The R2R3-type MYB transcription factor MdMYB90-like is responsible for the enhanced skin color of an apple bud sport mutant

The anthocyanin content in apple skin determines its red coloration, as seen in a Fuji apple mutant. Comparative RNA-seq analysis was performed to determine differentially expressed genes at different fruit development stages between the wild-type and the skin color mutant. A novel R2R3-MYB transcription factor, MdMYB90-like, was uncovered as the key regulatory gene for enhanced coloration in the mutant. The expression of MdMYB90-like was 21.3 times higher in the mutant. MdMYB90-like regulates anthocyanin biosynthesis directly through the activation of anthocyanin biosynthesis genes and indirectly through the activation of other transcription factors that activate anthocyanin biosynthesis. MdMYB90-like bound to the promoters of both structural genes (MdCHS and MdUFGT) and other transcription factor genes (MdMYB1 and MdbHLH3) in the yeast one-hybrid system, electrophoretic mobility shift assay, and dual-luciferase assay. Transgenic analysis showed that MdMYB90-like was localized in the nucleus, and its overexpression induced the expression of other anthocyanin-related genes, including MdCHS, MdCHI, MdANS, MdUFGT, MdbHLH3, and MdMYB1. The mutant had reduced levels of DNA methylation in two regions (−1183 to −988 and −2018 to −1778) of the MdMYB90-like gene promoter, which might explain the enhanced expression of the gene and the increased anthocyanin content in the mutant apple skin.


Introduction
A bud sport is a naturally occurring mutation in a tree branch, and these have been widely used by breeders to select desirable characteristics for breeding 1,2 . The phenotypes of bud sport mutants mainly include early development, fruiting spurs, and coloring. In recent years, high-throughput sequencing technology has been widely used in studies to uncover the molecular mechanisms of these natural mutations. In one Fuji apple bud mutant, SNPs (single-nucleotide polymorphisms) and unique InDels (insertions or deletions) were detected by using whole-genome resequencing 3 . In a "Yanfu 6" apple spurtype mutant, microRNAs played important roles in regulating the shoot apical meristem, cell division, and internode length 4 . MYB transcription factors and epigenetic regulation were reported in an anthocyanindeficient yellow-skinned somatic mutant 'Blondee' 5 . In European pear, the methylation level of PcMYB10 is associated with the formation of green-skinned sports 6 . A large number of excellent varieties have been produced by utilizing bud mutants in fruit trees, such as apple, pear, cherry, orange, and grape 4,[7][8][9][10][11] .
Anthocyanins are secondary metabolites that have antioxidant and antitumor functions as well as activity against coronary heart disease and even help to defend against pathogens and ultraviolet radiation [12][13][14] . Apple fruit coloration not only determines fruit appearance and economic characteristics but also has beneficial value for human health. Therefore, it is considered an important trait for apple breeding 15,16 . Both structural genes and regulatory genes in the anthocyanin pathway have been proven to be important for fruit skin color. MdPAL (phenylalanine ammonia lyase), MdCHS (chalcone synthase), MdCHI (chalcone isomerase), MdF3H (flavanone 3-hydroxylase), MdDFR (dihydroflavonol 4-reductase), MdANS (anthocyanidin synthase), and MdUFGT (UDPglucose flavonoid 3-O-glucosyltransferase) have been discovered to have positive correlations with the accumulation of anthocyanin in apple skin [17][18][19][20][21] . Moreover, regulatory genes affect fruit skin color through the regulation of structural genes 22 . Ectopic expression of apple MdMYB1 can activate both DFR and UFGT structural genes involved in anthocyanin biosynthesis in tobacco 23 . MdMYBA can bind specifically to an anthocyanidin synthase (MdANS) promoter region to regulate anthocyanin synthesis in apple skin 24 .
In recent years, multiple omics technologies, such as mRNA sequencing, miRNA sequencing, metabolomic, and proteomic analyses, have been used in the analysis of apple mutants 5,[25][26][27] . A color mutant of a Fuji apple with early coloring and red skin pigmentation was discovered in Jiangsu Province, China 25 . Proteomics investigations identified 451 differentially expressed proteins in the fruit skin of this mutant. The mutant had significantly increased expression of photosynthesis-related proteins, stress-related proteins, and proteins in the anthocyanin biosynthesis pathway, but the expression of mitogenactivated protein kinase 4 (MAPK4) and mevalonate kinase (MVK) was substantially downregulated, indicating posttranscriptional regulation of skin color formation in the mutant. To understand the transcriptional regulation of anthocyanin biosynthesis in this mutant, comparative RNA-seq analysis of a Fuji apple and its color mutant was performed, a key regulatory gene, MdMYB90-like, was uncovered as a novel R2R3-type MYB transcription factor, and the mechanism of its regulation of anthocyanin biosynthesis was discussed.

Results
M_Fuji apple mutant skin color development occurred earlier than that of wild-type Fuji Mature apple fruits from the skin color mutant (M_Fuji) had redder skin than the original yellowishgreen skin with a red flush (Fig. 1A). Skin color development was light-dependent with little anthocyanin when bagged but attained the highest level at 6 days after bag removal (DABR) under continuous light treatment (Fig. 1B). In comparison, the anthocyanin content increased slowly in the wild-type Fuji apples. A significant difference was observed between the mutant and the wild-type apple at 2 DABR under continuous light treatment (Fig. 1A, B).  1D).
Structural genes in anthocyanin biosynthesis are largely regulated at the transcriptional level by the MYB-bHLH-WD40 protein complex 28 . The protein complex is composed of three types of transcription factors: the R2R3-MYB transcription factor, the basic helix-loop-helix (bHLH) transcription factor, and the WD40 protein. Among the differentially expressed MYB transcription factors, MdMYB1 (LOC103444202) and an unknown transcript, Tcons_00045044 were consistently upregulated in the mutant. MdMYB1 was the first identified R2R3-type MYB transcription factor regulating anthocyanin biosynthesis in apple 29 . Transcript Tcons_00045044 showed high homology with the PbMYB90-like of pear and was thus designated MdMYB90-like. In addition, a MdbHLH3 (LOC103449015) transcription factor was upregulated at stage 2, but no WD40 genes were detected among the DEGs.

Isolation and analysis of the MdMYB90-like transcription factor
Transcript (Tcons_00045044) was extremely upregulated during the three-phase periods in the mutant (Fig.  2). When its sequence was BLASTed against the apple genome (Malus×domestica HFTH1 V1.0 a1 transcripts), it best matched the gene HF36881-RA. The HF36881-RA gene showed high homology with PbMYB90-like of pears in the NCBI GenBank database and thus was designated MdMYB90-like. The coding sequence (CDS) of MdMYB90-like was 621 bp and encoded a putative protein of 206 amino acids with an ATG start codon at nucleotide position 1 and a TGA stop codon at position 3297 ( Supplementary Fig. S1A). Both PbMYB90-like and MdMYB90-like had similar gene structures, including three exons and two introns, and the first two exons were 130-bp long. Furthermore, the R2 domain of MdMYB90like consisted of exon 1 and part of exon 2, while the R3 domain was split over exons 2 and 3.

Subcellular localization of the MdMYB90-like protein
To determine the subcellular location of MdMYB90like, the full-length CDSs of MdMYB90-like were inserted into the pC29_35S:GFP5_his6 vector. An empty 35S:GFP vector was used as the negative control, while MdMYB1 was used as the positive control. The constructs were transformed into onion epidermal cells by biolistic transformation and into tobacco leaves by agroinfiltration. GFP fluorescence was observed in the MdMYB1-GFPand MdMYB90-like-GFP-transformed onion and tobacco nuclei, while fluorescent signals were observed in both the nucleus and cytoplasm of the empty 35S:GFP vector ( Supplementary Fig. S2). These results indicated that, similar to MdMYB1, MdMYB90-like proteins were also localized in the nucleus.

Analysis of cis-elements in gene promoters
To further characterize the function of this transcription factor, a 2018-bp region upstream of the translation start site in the MdMYB90-like gene (the putative promoter sequence) was cloned and analyzed through the PlantCARE program. Multiple MYB-binding elements, light-responsive elements (G-boxes), GT1 motifs, and several hormone-responsive elements, such as abscisic acid-responsive elements, auxin-responsive elements, and MeJA-responsive elements, were detected (Supplementary Fig. S3). These results suggested that the expression of the MdMYB90-like gene might be regulated by various factors, such as abscisic acid, jasmonic acid, gibberellin, and light. Mitosis-specific activator (MSA)-like elements were found in the MdMYB90-like gene promoter, which indicated that it might be involved in cell cycle regulation.
In addition, the promoter regions of MdMYB1, MdCHS, MdUFGT, MdANS, and MdbHLH3 were also isolated and analyzed by PlantCARE. Cis-elements, including hormone-responsive elements, light-responsive elements, low-temperature elements, and MYB-binding elements, were found in MdMYB1. MYB-binding elements, lightresponsive elements, and several hormone-responsive elements were found in the promoters of MdCHS, MdUFGT, MdANS, and MdbHLH3, such as abscisic acidresponsive elements and MeJA-responsive elements ( Supplementary Fig. S3).

Functional analysis of MdMYB90-like by overexpression in transgenic materials
To study the biological function of MdMYB90 in the regulation of anthocyanin biosynthesis, an overexpression vector (62SK-MdMYB90-like) was constructed and transformed into apple skins by transient agroinfiltration in both Fuji and Golden Delicious apple. Anthocyanin accumulated when 62SK-MdMYB90-like was transformed and cultured under continuous light for 3-5 days; however, no anthocyanin accumulation was observed when the empty vector 62SK was transformed or when 62SK-MdMYB90-like transformed apple was cultured in the dark, indicating the requirement of light for anthocyanin biosynthesis. Anthocyanin accumulation was faster in Fuji (3 days) than Golden Delicious (5 days) apple ( Fig. 3A, B and Supplementary Fig. S4).
The function of MdMYB90-like was also studied in stably transformed apple calli by the agrobacteriummediated transformation of "Orin" apple calli. Transgenic calli that overexpressed the MdMYB90-like gene under the 35S promoter began to show red spots 2 days after transfer to light conditions (Fig. 3C), but no visible changes were observed in wild-type calli or transgenic calli cultured in the dark. Light-dependent anthocyanin biosynthesis was again confirmed in the transgenic calli. The anthocyanin content was analyzed 5 days after light exposure and was found to be significantly higher in transgenic MdMYB90-like calli than in the WT control ( Fig. 3D).
Gene expression was analyzed in transgenic calli (Supplementary Fig. S4). The expression of the MdMYB90-like gene was more than 80 times higher in transgenic calli than in WT calli. In addition, the overexpression of the MdMYB90-like gene promoted the expression of both structural (MdCHS, MdCHI, MdANS, and MdUFGT) and regulatory genes (MdMYB1 and MdbHLH3) in the anthocyanin biosynthesis pathway (Fig. 3E). In contrast, the transcription level of the unrelated gene MdSIMYB1 was not induced by MdMYB90-like overexpression in apple calli (Fig. 3E). These results suggested that MdMYB90-like might promote apple anthocyanin accumulation directly by activating anthocyanin biosynthesis genes and indirectly by activating other transcription factors.

Transcriptional activity of MdMYB90-like
MdMYB1 activates the expression of downstream anthocyanin biosynthesis genes by interacting with MYB cis-elements in their promoters 23 . MYB-binding elements (MBS, CAACTG) were detected in the promoters of MdCHS, MdUFGT, MdANS and MdbHLH3, MdMYB1, and MdMYB90-like genes. To determine whether MdMYB90-like could interact with these genes, a yeast one-hybrid (Y1H) assay was performed. The results showed that MdMYB90-like could bind to the promoters of CHS and UFGT genes. MdMYB90-like also bound to the MdMYB1 and MdbHLH3 promoters but not to its own promoter. In contrast, MdMYB1 could bind to its own promoter as well as promoters of other genes (CHS, UFGT, and MdBHLH3) but not to the MdMYB90-like promoter (Fig. 4A). These results suggested possible Golden Delicious (B) apple skin after 3 and 5 days of treatment, respectively. Agrobacterium harboring the MdMYB90-like overexpression vector and the 62SK empty vector were infiltrated into apple skins and exposed to light and dark, respectively. C Accumulation of anthocyanin in MdMYB90-like overexpression calli (MdMYB90-like-OE) and wild-type calli (WT) after 5 days of light and dark treatment. WT calli were used as the control. D Anthocyanin contents of transgenic (MdMYB90-like-OE) and wild-type calli (WT). E Expression levels of MdMYB90-like, MdSIMYB1, MdMYB1, MdCHS, MdCHI, MdANS, and MdUFGT in transgenic and wild-type apple calli. Asterisks (*) and (**) denote significant differences between samples at P < 0.05 and P < 0.01 significance levels, respectively regulation of both structural and regulatory genes by MdMYB90-like.
Direct binding of MdMYB90-like protein to MYB-binding elements (MBS) in the MdCHS, MdUFGT, MdMYB1, and MdBHLH3 promoters was revealed by an electrophoretic mobility shift assay (EMSA). MdMYB90-like protein bound to all probes containing the MBS elements from different gene promoters. The binding could be reduced by competitors containing the MBS elements but not reduced by the MBS mutants, indicating that MdMYB90-like could specifically recognize these MBS elements in the promoters ( Supplementary Fig. S5). To further analyze the activation of anthocyanin biosynthesis genes by MdMYB90-like, MdMYB90-like was cotransformed into tobacco leaves with constructs containing the promoters of MdCHS, MdUFGT, MdbHLH3, and MdMYB1 fused to the firefly LUC gene (Fig. 4B). The results showed activation of LUC activity in all cotransformations with MdMYB90-like (Fig. 4C), demonstrating the regulatory activity of MdMYB90-like on these anthocyanin biosynthesis genes.

Mechanism of MdMYB90-like and MdMYB1 gene upregulation in the apple mutant
Sequence variations in the promoter region of an apple mutant have been reported to contribute to the differential expression of regulatory genes and fruit color 33 . To determine whether sequence variations existed in our mutant, MdMYB90-like and MdMYB1 gene sequences from Fuji and its mutant were analyzed. The 621-bp coding region and a 2018-bp region upstream of the MdMYB90-like gene were cloned from both the Fuji apple and the mutant and sequenced. No sequence difference was detected in the gene-coding region, while only two single-nucleotide polymorphisms (SNPs) were found in the promoter region at −1293 bp and −634 bp upstream of the translation start codon. Similarly, no sequence difference was found in either the promoter or CDS of MdMYB1 between the Fuji apple and the mutant.
DNA methylation has been reported as another reason for the regulation of gene expression in apple mutants 34 . To analyze DNA methylation in the promoters and the CDSs of the MdMYB1 and MdMYB90-like genes, McrBC-PCR analysis was conducted. McrBC is an endonuclease that cleaves DNA containing methylcytosine on one or both DNA strands. For the MdMYB1 gene, the promoter and the gene-coding regions of both the mutant and the normal Fuji apple showed low levels of methylation because all fragments were resistant to McrBC digestion, and no visible difference was detected after PCR amplification (Fig. 5A). However, differences were detected in two regions of the MdMYB90-like promoters (−1183 to −988 and −2018 to −1778) (Fig. 5B). DNA methylation levels in these two regions were high because they were sensitive to McrBC digestion. When we compared the normal Fuji apple and the mutant, we found evidence of relatively higher methylation in the normal Fuji apple because the fragments almost completely disappeared after McrBC digestion, while a weak band could be recognized in the mutant (Fig. 5B).
To confirm these results, bisulfite sequencing (BSP)-PCR was performed to detect cytosine methylation in the two regions of the MdMYB90-like promoter in the Fuji apple and the mutant. In the −2018 to −1778 region, 76 cytosine positions, including 52 CHH (68.42%), 14 CG (18.42%), and 10 CHG (13.16%) types of cytosine methylation sites, were detected in a 198-bp DNA fragment (−1997 to −1800) (Fig. 5C, G). High levels of methylation were detected in both the Fuji apple and the mutant; however, the methylation level in the Fuji apple (79.96%) was significantly higher than that in the mutant (70.56%) in the CHH-type cytosines (Fig. 5E, G). Similarly, in the −1183 to −988 region, 33 cytosine positions, including 24 CHH (80%), 6 CG (18.18%), and 3 CHG (9.09%) types of cytosine methylation sites, were detected in a 154-bp DNA fragment (−1162 to −1009) (Fig. 5D,  H). The overall methylation levels were high in both the Fuji apple and the mutant. Reduced methylation was detected in the mutant at three positions: CG at −1124, CHG at −1122, and CHH at −1093 (Fig. 5F, H). The bisulfite sequencing data were in good agreement with those of the McrBC-PCR analysis and confirmed the reduced methylation in the mutant in the two promoter regions.

Candidate genes for skin color mutation
Apple color reflects the anthocyanin content in the apple skin. In red delicious apples, the red mutant showed earlier coloration, and a higher anthocyanin content correlated with higher expression of genes related to anthocyanin biosynthesis 35 . The anthocyanin biosynthesis pathway has been wellstudied in various plants 36 . It begins with 4-coumaroylcoenzyme A (CoA), a metabolic intermediate from the phenylpropanoid pathway. The synthesis of naringenin chalcone from 4-coumaroyl-coenzyme A is the first commitment step for anthocyanin biosynthesis and is catalyzed by chalcone synthase (CHS). In this study, various genes in the anthocyanin biosynthesis pathway were detected as differentially expressed genes between Fuji apples and the mutant. These genes include two PAL genes (LOC103433222, LOC103430265) and one 4CL gene (LOC103426517) in the phenylpropanoid pathway, three CHS genes (LOC103443512, LOC103421794, LOC103443513), one CHI gene (LOC103430446), one DFR gene (LOC103448549), two ANS genes (LOC103437326, LOC103437327), and three UFGT genes (LOC103417897, LOC103420802, LOC103428842). PAL and 4CL work in the phenylpropanoid metabolism pathway to produce 4coumaroyl-coenzyme A from phenylalanine. Studies found that the expression of PAL was positively correlated with the synthesis of anthocyanin in strawberries and apples 37,38 . The increased expression of two PAL genes in this study was also positively correlated with the increased anthocyanin content in the apple mutant (Fig. 2). CHS is the key enzyme in anthocyanin biosynthesis. Three CHS transcripts (LOC103443512, LOC103421794, LOC103443513) were shown to be greatly upregulated in the mutant. Interestingly, these three CHS transcripts were also detected in Granny Smith apples during fruit development, and their activation by 5-aza-20-deoxycytidine (5-aza-dC) treatment enhanced apple coloration 39 . Silencing of CHS in transgenic Royal Gala apple significantly reduced the anthocyanin content 40 . CHI catalyzes the conversion of chalcone to flavanones. We observed the activated expression of a CHI (LOC103430446) in the mutant. The same gene was shown to be activated in Granny Smith apples after 5-aza-20-deoxycytidine treatment, and the anthocyanin content was upregulated 39 . The upregulated ANS (LOC103437326, LOC103437327) plays a role in the oxidation of colorless anthocyanidins to produce colored anthocyanins. UFGT catalyzes the glycosylation of anthocyanidins to anthocyanins and has been found to contribute to cyanidin 3-galactoside biosynthesis in apple skin 41 . Three UFGT genes (LOC103417897, LOC103420802, and LOC103428842) were upregulated in our apple mutant (Fig. 2), and they were also upregulated in Granny Smith apples after 5-aza-dC treatment, which promoted anthocyanin accumulation 39 . UFGT was also reported to be the key gene determining white or red grape phenotypes 42 . The increased expression of proteins in the anthocyanin biosynthesis pathway was also reported in the same mutant by proteomics 25 . The coordinately induced genes in the anthocyanin biosynthesis pathway in the apple mutant might suggest the involvement of transcription factors, which have been reported to directly regulate the expression of structural genes 43 . Indeed, we observed differentially expressed transcripts encoding MYB and bHLH transcription factors. A new apple MYB transcription factor, MdMYB90-like, was characterized in detail and was found to be the key regulatory gene for enhanced anthocyanin biosynthesis in the mutant.

MdMYB90-like is the key regulator in apple anthocyanin biosynthesis
The two-repeat (R2R3) MYB family is the largest family characterized in plants. A large number of these proteins have been isolated and proven to regulate anthocyanin biosynthesis in many plant species 44 . In apples, MdMYB1, MdMYB10, and MdMYBA were identified to be responsible for anthocyanin accumulation by regulating the expression of anthocyanin biosynthesis structural genes. For example, previous studies have shown that MdMYB10 mainly enhances anthocyanin content in apples by upregulating the expression of the DFR gene 47 . MdMYB1 can activate both DFR and UFGT structural genes to regulate anthocyanin biosynthesis, and MdMYBA can bind specifically to an anthocyanidin synthase (MdANS) promoter region to regulate anthocyanin synthesis in apple skin 23,24 .
RNA-seq data analysis detected two obviously upregulated R2R3-type MYB transcription factors in the apple color mutant. MdMYB1 has been shown to be involved in the regulation of anthocyanin biosynthesis in apple skin 29 . It has been reported that different methylation statuses in the MdMYB1 promoter region affect its expression and subsequently regulate anthocyanin biosynthesis in the Ralls apple mutant 34 . However, our results showed no difference in either the nucleotide sequence or methylation level in either the coding region or promoter between Fuji and the mutant. Thus, the increased MdMYB1 expression was more likely to be affected by factors other than the cause of the mutant phenotype.
A novel transcript that encoded a basic R2R3-MYB transcription factor was identified for its consistent upregulation in the mutant at all three stages. Its protein sequence had the highest homology with the pear PbMYB90-like protein and thus was designated MYB transcription factors have been reported to interact with bHLH transcription and form a complex with bHLH to regulate anthocyanin biosynthesis 45 . MdMYB1 has been reported to interact with MdbHLH3 in apples to form a MYB-bHLH complex 32 . MdMYB90like also contained the bHLH interaction motif [D/E] Lx2 [R/K]x3 Lx6 Lx3R in the R3 domain, indicating its potential to interact with the bHLH partner. Compared with MdMYB1, MdMYB90-like had a difference of only three amino acids in the bHLH motif sequence. Our results showed that MdbHLH3 expression could be induced by MdMYB90-like in transgenic apple calli and that MdMYB90-like could bind to the promoter of MdbHLH3 in the YIH assay (Fig. 4A). In addition, MdMYB90-like could activate the expression of MdbHLH3 in the dual-luciferase assay (Fig. 4B, C), indicating that MdMYB90-like could be another partner with MdbHLH3 in apples.
Analysis of the cis-acting element in the promoter regions showed multiple cis-elements in structural genes (MdCHS, MdUFGT, and MdANS) as well as regulatory genes (MdbHLH3, MdMYB1, and MdMYB90-like), including MYB-binding elements (MBS), light-responsive elements (G-box, ACE, GT1motif, and TCCC-motif), hormone-responsive elements (ABRE for abscisic acid response, CGTCA-motif for MeJA response, and GARE motif for gibberellin response), and elements for low-temperature (LTR) and cell cycle (MSA-like) responses ( Supplementary  Fig. S3). This explains the regulation of anthocyanin biosynthesis by various environmental and genetic factors and the light dependence of anthocyanin biosynthesis ( Figs. 1 and 3).
YIH assay, EMSA, and dual-luciferase assay showed that MdMYB90-like could bind and activate the expression of both structural genes (MdCHS and MdUFGT) and regulatory genes (MdbHLH3 and MdMYB1) (Fig. 4), which had MYB-binding elements in their promoters. One exception was the MdANS gene, which showed activation in transgenic apple calli but was not activated in the Y1H assay. Transgenic apple calli that overexpressed MdMYB90-like also activated the expression of these genes (Fig. 3C-E), demonstrating the regulatory role of MdMYB90-like in apples. Overexpression of MdMYB90like either in a transient assay in apple skins or in stably transformed apple calli resulted in the accumulation of anthocyanin under light conditions (Fig. 3).
MdMYB1 also bound to both structural genes (MdCHS and MdUFGT) and regulatory genes (MdbHLH3 and MdMYB1) in the Y1H assay (Fig. 4A). MdMYB1 activated its own expression in Y1H cells, indicating the possibility of self-activation in apples. Although MYB-binding elements were also present in the MdMYB90-like promoter, the Y1H assay showed no activation of MdMYB90-like by either MdMYB1 or itself (Fig. 4A). The interaction and activation of DFR and UFGT genes by MdMYB1 were reported in transgenic tobacco 23 , while MdMYBA was found to bind specifically to the MdANS promoter region for its activation 24 .
Our results indicated that MdMYB90-like activated anthocyanin biosynthesis in the apple mutant by both direct activation of anthocyanin biosynthesis genes (MdCHS and MdUFGT) and indirect activation of these genes through other transcription factors (MdMYB1 and MdbHLH3). This might explain the activation of MdANS in transgenic apple calli. Although MdMYB90like could not activate MdANS directly, it could activate MdMYB1. An identical protein of MdMYB1, MdMYBA, was reported to directly interact with the MdANS promoter and activate its transcription 24 . As an important regulator of anthocyanin biosynthesis in apple, the expression of MdMYB1 is affected by many environmental factors and plant hormones 23,29,[48][49][50] and regulated by other genes [48][49][50] . For example, MdEIL1 was found to directly bind to the promoter of MdMYB1 and transcriptionally activate its expression during ethyleneregulated fruit ripening and anthocyanin accumulation 50 . The regulatory role of MdEIL1 is similar to that of MdMYB90-like. Two other genes (MdTCP46 and MdBT2) were reported to regulate MdMYB1 expression in light-induced anthocyanin biosynthesis 49 . MdTCP46 binds to the MdMYB1 protein and promotes its transcriptional activity, while MdBT2 ubiquitinates and degrades the MdTCP46 and MdMYB1 proteins under low-light conditions.
Methylation of the MdMYB90-like promoter may be the reason for different skin pigmentation patterns DNA methylation or demethylation in gene promoters affects gene expression, and the expression of MYB transcription factors can have an impact on anthocyanin accumulation in apple skin 34,51 . In this study, neither significant sequence variation nor changes in DNA methylation levels were detected in MdMYB1 between the Fuji apple and the mutant. However, significant changes in DNA methylation were found in two regions of the MdMYB90-like promoters. Unlike the other regions of the MdMYB90-like promoter that had low levels of methylation (Fig. 5), two regions (−1183 to −988 bp and −2018 to −1778 bp) were hypermethylated, but differences in methylation levels were recognized between the Fuji apple and the mutant. The mutant showed significantly lower levels of methylation in the two regions (Fig. 5). The lower levels of methylation may explain the increased expression of the MdMYB90-like gene, the upregulation of other anthocyanin biosynthesis genes and their regulators, and the enhanced fruit color. Similar results were reported in Gala apples, in which the methylation levels in the MdMYB10 promoter were negatively correlated with anthocyanin contents in the yellow-skinned somatic mutant Blondee and its red-skinned parent Kidd's D-8 5 .

Regulatory network of anthocyanin biosynthesis in the apple mutant
The anthocyanin biosynthesis pathway is well known, and the key regulatory genes controlling the pathway have been studied in many plants. Anthocyanin biosynthesis is regulated by both developmental and environmental factors through specific activation or repression of MYB transcription factors. MYB-bHLH-WD40 regulatory complexes are thought to activate specific parts of the pathway by different MYB transcription factors 43 . To date, many MYB transcription factors have been identified. Here, we report the identification of a new MYB gene, MdMYB90-like, from an apple mutant. In this mutant, demethylation in two regions of its promoter correlated with increased expression and was probably the cause for its upregulation. MdMYB90-like could bind to cis-elements in other regulatory genes, such as MdMYB1 and MdbHLH3, as well as structural genes in the pathway. Activated MdMYB1 could activate its own expression, as well as structural genes in the pathway, and thus promote the biosynthesis of anthocyanins (Fig. 6). MdMYB90-like played a key role in the regulation of anthocyanin biosynthesis in two possible ways: direct activation of anthocyanin biosynthesis genes (MdCHS and MdUFGT) and indirect activation of these genes through other transcription factors (MdMYB1 and MdbHLH3).

Plant materials
The Fuji apple (Malus domestica Brokh cv. Fuji) and its bud mutant (M_Fuji) were previously observed on a branch at the experimental orchard of Nanjing Agricultural University in Shilaojia County, Jiangsu Province, China 25 . The bud mutant was clonally propagated by grafting, and the mutant phenotype was stably inherited. Flowers on both the wild-type and the mutant were pollinated on April 25, 2016. Young fruits 30 days after pollination (30 DAP) were bagged with two-layer paper bags (Hong Tai, Xi'an, China), which had an inner layer made of red paper with a wax coating and a brown bag as the outside layer. Bags were removed 30 days before harvesting (150 DAP), and samples were collected at five time points: 0, 4, 8, 12, and 16 days after bag removal (DABR) in 2016 for anthocyanin content analysis. Samples of fruits from Fuji (O1, O2, O3) and it's mutant (M1, M2, M3) were collected at 4, 8, and 12 DABR for RNA extraction and RNA-seq analyses. Nine fruits from each sample were randomly divided into three groups and analyzed as three biological replicates. To speed up anthocyanin accumulation, apples and calli were exposed to continuous light.

Anthocyanin analysis
Anthocyanin analysis was conducted according to methods reported by Pirie and Mullins 52 and Xu et al. 34 . Samples (0.2 g) were added to 10 mL precooled hydrochloric acid/methanol (1/99, v/v) solution and extracted in the dark at room temperature for 2 h. All samples were measured in triplicate, and the absorbance at 530 nm and 600 nm was determined by a spectrophotometer. The relative anthocyanin content (Q) was calculated as Q = OD530-OD600, and Q = 0.01 was defined as one unit of anthocyanin content for convenience.
DNA extraction, RNA extraction, library construction, and RNA-seq Total RNA was isolated from each sample by using a Mini BEST Plant RNA Extraction kit (Takara Biomedical Technology-Beijing Co., Ltd, China). mRNA was enriched by oligo(dT) beads, and then the enriched mRNA was fragmented randomly into short fragments and reverse transcribed into cDNA with random primers. Secondstrand cDNA was synthesized by DNA polymerase I, RNase H, dNTPs, and buffer. Then, the cDNA fragments were purified with a QiaQuick PCR extraction kit, endrepaired, poly(A)-added, and ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR-amplified and sequenced using Illumina HiSeqTM 2500 by Gene Denovo Biotechnology Co. (Guangzhou, China).
Genomic DNA (gDNA) was extracted from apple samples by using a Mini BEST Plant DNA Extraction kit (TaKaRa).
Mapping of reads to the reference genome, gene annotation, and gene expression analysis By base calling, the original image data produced by the sequencer were transferred into sequences, which were defined as "raw reads". Clean reads were obtained after the removal of adaptor sequences and reads with >10% unknown bases (N). Differentially expressed genes (DEGs) were selected by comparing the data of different samples according to the reported methods 53 . The threshold for DEGs was a false discovery rate (FDR) ≤0.001, an absolute value of log2 ratio ≥1, and at least one sample read >10. Web Gene Ontology Annotation Plot (WEGO) was used to perform GO classification of the DEGs and to understand the distribution of gene functions in the species at the macro level 54 . Pathway-based analysis was performed by searching the KEGG pathway-related database (https://www.kegg.jp/) 55 . A pathway with a Q value ≤0.05 was defined as significantly enriched in differentially expressed genes.

Gene expression analysis by qRT-PCR
Transcription levels obtained by RNA-seq of 20 selected DEGs were confirmed by qRT-PCR. The selected DEGs were chosen among the three libraries based on their relation to secondary metabolism, flavonoid metabolism, and transcription factors. 18S RNA was used as the reference gene. Gene-specific primers were designed using Primer3 software and are listed in Supplementary  Table S5. RNA was extracted from different stages of mutant (M1, M2, and M3) and wild-type apples (O1, O2, and O3). qRT-PCR was performed with an ABI 7300 Real-Time PCR System to analyze gene expression according to the manufacturer's instructions. All reactions were carried out using SYBR Green Master Mix (SYBR Premix EX TaqTM. TaKaRa) in a total volume of 20 µL, and PCR amplification was performed with the following parameters: 95°C denaturing for 5 min, followed by 40 cycles of 95°C denaturing for 5 s, 55°C annealing for 30 s, and 72°C extension for 30 s. All reactions and nontemplate controls were performed in triplicate. Relative transcription levels were calculated using the 2-ΔΔCt method 56 . Each measurement was performed with three biological replicates.
Analysis of the gene sequence and phylogenetic tree construction DNA fragments of the following gene promoters, (LOC103444202), and MdMYB90-like (HF36881-RA), and the CDSs of MdMYB90-like (621 bp) and MdMYB1 were PCR-amplified from Fuji apple. The primers are listed in Supplementary Table S5. The promoters were sequenced and analyzed using Plant-CARE online tools (http://bioinformatics.psb.ugent.be/ webtools/plantcare/html/) for cis-acting regulatory elements.

Subcellular localization
The PCR-amplified full-length CDSs of MdMYB90-like and MdMYB1 were sequenced and cloned into the pC29_35S:GFP5_his6 vector (Supplementary Fig. S6).
The recombinant plasmids MdMYB90-like-GFP and MdMYB1-GFP were used for transit gene expression analysis. Two-centimeter squares were cut from fresh onion and placed on MS hypertonic media. After dark culture at 25°C overnight, epidermal cells of the onion's inner skin were peeled off, placed on MS medium, and transformed by particle bombardment. Transformed tissue was incubated overnight at 25°C, and GFP expression was detected by LSM 710 NLO laser confocal microscopy (Zeiss, Germany). A 35S:GFP vector was used as a positive control.
For subcellular localization in tobacco leaves, Agrobacterium harboring the 35S:GFP, 35S:MdMYB90-like-GFP, and 35S:MdMYB1-GFP constructs were infiltrated and transiently expressed in tobacco leaves. GFP signals were captured under a laser confocal microscope.

Overexpression of MdMYB90-like in apple calli
The full-length CDS of MdMYB90-like was cloned into a pCAMBIA1301 vector 58 to generate a 35S:MdMYB90-like construct ( Supplementary Fig. S6). The recombinant plasmid was introduced into Agrobacterium strain LBA4404 and transformed into "Orin" apple calli according to the method described by An et al. 59 . Transgenic calli were screened based on hygromycin resistance. Transgenic calli and WT calli were grown at 24°C under dark conditions and subcultured every 15 days on media supplemented with hygromycin. Three lines of transgenic calli were harvested, transferred to new plates, and cultured for 5 days under light conditions. Anthocyanin contents and the expression of genes related to anthocyanin biosynthesis were analyzed in both transgenic and WT calli.

Transient expression of MdMYB90-like in apple skin
35S:MdMYB90-like was cloned into the pGreenII62-SK vector to generate 62SK-MdMYB90-like ( Supplementary  Fig. S6). The empty pGreenII62-SK vector was used as a negative control. The vectors were introduced into GV3101(p-soup). Cultured A. tumefaciens cells were injected into Fuji and Golden Delicious fruit skins, and the infiltrated fruits were cultured at room temperature under continuous light conditions.

Dual-luciferase assay
The 62SK-MdMYB90-like effectors (35S:MdMYB90-like cloned into pGreenII62-SK vector) and reporter constructs (the promoter fragments of MdMYB1, MdbHLH3, MdCHS, and MdUFGT cloned into the pGreenII 0800-LUC vectors, Supplementary Fig. S6) were transformed into A. tumefaciens GV3101(p-soup). The bacteria were mixed and coinjected into tobacco leaves and cultured for 2 days under light conditions. A living fluorescence imager was used to observe the fluorescence of the tobacco leaves, which were also sampled to measure LUC/REN activity.
Binding reactions contained 2 µL of 1× binding buffer, 1 µL of MYB90-like-MBP protein extract, 1 µL of biotinlabeled probes, and 1 µL of unlabeled competitors or mutant probes in a total volume of 10 µL. Reactions were electrophoresed, transferred, and detected as described in the Chemiluminescent EMSA Kit (Beyotime Biotechnology).

DNA methylation analysis
Genomic DNA (1 µg) from apple skins of both Fuji and its colored mutant was digested separately with the methylation-specific endonuclease enzyme McrBC (New England Biolabs) in a 100 µl total volume including 1 µg DNA, 1× NEB2 buffer, 1× BSA (bovine serum albumin), 20 U McrBC and 1 mM GTP or ddH 2 O (as a negative control). The reaction was performed at 37°C overnight and was stopped by heating at 65°C for 20 min. The digested gDNA was used as a template for PCR analysis. Primers were designed to divide the promoter and exon sequences of both MdMYB1 and MdMYB90-like into 14 fragments (Supplementary Table S5). PCRs were performed with the following parameters: 95°C denaturing for 1 min, followed by 35 cycles of denaturing at 95°C for 30 s, annealing at 55°C for 30 s and extension at 72°C for 1 min, and a final 5 min extension at 72°C. PCR amplification products were checked by agarose gel (1.2%) electrophoresis.
To analyze methylated nucleotides in the promoter region of MdMYB90-like, bisulfite sequencing was performed. Genomic DNA (1 µg) from apple skins of both Fuji and its colored mutant was treated with a DNA Methylation Kit (www.cwbiotech.com). Treated gDNA and untreated gDNA (control) were used as templates to amplify two regions of the MdMYB90-like gene promoter (−1183 to −988 and −2018 to −1778), which showed different levels of methylation between the wild-type and the mutant. The primers are listed in Supplementary Table  S5. Amplified PCR products were cloned, sequenced, and analyzed using cytosine methylation analysis online tools CyMATE (http://www.cymate.org/). Methylation levels as the percentage of methylated nucleotides were calculated from nine independent clones.

Statistical analysis
For statistical analysis, three replicates were performed. Statistical analysis was performed using Microsoft Excel 2010. Each value represents the mean ± SE of three independent biological replicates. The differences between data were analyzed with t tests. A P value <0.05 was considered statistically significant.