Natural variations in the Cis‐elements of GhRPRS1 contributing to petal colour diversity in cotton

Summary

The cotton genus comprises both diploid and allotetraploid species, and the diversity in petal colour within this genus offers valuable targets for studying orthologous gene function differentiation and evolution. However, the genetic basis for this diversity in petal colour remains largely unknown. The red petal colour primarily comes from C, G, K, and D genome species, and it is likely that the common ancestor of cotton had red petals. Here, by employing a clone mapping strategy, we mapped the red petal trait to a specific region on chromosome A07 in upland cotton. Genomic comparisons and phylogenetic analyses revealed that the red petal phenotype introgressed from G. bickii. Transcriptome analysis indicated that GhRPRS1, which encodes a glutathione S‐transferase, was the causative gene for the red petal colour. Knocking out GhRPRS1 resulted in white petals and the absence of red spots, while overexpression of both genotypes of GhRPRS1 led to red petals. Further analysis suggested that GhRPRS1 played a role in transporting pelargonidin‐3‐O‐glucoside and cyanidin‐3‐O‐glucoside. Promoter activity analysis indicated that variations in the promoter, but not in the gene body of GhRPRS1, have led to different petal colours within the genus. Our findings provide new insights into orthologous gene evolution as well as new strategies for modifying promoters in cotton breeding.

Introduction

Cotton is an important economic crop and an ideal model for studying species evolution and cell elongation. The genus Gossypium comprises over 50 species, including both diploids and tetraploids, exhibiting a wide range of phenotypic variation. Petal colour is one of the many distinctive traits in the genus. It serves not only as a key morphological marker for studying the evolution of closely related species, but also as a vital resource for investigating functional differentiation in orthologous genes. The petal colour of cotton can vary greatly, including red, pink, white, and yellow. Such diverse coloration indicates high genetic diversity; however, the evolutionary origins of this variation remain unclear, which hinders our ability to protect and utilize wild cotton resources effectively.

Petal colour is produced by various pigments, including chlorophyll, carotenoids, and flavonoids (Hao et al., 2020; Zhao et al., 2021; Zheng et al., 2022). Chlorophyll is primarily associated with green petals, such as the indoor ornamentals Dianthus caryophyllus (Shi et al., 2021) and Arabidopsis (Zheng et al., 2022). Conversely, carotenoids produce a wide range of petal colours from yellow to orange and red (Berardi et al., 2021). Notably, flavonoids, particularly anthocyanidin glycosides, serve as primary pigments (Sun et al., 2016). Anthocyanins can produce a spectrum of colours, from pale yellow to blue (Stanley et al., 2020). In cotton, several reports have demonstrated that flavonoids are the major pigments in petals (Cai et al., 2023; Li et al., 2020; Lu et al., 2017). Cotton petals undergo remarkable colour transitions during their short lifespan, with all the petals changing colours post‐anthesis. The mechanisms responsible for these dramatic colour changes remain largely unknown.

The biosynthetic pathways of anthocyanins are highly conserved among plant species, with enzymes catalysing a metabolic pathway from phenylalanine to anthocyanins (Sunil and Shetty, 2022). Anthocyanins are synthesized on the cytosolic side of the endoplasmic reticulum, and colourless anthocyanins undergo glycosylation modifications to generate stable anthocyanins. Anthocyanidin glycosides are transported to vacuoles for storage. The same anthocyanin can display different colours in the vacuole due to differences in pH value (Liang et al., 2020). However, a study of cotton showed that petal colour changes were not primarily associated with pH variation (Ke et al., 2022). A well‐described triple complex consisting of R2R3‐MYB (Cai et al., 2023; Liang et al., 2020; Lu et al., 2017), bHLH (Park et al., 2007; Qi et al., 2011), and WD40 (Tian et al., 2020; Walker et al., 1999) proteins regulates anthocyanin structural genes and promotes the final biosynthesis and accumulation of anthocyanin. In cotton, extensive efforts have been made to clone colour‐related genes. Mapping‐based cloning revealed that the MYB gene GhPAP1D on chromosome A07 was the causal gene underlying the R1 locus for red cotton plants and leaves (Li et al., 2019). In another study, a carotenoid synthase gene GhPSY was identified as a candidate gene from A07 in the R1 phenotype (Cai et al., 2014). Purple spot petals were also mapped to A07, and a MYB113 transcription factor GbBM was identified as the causal gene (Abid et al., 2022). Apart from transcription factors, glutathione S‐transferases play a role in anthocyanin transport and storage in anthocyanin‐producing plants, including maize, Arabidopsis, and grapevine (Alfenito et al., 1998; Jiang et al., 2010; Perez‐Diaz et al., 2016). Virus‐induced silencing of GhGSTF12 in red cotton leaves turned their red colour to green (Shao et al., 2021). In the diploid cotton G. arboreum, red‐spot petals were also mapped to the A07 chromosome, and a TT19‐like gene, Gar07G08900, was identified as a candidate gene (Zhang et al., 2022). Furthermore, a TT19‐like GST and the R2R3‐MYB GhPF from the A07 were identified as causal genes for the same red petal phenotype in upland cotton (Chai et al., 2022). However, the roles of GSTs in petal colour formation and its evolution in cotton genus remain poorly understood.

While forward genetics has facilitated the initial characterization of the Beauty Gene in cotton, limitations imposed by the genotype of transgenic cotton methods have resulted in a scarcity of published work directly targeting gene editing for petal colour mutations. This has led to different candidate genes being proposed for the same trait in various studies. Here we investigated the evolution of petal and petal‐spot colours in cotton. To identify the causal genes responsible for the red petal colour, we constructed different artificial populations for both preliminary and fine mapping. Gene editing and overexpression experiments were carried out in red and white petal cotton, ultimately confirming GhRPRS1 as the causal gene. We further conducted an introgression analysis to elucidate the origins of this gene, which was traced back to wild diploid cotton. To gain a deeper understanding of the role of GhRPRS1, we employed transcriptome, metabolome, and qPCR analyses to investigate its activity in transporting anthocyanins and its involvement in petal colour changes post‐anthesis. Finally, we analysed promoter activity and sequence variations to uncover the functional divergence of orthologs via promoter variation within the genus.

Methods

Plant materials

Gossypium hirsutum ZYH with red petals and red spots and ZM24 only with white petals were collected from our in‐house herbarium. For fine mapping, four F₂ populations consisting of 3182, 1891, 820, and 1651 individuals were obtained by self‐fertilization of the F₁ population derived from hybridization of ZYH and ZM24. Materials for mapping, mRNA‐seq, and qPCR analysis were also grown in the experimental field at Anyang (36.06° N, 114.81° E) and Sanya (18.33° N, 109.65° E) in 2020. Materials for qPCR and N. benthamiana for GFP, GUS, and LUC investigation were grown in a greenhouse at 25 °C in periods of 16 h light and 8 h dark. All plant tissues were immediately frozen in liquid nitrogen after collection and stored at −80 °C until analysis.

Sampling and sequencing

For bulked segregation analysis, fresh young leaves were collected and immediately frozen in liquid nitrogen from individuals consisting of two parents (ZYH and ZM24), 30 F₂ individuals with red petals, and 30 F₂ individuals with white petals. Genomic DNA was extracted with a previously reported method (Hu et al., 2020). Equal amounts of DNA from 30 extractions of red or white petals were mixed to generate two sequencing pools. The bulk segregation analysis (BSA) sequencing and analysis workflow was performed following previously reported protocols (Hu et al., 2020; Yang et al., 2019). The single‐nucleotide polymorphisms (SNP) index was calculated as (SNP index of the red petal pool)‐ (SNP index of the white petal pool). The average ΔSNP index was calculated using a 1000‐kb sliding window with a 10‐kb step size, and 99% confidence intervals of the ΔSNP index were calculated under the null hypothesis (no QTL) as the ΔSNP index threshold (Takagi et al., 2013). For mRNA‐seq, three petal growth stages were collected and total RNA was extracted with a previously reported workflow (Yuan et al., 2023; Zhan et al., 2021; Zhang et al., 2019). To clarify, the timing of petal sampling was defined as follows: S stage represents 3 days post‐budding, M stage represents 15 days post‐budding, and B stage represents 25 days post‐budding. The prefixes P or W before S, M, or B indicate ZYH and ZM24, respectively.

Identification of introgression fragment

To identify introgression derived from G. bickii in the G. hirsutum genome, we first used re‐sequencing data from a total of 109 accessions, including 20 G. barbadense, 5 G. darwinii, 60 G. hirsutum, 10 G. arboreum, 4 G. mustelinum, 6 G. tomentosum, and 12 wild diploid cottons (Table S4), to call SNPs against the G. hirsutum genome (TM‐1_ICR v1). To analyse their genetic relationships, a phylogenetic tree of all accessions was constructed using FastTreeMP based on the SNP matrix (Price et al., 2009).

Overexpression and knockout of GhRPRS1 in upland cotton

The full‐length GhRPRS1 (645 bp) CDS fragments were amplified by PCR from the cDNA of ZYH and ZM24 using the primers GhRPRS1‐067‐F and GhRPRS1‐067‐R (Table S3). The WMV067‐AADA overexpression vector was digested with SalI and BamHI and the fragments of purification were inserted into this plant binary vector driven by the cauliflower mosaic virus (CaMV) 35S promoter. The WMV067‐AADA‐GhRPRS1 vectors were introduced into Agrobacterium strain EHA105 and transformed into the cotton cultivar ZM24. Transgenic workflows were performed with a previously reported method (Ge et al., 2023).

For gene edit analysis, two targets of GhRPRS1 5′‐GGTTGACGGAGGAGG‐3′ and 5′‐TTGAGACACCATGTGCCACAAAT‐3′ were evaluated and selected. The pAtU6‐sgRNA‐gRNAU6t fragment was amplified using target primers (Table S3) from the AT‐26‐PSG vector and inserted into the AADA‐C016 vector using the single restriction endonucleases BsaI. For gene editing materials, we first detected the positive Cas protein (Table S3) and then used next‐generation sequencing (NGS) to sequence the target sites. Generally, the target region was amplified from genomic DNA with site‐specific primers in the first round of PCR (Table S3). In the next round, both forward and reverse barcodes were added to the ends of the PCR products (Table S3). Equal amounts of PCR product were sent for NGS at the State Key Laboratory of Rice Biology using the Hi‐TOM platform (China National Rice Research Institute, Chinese Academy of Agricultural Sciences, Hangzhou, China) (Liu et al., 2019).

Transient expression of GhRPRS1

The basic vector pCambia2300 was used for GhRPRS1 transient overexpression in cotton cotyledons. The full‐length CDS fragments of GhRPRS1‐red and GhRPRS1‐white were amplified and were driven by the cauliflower mosaic virus (CaMV) 35S promoter. The vectors were transformed into Agrobacterium strain GV3101 for the transformation, and the resulting bacterial suspension was adjusted to an optical density (OD) of 0.6. An equal volume of Agrobacterium strain GV3101 (0.5 mL) for each vector was then injected into fully flattened DC109 leaves. Cotton seedlings with fully flattened cotyledons which were about 10 days old were used for Agrobacterium injection. Light was avoided for 2 days. The suspensions are described in detail in our previous study (Zhang et al., 2023). Leaves were sampled for determination of anthocyanin content and pictures.

Fine mapping of red petals

We analysed InDel markers within the candidate region (A07: 8−15 Mb) to designed 48 pairs of InDel primers using Primer3, and selected four SNPs for KASP markers. Two parents and 10 randomly selected F₂ individuals were used to test the primers for polymorphism detection efficiency. Then a small population, consisting of 94 F₂ individuals and two parents, was used to test the linkage between markers and phenotypes. Finally, the linked markers were used to calculate recombination values in a large population consisting of 7544 F₂ or BC₁F₁ individuals. The extraction of genomic DNA, PCR conditions, and genotyped by polyacrylamide gel electrophoresis followed previous research (Hu et al., 2020).

RNA‐seq and qPCR analysis

A total of 11 different tissues were sampled for qPCR analysis for GhRPRS1 expression, including stamen, root, stem, leaf, cotyledon, hypocotyl, ovule, fibre (3 DPA), flowering bud, sepal, anther, and petal. To further analyse the expression of GhRPRS1 in different petal development stages, a total of 17 different development stage petals were sampled. Total RNA was extracted using a FastPure Universal Plant Total RNA Isolation Kit (Vazyme, Nanjing, China) and cDNA was synthesized using HiScript III 1st Strand cDNA Synthesis Kit (+gDNA wiper; Vazyme, Nanjing, China). The GhUBQ7 (Gh_D11G11500) gene was used as an internal control for cotton. qPCR was performed following the SYBR® Premix Ex Taq™ (Takara, China) manufacturer's protocols on a Roche 480 Real‐Time PCR System (Applied Biosystems, USA). The three‐step method involved the following PCR conditions: 40 cycles of 95 °C for 30 s, 95 °C for 5 s, and 60 °C for 30 s. The 2^−ΔΔCT method was used to calculate the relative expression levels of each target gene. Three replicates were conducted for each sample.

Promoter activity analysis

The 2 Kb promoters of GhRPRS1 in ZYH and ZM24 were isolated and cloned into the pBI121 vector by replacing the 35S promoter, and then the constructed vector was introduced into Agrobacterium strain GV3101. Various promoter deletions of GhRPRS1 were also subcloned into the pBI121 plasmid. Further, we used the pGreen‐35Smini‐0800 system to verify the promoter element activity. The promoter element of the short fragment was repeated three times in tandem and directly synthesized and ligated into the pGreen‐35Smini‐0800 vector. Then they were introduced into Agrobacterium strain GV3101 with pSoup‐p19. Nicotiana benthamiana plants were grown under 16 h light and 8 h dark periods at 28 °C for 4 weeks and used for Agrobacterium infiltration for GUS and LUC transient expression analysis. Transformed Agrobacterium cells were resuspended in 10 mm 2‐(N‐Morpholino)‐ethanesulfonic acid (MES) buffer (pH 5.7) with 10 mm MgCl₂ and 100 μm acetosyringone, and kept at dark for 3–6 h at room temperature. After 48–72 h, the infected areas were harvested to determine the expression of GUS, GUS staining, and LUC activity assays.

Targeted metabolomic analysis of anthocyanin

Targeted metabolomic analysis was performed on two stages of petal development in ZYH and ZM24 before and after colour of the petal changed. First, the petal sample was vacuum freeze‐dried and ground (30 Hz, 1.5 min) to a powder using a mill (MM 400, RETSCH, Germany), and then 100 mg of powder was dissolved in 1.0 mL of extract solution (70% methanol in water). The dissolved sample was refrigerated overnight at 4 °C, and vortexed three times during the period to increase the extraction rate. After centrifugation (10 000 g for 10 min), the supernatant was aspirated, filtered with a microporous membrane (0.22 μm pore size), and stored for LC–MS/MS analysis. Three replicates were conducted for each sample.

Protein extraction and western blotting

For protein extraction and western blotting for transient expression analysis, 0.5 g transgenic cotton leaves expressing the GhRPRS1 proteins with a GFP tag were sampled and ground into a powder with liquid nitrogen. An equal volume of protein isolation buffer (1 mm EDTA pH 8.0, 20 mm Tris–HCl pH 7.5, 5 mm dithiothreitol, 150 mm NaCl, 0.1% sodium dodecyl sulphate [SDS], 10% glycerol, and 1 × protease inhibitor cocktail) was then added to the powder and incubated for 30 min at 4 °C. The mixture was centrifuged at 4 °C for 10 min at 13 000 g . The supernatant was transferred to a new tube and boiled in protein sample buffer (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) for 5 min. Proteins were separated on 10% SDS–SDS‐polyacrylamide gel electrophoresis (SDS‐PAGE) and immunoblotted with different antibodies according to methods used by a previous study (Liu et al., 2021; Zhang et al., 2023).

Extraction and quantify of anthocyanins

Anthocyanin content is calculated by measuring absorbance values at 530 nm and 700 nm at different pH values. A total of 1 g cotton petal was harvested and ground into a powder with liquid nitrogen. Approximately, 0.1 g sample was mixed with 1 mL of extraction solution buffer, and this was fully homogenized and transferred to an EP tube for incubation at 60 °C for 30 min. The resulting extract was diluted to 1 mL. Finally, this was centrifuged at 13 400 g at room temperature for 10 min, and the supernatant was aspirated for testing. Determination and calculation procedures followed manufacturer instructions for the plant anthocyanin content assay kit (Beijing Solarbio Science & Technology Co., Ltd., Beijing, China).

Dimethylaminocinnamaldehyde (DMACA) staining of different cotton tissues

The DMACA (p‐Dimethylaminocinnamaldehyde) staining method is commonly used to detect the presence of anthocyanins in plant tissues. Samples were immersed in 0.1% DMACA solution in 6 N HCl and 95% ethanol (1:1 ratio). After staining for 10–30 min, the tissues were washed three times with distilled water. Photographs of the stained samples were taken while they were immersed in anhydrous ethanol.

Statistical analyses

Statistical analyses were conducted using R software (version 4.0.2). The data were presented as mean ± standard deviation (SD). Group differences were evaluated using Student's t‐test or one‐way ANOVA followed by Tukey's post hoc test for multiple comparisons. A significance level of P < 0.05 was considered statistically significant. GraphPad Prism (version 9.0) was utilized for additional data visualization and statistical analyses.

Results

High diversity of petal colours in cotton

To investigate cotton petal colour differentiation, we conducted a comprehensive analysis across 29 species spanning the eight diploid cotton chromosome groups (A through G, plus K) and the allotetraploid (AD) group (Table S1). Within the American‐clade cotton species, a wide variety of petal colours were observed. Species like D₄, D₆, D₇, D₉, and D₁₁ exhibited red petals, while the remaining species primarily displayed yellow or white petals. The Asian‐African clades predominantly exhibited varying hues of yellow petals. On the other hand, the Australian clade, which includes the C₁, G₁, G₂, G₃, and K₂ genome species, mainly featured red or pink petals. These results suggest a strong correlation between the geographic distribution of cotton species and their petal colours (Figure 1a).

Figure 1.

Phenotyping and genetic analysis of petal colour in cotton genus. (a) Investigation of petal colour and geographic distribution in 29 diverse cotton species. The 29 species are categorized into 8 diploid chromosomal groups (A–G plus K) and the allotetraploid AD group. G. aridum (D₄), G. gossypioides (D₆), G. lobatum (D₇), G. laxum (D₉), G. schwendimanii (D₁₁), G. sturtianum (C₁), G. bickii (G₁), G. australe (G₂), G. nelsonii (G₃), and G. rotundifolium (K₂) exhibit a red petal colour. (b) Evolutionary relationships among cotton species with different petal colours. (c) Genetic analysis of the red petal by crossing G. hirsutum (ZYH) with G. hirsutum (ZM24), and G. barbadense (3–79), respectively.

In comparison to wild diploid cotton, we observed that cultivated cotton, including G. hirsutum and G. arboreum, exhibited a broader range of petal colours. G. arboreum, for instance, displayed at least three petal colours, including white, yellow, and red. Upland cotton, on the other hand, had a minimum of four distinct petal colours, encompassing white, cream, yellow, and red. This indicates that artificial selection has contributed to an increased diversity of petal colours in cultivated cotton.

Molecular phylogenetic analysis suggests that Gossypium diverged from Hawaiian Kokia and the African‐Madagascaran genus Gossypioides ~ 5–10 MYA (Hu et al., 2021; Wang et al., 2022a; Wendel et al., 2010). Notably, both Kokia and G. gossypioides exhibit red petals, implying that the ancestor of Gossypium likely had red petals, with non‐red petal colours subsequently evolving from this red‐petal ancestor (Figure 1b).

Given that red and white are the primary petal colours within the genus, we selected two upland cotton accessions, one with red petals (ZYH) and the other with white petals (Figure S1), to study the genetics and evolution of petal colours. We constructed three F₂ populations and one BC₁F₁ population for genetic analysis. All F₁ plants exhibited red petals with red spots, and the F₂ population exhibited a ratio of 3:1 red to white petals (Table S2). This 1:1 ratio in the BC₁F₁ population reaffirmed that the red petal allele is dominant. Viewed together, our results suggest that petal colour is regulated by a single gene with alleles in a dominance relationship and can be utilized as a marker trait for investigating the differentiation of orthologous genes within the genus (Figure 1c).

Genetic analysis and fine mapping for red petals with red spots

To identify the causal gene responsible for red petals, we employed a BSA strategy on F₂ population. In line with our genetic analysis results, we pinpointed a specific QTL on chromosome A07 associated with both red petal colour and red spot, which has been designated as Red Petal and Red Spot 1 (RPRS1) (Figure 2a). This QTL is situated within a 5.4 Mb region spanning from 8.0 to 13.4 Mb on chromosome A07 (Figure 2b).

Figure 2.

Genetic mapping of red petal with red spot. (a) BSA analysis conducted on the red petal phenotype within an F₂ segregating population. The ΔSNP index (representing the SNP index difference between the red and white bulk populations) is illustrated by red lines, with the 99% confidence interval depicted in black. (b) Zoomed‐in the BSA signal on chromosome A07 from Figure 1a. (c) The polymorphism InDel markers locating on A07 signal interval. The digit within the bracket is the number of recombination individuals. (d) The signal interval was narrowed down between InDel9099 and InDel1535. (e) SNP density distribution on chromosome A07 for ZYH and ZM24. (f–h) The number of SNPs (f), Indels (g), and mapped reads (h) on chromosome A07. The TM‐1 genome was used as a reference genome for read mapping.

To further refine the mapping of GhRPRS1, we utilized four distinct populations consisting of three intra‐upland cotton F₂ populations and one G. barbadense–G. hirsutum F₂ population. We developed a set of 51 markers showing diversity between the parental lines ZYH (red petal) and ZM24 (white petal) across a genomic region ranging from 5 to 18 Mb. Using four KASP markers, we successfully narrowed down the RPRS1 region from an 8 820 350 bp segment to 13 400 000 bp when analysing a panel of 820 individuals (Figure S2).

Subsequently, by examining 1891 F₂ individual plants exhibiting recessive traits, we were able to further narrow down the RPRS1 region to a more specific interval located between InDel9099 and InDel1535 using a set of 47 markers (see Figures 2c,d and S3, Table S3). This refined region encompasses ~4.25 Mb and contains a total of 187 protein‐coding genes. Remarkably, no recombination events were observed when studying an additional 3182 individuals, indicating that recombination is highly restricted within this 4.25 Mb interval (Table S4).

To gain a deeper insight into the genetic characteristics of the parental lines ZYH and ZM24, we performed a comparative genomic analysis. Our findings revealed that ZYH exhibits a higher density of both SNPs and insertions/deletions (InDels) compared with ZM24 (Figures 2e and S4). Additionally, when we assessed the alignment of mapped reads to the TM‐1 reference genome, ZM24 displayed a substantially greater number of mapped reads compared with ZYH (Figure 2f–h). This suggests that ZYH has a more distant phylogenetic relationship with both ZM24 and TM‐1.

Introgression from the wild cotton G. bickii introduced the red petal to upland cotton

Many studies show that recombination is suppressed in alien introgression regions (Nagy et al., 2010). Given that the red or pink colour is predominantly found in species with the D or C, G, and K genomes, it is possible that the origin of RPRS1 involves introgression from wild cotton species. To trace the potential origin of RPRS1, we collected 109 accessions from 9 species to reconstruct the phylogenetic relationship of RPRS1 (Table S5). When we used the SNPs from RPRS1 to construct a phylogenetic tree, we observed that ZYH clustered together with G. bickii, while ZM24 clustered together with G. hirsutum accessions. However, when we utilized the SNPs from the border regions of the RPRS1 locus, both ZYH and ZM24 clustered together with G. hirsutum in both phylogenies (Figure 3a). Consequently, our results suggest that RPRS1 likely originated from G. bickii.

Figure 3.

Validation of the RPRS1 locus deriving from introgression. (a) Comparison of phylogenetic trees analysis among RPRS1 locus, its upstream border and downstream border. SNPs from of 109 accessions were used for the tree construction (b) The read mapping depth of ZYH, ZM24, red petal pool, and white petal pool within 8–16 Mb interval on chromosome A07. All the data were mapping to the TM‐1 reference genome. (c) Mapped read number on chromosome A07 of G. hirsutum within 8–16 Mb. (d) The relative read mapping depth of ZYH, ZM24, red petal pool, and white petal pool within 8–16 Mb interval on chromosome A07. All the data were mapping to the integrated sequences of G. bickii Chr07 and TM‐1. (e) Mapped read number on chromosome Chr07 of G. bickii within 8–14 Mb.

To further validate the introgression from G. bickii, we employed read mapping depth analysis. Initially, we merged the Chr07 of G. bickii with the TM‐1 genome to generate a new reference genome. Subsequently, we mapped the reads from both parent lines and descendant pools to this new reference genome. ZM24 and the white petal pool displayed a high read mapping ratio on the upland cotton A07, while ZYH and the red petal pool exhibited a lower read mapping ratio in the RPRS1 locus, forming a “U” pattern. Conversely, ZYH and the red petal pool displayed an “n” pattern when mapping against G. bickii Chr07 (Figures 3b,e and S5). The number of mapped reads further supports the notion that ZYH and the red petal pool are more closely related to G. bickii than to G. hirsutum in the RPRS1 locus (Figure 3c,d).

Furthermore, we employed 11 SSRs to assess genetic relationships of the RPRS1 locus among ZM24, ZYH, and G. bickii. The results revealed that all SSRs were found in the genomes of G. bickii and ZYH but were notably absent in ZM24 (Figure S6). Collectively, our findings strongly suggest that alien introgression from G. bickii has imparted the red petal trait to ZYH.

Glutathione‐S‐transferase is the causal gene for the red petal colour

To identify the causal gene in the RPRS1 locus, we collected petal samples for mRNA‐seq analysis from both ZYH and ZM24 at three different stages of petal development: 3 days after budding (referred to as the S stage), 15 days after budding (the M stage), and 26 days after budding (the B stage). Principal component analysis (PCA) indicated that the three replicates for each treatment clustered together, demonstrating the high reproducibility of our data (Figure S7a). We employed DESeq2 to identify differentially expressed genes (DEGs) by comparing samples at the same developmental stage between ZYH and ZM24. This analysis revealed a total of 1068 DEGs, consisting of 838 upregulated genes and 299 downregulated DEGs (Figure S7b). Since there was no discernible difference in petal colour between ZYH and ZM24 at the S stage, it suggests that the causal gene might not manifest differential expression during the initial growth phase. Consequently, our focus shifted to genes that displayed differential expression in both the B and M stages but not in the S stage. Among the DEGs identified, 16 exhibited stable differential expression across B and M stages (Figure S7c,d). It should be noted that 15 out of the 16 DEGs are located in the RPRS1 locus, with 10 genes being downregulated and 5 genes being upregulated and in ZYH compared with ZM24. To further investigate their roles, we next knocked down the 10 downregulated genes (GPL, TTF5A, S7E‐1, PP, HMGY, WAS2, COML, CAL, TPA, and SUS5) in white petal ZM24. The results indicated that silencing these genes did not alter petal colour (Figure S7e,g), suggesting that they were not involved in the pink petal coloration. Similarly, we knocked down 5 upregulated genes (PUP, GLU, FLT, VVM‐6, and GST) in red petal material ZYH (Figure S7f,h). Among these, only GST showed a change in petal colour upon reduction of its expression, suggesting that GST is a candidate gene influencing red petal coloration.

The association of red petals with R2R3 MYB transcription factors is well‐documented in many flowering plants (Cai et al., 2014; Li et al., 2019; Lu et al., 2017). In Gossypium barbadense, a homologue of MYB113, termed GbBM, has been identified as a causal gene regulating purple spot formation at the base of flower petals (Abid et al., 2022). Within the RPRS1 locus, there is a tandem duplication of three MYB113 genes (Gh_A07G083400, Gh_A07G083500, and Gh_A07G083600) as well as a MYB62 gene (Gh_A07G072200). Among these MYB113 genes, Gh_A07G083400 shows no expression in any of the three stages of petal development. Although Gh_A07G083600 displays significant differential expression, its expression level is higher in white petals compared with red petals. Given the positive role of MYB113 in petal colour, Gh_A07G083600 seems unlikely to be the causal gene in the RPRS1 locus. The remaining gene, Gh_A07G083500, is an ortholog of GbBM. Unlike the upregulated expression of GbBM in G. barbadense when compared with upland cotton, we found that Gh_A07G083500 does not exhibit significant differential expression between ZM24 and ZYH, suggesting that Gh_A07G083500 is not responsible for the petal colour difference between ZM24 and ZYH (Figure S8a). Furthermore, the function of the three MYB113 genes was validated through virus‐induced gene silencing (VIGS) experiments. These experiments revealed that silencing the genes did not alter the colour of the petals (Figure S8b,c). Despite MYB62 showing higher expression level in ZYH compared with ZM24 specifically in stage B, the Cas9 mutants did show any observable changes in petal colour (Figure S7i,j). Consequently, MYB113s and MYB62 were excluded as candidate genes for the RPRS1 locus.

These findings suggested that GST could be a candidate gene for RPRS1, and as a result, we named this gene as GhRPRS1. To further investigate the function of GhRPRS1, two different experimental approaches were employed: overexpression and gene editing. Initially, GhRPRS1 was cloned from both ZYH and ZM24, revealing the presence of only three nonsynonymous SNPs. Two of these SNPs caused a change from proline to alanine and valine to alanine, while the other resulted in lysine to methionine (Figure S9). In the overexpression experiment (Figure 4a), both genotypes of GhRPRS1 were introduced into the ZM24 background. Remarkably, both genotypes led to red petal colour (Figure 4b). Anthocyanin accumulation was also observed in other tissues in the overexpression lines (Figure S10a). This intriguing result suggests that the nonsynonymous SNPs did not affect the gene function. Analyses of anthocyanin content further supported the observation that both genotypes enhanced anthocyanin accumulation in the transgenic lines (Figure 4c). Furthermore, both the expression level and protein level of GhRPRS1 in the transgenic lines were highly correlated with the intensity of red colour and anthocyanin content (Figure 4d–f).

Figure 4.

Functional validation of two genotypes of GhRPRS1 in cotton. (a) The WMV067‐AADA vector was employed for overexpressing the GhRPRS1 in ZM24. (b) Two genotypes of GhRPRS1 cloned from white petal and red petal were overexpressed in ZM24. Lines RPRS1‐5 and RPRS1‐6 were generated by overexpression GhRPRS1 from white petal, while RPRS1‐9 and RPRS1‐10 were generated by the overexpression of GhRPRS1 from red petal. ZM24 served as controls for white petals. (c) The measurement of anthocyanin content in GhRPRS1 overexpression lines. (d–f) The expression levels of GhRPRS1 in overexpression lines were detected by qPCR and semi‐quantitative PCR. GhActin served as an internal control and western blotting of GhRPRS1 in transgene lines. (g) WMVC016 was utilized for gene editing of GhRPRS1 in ZYH. Three mutants with base deletion were observed at the target position. rprs1‐10, rprs1‐4, and rprs‐2 were the knockout lines in ZYH. (h) The phenotype of three GhRPRS1 knockout lines in ZYH, ZYH served as controls for red petals. (i) The measurement of anthocyanin content in GhRPRS1 knockout lines. (j) Comparison of petal colours between ZM24 and ZYH. A total of 13 stages were photographed from little bud to flowering. (k) Expression level of the GhRPRS1 in 13 stages as shown in panel j. (l) Comparison of petal colours before and after flowering between ZM24 and ZYH. (m, n) Comparison of expression level of GhRPRS1 between ZM24 and ZYH in four stages as showed in panel (k). Data were analysed using GraphPad Prism (v8.0.2, GraphPad Software, United States) software. Statistical testing was applied using a Student's t‐test, statistical significance is defined as P < 0.05.

Given that ZYH is controlled by a single dominant allele, we also employed CRISPR technology to generate GhRPRS1 mutants in ZHY background (Figures 4g and S10b). Three mutant rprs1‐10, rprs1‐4, and rprs1‐2 was obtained carrying a 17‐bp, 2‐bp, 1‐bp deletion leading to a stop‐gain mutation and expressing a white petal phenotype (Figure 4h). Anthocyanin content was significantly reduced in the lines rprs1‐10, rprs1‐4, and rprs1‐2, when compared with ZYH (Figure 4i). These findings collectively confirm that GhRPRS1 is indeed the causal gene responsible for red petals and the appearance of red spots.

GhRPRS1 enhanced anthocyanin accumulation in the red petal

Anthocyanidins are water‐soluble natural pigments that are widely distributed across plant taxa and play a crucial role in petal colour formation. We observed a substantial difference in anthocyanidin content between ZYH and ZM24 (Figure S11). To determine which specific types of anthocyanidins contribute to petal coloration, we selected two stages for targeted anthocyanin metabolomics analysis: petals without colour differences (PS and WS) and petals with colour differences (PB and WB). In our study, we detected a total of 227 anthocyanin‐related metabolites in petals. Of these, 18 were identified as differential metabolites that are associated with the formation of red petals (Figure S12). Apart from the well‐known anthocyanins, such as cyanidin‐3‐O‐glucoside, cyanidin‐3‐O‐galactoside, and pelargonidin‐3‐O‐glucoside, which are commonly associated with floral coloration, our study has identified an additional 15 metabolites contributing to the red pigmentation of petals. These include cyanidin‐3‐O‐rutinoside (keracyanin), cyanidin‐3‐O‐(6″‐O‐malonyl) glucoside, naringenin chalcone, 6‐O‐Malonylgenistin, quercetin‐3‐O‐glucosyl (1 → 3) rhamnosyl (1 → 6) galactoside, epicatechin, apigenin‐7‐O‐rutinoside, rhamnetin (7‐O‐Methxyl Quercetin), kaempferol‐3,7‐O‐dirhamnoside, epicatechin‐epiafzelechin, epicatechin glucoside, 6″‐O‐Malonylgenistin, quercetin‐3‐O‐(6″‐acetyl) glucoside, procyanidin B2, and procyanidin B4.

We also found a strong correlation between the gene expression levels of GhRPRS1 and the content of these 13 differential metabolites, highlighting the involvement of GhRPRS1 in regulating petal colour (Figure S13). Several other differential metabolites were identified as intermediate metabolites in the anthocyanin synthesis pathway, and they also exhibited a robust correlation with GhRPRS1. It is noteworthy that none of the key genes associated with anthocyanin biosynthesis exhibited significantly different expression levels within the anthocyanin biosynthesis pathway (Figure S12). This observation suggests that the GhRPRS1 gene plays a crucial role in the pigmentation of petals and the development of petal base spots.

Expression differentiation of GhRPRS1 resulted in different petal colours

Since overexpression of GhRPRS1 from both white and red petals led to red petal phenotype (Figure 4c), we next analysed the gene expression levels of GhRPRS1 in the ZYH and ZM24. Initially, we examined expression across 11 different tissues and observed a preferential expression in petals (Figure S14). Additionally, DMACA staining confirmed that petals had the highest anthocyanin content when compared with the other seven tissues (Figure S15). To gain insight into the expression patterns of GhRPRS1 during petal development, we collected 13 petal samples from both ZYH and ZM24 at various stages, ranging from bud to fully bloomed flowers. The results revealed that ZM24 maintained a consistently low expression level throughout all stages, while ZYH exhibited an increase in expression starting from the 4th stage and maintained a high expression level from the 9th to the 13th stage (Figure 4j,k).

An interesting phenomenon to note is the colour transformation in white petals, which remain white until they are pollinated on the day of flowering. After pollination, the edges of the petals gradually turn red until they eventually fall off. (Figure 4l). In both ZM24 and ZYH, we examined the petals at four stages before and after pollination to observe the changes in GhRPRS1 expression. The expression patterns of GhRPRS1 in ZYH and ZM24 are quite similar. GhRPRS1 starts with high expression before flowering (1st stage), decreases at the 2nd stage, and then gradually increases during the 3rd and 4th stages. ZM24 shows a greater increase compared with ZYH (Figure 4m,n). The anthocyanin content also rose from the 2nd to 4th in both ZYH and ZM24 (Figure S16a–d). These results suggest that GhRPRS1 plays a role in turning white petals red after pollination. We confirmed this by observing petal colour in gene‐editing GhRPRS1 lines, where the petals remained white after pollination (Figure S16e). Taken together, our results indicate that variation in promoters has led to changes in the expression profiles of GhRPRS1, which subsequently resulted in phenotypic variation.

Cis‐elements variations in promoter resulted in GhRPRS1 function differentiation

To gain a deeper understanding of GhRPRS1's role in cotton petal colour formation, we conducted a comparative analysis of the amino acid sequences of GhRPRS1 in 21 cotton species (Figure S17). The results from multiple sequence alignments indicated a high degree of conservation in the GhRPRS1 gene across the genus, with only six nonsynonymous mutations identified. Notably, these nonsynonymous mutations did not appear to be associated with petal colour variation since they did not distinguish red and non‐red petals. Phylogenetic analysis further supported this observation, showing a mixed distribution of red and non‐red petal species (Figures 5a and S18).

Figure 5.

Cis‐element variations in promoters of GhRPRS1 resulted in petal colour differentiation in cotton. (a) Six haplotypes were identified in 21 cotton species. (b) Transient expression of the two haplotypes of GhRPRS1 in the cotyledons of DC109. (c) Comparison promoter sequence similarities of GhRPRS1 from 26 cotton species. (d) A series of vectors carrying GUS were constructed for promoter activities analysis. F1, F2, F3, and F4 represent different length of truncated sequences. (e) Promoter activities of F1–F4 were tested in tobacco leaves by GUS staining. (f) Expression levels of GUS were detected 3 days after injection. Actin served as an internal control. (g) Dot plot analysis of promoter sequences between ZYH and ZM24. E1, E2, and E3 represent the Indels between the two promoters. (h) E1, E2, and E3 were identified −500 to −1500 regions. (i) E1, E2, and E3 were integrated into the pGreenII‐0800‐35smini‐LUC vector to test their regulated activity in tobacco leaves, respectively. (j–l) Transient expression E1 (j), E2 (k), E3 (l) using the pGreenII‐0800‐35smini‐LUC system in tobacco leaves. The luciferase (LUC) activity detected 3 days after injection. Fluorescence intensity is captured, and data represent means ± SE of three independent experiments. Data were analysed using GraphPad Prism (v8.0.2, GraphPad Software, United States) software. Statistical testing was applied using a Student's t‐test, statistical significance is defined as P < 0.05.

To investigate whether GhRPRS1 protein from ZYH and ZM24 is responsible for anthocyanin accumulation, we conducted transient expression assays in cotton cotyledons. Two variants of GhRPRS1 were evaluated in DC109. As illustrated in Figure 5b, a noticeable increase in red pigmentation was observed at the injection site for both GhRPRS1 haplotypes. This observation strongly suggests that GhRPRS1 functions similarly in both red and white petals. Thus, the differences in petal colour may be attributed to variants in the promoter region of the gene.

To explore the impact of promoter sequence variants on GhRPRS1 expression, we cloned the 2000‐bp promoter regions of ZM24 and ZYH. Sequence alignment revealed that the ZYH promoter exhibited a low similarity with the ZM24 promoter (Figures 5c and S19). A promoter sequence‐based phylogenetic tree revealed that ZYH is closely related to species with red petals, particularly G. bickii, while ZM24 clustered together with G. hirsutum TM‐1 (Figure S20), further confirming that the RPRS1 locus is derived from G. bickii. To identify the specific fragment responsible for the differential expression of GhRPRS1 between ZYH and ZM24, we systematically truncated the promoter sequence into three sections, each involving a 500 bp deletion from the 5′ end. These truncated fragments were designated as F1–F4 (Figure 5d,e). Subsequently, we cloned these fragments into the PBI121‐GUS vector and introduced these constructs into tobacco leaves to assess their expression. The blank vector (without the 35S promoter) served as the control, as shown in Figure S21. When F1, F2, or F3 were introduced into tobacco leaves, the ZYH promoter demonstrated significantly higher activity compared with the ZM24 promoter. In contrast, when F4 was introduced into tobacco leaves, the ZYH promoter exhibited activity similar to that of the ZM24 promoter (Figure 5f). Since there was only a subtle difference between the ZYH and ZM24 promoters in the case of F3, we inferred that the fragment spanning from −1500 to −500 likely plays a crucial role in regulating promoter activities. A dot plot analysis revealed the presence of four substantial InDels between the ZYH and ZM24 promoters. These InDels were suspected to influence the promoter activities of ZM24 (Figure 5g,h). Based on the locations of these InDels, we generated three pairs of fragments, some with the insertions and others without, and integrated them into the pGreen‐35Smini‐0800 vector. Subsequently, these constructs were transiently expressed in tobacco leaves. Both luciferase activity assays and fluorescence intensities showed that ΔE1 and ΔE2 exhibited very weak promoter activities, suggesting that these elements were not responsible for the differential expression of GhRPRS1 between ZYH and ZM24. ΔE3 displayed robust promoter activity, but it was weaker than E3 (Figures 5i–l and S22). This observation indicates that the deletions within the E3 region have a significant impact on the activity of the ZM24 promoter.

Moreover, overexpression experiments revealed that introducing orthologs of GhRPRS1 from both upland cotton and G. bickii resulted in similar red petal coloration. This strongly suggests that GhRPRS1 performs a conserved function in anthocyanin transport within the genus. To confirm this conserved function, we cloned GhRPRS1 from various cotton species and tested their ability to transport anthocyanin by transiently expressing them in the cotyledons of the DC109 cultivar. The results demonstrated that GhRPRS1 from all the species tested was capable of anthocyanin transport (Figure S23). Therefore, our findings indicate that genetic variation within GhRPRS1 have not altered its role in petal colour formation.

Discussion

A wide variety of petal colours, which are related to environmental adaptability, contribute to survivorship. The colour of petals is usually determined by the pigments on the petals. Petal colour selection is often related to the plant pollinators. Several studies have shown that plants with brightly coloured petals attract the attention of certain pollinating animals (Gómez et al., 2020; Grotewold, 2006). Petal colour is one of the most important aesthetic traits of ornamental plants, as well as an important breeding goal for horticulturists (Han et al., 2022). Cottons exhibit a broad range of petal colours, and due to either natural or artificial selection, these colours have varying geographical distributions and wide applications. Anthocyanins, such as cyanidin, pelargonidin, and delphinidin, have been reported to regulate petal colour. In cotton, petal colour is also influenced by the synthesis and distribution of anthocyanins (Li et al., 2023; Xing et al., 2022). Anthocyanins from cotton not only have significant antioxidant activity but also can be used as food colour additives in pastries, biscuits, candies, and other foods (Belwal et al., 2020). Recently, breeders use petal colour‐forming genes to create coloured fibres or in transgenic materials for target gene fusion expression (Ge et al., 2023; Wang et al., 2022a, 2022b). In this study, we identified a red petal‐related gene named GhRPRS1, which is abundantly and preferentially expressed in red petals. GhRPRS1 serves as a carrier involved in anthocyanin transport and promotes petal colour formation. Our study showed that RPRS1 locus is introduced by an introgression, given that many studies in cotton as well as in other species shows that alien fragments restrict genomic recombination. Our study also showed a similar situation that introgressions reduce the recombination events.

Due to limitations in genotype and transgenic platforms, only a few researchers have attempted gene editing in dominant cotton mutants. VIGS has traditionally been a favoured strategy for validating gene function in cotton, particularly for fine mapping. Our genotype‐independent transgenic platform offers distinct advantages for manipulating red petal and red‐spot petal traits in ZYH, enabling the rapid generation of both overexpression lines and gene editing mutant. This approach holds significant promise for validating gene function effectively.

In different species, GST has been reported to be related to petal colour formation, as structure variations can cause petal colour differences (Conn et al., 2008; Liu et al., 2022; Lu et al., 2021; Sun et al., 2012). There are studies on different cotton species reporting that the GST gene is associated with red petals (Chai et al., 2022), red‐spot petals (Zhang et al., 2022a), or the formation of red plants (Shao et al., 2021). In our study, employing forward genetic mapping, we identified an association between GST and petal colour. Through GST gene knockout, we successfully elucidated the resulting changes in petal colour. Similarly, overexpressing GST in white petals led to observed red pigment accumulation. In Arabidopsis, a point mutation in the splicing site affected intron splicing of the TT19 gene in the tt19 mutant, resulting in reduced anthocyanin accumulation in seedling hypocotyls (Sun et al., 2012). Unlike Arabidopsis, mutations in the gene coding region of cotton did not alter gene structure; rather, mutations in promoters affected GST (RPRS1) expression levels across different cotton species, thereby influencing GST function and petal coloration. In maize, a GST gene BZ2 adds a glutathione tag to cyanidin‐3‐glucoside, forming a glutathione conjugate recognized as the transfer intermediate (TI) by the vacuolar glutathione pump (Marrs et al., 1995). Additionally, GhRPRS1 in cotton can transport not only cyanidin‐3‐glucoside but also pelargonidin‐3‐glucoside. However, whether these glucosides are conjugated to glutathione before transport to the vacuole remains unknown in cotton. Unlike Arabidopsis and maize, GhRPRS1 primarily functions in petal coloration, with high expression levels in petals. Variations in cis‐elements are strongly correlated with petal colour across diverse species. However, due to challenges in collecting petal samples from various cotton species, assessing the relationship between promoter sequence variations and RPRS1 expression is currently difficult. Future studies could involve cloning the promoter sequence of RPRS1 from different species to drive GUS reporter or RPRS1 expression in white petal upland cotton as an alternative method to infer RPRS1 expression in diverse species.

Our research first identified that GST plays a crucial role in controlling the formation of red petals and red‐spot petals within the same cotton species. To determine whether our results can be extended across the genus, we verified the function of the GST in different cotton species and found all demonstrated the capacity to accumulate red pigments. Thus, our results illustrate that GST is conserved in regulating petal colour formation.

Anthocyanin is a plant pigment that is crucial to the determination of petal colour (Sunil and Shetty, 2022). The GST is located in the cytoplasm and can participate in the transport of anthocyanins (Figure S24). Some studies have shown that GST can bind to anthocyanin‐3‐O‐glucoside (A3G) to both promote its transport from the endoplasmic reticulum membrane to the tonoplast membrane and participate in its accumulation and maintenance (Sun et al., 2012). GST family members can be divided into four big categories: Phi (Fi), Tau (U), Lambda (L), and Theta. Only Phi members play important roles in anthocyanin accumulation (Jiang et al., 2019), and the GhRPRS1 we identified also belongs to Phi family. Many studies reported that Cyanidin‐3‐O‐glucoside is the main pigment that changes the red colour of petals, which could be transported by GST (Chai et al., 2022). We further determined that red pigment accumulation ability by GST is conserved on petal colour in different cotton species.

Although variants within gene sequences are the direct cause of phenotypic diversity, increasing evidence shows that variants in promoter regions are also an important cause of phenotypic changing (Liu et al., 2022). Researchers have compared the promoter sequences of members of the GST gene family in barley and found that their promoter sequences vary substantially, resulting in differences in expression and resulting plant physiological processes (Rezaei et al., 2013). The core regulatory elements in the promoter are the fundamental means for gene expression regulation. Variation in the element sequence will cause major differences in gene expression (Sloutskin et al., 2021). In our research, GST exhibited a striking similarity in sequence and function, providing additional evidence for it not being responsible for petal colour variations. Furthermore, GST promoters across cotton species were remarkably variable, substantiating their role in functional differentiation. We selected the GST promoter sequence and performed a series of truncation and element activity analyses in ZYH and ZM24. Consistent with our expectations, we observed that the GST promoter sequence exhibits significant variation in activity. Further, we identified a deletion in the promoter associated with white petals that resulted in reduced activity post‐flowering. Therefore, we conclude that variation in the promoter region of GST among various cotton species might influence the expression of genes and cause functional differentiation.

Interestingly, the accumulation of red pigment is not limited to red petals alone. Upland cotton with white petals gradually exhibits red coloration after pollination on the day of flowering. Previously, the cause of this phenomenon had not been reported. In our study, we confirmed this is also caused by differential expression of GST. The mechanisms driving the observed increase in GhRPRS1 expression post‐pollination in both white and red petals remain unclear. Pre‐pollination, the GhRPRS1 showed a higher expression level in red petal compared with white petal. Post‐pollination, the expression level of GhRPRS1 was slightly downregulated between 1st and 2nd stages. Studies suggest that substantial hormones were transported to the petals after pollination (Zúñiga‐Myao et al., 2023), and previous studies have demonstrated that GST expression can be induced by gibberellin or ethylene (Zhang et al., 2022b). Consequently, it is hypothesized that the differential expression of GST in petals following pollination may be influenced by these hormones in response to the changing floral state. Furthermore, significant morphological changes in the flower corolla before and after flowering represent distinct developmental stages. The downregulation of GhRPRS1 in red petals may be attributed to the transition between developmental stages, varying intensities of stressors, and factors, such as light quality and availability (Galle et al., 2018). Notably, the expression of GhRPRS1 in ZYH and ZM24 increased from the 2nd to the 4th stages, which aligns with the presence of an increase in anthocyanin levels during this period. The elevated anthocyanin content may serve protective roles against ultraviolet light or deter herbivores, as suggested in previous studies (Glover and Martin, 2012). Future research should focus on elucidating the temporal regulation of GST transcriptional control. In sum, we proposed a molecular model of red petal formation (Figure 6). Anthocyanin synthesis pathways between ZYH and ZM24 had no significant difference (Figure S12, Dataset S1), and differential accumulation of anthocyanins in the vacuole is responsible for red petals. GhRPRS1 is a key gene that controls red‐petal and red‐spot petal formation as a transport carrier. Our research analysed the molecular underpinnings of coloration of red petals in cotton from multiple perspectives and provided new genetic resources for petal colour‐assisted breeding.

Figure 6.

A work model for GhRPRS1 transport anthocyanin in both red and white petals. GhRPRS1 works as anthocyanin transporter in cotton. The presence of E3 element in promoter of GhRPRS1 from ZYH enhanced the expression of the GhRPRS1. As a result, the red petals have more anthocyanin transporter, which facilitates the transport of anthocyanins from the endoplasmic reticulum membrane to the tonoplast. While in ZM24, the absence of E3 resulted in weakly expressed of GhRPRS1, and only a small part of transporters is available for anthocyanin transporting. Since different amount of anthocyanin were deposited in vacuoles between ZHY and ZM24, they exhibited red and white colours in petals.

Author contributions

ZEY and FGL designed and managed this research; WH, YLC, BBG, LQL, DY, ZZX and MYL performed the experiments and/or analysed data; YHZ, QDY, CXG, and RJL analysed data; LY and MM performed data processing and bioinformatics analysis; YLC, WQQ, PW, and PFM performed the field trait investigation; WH wrote the manuscript and ZEY revised the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Figure S1 Comparison of petal colours between ZYH and ZM24.

Figure S2 Kompetitive Allele‐Specific PCR (KASP) marker development and polymorphism identification.

Figure S3 Fine mapping of red petals with red spot by Indel markers.

Figure S4 Comparison of SNP densities between ZM24 and ZYH.

Figure S5 Validation the introgression fragment by reads mapping.

Figure S6 PCR production of 11 SSR markers amplification using three different templates.

Figure S7 Transcriptome and metabolomic analysis.

Figure S8 Gene silencing experiment of MYB113‐835 in ZYH.

Figure S9 Protein sequence alignments of GhRPRS1 between ZYH and ZM24.

Figure S10 Anthocyanin accumulated in different tissues and organs by overexpression of GhRPRS1 in ZM24.

Figure S11 Targeted metabolomic analysis of anthocyanin in two developmental stages of petal.

Figure S12 Transcriptome and metabolome identified the key genes involved in anthocyanin biosynthesis and transporting.

Figure S13 Pearson correlation analysis between the expression of GhRPRS1 and contents of 13 differential metabolites.

Figure S14 Expression patterns of GhRPRS1 in different tissues.

Figure S15 Specific staining of anthocyanins in different cotton tissues.

Figure S16 Measurement anthocyanin content in petals post‐pollination.

Figure S17 Protein sequence alignments of GhRPRS1 retrieved from 21 cotton species.

Figure S18 Maximum likelihood (ML) phylogenetic tree of GhRPRS1 in cotton genus.

Figure S19 Promoter sequence alignments of GhRPRS1 cloned from ZYH and ZM24.

Figure S20 The phylogenetic relationship of GhRPRS1 promoter sequences in different cotton species.

Figure S21 Staining results of PBI121 empty vector without the 35S::GUS Sequence.

Figure S22 Measuring fluorescence intensities of different GhRPRS1 promoter element.

Figure S23 Functional validation of RPRS1 in DC109.

Figure S24 Subcellular localization of GhRPRS1 protein from ZYH.

Table S1 The geographic distribution of cotton species and their petal colours.

Table S2 The genetic patterns and statistics of red and white petals in the segregating population.

Table S3 The primers used in our study.

Table S4 The recombinant individual statistics of genetic markers in the population through segregation.

Table S5 The genomic sequencing data of different cotton varieties.

Dataset S1 The distribution of differentially expressed genes during the various sampling intervals.

Data availability

All data included in this study are available as Supporting Information. The raw re‐sequencing and transcriptome data have been deposited in the NCBI under BioProject PRJNA1041941.