• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of plntphysLink to Publisher's site
Plant Physiol. Nov 2012; 160(3): 1613–1629.
Published online Sep 21, 2012. doi:  10.1104/pp.112.203786
PMCID: PMC3490610

Allelic Variation in Paralogs of GDP-l-Galactose Phosphorylase Is a Major Determinant of Vitamin C Concentrations in Apple Fruit1,[C][W][OA]


To identify the genetic factors underlying the regulation of fruit vitamin C (l-ascorbic acid [AsA]) concentrations, quantitative trait loci (QTL) studies were carried out in an F1 progeny derived from a cross between the apple (Malus × domestica) cultivars Telamon and Braeburn over three years. QTL were identified for AsA, glutathione, total antioxidant activity in both flesh and skin tissues, and various quality traits, including flesh browning. Four regions on chromosomes 10, 11, 16, and 17 contained stable fruit AsA-QTL clusters. Mapping of AsA metabolic genes identified colocations between orthologs of GDP-l-galactose phosphorylase (GGP), dehydroascorbate reductase (DHAR), and nucleobase-ascorbate transporter within these QTL clusters. Of particular interest are the three paralogs of MdGGP, which all colocated within AsA-QTL clusters. Allelic variants of MdGGP1 and MdGGP3 derived from the cultivar Braeburn parent were also consistently associated with higher fruit total AsA concentrations both within the mapping population (up to 10-fold) and across a range of commercial apple germplasm (up to 6-fold). Striking differences in the expression of the cv Braeburn MdGGP1 allele between fruit from high- and low-AsA genotypes clearly indicate a key role for MdGGP1 in the regulation of fruit AsA concentrations, and this MdGGP allele-specific single-nucleotide polymorphism marker represents an excellent candidate for directed breeding for enhanced fruit AsA concentrations. Interestingly, colocations were also found between MdDHAR3-3 and a stable QTL for browning in the cv Telamon parent, highlighting links between the redox status of the AsA pool and susceptibility to flesh browning.

In plants, l-ascorbic acid (AsA; vitamin C) is essential for the detoxification of reactive oxygen species produced under stress or following exposure to pathogens. In addition to these antioxidant functions, AsA has been shown to be involved in a range of important cellular processes, including plant development and hormone signaling, cell cycle, cell expansion, senescence, and as a cofactor for a number of important enzymes (for review, see Davey et al., 2000; Smirnoff et al., 2001; Noctor, 2006). Fruit AsA concentrations have also been correlated with the maintenance of quality during postharvest storage (Davey and Keulemans, 2004; Davey et al., 2007) and have been linked to susceptibility to internal browning in both apple (Malus × domestica; Davey et al., 2006; Davey and Keulemans, 2009) and pear (Pyrus communis; Veltman et al., 1999; Franck et al., 2003). Finally, AsA is clearly an essential dietary component for humans, with a protective role proposed for many disorders and diseases (Diplock et al., 1998). Given its importance for all metabolically active tissues, there is widespread interest in unraveling the mechanisms underlying the genetic control of AsA concentrations in fruits as well as in how AsA interacts with other plant antioxidant pools.

The concentration of AsA will be determined by the net rates of biosynthesis, recycling, degradation, and/or intercellular and intracellular transport, but the relative contribution of these various processes depends on several factors, including genetics, tissue type (Bulley et al., 2009), developmental stage (Hancock et al., 2007; Bulley et al., 2009; Ioannidi et al., 2009), and light intensity (Yabuta et al., 2007; Gautier et al., 2009). The biosynthesis of AsA proceeds via l-Gal (Wheeler et al., 1998), although conclusive evidence for all steps has only relatively recently become available (Conklin et al., 2006; Laing et al., 2007). Alternative biosynthetic routes involving uronic acids (Davey et al., 1999; Agius et al., 2003), l-gulose (Wolucka and Van Montagu, 2003), or myoinositol (Lorence et al., 2004) have been proposed in several plant species, including apple (Davey et al., 2004; Razavi et al., 2005; Fig. 1), but their physiological relevance and contribution to the AsA pool is still far from clear in most plant species, with the possible exception of strawberry (Fragaria × ananassa; Agius et al., 2003; Cruz-Rus et al., 2011; Zorrilla-Fontanesi et al., 2011).

Figure 1.
AsA biosynthetic and recycling pathways in plants: l-Gal pathway, reactions 1 to 9; l-gulose pathway, reactions 1 to 5 and 10 to 13; d-galacturonate pathway, reactions 14 to 16; myoinositol/glucuronate pathway, reactions 17 to 21; recycling pathway, reactions ...

As an antioxidant, AsA is able to accept electrons from a wide range of radical substrates, and in this process it becomes oxidized first to monodehydroascorbate and then to dehydroascorbate (DHA). These oxidized forms of AsA can be regenerated by the ascorbate-glutathione (GSH) cycle, so that GSH and the activities of GSH reductase, dehydroascorbate reductase (DHAR), and monodehydroascorbate reductase (MDHAR) maintain the size and redox status of the AsA pool (Noctor and Foyer, 1998; Fig. 1). Indeed, overexpression of an Arabidopsis (Arabidopsis thaliana) MDHAR (Eltayeb et al., 2007) and a wheat (Triticum aestivum) DHAR (Chen et al., 2003) have both been shown to increase foliar AsA concentrations in tobacco (Nicotiana tabacum). MDHAR activity has also been positively correlated with both AsA and fruit firmness in tomato (Solanum lycopersicum) after chilling stress (Stevens et al., 2008).

Tissue AsA concentrations can also be maintained by intercellular transport, and there is evidence for the long-distance transport of AsA via the phloem from source (leaf) to sink (fruit) tissues (Franceschi and Tarlyn, 2002; Hancock et al., 2003). In apple, fruit AsA concentrations have been suggested to be partly dependent on the translocation of AsA from leaves (Li et al., 2009), but in black currant (Ribes nigrum), others concluded that the contribution of phloem AsA transport to fruit AsA concentrations was negligible (Hancock et al., 2007). While the actual mechanisms of long-distance transport of AsA have not been fully determined, attention has focused on the large family of Nucleobase-Ascorbate Transporters (NATs; de Koning and Diallinas, 2000), and NAT homologs have been found to be highly expressed in vascular tissues (Maurino et al., 2006).

Genes involved in several of these mechanisms have been proposed to be key regulators of fruit AsA concentrations, including GDP-l-Gal phosphorylase (GGP) or vitamin c defective2 (VTC2) in kiwifruit (Actinidia deliciosa; Bulley et al., 2009, 2012) as well as GDP-Man-3,5-epimerase (GME; Gilbert et al., 2009), l-Gal-1-P-phosphatase (GPP or VTC4; Ioannidi et al., 2009), and MDHAR (Stevens et al., 2007) in tomato. However, apart from GGP (Bulley et al., 2012), overexpression of these structural genes has to date had limited success in altering the fruit AsA pool (Agius et al., 2003; Bulley et al., 2009; Haroldsen et al., 2011).

In this work, we set out to identify potential genetic determinants of fruit AsA concentrations in apple fruit using a combination of molecular and genomic approaches. Initial quantitative trait loci (QTL) analyses of AsA concentrations (Davey et al., 2006) have been expanded to identify QTL for other antioxidants and fruit quality traits over three years, including results in 1 year comparing the concentrations of AsA in fruit and leaves. Alignments of the apple orthologs of genes involved in AsA biosynthesis, turnover, and transport against the whole genome sequence of cv Golden Delicious (Velasco et al., 2010) allowed us to identify candidate genes (CGs) colocating with stable QTL clusters. Using next-generation sequencing (RNA-Seq) data, polymorphic single-nucleotide polymorphism (SNP)-based markers were developed for these colocating CGs, and their positions on individual linkage groups were confirmed by linkage mapping in our mapping population. Finally, associations between allelic variants of these CGs and their expression levels in cultivars with contrasting AsA concentrations allowed us to develop allele-specific markers associated with high fruit AsA concentrations.


Population Variation in Tissue AsA and GSH Concentrations

Fruit AsA-total ascorbic acid (totAsA) and GSH-total glutathione (totGSH) concentrations showed a high degree of variability in the cv Telamon × Braeburn (TxB) mapping population, with mean values for flesh totAsA and totGSH on average varying 10- and 7-fold across the population over the three years of measurements, respectively (Table I). Total antioxidant activity (TAA) was only measured in 2006, with results indicating that the contribution of totAsA to TAA ranged from 4% to 25% in flesh and from 6% to 50% in skin tissues. Both AsA and totAsA (AsA + DHA) and GSH and totGSH (GSH + oxidized glutathione [GSSG]) concentrations in flesh and skin tissues showed normal distribution across progeny of the cv TxB population over the years, but concentrations of the oxidized forms (DHA and GSSG) displayed a statistically normal distribution in only one of the three years of measurement (2006).

Table I.
Overview of population mean values and distribution for fruit antioxidant traits of flesh (FL) and skin (SK) tissues, as well as fruit fresh weight over the years

Fruit AsA, totAsA, GSH, and totGSH concentrations in individuals of the TxB progeny were positively correlated (P < 0.0001) between flesh and skin tissues, such that fruit with high skin AsA and GSH concentrations also contained high flesh AsA and GSH concentrations (Supplemental Table S1). AsA and totAsA concentrations were also significantly correlated across the years (Supplemental Table S2; P < 0.0001); however, correlations for DHA, GSH, GSSG, and totGSH concentrations were lower or not significant at all. Comparing totAsA concentrations across the years indicated that in 2009, values were 27% and 42% higher in fruit flesh and 40% and 60% higher in fruit skin than the concentrations determined in 2005 and 2006, respectively. The percentage proportion of DHA in the totAsA pool, which is an indicator of the degree of oxidative stress experienced by the tissue, was found to be significantly higher in 2006 compared with 2005 or 2009. Finally, fruit skin tissues contained approximately 40% more totGSH than flesh tissues, and values in both tissues were around 50% higher in 2009 than in 2006.

These year-to-year differences in fruit AsA and GSH traits are probably related to significant differences in the climatic conditions (Supplemental Fig. S1). For example, the length of the harvest season ranged from 32 d in 2006 to 59 d in 2005, even though the first day of the harvesting period was the same each year. The highest mean population AsA and totAsA concentrations were measured in 2009, when both average daily temperature and hours of sun radiation were higher than those of the other years, suggesting an adaption to the high light and high temperature experienced in this year. Similarly, the lowest AsA and totAsA concentrations were measured in 2006, when hours of sunlight were lower than in either 2005 or 2009. A comparison of leaf and fruit totAsA and totGSH concentrations in 2009 indicated that apple leaves contained approximately 65-fold more totAsA and 12-fold more totGSH than fruit flesh tissues (Supplemental Table S3). No significant correlations were found between fruit and foliar AsA and totAsA concentrations in fruit of progeny from TxB, but there were significant correlations between foliar and both fruit flesh and skin GSH, totGSH, and GSSG concentrations and percentage GSSG (Supplemental Table S4).

Population Variation in Flesh Browning and Fruit Quality Traits

The susceptibility of fruit flesh to browning was assessed at harvest in 2006 and 2009 by measuring the time taken for cut surfaces to brown (Br_time) and the final color of the affected tissue (Br_color). Br_time ranged from 15 to 265 min and was negatively correlated with the percentage DHA values of flesh tissues (Supplemental Table S5) but not with the flesh totAsA values. Br_color was positively correlated with the flesh percentage DHA. A strong correlation (P < 0.0001) was also found between flesh AsA-totAsA concentrations and firmness and to a lesser extent with fruit percentage dry weight and soluble solids content (measured as °Brix [P < 0.01]; Supplemental Table S5).

QTL Mapping

We identified a total of 27 fruit AsA- and GSH-QTL that were stable for at least two of the three measurement years. These were clustered into four main genomic regions on linkage groups (LGs) 10, 11, 16, and 17 (Fig. 2). The total population variability explained by these AsA-QTL (QTL for AsA, totAsA, and DHA) ranged from 10.7% to 59.5% per trait but was much lower for GSH-QTL (QTL for GSH, totGSH, and GSSG), where population variability ranged between 9.5% and 13.1% per trait (Supplemental Table S6).

Figure 2.
Candidate gene and QTL mapping. Overview is shown for the locations of candidate genes and QTL for mean AsA, totAsA (AsA + DHA), DHA, GSH, totGSH (GSH + GSSG), GSSG, and TAA of fruit flesh and skin tissues as well as leaves and for the rate of flesh browning ...

QTL for Fruit and Leaf AsA Concentrations

Twenty flesh AsA-QTL were identified in the cv Telamon parent and 17 in cv Braeburn. These QTL individually explained from 7.6% to 30% of the total population variance (Supplemental Table S6). Clusters of stable (present in two or more years) and significant (log of the odds [LOD] > 3.5) AsA-QTL were identified at the bottom of LG 11 and in the middle of LG 10 in both parental maps (Table II). Additional AsA-QTL on LGs 3, 6, and 16 were not stable across all years, while parent-specific QTL were located on LGs 3 and 17. In the majority of cases, QTL for totAsA colocated with QTL for flesh AsA concentrations in both parents (Fig. 2; Supplemental Table S6). Of particular interest was the QTL cluster for flesh AsA located on LG 11, as these were stable over all three years and explained a relatively high proportion (up to 27.5%) of the total population variability.

Table II.
QTL for fruit flesh AsA concentrations

Stable clusters of skin AsA-QTL were detected on LGs 9, 10, and 16, which individually accounted for up to 17.3% of population variability, while additional, non-year-stable QTL were found on LGs 2, 3, 11, and 17 (Fig. 2; Supplemental Table S6). The skin AsA-QTL on LGs 11, 16, and 17 colocated with AsA-QTL of flesh, with only the QTL cluster on LG 9 specific for fruit skin AsA-totAsA concentrations.

QTL analysis of leaf AsA, totAsA, and DHA concentrations (2009 only) identified two QTL clusters on LG 6 and 15 (Table III), but only the QTL on LG 6 colocated with that of a fruit AsA-QTL derived from the Telamon parent.

Table III.
QTL mapping for leaf antioxidant concentrations

QTL for Fruit and Leaf GSH Concentrations

From measurements in 2006 and 2009, we identified nine skin and nine flesh GSH-QTL, distributed across nine LGs (Fig. 2; Supplemental Table S6). LOD scores for these QTL ranged from 3.0 to 3.4 and were generally lower than those observed for AsA-QTL. Only one region in the middle of LG 16 contained QTL that were stable across parents and tissues (fruit flesh, fruit skin, and leaf). In 2009, leaf GSH concentrations were measured, leading to the identification of QTL on LGs 6, 12, and 16 (Fig. 2; Table III), with the QTL for leaf GSSG on LG 16 colocating with fruit AsA-QTL in both parents (Fig. 2).

QTL for Fruit TAA

In cv Telamon, two significant TAA-QTL for flesh were detected on LGs 6 and 12, accounting for a total of 23.1% of the population variability (Supplemental Table S6). In cv Braeburn, a major TAA-QTL on LG 11 (LOD = 3.9; population variability of 20.6%), and two minor TAA-QTL on LGs 6 and 16 were identified in flesh tissues, all partially overlapping with QTL for flesh AsA-totAsA.

QTL for Flesh Browning and Fruit Quality Traits

QTL for susceptibility to flesh browning (Br_time and Br_color) were located on LGs 3, 10, 16, and 17 (Fig. 2; Supplemental Table S7). Of these, the QTL for Br_color on LG 16 of cv Braeburn and on LG 17 of cv Telamon were stable over both years of measurements and colocated with AsA-QTL on these LGs (Fig. 2). On LG 10 (both parents) and on LG 17 (cv Telamon), QTL for Br_time mapped to the same position as QTL for AsA, totAsA, and DHA concentrations of flesh tissues. Interestingly, stable QTL for flesh browning (Br_time and Br_color) at the bottom of cv Telamon LG 17 colocated with the stable QTL for flesh DHA concentrations, but no comparable QTL were detected in the “high-fruit-quality” cv Braeburn parent. A large cluster of fruit AsA-QTL in the middle of LG 10 in both parents colocated with a range of previously identified stable QTL for fruit traits, including fresh weight, flesh soluble solids content, and firmness, in this population (Supplemental Table S7; Kenis et al., 2008).

Mapping of Candidate Gene Orthologs and Fruit QTL Clusters

From the literature, we identified 22 structural genes involved in AsA biosynthesis, recycling, degradation, and transport. Using the apple reference consensus coding sequence (CDS) set (http://genomics.research.iasma.it/), 99 apple orthologs of these genes were identified, based on similarity to Arabidopsis sequences and by phylogenetic analysis using published sequences from several additional plant species (Supplemental Table S8). Following an in silico analysis of their positions in the apple genome (Supplemental Table S8), SNP-based markers (Supplemental Table S9) were developed for the subset of CGs that colocated within our QTL using RNA-Seq data of both parents, allowing these CGs to be mapped onto our cv TxB genetic linkage maps (Supplemental Fig. S2). A total of 10 AsA CGs were mapped onto the cv Telamon map and 13 onto the Braeburn map. Generally, these CGs mapped to the expected positions predicted from the Golden Delicious genome assembly. The only exception was for MdDHAR3-1 (MDP0000240690; LG 9), where the marker mapped to the homologous region of LG 17, which contains two other paralogous copies of MdDHAR3 (MdDHAR3-2, MDP0000175246 and MdDHAR3-3, MDP0000156763).

From the l-Gal biosynthetic pathway, one ortholog of GGP (MDP0000172222, VTC2, MdGGP1), with 71% protein sequence identity to Arabidopsis At4g26850.1, mapped to LG 11 of cv Telamon within the cluster of stable flesh AsA-QTL in both parents (Fig. 2). A clone of this gene from Malus sylvestris (GB CN915822) has been shown to have GGP activity when expressed in Escherichia coli (Supplemental Table S10). A paralog of MdGGP1, MdGPP2 (MDP0000288088; 69.5% identity to At4g26850.1 and 89.6% identity to MdGPP1), mapped to LG 3 of cv Braeburn within the flesh AsA-QTL cluster in cv Telamon that was only detected in 2006. Again, a homologous gene from apple (GB CN939721) has been demonstrated to be active (Supplemental Table S10). A third more distantly related ortholog of Arabidopsis GPP, MdGPP3 (MDP0000191488; 30.1% identity to At4g26850.1 and 34% identity to MdGGP1), mapped to LG 10 of cv Telamon and also colocated within a stable AsA-QTL. The markers developed for both MdGGP1 (2006 and 2009) and MdGGP3 (2005) were also the best markers describing the respective QTL (Table II). None of the other genes involved in the l-Gal biosynthetic pathway colocated with any of our AsA-QTL, apart from a putative GDP-mannose pyrophosphorylase (GMP or VTC1; MDP0000323050, MdGMP2-1) with 83.5% identity to At1g74910, which colocated with the non-year-stable flesh AsA-QTL on LG 6 (Fig. 2).

From the alternative AsA biosynthetic pathways, a marker for a putative d-GalUA reductase (MdGalUR1, MDP0000309417; 57.5% identity with At2g37790) mapped to LG 12 of cv Braeburn but did not colocate with any QTL (Fig. 2). However, two MdGalUR2 sequences (MDP0000135496 and MDP0000857724) mapped in silico within the flesh AsA-QTL on LG 16 (Fig. 2; Supplemental Table S8). It was not possible to develop markers for these others, however, as expression levels were too low to identify unique SNPs from our RNA-Seq data.

We identified six DHAR and six MDHAR apple orthologs involved in the recycling of oxidized DHA (Supplemental Table S8). Of these, MdDHAR3-3 (65% identity with AtDHAR3) mapped to the bottom of LG 17 in cv Telamon and colocated with a stable QTL for flesh DHA concentration (2005 and 2009 data) and a minor QTL for flesh AsA concentration (2005; Fig. 2). The other paralogs of MdDHAR3 located in silico to LG 9 (MdDHAR3-1, MDP0000240690 and MdDHAR3-4, MDP0000530903), also within the QTL intervals for skin AsA concentrations (2005 data). There were no colocations, however, between any of the MdMDHARs and antioxidant QTL.

Twenty-one apple orthologs of the NAT protein family were identified in the reference CDS set. Four of these are highly similar to AtNAT7 and AtNAT12, which are the two members considered to be specifically involved in AsA transport, and are located on LGs 11, 13, 15, and 16 (Supplemental Table S8). MdNAT7-2 (MDP0000320308; 58% identity with AtNAT7) mapped to LG 16 of cv Braeburn, colocating with the significant QTL for flesh AsA-totAsA concentrations in both parents (Fig. 2).

Finally, in silico alignment of genes of GSH metabolism (glutathione reductase and glutathione synthase; Supplemental Table S8) identified only one potentially interesting colocation on LG 4 between MdGS2-2 (MDP0000188669; 61% identity to At5g27380) and GSH-QTL for both flesh and skin tissues detected in cv Braeburn.

Sequence Variations in Candidate Regulatory Genes for Fruit AsA Concentrations

High-throughput RNA sequencing (RNA-Seq) was used to detect polymorphisms in the transcript sequences of AsA CGs and to identify allelic SNPs linked to enhanced fruit AsA-totAsA concentrations. Aligning the RNA-Seq reads to the cv Golden Delicious reference consensus CDS set (http://genomics.research.iasma.it/) resulted in a mean read coverage for MdGGP1 (LG 11) of 164.6- and 123.9-fold for cv Braeburn and Telamon, respectively, indicating high transcript expression and good read depth. A total of 10 different SNPs along the 1,704-bp MdGGP1 cv Golden Delicious reference CDS were identified (Supplemental Fig. S3; Supplemental Table S11). This corresponds to one SNP every 170 bp, which is consistent with previous reports in apple cDNA sequences (one SNP every 149 bp; Chagné et al., 2008). These alignments were used to correct the MdGPP1 reference gene model, which was found to contain an incorrectly predicted start codon and two extra codons (Supplemental Figs. S3 and S4). As a result, the majority of the SNPs in MdGGP1 shift into the 5′ untranslated region and only two SNPs are found within the CDS, at reference positions 414 and 885 bp. Neither SNP leads to an amino acid change, but SNP885 is unique to cv Telamon and was developed into a diagnostic SNP marker for mapping and genotyping.

For MdGGP2 (LG 3), mean read coverage following alignment to the cv Golden Delicious reference consensus CDS set was 122.5- and 111.2-fold for cv Braeburn and Telamon, respectively, again indicating high expression and good read depth. A total of eight SNPs were identified in the MdGGP2 CDS, of which three were heterozygous in cv Braeburn and six in cv Telamon (Supplemental Fig. S5; Supplemental Table S12). Both cv Telamon and Braeburn possess nonsynonymous SNPs at positions 74 and 169 bp. However, as both SNPs are present in both parents, they are unlikely to be responsible for the parent-specific allelic differences observed. Both also have a homozygous, but different, SNP at reference position 564, which was used for mapping and genotyping.

Mean read coverage of MdGGP3 following alignment to the cv Golden Delicious CDS was 34.8- and 31.3-fold for cv Braeburn and Telamon, respectively. Both the cv Telamon and Braeburn CDS were noticeably more divergent from the reference CDS than MdGGP1 and MdGGP2, and we identified 79 SNPs in cv Braeburn and 75 SNPs in cv Telamon (Supplemental Fig. S6). A unique nonsynonymous SNP in cv Telamon at reference position 211 was developed for mapping. Both parents contain nonsynonymous SNPs that could influence MdGGP3 protein function, and both also possess a homozygous 33-bp deletion at reference position 272 at the beginning of intron 4, leading to the loss of 11 amino acids. In addition, in cv Braeburn, a heterozygous single-nucleotide deletion at reference position 986 leads to a frame shift and a different and truncated N-terminal amino acid sequence (Supplemental Fig. S7). This suggests that cv Braeburn contains an MdGGP3 allele that codes for a truncated and possibly dysfunctional protein. Alignment of cv Braeburn de novo contigs supports this conclusion, with fragments of both cv Braeburn alleles apparently being present.

The mean read coverage for MdDHAR3-1 (LG 9) was 14.0- and 20.0-fold for cv Braeburn and Telamon, respectively; however, these alignments contained a large region of nonspecifically matched reads, so that only a single SNP could be reliably identified. This SNP-based marker mapped to the homologous region on LG 17, which contains both MdDHAR3-2 and MdDHAR3-3 (MDP0000175246 and MDP0000156763). However, the read coverage for these latter two sequences was too low to allow additional MdDHAR3-specific markers to be developed.

For MdNAT7-2 (LG 16), the mean read coverage was 33.1- and 34.9-fold for cv Braeburn and Telamon, respectively. However, coverage was not uniform, and it was not possible to generate a full-length consensus sequence in either case. In addition, the consensus read-mapping sequences from both parents showed a considerable degree of polymorphism from the cv Golden Delicious reference CDS, with 252 variants detected in cv Braeburn and 151 in cv Telamon.

Allele Association Studies

Orthologous CG sequences, which mapped within AsA-QTL, highlighted several genes with a potential role in regulating fruit AsA concentrations. Of particular interest are the paralogous copies of MdGGP, as all three were mapped within significant flesh AsA-QTL, and two of these QTL were further stable for at least 2 years. In addition, MdNAT7-2 colocated with QTL for flesh antioxidant concentrations on LG 16. To further investigate the involvement of these four genes in the control of apple AsA concentrations, correlations between the allele SNP variants were examined in the TxB progeny as well as in a set of commercial apple cultivars with a wide range of totAsA concentrations.

TotAsA concentrations of flesh tissues in progeny of the TxB population were significantly correlated with SNP variants for MdGGP1 (all three years), MdGGP2 (1 year), MdGGP3 (2 years), and MdNAT7-2 (2 years; Table IV). Specifically, genotypes homozygous for the T allele (T/T genotypes) of MdGGP1 at position 885 had statistically significantly higher flesh totAsA concentrations than the T/C genotypes over the three years of measurement. Similarly, G/A genotypes for MdGGP2 (SNP564) or the T/T genotypes for MdGGP3 (SNP211) also had higher flesh totAsA concentration than G/G or T/C, respectively, but only in 1 or 2 years of measurements (Table IV). For MdNAT7-2 (SNP92), the T/G genotypes had also higher fruit totAsA concentrations in 2006 and 2009 compared with G/G genotypes. Comparing the flesh totAsA concentrations of progeny with the “best” and “worst” SNP variant combinations for these genes showed that the best SNP allelic variants resulted in fruit mean totAsA concentrations that were up to 10-fold higher than those carrying the unfavorable allelic variants (data not shown). In all cases, cv Braeburn was the parent that transferred the beneficial allelic variants for high totAsA concentrations.

Table IV.
Association between fruit totAsA concentrations and candidate gene allelic variants in the TxB mapping population

To validate these associations outside the mapping population, a selection of 22 apple commercial cultivars with known fruit totAsA concentrations (Davey et al., 2004; I. Mellidou and M.W. Davey, unpublished data) were genotyped with the same SNP markers. In the case of MdGPP2 (LG 3) and MdNAT7-2 (LG 16), these markers yielded complex melting curve profiles in the high-resolution melting (HRM) SNP assay and so could not be used with this germplasm. This is presumably the result of additional polymorphisms present in the PCR amplicon region. However, the SNP variants in MdGGP1 and MdGPP3 were successfully determined for 22 and 11 of these cultivars, respectively (Table V). For the remaining 11 cultivars, the complex melting curve profiles for MdGPP3 again precluded us from establishing SNP genotypes. Results show that once again the MdGPP1 T allele and the MdGGP3 T allele were both associated with higher fruit totAsA concentrations in these cultivars. Furthermore, in the 11 cultivars for which we have the SNP genotypes for both MdGGP1 and MdGGP3, the favorable T+T allele combination was always associated with an average 3-fold higher fruit totAsA concentration and a maximum 6-fold higher concentration than those with the C+C combination.

Table V.
Association between fruit totAsA concentrations and candidate gene allelic variants in commercial cultivars

Gene Expression Analysis

To understand the basis for the differences in totAsA contents in fruit of cultivars with different SNP alleles, quantitative PCR expression analysis of CGs was carried out in the flesh of fruit from 10 apple genotypes with high or low AsA concentrations (Fig. 3). In all cases, the expression levels of MdGGP1, MdGGP3, and MdDHAR3-3 were positively and significantly correlated with fruit totAsA concentrations (Fig. 3, B–G), with the mean differences in their relative expression among the high and low AsA groups being 2.0-, 1.5-, and 2.0-fold, respectively. In contrast, there were no statistically significant differences in the mean relative expression of MdGGP2 and MdNAT7-2, although in both cases, cv Braeburn once again had a higher relative expression than cv Telamon tissues.

Figure 3.
Gene expression studies. Changes in flesh totAsA concentrations (nmol g−1 fresh weight; A) and expression levels of MdGGP1 (B), MdGGP2 (C), MdGGP3 (D), MdDHAR3-3 (E), MdNAT7-2 (F), and MdeIF1-4A (G) in the mature fruit of apple genotypes with ...

Analysis of the RNA-Seq data shows that the mean expression levels of MdGGP1 (LG 11) and MdGGP2 (LG 3) are much higher than that of MdGGP3 (LG 10). Reads per kilobase exon model per million (RPKM) values for MdGGP1, MdGGP2, and MdGGP3 in cv Braeburn and Telamon, respectively, were 84.1 and 56.8, 61.4 and 50.4, and 13.9 and 15.3. Although these values are derived from RNA samples isolated during bud and flower development, comparison with the RPKM values of these genes in the low-AsA cv Royal Gala at the mature fruit stage (132 d after full bloom) again shows that MdGGP1 (RPKM 58) and MdGGP2 (RPKM 52) are much more strongly expressed than MdGGP3 (RPKM 5; W.A. Laing, unpublished data). Together with the QTL results, this suggests that MdGGP1 expression is particularly important for the regulation of tissue flesh AsA concentrations.


Antioxidant Concentrations and QTL in the cv TxB Mapping Population: Genetics versus Environment

The absolute concentration of antioxidants varied considerably from year to year, with totAsA and totGSH concentrations being greatest in 2009. This is presumably due to the higher light and temperatures experienced prior to harvest in 2009, factors that are known to influence totAsA concentrations (Davey et al., 2000, 2007; Dumas et al., 2003). However, there was no clear correlation between totAsA concentrations and mean daily temperatures or sunlight hours across the years, underlying the fact that it is difficult to dissociate the individual contributions of temperature, light, development, and genetics from the regulation of antioxidant pools. Nonetheless, the fact that stable (present in at least two measurement years) fruit flesh AsA-QTL (LGs 10, 11, and 17) were found demonstrates a strong genetic component to the control of fruit AsA concentrations in apple.

Environmental factors such as light intensity are expected to have a larger impact on AsA concentrations in the skin compared with flesh tissues and therefore could mask the underlying genetic control mechanisms. Despite this, significant QTL for skin AsA and totAsA concentrations were identified, and most of these colocated with flesh AsA-QTL. Only the QTL cluster on LG 9 (both parents) and on the cv Telamon LG 16 were specific for skin AsA concentrations. These regions on LG 9 and LG 16 have both been shown to contain major QTL for apple fruit polyphenolic contents (Chagné et al., 2007, 2012; Khan et al., 2012). We have previously reported that the sun-exposed (red) side of apples has higher anthocyanin and AsA concentrations and is more resistant to both biotic and abiotic stress than the shaded (green/yellow) side (Davey et al., 2004, 2007), and recently, Bulley et al. (2012) demonstrated that transgenic fruits from tomato and strawberry with elevated AsA contents also had an approximately 50% larger polyphenolic pool. Finally, we report here positive correlations between TAA, AsA, and totAsA concentrations in both flesh and skin tissues and colocations between flesh AsA- and TAA-QTL on LGs 6, 11, and 16. Therefore, the skin-specific AsA-QTL on LGs 9 and 16 may be related to interaction with the polyphenol contents in apple skin tissues. In this regard, the recent report indicating that the accumulation of anthocyanins in Arabidopsis leaves is fine-tuned by the AsA redox state is of particular interest (Page et al., 2012).

For the first time, to our knowledge, we also present QTL for fruit GSH concentrations, but apart from QTL for flesh GSSG concentrations on LG 16 that colocalized with AsA-QTL in both parents, all the other QTL had low LOD scores, and no significant associations between AsA- and GSH-QTL were detected.

The active transport of foliar AsA to developing sink tissues has been demonstrated previously (Franceschi and Tarlyn, 2002; Hancock et al., 2003; Tedone et al., 2004). While both apple fruit and leaf tissues have been shown to be capable of AsA biosynthesis (Davey et al., 2004; Razavi et al., 2005; Li et al., 2010), there were no significant correlations between fruit and leaf totAsA concentrations and only one colocation between leaf and fruit AsA-QTL on LG 6. This would also need to be confirmed, as it is based on only 1 year of measurements. Regardless, these results indicate that screening progeny on the basis of foliar AsA concentrations to identify cultivars rich in fruit vitamin C concentrations will not be highly informative.

Links between Vitamin C Concentrations, Flesh Browning, and Other Fruit Quality Traits

Susceptibility to browning has been linked to flesh DHA concentrations in apple (Davey et al., 2006), and in pears, the development of internal postharvest browning disorder occurs when AsA concentrations decline below a certain threshold value (Franck et al., 2003). Here, the time taken for cut apple flesh to complete browning and the final color intensity following browning were both correlated with DHA concentrations and the percentage DHA of flesh tissues. In addition, stable QTL for browning colocated with stable QTL for flesh DHA concentrations on LG 17. This browning QTL was only detected in cv Telamon, which produces poor-quality fruit and in which the flesh percentage DHA was on average 3-fold higher than in cv Braeburn (data not shown). Taken together, these results suggest that factors that lead to an oxidized AsA pool are associated with susceptibility to flesh browning.

In cherry tomatoes, fruit firmness and AsA concentrations have been reported to be closely correlated (Gilbert et al., 2009), and here, we show that firm apples also generally contain higher flesh AsA concentrations. Flesh AsA-totAsA concentrations and percentage DHA are also significantly correlated with fruit flesh soluble solids content, suggesting a link to the supply of sugar substrates for AsA biosynthesis. However, the significance of this is not clear, as a large cluster of QTL for fruit physiological traits, including fruit firmness, soluble solids content, fresh weight, and flesh browning, all colocated with the fruit flesh AsA-QTL on LG 10; therefore, the correlation with fruit AsA-totAsA concentrations could be due to indirect effects on fruit physiology and quality.

Candidate Genes Regulating AsA Concentrations

Candidate Regulatory Genes from the AsA Biosynthetic Pathways

GMP (VTC1) catalyzes the conversion of d-Man-1-P to GDP-d-Man (Fig. 1), and we observed a colocation between a putative MdGMP (MdGMP2-1) and the flesh AsA-QTL cluster on LG 6. However, these QTL explained only a relatively low proportion of the population variability (10%) and were detected only in 1 year. In tomato (Ioannidi et al., 2009) and kiwifruit (Bulley et al., 2009), fruit AsA concentrations were also not correlated with GMP expression.

GME has been proposed to be critical for the regulation of plant AsA concentrations by several groups (Wolucka and Van Montagu, 2007; Gilbert et al., 2009), and in tomato, SlGME1 colocates with a cluster of stable fruit AsA-QTL (Stevens et al., 2007), while in apple, MdGME transcript levels are highly correlated with AsA concentrations during apple fruit development (Li et al., 2011). However, the data presented here support the results obtained in peach (Prunus persica; Imai et al., 2009) and tomato (Ioannidi et al., 2009), which suggest that this step is not a major point of control in mature apple fruit, at least under our field conditions.

GGP (VTC2; Fig. 1) catalyzes the first committed step of AsA biosynthesis and has also been suggested to be the rate-limiting step for AsA biosynthesis in plants (Linster and Clarke, 2008; Bulley et al., 2009, 2012). Our results here provide strong evidence that MdGGP1 is the only structural gene of AsA metabolism that is tightly linked to flesh AsA concentrations in apple and that this is independent of the environmental conditions, since significant AsA-QTL were detected over all three years of measurement. While we cannot exclude the possibility that these QTL regions contain other regulatory genes, it is interesting that the two other MdGGP paralogs (MdGGP2 on LG 3 and MdGGP3 on LG 10) also mapped within fruit AsA-QTL.

The next step of the l-Gal pathway is catalyzed by GPP (VTC4). Although GPP has been suggested to be important in regulating AsA concentrations in tomato fruit (Ioannidi et al., 2009) and apples (Li et al., 2011), these conclusions are not supported by our results. None of the other genes involved in the main AsA biosynthetic pathway (PMI, PMM, GalDH, or GLDH) mapped to positions located within our AsA-QTL, suggesting that these have no major control over AsA homeostasis in mature apple fruit under Belgian field conditions.

From the alternative (uronic acid) AsA biosynthetic routes (Fig. 1), the reaction catalyzed by GalUR (Enzyme Commission [EC] in strawberry has been the most extensively studied (Agius et al., 2003). d-Galacturonate has been suggested to be an alternative substrate for the synthesis of AsA in fruit skin by precursor feeding experiments in apple fruit tissues (Li et al., 2008), and biosynthesis proceeding via GalUR has been suggested to play a more significant role in the later stages of fruit ripening when breakdown of cell wall pectin takes place (Melino et al., 2009; Cruz-Rus et al., 2011; Badejo et al., 2012). The in silico colocation of two MdGalUR paralogs within flesh AsA-QTL clusters on LG 16 of both parents supports a possible role for GalUR in mature fruits, but there were no other colocations between genes of the alternative AsA biosynthetic pathways and AsA-QTL.

To summarize this section, except for paralogs of the first dedicated step of the l-Gal biosynthetic pathway (MdGGP1-3), our results do not support a key role for any of the other genes of the main l-Gal biosynthetic pathway. However, as reactions before the GGP step also supply substrates for cell wall polysaccharide biosynthesis, mutations in these genes could affect the flux of substrates through to GGP.

CGs from the AsA Recycling Pathway

DHA, the oxidized form of AsA, is unstable and can be readily degraded if not rereduced by the activity of MDHAR (EC and/or DHAR (EC In tomato, an ortholog of MDHAR (SlMDHAR3) has been found to colocate with a stable QTL for fruit AsA contents (Stevens et al., 2007), and MDHAR enzyme activity has been associated with elevated AsA concentrations under chilling stress (Stevens et al., 2008). In apple, we found no colocation between MdMDHARs and any antioxidant QTL. However, MdDHAR3-3 mapped to the bottom of cv Telamon LG 17 within a highly significant, stable QTL for flesh DHA concentrations, and MdDHAR3-1 mapped in silico within QTL for skin AsA concentrations in LG 9, suggesting that orthologs of DHAR could be important in helping to maintain the reduced AsA pool in apple fruit. The colocation of MdDHAR3-3 within a stable QTL for flesh browning suggests that genetic regulation of redox status of the AsA pool via DHAR is important for postharvest fruit quality traits in apple.

CGs Related to Intercellular AsA Transport

Intercellular AsA transport via the NATs could represent an important mechanism of maintaining sink tissue AsA concentrations (Maurino et al., 2006). MdNAT7-2 mapped to the same position as significant fruit AsA-QTL on LG 16 of the cv Braeburn genetic map, and another ortholog of NAT (MdNAT12-2) colocated in silico with a QTL for leaf DHA concentrations on LG 15 of both maps. As NATs are highly expressed in vascular tissues (Maurino et al., 2006) and they colocated with AsA-QTL in both fruit and leaf tissues, these results indicate that they may be involved in the long-distance transport of AsA in apples.

Molecular Markers for Fruit AsA Concentrations

Here, we show that allelic variations in the nucleotide sequences of MdGGP1, MdGGP3, and MdNAT7-2 are associated with up to 10-fold differences in totAsA concentrations of fruit from individuals of our cv TxB mapping population. The remaining MdGGP paralog, MdGGP2, also mapped within an AsA-QTL cluster on LG 3, but this cluster was only detected in 2006. Importantly, fruit totAsA concentrations of 22 commercial apple cultivars were also associated with the same allelic variations in MdGGP1 and MdGGP3, and the T+T combination of alleles for MdGGP1 and MdGGP3 is consistently associated with an up to 6-fold higher fruit totAsA concentration. Sequence analysis of the (corrected) MdGGP1 CDS showed that there are no nonsynonymous SNPs (Supplemental Table S11), while in MdGGP2, nonsynonymous SNPs result in three amino acid changes (Supplemental Table S12), but none of these are predicted to alter structure or function. By comparison, MdGGP3, MdDHAR3-3, and MdNAT7-2 CDS contain multiple synonymous and nonsynonymous SNPs. Further study of the impact of these sequence differences, however, was limited by the complexity and relatively low coverage of the MdDHAR3-3 and MdNAT7-2 transcript sequences. Since the SNPs in MdGGP1 and MdGGP2 are not associated with altered protein function, it is likely that the observed allelic associations are due to linkage with polymorphisms in the promoter region that alter allele expression. Gene expression analyses in a range of phenotypically diverse cultivars demonstrated that MdGGP1 and MdGGP3 as well as MdDHAR3-3 expression levels are all strongly correlated with fruit totAsA concentrations (Fig. 3). Expression levels of MdGGP2 (LG 3) or MdNAT7-2 (LG 16), in contrast, were not significantly correlated, but this could be because they mapped within a non-year-stable AsA-QTL cluster and the expression analyses were carried out in fruit harvested in a different year. Conceivably, therefore, these CGs might still be important only under certain environmental/developmental conditions. Finally, our RNA-Seq data demonstrate that mean expression levels of the third paralog of MdGGP (MdGGP3) are considerably lower than the expression of the other two gene copies in developing tissues as well as in mature fruit. This, in combination with the high number of SNPs and other polymorphisms, suggests that MdGGP3 is less important for the regulation of tissue AsA concentrations and that the translation product may even be nonfunctional.


High fruit AsA concentration is considered to be a desirable trait for consumers, but in apple, as in other hard fruit species, the long juvenile period of plants means that fruit-specific traits cannot be determined until progeny reach maturity around 3 to 5 years after germination. As such, there is much interest in the development of genetic markers for the early selection and/or screening of progeny for fruit AsA concentrations. Our QTL, CG mapping, and analysis of allelic variations across a selection of germplasm show that MdGGP paralogs are consistently linked to fruit totAsA concentrations, clearly indicating a major role for these genes. Gene expression analyses suggest that the regulation of flesh AsA concentrations in apple is primarily through control of the expression levels of MdGGP1, MdGGP3, as well as MdDHAR3-3. The SNP-based markers developed here are thus excellent candidates for the early screening of progeny for increased fruit AsA concentrations. Despite conflicting reports in the literature, our work does not support a major role for the other biosynthetic genes in apple fruit AsA homeostasis, at least under Belgian climatic conditions. However, we propose that MdDHAR3-3 helps regulate the redox state of the AsA pool in apple fruit and that increased flesh DHA concentrations are associated with susceptibility to flesh browning.


Mapping Population

An F1 mapping population consisting of 257 individuals was created from a cross between the apple (Malus × domestica) cultivars Telamon and Braeburn, using cv Telamon as the female parent (Kenis and Keulemans, 2005). For the 2005 and 2006 work, a population of mature trees more than 4 years old was used. Trees were grown on M.9 rootstock at the Rillaar Experimental Field Station in Aarschot, Belgium. For the 2009 work, a regenerated copy of the same population was propagated on M.27 rootstock, grown on the same field.

Harvest and Meteorological Measurements

Fruit from individual progeny were harvested up to twice per week when considered commercially ripe. Ripeness was assessed by experts based on ripening parameters. A minimum of 10 healthy apples per genotype were sampled for AsA, totAsA, GSH, totGSH, and TAA analysis and processed essentially as described previously (Davey et al., 2006). Fruit were immediately transported to the laboratory for analysis. Leaf tissue from 180 genotypes was collected before fruit harvest and stored at −80°C until further analysis. Mature apple fruit were also collected in 2011, pooled into three biological replications, each one consisting of at least three healthy apples, and processed immediately for AsA measurement or stored at −80°C for gene expression studies. Data for mean daytime air temperature (°C) and sun radiation (W m−2) during a period of 5 weeks before the harvest period, starting on August 1, were obtained from the university meteorological station (Heverlee), located 25 km from the experimental field.

Fruit Physiological Measurements and Flesh Browning

For all years of QTL analysis, fruit fresh and dry weight were measured immediately after harvest. Flesh firmness was determined using a penetrometer fitted with an 11-mm-diameter plug, while at the same time, the juice released by penetration of the probe was used to measure the soluble solids concentration (mainly sugars; °Brix) using a digital refractometer (Pocket PAL-1; Atago), essentially as described by Kenis et al. (2008). In 2006 and 2009, the susceptibility of the apple flesh to browning was assessed as Br_time, and Br_color was also evaluated (Davey and Keulemans, 2009).

AsA and GSH Analysis

Samples for metabolite analysis were obtained from skin and flesh fractions from 10 individual fruit by blending using a metaphosphoric acid/EDTA/polyvinylpolypyrrolidone extraction, essentially as described previously (Davey et al., 2003, 2006). Extracts were filtered and injected into an HPLC system (Waters 2690 separation module) equipped with a C18 rocket column (Grace; 53 mm × 7 mm). AsA and GSH were detected at 243 and 197 nm, respectively (Waters 996 photodiode array detector). AsA analysis usually occurred on the same day within a few hours of harvest, although at peak harvest time it was sometimes necessary to store apples for a maximum of up to 48 h in a cool cell at 2°C before extraction. In total, the fruit from 136, 140, or 163 individual progeny were analyzed for AsA, totAsA, and DHA concentrations in the season of 2005, 2006, and 2009, respectively, whereas GSH, totGSH, and GSSG measurements of the same samples were carried out in fruit only in 2006 and 2009.


The acid-soluble TAA of the fruit tissues harvested in 2006 was determined as the sum of the 2,2-azinobis (3-ethylbenzthiazoline-6-sulfonic acid)-reactive substances present in aliquots of the same extracts used to measure AsA and GSH concentrations (Davey et al., 2007). A734 was measured in three technical replicates of all samples using the Multiskan Spectrum Microplate spectrophotometer (Thermo Labsystems). AsA was used as an external standard, and the TAA of each extract was expressed as AsA equivalents.

Linkage Map Analysis and Development of a Bin Mapping Set

The available genetic maps of cv Telamon and Braeburn (Kenis et al., 2008) were further saturated with 14 additional microsatellite loci and CG-based markers to improve marker density in key chromosome regions. The updated linkage maps of cv Telamon and Braeburn were constructed using the JoinMap 3.0 software (Van Ooijen and Voorrips, 2001), essentially as described previously (Davey et al., 2006; Kenis et al., 2008).

A total of 324 and 339 molecular markers were initially ordered in 17 linkage groups, corresponding to the 17 apple chromosomes. To improve the resolution of QTL analysis, markers that were less informative (scored on less than 50% of genotypes), markers with distorted segregation, or markers with two alleles heterozygous in both parents (hk × hk segregation) were discarded, unless the gap between neighbor markers was greater than 10 centimorgans. A subset of 298 markers was used to construct the cv Telamon genetic map, with 96% of the markers highly informative (ab × aa, ef × eg, ab × cd), whereas 313 markers were employed for the cv Braeburn genetic map, which contained less than 10% less informative markers (hk × hk). A bin set of 14 highly informative genotypes was created essentially as described by Chagné et al. (2008). Successful markers were considered those that were heterozygous in at least one parent, had a normal segregation ratio in the bin set, and mapped onto the expected position on the bin maps.

QTL Analysis

QTL for AsA, GSH, and fruit physiological traits were identified using phenotypic data from 2005, 2006, and 2009 and updated genetic maps of the TxB population. QTL mapping was carried out using the MapQTL 4.0 software (Van Ooijen et al., 2002). QTL were detected using interval mapping in combination with restricted multiple QTL model mapping (rMQM) for normally distributed traits with a step size of 1 centimorgan or using the Kruskal-Wallis nonparametric test for nonnormally distributed ones. In rMQM, the most informative marker with the highest LOD score within the intervals of a QTL was selected as a cofactor, and multiple rMQM runs were performed until the cofactors selected were stable, as described previously (Davey et al., 2006). The LOD threshold for the significance of a QTL was selected based on the genome-wide level, using 1,000 permutations at 90% for each trait (around 3.5 or 3.0 for fruit and leaf tissues, respectively). A QTL was classified as significant/nonsignificant and major/minor based on a LOD threshold of 3.5 and 20% of explained population variability. QTL identified were described by the marker with the highest LOD score in the corresponding QTL region, and QTL regions were defined as the 1.5-LOD support interval. Linkage maps and QTL were drawn using MapChart 3.0 software (Voorrips, 2002).

Candidate Gene Mapping

The sequences of AsA-related CGs from Arabidopsis or other plant species were retrieved from GenBank (http://www.ncbi.nlm.nih.gov/genbank/), PLAZA 2.0 (http://bioinformatics.psb.ugent.be/plaza/), Phytozome Genome Browser (http://www.phytozome.net), or the SOL Genomics Network (http://solgenomics.net). Orthologous sequences were retrieved from the cv Golden Delicious assembly (Velasco et al., 2010) using BLASTX versus the apple consensus CDS set. Gene positions on the genome assembly and their sequences were extracted from the IASMA genome browser (http://genomics.research.iasma.it). The amino acid sequences for each CG were aligned using the BLOSUM62 matrix, and phylogenetic analysis was conducted using the neighbor-joining method and 1,000 bootstrap replicates in Geneious software version 5.4.6 (Drummond et al., 2011) in order to determine the number of copies present in apple for each Arabidopsis accession. The positions of these homologs were aligned in silico across the 17 linkage groups of the cv TxB map, based on the physical positions of framework simple sequence repeats.

SNP Marker Development

RNA-Seq data from cv Telamon and Braeburn were used to identify SNPs and polymorphisms in CG sequences. In total, 48.1 and 73.6 million 75-bp paired end reads, respectively, were generated from RNA pooled from eight different developmental stages of tree bud development and two tissue types. Reads were trimmed and then mapped against the apple consensus CDS set, downloaded from the GDR Web site (http://www.rosaceae.org/), using default mapping parameters within CLCBio Genomics Workbench version 5.5. CLCBio software was also used to carry out the de novo assembly of reads into contigs for each cultivar separately, again using default alignment parameters and the automatic scaffolding function. SNPs were called relative to the cv Golden Delicious reference CDS on the basis of using a minimum read depth of 6-fold and a percentage proportion coverage of greater than 25%. Finally, to control gene annotation, the consensus sequences from both the mapped reads and the de novo contigs were extracted and aligned against the annotated genomic DNA contigs used to create the apple genome assembly, again downloaded from GDR and using the Geneious version 5.6 software package. Functional descriptions of consensus CDS and translated protein sequences were confirmed using Blast2GO software (http://www.blast2go.com/b2ghome) as well InterPro (Geneious version 5.6). To validate gene function, clones of GGPs were expressed in Escherichia coli, and GGP activity was measured essentially as described previously (Laing et al., 2007).

PCR primers were developed to identify length polymorphisms, or flanking SNPs, to allow progeny genotyping using the HRM technique (Chagné et al., 2008; Studer et al., 2009). Gene-specific SNP primer pairs were designed using Primer3 software version 4.0 (http://frodo.wi.mit.edu/primer3/) employing the following conditions: product size between 75 and 200 bp, with optimum 120 bp spanning the identified SNPs; GC% ranging from 40% to 55%; self-complementarity and 3′ self-complementarity were set at 4 and 1, respectively; and all the other parameters were set at default. All CG-derived markers developed were screened in the cv TxB bin mapping set and both parents.

Genomic DNA from leaves was extracted from approximately 100-mg powdered aliquots using the Genomic DNA Purification kit (MBI Fermentas). HRM conditions were as follows: the PCR for each sample was carried out in 15 μL containing the HRM Master Mix (Type-it HRM PCR Kit; Qiagen), 10 to 20 μm of each primer, and 5 ng of genomic DNA. Cycling conditions were incubation at 95°C for 5 min, followed by 40 cycles of 95°C (10 s) and 55°C (30 s). HRM analysis was followed immediately by denaturation from 65°C to 95°C, rising by 0.1°C in each step. The HRM analysis was carried out in a Rotorgene Q thermocycler software (Qiagen). Markers that mapped to their expected positions and gave an expected segregation rate were further screened over 100 individuals from the cv TxB population to construct new genetic maps, as described previously.

Allelic Typing and Gene Expression Studies

Gene expression studies were performed following Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines (Bustin et al., 2009). Total RNA was isolated from flesh fruit tissues using a modified cetyl-trimethyl-ammonium bromide method (Gasic et al., 2004) and the RNeasy Plus Mini Kit following the manufacturer’s instructions (Qiagen). The purity of total RNA extracted was determined as the 260/280-nm or 260/230-nm ratio with NanoDrop 2000 (Thermo Scientific), and the integrity was checked by electrophoresis on a 1% agarose gel (Gel Doc EZ Imager; Bio-Rad). One microgram of total RNA was reverse transcribed to cDNA with SuperScript II (Invitrogen) using random oligo(dT) primers. The expression profiles of CGs considered to be key control points for AsA accumulation (MdGGP1, MdGGP2, MdGGP3, MdDHAR3-3, and MdNAT7-2; primer pair sequences are listed in Supplemental Table S9) were validated by real-time quantitative PCR using SYBR Green I technology on a Rotor Gene Q (Qiagen). 18S-rRNA (5′-GTTACTTTTAGGACTCCGCC-3′ and 5′-TTCCTTTAAGTTTCAGCCTTG-3′), Malus spp. eIF-4A (5′-ATCAGGCTCATCCCGTGT-3′ and 5′-AGCAACACCCTTCCTTCC-3′; Zubini et al., 2007), and Arabidopsis Actin11 (5′-GGACCTTGCAGGCCGTGACC-3′ and 5′-AACCTCCGGGCAGCGGAATC-3′) were used as reference genes. The stability of the expression of the references genes was assessed with NormFinder software (http://www.mdl.dk/publicationsnormfinder.htm). All reactions were set up in duplicate, containing a 1-μL cDNA template (50 ng), 7.5 μL of Absolute QPCR SYBR Green Mix, and 1 μL of each primer (3.75 mm) in a final volume of 15 μL. The cycling conditions were as follows: denaturation step at 95°C for 10 min, followed by 45 cycles of denaturation at 95°C for 20 s, annealing at 63°C for 20 s, and extension at 72°C for 20 s. A melting curve analysis was performed, ranging from 55°C to 95°C, with temperature increasing in steps of 0.5°C s−1. The criteria of acceptance for reaction efficiency ranged from 0.83 to 1.03 and both r and r2 > 0.97. For each run, a standard curve was included based on cDNA pooled from all samples to be analyzed in a range of different dilutions. The relative quantification of expression levels was performed using the comparative cycle threshold method (Pfaffl, 2001). Two technical replications were performed for each of the three biological replications per sample. All expression data were calculated as an expression ratio relative to the geometric mean of the expression of 18S-rRNA and Actin, as these showed the highest stability in our sample set, with stability values (Vandesompele et al., 2002) of 0.34 and 0.47, respectively.

Descriptive Statistical Analysis

Statistical analysis of all traits was carried out using the SAS 9.2 software package (SAS Institute). Mean values, sd values, ranges, and skewness of the F1 progeny were calculated for each year and trait, while the normality of the distributions was tested using the Shapiro-Wilk test (Shapiro and Wilk, 1965). Pearson correlation coefficients were employed to compare mean values of the traits over the three years. The significance of the mean fold difference in totAsA concentrations of genotypes carrying different alleles/allele combinations, as well as the significance of the mean relative expression of CGs of high/low AsA genotypes, were tested with Student’s t test (*P < 0.05, **P < 0.01, and ***P < 0.001).

Supplemental Data

The following materials are available in the online version of this article.

Supplementary Material

Supplemental Data:


We acknowledge stimulating discussions with Otto van Poeselaere.


quantitative trait loci
l-ascorbic acid
dehydroascorbate reductase
monodehydroascorbate reductase
candidate gene
single-nucleotide polymorphism
total ascorbic acid
total glutathione
Telamon × Braeburn
total antioxidant activity
oxidized glutathione
time taken for cut surfaces to brown
final color of the affected tissue
linkage groups
log of the odds
coding sequence
high-resolution melting
reads per kilobase exon model per million
restricted multiple QTL model mapping
Enzyme Commission


  • Agius F, González-Lamothe R, Caballero JL, Muñoz-Blanco J, Botella MA, Valpuesta V. (2003) Engineering increased vitamin C levels in plants by overexpression of a D-galacturonic acid reductase. Nat Biotechnol 21: 177–181 [PubMed]
  • Badejo AA, Wada K, Gao Y, Maruta T, Sawa Y, Shigeoka S, Ishikawa T. (2012) Translocation and the alternative D-galacturonate pathway contribute to increasing the ascorbate level in ripening tomato fruits together with the D-mannose/L-galactose pathway. J Exp Bot 63: 229–239 [PMC free article] [PubMed]
  • Bulley SM, Rassam M, Hoser D, Otto W, Schünemann N, Wright M, MacRae E, Gleave A, Laing W. (2009) Gene expression studies in kiwifruit and gene over-expression in Arabidopsis indicates that GDP-L-galactose guanyltransferase is a major control point of vitamin C biosynthesis. J Exp Bot 60: 765–778 [PMC free article] [PubMed]
  • Bulley SM, Wright M, Rommens C, Yan H, Rassam M, Lin-Wang K, Andre C, Brewster D, Karunairetnam S, Allan AC, et al. (2012) Enhancing ascorbate in fruits and tubers through over-expression of the L-galactose pathway gene GDP-L-galactose phosphorylase. Plant Biotechnol J 10: 390–397 [PubMed]
  • Bustin SA, Benes V, Garson JA, Hellemans J, Huggett J, Kubista M, Mueller R, Nolan T, Pfaffl MW, Shipley GL, et al. (2009) The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin Chem 55: 611–622 [PubMed]
  • Chagné D, Carlisle CM, Blond C, Volz RK, Whitworth CJ, Oraguzie NC, Crowhurst RN, Allan AC, Espley RV, Hellens RP, et al. (2007) Mapping a candidate gene (MdMYB10) for red flesh and foliage colour in apple. BMC Genomics 8: 212. [PMC free article] [PubMed]
  • Chagné D, Gasic K, Crowhurst RN, Han Y, Bassett HC, Bowatte DR, Lawrence TJ, Rikkerink EHA, Gardiner SE, Korban SS. (2008) Development of a set of SNP markers present in expressed genes of the apple. Genomics 92: 353–358 [PubMed]
  • Chagné D, Krieger C, Rassam M, Sullivan M, Fraser J, André C, Pindo M, Troggio M, Gardiner SE, Henry RA, et al. (2012) QTL and candidate gene mapping for polyphenolic composition in apple fruit. BMC Plant Biol 12: 12–28 [PMC free article] [PubMed]
  • Chen Z, Young TE, Ling J, Chang SC, Gallie DR. (2003) Increasing vitamin C content of plants through enhanced ascorbate recycling. Proc Natl Acad Sci USA 100: 3525–3530 [PMC free article] [PubMed]
  • Conklin PL, Gatzek S, Wheeler GL, Dowdle J, Raymond MJ, Rolinski S, Isupov M, Littlechild JA, Smirnoff N. (2006) Arabidopsis thaliana VTC4 encodes L-galactose-1-P phosphatase, a plant ascorbic acid biosynthetic enzyme. J Biol Chem 281: 15662–15670 [PubMed]
  • Cruz-Rus E, Amaya I, Sánchez-Sevilla JF, Botella MA, Valpuesta V. (2011) Regulation of L-ascorbic acid content in strawberry fruits. J Exp Bot 62: 4191–4201 [PMC free article] [PubMed]
  • Davey MW, Auwerkerken A, Keulemans J. (2007) Relationship of apple vitamin C and antioxidant concentrations to harvest date and postharvest pathogen infection. J Sci Food Agric 87: 802–813
  • Davey MW, Dekempeneer E, Keulemans J. (2003) Rocket-powered high-performance liquid chromatographic analysis of plant ascorbate and glutathione. Anal Biochem 316: 74–81 [PubMed]
  • Davey MW, Franck C, Keulemans J. (2004) Distribution, developmental and stress responses of antioxidant metabolism in Malus. Plant Cell Environ 27: 1309–1320
  • Davey MW, Gilot C, Persiau G, Ostergaard J, Han Y, Bauw GC, Van Montagu MC. (1999) Ascorbate biosynthesis in Arabidopsis cell suspension culture. Plant Physiol 121: 535–543 [PMC free article] [PubMed]
  • Davey MW, Kenis K, Keulemans J. (2006) Genetic control of fruit vitamin C contents. Plant Physiol 142: 343–351 [PMC free article] [PubMed]
  • Davey MW, Keulemans J. (2004) Determining the potential to breed for enhanced antioxidant status in Malus: mean inter- and intravarietal fruit vitamin C and glutathione contents at harvest and their evolution during storage. J Agric Food Chem 52: 8031–8038 [PubMed]
  • Davey MW, Keulemans J. (2009) Nutritional enhancement in apple: identification of QTL for fruit vitamin C concentration and their stability over different production years. Acta Hortic 814: 591–598
  • Davey MW, Van Montagu M, Inze D, Sanmartin M, Kanellis A, Smirnoff N, Benzie IJJ, Strain JJ, Favell D, Fletcher J. (2000) Plant L-ascorbic acid: chemistry, function, metabolism, bioavailability and effects of processing. J Sci Food Agric 80: 825–860
  • de Koning H, Diallinas G. (2000) Nucleobase transporters (review). Mol Membr Biol 17: 75–94 [PubMed]
  • Diplock AT, Charleux JL, Crozier-Willi G, Kok FJ, Rice-Evans C, Roberfroid M, Stahl W, Viña-Ribes J. (1998) Functional food science and defence against reactive oxidative species. Br J Nutr (Suppl 1) 80: S77–S112 [PubMed]
  • Drummond AJ, Ashton B, Buxton S, Cheung M, Cooper A, Duran C, Field M, Heled J, Kearse M, Markowitz S, et al. (2011) Geneious v5.4. http://www.geneious.com/ (January 20, 2012)
  • Dumas Y, Dadomo M, Di Lucca G, Grolier P. (2003) Effects of environmental factors and agricultural techniques on antioxidant concentration of tomatoes. J Sci Food Agric 83: 369–382
  • Eltayeb AE, Kawano N, Badawi GH, Kaminaka H, Sanekata T, Shibahara T, Inanaga S, Tanaka K. (2007) Overexpression of monodehydroascorbate reductase in transgenic tobacco confers enhanced tolerance to ozone, salt and polyethylene glycol stresses. Planta 225: 1255–1264 [PubMed]
  • Franceschi VR, Tarlyn NM. (2002) l-Ascorbic acid is accumulated in source leaf phloem and transported to sink tissues in plants. Plant Physiol 130: 649–656 [PMC free article] [PubMed]
  • Franck C, Baetens M, Lammertyn J, Verboven P, Davey MW, Nicolaï BM. (2003) Ascorbic acid concentration in cv. Conference pears during fruit development and postharvest storage. J Agric Food Chem 51: 4757–4763 [PubMed]
  • Gasic K, Hernandez A, Korban SS. (2004) RNA extraction from different apple tissues rich in polyphenols and polysaccharides for cDNA library construction. Plant Mol Biol Rep 22: 437–438
  • Gautier H, Massot C, Stevens R, Serino S, Genard M. (2009) Regulation of tomato fruit ascorbate concentration is more highly dependent on fruit irradiance than leaf irradiance. Ann Bot (Lond) 103: 495–504 [PMC free article] [PubMed]
  • Gilbert L, Alhagdow M, Nunes-Nesi A, Quemener B, Guillon F, Bouchet B, Faurobert M, Gouble B, Page D, Garcia V, et al. (2009) GDP-D-mannose 3,5-epimerase (GME) plays a key role at the intersection of ascorbate and non-cellulosic cell-wall biosynthesis in tomato. Plant J 60: 499–508 [PubMed]
  • Hancock RD, McRae D, Haupt S, Viola R. (2003) Synthesis of l-ascorbic acid in the phloem. BMC Plant Biol 3: 7. [PMC free article] [PubMed]
  • Hancock RD, Walker PG, Pont SDA, Marquis N, Vivera S, Gordon SL, Brennan RM, Viola R. (2007) L-Ascorbic acid accumulation in fruit of Ribes nigrum occurs by in situ biosynthesis via the L-galactose pathway. Funct Plant Biol 34: 1080–1091
  • Haroldsen VM, Chi-Ham CL, Kulkarni S, Lorence A, Bennett AB. (2011) Constitutively expressed DHAR and MDHAR influence fruit, but not foliar ascorbate levels in tomato. Plant Physiol Biochem 49: 1244–1249 [PMC free article] [PubMed]
  • Imai T, Ban Y, Terakami S, Yamamoto T, Moriguchi T. (2009) L-Ascorbate biosynthesis in peach: cloning of six L-galactose pathway-related genes and their expression during peach fruit development. Physiol Plant 136: 139–149 [PubMed]
  • Ioannidi E, Kalamaki MS, Engineer C, Pateraki I, Alexandrou D, Mellidou I, Giovannonni J, Kanellis AK. (2009) Expression profiling of ascorbic acid-related genes during tomato fruit development and ripening and in response to stress conditions. J Exp Bot 60: 663–678 [PMC free article] [PubMed]
  • Kenis K, Keulemans J. (2005) Genetic linkage maps of two apple cultivars (Malus × domestica Borkh.) based on AFLP and microsatellite markers. Mol Breed 15: 205–219
  • Kenis K, Keulemans J, Davey MW. (2008) Identification and stability of QTLs for fruit quality traits in apple. Tree Genet Genomes 4: 647–661
  • Khan SA, Chibon P-Y, de Vos RCH, Schipper BA, Walraven E, Beekwilder J, van Dijk T, Finkers R, Visser RGF, van de Weg EW, et al. (2012) Genetic analysis of metabolites in apple fruits indicates an mQTL hotspot for phenolic compounds on linkage group 16. J Exp Bot 63: 2895–2908 [PMC free article] [PubMed]
  • Laing WA, Wright MA, Cooney J, Bulley SM. (2007) The missing step of the L-galactose pathway of ascorbate biosynthesis in plants, an L-galactose guanyltransferase, increases leaf ascorbate content. Proc Natl Acad Sci USA 104: 9534–9539 [PMC free article] [PubMed]
  • Li M, Chen X, Wang P, Ma F. (2011) Ascorbic acid accumulation and expression of genes involved in its biosynthesis and recycling in developing apple fruit. J Am Soc Hortic Sci 136: 231–238
  • Li M, Ma F, Guo C, Liu J. (2010) Ascorbic acid formation and profiling of genes expressed in its synthesis and recycling in apple leaves of different ages. Plant Physiol Biochem 48: 216–224 [PubMed]
  • Li M, Ma F, Shang P, Zhang M, Hou C, Liang D. (2009) Influence of light on ascorbate formation and metabolism in apple fruits. Planta 230: 39–51 [PubMed]
  • Li M-J, Ma F-W, Zhang M, Pu F. (2008) Distribution and metabolism of ascorbic acid in apple fruits (Malus × domestica Borkh cv. Gala). Plant Sci 174: 606–612
  • Linster CL, Clarke SG. (2008) L-Ascorbate biosynthesis in higher plants: the role of VTC2. Trends Plant Sci 13: 567–573 [PMC free article] [PubMed]
  • Lorence A, Chevone BI, Mendes P, Nessler CL. (2004) Myo-inositol oxygenase offers a possible entry point into plant ascorbate biosynthesis. Plant Physiol 134: 1200–1205 [PMC free article] [PubMed]
  • Maurino VG, Grube E, Zielinski J, Schild A, Fischer K, Flügge UI. (2006) Identification and expression analysis of twelve members of the nucleobase-ascorbate transporter (NAT) gene family in Arabidopsis thaliana. Plant Cell Physiol 47: 1381–1393 [PubMed]
  • Melino VJ, Soole KL, Ford CM. (2009) Ascorbate metabolism and the developmental demand for tartaric and oxalic acids in ripening grape berries. BMC Plant Biol 9: 145–159 [PMC free article] [PubMed]
  • Noctor G. (2006) Metabolic signalling in defence and stress: the central roles of soluble redox couples. Plant Cell Environ 29: 409–425 [PubMed]
  • Noctor G, Foyer CH. (1998) Ascorbate and glutathione: keeping active oxygen under control. Annu Rev Plant Physiol Plant Mol Biol 49: 249–279 [PubMed]
  • Page M, Sultana N, Paszkiewicz K, Florance H, Smirnoff N. (2012) The influence of ascorbate on anthocyanin accumulation during high light acclimation in Arabidopsis thaliana: further evidence for redox control of anthocyanin synthesis. Plant Cell Environ 35: 388–404 [PubMed]
  • Pfaffl MW. (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 29: e45. [PMC free article] [PubMed]
  • Razavi F, Keulemans J, Davey MW. (2005) A study of the L-ascorbate biosynthetic capacity of apple fruit. Commun Agric Appl Biol Sci 70: 213–216 [PubMed]
  • Shapiro SS, Wilk MB. (1965) An analysis of variance test for normality. Biometrika 52: 591–599
  • Smirnoff N, Conklin PL, Loewus FA. (2001) Biosynthesis of ascorbic acid in plants: a renaissance. Annu Rev Plant Physiol Plant Mol Biol 52: 437–467 [PubMed]
  • Stevens R, Buret M, Duffé P, Garchery C, Baldet P, Rothan C, Causse M. (2007) Candidate genes and quantitative trait loci affecting fruit ascorbic acid content in three tomato populations. Plant Physiol 143: 1943–1953 [PMC free article] [PubMed]
  • Stevens R, Page D, Gouble B, Garchery C, Zamir D, Causse M. (2008) Tomato fruit ascorbic acid content is linked with monodehydroascorbate reductase activity and tolerance to chilling stress. Plant Cell Environ 31: 1086–1096 [PubMed]
  • Studer B, Jensen LB, Fiil A, Asp T. (2009) ‘Blind’ mapping of genic DNA sequence polymorphisms in Lolium perenne L. by high resolution melting curve analysis. Mol Breed 24: 191–199
  • Tedone L, Hancock RD, Alberino S, Haupt S, Viola R. (2004) Long-distance transport of l-ascorbic acid in potato. BMC Plant Biol 4: 16. [PMC free article] [PubMed]
  • Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. (2002) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 3: H0034 [PMC free article] [PubMed]
  • Van Ooijen JW, Boer MP, Jansen RC, Maliepaard C. (2002) MapQTL 4.0: Software for the Calculation of QTL Positions on Genetic Maps. Plant Research International, Wageningen, The Netherlands
  • Van Ooijen JW, Voorrips RE. (2001) JoinMap 3.0, Software for the Calculation of Genetic Linkage Maps. Plant Research International, Wageningen, The Netherlands
  • Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, Kalyanaraman A, Fontana P, Bhatnagar SK, Troggio M, Pruss D, et al. (2010) The genome of the domesticated apple (Malus × domestica Borkh.). Nat Genet 42: 833–839 [PubMed]
  • Veltman RH, Sanders MG, Persijn ST, Peppelenbos HW, Oosterhaven J. (1999) Decreased ascorbic acid levels and brown core development in pears (Pyrus communis L cv. Conference). Physiol Plant 107: 39–45
  • Voorrips RE. (2002) MapChart: software for the graphical presentation of linkage maps and QTLs. J Hered 93: 77–78 [PubMed]
  • Wheeler GL, Jones MA, Smirnoff N. (1998) The biosynthetic pathway of vitamin C in higher plants. Nature 393: 365–369 [PubMed]
  • Wolucka BA, Van Montagu M. (2003) GDP-mannose 3′,5′-epimerase forms GDP-L-gulose, a putative intermediate for the de novo biosynthesis of vitamin C in plants. J Biol Chem 278: 47483–47490 [PubMed]
  • Wolucka BA, Van Montagu M. (2007) The VTC2 cycle and the de novo biosynthesis pathways for vitamin C in plants: an opinion. Phytochemistry 68: 2602–2613 [PubMed]
  • Yabuta Y, Mieda T, Rapolu M, Nakamura A, Motoki T, Maruta T, Yoshimura K, Ishikawa T, Shigeoka S. (2007) Light regulation of ascorbate biosynthesis is dependent on the photosynthetic electron transport chain but independent of sugars in Arabidopsis. J Exp Bot 58: 2661–2671 [PubMed]
  • Zorrilla-Fontanesi Y, Cabeza A, Domínguez P, Medina JJ, Valpuesta V, Denoyes-Rothan B, Sánchez-Sevilla JF, Amaya I. (2011) Quantitative trait loci and underlying candidate genes controlling agronomical and fruit quality traits in octoploid strawberry (Fragaria × ananassa). Theor Appl Genet 123: 755–778 [PubMed]
  • Zubini P, Baraldi E, De Santis A, Bertolini P, Mari M. (2007) Expression of anti-oxidant enzyme genes in scald-resistant ‘Belfort’ and scald-susceptible ‘Granny Smith’ apples during cold storage. J Hortic Sci Biotechnol 82: 149–155

Articles from Plant Physiology are provided here courtesy of American Society of Plant Biologists
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...