CRISPR/Cas9-mediated mutation of GhCAD decreases the gossypol content of cottonseed

Abstract

Cottonseed is the most important byproduct of cotton production. However, high free gossypol contents limit the application of cottonseed in the food or feed industry. In this study, CRISPR/Cas9 technology was used to knock out the (+)-δ-cadinene synthase gene (GhCAD) to decrease the gossypol content. Gossypol levels decreased approximately 64% in cottonseeds and leaves following the targeted mutation of GhCAD. If only GhCAD1-A was edited, the seed gossypol content decreased by approximately 46%, but there were no major changes in the leaf gossypol content. In addition, the protein and fatty acid (C16:0, C18:1, and C18:2) profiles of the transgenic cotton seeds were similar to those of the control cotton seeds. Furthermore, transcriptome analysis revealed that the jasmonic acid signal transduction pathway was significantly enriched among the DEGs, and GhMYC2-D09 expression was down-regulated. Silencing of GhMYC2-D09 via virus-induced gene silencing decreased the expression of gossypol biosynthesis-related genes, ultimately restricting the accumulation of gossypol in cotton leaves. In contrast, the overexpression of GhMYC2-D09 in hairy roots had the opposite effect. Dual-luciferase assays revealed that GhMYC2-D09 can activate the expression of GhCAD1-A and GhCAD1-C, but Y1H assays revealed that GhMYC2-D09 cannot bind directly to GhCAD promoters. In conclusion, we used CRISPR/Cas9 technology to silence GhCAD expression and developed new genetic resources for generating low-gossypol cotton materials. Furthermore, we characterized GhMYC2-D09 as a transcription factor that increases gossypol biosynthesis. These findings may provide new insights to further elucidate the regulatory network of gossypol biosynthesis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13036-025-00556-2.

Introduction

Cotton has diverse applications worldwide. Fibers are widely employed in the textile industry, whereas seeds are utilized in food products and animal feed. Cottonseed kernels are rich in protein and oil (38% and 35% by mass, respectively), making them valuable nutritional additives in food and feed [1]. Cottonseed oil, which is 70% unsaturated fatty acid (linoleic acid represents a significant portion), can lower cholesterol and blood lipid levels [2, 3]. In certain countries (e.g., Pakistan), cottonseed oil is one of the main edible oils [4, 5]. With the intensive development of animal husbandry, satisfying the increasing demand for soybean meal in China is a considerable challenge. Cottonseed meal, which is produced after husking and extracting oil, has a higher crude protein content (43.5–47.0%) than soybean meal and can serve as a viable alternative in livestock and poultry diets [6].

Although cottonseed is a rich source of nutrients, its high gossypol content can be harmful to monogastric animals, such as humans, thereby limiting its utility [7]. Some studies have shown that gossypol has potentially fatal toxic effects on various animals, including chickens [8, 9], pigs [10, 11], and dogs [12, 13]. In contrast, ruminants exhibit some tolerance to gossypol, making cottonseed useful for the production of feed for cows, sheep, goats, and other similar animals [14–17]. Notably, gossypol and its associated terpenoids are natural phytochemicals that contribute to cotton defense responses to pests and pathogens [18–22]. In earlier studies on cotton bollworm larvae, the addition of gossypol to the diet resulted in decreases in body weight and eclosion rates as well as delayed growth [23, 24]. Additionally, gossypol inhibits the germination of spores produced by certain pathogens, including Verticillium dahliae and Fusarium oxysporum f. sp. vasinfectum [25]. Therefore, developing improved cotton varieties that produce seeds and related materials with low gossypol contents is critical for mitigating the toxicity of gossypols to animals while preserving the protective effects of gossypols in cotton plants.

Researchers have devoted considerable effort to breeding low-gossypol cotton. Gossypol can be divided into (+)-gossypol and (−)-gossypol, of which (−)-gossypol is toxic to humans and nonruminant animals. Knocking out GhDIR5, which is a (−)-gossypol synthesis gene, via CRISPR/Cas9 technology prevents gossypol formation in cottonseeds, but the contents of (+)-gossypol and other similar terpenoids are unaffected, thereby maintaining insect resistance to cotton [26]. Gossypol is stored in pigment glands in various cotton plant parts, including roots, stems, leaves, seeds, and bracts. By using CRISPR/LbCpf1 to knock out PIGMENT GLAND FORMATION (PGF), researchers created cotton plants without glands [27]. However, these glandless cotton varieties, which have diminished gossypol-based defense mechanisms, are highly susceptible to pests [28]. Subsequently, GoPGF was knocked down via RNA interference (RNAi) involving a seed-specific promoter, resulting in the development of a cotton germplasm with seed-specific low-gossypol contents (98% decrease) and transgenic cotton plants that remained resistant to insect pests [29]. The gossypol biosynthesis pathway has been elucidated [30, 31], with (+)-δ-cadinene synthase (CAD) identified as the rate-limiting enzyme for gossypol biosynthesis. Researchers used an RNAi expression cassette containing a seed-specific promoter to downregulate CAD expression, which decreased the seed gossypol content [32]. The resulting stable transgenic RNAi line produced seeds with ultralow gossypol contents while maintaining normal terpenoid levels in non-seed tissues as well as normal fiber quality and yield [33–35].

CRISPR/Cas9 technology is extremely useful for precisely modifying target genes. This technology has been successfully used to characterize certain genes functionally and improve the traits of many crops, including maize [36, 37], wheat [38, 39], rice [40], tomato [41, 42], soybean [43, 44] and cotton [45]. In cotton, there are two CAD subfamilies (CAD1-A and CAD1-C) [46, 47], with CAD1-A genes expressed mainly in seeds and roots, whereas CAD1-C genes are more commonly expressed in various tissues and organs [48]. In Gossypium hirsutum, two copies belong to CAD1-A and six copies belong to CAD1-C [49]. Therefore, knocking out CAD genes via CRISPR/Cas9 technology may decrease the gossypol content in seeds.

In this study, we aimed to develop new cotton germplasm resources by editing GhCAD1-A and GhCAD1-C via CRISPR/Cas9 technology to decrease the gossypol content and revealed a transcription factor (GhMYC2-D09) that is responsive to methyl jasmonate (MeJA) and promotes gossypol biosynthesis. This study may be useful for further clarifying how gossypol biosynthesis is regulated and identifying new genetic targets for modulating the gossypol content of cotton-derived resources.

Materials and methods

GhCAD expression profile analysis of different tissues

Transcriptomic data were downloaded from the Cotton Omics Database (http://cotton.zju.edu.cn/10.rnasearch) [50]. A heatmap was created based on fragments per kilobase of exon per million mapped (FPKM) values via TBtools software (v1.108).

Construction of CRISPR/Cas9 gene editing vectors

The design principle of target sites followed 5’-G-(20 bp)-NGG-3’ [51]. The alignment of genomic sequences of GhCAD genes were performed, and the specific target site for GhCAD1-A or GhCAD-C were designed in low similarity region, and common target sites for GhCAD genes were designed in high similarity region. CRISPR/Cas9 vectors were provided by Prof. Jiankang Zhu. Guide RNA oligonucleotides were synthesized and annealed to double-stranded DNA fragments, which were then cloned and inserted into the BbsI site of pAtU6-26-SK. The recombinant vector (pAtU6-sgRNA-SK) was digested with KpnI and SalI (New England Biolabs, MA, USA), and the resulting fragment (AtU6-sgRNA) was inserted into p35S-Cas9-SK to generate pAtU6-sgRNA-35 S-Cas9-SK. This vector was then digested with KpnI and EcoRI, and the resulting fragment was inserted into pCAMBIA2301. The binary vector was introduced into A. tumefaciens strain LBA4404 cells (Vazyme, Nanjing, China) via a heat shock procedure.

Plant materials

The control cotton line (R15) used for genetic transformation and gene editing was derived from Gossypium hirsutum cv. Coker 312. The Agrobacterium-mediated transformation method was reported by Wang et al. [52]. Gene-edited cotton and R15 were grown in a greenhouse at 30 °C (day)/25°C (night) with a 16-h light/8-h dark photoperiod. Mature leaves and seeds were collected and examined to determine gossypol contents.

Gossypol, oil, fatty acid and protein measurements

The gossypol contents were determined as previously described [53]. Briefly, cotton leaf samples (50 mg) were ground into powder and transferred to 2 mL tubes, after which 1 mL of extraction buffer consisting of a 1:1 (v: v) mixture of acetonitrile and 0.1% (v/v) formic acid aqueous solution was added. Ethanol (1 mL) was used to prepare cottonseed powder (20 mg) extracts. Gossypol was analysed via a liquid chromatography‒mass spectrometry (LC‒MS) system (1260 Infinity II; Agilent Technologies, Santa Clara, CA, USA) with an Agilent C18 chromatography column (2.1 mm × 75 mm, 2.7 μm). Mobile phase: A-acetonitrile and B-0.1% formic acid. Gradient elution: 0–5 min:50% A and 50% B. Flow rate: 0.5 mL/min. Injection volume: 10 µL, Column temperature: 30℃. Electron impact ionization: ElectroSpray Ionization (ESI). Scanning mode: ESI+. Detection method: Multi-React Monitoring (MRM). Ion spray voltage: 4000 V. Ionization source temperature: 300℃. Sheath gas: High purity nitrogen, 11 L/min, 350℃. Collision energy: m/z 517.4 > 231.3, 69 V, m/z 517.4 > 259.3, 45 V.

Mature cottonseed oil contents were measured as previously described by Hovav et al. [54]. Cottonseeds were weighed and ground into powder, which was transferred to preweighed 2 mL tubes. After adding 1.5 mL hexane, the mixture was incubated at room temperature with shaking for 1 h and then centrifuged (12,000 rpm for 10 min). The supernatant was transferred to a new preweighed tube. The process was repeated for the residue. Hexane was removed via evaporation, and the tube was subsequently weighed.

To measure fatty acid contents, mature cottonseeds were ground into powder in liquid nitrogen, and then 20 mg of powdered material was transferred to a 2 mL glass tube. After 0.5 mL of 5% sulfuric acid–methanol solution and 400 µL of C19-nonadecanoic acid (4 mg/mL) internal standard were added, the resulting mixture was mixed and incubated at 70 °C with shaking for 30 min and then cooled to room temperature. Next, 200 µL H₂O and 800 µL hexane were added to the tube. The solution was mixed and then centrifuged (1,500 rpm for 10 min). The upper phase (200 µL) was transferred to a sample vial and diluted with 180 µL hexane. Methyl-esterified fatty acids were analysed via a gas chromatography‒mass spectrometry system (GCMS-7890 A-7000; Agilent, USA).

To measure protein contents, mature cottonseeds (0.1 g) were ground into powder in liquid nitrogen. The total protein content of the cottonseed was extracted via the Plant Total Protein Extraction Kit (Coolaber, Beijing, China) and then quantified via the BCA Protein Assay Kit (Coolaber, Beijing, China) according to the manufacturer’s instructions.

Agronomic traits analysis

To determine the seed germination rate, 20 cottonseeds were placed on filter paper moistened with water, covered with another layer of filter paper, and then incubated in a chamber at 28 °C for 3 days. Seed germination rates (number of germinated seeds/20) were calculated. To measure plant height, the distance from the cotyledon node to the field seedling apex was measured via a tape line. Fiber quality traits in the field condition, including micronaire, average length of the upper half, uniform index, and breaking tenacity were measured by the Test Center of Cotton Quality, Ministry of Agriculture, China.

Phylogenetic relationship and conserved domain analyses of GhMYC2

Multiple full-length MYC2 amino acid sequences from cotton and other plants were aligned, and a neighbor-joining phylogenetic tree was constructed using MEGA7.0. TBtools software was used to analyse the conserved domains of MYC2 from different species. The accession numbers of MYC2 from other plants are as follows: AtMYC2 (AT1G32640), TcMYC2a (ATY38591.1), SmMYC2 (AIO09733.1), NtMYC2 (ADH04270.1), AaMYC2 (AKO62850.1), AsMYC2 (AKO62850.1), OsMYC2 (XP_015614012.2), and CrMYC2 (AAQ14332.1).

Virus-induced gene silencing (VIGS) procedure

On the basis of an alignment of homologous gene sequences, specific fragments of GhMYC2-A08/GhMYC2-D08, GhMYC2-A09/GhMYC2-D09, and GhMYC2-A12/GhMYC2-D12 were cloned and separately inserted into VIGS vectors to generate pTRV2:GhMYC2-AD08, pTRV2:GhMYC2-AD09, and pTRV2:GhMYC2-AD09. A. tumefaciens strain GV3101 cells were transformed with pTRV2:GhMYC2, pTRV2:CLA1, pTRV2, and pTRV1. Positive clones were selected and cultured for an OD₆₀₀ of 1.2. The bacterial cells were collected and suspended in a solution containing 200 µM acetosyringone, 10 mM MES, and 10 mM MgCl₂. The pTRV2:GhMYC2, pTRV2:CLA1, and pTRV2 suspensions were mixed with the pTRV1 suspension (v/v = 1:1), and the resulting mixture was used for the infiltration of the cotyledons of 8-day-old cotton seedlings. pTRV2:CLA1 and the empty pTRV2 vector were used as positive and negative controls, respectively. After the positive control plants presented an albino phenotype, the leaves were harvested for analysis of gene silencing efficiency and gossypol contents.

Subcellular localization and transactivation activity analyses of GhMYC2-D09

The GhMYC2-D09 coding sequence (CDS) was fused with a GFP-encoding sequence in the pCAMBIA1302 vector to generate p35S-MYC2-D09::eGFP. Additionally, p35S-H2B::mCherry [55] was constructed and co-expressed with p35S-MYC2-D09::eGFP to label the nucleus. All the recombinant plasmids were introduced into Arabidopsis protoplasts, which were then examined for fluorescence using a laser confocal microscope (Zeiss, Germany).

The GhMYC2-D09 CDS was inserted into pGBKT7. Y1H Gold yeast cells were transformed with the recombinant vector and plated on solid SD/−Trp (SDO) medium to screen for transformants. Cultures of positive clones were diluted prior to the inoculation of solid SD/−Trp/−His/−Ade (TDO) medium for transcriptional activation assays. An empty BD vector was used as a negative control. After a 3-day incubation at 30 °C, the colonies were photographed.

Cotton hairy root acquisition

The p35S-MYC2-D09::eGFP plasmid or CRISPR/Cas9 plasmid was transformed into Agrobacterium rhizogenes K599. The hairy root induction procedure was conducted according to the previous report [56]. Genomic DNA was extracted from each hairy root using Plant Genomic DNA Kit (TIANGEN, Beijing, China). Segments with target sites were amplified with specific primers (Table S1), and the amplified products were analyzed by Hi-TOM sequencing [62]. The target site mutation ratio (target site mutation reads/total reads) > 1% was considered as edited hairy root. Editing efficiency of each target site was evaluated in 20 hairy roots. Editing efficiency (%) = Number of hairy roots with editing events/ 20.

Dual-luciferase (LUC) assay

The GhMYC2-D09 CDS and the promoter of GhCAD were cloned and inserted into pGreenII 62-SK and pGreenII 0800-LUC, respectively. A. tumefaciens cells were transformed with promoter-LUC or GhMYC2-D09-62 SK/promoter-LUC for the subsequent infiltration of Nicotiana benthamiana leaves. After 48 h, the luciferase substrate was smeared on the leaves, which were then placed in darkness for 2 min. The leaves were examined and photographed via a multifunction imager (Berthold, Germany).

RNA-seq and qRT‒PCR analyses

Three replicates of young leaves were collected from sgAC1-2, sgC2-1, and R15 plants and immediately frozen in liquid nitrogen. Total RNA was extracted via the RNAprep Pure Plant Kit (Tiangen, Beijing, China). The RNA quality was evaluated via the RNA Nano 6000 Assay Kit and the 2100 Bioanalyzer system (Agilent Technologies). High-quality RNA samples were sequenced via the Illumina NovaSeq platform at Novogene (Beijing, China). Clean reads were mapped to the TM-1 reference genome [57] using HISAT2 [58]. Differentially expressed genes (DEGs) were analysed using DESeq2 [59], with log₂ (fold-change) ≥ 1 and P < 0.05 used as the criteria for identifying significant DEGs. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed via clusterProfiler [60].

Taq Pro Universal SYBR qPCR Mix (Vazyme, Nanjing, China) and an ABI 7500 Fast Real-Time PCR System (Thermo Fisher Scientific, MA, USA) were used for qRT‒PCR analysis. GhUBQ7 (GenBank ID: DQ116441) was selected as the internal control for normalizing gene expression levels. Details regarding the qRT‒PCR primers used are listed in Table S1. Three biological replicates were analysed for each sample. Relative gene expression was calculated using the 2^−ΔΔCT method [61].

Results

Design and selection of SgRNAs for CRISPR/Cas9

The G. hirsutum genome (v. NAU) harbors six GhCAD1-C family genes and two GhCAD1-A family genes [49]. Transcriptional analyses and relative expression analyses of the calycle, leaf, petal, stem, torus, ovule, root, and seed at different stages revealed that two GhCAD1-A genes were highly expressed in the developing ovule and roots, whereas the GhCAD1-C genes were expressed in various tissues (Fig. 1A, Figure S1). Four sgRNAs (sgA1, sgA2, sgA3, and sgA4) and three sgRNAs (sgC1, sgC2, and sgC3) were designed to target the GhCAD1-A and GhCAD1-C subfamily genes, respectively, two sgRNAs (sgAC1 and sgAC2) were designed to target both the GhCAD1-A and GhCAD1-C subfamily genes (Fig. 1B, Figure S2). The editing efficiencies of the sgRNAs in cotton hairy roots were determined via Hi-TOM sequencing [62]. The results indicated that the editing efficiency of individual sgRNAs varied. More specifically, sgA3, sgA4, sgAC1, sgAC2, sgC1, and sgC2 yielded targeted mutations. Notably, sgA4 had the highest editing efficiency (80%), followed by sgAC2 (50%). In contrast, targeted mutations were undetectable for sgA1, sgA2, and sgC3 (Fig. 1C). Gene-editing vectors for sgA3, sgA4, sgAC1, sgAC2, sgC1, and sgC2 targets were introduced into separate R15 samples via Agrobacterium-mediated transformation.

Fig. 1.

Construction of vectors for transformation and mutation analysis of T₀ plants. (A) GhCAD1-A and GhCAD1-C expression profiles in the TM-1 calycle, leaf, petal, stem, torus, ovule, root, and seed determined via transcriptome sequencing. (B) GhCAD1-A and GhCAD1-C gene structure and target sites as well as the T-DNA region of the CRISPR/Cas9 vector used for targeted mutations in cotton. (C) Editing efficiency of nine sgRNAs in hairy roots using the CRISPR/Cas9 system. Twenty hairy roots were used for each target site. (D) Mutation analysis of T₀ gene-edited plants

Mutation analysis in T₀ plants

No plants were recovered from the sgA3 callus, but three T₀ transgenic plants (sgA4-1, sgA4-2, and sgA4-3) were recovered from the sgA4 callus in which GhCAD1-A subfamily genes were edited. Four plants (sgC1-1, sgC1-2, sgC1-3, and sgC1-4) were recovered from the sgC1 callus, and three plants (sgC2-1, sgC2-2, and sgC2-3) were recovered from the sgC2 callus (GhCAD1-C subfamily). Four plants (sgAC1-1, sgAC1-2, sgAC2-1, and sgAC2-2) were recovered from sgAC1 and sgAC2 calli (both the GhCAD1-A and GhCAD1-C subfamilies). Editing events were detected via Hi-TOM sequencing. The results revealed that different mutations were independently generated at the respective target sites for sgA4, sgC1, sgC2, sgAC1, and sgAC2 (Fig. 1D, Table S2). In plant sgA4-1, GhCAD1-A was edited in both genomes, but in plants sgA4-2 and sgA4-3, only GhCAD1-A in the A genome was edited. In plant sgC1-2, GhCAD1-C in each genome was edited. In contrast, in plants sgC1-1, sgC1-3, sgC2-2, and sgC2-3, only GhCAD1-C in the A genome was edited, whereas in plants sgC1-4 and sgC2-1, only GhCAD1-C in the D genome was edited. In plant sgAC2-1, GhCAD1-A in both genomes and GhCAD1-C in the D genome were edited. In plants sgAC1-1 and sgAC1-2, only GhCAD1-A in the A genome was edited, whereas in plant sgAC2-2, only GhCAD1-C in the D genome was edited. Deletions were the most common mutations induced by CRISPR/Cas9 in the targeted genes. On the basis of the results of the mutational analysis, we selected plants sgA4-1, sgC1-2, and sgAC2-1 for further analysis.

Analysis of the mutation and gossypol contents of T₁ and T₂ plants

The selected T₀ plants (sgA4-1, sgAC2-1, and sgC1-2) were self-pollinated to generate T₁ seeds. Ten T₁ seedlings derived from each T₀ plant were grown and selected because of their healthy and consistent growth. Editing events were detected in four of the 10 sgA4-1 T₁ plants, five of the 10 sgC1-2 T₁ plants, and four of the 10 sgAC2-1 T₁ plants. The results of the mutation analysis of these plants are provided in Table S3. In sgA4-1 plants 1, 3, 5, and 6, GhCAD1-A was edited in both genomes. In sgC1-2 plants 1, 3, and 4, two of three GhCAD1-C genes in the A and D genomes were edited, whereas in plants 8 and 10, one GhCAD1-C gene in the A genome and one GhCAD1-C gene in the D genome were edited. In the four sgAC2-1 T₁ plants, GhCAD1-A in both genomes and all the GhCAD1-C genes in the D genome were edited. In contrast, in plants 2 and 3, all GhCAD1-C genes in the A genome were not mutated, whereas in plants 7 and 10, one GhCAD1-C gene in the A genome was edited. The gossypol contents in the leaves and seeds of T₁ plants were determined via an LC‒MS system. The seed gossypol contents were significantly lower in the gene-edited T₁ plants than in the R15 control plants (Fig. 2A). Significant decreases in the leaf gossypol contents were also detected for the sgAC2-1 and sgC1-2 plants but not for the sgA4-1 plants (Fig. 2B).

Fig. 2.

Gossypol contents and agronomic traits of gene-edited cotton plants and wild-type (WT) R15. (A) Comparison of seed gossypol contents of WT and T₁ mutants (n = 4, P < 0.05, Tukey’s HSD test). (B) Comparison of leaf gossypol contents of WT and T₁ mutants (n = 4, P < 0.05, Tukey’s HSD test). (C–F) Gossypol levels in cotton lines determined by LC‒MS. (C, D) Seeds of WT and T₂ mutants (n = 4, P < 0.05, Tukey’s HSD test). Black arrows indicate gossypol peaks for WT and mutant plants. (E, F) WT and T₂ mutant leaves. (G) Protein and fatty acid contents in WT and gene-edited plants. (H) Germination of seeds from WT and gene-edited plants. (I) Plant height and fiber quality (upper half mean length, breaking tenacity, uniformity index, and micronaire) of WT and gene-edited plants. (J) PCR analysis of Cas9 in T₂ cotton plants. Cas9 was detected in 22 sgAC2-1 progeny plants. Cas9-free plants are indicated in red. Negative control: R15; positive control: plasmid containing Cas9

T₂ plants were grown from seeds collected from four sgA4-1 T₁ plants, sgAC2-1 T₁ plants 7 and 10, and sgC1-2 T₁ plants 1, 3, and 4. These plants were screened for new editing events. New mutations were detected in the progenies of sgA4-1 plants 1, 3, and 5; sgC1-2 plants 1 and 4; and sgAC2-1 plant 7 (Table S4). A 1-bp insertion was detected in the GhCAD1-A target site in the D genome of the progenies of sgA4-1 plant 1, whereas 4-bp and 5-bp deletions were detected in the progenies of plants 3 and 5, respectively. A 4-bp deletion was detected in the GhCAD1-C target site in the A genome of progenies of sgC1-2 plant 1, whereas a 2-bp deletion was detected in the GhCAD1-C target site in the D genome of progenies of plant 4. A large deletion and a 3-bp deletion were detected in GhCAD1-A in the A and D genomes, respectively, in the progenies of sgAC2-1 plant 7. These new mutations were not detected in the T₁ plants, which indicated that Cas9-mediated editing continued in the T₂ generation, resulting in new editing events. The leaf and seed gossypol contents were considerably lower for these T₂ plants than for the R15 plants. Gossypol levels decreased in the seeds of sgA4-1 but not in the leaves (Fig. 2C–F). Overall, the gossypol levels in T₂ plants were similar to those in T₁ plants, implying that the CRISPR/Cas9-induced mutations in GhCAD and their biochemical effects are heritable. These results were consistent with the finding that GhCAD1-A is expressed mainly in seeds and roots and that GhCAD1-C is more widely expressed in various tissues and organs.

Next, we determined the oil, protein, and fatty acid contents of mature T₂ cottonseeds from sgA4-1, sgC1-2, and sgAC2-1 plants. The seed protein contents were similar among the gene-edited lines and R15. Cottonseed oil contains mainly palmitic acid, oleic acid, and linoleic acid. The contents of these fatty acids and the total oil contents in the gene-edited plants were similar to those in the R15 control plants (Fig. 2G). Accordingly, a decrease in the gossypol content did not adversely affect the nutritional quality of cottonseed. We also analysed the seed germination rates in chamber and fiber quality in the filed condition of the T₂ gene-edited plants. The results revealed that the gene-edited plants (sgA4-1, sgC1-2, and sgAC2-1) and R15 did not differ significantly in terms of their seed germination rates (Fig. 2H) or fiber quality (Fig. 2I).

If the CRISPR/Cas9 construct cannot be completely removed by segregation, it will remain constitutively active in plants, which may increase the risk of off-target mutations. PCR amplification of the Cas9 gene in T₂ plants was performed to determine whether Cas9 was segregated to produce Cas9-free plants. The Cas9-containing plasmid was used as a positive control, whereas DNA from the wild-type control (R15) was used as a negative control. For sgAC2-1 progenies, three of 22 plants lacked Cas9 (Fig. 2J). Conversely, Cas9 was detected in all of the sgA4-1 and sgC1-2 progenies. The off-targets of sgAC2-1, sgA4-1 and sgC1-2 were predicted by CRISPR-P (V2.0) [63], and the sequencing results indicated that no off-target occurred (Table S5). These Cas9-free plants may be ideal for promoting the application and adoption of new cotton germplasm resources in China.

Comparative transcriptome analysis of leaves from gene-edited and R15 plants

To identify the genes associated with GhCAD that are also involved in gossypol biosynthesis, we conducted a comparative transcriptome analysis of sgAC2-1, sgC1-2, and R15 leaves. A total of 332 million clean reads were obtained, of which more than 90% were mapped to the reference genome (G. hirsutum TM-1 CRI_v1). We identified 28,218 DEGs, 14,826 and 13,392 of which presented upregulated and downregulated expression, respectively. Among the downregulated genes, 1,294 were common to both transgenic lines (Fig. 3A). Among the 14,826 genes whose expression levels were upregulated, 3,640 were common to sgAC2-1 and sgC1-2 (Fig. 3B). According to the KEGG pathway enrichment analysis, the common downregulated genes were significantly associated with phenylpropanoid biosynthesis, terpenoid biosynthesis, the MAPK signalling pathway, and amino sugar and nucleotide sugar metabolism (Fig. 3C). The upregulated genes were related to amino acid metabolism, carbon fixation in photosynthetic organisms, the spliceosome, and peroxisomes (Fig. 3D). To verify the reliability of the RNA-seq data, qRT‒PCR analysis was conducted to determine the expression levels of 12 downregulated genes related to terpenoid biosynthesis and hormone signal transduction in R15 and the mutants (Fig. 3E). The trends in the qRT‒PCR data were similar to those in the RNA‒seq data (Figure S3A). In addition, a Pearson correlation analysis revealed a positive correlation between the qRT‒PCR and RNA‒seq data (Figure S3B). The expression levels of key genes in the ABA and JA signal transduction pathway were significantly downregulated, especially MYC2 gene (Fig. 3F). ABA signal transduction pathway is associated to many abiotic stresses, such as low temperature, drought and wound. MYC2, one of the target proteins of JAZ, is the key transcriptional factor regulating the gene responsive to JA, and MYC2 plays an important role in regulating the defense response against plant diseases and pests. Therefore, we focused on the role of MYC2 in the gossypol biosynthesis.

Fig. 3.

Comparative analysis of sgAC2-1, sgC1-2, and R15 leaves. (A, B) Venn diagrams showing shared and unique (A) down-regulated and (B) up-regulated DEGs in CAD1-C (sgC1-2) and CAD1-AC (sgAC2-1). (C, D) KEGG enrichment analysis of (C) down-regulated and (D) up-regulated DEGs. (E) qRT‒PCR analysis of expression levels of 12 genes selected according to RNA-seq data. Values are presented as the mean ± SD of three independent biological replicates. GhUBQ7 was used as a internal control. ^*p < 0.05, ^**p < 0.01, Student’s t-test. (F) ABA and JA signal transduction pathway and a heatmap of the transcript levels of selected genes (according to transcriptome data)

Silencing of GhMYC2-D09 via VIGS resulted in decreased gossypol contents

We analysed the phylogenetic relationships and conserved domains of MYC2 genes in cotton and eight other species. Six GhMYC2 genes (Gh_A08G179900, Gh_D08G177800, Gh_A09G218700, Gh_D09G210300, Gh_A12G233300, and Gh_D12G227600) were clustered with MYC2 genes from eight other species and shared the same bHLH domain. Gh_A12G132000 and Gh_D12G132300 formed a separate cluster, and most of their amino acid sequences were missing. Therefore, after excluding Gh_A12G132000 and Gh_D12G132300, the other six GhMYC2 genes were analysed further (Fig. 4A). To explore the contributions of these six GhMYC2 genes to gossypol biosynthesis, the genes were divided into three groups (Gh_A08G179900/Gh_D08G177800, Gh_A09G218700/Gh_D09G210300, and Gh_A12G233300/Gh_D12G227600) according to sequence similarity and were respectively named as GhMYC2-AD08, GhMYC2-AD09, and GhMYC2-AD12. The three groups of genes were silenced in cotton seedlings via VIGS, and the resulting changes in gossypol contents were analysed via an LC‒MS system. The silencing of GhMYC2-AD09 decreased gossypol levels (Fig. 4B). We also silenced GhMYC2-A09 and GhMYC2-D09 separately via VIGS (Figs. 4C, Figure S4), which resulted in decreased gossypol contents in the GhMYC2-D09-silenced plants (Fig. 4D). The silencing of GhMYC2-D09 was accompanied by significant decreases in the expression of gossypol biosynthesis-related genes (CAD, CYP706B1, DH1, CYP82D113, CYP71BE79, and 2-ODD-1) (Fig. 4E). Therefore, GhMYC2-D09 was selected as a candidate gene that may regulate gossypol biosynthesis.

Fig. 4.

Effect of silencing GhMYC2 on gossypol content. (A) Conserved domain analysis of MYC2 proteins from nine species. Six GhMYC2 genes were indicated by a red frame. (B) Gossypol contents after three GhMYC2 gene groups were silenced by VIGS. (C) Expression level of GhMYC2-A09 and GhMYC2-D09 after silencing. (D) Gossypol contents after GhMYC2-A09 and GhMYC2-D09 were silenced. (E) Expression level of gossypol biosynthesis-related genes in GhMYC2-D09-silenced plants. Gene accession numbers: CAD (Gh_D05G3506), CYP706B1 (Gh_D03G176100), CYP82D113 (Gh_D05G199000), CYP71BE79 (Gh_A13G153200), DH1 (Gh_A01G228800), 2-ODD-1 (Gh_D13G235700). Values are presented as the mean ± SD of three independent biological replicates. Statistical significance: *P ˂ 0.05 (Student’s t test)

GhMYC2-D09 is a JA-induced transcription factor

To clarify the subcellular localization of GhMYC2-D09, the GhMYC2-D09:eGFP fusion protein was transiently expressed in Arabidopsis protoplasts. eGFP fluorescence was observed in the nuclei (Fig. 5A), indicating that GhMYC2-D09 is a nucleus-localized protein. Furthermore, the expression patterns of GhMYC2-D09 in different tissues were detected via qRT‒PCR. GhMYC2-D09 was expressed in various tissues, including the root, stem, leaf, petal, bract, sepal, torus, and ovule (20 and 35 dpa), but it was highly expressed in the root, stem, bract, sepal, and torus (Fig. 5B). We examined the expression of GhMYC2-D09 and CAD after treatment with 100 µM MeJA. The qRT‒PCR data revealed that CAD and GhMYC2-D09 are JA-responsive genes whose expression levels spike at 12 h and 3 h post treatment, respectively (Fig. 5C). These results indicate that GhMYC2-D09 is a JA-responsive transcription factor. We speculated that it may serve as an upstream regulator of CAD genes.

Fig. 5.

Validation of GhMYC2-D09 as a transcription factor. (A) Subcellular localization of GhMYC2-D09 in Arabidopsis protoplasts. (B) GhMYC2-D09 expression profiles in different tissues (root, leaf, stem, petal, bract, sepal, torus, and ovule at different stages). (C) CAD (Gh_D05G3506) and GhMYC2-D09 expression at specific time points (0, 1, 3, 6, and 12 h) after a 100 µM MeJA treatment. Values are presented as the mean ± SD of three independent biological replicates

GhMYC2-D09 overexpression promoted gossypol accumulation in hairy roots

To further characterize GhMYC2-D09 functionally, it was overexpressed in hairy roots (Fig. 6A). Specifically, three independent overexpression (OE) lines were obtained (Fig. 6B). We found that the expression of gossypol biosynthesis-related genes, including CAD, increased significantly in OE-GhMYC2-D09 hairy roots (Fig. 6C). Moreover, the gossypol content increased substantially in these OE lines (Fig. 6D). Hence, GhMYC2-D09 may be a positive regulator of gossypol biosynthesis.

Fig. 6.

Gossypol contents of OE-GhMYC2-D09 plants. (A) Overexpression of GhMYC2-D09 in hairy roots, with the empty pCAMBIA1302 vector serving as a control (CK). (B) Expression of GhMYC2-D09 in three OE-GhMYC2-D09 hairy root lines. (C) Expression of gossypol biosynthesis-related genes in OE-GhMYC2-D09 hairy roots. Gene accession numbers: CAD (Gh_D05G3506), CYP706B1 (Gh_D03G176100), CYP82D113 (Gh_D05G199000), CYP71BE79 (Gh_A13G153200), DH1 (Gh_A01G228800), 2-ODD-1 (Gh_D13G235700). (D) Gossypol content in OE-GhMYC2-D09 hairy roots. Bar: 1 cm. Values are presented as the mean ± SD of three independent biological replicates. Statistical significance: *P ˂ 0.05 (Student’s t test)

GhMYC2-D09 promoted GhCAD transcription

To determine whether GhMYC2-D09 directly regulates GhCAD expression, we first used a yeast assay system to analyse the transcriptional activity of GhMYC2-D09. Yeast cells harboring pGBKT7-GhMYC2-D09 grew well on solid SDO and TDO media, whereas yeast cells harboring an empty vector grew only on TDO medium (Fig. 7A), suggesting that GhMYC2-D09 confers transcriptional activity. Previous studies have shown that MYC2 can bind directly to G-box (CACGTG) and G-box-like (AACGTG/CATGTG/CACATG) motifs [64–67]. In a broader sense, the E-box motif (CANNTG) is the binding site of bHLH transcription factors [68]. The presence of E-box motifs in the 2 kb GhCAD promoter region was predicted using the PlantCARE database (Table S6). A dual-LUC assay was performed to determine whether GhMYC2-D09 activates GhCAD transcription. Because there are six GhCAD1-C subfamily members, we selected Gh_A04G1295 as a representative gene. Tobacco leaves were co-infiltrated with GhMYC2-D09 and three reporters, the activities of the GhCAD1-A-D, GhCAD1-A-D, and GhCAD1-C promoters increased significantly (relative to the corresponding control levels) (Fig. 7B–D). We also analysed whether GhMYC2-D09 can interact directly with GhCAD promoters. The direct binding of GhMYC2-D09 to GhCAD promoters was not detected via yeast one-hybrid (Y1H) assays (Fig. 7E).

Fig. 7.

Regulatory effect of GhMYC2-D09 on GhCAD expression. (A) Transcriptional activation assay in yeast. SDO (SD/−Trp medium); TDO (SD/−His/−Trp/−Ade medium). (B) Schematic representation of the reporter and effector. GhCAD1-A and GhCAD1-C promoter fragments (2,000 bp) were fused with the LUC gene. The effector plasmid contained the full-length GhMYC2-D09 sequence under the control of the CaMV 35 S promoter. (C) Effect of GhMYC2-D09 on the GhCAD promoter. LUC/REN, Luciferase/Renilla ratio. Values are presented as the mean ± SD of three independent biological replicates. Statistical significance: *P ˂ 0.05 (Student’s t test). (C) Effect of GhMYC2-D09 on GhCAD1-A and GhCAD1-C promoters in N. benthamiana leaves (E) Binding of GhMYC2-D09 to GhCAD1-A and GhCAD1-C promoters assessed in a Y1H assay

Discussion

CRISPR/Cas9 has been widely used to improve crop traits and elucidate gene functions [69–71]. To edit a target gene using CRISPR/Cas9 technology, multiple sgRNAs are typically designed and assessed for their application [72] because not all designed target sites work. Methods for evaluating target site editing efficiency have been reported in some plants, such as Chinese kale, soybean, and cotton [56, 73, 74]. In this study, the GhCAD gene family, which comprises several members, evaluation of editing efficiency at target sites before stable transformation was performed in the hairy root system, and ineffective target sites were detected and excluded. The selection of effective target sites was helpful for generating mutants and reducing the workload of genetic transformation.

Cotton is an economically important crop, with various byproducts of cotton processing, such as cottonseed meal, oil, and protein, contributing 15–25% of the overall value of cotton [75]. Cottonseed oil is commonly used for frying food because of its resistance to high temperatures, but it also has other uses, including in the energy (e.g., biofuel production), chemical, and agricultural industries [76]. Cottonseed protein and meal are included in food additives and are used for producing bioplastic films and coatings [77, 78]. The presence of gossypol, a toxic compound, has prevented the widespread use of cottonseeds. Thus, researchers have created low-gossypol cotton materials, including seeds, in which the gossypol content decreases by 98–99% [29, 32]. The gossypol contents in the mutants produced in the present study also decreased significantly, although the extent of the decrease was less than that reported previously.

Previous studies have shown no significant correlations among cottonseed gossypol, amino acid, or fatty acid contents [79, 80]. In the present study, cotton plants with decreased gossypol contents presented protein and oil contents that were similar to those in the R15 control (Fig. 2). However, free gossypol can bind to the free epsilon-amino group of amino acids, thereby affecting seed protein content and quality [81]. Thus, cottonseed protein content and distribution need to be more thoroughly investigated. Seeds collected from the gene-edited plants created in this study may be useful as feed additives. Notably, some of the sgAC2-1 T₂ plants that produced seeds with low gossypol levels lacked Cas9 (Fig. 2J), making them potentially useful for expediting the commercialization of low-gossypol cotton varieties.

Many studies have clarified the regulatory effects of MYC2 on the biosynthesis of secondary metabolites [82–86]. TaMYC2a plays a positive role in regulating Taxol biosynthesis by directly binding the T/G-box, G-box and E-box of the TASY and TcERF12/15 promoters [87]. In Salvia miltiorrhiza, SmMYC2 positively regulates phenolic acid biosynthesis by binding the E-box of genes in the phenolic acid biosynthetic pathway [88]. AaMYC2 binds the G-box-like motif within the promoter of key structural genes in the artemisinin biosynthetic pathway to positively regulate artemisinin biosynthesis [89]. In this study, GhMYC2-D09 was shown to positively regulate gossypol biosynthesis, but direct regulation was not detected. A previous study revealed that GhMYC2 positively regulates gossypol biosynthesis by binding to the promoter of the cytochrome P450 gene CYP71BE79 [90]. We determined that GhMYC2-D09 was homologous to GhMYC2. Both of these genes encode proteins that promote gossypol biosynthesis, albeit in different ways. GhMYC2 directly regulates gossypol biosynthesis, whereas our findings indicated that GhMYC2-D09 changed the GhCAD expression in unknown way to regulates gossypol biosynthesis, which warrants further study. We speculate that GhMYC2 may be involved, but this possibility needs to be experimentally verified.

Conclusions

In summary, we developed new low-gossypol resources by knocking out GhCAD genes via the CRISPR/Cas9 system and determined that GhMYC2-D09 is a JA-induced transcription factor that involved in regulating gossypol biosynthesis.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Lili Zhou: Writing-original draft, Visualization, Validation, Data curation, Software, Formal analysis; Yali Wang: Investigation, Formal analysis; Jiamin Wang: Investigation, Formal analysis; Peilin Wang: Writing-review & editing ; Guoqing Lu: Investigation, Formal analysis; Xiaofeng Su: Resources; Mahideen Afridi: Writing-review & editing; Huiming Guo: Writing-review & editing, Supervision, Resources; Hongmei Cheng: Writing-review & editing, Supervision, Project administration.

Funding

This research was supported by Biological Breeding-National Science and Technology Major Project (2023ZD04039), Science and Technology Project of the Ministry of Agriculture, Nanfan special project, CAAS (YBXM2516), Hainan Seed Industry Laboratory (project of B23CJ0208), Winall Hi-tech Seed Co., Ltd. (GMLM2023) and Hebei Key Technology R&D program (21326314D).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.