The expression profiles of carotenoid metabolism pathway genes are one of the important factors that regulate carotenoid content and composition in the fruit of Citrus cultivars. To identify regulatory factors of gene expressions in the carotenoid pathway, expression quantitative trait loci (eQTL) analysis was applied using a hybrid population. In the analyzed population, significant eQTLs for phytoene synthase (PSY), phytoene desaturase (PDS), ζ-carotene desaturase (ZDS), β-ring hydroxylase (HYb), and zeaxanthin epoxidase (ZEP) were detected. Among these eQTLs, eQTLs for HYb and ZEP were located on their responsible gene loci, respectively. This result indicates that the expression level of HYb and ZEP was influenced by cis-elements in their up-stream regions. Conversely, eQTLs for PDS and ZDS were located in different loci from those of the responsible genes, indicating that the loci were trans-regulating factors. It also suggested that expression levels of PDS and ZDS were regulated by a common transcription factor because their eQTLs were co-localized. Significant eQTL of lycopene β-cyclase (LCYb) and 9-cis-epoxycarotenoid dioxygenase (NCED) were not detected in this population. The major factors to control gene expression depend on each carotenoid metabolism gene, and their combination might cause complex carotenoid metabolism and extend the variation of content and composition in citrus varieties. eQTL analysis was demonstrated as a powerful tool to identify cis- or trans-regulation for carotenoid metabolism genes, as well as in generating functionally defined markers for marker-assisted selection to increase the carotenoid content of citrus in breeding programs.
Citrus fruits contain significant amounts of various carotenoids and some of them are known to benefit human health. Approximately 115 different carotenoids have been reported in citrus fruits, and the color of the fruit and peel are caused by carotenoid accumulation (Stewart and Wheaton, 1973). The carotenoid content and composition in Citrus vary greatly among varieties (Gross, 1987; Ikoma et al., 2001; Kato et al., 2004). As a typical example, the Satsuma mandarin (Citrus unshiu Marc.) and other mandarins predominantly accumulate β-cryptoxanthin (β-Cry) in the flavedo and juice sacs in mature fruit (Goodner et al., 2001; Ikoma et al., 2001), while sweet orange (Citrus sinensis Osbeck) accumulates violaxanthins, predominantly 9-cis-violaxanthin (Lee and Castle, 2001; Molnár and Szabolcs, 1980) (Fig. 1). The enrichment of carotenoids with health-promoting functions, especially the β-Cry content, is an important objective in the Japanese citrus breeding program. Therefore, elucidation of the molecular mechanisms that regulate carotenoid content and composition among citrus varieties is of great importance when attempting to produce a new cultivar rich in health-promoting carotenoids. Moreover, the development of DNA markers is required to progress molecular breeding to increase carotenoid content or alter carotenoid composition.
Carotenoid metabolic pathway in plants.
Many reports have been published to characterize the physiological and molecular features of carotenoid biosynthesis in Citrus. Phytoene synthase (PSY) and β-carotene hydroxylase (HYb) are the key enzymes that regulate carotenoid contents in the metabolic pathway. Elevation of the transcript level of the PSY gene significantly influenced the onset of coloration in the flavedo and juice sac of fruit (Ikoma et al., 2001; Kim et al., 2001). In addition to PSY, the transcription levels of carotenoid metabolism genes, such as lycopene β-cyclase (LCYb), HYb, ζ-carotene desaturase (ZDS), and zeaxan-thin epoxidase (ZEP), influence carotenoid components in citrus fruit (Fanciullino et al., 2007; Kato et al., 2004; Rodrigo et al., 2004). Thus, the physiological features of carotenoid accumulation in citrus fruit during fruit maturation have been characterized. However, little is known about the transcription regulation of these genes and mechanism causing carotenoid variations of content and composition among citrus varieties. Therefore, further study is required for new insight into citrus carotenoid metabolism to find the genetic factors influencing the transcription of each carotenoid biosynthesis gene.
Recently, the method of expression quantitative trait locus (eQTL) analysis was developed, which has great potential to elucidate transcriptional regulation (Doerge, 2002; Jansen and Nap, 2001). In eQTL analysis, transcript levels, when assessed in a mapping population, are analyzed as quantitative traits and their variation is used to map them as eQTLs. The eQTLs investigated in a population provide the necessary information required for identifying genes or loci that control quantitative traits. In plants, global eQTL analyses of gene expression have been applied to detect cis-polymorphisms controlling individual genes as well as to identify transeQTLs that regulate individual genes from remote loci (DeCook et al., 2006; Keurentjes et al., 2007; Potokina et al., 2008; West et al., 2007). Initial observations from global transcriptome QTL mapping studies indicated that eQTLs are located cis- and trans-relative to the physical positions of the genes. It has been reported that the locations of eQTLs that regulate a gene’s activity could be correlated with those of QTLs for traditional phenotypic traits and so provide additional clues as to the genetic basis of quantitative genetic variation (Bystrykh et al., 2005; Hubner et al., 2005; Kirst et al., 2004). So far, the transcription factors regulating the carotenoid metabolism genes have not been identified, except for the RAP2.2 transcription factor, which regulates PSY gene expression in Arabidopsis leaves (Welsch et al., 2007). Therefore, the identification of novel transcriptional factors by eQTL analysis would help our understanding of the genetic regulation of the variations in the quantities and compositions of carotenoid accumulation in citrus fruit. The eQTL approach could also provide stable functional markers based on molecular information to select breeding progeny with higher carotenoid contents.
In this report, eQTL analysis was applied to identify possible cis- and trans-regulatory regions related to citrus carotenoid metabolism. A possible procedure to generate haplotype-specific markers that could be practically applied to marker-assisted selection (MAS) in breeding programs was also discussed.
All plants used in the experiments were cultivated in the research field of the National Institute of Fruit Tree Science, Citrus Research Center, Okitsu, Shimizu, Shizuoka, Japan. For genetic analysis, a hybrid population, designated as the AG population, was generated by crossing ‘Okitsu-46 (A255)’ and ‘Nou-5 (G434)’, which were top-grafted on Satsuma mandarin interstocks to promote flowering and fruiting. The female parent of the AG population, ‘Okitsu-46’, was derived from hybridization between ‘Sweet Spring’ (‘Ueda unshiu’ (C. unshiu) × ‘Hassaku’ (C. hassaku)) × ‘Trovita’ orange (C. sinensis); the male parent, ‘Nou-5’, was derived from hybridization between ‘Lee’ (‘Clementine’ mandarin × ‘Orlando’ tangelo) × ‘Mukaku kishu’ (C. kinokuni Hort. ex Tanaka). The two parental clones were used for evaluation of the parental carotenoid levels.
The AG population consisted of 93 individuals and was used to construct a linkage map by DNA markers, as described previously (Omura et al., 2003). Among the population, juice sacs of fruits were sampled from 51 individuals on December 7, 2003 and used for gene expression analysis of carotenoid metabolism genes. The fruits analyzed in the experiment were the same specimens analyzed for carotenoid contents by QTL analysis (Sugiyama et al., 2011).
RNA extraction and quantification of expression levels using the TaqMan probe/primer systemTotal RNA was extracted from juice sacs of harvested fruit (more than three fruits per individual) of the AG population by the method described by Ikoma et al. (1996). An RNeasy Minikit (Qiagen, Hilden, Germany) with on-column DNase digestion purified the RNA samples.
Total RNAs (0.2 μg) were used to synthesize cDNAs with random hexamers at 37°C for 60 min using TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA, USA). Expression levels were estimated using quantitative RT-PCR (qRT-PCR) on the following carotenoid biosynthesis genes: PSY, PDS, ZDS, LCYb, HYb, ZEP, 9-cis-epoxycarotenoid dioxygenase2 (NCED2), and 9-cis-epoxycarotenoid dioxygenase3 (NCED3). The total expression level of each gene was evaluated using a TaqMan minor groove binder (MGB) probe with a set of primers/probe. Primer Express software was used to design the primers and probes (Applied Biosystems) to amplify common cDNA sequences from the Citrus cultivars examined. The TaqMan Ribosomal RNA Control Reagents 4,7,2′-trichloro-7′-phenyl-6-carboxyfluorescein (VIC) Probe (Applied Biosystems) was used as an endogenous control. Each reaction contained 900 nM primers, 250 nM TaqMan MGB Probe, and 2.5 μL template cDNA. The thermal cycling conditions were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. The ABI PRISM 7300 Sequence Detection System Software (Applied Biosystems) analyzed the levels of gene expression and relative quantities of gene expression were normalized against the expression of 18S ribosomal RNA. The qRT-PCR score of each hybrid was described relative to the score of ‘Okitsu-46’, which had an estimated value of 1.0.
Analyses of correlation coefficientsThe data of carotenoid contents described in QTL analysis (Sugiyama et al., 2011) and the qRT-PCR scores in the present study were used to analyze the correlation coefficients between carotenoid contents and relative expression levels of carotenoid metabolism genes in the hybrid AG population. The software JMP 6 (JMP release 6.0; SAS Institute Inc., Cary, NC, USA) was used to analyze the correlation coefficients.
eQTL analysisAs the basis of QTL analysis, the expressed sequence tag (EST)-based cleaved amplified polymorphic sequence (CAPS) markers (Omura et al., 2003) and single nucleotide polymorphism (SNP) markers (Fujii et al., 2013) were used to construct a linkage map. The CAPS and SNP markers included markers derived from the genomic regions of all carotenoid metabolism genes. In eQTL analysis, Windows QTL Cartographer ver. 2.5 (http://statgen.ncsu.edu/qtlcart/WQTLCart.htm) was used to perform QTL analysis on the linkage map of DNA markers constructed by JoinMap ver. 3.0 (Kyazma; Wageningen, the Netherlands). Linkage analysis was performed with the backcross (BC1) model set independently for the female parent, ‘Okitsu-46’ (A255) and male parent, ‘Nou-5’ (G-434). The ‘Okitsu-46’ map consisted of 345 markers linked into nine groups (A-01 to A-09) and covered 660 cM (66.1% of the 998 cM integration map for the AG population). The‘Nou-5’ map consisted of 254 markers linked in 12 groups (G-01 to G-09-2) and covered 642 cM (64.3%). The remaining uncovered regions in the individual maps were derived from the non-segregating marker loci for either parent. Some of the linkage groups were separated into two fragmented subgroups. The marker locations were also confirmed on the integration map of all markers segregated in the population calculated by the cross pollination (CP) model of JoinMap.
The qRT-PCR score for the expression level of each gene was subjected to QTL analysis using Interval mapping (IM) and composite interval mapping (CIM) by Windows QTL Cartographer ver. 2.5. Logarithm of odds (LOD) scores obtained by IM and CIM were checked by a permutation test with 1,000 permutations.
Association analysis of carotenoid metabolism genes and QTLsWhen efficient DNA markers were not available in the literature (Fujii et al., 2013; Omura et al., 2003) for association analysis of carotenoid metabolism genes and QTLs, new SNP markers or sequence tagged site (STS) markers were generated. SNPs were detected by comparing direct sequences of parental cultivars of carotenoid metabolism genes, and SNP genotyping was performed using the TaqMan SNP detection system (Applied Biosystems 7300 Real Time PCR System). STS sequences were also aligned to the scaffold of the genomic sequence of Clementine (http://www.citrusgenomidb.org/) to confirm the copy number and PCR products of STSs for the genes. The carotenoid content and the expression levels between genotypes of candidate genes were compared by statistical analysis (t-test). The data on carotenoid contents were obtained from a previous report (Sugiyama et al., 2011).
The expression levels of eight carotenoid metabolism genes were investigated in 51 individuals of the AG population using the TaqMan probe/primer system. Between the two parental lines, ‘Okitsu-46’ and ‘Nou-5’, the expression levels of almost all carotenoid metabolism genes were higher in ‘Okitsu-46’. The expression levels of PSY and NCED2 of ‘Nou-5’ were less than 20% and 10% of ‘Okitsu-46’, respectively, while those of the other genes were approximately 40% (Table 1; Fig. 2).
Relative expression levels of carotenoid metabolism genes in parental Citrus cultivars ‘Okitsu-46 (A255)’ and ‘Nou-5 (G434)’ and their progeny, the hybrid AG population.
Histogram of expression levels of each carotenoid metabolic gene in parental citrus cultivars ‘Okitsu-46 (A255)’ and ‘Nou-5 (G434)’ and their progeny, the hybrid AG population.
Among the AG population, wide variations were observed in expression levels of PDS, ZDS, LCYb, ZEP, NCED2, and NCED3, some of which exceeded the parental differences. For example, some individuals showed higher expression levels than ‘Okitsu-46’ and some individuals showed lower expression levels than ‘Nou-5’. These results indicated transgressive variation for these genes. The expression levels of PSY and HYb in the AG population were intermediate between the parents.
2. Correlation coefficients between carotenoid contents and relative expression levels of carotenoid metabolism genesThe contents of four abundant carotenoids (β-Cry, violaxanthin, phytoene, and ζ-carotene) in the juice sacs of the hybrid AG population and total carotenoids (sum of abundant carotenoids, lutein, β-carotene, and zeaxan-thin) were used to analyze the correlation coefficients (Table 2). The content of β-Cry, which is the most abundant carotenoid, significantly correlated with the content of β-carotene and zeaxanthin, and with the qRT-PCR scores of PSY, PDS, ZDS, LCYb, and ZEP. The total carotenoid content was also correlated with the expression levels of PSY, PDS, ZDS, and ZEP. Conversely, the contents of violaxanthin, phytoene, and ζ-carotene did not show significant correlations.
Carotenoid contents and coefficients of correlation (r) between carotenoid contents and relative expression levels of carotenoid metabolism genes in the hybrid AG populationz.
Based on the linkage maps of 93 individuals of the AG population, QTL mapping was performed on the expression levels of each carotenoid metabolism gene (Table 3). As shown in Table 3, 10 eQTL regions for PSY, PDS, ZDS, HYb, and ZEP were detected by the IM method. However, no significant eQTLs were detectable for LCYb, NCED2, and NCED3. The eQTL with the greatest effect on PSY was broadly detected from 0.0 to 60.8 cM on linkage group (LG) 4 of the ‘Nou-5’ map (G-04), with the peak at the Lp0206 marker locus. The LOD score determined by IM was 4.4 at the peak, explaining 32.4% of the variance. The peak of the second eQTL for PSY was located on LG8 of the ‘Okitsu-46’ map (A-08) at Gn0029, with a LOD of 3.5 at the peak, explaining 27.8% of the variance. Moreover, by CIM analysis, a new QTL was detected on the G-08.2 group.
Major eQTLs for the expression of carotenoid metabolism genes analyzed in the Citrus hybrid [‘Okitsu-46 (A255)’ × ‘Nou-5 (G434)’] AG population.
In IM analysis, the eQTL region with the highest LOD for PDS was detected in the same region as that for ZDS on A-08. At the 96.4 cM position of A-08, where the Tf0271 marker was placed nearest to the top of the peak eQTL, the LODs for PDS and ZDS were 7.1 and 3.6, respectively. At this locus, the percentages of explainable variance for these genes were 46.2% and 26.9%, respectively. These major eQTLs were also confirmed to be located at the same locus by CIM. CIM revealed new QTLs for PDS on A-01 and G-051 with significant LOD scores of 3.8 and 4.0, respectively.
The highest eQTLs for HYb and ZEP were located at G-03 (LOD 3.5, explainable variance 26.7%) and A-02 (LOD 3.4, explainable variance 25.9%), respectively. The major eQTLs for each gene were located independently on different linkage groups and did not overlap the others, except the eQTLs for PDS and ZDS.
4. Association of eQTLs with candidate gene lociEach locus for carotenoid metabolism-related genes, PSY, PDS, ZDS, LCYb, LCYe, HYb, ZEP, violaxan-thin de-epoxidase (Vdep), carotenoid cleavage enzyme (CCD), NCED2, and NCED3, were mapped on the linkage map for the AG population (Table 4). LCYb and LCYe were not mapped directly in this experiment because DNA markers have not been generated for segregation analysis of the AG population. However, their locations could be estimated with help from the genomic sequence of Clementine and STS markers linked to LCYb and LCYe. The putative positions of LCYb and LCYe were assigned on G-031 (from 50.2 to 77.0 cM) and G-09 (from 0 to 16.2 cM), respectively. Blast search was carried against the Clementine genome sequence for the 10 carotenoid metabolism genes that were mapped on this linkage map. PSY, PDS, LCYe, HYb, Vdep, NCED2, and NCED3 existed as single loci in the genome sequence, while ZDS and ZEP consisted of plural copies arranged in a tandem repeat. LCYb had duplicated copies at different loci. Based on the copy number information on these genes, the association between eQTL locus and its corresponding gene locus was evaluated for each carotenoid metabolism gene, except for LCYb.
Allele discriminating markers of Citrus carotenoid biosynthesis genes and their locations on the phase maps of each parental cultivar, ‘Okitsu-46 (A255)’ and ‘Nou-5 (G434)’, and on the integrated link-age map of the hybrid progeny AG population.
STS marker (Gn0068) derived from the ZEP gene sequence was located at 64.5 cM on A-02, where the peak eQTL for ZEP was detected with an LOD score of 3.4 (IM). A significant difference in the ZEP expression level was observed among individuals with different genotypes in the AG population (t-test, P < 0.01) (Fig. 3). The mean expression level was 0.4 for individuals with homozygous genotype (aa) for Gn0068 marker and 0.6 for heterozygous genotype (ab). ‘Okitsu-46’ was heterozygous for this marker and ‘Nou-5’ was homo-zygous. The haplotype unique for ‘Okitsu-46’ would have a positive effect, elevating ZEP expression. In order to confirm this positive effect precisely, a set of SNP markers was generated from the ZEP locus to discriminate the four haplotypes derived from both parents, because Gn0068 marker is a CAPS marker and cannot discriminate them. The result indicated that no eQTL for ZEP was detected in G-02, although the ZEP locus of SNP markers was located at 41.0 cM position of G-02 (Table 4). Therefore, only the haplotype with an eQTL linked to the Gn0068 marker in ‘Okitsu-46’ had a positive effect on ZEP expression.
eQTL map of linkage group 2 (A-02) from the hybrid AG progeny of the ‘Okitsu-46 (A255)’ × ‘Nou-5 (G434)’ cross for the expression level of zeaxanthin epoxidase (ZEP) in Citrus. (A) Linkage map of A-02 with LOD scores from eQTL analysis of ZEP expression. Arrow shows the region with the highest LOD score by IM analysis. (B) Segregation of ZEP expression in the AG population at the Gn0068 marker locus. Gn0068 is an STS marker based on the ZEP gene. The mean expression level was 0.4 for individuals of the population that were genotypically homozygous at marker Gn0068 and 0.6 for heterozygous individuals. White and black arrows indicate the expression levels of parental cultivars. The genotype of the parental cultivars showed that ‘Okitsu-46’ was heterozygous and ‘Nou-5’ was homozygous.
HybG-NC marker (Gn0077) derived from the HYb gene sequence was located at 86.3 cM on G-03, where the peak eQTL for HYb was detected with an LOD score of 3.6 (IM) (Fig. 4). The haplotype with an eQTL linked to the HybG-NC marker in ‘Nou-5’ had a negative effect on HYb expression.
eQTL map of linkage group 3 (G-03) from the hybrid AG progeny of the ‘Okitsu-46 (A255)’ × ‘Nou-5 (G434)’ cross for the expression level of β-ring hydroxylase (HYb) in Citrus. (A) Linkage map of G-03 with LOD scores from eQTL analysis of HYb expression. Arrow shows the region with the highest LOD score by IM analysis. (B) Segregation of HYb expression in the AG population at the Gn0077 marker locus. Gn0077 is an STS marker based on the HYb gene. The mean expression level was 0.6 for individuals of the population that were genotypically homozygous for HybG-NC and 0.4 for heterozygous individuals. White and black arrows indicate the expression levels of parental cultivars. The genotype of the parental cultivars showed that ‘Okitsu-46’ was heterozygous and ‘Nou-5’ was homozygous.
PsyG-CT marker (Gn0009) derived from the PSY gene sequence was located at 25.0 cM on G-04, where the peak eQTL for PSY was detected with an LOD score of 3.8 (IM) (Fig. 5). As shown in Figure 5, PSY was located in the central part of the eQTL region of PSY expression by IM method. CIM analysis also indicated a significant LOD for the eQTL on the PSY locus; however, the PSY locus was 4.1 cM from the peak of the eQTL. Thus, eQTL analysis provided us with three novel eQTLs that regulate the transcription of PSY, HYb, and ZEP.
eQTL map on linkage group 4 (G-04) from the hybrid AG progeny of the ‘Okitsu-46 (A255)’ × ‘Nou-5 (G434)’ cross for the expression level of phytoene synthase (PSY) in Citrus. (A) Linkage map of G-04 with LOD scores of eQTL analysis of PSY expression. Arrow shows the region with the highest LOD scores by IM analysis. (B) Segregation of PSY expression in the AG population at the Gn0009 marker. Gn0009 is a SNP marker based on the PSY gene. The mean expression level was 0.5 for individuals of the population that were genotypically homozygous for PsyG-CT and 0.3 for heterozygous individuals. White and black arrows indicate expression levels of parental cultivars. The genotype of the parental cultivars showed that ‘Okitsu-46’ was homozygous and ‘Nou-5’ was heterozygous.
The common eQTLs for PDS and ZDS were detected from 95.3 to 100 cM in region A-08; however, there were no detectable eQTLs for PDS and ZDS in the ‘Nou-5’ map (Fig. 6). The loci for PDS and ZDS were mapped on G-03 and G-091, respectively. Interestingly, the Tf0271 marker, which was generated from a transcription factor gene, was located at the peak of this eQTL. This indicated that the transcription factor would regulate PDS and ZDS transcription and only the haplotype with an eQTL linked to the Tf0271 marker in ‘Okitsu-46’ had a positive effect on PDS and ZDS expression. In this locus, weak eQTLs for ZEP (LOD 1.9) and LCYb (LOD 1.8) were overlapped.
eQTL map on linkage group 8 (A-08) from the hybrid AG progeny of the ‘Okitsu-46 (A255)’ × ‘Nou-5 (G434)’ cross for the expression level of phytoene desaturase (PDS). (A) Linkage map of A-08 with LOD scores of eQTL analysis of PDS expression. Arrow shows the region with the highest LOD scores by IM analysis. (B) Segregation of PDS expression in the AG population at the Tf0271 marker locus. The mean expression level was 0.4 for homozygous individuals of the population and 0.8 for heterozygous individuals. White and black arrows indicate expression levels of parental cultivars. The genotype of the parental cultivars showed that ‘Okitsu-46’ was heterozygous and ‘Nou-5’ was homozygous.
There has been much research in citrus on the physiological roles of carotenoid metabolism genes in influencing the carotenoid contents and composition using typical varieties. Expression quantity variance among carotenoid metabolism genes would influence the carotenoid contents and composition in citrus fruits (Kato et al., 2004). However, it is still unknown how the expressions of these genes are regulated to explain the variation in carotenoid content and composition. Compared with physiological research, genetics research using mapping populations has been quite limited in plants. In maize, two QTLs for carotenoid components in kernels were associated with loci of PSY and ZDS (Wong et al., 2004). Their results indicated that the carotenoid content was largely regulated by expression levels of the carotenoid metabolism genes PSY and ZDS, and suggested the contribution of cis-factors to carotenoid accumulation. In citrus, a limited study was reported on QTL mapping of carotenoid contents in fruits (Sugiyama et al., 2011). Therefore, the present study provides new insight into the regulatory factors controlling the transcription of carotenoid metabolism genes.
Analysis of eQTLs using segregating populations is a popular approach for identifying factors involved in cis- and trans-regulation of gene expression levels. In the present study, eQTL analysis was applied to explore the candidate factors regulating carotenoid metabolism gene expression in citrus. eQTL analysis showed that HYb and ZEP loci were located in the eQTL region responsible for their gene expression levels. The differences in the expressions of HYb and ZEP between genotypes were more significant than in the genotypes of markers located in the flanking regions. The result for ZEP was consistent with a previous report that the 5′-UTR diversity of ZEP haplotypes was associated with differences in the expression levels among alleles (Sugiyama et al., 2010). These results suggest that structural differences, such as cis-elements, might cause differences in the gene expression levels. Genomic sequences of Citrus species (Terol et al., 2008) will contribute to further studies that focus on investigating such differences among alleles.
Conversely, the eQTLs for PDS and ZDS were not mapped near the responsible genes, suggesting that their major transcription regulation might be controlled by trans-factors. In the case of the eQTLs for PDS and ZDS, a transcription factor-derived marker, Tf0271, was located just in the peak region on A-08. PDS and ZDS catalyze the conversion of phytoene to lycopene in the carotenoid biosynthetic pathway, and they have similar expression profiles in Citrus in response to ethylene treatment (Fujii et al., 2007), which suggests they might be regulated by the same transcription factor. Interestingly, the region around the Tf0271 marker also overlapped the weak eQTLs of ZEP and LCYb (LOD 1.9 and 1.8, respectively). The overlapping of eQTLs for different genes indicates the presence of a novel transcription factor gene regulating a wide spectrum of carotenoid metabolism genes. This would be a different type of transcription factor than RAP2.2 from Arabidopsis (Welsch et al., 2007). Thus, eQTL analysis was demonstrated as a powerful tool to explore the regulatory factors of carotenoid metabolism. Actually, eQTLs for the carotenoid metabolism genes had much higher LOD scores than previously published QTLs for carotenoid content using the same population (Sugiyama et al., 2011). For example, the significant QTLs for 9-cis-violaxanthin (cis-Vio) and total violaxanthins (Vios) were mapped on A-02. The weak peak of cis-Vio and Vios contents by IM analysis with weak LODs of 1.8 and 1.6, respectively, was mapped on an STS marker (Gn0068) based on the ZEP gene (A-02). It was expected that the association of the genotypes of the ZEP locus with the expression level of ZEP and the violaxanthin contents would help to select individuals with different levels of Zea/Vios ratio generated by different levels of ZEP expression.
In the present study, there were significant correlation coefficients between the content of β-Cry and relative expression levels of PSY, PDS, ZDS, LCYb, and ZEP genes. This result means that the β-Cry content might be influenced by plural gene expressions of PSY, PDS, ZDS, and LCYb, which are upstream enzymes for β-Cry synthesis, while in the comparison between the eQTLs of related gene expression and the QTL of carotenoid content, it was expected that some eQTLs would contribute to carotenoid contents, suggesting overlaps between the locations of the eQTLs in this experiment and QTLs of carotenoid contents in the previous research (Sugiyama et al., 2011). However, the eQTLs for carotenoid biosynthesis genes did not overlap with the QTLs for carotenoid contents. This lack of a clear relationship suggested that carotenoid contents are regulated not only by the expression levels of the carotenoid metabolism genes investigated in the present study, but also by other factors, such as post-transcriptional regulation of carotenoid metabolism genes and other genes, which were not investigated in the present study. In fact, QTL analysis in the previous report (Sugiyama et al., 2011) identified a major QTL for β-Cry content and total carotenoid content located at Gn0005, which was a marker derived from a non-carotenoid metabolism gene, CitPAP (DDBJ Acc. No. AB011797). In bell pepper (Capsicum annuum L.), PAP is thought to encode a carotenoid-associated protein that is involved in carotenoid accumulation (Pozueta-Romero et al., 1997). Therefore, we supposed that the difference in the carotenoid content and composition in the AG population was strongly influenced by various factors other than carotenoid metabolism genes.
The eQTL analysis in this experiment was only a preliminary step in the goal of increasing carotenoid contents in citrus breeding program; however, it was confirmed that this approach was effective in identifying candidate loci, which could generate functionally supported selectable markers for MAS, leading to improved metabolic design with favorable genotype combinations by allele mining in the future (Wolters et al., 2010).
