INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

The glucosinolates, a large group of naturally occurring plant defense compounds, are almost exclusively limited to the order Capparales. These nitrogen- and sulfur-containing secondary metabolites are derived from a variety of protein amino acids (Met, Leu, iso-Leu, Val, Trp, and Phe) through a three-part biosynthetic pathway (Halkier and Du, 1997) comprising: the elongation of the amino acid carbon chain, the formation of the basic glucosinolate skeleton, and further side chain modification (Fig. 1). Elongation of the carbon chain occurs via a threestep process that is similar to elongation of 2-oxoisovalerate in Leu biosynthesis (Chisholm and Wetter, 1964; Graser et al., 2000). First, the 2-oxo acid formed by de-amination of the amino acid is condensed with acetyl-coenzyme A by an isopropylmalate synthase-like enzyme (GS-elongase; de Quiros et al., 2000). Then, isomerization of the resulating alkylmalate followed by oxidative decarboxylation leads to a new 2-oxo acid with an additional methylene group. This new 2-oxo acid can either proceed through an additional carbon elongation cycle or undergo transamination to form a chain-extended amino acid (Fig. 1).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Glucosinolate biosynthetic pathway. I, 2-Alkylmalic acid; II, 3-alkylmalic acid; III, 2-oxo acid. A, Basic glucosinolate structure. B, Outline of the pathway which can be divided into three parts: elongation of the amino acid side chain, formation of the basic glucosinolate skeleton, and further side chain modification. Each chain elongation cycle adds an additional methylene group (Graser et al., 2000).

The committed step in formation of the basic glucosinolate skeleton is conversion of the amino acid (chain extended or unmodified) into the corresponding oxime by an amino-specific cytochrome P450 mono-oxygenase (Wittstock and Halkier, 2000). The oxime is then converted to a thiohydroximate intermediate, followed by sequential Glc and sulfate transfer to complete the basic glucosinolate skeleton (Fig. 1; Halkier and Du, 1997). The initially formed glucosinolate can undergo a variety of subsequent transformations that modify the side chain. These side-chain modifications are specific for the precursor amino acid utilized in the formation of the glucosinolate. Figure 2 shows the proposed biosynthetic sequence for the modifications of the chain-elongated Met-derived glucosinolates, which are the major glucosinolates in Arabidopsis and many other Brassicaceae species.

View larger version (18K): [in this window] [in a new window]   Figure 2.   Side chain modifications of Met-derived glucosinolates in Arabidopsis. Potential side chain modifications for the elongated Met derivative, C4 dihomo-Met, are shown. Steps with natural variation identified in this study are shown in bold to the right or left of each enzymatic arrow.

The principal biological activities of glucosinolates are mediated by hydrolysis products formed when tissue disruption brings glucosinolates into contact with myrosinase, a thioglucosidase. After Glc cleavage, the resulting unstable aglycone generates numerous compounds (isothiocyanates, nitriles, epithiocyanates, and thiocyanates) with diverse biological activities. Myrosinase hydrolysis products can serve as oviposition and feeding stimulants for insects specialized on glucosinolate-containing plants, but act as toxins or feeding deterrents toward generalist insect herbivores (Giamoustaris and Mithen, 1995). Thus, any given glucosinolate may have positive or negative impacts on plant fitness depending upon the insect herbivores present. Previous research has shown that such heterogenous selection due to insect herbivory occurs on glucosinolate concentration in both Arabidopsis and other Brassicaceae (Mauricio and Rausher, 1997; Mauricio, 1998; Stowe, 1998a, 1998b). In the face of such heterogeneous selection pressures, it is not surprising that glucosinolates show extensive genetic variation within and among plant species (Rodman, 1980; Daxenbichler et al., 1991).

We analyzed glucosinolate profiles from the leaves and seeds of 39 different Arabidopsis ecotypes representing a diverse sample of the geographical and environmental range of this species. Extensive variation was found in both the composition and total concentration of glucosinolates in these Arabidopsis ecotypes. The structural variety can be explained by polymorphism at only five genetic loci, creating a modular system for generation of biosynthetic diversity that may be a response to heterogeneous natural selection.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Genetic Control of Natural Variation in Glucosinolate Profiles and Concentrations in Arabidopsis Leaves

To survey natural variation in the glucosinolates of Arabidopsis, we identified and quantified 22 different glucosinolates present in the leaves of 39 different ecotypes (Tables I and II; Hogge et al., 1988). Analysis of the presence or absence of specific glucosinolates allowed us to identify several genetic polymorphisms regulating the composition of leaf glucosinolates. Four loci have been described previously: GS-Elong controls production of glucosinolates with three carbon or four carbon side chains. GS-Alk is responsible for production of alkenyl glucosinolates. GS-OHP catalyzes production of 3-hydroxypropyl glucosinolates. Finally, GS-OH controls production of 2-hydroxy-3-butenyl glucosinolate (Figs. 1 and 2; Magrath et al., 1994; Mithen et al., 1995). In addition to these four loci, evidence for a previously unknown Arabidopsis locus was found in this collection of ecotypes. This locus, designated GS-OX, controls the conversion of methylthioalkyl to methylsulfinylalkyl glucosinolates (Fig. 2 and Table I; Giamoustaris and Mithen, 1996). Most ecotypes carry out this conversion efficiently and typically contain at least twice as much methylsulfinylalkyl as methylthioalkyl glucosinolate in the leaves. However, the Bla-10, Can-0, and Su-0 ecotypes all have higher concentrations of methylthioalkyl than methylsulfinylalkyl glucosinolates, indicating that they are impaired in this conversion and presumably contain a different GS-OX allele than the other ecotypes (Fig. 3A).

View this table: [in this window] [in a new window]   Table I.   Glucosinolates in the leaves of arabidopsis ecotypes Quantities are given in µmol g dry wt1 and are the mean of three extractions of each ecotype. The nos. at the top of each column refer to the list of glucosinolate in "Materials and Methods." The values underneath the totals section are as follows: Aliph, sum of aliphatic glucosinolates; Indole, sum of indolic glucosinolates; MT, sum of methylthio glucosinolates; MSO, sum of methylsulfinyl glucosinolates; AOP, sum of alkenyl and hydroxy aliphatic glucosinolates; C3, sum of three carbon aliphatic glucosinolates; C4, sum of four carbon aliphatic glucosinolates; C7, sum of seven carbon aliphatic glucosinolates; C8, sum of eight carbon aliphatic glucosinolates; and C4 per, C4/(C3 + C4), C8 per = C8/(C7 + C8).

View this table: [in this window] [in a new window]   Table II.   Glucosinolates in the seeds of Arabidopsis ecotypes Quantities are given in µmol g dry wt1 and are the mean of three extractions of each ecotype. The nos. at the top of each column refer to the list of glucosinolate in "Materials and Methods." The values underneath the totals section are as follows: Aliph, sum of aliphatic glucosinolates; Indole, sum of indolic glucosinolates; Benzyl, sum of benzyl glucosinolates; MT, sum of methylthioalkyl glucosinolates; MSO, sum of methylsulfinyl glucosinolates; AOP, sum of alkenyl and hydroxy aliphatic glucosinolates; Benzoxy, sum of benzoyloxy aliphatic glucosinolates; C3, sum of three carbon aliphatic glucosinolates; C4, sum of four carbon aliphatic glucosinolates; C7, sum of seven carbon aliphatic glucosinolates; C8, sum of eight carbon aliphatic glucosinolates; and C4 per, C4/(C3 + C4), C8 per = C8/(C7 + C8).

View larger version (36K): [in this window] [in a new window]   Figure 3.   Effect of GS-OX on conversion of methylthioalkyl to methylsulfinylalkyl glucosinolates. Bar diagrams show the ratio of methylthioalkyl (MT) glucosinolate to methylsulfinylalkyl (MSO) glucosinolate content in leaf (A) and seeds (B) for each ecotype analyzed.

In addition to the variation in glucosinolate profile, there were dramatic differences in the total concentration of the Met-derived glucosinolates in leaves of the 39 ecotypes. There was a nearly 20-fold range between the ecotype with the highest accumulation, Cvi, with about 37.81 µmol g dry weight¹, and the lowest accumulating ecotype, Pi-0, with 2.83 µmol g dry weight¹ (Table I). Comparing the accumulation of aliphatic glucosinolates in conjunction with the allelic status of the inferred genes indicated that only the GS-Alk and GS-Ohp loci (collectively referred to as Alkohp) had significant control over accumulation of glucosinolates in Arabidopsis (Fig. 4). These two loci are tightly linked and are responsible for converting methylsulfinylalkyl glucosinolates into alkenyl (GS-Alk) or hydroxyalkyl (GS-Ohp) glucosinolates. Ecotypes that accumulate significant levels of the precursor methylsulfinylalkyl glucosinolate are null for GS-Alk and GS-Ohp reactions, and therefore can be classified as containing a third AOP allele (Kliebenstein et al., 2001). Further, mapping in mul-tiple ecotypes suggested that all three variants are due to different alleles at a single genetic locus (Kliebenstein et al., 2001). Therefore, we contrasted the GS-Alk, GS-Ohp, and GS-AOP^null variants by ANOVA. Ecotypes with GS-Alk had aliphatic glucosinolates concentrations 2.5 times higher than ecotypes with GS-Ohp, and 4.0 times that of the GS-AOP^null (F = 25.9, df_factor = 2, df_error = 32, and P = 2.1 × 10⁷ by ANOVA). In addition, this locus or closely linked loci explained 61% of the variation among ecotypes for accumulation of leaf aliphatic glucosinolates. Previous reports have also shown that other enzymes involved in modifying glucosinolate chain structure can control glucosinolate accumulation (de Quiros, et al., 2000).

View larger version (24K): [in this window] [in a new window]   Figure 4.   GS-AOP regulates the accumulation of aliphatic glucosinolates in the leaves of Arabidopsis. The bars depict the average total aliphatic glucosinolate accumulation in leaves of 39 ecotypes. The ecotypes are classified based on the inferred genotype at three biosynthetic loci: GS-Elong, either C3- or C4-accumulating ecotypes; GS-AOP, alkenyl-, hydroxypropyl-, or methylsulfinylalkyl-containing ecotypes; and GS-OH, functional or nonfunctional alleles.

Genetic Control of Natural Variation in Glucosinolate Profiles and Concentrations in Arabidopsis Seeds

We analyzed the same collection of 39 ecotypes for accumulation of seed glucosinolates and identified 34 different glucosinolates (Table II). This included the 22 glucosinolates present in the leaves, plus benzoate esters of the three to eight carbon hydroxyalkyl and 2-hydroxy-3-butenyl glucosinolates (Hogge, et al., 1998). In addition to the polymorphisms previously described for the leaf tissue, we identified two new polymorphisms. In the ecotypes producing significant levels of four carbon aliphatic glucosinolates, there was variation for the accumulation of 4-hydroxybutyl glucosinolate. We named this new polymorphism GS-OHB.

We also identified a new allele of the GS-OH locus. The Kas and Sorbo ecotypes contain significant amounts of 2-hydroxy-3-butenyl glucosinolate in the seeds, but accumulate only minimal levels in the leaves, whereas Cvi produces no detectable amounts of this glucosinolate in either tissue. This suggests that there are three tissue-specific alleles of GS-OH: positive in leaf and seed, positive in seed only, and negative in all tissues (as displayed by Cvi). This is further supported by mapping data which indicates that all three phenotypic states of GS-OH are in fact due to alleles either at a single locus or closely linked loci (D.J. Kliebenstein, unpublished data).

Another difference in glucosinolate profiles between leaves and seeds is the ratio of methylthioalkyl to methylsulfinylalkyl glucosinolates. The leaves of most ecotypes typically contained at least twice the level of methylsulfinylalkyl as methylthioalkyl, except for three ecotypes (Bla-10, Can-0, and Su-0) that reversed this ratio (Fig. 3A). It is interesting that seeds of all ecotypes contained more methylthioalkyl than methylsulfinylalkyl, like the leaves of Bla-10, Can-0, and Su-0 (Fig. 3B).

Coordinate Control of Glucosinolate Accumulation

Although GS-AOP explained 61% of the variation in leaf aliphatic glucosinolates, this locus exhibited no significant association with concentration of seed aliphatic glucosinolates. This indicates that there may be different control mechanisms for the accumulation of glucosinolates in the two tissues. To test for coordinate regulation of seed and leaf glucosinolate accumulation, we compared the total accumulation of indolic or aliphatic glucosinolates between the two tissues. The levels of aliphatic glucosinolates in leaves and seeds of the different ecotypes showed a significant positive correlation (r = 0.35, P = 0.04, and n = 34; Fig. 5A). This suggests coordinate regulation of glucosinolate accumulation in these organs, which is presumably under the control of loci other than GS-AOP. One possible explanation is the transport of leaf glucosinolates into the seed. In contrast, the indolic glucosinolates showed no significant correlation between seed and leaf accumulation (r = 0.06, P = 0.70, and n = 34; Fig. 5B). Further, we observed no such negative correlations among the levels of the different glucosinolate classes (aliphatic, indolic, and aromatic) in either seeds or leaves (data not shown).

View larger version (11K): [in this window] [in a new window]   Figure 5.   Correlation of glucosinolate accumulation between the leaves and seeds. A, Scatter plot depicting the relationship between aliphatic glucosinolate accumulation in leaves (ALIPHL) and seeds (ALIPHS) of the ecotypes tested. The 90% confidence ellipse interval is drawn for reference. The values are in µmol g dry weight1. B, Scatter plot depicting the relationship between indolic glucosinolate accumulation in the leaves (INDOLEL) and seeds (INDOLES) of the ecotypes tested. The values are in µmol g dry weight1.

Genetics of Side-Chain Modification in Arabidopsis

We identified six aliphatic side chain modifications with apparent natural variation. These are: GS-OX (conversion of methylthioalkyl to methylsulfinylalkyl glucosinolate), GS-Alk (production of alkenyl glucosinolates), GS-Ohp (production of 3-hydroxypropyl glucosinolate), GS-null (absence of GS-Alk or GS-Ohp leading to the accumulation of methylsulfinylalkyl glucosinolates), GS-OHB (production of 4-hydroxybutyl glucosinolate), and GS-OH (production of 2-hydroxy-3-butenyl glucosinolate; Fig. 2). Previous analysis had shown that GS-Alk, GS-Ohp, and GS-AOP^null are allelic variants of a single genetic locus and that the three putative alleles of GS-OH are also all variants of a single locus (Parkin et al., 1994; Mithen et al., 1995; Giamoustaris, et al., 1996; Kliebenstein et al., 2001). We examined the newly described GS-OX polymorphism to see if the variation was caused by segregation at a single genetic locus. We used the ecotype, Wei-0, which has a GS-OX phenotype similar to Bla-10, Can-0, and Su-0 (data not shown). Wei-0 was crossed to Ler, which is able to efficiently convert methylthioalkyl to methylsulfinylalkyl glucosinolates. HPLC analysis of the F₁ progeny showed that GS-OX^Wei-0 is recessive to GS-OX^Ler (data not shown). HPLC analysis of 92 random F₂ progeny showed that 71 individuals had the Ler-like variant and 21 individuals had the Wei-0 phenotype. This is indistinguishable from a 3:1 segregation pattern and suggests that the GS-OX variation is due to segregation of a single genetic locus in this cross. The GS-OX locus mapped to the bottom of chromosome I, 3 cM teleomeric of the microsatellite AthGeneA and 20 cM centromeric from nga692 (Fig. 6). Further work is required to identify the gene responsible for this variation. In combination with the previously published data, this suggests that most of the natural variation in side-chain modifications in Arabidopsis can be explained by segregation of a small number of loci with large, discrete effects on glucosinolate profiles.

View larger version (11K): [in this window] [in a new window]   Figure 6.   Map of GS-OX on Chromosome I. Ninety-two F2 progeny were scored for the microsatellites indicated and for the GS-OX biochemical phenotype. The distance from the AthGeneA and nga692 markers to GS-OX is shown in cM to the right of the arrows. The approximate location of GS-OX is shown to the left of the chromosome.

Genetics of Chain Elongation in Arabidopsis

The elongation of the carbon side chain of the base amino acid is the first step in the biosynthesis of a majority of Arabidiopsis glucosinolates. Met-derived glucosinolates in Arabidopsis have side-chain lengths of three to eight carbon atoms (one to six additional carbon atoms) with the three- (C₃) and four- (C₄) carbon chain lengths being predominant (Hogge et al., 1988). Previous work showed that a single genetic locus (GS-Elong) determines the conversion of C₃ to C₄ aliphatic glucosinolates in Arabidopsis (Magrath et al., 1994). Map-based cloning studies have suggested that the gene responsible for this variation is a homolog of isopropylmalate synthase, which condenses acetyl-coenzyme A with an oxo acid (Fig. 1; de Quiros et al., 2000). It is interesting that plants with the C₃ GS-Elong allele also accumulate glucosinolates with side chains of seven (C₇) and eight carbon (C₈) atoms (Tables I and II). This is unexpected because the biosynthesis of C₇ and C₈ glucosinolates is assumed to require the full series of shorter-chain intermediates (C₃, C₄, C₅, and C₆), one of which could also give rise to C4 glucosinolates (Fig. 1). Two models could account for this phenomenon: (a) The C₃ GS-Elong allele does not block the chain elongation process, but resists diversion of metabolic flow at the C₄ stage, and so channels intermediates toward longer chain length products; or (b) the C₃ GS-Elong allele does block the chain elongation process prior to the formation of C₄ intermediates, requiring the C₇ and C₈ glucosinolates to be produced in a different pathway.

To differentiate between these two models, we compared the efficiency of conversion of C₃ with C₄ and C₇ with C₈ intermediates in glucosinolate biosynthesis. This was done by generating functions that measured the efficiency of each reaction. For the three- to four-carbon reaction, the function is C₄per = C₄/(C₃ + C₄). For the seven- to eight-carbon reaction, the function is C₈per = C₈/(C₇ + C₈). Both of these functions utilize the ratio of the accumulation of the precursor and product chain lengths as a gauge of reaction efficiency. As shown in Figure 7A, the efficiency of the C₃ to C₄ reaction is highly correlated between the leaf and seed tissues of the same ecotype (r = 0.98, P < 10⁸, and n = 34). The same is also true for the C₇ to C₈ conversion (r = 0.77, P < 10⁸, and n = 34; Fig. 7B). This implies that these reactions are under strict genetic control with minimal differences between the two tissues.

View larger version (22K): [in this window] [in a new window]   Figure 7.   Correlation of C7 to C8 and C3 to C4 intermediates in the biosynthesis of chain-elongated Met-derived glucosinolates. A, Scatter plot showing the correlation of the conversion of C3 to C4 in the seeds and leaves. B, Scatter plot showing the correlation of the conversion of C7 to C8 in the seeds and leaves. C, Scatter plot showing the correlation of the conversion of C3 to C4 with C7 to C8 in the leaves. D, Scatter plot showing the correlation of the conversion of C3 to C4 with C7 to C8 in the seeds.

Furthermore, the efficiency of C₃ to C₄ conversion showed a strong negative correlation with the C₇ to C₈ conversion efficiency (for leaf, r = 0.78, P < 10⁸, and n = 34; for seed, r = 0.92, P < 10⁸, and n = 34; Fig. 7, C and D). Ecotypes with higher C₃ ratios had lower C₇ ratios (Fig. 7C and D). Thus, ecotypes containing the GS-Elong C₃ allele are more efficient at elongating C₇ intermediates to C₈ intermediates than ecotypes containing the GS-Elong C₄ allele (Fig. 7, C and D). This correlation is independent of the total level of C₃, C₄, C₇, or C₈ glucosinolates. The involvement of the GS-Elong C₃ allele in altering the production of C₇ and C₈ glucosinolates supports the hypothesis that this allele does not block chain elongation after C₃ but instead prevents the diversion of metabolites at the C₄ chain length.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Modular Variation in Glucosinolate Profiles

One possible effect of heterogeneous natural selection on plant defense products is the rapid evolution of new compounds, new mixtures of compounds, or new patterns of gene regulation controlling compound accumulation. New compounds may increase resistance to herbivores that have become adapted to existing defenses, whereas mixtures may provide a unique complement of defenses, retarding counteradaptation of enemies. In this study, the 39 ecotypes sampled could be divided into seven different classes, each with its own unique mixture of leaf glucosinolates (Table III). These seven different classes are created by variation at only three of the five identified genetic loci, GS-Elong, GS-AOP, and GS-OH (Fig. 1 and 2 and Table III). If the GS-OX locus is also used to separate the glucosinolate profiles, Arabidopsis theoretically could have up to 14 different glucosinolate profiles. Thus, a small set of polymorphic loci in Arabidopsis generates a modular alteration in glucosinolate profile. This may enable rapid responses to changing selective pressures and could allow evolution in response to altered herbivore abundance.

Arabidopsis undoubtedly contains additional glucosinolates (and polymorphic loci controlling their formation) that remain to be discovered. Several minor peaks in our extracts were not identified. Furthermore, because only 39 ecotypes were analyzed and several glucosinolates were only found in a single accession (Tables I and II), much of the variation remains to be sampled. Given the modular nature of glucosinolate profiles in Arabidopsis, a single new locus could produce a whole series of novel glucosinolate chemotypes.

Large Variation in the Level of Glucosinolate Accumulation

In addition to altering the types of glucosinolates, heterogeneous natural selection might also favor a broad range of glucosinolate concentrations. In the leaves of the 39 ecotypes tested, there was a 20-fold difference in the total concentration of aliphatic glucosinolates (Table I and Fig. 3). More than 60% of this variation was due to the GS-AOP locus. However, even after accounting for the effect of GS-AOP, there is still a large amount of variation that is due to other factors. In the seeds of these ecotypes, the high and low ecotypes showed only a 3-fold range of aliphatic glucosinolate concentrations (Table II). This again suggests that independent factors regulate the accumulation of aliphatic glucosinolates in different tissues. Quantitative trait locus mapping of the loci influencing the accumulation of aliphatic glucosinolates in Arabidopsis will enable us to identify the factors regulating the difference between leaf and seed accumulation.

Aliphatic glucosinolate concentration was more variable in leaves than in seeds, whereas indolic glucosinolate variation showed the opposite pattern. The indolic glucosinolates exhibited only a 3-fold quantitative difference from high to low ecotype in the leaf, but had a 20-fold range from high to low ecotype in the seed (Tables I and II). This finding suggests that indolic glucosinolate accumulation in the leaves and seeds is under different modes of regulation. Further, this regulation is separate from that controlling the aliphatic glucosinolates. This could be a possible consequence of the different herbivore pressures on seeds and leaves and the fact that aliphatic and indole glucosinolates have different effects on herbivores (Bartlet et al., 1994).

Multiple Effects of GS-Elong on Aliphatic Glucosinolate Chain Elongation

Previous analysis had identified two different alleles of GS-Elong in Arabidopsis. One allele was associated with the accumulation of C₃ side-chain aliphatic glucosinolates and the other with C₄ side-chain glucosinolates (Fig. 1). This led to the hypothesis that ecotypes containing the C₃ allele are blocked in their ability to elongate the side chain past three carbons, and thus lack the potential to produce C₄ glucosinolates (Magrath et al., 1994). However, ecotypes with both alleles are able to produce the C₇ and C₈ side-chain glucosinolates, which theoretically require the C₄ intermediate. In our ecotype analysis, we found that the allelic status at GS-Elong controls the production of C₇ and C₈ glucosinolates in addition to C₃ and C₄ glucosinolates. Ecotypes with the C₃ allele contain a higher ratio of C₈ to C₇ glucosinolates than the C₄ allele (Fig. 7). This close correlation between C₃ and C₄ glucosinolates and C₇ and C₈ glucosinolates suggests that all of these glucosinolates are formed in the same pathway. This linkage could be explained alternatively by the biosynthesis of the different carbon length glucosinolates occurring via separate pathways employing some of the same enzymatic machinery or employing enzymes encoded by closely linked genes. It is interesting that the Kas and Cvi ecotypes are intermediate in all of the distributions examined (Fig. 7). This could be explained by either an intermediate GS-Elong allele or a modifying second locus. Further map-based cloning of this function will help to elucidate this difference.

Future Work

Our survey of the natural variation in glucosinolate composition of 39 ecotypes of Arabidopsis revealed variation in seven distinct side-chain modifications of aliphatic glucosinolates representing five polymorphic loci.We have already begun to use this variation to clone the GS-OX and GS-OH loci as well as to identify and clone quantitative trait loci controlling glucosinolate concentration. In addition, we are investigating the susceptibility of these 39 ecotypes to insects and pathogens to explore the importance of glucosinolates in plant defense. The study of natural variation in Arabidopsis provides a valuable set of tools for answering questions about the biosynthesis, evolution and function of these interesting natural products.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Growth

One hundred plants were grown in 3.25- × 3.25- × 2.25-inch pots at 18 pots to a flat for 3 weeks in a standard soil/vermiculite mixture. They were placed 10 inches from 4 × 60-W cool-white bulbs and 4 × 60-W wide-spectrum bulbs (GE, Fairfield, CT) in a 16-h-light, 8-h-dark photoperiod. At 3 weeks, the tissue was harvested and the leaf material was lyophilyzed. Plant growth continued for seed production, and seeds were harvested after 2 months. Ten milligrams of leaf material and 5 mg of seeds were used for the glucosinolate extraction. All samples and ecotypes were done in triplicate.

Ecotypes

This study analyzed the following ecotypes (with abbreviation, location, and Nottingham Stock Center no.): Aa-0, Rhon, Germany, N900; Ag-0, Argentat, France, N901; Bla-10, Bl-1, Bologna, Italy, N968; Bs-1, Basel, N996; Cal, Calver, UK, N1062; Can, Canary Islands, N1064; Cnt, Canterbury, UK, N1635; Col-0, Columbia, MO, N1092; Cvi, Cape Verde Islands, N1096; Di-1, Dijon, France, N1108; Di-g, Dijon, France, N910; Ei-2, Eifel, Germany, N1124; Ema-1, East Malling, UK, N1637; Hodja, Khurmatov, Tadjikistan, N922; Ita-0, Ibel Tazekka, Morocco, N1244; Ka-0, Karnten, Austria, N1266; Kas, Kasmir, India, N903; Kil-0, Killean, UK, N1270; Kondara, Khurmatov, Tadjikistan, N916; Ler, Landsberg, Germany, N1642; Lip-0, Lipowiec, Poland, N1336; Ma-0, Marburg, Germany, N1356; M_r-0, Monte Tosso, Italy, N1372; Mrk-0, Markt, Germany, N1374; Mt, Martuba, Libya, N1380; No, N1394; Oy-0, Oystese, Norway, N1643; Petergof, Russia, N926; Pi-0, Pitztal, Austria, N1456; Pog-0, Point Gray, BC, N1476; Rsch-0, Rschew, Russia, N1490, Sei-0, Seis am Schlern, Italy, N1504; Su-0, Southport, UK, N1540;Tsu-1, Tsu, Japan, N1640; Wl-0, Wildbad, Germany, N1630; and Yo-0, Yosemite, CA, N1622.

96-Well Glucosinolate Extraction and Purification

This purification technique follows the basic sephadex/sulfatase Arabidopsis protocol previously described (Hogge et al., 1988). Samples were harvested into deep-well micro-titer tubes (either 10 fresh leaf discs frozen in liquid nitrogen, 10 mg freeze-dried leaf material, or 5 mg dried seeds). Four 2.3-mm ball bearings were added and the samples ground into a fine powder in a paint shaker by high speed agitation. To extract glucosinolates, 400 µL of methanol, 10 µL of 0.3 M lead acetate, and either 120 µL of water for seeds and freeze-dried material or 80 µL of water for fresh tissue was added. The samples were mixed for 1 min in the paint shaker and allowed to incubate for 60 min at 180 rpm on a rotary shaker. The tissue and protein were pelleted by centrifuging for 10 min at 2,500g and the supernatant used for anion-exchange chromatography. Ninety-six well filter plates (Millipore, Tempe, AZ, catalogue no. MAHVN4550) were loaded with 45 µL of diethylaminoethyl Sephadex A-25 using the Millipore multiscreen column loader. Then 300 µL of water was added and allowed to equilibrate for 2 to 4 h. After water was removed with 2 to 4 s of vacuum on the vacuum manifold (Qiagen, Valencia, CA), 150 µL of the supernatant was added to the 96-well columns and the liquid removed by 2 to 4 s of vacuum. This step was repeated once to bring the total volume of plant extract to 300 µL. The columns were washed four times with 150 µL of 67% (v/v) methanol, three times with 150 µL of water, and three times with 150 µL of 1 M sodium acetate. To desulfate glucosinolates on the column, 10 µL of water and 10 µL of sulfatase solution was added to each column and the plates incubated overnight at room temperature (Hogge et al., 1988). Desulfo-glucosinolates were eluted by placing a deep-well 2-mL 96-well plate in the bottom of the 96-well vacuum manifold and aligning the 96-well column plate. The DEAE Sephadex was then washed twice with 100 µL of 60% (v/v) methanol and twice with 100 µL of water.

HPLC

Forty microliters of the glucosinolate extract was run on a 5-µm column (Lichrocart 250-4 RP18e, Hewlett-Packard, Waldbronn, Germany) on a Hewlett-Packard 1100 series HPLC. Compounds were detected at 229 nm and separated utilizing either of the two following programs with aqueous acetonitrile. For seeds, the program was an 8-min gradient from 1.5% to 5.0% (v/v) acetonitrile, a 2-min gradient from 5% to 7% (v/v) acetonitrile, a 32-min gradient from 7% to 52% (v/v) acetonitrile, a 2-min gradient from 52% to 92% (v/v) acetonitrile, 5 min at 92% (v/v) acetonitrile, a 3-min gradient from 92% to 1.5% (v/v) acetonitrile, and a final 8 min at 1.5% (v/v) acetonitrile. For leaf material, the program was a 6-min gradient from 1.5% to 5.0% (v/v) acetonitrile, a 2-min gradient from 5% to 7% (v/v) acetonitrile, a 7-min gradient from 7% to 25% (v/v) acetonitrile, a 2-min gradient from 25% to 92% (v/v) acetonitrile, 6 min at 92% (v/v) acetonitrile, a 1-min gradient from 92% to 1.5% (v/v) acetonitrile, and a final 5 min at 1.5% (v/v) acetonitrile.

Glucosinolate Identification and Quantification

All glucosinolates had been previously isolated and identified Arabidopsis (P. Brown and T. Gershenzon, unpublished data; Hogge et al., 1988). The identity of HPLC peaks was based on a comparison of retention time and UV absorption spectrum as deterimined on a diode-array detector with those of standards. Results are given as µmol g dry weight¹ tissue calculated from HPLC peak areas using response factors computed for pure de-sulfo glucosinolate standards at A229 nm (P. Brown and T. Gershenzon, unpublished data). Each ecotype was run in triplicate. The list of compound identities is as follows: 1, 3-hydroxypropyl; 2, 3- methylsulfinylpropyl; 3, 4-hydroxybutyl; 4, 2(S)-hydroxy-3-butenyl; 5, 4-methylsulfinylbutyl; 6, 2(R)-hydroxy-3-butenyl; 7, allyl; 8, 2-methylthioethyl; 9, 5-methylsulfinylpentyl; 10, 2-hydroxy-4-pentenyl; 11, 3-butenyl; 12, 1-methylethyl; 13, 6-methylsulfinylhexyl; 14, 3-methylthiopropyl; 15, 4-hydroxy-indolyl-3-methyl; 16, 7-methylsulfinylheptyl; 17, 2-methylpropyl; 18, 4-pentenyl; 19, 4-methylthiobutyl; 20, 8-methylsulfinyloctyl; 21, indolyl-3-methyl; 22, 4-methoxy-indolyl-3-methyl; 23, 5-methylthiopentyl; 24, 6-methylthiohexyl; 25, benzyl; 26, 1-methoxy-indolyl-3-methyl; 27, 3-benzoyloxypropyl; 28, 2-benzoyloxy-3-butenyl; 29, 4-benzoyloxybutyl; 30, 7-methylthioheptyl; 31, 5-benzoyloxypentyl; 32, 8-methylthiooctyl; 33, 6-benzoyloxyhexyl; and 34, 8-benzoyloxyoctyl.

Statistics

Means are given for each ecotype. Systat (version 7) was utilized for statistical analysis. ANOVA utilized for GS-AOP is as follows: leaf aliphatic glucosinolate = constant + GS-AOP.

Mapping and Microsatellites

DNA was isolated in a 96-well format as previously described (Kliebenstein et al., 2001). Five microliters of the resuspended DNA was added to 20 µL of PCR reaction mixture (2.5 mM MgCl₂, 200 pM primers, and 0.5 units TAQ polymerase) containing primers for the microsatellite markers listed. Microsatellite primer sequences were obtained from The Arabidopsis Information Resource (www.Arabidopsis.org). The reactions were run with the following cycle program: 95°C for 3 min; 40 cycles of 95°C at 20 s, 56°C at 20 s, and 72°C for 1 s; and 72°C for 3 min and 4°C final). The polymorphisms were scored on 4% (w/v) agarose. The allelic state at the GS-OX locus was scored by analyzing the production of 3-methylthiopropyl glucosinolate by HPLC as described above. Mapping of the markers and GS-OX utilized Mapmaker (version 3; Lander et al., 1987).