The Metabolism of a Novel Cytochrome P450 (CYP77B34) in Tribenuron-Methyl-Resistant Descurainia sophia L. to Herbicides with Different Mode of Actions

Descurainia sophia L. (flixweeds) is a noxious broad-leaf weed infesting winter wheat fields in China that has evolved high resistance to tribenuron-methyl. In this work, a brand new gene CYP77B34 was cloned from tribenuron-methyl-resistant (TR) D. sophia and transferred into Arabidopsis thaliana, and the sensitivities of Arabidopsis with or without the CYP77B34 transgene to herbicides with a different mode of actions (MoAs) were tested. Compared to Arabidopsis expressing pCAMBIA1302-GFP (empty plasmid), Arabidopsis transferring pCAMBIA1302-CYP77B34 (recombinant plasmid) became resistant to acetolactate synthase (ALS)-inhibiting herbicide tribenuron-methyl, protoporphyrinogen oxidase (PPO)-inhibiting herbicides carfentrazone-ethyl and oxyfluorfen. Cytochrome P450 inhibitor malathion could reverse the resistance to tribenuron-methyl, carfentrazone-ethyl and oxyfluorfen in transgenic Arabidopsis plants. In addition, the metabolic rates of tribenuron-methyl in Arabidopsis expressing CYP77B34 were significantly higher than those in Arabidopsis expressing pCAMBIA1302-GFP. Other than that, the transgenic plants showed some tolerance to very-long-chain fatty acid synthesis (VLCFAs)-inhibiting herbicide pretilachlor and photosystem (PS) II-inhibiting herbicide bromoxynil. Subcellular localization revealed that the CYP77B34 protein was located in the endoplasmic reticulum (ER). These results clearly indicated that CYP77B34 mediated D. sophia resistance to tribenuron-methyl and may have been involved in D. sophia cross-resistance to carfentrazone-ethyl, oxyfluorfen, pretilachlor and bromoxynil.


Introduction
Herbicides have been the most effective way for controlling weeds for several decades. Weeds have evolved serious resistance to a large number of herbicides due to extensive and intensive use of herbicides, which leads to serious loss of crop yields [1]. The mechanisms of weed resistance are typically classified into two broad categories. Target-site-based resistance (TSR) is endowed by gene mutation or deletion, or (and) gene overexpression or amplification of the herbicide target protein. Meanwhile, non-target-site-based resistance (NTSR) is achieved by enhanced herbicide metabolism (or) and sequestration, reduced penetration and (or) translocation [2,3]. In comparison with the well-known TSR, NTSR is focused on rarely and often covered by TSR or purposely ignored by researchers due to its complexities and difficulties. Metabolic resistance, one of the most common and important NTSRs, is caused by enhanced herbicide metabolism in weeds. Herbicide metabolism in plants usually experiences phases of conversion, conjugation and compartmentation, which are mediated by different metabolic enzymes. These metabolic enzymes include cytochrome P450 monooxygenase (P450s), glucosyltransferases (GTs), glutathione S-transferases (GSTs), aldo-keto reductase, ATP binding cassette transporters (ABC transporters) and esterases. R), Helix I (A-G-x-D-T-S), Helix K (E-x-x-R) and meander (P-x-x-F-x-P-x-x-F). These typical structural domains indicated that CYP77B34 was a member of P450 proteins ( Figure  S1). In addition, CYP77B34 displayed high identity with CYP77A3 in Brassia napus (73.95%), CYP77B1 in A. thaliana (72.56%), CYP77A3 in Eutrema salsugineum (77.67%) and CYP77B1-1 in Isatis tinctoria (90.54%) ( Figure S2).

Transcript Levels of the CYP77B34 Gene in D. sophia
Constitutive and inducible expression of the CYP77B34 gene in D. sophia were determined in order to explore its roles in D. sophia resistance to tribenuron-methyl ( Figure 1). The expression level of CYP77B34 in pHB23 and N11 D. sophia populations was 6.21-and 2.16-fold higher, respectively, than that in tribenuron-methyl-susceptible (TS) D. sophia (SD8) before tribenuron-methyl treatment (BT). In addition, CYP77B34 could be induced by tribenuron-methyl, and the expression level in TR D. sophia was always higher than that in TS D. sophia at 1, 3, 5 and 7 days after treatment (DAT). The expression levels increased about 9.00-fold in pHB23 at 1 DAT and 21.25-fold in N11 populations at 7 DAT comparing with that of TS D. sophia. Other than that, we also noted that the expression levels of CYP77B34 in TS and TR D. sophia decreased at 3 and 5 DAT comparing with that at 1 DAT. This may be due to the enhanced concentration of tribenuron-methyl absorption by plants at 3 and 5 DAT, which inhibited the expression of CYP77B34 TS and TR D. sophia. With the metabolism of tribenuron-methyl in two TR D. sophia, the inhibiting effect of tribenuron-methyl on CYP77B34 expression disappeared and even exhibited an inducing effect on CYP77B34 expression at 7 DAT. Expression levels of CYP77B34 in TS (SD8) and TR (pHB23 and N11) D. sophia plants before (BT) or 1, 3, 5 and 7 days after treatment (DAT) with or without tribenuron-methyl. 18sRNA was the reference gene. All data are mean ± SE of three replicates. Columns with different letters indicate that the relative expression level had a significant difference at the same time point (using one-way analysis of variance, combined with Dunnett's post-test comparison, p < 0.05).

Subcellular Localization of the CYP77B34 Protein
The results of transient expression indicated that the green fluorescence associated with the empty vector pCAMBIA1302-GFP distributed throughout the nucleus and cytoplasm (Figure 2A), and the red fluorescence emitted by the endoplasmic reticulum (ER) marker mCherry-HDEL was observed in the ER ( Figure 2B). Merging of the green and red fluorescence was not unanimous, which suggested that the empty vector pCAMBIA1302-GFP was localized in the cytoplasm and the nucleus rather than in the ER ( Figure 2D). When pCAMBIA1302-CYP77B34-GFP and mCherry-HDEL constructs were co-infiltrated Figure 1. Expression levels of CYP77B34 in TS (SD8) and TR (pHB23 and N11) D. sophia plants before (BT) or 1, 3, 5 and 7 days after treatment (DAT) with or without tribenuron-methyl. 18sRNA was the reference gene. All data are mean ± SE of three replicates. Columns with different letters indicate that the relative expression level had a significant difference at the same time point (using one-way analysis of variance, combined with Dunnett's post-test comparison, p < 0.05).

Subcellular Localization of the CYP77B34 Protein
The results of transient expression indicated that the green fluorescence associated with the empty vector pCAMBIA1302-GFP distributed throughout the nucleus and cytoplasm (Figure 2A), and the red fluorescence emitted by the endoplasmic reticulum (ER) marker mCherry-HDEL was observed in the ER ( Figure 2B). Merging of the green and red fluorescence was not unanimous, which suggested that the empty vector pCAMBIA1302-GFP was localized in the cytoplasm and the nucleus rather than in the ER ( Figure 2D). When pCAMBIA1302-CYP77B34-GFP and mCherry-HDEL constructs were co-infiltrated in N. benthamiana leaves, the co-localization signals were observed in the ER network ( Figure 2E-H). Pearson's correlation coefficients between mCherry-HDEL and CYP77B34-GFP or mCherry-HDEL and pCAMBIA1302-GFP were 0.84 ± 0.021 and 0.27 ± 0.067, respectively ( Figure S4). These results demonstrated that CYP77B34 was localized in the ER, which is similar to most P450s.
in N. benthamiana leaves, the co-localization signals were observed in the ER network (Figure 2E-H). Pearson's correlation coefficients between mCherry-HDEL and CYP77B34-GFP or mCherry-HDEL and pCAMBIA1302-GFP were 0.84 ± 0.021 and 0.27 ± 0.067, respectively ( Figure S4). These results demonstrated that CYP77B34 was localized in the ER, which is similar to most P450s.

Herbicide Sensitivity of Transgenic Arabidopsis Expressing CYP77B34
In order to characterize the metabolic ability of CYP77B34, the herbicide sensitivities of transgenic Arabidopsis expressing CYP77B34-GFP or GFP were compared. Four individual transgenic lines were obtained with different transcript levels of CYP77B34 ( Figure  S5). Obviously, the transgenic Arabidopsis carrying CYP77B34-GFP displayed different tolerance of herbicides with different MoAs. Compared with Arabidopsis expressing GFP, three lines of Arabidopsis carrying CYP77B34-GFP exhibited higher tolerance to tribenuron-methyl (ALS), carfentrazone-ethyl (PPO) and oxyfluorfen (PPO) (

Herbicide Sensitivity of Transgenic Arabidopsis Expressing CYP77B34
In order to characterize the metabolic ability of CYP77B34, the herbicide sensitivities of transgenic Arabidopsis expressing CYP77B34-GFP or GFP were compared. Four individual transgenic lines were obtained with different transcript levels of CYP77B34 ( Figure S5). Obviously, the transgenic Arabidopsis carrying CYP77B34-GFP displayed different tolerance of herbicides with different MoAs. Compared with Arabidopsis expressing GFP, three lines of Arabidopsis carrying CYP77B34-GFP exhibited higher tolerance to tribenuron-methyl (ALS), carfentrazone-ethyl (PPO) and oxyfluorfen (PPO) ( Table 1, Figures S6 and S7).
Among the three lines, the line of 77B34#7 displayed the highest tolerance to tribenuronmethyl, carfentrazone-ethyl and oxyfluorfen with RIs of 3.08, 1.82 and 2.13, respectively. In addition, Arabidopsis with CYP77B34-GFP had tolerance of herbicides of bromoxynil (PSII) and pretilachlor (ALCFAs) ( Figure S7F,G). Most seedlings of Arabidopsis with GFP grown in 20 µM bromoxynil were etiolated, while there were few etiolated seedlings for Arabidopsis with CYP77B34-GFP ( Figure S7F). The seedlings' growth of Arabidopsis carrying GFP was completely inhibited by pretilachlor at the concentration of 80 µM, while that of the Arabidopsis carrying CYP77B34-GFP was much better ( Figure S7G). In addition, the sensitivities of Arabidopsis with or without the CYP77B34 gene to other herbicides with different concentrations were tested in this work. These herbicides (MoAs) included imazamox and flumetsulam (ALS), napropamide (unknown MoA), glyphosate (enolpyruvyl shikimate phosphate synthase, EPSPS), glufosinate ammonium (glutamine synthetase, GS), diclofop-methyl (acetyl-coenzyme A carboxylase, ACCase), atrazine (PSII), propyzamide and pendimethalin (microtubule assembly), fluroxypyr (auxin mimics) and fomesafen (PPO). However, Arabidopsis with or without CYP77B34 gene showed no obvious differences to the herbicides mentioned above ( Figure S8).
More importantly, the P450 inhibitor malathion could reverse the resistance to those three herbicides. Treatment with malathion (55 µM) had no response to the control line CK-GFP and three 77B34 lines. Synergism of herbicides and malathion could reduce the GR 50 of three 77B34 lines, while it had no effect on CK-GFP (Table 1, Figure S6).

Absorption and Metabolism of Tribenuron-Methyl by Transgenic Arabidopsis Expressing CYP77B34
The Retention time of tribenuron-methyl was 1.366 min ( Figure S9). The determination coefficients (R 2 ) of two linear curves were 0.9951 (elution matrix) and 0.9914 (extract matrix) ( Figure S10). Average recoveries were 90.93-105.19% (elution matrix) and 81.36-91.80% (extract matrix) at three different addition levels (Table S1). Relative standard deviation (RSD) of repeatability was below 5% (n = 4) ( Table S1). The result showed that this method had high accuracy and good repeatability and was suitable for the absorption and metabolism analysis of tribenuron-methyl in Arabidopsis.
The reduced dosage between what was applied to and eluted from leaf surface was considered the absorption dosage of tribenuron-methyl by transgenic Arabidopsis plants.

Discussion
The roles of P450s in herbicide metabolism and weed resistance are well-known and have been reviewed in detail elsewhere [7,26]. Nevertheless, most of the existing evidence on P450's involvement in herbicide resistance is indirect according to the results of synergism of P450 inhibitors, RNA-seq, transcript level or contents of P450s [27][28][29]. The strongest evidence on P450 conferring herbicide resistance is the expressing P450 allele in a herbicide-sensitive individual or a substituted organism (such as Arabidopsis), and observing a decrease of susceptibility to herbicides [3]. However, identification of herbicide-metabolizing and resistance-endowing genes is slow, and only a few P450 genes have been iden- The difference between the absorption and residue in transgenic Arabidopsis plants was counted as tribenuron-methyl metabolism. The metabolic rate of tribenuron-methyl in three transgenic lines (77B34#2, 77B34#7 and 77B34#8) was significantly higher than that in line CK-GFP at 1, 3, 5 and 7 DAT ( Figure 3B). Three transgenic Arabidopsis lines exhibited no significant differences in tribenuron-methyl metabolism at 1, 3 and 5 DAT. However, the metabolic ability for tribenuron-methyl of 77B34#7 and 77B34#8 was significantly higher than that in 77B34#2 at 7 DAT (Table S2).
We noticed that studies on metabolic herbicide resistance in broad-leaf weed species are limited, and genes related to herbicide metabolism almost remain unknown [24,[36][37][38][39]. In recent years, our team has proved the important role of P450s in D. sophia resistance to tribenuron-methyl through a series of experiments, such as synergism of P450 inhibitors, absorption/metabolism, RNA-seq and qRT-PCR. A batch of metabolic enzyme genes including P450s (CYP77B34, CYP709B8, CYP81F5, CYP96A146 and CYP96A147) were identified in TR D. sophia plants [24,25]. These P450s of CYP77B34, CYP709B8, CYP81F5, CYP96A146 and CYP96A147 may confer D. sophia resistance to tribenuron-methyl. Nevertheless, the function of a specific P450 gene in the metabolism and resistance to tribenuron-methyl is still ambiguous. In this work, the CYP77B34 gene was transferred into Arabidopsis, and the sensitivities to tribenuron-methyl and other herbicides with different MoAs were observed. The results indicated that P450 of CYP77B34 exhibited metabolic ability for herbicides with MoAs of an ALS-inhibiting herbicide (tribenuron-methyl), PPO-inhibiting herbicides (carfentrazone-ethyl and oxyfluorfen), a VLCFAs-inhibiting herbicide (pretilachlor) and a PSII-inhibiting herbicide (bromoxynil) (Figures S6 and S7). For three lines of Arabidopsis expressing CYP77B34, the RIs to tribenuron-methyl were 1.87, 3.08 and 2.17, respectively ( Table 1). The P450 inhibitor malathion reduced the RI values of the three lines to 1.38, 1.62 and 1.67, which demonstrated that P450 (CYP77B34) mediated tribenuron-methyl metabolism (Table 1). In addition, tribenuron-methyl metabolism in transgenic Arabidopsis expressing CYP77B34 was faster than that in CK-GFP (Figure 3). The present results indicated that CYP77B34 mediated the metabolism and resistance to tribenuron-methyl in D. sophia. However, the proportion of resistant contribution caused by CYP77B34 is unclear because of the enormous quantity of P450s and their unclear function in D. sophia. In the meantime, both the constitutive and induced expression levels of CYP77B34 in TR D. sophia were significantly higher than that in TS D. sophia (Figure 1). This demonstrated that CYP77B34 could be involved in resistance evolution to tribenuron-methyl in TR D. sophia. What we can not ignore is that the TR D. sophia had potential cross-resistance risk to carfentrazone-ethyl, oxyfluorfen, pretilachlor and bromoxynil due to its metabolism of these herbicides. Compared with those of TSR, the biggest threat is that NTSR causes weeds to evolve unpredictable cross-resistance to herbicides with different MoAs, even including herbicides that are not yet in the market. In addition, resistance management strategies of herbicide mixtures and rotations which are effective in managing TSR may have little or no effects on NTSR metabolic resistance [7,40]. This situation has been confirmed in many resistance cases mediated by P450. For example, O. sativa transferring the L. rigidum CYP81A10v7 gene became highly resistant to ACCase-and ALS-inhibiting herbicides (diclofop-methyl, tralkoxydim, chlorsulfuron) and moderately resistant to HPPD-inhibiting herbicide trifluralin [35].
The evolution of new functions in plant metabolism usually experiences a very long process. However, the comparative analysis of plant P450s is helpful to understanding P450 function and evolution in plant metabolism [41]. Phylogenetic analysis of nucleotide sequences of CYP77B34 was inferred using the neighbor-joining method employing MEGA7 software. The CYP77B34 gene exhibited maximum similarity with members of theCYP77A and CYP77B subfamilies ( Figure S3). The results of recombinant enzymes demonstrate that CYP77s can either in-chain hydroxylate or epoxidize fatty acids, in particular with 16and 18-carbon chain lengths to form multiple precursors of cutin and suberin [42,43]. For example, Arabidopsis CYP77A6 was confirmed to be involved in the synthesis of flower cutin monomers [44], and Arabidopsis CYP77A4 is involved in signaling defense with epoxidation and in-chain hydroxylation [45,46]. CYP77A1 and CYP77B1 in petunia were related to the development of ornamental organs [42,47]. In P450-catalyzed herbicide metabolism, alkyl-or aryl-hydroxylation is one of the most common reactions [48]. Hydrophobic herbicide molecules can be oxidized to more hydrophilic metabolites by the hydroxylation reaction catalyzed by P450, which is the key step of herbicide degradation. CYP77B34 may then play this pivotal role in the metabolism of herbicides. Meanwhile, the CYP77B34 protein is localized in the endoplasmic reticulum, which encompasses various kinds of molecular machines of proteins, including folding, quality control, signal transduction and degradation [49]. CYP77B34 may play one or more roles in that.

Plant Materials
The tribenuron-methyl-susceptible (TS) D. sophia population (SD8) was originally collected from remote areas at Linyi city of Shandong province in China (35 • [25,50]. Seeds of D. sophia were sterilized with 20% H 2 O 2 for 30 min, then soaked in a 0.3% gibberellin solution overnight after rinsing with distilled water. Next, these seeds were germinated on wet paper for 3-5 days in a climate chamber, then transplanted into plastic pots containing moist loam soil, then grown in a climate chamber under conditions of 25 • C/23 • C (day/night) temperature, 16 h photoperiod with light intensity of 20,000 lux.
Arabidopsis thaliana ecotype Columbia (Col-0) and Nicotiana benthamiana seeds were sterilized with 75% ethyl alcohol for 30 s and further sterilized in a 5% sodium hypochlorite solution for 8 min after rinsing with sterile water. Finally, the seeds were washed with sterile water and germinated on Murashige and Skoog solid medium (MS) for 2-3 weeks in a climate chamber with the same conditions as above after vernalization in 4 • C for 12-24 h [51]. These operations were performed under sterile conditions.  Table 2. The qPCR program was as follows: 95 • C for 3 min and 40 cycles of (95 • C for 3 s, 60 • C for 20 s and 72 • C for 32 s). Data was collected at the stage of 72 • C for 32 s, and melting analysis was carried out to verify the absence of nonspecific amplification at the end of the cycles. The data of gene expression was analyzed with 7500 Software v2.3 (Applied Biosystems, Foster City, CA, USA). Relative expression ratio (as 2 −∆∆Ct ) was calculated by the comparative C T method, where ∆C T = [C T target gene-C T internal control gene]. One-way analysis of variance (ANOVA) with Dunnett's post-test (α = 5%) was conducted to assess significant differences by IBM SPSS Statistics 21 (International Business Machines Corporation, Armonk, NY, USA). Three biological replicates and four technical replicates were performed for each time point of TS and TR D. sophia populations [52]. Table 2. Information of primers for cloning, vector construction and qRT-PCR of the CYP77B34 gene.

Cloning and Phylogenetic Analysis
Five plants of TR D. sophia (N11) at the four-leaf stage were used for RNA extraction. Total RNA extraction and cDNAs synthesis were the same as in the above-mentioned methods. The amplification of the CYP77B34 gene was performed using TIANSeq HiFi Amplification Mix (TIANGEN, Beijing, China). The PCR reaction mixtures contained 100 ng of cDNA, 0.25 µM of forward and reverse primers ( Table 2) and 1 × HiFi amplification mix. The PCR conditions were as follows: 94 • C for 2 min, 35 cycles (98 • C for 10 s, 60 • C for 30 s and 68 • C for 45 s) and an extension step of 68 • C for 5 min. The PCR products were sequenced (Sangon Biotech, Shanghai, China) and aligned with DNAMAN 6.0 software (Lynnon Biosoft, San Ramon, CA, USA).
The phylogenetic analyses of CYP77B34 were performed with the Mega 7 program using available nucleotide sequences of plant P450s in the CYP77 family by the neighborjoining method [53]. Bootstrap analysis with 1000 replicates was conducted in order to obtain confidence levels for the branches. In order to facilitate and simplify the result, only major groups were shown.

pCAMBIA1302-CYP77B34-GFP Vector Construction and Agrobacterium Transformation
The cDNA of CYP77B34 with a full-length sequence was inserted in the pCAMBIA1302-GFP vector with 35S promoters according to the instructions of a Clone ExpresII One-Step Cloning Kit (Vazyme Biotech, Nanjing, China). The pCAMBIA1302-CYP77B34-GFP vector was transformed into Escherichia coli, and the insert fragment sequence was confirmed by sequencing. The pCAMBIA1302-CYP77B34-GFP plasmid was purified and then transferred into the Agrobacterium tumefaiciens strain GV3101 (Biomed, Beijing, China).

Subcellular Localization
Tobacco leaves' transient expression systems were used to determine the subcellular localization of the CYP77B34 protein. The subcellular localization was conducted according to the methods described by Pang et al. [54]. pCAMBIA1302-CYP77B34-GFP, the empty vector pCAMBIA1302-GFP and endoplasmic-reticulum (ER)-specific marker (mCherry-HDEL) were co-expressed in tobacco leaves. A. tumefaciens (GV3101 strain) transformed with pCAMBIA1302-CYP77B34-GFP, pCAMBIA1302-GFP or mCherry-HDEL was incubated at 28 • C with shaking (200 rpm) for 18 h. Bacteria were collected by centrifugation at 4000 rpm for 4 min and were resuspended to a final OD 600 of 1.5 in an MES buffer (10 mM of MgCl 2 and 10 mM of MES, pH 5.6) containing 150 nM of acetosyringone and were rested for 3 h. The bacteria suspensions mixed with a p19 silencing suppressor were infiltrated into leaves of N. benthamiana. Three days later, leaves of N. benthamiana were collected for microscopic observation using a Leica Stellaris5 confocal microscope (Leica, Heidelberg GmbH, Heidelberg, Germany). The green fluorescent protein (GFP) signal was excited at 488 nm and detected in the 500-550 nm range. The mCherry signal was excited at 561 nm and detected at 590-640 nm. The ImageJ plugin coloc 2 was used to measure Pearson's correlation coefficient. More than 10 pictures were quantified for each co-localization.

Arabidopsis Transformation and Transcript Analysis of CYP77B34
A. tumefaiciens-transformed pCAMBIA1302-CYP77B34-GFP or pCAMBIA1302-GFP were transformed into A. thaliana by the standard floral dip method [55]. Transgenic Arabidopsis were selected on MS plates containing 25 µg mL −1 of hygromycin B, and homozygous plants were obtained by single plant reproduction for 3 consecutive generations. At least 4 independent transgenic lines, which was confirmed by PCR, were collected for each construct. The expression levels of CYP77B34 in each line of transgenic Arabidopsis were determined by qRT-PCR.
Transgenic Arabidopsis at the 6-leaf stage was used for transcript analysis. The experiment was carried out on three biological and three technical replications. The GAPDH gene was selected as a reference gene, and the primer information is shown in Table 2. The qRT-PCR program was the same as that in expression levels' determination of CYP77B34 in D. sophia.
where b is the slope at GR 50 , C and D are the lower and upper limits, respectively. The resistance index (RI) was calculated by GR 50 of the transgenic Arabidopsis expressing CYP77B34-GFP divided by that of the transgenic Arabidopsis expressing GFP to estimate the resistance levels.

Tribenuron-Methyl Metabolism Analysis in Transgenic Arabidopsis Using LC-MS
Transgenic Arabidopsis plants expressing GFP and CYP77B34 were treated with tribenuronmethyl (0.16 µg/plant) at the 6-leaf stage. Above-ground parts of transgenic Arabidopsis plants were harvested at 0, 1, 3, 5 and 7 DAT. For each time point, five plants were harvested as one replicate per line, and four replicates were applied.
The residual tribenuron-methyl on the surfaces of transgenic Arabidopsis plants was washed with 8 mL of acetonitrile. The elution of acetonitrile was filtered with a 0.22 µm syringe filter, and tribenuron-methyl was quantified by LC-MS (Agilent 1290 InfinityII-6470, Palo Alto, CA, USA). Above-mentioned plant tissues were used for tribenuron-methyl metabolism measurement. The plants were ground into powder in liquid nitrogen and sonicated for 10 min with 5 mL of acetonitrile (with 1% acetic acid). After adding 0.5 g of NaCl, the homogenate was vibrated by a vortex for 2 min and centrifuged at 5000 rpm for 5 min. A 1 mL supernatant acetonitrile layer was added into a 2 mL centrifuge tube (containing 50 mg of anhydrous MgSO 4 , 25 mg of PSA and 7 mg of GCB) and vibrated by a vortex for 1 min. Then, the supernatant was filtered with a 0.22 µm syringe filter after centrifugation at 12,000 rpm for 3 min and analyzed by LC-MS.
The analysis of tribenuron-methyl was conducted by LC-MS with Agilent HPLC packed Shim-pack GIST C18 (50 × 2.1 mm i.d., 2 µm). The mobile phase was composed of 80% water and 20% acetonitrile, and the flow rate was 0.4 mL/min. Injection volume was 2 µL, and run time was 2.5 min. MS ran at conditions of DL temperature of 250 • C, heat block temperature of 300 • C, nebulizing gas flow of 5.0 L/min and drying gas flow of 11.0 L/min. The multiple reaction monitoring mode (MRM) for tribenuron-methyl was optimized at 396.2 > 155.1 under the above-mentioned conditions. Linearity, recovery and precision were evaluated to ensure the quality of the analytical method. Linearity was determined by injecting tribenuron-methyl working standard solutions at concentrations of 0.001, 0.005, 0.01, 0.05, 0.1 and 0.2 mg/L followed by linear regression analysis. Precision was calculated as relative standard deviation (RSD) from recovery studies with standard spiked samples (n = 4) at levels of 0.01, 0.1 and 0.2 mg/L (for the elution matrix) and 0.016, 0.16 and 0.32 mg/L (for the extract matrix).
Data of tribenuron-methyl absorption and metabolism were subjected to ANOVA with Duncan's test by IBM SPSS Statistics 21. Four replicates were conducted for each line.

Conclusions
In summary, CYP77B34 from D. sophia is a typical P450 protein localized in the ER. The constitutive and induced expression levels of the CYP77B34 gene in TR D. sophia was significantly higher than that in TS plants. Arabidopsis expressing CYP77B34 exhibited obvious resistance to tribenuron-methyl, carfentrazone-ethyl and oxyfluorfen. The P450 inhibitor malathion could reverse the transgenic Arabidopsis resistance to three herbicides. Compared to that of CK-GFP, three 77B34 lines possessed stronger metabolic ability for tribenuron-methyl. In addition, transgenic Arabidopsis expressing CYP77B34 also displayed tolerance of bromoxynil and pretilachlor. The results provided direct evidence on CYP77B34 endowing D. sophia resistance to tribenuron-methyl. At the same time, it also proved that cross-resistance patterns caused by P450 could be unpredictable.