Genome Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/15/13 For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
61
ARTICLE Identification of QTLs for capsaicinoids, fruit quality, and plant architecture-related traits in an interspecific Capsicum RIL population Shawn C. Yarnes, Hamid Ashrafi, Sebastian Reyes-Chin-Wo, Theresa A. Hill, Kevin M. Stoffel, and Allen Van Deynze
Abstract: Quantitative trait loci (QTL) analyses in pepper are common for horticultural, disease resistance, and fruit quality traits; although none of the studies to date have used sequence-based markers associated with genes. In this study we measured plant architectural, phenological, and fruit quality traits in a pepper mapping population consisting of 92 recombinant inbred lines derived from a cross between Capsicum frutescens acc. 2814-6 and C. annuum var. NuMexRNAKY. Phenotypic measurements were correlated to loci in a high-density EST-based genetic map. In total, 96 QTL were identified for 38 traits, including 12 QTL associated with capsaicinoid levels. Twenty-one loci showed correlation among seemingly unrelated phenotypic categories, highlighting tight linkage or shared genetics between previously unassociated traits in pepper. Key words: Capsicum frutescens, Capsicum annuum, pepper, phenology, capsaicinoids, fruit quality. Résumé : Les analyses pour identifier les locus de caractères quantitatifs (QTL) chez le poivron sont chose courante pour des caractères horticoles, de résistance aux maladies et de qualité. Cependant, aucune des études réalisées a` ce jour n'a fait appel a` des marqueurs d'ADN associés a` des gènes. Dans ce travail, les auteurs ont mesuré des caractères en lien avec l'architecture de la plante, sa phénologie et la qualité des fruits au sein d'une population de cartographie composée de 92 lignées recombinantes dérivées du croisement entre l'accession 2814-6 du Capsicum frutescens et le cultivar NuMexRNAKY du C. annuum. Les mesures phénotypiques ont été corrélées avec les locus d'une carte génétique a` haute densité fondée sur des marqueurs d'ADN provenant d'EST. Au total, 96 QTL ont été identifiés pour 38 caractères, incluant 12 QTL associés a` la teneur en capsaicinoïdes. Vingt-et-un locus présentaient une corrélation entre des phénotypes sans lien apparent, ce qui suggère soit une liaison étroite ou une assise génétique partagée entre caractères présumés sans lien chez le poivron. [Traduit par la Rédaction] Mots-clés : Capsicum frutescens, Capsicum annuum, poivron, phénologie, capsaicinoïdes, qualité des fruits.
Introduction The genus Capsicum is comprised of five domesticated species: C. annuum, C. baccatum, C. chinense, C. frutescens, and C. pubescens (Bosland 1996). This genus represents one of the fastest growing agricultural crops in the United States, primarily chili peppers and sweet bell peppers of the species C. annuum. Peppers are grown for use as food, spice, food coloring, ornament, and for their pain-killing and medicinal properties (Lucier and Jerardo 2006). Pungency is an important agronomic pepper trait, as the pungency of a pepper variety determines its appropriate use in food, spice, and medicine. The capsaicinoid biochemical pathway is unique to the genus Capsicum, and it is responsible for the compounds, particularly capsaicin, perceived as burning pain by stimulating the vanilliod receptors on primary afferent neurons on mammal taste papillae (Caterina et al. 1997; Liu and Simon 2000; Kido et al. 2003). Capsaicinoids are synthesized in the placental epidermis and accumulate in placental blisters (Fujiwake et al. 1980). Genetic regulation of the biosynthesis and accumulation of capsaicinoids has not been elucidated (Aza-Gonzalez et al. 2011). In C. annuum, a single recessive allele of the Pun1 locus, generated by a large deletion, is responsible for the absence of pungency in peppers homozygous for this allele (Stewart et al. 2005). However, the role of the Pun1 acyltransferase in relation to the capsaicinoid biochemical pathway remains unknown.
One approach to identify genetic elements involved in pepper pungency is to identify quantitative trait loci (QTL) associated with pungency phenotypes. Although influenced by environmental factors, capsaicinoid content is a heritable quantitative trait (Zewdie and Bosland 2000). Seven QTL associated with capsaicinoid content have previously been identified on pepper chromosomes 3, 4, and 7 using F2 mapping populations and relatively low DNA marker densities (Blum et al. 2003; Ben Chaim et al. 2006). In this study we identify QTL correlating to fruit capsaicinoid content and a variety of other agronomic and fruit quality traits using a high-density genetic map derived from a RIL population generated from an interspecific cross between C. frutescens and C. annuum. In this cross, C. annuum var. NuMexRNAKY, is mildly pungent, and like other C. annuum varieties, is a herbaceous, annual, large-fruited, domesticated pepper cultivar. In contrast, C. frutescens acc. 2814-6 is a short-lived, woody perennial with small pungent fruits (Fig. 1). Although C. frutescens is cultivated for tabasco sauce, it is similar to semidomesticated weedy varieties (Pickersgill 1971).
Materials and methods Field trial design The mapping population used in this study consisted of 105 recombinant inbred lines (RILs) derived by single-seed descent from a cross between C. frutescens var. 2814-6 and C. annuum var. NuMexRNAKY from Cornell University (same cross as Wu et al. 2009). Seedlings were established in a greenhouse in early April
Received 31 May 2012. Accepted 17 September 2012. Corresponding Editor: P. Gulick. S.C. Yarnes, H. Ashrafi, S. Reyes-Chin-Wo, T.A. Hill, K.M. Stoffel, and A. Van Deynze. Seed Biotechnology Center, University of California, Davis, CA 95616, USA. Corresponding author: Allen Van Deynze (e-mail:
[email protected]).
Genome 56: 61–74 (2013) dx.doi.org/10.1139/gen-2012-0083
Published at www.nrcresearchpress.com/gen on 6 December 2012.
Genome Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/15/13 For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
62
Genome Vol. 56, 2013
Fig. 1. Fruit shape diversity in parental and recombinant inbred lines.
C. annum var. X C. frutescens NuMexRNAKY
Selection of RILs 2010 and transplanted to two field locations in Davis, California, on 4 June 2010. Location 1 was a Reif, very sandy, loam series soil and location 2 was a Yolo, silt, loam series soil. Each location consisted of two randomized replications of plots containing six plants each of 105 RILs along with the parents. Throughout the growing season the developing plants were measured for a variety of phenotypic traits for use in QTL analysis. A summary of traits measured are given in Table 1 and described further below. Capsaicinoid measurements Fruits were harvested at breaker stage. Six to 20 fruits were harvested from at least three plants per plot, dried for 1 week at 85 °C, and ground to a fine powder. Acetonitrile (2.5 mL of 100%) was mixed with each ground tissue sample (0.5 g per plot) and placed in a sonicating water bath at 65°C for 1 h. Samples were centrifuged (3500 rpm for 5 min) and the supernatant filtered through 45 m syringe filters. A 20 L aliquot of the filtered extract was injected via an autosampler for HPLC separation (Agilent Technologies 1200 series). The mobile phase for HPLC was 47% filtered acetonitrile at a flow rate of 1 mL/min. The solid phase consisted of a precolumn (Agilent Eclipse XDB-C18, 5 m, 4.6 mm × 12.5 mm) and a reverse phase HPLC column (Agilent Eclipse XDB-C18, 5 m, 4.6 mm × 150 mm). Two technical replicates were run for each sample. Samples with inconsistent measurements were run a third time and the outlying value dropped. Parts per million (ppm) were calculated from the mean of two technical replicates. Capsaicinoids were detected by absorption at 280 nm. Retention times averaged 7.6 min for nordihydrocapsaicin (NDHC), 8.3 min for capsaicin, and 12.2 min for dihydrocapsaicin (DHC). An external standard of natural capsaicinoids containing 55% capsaicin (Sigma, 360376) was prepared in 100% acetonitrile for a linear standard curve of combined capsaicinoids ranging from 15 to 4000 ppm. The integrated peak areas of combined capsaicinoids 1Supplementary
in unknown samples were converted to ppm using the equation of the line describing the combined capsaicinoids in the standard (R2 = 0.99). The relative contribution of capsaicin in the standard was calculated as 55% ppm of total capsaicinoids (ppm), and a linear equation was derived describing the relative contribution of capsaicin to the standard. Capsaicin peaks in unknown samples were converted to ppm using the equation of the line describing capsaicin in the standard. The relative contribution of NDHC and DHC in the standard was calculated as 45% ppm of total capsaicinoids, and a linear equation was derived describing the combined contribution of these two compounds. Assuming NDHC and DHC have similar molar absorption coefficients, their peaks in unknown samples were converted to ppm using the equation describing the contribution of both compounds. Leaf morphology measurements In mid-July 2010, leaves with petioles attached were collected from the lowest node of each plant in every plot where at least three plants survived (3–6 leaves/plot). Petiole lengths were recorded and the petioles cut from the leaves. Leaves without petioles were imaged using a scanner (Canon K13018) and the images analyzed with the Tomato Analyzer 3.0 (Brewer et al. 2006; Gonzalo et al. 2009). The Tomato Analyzer's basic measurements (perimeter, area, width mid-height, maximum width, height midwidth, and maximum height) were recorded for each leaf. Fruit morphology measurements Fruits (6–20) were collected at the breaker stage of development from all plots from at least three plants. Latitudinal and longitudinal sections were imaged and analyzed as previously described for leaves. A subset of the measurements taken by the Tomato Analyzer 3.0 was used in subsequent QTL analysis (Table 1). Two traits, pericarp area and thickness, derived from latitudinal sectioned fruit were modified algebraically from the Tomato Analyzer 3.0 output to yield units of square centimetre (cm2) and centimetre (cm), respectively. Molecular map DNA was extracted from 119 RILs and parents as described by Stoffel et al. (2012), of which a subset of 105 RILs were planted in the field for this study. The DNA of each sample was hybridized and processed in duplicate with the Pepper GeneChip consisting of 30 815 unigenes. The hybridization intensities were background corrected and normalized using the Affymetrix package of Bioconductor and R. The data were then analyzed using a modified algorithm from West et al. (2006) by custom Perl scripts. The haplotypes were determined for each RIL individual in the population using the methodology of M.J. Truco and R.W. Michelmore (unpublished). Genetic loci (unigenes) were assigned to linkage groups using MadMapper, ordered using Record (van Os et al. 2005), and genetic distances (cM) were calculated with JoinMap 4.0. (van Ooijen 2006). In this population, 16 188 unigenes were mapped in 2886 genetic bins. Linkage groups were associated and orientated with chromosomes P1–P12 based on common COSII markers (Wu et al. 2009). The average distance between markers in our map was 0.46 cM (unpublished), a subset of those markers averaging 5 cM apart, as well as markers at the peak of QTL, are reported in Supplemental Table S1.1 The marker sequences represent unigenes on the Pepper GeneChip. The exact nucleotide changes that identify the SPP markers are unknown. Statistical and QTL analyses The statistical software JMP 8.0.2 (SAS, Cary, N.C.) was used for all statistical analyses, except QTL identification. ANOVA were used to test for differences among treatments (RILs) and locations.
data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/dx.doi.org/10.1139/gen-2012-0083.
Published by NRC Research Press
Genome Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/15/13 For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Yarnes et al.
63
Table 1. Summary of phenotypic traits and their measurements. Trait category
Trait
Measurement description
Capsaicinoids
Total capsaicinoids Nordihydrocapsaicin Capsaicin Dihydrocapsaicin
Determined by HPLC
Leaf traits
Petiole length Perimeter Area Width mid-height Maximum width Height mid-width Maximum height
Directly measured Measured from image using the Tomato Analyzer 3.0
Phenology and floral traits
Days to flower Days to breaker Stigma exertion
Days after sowing until three plants per plot had open flowers Days after sowing until three plants had red ripe fruit Each plot assessed for stigma exertion at anthesis relative to petals: inserted, same, exerted Each plot assessed for flower position at anthesis relative to the stem: pendent, intermediate, erect
Penduncle position Whole plant morphology
Early habit Late habit Vigor Plant height Branching density Leaf density
Fruit traits Latitudinal
Longitudinal
Perimeter Area Width mid-height Maximum width Height mid-width Maximum height Lobedness degree Pericarp area Pericarp thickness Perimeter Area Width mid-height Maximum width Height mid-width Maximum height Curved height Width widest
Each plot assessed for plant form on 19 July 2010: prostrate, compact, erect Each plot assessed for plant form on 13 September 2010: prostrate, compact, erect Each plot assessed for plant vigor on 19 July 2010: poor, fair, intermediate, good, excellent Heights of first three plants per plot measured on 16 September 2010 Each plot assessed for plant branching on 13 September 2010: sparse, intermediate, dense Each plot for leaf density on 13 September 2010: sparse, intermediate, dense Measured from image using the Tomato Analyzer 3.0
Least squared means were calculated and used to identify RILs exhibiting transgressive segregation: phenotypic values significantly higher than the highest parental value or significantly lower than the lowest parental value. Correlations between phenotypes were determined by pairwise correlations of the combined means of each RIL (␣ = 0.05). QTL analyses were performed using Windows QTL Cartographer 2.5 composite interval map (CIM) function (Wang et al. 2011). Each trait was analyzed using three means: location 1 specific means, location 2 specific means, and the combined location means. Mean-specific thresholds were determined by permuting 1000 times. The likelihood ratio (LR) values reported for the permutations were converted to log of the odds ratio (LOD) scores (1 LOD = LR × 0.217). Only QTL with LOD scores equaling or exceeding the mean-specific threshold at ␣ = 0.03 were considered significant. QTL analyses were performed in six sets
corresponding to six trait categories (Table 1), and the data from the six sets were combined by consolidating overlapping QTL (see Fig. 2 for a visual example of QTL consolidation). QTL were consolidated when the width of two LOD intervals overlapped, and defined by the position (cM) furthest right and left of the second LOD intervals.
Results Field trial and QTL analysis Of the 105 RILs planted, only 92 performed well enough in the field for phenotypic evaluation. Plant growth in the Reiff series soil at location 1 was less vigorous than in the Yolo series soil at location 2. Measurements associated with plant size, leaf size, and fruit sizes were significantly higher at location 2, with the exceptions of longitudinal fruit area and height mid-width, which were Published by NRC Research Press
Genome Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/15/13 For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
64
Genome Vol. 56, 2013
Fig. 2. Capsaicinoid QTL on chromosome 4. Grey bars indicate QTL at ␣ ≥ 0.03. QTL 4.14 and 4.15 illustrate how QTL with overlapping LOD intervals are consolidated within trait category.
significantly smaller at location 2 (Table 2). Plants also displayed higher leaf densities at location 2. None of the capsaicinoid or latitudinal fruit traits means differed between locations (Table 2). Before QTL were consolidated, there were 262 mean–loci correlations identified at ␣ = 0.03 (Table S2), of which 40% were based on location 1 specific means, 31% were based on location 2 specific means, and 29% were from the combined location means. Location 1 contributed to the identification of more mean–loci correlations, mainly among latitudinal fruit traits (Fig. S1). The 262 mean–loci correlations identified at ␣ = 0.03 consolidated, based on two LOD interval overlaps, to a total of 96 QTL (Fig. 3; Table 3). Of the 96 consolidated QTL, 58% were based on the correlation of a single phenotypic trait and 42% were based on multiple phenotypic traits. The width (two LOD interval extremes) of consolidated QTL ranged from 0.4 to 15.8 cM, with a mean width of 3.4 cM. Twenty-one of the 96 consolidated QTL correlated to multiple trait categories. Capsaicinoids All four capsaicinoid traits displayed positive transgressive segregation (Table 4), with some RILs having significantly higher capsaicinoid levels than the most pungent parent, C. frutescens. For example, the most pungent RIL had mean total capsaicin of 6282 ppm, more than double the concentration found in the most pungent parent, C. frutescens (2630 ppm, Fig. 4A; Table 4). Negative transgressive segregation could not be determined for the capsaicinoids because the least pungent parent, C. annuum, displayed capsaicinoid levels at the low end of our detection limit (⬃15 ppm). Pairwise correlations between and among traits (Table S3) reveal some significant correlations for capsaicinoid traits. The levels of capsaicin, DHC, and NDHC were positively correlated to one another, with RILs displaying high levels of one capsaicinoid tending to have high levels of the other two. Capsaicinoid traits were negatively correlated to latitudinal (cross-section) fruit size traits
and leaf size traits. However, longitudinal (lengthwise) fruit size traits were not correlated to pungency traits. Twelve QTL were found in association with capsaicinoid traits on six chromosomes (QTL 3.1, 4.2, 4.13, 4.14, 4.15, 4.16, 5.4, 6.8, 7.3, 10.2, 10.3, and 11.8) (Fig. 3; Table 3). Leaf traits All leaf traits displayed transgressive segregation. All traits, at both locations, had some RILs displaying means significantly greater than, C. annuum, the parent with the largest leaves and petioles (Table 4). Petiole length and leaf perimeter also displayed negative transgressive segregation, with some RILs being significantly smaller than C. frutescens, the smallest parent (Table 4). Pairwise correlations between and among traits (Table S3) reveal some significant correlations for leaf traits. There was a significant correlation between large leaves and short petioles. Plant height and vigor were correlated with small leaves and long petioles. Plants with large leaves and long petioles tended to have large fruit and wide leaves, and they were found significantly more often on plants with low branching density. Twenty-three QTL were identified in association with leaf traits (Fig. 3; Table 3). Phenology and floral traits Phenology and floral traits did not display transgressive segregation (Table 4). No RILs had significantly earlier flowering or breaker date than C. annuum. In fact, C. annuum had significantly earlier date to breaker than all RILs (Fig. 3). Three QTL were found associated with days to breaker (QTL 2.6, 2.8, and 8.1). Collectively these three QTL (2 LOD intervals) span 35 SPP markers. Seventyfive percent of the RIL population had haplotypes identical to C. annuum over these 35 SPP markers. Pairwise correlations between traits (Table S3) did not reveal surprising results about phenology and floral traits. Plants that flowered early tended to have earlier maturing fruits, and small plants tend to be late in fruit maturaPublished by NRC Research Press
Genome Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/15/13 For personal use only. This Just-IN manuscript is the accepted manuscript prior to copy editing and page composition. It may differ from the final official version of record.
Yarnes et al.
65
Table 2. Summary of location differences among phenotypic traits (␣ = 0.05). Location 1
Location 2
ANOVA between locations
Trait category
Trait
Capsaicinoids
Total capsaicinoids (ppm) Nordihydrocapsaicin (ppm) Capsaicin (ppm) Dihydrocapsaicin (ppm)
1072.1 156.3 444.3 552.9
986.0 157.3 408.7 520.8
1046.6 159.5 435.9 529.0
989.7 168.5 432.0 495.1
1,151 1,151 1,151 1,151
0.8643 0.8989 0.8964 0.7504
Leaf
Petiole length (cm) Perimeter (cm) Area (cm2) Width mid-height (cm) Maximum width (cm) Height mid-width (cm) Maximum height (cm)
2.0 5.7 2.0 1.5 1.5 2.3 2.5
1.3 2.3 2.0 1.8 1.8 1.7 1.7
2.9 6.9 2.5 1.6 1.7 2.2 2.9
1.5 1.7 1.8 1.7 1.6 1.5 1.5
1,2395 1,1937 1,1937 1,1937 1,1937 1,1937 1,1937
258.4432 154.3518 38.6497 4.2503 3.8468 22.8989 37.8539
Phenology and floral
Stigma exertion (0–3) Penduncle position (0–3) Time to flower (days) Time to breaker (days)
2.8 2.2 123.8 206.0
0.4 0.7 13.5 18.2
2.8 2.2 124.8 207.8
0.5 0.7 9.9 17.3
1372 1378 1380 1310
0.0092 0.1842 0.7518 0.5421
0.9236 0.6680 0.3865 0.4621
Whole plant morphology
Early habit (0–3) Late habit (0–3) Vigor (0–3) Plant height (cm) Branching density (0–3) Leaf density (0–3)
2.3 2.2 2.9 41.0 2.0 2.0
0.7 0.6 1.0 19.2 0.7 0.6
2.0 2.2 3.2 60.5 1.9 2.2
0.7 0.6 0.8 18.3 0.7 0.6
1390 1393 1,416 1,1202 1391 1327
22.8049 0.7299 15.0157 326.6254 1.3720 10.5816
