A consensus map for Cucurbita pepo
- First Online:
- Cite this article as:
- Zraidi, A., Stift, G., Pachner, M. et al. Mol Breeding (2007) 20: 375. doi:10.1007/s11032-007-9098-6
Using random amplified polymorphic DNA (RAPD), amplified fragment length polymorphism (AFLP), simple sequence repeats (SSR), and morphological traits, the first genetic maps for Cucurbita pepo (2n=2x=40) were constructed and compared. The two mapping populations consisted of 92 F2 individuals each. One map was developed from a cross between an oil-seed pumpkin breeding line and a zucchini accession, into which genes for resistance to Zucchini Yellow Mosaic Virus (ZYMV) from a related species, C. moschata, had been introgressed. The other map was developed from a cross between an oil-seed pumpkin and a crookneck variety. A total of 332 and 323 markers were mapped in the two populations. Markers were distributed in each map over 21 linkage groups and covered an average of 2,200 cM of the C. pepo genome. The two maps had 62 loci in common, which enabled identification of 14 homologous linkage groups. Polyacrylamide gel analyses allowed detection of a high number of markers suitable for mapping, 10% of which were co-dominant RAPD loci. In the Pumpkin-Zucchini population, bulked segregant analysis (BSA) identified seven markers less than 7 cM distant from the locus n, affecting lignification of the seed coat. One of these markers, linked to the recessive hull-less allele (AW11-420), was also found in the Pumpkin-Crookneck population, 4 cM from n. In the Pumpkin-Zucchini population, 24 RAPD markers, previously introduced into C. pepo from C. moschata, were mapped in two linkage groups (13 and 11 markers in LGpz1 and LGpz2, respectively), together with two sequence characterized amplified region (SCAR) markers linked to genes for resistance to ZYMV.
KeywordsDNA markersCucurbita moschataBSASeed coatZYMVSCAR-marker
The genus Cucurbita includes some of the oldest domesticated plant species. Three of them, Cucurbitapepo L., Cucurbita maxima Duchesne, and Cucurbita moschata Duchesne, are economically important crop species worldwide (Robinson and Decker-Walters 1997), with C. pepo exhibiting the widest variation, especially with respect to fruit characteristics. Duchesne (1786) and Naudin (1856) concluded that C. pepo L. could be the most polymorphic species in the plant kingdom. Botanical classification based on allozyme variation recognized three subspecies: C. pepo subsp. pepo, C. pepo subsp. ovifera (L.) D.S. Decker [syn. C. pepo subsp. texana (Scheele) Filov], and C. pepo subsp. fraterna (Bailey) Lira, Andres and Nee (Andres 1987; Decker 1988; H. S. Paris personal communication). Using mainly fruit shape, Paris (1986) classified edible-fruited C. pepo into eight cultivar-groups: Pumpkin, Zucchini, Cocozelle, Vegetable Marrow, Acorn, Crookneck, Scallop, and Straightneck. The first four of these cultivar-groups named above belong to subsp. pepo and the latter four to subsp. ovifera (Paris 2001). Polymorphisms of DNA markers, including RFLP, RAPD, ISSR, AFLP, and SRAPD, have been used to study genetic relationships within C. pepo (Torres-Ruiz and Hemleben 1991; Katzir et al. 2000; Paris et al. 2003; Ferriol et al. 2003) and the results support these classifications. Also, the results obtained, using AFLP, ISSR, and SSR were very highly correlated (P < 0.000001).
Genetic maps have been constructed from inter-specific crosses in the genus Cucurbita (Lee et al. 1995; Brown and Myers 2002), and we recently reported the initiation of a map for C. pepo (Zraidi and Lelley 2004). Compared to other economically important Cucurbitaceae, genetic mapping of Cucurbita (2n = 2× = 40) is in its infancy. Mapping efforts in melon, Cucumis melo L., resulted in a large number of DNA-based maps (e.g., Baudracco-Arnas and Pitrat 1996; Wang et al. 1997; Périn et al. 1998; Oliver et al. 2001; Danin-Poleg et al. 2002). Similarly, in cucumber, Cucumis sativus L., several maps have been developed (e.g., Serquen et al. 1997; Park et al. 2000; Fazio et al. 2002). Efforts to anchor, align, and merge maps with SSR, RFLP or STS markers are underway for cucumber (Bradeen et al. 2001), and melon (Gonzalo et al. 2005).
In Austria, C. pepo has been cultivated for centuries primarily for seed oil production (Teppner 2000). Pumpkin seed-oil is a delicious salad oil (Murkovic et al. 1996). It is highly valued for its excellent nutritional quality and medicinal value, especially for preventing and curing benign prostate hyperplasia in its early stage (Blumenthal et al. 1998; Kreuter 2000; Schmidlin and Kreuter 2003), although the chemical agent responsible for this property of the seeds has not been identified yet (Murkovic et al. 2004). A special feature of Austrian oil-pumpkin varieties is their completely hull-less (non-lignified) seed coat (Teppner 2000; Zraidi et al. 2003), which is conferred largely by a single recessive gene, n (naked seed) (Paris and Brown 2005).
The principal objective of the present study was to construct a genetic map for C. pepo, using RAPD, AFLP, SSR, and morphological markers. Co-linear common markers segregating in two F2 populations that mapped to the same linkage group were considered to be locus specific, and were used to align and merge the linkage maps that were constructed. Another objective was the identification of markers tightly linked to the seed coat locus n, believed to be controlled by a single major gene (Grebenščikov 1954; Prym von Becherer 1955; Zraidi et al. 2003). For this bulked segregant analysis (BSA) (Michelmore et al.1991) was used.
Materials and methods
Two C. pepo F2 populations, each derived from a single F1 fruit, were used for mapping. The first population, which consisted of 92 plants, is here designated as the Pumpkin–Zucchini population, following the horticultural classification of Paris (1986). This population was derived from an intra-subspecific (C. pepo subsp. pepo) cross between an oil-pumpkin breeding line (SZG1), developed by “Saatzucht Gleisdorf” (Gleisdorf, Austria), and a zucchini accession, “True French Resistant”, a near-isogenic line of the UK zucchini “True French” carrying three complementary dominant genes for resistance to Zucchini Yellow Mosaic Virus (ZYMV). These were introgressed from the Portuguese C. moschata “Menina” (Paris and Cohen 2000). Seeds of “True French Resistant” were obtained from H. S. Paris of the A. R. O. Newe Ya’ar Research Center, Ramat Yishay, Israel. The second population, which also consisted of 92 plants, is here designated as Pumpkin–Crookneck population. This population was derived from an inter-subspecific cross (C. pepo subsp. pepo × C. pepo subsp. ovifera) between the US oil-pumpkin variety “Lady Godiva” and the Italian crookneck variety “Bianco Friulano.”
Ten seeds from each of the 92 selfed F2 plants of the Pumpkin–Zucchini population were sown to produce a segregating F3 progeny. This was used to identify homozygous hull-less (n/n) and homozygous hulled (N/N) F2 genotypes to create contrasting DNA pools for BSA (Michelmore et al. 1991).
Three qualitative gene loci, M (Mottled leaves), n (naked seeds), and Bu (Bush habit) (Paris and Brown 2005), showed clear segregation in the F2 populations and were scored. The leaf-mottle trait was scored as the presence (M/–) or absence (m/m) of silver patches in the axils of the leaf veins, without regard to the extent or intensity of silver. The seed coat was scored as hulled (N/–) or hull-less (n/n), irrespective of an occasional residual lignification observed in the hull-less seeds (Teppner 2000). The growth habit was scored as plants being either bush (Bu/–) (length of the vine maximum 1.5 m) or vining (bu/bu) (length of vine 2–6 m) types.
DNA was isolated from fresh young leaves using the Wizard Genomic DNA Purification Kit (Promega Corp., Madison, WI, USA). DNA concentration was determined by the GenQuant RNA/DNA Calculator (Amersham Biosciences Europe, Freiburg, Germany) according to the manufacturer’s instructions.
The AFLP analysis (Vos et al. 1995) was done as described by Hartl et al. (1999) and Buerstmayr et al. (2002), using MseI and EcoRI enzymes. Selective amplification was performed using 28 EcoRI and MseI primer combinations with two or three randomly chosen selective nucleotides. RAPD and AFLP gels from the two populations were compared and bands, showing identical molecular weights in the two populations, were considered to be common markers.
A screen for potential SSR polymorphisms between the four parental genotypes used in the two crosses was done by J. E. Staub, University of Wisconsin, Madison, USA, using 102 Cucumis-SSR primers (Fazio et al. 2002). Eighteen C. pepo SSR primer pairs were developed by our laboratory, and were tested on the parents of the Pumpkin–Zucchini population.
Bulked segregant analysis (BSA)
Analysis of F3 progenies allowed the identification of N/N homozygotes from the F2 in the Pumpkin–Zucchini population. Two DNA pools were made, each containing a mixture of 12 DNA samples from homozygous dominant (hulled) and homozygous recessive (hull-less) plants for the n gene (Zraidi et al. 2003). A total of 1,068 RAPD markers were screened for polymorphism using the two DNA-bulks, and markers showing association with the n locus were tested on DNA from single plants of the two pools. Ultimately, markers associated with the trait, found through BSA, were used on the entire F2 population to calculate the degree of linkage.
Marker segregation was tested for fit to 3:1 (dominant loci) and 1:2:1 (SSR and co-dominant RAPD and AFLP markers) ratios at P = 0.05 probability. Linkage maps were constructed independently for each population using MAPMAKER/EXP 3.0 (Lander et al. 1987; Lincoln et al. 1992). Markers were associated with the group command with a likelihood of LOD ≥5 and with a maximum distance of 30. Marker-ordering within linkage groups was initially performed using the order command with a likelihood of LOD ≥3. Linkage groups containing more than six markers, which could not be ordered with this likelihood level, were ordered manually, using the compare command. The remaining markers were then integrated into these linkage groups, using the try command. Distances were calculated with the Kosambi function (Kosambi 1944). The order of the markers within linkage groups was double checked, using the ripple command. Linkage groups in the Pumpkin–Zucchini map are named LGpz, and those of the Pumpkin–Crookneck map LGpc.
Of the 495 tested RAPD primers, 227 (46%) and 296 (60%) showed polymorphism between the parents of the Pumpkin–Zucchini and the Pumpkin–Crookneck populations, respectively. Finally, 101 primers were selected for mapping because they showed polymorphism in both populations. These primers produced an average of 15 bands, each ranging in size from 150 to 1,900 bp. The average number of polymorphic loci per primer was 2.6 and 2.7 in the Pumpkin–Zucchini and Pumpkin–Crookneck populations, respectively.
AFLP primer combinations generated up to 60 bands, ranging in size from 50 to 700 bp. The 28 AFLP EcoRI/MseI primer combinations produced 118 polymorphic loci, with an average of 4.2 loci per primer combination, in the Pumpkin–Zucchini and 180 polymorphic loci, with an average of 6.4 loci per primer combination in the Pumpkin–Crookneck population.
Construction and characteristics of the maps
A total of 333 markers (247 RAPD, 82 AFLP, 3 SSR, and the n locus), 29 of which were co-dominant (21 RAPDs, 5 AFLPs, and 3 SSRs), were integrated in the Pumpkin–Zucchini map covering 2,140 cM of the genome and distributed over 21 major and 3 minor linkage groups (Fig. 2). The average size of the linkage groups was 90 cM, ranging from 10 cM in LGpz21 to 191 cM in LGpz4. The average number of markers per linkage group was 16 and the average distance between markers was 6.4 cM. Except linkage group LGpz21, all linkage groups contained at least one co-dominant marker, with a maximum of six in LGpz1, or one common marker, with a maximum of 11 in LGpz3 (Fig. 2).
The Pumpkin–Crookneck map contained 323 markers consisting of 196 RAPD markers, of which 20 were co-dominant, 125 AFLP with 12 co-dominant markers, and two gene loci, n on LGpc9 and Bu on LGpc12. This map covers 2,234 cM, distributed over 21 major and 3 minor linkage groups ranging in size from 38 cM in LGpc21 to 176 cM in LGpc3, with an average of 93 cM. Except LGpc20 which, however, contained 17 markers, all other linkage groups contained at least one co-dominant and/or one common marker (Fig. 3). The average number of markers was 15.5 per linkage group, and the distance between markers was 6.9 cM (Fig. 3). The distribution of RAPD and AFLP markers throughout the genome was markedly different (Figs. 2, 3). While the distribution of RAPD markers was relatively even in both maps, AFLP markers tended to cluster in most cases.
Chi-square tests (P = 0.05) revealed that the segregation of 20 (4.7%) loci in the Pumpkin–Zucchini and 26 (5.6%) loci in the Pumpkin–Crookneck map were distorted. Markers having distorted segregation were at first excluded from grouping and ordering analyses. They were mapped after creating linkage groups using only markers without segregation distortion. Ten distorted markers in the Pumpkin–Zucchini and 15 in the Pumpkin–Crookneck map were grouped in doublets or triplets and were not included in the final maps. Eleven distorted markers, distributed over five linkage groups, were included in the Pumpkin–Zucchini map, and ten, distributed over four linkage groups, in the Pumpkin–Crookneck map. Some of these markers were grouped together in the same linkage group in both maps, namely LGpz/LGpc5 and LGpz/LGpc8. The Pumpkin–Zucchini map contains 31 RAPD markers originating from the source of ZYMV resistance, C. moschata. None of these 31 markers showed segregation distortions.
Markers for ZYMV resistance and hull-less seed coat
About 24 of the 31 C. moschata specific markers were mapped onto two linkage groups in clusters of the Pumpkin–Zucchini map. LGpz1 contains 11 and LGpz2 contains 13 of these markers, two of which are closely linked to the ZYMV-resistance loci. These two markers have been converted to SCARs (unpublished results). The remaining seven markers were distributed over six linkage groups, namely two in LGpz7, and one in each of the following linkage groups: LGpz9, LGpz11, LGpz13, LGpz14b, and LGpz19 (see italic font in Fig. 2).
Alignment of the two maps
Gels of both populations, containing either RAPD or AFLP fragments, were compared to identify common markers based on their identical fragment size. Fifty-two RAPD (19.4%) and 40 AFLP loci (33.8%), were detected. Of the 92 common markers scored, 62 were used in mapping to recognize identical linkage groups (Fig. 4). We initially identified ten linkage groups that contained at least three common markers. When using relaxed grouping, i.e., LOD ≥3 and recombination frequency of higher than 35, the two linkage groups LGpz8a and LGpz8b were joined in the Pumpkin–Zucchini map (Fig. 2). In the same way the two linkage groups LGpc6a and LGpc6b were joined in the Pumpkin–Crookneck map (Fig. 3). Based on the presence of common markers, LGpz14 and LGpz16 in the Pumpkin–Zucchini map and LGpc5 and LGpc13 in the Pumpkin–Crookneck map were connected (Fig. 4). Based on the presence of at least two common markers in each of the two maps, 14 linkage groups were aligned with an average of four common markers and a maximum of ten (LG3 in Fig. 4). LG2 contained three common markers in the Pumpkin–Zucchini map, yet only one of these markers could be mapped in the Pumpkin–Crookneck population. Therefore the alignment of this linkage group is based on only one marker. Although the order of common markers was, in most cases, conserved, markers in LG3 and LG4 had to be rearranged (Fig. 4). The average distances between common markers in the two populations were, in most cases, similar.
By virtue of their marker coverage, density and consistency, the maps presented herein are suggested as reference maps for C. pepo. We have used mainly RAPD and AFLP markers as they are universal in their application and require no sequence information for their development. The combination of these markers allowed the construction of moderately saturated maps in both populations. Although they are, in general, dominant markers which limits their use in comparative mapping, they were successfully used here and in other studies for comparison of linkage groups and to align genetic maps (Plomion et al. 1995; Krutovskii et al. 1998; Laucou et al. 1998; Costa et al. 2000). Map alignment and map merging have generally been done on different populations, derived from the same two parental genotypes, or on two populations having one parental genotype in common. In our case, we used four different parental genotypes to create the two mapping populations, but two of the four parents are closely related, both being oil pumpkins. The number of common markers (92 in total, i.e., almost 30%) was relatively high. They allowed the establishment of a skeletal map for C.pepo, in which 62 mapped common loci were used to align 15 linkage groups. We consider these 15 linkage groups to represent different chromosomes, that is three quarters of the haploid number of chromosomes of C. pepo. The 62 common loci, represented by fragments of identical size in the two populations and therefore considered to be orthologous, were contributed in almost equal numbers by the four parents. This suggests an even distribution of alleles in the four parents. The order of these markers was conserved in most cases. Only in two cases rearrangements were observed, which could be due to the size of the mapping population, and/or to the algorithm used for arranging the markers (Périn et al. 2000). Other reasons could be genotyping errors, missing values and mapping distorted markers (Hackett and Broadfoot 2003).
The significantly higher level of polymorphism, as revealed by both RAPD and AFLP markers, between the parents of the pumpkin/crookneck population than between the parents of the pumpkin/zucchini population clearly reflects the taxonomical position of the three cultivar-groups within the species. Pumpkin and zucchini belong to the same subspecies, C. pepo subsp. pepo. They were most probably domesticated in Mexico. The crookneck variety, on the other hand, belongs to C. pepo subsp. ovifera and was domesticated most likely in eastern US. The genetic relationship among these cultivar groups is solidly established by isozyme and DNA-markers (Decker 1988, Sanjur et al. 2002, Paris et al. 2003).
The level of polymorphism of RAPD in our mapping populations was higher than that found by Brown and Myers (2002), even though they used an inter-specific (C. pepo × C. moschata) mapping population. The greater amount of polymorphism detected in our populations may partly be due to the use of polyacrylamide instead of agarose for fragment separation (Stift et al. 2003).
Results of mapping using AFLP markers have not been previously published for Cucurbita and therefore comparison of the polymorphism level is not possible. Nevertheless, this polymorphism level in our study was similar to that found in C. melo (Wang et al. 1997), which is probably in part due to the similar genome size of the two species (Bennett and Leitch 2004). The AFLP technique produced twice as many polymorphic bands as the RAPD marker system (average polymorphism of the two populations: 5.4 vs. 2.7). Similar results have been reported in other species (Powell et al. 1996; Milbourne et al. 1997; Costa et al. 2000; Haussmann et al. 2002; Zhong et al. 2004).
SSRs are versatile, highly informative, co-dominant markers for many eukaryotic genomes (Wang et al. 1994). They are locus specific, abundant and evenly distributed throughout the genome, and were recommended as standard markers to be used in the construction of highly saturated genetic maps (Beckmann and Soller 1990). Unfortunately, they showed poor transferability among cucurbit genera, as predicted by Katzir et al. (1996). The four parents of the two maps were genotyped for more than 100 melon SSR markers. Only a few of the markers generated bands, and none of them were polymorphic between the parents used to create the mapping populations. This was not the case when using Cucurbita SSR markers. Five of the available 18 C. pepo specific SSR markers showed polymorphism, and three of them could be mapped.
The genome coverage in the Pumpkin–Zucchini map (2,140 cM) and in the Pumpkin–Crookneck map (2,234 cM) was slightly higher than the genome coverage (1,954 cM) of the inter-specific map reported by Brown and Myers (2002). The average distance between markers they found (13 cM) was, however, twice as large as that of our study (∼6.6 cM), and is probably due to the low number of polymorphic markers they could use for mapping (148 vs. 330 markers). This is, however, understandable considering that Brown and Myers (2002) used a back-cross population for mapping, which effectively reduces the number of potential polymorphisms to half.
The difference in distribution among RAPD and AFLP markers described here are in good agreement with several reports in the literature, especially on clustering of AFLP markers in other species (Qi et al. 1998; Bert et al. 1999; Boivin et al. 1999; Haanstra et al. 1999; Vuylsteke et al. 1999; Young et al. 1999; Haussmann et al. 2002). This can be attributed to the use of EcoRI that cuts in AT-rich domains mainly located in the centromeric and telomeric regions. In contrast, however, De la Rosa et al. (2003) reported even distribution of both marker types, AFLP (using EcoRI) and RAPD in olive.
Polyacrylamide gel electrophoresis, due to its higher resolution, resulted in additional scorable RAPD fragments and identification of co-dominant RAPD markers (Fig. 1). In the two populations RAPD primers produced approximately 10% co-dominant loci, which was twice that found in the AFLP marker system. The number of RAPD-based co-dominant loci was considerably greater than that reported for other species (Grattapaglia and Sederoff 1994; Laucou et al. 1998; Krutovskii et al. 1998), likely due to the use of polyacrylamide instead of agarose gel for fragment separation. Two co-dominant RAPD markers, linked to ZYMV resistance, were isolated, sequenced, and successfully converted into SCAR markers. One allele pair differed in length by six nucleotides, the other pair differed in three base substitutions (SNPs). It is well known, that electrophoretic mobility of DNA in polyacrylamide is also affected by its base composition causing structural changes of the molecules. Thus, molecules of the same size can differ in mobility and can be distinguished using polyacrylamide gel electrophoresis.
In comparison to other crop species (Laucou et al. 1998; Haanstra et al. 1999; Bert et al. 1999; Cai et al. 2004), the number of distorted loci found in this study was relatively low (∼5%). It is worth mentioning the absence of any segregation distortion among the 31 markers originating from the source of resistance against ZYMV, C. moschata. Obviously, the integration of the genetic material from C. moschata into C. pepo, after backcrossing with the latter, was without any negative consequences. Segregation distortions have been reported in numerous genetic mapping studies, especially in populations derived from inter-specific crosses (e.g., Paterson et al. 1988; Keim et al. 1990; Heun et al. 1991; Echt et al. 1994). The mechanism of segregation distortion is not clearly understood, but it may have a number of different causes, such as chromosomal imbalance (aneuploidy), chromosomal rearrangements (translocations), competition among gametes, or genes affecting the viability of the embryo, as was suggested recently by Gonzalo et al. (2005). This can be the reason for the higher frequency (14%) of distorted markers in the inter-specific hybrid between C. pepo and C. moschata reported by Brown and Myers (2002).
Segregation distortion in C. pepo has been observed by Wilson and Payne (1994) and Paris (2000). In the first study, different pollen mixtures were applied in crosses between cultivated and free-living genotypes of C. pepo to study the competition of microgametophytes in fertilization. In the latter publication inter-subspecific crosses between two C. pepo genotypes were studied. Results of both experiments suggested an interaction between microgametophyte and gynoecium, which could, however, also be partly caused by the structure of gynoecium (Wilson and Payne 1994). Paris (2000) put forward the idea that interaction of microgametophyte with gynoecium and consequently gametophytic selection could partly be responsible for the continuous existence of groups of particular fruit shapes (cultivar-groups) throughout the centuries (Paris 2000). In our case, the average of 5% distorted markers in both populations was relatively low. Some of these markers were grouped in the same linkage groups in both maps, e.g., two in LGpz/LGpc5 and three in LGpz/LGpc8. The remaining were scattered individually in different linkage groups. No discernible difference between the intra- and inter-subspecific cross, with respect to segregation distortion, was observed.
Tagging the seed coat character
Of the seven markers, which were found linked to the n locus in the Pumpkin–Zucchini map, only two were in coupling phase with the hull-less allele. The marker AW11-420, which was 7 cM away from the hull-less allele in the Pumpkin–Zucchini map, was also found linked, at a distance of 3 cM, to the n/n (hull-less) allele in the Pumpkin–Crookneck map. This marker could play an important role in map-based cloning experiments, studying the genetics of the lignification of the seed-coat in C. pepo. The usefulness of this marker in selecting for hull-less genotypes has to be validated. Using BSA, Haley et al. (1994) identified RAPD markers in common bean, linked in both, coupling and repulsion, to a resistance allele against Bean Common Mosaic Virus (BCMV). In a case study they demonstrated that repulsion-phase linkages provided greater selection efficiency than coupling-phase linkage, even when the former had a greater linkage distance from the pest resistance allele. Repulsion-phase markers greatly improved selection efficiency for homozygous resistant genotypes. This we shall test too for the repulsion-phase markers in hull-less pumpkin.
Tagging ZYMV resistance
Three independent complementary genes confer resistance to ZYMV in C.pepo (Paris and Cohen 2000). About 24 of 31 bands introduced into C. pepo from the source of resistance, C. moschata, were mapped to LGpz1 and LGpz2. The remaining seven markers were distributed over six other linkage groups, and conceivably one of them is linked to the third putative resistance gene. Increasing map density may join some linkage groups into one, carrying the third independent resistance gene.
This research was financially supported by the Austrian Science Fund (FWF Project, No. P15773) and by the State of Lower Austria. Our thanks to J. E. Staub, USDA–ARS and University of Wisconsin, Madison, for testing the Cucumis-SSRs on the four parents used in our crosses. We are grateful to J. E. Staub and H. S. Paris for critically reading the manuscript.