Different Biosynthesis Patterns among Flavonoid 3-glycosides with Distinct Effects on Accumulation of Other Flavonoid Metabolites in Pears (Pyrus bretschneideri Rehd.)

Flavonoid biosynthesis profile was clarified by fruit bagging and re-exposure treatments in the green Chinese pear ‘Zaosu’ (Pyrus bretschneideri Rehd.) and its red mutant ‘Red Zaosu’. Two distinct biosynthesis patterns of flavonoid 3-glycosides were found in ‘Zaosu’ pear. By comparison with ‘Red Zaosu’, the biosynthesis of flavonoid 3-galactosides and flavonoid 3-arabinosides were inhibited by bagging and these compounds only re-accumulated to a small degree in the fruit peel of ‘Zaosu’ after the bags were removed. In contrast, the biosynthesis of flavonoid 3-gluctosides and flavonoid 3-rutinosides was reduced by bagging and then increased when the fruits were re-exposed to sunlight. A combination of correlation, multicollinearity test and partial-correlation analyses among major flavonoid metabolites indicated that biosynthesis of each phenolic compound was independent in ‘Zaosu’ pear, except for the positive correlation between flavonoid 3-rutincosides and flavanols. In contrast with the green pear cultivar, almost all phenolic compounds in the red mutant had similar biosynthesis patterns except for arbutin. However, only the biosynthesis of flavonoid 3-galactosides was relatively independent and strongly affected the synthesis of the other phenolic compounds. Therefore, we propose a hypothesis that the strong accumulation of flavonoid 3-galactosides stimulated the biosynthesis of other flavonoid compounds in the red mutant and, therefore, caused systemic variation of flavonoid biosynthesis profiles between ‘Zaosu’ and its red mutant. This hypothesis had been further demonstrated by the enzyme activity of UFGT, and transcript levels of flavonoid biosynthetic genes and been well tested by a stepwise linear regression forecasting model. The gene that encodes flavonoid 3-galacosyltransferase was also identified and isolated from the pear genome.


Introduction
Flavonoid secondary metabolites are associated with a multitude of biological functions. The pigments that color fruit peels are anthocyanins and proanthocyanins which are flavonoid metabolites [1]. The colorless flavonoid compounds (e.g. flavanols) are hidden by the ubiquitous green of the chlorophylls in epidermal cells of leaves, buds or fruits, but they, as well as anthocyanins, make significant contributions to resistance to UV light and pathogens [2,3]. With their antioxidant activity, these flavonoids also have a health benefit to humans [2].
The Chinese pear 'Red Zaosu' originated as a spontaneous mutation from 'Zaosu' (Pyrus bretschneideri Rehd.). The young fruit of 'Red Zaosu' is dark red, but then its color intensity continuously decreases accompanied with the appearance of green stripes on its surface during the maturation. In 'Red Zaosu' fruit, growth parameters and fruiting characters are similar with those in 'Zaosu' [4].
The flavonoid pathway in pear fruits is well established (Fig. 1) [5]. Phenylalanine is the precursor for the synthesis of many polyphenols, and is enzymatically converted to flavonoids via many steps. The enzymes involved in the pathway include phenylalanine ammonialyse (PAL), chalone synthase (CHS), flavanone 3-hydroxylase (F3H), dihydroflavonols 4-reductase (DFR), anthocyanin reductase (ANR), anthocyanin synthase (ANS), flavonol synthase (FLS), and UDP-glucose: flavonoid 3glucosyltransferase (UFGT) [5,6]. Changes in the transcriptional levels of flavonoid biosynthetic genes could cause systematic variation in flavonoid biosynthesis. The silencing of MdANS in a transgenic red Malus hybrid significantly reduced anthocyanin content accompanied by rises in flavanols and flavonols [7]. Transcriptional factors in the regulation of flavonoid biosynthesis in pear fruits, PyMYB10 and PcMYB10 were isolated via homology with apple sequences [8][9][10]. In the cultivar 'Early Red Doyenne du Comice' and its green variant strain, most of the structural genes were up-regulated in the red cultivar during fruit development, but they were not the key factors controlling the red/green color mutant [11]. The high methylation level of the promoter of PcMYB10, which caused low expression levels of PyUFGT, was associated with the formation of the green-skinned sport in the 'Max Red Bartlett' pear [10].
Fruit bagging can provide ideal fruits for studying flavonoid biosynthesis, by ensuring fruit of similar developmental stage and physiological condition [12]. Bagging treatments can significantly inhibit the biosynthesis of most flavonoid compounds in apple peels. Compared with unbagged apple fruits, the bagged ones contain less anthocyanins, procyanidins, quercetin glycosides, and simple phenols in their peels [3,13,14]. The effect of fruit bagging differs among cultivars [3]. In 'Fuji' apple, anthocyanins gradually accumulate after the bag removal and the anthocyanin content was the highest when the bags had been removed for 20 days [15]. In red Chinese sand pears, the anthocyanins accumulated rapidly for more than 10 days after the bag was removed, then the rate of accumulation decreased as the fruits matured [16]. Although the synthesis pattern of anthocyanins has been studied by bagging and re-exposure treatments in pears and apples, the synthesis patterns of flavonols, flavanols and phenolic acids are still unclear.
In the present study, we studied flavonoid biosynthesis in 'Zaosu' and 'Red Zaosu' in terms of metabolites and gene expression. We identified two UF3GT genes related to different flavonoid 3-glycoside biosynthesis patterns. Interrelationships in the biosynthesis of different flavonoid compounds were studied by a combination of correlation and partial-correlation analyses. Moreover, we built a forecasting model based on stepwise linear regression and analyzed the enzyme activity of UF3GT and the expression patterns of flavonoid biosynthetic genes to confirm these interrelationships.

Plant materials
The Chinese pear cultivars 'Zaosu' (Pyrus bretschneideri Rehd.) and its red mutant 'Red Zaosu', the fruit of which matured in end of July, were used for materials. The trees were 5-year-old on Pyrus betulaefolia Bunge rootstocks which grew in a commercial plantation of Dali County (lat: 34.8542, lng: 109.8378; elevation: 368 m), Shaanxi, China. No specific permissions are required for this location. In this study, no endangered or protected species were involved.
Considering the effects of the different transcriptional levels of flavonoid biosynthetic genes during the fruits maturation, we extended the duration of re-exposure time from less a month in previous studies [13,17] to almost two months. In early May 2012, about 30 well exposed fruits per tree were bagged for each treatment. About 30 days later, when the color of bagged fruits turned to light yellow without any redness on its surface, we started to remove bags every 8 days until the fruits matured. Eight different re-exposure duration treatments were used: 0 days (bagged fruits), 8 days, 16 days, 24 days, 32 days, 40 days, 48 days and 56 days. Non-bagged fruits were treated as the control. For each treatment, 5 replications (i.e. 5 trees) were used. To ensure the similar developmental stages, sizes and physiological conditions of the fruit, all the mature fruits (bagged,re-exposed and unbagged) were collected on the same day (24 July, 2012). For each tree replication, random samples of 20 fruit were taken. The bagged fruits were harvested without taking off the bags to avoid exposure to light before sample collections. Fruits in different natural developmental stage (25 days, 40 days, 55 days, 70 days, 85 days and 100 days after flower full bloom) were also obtained. Fruit peels (about 1 mm thick) were collected with a peeler. The samples were immediately frozen in liquid nitrogen and stored at 280uC until analysis.

Fruit color measurement
Fruit skin color was measured on the most colored part of fruit using a colorimeter (CR-400, Minolta, Japan), which provided CIE L*, a*, and b* value. L* represents the relative lightness of color with a range from 0 to 100, being small for dark color and large for light color. Both a* and b* scales extend from 260 to 60. Negative a* value indicates greenness and positive for redness, while b* is negative for blueness and positive for yellowness [18]. For this study, we used a* value to evaluate the redness/greenness on fruit peels.

Phenolic compounds analysis
The extraction and analysis of phenolic compounds were carried out as described by Zhang et al. [19]. Briefly, the phenolics were extracted with 70% methanol containing 2% formic acid at 0-4uC.The supernatant was filtered through a 0.45 mm syringe filter prior to HPLC analysis.

Identification of candidate UFGT genes, phylogeny analysis and cDNA cloning
The identification and analysis of candidate genes from genome database were carried out as described by Li et al. [20]. Candidate UF3GT genes were identified by performing blastn analysis against the Pear Genome Database from Centre of Pear Engineering Technology Research, Nanjing Agricultural University, China (http://peargenome.njau.edu.cn/default.asp?d = 4&m = 2) [21] using MdUFGT1 (AF117267.1) and PcUFGT (KC460398.1) in GenBank as queries and an E-value of 1.00E-09 as the threshold. The putative candidate gene sequences were retrieved from the Pear Genome Database: http://peargenome.njau.edu.cn/default. asp?d = 4&m = 2/. The corresponding sequences of candidate genes were then used for a blastn search against the Pear EST database from Centre of Pear Engineering Technology Research, Nanjing Agricultural University, China (http://peargenome.njau. edu.cn/default.asp?d = 4&m = 2) to confirm that each predicted gene was expressed in the pear transcriptome when there is a high similarity EST sequence (score .300 bp, and identity .99%). Two putative candidate UF3GT genes, which were named as PyUFGT1 and PyUFGT2, were screened for expression analysis. The fulllength PbUFGT1 cDNA was isolated by using the primer set (forward, 59-ATGGCACCGCCGCCGCC-39, and reverse, 59-CTATGCTTTCTTGGATCCTGATATA -39) designed according to its predicted CDs (from Coding DNA Sequence) sequence data in the Pear Genome Database (http://peargenome.njau.edu. cn/default.asp?d = 4&m = 2). Phylogenetic analysis of Malus and Pyrus amino acid sequences was performed using maximum likelihood (http://www.phylogeny.fr).

Total RNA extraction and RT-PCR
Total RNA was extracted by using SDS-phenol method as described by Fonseca et al. [22]. The RNA concentration and quality were detected by UV spectrophotometry and by running on a 1.2% agar/EB gel.

Enzyme measurement of UFGT
One gram of frozen tissue was homogenized using a pestle and mortar in 5% PVPP and 2.4 ml 100 mM Tris-HCl buffer (pH 8.0) containing 14 mM b-mercaptoethanol, 5 mM DTT, 2 mM EDTA, 15 mM MgCl2, 0.5% Triton X-100, 10% glycerol, and 1% BSA. The homogenate was centrifuged at 14,000 g for 20 min at 4uC. The supernatant was applied to a PD10column equilibrated with the elution buffer containing 100 mM Tris-HCl buffer (pH 8.0) and 2 mM DTT. Once the dissolved precipitate was completely absorbed onto the column, the protein was eluted with 2 mL elution buffer.
UFGT (EC 2.4.1.91): was assayed in a modification of the buffer system reported by Do et al. (1995). In a final assay volume of 200 mL, the reaction conditions were 100 mM buffer (Tris-HCl, pH 8.0), 250 mM MgCl2, 2 mM dithiothreitol, 0.9 mM UDPglucose(or galactose), 100 mM quercetin or cyanidin chloride, and 100 mL enzyme extract. The reaction mixture was incubated at 37 uC for 10 min and extracted twice with ethylacetate. The ethylacetate was then evaporated completely. The reaction product was dissolved with methanol and analyzed by HPLC at 525 nm or 365 nm.

Statistical analysis
Analysis of variance and significant difference tests were conducted to identify differences among means by one-way ANOVA (with Turkey's HSD test). Correlation and partial-Correlation analyses were conducted by determining the Pearson product moment correlation and its level of probability. A multicollinearity test was conducted by the Variance Inflation Factor (VIF). Stepwise linear regression analysis (SLRA) was applied to select the most suitable variables from major phenolic compounds (arbutin, phenolic acids, flavanols, flavonoid 3galactosides, flavonoid 3-gluctosides, flavonoid 3-rutinosides, flavonoid 3-arabinosides) to forecast the hypothetical changes of flavonoid profiles. Statistical analyses and SLRA were conducted using SPSS 16.0 (SPSS, Chicago, IL, USA).

Red color development and changes in anthocyanin levels
Fruit bagging significantly affected color parameters. Natural 'Red Zaosu' fruits presented the highest values for red color (component 'a*' after Minolta colorimeter analysis). The 'a*' values of bagged fruits was almost 0 in the two cultivars, which indicated no red pigment in their skins. The re-exposure fruits of 'Red Zaosu' had increasing 'a*' values from 8 days to 56 days, coincident with the gradually changed colors from light yellow to red. With the same treatments, the light yellow 'Zaosu' fruits gradually turned to green. The anthocyanin level corresponded to the changing patterns of 'a*' values in 'Red Zaosu' pear, and we did not find any anthocyanins in bagged, re-exposed or natural fruits of 'Zaosu' pear ( Fig. 2).

Biosynthesis patterns of phenolic compounds
Arbutin and three major phenolic acid compounds, i.e. gallic acid, chlorogenic acid and caffeic acid, were identified and quantified in the fruit peel of 'Zaosu' pear and its red mutant (Fig.3, Table.1). In response to fruit bagging/re-exposure treatments, the concentration of arbutin remained unchanged in the both cultivars (Fig.3, Table.1). The biosynthesis pattern of phenolic acids for the two cultivars did not differ significantly. The levels were slightly inhibited by bagging and gradually recovered after bag removal (Fig.3, Table.1 and Table.2). As a result, 'Zaosu' and its red mutant showed similar biosynthesis patterns in upstream flavonoid pathway. The fruits were showed in graph A, the anthocyanidin concentrations were showed in graph B and the color data in pear fruit peel were showed in graph C. 'a*' value represents peel color from greenness (negative value) to redness (positive value). The treatments of re-exposed to sunlight for 0 day represented bagged fruits, natural fruits were used as controls. Error bars are SE for 5 replicates. doi:10.1371/journal.pone.0091945.g002  Four types of flavanols, i.e. catechin, epicatechin, procyanidin B1 and procyanidin B2, seven types of flavonols, i.e. quercetin 3galactoside, quercetin 3-gluctoside, quercetin 3-arabinoside, quercetin 3-rutinoside, isrohamnetin 3-galactoside, isrohamnetin 3gluctoside, isrohamnetin 3-rutinoside and one anthocyanin, i.e. cyanidin 3-galactoside were detected in the pear fruit peels. The concentration of flavanols in the both cultivars was significantly reduced by the bagging treatment. When the fruits were reexposed to sunlight, flavanols gradually re-accumulated for 40 days in 'Red Zaosu' pear, but interestingly, it just slightly reaccumulated for 24 days in 'Zaosu' pear ( Fig.3, Table. 1 and  Table.2). Fruit bagging significantly reduced the concentration of all types of flavonoid 3-glycosides, i.e. anthocyanidin and flavonols in 'Red Zaosu'. All of these compounds increasingly re-accumulated for 56 days when the fruits were re-exposed to sun light ( Fig.3 and Table. 1). In 'Zaosu', cyanidin-3-galactoside was not detectable, only rare quecertin-3-galactoside and isrohamnetin-3galactoside were detected. Two different flavonoid 3-glycosides biosynthesis patterns were found in 'Zaosu'. In the first pattern, the concentration of flavonol gluctosides and rutinosides was reduced by bagging, and then continuously increased after bag removal, but the concentration of flavonoid 3-galactosides and quecertin 3arabinoisde were hardly affected by bagging or re-exposure treatment (Fig.3, Table.1). These results indicated that the main variation of flavonoid profile between 'Zaosu' and its red mutant 'Red Zaosu' were in the flavonoid 3-glycosides pathway and the flavanol pathway. When the bagged fruits were re-exposed to sunlight, the ability to synthesize flavanols and flavonoid 3glycosides for 'Red Zaosu' could completely recover, but that was much more difficult for 'Zaosu'. Moreover, two distinct biosynthesis patterns of flavonoid 3-glycosides were found in 'Zaosu' pear.
During the fruit developmental process, the concentration of all major flavonoid compounds continuously decreased in 'Red Zaosu' and 'Zaosu'.The main variances of the flavonoid profile between 'Zaosu' and 'Red Zaosu' were in flavonoid 3-glycosides pathway and flavanol pathway, which were quite similar with that in the bagging and re-exposure treatments. Compared with 'Red Zaosu', only flavonoid 3-galactosides were severely suppressed in 'Zaosu' during the whole coloring process, which were also consistent with our observation during the re-exposure treatments (Fig.S1).

Correlations among the biosynthesis patterns of phenolic compounds
Correlations among the synthesis patterns of flavonoid 3glycosides, flavanols, arbutin and phenolic acids were conducted to define the similarities and the interactions among the synthesis patterns of these flavonoid metabolites. There was no significant difference between the two cultivars in arbutin biosynthesis (p = 0.806) and phenolic biosynthesis (p = 0.084). The synthesis patterns of flavonoid glycosides and flavanols between 'Red Zaosu' and 'Zaosu' were both correlated but at different levels (positive correlation for flavonoid glycosides, r = 0.828, p,0.01, weak correlation for flavanol, r = 0.636, p,0.01)( Table 2).
A combination analysis of correlation, partial-correlation and multicollinearty tested differences among flavanols, and 4 types of flavonoid 3-glycosides, e.g. flavonoid 3-galactosides, flavonoid 3gluctosides, flavonoid 3-rutinosides and flavonoid 3-arabinoside in 'Red Zaosu' and 'Zaosu', respectively. The correlation analysis and multicollinearity test by VIF indicated that flavanol biosynthesis pattern correlated with the biosynthesis patterns of flavonoid 3-gluctosides and flavonoid 3-rutinosides, but the biosynthesis patterns of other flavonoid metabolites were independent, respectively ( Table 3). In 'Red Zaosu', all biosynthesis patterns of major flavonoid metabolites were correlated, except for arbutin (Table 3). However, in 'Red Zaosu', when we used flavonoid 3galactosides as the control variable in a combination analysis of partial-correlation and multicollinearity test, the correlations coefficients and VIF among other flavonoid severely reduced ( Table 3; VIF,10, data not shown), but when we used other components as control variables, the correlation coefficients involved flavonoid 3-galactosides were just slightly affected (Table. 3). This result indicated that the synthesis patterns of flavonoid 3galactosides was relatively independent, but strongly affected the synthesis patterns of other flavonoid metabolites in the red pear. More interestingly, these results implied that the strongly accumulation of flavonoid 3-galactosides stimulated the biosynthesis of other flavonoid compounds in the red mutant and, therefore, caused systemic variation of flavonoid biosynthesis profiles between 'Zaosu' and its red mutant.

Stepwise linear regression model for forecasting flavonoid biosynthesis profiles in 'Red Zaosu'
To forecast the hypothetical changes of flavonoid profiles when the flavonoid 3-galactosides pathway was inhibited (as well as that in 'Zaosu') in 'Red Zaosu', regression equations for biosynthesis patterns of major flavonoid compounds, e.g. phenolic acids, flavanols, flavonoid 3-galactosides, flavonoid 3-gluctosides, flavonoid 3-rutinosides, and flavonoid 3-arabinosides, were determined by stepwise linear regression analysis (SLRA), respectively (Table.  4). Suitable independent variables were selected from external cause (re-accumulation duration) and internal causes (major flavonoid compounds). The biosynthesis patterns of flavonoid 3glycosides, flavanols, and phenolic acids in the forecasting model were similar with that in 'Zaosu' (Fig. 3). The results showed that when flavonoid 3-galactosides pathway was inhibited in 'Red Zaosu', the biosynthesis patterns of all other major flavonoid compounds were inhibited like 'Zaosu', which further proved the hypothesis we mentioned before.

Expression of key genes in flavonoid biosynthesis and UFGT activity
The transcript levels of PbANR, PbANS, PbCHS, PbF3H, PbDFR, and PbMYB10 in bagged fruits were significantly lower than that in control fruits and increased to different levels after the bag removal. Compared to its red mutant, 'Zaosu' had lower transcript levels but similar synthesis patterns of these genes. Two UFGT genes showed different expression patterns in the both cultivar. PbUFGT2 showed a quite low transcript level and was not sensitive to bagging and re-exposure treatments in the both cultivars. After the bags were removed, PbUFGT1 was firstly activated and rapidly increased to a high transcript level, as was PbMYB10. Although its expression decreased later, it still remained a quite high level  Table 3. The correlation and partial correlation coefficients of the synthesis patterns of flavonoid 3-glycosides and flavanols in the two cultivars. compared with the bagged fruits. The results indicated that PbUFGT1 and PbUFGT2 were associated with the two different biosynthesis types of flavonoid 3-glycosides. The expression of other flavonoid biosynthetic genes also increased after the fruits reexposed to the sunlight, but they reached their maxima later than PbUFGT1 (Fig.4). Moreover, compared with the high UFGT activity in 'Red Zaosu', UFGT was rarely activated in 'Zaosu' (Fig.5). These results explained that why the accumulation of flavonoid 3-galactosides was inhibited in 'Zaosu' in transcript levels and further demonstrated the strong accumulation of flavonoid 3-galactosides stimulated the biosynthesis of other flavonoid compounds in the red mutant and, therefore, caused systemic variation of flavonoid biosynthesis profiles between 'Zaosu' and its red mutant.

Candidate genes which encoding flavonoid 3glycosyltransferase
Two UFGT genes were identified in the Pyrus genome. PbUFGT1, same as PcUFGT (GenBank accession: JX403956) reported by Wang et al. [9], shared high similarity with MdUFGT1 (GenBank accession: AF117267); PbUFGT2 had high homology with PpUFGT (GenBank accession: GU390548) reported by Zhang et al. [17] (Fig. 6). The cDNA sequences of PbUFGT1 containing the complete coding region were isolated from the fruit peels of 'Zaosu' and its red mutant, respectively, and showed high homology (at 100%) between 'Zaosu' and its red mutant.

Discussion
Clearly, the bagging treatment lowered most phenolic compound concentrations in the peel of the two cultivars (Figs. 3-4, Table.1). This was consistent with the observation that the activity/expression of key enzymes/genes involved in the flavonoid metabolism pathway could be up-regulated by light irradiations [14, 24 and 25]. Different types of flavonoid 3glycosides in 'Zaosu' had different synthesis patterns. Flavonol arabinosides and flavonol galactosides were hardly affected by bagging and re-exposure treatments; flavonol gluctosides and flavonol rutinosides re-accumulated after the bag removal (Figs. [4][5][6]. Previous studies reported that anthocyanins and flavonols can protect leaves or fruits from photodamage by absorbing visible / UV-B light [2,3]. The concentration of anthocyanins and flavonols decreased most among flavonoid metabolites by fruit bagging and increased most when the fruits re-exposed to visible and UV-B light [3,[26][27][28]. But in our study, the synthesis pattern of quercetin 3-arabinoside in 'Zaosu' showed that it might be not sensitive to light. Its continuous accumulation in the red mutant could be explained: When the bagged 'Red Zaosu' fruits were re-exposed to sunlight, quercetin 3-arabinoside was stimulated to continuously accumulate by large amounts of its new synthesized substrate, quecertin, which in response to accumulating other quercetin 3glycosides, especially quercetin 3-galactosides. Two UFGT genes, i.e. PbUFGT2 and PbUFGT1 were isolated from pear genome. Their expressions were totally different and corresponded to synthesis patterns of quercetin 3-arabinoside and other flavonoid glycosides, respectively. In this case, suitable genes should be first identified before further research on its relation with the synthesis pattern of relevant metabolites. Whether the accumulation of flavanols in fruit peels was independent or dependent of the light conditions to which fruits were exposed were still in dispute in previous studies [3,29,30]. Distinct biosynthesis patterns of were found between 'Zaosu' and its red mutant, flavanols synthesis was strongly activated when bag removed in the red mutant, but rarely affected by the re-exposure treatments in 'Zaosu'. The correlation coefficient (r = 0.861) between flavonoid 3-glycosides and flavonols in 'Red Zaosu' was higher than that (r = 0.476) in 'Zaosu'. This variation also could be caused by the biosynthesis flavonoid 3-galactosides in 'Red Zaosu'. All flavonoid 3-glycosides synthesis patterns were similar in the red mutant. In this multicollinear system, interferences should be erased by a combination analysis of partial correlation and variance inflation factor. The results showed that only flavonoid 3galactosides made a significant contribution to accumulation of flavanols and other flavonols in 'Red Zaosu'.
Above all, the main distinctions in flavonoid biosynthesis patterns between 'Zaosu' and its red mutant were in the flavonol/anthocyanin pathways and the flavanol pathway. When the fruit bags were removed, the re-accumulation of other flavonoid compounds was probably caused by flavonoid 3galactosides synthesis in the red mutant. In order to confirm this, the expression patterns of flavonoid structural genes have been determined. PbUFGT1 and PbMYB10 were first activated and rapidly increased to a high transcript level in 'Red Zaosu', but stayed in a relatively low in 'Zaosu'. The expression of other flavonoid biosynthetic genes also increased after the fruits were reexposed to the sunlight, but they reached their maxima later than PbUFGT1. This result indicated that the inhibition of PbUFGT1 maybe the key factor that caused the main distinctions of flavonoid biosynthesis patterns between 'Zaosu' and its red mutant, but the similarity of PbUFGT1 coding sequences in the two cultivars were 100%. Therefore, the ultimate reason could be a transcription factor that can regulate UFGT expression, for example the MYB10 or another gene in the R2R3-MYB family.  Figure 4. The relative expression patterns of flavonoid structure genes and one transcriptional factor PcMYB10 in the peels of 'Zaosu' and its mutant 'Red Zaosu' by bagging and re-exposure treatments. The treatments of re-exposed to sunlight for 0 day represented bagged fruits, natural fruits were used as controls. Error bars are SE for 5 replicates. doi:10.1371/journal.pone.0091945.g004 Supporting Information Figure S1 The biosynthesis patterns of flavonoid metabolites in the fruit peels of 'Zaosu' and its mutant 'Red Zaosu' during the fruit coloring process. Error bars are SE for 5 replicates. (TIF)