Eleven spontaneous mutations of acetyl-CoA carboxylase have been identified in many herbicide-resistant populations of 42 species of grassy weeds, hampering application of aryloxyphenoxypropionate, cyclohexadione and phenylpyrazoline herbicides in agriculture.
IC50 shifts (resistance indices) caused by herbicide-resistant mutations were determined using a recombinant yeast system that allows comparison of the effects of single amino acid mutations in the same biochemical background, avoiding the complexity inherent in the in planta experiments. The effect of six mutations on the sensitivity of acetyl-CoA carboxylase to nine herbicides representing the three chemical classes was studied.
A combination of partially overlapping binding sites of the three classes of herbicides and the structure of their variable parts explains cross-resistance among and between the three classes of inhibitors, as well as differences in their specificity. Some degree of resistance was detected for 51 of 54 herbicide/mutation combinations.
Introduction of new herbicides targeting acetyl-CoA carboxylase will depend on their ability to overcome the high degree of cross-resistance already existing in weed populations.
Aryloxyphenoxypropionate (fop), cyclohexadione (dim) and phenylpyrazoline (pinoxaden) herbicides targeting plastid acetyl-CoA carboxylase (ACCase) have been widely used to control grassy weeds. The successful application of these herbicides is significantly hampered by herbicide-resistant weed populations of 42 species (Heap, 2012). Eleven mutations affecting ACCase herbicide sensitivity have been identified (Table 1): I1781L and I1781V, W1999C and W1999L, W2027C, I2041N and I2041V, D2078G, C2088R, and G2096A and G2096S (amino acid residue numbering as in the Alopecurus myosuroides ACCase) with nontarget-site resistance further aggravating the problem (Delye, 2005; Delye et al., 2005, 2008; Zhang & Powles, 2006a,b; Liu et al., 2007; Yu et al., 2007; Kaundun, 2010; Petit et al., 2010; Powles & Yu, 2010; Collavo et al., 2011; Cruz-Hipolito et al., 2011; Scarabel et al., 2011; Beckie & Tardif, 2012).
Table 1. Amino acid residues (G, A, I, L, S & V) involved in herbicide resistance (shaded in green) and binding (conservative substitutions are indicated by orange shading)
All of the known herbicide-resistant mutations occur in the carboxyl transferase (CT) domain of the multidomain plastid ACCase found in grasses (Fig. 1). Plastids of dicotyledonous plants contain herbicide-resistant multisubunit ACCases (Tong, 2012). The multidomain eukaryotic ACCases (plant cytosolic, yeast, human) are mostly resistant to fops and dims. The herbicide-binding sites have been determined by X-ray crystallography of their complexes with the yeast CT domain (Zhang et al., 2004; Xiang et al., 2009; Yu et al., 2010). Fourteen of the 32 amino acid residues found in the partially overlapping binding sites are conserved in all species. Conservative substitutions are allowed for the remaining residues (Table 1). The fitness penalty for mutating the conserved binding-site residues could be high, if not lethal, decreasing ACCase activity below the threshold level required for de novo fatty acid synthesis and plant growth. Only three herbicide-resistant mutations are found for the binding-site residues: residue 1781 in the binding sites of all three classes of herbicides, and 1999 and 2041 in the fop-binding site (Table 1). The remaining four residues that give rise to herbicide resistance are located in the immediate vicinity of 1781, 1999 or 2041 (Fig. 1). The herbicide-binding sites overlap to different degrees with the acetyl-CoA binding site in the CT active site.
The herbicide-resistant mutations have been studied in several grass species, both in planta and using ACCase enzymatic assays (Delye, 2005; Delye et al., 2005, 2008; Zhang & Powles, 2006a,b; Liu et al., 2007; Yu et al., 2007; Kaundun, 2010; Petit et al., 2010; Powles & Yu, 2010; Collavo et al., 2011; Cruz-Hipolito et al., 2011; Scarabel et al., 2011). Although a link between the mutations and plant herbicide resistance has been established, analysis of plant resistance was often complicated by a lack of proper characterization of the resistant plant populations. Several studies have addressed these problems (Liu et al., 2007; Yu et al., 2007; Delye et al., 2008; Kaundun, 2010; Petit et al., 2010; Scarabel et al., 2011). Homozygous mutants have a higher degree of resistance than heterozygous mutants, leading to an increased survival rate at higher herbicide doses. Multiple resistance mutations in a single population or individual plant affect both the response to specific herbicides and their application rates in field tests. Sequences of all ACCase homoeologs have to be analyzed for possible mutations. ACCase sensitivity to inhibitors has been established as the major molecular determinant of the plant response to herbicide application in the field, but there are many steps that affect the agricultural outcome in a species-dependent manner, such as effective herbicide concentration and formulation, foliar uptake and penetration, translocation and transport to plastids, and finally chemical stability, modification and detoxification in plants.
Here we report the effect of six of the 11 mutations on the sensitivity of wheat (Triticum aestivum) plastid ACCase towards five fops, three dims and pinoxaden, studied by the comparison of IC50 shifts (resistance indices, RIs) determined using yeast ACCase gene-replacement strains to avoid the complexity inherent in the whole-plant experiments.
Materials and Methods
Preparation and application of yeast gene-replacement strains that depend on foreign ACCases for growth to study the mode of action and sensitivity to inhibitors (to measure IC50 values), in vitro measurements of ACCase activity and pulse-labeling of de novo synthetized fatty acids in wheat plants were described previously (Gornicki & Haselkorn, 1993; Joachimiak et al., 1997; Nikolskaya et al., 1999; Zagnitko et al., 2001; Jelenska et al., 2002; Liu et al., 2007; Marjanovic et al., 2010). Yeast strains were grown in 96-well culture plates with herbicide concentrations ranging from 0.001 to 100 μM. Cell density was measured at 600 nm. Measurements for the time-point when the yeast density in the absence of the inhibitor was between OD 0.8 and 1.1 reflected inhibition of the ACCase most closely. Conversion of 14C-NaHCO3 ito acid-stable malonyl-CoA was measured in the enzymatic ACCase assay with herbicide concentrations ranging from 0.001 to 30 μM. Additional information is provided in Supporting Information, Methods S1.
The yeast ACC1 null mutation can be complemented by expression of ACCases from other species (Joachimiak et al., 1997; Nikolskaya et al., 1999; Zagnitko et al., 2001; Jelenska et al., 2002; Liu et al., 2007; Marjanovic et al., 2010). We have shown that IC50 values measured by growth inhibition of the resulting gene-replacement yeast strains and by ACCase in vitro enzymatic assay are very similar for many classes of compounds. We have prepared gene-replacement yeast strains with chimeric wheat ACCase in which the herbicide-binding domain comes from the herbicide-sensitive plastid isozyme (strain C50P50) and six mutants carrying single amino acid substitutions corresponding to ACCase spontaneous mutations found in different populations of grassy weeds resistant to fops and/or dims selected after agricultural applications of the herbicides (strains named after the mutations). Another three ACCases with mutations found in resistant grasses did not complement the yeast mutant; these were not studied further.
Fitness and resistance
The growth properties of the gene-replacement yeast strains depend on the ACCase variant (Table S1). The doubling time of strain C50P50 is longer than that of the strain carrying full-length wheat cytosolic ACCase (C100) and the corresponding haploid wildtype yeast, indicating that activity of the chimeric ACCase is growth-limiting. Doubling times of strains Q1756E, I1781V, W2027C and I2041N are similar to the doubling time of C50P50: apparently these mutations do not lower the ACCase activity further. Doubling times of strains D2078G and I1781V are shorter. Prolonged propagation of the yeast gene-replacement strains was avoided in order to prevent selection of faster-growing secondary mutants (Marjanovic et al., 2010), but it is possible that the shorter doubling time for strain D2078G is caused by such mutation(s). Similar fast-growth phenotype was observed for multiple I1781L strains (Zagnitko et al., 2001). The W1999C strain grows significantly slower than C50P50, suggesting that Trp at position 1999 is essential for full enzymatic activity. Mutation W1999S did not complement the yeast ACC1 null mutation, indicating that Ser at this position reduces the ACCase activity to a level insufficient to sustain fast-growing yeast. ACCase with the G2096A and C2088R mutations did not complement the null mutation (Liu et al., 2007), probably for the same reason.
Mutations at residues 2078 and 2088, but not at residue 1781, have been previously associated with a fitness penalty imposed on herbicide-resistant plants (Delye, 2005; Yu et al., 2007; Powles & Yu, 2010). Our results suggest that in addition to C2088R, mutations W1999C, W1999S, and G2096A lower ACCase activity and could impose a fitness penalty on herbicide-resistant plants as well. Our result for the 2078 mutant cannot be interpreted unequivocally (see earlier).
The Q1756E mutation was found in some Lolium rigidum biotypes where it was proposed to play a secondary role in sethoxydim resistance (Zhang & Powles, 2006a). Our results show that this mutation, located in the yeast CT structure > 30Å from the herbicide-binding sites, alone is not sufficient to make ACCase resistant to any of the herbicides tested – RIs for Q1756E vary between 0.8 and 2.1 (Table 2). The RIs for other mutations and herbicides (Table 2) are equal to or larger than 5, with four exceptions, all for the W1999C mutation: RI of 1.1 ± 0.3 for clodinafop and 2 ± 1 for sethoxydim show no effect on sensitivity, and 3 ± 1 for tralkoxydim suggests only a marginal resistance, while 0.3 ± 0.1 for haloxyfop indicates possible increased sensitivity.
Table 2. Resistance index (± SE) for seven ACCase mutations and cytosolic ACCase (C100) relative to wheat plastid ACCase
In all cases other than W1999C, mutated ACCase becomes resistant to all the herbicides, but the degree of resistance depends on the herbicide structure (Table 2). All of the spontaneous herbicide-resistant mutations analyzed here occurred following repeated applications of dims or fops, but not of pinoxaden, which was introduced to agriculture more recently. A combination of partially overlapping binding sites of fops, dims and pinoxaden (Fig. 1, Table 1), and the structure of the variable parts of fops and dims (Fig. S1) explain both the cross-resistance and differences in specificity, as discussed in the following section.
Correlation between inhibition in vivo and in vitro
Wheat germ extract was used to study herbicide inhibition of wheat plastid ACCase in vitro. The IC50 values, from the low of 0.07 μM for quizalofop to the high of 3.3 μM for haloxyfop (Table S1), were within the ranges previously reported for wildtype plastid ACCases from various grasses (Delye, 2005; White et al., 2005; Yu et al., 2007; Cruz-Hipolito et al., 2011). The in vitro IC50 values for pinoxaden and fops are very similar to the corresponding values measured in vivo using the gene-replacement strain, but for dims, the in vivo values are significantly lower (Table S1). The molecular mechanism of this increased sensitivity is unknown.
High herbicide sensitivity of fatty acid biosynthesis in vivo was previously observed for wheat, where haloxyfop IC50 values measured in vitro using wheat germ and partially purified chloroplast ACCase were 4 and 2 μM, respectively, but IC50 for fatty acid synthesis in leaves was one order of magnitude lower (Gornicki & Haselkorn, 1993). Fatty acid synthesis was measured by incorporation of radiolabeled acetate taken up by leaf cuts in a pulse-labeling experiment. Strong inhibition of fatty acid synthesis indicates that the herbicides are efficiently translocated by transpiration and then transported to chloroplasts where de novo fatty acid synthesis occurs. Foliar uptake is required in agricultural application of the herbicides. The IC50 values for in planta fatty acid biosynthesis are substantially lower than the IC50 values for ACCase measured in vitro (Table S1). The molecular mechanism of this apparent increase in sensitivity is unknown.
Our findings are in agreement with the consensus pattern of herbicide resistance derived from earlier studies (Table 2). With the wide dynamic range of the IC50 shift measurements (better than 100-fold for most herbicides), we can correlate herbicide structure (Fig. S1) with the location of the mutated amino acid residues relative to the herbicide-binding site (Fig. 1) and RI imposed by a particular mutation (Table 2).
The ACCase RI range underlying plant resistance observed in whole plants and in the field is extremely broad. Biotypes with the I1781L mutation exhibit resistance to fops, dims and pinoxaden (Zagnitko et al., 2001; Christoffers et al., 2002; Delye, 2005; White et al., 2005; Liu et al., 2007; Beckie & Tardif, 2012). Their RIs (Table 2) vary from 6 to 17. For the I1781V mutant, RIs vary from 8 to 35. The twofold increase in IR for sethoxydim and tralkoxydim, the only difference compared with I1781L (Table 2), may be sufficient to explain the higher resistance of Phalaris paradoxa carrying the I1781V mutation to dims (Collavo et al., 2011). At the other end of the spectrum, all of the RIs for D2078G exceed 100 (Table 2). Biotypes with this mutation are resistant to all three classes of herbicides (Delye, 2005; Liu et al., 2007; Yu et al., 2007; Kaundun, 2010; Collavo et al., 2011; Cruz-Hipolito et al., 2011; Beckie & Tardif, 2012).
I1781 is part of the binding site of all three classes of herbicides, which explains the cross-resistance and the low RIs caused by replacement of Ile with Leu or Val. D2078 is not a part of either herbicide-binding site, but it is located next to and in contact with I1781. Substituting Asp with Gly at position 2078 can be expected to affect binding of all herbicides with larger RIs as a result of a more substantial structural change.
The W1999C mutation has been reported to affect sensitivity to fenoxaprop (Liu et al., 2007; Beckie & Tardif, 2012), consistent with the RI of 86 (Table 2). Quizalofop resistance has not been tested on W1999C mutant plants, but we recorded an RI of 122 (Table 2). W1999 contributes to the fop-binding site. Fenoxaprop and quizalofop have large two-ring aromatic moieties (Fig. S1) pointing towards W1999 (Fig. 1), explaining why the W1999C mutation affects their binding. Sensitivity to clodinafop is not affected and resistance to diclofop is only modest (RI = 6; Table 2). The W1999C mutation increases sensitivity to haloxyfop (RI = 0.3; Table 2), which differs from clodinafop and diclofop primarily by the presence of a 3-fluoromethyl group (Fig. S1), which is in close proximity to W1999 (Fig. 1). RIs for the three dims are low (2–5) (Table 2), which is consistent with the position of their binding site at a distance from W1999 (Fig. 1). The pinoxaden-binding site is also further away from W1999; the W1999C mutation leads to only moderate resistance to pinoxaden (RI = 7; Table 2).
The W2027C mutation has the strongest effect on sensitivity to fops, with RI of 39 for haloxyfop and over 200 for the remaining fops (Table 2), consistent with the high resistance of W2027C mutant biotypes to fops. RIs for dims and pinoxaden are much lower (8–15; Table 2), consistent with the greater distance between their binding sites and the W1999/W2027 stack (Fig. 1). W2027C mutant biotypes have been reported to be resistant to pinoxaden but either resistant or sensitive to dims (Delye, 2005; Liu et al., 2007; Yu et al., 2007).
A similar pattern of resistance is observed for I2041N: RIs > 100 for fops, 13 for pinoxaden and 5–7 for dims (Table 2). Residue I2041 contributes to the fop-binding site, which explains the strong effect of the Ile to Asn substitution on sensitivity to these herbicides. I2041N mutant biotypes are resistant to fops but either resistant or sensitive to pinoxaden and dims (Delye, 2005; Zhang & Powles, 2006b; Liu et al., 2007; Petit et al., 2010; Cruz-Hipolito et al., 2011; Beckie & Tardif, 2012).
An additive effect of amino acid substitutions at residues critical for herbicide binding explains sensitivity/resistance of other eukaryotic multidomain ACCase. Two amino acid differences, I1781L and D2078G (Table 1), are sufficient to explain resistance of wheat cytosolic ACCase to most of the herbicides tested. Based on the yeast test, wheat cytosolic ACCase (strain C100) is sensitive to only tepraloxydim and modestly to pinoxaden (Table S1).
Toxoplasma gondii ACC1 (apicoplast ACCase) differs from the sensitive grass ACCases at only two of the binding site/resistance residues: Leu and Met, at residues equivalent to wheat plastid ACCase residues 1781 and 2088, respectively (Table 1). The IC50 for T. gondii ACC1 inhibition with haloxyfop and clodinafop is c. 5 μM (Zuther et al., 1999; Jelenska et al., 2002), significantly higher than the corresponding IC50 for wheat plastid ACCase, but similar to the IC50 of the I1781L mutant, 3.2 and 7.5 μM, respectively (Table S1). Met at the residue equivalent to 2088 is clearly not critical for fop sensitivity, but may contribute to T. gondii ACC1 resistance to dims. Arg at position 2088 causes resistance to all three classes of herbicides (Powles & Yu, 2010).
The yeast ACCase is resistant to all three classes of herbicides, as indicated by the lack of growth inhibition of wildtype yeast up to 100 μM (Joachimiak et al., 1997; Nikolskaya et al., 1999) and their weak binding to the yeast CT domain (Zhang et al., 2003, 2004; Xiang et al., 2009; Yu et al., 2010). Yeast ACCase contains three resistance-causing residues, L1781, V2041 and A2096 (Table 1). We have not studied I2041V and G2096A mutants in our yeast system. These mutations were reported to cause resistance to fops but not dims or pinoxaden (Powles & Yu, 2010). Met found in yeast at a residue equivalent to grass residue 2088 probably adds to dim resistance, as suggested by the properties of T. gondii ACC1 (see earlier). Finally, human ACC1 (cytosolic ACCase) and ACC2 (mitochondrial ACCase), which have Ile, Val and Met at positions equivalent to residues 1781, 2041 and 2088, respectively, are resistant to fops and dims (Marjanovic et al., 2010).
Many mutations in the herbicide-binding site or in its immediate vicinity cause different degrees of cross-resistance among and between the three classes of inhibitors. At least some degree of resistance was detected for 51 of 54 herbicide–mutation combinations (Table 2). Resistant populations that arose in many areas of the globe as a result of extensive use of fops and/or dims may be difficult to combat with new chemical classes of herbicides targeting the same enzyme with the same mode of action, as the pinoxaden case illustrates (Kaundun, 2010; Petit et al., 2010; Collavo et al., 2011; Cruz-Hipolito et al., 2011; Scarabel et al., 2011; Beckie & Tardif, 2012). Pinoxaden was introduced in 2006, some 30 yr after introduction of fops and dims. It belongs to a different chemotype, but has the same mode of action. Many weed populations harbor mutant ACCases resistant to pinoxaden with RI in the range of 7–120, for the six mutations reported here (Table 2). Heavy use of pinoxaden, in turn, will accelerate selection of multiple mutants and the spread of resistance to fops and dims. The utility of new herbicides targeting ACCase will depend on their ability to overcome the high degree of cross-resistance already existing in weed populations. Inhibitors targeting ACCase parts other than the CT domain have been developed (Marjanovic et al., 2010; Tong, 2012). Our yeast system can facilitate the discovery and analysis of new herbicides targeting ACCase.
We thank R. Haselkorn (University of Chicago) for support, advice and help in manuscript preparation. We thank Rafael G. Galdames (Unidad de Biotecnología INIA Carillanca, Temuco, Chile) for discussions and help with initial experiments.