Key role for a glutathione transferase in multiple-herbicide resistance in grass weeds

Abstract

Multiple-herbicide resistance (MHR) in black-grass (Alopecurus myosuroides) and annual rye-grass (Lolium rigidum) is a global problem leading to a loss of chemical weed control in cereal crops. Although poorly understood, in common with multiple-drug resistance (MDR) in tumors, MHR is associated with an enhanced ability to detoxify xenobiotics. In humans, MDR is linked to the overexpression of a pi class glutathione transferase (GSTP1), which has both detoxification and signaling functions in promoting drug resistance. In both annual rye-grass and black-grass, MHR was also associated with the increased expression of an evolutionarily distinct plant phi (F) GSTF1 that had a restricted ability to detoxify herbicides. When the black-grass A. myosuroides (Am) AmGSTF1 was expressed in Arabidopsis thaliana, the transgenic plants acquired resistance to multiple herbicides and showed similar changes in their secondary, xenobiotic, and antioxidant metabolism to those determined in MHR weeds. Transcriptome array experiments showed that these changes in biochemistry were not due to changes in gene expression. Rather, AmGSTF1 exerted a direct regulatory control on metabolism that led to an accumulation of protective flavonoids. Further evidence for a key role for this protein in MHR was obtained by showing that the GSTP1- and MDR-inhibiting pharmacophore 4-chloro-7-nitro-benzoxadiazole was also active toward AmGSTF1 and helped restore herbicide control in MHR black-grass. These studies demonstrate a central role for specific GSTFs in MHR in weeds that has parallels with similar roles for unrelated GSTs in MDR in humans and shows their potential as targets for chemical intervention in resistant weed management.

The evolution of herbicide resistance in weeds is a global problem with serious implications to sustainable arable agriculture (1–3). The best-characterized resistance mechanisms arise from mutations in the proteins targeted by herbicides that lead to a reduced sensitivity to inhibition [target site-based resistance (TSR)]. Mutations leading to TSR have been well described for the plastoquinone-binding protein of photosystem II (PSII) and the acetyl CoA carboxylases (ACCases) and acetolactate synthases involved in fatty acid and branched chain amino acid biosynthesis, respectively (4). Whereas TSR in weeds is widespread, chemical control can be restored by alternating the use of herbicides with differing modes of action (1, 2, 4). A second and more problematic mechanism is based on weeds evolving multiple-herbicide resistance (MHR), which is distinct from herbicide cross-resistance arising from the pyramiding of multiple-TSR traits (4, 5). In MHR, weeds deploy a central defense system, which counteracts herbicide-imposed toxicity irrespective of their mode of action (3). MHR is linked to an enhanced ability of the weeds to detoxify herbicides and has also been termed metabolism-based resistance (3, 6). MHR is most problematic in black-grass (Alopecurus myosuroides) and annual rye-grass (Lolium rigidum), which compete with cereal crops (1). In these weeds an enhanced ability to metabolize herbicides is a powerful route to resistance to graminicides, as the differential rates of detoxification between grasses and cereals represents one of the few biochemical features that can be exploited in selective chemical weed control (6).

MHR was first reported in black-grass in 1982 at Peldon in Essex, England, with independent outbreaks subsequently recorded across Europe (6, 7). Similarly, many populations of MHR annual rye-grass have arisen independently around the world, in some cases pyramiding with TSR traits (8). In grass weeds, MHR is associated with elevated levels of herbicide-detoxifying enzymes, including cytochrome P450 mixed-function oxidases (CYPs), family 1 UDP-glucose-dependent glycosyltransferases (UGTs), and glutathione transferases (GSTs) (9–11), as well as membrane-associated ATP-binding cassette (ABC) drug transporter proteins (12). Collectively, we have termed these xenobiotic detoxifying enzymes and transporters the “xenome” (13). In MHR black-grass the coordinated up-regulation of the xenome is associated with resistance to several graminicides (14, 15), such as chlorotoluron (PSII-inhibiting phenylurea) and fenoxaprop-p-ethyl (ACCase inhibitor). Previous studies in black-grass identified a specific GST as a highly expressed xenome component in the independent MHR black-grass populations, biotypes “Peldon” and “Spain”, but not in TSR or wild-type sensitive (WTS) plants (14, 15). As a member of the plant-specific phi (F) class of GST (GSTF1), this enzyme has been renamed A. myosuroides (Am) AmGSTF1 (13). Unlike other GSTFs in crops and weeds, AmGSTF1 showed little activity in detoxifying herbicides (15) but was highly active as a glutathione peroxidase (GPOX), catalyzing the reduction of organic hydroperoxides (15). The up-regulation of AmGSTF1 in MHR black-grass shows several intriguing parallels with the enhanced expression of unrelated pi (P) class GSTs (GSTPs) in multiple-drug-resistant (MDR) tumors in humans. In addition to directly detoxifying therapeutic drugs and quenching the formation of toxic hydroperoxides formed during treatment, GSTPs directly regulate signaling pathways that promote cellular defense (16). On the basis of this precedence for a GST to orchestrate MDR in humans, we have investigated the potential for the unrelated GSTF1s to have a regulatory role in MHR in weeds, through a combination of transgenesis and chemical inhibition studies.

Results

AmGSTF1 and Orthologs Are Highly Expressed in MHR Black-Grass and Annual Rye-Grass.

To determine how MHR affected protein expression, crude extracts from Peldon and WTS black-grass were fractionated on phenyl Sepharose to remove photosynthetic components and the polypeptides present visualized by staining following two-dimensional gel electrophoresis. Protein profiles were essentially identical, except for seven polypeptides with molecular masses around 28 kDa, which were enhanced in the MHR Peldon plants (Fig. 1A). The 28-kDa polypeptides were all identified as isoforms of AmGSTF1 on the basis of their identical peptide fingerprints and the presence of a diagnostic 1,038-Da fragment that yielded the common sequence VFGPAMSTNV following tandem MS (15). Analysis of leaf extracts showed that AmGSTF1 corresponded to 0.2% of the total protein in the MHR plants, being over 20 times more abundant than in the respective WTS weeds. These differences in expression levels were confirmed by Western blotting using an anti-GST serum (Fig. 1B) and by quantitative PCR (qPCR), with transcripts encoding AmGSTF1 isoenzymes being 13 times more abundant in MHR Peldon compared with WTS plants (Fig. 1C). To determine whether orthologs of AmGSTF1 were present in annual rye-grass, the MHR biotype SLR 31, which shows enhanced expression of herbicide-detoxifying CYP enzymes (17), was analyzed. Using conserved GSTF1-like sequences found in black-grass and cereals, eight closely related (83–91% amino acid identity) PCR amplification products were generated. Based on qPCR, these products were 4 times more abundant in the MHR SLR 31 plants than in the corresponding WTS weeds (Fig. 1C). As determined by Western blotting, GSTF1-like polypeptides were also more abundant in extracts from SLR 31 plants and the independent MHR biotype VLR 69 (18), compared with WTS plants (Fig.1B and Fig. S1A). A cDNA encoding a L. rigidum (Lr) LrGSTF1 with 91% identity to AmGSTF1 was then isolated from annual rye-grass by PCR (Fig. 2) and then cloned and expressed in Escherichia coli alongside AmGSTF1. Both recombinant enzymes had similar enzyme activities (Table 1), being highly active GPOXs and showing a limited ability to catalyze the glutathione conjugation of the model GST substrate 1-chloro-2,4-dinitrobenzene (CDNB). Consistent with these findings, GPOX activities determined in the MHR annual rye-grass populations were two- to fourfold higher than those determined in WTS plants (Fig. S1B).

Fig. 1.

Analysis of the GSTs in MHR grass weeds and transgenic Arabidopsis. (A) Analysis of soluble hydrophobic protein extracts from WTS and MHR Peldon black-grass biotypes by 2D gel electrophoresis. The red arrows refer to polypeptides identified by proteomics as AmGSTF1 subunits. (B) Western blots showing immunodetectable GSTF1 polypeptides in MHR black-grass (Peldon) and annual rye-grass (SLR31) and Arabidopsis plants expressing AmGSTF1 (lines 8 and 12) relative to WTS weeds and vector only (V) controls. (C) GSTF1 transcript abundance in WTS and MHR biotypes of black-grass and annual rye-grass as determined by qPCR.

Fig. 2.

Aligned sequences of GSTF1 orthologs from black-grass (UniProt accession no. Q9ZS17) and annual rye-grass. Residues within boxes represent hypothetical active-site residues inferred from the known maize GSTF1 crystal structure (Protein Data Bank accession no. 1axd). Cys120 residue is shaded in gray. Asterisks (*) denote identical amino acid residues. Nonidentical amino acids with similar properties are denoted by : and . .

Table 1.

Activities of black-grass and annual rye-grass GSTF1 enzymes

	Mean enzyme-specific activity, nmol⋅s−1⋅mg−1 protein
Substrate	AmGSTF1	LrGSTF1
CDNB	22.7 ± 0.42	26.3 ± 0.42
Cu-OOH	21.9 ± 0.42	43.5 ± 0.71
Lin-OOH	98.6 ± 15.9	272.6 ± 22.5

CDNB: 1-chloro-2,4-dinitrobenzene. Cu-OOH: cumene hydroperoxide. Lin-OOH: linoleic acid hydroperoxide. Measurements were performed in technical triplicate. Mean enzyme-specific activities are shown ± SD, n = 3.

Expression of AmGSTF1 in Arabidopsis Results in an MHR Phenotype.

The high levels of expression of GSTF1 orthologs in two species of MHR weeds suggested an important role for these proteins in herbicide resistance. To investigate the function of black-grass AmGSTF1, the enzyme was expressed in Arabidopsis, using the constitutive 35S promoter. The plants were similarly transformed with AmGSTL1, a lambda class member of the GST superfamily that is also constitutively enhanced in MHR black-grass (14). The expression of the black-grass GSTs in the homozygote immediate (T1) progeny was confirmed by Western blotting. With AmGSTL1, low levels of a 27-kDa polypeptide were identified in the transgenics that was absent in the controls (Fig. S2A), with the highest-expressing plants (line 16) used for further characterization. The anti-GSTF serum identified 26-kDa polypeptides in control plants (Fig. 1B), corresponding to the expression of known endogenous GSTFs in Arabidopsis (19). In the AmGSTF1 transgenics, additional 28-kDa immunoreactive polypeptides were also determined (Fig. 1B). Two lines showing intermediate (line 8) and high (line 12) expression of the transgene were selected for further analysis. All AmGST transformants, together with the respective controls, were tested for herbicide tolerance, using a combination of spraying whole plants, as well as germination phytotoxicity studies on agar (Fig. 3). The herbicides selected were the chloroacetanilide alachlor that inhibits fatty acid elongation and hence cell division and the PS II inhibitors atrazine (chloro-s-triazine) and chlorotoluron (phenylurea). Other classes of graminicidal herbicides used to control black-grass could not be tested in Arabidopsis, as they were either too toxic (e.g., the sulfonylurea thifensulfuron-ethyl and the diphenyl ether fluoroglycofen-ethyl) or inactive, due to inherent differences in herbicide target sensitivities in monocots and dicots (e.g., the aryloxyphenoxypropionate fenoxaprop-p-ethyl). AmGSTL1 transformants were as susceptible to herbicides as the vector-only controls (Fig. S2B). In contrast, the AmGSTF1 transformants were considerably more resistant to all three herbicides, both in spray and in germination trials (Fig. 3). Importantly, although the transgenic expression of a GST could enhance tolerance to alachlor and atrazine, which both undergo S-glutathionylation as primary steps in their metabolism (13), this is not the case with chlorotoluron, which is detoxified in plants by the combined action of CYPs and UGTs (20). To determine whether the detoxifying glutathione-conjugating activity of AmGSTF1 could contribute to the increased tolerance toward alachlor and atrazine, plant extracts were assayed for GST-conjugating activities. Consistent with the activity profile of the enzyme, both AmGSTF1 transgenic lines showed a 3- to 4.5-fold enhancement in GPOX activity and increased glutathione conjugation of CDNB and the herbicide alachlor (Table 2). The AmGSTF1 expressors also showed increased conjugating activity toward the herbicide atrazine (Table 2). Because AmGSTF1 had no detectable activity with atrazine as a substrate (15), these increases in GST activity had to result from the increased expression of endogenous Arabidopsis enzymes. Similarly, AmGSTF1 expression was also associated with an enhancement in unrelated glycosylating activities in Arabidopsis toward the xenobiotic 2,4,5-trichlorophenol (Table 2). Thus, although the exact route of atrazine detoxification in the AmGSTF1 transformants was not determined, it was demonstrated that two independent routes of bioconjugation known to be involved in the metabolism of this herbicide were both enhanced (13). Recent studies have shown that MHR in grasses is also associated with changes in endogenous antioxidant and secondary metabolism, notably an accumulation of cytoprotectants such as glutathione, flavonoids, and anthocyanins (14). When the transgenic Arabidopsis plants were analyzed using liquid chromatography coupled to MS detection (LC-MS), a range of UV-absorbing metabolites were determined (Fig. 4), which accumulated at higher levels in the AmGSTF1 expressors compared with vector-only controls (Table 2). On the basis of their UV and MS spectra, these compounds were identified as conjugates of the flavonol kaempferol and the anthocyanin cyanidin, respectively (Fig. 4). Levels of glutathione (GSH) were also shown to be modestly enhanced in the AmGSTF1 expressors, although the ratio of reduced to oxidized forms was unaffected relative to controls (Table 2). To determine whether these biochemical changes were associated with a perturbation in gene expression, the transcriptome of the line 12 transgenics was compared with that of wild-type plants, using the Affymetrix GeneChip platform. Ranking the top 50 up- and down-regulated genes showed only minor changes in the transcriptome (Dataset S1). When the top 12 most perturbed genes were used as qPCR biomarkers in the two independent AmGSTF1-expressing plant lines (lines 8 and 12), no consistent changes in gene expression were determined (Fig. S3). This inferred that the changes in biochemistry determined following transformation with AmGSTF1 were not regulated at the level of transcription.

Fig. 3.

Herbicide resistance of transgenic Arabidopsis expressing AmGSTF1. (A) AmGSTF1 expressors and vector-only controls were germinated on agar containing 2 µM chlorotoluron, alachlor, atrazine, or acetone and maintained for 30 d. (B) AmGSTF1-expressing and vector-only control plants were sprayed with chlorotoluron, alachlor, atrazine, or formulation only at rates of 30 g ai per hectare, 1200 g ai per hectare, and 30 g ai per hectare, respectively, and assessed 9 d after herbicide application.

Table 2.

Biochemical phenotype of two independent lines of Arabidopsis transformed with AmGSTF1 compared with plants transformed with the respective empty vector

	Vector only	Line 8	Line 12
Enzyme activity, nkat⋅mg−1
GST	0.26 ± 0.01	1.01 ± 0.01	1.66 ± 0.02
GPOX	0.02 ± 0.00	0.06 ± 0.00	0.09 ± 0.00
Glutathione reductase	0.434 ± 0.00	0.456 ± 0.01	0.406 ± 0.00
Thiol transferase	0.029 ± 0.01	0.088 ± 0.01	0.097 ± 0.01
Catalase	1,949 ± 33	1,725 ± 163	1,718 ± 264
Antioxidant content, nmol⋅g−1 FW
GSH	202 ± 5	252 ± 8	299 ± 6
GSSG	9 ± 1	10 ± 1	12 ± 2
Flavonol	564 ± 32	1,086 ± 59	1,237 ± 67
Anthocyanin
Peak 2	115 ± 9	454 ± 32	466 ± 47
Peak 3	122 ± 5	432 ± 9	483 ± 23
Herbicide-conjugating activity, pkat⋅mg−1
Alachlor	3.5 ± 0	7.1 ± 0.2	10.7 ± 0.2
Atrazine	0.015 ± 0.00	0.061 ± 0.01	0.056 ± 0.00
Glucosyl transferase activity, fkat⋅mg−1
2,4,5-Trichlorophenol	15.4 ± 3.1	22.9 ± 1.3	21.1 ± 2.0
Quercetin	16.7 ± 0.8	19.9 ± 2.2	20.9 ± 0.7

Measurements are shown ± SEM, n = 3. FW: fresh weight.

Fig. 4.

(A and B) Polyphenol content of Arabidopsis plants transformed with (A) vector only or (B) AmGSTF1. Flavonols and anthocyanins were identified by HPLC-MS with reference to earlier published work. Peaks 1 and 4 were rhamnosylated conjugates of the flavonol kaempferol, whereas peaks 2 and 3 were derivatives of the anthocyanin cyanidin.

MHR in Weeds Is Chemically Reversible.

The results of the transgenesis studies showed that AmGSTF1 played a causative role in MHR. It was therefore of interest to identify potential chemical intervention strategies that could disrupt the function of the GST and help restore herbicide sensitivity. Drawing on parallels with MDR in humans, the inhibition of drug-detoxifying GSTs has been a productive target for medicinal chemistry programs (21). Such chemical interventions have also been shown to disrupt GSTs functioning in signaling roles, for example in modulating the c-Jun-N-terminal kinase (JNK) and apoptosis signal-regulating kinase (ASK1) signaling pathways (16, 22). As such, there is good precedence for using inhibitors to disrupt GSTs eliciting resistance through multiple mechanisms. These inhibitors can be subdivided into GSH conjugates and related peptidomimetics that bind in the related glutathione (G) binding site of GSTs and compounds acting on the large hydrophobic (H) binding domain (22). The latter simpler H-site inhibitors developed for cancer chemotherapy (23) were tested for their ability to inhibit GSTF1 activity and to augment herbicide efficacy. These included the classic GST inhibitor ethacrynic acid, as well as compounds based on bromoenol lactone and benzoxadiazole chemistries (Fig. 5A). Each compound was tested for its ability to inhibit the conjugating activity of AmGSTF1 toward CDNB. In parallel, each compound was sprayed onto WTS and MHR Peldon black-grass 48 h before an application of herbicide. By combining the in vitro and in planta screens, several compounds that inhibited GST activities could be discounted from further exploration due to their innate phytotoxicity (e.g., compounds 2 and 3) or a lack of an observable potentiation in herbicide activity (compound 4, Fig. S4). The compound 4-chloro-7-nitrobenzoxadiazole [1, nitrobenzoxadiazole (NBD)-Cl], derivatives of which target GSTPs in tumor cell lines (16, 24), was shown to both inhibit AmGSTF1 (Fig. 5B) and enhance the phytotoxicity of chlorotoluron when presprayed on MHR Peldon black-grass (Fig. 6 A and B). Similarly, NBD-Cl enhanced the herbicidal activity of the ACCase-inhibiting graminicides fenoxaprop-p-ethyl and clodinafop-propargyl when applied to Peldon plants, with similar results obtained with the independent MHR Spain biotype (Fig. 6 C–F). Consistent with results obtained with this class of chemistry in MDR cells (16, 24), NBD-Cl exerted its effects without causing overt secondary toxicity. Thus, both black-grass (Fig. 6 A and B) and wheat (Table S1), showed no leaf damage or growth inhibition when exposed to NBD-Cl. Using simple nucleophilic aromatic substitution chemistry, a series of NBD derivatives were prepared, bearing a variety of leaving and activating groups, and used in spray trials (Fig. 7). On the basis of the results of this limited screen (Fig. S5), a nitro group proved to be the most viable activating group. Importantly for future agrochemical optimization, the enhancement of herbicide activity showed a dependence on chemical structure, notably on the nature of the leaving groups (alkoxy and thiol > amino). Intriguingly, modifying the NBD pharmacophore with similar modifications to those shown to potentiate activity in MDR proved ineffective in counteracting MHR. For example, the active GSTP1 inhibitor 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)-hexanol (NBDHEX) (compound 5, Fig. S5A) (16) proved ineffective when used with chlorotoluron against MHR black-grass, possibly due to its inability to be taken up by intact leaves.

Fig. 5.

Selected GST inhibitors tested in this study. (A) Inhibitor chemical structures. (B) Their efficacy against black-grass and annual rye-grass GSTF1s as determined from IC₅₀ values. 1, NBD-Cl; 2, ethacrynic acid; 3, cyanuric chloride; 4, bromoenol lactone.

Fig. 6.

Effect of NBD-Cl on herbicide resistance in black-grass. (A and B) For studies with chlorotoluron, (A) WTS and (B) MHR Peldon black-grass plants were treated at 12 d with either formulation or NBD-Cl (270 g ai per hectare), before an application of either formulation only (Form) or 500 g ai per hectare of herbicide (chlorotoluron; CHL). (C and D) For studies with fenoxaprop-p-ethyl formulated as Cheetah Super, (C) WTS or (D) MHR Peldon plants were pretreated with NBD-Cl (80 g ai per hectare), before spraying with formulation control (Form) or 85 g ai per hectare of herbicide [fenoxaprop-p-ethyl (FXP)]. (E and F) For studies with the independent MHR Spain black-grass biotype, WTS and MHR Spain black-grass plants were pre-treated with NBD-Cl (270 g ai per hectare) or formulation only followed by a treatment with (E) 165 g ai per hectare of fenoxaprop-p-ethyl (FXP) or formulation only or (F) 250 mL per hectare of clodinafop-propargyl (CDF), as the commercial formulation Topik, or formulation only. In all cases plants were evaluated for phytotoxic injury 21 d postherbicide application.

Fig. 7.

Nucleophilic aromatic substitution chemistry approaches to analogs explored in this study (EWG, electron withdrawing group; LG, leaving group; Nu, nucleophile).

Discussion

Our results demonstrate a central role for AmGSTF1 in MHR in black-grass, with the enhancement of the orthologous LrGSTF1 in resistant annual rye-grass, suggesting similar functions in other weeds. The AmGSTF1 transgenesis studies in Arabidopsis showed that increased tolerance to herbicides was associated with an enhancement in two classes of xenobiotic detoxifying enzymes and the accumulation of GSH, flavonols, and anthocyanins (Table 2 and Fig. 4). These changes in biochemical phenotype closely resemble those determined in MHR, relative to WTS black-grass (15). Single GSTs have previously been shown to confer tolerance to herbicides as a direct consequence of their xenobiotic-detoxifying (13) or protective GPOX functions (25). Plant GSTs have also been shown to orchestrate tolerance to abiotic stress through their ability to regulate redox signaling pathways that transcriptionally activate defense genes (26). On the basis of our earlier studies, we had proposed that AmGSTF1 primarily conferred its protective activity against graminicides in black-grass by acting as a scavenger of hydroperoxides released as a downstream consequence of herbicide injury (15). However, the current studies in the transformed Arabidopsis show that AmGSTF1 effects its changes in protective antioxidant and detoxification metabolism through an as yet uncharacterized mechanism that does not involve the transcriptional activation of oxidative stress response pathways or measurable changes in the redox ratio of GSH:GSSG.

Instead we propose that AmGSTF1 must be exerting its metabolic effects either by binding and stabilizing key pathway intermediates or by posttranslationally modulating the activities of specific regulatory proteins and enzymes. With respect to the metabolite binding or “ligandin” function there is now strong evidence linking GSTs to flavonoid and anthocyanin metabolism apparently through binding and shuttling pathway intermediates from their sites of synthesis in the cytosol for deposition in the tonoplast (27). Whereas in previous studies a role for GSTs in flavonoid metabolism/transport has been inferred from loss-of-function studies with mutants, our current results uniquely show that constitutive expression of a GST from black-grass leads to the increased accumulation of these metabolites. In recent studies, we have demonstrated that GSTs selectively bind a range of biologically active natural products, including flavonoids, and can protect the more unstable intermediates from auto-oxidation (28). In the current study, as determined by HPLC-MS, AmGSTF1 was found to bind several flavonoids when exposed to extracts from the grass weeds and Arabidopsis (Fig. S6). Therefore, we postulate that AmGSTF1 could effectively act as a flavonoid ligandin binding protein, which, by stabilizing key intermediates, reduces their turnover and results in a steady accumulation of flavonol and anthocyanin end products. AmGSTF1 also shows several functional similarities to the distantly related GSTP1 linked to MDR. In particular, both enzymes are acted on by chemicals containing the NBD pharmacophore, resulting in enzyme inhibition and the suppression of resistance. In the case of GSTP1, NBD-containing chemistries form glutathionylated derivatives that bind to both the G and H sites of the enzyme (16, 24). Additionally, NBD compounds also alkylate cysteines in GSTP1 away from the active site, which results in a disruption in the enzyme’s ability to S-glutathionylate regulatory proteins (22, 23). Intriguingly, AmGSTF1 and LrGSTF1 both contain a conserved cysteinyl residue (Cys-120) (Fig. 2) that is readily covalently modified by both NBD-Cl and GSH (Fig. S7).

These studies also shed light on the relative importance of enhanced xenobiotic metabolism in determining resistance to multiple herbicides in plants. Previous studies demonstrated that safeners, which are agrochemicals that enhance herbicide selectivity in cereal crops through enhanced xenobiotic detoxification, cause major increases in the expression of CYPs, GSTs, UGTs, and other xenome components on feeding to plants and cultures of Arabidopsis (29). However, unlike that in cereal crops, the large-scale enhancement of the xenome in Arabidopsis did not lead to increased herbicide tolerance. In the current study, we show that the up-regulation of a single xenome component, AmGSTF1, can confer increased tolerance to multiple herbicides. Previously we had proposed that safening and MHR in grass weeds were functionally similar (14). The current findings suggest that although the two responses show similarities in xenome responsiveness, the number of key changes required to elicit MHR may be more restricted than those associated with safening.

The finding that a single protein, AmGSTF1, has such a major role in controlling the complex phenotype of MHR is unexpected. Studies in annual rye-grass provide evidence that MHR is a polygenic trait that sequentially evolves under repeated herbicide application (1, 17). This suggests that there are additional signaling mechanisms regulating MHR in grass weeds. Intriguingly, the enhancement of expression of the GSTF1 ortholog seen in MHR annual rye-grass was less marked than that determined in the resistant black-grass. This indicates that additional regulatory systems are particularly likely to have evolved to control MHR in annual rye-grass relative to black-grass, which is consistent with the relative greater diversity of resistance mechanisms described in this weed (1). When functioning in a regulatory role, plant GSTs represent a tractable central mechanism for controlling plant stress responses and are already implicated in assuming similar roles in development processes (13, 19). In black-grass, AmGSTF1 is known to be induced by multiple inputs, which include both biotic (drought, heat) and abiotic (herbicide, safener) treatments (14, 15). We speculate that in black-grass, biotypes showing elevated levels of expression of genes regulating each of these signaling pathways could be sequentially selected for through a combination of repeated herbicide treatments or adverse weather conditions, effectively ramping up the expression of AmGSTF1 over generations to the point that MHR is invoked. Although the exact mechanism of AmGSTF1 induction is yet to be determined, we can eliminate the possibility of the amplification of the respective gene having a causative role on the basis of Southern analysis of genomic DNA from Peldon vs. WTS black-grass (15).

The identification of compounds that when used with existing graminicides can help restore weed control in MHR weeds is potentially a very important development. Compounds such as piperonyl butoxide that inhibit detoxifying CYPs and esterases have been used for many years to counteract metabolic-based insecticide resistance (30). Similarly, in weed control, inhibitors of detoxifying enzymes can assist in the chemical control of plants showing metabolism-based resistance (31). For example, tridiphane is a known synergist of herbicides in weeds that are detoxified by GSTs by acting as a potent product-based inhibitor of these enzymes following its S-glutathionylation (31). In the case of the MHR grass weeds, tridiphane proved inactive as a synergist and a weak inhibitor of AmGSTF1 (IC₅₀ > 0.1 mM). In contrast, we were able to demonstrate that a simple compound like NBD-Cl was surprisingly effective in enhancing herbicidal efficacy in MHR grass weeds. This observation has several parallels in using similar chemistries to disrupt GSTP1 functioning to control apoptosis in tumor cells, thereby counteracting MDR (16, 23, 24). Although we cannot exclude the possibility that NBD-Cl is acting on additional protein targets that underpin MHR, the structure activity studies with the related chemical series and lack of discernable phytotoxic side effects support the conclusion that AmGSTF1 is the primary target of this pharmacophore. Although the innate reactivity of NBD-Cl in the environment poses serious limitations in developing this compound for practical use in the field, this study shows that by identifying a key component regulating MHR in grass weeds it is possible to develop small molecules that can help to reverse the resistance phenotype and restore control by existing graminicides. In view of the rapid spread of MHR in grass weeds and the restricted development of new herbicides, such compounds offer a viable alternative strategy in counteracting resistance in the field.

Materials and Methods

Plant Studies.

Chemical treatments (herbicides and inhibitors), MHR and WTS black-grass and annual rye-grass and control and transformed lines of Arabidopsis thaliana were prepared as detailed in SI Materials and Methods.

Plant Analysis.

Protein and metabolite analysis and enzyme assays were conducted as described in SI Materials and Methods, using previously described methods. Alterations in the transcriptome of Arabidopsis plants transformed with AmGSTF1 (line 12) relative to untransformed controls were determined in triplicate, using Affymetrix arrays, and the results confirmed by qPCR as described in SI Materials and Methods.

Accessions.

LrGSTF1 nucleotide sequences were deposited with the European Nucleotide Archive (accession no. HF548530). Microarray data files have been deposited with the Gene Expression Omnibus (accession no. GSE42065) (32).