Polyacetylene quantification in a carrot F2 family
In the present study a carrot F2 family derived from a cross of a cultivated carrot breeding line and the wild relative D. c. commutatus was used for quantification of major PAs FaOH and FaDOH. Since the roots of the original parental plants were not available for PA analysis, we measured the PA contents of each six siblings raised for the pilot experiment with the aim to identify parental lines differing in PA contents and to justify a bi-parental QTL approach. The average PA contents in the wild parent appeared to be considerably higher than the average PA contents in the cultivated breeding line. in the rare PA studies based on carrot wild relatives it was shown, that Daucus species and subspecies can contain much higher PA contents than cultivated carrot forms . Accessions of D. c. commutatus were among the (sub)species with the highest FaOH and FaDOH contents . PA data collected from the pilot experiment as well as those from the main experiment showed that - with the exception of three plants - all investigated plants possessed a higher FaDOH level compared with FaOH. According to Czepa and Hofmann  the most abundant PA in cultivated orange carrots is FaDOH, and the highest total PA levels were found in the periderm tissue [5, 9, 38]. In our study we focused for QTL analysis on total FaOH and FaDOH contents measured in samples representing a mixture of the different root tissues (periderm, phloem, xylem). Considering the large number of F2 individuals involved in this study, this approach appeared to be appropriate to detect QTLs relevant for carrot breeding. A further dissection of the root tissues for QTL analysis might be applied in future studies to reveal tissue-specific changes in PA biosynthesis.
A large phenotypic variability was observed for the contents of both PAs in the F2 population indicating that the selection of the crossing parents was appropriate for the bi-parental QTL approach. Carrot is a highly heterozygous species, and genetic uniformity even in advanced breeding lines appears to be not the rule, as it is demonstrated by the phenotypic variation of the PA contents in the six individuals of the 345B parental line. The observed frequency distributions in the F2 progeny indicate a polygenic inheritance of the PA production in carrot. This is not surprising considering the complex biochemical pathway from which the falcarinol-type PAs derive. Nevertheless, the high phenotypic variation in the F2 family, which contains plants with no measurable contents of PAs and on the other side plants with extremely high contents above 1000 µg/g FaOH or even 3000 µg/g FaDOH suggest, that major structural or regulatory genes are involved in the control of PA biosynthesis. Generally, the inheritance of PA concentrations and distribution patterns are poorly understood. The only study on carrot PAs based on a segregating population showed that PA contents are a heritable trait . In the carrot F2 family used in the 2-years experiment for QTL analyses of falcarinol-type PAs, a comparatively low impact of the environment on the accumulation of PAs was found, and the broad-sense heritability for, e.g. FaDOH was estimated with 0.88. Transgressive segregation was also observed in this study .
Linkage mapping and QTL analysis
Carrot belongs to the vegetable species for which already several partial or saturated linkage maps have been developed and published (for review, see ). For the current study publicly available SNP markers  were used for linkage mapping, allowing integration with other published genetic maps. Overall the genetic map is smaller compared to the first high-resolution SNP-based carrot map published by Cavagnaro et al. , but the marker order is similar, giving no indication of regions with recombination suppression. It has to be noted, that several regions of segregation distortions were identified, which could have an effect, though limited, on the general lower number of observed recombinations in this study. While segregation distortion does not affect QTL mapping especially in dense marker maps, it may lead to erroneous interpretations of the number of loci controlling a trait of interest .
In the present study we dissected the genetic basis of two key PAs, FaOH and FaDOH, by QTL analysis. Several major genomic regions seem to be involved in PA production in carrots. Overall, QTLs were localized on six out of the nine carrot chromosomes, which is a sign for the complexity of the genetic control of PA accumulation in carrot. Strong QTLs for FaOH and FaDOH (LOD > 10) were identified with overlapping 2-LOD confidence intervals on chromosomes chr_4 and chr_9 indicating a major involvement of these regions in PA biosynthesis. We propose that these QTL regions include major genes that control the primary or secondary PA metabolism in carrot. The QTL with the strongest statistical support was found for FaDOH in the upper region of chr_9. Interestingly, in the study of Le Clerc et al.  the biggest QTL for FaDOH was also located on chr_9, but according to their linkage map of chr_9 the QTL was on the opposite side. Due to the usage of proprietary markers used by these researchers (V. le Clerc, pers. comm.) it was not possible to verify the orientation of chr_9. However, it seems possible that both strong QTLs for FaDOH share a common biological origin. The QTL identified by us for both PAs in the lower region of chr_4 could be the same QTL identified for FaDOH by Le Clerc et al.  in this genomic region. A QTL for falcarindiol-3-acetate, a minor PA probably metabolized from FaDOH, was detected by latter researchers on top of chr_8, where we identified a QTL for the FaDOH/FaOH ratio. QTLs for PA ratio might be useful to identify loci controlling the interconversions among the PAs. It is presumed, that FaDOH is produced from the precursor substance FaOH [4, 6]. As the QTL effects of allele 'B' from the wild parent P_1870 are responsible for a lower PA ratio in 'BB' genotypes, it is probably that the QTLs for the PA ratio indicate impaired conversion of FaOH into FaDOH (i.e. accumulation of FaOH). The generally increasing QTL effects of the 'B' allele for all other QTLs associated with total amounts of single PAs indicate, that rather the primary metabolism of common precursors for all falcarinol-type PAs, as for instance crepenynic and dehydrocrepenynic acid, is controlled by genetic factors underlying these QTLs. Likely there was a selection during the domestication process against unfavourable "wild" alleles associated with high FaDOH contents to avoid a too high level of bitterness in cultivated carrots.
Overall, the results from our QTL analysis confirmed the utility of a carrot family derived from a wide cross of a cultivated carrot with a D. c. commutatus accession rich in PAs. Previous studies based on the genetics of carotenoid accumulation used a mapping population derived from a cross between a cultivated orange carrot and a wild white carrot collected in North America [44, 45]. Important advances in the understanding of the genetic control of carrot anthocyanin pigmentation have been made by using a purple wild carrot parent from Turkey [41, 42].
To dissect the genetic basis of a quantitative trait, such as the content of a natural product, two main methods have been used in the past, bi-parental QTL mapping and genome-wide association studies (GWAS). Conventional QTL mapping depends on a diverging genetic diversity of two parents and is very time-consuming since in several crop plants including carrot a F2 mapping population has to be developed. Moreover, QTL regions can be quite large and may include many potential candidate genes. Nevertheless, this method has been used in carrot research and breeding to elucidate the genetics of important secondary metabolites determining root quality such as β-carotene, anthocyanins, polyacetylenes, and volatile terpenes [32, 33, 41, 44]. GWAS can overcome the limitations of bi-parental QTL mapping and has great potential for detection of QTLs with high resolution in diverse sequenced genotypes. Combinatorial approaches are also very useful to compensate the limitations of each method. Both QTL analysis and GWAS have been used for the detection of DcTPS54, a sabinene synthase gene putatively involved in carrot flavour [32, 46] and discovered the Or gene controlling carotene accumulation [47, 48]. GWAS aimed at PA compounds in large sets of carrot cultivars, land races and wild relatives would allow the identification of additional QTLs, but might be especially helpful to narrow down the QTL intervals for both FaOH and FaDOH on chromosome chr_4 allowing the more precise association with candidate genes.
Candidate genes associated with QTLs
In this work, the combination of a PA metabolite quantification, bi-parental QTL analysis and the discovery of putative candidate genes from two different gene families involved in fatty acid metabolism have been used to get insights into the genetic control of PA biosynthesis. Few biochemical pathways have been studied in carrot by a combinatorial approach based on linkage mapping, QTL analysis and candidate gene identification. In contrast to previous investigations on carrot carotenoids, anthocyanins and terpenes (for review, see [49, 50, 51]), no studies have been, to date, reported on comparable candidate gene approaches for PA compounds. The reason is, that in higher plants little is known on the genetic control and the enzymes involved in the biosynthesis of these compounds. Until recently, no carrot genes involved in PA production have been described. Work in the past on parsley (Petroselinum crispum, Apiaceae) identified a divergent form of FAD2 that was upregulated in response to pathogen attack and, when expressed in soybean embryos, resulted in the production of crepenynic acid and dehydrocrepenynic acid [36, 52]. These results of the parsley studies are consistent with a pathogen-responsive, divergent FAD2-mediated pathway leading to the accumulation of acetylenic fatty acids needed for PA formation . The FAD2 inventory work of Busta et al.  resulted in the identification of 24 carrot FAD2 members, and it was shown after functional analysis of six genes in yeast and Arabidopsis thaliana, that these genes represented the major entry point into carrot PA biosynthesis. We performed a reannotation of the carrot genome sequence  and detected seven new FAD2 gene models, which increases the total number of carrot FAD2 family member to the large total number of 31. It is a striking feature, that carrots seem to contain the highest number of FAD2 family members reported so far in the plant kingdom. Knowledge about the FAD2 family including the divergent forms is based largely on research on A. thaliana and on oil seed crops. Despite the evidence for functional diversification of FAD2 in multiple plant families, the potential of such diversification in major crops including possible roles in biotic stress resistance has been largely unexplored .
Recently a biosynthetic gene cluster was discovered in tomato, that was required for FaDOH production. Interestingly, among the four highly co-expressed clustered genes there was a CER1 decarbonylase . To our knowledge, this was the first time, that a possible CER1 gene function was reported, that is different from the so far recognized function in the alkane biosynthetic pathway. The Arabidopsis CER1 protein is known as a decarbonylase that converts fatty acid metabolites into alkanes. Alkanes are components of waxes in the plant cuticle, a waterproof barrier serving to protect land plants from both biotic and abiotic stimuli . We searched for carrot orthologous genes and found six putative CER1 genes showing a close phylogenetic relationship with the previously characterized tomato CER1 and additionally six putative CER3 gene models. Few reports are available about CER gene families in plants. Characterization of the CER1 family genes in Brachypodium distachyon identified eight CER1 homologs , and in rice seven putative CER1 paralogs were identified .
We identified two genomic regions on two different carrot chromosomes around QTLs for FaOH and FaDOH with high LOD values of > 14 and > 21, respectively, that contain several FAD2 and CER1/3 genes. The six FAD2 genes clustering on chr_4 and the three CER1 genes on chr_9 are preliminary considered as candidate genes, based on their location within 2-LOD QTL confidence intervals. However, a causal relationship between CER1 genes and the production of FaOH and FaDOH remains to be biochemically explained. Concerning the six FAD2 genes associated with the major QTLs on chr_4 their distance to the QTL main peak is larger. However, in the upper region of this chromosome (outside of the 2-LOD interval) the gene DcFAD2-31 is located, which might have an influence on the calculated QTL peak positions. The six carrot FAD2 genes associated with the strong QTL on chr_4 reside close to one another within a small region (29.3 – 29.4 Mbp) and might have been originated from local tandem duplications, as data from microsyntenic and phylogenomic analyses suggest . This cluster also contains the two genes DcFAD2-7 and DcFAD2-8, which have been functionally characterized as Δ12-fatty acid acetylenases, and additionally the gene DcFAD2-19 which was described as a bifunctional enzyme capable of catalysing both Δ12 desaturation of oleate and Δ14 desaturation of crepenynate . Interestingly, in the plants of BRL, which showed high PA levels in comparison to other carrot cultivars, a co-expression seem to have occurred in the periderm tissue samples for five out of the six FAD2 candidates including DcFAD2-7, DcFAD2-8, and DcFAD2-19. Taken together, it is likely that the QTL region on chr_4 plays a major role in carrot PA production. Nevertheless, the situation, that 10 out of the 12 QTLs, with the exception of those on chr_3, are associated with FAD2 candidate genes, support the suggested importance of the FAD2 family on the PA biosynthesis in Apiaceae. Further functional and biochemical studies are needed to identify the most relevant FAD2s. Even beyond identifying FAD2 and CER candidate genes for QTLs identified in this study, more genes involved in PA biosynthesis might be identified elsewhere in the carrot genome. Since the PAs in this study are related in terms that they represent compounds produced at different steps in the same biosynthetic pathway, it is also possible that a single regulatory gene like a transcription factor affects the compounds produced in this pathway.
Implications for carrot breeding
Bitterness is considered as an undesirable taste of carrot roots, which can cause consumer rejection and is one of the main reasons for low preference scores in sensory evaluations of carrots . It is a very complex quality trait because numerous chemically different compounds may contribute to bitter taste. Potential bitter compounds in carrot are volatile mono- and sesquiterpenes, PAs, phenylpropanoids, and isocoumarins. Schmid et al.  recently listed 14 known bitter off-taste compounds including the major PAs FaOH and FaDOH. In the study of Le Clerc et al.  the total PA content was closely related to bitterness, and the highest quantities accumulated in the most bitter genotypes, whereas the lowest amounts were measured in the least bitter genotypes. There is some evidence from quantitative chemical analyses combined with sensory analysis, that FaDOH is highly correlated with bitterness, whereas FaOH is not [7, 13]. On the other hand, several in vitro studies have confirmed, that FaOH is one of most cytotoxic PAs in Apiaceae vegetables and, according the current state of knowledge, has a higher bioactivity than FaDOH . A better understanding of the genetics of the PA levels present in carrot roots might support breeding carrot cultivars with low bitterness but high health potential for the consumers. To reach this goal it will be necessary to reveal the genes that control the decisive steps in the complex PA biosynthesis pathway, i.e. the formation of FaDOH. Carrot chemotypes with acceptable amounts of bioactive PAs may contribute significantly to the known positive effects of carrots on human health.