CYP81A P450s are involved in concomitant cross‐resistance to acetolactate synthase and acetyl‐CoA carboxylase herbicides in Echinochloa phyllopogon

Summary

• Californian populations of Echinochloa phyllopogon have evolved multiple‐herbicide resistance (MHR), posing a threat to rice production in California. Previously, we identified two CYP81A cytochrome P450 genes whose overexpression is associated with resistance to acetolactate synthase (ALS) inhibitors from two chemical groups. Resistance mechanisms to other herbicides remain unknown.
• We analyzed the sensitivity of an MHR line to acetyl‐CoA carboxylase (ACCase) inhibitors from three chemical groups, followed by an analysis of herbicide metabolism and segregation of resistance of the progenies in sensitive (S) and MHR lines. ACCase herbicide metabolizing function was investigated in the two previously identified P450s.
• MHR plants exhibited resistance to all the ACCase inhibitors by enhanced herbicide metabolism. Resistance to the ACCase inhibitors segregated in a 3 : 1 ratio in the F₂ generation and completely co‐segregated with ALS inhibitor resistance in F₆ lines. Expression of the respective P450 genes conferred resistance to the three herbicides in rice, which is in line with the detection of hydroxylated herbicide metabolites in vivo in transformed yeast.
• CYP81As are super P450s that metabolize multiple herbicides from five chemical classes, and concurrent overexpression of the P450s induces metabolism‐based resistance to the three ACCase inhibitors in MHR E. phyllopogon, as it does to ALS inhibitors.

Introduction

Herbicide resistance in weeds exemplifies plant adaptive evolution in response to human‐driven selection in nature. Significant selection pressure by herbicides has led to convergent evolution of resistance across species all over the world, resulting in a great threat to agriculture (Gould et al., 2018; Kreiner et al., 2018). Considerable efforts toward elucidating the adaptation mechanism showed that a single nucleotide substitution in genes encoding herbicide target sites often accompanies resistance to herbicides (Powles & Yu, 2010). In addition, remarkable resistance mechanisms such as massive gene copy number amplifications (Gaines et al., 2010; Koo et al., 2018), deletion of a codon triplet (Patzoldt et al., 2006) and double amino acid substitutions in a single gene (Yu et al., 2015) have also been reported. The above mutations have been found in genes encoding herbicide target sites, and thus called target‐site resistance (TSR). The limited number of target‐site encoding genes in plant genomes facilitates analyses of the molecular mechanism of TSR, leading to discoveries like the above. Resistance mechanisms other than TSR are categorized as non‐target‐site resistance (NTSR), and include herbicide sequestration, limited translocation and rapid metabolism (Délye, 2013). Compared to TSR, most genes related to NTSR comprise a superfamily, hampering the identification of genes involved in resistance (Yuan et al., 2007). Therefore, only a few genes have been shown to be involved in NTSR (Cummins et al., 1999, 2013; Iwakami et al., 2014).

Among all resistance mechanisms, enhanced herbicide metabolism is the most threatening in agriculture because metabolism‐based resistant plants often exhibit resistance not only to the herbicides for which resistance was selected, but also to herbicides with other modes of action that were never used for selection, including yet‐to‐be‐discovered herbicides. Despite its serious threat to agriculture, the molecular mechanisms of multiple‐herbicide resistance (MHR), including herbicide‐metabolizing enzymes, remain poorly understood. Several studies have postulated the idea of metabolism‐based cross‐resistance (for definitions, see Hall et al., 1994), where the same enzyme serendipitously metabolizes a diverse range of herbicides (Beckie & Tardif, 2012; Yu & Powles, 2014). By contrast, more and more studies have suggested the involvement of multiple genes in MHR, especially in allogamous weeds such as annual ryegrass (Lolium rigidum Gaud.) and blackgrass (Alopecurus myosuroides Huds.) (Preston et al., 1996; Petit et al., 2010; Busi et al., 2011). Multiple alleles are speculated to accumulate in the process of stepwise resistance selection to a single herbicide, as exemplified by artificial resistance selection (e.g. Neve & Powles, 2005; Busi et al., 2013). Poor understanding of the mechanism makes it impossible to predict the resistance profile of metabolism‐based resistant weeds, further complicating resistance management.

Cytochrome P450 monooxygenases (P450s) are known to play a major role in herbicide metabolism in plants (Werck‐Reichhart et al., 2000). P450s are ubiquitous in nature, appearing in almost all living organisms. Animals possess dozens of P450 genes in their genomes, some of which are known to be versatile ‘super P450s’ that metabolize various xenobiotics including herbicides (Ohkawa & Inui, 2015). In plants, meanwhile, P450 genes number in the hundreds (Nelson & Werck‐Reichhart, 2011). As expected with this number of genes, they are often functionally differentiated with narrow substrate specificity (Morant et al., 2003). This has challenged the idea that MHR can be caused by a super P450 that metabolizes multiple herbicides.

Echinochloa phyllopogon (Stapf.) Koss., also known as Echinochloa oryzicola, has evolved MHR in the Sacramento Valley in California, USA (Fischer et al., 2000). The resistant plants exhibit resistance to herbicides from at least five modes of action such as acetolactate synthase (ALS) inhibitors, acetyl‐CoA carboxylase (ACCase) inhibitors, very‐long‐chain fatty acid elongase inhibitors, a 1‐deoxy‐d‐xylulose 5‐phosphate synthase (DXS) inhibitor and an auxin‐like herbicide (Fischer et al., 2000; Osuna et al., 2002; Ruiz‐Santaella et al., 2006; Bakkali et al., 2007; Yasuor et al., 2008, 2009, 2012). The potential for effective herbicides being limited has caused a severe problem in rice production in the Sacramento Valley. Biochemical and physiological work revealed that the mechanism of resistance is enhanced herbicide metabolism at least for resistance to ALS, ACCase and DXS inhibitors (Osuna et al., 2002; Yasuor et al., 2009, 2010). Previously, we found that two P450s, CYP81A12 and CYP81A21, which metabolize ALS inhibitors, were overexpressed in MHR E. phyllopogon (Iwakami et al., 2014). The overexpression of the two P450s perfectly co‐segregated with ALS inhibitor resistance in crossed progenies of MHR and sensitive (S) lines, strongly suggesting that the ALS resistance in the MHR line is caused by overproduction of the P450s. It is also suggested that a single trans element is responsible for overexpression of the two P450s. However, the molecular players involved in the resistance to other herbicides have yet to be disclosed in MHR E. phyllopogon.

ACCase inhibitors belong to one of the major categories of herbicides. Among the several forms of ACCase in plants, herbicides target the plastid‐localized homomeric form (Délye, 2005). As the plastid‐localized homomeric form of ACCase is found exclusively in Poaceae, these are the only plants in which ACCase‐inhibiting herbicides are effective. ACCase inhibitors have one of three chemical backbones, namely aryloxyphenoxypropionates (APPs), also known as ‘FOPs’, cyclohexanediones (CHDs), also known as ‘DIMs’, or phenylpyrazoline (PPZ), also known as ‘DEN’. MHR E. phyllopogon is known to be resistant to FOP group ACCase inhibitors by an enhanced herbicide metabolism (Ruiz‐Santaella et al., 2006; Bakkali et al., 2007) and not by TSR as no difference in activity was observed between MHR and S lines, and there was no ACCase‐inhibitor sensitivity in the target site or the ACCase gene sequences in all four copies (Bakkali et al., 2007; Iwakami et al., 2012). It is not known whether there is cross‐resistance to other classes of ACCase inhibitors in MHR E. phyllopogon.

In this study, we evaluated the cross‐resistance characteristics of three classes of ACCase inhibitors in MHR E. phyllopogon and investigated whether the identical mechanism of resistance to ALS inhibitors is behind the resistance to ACCase inhibitors. Our study revealed that MHR E. phyllopogon exhibits resistance to ACCase inhibitors from the three chemical classes. Genetic inheritance studies and functional characterization of the two genes showed that the resistance to ACCase inhibitors was associated with the overexpression of the genes CYP81A12 and CYP81A21, similar to ALS inhibitor resistance. We provide strong evidence that the two CYP81A P450s are involved in metabolism‐based cross‐resistance to both ALS and ACCase inhibitors in MHR E. phyllopogon. The current work underpins the idea that a single evolutionary step, not stepwise evolution, may lead to concomitant cross‐resistance to multiple herbicides by activating one or more super P450s that metabolize multiple herbicides.

Materials and Methods

Echinochloa phyllopogon

The S line, 401, and the MHR line, 511, of E. phyllopogon were originally collected in the Sacramento Valley, California, in 1997 (Fischer et al., 2000). They were self‐pollinated three times before use (Iwakami et al., 2012). The F₂ and F₆ generations of these lines are described elsewhere (Iwakami et al., 2014).

Herbicide sensitivity of E. phyllopogon

Seeds of E. phyllopogon were incubated in distilled water at 30°C in the dark for 2 d. Germinated seeds were grown hydroponically in distilled water at 25°C for six more days with a 12‐h photoperiod supplied by fluorescent light (approximately 300 μmol m⁻² s⁻¹). The hydroponic solution was changed to Kasugai nutrient solution at 1 : 10 concentration (Ohta, 1970) for 3 d, followed by Kasugai nutrient solution at full concentration. The nutrient solution was changed every 3 d. Diclofop‐methyl (DM), tralkoxydim (TD) and pinoxaden (PD) were applied at day 13 (2.5‐leaf stage) by dipping a shoot of E. phyllopogon for 30 min in each herbicide solution with 0.01% Tween 20 (Wako, Osaka, Japan). The doses of DM, TD and PD were 0, 1, 3, 10, 30, 100 and 300 μM each. Shoot fresh weight (FW) was measured 6 d after herbicide application. Each pot contained three plants and the experiment was performed with four replicates. Dose–response curves were drawn using two‐parameter log‐logistic regression (Ritz et al., 2015) with the drc package in R v.3.3.3 (R Core Team, 2016). Each experiment was performed at least twice with similar results, and therefore, only one of the results is shown.

Sensitivities to DM, TD and PD were evaluated in F₂ and F₆ plants. To evaluate the F₂ plants, 160, 20 and 20 individuals of the F₂ population, S and MHR lines were used, respectively. In the evaluation of the F₆ plants, eight plants per line were used. The herbicide concentrations were 30 μM for DM and 10 μM for TD and PD. The sensitivity to bensulfuron‐methyl (BSM), an ALS inhibitor, of 10 F₆ lines was also evaluated, as reported previously (Iwakami et al., 2014).

Herbicide metabolism analysis in E. phyllopogon

Plants were grown and treated with DM, TD and PD as described above. The concentration of each herbicide was the same as in the experiment with crossed progenies. Shoots of six plants (c. 300 mg) of each line were harvested at 0, 3, 6, 12 and 24 h after herbicide treatment, snap‐frozen in liquid nitrogen and stored at −80°C until extraction. The shoots were ground into powder in liquid nitrogen with a mortar and pestle and homogenized with 80% (v/v) of cold methanol (5 ml g⁻¹ FW). The homogenate was ultrasonicated for 10 min followed by 10 min of centrifugation at 20 000 g. The supernatant was used for the following analysis.

DM, TD, PD and their metabolites were analyzed using an LC‐MS/MS system (Tripple TOF 5600; AB SCIEX, Tokyo, Japan) connected to a NexeraX2 UHPLC unit (Shimadzu, Kyoto, Japan) equipped with a SunFire C18 column (150 mm × 2.1 mm, 3.5 μm; Waters, Tokyo, Japan). The analysis was performed under the following conditions: mobile phase A, 0.1% ammonium formate + 0.1% formic acid in water, mobile phase B, 0.1% methanol, and a 2% linear gradient of B for 15 min and 95% of B for 10 min in the positive‐ion mode and 15 min for the negative‐ion mode.

DM was detected in positive‐ion mode with m/z 341.0 and m/z 281.013 for the precursor and product ions, respectively. Diclofop acid was detected in negative‐ion mode with m/z 325.0 and m/z 252.983 for the precursor and product ions, respectively. TD was detected in positive‐ion mode with m/z 330.2 and m/z 284.165 for the precursor and product ions, respectively. PD was detected in positive‐ion mode with m/z 341.0 and m/z 281.013 for precursor and product ions, respectively. Pinoxaden acid was detected in positive‐ion mode with m/z 317.186.

Statistical analyses were carried out on square root‐transformed data. Student's t‐tests were performed using R v.3.3.3 (R Core Team, 2016).

Heterologous production of CYP81As in Saccharomyces cerevisiae

CYP81A12 and CYP81A21 derived from the MHR line were expressed together with a gene encoding their redox partner, cytochrome P450 reductase (ATR1) from Arabidopsis thaliana (L.) Heynh. in budding yeast. Both CYP81A genes were synthesized and the codon was optimized to be expressed in yeast by the GenPart DNA fragment service (GenScript, Piscataway, NJ, USA). Both DNA fragments were re‐amplified by PCR using the PrimeSTAR MAX DNA polymerase (Takara, Shiga, Japan) and appropriate primer sets (Supporting Information Table S1). Each forward PCR primer contained a 16‐bp vector homologous sequence and the yeast Kozak sequence, and each reverse primer contained a 16‐bp vector homologous sequence after the stop codon. PCR conditions were 30 cycles at 98°C for 10 s, 55°C for 5 s and 72°C for 10s. The amplified fragments were cloned into the KpnI‐EcoRI site of the pYeDP60 vector (Urban et al., 1990) using the NEBuilder HiFi DNA Assembly Master Mix (New England BioLabs, Ipswich, MA, USA). After confirmation of the insert DNA sequences by sequencing, resultant plasmids were used to transform S. cerevisiae WAT11 (Pompon et al., 1996), which has a genomically integrated ATR1, using the lithium acetate method (Ito et al., 1983). Transformants were inoculated into 2 ml of SGI medium (0.67% yeast nitrogen base without amino acids, 0.1% casamino acid, 40 μg ml⁻¹ l‐tryptophan and 2% d‐glucose) and were grown at 30°C for 24 h. Precultures (80 μl) were transferred to 4 ml of YPD medium (1% yeast extract, 2% polypeptone and 2% d‐glucose). After culture at 30°C for 24 h, 440 μl of 20% galactose was added to induce expression of CYP81As, and the culture was continued at 28°C for 14 h. Cells from 2 ml cultures were harvested and resuspended in 200 μl of whole‐cell assay buffer (100 mM potassium phosphate buffer (pH 7.4), 7 mM magnesium sulfate, 10 mM d‐glucose). The reaction was started by adding 500 μM DM, TD or PD. After incubation at ambient temperature with shaking for 24 h, the reaction was stopped by adding 200 μl of 0.2% formic acid and 40% acetonitrile in water. The resulting insoluble materials and yeast cells were precipitated by centrifugation at 20 000 g and 4°C for 5 min. DM metabolites were analyzed as described above. The metabolites of PD and TD were analyzed using an LC‐MS system equipped with a COSMOCIL 5C₁₈‐MS‐II column (150 mm × 3.0 mm, 5 μm; Nacalai Tesque, Kyoto, Japan) under the following conditions: mobile phase A, 0.1% formic acid in water, mobile phase B, 0.1% formic acid in acetonitrile, and a 20–90% linear gradient of B for 12 min and 90% B for 5 min delivered at 0.4 ml min⁻¹. MS was simultaneously performed in positive‐ and negative‐ion modes using an LCMS‐8030 device (Shimadzu).

Real‐time PCR

The S and MHR lines were grown until the 2.5‐leaf stage as described above. RNA was extracted from the shoots of four plants using an RNeasy Plant Mini Kit (Qiagen) and then treated with DNase I as described previously (Iwakami et al., 2014). The RNAs were reverse transcribed using the oligo(dT)20 primer with ReverTra Ace (Toyobo, Osaka, Japan) according to the manufacturer's instructions. Real‐time PCR was performed using SYBR Green Realtime PCR Master Mix (Toyobo) on a CFX Connect Real‐Time PCR Detection System (Bio‐Rad) with PCR cycles of 95°C for 15 s and 60°C for 30 s. Eukaryotic translation initiation factor 4B (EIF4B) was used as an internal control gene and validated as a stably expressed gene in lines 401 and 511 (Iwakami et al., 2014). The primers for CYP81A12, CYP81A21 and EIF4B are described in Iwakami et al. (2014). The experiment was conducted with three biological and two technical replications. Data were analyzed using the ΔΔCT method.

Rice transformation with CYP81A genes

A binary vector, pCAMBIA1390‐sGFP (Toki et al., 2006), was digested with SalI and BsrGI to remove synthetic green fluorescent protein (sGFP). RNA was extracted from seedlings of the MHR line and of barley (Hordeum vulgare L. ‘Golden Promise’) and cDNA was synthesized as described above. Full‐length P450 genes of E. phyllopogon were amplified from the cDNA of the MHR line as previously reported (Iwakami et al., 2014). Four genes encoding CYP81A genes were found in the genome of barley ‘Morex’ (Colmsee et al., 2015) and were named by the Cytochrome P450 Nomenclature Committee as follows: HORVU2Hr1G063220, CYP81A61; HORVU5Hr1G096930, CYP81A62; HORVU5Hr1G096940, CYP81A63; HORVU5Hr1G096950, CYP81A64. The full‐length HvCYP81A3 was amplified with PrimeSTAR GXL DNA Polymerase (Takara) using a forward primer (5′‐CATCCAATCAGCCTCAGACC‐3′) and a reverse primer (5′‐TATGATGCTATAACAAGGAAGTCG‐3′) from the cDNA of Golden Promise. The PCR conditions were as follows: 38 cycles at 98°C for 10 s, 64°C for 15 s and 68°C for 90 s. The amplicons were subcloned into pGEM‐T Easy vector (Promega) and were used as a template for the PCR using PrimeSTAR MAX (Takara) with the primers listed in Table S2. The PCR conditions were 30 cycles at 98°C for 10 s, 62°C for 5 s and 72°C for 60 s. The amplicons were inserted under the Cauliflower mosaic virus (CaMV) 35S promoter of the linearized pCAMBIA1390 vector by the ligation reaction using the In‐Fusion HD Cloning Kit (Takara) according to the cloning kit protocol.

Each binary vector was transformed into Agrobacterium tumefaciens strain EHA105 following the protocol in Hofgen & Willmitzer (1988). Scutellum‐derived calli of rice (Oryza sativa L. ‘Nipponbare’) were transformed according to Toki et al. (2006). The copy number of transgenes for regenerated plants (T₀ generation) was estimated by genomic DNA blot analysis as described previously (Endo et al., 2016). The DNA probe for the genomic DNA blot was designed on the selection marker gene, hygromycin phosphotransferase, and was prepared using the PCR DIG Probe Synthesis Kit (Roche) according to the DIG Application Manual (Roche) with the forward primer (5′‐GATGTTGGCGACCTCGTATT‐3′) and the reverse primer (5′‐GTGCTTGACATTGGGGAGTT‐3′). Seeds of single‐copy lines selected by DNA blot analysis were collected. The single copies were further confirmed in the T₁ generation by segregation of hygromycin B (50 mg l⁻¹) sensitivity. The hygromycin B‐resistant plants were further grown for proliferation. T₂ homozygous lines, where hygromycin B resistance was fixed, were used in the herbicide sensitivity test.

Herbicide sensitivity of transgenic rice

Nine independent transgenic calli selected by hygromycin B were placed on plates of N6D medium containing the respective herbicides (Toki, 1997). The concentrations of herbicides were chosen where wild‐type calli almost stopped growing: 0.4 μM for DM, 0.3 μM for TD and 0.05 μM for PD. The plates were incubated for 3 wk at 25°C for the calli of CYP81A12 and CYP81A21 and at 30°C for the calli of the other genes. Sensitivity was evaluated based on the growth of the calli such as diameter and proliferation. Experiments were repeated four times for CYP81A12 and CYP81A21 and three times for other genes with similar results. Therefore, only the results from a single experiment are shown.

Rice seeds were dehusked and sterilized twice in 2.5% sodium hypochlorite for 15 min. The seeds were germinated on wet filter paper for 3 d. Germinated seeds were placed on MS medium with the respective herbicides and grown for 6 d.

Phylogenetic analysis of P450

Amino acid sequences of the CYP81 family of rice, Arabidopsis, tomato (Solanum lycopersicum L.) and Medicago truncatula Gaertn. were obtained from the Rice Cyt P450 Database (http://ricephylogenomics.ucdavis.edu/p450/index.shtml), the Arabidopsis Cytochrome P450, Cytochrome b₅, P450 Reductase, β‐Glucosidase and Glycosyltransferase Site (http://www.p450.kvl.dk/), and the Cytochrome P450 Homepage (http://drnelson.uthsc.edu/CytochromeP450.html). The P450 sequences of E. phyllopogon are CYP81A12 (AB818461), CYP81A14 (AB733994), CYP81A15 (AB733995), CYP81A18 (AB733996), CYP81A21 (AB818462); CYP81A23 (AB734000), CYP81A24 (AB734001) and CYP81A26 (AB734003). The gene IDs for the barley CYP81A are given in the ‘Rice transformation with CYP81A genes’ section above.

Sequences were aligned using Muscle (Edgar, 2004). Trees were constructed from the aligned sequences by the maximum likelihood method using Mega6 (Tamura et al., 2013) under the WAG model for amino acid substitution with a gamma distribution for the rates. One thousand bootstrap replicates were performed. The resulting tree was visualized using iTOL (https://itol.embl.de/).

Results

Sensitivity and metabolism of ACCase herbicides in E. phyllopogon

MHR plants show resistance to ACCase inhibitors from different chemical classes: DM from the FOP group, TD from the DIM group and PD from the DEN group (Fig. 1). The dose ratio (MHR line : S line) that reduced the growth by 50%, hereafter referred to as resistance index, was 115.3 for DM, 10.9 for TD and 36.4 for PD. The results indicate that the MHR line is resistant to various ACCase herbicides. We then compared the metabolism of these herbicides between the lines.

Figure 1

Sensitivity to acetyl‐CoA carboxylase inhibitors in sensitive (S) and multiple‐herbicide resistance (MHR) Echinochloa phyllopogon. (a) S and MHR lines treated with each ACCase inhibitor. Diclofop‐methyl (DM), 0, 10, 30 μM (left to right); tralkoxydim (TD), 0, 3, 10 μM (left to right); pinoxaden (PD), 0, 3, 10 μM (left to right). (b) Dose response to each ACCase inhibitor. Blue, S line; red, MHR line.

Among the three herbicides, two of them, DM and PD, are known to be prodrugs, and once they are absorbed in the plants they are converted into diclofop acid and pinoxaden acid, respectively, by plant endogenous esterase activity (Jeschke, 2016). By contrast to the slight ACCase‐inhibiting activity of DM, both PD and pinoxaden acid have significant ACCase‐inhibiting activity (Wenger et al., 2012; Jang et al., 2013). The active bodies of the respective herbicides, diclofop acid, TD and pinoxaden acid, can be inactivated by hydroxylation, most likely via P450s in plants (Shimabukuro et al., 1979; Zimmerlin & Durst, 1990; Hadfield et al., 1994; Wenger et al., 2012).

The amount of DM decreased rapidly in a similar way in both the S and the MHR lines during the first 6 h (Fig. 2; Table S3). By 3 h, the amount of DM became < 15% of that at 0 h. Diclofop acid, the active form of DM, drastically increased during the first 3 h, however. Notably, the amount of diclofop acid was significantly lower in the MHR line. The amount of TD drastically decreased in the first 6 h in both lines. A significantly lower amount of TD was observed in the MHR line by 6 h. The amount of PD detected was far lower than that of the two other herbicides. The maximum amounts were 15.6, 23.8 and 0.311 μg g⁻¹ FW for DM, diclofop acid and TD, respectively, while it was 0.081 μg g⁻¹ FW for PD, lowest among the three herbicides. In addition, the amount of PD did not sharply decrease, in contrast to DM and TD. This suggests that PD was rapidly metabolized into pinoxaden acid and further metabolized during the 30 min of herbicide application. Therefore, the significantly higher amount of PD observed in the MHR line at 3, 6 and 24 h (1.2‐ to 1.4‐fold) would be too small to influence herbicide sensitivity. By contrast to the slight difference in the amount of PD, a larger difference was observed in the amount of pinoxaden acid at 12 h (3.2‐fold) and 24 h (4.7‐fold), while no significant difference was observed by 3 h.

Figure 2

Time course of acetyl‐CoA carboxylase inhibitor metabolism in sensitive (S) and multiple‐herbicide resistance (MHR) Echinochloa phyllopogon. Plants were treated with 30 μM diclofop‐methyl (DM), 10 μM tralkoxydim (TD) or 10 μM pinoxaden (PD). The amount of each compound was monitored by LC‐MS/MS. Blue, sensitive line; red, MHR line. Bars represent ± SE (n = 3). Putative pinoxaden acid was monitored by LC‐MS/MS at m/z 317.186. *, P < 0.01 (Student's t‐test).

The relative amount of activity in the MHR and S lines mostly reflects the relative order of their resistance to the three herbicides. The largest difference was observed in diclofop acid (up to 4.9‐fold), followed by pinoxaden acid (up to 4.7‐fold) and TD (up to 2‐fold) (Table S3). Notably, our LC‐MS/MS study also showed that the absorption of DM and PD was similar in the S and MHR lines, as the herbicide content was not prominently different at 0 h. In summary, the MHR line inactivates the ACCase inhibitors more promptly than the S line.

Resistance segregation in crossed progenies of MHR and S lines

In the F₂ population, herbicide sensitivity to all three ACCase inhibitors is segregated (Fig. 3a). For the three herbicide resistances, the segregation ratios of the resistant + intermediate : sensitive fitted a 3 : 1 ratio: DM, 121 : 39 (χ² = 0.0333, P = 0.8551), TD, 118 : 42 (χ² = 0.1333, P = 0.715), PD, 117 : 43 (χ² = 0.3, P = 0.5839). The higher resistance factor in DM allowed intermediate individuals to be discriminated, similar to our previous research on ALS inhibitor resistance in the MHR line, where we could distinguish the intermediates only for BSM, a herbicide with a high resistance index (Iwakami et al., 2014). The DM segregation of resistant : intermediate : sensitive fitted a 1 : 2 : 1 ratio (χ² = 0.3333, P = 0.3267). These results suggest that, as for the ALS inhibitor, resistance is under the control of a single element.

Figure 3

Herbicide sensitivity in crossed progenies of sensitive (S) and multiple‐herbicide resistance (MHR) Echinochloa phyllopogon. (a, b) Sensitivity to acetyl‐CoA carboxylase (ACCase) inhibitors was evaluated by the plant's response to 30 μM diclofop‐methyl (DM), 10 μM tralkoxydim (TD) and 10 μM pinoxaden (PD). (a) F₂ segregation based on sensitivity to ACCase inhibitors. Blue, 20 S individuals; red, 20 MHR individuals; green, 160 F₂ individuals. (b) mRNA levels of CYP81A12 and CYP81A21 quantified by real‐time PCR. Sixteen F₆ lines were analyzed. Four biological replications for S and MHR parents. Each boxplot shows median and interquartile range. Whiskers correspond to 1.5 times the interquartile range. Colored dots represent alleles of CYP81A12 and CYP81A21. Blue, allele of S line; red, allele of MHR line; green, heterozygous; F6‐S, sensitive F₆ lines; F6‐R, resistant F₆ lines. (c) F₆ segregation in sensitivity to bensulfuron‐methyl (BSM), an acetolactate synthase (ALS) inhibitor, and ACCase inhibitors (DM, TD, PD). BSM sensitivity was evaluated as described previously (Iwakami et al., 2014). The F₆ lines include 30 ALS‐sensitive lines, 19 ALS‐resistant lines and one segregating line. R, resistant.

Next, we investigated the sensitivity of 50 F₆ recombinant inbred lines of the MHR and S lines to ACCase inhibitors. In our previous study, the ALS inhibitor sensitivity of each line was characterized and the higher expression of CYP81A12 and CYP81A21 in the roots of plants at the one‐leaf stage was perfectly associated with ALS resistance (Iwakami et al., 2014). Higher expressions of CYP81A12 and CYP81A21 were completely associated with ACCase inhibitor resistance in the shoots of F₆ lines at the 2.5‐leaf stage (Fig. 3b). Intriguingly, all the ALS‐sensitive lines were sensitive to all ACCase inhibitors and all the ALS‐resistant lines were resistant to all ACCase inhibitors (Fig. 3c).

The two genes are known to have single nucleotide polymorphisms in coding sequences in the MHR and S lines (Iwakami et al., 2014). One polymorphism in particular causes amino acid substitution in CYP81A21. The polymorphism, however, did not co‐segregate with ACCase inhibitor resistance as in our previous research on ALS inhibitor resistance (Fig. 3b), indicating that the substitutions are not involved in this resistance. Therefore, further characterization of the two genes was conducted using MHR alleles.

ACCase inhibitor metabolism by CYP81A12 and CYP81A21

The data so far suggest that the same mechanism behind the resistance to ACCase inhibitors is also behind resistance to ALS inhibitors. To investigate the function of CYP81A12 and CYP81A21, the respective genes were introduced into rice calli. Nine independent lines of calli transformed with CYP81A12 and CYP81A21 under the control of the CaMV35S promoter were compared with calli expressing the gene for green fluorescent protein (GFP) in their sensitivity to the four ACCase inhibitors. At least five of the nine independent calli with CYP81A12 and CYP81A21 grew vigorously on medium containing 0.4 μM DM, 0.3 μM TD or 0.05 μM PD, while the calli with GFP completely stopped growing (Fig. S1). The calli with both genes proliferated at high concentrations (at least three times) of each herbicide, and some calli even proliferated on medium with 10 times higher herbicide concentrations. We selected single‐copy lines from regenerated transformed plants by Southern blotting (Fig. S2) and T₂ homozygous lines of CYP81A12 and CYP81A21. Rice plants of T₂ homozygous lines grew well on medium containing 10 μM DM, 1 μM TD or 0.1 μM PD (Fig. 4).

Figure 4

Sensitivity to acetyl‐CoA carboxylase inhibitors in rice transformed with CYP81A12 or CYP81A21. Germinated seeds of T₂ homozygous rice lines, CYP81A12#3 and CYP81A21#4, were placed on medium containing 10 μM diclofop‐methyl (DM), 1 μM tralkoxydim (TD) or 0.1 μM pinoxaden (PD). Plants were grown for 6 d.

Next, we heterologously expressed CYP81A12 and CYP81A21 in a yeast WAT11 strain harboring Arabidopsis cytochrome P450 reductase in its genome (Pompon et al., 1996). Each herbicide was added separately to the yeast culture medium and 3 h later the solution was analyzed using LC‐MS/MS.

DM was converted into its active form, diclofop acid, in all the yeast solutions, probably because of endogenous esterase activity of the yeast. The putative hydroxylated product of diclofop acid was detected with a retention time of 15.4 min in the media of yeast expressing CYP81A12 and CYP81A21, while no putative hydroxylated metabolites were detected in empty vector medium (Fig. 5). Similarly, putative hydroxylated TD was only detected in the culture solution of CYP81A12‐ and CYP81A21‐expressing yeast with a retention time of 10.9 min. PD was converted to its active form, pinoxaden acid, in all the yeast solutions, but was further metabolized into hydroxylated product only in the P450‐expressing yeast media (retention time 4.9 min).

Figure 5

Metabolism of acetyl‐CoA carboxylase (ACCase) inhibitors in yeast transformed with CYP81A12 or CYP81A21. Yeast harboring empty vector (pYeDP60), CYP81A12 or CYP81A21 was incubated with ACCase inhibitors: diclofop‐methyl (DM); tralkoxydim (TD); pinoxaden (PD). Putative hydroxylated products of the active bodies, diclofop acid, TD and pinoxaden acid, were monitored by LC‐MS/MS at m/z 317.2, 346.2 and 322.2, respectively.

Function of the CYP81A subfamily in E. phyllopogon and barley

The CYP81A subfamily in E. phyllopogon comprises at least 12 CYP81A genes, including three putative pseudogenes due to frame shifts (Iwakami et al., 2014). The other presumably functional CYP81A genes of the MHR line other than CYP81A12 and CYP81A21 were used to transform rice calli (Fig. 6a). The calli expressing CYP81A15 and CYP81A24 survived on medium with 0.05 μM PD. Decreased sensitivity to TD was also observed in CYP81A24‐expressing calli. By contrast, no significant decrease in DM sensitivity was observed in all the transgenic calli expressing the seven CYP81A genes. The results suggest that small differences in amino acid sequence influence ACCase herbicide‐metabolizing function. The absence of a difference in the transcript levels and sequences of CYP81A15 and CYP81A24 between the MHR and S lines (Iwakami et al., 2014) implies that these genes are not involved in resistance, despite carrying a herbicide‐metabolizing function.

Figure 6

Characterization of CYP81A genes of Echinochloa phyllopogon and barley. (a, b) Nine independent calli were placed on medium containing herbicide: 0.4 μM diclofop‐methyl (DM); 0.3 μM tralkoxydim (TD); 0.05 μM pinoxaden (PD). The calli were cultivated on the media for 3 wk. (a) Response of calli transformed with E. phyllopogon CYP81A genes to herbicides. (b) Response of calli transformed with barley CYP81A gene to herbicides. (c) Phylogenetic tree of the CYP81 family of P450s of E. phyllopogon, barley, rice, Arabidopsis, tomato and Medicago truncatula. Only CYP81As are shown for E. phyllopogon and barley.

The partially conserved ACCase inhibitor‐metabolizing function of E. phyllopogon CYP81As motivated us to analyze CYP81A P450 in other species. We chose barley genes for the analysis, as barley is known to be naturally tolerant to these herbicides. In a search of the barley genome, four CYP81A genes were found (CYP81A61 to CYP81A64). We isolated one of the highly expressed genes, CYP81A63 (Fig. S3), from H. vulgare ‘Golden Promise’ and expressed it in rice. The transformed rice exhibited marked resistance to DM and PD (Fig. 6b). We observed better proliferation in some calli of CYP81A63 on medium with TD than in the negative control, although the level of resistance was relatively low.

Discussion

A first case of suspected metabolism‐based cross‐resistance was reported in 1986 in an Australian population of annual ryegrass where the population exhibited resistance to both ALS and ACCase inhibitors (Heap & Knight, 1986). Since then, similar cases of resistance to herbicides with more than one mode of action have often been observed in metabolism‐based resistant weeds (Beckie & Tardif, 2012). These observations have led to the notion that metabolism‐based resistance poses a serious threat to agriculture. However, the molecular mechanism of MHR in metabolism‐based resistance has eluded scientists for a long time. The absence of evidence for molecular players made the interpretation of MHR complicated especially in cross‐pollinating weeds where the accumulation of distinct herbicide‐metabolizing alleles can easily occur. In fact, the accumulation of multiple genes is usually emphasized in describing metabolism‐based resistance (Délye, 2013). In this study, however, we present confirmatory evidence that the mechanism of resistance to ALS and some ACCase inhibitors in MHR E. phyllopogon is identical by analyzing resistance co‐segregation in the crossed progenies of S and MHR lines and also by functional characterization of CYP81A12 and CYP81A21. The concurrent overexpression of the two P450 genes that completely link ALS and ACCase herbicide resistance suggests the presence of a single trans element regulating the P450 genes as shown in our previous paper, where we demonstrated that the overexpression of two genetically unlinked genes is inherited in a single‐locus control model (Iwakami et al., 2014). Thus, our results indicate that the evolution of metabolism‐based resistance occurring in a single event may lead to simultaneous nullification of multiple herbicides, which underpins the threat of metabolism‐based resistance reported by researchers (Yuan et al., 2007; Yu & Powles, 2014; Hicks et al., 2018).

In this study, we observed relatively large differences between the resistance indices for the three herbicides and the rates of herbicide metabolism. However, these differences do not invalidate our claim that enhanced metabolism plays a major role in resistance to the three ACCase inhibitors in the MHR line. Differences in metabolism rates of a few fold resulting in variable resistance indices in whole plants have been observed in other well‐characterized cases of metabolism‐based resistant weeds (Christopher et al., 1991; Ma et al., 2013; Yu et al., 2013; Kaiser & Gerhards, 2015). Meanwhile, our study does not exclude the possible involvement of other mechanisms, although most of the mechanisms proposed for weed resistance to ACCase inhibitors, such as insensitivity or overproduction of target site (Kaundun, 2014; Laforest et al., 2017) and reduced absorption of herbicide (De Prado et al., 2005), have been dismissed. The only mechanism that we have not investigated among the proposed resistance mechanisms to ACCase inhibitors is protection from herbicidal stress (Cummins et al., 2013). Further investigations that include protecting samples from herbicidal stress will help to provide a complete understanding of resistance to multiple herbicides in MHR E. phyllopogon.

The approximately 10‐fold resistance to DM that was observed in the calli is somewhat low considering the high resistance index of the MHR line. The discrepancy may be a consequence of the evaluation method, possibly due to temporary and permanent exposure in E. phyllopogon and rice calli, respectively. Differences between the two species could be another reason for the discrepancy. In the case of DM metabolism in plants, glycosylation after hydroxylation of diclofop acid is known to play an important role (Shimabukuro et al., 1987) and may differ between E. phyllopogon and rice. Other factors, such as relatively weak activity of CaMV35S promoter in rice and transformation of only one of the two genes, may also have some influence. Thorough study of DM metabolism, including co‐transformation of the two CYP81A genes, will be an important next step.

P450 genes participating in ACCase metabolism have not been reported, although the involvement of P450s in DM metabolism in wheat and barley is well known (Zimmerlin & Durst, 1990; Romano et al., 1993). The involvement of P450 in TD and PD metabolism in plants has also been suggested by analyses of herbicide metabolites in plants (Hadfield et al., 1994; Wenger et al., 2012). We identified several CYP81A members from E. phyllopogon and barley that endow resistance to ACCase inhibitors when expressed in rice. The CYP81 family of P450s is known to be prone to blooming, and lineage‐specific evolution of the subfamily has occurred (Fig. 6c) (Nelson & Werck‐Reichhart, 2011). Of particular interest here is the exclusive existence of CYP81A in Poaceae (Nelson, 2009). It may not be a coincidence that the genes metabolizing the selective ALS and ACCase herbicides used on Poaceae crops, in which the herbicides are rapidly metabolized, were found to originate from a Poaceae‐specific P450 subfamily, namely CYP81A. The ACCase inhibitor‐metabolizing function of CYP81A P450s corroborates the importance of CYP81A in herbicide sensitivity in Poaceae.

Plant P450s are often highly differentiated in substrate specificity, reflecting the large number of genes in the plant kingdom (Morant et al., 2003; Nielsen & Møller, 2005). Here, however, we show that at least some CYP81A P450s carry extremely broad substrate specificity: CYP81A12 and CYP81A21 metabolize as many as five chemically unrelated groups of herbicides, namely sulfonylureas (BSM), triazolopyrimidines (penoxsulam) (Iwakami et al., 2014), FOP (DM), DIM (TD) and DEN (PD). Notably, our previous report on ALS inhibitor metabolism showed that two P450s metabolize BSM by de‐methylation (Iwakami et al., 2014), which differs from the current study where hydroxylation was observed for the three ACCase herbicides. Homology modeling or docking simulation of the P450s may elucidate why these P450s metabolize diverse sets of herbicides, including the different metabolizing reactions between ALS and ACCase herbicides. Such analyses would also explain the different functions among CYP81As in ACCase herbicide metabolism.

Our research presents strong evidence that overexpression of multi‐herbicide metabolizing P450s is behind the concomitant resistance to multiple herbicides in MHR E. phyllopogon. MHR in E. phyllopogon, however, is likely to be more complicated. The MHR line exhibits decreased sensitivity to fenoxaprop‐ethyl, an ACCase inhibitor (Bakkali et al., 2007), and the MHR line also accumulates more rapidly glutathione‐conjugated metabolites of fenoxaprop‐ethyl (Bakkali et al., 2007). The reaction of glutathione conjugation of fenoxaprop‐ethyl is mediated by glutathione S‐transferase (GST) in wheat (Thom et al., 2002), a crop that exhibits a high level of tolerance to fenoxaprop‐ethyl. It is reasonable to presume that the enhanced fenoxaprop‐ethyl metabolism in the MHR line is caused by the enhanced activity of GST, which is upregulated in coordination with CYP81A12 and CYP81A21. Further research to identify the GST gene involved in fenoxaprop‐ethyl resistance is under way in our laboratory. The ultimate challenge will be identification of the gene causing metabolism‐based resistance. Our previous research suggested that the overexpression of CYP81A12 and CYP81A21 is regulated by a single trans element (Iwakami et al., 2014). Identification of the causal element will provide great insight for our understanding of the evolution of MHR in weeds.

Acknowledgements

We thank Dr David Nelson for naming barley P450s. We also thank Dr Rintaro Suzuki for discussion. The seeds of barley were provided by the Institute of Plant Science and Resources, Okayama University, with support in part by the National Bio‐Resource Project of the MEXT, Japan. This work was supported by JSPS KAKENHI grant no. 15H06072, 17K15234 (to SI). The authors declare no conflict of interest.

Author contributions

SI, YK, TY, MI, YS, ST, AU, TT and HM designed the research. SI, YK, TY, MI, ME, HS and KN performed experiments. SI, YK, TY and MI analyzed the data. SI, TY and AU wrote the paper.