Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Dissecting genetic architecture of grape proanthocyanidin composition through quantitative trait locus mapping

Dissecting genetic architecture of grape proanthocyanidin composition through quantitative trait... Background: Proanthocyanidins (PAs), or condensed tannins, are flavonoid polymers, widespread throughout the plant kingdom, which provide protection against herbivores while conferring organoleptic and nutritive values to plant-derived foods, such as wine. However, the genetic basis of qualitative and quantitative PA composition variation is still poorly understood. To elucidate the genetic architecture of the complex grape PA composition, we first carried out quantitative trait locus (QTL) analysis on a 191-individual pseudo-F1 progeny. Three categories of PA variables were assessed: total content, percentages of constitutive subunits and composite ratio variables. For nine functional candidate genes, among which eight co-located with QTLs, we performed association analyses using a diversity panel of 141 grapevine cultivars in order to identify causal SNPs. Results: Multiple QTL analysis revealed a total of 103 and 43 QTLs, respectively for seed and skin PA variables. Loci were mainly of additive effect while some loci were primarily of dominant effect. Results also showed a large involvement of pairwise epistatic interactions in shaping PA composition. QTLs for PA variables in skin and seeds differed in number, position, involvement of epistatic interaction and allelic effect, thus revealing different genetic determinisms for grape PA composition in seeds and skin. Association results were consistent with QTL analyses in most cases: four out of nine tested candidate genes (VvLAR1, VvMYBPA2, VvCHI1, VvMYBPA1) showed at least one significant association with PA variables, especially VvLAR1 revealed as of great interest for further functional investigation. Some SNP-phenotype associations were observed only in the diversity panel. Conclusions: This study presents the first QTL analysis on grape berry PA composition with a comparison between skin and seeds, together with an association study. Our results suggest a complex genetic control for PA traits and different genetic architectures for grape PA composition between berry skin and seeds. This work also uncovers novel genomic regions for further investigation in order to increase our knowledge of the genetic basis of PA composition. Background diverse qualities are directly linked to PA chemical Proanthocyanidins (PAs), or condensed tannins, are fla- structures. As polymers, PA structure varies depending vonoid polymers widespread throughout the plant king- on the degree of polymerisation and the nature of build- ing blocks, the flavan-3-ols (differences in stereochemis- dom. They accumulate in many organs and tissues to provide protection against pests [1]. They are also deter- try, hydroxylation pattern on the B-ring and presence/ minant in food quality and their beneficial effects on absence of a galloyl group, Figure 1). Our understanding human health are increasingly investigated [1,2]. These of PA biosynthesis has been significantly improved through the isolation of two genes coding for leu- coanthocyanidin reductase (LAR, [3]) and anthocyanidin * Correspondence: doligez@supagro.inra.fr reductase (ANR, [4,5]), two specific enzymes for the for- UMR AGAP, INRA, 2, place Viala, 34060 Montpellier, France mation of flavan-3-ols, respectively (+)-(gallo)catechin Full list of author information is available at the end of the article © 2012 Huang et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Huang et al. BMC Plant Biology 2012, 12:30 Page 2 of 19 http://www.biomedcentral.com/1471-2229/12/30 and (-)-epi(gallo)catechin. However, several issues con- [13-16] while (+)-gallocatechin and (-)-epigallocatechin- cerning PA composition require further study, such as 3-O-gallate are present in trace amounts only [15]. PAs the synthesis of galloylated units, the genetic mechanism are abundant in grape berries with drastic differences in of polymerisation, and the origin of extension units, composition between skin and seeds: total content is since all flavonoid intermediates are believed to assume usually higher in seeds while polymer size is much lar- a2,3-trans configuration, similar to the 2,3-configura- ger in skin [15,16]. In terms of constitutive building tion of (+)-(gallo)catechin (Figure 1), while major PA blocks, (-)-epigallocatechin is a major component of grape skin PAs [15] while it is not detectable in seeds extension blocks assume a 2,3-cis configuration (e.g. [16]; (-)-epicatechin-3-O-gallate is present in large pro- (-)-epicatechin, Figure 1). Moreover, few studies are available on the genetic basis of PA composition quanti- portion as both extension and terminal subunits in tative variation [6,7]. seeds while it is present only in small amounts in skin It is of great interest to understand PA genetics in [15,16]. Advances in understanding grape PA synthesis grape since PAs are involved in grapevine self-defence have been mainly obtained through homologous cloning mechanisms and are responsible for major organoleptic [14,17-21]. However, the complex PA composition properties of red wine [8-10]. Because of its rich PA within a tissue and the contrasted composition between composition and the multiple genetic and genomic tools tissues suggest a complex interaction of many factors in available for this species, such as the whole genome the determinism of grape PA composition. sequence [11,12], grape could represent also an interest- One way to assess genetic determinism of trait varia- ing model for PA genetic study. Indeed, in Arabidopsis, tion without apriori knowledge is quantitative trait loci a major model for PA studies, PAs are only detected in (QTL) mapping. QTL mapping makes use of segregating the seed coat with the presence of (-)-epicatechin as populations and gives global insights into the genetic sole building block. By contrast, PAs are present in dif- architecture of the target phenotype, i.e.the number, ferent organs of grapevine and are composed of four position and effects of genomic regions [22]. Among all major building blocks: (+)-catechin, (-)-epicatechin, available mapping methods, the multiple-QTL approach (-)-epigallocatechin and (-)-epicatechin-3-O-gallate is particularly suitable for complex trait analysis since it OH HO O OH Extension subunits OH 3’ OH 2’ OH 4’ 1’ HO O 7 2 5’ OH OH 6’ A C 6 2 OH 4 1 Terminal subunits galloyl OH PA monomeric units R R1 R2 2, 3 configuration Subunits studied in this work (+)-catechin H OH H trans catEx/catT (-)-epicatechin H H OH cis epiEx/epiT (-)-epicatechin-3-O-gallate H H O-galloyl cis galEx/galT (+)-gallocatechin OH OH H trans n.d. (-)-epigallocatechin OH H OH cis egcEx (-)-epigallocatechin-3-O-gallate OH H O-galloyl cis n.d. Figure 1 Structures of proanthocyanidins and monomeric subunits. A generic structure of proanthocyanidin is shown and the possible configurations are highlighted in colour. “n” indicates the number of extension units, variable according to plant species and tissues. The general chemical structure of PA monomeric subunit includes a C6-C3-C6 skeleton which is called the A-C-B rings. The carbon nomenclature is indicated as numbers next to the corresponding carbon. The B-ring generally bears two or three hydroxyl groups. According to the stereochemistry of carbons 2 and 3 on the C-ring, the PA monomeric subunit could be in 2,3-trans (e.g. (+)-catechin) or in 2,3-cis configuration (e.g.(-)-epicatechin). The structure of galloyl is shown next to PA generic structure. The right column denotes the subunits studied in this work where “Ex” indicates “extension” units and “T”, terminal subunits/monomers. n.d., not detected. Huang et al. BMC Plant Biology 2012, 12:30 Page 3 of 19 http://www.biomedcentral.com/1471-2229/12/30 uses simultaneously multiple marker intervals with pos- [25], additional SSR markers, heterozygous both in Syrah sible inclusion of epistasis terms in QTL mapping and Grenache, were chosen from recent grapevine refer- model [23]. Instead of creating segregating populations, ence maps [29]. Based on the 12X grapevine reference one can also explore the existing diversity through asso- sequence (Grape Genome Browser http://www.genoscope. ciation mapping to identify loci involved in phenotypic cns.fr), we designed primers for candidate gene amplifica- variation [24]. According to the genetic architecture of tion using Primer3 (http://www.bioinformatics.nl/cgi-bin/ target traits, one can build appropriate breeding strategy primer3plus/primer3plus.cgi) with default parameters. Pri- and/or develop further gene function studies. mers used in this study are listed in Additional file 1. For The aim of this work was to investigate the genetic SNP analysis, gene fragments were amplified, sequenced determinism of PA composition variation in both skin and and analysed as described in [30]. seeds of grape. For this purpose, we first characterised skin and seed grape PA composition in a pseudo-F1 pro- Linkage map construction geny derived from a cross between Syrah and Grenache Framework maps were constructed based on the 97- cultivars. Three categories of PA variables were con- marker linkage map of [25] with 56 additional SSRs. All structed in order to capture the complex PA profiles: total 153 markers had a genotypic error rate lower than 1.5%, content variables, percentages of constitutive subunits, using Tmap as check [31]. Linkage maps were con- which assessed the biosynthesis efficiency among building structed using CarthaGène 0.999R [32] as described in blocks, and composite ratio variables, which included esti- [33] with Haldane mapping function. The “Syrah” and mation of polymer size and metabolite flux between build- “Grenache” framework maps were composed of 121 ing blocks. We then applied a multiple-QTL genome scan SSRs (total length 1118.8 cM), and 133 SSRs (1349.4 cM to identify main effect QTLs and pairwise epistatic interac- total length) respectively. The “Consensus” framework tions for PA variables. Nine functional candidate genes, map spanned 1256.4 cM based on 153 SSRs among among which eight co-located with QTLs, were sequenced which over 70% allowed segregation in four genotypic and their SNP-phenotype associations were investigated in classes in the pseudo F1 progeny (ab × ac and ab × cd). a grapevine diversity panel. We present here the first Marker order reliability was ensured at LOD 2 thresh- extensive study of genetic architecture of PA composition old. Segregation distortion on genotypic classes was ver- in grape and confirm the involvement of some candidate ified by a c test according to the segregating type of genes in PA composition variation. each marker for the different maps (e.g. for markers seg- regating as ab × cd in the consensus map, the H Methods hypothesis was ac:bc:ad:bd = 1:1:1:1). Twenty-five mar- Plant material kers out of 153 exhibited distorted segregation (P < The two grapevine populations used in this study have 0.05) and were mainly grouped on chromosomes 3 (4 been previously described [25-27]. Briefly, the QTL markers), 4 (6 markers) and 10 (5 markers). Markers on mapping population (S × G) consisted of a pseudo-F1 chromosome 4 exhibited the most significant allelic progeny of 191 individuals from a cross between two deviation (P < 0.001) due to segregating distortion wine grape cultivars, Syrah (S) and Grenache (G) and between Syrah alleles (aa:ab ~2/3:1/3). was maintained under classical local training system (3300 plants/ha plant density) at Montpellier SupAgro Phenotyping and PA variable construction Domaine du Chapitre (Hérault, France). The S × G Grapes were harvested at maturity (20° Brix). For each population was planted in two blocks. Each individual genotype, eight representative berry clusters were har- from the progeny was planted in two elementary plots vested from the five plants of the elementary plot. Sam- (one per block) comprising five plants each. Parental ple homogenisation was based on the accumulation of cultivars were also planted in each block with nine and total solutes (principally sucrose), a major marker of 43 elementary plants for Syrah and Grenache, respec- berry development during ripening. Berry density was tively. The association mapping population (CC) con- assessed by floatation in salt solutions [34]. Twenty-five sisted of a core-collection of 141 cultivars maximising berries with a density between 130 and 160 g NaCl/L agro-morphological diversity for 50 quantitative and were randomly selected. In the present study, we qualitative traits, and maintained at INRA Domaine de focused on the analysis of berry skin and seeds since Vassal (Hérault, France) [26]. PA concentration is quite low in flesh, flesh PAs accounting for only 2-6% of the total berry PA content DNA extraction, marker genotyping and gene sequencing [35]. Berry skin and seeds were separated, ground in DNA extraction and marker genotyping were already liquid nitrogen and stored at-80°C until analysis. PAs described in Adam-Blondon et al. [28]. In order to densify were extracted and analysed by high performance liquid a previous 97-SSR linkage map of the S × G population chromatography (HPLC) after acid-catalysed cleavage in Huang et al. BMC Plant Biology 2012, 12:30 Page 4 of 19 http://www.biomedcentral.com/1471-2229/12/30 the presence of phloroglucinol according to [36]. For sum of terminal subunits and monomers. We also con- the S × G population, both skin and seeds were ana- structed composite ratio variables, including mean degree lysed for 2 consecutive years: skin was analysed in 2005 of polymerisation (mDP), ratio of 2,3-trans-to2,3-cis- (1 block) and 2006 (2 blocks) while seeds were analysed subunits in extension position (Ftranscis_Ex), terminal in 2006 (2 blocks) and 2007 (2 blocks). For the CC position (Ftranscis_T) and overall subunits (Ftranscis_all) population, skin was analysed in 2005 and 2006 and and the ratio of B-ring di-hydroxylated to B-ring tri- seeds in 2006. hydroxylated subunits (F3pr35). Variables studied in this In order to obtain an exhaustive view of PA composi- work are summarised in Table 1. tion, three categories of PA variables were studied in this work: total PA content, subunit percentage and compo- Phenotypic data analysis site ratio variables. For total content variables, concP All statistical analyses were performed with R software (mg/g fresh weight) reflects the biosynthesis intensity in [37]. We identified the best-fit mixed model for each each tissue, concB (mg/berry) brings total content to sin- PA variable through Bayesian information criterion gle berry level by taking into account berry size while (BIC) in order to extract the best linear unbiased predic- concK (mg/kg berries) is a common enological measure- tors (BLUPs) for genotypic values and to estimate the ment taking into account yield-related traits. Since all PA broad sense heritability (H ). Mixed model fit was per- building blocks are derived from the same intermediate formed with lme4 package [38]. The mixed model structure, naringenin chalcone, we used the percentage assumption of normality of residual and BLUPs was of each PA subunit to total subunit quantity to assess checked after model fitting by quantile-quantile plot partitioning efficiency between PA building blocks. Our comparing the distribution of residual and random PA characterisation method did not distinguish between effect predictors to a theoretical normal distribution terminal units of polymers and flavan-3-ol monomers. (Additional file 2). No data transformation was The notation ending with “T” corresponds thus to the performed for PA variables measured in the two Table 1 PA variables used in this study and their description PA traits Skin/ Definition Biological/biochemical significance Seed Total content concP +/+ mg/g fresh tissue Biosynthesis intensity per gram of tissue concB +/+ mg/berry Taking berry size into account concK +/+ mg/Kg berries Taking yield related-trait into account Subunit 100·(subunit content)/(total content) Assessment of partitioning efficiency of percentage catEx +/+ (+)-catechin Extension subunit PA biosynthesis among different subunits epiEx +/+ (-)-epicatechin Extension subunit galEx +/+ (-)-epicatechin-3-O-gallate Extension subunit egcEx +/- (-)-epigallocatechin Extension subunit catT +/+ (+)-catechin Terminal subunit/monomer epiT +/+ (-)-epicatechin Terminal subunit/monomer galT +/- (-)-epicatechin-3-O-gallate Terminal subunit/monomer Composite variables mDP +/+ mean Degree of Polymerisation(Total number of extension and Assessment of PA polymer size terminal/monomer subunits)/(Number of terminal subunit/monomer) F3pr35 +/- (catEx + epiEx + galEx + catT + epiT)/(egcEx) Assessment of flux between B-ring di-OH and tri-OH subunits Ftranscis_Ex +/+ Skin:catEx/(epiEx + galEx + egcEx) Assessment of flux between 2,3-trans subunit and Seed: catEx/(epiEx + galEx) 2,3-cis subunit in extension part Ftranscis_T +/+ Skin: catT/epiT Assessment of flux between 2,3-trans subunit and Seed: catT/(epiT + galT) 2,3-cis subunit in terminal part Ftrancis_all +/+ Skin: (catEx + catT)/(epiEx + galEx + egcEx + epiT) Global assessment of flux between 2,3-trans subunit Seed: (catEx + catT)/(epiEx + galEx + epiT + galT) and 2,3-cis subunit(extension + terminal/monomer) Presence/Absence (indicated by +/-) of a given trait in grape berry tissues. based on PA content expressed in mg/g fresh tissue F for Flux. Huang et al. BMC Plant Biology 2012, 12:30 Page 5 of 19 http://www.biomedcentral.com/1471-2229/12/30 populations. More information regarding phenotypic is the probability of a greater F valueunder thenull data analysis and best-fit model for each PA variable is hypothesis that polymorphism was independent of the in Additional file 3. phenotype. The adjusted P value (called p_adj_Marker in TASSEL), is the site-wise P value adjusted for multi- QTL analysis ple tests which takes into account the dependence QTL analysis was performed on the genotypic BLUPs between SNPs due to linkage disequilibrium. Because with R/qtl package [39]. Multiple QTL regression was each gene was tested independently, we used an addi- carried out with “stepwiseqtl” function. This approach tional Bonferroni correction to correct for the number of studied genes (nine) which led to a threshold of uses forward/backward selection to identify a multiple- QTL model with inclusion of both main effect QTLs and 0.0056 for the adjusted P value. As the permutation pairwise interactions. Maximum QTL number was set to method is not available for MLM, we used the thresh- 10 for forward selection (max.qtl = 10). Model choice old proposed by Benjamini and Hochberg [45] with q was made via a penalized LOD score (pLOD) which is equal to 0.05 which led to a threshold of 0.0039. The the LOD score for the model (the log likelihood ratio effect of minor genotypic frequency and non-normality comparing the full model to the null model without of observed trait distribution was checked (details in QTL) with penalties on the number of QTLs and pair- Additional file 4). wise QTL × QTL interactions [40]. For each PA variable, specific penalties for main effect and digenic pairwise Results interaction terms were derived from 1000 permutations Phenotype analysis of two-dimensional scan (the “scantwo” function, method PA variable distribution and heritability = “hk”, n.perm = 1000) and penalties at genome-wide For the S × G population, all PA variables showed contin- error rate of 0.05 were used for multiple-QTL model fit- uous distribution and transgressive segregation and varia- ting. The QTL model with the largest pLOD was identi- tion extent was equivalent in the S × G and CC fied as the most probable one. Once determined the populations (Figure 2, Additional file 5). In agreement multiple QTL model, we refined QTL position ("refi- with previous studies [15,16,46], samples taken in 2006, neqtl” function) and estimate R for the whole model and for which both berry skin and seeds were analysed, dis- each term of the model, the individual LOD score of played different PA composition between tissues as illu- each term and the genotypic effect ("fitqtl” function). The strated in Figure 2A with the mean values in the S × G “lodint” function was used to derive LOD-1 QTL location population. For both S × G and CC populations, (-)-epica- confidence interval. Allelic effects for consensus QTLs techin (epiEx) was the predominant extension subunit in were estimated as described by Segura et al.[41].Gen- all tissues while (-)-epigallocatechin (egcEx) was only ome scan was performed with a 1 cM step. detected in skin. (+)-catechin (catT) was the predominant terminal subunit/monomer in both skin and seed while Association analysis galloylated units (galEx and galT) were more abundant in Nine candidate genes were selected for association test seed PA. Each subunit exhibited large variation according according to their function and co-localisation with to genotype. For instance in skin, (-)-epigallocatechin QTLs. Prior to association test, we used R kinship pack- (instead of (-)-epicatechin) could be the predominant sub- age [42] to perform model comparison among different unit in the extension position (67.9%) (Additional file 5). nested models according to [43] in order to select the PA content variables (concP, concB, concK) reached best fitted model for association test for each PA vari- higher values in seeds than in skin regardless of the able. Ancestry structure and kinship matrix were esti- unit, as illustrated for concP (Figure 2B), which exhib- mated based on 20 SSR markers located throughout the ited the largest difference between tissues. Comparison whole genome as described in [25]. of composite PA ratio variables showed different range After model comparisons, we used TASSEL package of variation between skin and seed. PAs were on average to perform association tests [44]. Two models were 8-times shorter in seeds than in skin with wide variation used: one accounting for ancestry structure effect (with in skin (Figure 2C). All three ratio variables assessing General Linear Model, or GLM in TASSEL) the other the flux between 2,3-trans and 2,3-cis forms (Ftrancis for both ancestry structure effect and random genetic series) pointed to different kinetics for extension and background effect (with Mixed Linear Model, or MLM terminal positions: trans subunits were more abundant in TASSEL). Association tests were performed on in skin for terminal units/monomers (Ftranscis_T) while BLUPs for skin variables and raw data for seed vari- they were much reduced in seeds when considering ables since seed data were available for 2006 only. For extension positions alone (Ftranscis_Ex) or extension GLM analyses, tests were run with 1000 permutations plus terminal subunits/monomers (Ftranscis_all, Figure 2D-F). Since the major extension blocks were in cis- allowing the determination of site-wise P value, whi ch Huang et al. BMC Plant Biology 2012, 12:30 Page 6 of 19 http://www.biomedcentral.com/1471-2229/12/30 A B SxG seed egcEx SxG skin galEx CC seed 80 60 epiEx CC skin Grenache catEx Syrah galT epiT % 40 catT skin seed concP (mg/g fresh tissue) C D Ftranscis_Ex mDP 0 0 Ftranscis_T Ftranscis_all Figure 2 Comparative composition of skin and seed PA in 2006 (A) and distribution of PA variables of S × G and CC populations in 2006 for concP (B), mDP (C), Ftranscis_Ex (D) Ftranscis_T (E) and Ftranscis_all (F).(A) PA composition in skin and seeds based on the S × G offspring average is shown. Each building block is presented as the offspring average percentage of total content. (B-F) Distribution of PA variables in S × G and CC populations in 2006. Upper limits of data interval are indicated under the x-axis. Full symbols near x-axis show mean parental values for S × G population, circle for Grenache and triangle for Syrah, in pink for skin values and in blue for seed value as for the offspring histograms. N° of Individuals N° of Individuals 0.5 2.5 0.7 3.5 0.9 4.5 1.1 5.5 1.3 1.5 6.5 1.7 7.5 1.9 8.5 2.1 9.5 2.7 2.9 N° of Individuals N° of Individuals N° of Individuals 0.015 0.007 0.025 0.013 0.035 0.019 0.045 0.025 0.055 0.031 0.065 0.037 0.075 0.05 0.085 0.07 60 0.095 0.09 0.12 0.11 0.17 0.13 0.23 0.15 0.28 0.17 0.33 0.19 0.38 0.21 0.43 150 Huang et al. BMC Plant Biology 2012, 12:30 Page 7 of 19 http://www.biomedcentral.com/1471-2229/12/30 configuration, i.e. (-)-epi(gallo)catechin, Ftranscis_Ex was epiT, F3pr35 and seed concP where additional QTLs always less than 1 both in skin and seeds. The higher were identified through parental detection (Additional Ftranscis_T in skin conformed to the fact that (+)-cate- file 6). More QTLs and digenic pairwise interactions chin was the predominant terminal subunits/monomer were identified on the consensus map than on parental in skin. (Figure 2B). Means of each PA variable mea- maps, allowing some QTL models to explain more than sured in both skin and seeds were systematically differ- 80% of the BLUP variance in consensus mapping (Addi- ent (paired t-test, P < 0.001, data not shown). tional file 6), as illustrated in the case of epiT in seeds For S × G population, average H of PA variables was (Figure 5D). Some loci were involved in phenotypic var- 0.56 (from 0.24 to 0.82) and 0.44 (from 0.26 to 0.54) in iation almost exclusively through digenic epistasis such skin and seeds, respectively. No significant difference in as locus 10@32 for seed concB or locus 14@16.0 for H magnitude was detected between skin and seeds (t- seed epiT (Figure 5). Loci were mainly of additive effect test, P = 0.053). Nevertheless, higher H were observed while dominance was predominant at some loci for for skin variables, especially catT and mDP (0.76 and concK, epiEx, mDP and Ftranscis_T in skin and galEx, 0.82, respectively, Additional file 5). A high H value epiT, Ftranscis_T and Ftranscis_all in seeds (Figure 4). was also found for these two traits in CC (0.86 and 0.72 Among all detected QTLs, only 10 main effect loci over- for catT and mDP, respectively, Additional file 5). lapped for the same variable in both tissues: 1 for concB, PA variable correlation 2for concK, 1for epiEx, 1for galEx, 2 for catT, 1for We performed PA variable correlation on genotypic epiT, 1 for mDP and 1 for Ftranscis_T (see Figure 5 for values (BLUPs) from S × G population because the two- some examples). Parental alleles contribution to these year data available both in skin and seeds allowed us to common loci was not always consistent across tissues work with a much reduced environmental effect. All (Additional file 6), which could be an indication of tis- three total content variables were highly correlated sue-specific genetic mechanisms. Different genetic archi- within a given tissue while significant correlations tectures were observed for the same PA variable between tissues were only observed for concB and between berry skin and seeds as illustrated in Figure 5: concK (Figure 3). Among subunit percentage variables few moderate QTLs (< 3) or no QTL in skin vs several in skin (Figure 3), the most noticeable features were the (> 5) small to moderate QTLs with possible involvement significant negative correlation between egcEx (B-ring of epistasis in seeds (for concP, concB, catEx, galEx, tri-hydroxylated subunit) and all other units (B-ring di- Ftranscis_Ex and Ftranscis_T, illustrated by concB in hydoxylated subunits) and the positive correlation Figure 5A); many QTLs with involvement of epistasis in between (-)-epicatechin and (+)-catechin either in exten- skin vs a small number (2) of main effect QTLs in seeds (epiEx in Figure 5B); a major QTL (R sion position (Ex) or in terminal/monomer position (T). > 50%) and some In seeds, the most noticeable feature was the significant QTLs of moderate effect in skin v.s.manyQTLsof correlation of epiT with all other variables (negative with small to moderate effect in seeds (for catT and Ftransci- extension units and positive with terminal units) while s_all, illustrated by catT in Figure 5C), or few moderate epiEx was negatively correlated with other subunits. For QTLs in skin vs many moderate QTLs in the presence the same subunit in a given tissue, no highly significant of a QTL of large effect and epistasis in seeds (epiT, Fig- correlation was observed between extension and terminal ure 5D). Conversely, similar genetic architecture position, except the negative correlation between epiEx between skin and seeds was observed for concK (only and epiT in seeds (P < 0.001). Significant correlation moderate main effect QTLs, Figure 5E) and mDP (a between subunit percentage variable and composite vari- major QTL and a few QTLs of moderate effect, Figure ables inside a tissue reflected the variable construction 5F). Details regarding position, major allelic effect, LOD (Table 1). Between tissues, significant positive correlations score, LOD-1 confidence interval and percentage of were observed for concK, most of terminal subunits/ explained variation (R ) for each QTL are given in Addi- monomers pairs, galEx and also mDP and Ftranscis_all. tional file 6. PA total content QTL analysis In skin, 1, 1 and 3 QTLs were identified on the consen- Global features of PA QTLs sus map for concP, concB and concK, respectively. One We performed QTL detection with genotypic BLUPs additional QTL for concP was identified through paren- both on consensus and parental maps. In total, 103 vs tal detection on the Syrah map. Conversely, for concK, 43 QTLs and 24 vs 2 digenic epistatic interactions were all QTLs exhibited a major Grenache allelic effect and identified on the consensus map for seed and skin PA one additional QTL was identified on chromosome 9 on variables, respectively (Figure 4). QTLs detected on par- the Grenache map. ental maps were generally also detected on the consen- In seeds, 6, 10 and 5 QTLs were identified for concP, sus map except for skin concP, concK, catEx, galEx, concB, and concK, respectively. The locus positioned at Huang et al. BMC Plant Biology 2012, 12:30 Page 8 of 19 http://www.biomedcentral.com/1471-2229/12/30 skin seed 1.0 0.7 0.7 0.1 -0.2 0.0 0.1 0.1 -0.2 0.0 0.1 0.3 0.1 -0.1 -0.1 0.1 0.1 0.1 -0.1 0.1 -0.2 0.0 0.1 0.1 0.1 -0.2 -0.1 concP 1.0 0.7 0.1 -0.1 -0.1 0.1 -0.1 -0.2 0.1 0.1 0.1 0.0 -0.1 0.0 0.2 -0.1 0.1 0.0 0.1 -0.1 -0.1 -0.1 0.1 0.0 0.0 0.0 concB concK 1.0 0.1 -0.2 0.1 0.1 0.1 -0.1 0.0 0.1 0.2 0.1 -0.1 0.1 0.0 0.3 0.1 -0.1 0.1 0.0 -0.1 0.1 0.0 0.1 0.0 0.0 catEx 1.0 0.3 0.1 -0.4 0.2 0.0 -0.2 1.0 0.1 0.5 0.4 -0.1 -0.1 0.0 0.2 -0.1 0.1 0.0 -0.1 0.0 0.1 0.2 0.1 0.1 epiEx 1.0 0.2 -1.0 0.1 0.2 -0.1 0.3 -0.2 0.2 0.9 0.0 -0.1 -0.1 0.0 -0.2 0.1 0.1 -0.1 0.0 0.0 0.0 0.2 0.1 galEx 1.0 -0.4 0.2 0.1 -0.1 0.2 0.0 0.2 0.3 0.0 0.1 0.3 0.0 -0.2 0.3 -0.1 -0.1 0.3 0.1 0.0 -0.1 -0.1 1.0 -0.2 -0.3 0.2 -0.4 0.1 -0.4 -1.0 0.0 0.1 0.0 0.0 0.2 -0.2 -0.1 0.1 -0.1 0.0 -0.1 -0.1 -0.1 skin egcEx 1.0 0.6 -0.9 0.2 0.4 0.9 0.2 -0.1 -0.1 0.1 0.0 -0.4 -0.2 0.3 0.4 0.5 -0.5 0.1 0.0 0.3 catT 1.0 -0.6 0.1 -0.3 0.5 0.3 0.0 0.0 0.2 0.0 -0.3 -0.2 0.3 0.2 0.3 -0.3 0.1 0.1 0.3 epiT 1.0 -0.2 -0.3 -0.9 -0.2 0.1 0.2 -0.1 0.1 0.4 0.3 -0.3 -0.4 -0.4 0.5 -0.1 0.1 -0.3 mDP Ftranscis_Ex 1.0 0.1 0.6 0.4 -0.1 -0.1 0.0 0.2 -0.1 0.1 0.0 -0.1 0.0 0.0 0.2 0.1 0.1 Ftranscis_T 1.0 0.4 -0.1 -0.1 -0.2 -0.1 0.0 -0.1 -0.1 0.0 0.2 0.2 -0.2 0.0 -0.2 0.0 Ftranscis_all 1.0 0.3 -0.1 -0.2 0.1 0.1 -0.3 -0.2 0.3 0.3 0.4 -0.4 0.2 0.0 0.3 F3pr35 1.0 0.0 -0.1 0.0 0.0 -0.2 0.2 0.1 -0.1 0.1 0.0 0.0 0.1 0.1 concP 1.0 0.6 0.5 0.2 0.0 -0.2 0.4 -0.2 -0.1 -0.1 0.2 0.5 0.4 1.0 0.6 0.3 0.0 0.0 0.1 -0.2 -0.1 0.0 0.3 0.3 0.2 concB 1.0 0.3 -0.2 0.0 0.1 -0.1 0.2 -0.1 0.3 0.2 0.3 concK 1.0 -0.1 0.1 -0.1 -0.5 -0.2 0.3 1.0 0.3 0.5 catEx 1.0 -0.2 -0.5 -0.4 -0.5 0.6 -0.2 -0.1 -0.4 epiEx galEx P<0.05 1.0 -0.6 -0.5 0.0 0.5 -0.1 -0.2 -0.5 seed catT P<0.01 1.0 0.4 0.1 -0.8 0.1 0.7 0.8 epiT P<0.001 1.0 0.4 -0.8 -0.4 -0.4 0.0 1.0 -0.4 -0.1 -0.4 0.0 galT Self-correlation mDP 1.0 0.1 -0.1 -0.5 1.0 0.3 0.6 Ftranscis_Ex 1.0 0.7 Ftranscis_T 1.0 Ftranscis_all Figure 3 PA variable correlation based on genotypic BLUP of S × G population. The Pearson pairwise correlation coefficient (r) is shown and colour codes give the significance of correlation tests. Skin variables are indicated in pink and seed variables are indicated in blue. The bold black lines delimit the pairwise correlation inside a tissue for a given variable category, i.e. total content, subunit percentage and composite variables. The bold green lines delimit the pairwise correlation between tissues for a given variable category. Simple variables: percentage of constitutive units 40-50 cM on chromosome 2 was identified for all three In skin, 1, 9, 3 and 5 QTLs respectively were identified on variables. Epistasis was strongly involved in genetic architecture of concB and in total accounted for around consensus map for catEx, epiEx, galEx, and egcEx, with 30% of the BLUP variance (Figure 1). several overlapping QTLs (Figure 4). For catEx, two addi- In summary, fewer total content QTLs were detected in tional QTLs on chromosomes 14 and 18, were specifically skin than in seeds. Common loci between skin and seeds identified on the Grenache map. Pairwise interactions for concB and concK were identified on chromosomes 8, were identified for epiEx. For terminal subunits/mono- 13, and 17. For each tissue, one locus was identified to mers, 2 and 4 consensus QTLs were detected for epiT and be common to the three total content variables: the QTL catT, respectively. These two traits had co-locating loci on on chromosome 8 for skin and the QTL on chromosome chromosomes 8 and 17 with an especially large R for the 2 for seeds. locus on chromosome 17 (55.8% for catT). concP concB concK catEx epiEx galEx egcEx catT epiT mDP Ftranscis_Ex Ftranscis_T Ftranscis_all F3pr35 concP concB concK catEx epiEx galEx catT epiT galT mDP Ftranscis_Ex Ftranscis_T Ftranscis_all Huang et al. BMC Plant Biology 2012, 12:30 Page 9 of 19 http://www.biomedcentral.com/1471-2229/12/30 14 16 12 3 4 56 7 8 9 10 11 12 13 15 17 181920 21 concP skin concB concK concP concB seed concK 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 catEx epiEx galEx skin egcEx catT epiT catEx epiEx galEx seed catT epiT galT 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 mDP Ftranscis_Ex skin Ftranscis_T Ftranscis_all F3pr35 mDP Ftranscis_Ex seed Ftranscis_T Ftranscis_all 20 cM 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 As Ag D Digenic interaction Figure 4 Overview of skin and seed PA QTLs identified on the consensus map for total content (A), subunit percentage (B) and composite variables (C). For each variable category, two panels are shown: the upper one for QTLs in skin and the lower one for seeds. The x- axis of each panel spans the whole genome where chromosome sizes are proportional to genetic distance of consensus map and the chromosome numbers are indicated under the x-axis of lower panels. QTLs are indicated by horizontal lines with width corresponding to LOD-1 confidence interval. As, Ag and D respectively indicate additive effect from Syrah alleles, additive effect from Grenache alleles and dominance effect which were estimated according to [41]. Color codes correspond to major effects for each QTL, estimated as (|As| or |Ag| or|D|)/(|As|+|Ag|+| D|) > 0.30. Triangles indicate loci involved in digenic pairwise interactions. Grape candidate genes for PA synthesis are indicated on the upper black line of (A) where bar size is proportional to the flanking marker interval of the gene. Green bars are for genes coding for synthetic enzymes while red bars are for genes coding for transcription factors. The number above the flanking marker interval indicates the corresponding candidate gene: 1, VvLAR1 (leucoanthocyanidin reductase) [14]; 2, VvLDOX (leucoanthocyanidin dioxygenase) [17]; 3, VvF3H (flavanone 3-hydroxylase) [17]; 4, VvMYB5b [47]; 5, VvC4H (cinnamate 4-hydroxylase); 6, VvF3’5’Hs (flavonoid 3’-5’ hydroxylases) [18-20]; 7, VvMYC1 [48], 8, VvPAL (phenylalanine ammonia-lyse) [17]; 9, VvMYB5a [49]; 10, VvMYBPA2 [50]; 11, VvCHIs (chalcone isomerases) [17,21]; 12, VvCHS (chalcone synthase) [17]; 13, VvWDR2 [51]; 14, VvMYBPA1 [52]; 15, VvMYCA1 [51]; 16, VvPAL (phenylalanine ammonia-lyse) [17]; 17, Vv4CL (4- coumaroyl CoA ligase); 18, VvWDR1 [51]; 19, VvLAR2 (leucoanthocyanidin reductase) [14]; 20, VvF3’Hs (flavonoid 3’-hydroxylases) [18-20], 21, VvDFR (dihydroflavonol reductase) [17]. Detailed genetic maps with marker names are available in Additional file 7. Huang et al. BMC Plant Biology 2012, 12:30 Page 10 of 19 http://www.biomedcentral.com/1471-2229/12/30 A concB B epiEx 60 60 60 60 Global R2=75.9% Global R²=79.79% Global R²=21.9% Global R2=16.1% 50 50 50 40 40 40 40 R²30 R² R² 30 30 R²30 20 20 20 10 10 10 10 0 0 0 C D catT epiT 60 60 60 Global R2=66.57% Global R2=71.5% Global R2=27.8% Global R2=87.71% 50 50 50 40 40 40 R² 30 30 R² 30 R² R²30 20 20 20 10 10 10 0 0 0 concK mDP E F 60 60 60 Global R2=70.1% Global R2=27.9% Global R²=53.48% Global R²=48.53% 50 50 50 50 40 40 40 R²30 30 R² 30 30 R² R² 20 20 20 10 10 10 10 0 0 0 Figure 5 R distribution of skin and seed PA QTLs identified on the consensus map for concB (A), epiEx (B), catT (C), epiT (D), concK 2 2 (E) and mDP (F). R of main effect QTLs (solid bar) and R of digenic epistatic interaction (hatched bars) are sorted according to their magnitude. Skin variables are indicated in darkpink and seed variables in blue. Locus names are indicated on the x-axis and should be read as chromosome@position_on_the_chromosome. Locus names are highlighted in pink for loci identified in both skin and seed for the same variable; we considered loci as “common” loci when their LOD-1 confidence interval overlapped. Loci involved in digenic epistasis are indicated by a dark dot under the locus names for which R was estimated without inclusion of the associated interaction. In seeds, 2 QTLs for epiEx were identified on the con- subunits of the same nature but with different positions sensus map while two additional loci on chromosomes 8 in the polymer (e.g. epiEx and epiT). The locus posi- and 12 were identified solely on Grenache and Syrah tioned at approx. 7 cM on chromosome 17 was identi- maps, respectively (Figure 4, Additional file 6). For all fied for all PA simple variables in both tissues except for other simple variables, at least 7 QTLs and 1 pairwise egcEx, galEx and catEx in skin. interaction were involved in multiple-QTL models on the Composite ratio variables consensus map. For galEx, additional loci were identified In skin, the best QTL model for mDP, Ftranscis_T and on chromosomes 3, 5 and 9 through parental detection. Ftranscis_all included only a few main effect QTLs (2 to Joint consideration of the results obtained from both 4 QTL, Additional file 6) without digenic interaction. tissues showed that different QTLs were identified for The major locus on chromosome 17 was also identified 17@21.9 17@7.0 8@79.1 6@45.0 8@78.0 13@5.0 1@25.0 8@56.0 13@14.0 17@6.0 2@44.0 17@12.0 2@45.0:19@35.0 2@14.0:14@33.0 2@49.0 18@17.0 4@60.0 18@33.0 4@48.0 4@60.0:8@69.0 8@12.0 2@45.0 1@11.0:10@32.0 19@35.0 17@38.0 14@17.0 2@14.0 8@79.1 1@11.0 14@33.0 8@56.0 13@1.0 8@69.0 10@32.0 17@6.0 17@4.0 10@2.0 1@18.0 8@79.1 8@77.0 6@44.0 6@45.0 8@54.0 17@7.0 18@76.0 6@45.0:18@18.4 3@0.0:13@0.0 3@0.0 13@0.0 5@47.0 18@18.4 17@6.0 17@6.0 17@7.0 4@35.0 18@4.0 5@7.0 14@16.0:19@29.0 14@23.0 5@41.0 4@52.0:17@7.0 9@57.0:18@4.0 4@52.0 8@17.0 19@29.0 1@0.0 11@57.3 9@57.0 14@16.0 Huang et al. BMC Plant Biology 2012, 12:30 Page 11 of 19 http://www.biomedcentral.com/1471-2229/12/30 for these 3 variables: in the case of mDP, it explained significantly associated to catT and mDP in skin while more than 50% of total BLUP variance (Figure 5F). Six the confidence interval of the QTLs for these two vari- QTLs were detected for F3pr35 on the consensus map ables overlapped. Conversely, we observed some SNP- while one additional Syrah-specific QTL was identified phenotype associations only in the diversity panel: through parental detections on chromosome 13. For VvLAR1-skin Ftranscis_all, VvMYBPA2-skin concP, Ftranscis_Ex, QTLs were solely identified through par- VvMYBPA2-skin concK, VvMYBPA2-skin mDP, ental detections (1 QTL for Syrah map and 2 QTLs for VvMYBPA2-seed galT, VvCHI1-skin concP, VvMYBPA1- skin epiT and VvMYBPA1-seed Ftranscis_T. Grenache map). In seeds, a three-additive-QTL model was identified for mDP while models with 10 QTLs and one to four digenic Discussion interactions were the best ones for Ftranscis_Ex, Ftrans- PA variation extent compared to previous studies cis_T and Ftranscis_all. In addition, digenic interaction A first characterisation of PA composition in a grape- accounted for about 30% of BLUP variance for Ftransci- vine pseudo-F1 population was provided by Hernandez- s_Ex and Ftranscis_all (Figure 5 and Additional file 6). Jimenez and co-workers [46]. Their population was In summary, different QTLs were identified for the composed of 42 offsprings, derived from a cross same variables, depending on berry tissues. Among all between Syrah and Monastrell. In all tissues, the subunit composite variables, only the large effect QTL for mDP percentage in extension position and Ftranscis-series and Ftranscis_T on chromosome 17 was common to variables of the Syrah × Monastrell population were of a skin and seeds. Comparison of multiple-QTL models magnitude and extent equivalent to those of the present between both tissues showed that more digenic interac- study. More divergent results were observed for 1) epiT, tions were involved in seed variables than in skin which is more abundant in the Syrah × Monastrell variables. population; 2) mDP, which is higher in our study and 3) total content variables, for which the population mean Association analyses on candidate genes and variation extent was three-to two-fold larger in the We positioned 21 known grape PA functional candidate present study than in [46]. Syrah, the common parent, genes on the genetic map using their relative position to behaved similarly in both studies although we observed SSR markers on the grape genome ([11], http://www. 10-fold and two-fold higher total PA contents in our genoscope.cns.fr, Figure 4, see Additional file 7 for the study for skin and seeds, respectively. The difference names of flanking markers of candidate genes). Associa- observed in offsprings may result from the fact that the tion tests were performed for nine functional candidate two populations differed by one parent but environmen- tal differences as well as PA extraction and quantifica- genes, eight of them co-locating with QTLs. Among can- didates, there were both genes encoding flavonoid path- tion methods might also have affected PA variables, as way enzymes and putative regulators. Genes were suggested by the different PA composition of the same partially to totally sequenced (gene coverage from 25 to parental cultivar. Indeed, mixed model fit suggested that 100%), mainly in exons (Table 2). Two models were used year had a major effect on PA-related variables for both for association studies since model comparison showed tissues, which was consistent with a previous study an equivalent fit (Additional file 8): one accounted for where PA content and composition were measured in fixed ancestry structure effect (GLM in TASSEL), the two cultivars for two consecutive years [53]. The large other for both fixed ancestry structure effect and random quantitative variation in PA variables in S × G, of genetic background effect (MLM in TASSEL). four out of equivalent extent in the CC diversity panel, underlines nine genes showed at least one significant association the interest to implement a quantitative genetic with PA variables with consistent results between GLM approach on a F1 population for grape PA studies. and MLM (Table 3). Seventy-eight percent of significant Two studies only have characterised PA composition tests (21 out of 27 tests) were common to GLM and in different grape cultivars [54,55] with at most a 37- MLM while 6 additional associations were only signifi- cultivar sample [54]. Different biochemical analyses did cant with MLM model. The reason for this discrepancy is not allow for result comparison between this latter probably that the adjusted P-value in GLM was estimated study and the present work. Nevertheless, CC in our by taking into account dependence between tests due to work was composed of 141 grapevine cultivars of broad linkage disequilibrium [44] while in MLM, each SNP is geographical origin (from East to West Europe) and was tested under an hypothesis of independence. Association initially defined to maximise the diversity of 50 agro- results were consistent with QTL analyses for following morphological traits [26]. The PA composition variation gene-phenotype pairs: VvLAR1-skin catT and VvLAR1- in the diversity panel provides thus the potential to refine PA QTLs in a population of larger genetic skin mDP (Table 3). In particular, several SNPs in linkage disequilibrium for VvLAR1 (data not shown) were background. Huang et al. BMC Plant Biology 2012, 12:30 Page 12 of 19 http://www.biomedcentral.com/1471-2229/12/30 Table 2 Summary of candidate genes for association tests Sequence (size and localisation) Number of SNPs chr Gene References 5’- Exon Intron 3’- seq/gene 5’- Exon Intron 3’- total QTL UTR UTR size UTR UTR 1 VvLAR1 [14] - 1008 412 65 1420/2980 - 21 9 - 30 Skin: catT, mDP. Seed: concB, galEx, catT, epiT, Ftranscis_Ex, Ftranscis_all 6 VvF3’5’H [19,20] - 1296 50 - 1346/2325 - 4 - - 4 Skin: epiEx, egcEx, catT, mDP, F3pr35, 1.1 Ftranscis_all 6 VvF3’5’H [19,20] - 630 12 24 642/1932 - 3 0 0 3 Skin: epiEx, egcEx, catT, mDP, F3pr35, 2.1 Ftranscis_all 8 VvMYB5a [49] 66 687 382 36 1069/1069 1 12 2 1 16 Skin: concP, concB, concK, epiEx, egcEx, epiT, F3pr35. Seed: concB, catEx, catT, Ftranscis_T 11 VvMYBPA2 [50] 93 855 263 37 1148/1479 - 14 5 - 19 No QTL 13 VvCHI1 [17,21] - 440 216 54 656/1486 - 3 5 1 9 Skin: concK, epiEx. Seed: concK, catT, Ftranscis_Ex, Ftranscis_all 13 VvCHI2 [17,21] - 206 452 46 658/2524 - 1 6 - 7 Skin: concK, epiEx. Seed: concK, catT, Ftranscis_Ex, Ftranscis_all 15 VvMYBPA1 [52] 384 861 87 1 0 948/948 8 11 - - 19 Seed: concP, catEx, Ftranscis_Ex 18 VvDFR [17] - 425 194 111 619/2469 - 1 - 2 3 Skin: egcEx. Seed: concP, concK, epiT Chr, chromosome; seq/gene size, total length of sequenced exons and introns/predicted gene size. Multiple QTL mapping in a pseudo-F1 population for alleles (Additional file 6). Parental detection allowed grape PA composition the identification of smaller additive QTLs because of To our knowledge, this study presents the first QTL a greater power due to more individuals in each geno- analysis on grape PA composition with comparisons typic class compared to consensus detection (i.e., 2 between skin and seeds of grape berry. This is also the and 4 genotypic classes for parental and consensus first work on grape using multiple QTL models taking detection, respectively). On the other hand, QTL into account both main effects and digenic epistasis dur- detection on the consensus map allowed us to estimate ing the mapping procedure. QTL mapping in animals QTL dominance effect s, i.e. the interaction between has shown that epistasis effects are often large enough allelic classes, but not necessarily with the assumption to be detected and thus merit a systematic scan regard- of a dominant-recessive relationship [59]. In the pre- less of population size, although larger populations (> sent work, 9.8% and 30% of QTLs detected in skin and 500 individuals) allow a more powerful epistasis detec- seeds, respectively, had dominance as the major allelic tion [56]. By employing the multiple QTL mapping effect (D in Figure 4 and Additional file 6). For exam- approach, we actually showed the important involve- ple, the locus 8@69 of concB in seed was involved in ment of epistatic interaction in shaping PA composition phenotypic variation almost exclusively through domi- variation; indeed, some loci were involved in phenotypic nance (Additional file 6) and this information would variation almost exclusively through pairwise interaction. have been overlooked if parental mapping only had Our mapping population is of sufficient size (191 indivi- been performed. duals) to allow identification of small effect QTLs. How- ever, one should keep in mind that the R estimate of Genetic architecture of grape PA composition individual QTLs is usually overestimated [57] and may PA total content have a wide confidence interval [58]. Some of the identi- QTL results were consistent with the results of PA vari- fied QTLs may therefore be of smaller effect in reality. able correlation: no significant correlation was observed One should thus be cautious in result interpretation and for concP between tissues and no co-locating QTL was further identification of causal polymorphism although identified for this variable between skin and seeds, while we did check initially the genome-wide first type common QTLs were identified for both concB and error rate. concK across tissues. Since concB and concK take into Allele contribution to individual QTL was mainly account berry size- and yield-related traits, these co- due to additive effects between Syrah and/or Grenache located QTLs for concB and concK may be involved Huang et al. BMC Plant Biology 2012, 12:30 Page 13 of 19 http://www.biomedcentral.com/1471-2229/12/30 Table 3 Results of the association study: significant SNP-phenotype associations along with the co-located QTL Chr Gene Marker Position Syn/Ns Tissue Trait n.obs p.MLM p.adj.GLM QTL 1 VvLAR1 int2687 intron 4 skin catT 112 2.92E-04 9.99E-04 Yes e5-2734 exon 5 Syn skin catT 112 1.22E-05 9.99E-04 Yes e1-82 exon 1 Ns (Ala ↔ Thr) skin mDP 115 7.08E-04 0.025 Yes e1-132 exon 1 Syn skin mDP 113 5.44E-04 9.99E-04 Yes e1-138 exon 1 Syn skin mDP 105 0.0013 0.0021 Yes e1-156 exon 1 Ns (Asn ↔ Lys) skin mDP 111 5.19E-04 9.99E-04 Yes e3-665 exon 3 Syn skin mDP 117 3.41E-04 9.99E-04 Yes e3-734 exon 3 Syn skin mDP 110 5.66E-04 9.99E-04 Yes int2405 intron 3 skin mDP 94 1.85E-04 9.99E-04 Yes e4-2524 exon 4 Syn skin mDP 103 5.58E-04 9.99E-04 Yes int2636 intron 4 skin mDP 104 6.49E-04 9.99E-04 Yes e5-2722 exon 5 Syn skin mDP 107 4.82E-04 9.99E-04 Yes e5-2776 exon 5 Syn skin mDP 107 4.82E-04 9.99E-04 Yes e5-2779 exon 5 Syn skin mDP 107 5.61E-04 9.99E-04 Yes e5-2785 exon 5 Syn skin mDP 107 5.61E-04 9.99E-04 Yes e5-2872 exon 5 Ns (Ile ↔ Met) skin mDP 104 6.37E-04 9.99E-04 Yes e5-2896 exon 5 Syn skin mDP 104 6.37E-04 9.99E-04 Yes e5-2902 exon 5 Syn skin mDP 104 6.37E-04 9.99E-04 Yes e1-156 exon 1 Ns (Asn ↔ Lys) skin Ftranscis_all 109 0.0032 0.0509 No int2687 intron 4 skin Ftranscis_all 114 0.0025 0.042 No e5-2734 exon 5 Syn skin Ftranscis_all 114 1.84E-04 9.99E-04 No 11 VvMYBPA2 intron06Y intron skin concP 117 9.77E-04 9.99E-04 No p19_GA promoter skin concK 55 4.37E-04 9.99E-04 No p18 promoter skin mDP 54 1.76E-04 9.99E-04 No p19_GA promoter skin mDP 55 3.36E-06 9.99E-04 No intron05M intron seed galT 82 0.0015 0.3377 No 1293 W exon 3 Syn seed galT 93 0.0014 0.3067 No 1322 W exon 3 Ns (Leu ↔ His) seed galT 92 0.0013 0.3387 No 1398Y exon 3 Syn seed galT 92 0.0029 0.6773 No 1473Y exon 3 Syn seed galT 93 0.0026 0.5135 No 13 VvCHI1 Y183 exon 4 Syn skin concP 108 0.0025 0.049 No 15 VvMYBPA1 p277R promoter skin epiT 125 0.0018 0.03 No 702W exon 2 Ns (Ser ↔ Thr) seed Ftranscis_T 68 0.0036 0.6783 No p.MLM, p-value from mixed model, p.adj.GLM, adjusted p-value from GLM. Bold cases indicate significant associations in both MLM and GLM results. QTL, the candidate genes were under QTLs of the same PA variables as those associated with SNP. indirectly in PA total content through alteration of berry variables. (Table 3). VvMYBPA2 is mainly expressed in development or yield-related traits. Indeed, several QTLs berry skin at green stage and its overexpression in grape for concK co-located with QTLs for yield-related traits, hairy root significantly increased PA production [50]. especially loci on chromosomes 8, 13, 17 and 18 where The significant associations of VvMYBPA2 were posi- QTLs for yield related and berry size related-trais were tioned in promoter and intron and might be involved also identified in the same S × G population (Doligez et either in transcription level alteration or through linkage al., unpublished data). Unlike these yield-related loci, the disequilibrium with other causal mutations. A minor loci identified for concP, which were also identified for association was identified between a non-synonymous concB and concK could be specific targets for a better polymorphism of VvCHI1 and skin concP (Table 3). understanding of the contrasting PA content in berry VvCHI encodes an upstream enzyme in the grape flavo- compartments. noid pathway. This gene may be involved in PA content Association tests were in accordance with the involve- variation through the control of the flux of intermediate ment of VvMYBPA2 in PA content variation, as sug- substrate. However, precise involvement of VvCHI1 in gested by a previous study [50]. Two VvMYBPA2 SNPs PA content variation needs further genetic and func- were significantly associated with skin PA content tional confirmation. Huang et al. BMC Plant Biology 2012, 12:30 Page 14 of 19 http://www.biomedcentral.com/1471-2229/12/30 Among tested candidate genes, VvMYB5a, located on 3-O-gallate was probably under the control of many chromosome 8, is under several QTLs, especially total genomic regions and digenic epistatic interactions content-related QTLs. A previous physiological study (Figures 4 and Additional file 6). Additional information showed that ectopic expression of VvMYB5a in tobacco was provided by association tests which revealed 4 weak induced expression of flavonoid genes and significantly but significant associations between galT in seeds and increased both PA and anthocyanin production [49]. SNPs of VvMYBPA2. These associated SNPs are located The authors therefore proposed VvMYB5a as an in introns or in C-terminal of the proteins which could upstream regulator of flavonoid pathway. In our work, contain protein-protein interaction domains [66] (Table no significant association was found for VvMYB5a while 2). This result suggests that the associated SNPs might the whole gene was sequenced (Table 2). Further inves- lead to alteration of transcriptional complex recruitment tigations would be necessary to figure out if this gene is or interaction with other proteins.[50].Glucosyltrans- involved in grape PA content variation. ferases were recently identified as putative candidates PA subunit synthesis: the hydroxylation patterns of flavan- involved in the first enzymatic step of PA galloylation 3-ols [67]. Since they are located on chromosome 3 where All flavonoids carry a hydroxyl group at the 4’ position QTLs for galEx in skin and seed are positioned (Figure of B-ring (Figure 1). The flavonoid hydroxylation pat- 4 and Additional file 7), they may be good candidates to terns of B-ring were first studied in ornamental plants be tested by association genetics in the next future. for colour engineering because it is a major colour PA subunit synthesis: the trans- and cis- subunits determinant for anthocyanins, another class of flavo- Synthesis of PA trans-and cis-subunits is tightly related noids sharing a similar C6-C3-C6 skeleton with PA to PA polymerisation since intermediate substrates in monomeric subunits [60,61]. Links between F3’H and the flavonoid pathway are assumed to take up a trans- F3’5’H gene activities and their relative flavan-3-ols are configuration while major extension subunits assume a less obvious due to the lack of easily assessed reporters. cis-configuration (e.g. (-)-epicatechin) [68]. Major Our results for grape skin variables showed that five advances in understanding PA subunits biosynthesis genomic regions (on chromosomes 3, 6, 8, 10 and 18) were made through the isolation of two genes coding harboured co-located QTLs for epiEx, egcEx and for specific enzyme activities for the formation of term- F3pr35. This co-localisation is not surprising since epiEx inal/monomers: 2,3-trans-(gallo)catechin and 2,3-cis-epi and egcEx were the major components for F3pr35 vari- (gallo)catechin [3-5]. Recently, another dynamic view of able construction and were therefore highly correlated the flux between trans-and cis-terminal units/monomers (Figure 2). These loci are probably involved in the flux was provided by Gargouri and co-workers who demon- between di-hydroxylated and tri-hydroxylated PA build- strated the ability of grape ANR to epimerise (+)-cate- ing blocks. An interesting point is that the QTL on chin to (-)-epicatechin [69]. Our results seem to be in chromosome 6 for both egcEx and F3pr35 co-located accordance with this work since a major locus on chro- with a genomic region corresponding to the F3’5’H gene mosome 17 was identified for catT and epiT, the two family. chiral flavan-3-ols, which was also the major locus for However, no significant association was detected Ftranscis_T and Ftranscis_all in skin. This locus further- between the two tested VvF3’5’H isogenes and hydroxy- more co-located with VvLAR2,anisogene of LAR, lation pattern variables in this work. F3’5’H is present as which belongs to the Reductase-Epimerase-Dehydrogen- a multigenic family in the grapevine genome in which at ase (RED) family, as ANR, and thus might display both least 15 isogenes have been identified [62]. For dupli- epimerase and reductase activity. Similarly, three cated genes, neofunctionalisation and/or subfunctionali- VvLAR1 SNPs were significantly associated to Ftransci- sation could conduct to specialisation of each isoform in s_all in skin and therefore merit further functional a spatio-temporal manner [63-65]. Actually, the isogene investigation to understand its involvement in in vivo VvF3’5’H1.1 (or VvF3’5’Hn in [62]) was shown to be PA subunit synthesis. expressed only in vegetative organs, while VvF3’5’H2.1 On the other hand, the origin of extension subunits is (or VvF3’5’Hf in [62]) is expressed in berry skin. The still uncertain: are extension subunits derived from assessment of the polymorphisms of all isogenes may intermediate substrates in the pathway or from end pro- give more insights for links between F3’5’H and hydro- ducts such as (-)-epicatechin and (+)-catechin? [1,68,70]. xylation variation. In our work, significant correlation was not systemati- PA subunit synthesis: the galloylated flavan-3-ols cally observed for subunits of the same nature but dif- To date, the underlying genetic determinism for the fering in position in the polymer, and few QTLs co- production of PA galloylated building blocks is still located. In addition, the QTLs for flux between trans- unclear. Our results in both grape berry skin and seeds and cis-subunits were most often different between extension position (Ftranscis_Ex) and terminal subunits/ showed that the quantitative variation of (-)-epicatechin- Huang et al. BMC Plant Biology 2012, 12:30 Page 15 of 19 http://www.biomedcentral.com/1471-2229/12/30 monomes (Ftranscis_T). All these results argue in favour differences between tissues were also observed. Globally, of the involvement of different loci in PA building for the same PA subunit percentage variables in both blocks synthesis and in the control of flux between tissues, only 5 QTLs among 74 had overlapping intervals trans-and cis-subunits according to their position in the between skin and seeds. Another contrasting feature was polymer. Stafford et al. already suggested from radioac- observed: in skin, 54% of all QTLs accounted for Syrah tive labelling experiments that upper and lower units additive effect and 78% for Grenache effect whereas in arise from different steps of the pathway rather than seeds, 74% of all QTLs accounted for Syrah additive from the condensation of similar units [71]. effect and 53% for Grenache effect (Additional file 6 and PA polymerisation Figure 4). Even for loci identified in both tissues for a An aspect of the debate about PA polymerisation con- given variable, major allelic effect and R differed (e.g. cerns the enzymatic or nonenzymatic polymerisation the QTL in chromosome 17 for catT, Additional file 6). (reviewed by [1,70]). The existence of a polymerase is Our results suggest that seed PA variation is controlled supported by the barley PA mutant ant26, containing by QTLs of moderate and equivalent magnitude with amounts of (+)-catechin equivalent to wild-type content involvement of epistasis. On the other hand, skin vari- but only trace amounts of PAs [72]. On the other hand, ables are mainly under the genetic control of a few large in vitro chemical synthesis of PAs has also been effect loci with a fluctuating variance unexplained by reported [73] and these authors observed a modulation QTLs. of polymer size through the modulation of the relative The different genetic architectures between tissues amounts of extension unit intermediates and monomers. could result from divergent functional evolution of PAs One can thus hypothesise that instead of a polymerase, in these two berry compartments. For fruits in general, the ant26 mutation could directly affect the suitable ensuring protection of the embryos is essential. Because conditions for spontaneous PA polymerisation, such as of their abundance and their ability to protect plants appropriate pH [70]. Further investigation of the QTLs against biotic stresses, PAs and flavan-3-ols might be identified in this study would bring more insights into the major molecules involved in grape embryo protec- this polymerisation issue. Indeed, in the case of skin tion. In fact, their influence in maintaining seed dor- mDP, H was high (0.82) and the multiple QTL model mancy has been demonstrated in Arabidopsis [75] and accounted for 70% of the genotypic variance, corre- their interaction with phytohormones has also been sponding to 57% of the total phenotypic variance. The reported [76,77]. Therefore, to prevent biological fluc- largest QTL on chromosome 17 explained alone 55% of tuationdueto asinglepolymorphism mutation, a net- genotypic variance and was also the major locus for work with multiple cross-talking actors as a product of evolution without human selection could be postulated seed mDP (Figure 5F). This QTL is therefore an inter- esting target for mDP genetic mechanism investigation. in the case of seed PAs, as suggested by the identifica- Another mDP QTL located on chromosome 1 might tion of numerous small effect QTLs and the involve- also be an interesting target for understanding PA poly- ment of epistasis. Conversely, skin is the first protective merisation. This QTL co-located with a gene encoding a barrier of the grape berry against its environment. In PA-specific synthetic enzyme, VvLAR1, for which several plants, PAs are thought to be involved in self-defence SNPs in linkage disequilibrium (data not presented) mechanisms [1]. For berry consumers, skin PAs confer were significantly associated to mDP and catT in skin, flavour to berries and are also responsible for major consistent with the corresponding QTL. Interestingly, organoleptic qualities of wine. Consumers in turn could VvLAR1 is highly polymorphic in Grenache while almost help the plant in seed dispersion or vegetative propaga- homozygous in Syrah (1 SNP in the coding region, data tion. However, high quantities of PAs would confer to not presented), in accordance with the fact that the berries too much astringency and bitterness, which QTLfor mDP inthisregionismainly due to aGre- would lead consumers (specially humans) to reject nache allelic effect. grapevines producing such berries for direct consump- tion (or wine-making). Human selection in particular Tissue-specific genetic architecture for PA composition could therefore have narrowed down the genetic basis In accordance with a previous study which demon- of skin PAs over time and consequently led to a specific strated the tissue specificity of transcriptional profiles in genetic architecture with a few large effect QTLs for grape berry [74], the present work illustrates different skin PA variables. genetic mechanisms for grape PA composition between skin and seeds: QTLs differed in terms of number, posi- QTL mapping and association analysis as complementary tion, R and allelic effects. For total content variables, approaches for candidate locus identification the major QTLs differed in skin and seeds. For sub- In this work, we used both QTL mapping and associa- unit percentage and composite variables, important tion analyses to identify phenotype-marker associations. Huang et al. BMC Plant Biology 2012, 12:30 Page 16 of 19 http://www.biomedcentral.com/1471-2229/12/30 For identification of phenotype-associated markers, G population. For each PA variable, 4 panels are shown: distribution of grapevine segregating populations have a greater diver- residuals of the best fitted model (box-and-whisker plot, topleft), quantile-quantile plot of model residuals against a theoretical normal sity than populations derived from inbred lines due to distribution (topright), distribution of BLUPs of the best fitted model heterozygous parental cultivars. However, their genetic (box-and-whisker plot, bottomleft), quantile-quantile plot of BLUPs background remains relatively narrow compared to against a theoretical normal distribution (bottomright). diversity panels. On the other hand, QTL mapping may Additional file 3: Phenotypic data analysis and best fitted models for variance component estimation. Analysis method and effects reveal associations undetected in diversity panels due to included in the best fitted model. low allelic frequency. The inconsistency between both Additional file 4: Effect of minor genotypic frequency and non- approaches sometimes encountered in this work could normalty of observed phenotype on the association test. Two therefore result from the fact that the available genetic sections are in this file. 1. Test for the enrichment of low frequency polymorphisms among associated markers. 2. Test for the effect of the polymorphisms were different between the two popula- non-normalty of the trait in the association tests. tions: causal polymorphisms in one population might be Additional file 5: Summary of PA variable distributions and broad monomorphic in the other. In addition, our analysis sense heritability (H ) in S × G and CC populations. Two tables focused on genes of known function co-locating with inside: Table A, summary of S × G population; Table B, summary of CC population. Skin data were collected in 2005 and 2006 and seed data QTL while other candidates could underlie QTL inter- were collected in 2006 and 2007. Parental values are indicated as mean vals. Besides time-consuming fine mapping, candidate ± standard error. Broad sense heritability (H ) was estimated based on genes can also be selected by combining QTL results the best fitted model as the percentage of phenotypic variance explained by the genotypic variance. with other data such as transcriptomics. Nevertheless, Additional file 6: QTL summary for consensus and parental LAR1 gene evoked a particular interest through associa- detection. Summary for consensus detection and parental detection are tion test. Complementing functional studies performed in two separate sheets. Term: main effect QTLs and pairwise epistatic on a single cultivar [14], we provide here additional con- interactions. Main effect QTLs are indicated by chromosome@position of LOD peak while interaction terms are indicated by “:” linking main effect firmation of LAR1 gene involvement in grape PA com- QTLs. Map (for QTL summary of parental detection): the parental map position through a diversity panel study. used for QTL detection. LOD score and R were estimated by dropping the considered term from the full model. For loci involved in pairwise interaction, their LOD and R were estimated by dropping both the main Conclusions effect and the associated interaction effect. df: degree of freedom The present work confirmed presumptions about the dropped for QTL effect estimation. Type III SS: type III sum of squares. CI: complex genetic architecture of PA composition in LOD-1 confidence interval. For consensus detection, As, Ag and D indicate additive effect from Syrah alleles, additive effect from Grenache grape berries. QTLs for PA total content, PA building alleles and dominance effect which were estimated according to Segura blocks, degree of polymerisation and ratio between et al. [41]: As: 1/4[(μ +μ )-(μ +μ )], Ag: 1/4[(μ +μ )-(μ +μ )], D: 1/4 ad ac bd bc ac bc ad bd building blocks were identified. Berry PA composition [(μ +μ )-(μ +μ )], where μ , μ and μ are phenotypic means for ac bd bc ad bd bc ad corresponding genotypes relative to phenotypic mean of ac genotype. offers a case study for tissue-specific genetic architec- Thus here, μ = 0. effect column in QTL summary for consensus ac ture: in skin, the same major loci were involved in sev- detection indicates major effects of the considered locus involved in eral PA variables while multiple and moderate QTLs phenotype variation, satisfying the following condition: (|As| or |Ag| or | D|)/(|As|+|Ag|+|D|) > 0.30. “Estimate” column in QTL summary for parental with strong epistasis were the principal genetic factors detection gives the allelic effect between parental alleles or the for seed PA composition. These differences might be corresponding interaction effect, the sign is arbitrary. The type III sum of due to human selection on skin PA leading to a reduced squares, df, LOD and R of the QTL model are indicated at the last row of each variables, highlighted on yellow. The loci identified in both skin diversity of related genes while a multi-factor network and seeds for the same variable (i.e. having overlapping LOD-1 controlling seed PA synthesis would be necessary to confidence interval) are highlighted in pink. protect grapevine embryos. Association tests confirmed Additional file 7: QTL maps with positioned grape PA candidate the interest of VvLAR1 as a candidate gene in modulat- genes. Parental and consensus maps are presented in parallel: left, Syrah map, indicated by S; centre, consensus map, indicated by C and right, ing catT and mDP in berry skin, as well as VvMYBPA2 Grenache map, indicated by G. QTLs are presented as vertical lines at the for total content and probably for subunit galloylation. right side of each map: the line length corresponds to LOD-1 confidence This study provides the first assessment of genetic interval and the LOD peak is indicated by a small horizontal bar in the confidence interval. Known candidate genes are positioned between mechanism underlying grape PA composition which flanking markers, indicated by red-filled bars, according to 12X grape opens doors for further PA genetic studies. genome sequence (http://www.genoscope.cns.fr). Red-filled bars indicated flanking marker interval of regulatory genes and green-filled bars for enzyme-coding genes. Additional material Additional file 8: Model comparison prior to association analyses. -2lnlikelihood is shown. Model comparison was performed by likelihood Additional file 1: Primers used for the amplification and sequencing ratio comparing each model to the most complete model. Significance of the candidate genes. was assessed using the c distribution with degree of freedom as the Additional file 2: Distribution of residuals and BLUPs and Quantile- difference in the number of parameter between two models. quantile plot of residual and BLUPs of the best fitted model for S × Significance level is indicated as *, P < 0.05, **, P < 0.01, ***, P < 0.001. Huang et al. BMC Plant Biology 2012, 12:30 Page 17 of 19 http://www.biomedcentral.com/1471-2229/12/30 7. Caldas G, Blair M: Inheritance of seed condensed tannins and their Abbreviations relationship with seed-coat color and pattern genes in common bean ANR: Anthocyanidin reductase; BLUP: Best linear unbiased predictor; 4CL: 4- (Phaseolus vulgaris L.). TAG Theor Appl Genet 2009, 119(1):131-142. coumaroyl CoA ligase; C4H: Cinnamate 4-hydroxylase; catEx: (+)-catechin 8. Dai GH, Andary C, Mondolot-Cosson L, Boubals D: Involvement of phenolic extension subunits; catT: (+)-catechin terminal subunits/monomers; CHI: compounds in the resistance of grapevine callus to downy mildew Chalcone isomerase; CHS: Chalcone synthasel; concP: Total content in mg/g (Plasmopara viticol). Eur J Plant Pathol 1995, 101(5):541-547. fresh weight; concB: Total content in mg/berry; concK: Total content in mg/ 9. del Río JA, Gómez P, Báidez A, Fuster MD, Ortuño A, Frías V: Phenolic kg berries; DFR: Dihydroflavonol reductase; egcEx: (-)-epigallocatechin compounds have a role in the defence mechanism protecting grapevine extension units; epiEx: (-)-epicatechin extension subunits; epiT: (-)-epicatechin against the fungi involved in Petri disease. Phytopathologia Mediterranea terminal subunits/monomers; F3H: Flavanone 3-hydroxylase; F3’H: Flavonoid 2004, 43(1):87-94. 3’ hydroxylase; F3’5’H: Flavonoid 3’-5’ hydroxylase; galEx: (-)-epicatechin-3-O- 10. Arnold RA, Noble AC, Singleton VL: Bitterness and astringency of phenolic gallate extension subunits; galT: (-)-epicatechin-3-O-gallate terminal subunits/ monomers; LOD: Logarithm of odds; LDOX: Leucoanthocyanidin fractions in wine. J Agric Food Chem 1980, 28(3):675-678. dioxygenase; LAR: Leucoanthocyanidin reductase; PA: Proanthocyanidin; PAL: 11. Jaillon O, Aury JM, Noel B, Policriti A, Clepet C, Casagrande A, Choisne N, Phenylalanine ammonia-lyse; QTL: Quantitative trait locus; SNP: Single Aubourg S, Vitulo N, Jubin C, et al: The grapevine genome sequence nucleotide polymorphism; SSR: Simple sequence repeat. suggests ancestral hexaploidization in major angiosperm phyla. Nature 2007, 449(7161):463-U465. Acknowledgements 12. Velasco R, Zharkikh A, Troggio M, Cartwright DA, Cestaro A, Pruss D, We thank A.-A. Saïdou for the helpful discussions and kind assistance in Pindo M, FitzGerald LM, Vezzulli S, Reid J, et al: A high quality draft association model selection. We thank all the research team members for consensus sequence of the genome of a heterozygous grapevine their participation in sample preparation. We thank M. Farnos and P. variety. PLoS One 2007, 2(12):e1326. Ortigosa for collecting diversity panel samples and the staff of Domaine du 13. Souquet J-M, Labarbe B, Le Guernevé C, Cheynier V, Moutounet M: Chapitre and Domaine de Vassal for grapevine cultivation. We thank three Phenolic composition of grape stems. J Agric Food Chem 2000, anonymous reviewers for helpful suggestions and P. Chatelet for helpful 48(4):1076-1080. English revision. This work was funded in part by the European project, 14. Bogs J, Downey MO, Harvey JS, Ashton AR, Tanner GJ, Robinson SP: FLAVO (no. 513960) and a Ph.D. grant for YFH from INRA and Languedoc- Proanthocyanidin synthesis and expression of genes encoding Roussillon Region. leucoanthocyanidin reductase and anthocyanidin reductase in developing grape berries and grapevine leaves. Plant Physiol 2005, Author details 139(2):652-663. 1 2 UMR AGAP, INRA, 2, place Viala, 34060 Montpellier, France. INRA, UMR1083 15. Souquet J-M, Cheynier V, Brossaud F, Moutounet M: Polymeric 3 ® SPO, 2, place, Viala, 34060 Montpellier, France. UMT Geno-Vigne ®, IFV, 2, proanthocyanidins from grape skins. Phytochemistry 1996, 43(2):509-512. place Viala, 34060 Montpellier, France. UMR Génomique Végétale, INRA 16. Prieur C, Rigaud J, Cheynier V, Moutounet M: Oligomeric and polymeric UEVE ERL CNRS, 2, rue Gaston Crémieux, 91057 Evry, France. Department of procyanidins from grape seeds. Phytochemistry 1994, 36:781-784. Ecology and Evolutionary Biology, Brown University, 80 Waterman Street, 17. Sparvoli F, Martin C, Scienza A, Gavazzi G, Tonelli C: Cloning and molecular Box G-W, Providence, RI 02912, USA. analysis of structural genes involved in flavonoid and stilbene biosynthesis in grape (Vitis-Vinifera L). Plant Mol Biol 1994, 24(5):743-755. Authors’ contributions 18. Bogs J, Ebadi A, McDavid D, Robinson SP: Identification of the flavonoid YFH carried out the gene sequencing and alignment, performed data hydroxylases from grapevine and their regulation during fruit analyses, prepared tables and figures and drafted the manuscript. AD development. Plant Physiol 2006, 140(1):279-291. checked phenotyping and genotyping data, performed linkage map 19. Castellarin SD, Di Gaspero G, Marconi R, Nonis A, Peterlunger E, Paillard S, construction and first statistical analysis and participated in manuscript Adam-Blondon AF, Testolin R: Colour variation in red grapevines (Vitis preparation. AFL participated in gene sequencing, sequence alignment, vinifera L.): genomic organisation, expression of flavonoid 3’- association test analysis and in manuscript preparation. LLC participated in hydroxylase, flavonoid 3’,5’-hydroxylase genes and related metabolite upstream data analyses, data interpretation and in manuscript preparation. profiling of red cyanidin-/blue delphinidin-based anthocyanins in berry YB conducted field experimentation and sample collection. AC performed skin. Bmc Genomics 2006, 7:12. genotyping for linkage map construction. FV, VM, CM, and JMS carried out 20. Jeong ST, Goto-Yamamoto N, Hashizume K, Esaka M: Expression of the biochemical analysis. NT directed PA variable conception, interpreted the flavonoid 3’-hydroxylase and flavonoid 3’,5’-hydroxylase genes and result and participated in manuscript preparation. VC and PT conceived the flavonoid composition in grape (Vitis vinifera). Plant Sci 2006, 170(1):61-69. study, participated in its design, coordination, data interpretation and 21. Jeong ST, Goto-Yamamoto N, Kobayashi S, Esaka A: Effects of plant manuscript preparation. All authors read and approved the final manuscript. hormones and shading on the accumulation of anthocyanins and the expression of anthocyanin biosynthetic genes in grape berry skins. Plant Received: 13 July 2011 Accepted: 27 February 2012 Sci 2004, 167(2):247-252. Published: 27 February 2012 22. Lynch M, Walsh B: Genetics and analysis of quantitative traits Sunderland: Sinauer Associates; 1998. 23. Kao C, Zeng Z, Teasdale R: Multiple interval mapping for quantitative trait Dixon RA, Xie DY, Sharma SB: Proanthocyanidins–a final frontier in fla- Loci. Genetics 1999, 152:1203-1216. vonoid research? New Phytol 2005, 165(1):9-28. 24. Buckler ES, Thornsberry JM: Plant molecular diversity and applications to 2. Aron PM, Kennedy JA: Flavan-3-ols: Nature, occurrence and biological genomics. Curr Opin Plant Biol 2002, 5(2):107-111. activity. Mol Nutr Food Res 2008, 52(1):79-104. 25. Fournier-Level A, Le Cunff L, Gomez C, Doligez A, Ageorges A, Roux C, 3. Tanner GJ, Francki KT, Abrahams S, Watson JM, Larkin PJ, Ashton AR: Bertrand Y, Souquet J-M, Cheynier V, This P: Quantitative genetic bases of Proanthocyanidin biosynthesis in plants–Purification of legume anthocyanin variation in grape (Vitis vinifera L. ssp. sativa) berry: a leucoanthocyanidin reductase and molecular cloning of its cDNA. J Biol quantitative trait locus to quantitative trait nucleotide integrated study. Chem 2003, 278(34):31647-31656. Genetics 2009, 183(3):1127-1139. 4. Devic M, Guilleminot J, Debeaujon I, Bechtold N, Bensaude E, Koornneef M, 26. Barnaud A, Lacombe T, Doligez A: Linkage disequilibrium in cultivated Pelletier G, Delseny M: The BANYULS gene encodes a DFR-like protein grapevine, Vitis vinifera L. Theor Appl Genet 2006, 112(4):708-716. and is a marker of early seed coat development. Plant J 1999, 27. Doligez A, Adam-Blondon A, Cipriani G, Di Gaspero G, Laucou V, 19(4):387-398. Merdinoglu D, Meredith C, Riaz S, Roux C, This P: An integrated SSR map 5. Xie DY, Sharma SB, Paiva NL, Ferreira D, Dixon RA: Role of anthocyanidin of grapevine based on five mapping populations. TAG Theor Appl Genet reductase, encoded by BANYULS in plant flavonoid biosynthesis. Science 2006, 113(3):369-382. 2003, 299(5605):396-399. 28. Adam-Blondon AF, Roux C, Claux D, Butterlin G, Merdinoglu D, This P: 6. Guzman-Maldonado SH, Martinez O, Acosta-Gallegos JA, Guevara-Lara F, Mapping 245 SSR markers on the Vitis vinifera genome: a tool for grape Paredes-Lopez O: Putative quantitative trait Loci for physical and genetics. Theor Appl Genet 2004, 109(5):1017-1027. chemical components of common bean. Crop Sci 2003, 43(3):1029-1035. Huang et al. BMC Plant Biology 2012, 12:30 Page 18 of 19 http://www.biomedcentral.com/1471-2229/12/30 29. Cipriani G, Di Gaspero G, Canaguier A, Jusseaume J, Tassin J, Lemainque A, 51. Matus J, Poupin M, Cañón P, Bordeu E, Alcalde J, Arce-Johnson P: Isolation Thareau V, Adam-Blondon A, Testolin R: Molecular linkage maps: strategies, of WDR and bHLH genes related to flavonoid synthesis in grapevine resources and achievements Enfield, USA: Science Publishers; 2011. (Vitis vinifera L.). Plant Mol Biol 2010, 72(6):607-620. 30. Le Cunff L, Fournier-Level A, Laucou V, Vezzulli S, Lacombe T, Adam- 52. Bogs J, Jaffe FW, Takos AM, Walker AR, Robinson SP: The grapevine Blondon A-F, Boursiquot J-M, This P: Construction of nested genetic core transcription factor VvMYBPA1 regulates proanthocyanidin synthesis collections to optimize the exploitation of natural diversity in Vitis during fruit development. Plant Physiol 2007, 143(3):1347-1361. vinifera L. subsp. sativa. BMC Plant Biol 2008, 8(1):31. 53. Chira K, Schmauch G, Saucier Cd, Fabre S, Teissedre P-L: Grape variety 31. Cartwright DA, Troggio M, Velasco R, Gutin A: Genetic mapping in the effect on proanthocyanidin composition and sensory perception of skin presence of genotyping errors. Genetics 2007, 176(4):2521-2527. and seed tannin extracts from Bordeaux wine grapes (Cabernet 32. de Givry S, Bouchez M, Chabrier P, Milan D, Schiex T: CAR(H)(T)AGene: Sauvignon and Merlot) for two consecutive vintages (2006 and 2007). J multipopulation integrated genetic and radiation hybrid mapping. Agric Food Chem 2008, 57(2):545-553. Bioinformatics 2005, 21(8):1703-1704. 54. Travaglia F, Bordiga M, Locatelli M, Coïsson JD, Arlorio M: Polymeric 33. Doligez A, Audiot E, Baumes R, This P: QTLs for muscat flavor and proanthocyanidins in skins and seeds of 37 Vitis vinifera L. Cultivars: a monoterpenic odorant content in grapevine (Vitis vinifer L.). Mol Breed methodological comparative study. J Food Sci 2011, 76(5):C742-C749. 2006, 18:109-125. 55. Bordiga M, Travaglia F, Locatelli M, Coisson JD, Arlorio M: Characterisation 34. Singleton VL, Ough CS, Nelson KE: Density separations of wine grape of polymeric skin and seed proanthocyanidins during ripening in six berries and ripeness distribution. Am J Enol Vitic 1966, 17(2):95-105. Vitis vinifera L. cv. Food Chem 2011, 127(1):180-187. 35. Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y: RNA-seq: an 56. Carlborg O, Haley CS: Epistasis: too often neglected in complex trait assessment of technical reproducibility and comparison with gene studies? Nat Rev Genet 2004, 5(8):618-625. expression arrays. Genome Res 2008, 18:1509-1517. 57. Beavis W: The power and deceit of QTL experiments: lessons from 36. Fournand D, Vicens A, Sidhoum L, Souquet JM, Moutounet M, Cheynier V: comparative QTL studies. Proceedings of the Forty-Ninth Annual Corn and Accumulation and extractability of grape skin tannins and anthocyanins Sorghum Industry Research Conference 1994, 250-266. at different advanced physiological stages. J Agric Food Chem 2006, 58. Charcosset A, Gallais A: Estimation of the contribution of quantitative trait 54(19):7331-7338. loci (QTL) to the variance of a quantitative trait by means of genetic 37. R-Development-Core-Team: R: A Language and Environment for Statistical markers. TAG Theor Appl Genet 1996, 93(8):1193-1201. Computing Vienna: R Foundation for Statistical Computing; 2008, [http:// 59. Lippman ZB, Zamir D: Heterosis: revisiting the magic. Trends Genet 2007, www.R-project.org/]. ISBN 3-900051-07-0. 23(2):60-66. 38. Bates D, Maechler M, Dai B: lme4: Linear mixed-effects models using S4 60. Mol J, Grotewold E, Koes R: How genes paint flowers and seeds. Trends classes 2008 [http://lme4.r-forge.r-project.org/]. Plant Sci 1998, 3(6):212-217. 39. Broman KW, Wu H, Sen Å, Churchill GA: R/qtl: QTL mapping in 61. Holton TA, Cornish EC: Genetics and biochemistry of anthocyanin experimental crosses. Bioinformatics 2003, 19(7):889-890. biosynthesis. The Plant Cell Online 1995, 7(7):1071-1083. 40. Manichaikul A, Moon JY, Sen S, Yandell BS, Broman KW: A model selection 62. Falginella L, Castellarin S, Testolin R, Gambetta G, Morgante M, Di approach for the identification of quantitative trait Loci in experimental Gaspero G: Expansion and subfunctionalisation of flavonoid 3’,5’- crosses, allowing epistasis. Genetics 2009, 181(3):1077-1086. hydroxylases in the grapevine lineage. BMC Genomics 2010, 11(1):562. 41. Segura V, Denance C, Durel CE, Costes E: Wide range QTL analysis for 63. Ruzicka DR, Kandasamy MK, McKinney EC, Burgos-Rivera B, Meagher RB: The complex architectural traits in a 1-year-old apple progeny. Genome 2007, ancient subclasses of Arabidopsis actin depolymerizing factor genes 50(2):159-171. exhibit novel and differential expression. Plant J 2007, 52(3):460-472. 42. Atkinson B, Therneau T: Kinship: mixed kinship: mixed-effects Cox 64. Taylor JS, Raes J: Duplication and divergence: the evolution of new models, sparse matrices, and modeling data from large pedigrees. R genes and old ideas. Annu Rev Genet 2004, 38(1):615-643. package, Version 1.1.0-23., 1.1.0-23 2009 [http://cran.r-project.org]. 65. Rodin S, Riggs A: Epigenetic Silencing May Aid Evolution by Gene 43. Saïdou A-A, Mariac C, Luong V, Pham J-L, Bezançon G, Vigouroux Y: Duplication. J Mol Evol 2003, 56(6):718-729. Association studies identify natural variation at PHYC linked to flowering 66. Dubos C, Stracke R, Grotewold E, Weisshaar B, Martin C, Lepiniec L: MYB time and morphological variation in pearl millet. Genetics 2009, transcription factors in Arabidopsis. Trends Plant Sci 2010, 15(10):573-581. 182(3):899-910. 67. Khater F, Fournand D, Vialet S, Meudec E, Cheynier V, Terrier N: Identification 44. Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES: and functional characterization of cDNAs coding for hydroxybenzoate/ TASSEL: software for association mapping of complex traits in diverse hydroxycinnamate glucosyltransferases co-expressed with genes related samples. Bioinformatics 2007, 23(19):2633-2635. to proanthocyanidin biosynthesis. J Exp Bot 2012, 63:1201-1214. 45. Benjamini Y, Hochberg Y: Controlling the false discovery rate: a pratical 68. Xie DY, Dixon RA: Proanthocyanidin biosynthesis–still more questions and powerful approach to multiple testing. Journal of the Royal Statistical than answers? Phytochemistry 2005, 66(18):2127-2144. Society Series B (Statistical Methodology) 1995, 57(1):289-300. 69. Gargouri M, Manigand C, Mauge C, Granier T, Langlois d’Estaintot B, Cala O, 46. Hernandez-Jimenez A, Gomez-Plaza E, Martinez-Cutillas A, Kennedy JA: Pianet I, Bathany K, Chaudiere J, Gallois B: Structure and epimerase activity Grape skin and seed proanthocyanidins from Monastrell × Syrah Grapes. of anthocyanidin reductase from Vitis vinifera. Acta Crystallographica J Agric Food Chem 2009, 57(22):10798-10803. Section D 2009, 65(9):989-1000. 47. Deluc L, Bogs J, Walker AR, Ferrier T, Decendit A, Merillon JM, Robinson SP, 70. Terrier N, Ollé D, Verriès C, Cheynier V: Biochemical and molecular aspects Barrieu F: The transcription factor VvMYB5b contributes to the regulation of flavan-3-OL synthesis during berry development. In Grapevine of anthocyanin and proanthocyanidin biosynthesis in developing grape Molecular Physiology and Biotechnology. Edited by: Roubelakis-Angelakis KA, berries. Plant Physiol 2008, 147(4):2041-2053. Roubelakis-Angelakis KA. Netherlands: Springer; 2009:365-388. 48. Hichri I, Heppel SC, Pillet J, Leon C, Czemmel S, Delrot S, Lauvergeat V, 71. Stafford HA, Shimamoto M, Lester HH: Incorporation of [14 C] Bogs J: The basic Helix-Loop-Helix transcription factor MYC1 is involved Phenylalanine into Flavan-3-ols and Procyanidins in Cell Suspension in the regulation of the flavonoid biosynthesis pathway in grapevine. Cultures of Douglas Fir. Plant Physiol 1982, 69:1055-1059. Mol Plant 2010, 3(3):509-523. 72. Jende-Strid B: Genetic Control of Flavonoid Biosynthesis in Barley. 49. Deluc L, Barrieu F, Marchive C, Lauvergeat V, Decendit A, Richard T, Carde J- Hereditas 1993, 119(2):187-204. P, Merillon J-M, Hamdi S: Characterization of a grapevine R2R3-MYB 73. Delcour JA, Ferreira D, Roux DG: Synthesis of condensed tannins part 9. transcription factor that regulates the Phenylpropanoid pathway. Plant The condensation sequence of leucocyanidin with (+)-catechin and with Physiol 2006, 140(2):499-511. the resultant procyanidins. Journal of the Chemical Society, Perkin 50. Terrier N, Torregrosa L, Ageorges A, Vialet S, Verries C, Cheynier V, Transactions 1983, 1:1711-1717. Romieu C: Ectopic expression of VvMybPA2 promotes proanthocyanidin 74. Grimplet J, Deluc L, Tillett R, Wheatley M, Schlauch K, Cramer G, Cushman J: biosynthesis in grapevine and suggests additional targets in the Tissue-specific mRNA expression profiling in grape berry tissues. BMC pathway. Plant Physiol 2009, 149(2):1028-1041. Genomics 2007, 8(1):187. Huang et al. BMC Plant Biology 2012, 12:30 Page 19 of 19 http://www.biomedcentral.com/1471-2229/12/30 75. Debeaujon I, Peeters AJM, Léon-Kloosterziel KM, Koornneef M: The TRANSPARENT TESTA12 Gene of Arabidopsis Encodes a Multidrug Secondary Transporter-like Protein Required for Flavonoid Sequestration in Vacuoles of the Seed Coat Endothelium. The Plant Cell Online 2001, 13(4):853-872. 76. Corcoran MR, Geissman TA, Phinney BO: Tannins as Gibberellin Antagonists. Plant Physiol 1972, 49(3):323-330. 77. Green FB, Corcoran MR: Inhibitory action of five tannins on growth induced by several Gibberellins. Plant Physiol 1975, 56(6):801-806. doi:10.1186/1471-2229-12-30 Cite this article as: Huang et al.: Dissecting genetic architecture of grape proanthocyanidin composition through quantitative trait locus mapping. BMC Plant Biology 2012 12:30. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Plant Biology Springer Journals

Loading next page...
 
/lp/springer-journals/dissecting-genetic-architecture-of-grape-proanthocyanidin-composition-JItE8CcuN2

References (154)

Publisher
Springer Journals
Copyright
Copyright © 2012 by Huang et al; licensee BioMed Central Ltd.
Subject
Life Sciences; Plant Sciences; Agriculture; Tree Biology
eISSN
1471-2229
DOI
10.1186/1471-2229-12-30
pmid
22369244
Publisher site
See Article on Publisher Site

Abstract

Background: Proanthocyanidins (PAs), or condensed tannins, are flavonoid polymers, widespread throughout the plant kingdom, which provide protection against herbivores while conferring organoleptic and nutritive values to plant-derived foods, such as wine. However, the genetic basis of qualitative and quantitative PA composition variation is still poorly understood. To elucidate the genetic architecture of the complex grape PA composition, we first carried out quantitative trait locus (QTL) analysis on a 191-individual pseudo-F1 progeny. Three categories of PA variables were assessed: total content, percentages of constitutive subunits and composite ratio variables. For nine functional candidate genes, among which eight co-located with QTLs, we performed association analyses using a diversity panel of 141 grapevine cultivars in order to identify causal SNPs. Results: Multiple QTL analysis revealed a total of 103 and 43 QTLs, respectively for seed and skin PA variables. Loci were mainly of additive effect while some loci were primarily of dominant effect. Results also showed a large involvement of pairwise epistatic interactions in shaping PA composition. QTLs for PA variables in skin and seeds differed in number, position, involvement of epistatic interaction and allelic effect, thus revealing different genetic determinisms for grape PA composition in seeds and skin. Association results were consistent with QTL analyses in most cases: four out of nine tested candidate genes (VvLAR1, VvMYBPA2, VvCHI1, VvMYBPA1) showed at least one significant association with PA variables, especially VvLAR1 revealed as of great interest for further functional investigation. Some SNP-phenotype associations were observed only in the diversity panel. Conclusions: This study presents the first QTL analysis on grape berry PA composition with a comparison between skin and seeds, together with an association study. Our results suggest a complex genetic control for PA traits and different genetic architectures for grape PA composition between berry skin and seeds. This work also uncovers novel genomic regions for further investigation in order to increase our knowledge of the genetic basis of PA composition. Background diverse qualities are directly linked to PA chemical Proanthocyanidins (PAs), or condensed tannins, are fla- structures. As polymers, PA structure varies depending vonoid polymers widespread throughout the plant king- on the degree of polymerisation and the nature of build- ing blocks, the flavan-3-ols (differences in stereochemis- dom. They accumulate in many organs and tissues to provide protection against pests [1]. They are also deter- try, hydroxylation pattern on the B-ring and presence/ minant in food quality and their beneficial effects on absence of a galloyl group, Figure 1). Our understanding human health are increasingly investigated [1,2]. These of PA biosynthesis has been significantly improved through the isolation of two genes coding for leu- coanthocyanidin reductase (LAR, [3]) and anthocyanidin * Correspondence: doligez@supagro.inra.fr reductase (ANR, [4,5]), two specific enzymes for the for- UMR AGAP, INRA, 2, place Viala, 34060 Montpellier, France mation of flavan-3-ols, respectively (+)-(gallo)catechin Full list of author information is available at the end of the article © 2012 Huang et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Huang et al. BMC Plant Biology 2012, 12:30 Page 2 of 19 http://www.biomedcentral.com/1471-2229/12/30 and (-)-epi(gallo)catechin. However, several issues con- [13-16] while (+)-gallocatechin and (-)-epigallocatechin- cerning PA composition require further study, such as 3-O-gallate are present in trace amounts only [15]. PAs the synthesis of galloylated units, the genetic mechanism are abundant in grape berries with drastic differences in of polymerisation, and the origin of extension units, composition between skin and seeds: total content is since all flavonoid intermediates are believed to assume usually higher in seeds while polymer size is much lar- a2,3-trans configuration, similar to the 2,3-configura- ger in skin [15,16]. In terms of constitutive building tion of (+)-(gallo)catechin (Figure 1), while major PA blocks, (-)-epigallocatechin is a major component of grape skin PAs [15] while it is not detectable in seeds extension blocks assume a 2,3-cis configuration (e.g. [16]; (-)-epicatechin-3-O-gallate is present in large pro- (-)-epicatechin, Figure 1). Moreover, few studies are available on the genetic basis of PA composition quanti- portion as both extension and terminal subunits in tative variation [6,7]. seeds while it is present only in small amounts in skin It is of great interest to understand PA genetics in [15,16]. Advances in understanding grape PA synthesis grape since PAs are involved in grapevine self-defence have been mainly obtained through homologous cloning mechanisms and are responsible for major organoleptic [14,17-21]. However, the complex PA composition properties of red wine [8-10]. Because of its rich PA within a tissue and the contrasted composition between composition and the multiple genetic and genomic tools tissues suggest a complex interaction of many factors in available for this species, such as the whole genome the determinism of grape PA composition. sequence [11,12], grape could represent also an interest- One way to assess genetic determinism of trait varia- ing model for PA genetic study. Indeed, in Arabidopsis, tion without apriori knowledge is quantitative trait loci a major model for PA studies, PAs are only detected in (QTL) mapping. QTL mapping makes use of segregating the seed coat with the presence of (-)-epicatechin as populations and gives global insights into the genetic sole building block. By contrast, PAs are present in dif- architecture of the target phenotype, i.e.the number, ferent organs of grapevine and are composed of four position and effects of genomic regions [22]. Among all major building blocks: (+)-catechin, (-)-epicatechin, available mapping methods, the multiple-QTL approach (-)-epigallocatechin and (-)-epicatechin-3-O-gallate is particularly suitable for complex trait analysis since it OH HO O OH Extension subunits OH 3’ OH 2’ OH 4’ 1’ HO O 7 2 5’ OH OH 6’ A C 6 2 OH 4 1 Terminal subunits galloyl OH PA monomeric units R R1 R2 2, 3 configuration Subunits studied in this work (+)-catechin H OH H trans catEx/catT (-)-epicatechin H H OH cis epiEx/epiT (-)-epicatechin-3-O-gallate H H O-galloyl cis galEx/galT (+)-gallocatechin OH OH H trans n.d. (-)-epigallocatechin OH H OH cis egcEx (-)-epigallocatechin-3-O-gallate OH H O-galloyl cis n.d. Figure 1 Structures of proanthocyanidins and monomeric subunits. A generic structure of proanthocyanidin is shown and the possible configurations are highlighted in colour. “n” indicates the number of extension units, variable according to plant species and tissues. The general chemical structure of PA monomeric subunit includes a C6-C3-C6 skeleton which is called the A-C-B rings. The carbon nomenclature is indicated as numbers next to the corresponding carbon. The B-ring generally bears two or three hydroxyl groups. According to the stereochemistry of carbons 2 and 3 on the C-ring, the PA monomeric subunit could be in 2,3-trans (e.g. (+)-catechin) or in 2,3-cis configuration (e.g.(-)-epicatechin). The structure of galloyl is shown next to PA generic structure. The right column denotes the subunits studied in this work where “Ex” indicates “extension” units and “T”, terminal subunits/monomers. n.d., not detected. Huang et al. BMC Plant Biology 2012, 12:30 Page 3 of 19 http://www.biomedcentral.com/1471-2229/12/30 uses simultaneously multiple marker intervals with pos- [25], additional SSR markers, heterozygous both in Syrah sible inclusion of epistasis terms in QTL mapping and Grenache, were chosen from recent grapevine refer- model [23]. Instead of creating segregating populations, ence maps [29]. Based on the 12X grapevine reference one can also explore the existing diversity through asso- sequence (Grape Genome Browser http://www.genoscope. ciation mapping to identify loci involved in phenotypic cns.fr), we designed primers for candidate gene amplifica- variation [24]. According to the genetic architecture of tion using Primer3 (http://www.bioinformatics.nl/cgi-bin/ target traits, one can build appropriate breeding strategy primer3plus/primer3plus.cgi) with default parameters. Pri- and/or develop further gene function studies. mers used in this study are listed in Additional file 1. For The aim of this work was to investigate the genetic SNP analysis, gene fragments were amplified, sequenced determinism of PA composition variation in both skin and and analysed as described in [30]. seeds of grape. For this purpose, we first characterised skin and seed grape PA composition in a pseudo-F1 pro- Linkage map construction geny derived from a cross between Syrah and Grenache Framework maps were constructed based on the 97- cultivars. Three categories of PA variables were con- marker linkage map of [25] with 56 additional SSRs. All structed in order to capture the complex PA profiles: total 153 markers had a genotypic error rate lower than 1.5%, content variables, percentages of constitutive subunits, using Tmap as check [31]. Linkage maps were con- which assessed the biosynthesis efficiency among building structed using CarthaGène 0.999R [32] as described in blocks, and composite ratio variables, which included esti- [33] with Haldane mapping function. The “Syrah” and mation of polymer size and metabolite flux between build- “Grenache” framework maps were composed of 121 ing blocks. We then applied a multiple-QTL genome scan SSRs (total length 1118.8 cM), and 133 SSRs (1349.4 cM to identify main effect QTLs and pairwise epistatic interac- total length) respectively. The “Consensus” framework tions for PA variables. Nine functional candidate genes, map spanned 1256.4 cM based on 153 SSRs among among which eight co-located with QTLs, were sequenced which over 70% allowed segregation in four genotypic and their SNP-phenotype associations were investigated in classes in the pseudo F1 progeny (ab × ac and ab × cd). a grapevine diversity panel. We present here the first Marker order reliability was ensured at LOD 2 thresh- extensive study of genetic architecture of PA composition old. Segregation distortion on genotypic classes was ver- in grape and confirm the involvement of some candidate ified by a c test according to the segregating type of genes in PA composition variation. each marker for the different maps (e.g. for markers seg- regating as ab × cd in the consensus map, the H Methods hypothesis was ac:bc:ad:bd = 1:1:1:1). Twenty-five mar- Plant material kers out of 153 exhibited distorted segregation (P < The two grapevine populations used in this study have 0.05) and were mainly grouped on chromosomes 3 (4 been previously described [25-27]. Briefly, the QTL markers), 4 (6 markers) and 10 (5 markers). Markers on mapping population (S × G) consisted of a pseudo-F1 chromosome 4 exhibited the most significant allelic progeny of 191 individuals from a cross between two deviation (P < 0.001) due to segregating distortion wine grape cultivars, Syrah (S) and Grenache (G) and between Syrah alleles (aa:ab ~2/3:1/3). was maintained under classical local training system (3300 plants/ha plant density) at Montpellier SupAgro Phenotyping and PA variable construction Domaine du Chapitre (Hérault, France). The S × G Grapes were harvested at maturity (20° Brix). For each population was planted in two blocks. Each individual genotype, eight representative berry clusters were har- from the progeny was planted in two elementary plots vested from the five plants of the elementary plot. Sam- (one per block) comprising five plants each. Parental ple homogenisation was based on the accumulation of cultivars were also planted in each block with nine and total solutes (principally sucrose), a major marker of 43 elementary plants for Syrah and Grenache, respec- berry development during ripening. Berry density was tively. The association mapping population (CC) con- assessed by floatation in salt solutions [34]. Twenty-five sisted of a core-collection of 141 cultivars maximising berries with a density between 130 and 160 g NaCl/L agro-morphological diversity for 50 quantitative and were randomly selected. In the present study, we qualitative traits, and maintained at INRA Domaine de focused on the analysis of berry skin and seeds since Vassal (Hérault, France) [26]. PA concentration is quite low in flesh, flesh PAs accounting for only 2-6% of the total berry PA content DNA extraction, marker genotyping and gene sequencing [35]. Berry skin and seeds were separated, ground in DNA extraction and marker genotyping were already liquid nitrogen and stored at-80°C until analysis. PAs described in Adam-Blondon et al. [28]. In order to densify were extracted and analysed by high performance liquid a previous 97-SSR linkage map of the S × G population chromatography (HPLC) after acid-catalysed cleavage in Huang et al. BMC Plant Biology 2012, 12:30 Page 4 of 19 http://www.biomedcentral.com/1471-2229/12/30 the presence of phloroglucinol according to [36]. For sum of terminal subunits and monomers. We also con- the S × G population, both skin and seeds were ana- structed composite ratio variables, including mean degree lysed for 2 consecutive years: skin was analysed in 2005 of polymerisation (mDP), ratio of 2,3-trans-to2,3-cis- (1 block) and 2006 (2 blocks) while seeds were analysed subunits in extension position (Ftranscis_Ex), terminal in 2006 (2 blocks) and 2007 (2 blocks). For the CC position (Ftranscis_T) and overall subunits (Ftranscis_all) population, skin was analysed in 2005 and 2006 and and the ratio of B-ring di-hydroxylated to B-ring tri- seeds in 2006. hydroxylated subunits (F3pr35). Variables studied in this In order to obtain an exhaustive view of PA composi- work are summarised in Table 1. tion, three categories of PA variables were studied in this work: total PA content, subunit percentage and compo- Phenotypic data analysis site ratio variables. For total content variables, concP All statistical analyses were performed with R software (mg/g fresh weight) reflects the biosynthesis intensity in [37]. We identified the best-fit mixed model for each each tissue, concB (mg/berry) brings total content to sin- PA variable through Bayesian information criterion gle berry level by taking into account berry size while (BIC) in order to extract the best linear unbiased predic- concK (mg/kg berries) is a common enological measure- tors (BLUPs) for genotypic values and to estimate the ment taking into account yield-related traits. Since all PA broad sense heritability (H ). Mixed model fit was per- building blocks are derived from the same intermediate formed with lme4 package [38]. The mixed model structure, naringenin chalcone, we used the percentage assumption of normality of residual and BLUPs was of each PA subunit to total subunit quantity to assess checked after model fitting by quantile-quantile plot partitioning efficiency between PA building blocks. Our comparing the distribution of residual and random PA characterisation method did not distinguish between effect predictors to a theoretical normal distribution terminal units of polymers and flavan-3-ol monomers. (Additional file 2). No data transformation was The notation ending with “T” corresponds thus to the performed for PA variables measured in the two Table 1 PA variables used in this study and their description PA traits Skin/ Definition Biological/biochemical significance Seed Total content concP +/+ mg/g fresh tissue Biosynthesis intensity per gram of tissue concB +/+ mg/berry Taking berry size into account concK +/+ mg/Kg berries Taking yield related-trait into account Subunit 100·(subunit content)/(total content) Assessment of partitioning efficiency of percentage catEx +/+ (+)-catechin Extension subunit PA biosynthesis among different subunits epiEx +/+ (-)-epicatechin Extension subunit galEx +/+ (-)-epicatechin-3-O-gallate Extension subunit egcEx +/- (-)-epigallocatechin Extension subunit catT +/+ (+)-catechin Terminal subunit/monomer epiT +/+ (-)-epicatechin Terminal subunit/monomer galT +/- (-)-epicatechin-3-O-gallate Terminal subunit/monomer Composite variables mDP +/+ mean Degree of Polymerisation(Total number of extension and Assessment of PA polymer size terminal/monomer subunits)/(Number of terminal subunit/monomer) F3pr35 +/- (catEx + epiEx + galEx + catT + epiT)/(egcEx) Assessment of flux between B-ring di-OH and tri-OH subunits Ftranscis_Ex +/+ Skin:catEx/(epiEx + galEx + egcEx) Assessment of flux between 2,3-trans subunit and Seed: catEx/(epiEx + galEx) 2,3-cis subunit in extension part Ftranscis_T +/+ Skin: catT/epiT Assessment of flux between 2,3-trans subunit and Seed: catT/(epiT + galT) 2,3-cis subunit in terminal part Ftrancis_all +/+ Skin: (catEx + catT)/(epiEx + galEx + egcEx + epiT) Global assessment of flux between 2,3-trans subunit Seed: (catEx + catT)/(epiEx + galEx + epiT + galT) and 2,3-cis subunit(extension + terminal/monomer) Presence/Absence (indicated by +/-) of a given trait in grape berry tissues. based on PA content expressed in mg/g fresh tissue F for Flux. Huang et al. BMC Plant Biology 2012, 12:30 Page 5 of 19 http://www.biomedcentral.com/1471-2229/12/30 populations. More information regarding phenotypic is the probability of a greater F valueunder thenull data analysis and best-fit model for each PA variable is hypothesis that polymorphism was independent of the in Additional file 3. phenotype. The adjusted P value (called p_adj_Marker in TASSEL), is the site-wise P value adjusted for multi- QTL analysis ple tests which takes into account the dependence QTL analysis was performed on the genotypic BLUPs between SNPs due to linkage disequilibrium. Because with R/qtl package [39]. Multiple QTL regression was each gene was tested independently, we used an addi- carried out with “stepwiseqtl” function. This approach tional Bonferroni correction to correct for the number of studied genes (nine) which led to a threshold of uses forward/backward selection to identify a multiple- QTL model with inclusion of both main effect QTLs and 0.0056 for the adjusted P value. As the permutation pairwise interactions. Maximum QTL number was set to method is not available for MLM, we used the thresh- 10 for forward selection (max.qtl = 10). Model choice old proposed by Benjamini and Hochberg [45] with q was made via a penalized LOD score (pLOD) which is equal to 0.05 which led to a threshold of 0.0039. The the LOD score for the model (the log likelihood ratio effect of minor genotypic frequency and non-normality comparing the full model to the null model without of observed trait distribution was checked (details in QTL) with penalties on the number of QTLs and pair- Additional file 4). wise QTL × QTL interactions [40]. For each PA variable, specific penalties for main effect and digenic pairwise Results interaction terms were derived from 1000 permutations Phenotype analysis of two-dimensional scan (the “scantwo” function, method PA variable distribution and heritability = “hk”, n.perm = 1000) and penalties at genome-wide For the S × G population, all PA variables showed contin- error rate of 0.05 were used for multiple-QTL model fit- uous distribution and transgressive segregation and varia- ting. The QTL model with the largest pLOD was identi- tion extent was equivalent in the S × G and CC fied as the most probable one. Once determined the populations (Figure 2, Additional file 5). In agreement multiple QTL model, we refined QTL position ("refi- with previous studies [15,16,46], samples taken in 2006, neqtl” function) and estimate R for the whole model and for which both berry skin and seeds were analysed, dis- each term of the model, the individual LOD score of played different PA composition between tissues as illu- each term and the genotypic effect ("fitqtl” function). The strated in Figure 2A with the mean values in the S × G “lodint” function was used to derive LOD-1 QTL location population. For both S × G and CC populations, (-)-epica- confidence interval. Allelic effects for consensus QTLs techin (epiEx) was the predominant extension subunit in were estimated as described by Segura et al.[41].Gen- all tissues while (-)-epigallocatechin (egcEx) was only ome scan was performed with a 1 cM step. detected in skin. (+)-catechin (catT) was the predominant terminal subunit/monomer in both skin and seed while Association analysis galloylated units (galEx and galT) were more abundant in Nine candidate genes were selected for association test seed PA. Each subunit exhibited large variation according according to their function and co-localisation with to genotype. For instance in skin, (-)-epigallocatechin QTLs. Prior to association test, we used R kinship pack- (instead of (-)-epicatechin) could be the predominant sub- age [42] to perform model comparison among different unit in the extension position (67.9%) (Additional file 5). nested models according to [43] in order to select the PA content variables (concP, concB, concK) reached best fitted model for association test for each PA vari- higher values in seeds than in skin regardless of the able. Ancestry structure and kinship matrix were esti- unit, as illustrated for concP (Figure 2B), which exhib- mated based on 20 SSR markers located throughout the ited the largest difference between tissues. Comparison whole genome as described in [25]. of composite PA ratio variables showed different range After model comparisons, we used TASSEL package of variation between skin and seed. PAs were on average to perform association tests [44]. Two models were 8-times shorter in seeds than in skin with wide variation used: one accounting for ancestry structure effect (with in skin (Figure 2C). All three ratio variables assessing General Linear Model, or GLM in TASSEL) the other the flux between 2,3-trans and 2,3-cis forms (Ftrancis for both ancestry structure effect and random genetic series) pointed to different kinetics for extension and background effect (with Mixed Linear Model, or MLM terminal positions: trans subunits were more abundant in TASSEL). Association tests were performed on in skin for terminal units/monomers (Ftranscis_T) while BLUPs for skin variables and raw data for seed vari- they were much reduced in seeds when considering ables since seed data were available for 2006 only. For extension positions alone (Ftranscis_Ex) or extension GLM analyses, tests were run with 1000 permutations plus terminal subunits/monomers (Ftranscis_all, Figure 2D-F). Since the major extension blocks were in cis- allowing the determination of site-wise P value, whi ch Huang et al. BMC Plant Biology 2012, 12:30 Page 6 of 19 http://www.biomedcentral.com/1471-2229/12/30 A B SxG seed egcEx SxG skin galEx CC seed 80 60 epiEx CC skin Grenache catEx Syrah galT epiT % 40 catT skin seed concP (mg/g fresh tissue) C D Ftranscis_Ex mDP 0 0 Ftranscis_T Ftranscis_all Figure 2 Comparative composition of skin and seed PA in 2006 (A) and distribution of PA variables of S × G and CC populations in 2006 for concP (B), mDP (C), Ftranscis_Ex (D) Ftranscis_T (E) and Ftranscis_all (F).(A) PA composition in skin and seeds based on the S × G offspring average is shown. Each building block is presented as the offspring average percentage of total content. (B-F) Distribution of PA variables in S × G and CC populations in 2006. Upper limits of data interval are indicated under the x-axis. Full symbols near x-axis show mean parental values for S × G population, circle for Grenache and triangle for Syrah, in pink for skin values and in blue for seed value as for the offspring histograms. N° of Individuals N° of Individuals 0.5 2.5 0.7 3.5 0.9 4.5 1.1 5.5 1.3 1.5 6.5 1.7 7.5 1.9 8.5 2.1 9.5 2.7 2.9 N° of Individuals N° of Individuals N° of Individuals 0.015 0.007 0.025 0.013 0.035 0.019 0.045 0.025 0.055 0.031 0.065 0.037 0.075 0.05 0.085 0.07 60 0.095 0.09 0.12 0.11 0.17 0.13 0.23 0.15 0.28 0.17 0.33 0.19 0.38 0.21 0.43 150 Huang et al. BMC Plant Biology 2012, 12:30 Page 7 of 19 http://www.biomedcentral.com/1471-2229/12/30 configuration, i.e. (-)-epi(gallo)catechin, Ftranscis_Ex was epiT, F3pr35 and seed concP where additional QTLs always less than 1 both in skin and seeds. The higher were identified through parental detection (Additional Ftranscis_T in skin conformed to the fact that (+)-cate- file 6). More QTLs and digenic pairwise interactions chin was the predominant terminal subunits/monomer were identified on the consensus map than on parental in skin. (Figure 2B). Means of each PA variable mea- maps, allowing some QTL models to explain more than sured in both skin and seeds were systematically differ- 80% of the BLUP variance in consensus mapping (Addi- ent (paired t-test, P < 0.001, data not shown). tional file 6), as illustrated in the case of epiT in seeds For S × G population, average H of PA variables was (Figure 5D). Some loci were involved in phenotypic var- 0.56 (from 0.24 to 0.82) and 0.44 (from 0.26 to 0.54) in iation almost exclusively through digenic epistasis such skin and seeds, respectively. No significant difference in as locus 10@32 for seed concB or locus 14@16.0 for H magnitude was detected between skin and seeds (t- seed epiT (Figure 5). Loci were mainly of additive effect test, P = 0.053). Nevertheless, higher H were observed while dominance was predominant at some loci for for skin variables, especially catT and mDP (0.76 and concK, epiEx, mDP and Ftranscis_T in skin and galEx, 0.82, respectively, Additional file 5). A high H value epiT, Ftranscis_T and Ftranscis_all in seeds (Figure 4). was also found for these two traits in CC (0.86 and 0.72 Among all detected QTLs, only 10 main effect loci over- for catT and mDP, respectively, Additional file 5). lapped for the same variable in both tissues: 1 for concB, PA variable correlation 2for concK, 1for epiEx, 1for galEx, 2 for catT, 1for We performed PA variable correlation on genotypic epiT, 1 for mDP and 1 for Ftranscis_T (see Figure 5 for values (BLUPs) from S × G population because the two- some examples). Parental alleles contribution to these year data available both in skin and seeds allowed us to common loci was not always consistent across tissues work with a much reduced environmental effect. All (Additional file 6), which could be an indication of tis- three total content variables were highly correlated sue-specific genetic mechanisms. Different genetic archi- within a given tissue while significant correlations tectures were observed for the same PA variable between tissues were only observed for concB and between berry skin and seeds as illustrated in Figure 5: concK (Figure 3). Among subunit percentage variables few moderate QTLs (< 3) or no QTL in skin vs several in skin (Figure 3), the most noticeable features were the (> 5) small to moderate QTLs with possible involvement significant negative correlation between egcEx (B-ring of epistasis in seeds (for concP, concB, catEx, galEx, tri-hydroxylated subunit) and all other units (B-ring di- Ftranscis_Ex and Ftranscis_T, illustrated by concB in hydoxylated subunits) and the positive correlation Figure 5A); many QTLs with involvement of epistasis in between (-)-epicatechin and (+)-catechin either in exten- skin vs a small number (2) of main effect QTLs in seeds (epiEx in Figure 5B); a major QTL (R sion position (Ex) or in terminal/monomer position (T). > 50%) and some In seeds, the most noticeable feature was the significant QTLs of moderate effect in skin v.s.manyQTLsof correlation of epiT with all other variables (negative with small to moderate effect in seeds (for catT and Ftransci- extension units and positive with terminal units) while s_all, illustrated by catT in Figure 5C), or few moderate epiEx was negatively correlated with other subunits. For QTLs in skin vs many moderate QTLs in the presence the same subunit in a given tissue, no highly significant of a QTL of large effect and epistasis in seeds (epiT, Fig- correlation was observed between extension and terminal ure 5D). Conversely, similar genetic architecture position, except the negative correlation between epiEx between skin and seeds was observed for concK (only and epiT in seeds (P < 0.001). Significant correlation moderate main effect QTLs, Figure 5E) and mDP (a between subunit percentage variable and composite vari- major QTL and a few QTLs of moderate effect, Figure ables inside a tissue reflected the variable construction 5F). Details regarding position, major allelic effect, LOD (Table 1). Between tissues, significant positive correlations score, LOD-1 confidence interval and percentage of were observed for concK, most of terminal subunits/ explained variation (R ) for each QTL are given in Addi- monomers pairs, galEx and also mDP and Ftranscis_all. tional file 6. PA total content QTL analysis In skin, 1, 1 and 3 QTLs were identified on the consen- Global features of PA QTLs sus map for concP, concB and concK, respectively. One We performed QTL detection with genotypic BLUPs additional QTL for concP was identified through paren- both on consensus and parental maps. In total, 103 vs tal detection on the Syrah map. Conversely, for concK, 43 QTLs and 24 vs 2 digenic epistatic interactions were all QTLs exhibited a major Grenache allelic effect and identified on the consensus map for seed and skin PA one additional QTL was identified on chromosome 9 on variables, respectively (Figure 4). QTLs detected on par- the Grenache map. ental maps were generally also detected on the consen- In seeds, 6, 10 and 5 QTLs were identified for concP, sus map except for skin concP, concK, catEx, galEx, concB, and concK, respectively. The locus positioned at Huang et al. BMC Plant Biology 2012, 12:30 Page 8 of 19 http://www.biomedcentral.com/1471-2229/12/30 skin seed 1.0 0.7 0.7 0.1 -0.2 0.0 0.1 0.1 -0.2 0.0 0.1 0.3 0.1 -0.1 -0.1 0.1 0.1 0.1 -0.1 0.1 -0.2 0.0 0.1 0.1 0.1 -0.2 -0.1 concP 1.0 0.7 0.1 -0.1 -0.1 0.1 -0.1 -0.2 0.1 0.1 0.1 0.0 -0.1 0.0 0.2 -0.1 0.1 0.0 0.1 -0.1 -0.1 -0.1 0.1 0.0 0.0 0.0 concB concK 1.0 0.1 -0.2 0.1 0.1 0.1 -0.1 0.0 0.1 0.2 0.1 -0.1 0.1 0.0 0.3 0.1 -0.1 0.1 0.0 -0.1 0.1 0.0 0.1 0.0 0.0 catEx 1.0 0.3 0.1 -0.4 0.2 0.0 -0.2 1.0 0.1 0.5 0.4 -0.1 -0.1 0.0 0.2 -0.1 0.1 0.0 -0.1 0.0 0.1 0.2 0.1 0.1 epiEx 1.0 0.2 -1.0 0.1 0.2 -0.1 0.3 -0.2 0.2 0.9 0.0 -0.1 -0.1 0.0 -0.2 0.1 0.1 -0.1 0.0 0.0 0.0 0.2 0.1 galEx 1.0 -0.4 0.2 0.1 -0.1 0.2 0.0 0.2 0.3 0.0 0.1 0.3 0.0 -0.2 0.3 -0.1 -0.1 0.3 0.1 0.0 -0.1 -0.1 1.0 -0.2 -0.3 0.2 -0.4 0.1 -0.4 -1.0 0.0 0.1 0.0 0.0 0.2 -0.2 -0.1 0.1 -0.1 0.0 -0.1 -0.1 -0.1 skin egcEx 1.0 0.6 -0.9 0.2 0.4 0.9 0.2 -0.1 -0.1 0.1 0.0 -0.4 -0.2 0.3 0.4 0.5 -0.5 0.1 0.0 0.3 catT 1.0 -0.6 0.1 -0.3 0.5 0.3 0.0 0.0 0.2 0.0 -0.3 -0.2 0.3 0.2 0.3 -0.3 0.1 0.1 0.3 epiT 1.0 -0.2 -0.3 -0.9 -0.2 0.1 0.2 -0.1 0.1 0.4 0.3 -0.3 -0.4 -0.4 0.5 -0.1 0.1 -0.3 mDP Ftranscis_Ex 1.0 0.1 0.6 0.4 -0.1 -0.1 0.0 0.2 -0.1 0.1 0.0 -0.1 0.0 0.0 0.2 0.1 0.1 Ftranscis_T 1.0 0.4 -0.1 -0.1 -0.2 -0.1 0.0 -0.1 -0.1 0.0 0.2 0.2 -0.2 0.0 -0.2 0.0 Ftranscis_all 1.0 0.3 -0.1 -0.2 0.1 0.1 -0.3 -0.2 0.3 0.3 0.4 -0.4 0.2 0.0 0.3 F3pr35 1.0 0.0 -0.1 0.0 0.0 -0.2 0.2 0.1 -0.1 0.1 0.0 0.0 0.1 0.1 concP 1.0 0.6 0.5 0.2 0.0 -0.2 0.4 -0.2 -0.1 -0.1 0.2 0.5 0.4 1.0 0.6 0.3 0.0 0.0 0.1 -0.2 -0.1 0.0 0.3 0.3 0.2 concB 1.0 0.3 -0.2 0.0 0.1 -0.1 0.2 -0.1 0.3 0.2 0.3 concK 1.0 -0.1 0.1 -0.1 -0.5 -0.2 0.3 1.0 0.3 0.5 catEx 1.0 -0.2 -0.5 -0.4 -0.5 0.6 -0.2 -0.1 -0.4 epiEx galEx P<0.05 1.0 -0.6 -0.5 0.0 0.5 -0.1 -0.2 -0.5 seed catT P<0.01 1.0 0.4 0.1 -0.8 0.1 0.7 0.8 epiT P<0.001 1.0 0.4 -0.8 -0.4 -0.4 0.0 1.0 -0.4 -0.1 -0.4 0.0 galT Self-correlation mDP 1.0 0.1 -0.1 -0.5 1.0 0.3 0.6 Ftranscis_Ex 1.0 0.7 Ftranscis_T 1.0 Ftranscis_all Figure 3 PA variable correlation based on genotypic BLUP of S × G population. The Pearson pairwise correlation coefficient (r) is shown and colour codes give the significance of correlation tests. Skin variables are indicated in pink and seed variables are indicated in blue. The bold black lines delimit the pairwise correlation inside a tissue for a given variable category, i.e. total content, subunit percentage and composite variables. The bold green lines delimit the pairwise correlation between tissues for a given variable category. Simple variables: percentage of constitutive units 40-50 cM on chromosome 2 was identified for all three In skin, 1, 9, 3 and 5 QTLs respectively were identified on variables. Epistasis was strongly involved in genetic architecture of concB and in total accounted for around consensus map for catEx, epiEx, galEx, and egcEx, with 30% of the BLUP variance (Figure 1). several overlapping QTLs (Figure 4). For catEx, two addi- In summary, fewer total content QTLs were detected in tional QTLs on chromosomes 14 and 18, were specifically skin than in seeds. Common loci between skin and seeds identified on the Grenache map. Pairwise interactions for concB and concK were identified on chromosomes 8, were identified for epiEx. For terminal subunits/mono- 13, and 17. For each tissue, one locus was identified to mers, 2 and 4 consensus QTLs were detected for epiT and be common to the three total content variables: the QTL catT, respectively. These two traits had co-locating loci on on chromosome 8 for skin and the QTL on chromosome chromosomes 8 and 17 with an especially large R for the 2 for seeds. locus on chromosome 17 (55.8% for catT). concP concB concK catEx epiEx galEx egcEx catT epiT mDP Ftranscis_Ex Ftranscis_T Ftranscis_all F3pr35 concP concB concK catEx epiEx galEx catT epiT galT mDP Ftranscis_Ex Ftranscis_T Ftranscis_all Huang et al. BMC Plant Biology 2012, 12:30 Page 9 of 19 http://www.biomedcentral.com/1471-2229/12/30 14 16 12 3 4 56 7 8 9 10 11 12 13 15 17 181920 21 concP skin concB concK concP concB seed concK 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 catEx epiEx galEx skin egcEx catT epiT catEx epiEx galEx seed catT epiT galT 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 mDP Ftranscis_Ex skin Ftranscis_T Ftranscis_all F3pr35 mDP Ftranscis_Ex seed Ftranscis_T Ftranscis_all 20 cM 1 2 3 4 5 6 7 8 9 1011 12 13 141516 17 18 19 As Ag D Digenic interaction Figure 4 Overview of skin and seed PA QTLs identified on the consensus map for total content (A), subunit percentage (B) and composite variables (C). For each variable category, two panels are shown: the upper one for QTLs in skin and the lower one for seeds. The x- axis of each panel spans the whole genome where chromosome sizes are proportional to genetic distance of consensus map and the chromosome numbers are indicated under the x-axis of lower panels. QTLs are indicated by horizontal lines with width corresponding to LOD-1 confidence interval. As, Ag and D respectively indicate additive effect from Syrah alleles, additive effect from Grenache alleles and dominance effect which were estimated according to [41]. Color codes correspond to major effects for each QTL, estimated as (|As| or |Ag| or|D|)/(|As|+|Ag|+| D|) > 0.30. Triangles indicate loci involved in digenic pairwise interactions. Grape candidate genes for PA synthesis are indicated on the upper black line of (A) where bar size is proportional to the flanking marker interval of the gene. Green bars are for genes coding for synthetic enzymes while red bars are for genes coding for transcription factors. The number above the flanking marker interval indicates the corresponding candidate gene: 1, VvLAR1 (leucoanthocyanidin reductase) [14]; 2, VvLDOX (leucoanthocyanidin dioxygenase) [17]; 3, VvF3H (flavanone 3-hydroxylase) [17]; 4, VvMYB5b [47]; 5, VvC4H (cinnamate 4-hydroxylase); 6, VvF3’5’Hs (flavonoid 3’-5’ hydroxylases) [18-20]; 7, VvMYC1 [48], 8, VvPAL (phenylalanine ammonia-lyse) [17]; 9, VvMYB5a [49]; 10, VvMYBPA2 [50]; 11, VvCHIs (chalcone isomerases) [17,21]; 12, VvCHS (chalcone synthase) [17]; 13, VvWDR2 [51]; 14, VvMYBPA1 [52]; 15, VvMYCA1 [51]; 16, VvPAL (phenylalanine ammonia-lyse) [17]; 17, Vv4CL (4- coumaroyl CoA ligase); 18, VvWDR1 [51]; 19, VvLAR2 (leucoanthocyanidin reductase) [14]; 20, VvF3’Hs (flavonoid 3’-hydroxylases) [18-20], 21, VvDFR (dihydroflavonol reductase) [17]. Detailed genetic maps with marker names are available in Additional file 7. Huang et al. BMC Plant Biology 2012, 12:30 Page 10 of 19 http://www.biomedcentral.com/1471-2229/12/30 A concB B epiEx 60 60 60 60 Global R2=75.9% Global R²=79.79% Global R²=21.9% Global R2=16.1% 50 50 50 40 40 40 40 R²30 R² R² 30 30 R²30 20 20 20 10 10 10 10 0 0 0 C D catT epiT 60 60 60 Global R2=66.57% Global R2=71.5% Global R2=27.8% Global R2=87.71% 50 50 50 40 40 40 R² 30 30 R² 30 R² R²30 20 20 20 10 10 10 0 0 0 concK mDP E F 60 60 60 Global R2=70.1% Global R2=27.9% Global R²=53.48% Global R²=48.53% 50 50 50 50 40 40 40 R²30 30 R² 30 30 R² R² 20 20 20 10 10 10 10 0 0 0 Figure 5 R distribution of skin and seed PA QTLs identified on the consensus map for concB (A), epiEx (B), catT (C), epiT (D), concK 2 2 (E) and mDP (F). R of main effect QTLs (solid bar) and R of digenic epistatic interaction (hatched bars) are sorted according to their magnitude. Skin variables are indicated in darkpink and seed variables in blue. Locus names are indicated on the x-axis and should be read as chromosome@position_on_the_chromosome. Locus names are highlighted in pink for loci identified in both skin and seed for the same variable; we considered loci as “common” loci when their LOD-1 confidence interval overlapped. Loci involved in digenic epistasis are indicated by a dark dot under the locus names for which R was estimated without inclusion of the associated interaction. In seeds, 2 QTLs for epiEx were identified on the con- subunits of the same nature but with different positions sensus map while two additional loci on chromosomes 8 in the polymer (e.g. epiEx and epiT). The locus posi- and 12 were identified solely on Grenache and Syrah tioned at approx. 7 cM on chromosome 17 was identi- maps, respectively (Figure 4, Additional file 6). For all fied for all PA simple variables in both tissues except for other simple variables, at least 7 QTLs and 1 pairwise egcEx, galEx and catEx in skin. interaction were involved in multiple-QTL models on the Composite ratio variables consensus map. For galEx, additional loci were identified In skin, the best QTL model for mDP, Ftranscis_T and on chromosomes 3, 5 and 9 through parental detection. Ftranscis_all included only a few main effect QTLs (2 to Joint consideration of the results obtained from both 4 QTL, Additional file 6) without digenic interaction. tissues showed that different QTLs were identified for The major locus on chromosome 17 was also identified 17@21.9 17@7.0 8@79.1 6@45.0 8@78.0 13@5.0 1@25.0 8@56.0 13@14.0 17@6.0 2@44.0 17@12.0 2@45.0:19@35.0 2@14.0:14@33.0 2@49.0 18@17.0 4@60.0 18@33.0 4@48.0 4@60.0:8@69.0 8@12.0 2@45.0 1@11.0:10@32.0 19@35.0 17@38.0 14@17.0 2@14.0 8@79.1 1@11.0 14@33.0 8@56.0 13@1.0 8@69.0 10@32.0 17@6.0 17@4.0 10@2.0 1@18.0 8@79.1 8@77.0 6@44.0 6@45.0 8@54.0 17@7.0 18@76.0 6@45.0:18@18.4 3@0.0:13@0.0 3@0.0 13@0.0 5@47.0 18@18.4 17@6.0 17@6.0 17@7.0 4@35.0 18@4.0 5@7.0 14@16.0:19@29.0 14@23.0 5@41.0 4@52.0:17@7.0 9@57.0:18@4.0 4@52.0 8@17.0 19@29.0 1@0.0 11@57.3 9@57.0 14@16.0 Huang et al. BMC Plant Biology 2012, 12:30 Page 11 of 19 http://www.biomedcentral.com/1471-2229/12/30 for these 3 variables: in the case of mDP, it explained significantly associated to catT and mDP in skin while more than 50% of total BLUP variance (Figure 5F). Six the confidence interval of the QTLs for these two vari- QTLs were detected for F3pr35 on the consensus map ables overlapped. Conversely, we observed some SNP- while one additional Syrah-specific QTL was identified phenotype associations only in the diversity panel: through parental detections on chromosome 13. For VvLAR1-skin Ftranscis_all, VvMYBPA2-skin concP, Ftranscis_Ex, QTLs were solely identified through par- VvMYBPA2-skin concK, VvMYBPA2-skin mDP, ental detections (1 QTL for Syrah map and 2 QTLs for VvMYBPA2-seed galT, VvCHI1-skin concP, VvMYBPA1- skin epiT and VvMYBPA1-seed Ftranscis_T. Grenache map). In seeds, a three-additive-QTL model was identified for mDP while models with 10 QTLs and one to four digenic Discussion interactions were the best ones for Ftranscis_Ex, Ftrans- PA variation extent compared to previous studies cis_T and Ftranscis_all. In addition, digenic interaction A first characterisation of PA composition in a grape- accounted for about 30% of BLUP variance for Ftransci- vine pseudo-F1 population was provided by Hernandez- s_Ex and Ftranscis_all (Figure 5 and Additional file 6). Jimenez and co-workers [46]. Their population was In summary, different QTLs were identified for the composed of 42 offsprings, derived from a cross same variables, depending on berry tissues. Among all between Syrah and Monastrell. In all tissues, the subunit composite variables, only the large effect QTL for mDP percentage in extension position and Ftranscis-series and Ftranscis_T on chromosome 17 was common to variables of the Syrah × Monastrell population were of a skin and seeds. Comparison of multiple-QTL models magnitude and extent equivalent to those of the present between both tissues showed that more digenic interac- study. More divergent results were observed for 1) epiT, tions were involved in seed variables than in skin which is more abundant in the Syrah × Monastrell variables. population; 2) mDP, which is higher in our study and 3) total content variables, for which the population mean Association analyses on candidate genes and variation extent was three-to two-fold larger in the We positioned 21 known grape PA functional candidate present study than in [46]. Syrah, the common parent, genes on the genetic map using their relative position to behaved similarly in both studies although we observed SSR markers on the grape genome ([11], http://www. 10-fold and two-fold higher total PA contents in our genoscope.cns.fr, Figure 4, see Additional file 7 for the study for skin and seeds, respectively. The difference names of flanking markers of candidate genes). Associa- observed in offsprings may result from the fact that the tion tests were performed for nine functional candidate two populations differed by one parent but environmen- tal differences as well as PA extraction and quantifica- genes, eight of them co-locating with QTLs. Among can- didates, there were both genes encoding flavonoid path- tion methods might also have affected PA variables, as way enzymes and putative regulators. Genes were suggested by the different PA composition of the same partially to totally sequenced (gene coverage from 25 to parental cultivar. Indeed, mixed model fit suggested that 100%), mainly in exons (Table 2). Two models were used year had a major effect on PA-related variables for both for association studies since model comparison showed tissues, which was consistent with a previous study an equivalent fit (Additional file 8): one accounted for where PA content and composition were measured in fixed ancestry structure effect (GLM in TASSEL), the two cultivars for two consecutive years [53]. The large other for both fixed ancestry structure effect and random quantitative variation in PA variables in S × G, of genetic background effect (MLM in TASSEL). four out of equivalent extent in the CC diversity panel, underlines nine genes showed at least one significant association the interest to implement a quantitative genetic with PA variables with consistent results between GLM approach on a F1 population for grape PA studies. and MLM (Table 3). Seventy-eight percent of significant Two studies only have characterised PA composition tests (21 out of 27 tests) were common to GLM and in different grape cultivars [54,55] with at most a 37- MLM while 6 additional associations were only signifi- cultivar sample [54]. Different biochemical analyses did cant with MLM model. The reason for this discrepancy is not allow for result comparison between this latter probably that the adjusted P-value in GLM was estimated study and the present work. Nevertheless, CC in our by taking into account dependence between tests due to work was composed of 141 grapevine cultivars of broad linkage disequilibrium [44] while in MLM, each SNP is geographical origin (from East to West Europe) and was tested under an hypothesis of independence. Association initially defined to maximise the diversity of 50 agro- results were consistent with QTL analyses for following morphological traits [26]. The PA composition variation gene-phenotype pairs: VvLAR1-skin catT and VvLAR1- in the diversity panel provides thus the potential to refine PA QTLs in a population of larger genetic skin mDP (Table 3). In particular, several SNPs in linkage disequilibrium for VvLAR1 (data not shown) were background. Huang et al. BMC Plant Biology 2012, 12:30 Page 12 of 19 http://www.biomedcentral.com/1471-2229/12/30 Table 2 Summary of candidate genes for association tests Sequence (size and localisation) Number of SNPs chr Gene References 5’- Exon Intron 3’- seq/gene 5’- Exon Intron 3’- total QTL UTR UTR size UTR UTR 1 VvLAR1 [14] - 1008 412 65 1420/2980 - 21 9 - 30 Skin: catT, mDP. Seed: concB, galEx, catT, epiT, Ftranscis_Ex, Ftranscis_all 6 VvF3’5’H [19,20] - 1296 50 - 1346/2325 - 4 - - 4 Skin: epiEx, egcEx, catT, mDP, F3pr35, 1.1 Ftranscis_all 6 VvF3’5’H [19,20] - 630 12 24 642/1932 - 3 0 0 3 Skin: epiEx, egcEx, catT, mDP, F3pr35, 2.1 Ftranscis_all 8 VvMYB5a [49] 66 687 382 36 1069/1069 1 12 2 1 16 Skin: concP, concB, concK, epiEx, egcEx, epiT, F3pr35. Seed: concB, catEx, catT, Ftranscis_T 11 VvMYBPA2 [50] 93 855 263 37 1148/1479 - 14 5 - 19 No QTL 13 VvCHI1 [17,21] - 440 216 54 656/1486 - 3 5 1 9 Skin: concK, epiEx. Seed: concK, catT, Ftranscis_Ex, Ftranscis_all 13 VvCHI2 [17,21] - 206 452 46 658/2524 - 1 6 - 7 Skin: concK, epiEx. Seed: concK, catT, Ftranscis_Ex, Ftranscis_all 15 VvMYBPA1 [52] 384 861 87 1 0 948/948 8 11 - - 19 Seed: concP, catEx, Ftranscis_Ex 18 VvDFR [17] - 425 194 111 619/2469 - 1 - 2 3 Skin: egcEx. Seed: concP, concK, epiT Chr, chromosome; seq/gene size, total length of sequenced exons and introns/predicted gene size. Multiple QTL mapping in a pseudo-F1 population for alleles (Additional file 6). Parental detection allowed grape PA composition the identification of smaller additive QTLs because of To our knowledge, this study presents the first QTL a greater power due to more individuals in each geno- analysis on grape PA composition with comparisons typic class compared to consensus detection (i.e., 2 between skin and seeds of grape berry. This is also the and 4 genotypic classes for parental and consensus first work on grape using multiple QTL models taking detection, respectively). On the other hand, QTL into account both main effects and digenic epistasis dur- detection on the consensus map allowed us to estimate ing the mapping procedure. QTL mapping in animals QTL dominance effect s, i.e. the interaction between has shown that epistasis effects are often large enough allelic classes, but not necessarily with the assumption to be detected and thus merit a systematic scan regard- of a dominant-recessive relationship [59]. In the pre- less of population size, although larger populations (> sent work, 9.8% and 30% of QTLs detected in skin and 500 individuals) allow a more powerful epistasis detec- seeds, respectively, had dominance as the major allelic tion [56]. By employing the multiple QTL mapping effect (D in Figure 4 and Additional file 6). For exam- approach, we actually showed the important involve- ple, the locus 8@69 of concB in seed was involved in ment of epistatic interaction in shaping PA composition phenotypic variation almost exclusively through domi- variation; indeed, some loci were involved in phenotypic nance (Additional file 6) and this information would variation almost exclusively through pairwise interaction. have been overlooked if parental mapping only had Our mapping population is of sufficient size (191 indivi- been performed. duals) to allow identification of small effect QTLs. How- ever, one should keep in mind that the R estimate of Genetic architecture of grape PA composition individual QTLs is usually overestimated [57] and may PA total content have a wide confidence interval [58]. Some of the identi- QTL results were consistent with the results of PA vari- fied QTLs may therefore be of smaller effect in reality. able correlation: no significant correlation was observed One should thus be cautious in result interpretation and for concP between tissues and no co-locating QTL was further identification of causal polymorphism although identified for this variable between skin and seeds, while we did check initially the genome-wide first type common QTLs were identified for both concB and error rate. concK across tissues. Since concB and concK take into Allele contribution to individual QTL was mainly account berry size- and yield-related traits, these co- due to additive effects between Syrah and/or Grenache located QTLs for concB and concK may be involved Huang et al. BMC Plant Biology 2012, 12:30 Page 13 of 19 http://www.biomedcentral.com/1471-2229/12/30 Table 3 Results of the association study: significant SNP-phenotype associations along with the co-located QTL Chr Gene Marker Position Syn/Ns Tissue Trait n.obs p.MLM p.adj.GLM QTL 1 VvLAR1 int2687 intron 4 skin catT 112 2.92E-04 9.99E-04 Yes e5-2734 exon 5 Syn skin catT 112 1.22E-05 9.99E-04 Yes e1-82 exon 1 Ns (Ala ↔ Thr) skin mDP 115 7.08E-04 0.025 Yes e1-132 exon 1 Syn skin mDP 113 5.44E-04 9.99E-04 Yes e1-138 exon 1 Syn skin mDP 105 0.0013 0.0021 Yes e1-156 exon 1 Ns (Asn ↔ Lys) skin mDP 111 5.19E-04 9.99E-04 Yes e3-665 exon 3 Syn skin mDP 117 3.41E-04 9.99E-04 Yes e3-734 exon 3 Syn skin mDP 110 5.66E-04 9.99E-04 Yes int2405 intron 3 skin mDP 94 1.85E-04 9.99E-04 Yes e4-2524 exon 4 Syn skin mDP 103 5.58E-04 9.99E-04 Yes int2636 intron 4 skin mDP 104 6.49E-04 9.99E-04 Yes e5-2722 exon 5 Syn skin mDP 107 4.82E-04 9.99E-04 Yes e5-2776 exon 5 Syn skin mDP 107 4.82E-04 9.99E-04 Yes e5-2779 exon 5 Syn skin mDP 107 5.61E-04 9.99E-04 Yes e5-2785 exon 5 Syn skin mDP 107 5.61E-04 9.99E-04 Yes e5-2872 exon 5 Ns (Ile ↔ Met) skin mDP 104 6.37E-04 9.99E-04 Yes e5-2896 exon 5 Syn skin mDP 104 6.37E-04 9.99E-04 Yes e5-2902 exon 5 Syn skin mDP 104 6.37E-04 9.99E-04 Yes e1-156 exon 1 Ns (Asn ↔ Lys) skin Ftranscis_all 109 0.0032 0.0509 No int2687 intron 4 skin Ftranscis_all 114 0.0025 0.042 No e5-2734 exon 5 Syn skin Ftranscis_all 114 1.84E-04 9.99E-04 No 11 VvMYBPA2 intron06Y intron skin concP 117 9.77E-04 9.99E-04 No p19_GA promoter skin concK 55 4.37E-04 9.99E-04 No p18 promoter skin mDP 54 1.76E-04 9.99E-04 No p19_GA promoter skin mDP 55 3.36E-06 9.99E-04 No intron05M intron seed galT 82 0.0015 0.3377 No 1293 W exon 3 Syn seed galT 93 0.0014 0.3067 No 1322 W exon 3 Ns (Leu ↔ His) seed galT 92 0.0013 0.3387 No 1398Y exon 3 Syn seed galT 92 0.0029 0.6773 No 1473Y exon 3 Syn seed galT 93 0.0026 0.5135 No 13 VvCHI1 Y183 exon 4 Syn skin concP 108 0.0025 0.049 No 15 VvMYBPA1 p277R promoter skin epiT 125 0.0018 0.03 No 702W exon 2 Ns (Ser ↔ Thr) seed Ftranscis_T 68 0.0036 0.6783 No p.MLM, p-value from mixed model, p.adj.GLM, adjusted p-value from GLM. Bold cases indicate significant associations in both MLM and GLM results. QTL, the candidate genes were under QTLs of the same PA variables as those associated with SNP. indirectly in PA total content through alteration of berry variables. (Table 3). VvMYBPA2 is mainly expressed in development or yield-related traits. Indeed, several QTLs berry skin at green stage and its overexpression in grape for concK co-located with QTLs for yield-related traits, hairy root significantly increased PA production [50]. especially loci on chromosomes 8, 13, 17 and 18 where The significant associations of VvMYBPA2 were posi- QTLs for yield related and berry size related-trais were tioned in promoter and intron and might be involved also identified in the same S × G population (Doligez et either in transcription level alteration or through linkage al., unpublished data). Unlike these yield-related loci, the disequilibrium with other causal mutations. A minor loci identified for concP, which were also identified for association was identified between a non-synonymous concB and concK could be specific targets for a better polymorphism of VvCHI1 and skin concP (Table 3). understanding of the contrasting PA content in berry VvCHI encodes an upstream enzyme in the grape flavo- compartments. noid pathway. This gene may be involved in PA content Association tests were in accordance with the involve- variation through the control of the flux of intermediate ment of VvMYBPA2 in PA content variation, as sug- substrate. However, precise involvement of VvCHI1 in gested by a previous study [50]. Two VvMYBPA2 SNPs PA content variation needs further genetic and func- were significantly associated with skin PA content tional confirmation. Huang et al. BMC Plant Biology 2012, 12:30 Page 14 of 19 http://www.biomedcentral.com/1471-2229/12/30 Among tested candidate genes, VvMYB5a, located on 3-O-gallate was probably under the control of many chromosome 8, is under several QTLs, especially total genomic regions and digenic epistatic interactions content-related QTLs. A previous physiological study (Figures 4 and Additional file 6). Additional information showed that ectopic expression of VvMYB5a in tobacco was provided by association tests which revealed 4 weak induced expression of flavonoid genes and significantly but significant associations between galT in seeds and increased both PA and anthocyanin production [49]. SNPs of VvMYBPA2. These associated SNPs are located The authors therefore proposed VvMYB5a as an in introns or in C-terminal of the proteins which could upstream regulator of flavonoid pathway. In our work, contain protein-protein interaction domains [66] (Table no significant association was found for VvMYB5a while 2). This result suggests that the associated SNPs might the whole gene was sequenced (Table 2). Further inves- lead to alteration of transcriptional complex recruitment tigations would be necessary to figure out if this gene is or interaction with other proteins.[50].Glucosyltrans- involved in grape PA content variation. ferases were recently identified as putative candidates PA subunit synthesis: the hydroxylation patterns of flavan- involved in the first enzymatic step of PA galloylation 3-ols [67]. Since they are located on chromosome 3 where All flavonoids carry a hydroxyl group at the 4’ position QTLs for galEx in skin and seed are positioned (Figure of B-ring (Figure 1). The flavonoid hydroxylation pat- 4 and Additional file 7), they may be good candidates to terns of B-ring were first studied in ornamental plants be tested by association genetics in the next future. for colour engineering because it is a major colour PA subunit synthesis: the trans- and cis- subunits determinant for anthocyanins, another class of flavo- Synthesis of PA trans-and cis-subunits is tightly related noids sharing a similar C6-C3-C6 skeleton with PA to PA polymerisation since intermediate substrates in monomeric subunits [60,61]. Links between F3’H and the flavonoid pathway are assumed to take up a trans- F3’5’H gene activities and their relative flavan-3-ols are configuration while major extension subunits assume a less obvious due to the lack of easily assessed reporters. cis-configuration (e.g. (-)-epicatechin) [68]. Major Our results for grape skin variables showed that five advances in understanding PA subunits biosynthesis genomic regions (on chromosomes 3, 6, 8, 10 and 18) were made through the isolation of two genes coding harboured co-located QTLs for epiEx, egcEx and for specific enzyme activities for the formation of term- F3pr35. This co-localisation is not surprising since epiEx inal/monomers: 2,3-trans-(gallo)catechin and 2,3-cis-epi and egcEx were the major components for F3pr35 vari- (gallo)catechin [3-5]. Recently, another dynamic view of able construction and were therefore highly correlated the flux between trans-and cis-terminal units/monomers (Figure 2). These loci are probably involved in the flux was provided by Gargouri and co-workers who demon- between di-hydroxylated and tri-hydroxylated PA build- strated the ability of grape ANR to epimerise (+)-cate- ing blocks. An interesting point is that the QTL on chin to (-)-epicatechin [69]. Our results seem to be in chromosome 6 for both egcEx and F3pr35 co-located accordance with this work since a major locus on chro- with a genomic region corresponding to the F3’5’H gene mosome 17 was identified for catT and epiT, the two family. chiral flavan-3-ols, which was also the major locus for However, no significant association was detected Ftranscis_T and Ftranscis_all in skin. This locus further- between the two tested VvF3’5’H isogenes and hydroxy- more co-located with VvLAR2,anisogene of LAR, lation pattern variables in this work. F3’5’H is present as which belongs to the Reductase-Epimerase-Dehydrogen- a multigenic family in the grapevine genome in which at ase (RED) family, as ANR, and thus might display both least 15 isogenes have been identified [62]. For dupli- epimerase and reductase activity. Similarly, three cated genes, neofunctionalisation and/or subfunctionali- VvLAR1 SNPs were significantly associated to Ftransci- sation could conduct to specialisation of each isoform in s_all in skin and therefore merit further functional a spatio-temporal manner [63-65]. Actually, the isogene investigation to understand its involvement in in vivo VvF3’5’H1.1 (or VvF3’5’Hn in [62]) was shown to be PA subunit synthesis. expressed only in vegetative organs, while VvF3’5’H2.1 On the other hand, the origin of extension subunits is (or VvF3’5’Hf in [62]) is expressed in berry skin. The still uncertain: are extension subunits derived from assessment of the polymorphisms of all isogenes may intermediate substrates in the pathway or from end pro- give more insights for links between F3’5’H and hydro- ducts such as (-)-epicatechin and (+)-catechin? [1,68,70]. xylation variation. In our work, significant correlation was not systemati- PA subunit synthesis: the galloylated flavan-3-ols cally observed for subunits of the same nature but dif- To date, the underlying genetic determinism for the fering in position in the polymer, and few QTLs co- production of PA galloylated building blocks is still located. In addition, the QTLs for flux between trans- unclear. Our results in both grape berry skin and seeds and cis-subunits were most often different between extension position (Ftranscis_Ex) and terminal subunits/ showed that the quantitative variation of (-)-epicatechin- Huang et al. BMC Plant Biology 2012, 12:30 Page 15 of 19 http://www.biomedcentral.com/1471-2229/12/30 monomes (Ftranscis_T). All these results argue in favour differences between tissues were also observed. Globally, of the involvement of different loci in PA building for the same PA subunit percentage variables in both blocks synthesis and in the control of flux between tissues, only 5 QTLs among 74 had overlapping intervals trans-and cis-subunits according to their position in the between skin and seeds. Another contrasting feature was polymer. Stafford et al. already suggested from radioac- observed: in skin, 54% of all QTLs accounted for Syrah tive labelling experiments that upper and lower units additive effect and 78% for Grenache effect whereas in arise from different steps of the pathway rather than seeds, 74% of all QTLs accounted for Syrah additive from the condensation of similar units [71]. effect and 53% for Grenache effect (Additional file 6 and PA polymerisation Figure 4). Even for loci identified in both tissues for a An aspect of the debate about PA polymerisation con- given variable, major allelic effect and R differed (e.g. cerns the enzymatic or nonenzymatic polymerisation the QTL in chromosome 17 for catT, Additional file 6). (reviewed by [1,70]). The existence of a polymerase is Our results suggest that seed PA variation is controlled supported by the barley PA mutant ant26, containing by QTLs of moderate and equivalent magnitude with amounts of (+)-catechin equivalent to wild-type content involvement of epistasis. On the other hand, skin vari- but only trace amounts of PAs [72]. On the other hand, ables are mainly under the genetic control of a few large in vitro chemical synthesis of PAs has also been effect loci with a fluctuating variance unexplained by reported [73] and these authors observed a modulation QTLs. of polymer size through the modulation of the relative The different genetic architectures between tissues amounts of extension unit intermediates and monomers. could result from divergent functional evolution of PAs One can thus hypothesise that instead of a polymerase, in these two berry compartments. For fruits in general, the ant26 mutation could directly affect the suitable ensuring protection of the embryos is essential. Because conditions for spontaneous PA polymerisation, such as of their abundance and their ability to protect plants appropriate pH [70]. Further investigation of the QTLs against biotic stresses, PAs and flavan-3-ols might be identified in this study would bring more insights into the major molecules involved in grape embryo protec- this polymerisation issue. Indeed, in the case of skin tion. In fact, their influence in maintaining seed dor- mDP, H was high (0.82) and the multiple QTL model mancy has been demonstrated in Arabidopsis [75] and accounted for 70% of the genotypic variance, corre- their interaction with phytohormones has also been sponding to 57% of the total phenotypic variance. The reported [76,77]. Therefore, to prevent biological fluc- largest QTL on chromosome 17 explained alone 55% of tuationdueto asinglepolymorphism mutation, a net- genotypic variance and was also the major locus for work with multiple cross-talking actors as a product of evolution without human selection could be postulated seed mDP (Figure 5F). This QTL is therefore an inter- esting target for mDP genetic mechanism investigation. in the case of seed PAs, as suggested by the identifica- Another mDP QTL located on chromosome 1 might tion of numerous small effect QTLs and the involve- also be an interesting target for understanding PA poly- ment of epistasis. Conversely, skin is the first protective merisation. This QTL co-located with a gene encoding a barrier of the grape berry against its environment. In PA-specific synthetic enzyme, VvLAR1, for which several plants, PAs are thought to be involved in self-defence SNPs in linkage disequilibrium (data not presented) mechanisms [1]. For berry consumers, skin PAs confer were significantly associated to mDP and catT in skin, flavour to berries and are also responsible for major consistent with the corresponding QTL. Interestingly, organoleptic qualities of wine. Consumers in turn could VvLAR1 is highly polymorphic in Grenache while almost help the plant in seed dispersion or vegetative propaga- homozygous in Syrah (1 SNP in the coding region, data tion. However, high quantities of PAs would confer to not presented), in accordance with the fact that the berries too much astringency and bitterness, which QTLfor mDP inthisregionismainly due to aGre- would lead consumers (specially humans) to reject nache allelic effect. grapevines producing such berries for direct consump- tion (or wine-making). Human selection in particular Tissue-specific genetic architecture for PA composition could therefore have narrowed down the genetic basis In accordance with a previous study which demon- of skin PAs over time and consequently led to a specific strated the tissue specificity of transcriptional profiles in genetic architecture with a few large effect QTLs for grape berry [74], the present work illustrates different skin PA variables. genetic mechanisms for grape PA composition between skin and seeds: QTLs differed in terms of number, posi- QTL mapping and association analysis as complementary tion, R and allelic effects. For total content variables, approaches for candidate locus identification the major QTLs differed in skin and seeds. For sub- In this work, we used both QTL mapping and associa- unit percentage and composite variables, important tion analyses to identify phenotype-marker associations. Huang et al. BMC Plant Biology 2012, 12:30 Page 16 of 19 http://www.biomedcentral.com/1471-2229/12/30 For identification of phenotype-associated markers, G population. For each PA variable, 4 panels are shown: distribution of grapevine segregating populations have a greater diver- residuals of the best fitted model (box-and-whisker plot, topleft), quantile-quantile plot of model residuals against a theoretical normal sity than populations derived from inbred lines due to distribution (topright), distribution of BLUPs of the best fitted model heterozygous parental cultivars. However, their genetic (box-and-whisker plot, bottomleft), quantile-quantile plot of BLUPs background remains relatively narrow compared to against a theoretical normal distribution (bottomright). diversity panels. On the other hand, QTL mapping may Additional file 3: Phenotypic data analysis and best fitted models for variance component estimation. Analysis method and effects reveal associations undetected in diversity panels due to included in the best fitted model. low allelic frequency. The inconsistency between both Additional file 4: Effect of minor genotypic frequency and non- approaches sometimes encountered in this work could normalty of observed phenotype on the association test. Two therefore result from the fact that the available genetic sections are in this file. 1. Test for the enrichment of low frequency polymorphisms among associated markers. 2. Test for the effect of the polymorphisms were different between the two popula- non-normalty of the trait in the association tests. tions: causal polymorphisms in one population might be Additional file 5: Summary of PA variable distributions and broad monomorphic in the other. In addition, our analysis sense heritability (H ) in S × G and CC populations. Two tables focused on genes of known function co-locating with inside: Table A, summary of S × G population; Table B, summary of CC population. Skin data were collected in 2005 and 2006 and seed data QTL while other candidates could underlie QTL inter- were collected in 2006 and 2007. Parental values are indicated as mean vals. Besides time-consuming fine mapping, candidate ± standard error. Broad sense heritability (H ) was estimated based on genes can also be selected by combining QTL results the best fitted model as the percentage of phenotypic variance explained by the genotypic variance. with other data such as transcriptomics. Nevertheless, Additional file 6: QTL summary for consensus and parental LAR1 gene evoked a particular interest through associa- detection. Summary for consensus detection and parental detection are tion test. Complementing functional studies performed in two separate sheets. Term: main effect QTLs and pairwise epistatic on a single cultivar [14], we provide here additional con- interactions. Main effect QTLs are indicated by chromosome@position of LOD peak while interaction terms are indicated by “:” linking main effect firmation of LAR1 gene involvement in grape PA com- QTLs. Map (for QTL summary of parental detection): the parental map position through a diversity panel study. used for QTL detection. LOD score and R were estimated by dropping the considered term from the full model. For loci involved in pairwise interaction, their LOD and R were estimated by dropping both the main Conclusions effect and the associated interaction effect. df: degree of freedom The present work confirmed presumptions about the dropped for QTL effect estimation. Type III SS: type III sum of squares. CI: complex genetic architecture of PA composition in LOD-1 confidence interval. For consensus detection, As, Ag and D indicate additive effect from Syrah alleles, additive effect from Grenache grape berries. QTLs for PA total content, PA building alleles and dominance effect which were estimated according to Segura blocks, degree of polymerisation and ratio between et al. [41]: As: 1/4[(μ +μ )-(μ +μ )], Ag: 1/4[(μ +μ )-(μ +μ )], D: 1/4 ad ac bd bc ac bc ad bd building blocks were identified. Berry PA composition [(μ +μ )-(μ +μ )], where μ , μ and μ are phenotypic means for ac bd bc ad bd bc ad corresponding genotypes relative to phenotypic mean of ac genotype. offers a case study for tissue-specific genetic architec- Thus here, μ = 0. effect column in QTL summary for consensus ac ture: in skin, the same major loci were involved in sev- detection indicates major effects of the considered locus involved in eral PA variables while multiple and moderate QTLs phenotype variation, satisfying the following condition: (|As| or |Ag| or | D|)/(|As|+|Ag|+|D|) > 0.30. “Estimate” column in QTL summary for parental with strong epistasis were the principal genetic factors detection gives the allelic effect between parental alleles or the for seed PA composition. These differences might be corresponding interaction effect, the sign is arbitrary. The type III sum of due to human selection on skin PA leading to a reduced squares, df, LOD and R of the QTL model are indicated at the last row of each variables, highlighted on yellow. The loci identified in both skin diversity of related genes while a multi-factor network and seeds for the same variable (i.e. having overlapping LOD-1 controlling seed PA synthesis would be necessary to confidence interval) are highlighted in pink. protect grapevine embryos. Association tests confirmed Additional file 7: QTL maps with positioned grape PA candidate the interest of VvLAR1 as a candidate gene in modulat- genes. Parental and consensus maps are presented in parallel: left, Syrah map, indicated by S; centre, consensus map, indicated by C and right, ing catT and mDP in berry skin, as well as VvMYBPA2 Grenache map, indicated by G. QTLs are presented as vertical lines at the for total content and probably for subunit galloylation. right side of each map: the line length corresponds to LOD-1 confidence This study provides the first assessment of genetic interval and the LOD peak is indicated by a small horizontal bar in the confidence interval. Known candidate genes are positioned between mechanism underlying grape PA composition which flanking markers, indicated by red-filled bars, according to 12X grape opens doors for further PA genetic studies. genome sequence (http://www.genoscope.cns.fr). Red-filled bars indicated flanking marker interval of regulatory genes and green-filled bars for enzyme-coding genes. Additional material Additional file 8: Model comparison prior to association analyses. -2lnlikelihood is shown. Model comparison was performed by likelihood Additional file 1: Primers used for the amplification and sequencing ratio comparing each model to the most complete model. Significance of the candidate genes. was assessed using the c distribution with degree of freedom as the Additional file 2: Distribution of residuals and BLUPs and Quantile- difference in the number of parameter between two models. quantile plot of residual and BLUPs of the best fitted model for S × Significance level is indicated as *, P < 0.05, **, P < 0.01, ***, P < 0.001. Huang et al. BMC Plant Biology 2012, 12:30 Page 17 of 19 http://www.biomedcentral.com/1471-2229/12/30 7. Caldas G, Blair M: Inheritance of seed condensed tannins and their Abbreviations relationship with seed-coat color and pattern genes in common bean ANR: Anthocyanidin reductase; BLUP: Best linear unbiased predictor; 4CL: 4- (Phaseolus vulgaris L.). TAG Theor Appl Genet 2009, 119(1):131-142. coumaroyl CoA ligase; C4H: Cinnamate 4-hydroxylase; catEx: (+)-catechin 8. Dai GH, Andary C, Mondolot-Cosson L, Boubals D: Involvement of phenolic extension subunits; catT: (+)-catechin terminal subunits/monomers; CHI: compounds in the resistance of grapevine callus to downy mildew Chalcone isomerase; CHS: Chalcone synthasel; concP: Total content in mg/g (Plasmopara viticol). Eur J Plant Pathol 1995, 101(5):541-547. fresh weight; concB: Total content in mg/berry; concK: Total content in mg/ 9. del Río JA, Gómez P, Báidez A, Fuster MD, Ortuño A, Frías V: Phenolic kg berries; DFR: Dihydroflavonol reductase; egcEx: (-)-epigallocatechin compounds have a role in the defence mechanism protecting grapevine extension units; epiEx: (-)-epicatechin extension subunits; epiT: (-)-epicatechin against the fungi involved in Petri disease. Phytopathologia Mediterranea terminal subunits/monomers; F3H: Flavanone 3-hydroxylase; F3’H: Flavonoid 2004, 43(1):87-94. 3’ hydroxylase; F3’5’H: Flavonoid 3’-5’ hydroxylase; galEx: (-)-epicatechin-3-O- 10. Arnold RA, Noble AC, Singleton VL: Bitterness and astringency of phenolic gallate extension subunits; galT: (-)-epicatechin-3-O-gallate terminal subunits/ monomers; LOD: Logarithm of odds; LDOX: Leucoanthocyanidin fractions in wine. J Agric Food Chem 1980, 28(3):675-678. dioxygenase; LAR: Leucoanthocyanidin reductase; PA: Proanthocyanidin; PAL: 11. Jaillon O, Aury JM, Noel B, Policriti A, Clepet C, Casagrande A, Choisne N, Phenylalanine ammonia-lyse; QTL: Quantitative trait locus; SNP: Single Aubourg S, Vitulo N, Jubin C, et al: The grapevine genome sequence nucleotide polymorphism; SSR: Simple sequence repeat. suggests ancestral hexaploidization in major angiosperm phyla. Nature 2007, 449(7161):463-U465. Acknowledgements 12. Velasco R, Zharkikh A, Troggio M, Cartwright DA, Cestaro A, Pruss D, We thank A.-A. Saïdou for the helpful discussions and kind assistance in Pindo M, FitzGerald LM, Vezzulli S, Reid J, et al: A high quality draft association model selection. We thank all the research team members for consensus sequence of the genome of a heterozygous grapevine their participation in sample preparation. We thank M. Farnos and P. variety. PLoS One 2007, 2(12):e1326. Ortigosa for collecting diversity panel samples and the staff of Domaine du 13. Souquet J-M, Labarbe B, Le Guernevé C, Cheynier V, Moutounet M: Chapitre and Domaine de Vassal for grapevine cultivation. We thank three Phenolic composition of grape stems. J Agric Food Chem 2000, anonymous reviewers for helpful suggestions and P. Chatelet for helpful 48(4):1076-1080. English revision. This work was funded in part by the European project, 14. Bogs J, Downey MO, Harvey JS, Ashton AR, Tanner GJ, Robinson SP: FLAVO (no. 513960) and a Ph.D. grant for YFH from INRA and Languedoc- Proanthocyanidin synthesis and expression of genes encoding Roussillon Region. leucoanthocyanidin reductase and anthocyanidin reductase in developing grape berries and grapevine leaves. Plant Physiol 2005, Author details 139(2):652-663. 1 2 UMR AGAP, INRA, 2, place Viala, 34060 Montpellier, France. INRA, UMR1083 15. Souquet J-M, Cheynier V, Brossaud F, Moutounet M: Polymeric 3 ® SPO, 2, place, Viala, 34060 Montpellier, France. UMT Geno-Vigne ®, IFV, 2, proanthocyanidins from grape skins. Phytochemistry 1996, 43(2):509-512. place Viala, 34060 Montpellier, France. UMR Génomique Végétale, INRA 16. Prieur C, Rigaud J, Cheynier V, Moutounet M: Oligomeric and polymeric UEVE ERL CNRS, 2, rue Gaston Crémieux, 91057 Evry, France. Department of procyanidins from grape seeds. Phytochemistry 1994, 36:781-784. Ecology and Evolutionary Biology, Brown University, 80 Waterman Street, 17. Sparvoli F, Martin C, Scienza A, Gavazzi G, Tonelli C: Cloning and molecular Box G-W, Providence, RI 02912, USA. analysis of structural genes involved in flavonoid and stilbene biosynthesis in grape (Vitis-Vinifera L). Plant Mol Biol 1994, 24(5):743-755. Authors’ contributions 18. Bogs J, Ebadi A, McDavid D, Robinson SP: Identification of the flavonoid YFH carried out the gene sequencing and alignment, performed data hydroxylases from grapevine and their regulation during fruit analyses, prepared tables and figures and drafted the manuscript. AD development. Plant Physiol 2006, 140(1):279-291. checked phenotyping and genotyping data, performed linkage map 19. Castellarin SD, Di Gaspero G, Marconi R, Nonis A, Peterlunger E, Paillard S, construction and first statistical analysis and participated in manuscript Adam-Blondon AF, Testolin R: Colour variation in red grapevines (Vitis preparation. AFL participated in gene sequencing, sequence alignment, vinifera L.): genomic organisation, expression of flavonoid 3’- association test analysis and in manuscript preparation. LLC participated in hydroxylase, flavonoid 3’,5’-hydroxylase genes and related metabolite upstream data analyses, data interpretation and in manuscript preparation. profiling of red cyanidin-/blue delphinidin-based anthocyanins in berry YB conducted field experimentation and sample collection. AC performed skin. Bmc Genomics 2006, 7:12. genotyping for linkage map construction. FV, VM, CM, and JMS carried out 20. Jeong ST, Goto-Yamamoto N, Hashizume K, Esaka M: Expression of the biochemical analysis. NT directed PA variable conception, interpreted the flavonoid 3’-hydroxylase and flavonoid 3’,5’-hydroxylase genes and result and participated in manuscript preparation. VC and PT conceived the flavonoid composition in grape (Vitis vinifera). Plant Sci 2006, 170(1):61-69. study, participated in its design, coordination, data interpretation and 21. Jeong ST, Goto-Yamamoto N, Kobayashi S, Esaka A: Effects of plant manuscript preparation. All authors read and approved the final manuscript. hormones and shading on the accumulation of anthocyanins and the expression of anthocyanin biosynthetic genes in grape berry skins. Plant Received: 13 July 2011 Accepted: 27 February 2012 Sci 2004, 167(2):247-252. Published: 27 February 2012 22. Lynch M, Walsh B: Genetics and analysis of quantitative traits Sunderland: Sinauer Associates; 1998. 23. Kao C, Zeng Z, Teasdale R: Multiple interval mapping for quantitative trait Dixon RA, Xie DY, Sharma SB: Proanthocyanidins–a final frontier in fla- Loci. Genetics 1999, 152:1203-1216. vonoid research? New Phytol 2005, 165(1):9-28. 24. Buckler ES, Thornsberry JM: Plant molecular diversity and applications to 2. Aron PM, Kennedy JA: Flavan-3-ols: Nature, occurrence and biological genomics. Curr Opin Plant Biol 2002, 5(2):107-111. activity. Mol Nutr Food Res 2008, 52(1):79-104. 25. Fournier-Level A, Le Cunff L, Gomez C, Doligez A, Ageorges A, Roux C, 3. Tanner GJ, Francki KT, Abrahams S, Watson JM, Larkin PJ, Ashton AR: Bertrand Y, Souquet J-M, Cheynier V, This P: Quantitative genetic bases of Proanthocyanidin biosynthesis in plants–Purification of legume anthocyanin variation in grape (Vitis vinifera L. ssp. sativa) berry: a leucoanthocyanidin reductase and molecular cloning of its cDNA. J Biol quantitative trait locus to quantitative trait nucleotide integrated study. Chem 2003, 278(34):31647-31656. Genetics 2009, 183(3):1127-1139. 4. Devic M, Guilleminot J, Debeaujon I, Bechtold N, Bensaude E, Koornneef M, 26. Barnaud A, Lacombe T, Doligez A: Linkage disequilibrium in cultivated Pelletier G, Delseny M: The BANYULS gene encodes a DFR-like protein grapevine, Vitis vinifera L. Theor Appl Genet 2006, 112(4):708-716. and is a marker of early seed coat development. Plant J 1999, 27. Doligez A, Adam-Blondon A, Cipriani G, Di Gaspero G, Laucou V, 19(4):387-398. Merdinoglu D, Meredith C, Riaz S, Roux C, This P: An integrated SSR map 5. Xie DY, Sharma SB, Paiva NL, Ferreira D, Dixon RA: Role of anthocyanidin of grapevine based on five mapping populations. TAG Theor Appl Genet reductase, encoded by BANYULS in plant flavonoid biosynthesis. Science 2006, 113(3):369-382. 2003, 299(5605):396-399. 28. Adam-Blondon AF, Roux C, Claux D, Butterlin G, Merdinoglu D, This P: 6. Guzman-Maldonado SH, Martinez O, Acosta-Gallegos JA, Guevara-Lara F, Mapping 245 SSR markers on the Vitis vinifera genome: a tool for grape Paredes-Lopez O: Putative quantitative trait Loci for physical and genetics. Theor Appl Genet 2004, 109(5):1017-1027. chemical components of common bean. Crop Sci 2003, 43(3):1029-1035. Huang et al. BMC Plant Biology 2012, 12:30 Page 18 of 19 http://www.biomedcentral.com/1471-2229/12/30 29. Cipriani G, Di Gaspero G, Canaguier A, Jusseaume J, Tassin J, Lemainque A, 51. Matus J, Poupin M, Cañón P, Bordeu E, Alcalde J, Arce-Johnson P: Isolation Thareau V, Adam-Blondon A, Testolin R: Molecular linkage maps: strategies, of WDR and bHLH genes related to flavonoid synthesis in grapevine resources and achievements Enfield, USA: Science Publishers; 2011. (Vitis vinifera L.). Plant Mol Biol 2010, 72(6):607-620. 30. Le Cunff L, Fournier-Level A, Laucou V, Vezzulli S, Lacombe T, Adam- 52. Bogs J, Jaffe FW, Takos AM, Walker AR, Robinson SP: The grapevine Blondon A-F, Boursiquot J-M, This P: Construction of nested genetic core transcription factor VvMYBPA1 regulates proanthocyanidin synthesis collections to optimize the exploitation of natural diversity in Vitis during fruit development. Plant Physiol 2007, 143(3):1347-1361. vinifera L. subsp. sativa. BMC Plant Biol 2008, 8(1):31. 53. Chira K, Schmauch G, Saucier Cd, Fabre S, Teissedre P-L: Grape variety 31. Cartwright DA, Troggio M, Velasco R, Gutin A: Genetic mapping in the effect on proanthocyanidin composition and sensory perception of skin presence of genotyping errors. Genetics 2007, 176(4):2521-2527. and seed tannin extracts from Bordeaux wine grapes (Cabernet 32. de Givry S, Bouchez M, Chabrier P, Milan D, Schiex T: CAR(H)(T)AGene: Sauvignon and Merlot) for two consecutive vintages (2006 and 2007). J multipopulation integrated genetic and radiation hybrid mapping. Agric Food Chem 2008, 57(2):545-553. Bioinformatics 2005, 21(8):1703-1704. 54. Travaglia F, Bordiga M, Locatelli M, Coïsson JD, Arlorio M: Polymeric 33. Doligez A, Audiot E, Baumes R, This P: QTLs for muscat flavor and proanthocyanidins in skins and seeds of 37 Vitis vinifera L. Cultivars: a monoterpenic odorant content in grapevine (Vitis vinifer L.). Mol Breed methodological comparative study. J Food Sci 2011, 76(5):C742-C749. 2006, 18:109-125. 55. Bordiga M, Travaglia F, Locatelli M, Coisson JD, Arlorio M: Characterisation 34. Singleton VL, Ough CS, Nelson KE: Density separations of wine grape of polymeric skin and seed proanthocyanidins during ripening in six berries and ripeness distribution. Am J Enol Vitic 1966, 17(2):95-105. Vitis vinifera L. cv. Food Chem 2011, 127(1):180-187. 35. Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y: RNA-seq: an 56. Carlborg O, Haley CS: Epistasis: too often neglected in complex trait assessment of technical reproducibility and comparison with gene studies? Nat Rev Genet 2004, 5(8):618-625. expression arrays. Genome Res 2008, 18:1509-1517. 57. Beavis W: The power and deceit of QTL experiments: lessons from 36. Fournand D, Vicens A, Sidhoum L, Souquet JM, Moutounet M, Cheynier V: comparative QTL studies. Proceedings of the Forty-Ninth Annual Corn and Accumulation and extractability of grape skin tannins and anthocyanins Sorghum Industry Research Conference 1994, 250-266. at different advanced physiological stages. J Agric Food Chem 2006, 58. Charcosset A, Gallais A: Estimation of the contribution of quantitative trait 54(19):7331-7338. loci (QTL) to the variance of a quantitative trait by means of genetic 37. R-Development-Core-Team: R: A Language and Environment for Statistical markers. TAG Theor Appl Genet 1996, 93(8):1193-1201. Computing Vienna: R Foundation for Statistical Computing; 2008, [http:// 59. Lippman ZB, Zamir D: Heterosis: revisiting the magic. Trends Genet 2007, www.R-project.org/]. ISBN 3-900051-07-0. 23(2):60-66. 38. Bates D, Maechler M, Dai B: lme4: Linear mixed-effects models using S4 60. Mol J, Grotewold E, Koes R: How genes paint flowers and seeds. Trends classes 2008 [http://lme4.r-forge.r-project.org/]. Plant Sci 1998, 3(6):212-217. 39. Broman KW, Wu H, Sen Å, Churchill GA: R/qtl: QTL mapping in 61. Holton TA, Cornish EC: Genetics and biochemistry of anthocyanin experimental crosses. Bioinformatics 2003, 19(7):889-890. biosynthesis. The Plant Cell Online 1995, 7(7):1071-1083. 40. Manichaikul A, Moon JY, Sen S, Yandell BS, Broman KW: A model selection 62. Falginella L, Castellarin S, Testolin R, Gambetta G, Morgante M, Di approach for the identification of quantitative trait Loci in experimental Gaspero G: Expansion and subfunctionalisation of flavonoid 3’,5’- crosses, allowing epistasis. Genetics 2009, 181(3):1077-1086. hydroxylases in the grapevine lineage. BMC Genomics 2010, 11(1):562. 41. Segura V, Denance C, Durel CE, Costes E: Wide range QTL analysis for 63. Ruzicka DR, Kandasamy MK, McKinney EC, Burgos-Rivera B, Meagher RB: The complex architectural traits in a 1-year-old apple progeny. Genome 2007, ancient subclasses of Arabidopsis actin depolymerizing factor genes 50(2):159-171. exhibit novel and differential expression. Plant J 2007, 52(3):460-472. 42. Atkinson B, Therneau T: Kinship: mixed kinship: mixed-effects Cox 64. Taylor JS, Raes J: Duplication and divergence: the evolution of new models, sparse matrices, and modeling data from large pedigrees. R genes and old ideas. Annu Rev Genet 2004, 38(1):615-643. package, Version 1.1.0-23., 1.1.0-23 2009 [http://cran.r-project.org]. 65. Rodin S, Riggs A: Epigenetic Silencing May Aid Evolution by Gene 43. Saïdou A-A, Mariac C, Luong V, Pham J-L, Bezançon G, Vigouroux Y: Duplication. J Mol Evol 2003, 56(6):718-729. Association studies identify natural variation at PHYC linked to flowering 66. Dubos C, Stracke R, Grotewold E, Weisshaar B, Martin C, Lepiniec L: MYB time and morphological variation in pearl millet. Genetics 2009, transcription factors in Arabidopsis. Trends Plant Sci 2010, 15(10):573-581. 182(3):899-910. 67. Khater F, Fournand D, Vialet S, Meudec E, Cheynier V, Terrier N: Identification 44. Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES: and functional characterization of cDNAs coding for hydroxybenzoate/ TASSEL: software for association mapping of complex traits in diverse hydroxycinnamate glucosyltransferases co-expressed with genes related samples. Bioinformatics 2007, 23(19):2633-2635. to proanthocyanidin biosynthesis. J Exp Bot 2012, 63:1201-1214. 45. Benjamini Y, Hochberg Y: Controlling the false discovery rate: a pratical 68. Xie DY, Dixon RA: Proanthocyanidin biosynthesis–still more questions and powerful approach to multiple testing. Journal of the Royal Statistical than answers? Phytochemistry 2005, 66(18):2127-2144. Society Series B (Statistical Methodology) 1995, 57(1):289-300. 69. Gargouri M, Manigand C, Mauge C, Granier T, Langlois d’Estaintot B, Cala O, 46. Hernandez-Jimenez A, Gomez-Plaza E, Martinez-Cutillas A, Kennedy JA: Pianet I, Bathany K, Chaudiere J, Gallois B: Structure and epimerase activity Grape skin and seed proanthocyanidins from Monastrell × Syrah Grapes. of anthocyanidin reductase from Vitis vinifera. Acta Crystallographica J Agric Food Chem 2009, 57(22):10798-10803. Section D 2009, 65(9):989-1000. 47. Deluc L, Bogs J, Walker AR, Ferrier T, Decendit A, Merillon JM, Robinson SP, 70. Terrier N, Ollé D, Verriès C, Cheynier V: Biochemical and molecular aspects Barrieu F: The transcription factor VvMYB5b contributes to the regulation of flavan-3-OL synthesis during berry development. In Grapevine of anthocyanin and proanthocyanidin biosynthesis in developing grape Molecular Physiology and Biotechnology. Edited by: Roubelakis-Angelakis KA, berries. Plant Physiol 2008, 147(4):2041-2053. Roubelakis-Angelakis KA. Netherlands: Springer; 2009:365-388. 48. Hichri I, Heppel SC, Pillet J, Leon C, Czemmel S, Delrot S, Lauvergeat V, 71. Stafford HA, Shimamoto M, Lester HH: Incorporation of [14 C] Bogs J: The basic Helix-Loop-Helix transcription factor MYC1 is involved Phenylalanine into Flavan-3-ols and Procyanidins in Cell Suspension in the regulation of the flavonoid biosynthesis pathway in grapevine. Cultures of Douglas Fir. Plant Physiol 1982, 69:1055-1059. Mol Plant 2010, 3(3):509-523. 72. Jende-Strid B: Genetic Control of Flavonoid Biosynthesis in Barley. 49. Deluc L, Barrieu F, Marchive C, Lauvergeat V, Decendit A, Richard T, Carde J- Hereditas 1993, 119(2):187-204. P, Merillon J-M, Hamdi S: Characterization of a grapevine R2R3-MYB 73. Delcour JA, Ferreira D, Roux DG: Synthesis of condensed tannins part 9. transcription factor that regulates the Phenylpropanoid pathway. Plant The condensation sequence of leucocyanidin with (+)-catechin and with Physiol 2006, 140(2):499-511. the resultant procyanidins. Journal of the Chemical Society, Perkin 50. Terrier N, Torregrosa L, Ageorges A, Vialet S, Verries C, Cheynier V, Transactions 1983, 1:1711-1717. Romieu C: Ectopic expression of VvMybPA2 promotes proanthocyanidin 74. Grimplet J, Deluc L, Tillett R, Wheatley M, Schlauch K, Cramer G, Cushman J: biosynthesis in grapevine and suggests additional targets in the Tissue-specific mRNA expression profiling in grape berry tissues. BMC pathway. Plant Physiol 2009, 149(2):1028-1041. Genomics 2007, 8(1):187. Huang et al. BMC Plant Biology 2012, 12:30 Page 19 of 19 http://www.biomedcentral.com/1471-2229/12/30 75. Debeaujon I, Peeters AJM, Léon-Kloosterziel KM, Koornneef M: The TRANSPARENT TESTA12 Gene of Arabidopsis Encodes a Multidrug Secondary Transporter-like Protein Required for Flavonoid Sequestration in Vacuoles of the Seed Coat Endothelium. The Plant Cell Online 2001, 13(4):853-872. 76. Corcoran MR, Geissman TA, Phinney BO: Tannins as Gibberellin Antagonists. Plant Physiol 1972, 49(3):323-330. 77. Green FB, Corcoran MR: Inhibitory action of five tannins on growth induced by several Gibberellins. Plant Physiol 1975, 56(6):801-806. doi:10.1186/1471-2229-12-30 Cite this article as: Huang et al.: Dissecting genetic architecture of grape proanthocyanidin composition through quantitative trait locus mapping. BMC Plant Biology 2012 12:30. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

BMC Plant BiologySpringer Journals

Published: Feb 27, 2012

There are no references for this article.