Integrated genomic and transcriptomic analysis reveals candidate genes underlying herbicide resistance in Sorghum

Abstract

Background

Herbicide-resistant germplasms provide critical genetic resources for improving weed control and understanding resistance mechanisms in crops.

Objective

To screen sorghum accessions for tolerance to ACCase and ALS inhibitor herbicides at the seedling stage, identify the major locus and strong candidate gene associated with feproxydim resistance, and verify the gene expression pattern and genetic variation by quantitative real‑time PCR (qRT‑PCR) and KASP genotyping.

Method

A total of 316 sorghum accessions were screened for seedling-stage herbicide tolerance using gradient herbicide treatments. Bulked segregant analysis sequencing (BSA-Seq) was performed on resistant and susceptible gene pools constructed from the F₂ population derived from IS1219 × RTx430. Transcriptome sequencing (RNA-Seq) was conducted on leaf tissues after feproxydim treatment to identify candidate genes within the mapped interval. KASP markers were developed for the functional variation site of the key candidate gene for genotyping validation. Quantitative real‑time PCR (qRT-PCR) was used to measure the relative expression level of the target gene and compare it with the susceptible control line. Protein sequence comparison was used to detect variations in the key candidate gene between resistant and susceptible lines.

Result

In the screening with the ACCase inhibitor 10% feproxydim, IS1219 exhibited high-level resistance. Preliminary screening under the ALS inhibitor mesosulfuron-methyl treatment identified only SJ304 with visible tolerance, which was not subjected to further mapping or validation. BSA-Seq identified a major feproxydim resistance QTL on chromosome 1. RNA-Seq revealed five co-expressed candidate genes in the target interval, among which Sobic.001G431500 (encoding carboxylesterase 17, an α/β‑hydrolase) was markedly upregulated in the resistant line IS1219 but not in the susceptible line RTx430. Quantitative real‑time PCR (qRT-PCR) analysis confirmed that Sobic.001G431500 was significantly upregulated in the resistant line IS1219 compared with the susceptible control. KASP genotyping demonstrated that the IS1219 allele cosegregated with feproxydim resistance. Protein sequence comparison showed that the IS1219 allele carried a missense mutation V300A and a deletion P301_P303 at and after position 300.

Conclusion

These findings identify a major QTL and a strong candidate gene Sobic.001G431500 associated with feproxydim resistance in the sorghum line IS1219, based on differential expression, genetic variation, and genotype–phenotype cosegregation. This study provides valuable genetic resources and functional markers for marker-assisted selection and breeding of feproxydim-tolerant sorghum varieties.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08624-5.

Introduction

Sorghum [Sorghum bicolor (L.) Moench] is a globally important grain, forage, and bioenergy crop with remarkable adaptability to harsh environments such as drought, heat, and nutrient-poor soils [1–3]. However, its production is often severely constrained by weed infestation [4]. Weeds compete with sorghum for essential growth resources such as light, water, and nutrients, while also reducing canopy ventilation and light penetration, which in turn increases the incidence of diseases and insect pests. These effects ultimately cause substantial yield losses and deterioration of grain quality [5].

At present, chemical herbicide application remains the predominant method for weed control in sorghum production. ACCase (acetyl CoA carboxylase) inhibitors are economically critical herbicides for sorghum production, as they provide highly effective post‑emergence control of problematic grass weeds (e.g., johnsongrass, barnyardgrass) that are poorly managed by ALS (acetolactate synthase) inhibitors or conventional tactics [6]. Historically, sorghum’s close genetic relatedness to weedy grasses prevented safe use of ACCase herbicides in conventional systems, but the development of ACCase‑resistant germplasm has unlocked this mode of action as a foundational tool for sustainable weed management. These herbicides deliver > 91% control of key monocot weeds, reduce yield losses from grass competition, and offer a cost‑effective alternative to intensive tillage or labor‑intensive hand weeding [7]. Nevertheless, chemical weeding is frequently associated with issues such as phytotoxicity, the evolution of herbicide-resistant weed populations, and environmental contamination of farmland ecosystems [8]. Alternative strategies, including agronomic and biological control, generally fail to achieve effective and sustainable weed management [9]. Therefore, the identification and utilization of genetic resources and germplasm materials with herbicide resistance have become critical priorities for sorghum improvement [10].

Thanks to recent advances in ALS resistance genetics in sorghum, herbicide-resistant germplasms have been developed through approaches such as mutation breeding, natural population screening, and molecular marker-assisted selection, providing key genetic resources for the breeding of herbicide-tolerant sorghum varieties. The herbicide resistance mechanisms identified to date can be broadly classified into two categories: target-site resistance (TSR) and non-target-site resistance (NTSR) [11]. TSR results from mutations in herbicide target genes that reduce the affinity between the herbicide molecule and its target protein, thereby conferring tolerance [12]. For example, Zhang et al. [13] identified a novel OsEPSPS allele in rice harboring an Asp-213-Asn substitution within the predicted glyphosate-binding domain, which conferred tolerance to glyphosate at four times the recommended field concentration. In sorghum, two allelic variants of the SbALS gene—sbals-1 (A93T) and sbals-2 (S624N)—have been demonstrated to confer strong resistance to imidazolinone herbicides, enabling plants to survive treatments up to 16-fold the standard application rate [14]. NTSR, by comparison, minimizes the effective herbicide concentration at its site of action through physiological and biochemical mechanisms including reduced absorption/translocation, enhanced metabolic detoxification, and active efflux [15]. The major molecular systems underlying NTSR involve cytochrome P450 monooxygenases, glutathione S-transferases (GSTs), and ATP-binding cassette (ABC) transporters [16]. For instance, CYP81A6, a cytochrome P450 gene in rice, has been shown to confer broad-spectrum tolerance to bentazone and metsulfuron-methyl—two herbicides widely used in rice and wheat cultivation [17]. Although wild sorghum populations have evolved resistance to ALS [18] and DIM class ACCase inhibitors, clethodim and sethoxydim [19], a new DIM class herbicide, feproxydim, has been developed by CYNDA for controlling grass weeds resistant to ALS inhibitors and the FOP class of ACCase inhibitors [20]. Previous studies have demonstrated the involvement of carboxylesterases in herbicide/pesticide resistance. For example, knockout of a major Arabidopsis carboxylesterase increases resistance to the Phenoxy class herbicide 2,4-D-methyl [21] and lower expression another carboxylesterase is associated with increased resistance to haloxyfop-P-methyl, an ACCase-inhibiting herbicide, in crabgrass [22]. Microbial studies have shown that increased carboxylesterase expression can degrade chlorimuron-ethyl (an ALS inhibitor) [23] and increase tolerance to fungicides azoxystrobin, kresoxim-methyl, pyraclostrobin, trifloxystrobin, iprodione, and carbendazim [24]. However, no report of carboxylesterase in sorghum herbicide resistance has been found [20].

To fill these defined knowledge gaps and address sorghum weed control challenges, we set three specific objectives: to identify naturally occurring feproxydim-resistant sorghum germplasm, to map the genetic locus controlling feproxydim resistance using BSA-Seq, and to integrate RNA-Seq to pinpoint candidate metabolic resistance genes. Herbicide tolerance was evaluated at the seedling stage among diverse sorghum accessions, and the identified resistant and susceptible genotypes were used to construct an F₂ segregating population. Whole-genome resequencing of resistant and susceptible bulks via BSA-Seq identified genomic regions associated with herbicide resistance. By integrating transcriptomic data from resistant and susceptible parents under feproxydim stress, key candidate genes were identified. These findings provide valuable insights and technical support for the molecular breeding of herbicide-resistant sorghum.

Materials and methods

Plant materials

A total of 316 sorghum germplasm accessions were used in this study (Table S1), all provided by the Anhui Provincial Key Laboratory of Forage Breeding and Utilization. Among them, the herbicide-resistant accession IS1219 and the susceptible accession RTx430 were selected for further analysis. An F₁ hybrid population was generated by crossing IS1219 (♀) with RTx430 (♂). The harvested F₁ plants were self-pollinated to produce an F₂ segregating population.

Screening of herbicide-resistant Sorghum at the seedling stage and evaluation of resistance levels

For each sorghum accession, five seeds were sown per hole, with four holes per accession, and the experiment was performed in two biological replicates. Germinated seeds with visible radicles were transplanted into seedling trays filled with nutrient soil (hole size: 4.8 × 4.8 × 4.8 cm). The trays were maintained in a greenhouse under controlled conditions and watered from the bottom to ensure adequate moisture. At the five-leaf stage, seedlings were sprayed with 2× the field-recommended concentration of herbicide using a manual pressure nozzle at 0.2 MPa, with a total spray volume of 225 L per hectare (L/ha) and no adjuvants or surfactants added. For feproxydim treatment, the commercial formulation used was 10% feproxydim emulsifiable concentrate (EC), applied at a rate of 3 L/ha. For mesosulfuron-methyl treatment, the commercial formulation used was 30 g/L mesosulfuron-methyl oil-based dispersible concentrate (OD), applied at a rate of 2.4 L/ha. Phytotoxicity symptoms were evaluated according to the five-grade classification method specified in the Agricultural Industry Standards of the People’s Republic of China (Table S2). This scale ranges from grade 0 (no visible phytotoxicity, with growth consistent with the water-treated control) to grade 5 (complete plant death). The corresponding survival rate ranges for each grade are provided in Table S2 for reference.

At the five-leaf stage, seedlings were exposed to a gradient of herbicide concentrations (Table S3), including the recommended field rate (1×), a control (0×, sprayed with water), and higher dosages. Two types of herbicides—an ALS inhibitor (mesosulfuron-methyl) and an ACCase inhibitor (10% feproxydim)—were tested. The experiment was performed in three biological replicates per dose. Twenty-one days after spraying, plant mortality was recorded, and the LD₅₀ (herbicide dose causing 50% mortality) and resistance index (RI) were calculated.

LD₅₀ values (referred to as IC50 in the software output) were estimated via nonlinear regression in GraphPad Prism 9 (GraphPad Software, USA) using a four-parameter logistic model:

where: y = observed mortality (%), χ = log₁₀-transformed herbicide dose, Bottom = minimum response, Top = maximum response, HillSlope = slope of the curve, and LD₅₀ = dose corresponding to 50% mortality. 95% confidence intervals (CIs) for LD₅₀ were computed using the profile likelihood method implemented in GraphPad Prism, and the resistance index (RI) was calculated as the ratio of the LD₅₀ of the resistant accession to the LD₅₀ of the susceptible control.

RNA-Seq analysis

Resistant sorghum line IS1219 and susceptible line RTx430 were treated with 10% feproxydim at an application rate of 1.5 L/ha when plants reached the five-leaf stage. Leaf samples were collected from both genotypes 48 h after herbicide application, with three biological replicates per accession. Total RNA was extracted and sent to BGI-Wuhan (Wuhan, China) for transcriptome sequencing. Sequencing reads were aligned to the Sorghum bicolor reference genome (version 5.1) for downstream analysis. Total RNA integrity was assessed using an Agilent 2100 Bioanalyzer, and only samples with RIN values ≥ 8.0 were used for sequencing. Sequencing was performed with a read length of 150 bp paired-end, and the average sequencing depth per library was 9.93× (Table S10). Raw reads were quality-checked using FastQC v0.11.9, and clean reads were aligned to the reference genome. Gene expression levels were normalized using the TMM (Trimmed Mean of M-values) method with FeatureCounts v2.0.0. Differentially expressed genes (DEGs) were identified with DESeq2 package (version 1.36.0) under the criteria of |log₂FoldChange| ≥ 1 and FDR < 0.05. Functional enrichment analysis of DEGs was performed for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) terms in the R environment, with p < 0.05 considered significant.

Bulked Segregant Analysis Sequencing (BSA-Seq)

The F₂ population was grown in pots under greenhouse conditions. Leaf tissues were sampled from each individual at the 5-leaf stage for DNA extraction, after which the plants were immediately treated with 10% feproxydim at an application rate of 3 L/ha. Two weeks after herbicide application, resistant and susceptible individuals were identified based on their phenotypic responses. According to phenotypic classification, 19 extremely resistant and 19 susceptible plants were selected to construct two bulks for bulked segregant analysis (BSA). High-quality genomic DNA was extracted from individual plants using the DNAsecure Plant Genomic DNA Extraction Kit (Tiangen, Beijing, China; Catalogue No. DP320-03). DNA concentrations were quantified using a Qubit 4 Fluorometer (Thermo Fisher Scientific, USA) and diluted to 20 ng/µL. Sequencing libraries were prepared and sequenced on the DNBSEQ-T7 platform (MGI, Wuhan, China). The achieved mean sequencing depths were 38.84× for the parent bulk, 36.69× for the resistant bulk (R-bulk), and 36.74× for the susceptible bulk (S-bulk) (Table S11). Raw sequencing reads were quality-checked using FastQC v0.11.9, and high-quality reads were aligned to the Sorghum bicolor reference genome (v5.1) using BWA-MEM. Single nucleotide polymorphisms (SNPs) were identified with samtools (v1.10) and filtered with the default criteria of QTL-seq v2.2.9: minimum read depth ≥ 8× per site, minimum base quality (Phred score) ≥ 18, and minimum mapping quality ≥ 40. SNP-index values for each bulk were calculated using QTL-seq v2.2.9 with a sliding window approach using the default parameters (2000 kb for window size, 100 kb for step size). Δ(SNP-index) was computed as the absolute difference between the SNP-index of the resistant bulk and susceptible bulk. Confidence intervals were computed via the --N-rep parameter (set to the default value of 5000 replicates in QTL-seq v2.2.9) to generate a null distribution of Δ(SNP-index), and candidate genomic regions associated with herbicide resistance were identified based on the 95% and 99% thresholds derived from this distribution.

Kompetitive Allele-Specific PCR (KASP) genotyping

KASP primers (Table S4) targeting the candidate SNP were designed using Primer Premier 5 software (PREMIER Biosoft, USA). The genotyping population comprised an F₂ segregating population derived from the cross IS1219 (♀) × RTx430 (♂). Leaf tissues were sampled at the 5-leaf stage, after which the plants were immediately treated with 10% feproxydim at an application rate of 3 L/ha. Genomic DNA was isolated from the leaf samples using a magnetic bead-based extraction method. KASP genotyping was performed using the FLU-ARMS Master Mix (Videger, China; Catalogue No. BGH1001RV4) following the manufacturer’s standard protocol. Fluorescence signals were detected on an Applied Biosystems ViiA 7 Real-Time PCR System (Thermo Fisher Scientific, USA), an instrument equipped with an OptiFlex™ optical system that supports multi-channel fluorescence detection for simultaneous reading of FAM and HEX signals. Allele calling was conducted using the instrument’s built-in genotyping software to distinguish homozygous (IS1219 or RTx430 allele) and heterozygous genotypes.

Quantitative PCR (qPCR)

The IS 1219 and RTx430 plants cultivated to the 5-leaf stage were initially sampled, immediately followed by application of 10% feproxydim at a rate of 1.5 L/ha. Leaf tissue samples were collected at 6, 12, 24, and 48 h after treatment. RNA was extracted from leaves using the RNAprep Pure Plant Total RNA Extraction Kit (TIANGEN, Beijing, China; Catalogue No. DP432) following the manufacturer’s instructions. ToloScript all-in-one RT EasyMix for qPCR Kit (TOLOBIO, Shanghai, China; Catalogue No. 22107-01) was used to reverse transcribe mRNA into cDNA. Applied Biosystems real-time fluorescence quantitative PCR reagent (Thermo Fisher Scientific, Waltham, MA, USA) was used. qPCR primers (Table S5) were designed using the primer design website https://www.idtdna.com/pages/tools/primerquest (accessed on 26 January 2026), with PP2A as the reference gene [25]. qPCR reaction was in a 20 µL volume and each sample was replicated three times. The reaction contained 5 µL cDNA, 2 µL each primer (0.1 nmol/µL), 10 µL 2 × Q3 SYBR Qpcr Master mix (TOLOBIO, Shanghai, China; Catalogue No. 22204-01) and 3 µL ddH₂O. The samples were run with one cycle of 95 °C for 30 s, 40 cycles of 95 °C for 10 s, 60 °C for 30 s, and one cycle each of 95 °C for 15 s, 60 °C for 1 min, and 95 °C for 15 s. Melting curves, melting temperatures and Ct values were analyzed with QuantStudio™ Real-Time PCR software v1.6.1, where Ct values were used to calculate relative expression. qPCR was used to validate candidate gene expression.

Data analysis

For the dose–response assay, plants were treated with a series of herbicide concentrations, and plant mortality was recorded after treatment. Dose–response curves were constructed using GraphPad Prism (v9.0; GraphPad Software, San Diego, CA, USA) to assess the herbicide sensitivity of each genotype and to estimate the lethal dose causing 50% mortality (LD₅₀). For the F₂ segregation population, the numbers of resistant and susceptible plants were recorded based on phenotypic evaluation. A chi-square (χ²) test was performed in Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA) to assess the goodness of fit between the observed segregation ratio and the expected Mendelian ratio. Specifically, two hypotheses were tested: (1) resistance is dominant, with an expected 3:1 (resistant: sensitive) ratio, yielding a χ² value of 0.690 (P > 0.05); and (2) resistance is recessive, with an expected 1:3 ratio, yielding a χ² value of 312.690 (P < 0.05). The degrees of freedom (df = 1) were calculated as the number of phenotypic classes minus one, and a significance threshold of P < 0.05 was used to evaluate the fit.

Results

Screening of herbicide-resistant Sorghum materials at the seedling stage

As shown in Table 1, among the 316 sorghum accessions screened, only a few exhibited noticeable tolerance to herbicide treatment. Under application of the ACCase inhibitor 10% feproxydim, five accessions (1.59% of the total) showed reduced phytotoxicity: IS1219 (grade 2) displayed the highest resistance, while IS 7987, SJ72, SJ85, and PI 586,058 (all grade 4) exhibited moderate tolerance. It should be noted that a higher grade corresponds to greater herbicide-induced damage. For the ALS inhibitor mesosulfuron-methyl, only two accessions—SJ304 (grade 2) and PI 564,001 (grade 4)—showed visible tolerance, accounting for 0.64% of all materials tested. These results indicate that herbicide resistance is rare among the tested germplasm, and that IS1219 is the most promising resistant materials for subsequent dose–response and molecular analyses.

Table 1.

Identification of phytotoxicity grades of 316 sorghum accessions treated with two herbicides during seedling-stage screening

Herbicide (Target enzyme)	Phytotoxicity Grade	Accession Number	Accessions	Percentage (%)
10% Feproxydim(ACCase)	2	1	IS 1219	0.32
	4	4	IS 7987, SJ72, SJ85,  PI 586,058	1.27
Mesosulfuron-methyl(ALS)	2	1	SJ304	0.32
	4	1	PI 564,001	0.32

Analysis of resistance levels of Sorghum accessions at the seedling stage under different herbicide concentrations

As shown in Table 2, significant differences in herbicide sensitivity were observed among the tested sorghum genotypes. For each genotype, 10 plants were used in each of three biological replicates, and the survival rate was recorded after herbicide application to assess herbicide response. For the ACCase inhibitor feproxydim, the resistant line IS1219 exhibited markedly higher tolerance than the susceptible control RTx430. The LD₅₀ value of IS1219 (6.356 L/ha) was approximately 4.65 times greater than that of RTx430 (1.368 L/ha), indicating a high level of resistance to feproxydim. The corresponding regression model showed an excellent fit (r = 0.9962), confirming the reliability of the dose–response relationship. For the ALS inhibitor mesosulfuron-methyl, the resistant accession SJ304 displayed moderate resistance relative to RTx430, with an LD₅₀ value of 4.468 L/ha and a resistance index (RI) of 3.86. The correlation coefficients (r = 0.9875–0.9990) further demonstrated that the fitted regression models were robust and statistically sound.

Table 2.

Resistance Levels of Sorghum Resistant Accessions to Different Herbicides

Herbicide Name	Sorghum Material(R/S)	Regression Equation (y=)	Correlation Coefficient r	LD50Value (95%CL)	Resistance Index (RI)
10% Feproxydim(ACCase)	IS 1219		0.9962	6.356(5.282 ~ 8.342)	4.65
	RTx430		0.9889	1.368(1.092 ~ 1.806)
Mesosulfuron-methyl(ALS)	SJ304		0.9990	4.468(4.054 ~ 4.985)	3.86
	RTx430		0.9875	1.157(0.5597 ~ 1.699)

Overall, these results reveal that sorghum germplasm exhibits distinct resistance levels to herbicides targeting different enzymatic pathways. The strong resistance of IS1219 to feproxydim may be attributed to mutations in the ACCase target site and/or enhanced metabolic detoxification, while the moderate resistance of SJ304 to mesosulfuron-methyl likely involves ALS gene variation or non–target-site metabolic resistance mechanisms.

Genetic analysis of herbicide resistance and BSA-Seq of the F₂ population

To identify the genetic basis underlying herbicide resistance, an F₂ population of 213 plants was developed by crossing the resistant sorghum line IS1219 with the susceptible line RTx430. Figure 1 shows the phenotypic differences between RTx430 (sensitive control) and IS1219 (resistant accession) 14 days after treatment with 10% feproxydim at an application rate of 3 L/ha. RTx430 exhibited severe phytotoxicity (e.g., plant dwarfing, leaf yellowing and withering), while IS1219 showed no obvious phytotoxicity and maintained good growth status. Herbicide treatments were applied to both the F₁ plants and the F₂ population to evaluate resistance segregation. The F₁ plants exhibited resistance similar to the resistant parent, suggesting that the resistance trait is dominant. A chi-square (χ²) test of the F₂ population indicated that the segregation of resistant and susceptible phenotypes fit a 3:1 Mendelian ratio (χ² = 0.6901, P = 0.406 > 0.05; Table S6), consistent with inheritance controlled by a single dominant gene. These results suggest that herbicide resistance in IS1219 is primarily governed by a major allele conferring dominant resistance.

Fig. 1.

Phenotypes of RTx430 (sensitive control) and IS1219 (resistant accession) 14 days after treatment with 10% feproxydim at an application rate of 3 L/ha

To further validate the genetic locus controlling herbicide resistance, a KASP marker was developed based on the candidate SNP identified by BSA-Seq and used to genotype the F₂ population. The allelic discrimination plot (Fig. 2D) clearly separated the F₂ individuals into three distinct clusters: blue dots representing homozygous Allele 1/Allele 1 (resistant parent type, AA), red dots representing homozygous Allele 2/Allele 2 (susceptible parent type, aa), and green dots representing heterozygous Allele 1/Allele 2 (Aa), with no undetermined calls, indicating high genotyping quality. The correspondence between KASP genotypes and herbicide resistance phenotypes is summarized in Table S7: all individuals with homozygous resistant (AA) or heterozygous (Aa) genotypes exhibited resistance to feproxydim, while all individuals with homozygous susceptible (aa) genotype were susceptible. Specifically, the genotyping results revealed a segregation ratio of 37 homozygous resistant (AA), 66 heterozygous (Aa), and 37 homozygous susceptible (aa) individuals. A Chi-square (χ²) test confirmed that the genotype distribution fit the expected 1:2:1 Mendelian ratio (χ² = 0.4571, P > 0.05; Table S8). Notably, the combined resistant genotypes (AA + Aa) accounted for 103 plants, while the susceptible genotype (aa) comprised 37 plants, consistent with the 3:1 phenotypic segregation ratio observed in the F₂ population. This co-segregation of the KASP marker genotype with the herbicide resistance phenotype strongly supports the presence of a single dominant resistance locus.

Fig. 2.

BSA-Seq analysis for mapping feproxydim resistance loci in the F₂ population. A Genome-wide Δ(SNP-index) plot showing the distribution of SNP-index differences between resistant and susceptible bulks across all 10 sorghum chromosomes. The red line indicates the smoothed trend of Δ(SNP-index), and the green dashed lines represent the 95% confidence interval (p = 0.05). B Localized Δ(SNP-index) plot of chromosome 1 (Chr01), revealing a prominent candidate region associated with herbicide resistance located between 74.47 Mb and 75.37 Mb. C 117 genes annotated in the candidate gene region. D Allelic discrimination plot of the KASP marker used for genotyping the F₂ population. Blue dots represent homozygous Allele 1/Allele 1 (resistant parent type), red dots represent homozygous Allele 2/Allele 2 (susceptible parent type), and green dots represent heterozygous Allele 1/Allele 2

To identify the genomic region associated with feproxydim resistance, a bulked segregant analysis sequencing (BSA-Seq) approach was performed using F₂ population. As shown in Fig. 2A, most regions exhibited Δ(SNP-index) values fluctuating around zero, indicating random allele distribution. However, a distinct peak exceeding the 95% confidence threshold was detected on chromosome 1 (Chr01) ( Fig. 2B). Within this significant region (74.47–75.37 Mb), the Δ(SNP-index) reached its maximum value, suggesting strong association with the herbicide resistance trait. Gene annotation of this 0.9 Mb interval revealed 117 predicted genes ( Fig. 2C, Table S9).

RNA-Seq analysis of Sorghum response to feproxydim treatment

Six RNA-Seq libraries were constructed from the resistant line IS1219 and the susceptible line RTx430 with three biological replicates per genotype, sampled 48 h after treatment with 10% feproxydim at an application rate of 1.5 L/ha. Quality assessment confirmed that all sequencing datasets were of high quality and suitable for downstream analysis (Table S10). Cluster analysis showed that the three biological replicates of each genotype grouped closely together, indicating good experimental reproducibility ( Fig. 3A). A total of 2,578 differentially expressed genes (DEGs) were identified between IS1219 and RTx430, including 1,319 upregulated and 1,259 downregulated genes ( Fig. 3B). Gene Ontology (GO) enrichment analysis revealed that, under herbicide stress, upregulated DEGs were significantly enriched in terms such as intrinsic component of plasma membrane, integral component of plasma membrane, intermediate filament-based process, intermediate filament cytoskeleton organization, and 2-oxoglutarate-dependent dioxygenase activity ( Fig. 3C; Table S12).

Fig. 3.

Transcriptomic analysis of sorghum response to feproxydim treatment

KEGG pathway enrichment analysis further showed that upregulated DEGs were predominantly involved in flavonoid biosynthesis and stilbenoid, diarylheptanoid and gingerol biosynthesis, suggesting that enhanced secondary metabolism contributes to detoxification and antioxidant defense in resistant plants. In contrast, downregulated DEGs were enriched in zeatin biosynthesis, MAPK signaling, FoxO signaling, and plant–pathogen interaction pathways ( Fig. 3D; Table S13), indicating that growth- and defense-related processes are suppressed under herbicide stress. Collectively, these results suggest that feproxydim resistance in sorghum involves a complex regulatory network integrating hormone signaling, membrane-associated transport, defense responses, and secondary-metabolite biosynthesis. These transcriptional adjustments likely enable the resistant line IS1219 to mitigate herbicide-induced oxidative damage and maintain cellular homeostasis.

Integrated analysis of BSA-Seq and RNA-Seq

To identify key candidate genes associated with herbicide resistance, an integrated analysis combining BSA-Seq and RNA-Seq data was performed. Genes located within the candidate interval identified by BSA-Seq (Chr01: 74.47–75.37 Mb) were intersected with differentially expressed genes (DEGs) between the resistant line IS1219 and the susceptible line RTx430 collected 48 h after treatment with 10% feproxydim at an application rate of 1.5 L/ha.

This integrative analysis identified five genes responsive to feproxydim stress within the candidate region (Table 3). dividing the genes into two groups: one upregulated gene and four downregulated genes. The upregulated gene, Sobic.001G431500, encodes an α/β-hydrolase fold enzyme/carboxylesterase 17, which may be involved in herbicide metabolism or detoxification. The four downregulated genes included Sobic.001G425500 (encoding an HSP20-like chaperone involved in protein processing in the endoplasmic reticulum), Sobic.001G430500 (a dynamin-related GTPase effector), and two genes (Sobic.001G432500 and Sobic.001G432600) containing PGG domains, which may participate in cellular stress response and membrane remodeling.

Table 3.

Upregulated and downregulated genes identified by integrated BSA-Seq and RNA-Seq analysis

Gene Number	Chromosomal Location (Chr01)		Function Annotation	log2FoldChange		FDR	Expression direction
Sobic.001G430500		Chr01:75097727.75100237	Dynamin GTPase effector		-10.82675352	2.55E-36	down
Sobic.001G431500		Chr01:75194400.75196997	Alpha/Beta hydrolase fold		1.652600515	0.001978	up
Sobic.001G425500		Chr01:74680740.74681565	HSP20-like chaperone		-1.341917491	0.039719	down
Sobic.001G432500		Chr01:75266077.75267585	PGG domain		-1.822190403	0.037776	down
Sobic.001G432600		Chr01:75270555.75271610	PGG domain		-2.167617881	6.04E-05	down

The qRT-PCR analysis of Sobic.001G431500 expression over a 48-hour time course (0, 6, 12, 24, and 48 h) following feproxydim treatment revealed distinct expression patterns between the resistant sorghum line IS1219 and the susceptible line RTx430 ( Fig. 4). At 0 h, no significant difference in basal expression was observed between the two genotypes (ns). By 6 h post-treatment, IS1219 exhibited significantly higher expression than RTx430 (P < 0.01), a trend that continued at 12 h (P < 0.01). The expression difference peaked at 24 h, with IS1219 showing a highly significant upregulation (P < 0.001) relative to RTx430. Although the absolute expression level in IS1219 declined by 48 h, the difference remained highly significant (P < 0.001), confirming a sustained and robust induction of the target gene in the resistant genotype throughout the treatment period. These results validate the RNA-Seq data, demonstrating that the Sobic.001G431500 is strongly and persistently activated in IS1219 in response to herbicide stress.

Fig. 4.

qRT-PCR validation of target gene expression in resistant (IS1219) and susceptible (RTx430) sorghum lines following feproxydim treatment

Together, these results highlight Sobic.001G431500 as the most likely putative candidate gene conferring feproxydim resistance in IS1219, while the co-regulated downregulated genes may function in associated stress or transport pathways contributing to the overall resistance phenotype.

Sobic.001G431500 comparison between IS1219 and RTx430

To further explore how the gene confers resistance in IS1219, we compared its protein sequence with RTx430 and found that compared with the RTx430 allele, the IS1219 allele contained a missense mutation (V300A) and a deletion of three amino acids (P301_P303del) ( Fig. 5). It is possible that these changes explain the resistance in IS1219.

Fig. 5.

Protein sequence alignment between IS1219 and RTx430 (SbiRTX430.01G453700) alleles of Sobic.001G431500. “.” Indicates missense and “---” deletion mutations. Sequence color coding: Red-Small amino acids (AVFPMILW); Blue-Acidic (DE); Magenta-Basic (RHK); Green-Hydroxly/sulfhydryl/amine (STYHCNGQ)

Discussion

Herbicide-resistant accessions derived from natural populations represent valuable resources for herbicide-resistance breeding and genetic research. Such populations harbor abundant natural genetic variation and preserve resistance-associated loci that have adapted to local ecological conditions, while maintaining favorable agronomic traits through long-term natural selection. For example, Bao et al. [26] screened 854 maize inbred lines using 1 g/L glufosinate at the three-leaf stage and identified the line L336R, a naturally derived variant of L336 that displayed over twice the tolerance of the parental line. Similarly, in the present study, herbicide-resistant sorghum accessions were screened from a natural population. By applying 10% feproxydim at more than twice the recommended field dosage at the four- to five -leaf stage—consistent with the seedling treatment stage used by Tang et al. [14], the survival rate of IS1219 reached 69%, exceeding the maximum population survival rate of 51.9% reported by Rizwan et al. [27] using an Atlantis-type herbicide on lentil populations. Compared with earlier growth stages, sorghum seedlings at the four- to five -leaf stage exhibit stronger physiological metabolism and stress tolerance [28, 29], thereby improving the precision and reliability of herbicide resistance evaluation.

TSR, one of the most common herbicide resistance mechanisms, arises from point mutations or overexpression of herbicide target enzymes. Feproxydim is a typical ACCase-inhibiting herbicide, and target-site resistance (TSR) to ACCase inhibitors is frequently conferred by amino acid mutations in the ACCase coding region in grass species [30]. Notably, the present study did not perform full-length sequencing or mutation analysis of the ACCase gene, which represents a key experimental limitation. Nine single-amino-acid substitutions in the ALS gene (Ala122, Pro197, Ala205, Phe206, Asp376, Arg377, Trp574, Ser653, and Gly654) have been confirmed to confer resistance to ALS-inhibiting herbicides [31–33]. In contrast, NTSR involves detoxification and sequestration processes mediated by plant metabolic systems. Two major pathways underlie NTSR: (i) cytochrome P450 monooxygenases (P450s) and glutathione S-transferases (GSTs) detoxify herbicide molecules through oxidative or conjugative metabolism, and (ii) ATP-binding cassette (ABC) transporters reduce cellular herbicide concentrations by compartmentalization and transport. These systems function synergistically to mitigate herbicide toxicity [12]. In this study, five genes were identified within the candidate interval on chromosome 1 through RNA-seq, showing herbicide-responsive expression patterns. Integration of transcriptomic profiling, functional annotation, segregation and physiological analysis suggests that Sobic.001G431500, encoding carboxylesterase, encoding carboxylesterase, is a strong candidate gene for feproxydim resistance. This gene was markedly upregulated in the resistant line (IS1219) but not in the susceptible line (RTx430), indicating that enhanced hydrolytic or metabolic activity may contribute to resistance, although direct biochemical validation remains lacking.

The α/β-hydrolase family is widely known for mediating Phase I detoxification reactions, in which active herbicide molecules are transformed into less toxic or more conjugation-ready intermediates through hydrolysis, deesterification, or oxidation. Carboxylesterases (CXEs/CarEs), a key subgroup of α/β-hydrolases [34], hydrolyze carboxylate esters, thioesters, and amide bonds, and their role in herbicide metabolism is dual—governed by herbicide chemical structure and species-specific enzyme traits—rather than being limited to detoxification [35, 36]. Many herbicides—such as 2,4-D methyl ester and aryloxyphenoxypropionate (AOPP) herbicides (e.g., diclofop-methyl, clodinafop-propargyl, fenoxaprop-ethyl, fenthioprop-ethyl)—are applied as inactive ester forms that readily penetrate plant cuticles. Once inside, carboxylesterases catalyze their hydrolysis into active carboxylic acid forms, conferring both herbicidal selectivity and plant tolerance [37–39]. For instance, in the AOPP-susceptible weed black-grass, a GDSL family carboxylase was confirmed to drive AOPP bioactivation, highlighting the enzyme’s functional divergence across species [40]. In resistant sorghum, in contrast to this activation function in susceptible weeds, the elevated expression of Sobic.001G431500—markedly upregulated in the feproxydim-resistant line IS1219 but not in the susceptible line RTx430 and likely functioning as a carboxylesterase or related hydrolase—is hypothesized to mediate the Phase I detoxification of feproxydim rather than its bioactivation, thereby lowering its intracellular accumulation and preventing the inhibition of key biosynthetic pathways (e.g., fatty acid synthesis, consistent with feproxydim’s target of acetyl-CoA carboxylase [ACCase]). It is possible that the missense and deletion mutations in the IS1219 allele (Fig. 4) further enhance this putative detoxification process, though direct biochemical evidence (e.g., in vitro enzyme assays, substrate specificity testing) is required to validate this hypothesis [40]. The detoxification intermediates produced can subsequently undergo Phase II conjugation—catalyzed by UDP-glycosyltransferases (UGTs) and GSTs—and Phase III sequestration, where conjugates are transported into vacuoles by ABC transporters [41–43].

Together, these findings support the presence of a three-phase metabolic detoxification network in resistant sorghum, involving coordinated actions of hydrolytic enzymes, conjugative transferases, and transport proteins. This network effectively reduces the cellular concentration of active herbicides and represents a typical NTSR mechanism. The identification of Sobic.001G431500 as a candidate herbicide-resistant key α/β-hydrolase/carboxylesterase gene provides novel insights into metabolic resistance mechanisms in sorghum and offers a promising molecular target for breeding herbicide-tolerant forage and grain sorghum varieties.

Several limitations should be acknowledged in the present study. First, although qRT-PCR was performed to validate the RNA-Seq expression results of Sobic.001G431500, no direct functional validation—including in vitro carboxylesterase activity assays, transient overexpression, virus-induced gene silencing (VIGS), or CRISPR/Cas9-mediated knockout—was conducted to confirm the causal role of this gene in feproxydim resistance. All conclusions about its function remain supported only by genetic mapping, gene expression and cosegregation data. Second, all phenotypic evaluations were performed in a greenhouse from March to June, and potential environmental effects on resistance expression were not assessed. Since variable environmental conditions may affect the stability and magnitude of herbicide resistance under field conditions, multi-location and multi-environment field trials are needed to verify the robustness of resistance and its agronomic performance in practical production. Third, as feproxydim is an ACCase-inhibiting herbicide, we did not sequence the ACCase target gene or analyze its potential mutations in this study. It therefore remains unclear whether ACCase contributes to the observed resistance, and the complete molecular mechanism cannot be fully elucidated. In addition, some interpretations concerning the detoxification role of carboxylesterases are somewhat speculative without direct biochemical validation and should be treated as hypotheses requiring further experimental confirmation. We will conduct additional studies on ACCase gene sequencing, functional analysis, and biochemical validation of carboxylesterases to clarify the resistance mechanism in future projects.

Conclusions

This study used large-scale screening of sorghum germplasm for herbicide tolerance to identify multiple accessions resistant to ACCase-inhibiting feproxydim and ALS-inhibiting mesosulfuron-methyl. Among these, IS1219 exhibited prominent resistance to feproxydim, serving as a valuable genetic resource for dissecting resistance mechanisms. Through integrated bulked segregant analysis sequencing (BSA-Seq) and transcriptome (RNA-Seq) profiling, a major QTL associated with feproxydim resistance was mapped to a 0.9 Mb interval on chromosome 1, with five co-expressed candidate genes identified within this region.

A key finding is the characterization of Sobic.001G431500 (encoding carboxylesterase 17, an α/β-hydrolase) as a core candidate gene for feproxydim resistance in the QTL. This gene was markedly upregulated in the resistant line IS1219 compared to the susceptible RTx430, with quantitative real-time PCR (qRT-PCR) validating this expression pattern. Additionally, KASP genotyping demonstrated cosegregation of resistant allele with the resistant phenotype, while the presence of protein sequence variations in the IS1219 allele further supports the gene’s potential functional role in mediating the resistance. Collectively, these results suggest that enhanced hydrolytic or metabolic activity mediated by Sobic.001G431500 likely underpins the resistance phenotype, providing novel insights into the non-target-site resistance mechanisms of sorghum to ACCase-inhibiting herbicides. Importantly, the KASP marker developed for Sobic.001G431500 enables efficient, high-throughput discrimination of resistance alleles in early-generation breeding populations, overcoming the limitations of phenotypic screening (e.g., environmental variability, labor intensity). This marker can be directly integrated into marker-assisted selection (MAS) programs to accelerate the introgression of feproxydim resistance into elite sorghum and possibly other crop germplasms.

Authors’ contributions

Xing Zhichao: Investigation, Writing – original draft, Writing – review & editing. Cheng Zhengxiao: Writing – original draft, Investigation. Yang Xiaochun: Writing– original draft. Hu Lu: Investigation. Wang Kai: Writing– original draft. Wang Yongfei: Writing – original draft. Hu Die: Investigation. Wang Yi-Hong: Writing– review & editing. Du Junli: Investigation. Wang Lihua: Conceptualization, Funding acquisition, Investigation. Li Jieqin: Investigation, Supervision, Writing – review & editing.

Funding

The study was supported by the National Natural Science Foundation of China (32372134), Chuzhou “Star of Innovation and Entrepreneurship” Industrial Innovation Team.

Data availability

The datasets generated and/or analyzed during the current study are available in the China National GeneBank DataBase (CNGBdb) repository, [CNP0009093]

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.