The resistant A. aequalis population (KMN-R) was gathered in the Chinese Anhui wheat field where mesosulfuron-methyl was applied for at least 5 years. The susceptible population (KMN-S) was gathered from the vacant field where mesosulfuron-methyl was never applied. Seeds from 60 mature seedlings were gathered stochastically manually in each population. All collected seeds were dried and preserved in plastic bags until use.

Herbicides applied in the response test were shown in Table S1 . GST inhibitor 4-chloro-7-nitrobenzoxadiazole (NBD-Cl, 97%) and cytochrome P450 inhibitor malathion (95%) were bought from Sigma.

About 30 A. aequalis seeds were seeded in each pot containing nutrient soil, and the pot was moved to artificial climate chambers with a 14-h day of 20°C and 10-h night of 15°C at 75% humidity. After growing for 4 weeks, seedlings were thinned to 20 in each pot and sprayed with tested ALS-inhibiting herbicides ( Table S1 ) and returned to the greenhouse. After treating for 3 weeks, aboveground seedlings were cut to determine the fresh weight.

Fresh leaf tissue (100 mg) in each plant was used to extract genomic DNA (gDNA) using a Tiangen Biotech Co. DNA Kit with instructions. Primers ( Table S2 ) were designed and used to amplify A. aequalis ALS gene fragment involving all known ALS-inhibiting herbicide resistance-related mutations. Polymerase chain reaction (PCR) was conducted, and the PCR run was set as described (Wang et al., 2022). Amplified PCR products were also purified, cloned, and sequenced as described (Wang et al., 2022). Ten plants randomly selected from each population were used for sequencing. Sequences were aligned using BioEdit sequence alignment editor software and DNAMAN.

Seeds from A. aequalis populations were cultured by following the same method as above. A. aequalis seedlings were sprayed with mesosulfuron-methyl, NBD-Cl, NBD-Cl plus mesosulfuron-methyl, malathion, and malathion plus mesosulfuron-methyl. NBD-Cl (270 g a.i. ha^-1) was used 48 h before mesosulfuron-methyl treatment, and malathion (1,000 g a.i. ha^-1) at 2 h. Mesosulfuron-methyl was applied following the doses as above. Twenty-one days after mesosulfuron-methyl application, photographs were recorded for growth status and aboveground weight per pot was determined.

TaKaRa Biotech RNAiso Plus was used to extract total RNA from three KMN-R and three KMN-S seedlings with instructions. NanoPhotometer^® spectrophotometer was used to analyze the quantity and quality of total RNA. RNA sequencing (RNA-seq) was performed using Illumina platform to generate paired-end reads. Data from KMN-R and KMN-S seedlings were merged to assemble with Trinity (Grabherr et al., 2011) for generating the reference transcriptome. The clean read was filtered from raw read for downstream analyses, and clean data were mapped to reference for obtaining each gene read count. The gene read count for each transcript was normalized to fragments per kilobase of exon per million fragments mapped (FPKM). The expression level of each gene was estimated using RNASeq by expectation maximization (RSEM) (Li and Dewey, 2011). Differential expression between KMN-R and KMN-S was compared using t-test (p < 0.05). Differentially expressed genes (DEGs) were analyzed for Gene Ontology (GO) enrichment using GOseq R packages (Young et al., 2010).

The candidate NTSR contig was chosen based on the following criteria: upregulated for >2-fold in KMN-R than KMN-S (p < 0.05) and related to GST, ABC transporter, and GT annotation. Samples provided for RNA-seq were used to validate GST, ABC transporter candidate, and GT contig using quantitative real-time PCR (qRT-PCR). Primers ( Table S3 ) for GST, ABC transporter, and GT contig were designed using an online software (https://sg.idtdna.com/scitools/Applications/RealTimePCR/). The eukaryotic translation initiation factor 4A (eIF4A) gene was used as an internal control in A. aequalis (Zhao et al., 2018b). Only a particular PCR band was amplified in each primer, and the negative control did not amplify anything. The amplified PCR product was sequenced to confirm the anticipated gene. qRT-PCR was performed using ABI-7500 Fast Real-Time PCR System with TaKaRa SYBR^® Premix Ex Taq™ kits. Primer efficiencies of all studied genes were 94%–115%. qRT-PCR experiments and expression level analyses were conducted as described (Pan et al., 2019).

Each treatment contained three replicates, and every experiment was repeated twice. Herbicide effective rates reducing 50% plant inhibition in fresh weight (GR₅₀) were calculated using four-parameter log-logistic curve for fitting in SigmaPlot software as described (Wang et al., 2022). Resistance indexes (RIs) were determined as the GR₅₀ of KMN-R divided by the GR₅₀ of KMN-S to indicate resistant levels for the KMN-R population.

To remove surface fungi in A. aequalis, sterile water, 70% alcohol, and sodium hypochlorite solution (with 2.5% active Cl^-1) were used to separate soil from seedlings, then sterile water was used to wash the seedlings five times, and sterile absorbent papers were used to remove water. Ultimately, seedlings were cut into small fragments using flame-sterilized scalpels and stored at -80°C until use. All procedures were conducted under a sterile environment. Power Soil DNA Isolation Kit was used to extract microbial DNA from 1.0 g A. aequalis (wet weight). Primer pairs ITS1F (5’-CTTGGTCATTTAGAGGAAGTAA-3’) and ITS2R (5’-GCTGCGTTCTTCATCGATGC-3’) were used to generate fungal internal transcribed spacer (ITS) amplicon libraries. PCR was conducted, and the PCR run was set as described (Wang et al., 2019). Amplified PCR products were purified on 2% agarose gel. Illumina MiSeq platform was used to paired-end sequence (2 × 300 bp for fungi) purified amplicons by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Data were analyzed using Majorbio I-Sanger Cloud Platform (www.i-sanger.com).

Trimmomatic (Bolger et al., 2014) and FLASH (Magoč and Salzberg, 2011) were used to demultiplex, quality-filter, and merge raw fastq files with the following criteria: (a) Reads cut at any site can receive average quality scores <20 over 50-bp moving windows. (b) Primers used were correctly matched, permitting two nucleotide mismatches, and a read involving a blurred base was removed. (c) Sequences with an overlap of >10 bp were merged on account of the overlapped parts. Then, the remaining distinct sequences were calculated for pairwise distance and created a distance matrix. A 0.03 operational taxonomic unit (OTU) definition (97% sequence similarity cutoff level) was used to cluster OTUs using UPARSE, and OTUs contained multitude unity taxonomy. Singleton was removed from the dataset to minimize the effect of sequencing artifacts (Dickie, 2010). UCHIME (Edgar et al., 2011) was used to identify and remove chimeric sequence. The RDP Classifier algorithm against the UNITE v.7 ITS database (fungi) was used to analyze the taxonomy of ITS sequences with a confidence threshold of 70%.

Whole plant response tests showed that the KMN-R population evolved resistance to mesosulfuron-methyl ( Table 1 ). The GR₅₀ value of KMN-R was 131.03 g a.i. ha⁻¹, while the GR₅₀ value of KMN-S was 1.14 g a.i. ha⁻¹. The estimated RI for KMN-R was 114.94-fold by dividing the GR₅₀ value of KMN-R to that of KMN-S, indicating high-level mesosulfuron-methyl resistance. KMN-S was susceptible to five tested ALS-inhibiting herbicides ( Figure 1 ). Compared with KMN-S, the resistant KMN-R endowed resistance to the five ALS-inhibiting herbicides, including rimsulfuron, imazamox, pyroxsulam, bispyribac-sodium, and flucarbazone-sodium ( Figure 1 ).

The length of the A. aequalis ALS partial gene sequences was 1,917 bp. Sequence alignment between KMN-R and KMN-S samples assumed the TGG to TTG nucleotide variation, causing Trp-574-Leu mutation in KMN-R ( Figure 2 ), confirming that this known resistance-conferring substitution existed in the KMN-R population.

Malathion at 1,000 g a.i. ha^-1 or NBD-Cl at 270 g a.i. ha^-1 did not influence the development of A. aequalis seedlings. NBD-Cl or malathion had no impact on the susceptibility of the plants in susceptible population KMN-S to mesosulfuron-methyl ( Table 1 ). Treatment with malathion did not increase the toxicity of mesosulfuron-methyl to plants in resistant population KMN-R ( Table 1 ). However, treatment with NBD-Cl combined with mesosulfuron-methyl increased mesosulfuron-methyl toxicity to plants in KMN-R ( Table 1 ), with a markedly decreased GR₅₀ value from 131.03 to 47.16 g a.i. ha^-1.

Owing to the absence of the A. aequalis genome, the reference transcriptome in A. aequalis assembled the sequencing data from KMN-R and KMN-S. Then, 36,996 supposed genes with contig N50 size of 1,969 bp were identified from 42,875 transcripts. Assembled sequences were annotated using SwissProt, NCBI non-redundant protein sequences (Nr), protein family (PFAM), clusters of orthologous groups of proteins (KOG), GO, and KEGG ortholog (KO) databases ( Table S4 ). Based on FPKM at t-test (p < 0.05), DEGs in the KMN-R and KMN-S were assessed. A total of 1,484 differentially expressed contigs between KMN-R and KMN-S were confirmed ( Figure S1 ). DEG functions were evaluated using GO enrichment analyses, and metabolic process and cellular process were significantly enriched ( Figure S2 ).

Seven upregulated contigs of >2-fold (p < 0.05) in KMN-R than KMN-S in relation to GST, GT, and ABC transporter annotations were validated using prepared RNA-seq samples ( Table 2 ). Two contigs (TRINITY_DN14492_c0_g1 with a GSTZ2 annotation and TRINITY_DN438_c0_g1 with a GSTT3 annotation) were markedly upregulated using qRT-PCR, fold changes of which displayed no differences with RNA-seq results ( Table 2 ). These two highly expressed GST contigs appear to be involved in mesosulfuron-methyl resistance in KMN-R.

A total of 405,386 fungal sequences with 290 bp average length were obtained after filtering and qualifying raw reads. A total of 155 A. aequalis fungal OTUs with 97% similarity were inferred in 5 phyla, 19 classes, 37 orders, 67 families, and 90 genera ( Table S5 ). At the level of the phylum of A. aequalis fungal, the fungal community was dominated by Ascomycota (relative abundance 9.09%), Basidiomycota (relative abundance 3.07%), and Glomeromycota (relative abundance 0.12%) ( Figure 3A ). At the level of the class of A. aequalis fungal, the community was dominated by Dothideomycetes (relative abundance 2.96%), Sordariomycetes (relative abundance 2.05%), Tremellomycetes (relative abundance 1.69%), Eurotiomycetes (relative abundance 1.50%), and unclassified Ascomycota (relative abundance 1.30%) ( Figure 3B ). At the level of the family of A. aequalis fungal, the fungal community was dominated by Mycosphaerellaceae (relative abundance 1.77%), unclassified Ascomycota (relative abundance 1.30%), Trichosporonaceae (relative abundance 1.28%), and Nectriaceae (relative abundance 1.15%) ( Figure 3C ). At the level of the genus of A. aequalis fungal, the fungal community was dominated by unclassified Mycosphaerellaceae (relative abundance 1.77%), unclassified Ascomycota (relative abundance 1.30%), Apiotrichum (relative abundance 1.27%), and Fusarium (relative abundance 1.10%) ( Figure 3D ).

The alpha diversity indices (Simpson and Shannon) computed from fungal OTUs of KMN-R and KMN-S A. aequalis populations indicated that less varied fungal OTUs were found in KMN-S than KMN-R ( Table S6 ). At the phylum level of A. aequalis fungal, ANOVA was used to evaluate all detected phyla to test the relative abundance (%) of KMN-R and KMN-S ( Figure 4A ). Fungal community in KMN-R and KMN-S A. aequalis populations was dominated in four phyla, including Ascomycota, Basidiomycota, and Glomeromycota. It was observed that Ascomycota in KMN-R was significantly more abundant (p < 0.05) than that in KMN-S at the phylum level ( Table 3 ). Particularly, Mortierellomycota was only found with a high relative abundance in the KMN-R A. aequalis.

At the level of the class of A. aequalis fungal, ANOVA was used to evaluate all detected classes to test the relative abundance (%) of KMN-R and KMN-S ( Figure 4B ). The fungal community in KMN-R and KMN-S A. aequalis populations was dominated by 13 classes. Significant enrichments (p < 0.05) were observed in Dothideomycetes, Eurotiomycetes, and unclassified Ascomycota between the KMN-R and KMN-S A. aequalis at the class level ( Table 4 ). Specifically for Taphrinomycetes, Lecanoromycetes, Mortierellomycetes, Cystobasidiomycetes, Leotiomycetes, and Orbiliomycetes, we only observed a high relative abundance in the KMN-R A. aequalis ( Figure 5A ).

At the level of the family of A. aequalis fungal, ANOVA was used to evaluate all detected families to test the relative abundance (%) of KMN-R and KMN-S ( Figure 4C ). Fungal community in KMN-R and KMN-S A. aequalis populations was dominated in 19 families. Significant enrichments (p < 0.05) were observed in Mycosphaerellaceae and unclassified Ascomycota between the KMN-R and KMN-S A. aequalis at the family level ( Table 5 ). In addition, 34 families were only observed with a high relative abundance in the KMN-R A. aequalis, including Taphrinaceae, Didymellaceae, and unclassified Eurotiomycetes ( Figure 5B ), and 14 families were only observed with a high relative abundance in the KMN-S A. aequalis, including Taphrinaceae, Didymellaceae, unclassified Saccharomycetales, Bulleribasidiaceae, and Diatrypaceae ( Figure 6A ).

At the genus level of A. aequalis fungal, ANOVA was used to evaluate all detected genera to test the relative abundance (%) ( Figure 4D ). The fungal community in KMN-R and KMN-S A. aequalis populations was dominated by 17 genera. It was observed that unclassified Ascomycota in KMN-R was significantly more abundant (p < 0.05) than that in KMN-S at the genus level ( Table 6 ). In addition, 52 genera were only observed with a high relative abundance in the KMN-R A. aequalis, including unclassified Mycosphaerellaceae and unclassified Eurotiomycetes ( Figure 5C ), and 21 genera were only observed with a high relative abundance in the KMN-S A. aequalis, including unclassified Saccharomycetales, Humicola, and Dioszegia ( Figure 6B ).