Phenotypic evaluation.

Seeds of 143 F₉ Gy7 × H-19 recombinant inbred lines and two parental lines were planted on 5 June 2009 in an open-field [Plainfield loamy sand (Typic Udipasamment) soil] at the University of Wisconsin Agricultural Experiment Station in Hancock, WI. Ten non-replicated plants of each RIL were phenotyped for two morphological traits [little leaf (ll) versus standard-sized (LL) leaf; determinate (de) versus indeterminate (De)] 3 weeks after planting. Plants possessing main stems that ended in a flower cluster were classified as determinate. Likewise, plants were classified as little or standard leaf type when compared with parental lines by visual examination. Plant evaluations for leaf size and growth habit were conducted three times during the growing season (21, 42, and 70 d after sowing).

Molecular marker analysis.

Unexpanded young leaves of each RIL were collected from each of the 10 plants and pooled, and genomic DNA was extracted using the CTAB method (Murray and Thompson, 1980). A subset of 46 RILs was used for linkage analysis, in which 26 and 20 possessed standard-sized and little leaves and 21 and 25 were determinate and indeterminate, respectively.

Each polymerase chain reaction (PCR) contained 25 ng template DNA, 0.5 μM each of two primers, and 1 × PCR master mix (Fermentas, Glen Burnie, MD) in a total volume of 10.0 μL. A “touch-down” PCR program was used for all primer sets (Weng et al., 2005).

Two sets of molecular markers were used in this study. The first set included 821 SSR markers developed from a whole genome sequence of the Northern China fresh market-type cucumber inbred line 9930 (Huang et al., 2009). The second set included 15 SSR, 26 SCAR, and one singe nucleotide polymorphism markers previously mapped in the Gy7 × H-19 RIL population (Fazio et al., 2003a, 2003b; Robbins and Staub, 2009). Each primer pair amplified a single band in each sample template DNA, in which variation in PCR product size allowed for banding morphotypes to be resolved in 3% agarose gels stained with ethidium bromide.

Data analysis and map construction.

An RIL-based linkage map was constructed using the computer program JOINMAP 3.0 based on SSR and morphological marker segregation data that were analyzed for fit to the expected 1:1 ratios by χ² analyses. Marker data were assigned to LGs using a minimum logarithm of odds (LOD) score of 3.0, and the Kosambi map function (Kosambi, 1944) was used to calculate the genetic distance between markers.

Markers were assigned to LG according to shared molecular markers with the high-resolution cucumber genetic map of Ren et al. (2009). For synteny analysis between linkage maps, the order of 94 common markers on the map of Ren et al. (2009) and the Gy7 × H-19 RIL-based map was compared.

Segregation of leaf size and growth habit among recombinant inbred lines.

Of the 143 lines examined, 42 were homozygous little leaf (ll), 100 homozygous standard-sized leaf (LL), and one was heterozygous (Ll). Growth habit analysis of these RILs resulted in their classification as 57 homozygous determinate (dede) and 86 homozygous indeterminate (DeDe), respectively. The expression of each of these traits is controlled by single recessive genes in this mapping population (Bradeen et al., 2001; Fazio et al., 2003a; Serquen et al., 1997). However, χ² analysis revealed that although segregation at the little leaf locus deviated significantly from the expected 1:1 ratio (P < 0.001), segregation at the determinate locus marginally fit this ratio (P = 0.015).

The little leaf (ll) locus has pleiotropic associations with multiple lateral branching (Fazio et al., 2003a) and is located in a chromosome region that manifests relatively strong segregation distortion that supports the dominant action of Gy7 alleles (see subsequently). The action of the little leaf allele does not affect gynoecy conditioned by the F locus, a trait that also maps to the same LG as ll. It is sometimes difficult to discriminate between leaves that have an intermediate (50 to 70 cm²) and standard (≈80 cm²) leaf size (Staub et al., 2002). It is likely that the greater-than-expected standard leaf-sized plants identified here was the result of either segregation distortion or misclassification of leaf-type phenotypes.

Slight distortions in segregation, however, usually do not affect trait-marker association during linkage analysis (Hackett and Broadfoot, 2003). Nevertheless, pleiotrophy and epistatic interactions have been observed in cucumber and, thus, care must be taken during phenotypic classification for segregation analyses (Robbins et al., 2008).

Polymorphism screening and linkage analysis.

Of the 821 cucumber genomic SSR primer sets tested, 17.0% (143 of 821) were polymorphic between Gy7 and H-19. This degree of polymorphism was higher than that reported in previous DNA-based studies (3% to 12% polymorphism) examining diversity among elite (C. sativus var. sativus) and exotic (C. sativus var. hardwickii) cucumbers (Dijkhuizen et al., 1996; Staub and Bacher, 1997). This disparity can be explained by the fact that when compared with other elite commercial cucumber germplasm (lines and hybrids), Gy7 and H-19 are genetically distinct (Staub and Bacher, 1997). Moreover, the types of markers used here differ from those used by Dijkhuizen et al. (1996; used RFLP and isoenzyme markers) and Horejsi and Staub (1999; used RAPD markers). Moreover, the microsatellite markers used here were developed from a whole genome sequence of cucumber inbred line 9930 through pre-selection of long repeat motifs (Ren et al., 2009), which are generally more polymorphic than those having shorter motifs (Huang et al., 2009).

Forty-two markers (26 SCARs, 15 SSRs, and one SNP; Fazio et al., 2003a; Robbins and Staub, 2009) previously mapped in cucumber were also used for RIL analysis. Each primer pair amplified a single band, and, thus, 184 marker or gene loci were included in linkage mapping. One hundred seventy-six markers were mapped to seven LG using a LOD threshold of 3.0. Of these loci, 94 were shared with the cucumber genetic map constructed by Ren et al. (2009). Based on common-marker alignments between these maps, markers could be mapped to the seven LG (2n = 2x = 14), which corresponded to cucumber Chromosomes 1 to 7, respectively, as defined in Ren et al. (2009) and Table 1 (distribution information) and depicted in Figure 1.

Table 1.

Distribution of genetic markers among seven cucumber (Cucumis sativus L.) chromosomes mapped with a recombinant inbred line population derived from a Gy7 × H-19 cross.^z

View Table

View Full Size

A cucumber (Cucumis sativus L.) linkage map derived from a Gy7 × H-19 recombinant inbred line (RIL). Numbers on top of the map are linkage groups (LG) (1 through 7), which are also corresponding to chromosomes numbers as defined in Ren et al. (2009). Numbers in parentheses after LG numbers correspond to linkage groups on the map by Fazio et al. (2003a). Cumulative map distance (cM) is shown to the left of each linkage group and marker designation is on the right. SCAR markers are designated by –S or –M, L19-SNP-H19 is a single nucleotide polymorphism (SNP) marker, and the remaining are simple sequence repeat (SSR) markers. Two morphological genes are determinate (de) and little leaf (ll). Markers with asterisks showed segregation distortion.

Citation: HortScience horts 45, 6; 10.21273/HORTSCI.45.6.882

• Download Figure
• Download figure as PowerPoint slide

Of the 176 molecular marker loci examined by χ² test analysis, significant derivation from the expected 1:1 segregation (i.e., segregation distortion) was detected for 11 marker loci. Of these 11 markers, eight were concentrated in two chromosomal regions [segregation distortion region (SDR)], which included four (XSSR01609, XI1-S, XN6-A-S, and XNR66) in LG 3 (from 11.9 to 18.9 cM) and four (XSSR03940, XSSR16451, XSSR17624, and XSSR13884) in the middle of LG 6 adjacent to the ll locus (from 35.5 to 40.4 cM; Fig. 1). Three other loci showing segregation distortion (XSSR17023, XSSR14934, and XSSR06653) were scattered in different locations of LG 6 (Fig. 1). Although the SDR presented in LG 3 exhibited excess numbers of marker genotypes associated with H-19 (paternal parent), Gy7 maternal parent alleles were more frequently associated with ll and other four adjacent SSR loci. The SDR region adjacent the ll locus was also identified by Ren et al. (2009) using a Gy14 × PI 183967 mapping population. However, the preferentially transmitted alleles were from the exotic paternal PI 183967 parent.

The linkage map presented here spans 400.3 cM with a mean marker distance of 2.5 cM (Table 1). The genetic distance of this map is considerably less than the expected 750 to 1000 cM length of the cucumber genome as estimated from cytological observations (Ramachandran and Seshadri, 1986; Staub and Meglic, 1993). Given the relatively unsaturated condition of historic dominant marker-based cucumber maps, addition of codominant markers is desirable for MAS. Of particular consequence in this regard is LG 1, 2, 4, and 7, in which only 11, 6, 18, and 15 SSR markers were mapped here, respectively, in which each possessed varying estimated physical map length intervals (Table 1). There is, generally, a positive correspondence between genetic (LG) and physical (putative chromosomal) lengths in cucumber (Ren et al., 2009). Chromosomes 3 (LG 3) and 6 (LG 6) are the longest among the seven (Koo et al., 2005; Table 1), and these are, in fact, the LG having the longest map length in this study (i.e., LG 3 = 82.9 and LG 6 = 95.6 cM, respectively; Table 1). In contrast, such chromosome–LG associations cannot be extrapolated for the shortest chromosomes (5 and 7), although the marker concentrations on some LG (e.g., LG 5 having 46 markers) were relatively high. More accurate map length estimation must, therefore, await greater map saturation and intensive chromosome structure analysis (i.e., fluorescent in situ hybridization).

Marker-assisted selection can enhance conventional plant improvement efforts by increasing its efficiency (Collard and Mackill, 2007). The efficiency of MAS is dependent on many factors, including the strength of marker/trait associations (i.e., linkage) and a knowledge of epistasis, correlated traits, and genotype × environment interactions (Wang et al., 2007). The strategic use of ll and de trait/marker associations in MAS could increase the efficiency of selection to alter plant architecture for once-over harvest operations (Staub et al., 2008). The genes for little leaf (ll) and determinate (de) growth habit were both mapped onto LG 6 (Fig. 1), where the ll was flanked by SSR02355 and SSR03940 at 4.2 and 3.6 cM, respectively, and XCSWCTT14b and XSSR13251 flanked de at 1.4 and 4.2 cM, respectively. When compared with historic maps (Bradeen et al., 2001; Fazio et al., 2003a), the linkage map constructed here defines robust trait/marker relationships that could be exploited in MAS as has been the case with quantitative trait loci (Fan et al., 2006; Fazio et al., 2003b; Robbins and Staub, 2009). Moreover, because the whole genome sequence of cucumber (through Line 9930) is available (Huang et al., 2009), fine mapping of these genes should be feasible. For instance, intensive analysis of ll map position should be possible given the known whole genome scaffolds (sca) of flanking markers SSR02355 [resides in sca000028 (≈1.91 Mbp)] and SSR03849 [resides in sca000024 (≈2.22 Mbp) (Huang et al., 2009)]. As fine mapping is performed, putative candidate gene(s) may be identified and further evaluated.

Comparison of narrow and wide cross-based cucumber genetic maps.

Ren et al. (2009) used 995 SSR markers to saturate the Gy14 × PI 183967 wide-based map constructed by Kennard et al. (1994, a 70-point map of 480 cM in 10 LGs) using an F_2:3 population derived from the same initial parental cross. The map of Ren et al. (2009) spanned only 635.4 cM as a result of strong recombination suppression (LG 4, 5, and 7) and segregation distortion (LG 1, 4, and 6) in several genomic regions.

Ninety-four common marker loci were shared between the map constructed by Ren et al. (2009) and the one constructed here to facilitate synteny analysis. These maps shared seven, five, 18, 10, 24, 21, and nine markers in LG 1 to 7, respectively. Although the marker orders in LG 1, 2, 4, and 7 were not compared because of limited common markers, marker order concordance among 21 common markers in LG 6 indicated map colinearity (Fig. 1). Likewise, 16 of 18 common marker loci in LG 3 were collinear (SSR15419 and SSR15519 were not collinear). A comparative analysis of LG 5 is shown in Figure 2. In LG 5 of Ren et al. (2009), 99 markers clustered in two regions (63 at 6.0 cM and 36 at 7.9 cM). Among 11 markers defined in these two cluster regions, 10 were distributed across the Gy7 × H-19 RIL map (Fig. 2). For example, although two marker loci, XSSR14247 and XSSR15321, mapped to the same cluster in LG 5 (at 6.0 cM) in the wide-based map (i.e., Gy14 x PI 183967), their map interval was 50.8 cM in the narrow-based Gy7 × H-19 map, suggesting that recombination suppression did not occur in this region of the latter map. Moreover, the marker map locations across the two clusters in LG 5 of the wide-based map were distributed across the length of this LG in the narrow-based map (Fig. 2). The disparity between these maps may be caused by possible chromosomal structural changes (e.g., translocations, inversions) that have occurred in cultivated and wild cucumbers (Huang et al., 2009; Ren et al., 2009), although genotyping errors could not be excluded.

View Full Size

Comparative marker order analysis of cucumber (Cucumis sativus L.) Chromosome 5 as derived from mapping in a wide-based Gy14 × PI 183967 (left; var. hardwickii Alef. R.; redrawn from Ren et al., 2009) and narrow-based Gy7 × H-19 recombinant inbred line (RIL) populations (right; var. sativus). Only markers common to both maps are shown. Markers linked with solid lines define cross map colinearity.

Citation: HortScience horts 45, 6; 10.21273/HORTSCI.45.6.882

• Download Figure
• Download figure as PowerPoint slide

The position of marker SSR05819 was not concordant between the two maps examined (Fig. 2). Marker SSR05819 is located in the line 9930-derived cucumber whole genome scaffold designated sca000104. Several markers from this scaffold mapped to the 6.0 cM cluster in LG 5 in the wide-based cross map (Ren et al., 2009). Thus, it is likely that the location of this marker on the wide-based map was the result of an error during marker-map assignment. In addition to SSR05819, the order of four adjacent markers (SSR03529, SSR07100, SSR00772, and SSR17464) at the distal end of LG 5 of the Gy7 × H-19 map was not syntenic with that of the Gy14 × PI 183967 map. This marker order disparity may be indicative of chromosomal rearrangements that have occurred during domestication (i.e., cultivated versus feral cucumber comparisons). However, because these four and other two markers [SSR01498 (68.5 cM location) and SSR00170 (69.2 cM location)] in LG 5 (Fig. 2) belong to the same scaffold (sca000003) of the line 9930 whole genome assembly (Huang et al., 2009), it is also possible that the map position of these markers is incorrect in the narrow-based cross. Thus, additional fine map analysis will be needed to elucidate this apparent disparity.

The comparative mapping conducted here confirmed marker order differences between C. sativus var. sativus and var. hardwickii identified by Ren et al. (2009) and lent support to the hypothesis of chromosomal structure changes in cucumber. This study also identified the determinate (de) and little leaf (l) genes to be located in the long and short arms of chromosome 6, respectively, as tagged by flanking SSR markers.

The 174-point map developed here improves the marker saturation of the reference Gy7 × H-19 map developed by Fazio et al. (2003a) by 24.7%. It adds 140 codominant SSR markers to this narrow-based map. Moreover, unlike the case in the wide-based genetic map of Ren et al. (2009; Gy14 × PI 183967), no suppression of genetic recombination or significant segregation distortion was detected in the Gy7 × H-19 genetic map.

It has been long thought that determinate, gynoecious, multiple lateral branching U.S. processing cucumber phenotypes may improve yield potential in once-over mechanical harvest operations (Staub et al., 2008). Gain from selection for some yield and quality components have not been fully realized as a result of a lack of map saturation (weak trait/marker linkages) and a meager understanding of epistatic interactions (Robbins et al., 2008). For instance, the little leaf trait is pleiotrophic to multiple lateral branching, but also confers poor fruit quality. It may be possible, using previously identified marker/trait associations (Fazio et al., 2003b) and those identified here for de and ll, to apply MAS in the seedling stage to recover gynoecious, determinate, little leaf types that possess multiple lateral branching and then evaluate fruit quality characteristics in the open field and by salt stock evaluation, thus increasing selection efficiency and effectiveness.