In this research, four populations of E. crus-galli were investigated, each distributed in different geographical regions from China. The phenotypically resistant populations, JL-R (Tonghua city, Jilin province; 42.62°N, 126.07°E), AH-R (Hefei city, Anhui province; 31.25°N, 117.20°E), and HLJ-R (Hegang city, Heilongjiang province; 47.50°N, 130.86°E), were collected from rice fields where penoxsulam had proven ineffectiveness in weed control. In contrast, the susceptible population, YN-S (coordinates 30.91°N, 118.80°E), was obtained from a non-cultivated field close to rice paddies in Yunnan Province, where no herbicides had been previously used. For each of these four E. crus-galli populations, seeds were collected from a minimum of 40 individual plants and stored in a well-ventilated shelf. Penoxsulam (2-(2,2-difluoroethoxy)-N-(5,8-dimethoxy-[1,2,4]triazolo[1,5-c]pyrimidin-2-yl)-6-(trifluoromethyl)benzenesulfonamide, 25 g/L oil dispersion) was obtained from Dow AgroSciences.

Seeds from each of the JL-R, AH-R, HLJ-R, and YN-S populations were randomly selected and germinated in Petri dishes with damp filter papers, then incubated in a growth chamber with a temperature of 30°C/25°C and a 12-h photoperiod. After the shoot length reached approximately 1 cm, seven seedlings were transplanted into four replicate pots filled with a commercial potting soil mixture for each herbicide rate. These pots were placed in a growth chamber set at 30°C/25°C with a 12-h day/night cycle. Penoxsulam was applied at the rates 0 g a.i. ha⁻¹, 1.67 g a.i. ha⁻¹, 5 g a.i. ha⁻¹, 15 g a.i. ha⁻¹, 45 g a.i. ha⁻¹, 135 g a.i. ha⁻¹, and 405 g a.i. ha⁻¹ for the JL-R and HLJ-R populations; 0 g a.i. ha⁻¹, 5 g a.i. ha⁻¹, 15 g a.i. ha⁻¹, 45 g a.i. ha⁻¹, 135 g a.i. ha⁻¹, 405 g a.i. ha⁻¹, and 1,215 g a.i. ha⁻¹ for the AH-R population; and 0 g a.i. ha⁻¹, 0.19 g a.i. ha⁻¹, 0.56 g a.i. ha⁻¹, 1.67 g a.i. ha⁻¹, 5 g a.i. ha⁻¹, 15 g a.i. ha⁻¹, and 45 g a.i. ha⁻¹ for the YN-S population at three- to four-leaf stages of E. crus-galli using an experimental moving-boom sprayer (Model ASS-4, Beijing Research Center for Information Technology in Agriculture, China), equipped with a TeeJet XR8002VS flat-fan nozzle and a pressure of 0.275 MPa, delivering the volume of 450 L ha⁻¹ liquid. E. crus-galli growing in the pots were all harvested, and their dry weight biomass (after 3 days of 75°C oven-dried) was measured 3 weeks after herbicide treatment. The whole-plant dose–response experiment was conducted twice.

Seeds were cultured to three- to four-leaf stage as described in Section 2.2. Young plant leaf tissue, weighing 100 mg for each, was sampled from a minimum of 20 plants of each population. DNA was extracted using the DNAsecure Plant Kit (TianGen Biotech, Co. Ltd., Beijing, China), and total RNA was isolated using the RNA Easy Fast Plant Tissue Kit (TianGen Biotech, Co. Ltd., Beijing, China). First-strand cDNA was synthesized using the HiScript ^® III 1st Strand cDNA Synthesis Kit (Vazyme Biotech Co., Ltd., Nanjing, China).

E. crus-galli, an allohexaploid grass weed, harbors at least three ALS genes (Fang et al., 2022; Iwakami et al., 2015; Panozzo et al., 2021). Primers ALS-1 (forward, 5′-ATCCCCCATCCTCTCCTT-3′; reverse, 5′-GGTCCAGAGTTCACACCCTAG-3′) and primers ALS-2 (forward, 5′-CACCCTCCCCAAACCC-3′; reverse, 5′-CACGGAAACAACAGACTACAT-3′) were designed based on the ALS mRNA sequences of E. crus-galli LC006058.1 and LC006059.1 to amplify the complete ALS1 and ALS2 genes. Additionally, primers ALS-3 (forward, 5′-CCCCAATCCCCCATCCAT-3′; reverse, 5′-GCACCGCTCGCTGAATAC-3′) reported by Iwakami and coworkers were used to specific amplification of ALS3 gene (Iwakami et al., 2015). The fragments of ALS1-3 cover the known eight resistance-conferring mutation sites. For polymerase chain reaction (PCR), LA Taq^® was used with GC Buffer (Takara Biomedical Technology Co., Ltd., Beijing, China) according to the manufacturer’s instructions. The purification and TA cloning of the PCR products were conducted using the methods outlined in a previous report (Sun et al., 2023). A minimum of eight white colonies were carefully selected and then sent to Tsingke Biotechnology Co., Ltd. (Beijing, China) for Sanger sequencing. The resulting sequences were aligned with the documented ALS genes of E. crus-galli (accession numbers, LC006058.1, LC006059.1, and LC006063.1) using NCBI-BLAST and analyzed using DNAMEN 9.0.1 (Lynnon Corporation, Quebec, Canada) and SeqMan Pro (v7.1.0, DNASTAR Lasergene) (Clewley, 1995).

The extraction and evaluation of the ALS enzyme were conducted following the methods outlined by Yu with coworkers (Yu et al., 2004) and Han with coworkers (Han et al., 2012) with minor modifications. E. crus-galli seedlings (purified and ensured to carry the special mutation for all plants) were cultivated to the three- to four-leaf stage following the aforementioned method. Approximately 4 g of leaf tissue was sampled from each population, with a minimum of 30 seedlings harvested in each sample. The frozen leaf material was homogenized using a mortar and pestle in 8 mL of grinding buffer. Subsequently, an equal volume of 100% (NH4)₂SO₄ was added dropwise to the solution, followed by centrifugation. The resulting ALS pellet was redissolved in 3.5 mL resuspension buffer. ALS protein was desalted using a Sephadex G25 column with a 5-mL elution buffer.

For the enzyme assays, 100 μL of desalted enzyme extraction solution and 100 μL of penoxsulam at different concentrations were added to a 1.5-mL centrifugation tube. The dosage of penoxsulam was set at 0 μmol·L⁻¹, 0.00001 μmol·L⁻¹, 0.0001 μmol·L⁻¹, 0.001 μmol·L⁻¹, 0.01 μmol·L⁻¹, 0.1 μmol·L⁻¹, 1 μmol·L⁻¹, 10 μmol·L⁻¹, and 100 μmol·L⁻¹ for YN-S population, while it was set at 0 μmol·L⁻¹, 0.0001 μmol·L⁻¹, 0.001 μmol·L⁻¹, 0.01 μmol·L⁻¹, 0.1 μmol·L⁻¹, 1 μmol·L⁻¹, 10 μmol·L⁻¹, 100 μmol·L⁻¹, and 1,000 μmol·L⁻¹ for the JL-R, AH-R, and HLJ-R populations, respectively. The ALS enzyme concentration was determined using the Easy Protein Quantitative Kit (Bradford) (TransGen Biotech, Beijing, China), and the absorbance at 530 nm was measured using the FlexStation 3 full-wavelength scanning multifunctional enzyme marker (MD Electronics, USA) to determine ALS enzyme activity. The assay was conducted twice, with each treatment tested in three replicates.

The F1 generation of bagged selfed seeds from E. crus-galli mutated plants were all harvested for this experiment. After E. crus-galli seedlings reached three- to four-leaf stage, each tested population was divided into two subgroups. One subgroup was treated with 15 g a.i. ha⁻¹ penoxsulam, while the other was treated with water. The leaf tissues were collected at 0 d, 1 d, 2 d, and 3 d after treatment with 10 plants sampled each day, respectively. The samples were immediately flash frozen in liquid nitrogen. Total RNA extraction and first-strand cDNA was synthesized as previously described. The ALS gene and β-actin reference gene were according to previous reports (Fang et al., 2022, Fang et al., 2019b). qRT-PCR was conducted following the manufacturer’s instructions for Taq Pro Universal SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd., Nanjing, China), and the qRT-PCR program was executed using an ABI 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). The relative quantification of the ALS gene was calculated using the 2^−ΔΔCt method, where ΔCT =ALS (Mean CT) − Actin (Mean CT), ΔΔCT = ΔCT (treatment) − ΔCT (control) (Bustin et al., 2009; Livak and Schmittgen, 2001). ALS expression level was normalized to the Ct values for the YN-S population at 0 d (water treatment). Each process was repeated three times. Independent sample T-test was performed using SPSS v25.0 (IBM, Armonk, NY, USA) software to determine significant differences in the expression levels.

The ALS gene sequence alignment revealed three different mutations in the populations of E. crus-galli: JL-R (Pro-197-His, ALS1), AH-R (Trp-574-Leu, ALS1), and HLJ-R (Trp-197-Leu, ALS2). The amino acid sequences of these three mutated proteins, along with two non-mutated proteins (listed in Supplementary Table S1 ), were employed as query templates to the homology model using SWISS-MODEL (https://swissmodel.expasy.org/) (Waterhouse et al., 2018). The ALS protein (PDB ID 3e9y) from A. thaliana was used as the template protein. The homology modeling results were uploaded to SAVES (http://services.mbi.ucla.edu/SAVES/) to evaluate by PROCHECK. The chemical structure of penoxsulam was downloaded from the PubChem database (CID: 11784975) (https://pubchem.ncbi.nlm.nih.gov/). Molecule docking simulations were performed by Autodock Vina (Trott and Olson, 2010), according to previously reported method with a slight modification (Butt et al., 2020). The docking box was positioned at the active site, as reported in previous literature (Fang et al., 2022; Wang et al., 2009), where the protein bound with the penoxsulam ligand. The box size was set to 15 Å × 15 Å × 15 Å, with the active site coordinates as follows: center_x = 55.026, center_y = 50.923, and center_z = 46.274. The parameter settings were exhaustiveness = 400 and num_modes = 20, and other parameters were set to their default values.

Three mutant and two non-mutant ALS protein-coding sequences (CDS) were codon optimized, synthesized, and confirmed through Sanger sequencing by Sangon (Sangon Biotech (Shanghai) Co., Ltd.). The Bac-to-Bac baculovirus expression system and the purification of the five target proteins were conducted following the methods described in previous literature (Fang et al., 2022). The expression vector for the target gene was constructed based on Fang’s method and transformed into DH5α cells for amplification. Plasmid extraction and PCR verification were performed to confirm correctness. The plasmid was then transferred into DH10Bac competent cells via a transposition reaction, following the instructions of the Bac-to-Bac^® TOPO^® Expression System (Thermo Fisher Technology Co. Ltd, Shanghai, China). SF9 cells were inoculated at a density of 1 × 10⁶ cells mL⁻¹ in 1-L culture flasks, totaling 200 mL of cells. To infect the cells, 0.8 mL of P4 viral stock was added to the 200-mL cell suspension. After 3 d, cells were collected and centrifuged at 4°C. The supernatant was then collected for ALS protein purification. The quantification of the expressed ALS proteins (diluted in elution buffer) was performed as mentioned above. The protocol used for the in vitro heterologous ALS activity assay closely resembled that for ALS isolated from E. crus-galli, with the exception that in the former case, the penoxsulam concentration was set at 1 μM. Each treatment was performed with three replicates.

The whole-plant dose–response bioassay data were subjected to a four-parameter log-logistic equation analysis using SigmaPlot 12.5 software (Systat Software, Inc., San Jose, CA, USA) to calculate the GR₅₀ value (the dose causing a 50% reduction in above-ground dry weight). The regression equation used is as follows:

In this equation, y represents the ratio of dry weight at x herbicide dose to that of the control, x represents the herbicide treatment dose, C represents the lower limit of the dose–response, D represents the upper limit of the dose–response, and b denotes the slope of the curve.

Based on the GR₅₀ values obtained for each population, the resistance index (RI) was calculated using the following formula:

The same analysis calculated the herbicide concentration required to inhibit 50% of ALS activity (I₅₀) and its corresponding RI value. The resistance classification criteria were as follows: RI < 2, indicating a susceptibility; 2 ≤ RI < 5, indicating a low-level resistance; 5 ≤ RI < 10, indicating a moderate-level resistance; and RI > 10, indicating a high-level resistance.

All seedlings of the S population, YN-S, were completely inhibited at 15 g a.i. ha⁻¹ penoxsulam with a GR₅₀ value of 2.76 g a.i. ha⁻¹. The GR₅₀ values of the JL-R, HLJ-R, and AH-R populations (R) were 27.26 g a.i. ha⁻¹, 23.91 g a.i. ha⁻¹, and 164.81 g a.i. ha⁻¹, respectively, corresponding to R/S GR₅₀ ratios (RI) of 9.88-fold, 8.66-fold, and 59.71-fold ( Table 2 ; Figure 1 ) compared to the YN-S population, which revealed that the JL-R and HLJ-R populations had evolved moderate-level resistance and AH-R population showed high-level resistance to penoxsulam.

Sequencing confirmed that the ALS genes in E. crus-galli have no introns. Three ALS gene sequences were obtained, with lengths of 1,929 bp (ALS1), 1,932 bp (ALS2), and 1,935 bp (ALS3), encoding 643, 644, and 645 amino acids, respectively. Furthermore, sequence alignment analysis conducted using the COBALT tool on NCBI (https://www.ncbi.nlm.nih.gov/tools/cobalt/) identified five conserved amino acid mutations at specific positions within three distinct ALS copies ( Figures 2 , 3 ). Three nonsynonymous nucleotide mutations were identified in three resistant E. crus-galli populations. The Pro-197-His (CCC to CAC), Trp-574-Leu (TGG to TTG), and Pro-197-Leu (CCC to CTC) mutations caused by a nucleotide mutation were detected in ALS1 in JL-R population, in ALS1 in AH-R population, and in ALS2 in HLJ-R population, respectively ( Figure 4 ). The mutation frequency in all three populations was 100%, and all colonies tested from each population showed consistent results. Additionally, Pro-197-His/Leu mutations were first identified in penoxsulam resistant E. crus-galli, with Pro-197-His mutation being reported for the first time in Echinochloa spp.

The in vitro assay of E. crus-galli ALS activity showed no significant differences (p > 0.05) in total catalytic activity between the resistant populations JL-R (Pro-197-His), AH-R (Trp-574-Leu), and HLJ-R (Pro-197-Leu) compared to the susceptible YN-S population ( Table 3 ). As shown in the dose–response curves ( Figure 5 ), when the penoxsulam dose was 0.01 μmol·L⁻¹, the ALS activity in the YN-S population was significantly inhibited, with an inhibition rate exceeding 50%, while the R populations showed less inhibition. Specifically, the I₅₀ value for JL-R and HLJ-R population was 0.0118 μmol·L⁻¹ and 0.0126 μmol·L⁻¹, only 1.04- and 1.11-fold higher than that of the YN-S population, respectively. However, the I₅₀ value for AH-R population was 0.0470 μmol·L⁻¹, which was 4.12-fold higher than that of the YN-S, indicating a significant decrease in ALS sensitivity to penoxsulam in the AH-R population. Therefore, resistance in the HLJ-R and JL-R populations to penoxsulam may not be associated with a decrease in ALS sensitivity, whereas the reduced sensitivity of ALS in the AH-R population appears to be one of the key factors contributing to resistance development.

In this study, the transcriptional expression differences in ALS genes were compared among the resistant populations JL-R (Pro-197-His), AH-R (Trp-574-Leu), HLJ-R (Pro-197-Leu), and the susceptible population YN-S. The results of ALS gene expression levels, as shown in Figures 6A, B , indicated that there were no significant differences in ALS gene expression at 0 d, 1 d, 2 d, and 3 d after water treatment between the resistant populations JL-R, AH-R, HLJ-R, and the susceptible population YN-S ( Figure 6A ). Furthermore, there were no significant differences in ALS gene expression among these populations with the treatment of penoxsulam at 15 g a.i. ha⁻¹ ( Figure 6B ). Therefore, the resistance to penoxsulam in JL-R, AH-R, and HLJ-R populations appears to be unrelated to the transcriptional level of ALS gene.

To gain a deeper understanding of the potential mechanism driving the differences in herbicide binding specificity to penoxsulam between the non-mutated and the three mutated ALS proteins in E. crus-galli, homology models of the ALS proteins were constructed, and comparative structural analyses were performed. The template protein 3e9y exhibited the highest sequence identity with the target ALS proteins from E. crus-galli, at approximately 75.00% ( Table 4 ). The results of the Ramachandran plot indicated that more than 99% of the amino acid residues in the five different target ALS proteins were situated in the most favored and additional allowed regions ( Table 5 ; Supplementary Figure S1 ). This validation confirmed that the protein model structures were of high quality and could be reliably used for subsequent molecular docking studies.

The homology modeling of ALS proteins included six ligands: two FAD molecules, two magnesium ions (Mg²⁺), and two 2-(cyclohexylamino)ethanesulfonic acid (CHE) molecules. Molecular docking simulations demonstrated consistent binding interactions of penoxsulam with two non-mutated ALS proteins, YN-S_ALS1_643AA and YN-S_ALS2_644AA with binding energies of −7.5 kcal mol⁻¹ and −7.3 kcal mol⁻¹, respectively ( Table 6 ). The binding site of penoxsulam was situated close to the FAD-binding channel within the two subunits of E. crus-galli ALS. In this binding mode, the triazole-pyrimidine ring inserted into the interior of the channel, while the benzene ring structure was located on the outer surface of the channel ( Figure 7 ). However, mutations occurred at H170 (equivalent to A. thaliana H197) in the E. crus-galli ALS1 and L171 (equivalent to A. thaliana L197) in the E. crus-galli ALS2, resulting in changes to the docking mode of penoxsulam with ALS1 and ALS2, with binding energies of −6.3 kcal mol⁻¹ and −6.7 kcal mol⁻¹, respectively ( Table 6 ; Figure 8A ). An increase in binding energy values indicates a reduced affinity of the ALS protein for penoxsulam. Specifically, the H170 mutation resulted in the formation of hydrogen bonds between the H170 residue and the benzene ring, disrupting pre-existing hydrophobic interactions. The penoxsulam benzene ring and sulfonyl group rotated, leading to changes in interactions with residues like F179, V169, and S626 (equivalent to A. thaliana F206, V196, and S653) ( Figures 8A , 9A, B ). Similarly, the ALS2_L171 mutation resulted in changes in the hydrophobic interactions involving the P171 residue in ALS2, leading to alterations in the binding conformation of penoxsulam with ALS2 and changes in interactions with surrounding amino acid residues ( Figures 8B , 9D, E ). Additionally, the mutation at L547 (equivalent to A. thaliana L574) in E. crus-galli ALS1 resulted in a binding energy of −7.2 kcal mol⁻¹ with penoxsulam. The ALS1_L547 mutation disrupted the π–π stacking interaction between W547 and the pyrimidine ring, leading to altered interactions between penoxsulam and the surrounding residues. Notably, the pyrimidine ring of the penoxsulam molecule underwent a significant rotation ( Figures 8A , 9C ).

The recombinant ALS proteins were expressed by the Bac-to-Bac baculovirus expression system with His-tags. The theoretical molecular weight of ALS1 and ALS2 is approximately 69 kDa. Western blot analysis revealed a distinct band approximately 72 kDa, confirming the successful expression of the ALS proteins. The results of purified ALS protein activity to penoxsulam are presented in Table 7 . When treated with 1 μmol·L⁻¹ penoxsulam, the various ALS proteins exhibited distinct responses. Specifically, WT_ALS1_643AA and WT_ALS2_644AA displayed significant inhibition, with ALS activity inhibition rates of 86.81% and 79.26%, respectively. In contrast, ALS proteins with different mutations showed considerably lower activity inhibition rates compared to non-mutated ALS proteins. Notably, the ALS protein with the Trp-574-Leu mutation exhibited the lowest level of inhibition, consistent with our previous in vitro ALS enzyme activity results.