Introduction
Weeds play a major role in lowering productivity in arable fields, and their control has been a major task in farming since the beginning of agriculture. The availability of cost-effective and resource-saving organic herbicides has led to a shift from cultural and mechanical control to intensive utilization of chemical weed control since the middle of the last century. However, besides increasingly restrictive pesticide registration processes, emerging cases of herbicide-resistant weeds restrict the variety of herbicides available in different cropping systems and limit the number of possible rotations of active ingredients in the sense of good agricultural practice and optimal weed management.
Blackgrass (Alopecurus myosuroides Huds.) is one of the most noxious weeds in temperate Europe. Particularly in winter cereal–dominated cropping areas, A. myosuroides infestations can be severe, and due to this weed’s evolution of herbicide resistance to several chemical classes, effective control has become increasingly complex. To date, resistance to seven herbicide mechanisms of action (MoAs) has been described in A. myosuroides, occasionally occurring in multiple-resistant populations (Heap Reference Heap2019). Of particular concern, resistance to herbicides predominantly used in POST applications, such as inhibitors of acetyl-CoA synthase (WSSA Group 1, HRAC Group A) and acetolactate synthase (WSSA Group 2, HRAC Group B), is becoming more widespread (Peterson et al. Reference Peterson, Collavo, Ovejero, Shivrain and Walsh2018). Therefore, ensuring satisfactory efficacy of PRE herbicides by keeping the number of A. myosuroides plants at least below the estimated economic thresholds of 16 to 33 plants m−2 (Mennan et al. Reference Mennan, Bozoglu and Isik2003) has become increasingly important (Bailly et al. Reference Bailly, Dale, Archer, Wright and Kaundun2012). However, higher control rates may be desirable to keep seedbanks low, particularly in cases of known resistance.
As farmers increasingly rely on flufenacet, an inhibitor of the synthesis of very-long-chain fatty acids (VLCFAs), for the control of A. myosuroides (Hull and Moss Reference Hull and Moss2012), the selection pressure on the treated populations has also increased. Although resistance to this herbicide MoA is considered to evolve slowly (Moss et al. Reference Moss, Ulber and den Hoed2019), some A. myosuroides populations have shown the potential to evolve resistance against inhibitors of VLCFA synthesis (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b; Hull and Moss Reference Hull and Moss2012; Rosenhauer and Petersen Reference Rosenhauer and Petersen2015). In A. myosuroides populations from northern Germany and several Lolium spp. populations, flufenacet was shown to be detoxified by glutathione conjugation (Dücker et al. Reference Dücker, Zöllner, Lümmen, Ries, Collavo and Beffa2019a, Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b). Flufenacet tolerance in crops such as maize (Zea mays L.) was also linked to enhanced glutathione transferase (GST) activity (Bieseler et al. Reference Bieseler, Fedtke, Neuefeind, Etzel, Prade and Reinemer1997). Yet neither the regulation of resistance genes nor individual isoforms detoxifying flufenacet have been described. GSTs, however, have been identified as enzymes that detoxify various other herbicides such as alachlor, atrazine, EPTC, or metolachlor (Brabham et al. Reference Brabham, Norsworthy, Houston, Varanasi and Barber2019; Evans et al. Reference Evans, O’Brien, Ma, Hager, Riggins, Lambert and Riechers2017; Gronwald and Plaisance Reference Gronwald and Plaisance1998; Timmerman Reference Timmerman1989). This type of metabolism-based herbicide resistance can lead to unpredictable cross-resistance to other herbicide MoAs and chemical classes (Yu and Powles Reference Yu and Powles2014) and can even confer resistance to herbicides not yet on the market (Busi et al. Reference Busi, Gaines, Walsh and Powles2012). As a result, fewer herbicides have become available for effective weed control, making the choice of a suitable compound more challenging. In addition, potential reductions of the registered field rates in Europe may increase the risk of evolution of non–target site resistance (Neve and Powles Reference Neve and Powles2005) and survival of populations already showing a shift in sensitivity to flufenacet in the field.
The use of compounds described as GST inhibitors can partially reverse resistance to herbicides detoxified by GST activity. Several compounds, including ethacrynic acid and tridiphane, were identified as inhibitors of plant GSTs (Ezra et al. Reference Ezra, Dekker and Stephenson1985; Lamoureux and Rusness Reference Lamoureux and Rusness1986). Various GST inhibition studies were conducted with different compounds. These were found to be active on isoforms identified in crops such as maize or soybean [Glycine max (L.) Merr.] (Li et al. Reference Li, Gao, Xu, Pang, Liu, Wang and Tan2017; Skipsey et al. Reference Skipsey, Andrews, Townson, Jepson and Edwards1997), but also on isoforms identified in weeds, such as AmGSTF1, a phi class GST frequently associated with metabolism-based resistance in A. myosuroides. It has been suggested to play a key role in signaling response to abiotic stress (Cummins et al. Reference Cummins, Wortley, Sabbadin, He, Coxon, Straker, Sellars, Knight, Edwards, Hughes, Kaundun, Hutchings, Steel and Edwards2013; Tétard-Jones et al. Reference Tétard-Jones, Sabbadin, Moss, Hull, Neve and Edwards2018). However, GST inhibitors have not yet been tested in combination with flufenacet.
In this study, we characterized four A. myosuroides populations with different levels of flufenacet sensitivity in a dose–response bioassay. We compared 14C-flufenacet degradation rates in two populations that exhibit contrasting responses to the herbicide in the absence and presence of different GST-inhibiting compounds and finally identified differentially expressed GSTs as candidate genes in an RNA sequencing (RNA-seq) experiment.
Materials and Methods
Dose-Response of Different Alopecurus myosuroides Populations to Flufenacet
Four populations were selected for a dose–response bioassay: the sensitive A. myosuroides populations ‘Herbiseed-S’ (Herbiseed, Twyford, UK) and ‘Appel-S’ (Appels Wilde Samen, Darmstadt, Germany); an A. myosuroides field population from the northern Germany marsh region of Kehdingen (Kehdingen1), which was previously described to be significantly less susceptible to flufenacet due to enhanced metabolism (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b); and another population from the same region, suspected to be more difficult to control (Kehdingen2). The seedlings were sown in tissue culture containers (MP Biomedicals, Eschwege, Germany) on 0.7% agar Type A (Sigma-Aldrich, Steinheim, Germany) containing KNO3 (Sigma-Aldrich) at a concentration of 0.02 M. The seeds were kept in the dark at 4 C for 5 d and were then placed in a greenhouse for germination. The greenhouse was set to 22/16 C day/night conditions with a 14-h photoperiod provided by Master HPI-T plus 400W/645 E40 metal-halide lamps (Philips, Amsterdam, The Netherlands) at approximately 200 µmol m−2 s−1. As the primordial root emerged, 15 seedlings per population and treatment were transplanted into pots containing sandy loam with 2.2% organic matter (three pots with five seedlings each). The seedlings were covered with a thin layer of coarse sand and treated on the same day with 500, 250, 125, 62.5, 31.3, 15.6, 7.8, 4.0, and 0 g flufenacet ha−1 formulated as Cadou® SC using a laboratory track sprayer with a single flat spray nozzle (TeeJet® nozzle XR8001, 300 L ha−1, 200 kPa). The sprayed seedlings were irrigated from above after treatment and grown for 3 wk under the previously described greenhouse conditions. The aboveground fresh weight was assessed, and the dose-response data were analyzed as a completely randomized experiment using the three-parameter log-logistic model of the drc package in R software (RStudio v. 3.5.0) (Ritz et al. Reference Ritz, Baty, Streibig and Gerhard2015). The equation with parameters b (relative slope around the inflection point), d (upper limit), and e (inflection point) was used (Equation 1). The experiment was repeated once.
$$f\left( x \right) = {d \over {1 + \exp \{b\left[ {\log \left( x \right) - \log \left( e \right)} \right]\}}}$$
Flufenacet Degradation in Alopecurus myosuroides Seedlings with High and Reduced Flufenacet Sensitivity after Treatment with Selected GST Inhibitors
The seedlings of the populations Herbiseed-S, Appel-S, Kehdingen1, and Kehdingen2 were grown in tissue culture containers as described earlier until the first leaf reached about 2.5 cm in length. Initially, a phytotoxicity test was conducted to determine symptom-free dose rates of flufenacet and the GST inhibitors tridiphane (Dr. Ehrenstorfer, Augsburg, Germany), chloro-7-nitrobenz-2-oxa-1,3-diazole (NBD-Cl, Sigma-Aldrich), triphenyltin chloride (TPT-Cl, abcr GmbH, Kalsruhe, Germany), ethacrynic acid (Alfa Aesar, Wand Hill, MA, USA), and diethyl maleate (Acros Organics, Beijing, China), as well as malathion, a frequently used inhibitor of cytochrome P450 monooxygenases (CYPs; Sigma-Aldrich). Then, 32 seedlings per inhibitor treatment and population were incubated in 20-ml scintillation vials containing 1.2 ml of commercial water (Volvic™) with KNO3 at a concentration of 0.02 M and 14C-radiolabeled flufenacet at a concentration of 15 µM and 16.7 kBq ml−1. Treatments were tridiphane (10 mM), NBD-Cl (10 µM), triphenyltin chloride (1 µM), ethacrynic acid (1 mM), diethyl maleate (100 µM), and malathion (10 mM), with a flufenacet-only control. These 32 seedlings were incubated for 24 h in a growth chamber at 22/16 C day/night conditions with a 14-h photoperiod provided by Master TL-D 58W/840 REFLEX fluorescent lamps (Philips) at approximately 400 µmol m−2 s−1. The seedlings were washed two times in water and one time in 50% acetone. Four times for each treatment, including the control, eight of these 32 seedlings were pooled to one sample to obtain sufficiently high signals, leading to four pooled samples in total. These were extracted and analyzed by high-performance liquid chromatography as described by Dücker et al. (Reference Dücker, Zöllner, Lümmen, Ries, Collavo and Beffa2019a). The percentages of metabolized flufenacet recovered from differently treated seedlings were analyzed with a Kruskal-Wallis one-way ANOVA on ranks using SigmaPlot v. 13.0 (Systat Software, San José, CA, USA). Differences between two populations with the same treatment were analyzed using a Mann-Whitney test included in the R software (RStudio v. 3.5.0). The experiment was repeated once, and the data were pooled for the analysis.
Illumina Transcriptome Sequencing of Alopecurus myosuroides Seedlings with High and Reduced Flufenacet Sensitivity
For an RNA-seq study, the seeds of the populations Herbiseed-S, Appel-S, Kehdingen1, and Kehdingen2 were sterilized for 5 min in 5% sodium hypochlorite and rinsed five times with sterile demineralized water. The seeds were dried on filter paper and sown in tissue culture containers (MP Biomedicals) under sterile conditions. The containers were filled with 30 ml of sterile 2-mm glass beads, 30 ml of sterile 4-mm glass beads, and 12 ml of sterile commercial water (VolvicTM) with KNO3 at a concentration of 0.02 M. The containers were kept in the dark at 4 C for 5 d and were then transferred into a climate chamber under the conditions described earlier. The seedlings were grown until the first leaf reached a length of about 2.5 cm. To assess constitutive differences between the biotypes, the untreated seeds were removed from the containers, and each six individual whole seedlings per population were immediately frozen at −80 C.
The frozen seedlings were ground in a Tissue Lyser II swing mill (Qiagen, Hilden, Germany) at 30 Hz for 30 s in 2-ml reaction tubes each containing three tungsten carbide beads (3 mm). The ground tissue was suspended in standard buffer (RLT) provided with the Qiagen RNeasy Kit (Qiagen), and total RNA extraction was performed according to the manufacturer’s instructions, including an on-column DNase treatment with an RNase-Free DNase Set (Qiagen). High-quality RNA was assured (RIN scores > 7) using the RNA 6000 Nano Kit (Agilent Technologies, Waldbronn, Germany) as defined in the manufacturer’s instructions. cDNA libraries were obtained using the Illumina TruSeq Stranded mRNA Library Prep kit (Illumina, San Diego, CA, USA). The multiplexed cDNA libraries were sequenced using an Illumina HiSeq 4000 sequencer (Illumina). Each library was measured on three lanes to obtain paired-end reads of 150-bp length in high-output mode, providing three technical replicates per biological replicate.
RNA-seq Analysis
The reads were demultiplexed using an in-house script by Fasteris (Geneva, Switzerland). The obtained paired-end reads were trimmed and mapped against an A. myosuroides reference transcriptome assembled with Illumina reads by Gardin et al. (Reference Gardin, Gouzy, Carrère and Délye2015) using BWA with the maximal exact matches (MEM) algorithm (Li Reference Li2013) (BWA v. 7.12) within the Genedata Expressionist Refiner Genome software (v. 9.5; Genedata, Basel, Switzerland). The reads were trimmed mean of M values (TMM) normalized in order to robustly equate the overall expression levels of genes between samples under the assumption that the majority of them are not differentially expressed (Robinson and Oshlack Reference Robinson and Oshlack2010). A pairwise differential gene expression analysis was conducted using the exact test of edgeR provided by the Blast2GO PRO software v. 4.1.9 (Conesa et al. Reference Conesa, Götz, García-Gómez, Terol, Talón and Robles2005). Candidate genes were selected with the following cutoff criteria: false discovery rate (FDR) ≤ 0.05 and log fold-change ≥ 2. The probable function of the candidate genes was annotated using the Basic Local Alignment Search Tool X (BLASTx; Camacho et al. Reference Camacho, Coulouris, Avagyan, Ma, Papadopoulos, Bealer and Madden2009) against the National Center for Biotechnology Information non-redundant database. Protein sequence similarities between the candidate contigs and AmGSTF1 (Cummins et al. Reference Cummins, Wortley, Sabbadin, He, Coxon, Straker, Sellars, Knight, Edwards, Hughes, Kaundun, Hutchings, Steel and Edwards2013) were determined through global pairwise alignments using the EMBOSS Needle algorithms (https://www.ebi.ac.uk/Tools/psa/emboss_needle).
Results and Discussion
To better understand the mechanisms of flufenacet detoxification in A. myosuroides, we used dose−response bioassays to characterize the four A. myosuroides populations, Herbiseed-S, Appel-S, Kehdingen1, and Kehdingen2 (Figure 1). The sensitive reference populations Herbiseed-S and Appel-S had effective dose rates 50 (ED50 values) of 4.3 ± 0.8 (P < 0.01) and 5.7 g ± 1.3 (P < 0.01) flufenacet ha−1, respectively. They were more susceptible than the northern Germany field populations Kehdingen1 and Kehdingen2, which had ED50 values of 14.4 ± 3.4 (P < 0.01) and 24.0 g ± 2.2 (P < 0.01) flufenacet ha−1, respectively. A 5.6-fold increase in herbicide rate required to inhibit growth by 50% was calculated for population Kehdingen2. The sensitivity of this population to flufenacet was significantly lower than the sensitivity of the reference populations Herbiseed-S and Appel-S. Kehdingen1, however, only differed significantly from Herbiseed-S. Sufficient control with the field rate of 250 g flufenacet ha−1 was indicated with ED90 values between 16.2 and 102.9 g flufenacet ha−1 for all populations tested under favorable conditions. This relatively low level of reduction in sensitivity is in accordance with previously published studies (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b; Hull and Moss Reference Hull and Moss2012; Rosenhauer and Petersen Reference Rosenhauer and Petersen2015) and may result from cross-resistance to other herbicide chemistries or may directly be selected by flufenacet applications. Reduced sensitivity to various herbicides was previously described for populations from the marsh regions of northern Germany, the origin of the populations characterized in this study (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b; Rosenhauer and Petersen Reference Rosenhauer and Petersen2015) and may result from non–target site resistance.
Figure 1. Dose–response analysis of the fresh weight of four Alopecurus myosuroides populations treated with different dose rates of flufenacet estimated using a three-parameter log-logistic model (see Equation 1).
Enhanced metabolism has previously been found as the dominant form of non–target site resistance to flufenacet in A. myosuroides as well as in Lolium spp. (Dücker et al. Reference Dücker, Zöllner, Lümmen, Ries, Collavo and Beffa2019a, Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b). The results of the experiment with different potentially synergistic inhibitors in this study point in the same direction. To test the effects of different chemicals on the performance of flufenacet, seedlings of the sensitive A. myosuroides population Herbiseed-S and the less susceptible field population Kehdingen1 were treated with 14C-radiolabeled flufenacet in the absence and presence of different inhibitors of GSTs and the insecticide and inhibitor of CYPs and acetylcholinesterase malathion. After 24 h, on average, 37.9% of unmetabolized flufenacet was recovered from the sensitive population Herbiseed-S and 21.9% of unmetabolized flufenacet was recovered from seedlings of the less susceptible population Kehdingen1, treated with flufenacet only (Figure 2). Based on a t-test (P ≤ 0.05), population Kehdingen1 degraded flufenacet significantly faster than population Herbiseed-S in all treatments, except for the treatment with malathion.
Figure 2. Flufenacet degradation 24 h after treatment with different inhibitors in the sensitive Alopecurus myosuroides population Herbiseed-S (A) and population Kehdingen1 with reduced flufenacet efficacy (B). Different letters indicate significant differences in flufenacet degradation between treatments, and asterisks (*) indicate significant differences between the two populations for each treatment (P ≤ 0.05).
A comparison of flufenacet degradation rates revealed that the detoxification in the populations Herbiseed-S and Kehdingen1 was affected to different extents by treatments with selected GST inhibitors and the CYP inhibitor malathion (Figure 2). Only seedlings of population Herbiseed-S treated with triphenyltin-chloride degraded flufenacet on average faster than seedlings of this population treated with flufenacet only. In all other treatments, the average degradation rate was slower in comparison to treatments with flufenacet only. While the differences were not statistically significant for seedlings treated with diethyl maleate and NBD-Cl, slower flufenacet degradation resulting from treatments with the GST inhibitors tridiphane and ethacrynic acid was statistically supported by a Kruskal-Wallis test. Interestingly, treatments with the CYP inhibitor malathion also led to significantly reduced flufenacet degradation rates in population Kehdingen1, although hydroxylated phase I metabolites were not found in flufenacet-treated A. myosuroides (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b). Similar inhibitory effects were also described for Lolium spp. (Parcharidou Reference Parcharidou2019) and in a recombinant phi-class GST originating from maize. In addition, the authors hypothesized that malathion binds to the substrate binding site of the tested GST isoform (Kapoli et al. Reference Kapoli, Axarli, Platis, Fragoulaki, Paine, Hemingway, Vontas and Labrou2008). Diverse GST-inhibitory effects of malathion have previously been observed in other organisms such as rats (Hazarika et al. Reference Hazarika, Sarkar, Hajare, Kataria and Malik2003). However, further investigations are required to clarify whether the observed effect is a result of direct inhibition of GSTs or due to other effects of malathion.
In general, these results demonstrate the presence of an interaction between GST inhibitors and flufenacet as previously described for other herbicides detoxified by GSTs (Cummins et al. Reference Cummins, Wortley, Sabbadin, He, Coxon, Straker, Sellars, Knight, Edwards, Hughes, Kaundun, Hutchings, Steel and Edwards2013; Ezra et al. Reference Ezra, Dekker and Stephenson1985). Moreover, they support previous findings by Dücker et al. (Reference Dücker, Zöllner, Lümmen, Ries, Collavo and Beffa2019a, Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b) suggesting enhanced GST activity as a key driver for a reduction in flufenacet sensitivity. The use of GST inhibitors as synergists for field applications, however, is generally problematic, as compounds such as ethacrynic acid inhibit GST classes expressed in humans, for example, alpha or pi class (Huang et al. Reference Huang, Huang, Yeh, Lin and Yu2015) as well as plant-specific GSTs such as isoforms belonging to the phi or tau class.
While the inhibition of flufenacet degradation by tridiphane and ethacrynic acid consolidates the formulated hypothesis that flufenacet is detoxified by isoform(s) belonging to the superfamily of GSTs, it is not clear which isoforms of which class(es) are involved in the reduction of flufenacet sensitivity. Altogether 47 GST isoforms belonging to the classes tau (28), phi (13), theta (3), zeta (2), and lambda (2) were identified in the genome of Arabidopsis thaliana (Wagner et al. Reference Wagner, Edwards, Dixon and Mauch2002), and 79 putative GSTs belonging to the classes tau (52), phi (17), zeta (4), DHAR (2), EF1G (2), theta (1) and TCHQD (1) were identified in rice (Oryza sativa L.) (Jain et al. Reference Jain, Ghanashyam and Bhattacharjee2010). The large number of GST isoforms, as well as sequence similarities, complicates the identification of candidate GSTs in A. myosuroides with traditional methods like real-time PCR. This may particularly apply to the larger classes tau and phi, which have probably undergone extensive gene-duplication events after divergence of monocotyledonous and dicotyledonous plants (Monticolo et al. Reference Monticolo, Colantuono and Chiusano2017). Therefore, an RNA-seq approach was chosen for identification of candidate genes conferring reduction in flufenacet sensitivity in A. myosuroides by investigating differences in constitutive gene expression among populations with a shift in sensitivity.
A differential gene expression analysis was used to identify transcriptomic differences between the sensitive populations Herbiseed-S and Appel-S and the less susceptible populations Kehdingen1 and Kehdingen2. This revealed 319 gene-associated contigs with significantly higher expression and 218 gene-associated contigs with significantly lower expression as potential candidate genes conferring differences in flufenacet susceptibility. Annotations of detoxification-associated functions were most abundant among the set of more highly expressed contigs, indicating a constitutive upregulation of detoxification pathways. In addition to 15 glucosyltransferases and 7 CYPs, 6 GSTs (7 contigs) were found among the most frequent annotations (Figure 3). Furthermore, 9 significantly more highly expressed contigs were annotated as different transcription factors. Additionally, further transcription- and translation-related functions such as zinc finger domains, several kinases, a ribosome-binding factor, a mitochondrial RNA helicase, and chaperones were significantly more highly expressed. The higher expression of this group of genes indicates a complex change in gene expression patterns in the less sensitive populations that may play a role in the reduced sensitivity in the populations of interest.
Figure 3. Most frequent BLASTx annotations of contigs with differential expression. (A) Most frequent annotations with higher expression levels in Alopecurus myosuroides populations from Kehdingen, Germany, including a β-ketoacyl-CoA reductase1 (KCR1); and (B) most frequent annotations with higher expression levels in A. myosuroides populations from Kehdingen.
Interestingly, one of the significantly more highly expressed contigs, although with relatively low TMM values, was annotated as β-ketoacyl-CoA reductase1 (KCR1). KCR1 is part of the VLCFA elongase complex, which additionally consist of a enoyl-CoA-reductase, a β-hydroxyacyl-CoA dehydratase, and ketoacyl-CoA-synthases, the family of enzymes that represents the putative site of action of flufenacet and catalyze the first and rate-limiting step of the elongation of VLCFAs (Haslam and Kunst Reference Haslam and Kunst2013; Trenkamp et al. Reference Trenkamp, Martin and Tietjen2004). This observation leads to the hypothesis that a higher production rate of VLCFAs may lead to reduced sensitivity of weed populations treated with herbicides inhibiting the synthesis of VLCFAs. The combination of higher flufenacet detoxification rates and possibly a change in the VLCFA elongation complex can lead to a decrease in flufenacet sensitivity.
Among the set of 218 contigs with significantly lower expression levels, the largest group of contigs with similar BLASTx annotations contained contigs coding for receptor-like kinases, which may play a role in signal transduction or the regulation of transcription factors. In addition, each three of these contigs were annotated as mitochondrial elongation factors and GSTs, while all other annotations occurred less frequently. Contigs annotated as the typical housekeeping gene actin (e.g., contigs alomy11359 or alomy027178) were not among the set of differentially expressed genes and typically showed homogenous expression patterns.
Based on the results of the interaction study with GST inhibitors (Figure 2) and previous findings on the detoxification pathway of flufenacet in A. myosuroides (Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b), the GSTs with significantly higher expression levels in population Kehdingen2 were selected as candidate genes. The open reading frames of seven GSTs with significantly higher levels of expression were identified. Based on their amino acid sequences, four were identified as GSTs belonging to class tau, two as phi-class GSTs, and one as a theta-class GST (Table 1). As two of the contigs coding for tau-class GSTs had an amino acid similarity of 100%, only the longer contig was chosen for further analysis.
Table 1. Similarities of amino acid sequences of AmGSTF1 and the contigs of three tau-class glutathione transferases (GST1, GST2, GST3), two phi-class GSTs (GST4, GST5), and a theta-class isoform (GST6) with their contig names as published in the reference transcriptome by Gardin et al. (Reference Gardin, Gouzy, Carrère and Délye2015).
The analysis of the expression patterns of the six candidate GSTs revealed that all of them were more highly expressed in populations Kehdingen1 and Kehdingen2. In all six cases, the level of expression was significantly higher in population Kehdingen2 in comparison to the sensitive population Herbiseed-S. Only the level of expression of GST2 (tau) was also significantly higher in population Kehdingen1 in comparison to the sensitive populations. This suggests a generally higher expression level of detoxification-related genes or pathway(s) in population Kehdingen2 and is in accordance with the resistance level characterized in the dose–response bioassay. However, the expression levels differed from contig to contig. While GST1 (tau), GST2 (tau), GST4 (phi), and GST5 (phi) achieved relatively high expression levels, particularly in individuals of population Kehdingen2, GST3 (tau) and GST6 (theta) were expressed at low levels, with TMM values below 100 (Figure 4). Finally, AmGSTF1, which is frequently associated with metabolic herbicide resistance in A. myosuroides, was not among the candidate GSTs (sequence similarities <44.5% with local alignment; Table 1). This isoform was the only commonly upregulated gene in a set of multiple-resistant A. myosuroides analyzed in an RNA-seq experiment conducted by Tétard-Jones et al. (Reference Tétard-Jones, Sabbadin, Moss, Hull, Neve and Edwards2018), indicating upregulation of different pathways in different multiple-resistant populations. Yet all of these genes may play minor or major roles in flufenacet detoxification by glutathione conjugation, as indicated by the inhibitor tests and the metabolites identified in A. myosuroides and flufenacet-tolerant crops (Bieseler et al. Reference Bieseler, Fedtke, Neuefeind, Etzel, Prade and Reinemer1997; Dücker et al. Reference Dücker, Zöllner, Parcharidou, Ries, Lorentz and Beffa2019b).
Figure 4. Expression of three tau-class glutathione transferases (GST1, GST2, GST3), two phi-class GSTs (GST4, GST5), and a theta-class isoform (GST6) differentially expressed in the sensitive Alopecurus myosuroides populations Herbiseed-S and Appel-S and the populations Kehdingen1 and Kehdingen2 with reduced flufenacet efficacy. Different letters indicate significant differences between populations (false discovery rate ≤ 0.05). TMM, trimmed mean of M values.
However, in addition to their classical function of glutathione conjugation and their role in detoxification, GSTs can fulfill several other roles in plants that may indirectly decrease the activity of herbicides in A. myosuroides plants. These roles include their function as ligandins (e.g., for the auxin indole-3-acetic acid; Bilang and Sturm Reference Bilang and Sturm1995; Sylvestre-Gonon et al. Reference Sylvestre-Gonon, Law, Schwartz, Robe, Keech, Didierjean, Dubos, Rouhier and Hecker2019) and additionally include their glutathione-dependent hyperoxidase activity (Axarli et al. Reference Axarli, Dhavala, Papageorgiou and Labrou2009). The catalytic and noncatalytic functions of GSTs are involved in tolerance to abiotic stresses, regulation of antioxidants, pathogen defense, or signaling (Cummins et al. Reference Cummins, Wortley, Sabbadin, He, Coxon, Straker, Sellars, Knight, Edwards, Hughes, Kaundun, Hutchings, Steel and Edwards2013; Gullner et al. Reference Gullner, Komives, Király and Schröder2018; Kumar and Trivedi Reference Kumar and Trivedi2018; Marrs Reference Marrs1996; Nianiou-Obeidat et al. Reference Nianiou-Obeidat, Madesis, Kissoudis, Voulgari, Chronopoulou, Tsaftaris and Labrou2017). Thus, it has been confirmed that various plant GSTs are induced by environmental factors such as pathogens, xenobiotics, metals, drought, and cold, as well as phytohormone production or oxidative stress, which typically accompany the environmental factors (Gullner et al. Reference Gullner, Komives, Király and Schröder2018; Lallement et al. Reference Lallement, Brouwer, Keech, Hecker and Rouhier2014; Marrs Reference Marrs1996; Nianiou-Obeidat et al. Reference Nianiou-Obeidat, Madesis, Kissoudis, Voulgari, Chronopoulou, Tsaftaris and Labrou2017; Sylvestre-Gonon et al. Reference Sylvestre-Gonon, Law, Schwartz, Robe, Keech, Didierjean, Dubos, Rouhier and Hecker2019).
The regulation of GST expression in plants can be complex, as multiple transcription start points can be present (Thatcher et al. Reference Thatcher, Carrie, Andersson, Sivasithamparam, Whelan and Singh2007) and multiple regulatory elements can be present in the promoter regions of GSTs. For example, in the promoter region of the rice GST OsGSTL2, several stress-regulated cis elements such as Box-W1, EIRE, or LTR or elements responding to phytohormones such as TCA, CGTCA-motif, or ERE have been found. Similarly, the presence of ABRE and MYB in the promoter region of a tau-class GST of the succulent Salicornia brachiata Roxb. (Tiwari et al. Reference Tiwari, Patel, Chaturvedi, Mishra and Jha2016) indicates concurrent binding of different transcription factors (Hu et al. Reference Hu, He, Yang, Zeng, Wang, Chen and Huang2011).
The regulation of GST expression in weeds has so far only been described in a limited number of detailed studies (Brazier-Hicks et al. Reference Brazier-Hicks, Knight, Sellars, Steel and Edwards2018; Wei et al. Reference Wei, Zhu, Liu, Zhang, Zhu, Xu, Lin, Lu and Li2019). There is, however, evidence that posttranscriptional modifications play a role in resistance-related gene expression (Nandula et al. Reference Nandula, Riechers, Ferhatoglu, Barrett, Duke, Dayan, Goldberg-Cavalleri, Tétard-Jones, Wortley, Onkokesung, Brazier-Hicks, Edwards, Gaines, Iwakami, Jugulam and Ma2019; Tétard-Jones et al. Reference Tétard-Jones, Sabbadin, Moss, Hull, Neve and Edwards2018).
The regulation of GST expression, in particular the roles of the significantly more highly and less expressed transcription factors and their potential to interact with DNA binding domains for the expression of the identified candidate GSTs, remains to be investigated in the A. myosuroides populations studied here, which would require analyses of the genomic sequences. Further analysis of epigenetic modifications (e.g., methylation) of the GST promoters, which could modify gene expression (Gressel Reference Gressel2009; Schnekenburger et al. Reference Schnekenburger, Karius and Diederich2014), is also required.
Differentially expressed genes such as the set of receptor-like kinases with significantly lower expression levels may play a role in differential flufenacet susceptibility. With more than 600 isoforms identified in Arabidopsis and more than 1,100 isoforms identified in rice, these kinases belong to the largest family of receptors in plants (Shiu et al. Reference Shiu, Karlowski, Pan, Tzeng, Mayer and Li2004). The functions of these signaling proteins range from pathogen response via morphological development to functions involved in processes such as self-incompatibility (Morillo and Tax Reference Morillo and Tax2006; Shpak et al. Reference Shpak, Berthiaume, Hill and Torii2004; Stein et al. Reference Stein, Howlett, Boyes, Nasrallah and Nasrallah1991; Tang et al. Reference Tang, Wang and Zhou2017). Receptor-like kinase-mediated signaling can activate and repress signaling pathways and is affected by multiple interactions of extracellular domains (Jaillais et al. Reference Jaillais, Belkhadir, Balsemão-Pires, Dangl and Chory2011; Smakowska-Luzan et al. Reference Smakowska-Luzan, Mott, Parys, Stegmann, Howton, Layeghifard, Neuhold, Lehner, Kong, Grünwald, Weinberger, Satbhai, Mayer, Busch and Madalinski2018). While some kinases can affect gene expression (e.g., by phosphorylation of transcription factors; Sirichandra et al. Reference Sirichandra, Davanture, Turk, Zivy, Valot, Leung and Merlot2010), the role of these differentially expressed receptor-like kinases remains unclear.
Finally, the reduction of flufenacet detoxification in the presence of the GST inhibitors ethacrynic acid and tridiphane and the detection of six significantly more highly expressed GSTs provide further evidence concerning the role of GSTs in the detoxification of flufenacet and, thereby, the similarity of mechanisms in crops and grass weeds. Additional research is needed to validate the detected candidate genes involved in reduced flufenacet sensitivity in A. myosuroides. The availability of full cDNA sequences and genomic sequences will allow characterization of the biochemical function of the selected genes as well as a better understanding of regulation of their expression (Ravet et al. Reference Ravet, Patterson, Krähmer, Hamouzová, Fan, Jasieniuk, Lawton-Rauh, Malone, McElroy, Merotto and Westra2018). This will provide a better understanding of the evolution of herbicide resistance and the prediction of cross-resistance patterns. The implementation of agricultural practices to slow down or even avoid the evolution of flufenacet resistance is becoming increasingly important. This is particularly the case because the maximum field rate of flufenacet may be significantly reduced, as proposed in Europe, increasing the potential risk of sublethal dose rates selecting for non–target site resistance in the future. The knowledge of the detoxification pathways of the different herbicides and the genes involved will help to (1) set specific resistance diagnostics and (2) define the best mixture strategies, either by application of tank mixtures, premixed products, or sequential treatments.
Acknowledgments
The authors would like to thank Alberto Collavo and Joachim Kaiser for providing seed material and Falco Peter, Julia Unger, and Thomas Schubel for support with bioassays and sample shipment. The authors would like to express their great appreciation to Thomas Wolf, Michael Kohnen, Hans-Jürgen Albrecht, and Francesco Pulitano for providing the computational basis for bioinformatic analyses. Finally, the authors thank Andreas von Tiedemann, Lothar Lorentz, Anita Küpper, and Ralf Nauen for scientific discussions and Bayer AG, Crop Science Division for funding and access to equipment. No conflicts of interest have been declared.
Open access
