A glutathione S‐transferase GhTT19 determines flower petal pigmentation via regulating anthocyanin accumulation in cotton

Summary

Anthocyanin accumulations in the flowers can improve seed production of hybrid lines, and produce higher commodity value in cotton fibre. However, the genetic mechanism underlying the anthocyanin pigmentation in cotton petals is poorly understood. Here, we showed that the red petal phenotype was introgressed from Gossypium bickii through recombination with the segment containing the R ₃ ^bic region in the A07 chromosome of Gossypium hirsutum variety LR compared with the near‐isogenic line of LW with white flower petals. The cyanidin‐3‐O‐glucoside (Cy3G) was the major anthocyanin in red petals of cotton. A GhTT19 encoding a TT19‐like GST was mapped to the R ₃ ^bic site associated with red petals via map‐based cloning, but GhTT19 homologue gene from the D genome was not expressed in G. hirsutum. Intriguingly, allelic variations in the promoters between GhTT19 ^LW and GhTT19 ^LR , rather than genic regions, were found as genetic causal of petal colour variations. GhTT19‐GFP was found localized in both the endoplasmic reticulum and tonoplast for facilitating anthocyanin transport. An additional MYB binding element found only in the promoter of GhTT19 ^LR , but not in that of GhTT19 ^LW , enhanced its transactivation by the MYB activator GhPAP1. The transgenic analysis confirmed the function of GhTT19 in regulating the red flower phenotype in cotton. The essential light signalling component GhHY5 bonded to and activated the promoter of GhPAP1, and the GhHY5‐GhPAP1 module together regulated GhTT19 expression to mediate the light‐activation of petal anthocyanin pigmentation in cotton. This study provides new insights into the molecular mechanisms for anthocyanin accumulation and may lay a foundation for faster genetic improvement of cotton.

R ₃ ^bic locus encoded a GST‐like GhTT19 facilitating anthocyanin transport. R ₃ ^bic site was introgressed from G. bickii by a segment replacement. Deletion of a GhPAP1‐binding element in proGhTT19 ^LW results in white petals. GhHY5‐GhPAP1 modulates GhTT19‐mediated petal red pigmentation.

Introduction

Cotton has obvious heterosis, its coloured flowers not only provide convenient selection markers for effectively distinguishing the hybrids but also as a visual signal to attract pollinators. There has been a report on the purple spot increasing the visit of honeybees and improving the seed production in cotton hybrid lines (Abid et al., 2022). In addition, colourful cotton fibres are desirable agronomic traits that are targeted in cotton breeding (Hinchliffe et al., 2016; Ke et al., 2022; Yan et al., 2018). The pigments of most coloured petals and fibres of cotton are derived from anthocyanins, a major form of flavonoids (Tan et al., 2013). Thus, revealing the mechanism of anthocyanin biosynthesis and regulation is very important for cotton breeding.

It is reported that anthocyanins are synthesized on the cytosolic side of the endoplasmic reticulum (ER) by a multi‐enzyme complex (Grotewold, 2006; Winkel‐Shirley, 2001). After synthesis, they are transported into the vacuole for storage: the vesicle trafficking‐, membrane transporter‐ and glutathione S‐transferase (GST)‐mediated models are three distinct, but interrelated mechanisms for vacuolar sequestration of anthocyanins (Zhao, 2015). GST‐mediated transports and storages of anthocyanins, which is suggested to be coupled with vesicle trafficking‐ and ATP binding cassette (ABC) or multidrug and toxic extrusion (MATE) transporters for vacuolar sequestration of anthocyanins in Arabidopsis (Francisco et al., 2013; Gomez et al., 2009, 2011). Anthocyanin‐producing plants were found to have anthocyanin‐specific GST proteins, such as maize BZ2, Arabidopsis TT19 and grapevine VvGST1 (Marrs et al., 1995; Perez‐Diaz et al., 2016; Sun et al., 2012). The TT19‐like‐GSTs act as a carrier to transport anthocyanins from the endoplasmic reticulum to the vacuole. For instance, BZ2 is capable of conjugating cyanidin‐3‐O‐glucoside (Cy3G) with glutathione (Marrs et al., 1995), Arabidopsis TT19 and Grapevine VvGST1 bond to anthocyanins (Conn et al., 2008; Sun et al., 2012). Recent reverse genetic and molecular studies indicated that cotton GST was an essential gene for anthocyanin accumulation in the red leaf of upland cotton (Li et al., 2020; Shao et al., 2021). Also, Gar07G08900, a GST homologue was responsible for the red‐spot‐petals in Gossypium arboreum (Zhang et al., 2022). However, no forward genetic and genomic study was reported for more details about anthocyanin pigmentation in cotton petals. Light is the most important environmental factor that regulates anthocyanin synthesis and accumulation, by modifying the core MYB‐bHLH‐WD40 (MBW) transcriptional complex (Xu et al., 2014). The anthocyanin‐specific MYB transcription factors such as Arabidopsis PAP1 and its orthologs from other plants, such as tomato and apple, play a key role in the regulation of anthocyanin biosynthesis and accumulation (Ballester et al., 2010; Espley et al., 2007; Xu et al., 2014). Moreover, HY5, the light signalling component, could bind to the promoter of PAP1 and enhance its activity, mediating the light‐regulated anthocyanin production (An et al., 2017; Gangappa and Botto, 2016; Shin et al., 2013), indicating that anthocyanin accumulation in plants is widely regulated at transcriptional levels (Shao et al., 2022; Xu et al., 2015).

Five anthocyanin‐related loci R1, R2, R3, Re and Lc1 have been reported in cotton. R1 (Li et al., 2019) and Re (Wang et al., 2022) are related to red foliar organs and stem, R2 to red petal spots (Abid et al., 2022) and R3 to red petal and Lc1 to brown fibre (Hinchliffe et al., 2016). Most loci, except for R3, have been identified by map‐based cloning, and encode R2R3‐MYB transcription factors. In this work, using the genetic population between LR with the red flower and LW with the white flower, we mapped the R3 locus. By comparison of genome sequences, we confirmed it originated from the Gossypium bickii R₃ ^bic for red flower and encoded a GST protein. Biochemical, genetic, molecular and physiological studies confirmed the function of GhTT19 in anthocyanin transport and the allelic variations in promoters of GhTT19 ^LR and GhTT19 ^LW caused their differential transactivation by anthocyanin activator GhPAP1. Furthermore, the GhHY5‐GhPAP1 module regulation of GhTT19 expression could explain the redness variations of cotton flower petals under different light conditions in the field.

Results

Cy3G is the major anthocyanin in red petals of LR

Different flower colours existed in the three types of cotton (Figure S1). The red petals with a large basal spot in G. hirsutum LR variety were originally transferred from G. bickii (Liang et al., 1997). LR is the near‐isogenic line of G. hirsutum L. acc. LW with a pale white flower (Figure 1a,b). Liquid chromatography–mass spectrometry (LC–MS)/MS analysis of the water‐soluble anthocyanins extracted from LR's red flowers showed that one major peak was abundantly detected in LR extracts, but not detectable in extracts of LW flowers (Figure 1c). The fragment ions at m/z 287 and m/z 449 of the mother allowed the identification of Cy3G as the corresponding anthocyanin (Figure 1d). Quantification of anthocyanins confirmed the significantly higher Cy3G in petals of LR than LW (Figure 1e).

Figure 1.

Phenotypic differences of LR and LW. (a) Three‐month‐old LR and LW plants are grown in the field. (b) The flower phenotype at 2 days, 1 day before or at 0 day post‐anthesis (DPA). (c) HPLC chromatograms of anthocyanin in LR and LW lines. (d) Identification of anthocyanin molecule in red petals of LR flowers. An anthocyanin peak, corresponding to cyanidin‐3‐O‐glucoside (Cy3G) was identified from the red petals of LR using ESI–LC–MS/MS. (e) Quantification of total anthocyanins in LR and LW flower petals. Total anthocyanins were extracted with an acidic methanol solution and measured at 530 and 657 nm. Data are presented as means ± SE from at least three biological replicates. **P < 0.01, *P < 0.05 at Student's t‐test.

Map‐based cloning of the red flower R ₃ ^bic locus

To determine the genetic locus of anthocyanin pigments in cotton petals, we constructed a mapping population by crossing LR with LW plants. All F₁ plants of 76 offsprings exhibited red flower phenotypes of LR, but their F₂ generations were segregated into 252 red flower individuals of LR and 100 white flower individuals of LW, nearly fitting a 3 : 1 segregation ratio (χ ² = 2.202 < ${\chi }_{0.05,1}^2$ = 3.84). This result suggested that the phenotype of LR's red flower was controlled by a single nuclear dominant gene. Linkage analysis with mapping population including 352 F₂ plants revealed that the R₃ ^bic region, a chromosomal segment of a 33.2‐kb flanked between markers of A07‐0594 and A07‐0592 on Chr.A07, was responsible for the red petal phenotype (Figure 2a). To define which of four candidate genes determine the red petal phenotype, we cloned and analysed all four genes and analysed their expression patterns (Table S3, Figure 2b,c), and finally excluded the other three as unrelated ones to flower petal colour. Quantification of the transcript levels of candidate genes in petals with qRT‐PCR showed that the expression of ORF1 could not be detected in LW, but a considerable expressing level was detected in LR (Figure 2b,c,d). Therefore, ORF1 was considered the candidate gene for R ₃ ^bic .

Figure 2.

Map‐based cloning of the red petal gene R ₃ ^bic . (a) Physical map of the R ₃ ^bic locus. Four genes (ORF1‐4) in the region are indicated by green and grey thick arrows. (b) qRT‐PCR analysis of 4 candidate genes in the petals of LR and LW. ORF1 was strongly expressed in the red petals of LR but was almost not expressed in the white petals of LW. (c) FPKM expression values of candidate genes in the petals of LR and LW. (d) FPKM expression values of ORF1 (GH_A07G0814) and homologous gene of D subgenome (GH_D07G0816) in the petals of LR and LW. (e) Alignment of the ORF sequences of GhTT19 from LW and LR. A base combination of two positions could distinguish GhTT19 ^LR ‐At, GhTT19 ^LW ‐At and GhTT19‐Dt using pyrosequencing. (f) The number of base combinations of two positions in 30 positive clones. The genomic DNA of LR and LW were used for cloning. (g) Pyrophosphate sequencing using cDNA of LR petal. Data are presented as means ± SE from at least three biological replicates. **P < 0.01, *P < 0.05 at Student's t‐test.

Based on a high‐quality reference genome sequence of G. hirsutum L. acc. TM‐1 (http://cotton.zju.edu.cn/), this region contains four open reading frames (ORFs) that all have corresponding transcripts in LR (Tables S2 and S3). To further track the origin of the R ₃ ^bic site, we compared the genome sequences of G. hirsutum (both LR and LW) and G. bickii (Sheng et al., 2022) and found that this R ₃ ^bic site located in a large chromosomal segment of about 7 Mb in G. hirsutum‐LR (8–15 Mb segment of A07) that was introgressed from G. bickii (9.94–19.03 Mb segment on Chr07) by recombinant replacement of A07 regions (Figure S4). This region contains 274 genes (Table S4, Figure S5), including four genes in the R ₃ ^bic site. Among these genes, 83 genes showed more than twice the significant differential expression between petals of LR and LW: 10 up‐regulated genes in LR and 73 genes in LW. ORF1(GH_A07G0814) has the highest FPKM in LR and the highest fold‐change between the petals of LR and LW (Figure S5a). Except for ORF1, no other genes were reported to be related to anthocyanin metabolism (Figure S5b).

The ORF1 (GH_A07G0814) and their homologues in the D subgenome were amplified from LR and LW (Figure S2a). It encodes a putative GST protein of 214 amino acids highly homology to Arabidopsis TT19 and thus was named GhTT19. There were eight single nucleotide polymorphism (SNPs) in GhTT19‐At coding regions in LR and LW varieties (here named as GhTT19 ^LR ‐At and GhTT19 ^LW ‐At, respectively), resulting in three amino acid changes (Figure S2a). There were no differences between the sequences of GhTT19‐Dt. However, GhTT19 ^LR ‐At and GhTT19 ^bickii have identical ORF sequences (Figure S2a) as well as promoter sequences (Figure S3). Two SNPs in GhTT19 coding regions could distinguish GhTT19 ^LR ‐At, GhTT19 ^LW ‐At and GhTT19‐Dt using pyrosequencing (Figure 2e). A‐A combination represents GhTT19 ^LR ‐At, A‐G represents GhTT19 ^LW ‐At and G‐G represents GhTT19‐Dt. We cloned the sequence containing two loci from genomic DNA. Thirty positive clones from LR and LW were sequenced respectively. Only A‐A and G‐G combination exist in LR, A‐G and G‐G combination exist in LW (Figure 2f). Pyrosequencing of LR cDNA showed that although GhTT19 ^LR ‐Dt exists in LR/LW, it is not expressed (Figure 2g).

We developed a pair of SNP detection primers, based on point mutation at the 288^th position of the GhTT19 coding region that could specifically detect the 288^th‐A (LR) but not the 288^th‐G (LW) genotype (Figure S2b). An enlarged F₂ population of 1191 individuals, including 868 red‐flower individuals and 323 white‐flower individuals, was used for screening with the pair of SNP primers. All 868 red‐flower individuals were successfully amplified with the expected bands, while the 323 white‐flower individuals failed (Figure S2c). Thus, GhTT19 was the candidate gene attributed to the red flower of LR.

GhTT19 was required for anthocyanin accumulation

Among 12 GST genes in cotton plants belonging to the GSTF subgroup (Phi), GhTT19 from LR and LW shared the highest homology to AtTT19 in protein sequence, exhibiting 59% identity (Figure S6a). The GhTT19 ^LR over‐expressing lines of tt19 mutant, which had green hypocotyls similar to the wild‐type, showed purple hypocotyls (Figure S6b). When GhTT19 was suppressed in LR using the pCLCrV‐based virus‐induced gene silencing (VIGS) assay, the down‐regulation of GhTT19 in LR changed the petal colour from red into light pink (Figure 3a–e). We introduced GhTT19 ^LR into white cotton cultivar W0 via Agrobacterium‐mediated transformation. Overexpression of GhTT19 ^LR in the T₀ plant causes the new leaf, stem and flower to turn red (Figure 3f–i, Figure S9). The GhTT19‐dependent anthocyanin accumulation was further supported by qRT‐PCR analyses of GhTT19 expression in the F₂ segregated population developed from a cross between LW as the female parent and LR as the male parent. The F₂ population consisted of 39 white flower individuals (F₂, 1–39) and 97 red flower individuals (F₂, 40–136), of them, all the red flower individuals had highly expressed GhTT19, but all the white flower individuals had no detectable expressions of GhTT19 (Figure S10). In sum, GhTT19 was the important gene for anthocyanin accumulation in cotton.

Figure 3.

GhTT19 gene is responsible for anthocyanin accumulation in cotton petals. (a) The phenotypes of LR petal after GhTT19 were silenced by VIGS. The image below is an enlargement of the contents of the box. (b) Quantification of total anthocyanins in pCLCrVa:00 (CK) and GhTT19 silenced LR flower petals. Total anthocyanins were extracted with an acidic methanol solution and measured at 530 and 657 nm. (c) qRT‐PCR analysis of GhTT19 in pCLCrVa:00 and GhTT19 silenced LR flower petals. (d) The phenotype of LR after GhTT19 was silenced by VIGS. Part of the petal turns white. (e) qRT‐PCR analysis of GhTT19 in the red and white spot of the petal. The GhTT19 expression was strongly decreased in the white spot than in the red spot. (f) Overexpression of GhTT19 ^LR in W0 (white petal cotton) causes the flower (−1 and 0 day) to turn red. (g) PCR detection of GhTT19 ^LR transgenic cotton. (h) Expression of GhTT19 in OE1, OE2 and the wild‐type W0. (i) Quantification of total anthocyanins in OE1, OE2 and W0 flower petals. Total anthocyanins were extracted with an acidic methanol solution and measured at 530 and 657 nm.

GhTT19 bond to Cy3G and localized on ER and vacuole

To explain why this GST protein is critically required for anthocyanin accumulation in cotton, we examined whether GhTT19 can function as an anthocyanin‐binding protein carrier in the anthocyanin transport process. The bacterially expressed and purified GhTT19^LR recombinant proteins were examined by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblotting (Figure 4a,b). The affinities of Cy3G to respective GhTT19^LR and GhTT19^LW were measured by Biacore 8000, showing that both proteins could bind to Cy3G in vitro (Figure 4c). The values of affinities of the proteins GhTT19^LR and GhTT19^LW to Cy3G were 7.01 × 10⁻⁵ and 3.45 × 10⁻⁵, respectively, based on the steady‐state affinity model. Both proteins could bind to Cy3G in vitro and their affinities to Cy3G were comparative. The subcellular localization assay showed that GhTT19‐GFP was mostly localized on the membranes of ER and the vacuole, in line with the Arabidopsis TT19 (Sun et al., 2012), implying the GhTT19 functioned in anthocyanin transport. The results showed that both GhTT19^LR and GhTT19^LW could bind to Cy3G and localized on both ER and vacuole.

Figure 4.

GhTT19 binds to Cy3G in vitro. Both GhTT19^LR and GhTT19^LW proteins are able to bind Cy3G, and have the same subcellular localization. (a) Expression and purification of GhTT19^LR and GhTT19^LW proteins in E. coli. Arrow indicates the IPTG‐induced and purified target proteins, which were resolved in a standard SDS–PAGE. (b) Detection of the purified GhTT19 proteins by immunoblotting. (c) The binding affinity of Cy3G to GhTT19^LR and GhTT19^LW. The binding affinity was assessed using a Biacore 8000 biosensor system. (d) Subcellular localization of GhTT19^LR and GhTT19^LW. GhTT19 fusion proteins colocalize with ER, tonoplast marker‐mCherry fusion and show yellow signals after merging. GhTT19 protein localize in ER and vacuolar membrane. Bars = 50 mm.

An MYB binding element enhanced GhTT19 ^LR transcriptional activity via interaction with GhPAP1

To detect the cause of the differential expression levels of GhTT19 in varieties of LR and LW, we detected the tissue‐specific expression variations of GhTT19 in both plants. The transcriptional level of GhTT19 in petals and filaments was 240‐ and 12‐fold higher in LR than in LW, respectively. GhTT19 is expressed in roots, stems and leaves at comparatively lower levels both in LR and LW (Figure 5a). We further checked the promoter activities of GhTT19 ^LR and GhTT19 ^LW . The 1694‐bp and 1652‐bp promoter regions upstream of the GhTT19‐At gene were isolated using LR and LW genomic DNAs, respectively (Figure S11). Two promoters were subcloned into constructs to drive a GUS expression in tobacco leaves. The GUS expression was significantly higher in proGhTT19 ^LR‐transformant than in proGhTT19 ^LW‐transformant (Figure 5b,c). Hence, it is clear that proGhTT19 ^LR had stronger activity than proGhTT19 ^LW . To probe the cause of the high promoter activity of LR, we compared the cis‐acting elements in promoter sequences of both LR and LW through NewPLACE (www.dna.affrc.go.jp/PLACE/?action=newplace ). The GhTT19 promoters of both LR and LW had MYB binding element CAACTG. However, two AC elements (ACC[A/T]A[A/C][T/C] and ACC[A/T][A/C/T][A/C/T]), and one EEC element (GAAATTC) of MYB existed in the promoter of GhTT19 ^LR but absented in that of the GhTT19 ^LW (Figure S11). The MYB transcription factor GhPAP1 is a homologous gene of Arabidopsis PAP1 (Figure S12), which has been reported as the major controlling gene of the red plant phenotype in cotton (Li et al., 2019). It was shown that GhPAP1 had the same gene expression patterns as GhTT19. GhPAP1 was almost not expressed in roots, stems and leaves, but had higher expressing levels in the petals and filament both in LR and LW (Figure 5d). Y1H assays showed that both promoters of GhTT19 ^LR and GhTT19 ^LW could bind to GhPAP1, but the former had a stronger combination than the latter (Figure 5e). Besides, GhHY5 could bind to proGhTT19 ^LR , but not to proGhTT19 ^LW (Figure 5e). An electrophoretic mobility shift (EMSA) assay was employed to evaluate the effects of sequence mutations on GhTT19 promoter activity. The results showed that GhPAP1 could bind to the elements of CAACTG and EEC (GAAATTC), but not to the two AC elements (Figure 5f). We analysed promoter sequences of GhTT19 of different cotton species with diverse genetic backgrounds that had different flower colours. All of them have the CAACTG element, but only the cotton species with red flowers have the EEC (GAAATTC) element (Figure S13). Transactivation assays via proGhTT19 ^LR and proGhTT19 ^LW driven LUC reporter system further showed that GhPAP1 could activate proGhTT19 ^LR but not proGhTT19 ^LW (Figure 5g). The overall conclusion was that GhPAP1 could strongly interact with the promoter GhTT19 ^LR through GAAATTC and enhanced its transcript activity and anthocyanidin accumulation.

Figure 5.

GhPAP1 bound to the GhTT19 ^LR promoter but not efficiently bond to the GhTT19 ^LW promoter. (a) Expression patterns of GhTT19 in tissues of LR and LW plants. (b) Promoter activity of GhTT19 ^LR and GhTT19 ^LW genes. Vectors were introduced into Agrobacterium strain GV3101 and infiltrated into the Nicotiana benthamiana leaves. The activity of β‐glucuronidase (GUS) was examined using GUS staining. (c) The level of GUS gene expression using qRT‐PCR. (d) Expression of GhPAP1 in different tissues of LR and LW plants. (e) Y1H assay of the binding of GhPAP1 and GhHY5to promoters pGhTT19 ^LR and pGhTT19 ^LW . (f) Electrophoretic mobility shift assay (EMSA) result showing the interaction between GhPAP1 and labelled probes in the promoter of GhTT19 ^LR and GhTT19 ^LW . GhPAP1 could bind to CAACTG element and EEC element. While the EEC element only exists in the promoter of GhTT19 ^LW . (g) Relative luciferase activities were determined in the transient expression assays. Data are presented as means ± SE from at least three biological replicates. **P < 0.01, *P < 0.05 at Student's t‐test.

GhHY5‐GhPAP1 modulated anthocyanin accumulation relying on light intensity

Light intensity critically regulates flower coloration in cotton plants. When LR plants grew in the field under a strong light intensity of 80 000 Lux, the flowers displayed more anthocyanin accumulation. However, when LR grew in an artificial climate growth chamber where light intensity was 3200 Lux, the flowers were much less pigmented (Figure 6a). Correspondingly, the transcriptional levels of GhPAP1 and GhTT19 of LR cultivated under the stronger light intensity were drastically higher than those under weaker light intensity (Figure 6c). We then examined the cotton homologue genes of HY5, which is a positive regulator of light signalling and mediates the light‐regulated anthocyanin production (Lee et al., 2007; Xu, 2020). Among 163 bZIP transcription factors identified from the G. hirsutum genome, GH_D08G2693 was the closest homologue of Arabidopsis AtHY5 and was named GhHY5 (Figure S14a). Gene expression profiling revealed that GhHY5 was the only bZIP gene in cotton, appearing with similar expression patterns to GhTT19 and GhPAP1, displaying the highest expression levels in the petal and filaments (Figure S14b). qRT‐PCR analyses also validated the results (Figure 6b). Accordingly, GhTT19, GhPAP1 and GhHY5 of LR exhibited higher or lower expression levels, respectively, under stronger and weaker light intensity (Figure 6c). Y1H assays showed that GhHY5 protein bonds to the promoter of GhPAP1 (Figure 6d). Transactivation assays with the proGhPAP1::LUC reporter system showed that GhHY5 could activate proGhPAP1 (Figure 6e). EMSA assay, using bacterially expressed and purified GhHY5 recombinant protein, showed that GhHY5 could bind to the 332 bp sequence at the promoter of GhPAP1 (Figure 6f). Thus, GhHY5 was closely associated with GhPAP1‐GhTT19‐mediated anthocyanin accumulation in petals under different light conditions.

Figure 6.

GhHY5 can regulate GhPAP1, resulting in the change of petal colour under different light conditions. (a) The flower phenotypes of LR and LW under different light intensities (in the field or in the climatic chamber) on flowering day. (b) Expression patterns of GhHY5 in tissues of LR and LW plants. (c) Expression of GhTT19, GhPAP1 and GhHY5 in LR plant under different light intensities intensity (in the field or in the climatic chamber) on a flowering day. (d) Y1H assay of the binding of GhHY5 to the promoter of GhPAP1. (e) Relative luciferase activities were determined in the transient expression assays. GhHY5 can activate the promoter of GhPAP1. (f) Electrophoretic mobility shift assay (EMSA) result. Each of the DNA probes, which were generated by PCR, was labelled at one of its 5′‐ends with biotin incorporated during primer synthesis. Prob: 332 bp. For competitive experiments, DNA probes without biotin labeling. Data are presented as means ± SE from at least three biological replicates. **P < 0.01, *P < 0.05 at Student's t‐test. The representative photos were shown in Y1H assays.

Discussion

Colourful flowers, fruits, vegetables and seeds packed with natural plant pigments are able to attractive to pollinators, enhance stress tolerance and represent a visible selection marker for crop breeders (Su et al., 2019; Zhang et al., 2019; Zhu et al., 2017). We created four hybrid cotton varieties with red flowers, such as the national registered cotton variety Lu 05H9, Lu HB biaoza‐1, these varieties were planted and welcomed by cotton farmers. Anthocyanins, including pelargonidin, cyanidin, delphinidin and their derivatives from glycosylation, acylation and methylation, are widely found in higher plants (Grotewold, 2006). We showed here that the red pigments in the petals of LR flowers were mainly Cy3G, one of the most common anthocyanins present in the flower of cotton, which was absent in the LW variety possessing white flowers (Figure 1). There are four R2R3 MYB transcription factors, GhTT2 (Hinchliffe et al., 2016; Yan et al., 2018), GhPAP1 (Li et al., 2019), GhBM (Abid et al., 2022) and Re gene (Wang et al., 2022), which control flavonoid production in flowers and fibres of cotton. It has been reported that the genetic positions of red petals (R ₃ ^bic ) and large basal spots (R ₂ ^bic ) are different but closely linked (Liang et al., 1997). Our study confirmed that overexpression of GhTT19 ^LR in W0 (a white petal cultivar) could caused the petal but without the large basal spot (Figures 3f and S9b). Our genetic analysis revealed that the red petals of LR were controlled by a single dominant gene GhTT19 ^LR . A recent study reported that Gar07G08900, a GST homologue gene, is responsible for the red‐spot petals in G. arboreum (Zhang et al., 2022). An R2R3‐MYB113 could control base spot formation in Gossypium barbadense (Abid et al., 2022). These seem to indicate that polyploid cotton has multiple MYB and GST genes in regulating petal and base spot colour in different cotton varieties.

We mapped the R ₃ ^bic on chromosome A07 in a 33.2‐kb region containing four ORFs (Figure 2a) and found that ORF3 was a glutamine amidotransferase‐like gene. It has been reported that ORF4 is a methyltransferase MT‐A70 gene encoding an mRNA m6A writer protein. Gene functions of ORF1 and ORF2 are related to the transport or synthesis of plant pigments. ORF1 is a homologue of TT19 involved in the transport of anthocyanins from the ER to the vacuole (Sun et al., 2012). ORF2 is a phytoene synthase (PSY), a key enzyme catalysing two C20 geranylgeranyl diphosphate to form phytoene in the carotenoid pathway (Tanaka et al., 2008). However, ORF2 did not express in the petals of LW and LR (Figure 2b,c). In our results, Cy3G was the major anthocyanin molecule in the red petal of LR (Figure 1c,d), ORF1‐GST GhTT19 bond to Cy3G, and GhTT19's expression was positively associated with Cy3G accumulation in LR petals (Figures 2b–d and 5a). Thus, we suggested that GhTT19 of R ₃ ^bic might be the locus for the red pigmentation in cotton flowers. The SNP primers that could distinguish GhTT19 ^LR from GhTT19 ^LW clearly separated petal phenotypes between red and white in the genetic population (Figure S2). In addition, GhTT19 could rescue the anthocyanin accumulation in tt19, a mutant of Arabidopsis (Figure S6b), and its down‐regulation in LR could convert the petal colour from red into light pink (Figures 3a–e and S8). Overexpression of GhTT19 ^LR in W0 resulted in petals from white to pink (Figure 3f–i). In the F₂ independent segregated population, all red flower individuals had high expression levels of GhTT19, while all white flower individuals showed no detected GhTT19 expression (Figure S10). Moreover, gene and promoter cloning, whole genome re‐sequencing showed that the R₃ ^bic originated from G. bickii (Figures S2–S4). In LR, the GhTT19 ^LR ‐At gene and promoter fragment replace the GhTT19 ^LW ‐At. Pyrosequencing of LR cDNA showed that although GhTT19 ^LR ‐Dt exists in LR, it is not expressed (Figure 2e–g). These results demonstrated that GhTT19‐At located in R ₃ ^bic was responsible for the red pigment accumulation in LR flowers.

The biosynthesis, storage and function of anthocyanins usually take place at different subcellular locations (Zhao, 2015; Zhao and Dixon, 2010). TT19‐like GSTs are indispensable for anthocyanin pigment accumulation in many plants, likely by binding to anthocyanins and facilitating their translocation from the ER to the vacuole for storage (Sun et al., 2012; Zhao, 2015). We proved that GhTT19, which is mainly localized on membranes of both ER and vacuolar, could bind to Cy3G (Figure 4) and might facilitate the Cy3G stability and transport into the vacuole. This is consistent with TT19‐like GSTs from other plants. There are eight SNPs including three amino acid changes in GhTT19 from LR and LW (Figure S2), GhTT19^LR and GhTT19^LW proteins did not show differences in Cy3G binding capacities and subcellular localizations (Figure 4). Overexpressing either GhTT19 ^LR or GhTT19 ^LW in tt19 mutant appeared purple hypocotyl (Figure S6b). These results indicate that the different petal pigmentations between LR and LW flowers might be caused by their different expressing levels.

It was revealed that GhTT19 transcription in pigmented petals and filaments of LR was higher than that of LW and other tissues (Figure 5a). Indeed, the analysis showed that proGhTT19 ^LR had much stronger activity than proGhTT19 ^LW (Figure 5b,c). GhPAP1 is a homologous gene of Arabidopsis AtPAP1 (Figure S12) and is also involved in the regulation of anthocyanins in cotton (Li et al., 2019). Although the CAACTG element, a binding site of MYB, presented in both promoters of GhTT19 ^LW and GhTT19 ^LR , another MYB binding element EEC (GAAATTC), was absent in the GhTT19 ^LW promoter (Figure S11). EMSA experiments showed that GhPAP1 could bind to both CAACTG and EEC (GAAATTC) elements (Figure 5f). Respective Y1H and EMSA experiments showed that GhPAP1 could more efficiently bind to proGhTT19 ^LR than proGhTT19 ^LW (Figure 5e,f). CAACTG element was verified to exist in all cotton GST promoters with different petal colours, while the EEC (GAAATTC) element only existed in the GST promoters of red petal cotton species (Figure S13). More importantly, GhPAP1 could activate proGhTT19 ^LR but not proGhTT19 ^LW (Figure 5g). The conclusion was that GhPAP1 could enhance the proGhTT19 ^LR activity by interaction with the EEC (GAAATTC) element.

It is often observed that more anthocyanin accumulated in LR petals when cultivated at stronger than weaker light intensities (Figure 6a). The major light signalling component, HY5, has been reported to bind to the promoters of flavonol synthase and dihydroflavonol 4‐reductase and anthocyanin regulator PAP1 (An et al., 2017; Lee et al., 2007; Shin et al., 2013). GhHY5 had higher expressing levels in the petal and filament than in other organs and tissues (Figure 6b). The transcriptions of GhTT19, GhPAP1 and GhHY5 were all more up‐regulated under stronger light intensities (Figure 6c). The Y1H, EMSA and transactivation assays clearly proved that GhHY5 could bind to the promoters of GhPAP1 and activate GhPAP1 expression (Figure 6d–f). Moreover, an ACGT‐box, ACTCAT elements and core sequence of GT box (Figure S11) were found in the promoter of GhTT19 ^LR but not in that of GhTT19 ^LW , as it has been known that these cis‐elements are typical binding sites by bZIP TFs (Chen et al., 2016; Lee et al., 2007; Satoh et al., 2004). Y1H assay clearly showed that the GhHY5 only binds to proGhTT19 ^LR but not to proGhTT19 ^LW (Figure 5e). The transactivation assay showed that GhHY5 did not activate both proGhTT19 ^LR and proGhTT19 ^LW (Figure 5g). There has been reported that Arabidopsis HY5 can bind to the AtTT19 promoter (Lee et al., 2007) but GhHY5 bind to the GhTT19 promoter requires other partners, such as MYB transcription factors (An et al., 2017). In our study, GhPAP1 strongly bonded to proGhTT19 ^LR and activated its transcription, and the light‐regulated anthocyanin accumulation in the red petals was strongly correlated to the expressions of both GhHY5 and GhPAP1. Thus, we concluded that the differential expressions of GhTT19 ^LR and GhTT19 ^LW were caused by both GhHY5 and GhPAP1. Anthocyanin biosynthetic genes, including GhCHS, GhCHI, GhF3H, GhF3′H, GhDFR, GhANS1 and GhUFGT, and anthocyanin regulatory genes, including GhTT8, GhGL3 and GhWD40, had no significant difference in their expression levels (Figure S15). Anthocyanin transport genes, including GhTT12, GhGFSa, GhGFSb, GhAHA10 and GhABCC1 showed no significant difference in FPKM (Figure S16). Anthocyanin biosynthetic pathway genes were also not enriched in the GO enrichment analysis for the differentially expressed genes in the petals between LR and LW plants (Figure S17).

In summary, our study unveiled that the red‐petal phenotype of cotton cultivar LR containing one of the anthocyanins, Cy3G, was controlled by a single dominant locus, R3 ^bic , in which a candidate GST gene, GhTT19, was detected via map‐based cloning. The genetic experiments through both segregating populations and transgenic lines confirmed that GhTT19 functioned for anthocyanin accumulations in different genotypes of cotton, such as LR with red flowers and LW with white flowers, this was relative to the GhTT19 localizing mainly on both ER and tonoplast. We found that an allelic variation, an MYB binding site EEC element (GAAATTC), exists in proGhTT19 ^LR , but not in proGhTT19 ^LW . The EEC element was able to enhance the interaction of GhTT19 ^LR to GhPAP1 and hence resulted in higher expressions of GhTT19 ^LR than GhTT19 ^LW . This higher expression of GhTT19 ^LR was positively correlated with the Gy3G contents and petal colours in LR flowers. In addition, GhHY5, a major light signal component in cotton, was found to have a special interaction with GhTT19 ^LR and promoted its transcription under stronger light intensity. Taken together, our study verified that the GhHY5‐ and GhPAP1‐regulated expression of GhTT19 ^LR played essential roles in light‐mediated anthocyanin pigmentation in cotton red flower phenotypes (Figure 7). This allele is an excellent target to breed new cultivars of cotton.

Figure 7.

Proposed model for transcriptional regulation of GhTT19 and its role in anthocyanin accumulation. GhPAP1 bind to the CAACTG element and the EEC element (GAAATTC) from the promoter of GhTT19 ^LR and activate its expression. GhTT19 binds to Cy3G and facilitates its transport to the vacuole for storage and coloration. However, in LW plants, the EEC element is missing. GhPAP1 binds weakly to the promoter of GhTT19 ^LW and could not activate GhTT19 ^LW expression. Strong light irradiation (e.g., in the field) mediates the activation of GhHY5, which further induces the expression of the anthocyanin biosynthesis regulator GhPAP1. While, under weak light conditions, for example, in the climatic chamber, the GhHY5 expression level becomes lower than that under strong light, GhPAP1 expression is also reduced, and causes lower expression of GhTT19 ^LR , and thereby reduced red pigmentation in petals.

Experimental procedures

Plant materials

Gossypium hirsutum L. acc. LR (hereafter called LR) with red flowers originated from distant hybridization between G. hirsutum and wild diploid G. bickii. G. hirsutum L. acc. LW (here after called LW) with pale white flowers was the near‐isogenic line of LR. G. hirsutum L. acc. W0, with white petals, was the wild‐type receptor for overexpression GhTT19 ^LR . Three F₂ populations consisting of 352, 1191 and 136 individuals were developed from a cross between LW as the female parent and LR as the male parent. All offspring plants were grown in Linqing Breeding Station and in the greenhouse, the artificial climate chamber of Jinan. For low light intensity treatment, cotton plants were grown in an artificial climate growth chamber where light intensity was 3200 Lux. Genomic DNA was extracted from fresh leaves for genotyping. Wild‐type Arabidopsis Col‐0, tt19 mutant and transgenic Arabidopsis were grown in a 23/21 °C (day/night) chamber. All plant tissues were harvested and immediately frozen in liquid nitrogen, and stored at −80 °C for anthocyanin and RNA analyses.

Anthocyanin analysis

Around 200 mg of frozen petals were homogenized in liquid nitrogen and suspended in acidic methanol (99% methanol and 1% hydrochloric acid) for 1 h. After centrifugation, the absorbance of 200 μL supernatant at 530 and 657 nm was determined with a microplate reader (Omega, S/N 415‐2809, Germany). The relative concentration of anthocyanin was calculated using the equation: Q = [A530 – (0.25 × A657)] × M ⁻¹, where M and Q represent the fresh weight and anthocyanin relative concentration, respectively.

High‐performance LC–tandem electrospray–MS (LC–ESI–ToF–MS/MS) analysis

To identify anthocyanin molecules, ~2.0 g petals were ground in liquid nitrogen into powder and suspended in 20 mL methanol:formic acid (9 : 1) for extraction three times by ultrasonication. The extracts were concentrated to 5 mL and purified with C18 Sep‐pak solid phase extraction (500 mg/8 mL, GRACE, USA), with a mixed solution of methanol:water:formic acid (4 : 5 : 1) as eluent (10 mL). Elution solutions were subjected to Agilent 1260 UPLC‐DAD‐6530 ESI‐QTOF MS (Agilent Technologies, Palo Alto, CA) equipped with Agilent Poroshell 120 SB‐Aq C18 column (4.6 × 100 mm, 2.7 μm). LC–ESI–ToF–MS/MS analyses were carried out in accordance with a method described previously (Zhao et al., 2011). MS data were managed with the MassHunter B0.05.0 Workstation.

Map‐based cloning of R ₃ ^bic gene

To map the gene responsible for the phenotype of red petals in LR, an F₂ mapping population with 352 individuals was developed by a cross between LR and LW in 2015. Based on the genome sequence of TM‐1, we developed simple sequence repeat molecular markers (Wang et al., 2015). The R ₃ ^bic locus was mapped to a 33.2 kb region flanked by markers of A07‐0594 and A07‐0592 on Chromosome A07. The ORF and promoter sequences of the candidate genes were amplified from LR and LW varieties, respectively, and confirmed by sequencing. SNP primer S273 (Table S1), which could specifically amplify the 288th‐A genotype, but not the 288th‐G genotype, was developed based on point mutation and used in genotyping the enlarged F₂ population with 1191 individuals.

Genetic complementation of Arabidopsis tt19 mutant

The ORFs of GhTT19 genes from LR and LW cotton varieties (GhTT19 ^LR and GhTT19 ^LW , respectively) were cloned into the pBI121 vector under the control of a 35 S promoter. The vectors 35 S::GhTT19 ^LR and 35 S::GhTT19 ^LW were transformed into Agrobacterium strain GV3101 for the transformation of Arabidopsistt19 mutant using the floral dip method. Seeds from the T1 transgenic Arabidopsis were screened on 1/2 Murashige and Skoog medium containing 50 μg/mL kanamycin and 4% (w/v) sucrose. T₃ plants were used for phenotyping.

Expression and purification of recombinant GhTT19^LW and GhTT19^LR

The ORFs of GhTT19 ^LW and GhTT19 ^LR were subcloned into a pET‐30a expression vector. The bacterially expressing and purification of GhTT19^LW and GhTT19^LR recombinant proteins were conducted as described previously (Chai et al., 2017). Proteins were analysed using SDS–PAGE and immunoblotting. Horseradish peroxidase‐conjugated anti‐His antibody was used for immunoblotting.

Cy3G‐GhTT19 binding assay using Biacore 8000

Cy3G binding affinity to GhTT19 was assessed using a Biacore 8000 biosensor system (GE Healthcare). A CM5 sensor chip was activated with a 1 : 1 mixture of EDC (0.4 M 1‐ethyl‐3‐3‐dimethylaminopropyl carbodiimide) and NHS (0.1 M N‐hydroxysuccinimide in H₂O). Bacterially expressed and purified GhTT19^LW and GhTT19^LR proteins were diluted in 10 mm sodium acetate (pH 4.5) and were immobilized at a flow rate of 10 μL/min for 7 min. The remaining binding sites were blocked with 1 M ethanolamine hydrochloride (pH 8.5) at a flow rate of 10 μL/min for 7 min. Affinity measurement was performed at 25 °C with HBS‐EP (10 mm HEPES, 150 mM NaCl, 3 mM EDTA, 0.05% Tween 20, pH 7.4) as running buffer. A series of Cy3G concentrations were injected over the ligand surface consecutively as an association phase (contact time 180 s). Running buffer was injected as the dissociation phase (contact time 600 s). The flow rate was 30 μL/min. All data were processed using the Biacore Evaluation software version 1.1.

Promoter analysis

The promoters (about 1.6 kb) of GhTT19 ^LW and GhTT19 ^LR genes were isolated using primers S128F/R (Table S1). The promoter fragments were cloned into vector pBI121 by replacing the 35 S promoter and then sequencing‐confirmed constructs were introduced into Agrobacterium strain GV3101. The remaining steps were performed as described (Wan et al., 2016). For the trans‐activation assay, the ORFs of GhPAP1 and GhHY5 were recombined into the P2GW7 effector expression vector. The promoters of GhTT19 ^LR , GhTT19 ^LW and GhPAP1 were cloned into the pGreen‐0800‐LUC reporter vector. The effector expression and reporter constructs were transformed into Arabidopsis protoplasts, and the transactivation assay was performed as described previously (Zhao et al., 2021).

RNA extraction, quantitative RT‐PCR analysis

Total RNA was extracted from different tissues using the Biospin Plant Total RNA Extraction Kit (Hangzhou Bori Technology, China). First‐strand cDNA was generated using HiScrip Q RT SuperMix for qPCR (+gDNA wiper) (Vazyme, China) according to the manufacturer's instructions. Real‐time PCR was performed using a LightCycler FastStart DNA Master SYBR Green I kit (Roche) in a LightCycler 480 II (Roche) detection system. Histone 3 gene was selected as the internal reference. The relative expression levels were calculated using the cycle threshold (Ct) 2^−ΔΔCt method (Livak and Schmittgen, 2001). All primers are listed in Table S1.

Genome resequencing analysis and transcriptome sequencing

Young leaves of LR and LW were collected for DNA isolation. Genomic DNA was extracted using the CTAB method. The task of genome resequencing was performed on an Illumina HiSeq2000™ by Biomarker Technologies (Beijing, China). The sequences of clean reads were mapped to the TM‐1 genome (http://cotton.zju.edu.cn/). The difference sequences from the A07 chromosome of LR and LW were mapped to G. bickii genome (www.cottongen.org) (Sheng et al., 2022). Red and white petals at full bloom day from three different plants were chosen for transcriptome profiling. Sequencing was performed on the Illumina HiSeq™ system (LC Sciences). The sequences of clean reads were mapped to the TM‐1 genome (http://cotton.zju.edu.cn/). The expression levels of genes were calculated by quantifying the cDNA fragments per kilobase of transcript per million fragments mapped (FPKM).

Yeast‐one‐hybrid (Y1H) assay

The ORFs of GhHY5 and GhPAP1 were cloned into a yeast one‐hybrid vector pGADT7 in the reading frame with AD domain to make constructs AD‐GhHY5 and AD‐GhPAP1 fusions, respectively. Meantime, 1652 bp, 1694 bp and 1600 bp of promoter regions of GhTT19 ^LW , GhTT19 ^LR and GhPAP1 genes were cloned into a yeast vector pAbAi to generate MatchmakerTM Gold Yeast One‐Hybrid assay constructs proGhTT19 ^LW , proGhTT19 ^LR and proGhPAP1 constructs, respectively. Y1H Gold strain was used for the transformation of the above‐mentioned constructs for yeast one‐hybrid assays with URA3 as a selectable marker. The positive transformants were further texted on the YNB medium lacking uracil and leucine and containing 100 ng/mL Aureobasidin A (AbA). The empty vector was used as a control for excluding the self‐activation and false‐positive results.

EMSA assay

The ORF of GhPAP1 and GhHY5 were cloned into the pGS‐21a and pET‐30a respectively. The recombinant protein expression and purification were carried out as described (Chai et al., 2017). All DNA probes were obtained from Tsingke Biotechnology Company (Beijing, China). Images were obtained using a Chemiscope 3300 (Clinx Science Instruments Company).

Subcellular localization of GhTT19

The ORF of GhTT19 ^LR and GhTT19 ^LW was fused with GFP in the pBinGFP4 vector. The vector was transiently expressed in tobacco (Nicotiana benthamiana) leaf cells as described (Chai et al., 2017). An endoplasmic reticulum (ER) marker HDEL protein (pt‐rk‐CD3‐999) and a tonoplast intrinsic protein fused with mCherry (Nelson et al., 2007) were used as ER and tonoplast markers. Subcellular localization of GhTT19‐GFP was observed under a confocal laser scanning microscope (Zeiss LSM 780) with excitation and emission at 488 and 495–530 nm, respectively, for GFP and at 587 and 600–650 nm, respectively, for mCherry.

Virus‐induced gene silencing

A 407‐bp fragment (corresponding to bases 370–776) of GhTT19 was cloned into pCLCrVa and named pCLCrVa: GhTT19. pCLCrVa: GhTT19 and pCLCrVb were introduced into Agrobacterium strain LBA4404. The transformed Agrobacterium were grown overnight at 28 °C and then resuspended in infiltration media (10 mM MES, 10 mm MgCl₂ and 200 μm acetosyringone) and adjusted to OD600 = 1.0. Standing the cells at room temperature for 3 h. The Agrobacterium strain containing pCLCrVa: GhTT19 and pCLCrVb mixed at a ratio of 1 : 1. The LR cotton seedlings were infiltrated with the Agrobacterium cells on cotyledons. The plants were grown in an artificial climate chamber at 26 °C/23 °C (day/night) on a 16/8 h light–dark period. The light intensity was 22 000 Lux.

Cotton transformation

The full length of GhTT19 ^LR was cloned into the pBI121 vector under the control of the 35S promoter. The construct was introduced into wild‐type G. hirsutum L. acc.W0 (white petal) using Agrobacterium tumefaciens‐mediated transformation as described (Li et al., 2009).

Pyrosequencing analysis

The pyrosequencing analysis of the LR cDNA samples were performed according to the manufacturer's recommendations. Bisulphite‐converted cDNA were subjected to PCR amplification. The primers were designed using PyroMark Assay Design 2.0. One of the primers was biotin labelled. The primer sequences are listed in Table S1. KAPA2G Robust HotStart ReadyMix PCR Kit (KAPA KK5701) was used to perform the PCR reaction. To confirm the quality, the PCR products were checked by 1.5% agarose gel electrophoresis. A 25 μL of each PCR product was then subjected to quantitative pyrosequencing analysis using a PyroMark Q48 system (Qiagen) according to manufacturer's instruction.

Statistical analysis

All the experiments were repeated at least three times on three biological replicates. The data were subjected to Student's t‐test to detect statistical differences. *P < 0.05 and **P < 0.01.

Accession numbers

All sequences are available at DDBJ/EMBL/GenBank under the accession numbers: MN444040 (GhTT19‐At of G. hirsutum acc. LW), MN4444041(GhTT19‐Dt of G. hirsutum acc. LW), MN4444042 (GhTT19‐At of G. hirsutum acc. LR) and MN4444043 (GhTT19‐Dt of G. hirsutum acc. LR). The Illumina RNA‐seq and genome resequencing data are available at the Sequence read Archive under accession number PRJNA878950.

Competing interests

The authors declare no conflict of interest.

Author contributions

JSZ and QC designed the research; QC, XW, MG, XZ, YC, CZ, HJ, JW, YW, MZ, AMB and JZ performed research; QC, XW and XZ analysed the data; JSZ, QC and JZ wrote the paper. All authors discussed the results and commented on the manuscript.

Supporting information

Figure S1 Cotton flowers from G. bickii, LW and LR.

Figure S2 Alignment of the CDS sequences of GhTT19.

Figure S3 Alignment of the promoter sequences of GhTT19 from LR, LW and G. bickii.

Figure S4 SNP detection and prediction of variability between LR and LW using genome resequencing.

Figure S5 FPKM expression values and GO enrichment analysis of genes, from G. bickii introgressed segment of A07 regions, in the petals of LR and LW.

Figure S6 GhTT19 ^LR and GhTT19 ^LW can complement the anthocyanin deficiency phenotype of Arabidopsis tt19 mutant.

Figure S7 Amino acid alignments of homologue proteins of AtTT19 from different species.

Figure S8 The phenotype of LR after Sulphur (Su) silenced by VIGS.

Figure S9 The phenotype of GhTT19 overexpression lines.

Figure S10 qRT‐PCR analysis of GhTT19 in the F₂ population.

Figure S11 Alignment of the promoter sequences of GhTT19 ^LR and GhTT19 ^LW .

Figure S12 Identification and characterization of cotton PAP1 genes.

Figure S13 The CAACTG element and EEC element, the PAP1 binding sites, on the GhTT19 promoter of different cotton with diverse genetic backgrounds.

Figure S14 Transcriptional profiling of bZIP genes in different tissues in G. hirsutum acc.TM‐1.

Figure S15 Expression level of genes involved in anthocyanin biosynthesis, and regulation analysed by qRT‐PCR.

Figure S16 Transcript abundance of genes involved in anthocyanin transport analysed by RNA‐seq.

Figure S17 GO enrichment analysis of the differentially expressed genes in petals between LR and LW.

Table S1 All primers developed and used in the present research.

Table S2 Putative functions of candidate genes for red petal phenotype.

Table S3 ORFs and its Dt homologue gene sequences in LR, LW and G. bickii.

Table S4 Differential genes on chromosome A07 between LR and LW and their expression patterns in petals.