GRS1 expression in cv. Taibyosoubu-tori, which is a four-way cross cultivar, revealed anabundance of transcripts in the aerial tissues like leaves,but not in roots and flower buds (Fig. 6). In contrast,higher accumulation of aliphatic GSLs was detected inroots and flower buds (Table II). These results suggestthe long-distance transport of GSLs from leaves to rootsand flower buds. The observation of low levels of GRS1expression in the root during the thickening stage andthe abundant accumulation of GSLs in the rootprompted us to hypothesize that glucoraphasatin issynthesized in the leaf and accumulated in the root. Totest this hypothesis, we reciprocally grafted grs1-1 andwild-type radish plants on each other. For the graftingexperiment, cv. Karami199, which has a high concen-tration of glucoraphasatin in the root, was used. Scionsof grs1-1 were grafted on the stocks of both grs1-1and cv. Karami199. Likewise, scions of cv. Karami199were grafted onto the stocks of both grs1-1 and cv.Karami199. At the thickening stage of the graftedplants, there were no significant differences in themorphological traits between the heterografted andhomografted plant roots (Fig. 7A).

In the analysis of GSL levels in the homografts, thelevels in the leaves and roots were not significantlydifferent between the homografts and nongrafts (datanot shown). These results indicate that the distributionand accumulation of GSLs were not affected by graft-ing. In the leaf, glucoerucin was the dominant GSL inthe grafted GR/KA plants. Likewise, KA/GR plantsshowed high accumulation of glucoraphasatin in theleaf (Fig. 7B). These observations indicated the presence

of genotype-dependent GSL profiles in the leaf. Incontrast to the GSL profile of the leaf, the GSL profile ofthe root was affected by the genotype of the leaf (Fig.7C). Glucoraphasatin, which was not detected in root ofGR/GR plants, was accumulated to 157.7 mmol g21 dryweight in the root of KA/GR plants. The total amountof GSL in the root was not significantly different be-tween KA/KA and KA/GR plants (SupplementalTable S2). In the roots of KA/GR plants, glucor-aphasatin accounted for more than 85% of the totalGSLs. These results clearly showed that the majority ofthe aliphatic GSLs in the root were transported from theleaf.

DISCUSSION

Plants, including those of the Brassicaceae family,generally contain GSLs as the precursors of defensivesubstances against pests and pathogens. Nearly200 types of GSLs with different substituents areknown. Combinations of GSL-associated genes lead tothis variety of GSLs. A few transcription factors, MYBsand MYCs, control the gene expression of a large set ofGSL-associated genes responsive to environmentalcues (Gigolashvili et al., 2007, 2008; Hirai et al., 2007;Schweizer et al., 2013). In recent genome projects inBrassica spp. vegetables, the number of GSL-relatedgenes has been estimated and species-specific biosyn-thesis pathways have been elucidated (Wang et al.,2013; Liu et al., 2014; Mitsui et al., 2015). Mitsui et al.(2015) identified three METHYLTHIOALKYLMALATESYNTHASE1 (MAM1)-like genes in the radish genome,but MAM3-like genes were found to be absent. TheMAMs determine the side chain length of the aliphaticGSLs in Arabidopsis (Kroymann et al., 2001). Among

Figure 5. Proposed GSL biosynthesis pathway in radish. GRS1 is re-sponsible for the desaturation of the side chain of glucoerucin. FMOGS-OXs, Flavin monooxygenases, which catalyze the S-oxygenation ofthe side chain.

Figure 6. GRS1 expression. Quantitative analysis of GRS1 mRNAabundance was performed in various tissues. mRNA was extractedfrom the listed tissues of wild-type cv. Taibyosoubutori plants. ThemRNA levelswere analyzed by real-time PCR, and the expression levelswere normalized to that of ACTIN. Each data point represents themean 6 SD (n = 4).


Kakizaki et al.

Dow



http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1

the several aliphatic GSLs in Arabidopsis, only the four-carbon aliphatic GSLs, such as glucoraphasatin andglucoraphenin, have been detected in radish. TheAOPs, which function in the oxidation of glucor-aphenin in Arabidopsis, have not been identified in theradish genome (Kliebenstein et al., 2001; Mitsui et al.,2015). We identified a single genomic region that con-trols the GSL composition in radish. This region, whichwas further delimited in this study, contained a 2OGDinvolved in the biosynthesis of the radish-specific GSL,glucoraphasatin. In plants, 2OGDs have awide range ofbiochemical functions (Loenarz and Schofield, 2011;Kawai et al., 2014). Although genes with similarity toGRS1 have been identified in the genomes of both

Arabidopsis and cabbage, these species contain onlyglucoerucin and not glucoraphasatin in abundance. Thedivergence of the Arabidopsis lineage and the Brassica-Raphanus ancestor from a common ancestor has beenreported to have occurred 38.8 million years ago (Mitsuiet al., 2015). After the divergence, whole-genome triplica-tion in the Brassica-Raphanus ancestor may have occurredbetween 15.6 and 28.3 million years ago (Mitsui et al.,2015). Geneswith similarity toGRS1have been retained inBrassica spp., but the specificity of glucoraphasatin toradish indicates that GRS1 was generated only in theRaphanus genome after the whole-genome triplication.

In plants, 2OGDs belong to the second largest en-zyme family. Its members facilitate numerous oxidative

Table II. GSL content in different tissues

Numbers are averages 6 SD (n = 3). Values given are mmol g21 dry weight. Data within a column followed by the same letter are not significantlydifferent (P , 0.05). n.d., Not detected.

Tissue Glucoraphanin Glucoraphenin Glucoerucin Glucoraphasatin Glucobrassicin4OH-

Glucobrassicin

4OMe-

GlucobrassicinTotal

Leaf n.d. 2.3 6 1.3 a 0.1 6 0.2 a 24.8 6 16.4 a,b 0.1 6 0.1 a n.d. n.d. 27.3 6 17.8 a

Stem 0.3 6 0.5 a 5.2 6 1.5 a n.d. 41.7 6 9.6 a 0.6 6 0.3 b 0.1 6 0.1 a n.d. 47.9 6 11.8 a,b

Hypocotyl n.d. 1.3 6 0.7 a n.d. 14.0 6 4.3 b n.d. n.d. n.d. 15.2 6 4.6 a

Root n.d. 3.0 6 1.2 a 0.4 6 0.2 a 72.6 6 11.8 c n.d. 0.1 6 0.1 a 0.2 6 0.1 76.2 6 12.9 b

Flower bud 2.1 6 0.3 b 64.2 6 22.0 b 1.1 6 0.1 b 132.6 6 5.6 d 0.9 6 0.3 b 0.3 6 0.1 b n.d. 201.2 6 16.9 c

Figure 7. GSL concentrations inleaves and roots of grafted radishplants. A, Typical roots of recipro-cally grafted radish plants. Lettersindicate grafted plants of grs1-1 ongrs1-1 (GR/GR), cv. Karami199 ongrs1-1 (KA/GR), cv. Karami199on cv. Karami199 (KA/KA), andgrs1-1 on cv. Karami199 (GR/KA).The grafted plants were harvested60 d after grafting. Bars = 1 cm. Band C, GSL concentrations in theleaves (B) and roots (C) of graftedradish plants. DW, Dry weight; n.d.,not detected.



Dow



reactions, including hydroxylations, desaturations, di-merizations, and cyclizations (Loenarz and Schofield,2008; Farrow and Facchini, 2014). The Arabidopsis ge-nome contains 130 2OGD genes, corresponding to 0.5%of the total gene number (Kawai et al., 2014). Structuralanalysis of some 2OGDs revealed that the canonicalstructure contains a double-stranded b-helix core foldthat supports the residues coordinating iron (Cliftonet al., 2006). grs1-1 encodes a truncated protein con-taining an in-frame stop codon caused by the insertionof a retrotransposon, resulting in the deletion of iron-binding residues (Supplemental Fig. S2). In grs1-2, theiron-binding domain is present but the 2-oxoglutarate-binding domain is missing as a consequence of the 2-kbinsertion (Supplemental Fig. S2).

GRS1 belongs to the DOXC31 clade, which containsfunctionally diverse 2OGDs involved in various meta-bolic pathways, such as ZmBX6 from Z. mays involvedin the biosynthesis of benzoxiazinoid, which functionsin defense and allelopathy in graminaceous plants(Frey et al., 2003). This enzyme catalyzes the hydrox-ylation of DIBOA glucoside at position C7 (Jonczyket al., 2008). CrD4H from C. roseus is involved in thebiosynthesis of monoterpenoid indole alkaloids, whichcatalyze the hydroxylation of desacetoxyvindoline atposition C4 in a later step of vindoline biosynthesis(Vazquez-Flota et al., 1997). AtGSL-OH also belongsto DOXC31, which catalyzes the hydroxylation of3-butenyl glucosinolate (Hansen et al., 2008). All thepreviously characterized 2OGDs belong to DOXC31and catalyze the hydroxylation of the substrate. Inter-estingly, despite being a member of DOXC31, GRS1has a desaturation activity. To determine the in vitroactivity of GRS1, we tried to develop an assay usingrecombinant GRS1. When different expression condi-tions were considered (e.g. expression vectors, coex-pression with molecular chaperons, and temperatureduring induction), we could not detect the enzymaticactivity of GRS1 using commercial glucoerucin as asubstrate. Because glucoraphasatin was accumulated inthe transgenic Arabidopsis lines overexpressing theGRS1 cDNA, an additional factor might be necessaryfor the desaturation of glucoerucin or for the desatu-ration of the side chain occurring at the earlier stages ofthe GSL synthesis pathway.

Our grafting experiments showed that glucoraphasatinwas distributed from leaves to roots by the long-distancetransport machinery (Fig. 7, B and C). Two nitrate/peptide transporters, namely, GLUCOSINOLATETRANSPORTER1 (GTR1) and GTR2, are essential forGSL translocation in Arabidopsis (Nour-Eldin et al.,2012). In particular, short-chain aliphatic GSLs, suchas glucoerucin, have higher affinity for GTR than thelong-chain aliphatic or indolic GSLs (Andersen et al.,2013). In radish, some GSL synthesis-associated genes(e.g. RsBCAT4, RsUGT74B1, and RsGS-OX1) wereshown to be expressed weakly in roots, and abundanttranscripts were detected in leaves and stems (Wanget al., 2013). As shown in Figure 6, GRS1was stronglyexpressed in leaves but not in roots and flower buds.

These results suggest that similar GSL transport sys-tems are present in radish and Arabidopsis. The con-centrations of glucoerucin and glucoraphasatin in rootswere 3 times higher in cv. Karami199 than in grs1-1 (Fig.7C). The grafted plants with cv. Karami199 leaves andgrs1-1 roots showed 2 times higher concentration than theGR/KA plants. These findings suggest that not only thecomposition but also the concentration of GSLs in theroots is determined by the genotype of the leaf.

It is of interest to alter the composition of metabo-lites in vegetables to confer new flavors, pest protec-tion, or anticancer activity. Tattersall et al. (2001)reported that introducing the entire cyanogenic glu-coside pathway of sorghum (Sorghum bicolor) intoArabidopsis resulted in increased resistance to specificinsects. In many pharmacological studies, sulforaphane,an isothiocyanate derived from glucoraphanin(4-methylsulfinylbutyl glucosinolate), has been shownto have health-promoting activities (Juge et al., 2007).To increase the glucoraphanin composition, a MYB28allele derived from the wild species Brassica villosawas introgressed into broccoli (B. oleracea var italica;Traka et al., 2013). These metabolite-engineering ef-forts permit the breeding of new vegetables beneficialto human health. In previous studies, using a radishvariety whose GSL composition was known to bepoor, we found a single mutation leading to a quali-tative change in radish, resulting in the accumulationof glucoerucin instead of glucoraphasatin in the radishroot (Ishida et al., 2015). This drastic change providesan advantage for dishes containing radish. It is knownthat an isothiocyanate derived from glucoraphasatinproduces a yellow pigment, methanethiol, which im-parts color and flavor to the dishes (Takahashi et al.,2015). These phenomena are among the reasons citedfor avoiding the cooking of radish. Our results maybe of use in metabolite engineering for breeding ofhigh-value vegetables.

MATERIALS AND METHODS

Plant Materials

The radish (Raphanus sativus) mutant lacking glucoraphasatin, grs1-1, was asibling of NMR154N, described in our previous report (Ishida et al., 2015). Theallelic mutant, grs1-2, was an inbred line derived from Tohoku Karami daikon(R. sativus) and was identified by GSL profiling. The wild-type inbred lineHAGHN was used for genetic mapping. The commercial lines cv. Taibyosou-butori (Takii seed) and cv. Karami199 (Kaneko Seeds) were used for gene ex-pression analysis and for the grafting experiments.

Extraction and HPLC Analysis of GSLs

Extraction of GSLs from root and leaf was performed after lyophilizationand pulverization using aMulti-Beads Shocker following themethod describedby Ishida et al. (2015). GSLs were desulfurized with sulfatase (Sigma) accordingto themethod of Bjerg and Sørensen (1987), and desulfo-GSLs were subjected toHPLC, as reported in our previous study (Ishida et al., 2012). HPLC was per-formed on an LC-20A chromatograph (Shimadzu) fitted with a 5C 18-MS-IIcolumn (150 mm3 4.6 mm i.d., 5 mm; Nacalai Tesque). The HPLC analysis wasperformed with a flow rate of 1.5 mL min21 at a column oven temperature of30°C, and the absorbance was measured at a wavelength of 229 nm. The mobilephase consisted of ultrapure water (A) and 20% (v/v) acetonitrile (B). The


Kakizaki et al.

Dow



http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1

mobile phase program was as follows: 1% (v/v) solvent B for 1 min, followedby a linear elution gradient over the next 20 min to 99% (v/v) solvent B, then99% (v/v) solvent B for 3 min, which was changed to 1% (v/v) solvent B at24.1 min, and then 1% (v/v) solvent B for 10 min (total, 35 min). The individualGSL contents were calculated by the ratios of the individual desulfo-GSL peakareas to the peak areas of an internal standard, sinigrin (Sigma), and a responsefactor (ISO9167-1).

Genetic Mapping and BAC Screening

A population of 5,198 self-pollinated F2 and F3 progeny of grs1-1 andHAGHN was used for fine-mapping of grs1. The leaves of plants showing re-combination between the interval markers were used for GSL analysis. TheBAC libraries constructed using total DNA of the radish ‘Miyashige’ doubledhaploid line after partial digestion with HindIII were used for screening(Kitashiba et al., 2014).

Quantitative Real-Time PCR

Total RNA was extracted from various tissues with the RNeasy Plant MiniKit (Qiagen). The first-strand cDNAwas synthesized with 500 ng of RNA usingthe PrimeScript RT Reagent Kit with gDNA Eraser (TaKaRa) with randomhexamer and oligo(dT) primers in a volume of 20 mL. Quantitative real-timePCR was performed in a total volume of 25 mL, 1 mL of the cDNA, 0.4 mM gene-specific primers, and 12.5 mL of SYBR Premix ExTaq (TaKaRa) on a ThermalCycler Dice Real-Time System (TaKaRa) according to the manufacturer’s in-structions. The radish ACTIN gene was used as an internal control (Zouet al., 2013). All the expression data were based on at least three biologicalreplications.

Full-Length cDNA Cloning and Transformation

Total RNAwas extracted fromHAGHNwith the RNeasy PlantMini Kit, andRACEwas performedwith the FirstChoice RLM-RACE Kit (Life Technologies).The gene-specific primer sequences are listed in Supplemental Table S3. Theamplified fragments were cloned into a pCR2.1-TOPO vector (Invitrogen) andsequenced with the BigDye Terminator Version 3.1 Cycle Sequencing Kit (LifeTechnologies). A coding sequence of ORF3 was amplified from total RNA ofHAGHN leaves. The cauliflowermosaic virus 35S promoter and a terminator ofnopaline synthase were amplified from the pBI121 binary vector. Three am-plified fragments were fused into the HindIII/EcoRI-digested pZK3B binaryvector (Kuroda et al., 2010) using the In-Fusion HD Cloning Kit (TaKaRa). Thisconstruct was transformed intoArabidopsis (Arabidopsis thaliana) accession Ts-1using Rhizobium radiobacter GV3101::pMP90 by the floral dip method (Cloughand Bent, 1998). To synchronize the germination, all the seeds were kept at 4°Cfor 3 d after sowing. Plants were grown on soil (expanded vermiculite) or on0.8% agarmediumwith 0.53Murashige and Skoog salts under a 16-h/8-h cycleof white light (60–80 mmol m22 s21) and dark at 22°C in a growth chamber (FLI-2000; Eyela). The primer sequences used for vector construction are listed inSupplemental Table S3.

LC-MS Analysis of GSLs in Transformed Arabidopsis

The conditions for the detection and identification of desulfo-GSLs byLC-MSwere as follows. The samples (15 mL) were injected into 1200 Series HPLCequipment (Agilent) and desulfo-GSLs were separated on a TSKgel super-ODScolumn (2 3 100 mm, 3-mm particle size, 40°C) using a 0% to 20% acetonitrilegradient in water (46 min) with a flow rate of 0.4 mL min21. The detection wasdone online, first with a photodiode array detector at 230 nm (wavelength 190–950 nm) and subsequently with the LTQ Orbitrap XL MS/MS system (ThermoFisher Scientific) operated in electrospray ionization positive ion mode (m/z =100–800; spray voltage, 3.5 kV; temperature of the heated capillary, 300°C). Theflow rates for nitrogen sheath gas and auxiliary gas were set to 50 and 10 ar-bitrary units min21, respectively. Desulfo-glucoraphasatin (m/z = 178, 340, and362) was monitored by specific MS/MS scans in addition to the full scan.

Phylogenetic Analysis

The amino acid sequences were deduced from nucleotide sequences of thepredicted 2OGD genes and then aligned using the MUSCLE program (Edgar,2004). The number of amino acids substituted between each pair of 2OGD

proteins was estimated by the LG + G model (Le and Gascuel, 2008). From thenumber of estimated amino acid substitutions, a phylogenetic tree was recon-structed by the maximum likelihood method using MEGA version 6.06(Tamura et al., 2013). The bootstrap values were calculated with 1,000 repli-cations. Arabidopsis ALKBH2 (At2g22260), which was classified in the DOXA5class, was used as an outgroup.

Protein Assay of GRS1

The coding sequence ofGRS1was amplified fromwild-type HAGHN usingPCR with the primers mentioned in Supplemental Table S3. The amplifiedfragments were cloned into XhoI and SalI sites of the pColdI vector (TaKaRa)using the In-Fusion HD Cloning Kit (TaKaRa). The recombinant plasmids wereintroduced into Escherichia coli BL21 (DE3) pLysS strain.

Toproduce the recombinantGRS1protein, the bacteriawere cultured at 37°Cin 10 mL of Luria-Bertani medium supplemented with ampicillin (50 mg mL21)and chloramphenicol (30 mg mL21) until the optical density at 600 nm reached0.5. The protein synthesis was then induced by the addition of 0.1 mM iso-propylthio-b-galactoside and further culturing for 20 h at 15°C. The culture wasthen centrifuged at 5,000g for 5 min at 4°C. The cell pellets were resuspended in1 mL of 13 phosphate-buffered saline buffer, pH 7.4, sonicated by VP-300N(Taitec), and centrifuged at 15,000g for 10 min at 4°C. The reaction mixture con-tained 100 mM NaH2PO4, pH 6.8, 10 mM a-ketoglutaric acid, 10 mM ascorbate,0.25 mM ferrous sulfate, 0.25 mM glucoerucin, and 20 mL of crude protein solution.The assays were performed at 30°C for 30 min. The reactions were initiated by theaddition of the enzyme and terminated by extraction with methanol.

Reciprocal Grafting

Ten-day-old seedlings of grs1-1 and cv. Karami199 were used for grafting.Seedlings with cotyledon and the first true leaf were cut at the hypocotyl with arazor in a horizontal direction. The scions and stocks were joined with a holder(superwith14; Nasunics). The scions of grs1-1were grafted on the stocks of bothgrs1-1 and cv. Karami199, and the scions of cv. Karami199 were grafted on thestocks of both grs1-1 and cv. Karami199. The grafted plants were named GR/GR, GR/KA, KA/GR, and KA/KA. They were grown in plastic containers for10 d and transplanted into 15-cm pots in the greenhouse. For GSL analysis, theleaves and roots were harvested 60 d after grafting.

Chemical Compounds Studied

Chemical compounds studiedwere as follows: glucoraphasatin (PubChemCID: 6442557), glucoraphenin (PubChem CID: 15559531), glucoerucin(PubChem CID: 656539), and glucoraphanin (PubChem CID: 9548633).

Accession Numbers

The sequence reported in this article has been deposited in the GenBankdatabase with GenBank/EMBL/DNA Data Bank of Japan accession numberLC077856 (GRS1 full-length cDNA, HAGHN).

Supplemental Data

The following supplemental materials are available.

Supplemental Figure S1. Reverse transcription-PCR analysis in grs1-1 andthe wild type for seven ORFs linked to the grs mutation.

Supplemental Figure S2. Alignment of the deduced amino acid sequencesof GRS1 alleles.

Supplemental Table S1. Aliphatic GSL profile of the transgenic plants.

Supplemental Table S2. Aliphatic and indolic GSLs in grafted plants.

Supplemental Table S3. Primer sequences used in this work.

ACKNOWLEDGMENTS

We thank Y. Araki, S. Morimoto, Y. Kawamoto, C. Okuyama, Y. Niina, andS. Negoro for technical assistance; Dr. H. Fukuoka (Takii seed) for helpfuldiscussions; Dr. M. Kuroda (NARO) for providing the binary vector; The



Dow



http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1http://www.plantphysiol.org/cgi/content/full/pp.16.01814/DC1

Arabidopsis accession Ts-1was provided by RIKENBioResource Center, whichis participating in the National Bio-Resource Project of the Ministry of Educa-tion, Culture, Sports, Science and Technology, Japan.

Received November 30, 2016; accepted January 16, 2017; published January 18,2017.

LITERATURE CITED

Andersen TG, Nour-Eldin HH, Fuller VL, Olsen CE, Burow M, HalkierBA (2013) Integration of biosynthesis and long-distance transport es-tablish organ-specific glucosinolate profiles in vegetative Arabidopsis.Plant Cell 25: 3133–3145

Barillari J, Cervellati R, Paolini M, Tatibouët A, Rollin P, Iori R (2005)Isolation of 4-methylthio-3-butenyl glucosinolate from Raphanus sativussprouts (kaiware daikon) and its redox properties. J Agric Food Chem53: 9890–9896

Bjerg B, Sørensen H (1987) Quantitative analysis of glucosinolates and HPLC ofintact glucosinolates. In JP Wathelet, ed, Glucosinolates in Rapeseeds: Ana-lytical Aspects. Martinus Nijhoff, Dordrecht, The Netherlands, pp 125–150

Clifton IJ, McDonough MA, Ehrismann D, Kershaw NJ, Granatino N,Schofield CJ (2006) Structural studies on 2-oxoglutarate oxygenases andrelated double-stranded beta-helix fold proteins. J Inorg Biochem 100:644–669

Clough SJ, Bent AF (1998) Floral dip: a simplified method forAgrobacterium-mediated transformation of Arabidopsis thaliana. Plant J 16:735–743

De Carolis E, De Luca V (1994) 2-Oxoglutarate-dependent dioxygenaseand related enzymes: biochemical characterization. Phytochemistry 36:1093–1107

Edgar RC (2004) MUSCLE: multiple sequence alignment with high accur-acy and high throughput. Nucleic Acids Res 32: 1792–1797

Fahey JW, Zalcmann AT, Talalay P (2001) The chemical diversity anddistribution of glucosinolates and isothiocyanates among plants. Phy-tochemistry 56: 5–51

Farrow SC, Facchini PJ (2014) Functional diversity of 2-oxoglutarate/Fe(II)-dependent dioxygenases in plant metabolism. Front Plant Sci 5:524

Frey M, Huber K, Park WJ, Sicker D, Lindberg P, Meeley RB, SimmonsCR, Yalpani N, Gierl A (2003) A 2-oxoglutarate-dependent dioxygenaseis integrated in DIMBOA-biosynthesis. Phytochemistry 62: 371–376

Giamoustaris A, Mithen R (1995) The effect of modifying the glucosinolatecontent of leaves of oilseed rape (Brassica napus ssp oleifera) on its in-teraction with specialist and generalist pests. Ann Appl Biol 126: 347–363

Gigolashvili T, Engqvist M, Yatusevich R, Müller C, Flügge UI (2008)HAG2/MYB76 and HAG3/MYB29 exert a specific and coordinatedcontrol on the regulation of aliphatic glucosinolate biosynthesis inArabidopsis thaliana. New Phytol 177: 627–642

Gigolashvili T, Yatusevich R, Berger B, Müller C, Flügge UI (2007) TheR2R3-MYB transcription factor HAG1/MYB28 is a regulator ofmethionine-derived glucosinolate biosynthesis in Arabidopsis thaliana.Plant J 51: 247–261

Halkier BA, Gershenzon J (2006) Biology and biochemistry of glucosino-lates. Annu Rev Plant Biol 57: 303–333

Hansen BG, Kerwin RE, Ober JA, Lambrix VM, Mitchell-Olds T, GershenzonJ, Halkier BA, Kliebenstein DJ (2008) A novel 2-oxoacid-dependent dioxy-genase involved in the formation of the goiterogenic 2-hydroxybut-3-enylglucosinolate and generalist insect resistance in Arabidopsis. Plant Physiol148: 2096–2108

Hirai MY, Sugiyama K, Sawada Y, Tohge T, Obayashi T, Suzuki A, ArakiR, Sakurai N, Suzuki H, Aoki K, et al (2007) Omics-based identificationof Arabidopsis Myb transcription factors regulating aliphatic glucosino-late biosynthesis. Proc Natl Acad Sci USA 104: 6478–6483

Ishida M, Kakizaki T, Morimitsu Y, Ohara T, Hatakeyama K, Yoshiaki H,Kohori J, Nishio T (2015) Novel glucosinolate composition lacking4-methylthio-3-butenyl glucosinolate in Japanese white radish (Rapha-nus sativus L.). Theor Appl Genet 128: 2037–2046

Ishida M, Nagata M, Ohara T, Kakizaki T, Hatakeyama K, Nishio T(2012) Small variation of glucosinolate composition in Japanese cultivarsof radish (Raphanus sativus L.) requires simple quantitative analysis forbreeding of glucosinolate component. Breed Sci 62: 63–70

Jonczyk R, Schmidt H, Osterrieder A, Fiesselmann A, Schullehner K,Haslbeck M, Sicker D, Hofmann D, Yalpani N, Simmons C, et al(2008) Elucidation of the final reactions of DIMBOA-glucoside biosyn-thesis in maize: characterization of Bx6 and Bx7. Plant Physiol 146: 1053–1063

Juge N, Mithen RF, Traka M (2007) Molecular basis for chemopreventionby sulforaphane: a comprehensive review. Cell Mol Life Sci 64: 1105–1127

Kawai Y, Ono E, Mizutani M (2014) Evolution and diversity of the2-oxoglutarate-dependent dioxygenase superfamily in plants. Plant J 78:328–343

Kitashiba H, Li F, Hirakawa H, Kawanabe T, Zou Z, Hasegawa Y, Tonosaki K,Shirasawa S, Fukushima A, Yokoi S, et al (2014) Draft sequences of theradish (Raphanus sativus L.) genome. DNA Res 21: 481–490

Kliebenstein DJ, Lambrix VM, Reichelt M, Gershenzon J, Mitchell-OldsT (2001) Gene duplication in the diversification of secondary metabo-lism: tandem 2-oxoglutarate-dependent dioxygenases control glucosi-nolate biosynthesis in Arabidopsis. Plant Cell 13: 681–693

Kroymann J, Textor S, Tokuhisa JG, Falk KL, Bartram S, Gershenzon J,Mitchell-Olds T (2001) A gene controlling variation in Arabidopsisglucosinolate composition is part of the methionine chain elongationpathway. Plant Physiol 127: 1077–1088

Kuroda M, Kimizu M, Mikami C (2010) A simple set of plasmids for theproduction of transgenic plants. Biosci Biotechnol Biochem 74: 2348–2351

Kusznierewicz B, Iori R, Piekarska A, Namie�snik J, Bartoszek A (2013)Convenient identification of desulfoglucosinolates on the basis of massspectra obtained during liquid chromatography-diode array-electrosprayionisation mass spectrometry analysis: method verification for sprouts ofdifferent Brassicaceae species extracts. J Chromatogr A 1278: 108–115

Le SQ, Gascuel O (2008) An improved general amino acid replacementmatrix. Mol Biol Evol 25: 1307–1320

Liu S, Liu Y, Yang X, Tong C, Edwards D, Parkin IA, Zhao M, Ma J, Yu J,Huang S, et al (2014) The Brassica oleracea genome reveals the asym-metrical evolution of polyploid genomes. Nat Commun 5: 3930

Loenarz C, Schofield CJ (2008) Expanding chemical biology of 2-oxoglutarateoxygenases. Nat Chem Biol 4: 152–156

Loenarz C, Schofield CJ (2011) Physiological and biochemical aspects ofhydroxylations and demethylations catalyzed by human 2-oxoglutarateoxygenases. Trends Biochem Sci 36: 7–18

Mithen RF, Dekker M, Verkerk R, Rabot S, Johnson IT (2000) The nu-tritional significance, biosynthesis and bioavailability of glucosinolatesin human foods. J Sci Food Agric 80: 967–984

Mitsui Y, Shimomura M, Komatsu K, Namiki N, Shibata-Hatta M, ImaiM, Katayose Y, Mukai Y, Kanamori H, Kurita K, et al (2015) The radishgenome and comprehensive gene expression profile of tuberous rootformation and development. Sci Rep 5: 10835

Montaut S, Barillari J, Iori R, Rollin P (2010) Glucoraphasatin: chemistry,occurrence, and biological properties. Phytochemistry 71: 6–12

Nour-Eldin HH, Andersen TG, Burow M, Madsen SR, Jørgensen ME,Olsen CE, Dreyer I, Hedrich R, Geiger D, Halkier BA (2012) NRT/PTRtransporters are essential for translocation of glucosinolate defencecompounds to seeds. Nature 488: 531–534

Rask L, Andréasson E, Ekbom B, Eriksson S, Pontoppidan B, Meijer J(2000) Myrosinase: gene family evolution and herbivore defense inBrassicaceae. Plant Mol Biol 42: 93–113

Schweizer F, Fernández-Calvo P, Zander M, Diez-Diaz M, Fonseca S,Glauser G, Lewsey MG, Ecker JR, Solano R, Reymond P (2013) Ara-bidopsis basic helix-loop-helix transcription factors MYC2, MYC3, andMYC4 regulate glucosinolate biosynthesis, insect performance, andfeeding behavior. Plant Cell 25: 3117–3132

Stanke M, Morgenstern B (2005) AUGUSTUS: a web server for gene pre-diction in eukaryotes that allows user-defined constraints. Nucleic AcidsRes 33: W465–W467

Takahashi A, Yamada T, Uchiyama Y, Hayashi S, Kumakura K, Takahashi H,Kimura N, Matsuoka H (2015) Generation of the antioxidant yellow pigmentderived from 4-methylthio-3-butenyl isothiocyanate in salted radish roots(takuan-zuke). Biosci Biotechnol Biochem 79: 1512–1517

Tamura K, Stecher G, Peterson D, Filipski A, Kumar S (2013) MEGA6:Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol 30:2725–2729

Tattersall DB, Bak S, Jones PR, Olsen CE, Nielsen JK, Hansen ML, HøjPB, Møller BL (2001) Resistance to an herbivore through engineeredcyanogenic glucoside synthesis. Science 293: 1826–1828


Kakizaki et al.

Dow



Traka MH, Saha S, Huseby S, Kopriva S, Walley PG, Barker GC, Moore J,Mero G, van den Bosch F, Constant H, et al (2013) Genetic regulation ofglucoraphanin accumulation in Beneforté broccoli. New Phytol 198:1085–1095

Vazquez-Flota F, De Carolis E, Alarco AM, De Luca V (1997) Molecularcloning and characterization of desacetoxyvindoline-4-hydroxylase, a2-oxoglutarate dependent-dioxygenase involved in the biosynthesis ofvindoline in Catharanthus roseus (L.) G. Don. Plant Mol Biol 34: 935–948

Visentin M, Tava A, Iori R, Palmieri S (1992) Isolation and identification oftrans-4-(methylthio)-3-butenyl glucosinolate from radish roots (Rapha-nus sativus L). J Agric Food Chem 40: 1687–1691

Wang Y, Pan Y, Liu Z, Zhu X, Zhai L, Xu L, Yu R, Gong Y, Liu L (2013) Denovo transcriptome sequencing of radish (Raphanus sativus L.) and

analysis of major genes involved in glucosinolate metabolism. BMCGenomics 14: 836

Wittstock U, Halkier BA (2002) Glucosinolate research in the Arabidopsisera. Trends Plant Sci 7: 263–270

Wu G, Zhang L, Yin Y, Wu J, Yu L, Zhou Y, Li M (2015) Sequencing, denovo assembly and comparative analysis of Raphanus sativus tran-scriptome. Front Plant Sci 6: 198

Zhang Y, Kensler TW, Cho CG, Posner GH, Talalay P (1994) Anticarci-nogenic activities of sulforaphane and structurally related syntheticnorbornyl isothiocyanates. Proc Natl Acad Sci USA 91: 3147–3150

Zou Z, Ishida M, Li F, Kakizaki T, Suzuki S, Kitashiba H, Nishio T (2013) QTLanalysis using SNP markers developed by next-generation sequencing foridentification of candidate genes controlling 4-methylthio-3-butenyl glucosi-nolate contents in roots of radish, Raphanus sativus L. PLoS ONE 8: e53541



Dow



Documents

A 2-Oxoglutarate-Dependent Dioxygenase Mediates the...A 2-Oxoglutarate-Dependent Dioxygenase Mediates the Biosynthesis of Glucoraphasatin in Radish1[OPEN] Tomohiro Kakizaki*, Hiroyasu