The study of glucosinolates and their regulation has provided a powerful framework for the exploration of fundamental questions about the function, evolution, and ecological significance of plant natural products, but uncertainties about their metabolism remain. Previous work has identified one thiohydroximate S-glucosyltransferase, UGT74B1, with an important role in the core pathway, but also made clear that this enzyme functions redundantly and cannot be the sole UDP-glucose dependent glucosyltransferase (UGT) in glucosinolate synthesis. Here, we present the results of a nearly comprehensive in vitro activity screen of recombinant Arabidopsis Family 1 UGTs, which implicate other members of the UGT74 clade as candidate glucosinolate biosynthetic enzymes. Systematic genetic analysis of this clade indicates that UGT74C1 plays a special role in the synthesis of aliphatic glucosinolates, a conclusion strongly supported by phylogenetic and gene expression analyses. Finally, the ability of UGT74C1 to complement phenotypes and chemotypes of the ugt74b1-2 knockout mutant and to express thiohydroximate UGT activity in planta provides conclusive evidence for UGT74C1 being an accessory enzyme in glucosinolate biosynthesis with a potential function during plant adaptation to environmental challenge.
Secondary metabolism is a dominant factor that mediates evolutionary and ecological relationships between heterotrophs and autotrophs, and serves as the ultimate source of numerous drugs used in medicine, facts that explain why study in this area has been and continues to be of central importance in biology. Glucosinolates are a class of sulfur-rich metabolites almost exclusive to the Brassicales (Agerbirk and Olsen, 2012). Because of their importance for human health (Dinkova-Kostova and Kostov, 2012), plant defense against herbivores and innate immunity (Lipka et al., 2010; Mithöfer and Boland, 2012), glucosinolates have been the focus of major research efforts facilitated by the powerful genetic resources and molecular tools available for the reference plant, Arabidopsis thaliana (Sønderby et al., 2010).
Glucosinolate biosynthesis comprises up to three stages. First, the parental amino acid may undergo one or more rounds of chain elongation via a series of reactions analogous to those converting valine to leucine. The distribution of chain-length in the final products is controlled by genotype as well genotype × environment interactions (Burow et al., 2010). Next, the (homo)amino acids enter the core pathway common to all glucosinolates, which can be interpreted as an initial activation, wherein the (homo)amino acid is converted to a highly reactive nitrile oxide. The subsequent series of detoxification steps generates the stable glucosinolates, which are N-sulfated S-glucosyl thiohydroximates (Figure 1). Finally, the structural diversity of these compounds can be further increased by side-chain modifications, including oxygenation, hydroxylation, dehydrogenation, benzoylation and methoxylation, among others (Grubb and Abel, 2006; Halkier and Gershenzon, 2006; Sønderby et al., 2010). In Arabidopsis, the major glucosinolates are aliphatic, derived from methionine (Met,) or indolic derived from tryptophan (Trp); again, the distribution of final products is under genetic and environmental control (Burow et al., 2010).
Glucosylation of the thiohydroxamic acid intermediate is the penultimate reaction of the core pathway. We previously showed that Arabidopsis UGT74B1, a UDP-glucose-dependent glucosyltransferase (UGT) of plant Family 1 glycosyltransferases (GTs), is the major enzyme catalyzing this reaction in vivo (Grubb et al., 2004). Nevertheless, mutant plants lacking functional UGT74B1 still accumulate substantial levels of glucosinolates, indicating that at least one additional enzyme is capable of thiohydroximate glucosylation. Several reports of co-regulation of the closely related UGT74C1 with glucosinolate pathway-related genes, particularly with those involved in aliphatic glucosinolate synthesis, have predicted UGT74C1 as a candidate (Gachon et al., 2005; Hirai et al., 2007; Mikkelsen et al., 2010). However, there is as yet no direct evidence for its in vivo function. Arabidopsis thaliana contains 107 members of Family 1 UGTs, which share the highly conserved plant-specific UGT signature motif and glucosylate a large array of small molecules (Bowles et al., 2006; Osmani et al., 2009). We therefore set out to systematically search for and identify the additional UGT(s) that function(s) in glucosinolate synthesis. Our results from complementary biochemical and genetic approaches clearly demonstrate that UGT74C1 acts as an accessory UGT in the glucosinolate core pathway.
Screening recombinant Arabidopsis family 1 UGT enzymes
We expressed almost all A. thaliana Family 1 UGTs as recombinant C-terminal GST-fusions in E. coli, which we affinity-purified (Lim et al., 1998, 2002) and tested for activity against phenylacetothiohydroximate (PATH) as described previously for UGT74B1 (Grubb et al., 2004). As expected, amongst the 93 recombinant enzymes assayed (Table S1), the UGT74B1–GST (glutathione S-transferase) protein displayed the highest activity toward PATH under the three pH conditions tested (pH 6, pH 7 and pH 8). Interestingly, all analysed members of clade UGT74 showed activity toward PATH (Figure 2), ranging from approximately 50% (UGT74F1) to 5% (UGT74E2) relative to UGT74B1, with the exception of UGT74E1, which was not expressed. Minor activities (2–8%) were recorded for recombinant UGT enzymes 88A1, 75B1, 71B6, 73B1, and 73B3. The remaining 81 UGT–GST proteins expressed only trace (<2%) or no detectable activity toward this substrate (Table S1). Thus, we focused our analysis on members of the UGT74 clade as the most likely candidates for additional thiohydroximate UGTs in Arabidopsis.
Activity of recombinant UGT74 enzymes toward thiohydroximates
We used two thiohydroximate substrates to characterize the activity of affinity-purified (His)6–UGT74 proteins (Figure S1): 4-methylthiobutylthiohydroximate (MTBTH) and 1-methylindolyl-3-acetothiohydroximate (MIATH). The first compound is the natural precursor of the aliphatic 4-methylsulfinylbutyl glucosinolate (S4), the most abundant glucosinolate in leaves of Col-0, while the latter is a quasi-natural product and an analogue of the thiohydroximate precursor of all indolic glucosinolates, which can be incorporated into 1-methylindolyl-3-methyl glucosinolate (Pedras et al., 2009). Unfortunately, we could not further evaluate our lead candidate, UGT74C1, because bacterial preparations of the enzyme never consistently showed activity above background, despite extensive exploration of variations in culture conditions, purification protocols, metal co-factors, sugar-nucleotide donor substrates, and reaction buffer composition. The UGT74C1–GST construct of the initial screen expressed very low activity in subsequent experiments, and successful transient expression of (His)6–UGT74C1 in tobacco leaves (Nicotiana benthamiana) resulted in inactive preparations, while the corresponding UGT74B1 control was highly active in this system, results we cannot explain at present.
The remaining five purified (His)6–UGT74 enzymes showed significant activity above background and establishment of pseudo-single substrate conditions allowed the estimation of their kinetic parameters (Table 1). UGT74B1 was by far the most active (kcat > 30) and showed the highest affinity (Km< 20 μm) for both substrates, while the other four enzymes displayed low activities (kcat < 0.2) and affinities (Km > 100 μm). Nevertheless, a relatively inefficient enzyme could provide substantial activity if expressed at sufficiently high level. We therefore compared UGT74 gene expression in wild-type and ugt74b1 knockout plants to test if any clade member is overexpressed as a consequence of the metabolic block in ugt74b1. We did not observe significantly higher UGT74 mRNA levels, indicating that compensatory changes in UGT74 expression are not a likely explanation for the residual glucosinolate content of ugt74b1 plants (Figure S2a). Thus, while we could confirm that UGT74 members are particularly apt in thiohydroximate glucosylation relative to the other 86 UGTs tested, no particular candidate for a new glucosinolate biosynthetic enzyme emerged from this in vitro approach.
Table 1. Kinetic parameters of recombinant UGT74 proteins toward MIATH and MTBTH
Km app (μm)
Km app (μm)
The mean of three independent determinations is given ± standard deviation (SD); n.d. – activity not detectable.
36.5 ± 26.2
17.2 ± 7.1
2.28 ± 0.79
1.48 ± 0.55
0.11 ± 0.06
158 ± 90
0.0303 ± 0.0037
462 ± 67
0.009 ± 0.010
232 ± 136
0.0164 ± 0.0019
159 ± 36
0.182 ± 0.028
235 ± 39
0.0267 ± 0.0046
110 ± 39
0.021 ± 0.012
130 ± 161
Mutational analysis of clade UGT74 implicates UGT74C1 in glucosinolate synthesis
We next examined phenotypes caused by loss-of-function alleles of all UGT74 genes. Figure 3(a) illustrates the positions of insertional and point mutations. Analysis of mRNA levels in the wild-type and mutant lines by quantitative RT-PCR confirmed that all ugt74 lines are likely functional knockouts (Figure S2b). Initial characterization of the single mutants revealed, with the exception of ugt74b1 lines, no obvious morphological differences from the wild-type. Slightly elevated accumulation of aliphatic glucosinolates was measured for ugt74e2-1, inconsistent with the idea that this gene plays a direct role in thiohydroximate metabolism (Figure 4a). Thus, we concluded that if any UGT74 member plays such a role, it may be masked by the activity of UGT74B1. We therefore generated double homozygous mutants of each line with either ugt74b1-1 (Ws-0) in the case of ugt74f1 or with ugt74b1-2 (Col-0) for the remaining ugt74 mutants.
We previously characterized both ugt74b1 alleles, which cause dwarfism, partial sterility, and a low-glucosinolate chemotype (Grubb et al., 2004). Both wild-type accessions and their corresponding single ugt74b1 mutants are shown in Figure 3(b). All double mutants closely resemble ugt74b1, with one dramatic exception: the ugt74b1-2 ugt74c1-2 double mutant shows a strongly additive phenotype, producing extremely small plants with only a few small leaves (Figure 3b,c). While these plants persist for a few weeks on soil and eventually produce floral structures, they are completely sterile. The glucosinolate chemotype of ugt74b1-2 ugt74c1-2 plants also stands in stark contrast to those of the other double mutants (Figure 4b). While indolic glucosinolates are not affected relative to ugt74b1-2, the content of aliphatic glucosinolates is reduced an additional 70%, again implicating UGT74C1 specifically in the synthesis of Met-derived glucosinolates. Among the other double mutants, ugt74b1-2 ugt74d1 shows a approximately 2-fold increase in indolic glucosinolates. Although interesting, this result is inconsistent with direct a role of UGT74D1 in thiohydroximate metabolism and was not pursued further.
Loss of UGT74C1 in ugt74b1 plants causes seedling lethality
We maintained the ugt74b1-2 (+/−) ugt74c1-2 (−/−) line, which is fertile and indistinguishable from the wild-type, as a source of double homozygous mutant seedlings. We reasoned that their poor growth on soil might be caused by interactions with hostile organisms and therefore studied plant growth under axenic conditions. We included the sur1 line for comparison because loss of the SUR1 C-S lyase catalyzing the reaction prior to the glucosylation step causes failure of glucosinolate synthesis, overproduction of auxin, and seedling lethality (Boerjan et al., 1995; Mikkelsen et al., 2004). When grown on agar, ugt74b1-2 ugt74c1-2 plants are significantly smaller than ugt74b1-2 and similar in size to sur1 seedlings (Figure 5). The double mutant is extremely stunted and produces very small, curly and chlorotic leaves. Unlike sur1 seedlings, it occasionally develops a short inflorescence with petite sterile flowers. The double mutant maintains a plant-like habit with recognizable organs after an extended period of growth, while sur1 seedlings completely degenerate into callus tissue (Figure 5d).
Loss of UGT74C1 in ugt74b1 plants does not mimic the high-auxin phenotype of sur1
We reported previously that loss of UGT74B1 causes morphological alterations that are consistent with auxin overproduction, which we confirmed by indole-3-acetic acid (IAA) measurements (Grubb et al., 2004). Interestingly, although ugt74b1-2 ugt74c1-2 seedlings are significantly smaller than ugt74b1-2 and more similar to sur1 plants during early stages of development, they do not share the extreme high-auxin phenotype of sur1. The root system of the double mutant, albeit significantly underdeveloped relative to wild-type and ugt74c1-2 seedlings, is comparable with the root system of ugt74b1-2 plants, but clearly differs from that of the sur1 mutant (Figure 5b). In contrast with sur1 plants, ugt74b1-2 ugt74c1-2 seedlings produce excessive roots and callus-like protrusions at the hypocotyl-root junction only after extended growth on agar (>1 month). When grown under conditions to promote adventitious rooting (Grubb et al., 2004), the double mutant is again more similar to ugt74b1-2 than to sur1 seedlings (Figure S3). When germinated in the dark, ugt74b1-2 seedlings display a moderately de-etiolated phenotype, which is not exaggerated in the ugt74b1-2 ugt74c1-2 double mutant (Figure 5c). Although these data suggest that loss of UGT74C1 activity does not further elevate auxin production in the ugt74b1-2 background, we measured significantly elevated auxin levels relative to wild-type and ugt74b1-2. However, precursors of jasmonic acid and ethylene are also highly increased, suggesting general perturbation of hormone homeostasis in the extremely dwarfed double mutant (Table S2).
Loss of UGT74C1 in ugt74b1 plants affects primarily aliphatic glucosinolates
We determined glucosinolate composition in shoots of 3-week-old wild-type, ugt74c1-2, ugt74b1-2, ugt74b1-2 ugt74c1-2, and sur1 plants (Figure 6). Similar to the ugt74b1-1 allele (Grubb et al., 2004), the ugt74b1-2 mutation decreases total glucosinolates by about 30% due to lower content of both aliphatic and indolic derivatives. Interestingly, while the glucosinolate chemotype of the ugt74c1-2 single mutant is barely distinguishable from the wild-type, levels of all major Met-derived glucosinolates are dramatically reduced in the ugt74b1-2 ugt74c1-2 double mutant. The reduction of total aliphatic glucosinolates by about 80% in ugt74b1-2 ugt74c1-2 plants clearly exceeds the effect of the ugt74b1-2 mutation (30%). As in the single ugt74c1-2 mutant, loss of UGT74C1 in the ugt74b1-2 background differentially affects aliphatic glucosinolates depending on the length of the side chain (Figure 6b). While the content of 3-methylsulfinylpropyl glucosinolate (S3) is reduced by about half, 8-methylsulfinyloctyl glucosinolate (S8) is nearly abolished, with the S4 compound exhibiting an intermediate reduction. Thus, UGT74C1 seems to prefer aliphatic substrates with longer side-chains in vivo.
In contrast, content of Trp-derived glucosinolates is quite similar between the ugt74b1-2 and ugt74b1-2 ugt74c1-2 lines (Figure 6c). While the level of indolyl-3-methyl glucosinolate (IM) is lower by about 50% in both lines relative to wild-type, content of 4-methoxyindolyl-3-methyl glucosinolate (4IM) is largely unaffected and maintained at wild-type level. The content of 1-methoxyindolyl-3-methyl glucosinolate (1IM) is reduced significantly in the double mutant relative to ugt74b1-2 plants. However, the absolute magnitude of this change is miniscule compared with the change in aliphatic glucosinolates.
Because the genetic evidence implicates UGT74C1 in glucosinolate synthesis, we searched in more detail for a chemotype in the ugt74c1-2 single mutant. To this end, we examined the glucosinolate profiles of cauline leaves, flowers, fully expanded green siliques and seeds; none of these profiles revealed any significant differences compared with the wild-type (Figure S4). Therefore, the primary role of UGT74C1 in glucosinolate synthesis seems to be unrelated to the developmental stage of adult plants.
In summary, the genetic analysis points to a role of UGT74C1 in glucosinolate biosynthesis that is limited to its aliphatic branch. This conclusion is consistent with the phenotypic data that do not indicate enhancement of the high-auxin phenotype of ugt74b1-2 plants by loss of UGT74C1, as would be expected if UGT74C1 also glucosylates indolic thiohydroximates. Consistent with previous reports (Mikkelsen et al., 2004), glucosinolate production in sur1 plants is reduced severely (<3% of wild-type; Figure 6). As the ugt74b1-2 ugt74c1-2 double mutant accumulates total glucosinolates to about 25% of the level of the wild-type, it is possible that additional UGTs exist with roles in glucosinolate biosynthesis.
UGT74C1 overexpression complements all ugt74b1 phenotypes
The inactivity of recombinant UGT74C1 toward thiohydroximates in vitro and the lack of a glucosinolate chemotype in the single ugt74c1-2 mutant prompted us to test the biosynthetic capabilities of UGT74C1 by its overexpression in the ugt74b1-2 mutant, a genotype clearly deficient in thiohydroximate metabolism (Grubb et al., 2004; Figure 6). We generated transgenic ugt74b1-2 lines harboring UGT74C1 or UGT74B1 under the control of the Cauliflower Mosaic Virus 35S promoter. Overexpression of either enzyme in the wild-type did not obviously alter plant morphology (Figure 7a). As expected, overexpression of UGT74B1 in ugt74b1-2 plants complemented all morphological phenotypes as well as the glucosinolate chemotype (Figure 7a; Grubb et al., 2004). Seven of the eight isolated 35S::UGT74C1 (ugt74b1-2) lines showed mildly depressed, unchanged or modestly elevated UGT74C1 mRNA levels relative to the wild-type (Figure S5c–i). Interestingly, these are morphologically indistinguishable from ugt74b1-2 ugt74c1-2 plants (Figure 7a, line 2.10.8) and show the characteristic low-glucosinolate profile. Thus, these lines recapitulate all phenotypes of ugt74b1 plants that lack UGT74C1, possibly as a consequence of transgene-induced gene silencing (Figure S5b).
One 35S::UGT74C1 (ugt74b1-2) line (6.17.6) expressed nearly 200-fold higher UGT74C1 mRNA levels relative to the wild-type (Figure S5a), and almost entirely complements all phenotypes of ugt74b1-2 plants. Aside from a slight, vascular-associated chlorosis and a very modest reduction in size, this line is morphologically indistinguishable from the wild-type (Figure 7a). Moreover, the low-glucosinolate chemotype of ugt74b1-2 is largely alleviated, including the deficit in indolic glucosinolates (Figure 7b). Thus, while these results strongly imply that UGT74C1 is a thiohydroximate UGT, they also imply that its apparent specificity for aliphatic glucosinolates is a function of its expression pattern, rather than its inherent biochemical properties.
UGT74C1 shows thiohydroximate UGT activity in vitro
We next sought to directly verify that the 35S::UGT74C1 transgene expresses a thiohydroximate UGT in planta by comparing the PATH glucosylating activity of protein extracts prepared from ugt74b1-2 plants and ugt74b1-2 lines that overexpressed UGT74C1 or UGT74B1, which allows measurement of UGT74C1 activity in the absence of endogeneous UGT74B1. Total protein extract of the UGT74C1 overexpression line 6.17.6 exhibited five-fold higher specific activity toward PATH, very similar to results obtained for extracts of the respective 35S::UGT74B1 control line (Figure 7c). The presence of a N-terminal (His)6-tag facilitates purification of UGT74C1. While we were unable to recover thiohydroximate UGT activity from the affinity column, the level of activity in the flow-through was similar to the background (Figure 7c), indicating that the significantly higher activity in the crude extract is derived from the presence of (His)6–UGT74C1.
Finally, we determined the apparent affinity of the crude protein extract for PATH in pseudo-first order kinetics experiments with saturating UDP-glucose and varied PATH concentrations (Figure 7d). The activity of extracts prepared from ugt74b1-2 plants increases in a linear manner with PATH concentration, showing little sign of saturation, indicating activity of (a) low affinity protein(s). In contrast, the thiohydroximate UGT activity of extracts of the UGT74C1 overexpression line is clearly saturable in the same concentration range, with an apparent Km of 509 ± 250 μm. In conclusion, the 35S::UGT74C1 transgene complements ugt74b1-2 phenotypes because it encodes a functional thiohydroximate UGT.
Expression patterns of UGT74C1 and UGT74B1 overlap during plant development
We next compared tissue-specific patterns of UGT74C1- and UGT74B1-promoter directed GUS expression (Grubb et al., 2004). In our search for loss-of function ugt74c1 alleles, we identified line GT1305 in which the bacterial β-glucuronidase (GUS) gene uidA under the control of a minimal promoter is inserted (in the same orientation) 25 bp downstream of the predicted start codon of UGT74C1 (Figure 3a). Genetic analysis of this line is consistent with a single gene that confers kanamycin resistance and GUS expression. Thus, GT1305 (ugt74c1-1) plants, which are morphologically indistinguishable from the wild-type (Ler) and express nearly wild-type UGT74C1 transcript levels, likely report authentic UGT74C1 promoter activity (Figure 8). Histochemical analysis of ugt74c1-1 plants revealed a strong association of GUS expression with the shoot apical meristem (Figure 8 panel Ia), the plant vasculature in every organ examined, including the placenta and funiculus (Figure 8 panel Ia–g), and all tissue junctions, including the transition between different organs such as the shoot-root junction (Figure 8 panel IIe), the receptacle of the flower and silique (panel Ie, panel IIq), as well as branch points within organs (Figure 8 panel IId,o). GUS staining is not detectable in root tips and immature vascular tissues (Figure 8 panel Id, panel IIa–d) nor in floral organs (Figure 8 panel Ie, panel IIp). During the transition from vegetative to reproductive development, GUS expression gradually diminishes in rosette and cauline leaves (panel IIm) but is clearly detectable in developing inflorescences (Figure 8 panel IIo,q), flower stalks (Figure 8 panel Ie), and siliques (Figure 8 panel If,g, panel IIr). While both UGT74 promoters are relatively inactive in cauline leaves (Figure 8 panel IIm,s), both respond to wounding of these organs (compare Figure 8 panel IIm,s with n,f). It is interesting to note that the observed pattern of UGT74C1 promoter activity during plant development is strikingly similar to the expression domains reported for several genes encoding enzymes and regulatory proteins involved in glucosinolate synthesis (Reintanz et al., 2001; Tantikanjana et al., 2001; Grubb et al., 2004; Levy et al., 2005; Gigolashvili et al., 2007a,b, 2008). With a few exceptions, this similarity is particularly striking between UGT74C1 and UGT74B1 (Figure 8, panel II). However, in contrast with UGT74B1, promoter activity of UGT74C1 is not detectable in the root meristems (compare Figure 8a,c and g,i), developing root vasculature (compare Figure 8b,d and h,j), and in the anthers of mature flowers (compare Figure 8p and v).
UGT74 phylogeny is consistent with taxonomic glucosinolate distribution
Because the glucosinolate pathway arose in the Brassicales after their split from the rest of the rosids, we expected glucosinolate-related genes to be specific to that order. We therefore constructed a phylogeny of UGT74 proteins along with related sequences from flowering plants (Figures 9 and S6 and Table S7). Surprisingly, UGT74B1 resides in an ancient clade that includes UGTs from non-glucosinolate producing eudicots; however, within the eudicot-specific clade in which it resides, it occurs on a long branch rooted by a sequence from Carica papaya, consistent with a relatively early divergence from related eudicot genes and/or with neofunctionalization related to its new role. Interestingly, UGT74C1 similarly occurs on a long branch, but within a Brassica-specific clade. Given that Met-derived glucosinolates are a relatively modern evolutionary innovation (Mithen et al., 2010), this finding is consistent with the idea that UGT74C1 evolved to play a special role in the synthesis of these compounds. The absence of an obvious homolog in C. papaya, a species which does not synthesize aliphatic glucosinolates, lends additional support to this notion.
This study describes the results of a systematic search for additional UGTs that may act in glucosinolate biosynthesis. The unbiased search was motivated by the fact that plants that lack UGT74B1, which we previously demonstrated to function in the core pathway, nevertheless accumulate substantial quantities of glucosinolates (approximately 50% of the wild-type), indicating the existence of at least one additional enzyme that is capable of glucosylating thiohydroximate intermediates (Grubb et al., 2004).
UGT74C1 functions in (aliphatic) glucosinolate biosynthesis
The cumulative evidence strongly supports a role for UGT74C1 in glucosinolate biosynthesis that is largely limited to its aliphatic branch. The activity screen of a nearly complete set of Arabidopsis Family 1 UGTs clearly implicated the UGT74 clade as a group of enzymes with a propensity toward thiohydroximate glucosylation. While a recent report highlights the risks of assigning UGT functions based on in vitro activity screens (Morant et al., 2010), and we initially failed to characterize recombinant UGT74C1 activity in vitro, we obtained conclusive evidence that UGT74C1 is a thiohydroximate UGT from the genetic approach. The additive chemotype of further decreased aliphatic glucosinolate accumulation in the ugt74b1-2 ugt74c1-2 double mutant strongly argues for a direct involvement of UGT74C1 in Met-derived glucosinolate production (Figure 6). Although loss of UGT74C1 substantially augments the growth defects of ugt74b1-2 plants, resulting in extreme dwarfism and infertility of the double mutant, it does not exacerbate the high-auxin phenotypes observed for the ugt74b1-2 single mutant (Figures 5 and S3), again consistent with a primary role of UGT74C1 in the aliphatic branch.
Conversely, ectopic overexpression of UGT74C1 in ugt74b1-2 alleviates its morphological aberrations and low-glucosinolate chemotype, indicating that UGT74B1 and UGT74C1 are both thiohydroximate UGTs (Figure 7a,b). We obtained direct evidence for such a biochemical role of UGT74C1 by partial characterization of the thiohydroximate UGT activity expressed from a 35S::UGT74C1 transgene in ugt74b1-2 plant extracts (Figure 7c,d). Our unsuccessful attempts to detect thiohydroximate activity during transient expression assays in N. benthamiana raises the tantalizing prospect that UGT74C1 requires one or more protein partners for activity, possibly as part of a multienzyme complex, which are naturally not present in tobacco.
The availability of a gene trap line reporting UGT74C1 promoter activity allowed comparison of tissue-specific GUS expression patterns with the previously described UGT74B1::GUS line (Grubb et al., 2004). The expression domains of both UGT74::GUS genes coincide well, with the notable exceptions of the root meristem, the root elongation zone, anthers and cotyledons, where UGT74C1 is expressed to a lesser degree, or not at all (Figure 8). Low or undetectable UGT74C1::GUS expression in these tissues, which are important sites of auxin synthesis and action (Zhao et al., 2002; Ljung et al., 2005; Zhao, 2010), further supports a preferential role for UGT74C1 in aliphatic glucosinolate synthesis. Hierarchical cluster analysis of transcriptome co-expression data previously predicted a function of UGT74C1 in Met-derived glucosinolate formation (Gachon et al., 2005). This prediction is consistent with the observation that MYB28 overexpression specifically up-regulates aliphatic glucosinolate-related genes, including UGT74C1 (Gigolashvili et al., 2007b; Hirai et al., 2007). Likewise, genetic engineering of the aliphatic branch into tobacco (leading to the S4 compound) utilized UGT74C1 as the candidate UGT gene (Mikkelsen et al., 2010).
Although the activity screen implicated clade UGT74 as putative thiohydroximate UGTs, our data as well as reports by others do not support a major role for the remaining enzymes in glucosinolate production. First, their catalytic efficiency values (kcat/Km) for MIATH and MTBTH are 1000–10 000 times lower than those of UGT74B1 (Table 1). Second, interrogation of the ATTED-II co-expression database confirmed correlation of expression of UGT74B1 and UGT74C1, but not of the other five UGT74 members, with genes of the glucosinolate pathway (Table S3). Finally, the phenotypes and chemotypes of the corresponding ugt74 loss-of-function lines, either as single or double mutants with ugt74b1, are similar to the wild-type or ugt74b1 plants, respectively (Figures 3 and 4). Because UGT74D1 glucosylates IAA and IAA-derived metabolites (Jin et al., 2013; Tanaka et al., 2014), the two-fold accumulation of Trp-derived glucosinolates in the ugt74b1-2 ugt74d1 double mutant relative to ugt74b1-2 (Figure 4b) may be caused by diversion of excess indolyl-3-aldoxime (IOX), a consequence of deregulated IAA homeostasis, into the indolic glucosinolate branch (Grubb and Abel, 2006). While the function of UGT74E1 is still elusive, UGT74F1 and UGT74F2 modify aromatic acids, including salicylate and anthranilate (Lim et al., 2002; Quiel and Bender, 2003; Dean and Delaney, 2008), and UGT74E2 glucosylates indolyl-3-butyric acid (Tognetti et al., 2010).
The modest induction of UGT74F1 in the ugt74b1 background (−1.5-fold), which is statistically not significant (Figure S2), together with the fact that the encoded protein displays in vitro thiohydroximate UGT activity, may be interpreted as evidence for a role in glucosinolate synthesis. However, although we cannot completely rule out such a minor contribution, we deem this unlikely. First, the respective kcat values of UGT74F1 are −200 times lower than those of UGT74B1 (Table 1). Second, UGT74F1 transcripts are expressed at a much lower steady-state level (<5%) than either UGT74B1 or UGT74C1 mRNAs (Figure S7). Third, we showed previously that low thiohydroximate application (<15 μm) is lethal to plant roots (Grubb et al., 2004), suggesting that UGT74F1 operates in vivo at a very small fraction of its Vmax. As we have argued elsewhere (Kopycki et al., 2013) such an enzyme could not possibly support significant flux through the glucosinolate biosynthetic pathway. Considering the lack of genetic evidence from the double mutant analysis, we therefore chose not to pursue further transgenic approaches with the remaining members of the UGT74 clade.
Evolutionary innovation in metabolism of hormones and natural products
Phylogenetic analysis of the UGT74 clade supports the notion that UGT74B1 is the ancient thiohydroximate glucosyltransferase, whose divergence coincided with the origin of the pathway, while UGT74C1 diverged from a separate group of non-glucosinolate-related glucosyltransferases much later as part of evolutionary innovation in the glucosinolate pathway (Mithen et al., 2010). The tree in Figure 9 has two other interesting implications, however. First, UGT74D1, UGT74E1, and UGT74E2 are also Brassica-specific. Together with IOX-dependent auxin synthesis (Bak et al., 2001; Zhao et al., 2002), the function of UGT74D1 and UGT74E2 in IAA metabolism is another example of Brassicales-specific innovations in auxin metabolism, highlighting the ongoing dynamism in the evolution of pathways related to this plant hormone. Second, UGT74F1 and UGT74F2 reside in a well supported clade that, unexpectedly, includes no monocots. Given the ancient origin of the salicylate signaling pathway (Thaler et al., 2012), this situation implies that the ancestral gene of the clade may have been a salicylate glucosyltransferase that subsequently gave rise to proteins acting both in secondary metabolism and in the metabolism of another hormone. This finding is surprising, as we anticipated that enzymes of the stabilization phase of glucosinolate synthesis had been recruited from detoxification pathways.
Implications for glucosinolate biosynthesis
The identification of a UGT specific to one branch of the glucosinolate pathway is not surprising. At least 10 enzymes active in glucosinolate biosynthesis exhibit differential activity with respect to side-chain structure (Sønderby et al., 2010). SUR1 and GPP1 are the only genes believed to be common to the core pathway. Yet, even sur1 seedlings produce trace glucosinolate levels (Figure 6) and presence of another C-S lyase has been proposed (Hirai et al., 2007). While the chemotype of ugt74b1-2 ugt74c1-2 plants clearly places UGT74C1 in the aliphatic branch, analysis of the overexpression line indicates utilization of aliphatic and indolic thiohydroximates, implying that synthesis of both glucosinolate classes is compartmentalized, either in different cell types or biosynthetic complexes. The glucosinolate profile of the double mutant suggests additional UGTs contributing to glucosinolate production. While our study did not reveal any outstanding candidates for such a role, our in vitro activity screen pointed to several enzymes in other UGT clades (Table S1). While the relative activity of these UGTs was low compared with UGT74B1, we note that the screen was conducted with saturating thiohydroximate concentrations, which exaggerates the role of low affinity, high turnover enzymes while underestimating the contribution of high affinity, low turnover enzymes. Furthermore, as shown in Table 1, side-chain structure can make an enormous difference in activity. While the glucosinolates remaining in the ugt74b1 ugt74c1 mutant are largely indolic, the substrate used for the screen (PATH) is not. Thus, some UGT candidates remain to be examined.
An exciting new avenue in glucosinolate research has been the finding that 4IM is the precursor of a signaling molecule essential for resistance to nonhost pathogens (Bednarek et al., 2009; Clay et al., 2009). It is therefore worth noting that 4IM accumulation is not affected in any mutant examined here, despite the fact that ugt74b1-2 plants have less than half as much IM as the wild-type. IM is the precursor of 4IM, implying that control of 4IM level is actively regulated, even in the absence of pathogen attack. However, CYP81F2, which catalyzes the first step in the IM-to-4IM conversion, was identified based on natural variation of 4IM content (Pfalz et al., 2009), suggesting genetic control of baseline 4IM level, in addition to biochemical and environmental factors. The significant reduction of 1IM accumulation in ugt74b1-2 ugt74c1-2 plants (Figure 6c) suggests a direct effect on indolic glucosinolate synthesis. However, the levels of IM and 4IM are unchanged relative to the single mutant. A likely explanation for 1IM reduction is stress-induced 4IM synthesis, which can be elicited by biotic and abiotic factors (Bednarek et al., 2009; Clay et al., 2009). We demonstrated previously phytotoxicity of thiohydroximates (Grubb et al., 2004), a finding that suggests that ugt74b1-2 ugt74c1-2 plants suffer some degree of metabolic stress. As synthesis of 1IM and 4IM compete for the same precursor, the observed chemotype may result directly from competition or indirectly from downregulation of 1IM synthesis. Therefore, it is unlikely that the altered 1IM level represents an important clue to the function of UGT74C1.
We are left to speculate on the biological role of UGT74C1 in glucosinolate biosynthesis, which responds to numerous environmental and developmental cues (Burow et al., 2010). While UGT74B1 and most pathway genes are co-regulated, UGT74C1 is largely co-expressed with genes of the aliphatic branch, particularly in datasets that probe responses to hormone and stress treatments. In contrast, co-expression of UGT74B1 and aliphatic glucosinolate-related genes is limited to developmental datasets (Table S3), suggesting different ecophysiological functions of the two enzymes. Thus, UGT74C1 likely plays a role under some special set of conditions. Given that glucosinolate metabolism is organized at a cell type-specific scale (Shroff et al., 2008), we have may missed the detection of subtle chemotypes in ugt74c1 tissues due to insufficient spatial resolution (Figure S4). Definition of the precise conditions of UGT74C1 function in planta and elucidation of the intricate metabolic connections between glucosinolate homeostasis and associated biochemical networks remain challenges of current research, which will likely uncover new players. Sønderby et al. (2010) noted that ‘formation of the glucosinolate core structure seemed almost completely elucidated by 2006’, and that ‘a stunning number of >20 genes in glucosinolate biosynthesis in Arabidopsis has been discovered’ since. Indeed, the picture appears poised to get more complex yet.
Phenylacetothiohydroximate (PATH), 1-methylindolyl-3-acetothiohydroximate (MIATH), and 4-methylthiobutylthiohydroximate (MTBTH) were synthesized as described previously (Grubb et al., 2004; Pedras et al., 2009; Kopycki et al., 2011).
Activity screen of recombinant UGT polypeptides
Ninety-three cDNA sequences that encode Family 1 UGTs of A. thaliana were expressed in E. coli and the resulting recombinant C-terminal glutathione S-transferase (GST) fusion proteins were purified as described previously (Lim et al., 1998, 2002). Each glucosyltransferase activity reaction mix (200 μl) contained 1 μg of purified recombinant protein, 20 mm MES–KOH, pH 6.0 (or 20 mm Tris–HCl, pH 7.0; or 20 mm Tris–HCl, pH 8.0), 0.004% (v/v) β-mercaptoethanol, 5 mm MgSO4, 5 mm UDP-glucose, and 5 mm substrate (PATH). After incubation for 3 h at 30°C, the reactions were stopped by the addition of 20 μl of trifluoroacetic acid (24%, w/v), flash-frozen, and stored at −20°C before reverse-phase high pressure liquid chromatography (HPLC) analysis of desulfobenzylglucosinolate formation in accordance with Grubb et al. (2004).
Assay of (His)6-tagged recombinant UGT74 proteins
Construction of a vector for heterologous expression of UGT74B1 has been described previously (Grubb et al., 2004). An Arabidopsis cDNA that corresponded to UGT74C1 (At2g31790) was identified and a cDNA clone (U11123) was obtained from the Arabidopsis Biological Resource Center (ABRC) (Columbus, OH, USA); this clone served as the template for amplification of the corresponding cDNA. The coding sequences of the remaining UGT74 members were amplified with specific primers from reverse transcribed mRNA and cloned into the pQE30 or pQE31 vectors (Qiagen, Valencia, CA, USA, www.qiagen.com). The primer sequences and the restriction site details are given in the Table S4. All proteins were expressed as N-terminal (His)6-tag fusions that allowed single step purification. The fidelity of each clone was verified by DNA sequencing. The proteins were expressed in the E. coli strain M15[pREP4], affinity-purified, and their UGT activity and kinetic parameters determined (Grubb et al., 2004; Kopycki et al., 2013).
Protein concentration was determined with the Bio-Rad Protein Assay kit (Bio-Rad, Hercules, CA, USA, www.bio-rad.com/) following the manufacturer's protocol, and using bovine serum albumin as a standard. Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) analysis was performed using 13.3% (w/v) acrylamide gels and standard protocols for electrophoresis and protein staining with Coomassie R-250 stain (Bollag et al., 1996).
Wild-type Arabidopsis thaliana (L.) Heynh. accessions Columbia (Col-0), Landsberg erecta, Wassilewskija (Ws-0), mutants in UGT74D1, UGT74E1, UGT74E2, UGT74F1 and seeds for superroot1/rooty (Seo et al., 1998) were obtained from the ABRC. Gene trap line GT1305 (UGT74C1) was provided by J. Simorowski (CSHL; Cold Spring Harbor, NY, USA, www.cshl.edu; Sundaresan et al., 1995) and T-DNA insertion line (Col-0) GK-672-G10 (UGT74C1) by B. Weisshaar (GABI-Kat, Max-Planck-Institute for Plant Breeding Research, Cologne, Germany; Rosso et al., 2003). The UGT74F2 allele ugt74f2-i1a was a kind gift from J. Bender (Quiel and Bender, 2003). UGT74B1 T-DNA insertion alleles ugt74b1-1, ugt74b1-3 (Ws-0) and ugt74b1-2 (Col-0) have been described previously (Grubb et al., 2004). All insertion lines were backcrossed at least once prior to phenotypic analysis. Further details of the ugt74 mutant lines and primers used for genotyping are given in Table S5.
For expression analysis of transgenic and knockout lines, RNA was extracted using the RNeasy Plant Mini Kit, including DNase treatment (Qiagen). The Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific, Pittsburgh, PA, USA, www.thermoscientific.com) was applied with 2 μg total RNA as template and an oligo(dT) primer. Quantitation relied on the SYBR Green master mix (Applied Biosystems, Foster City, CA, USA, www.appliedbiosystems.com) applied with the Mx3005P QPCR System (Stratagene, La Jolla, CA, USA, www.agilent.com). The internal control for RNA equalization was PP2A (primers 5′-AGCCAACTAGGACGGATCTGGT-3′ and 5′-GCTATCCGAACTTCTGCCTCATTA-3′). UGT74B1 expression was quantified with primers 5′-CGTTTCGTATCCGTGGCTTA-3′ and 5′-CAGACTCACCATTTTCACAATCT-3′. UGT74C1 expression was quantified with primers 5′-CCTGACCGATTTCATCTCTAGTGC-3′ and 5′-TGGCTATGTCCAATGCAAAGGG-3′. Primers for validating T-DNA insertion lines, and for measuring expression of clade members in the ugt74b1-1 line, can be found in Table S6.
Analysis of desulfoglucosinolates by HPLC, both for plant-derived samples and for in vitro assays of UGT activity (desulfobenzylglucosinolate formation) was performed as previously described (Grubb et al., 2004; Kopycki et al., 2013). For kinetic analyses the formation of desulfoglucosinolates was monitored using the Waters (Milford, MA, USA, www.waters.com) Acquity ultraperformance liquid chromatography (UPLC) system [column: HSS T3 C18 1.8 μm, 2.1 × 100 mm; solvent: water (A) acetonitrile (B) gradient, 1% B 0.6 min, 20% B 4 min, 20% B 6 min, 100% B 6.5 min, 100% B 8 min; flow 0.5 ml min−1 at 40°C]. Desulfoglucosinolates were detected at 226 and 280 nm. The retention times and the ultraviolet (UV) light absorbance spectra of reaction products were compared with standard substances.
Histochemical analysis of the GUS reporter enzyme activity was adapted from Jefferson et al. (1987). Sample tissues were fixed in ice-cold 80% acetone (2 × 5 min), vacuum-infiltrated with an aqueous solution consisting of 50 mm NaH2PO4-Na2HPO4, pH 7.2, 0.5 mm K3Fe(CN)6, 0.5 mm K4Fe(CN)6, and 2 mm 5-bromo-4-chloro-3-indolyl-β-d-glucoronic acid and incubated overnight at 37°C. Tissues were cleared overnight in 70% ethanol and imaged with a stereomicroscope (Zeiss Stemi SV11, Jena, Germany, www.zeiss.com) equipped with a Cool Snap high-performance charge coupled device (CCD) camera.
All claims of statistical significance are based on a Student's one-tailed t-test, using a P-value 0.05 level of significance.
Construction of transgenic lines, plant growth conditions, determination of plant hormones, gene expression and phylogenetic analyses are described in Methods S1–S5.
We thank Birgit Ortel, Claudia Schramm and Katja Baumann-Kaschig for able technical assistance. This work was supported by the U.S. National Science Foundation (IBN 0344123 to S.A.) and core funding to the Leibniz Institute of Plant Biochemistry from the state of Saxony-Anhalt and the Federal Republic of Germany.