Research Article
Comprehensive profiling and natural variation of flavonoids in rice Running title: Rice flavonoid profiling and natural variation Xuekui Dong1, Wei Chen1, Wensheng Wang1, Hongyan Zhang2, Xianqing Liu3, Jie Luo1,*
1
National Key Laboratory of Crop Genetic Improvement and National Center of Plant Gene Research (Wuhan),
Huazhong Agricultural University, Wuhan 430070, China 2
Key Laboratory of Horticultural Plant Biology (Ministry of Education), Huazhong Agricultural University,
Wuhan 430070, China 3
College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, China
*Correspondence: [email protected] Edited by: Dabing Zhang, Shanghai Jiao Tong University, China
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1111/jipb.12204] This article is protected by copyright. All rights reserved. Received: March 10, 2014; Accepted: April 10, 2014
1
Abstract Flavonoids constitute a major group of plant phenolic compounds. While extensively studied in Arabidopsis, profiling and naturally occurring variation of these compounds in rice (Oryza sativa), the monocot model plant, are less reported. Using a collection of rice germplasm, comprehensive profiling and natural variation of flavonoids were presented in this report. Application of a widely targeted metabolomics method facilitated the simultaneous identification and quantification of more than 90 flavonoids using liquid chromatography tandem mass spectrometry (LC-MS/MS). Comparing flavonoid contents in various tissues during different developmental stages revealed tissue-specific accumulation of most flavonoids. Further investigation indicated that flavone mono-C-glycosides, malonylated flavonoid O-hexosides and some flavonoid O-glycosides accumulated at significant higher levels in indica than in japonica, while the opposite was observed for aromatic acylated flavone C-hexosyl-O-hexosides. In contrast to the highly differential accumulation between the two subspecies, relatively small variations within subspecies were detected for most flavonoids. Besides, an association analysis between flavonoid accumulation and its biosynthetic gene sequence polymorphisms disclosed natural variation of flavonoids was probably caused by sequence polymorphisms in coding region of flavonoid biosynthetic genes. Our work paves the way for future dissection of biosynthesis and regulation of flavonoid pathway in rice.
Keywords: Oryza sativa; flavonoids; metabolic profiling; natural variation; single-nucleotide polymorphisms (SNPs)
2
INTRODUCTION Flavonoids are one of the most widespread groups of plant secondary metabolites (Koes et al. 1994) that are involved in a wide range of physiological processes, such as pigmentation for fruits and flowers (Winkel-Shirley 2001), attracting pollinators and seeds dispersal (Lepiniec et al. 2006), interactions of plant with microbes and animals (Dixon and Paiva 1995), auxin transportation (Taylor and Grotewold 2005) and protecting the plant from UV-B damage (Ryan et al. 2001). Furthermore, their pharmaceutical properties are becoming more attractive for human health (Dixon et al. 1998). In plants, flavonoids are synthesized via the phenylpropanoid pathway and can be divided into six major subgroups: flavanones, flavones, flavonols, flavan-3-ols, anthocyanins and isoflavones. Their diversity arises further from a combination of modification reactions, such as glycosylation and acylation (Stobiecki 2000; Lepiniec et al. 2006). For example, flavonoid O-glycosides are widespread in higher plants (Lin and Harnly 2007), while C-glycosylated flavonoids formed by direct linkage of the sugar to the basic nucleus of the flavonoid via C-C bond are the major class of flavonoids accumulated in cereals (Besson et al. 1985; Ferreres et al. 2007). Besides, some flavonoids are further acylated by either aliphatic (acetyl, malonyl, succinyl) or aromatic (hydroxycinnamoyl such as p-coumaroyl and feruloyl) acyl groups (Harborne and Williams 2000; Cavaliere et al. 2005). Structural characterization has been highly developed using LC-MS/MS and/or nuclear magnetic resonance (NMR) (Lin and Harnly 2007; Routaboul et al. 2012) for the identification of flavonoids in different plant species (Saito et al. 2013). Extensive profiling of flavonoids has been carried out in dicots such as Arabidopsis (Tohge et al. 2005; Saito et al. 2013) and recent study has also reported the natural variation of flavonoids in the seeds of 41 Arabidopsis accessions (Routaboul et al. 2012). Unlike dicots that accumulate O-glycosylated flavonols as the major type of flavonoids, monocots such as the major cereal crops predominantly synthesize flavone C-glycosides (Yun et al. 2008). For instance, a number of glycosylflavones with different aglycones, apigenin, luteolin, chrysoeriol, tricetin and tricin, were detected in wheat stem and leaf extracts (Cavaliere et al. 2005; Wojakowska et al. 2013). Structure and accumulation of both flavone C-glycosides and flavone O-glycosides in flag leaf and germinating seed of rice, one of the major staple food for about half of the world’s population and the model plant for monocots (Izawa and Shimamoto 1996), has also been illustrated recently (Gong et al. 2013). However, comprehensive profiling of tissue and developmental specific accumulation of flavonoids in rice has received little attention. Asian cultivated rice could be divided into two major subspecies: japonica and indica, based on their morphological characters and geographical distribution (Morishima 1981). Investigation of flavonoids in rice leaf has revealed that there are significant difference in levels of four flavone C-glycosides between japonica and indica rice (Chen et al. 2013). In addition, studies of rice flavonoid biosynthetic pathway have enabled the 3
identification of a few genes responsible for production of flavonoids. OsCGT has been identified as a glycosyltransferase that produces flavone-6C-glucosides from 2-hydroxyflavanone substrates that were generated by the flavanone 2-hydroxylase CYP93G2 from flavanones (Brazier-Hicks et al. 2009; Du et al. 2010). Several flavonoid O-glycosyltransferases, such as UGT706C1, UGT707A3, and UGT706D1 were also reported (Ko et al. 2008). Recent genetic analysis of the metabolome using a rice population has also allowed the reconstruction of flavonoid biosynthetic pathway in rice (Gong et al. 2013). Despite of these progresses, naturally occurring variation of flavonoids and the genetic basis underlying these variations in rice remains elusive. DNA sequence polymorphisms can be a primary genetic cause of phenotypic variation in population and have different effects on these variations (Schmid et al. 2005; Huang et al. 2013). With the progress in high-throughput genotyping, direct association analysis of allelic status to phenotypic consequences has become a popular alternative for elucidating molecular mechanism underlying phenotypic variation (Chan et al. 2010; Li et al. 2013). Linking DNA sequence polymorphism within genes of interest to differential metabolite accumulation, such as flavonoids, could thus be performed to dissect the genetics of biosynthetic pathways. In this study, comprehensive metabolic profiling and natural variation analysis of flavonoids were carried out in rice and a total of 91 flavonoids were (tentatively) identified and quantified using an LC-MS-based widely targeted metabolomics method. Tissue-specific and developmentally controlled accumulations were observed for most flavonoids, together with their differential accumulations between the two major subspecies of rice. An association analysis between the accumulation of flavonoids and the allelic status of flavonoid biosynthetic downstream genes suggests that the natural variation of flavonoids in rice might be partially due to the sequence variations in the coding region of the key structural genes. RESULTS Tissue-specific accumulation of flavonoids in various tissues of rice Using LC-ESI-MS/MS, we applied a newly developed widely targeted metabolomics method (Chen et al. 2013) into the comprehensive profiling analysis of flavonoids in rice. Based on an analysis of flavonoid levels in 38 rice varieties, leaf samples from 10 indica rice varieties with the highest accumulations were obtained to construct a MS2 spectral tag (MS2T) library (Table S1). Analyzing data matrix generated by the library, six flavonoids were newly identified by a direct comparison of mass-to-charge-ratio (m/z) values, the retention time (RT) and fragmental patterns with that of the commercial standards. About 20 flavonoids were newly annotated in case no authentic standards are available, which was performed as Chen previously described (Table S2). A total of 91 flavonoids that belong to four categories (72 flavones, 9 flavanones, 6 flavonols and 4 anthocyanins) 4
were detected in our study (Table S2). Flavone derivatives were found to be the major flavonoid constituents in rice and were mainly discussed. To investigate accumulation of flavonoids in different tissues of rice, samples from six different tissues were collected, including flag leaf, mature culm, panicle at grain filling stage, mature grain, and seed at 72h after germination (termed germinating seed hereafter) and root at vegetative stage. For each tissue, a mixture of samples from 24 rice varieties (Table S3) was obtained. Visualization of the flavonoid profile was performed by hierarchical cluster analysis (HCA; Fig. 1A). Accumulation of flavonoids displayed a clear phenotypic variation in terms of their abundance in different tissues. Flag leaf contained the highest levels of most flavonoids, followed by culm, panicle and grain. Root, however, showed the lowest accumulation of most flavonoids. In addition to the higher amount of flavonoid accumulation, comparison of flavonoid profiles demonstrated that flag leaf also yielded the most complex profile of flavonoids in different rice tissues. Based on their tissue-specific accumulation patterns, flavonoids could be clearly grouped into four main clusters with nine subclusters (Fig. 1A). Flavonoids in cluster I showed higher levels in flag leaf and culm than other tissues, and were mainly represented by flavonoid O-glycosides, including some of the major flavonoids such as tricin 5-O-hexoside, tricin 7-O-hexoside, tricin O-malonylhexoside (m0723), chrysoeriol 5-O-hexoside (m0505) and apigenin 5-O-glucoside (m0447). Flavonoids in cluster II were mainly represented by flavonoid C-glycosides with higher levels detected in flag leaf and panicle, such as luteolin 6-C-glucoside (m0475), O-methylchrysoeriol C-hexoside, C-hexosyl-luteolin O-hexoside (m0762), chrysoeriol 6-C-hexoside (m0509) and C-hexosyl-chrysoeriol O-feruloylhexoside (m0926). The differential accumulation of flavonoids in grain, germinating seed, and root further divided each of the cluster I and II into three subclusters. Major flavonoid O-glycosides (s176, m0447, m0723, m0823, m0681) and flavonoid C-glycosides (m0545, m0509, m0926, m0443, m0887) were tightly grouped in subcluster 1 and subcluster 6, respectively. These flavonoids displayed higher levels in germinating seed than those in the grain, suggesting the increased amount of flavonoid glycosides synthesis during seed germination. The two quercetin O-glycosides belonging to cluster III (subcluster 7), however, stood away from other flavonoids due to their tissue-specific accumulation in panicle and grain. Three anthocyanins in cluster IV, including cyanidin 3-O-glucoside, cyanidin 3-O-rutinoside and peonidin 3-O-hexoside, showed a panicle-specific accumulation pattern. High levels of accumulation of anthocyanins and glycosylated quercetin in Arabidopsis seeds have been reported (Routaboul et al. 2006). Some of the flavonoids in cluster IV (subcluster 8-9) display highest levels in culm and panicle and were mainly represented by four flavonolignans which have been characterized in the bran of one important Indian medicinal rice variety (Rao et al. 2010), indicating that they also displayed high levels in culm and panicle of rice. 5
Quantification of eight major flavonoids revealed that tricin 5-O-hexoside, one of the major flavonoid O-glycosides belonging to cluster I, accumulated at the level of 369.4 μg/g dry weight (DW) in flag leaf, while only 2.0 μg/g DW was detected in root. Similar accumulation pattern was observed for tricin O-malonylhexoside and apigenin 5-O-glucoside, with the highest levels of 89.2 ug/g and 5.3 ug/g DW detected in flag leaf, respectively (Fig. 1B). In addition to the high levels in flag leaf, several major C-glycosylated flavonoids that belong to cluster II also showed relatively higher levels in panicle, with 73.3 μg/g, 71.6 μg/g, 54.1 μg/g and 92.7 ug/g DW detected for luteolin 6-C-glucoside, C-hexosyl-luteolin O-hexoside, chrysoeriol C-hexoside and C-hexosyl-chrysoeriol O-feruloylhexoside in panicle, respectively (Fig. 1C).
Flavonoid accumulation patterns during developmental processes To further clarify flavonoid accumulation patterns during different developmental stages, a total of 24 tissues/organs covering nine developmental stages of rice were used (Table S4), and the 91 flavonoids present in these tissues were quantified. The results showed that flavonoids with different modifications showed different accumulation patterns during developmental stages. Quantification of various forms of flavonoids was subsequently performed and major flavonoids were selected to uncover the range of the variations observed in seedling, leaf, culm, panicle and root during growing stages (Fig. 2). For most C-glycosylated and O-glycosylated flavonoids such as apigenin 6-C-glucoside (m0443), luteolin 6-C-glucoside (m0475), O-methylapigenin C-hexoside (m0467), tricin 7-O-hexosides (s176), chrysoeriol 7-O-hexoside (m0508) and luteolin 7-O-glucoside, there was a sharp increase in the seedlings during the first 10 days after germination (S2), followed by a slight decrease during the later stage (S2-S5), (m0478; Fig. 2A). Evaluation of the contents of these flavonoids indicated significantly elevated levels in leaves at tillering stage (L1-L2), while no big difference was observed at the reproductive stage. Decreased levels of these flavonoids were measured during growth in culms (Fig. 2A). The levels of C-pentosyl-flavone O-hexosides and their aromatic acylated derivatives, including C-pentosyl-chrysoeriol O-hexoside (m0741) and C-pentosyl-apeignin O-feruloylhexoside (m0885), showed marked increase during seedling stage, but sharp decreased levels were observed for most aromatic acylated C-hexosyl-flavone/flavanone O-hexoside, such as C-hexosyl-chrysoeriol O-feruloylhexoside (m0926) and C-hexosyl-naringenin O-coumaroylhexoside (m0892; Fig. 2B). For most of the malonlyated flavonoids such as chrysoeriol O-malonylhexoside (m0681), luteolin O-malonylhexoside (m1064) and quercetin O-malonylhexoside (m1068) in leaves, significant increase of their accumulations were observed during tillering stage (L2; Fig. 2C). Besides, differential accumulation of some flavonolignans during culm (C1-C4) and panicle stage (P1-P3) were also detected (Fig. 2D), suggesting developmentally controlled accumulation of these flavonoids. 6
Differential accumulation of flavonoids between japonica and indica Differences in genetic background and geographical distribution between japonica and indica may result in different accumulation of flavonoids in these two subspecies. Samples of six different tissues were collected, including flag leaf, culm, panicle, grain, germinating seed and root. For each tissue sample, a mixture of material from 12 varieties of each subspecies was obtained. In flag leaf, five mono-C-glycosylflavones exhibited significant difference in accumulation between japonica and indica. Among them, two flavone C-pentosides, apigenin C-pentoside and luteolin C-pentoside, displayed 200 times over-accumulation in indica than that in japonica, despite their overall low contents of less than 1.0 μg/g DW. Three major flavone C-hexosides, including apigenin 6-C-glucoside, luteolin 6-C-glucoside and chrysoeriol C-hexoside also showed up to 13, 13.5, 25-fold higher levels in indica compared to their japonica counterpart, respectively (Fig. 3A). Similar results were also obtained for three di-C-glycosylflavones (m0656, m0685, m0779; Table S5). Apart from flavone C-glycosides, significant elevated levels of major flavone O-hexosides could also be measured in indica when compared to japonica, including apigenin 5-O-glucoside, chrysoeriol 5-O-hexoside and tricin 5-O-hexoside, although no significant difference was found for most mono-O-glycosyl flavonoids between japonica and indica (Fig. 3A; Table S5). In contrast to the levels of flavones 5-O-glucoside, there was a significantly decreased level of luteolin 7-O-glucoside in indica than in japonica, suggesting the potential competition of the regio-specific glycosylation for the aglycone (Fig. 3A). Levels of malonylated flavonoid derivatives in various tissues were also compared between japonica and indica. Four major malonylated flavonoid glycosides in flag leaf, including O-malonylhexosides of tricin, chrysoeriol, quercetin and naringenin, exhibited significant higher levels in indica than that in japonica, with elevated levels up to 19.4, 36.4, 32.5, 13.1-fold, respectively (Fig. 3B). Metabolic analysis of aromatic acylated both flavone C-pentosyl-O-hexosides and flavone C-hexosyl-O-hexosides also revealed their broad variations between the two rice subspecies. Both C-pentosyl-apigenin-O-hexoside with its coumaroyl-, caffeoyl-, feruloylderivatives and C-pentosyl-chrysoeriol O-hexoside with its feruloyl derivative, were accumulated at significant higher levels in indica than in japonica, indicating their similar accumulation patterns as flavone mono-C-glycosides (Table S5). However, coumaroyl-, feruloyl- and caffeoyl- derivatives of flavone C-hexosyl-O-hexosides in flag leaf exhibited decreased levels in indica compared with japonica (Fig. 3C). Despite of tissue-specific accumulation of flavonoids in various tissues and significant variation of flavonoids in flag leaf, comparison of levels of flavonoids between japonica and indica in culm, panicle, grain and germinating seed showed similar pattern to that in flag leaf, except in the root where low levels of most flavonoids were detected (Table S5). This implied that the flavonoid differences exist between japonica and 7
indica are not tissue and developmental specific.
Natural variation for flavonoids in rice leaf To further study the natural variation in individual rice varieties, a collection of 156 rice accessions, representing the mini-core collection of Chinese rice varieties were used. Samples from flag leaf at heading stage were collected for metabolic profiling where the highest accumulations of most flavonoids occurred. Evaluation of flavonoid contents between japonica and indica by ANOVA revealed significant variation of 68 flavonoids (p
Comprehensive profiling and natural variation of flavonoids in rice Running title: Rice flavonoid profiling and natural variation Xuekui Dong1, Wei Chen1, Wensheng Wang1, Hongyan Zhang2, Xianqing Liu3, Jie Luo1,*
1
National Key Laboratory of Crop Genetic Improvement and National Center of Plant Gene Research (Wuhan),
Huazhong Agricultural University, Wuhan 430070, China 2
Key Laboratory of Horticultural Plant Biology (Ministry of Education), Huazhong Agricultural University,
Wuhan 430070, China 3
College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, China
*Correspondence: [email protected] Edited by: Dabing Zhang, Shanghai Jiao Tong University, China
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1111/jipb.12204] This article is protected by copyright. All rights reserved. Received: March 10, 2014; Accepted: April 10, 2014
1
Abstract Flavonoids constitute a major group of plant phenolic compounds. While extensively studied in Arabidopsis, profiling and naturally occurring variation of these compounds in rice (Oryza sativa), the monocot model plant, are less reported. Using a collection of rice germplasm, comprehensive profiling and natural variation of flavonoids were presented in this report. Application of a widely targeted metabolomics method facilitated the simultaneous identification and quantification of more than 90 flavonoids using liquid chromatography tandem mass spectrometry (LC-MS/MS). Comparing flavonoid contents in various tissues during different developmental stages revealed tissue-specific accumulation of most flavonoids. Further investigation indicated that flavone mono-C-glycosides, malonylated flavonoid O-hexosides and some flavonoid O-glycosides accumulated at significant higher levels in indica than in japonica, while the opposite was observed for aromatic acylated flavone C-hexosyl-O-hexosides. In contrast to the highly differential accumulation between the two subspecies, relatively small variations within subspecies were detected for most flavonoids. Besides, an association analysis between flavonoid accumulation and its biosynthetic gene sequence polymorphisms disclosed natural variation of flavonoids was probably caused by sequence polymorphisms in coding region of flavonoid biosynthetic genes. Our work paves the way for future dissection of biosynthesis and regulation of flavonoid pathway in rice.
Keywords: Oryza sativa; flavonoids; metabolic profiling; natural variation; single-nucleotide polymorphisms (SNPs)
2
INTRODUCTION Flavonoids are one of the most widespread groups of plant secondary metabolites (Koes et al. 1994) that are involved in a wide range of physiological processes, such as pigmentation for fruits and flowers (Winkel-Shirley 2001), attracting pollinators and seeds dispersal (Lepiniec et al. 2006), interactions of plant with microbes and animals (Dixon and Paiva 1995), auxin transportation (Taylor and Grotewold 2005) and protecting the plant from UV-B damage (Ryan et al. 2001). Furthermore, their pharmaceutical properties are becoming more attractive for human health (Dixon et al. 1998). In plants, flavonoids are synthesized via the phenylpropanoid pathway and can be divided into six major subgroups: flavanones, flavones, flavonols, flavan-3-ols, anthocyanins and isoflavones. Their diversity arises further from a combination of modification reactions, such as glycosylation and acylation (Stobiecki 2000; Lepiniec et al. 2006). For example, flavonoid O-glycosides are widespread in higher plants (Lin and Harnly 2007), while C-glycosylated flavonoids formed by direct linkage of the sugar to the basic nucleus of the flavonoid via C-C bond are the major class of flavonoids accumulated in cereals (Besson et al. 1985; Ferreres et al. 2007). Besides, some flavonoids are further acylated by either aliphatic (acetyl, malonyl, succinyl) or aromatic (hydroxycinnamoyl such as p-coumaroyl and feruloyl) acyl groups (Harborne and Williams 2000; Cavaliere et al. 2005). Structural characterization has been highly developed using LC-MS/MS and/or nuclear magnetic resonance (NMR) (Lin and Harnly 2007; Routaboul et al. 2012) for the identification of flavonoids in different plant species (Saito et al. 2013). Extensive profiling of flavonoids has been carried out in dicots such as Arabidopsis (Tohge et al. 2005; Saito et al. 2013) and recent study has also reported the natural variation of flavonoids in the seeds of 41 Arabidopsis accessions (Routaboul et al. 2012). Unlike dicots that accumulate O-glycosylated flavonols as the major type of flavonoids, monocots such as the major cereal crops predominantly synthesize flavone C-glycosides (Yun et al. 2008). For instance, a number of glycosylflavones with different aglycones, apigenin, luteolin, chrysoeriol, tricetin and tricin, were detected in wheat stem and leaf extracts (Cavaliere et al. 2005; Wojakowska et al. 2013). Structure and accumulation of both flavone C-glycosides and flavone O-glycosides in flag leaf and germinating seed of rice, one of the major staple food for about half of the world’s population and the model plant for monocots (Izawa and Shimamoto 1996), has also been illustrated recently (Gong et al. 2013). However, comprehensive profiling of tissue and developmental specific accumulation of flavonoids in rice has received little attention. Asian cultivated rice could be divided into two major subspecies: japonica and indica, based on their morphological characters and geographical distribution (Morishima 1981). Investigation of flavonoids in rice leaf has revealed that there are significant difference in levels of four flavone C-glycosides between japonica and indica rice (Chen et al. 2013). In addition, studies of rice flavonoid biosynthetic pathway have enabled the 3
identification of a few genes responsible for production of flavonoids. OsCGT has been identified as a glycosyltransferase that produces flavone-6C-glucosides from 2-hydroxyflavanone substrates that were generated by the flavanone 2-hydroxylase CYP93G2 from flavanones (Brazier-Hicks et al. 2009; Du et al. 2010). Several flavonoid O-glycosyltransferases, such as UGT706C1, UGT707A3, and UGT706D1 were also reported (Ko et al. 2008). Recent genetic analysis of the metabolome using a rice population has also allowed the reconstruction of flavonoid biosynthetic pathway in rice (Gong et al. 2013). Despite of these progresses, naturally occurring variation of flavonoids and the genetic basis underlying these variations in rice remains elusive. DNA sequence polymorphisms can be a primary genetic cause of phenotypic variation in population and have different effects on these variations (Schmid et al. 2005; Huang et al. 2013). With the progress in high-throughput genotyping, direct association analysis of allelic status to phenotypic consequences has become a popular alternative for elucidating molecular mechanism underlying phenotypic variation (Chan et al. 2010; Li et al. 2013). Linking DNA sequence polymorphism within genes of interest to differential metabolite accumulation, such as flavonoids, could thus be performed to dissect the genetics of biosynthetic pathways. In this study, comprehensive metabolic profiling and natural variation analysis of flavonoids were carried out in rice and a total of 91 flavonoids were (tentatively) identified and quantified using an LC-MS-based widely targeted metabolomics method. Tissue-specific and developmentally controlled accumulations were observed for most flavonoids, together with their differential accumulations between the two major subspecies of rice. An association analysis between the accumulation of flavonoids and the allelic status of flavonoid biosynthetic downstream genes suggests that the natural variation of flavonoids in rice might be partially due to the sequence variations in the coding region of the key structural genes. RESULTS Tissue-specific accumulation of flavonoids in various tissues of rice Using LC-ESI-MS/MS, we applied a newly developed widely targeted metabolomics method (Chen et al. 2013) into the comprehensive profiling analysis of flavonoids in rice. Based on an analysis of flavonoid levels in 38 rice varieties, leaf samples from 10 indica rice varieties with the highest accumulations were obtained to construct a MS2 spectral tag (MS2T) library (Table S1). Analyzing data matrix generated by the library, six flavonoids were newly identified by a direct comparison of mass-to-charge-ratio (m/z) values, the retention time (RT) and fragmental patterns with that of the commercial standards. About 20 flavonoids were newly annotated in case no authentic standards are available, which was performed as Chen previously described (Table S2). A total of 91 flavonoids that belong to four categories (72 flavones, 9 flavanones, 6 flavonols and 4 anthocyanins) 4
were detected in our study (Table S2). Flavone derivatives were found to be the major flavonoid constituents in rice and were mainly discussed. To investigate accumulation of flavonoids in different tissues of rice, samples from six different tissues were collected, including flag leaf, mature culm, panicle at grain filling stage, mature grain, and seed at 72h after germination (termed germinating seed hereafter) and root at vegetative stage. For each tissue, a mixture of samples from 24 rice varieties (Table S3) was obtained. Visualization of the flavonoid profile was performed by hierarchical cluster analysis (HCA; Fig. 1A). Accumulation of flavonoids displayed a clear phenotypic variation in terms of their abundance in different tissues. Flag leaf contained the highest levels of most flavonoids, followed by culm, panicle and grain. Root, however, showed the lowest accumulation of most flavonoids. In addition to the higher amount of flavonoid accumulation, comparison of flavonoid profiles demonstrated that flag leaf also yielded the most complex profile of flavonoids in different rice tissues. Based on their tissue-specific accumulation patterns, flavonoids could be clearly grouped into four main clusters with nine subclusters (Fig. 1A). Flavonoids in cluster I showed higher levels in flag leaf and culm than other tissues, and were mainly represented by flavonoid O-glycosides, including some of the major flavonoids such as tricin 5-O-hexoside, tricin 7-O-hexoside, tricin O-malonylhexoside (m0723), chrysoeriol 5-O-hexoside (m0505) and apigenin 5-O-glucoside (m0447). Flavonoids in cluster II were mainly represented by flavonoid C-glycosides with higher levels detected in flag leaf and panicle, such as luteolin 6-C-glucoside (m0475), O-methylchrysoeriol C-hexoside, C-hexosyl-luteolin O-hexoside (m0762), chrysoeriol 6-C-hexoside (m0509) and C-hexosyl-chrysoeriol O-feruloylhexoside (m0926). The differential accumulation of flavonoids in grain, germinating seed, and root further divided each of the cluster I and II into three subclusters. Major flavonoid O-glycosides (s176, m0447, m0723, m0823, m0681) and flavonoid C-glycosides (m0545, m0509, m0926, m0443, m0887) were tightly grouped in subcluster 1 and subcluster 6, respectively. These flavonoids displayed higher levels in germinating seed than those in the grain, suggesting the increased amount of flavonoid glycosides synthesis during seed germination. The two quercetin O-glycosides belonging to cluster III (subcluster 7), however, stood away from other flavonoids due to their tissue-specific accumulation in panicle and grain. Three anthocyanins in cluster IV, including cyanidin 3-O-glucoside, cyanidin 3-O-rutinoside and peonidin 3-O-hexoside, showed a panicle-specific accumulation pattern. High levels of accumulation of anthocyanins and glycosylated quercetin in Arabidopsis seeds have been reported (Routaboul et al. 2006). Some of the flavonoids in cluster IV (subcluster 8-9) display highest levels in culm and panicle and were mainly represented by four flavonolignans which have been characterized in the bran of one important Indian medicinal rice variety (Rao et al. 2010), indicating that they also displayed high levels in culm and panicle of rice. 5
Quantification of eight major flavonoids revealed that tricin 5-O-hexoside, one of the major flavonoid O-glycosides belonging to cluster I, accumulated at the level of 369.4 μg/g dry weight (DW) in flag leaf, while only 2.0 μg/g DW was detected in root. Similar accumulation pattern was observed for tricin O-malonylhexoside and apigenin 5-O-glucoside, with the highest levels of 89.2 ug/g and 5.3 ug/g DW detected in flag leaf, respectively (Fig. 1B). In addition to the high levels in flag leaf, several major C-glycosylated flavonoids that belong to cluster II also showed relatively higher levels in panicle, with 73.3 μg/g, 71.6 μg/g, 54.1 μg/g and 92.7 ug/g DW detected for luteolin 6-C-glucoside, C-hexosyl-luteolin O-hexoside, chrysoeriol C-hexoside and C-hexosyl-chrysoeriol O-feruloylhexoside in panicle, respectively (Fig. 1C).
Flavonoid accumulation patterns during developmental processes To further clarify flavonoid accumulation patterns during different developmental stages, a total of 24 tissues/organs covering nine developmental stages of rice were used (Table S4), and the 91 flavonoids present in these tissues were quantified. The results showed that flavonoids with different modifications showed different accumulation patterns during developmental stages. Quantification of various forms of flavonoids was subsequently performed and major flavonoids were selected to uncover the range of the variations observed in seedling, leaf, culm, panicle and root during growing stages (Fig. 2). For most C-glycosylated and O-glycosylated flavonoids such as apigenin 6-C-glucoside (m0443), luteolin 6-C-glucoside (m0475), O-methylapigenin C-hexoside (m0467), tricin 7-O-hexosides (s176), chrysoeriol 7-O-hexoside (m0508) and luteolin 7-O-glucoside, there was a sharp increase in the seedlings during the first 10 days after germination (S2), followed by a slight decrease during the later stage (S2-S5), (m0478; Fig. 2A). Evaluation of the contents of these flavonoids indicated significantly elevated levels in leaves at tillering stage (L1-L2), while no big difference was observed at the reproductive stage. Decreased levels of these flavonoids were measured during growth in culms (Fig. 2A). The levels of C-pentosyl-flavone O-hexosides and their aromatic acylated derivatives, including C-pentosyl-chrysoeriol O-hexoside (m0741) and C-pentosyl-apeignin O-feruloylhexoside (m0885), showed marked increase during seedling stage, but sharp decreased levels were observed for most aromatic acylated C-hexosyl-flavone/flavanone O-hexoside, such as C-hexosyl-chrysoeriol O-feruloylhexoside (m0926) and C-hexosyl-naringenin O-coumaroylhexoside (m0892; Fig. 2B). For most of the malonlyated flavonoids such as chrysoeriol O-malonylhexoside (m0681), luteolin O-malonylhexoside (m1064) and quercetin O-malonylhexoside (m1068) in leaves, significant increase of their accumulations were observed during tillering stage (L2; Fig. 2C). Besides, differential accumulation of some flavonolignans during culm (C1-C4) and panicle stage (P1-P3) were also detected (Fig. 2D), suggesting developmentally controlled accumulation of these flavonoids. 6
Differential accumulation of flavonoids between japonica and indica Differences in genetic background and geographical distribution between japonica and indica may result in different accumulation of flavonoids in these two subspecies. Samples of six different tissues were collected, including flag leaf, culm, panicle, grain, germinating seed and root. For each tissue sample, a mixture of material from 12 varieties of each subspecies was obtained. In flag leaf, five mono-C-glycosylflavones exhibited significant difference in accumulation between japonica and indica. Among them, two flavone C-pentosides, apigenin C-pentoside and luteolin C-pentoside, displayed 200 times over-accumulation in indica than that in japonica, despite their overall low contents of less than 1.0 μg/g DW. Three major flavone C-hexosides, including apigenin 6-C-glucoside, luteolin 6-C-glucoside and chrysoeriol C-hexoside also showed up to 13, 13.5, 25-fold higher levels in indica compared to their japonica counterpart, respectively (Fig. 3A). Similar results were also obtained for three di-C-glycosylflavones (m0656, m0685, m0779; Table S5). Apart from flavone C-glycosides, significant elevated levels of major flavone O-hexosides could also be measured in indica when compared to japonica, including apigenin 5-O-glucoside, chrysoeriol 5-O-hexoside and tricin 5-O-hexoside, although no significant difference was found for most mono-O-glycosyl flavonoids between japonica and indica (Fig. 3A; Table S5). In contrast to the levels of flavones 5-O-glucoside, there was a significantly decreased level of luteolin 7-O-glucoside in indica than in japonica, suggesting the potential competition of the regio-specific glycosylation for the aglycone (Fig. 3A). Levels of malonylated flavonoid derivatives in various tissues were also compared between japonica and indica. Four major malonylated flavonoid glycosides in flag leaf, including O-malonylhexosides of tricin, chrysoeriol, quercetin and naringenin, exhibited significant higher levels in indica than that in japonica, with elevated levels up to 19.4, 36.4, 32.5, 13.1-fold, respectively (Fig. 3B). Metabolic analysis of aromatic acylated both flavone C-pentosyl-O-hexosides and flavone C-hexosyl-O-hexosides also revealed their broad variations between the two rice subspecies. Both C-pentosyl-apigenin-O-hexoside with its coumaroyl-, caffeoyl-, feruloylderivatives and C-pentosyl-chrysoeriol O-hexoside with its feruloyl derivative, were accumulated at significant higher levels in indica than in japonica, indicating their similar accumulation patterns as flavone mono-C-glycosides (Table S5). However, coumaroyl-, feruloyl- and caffeoyl- derivatives of flavone C-hexosyl-O-hexosides in flag leaf exhibited decreased levels in indica compared with japonica (Fig. 3C). Despite of tissue-specific accumulation of flavonoids in various tissues and significant variation of flavonoids in flag leaf, comparison of levels of flavonoids between japonica and indica in culm, panicle, grain and germinating seed showed similar pattern to that in flag leaf, except in the root where low levels of most flavonoids were detected (Table S5). This implied that the flavonoid differences exist between japonica and 7
indica are not tissue and developmental specific.
Natural variation for flavonoids in rice leaf To further study the natural variation in individual rice varieties, a collection of 156 rice accessions, representing the mini-core collection of Chinese rice varieties were used. Samples from flag leaf at heading stage were collected for metabolic profiling where the highest accumulations of most flavonoids occurred. Evaluation of flavonoid contents between japonica and indica by ANOVA revealed significant variation of 68 flavonoids (p
