Introduction

Cucumber (Cucumis sativus L., 2n = 2x = 14) belongs to the family Cucurbitaceae (Jeffrey 1980; Mliki et al. 2003) and is one of the most important vegetable crops in many countries of the world. Total world production of cucumber in 2009 was 39.3 million tons, of which over 59% (23.2 million tons) were produced in China (FAOSTAT 2011, data available at http://faostat.fao.org/). Despite its economic importance, until very recently, genetic and genomic resources for cucumber are relatively scarce as compared with many field crops.

The most recent cucumber gene catalog documented approximately 140 mutants in cucumber for various morphological traits or disease resistances, but almost half of those mutants were unfortunately lost or unable to locate (Pierce and Wehner 1990; Wehner 2005). Inheritance and linkage relationships among some of these genes have been explored, with emphasis on disease resistance and fruit-related genes. For example, several genes for epidermal structures of cucumber fruits characteristics of different market classes have been identified and characterized (Fanourakis 1984; Fanourakis and Simon, 1987a, b; Walters et al. 2001, Wehner, 2005). All epidermal features seem to be controlled by simply inherited genes. Examples may include the gene u for uniform immature fruit color, H for heavy netting, Tu for tuberculate (warty fruit), D for dull fruit skin, ns for numerous spines, ss for small spines, and te for tender fruit (Wehner 2005). All these genes seem to be in the same linkage group (Fanourakis and Simon 1987a, b; Meglic and Staub 1996; Walters et al. 2001; Zhang et al. 2009). Interestingly, for several of these genes, it was believed that each gene in the recessive form (u, tu, d, and te) was consistent with European-type cucumbers and each in the dominant form (U, Tu, D and Te) was consistent with American-type cucumbers (Wehner 2005).

Marker-assisted selection (MAS) is playing an increasing role in accelerating conventional plant breeding. A moderately saturated or high-density linkage map and closely linked molecular markers for target genes are two requirements for efficient MAS. In cucumber, however, due to its narrow genetic base and thus low polymorphisms (Kennard et al. 1994), although nearly 30 linkage maps have been constructed, the majority of molecular markers placed on these low-resolution genetic maps are RAPDs (randomly amplified polymorphic DNAs) or AFLPs (amplified fragment length polymorphism) (e.g., Kennard et al. 1994; Serquen et al. 1997; Park et al. 2000; Fazio et al. 2003; Robbins et al. 2008; Fukino et al. 2008; Yuan et al. 2008) that are not breeder friendly. In additional, only a few horticulturally important genes or QTL (quantitative trait loci) of cucumber have been mapped with molecular markers (e.g., Kennard and Havey 1995; Sakata et al. 2006; Lui et al. 2008; Yuan et al. 2008). Most of the markers associated with target genes are not suitable for use in marker-assisted selection due to their dominant nature (RAPDs, AFLPs).

The sequencing of the whole cucumber genome (Huang et al. 2009) made it possible to use more breeder friendly molecular markers such as microsatellites (simple sequence repeats, SSRs) for genetic mapping and marker-assisted selection in cucumber. The usefulness of these cucumber microsatellite markers has already been demonstrated in several recent linkage mapping, and marker-trait association studies (for example, Zhang et al. 2009, 2010; Weng et al. 2010). Previously, we constructed a cucumber linkage map placing 995 SSR markers in seven linkage groups (Ren et al. 2009). However, this cucumber map was developed with only 77 recombinant inbred lines (RILs) from an inter-varietal (inter-subspecific) cross between cultivated cucumber Gy14 and wild cucumber (C. sativus var. hardiwickii) line PI 183967. Strong recombination suppression and thus clustering of molecular markers was found in several regions due to possible structural changes between cultivate and wild cucumbers. This necessitates development of linkage maps with intra-varietal mapping populations in cultivated cucumbers.

In the present study, we developed a recombinant inbred line (RIL) population from two cultivated cucumber inbred lines, the European glasshouse type 9110Gt and the Northern China type 9930 which are segregating for seven horticulturally important genes (see below). The objectives of this study were to develop a genetic map of cucumber with this intra-varietal cross RIL population with SSR markers and identify molecular markers for the seven genes.

Materials and methods

Plant materials

A set of 148 F9 recombinant inbred lines (RILs) was developed from the cross between two cultivated cucumber inbred lines 9930 and 9110Gt. 9110Gt was derived from the cross between a European greenhouse hybrid and a Northern Chinese line with dominant European glasshouse cucumber genetic background, while 9930 is a typical Northern China fresh market type cucumber. A single F1 plant from 9110Gt × 9930 mating was self-pollinated to produce F2 progeny which were then advanced to F9 by single-seed descent to generate 148 RILs. The two parental lines are different in seven leaf-, fruit- and sex expression-related traits including leaf color (virescent vs. normal green), foliage bitterness (bitter vs. non-bitter), sex expression (monoeocious vs. gynecious), immature fruit color (uniform vs. non-uniform), and fruit skin texture (heavy netting vs. non-netting; ribbing vs. non-ribbing and dull vs. glossy fruit skin). Details of these traits and methods of phenotyping of each trait are summarized in Table 1. For virescent leaf mutant plant, the cotyledons, first and second true leaves are yellow but return to normal green starting from third true leaf. Seedlings with virescent leaves usually grew slightly slower than the normal plant with green leaves. All these traits were relatively easy to differentiate among the segregating population.

Table 1 Description of seven cucumber horticultural characteristics segregating in 9110Gt × 9930 RIL population
Full size table

Phenotypic observations of morphological traits and data analysis

Phenotyping of the seven traits (Table 1) for the RIL mapping population was conducted in four greenhouse seasons in Beijing, China (2006 spring and autumn, 2007 autumn, and 2009 spring). Plants from the two parental lines, F1 and 148 RILs were grown in the greenhouse and phenotypic data were recorded. For each growing season, the experimental design was the same which was a randomized complete block design (RCBD), consisting of three blocks with five plants per plot for each RIL (in total, 15 plants per RIL).

Seeds of test materials were sown in pot soil in the greenhouse, and the seedlings were transplanted into greenhouse at the three-leaf stage. The morphological traits were evaluated at the time when they were fully expressed, which were generally straightforward. The details of evaluation for each trait are shown in Table 1. For bitterness, a small piece of tissue from the cotyledon of 5-day old seedling plants and leaf of adult plants was tasted for bitterness by three people who were sensitive to bitterness. Classification for the flower types was made when the plants reached a height with 20 nodes and were characterized as either monoecious (with both staminate and pistillate flowers) or gynoecious (with only pistillate flowers).

SSR markers analysis

In our previous studies, many SSR markers were developed from whole genome sequence of 9930 (Ren et al. 2009; Huang et al. 2009). In the present study, 2,304 SSR primer pairs from Ren et al. (2009) and 112 from Fazio et al. (2002, 2003) were employed for polymorphism screening between 9930 and 9110Gt parental lines. Polymorphic SSRs were applied to the RIL population for linkage analysis. Each polymerase chain reaction (PCR) was performed in a 15 μl volume containing 20 ng template DNA, 1 × PCR buffer, 0.5 unit Taq DNA polymerase (Tiangen Biological Company, Beijing, China), 1.0 μM of forward and reverse primers, and 0.2 mM dNTPs. Optimized PCR program included an initial denaturing step of 4 min at 94°C, followed by 35 cycles of 15 s at 94°C, 15 s at 55°C, 30 s 72°C, and a final extension at 72°C for 4 min. Subsequently, 4 μl of PCR product was used for electrophoresis in 8% polyacrylamide gels (120 V for 2 h).

Data analysis

Linkage analyses were performed with JoinMap 3.0 software (Van Ooijen and Voorrips 2001). Segregation of SSR markers and seven morphological traits was analyzed for conformation to expected ratios in RILs using a chi-square goodness-of-fit test, which was a built-in function of JoinMap 3.0. Marker data were assigned to linkage groups (LGs) using a minimum logarithm of odds (LOD)-likelihood score of 4.0. The Kosambi map function (Kosambi 1944) was used to calculate the genetic distance between markers.

Assignment of chromosomes of the seven linkage groups was based on common markers between the present map and that of Ren et al. (2009). The order of mapped loci between the two maps was compared to verify possible structural changes detected between cultivate and wild cucumbers (C. sativus var. hardwickii) (Ren et al. 2009).

Results

Inheritance of seven morphological characteristics in cucumber

Segregation of seven morphological traits among 148 RILs from 9930 × 9110Gt was evaluated in four greenhouse seasons over a three-year period. The data collected from the four greenhouse seasons were highly consistent, and no seasonal variation of expression of these traits was observed. Goodness-of-fit tests indicated that all the seven morphological loci demonstrated good agreement with expected 1:1(P = 0.05) segregation among the 148 F9 RILs (Table 2), which was consistent with simply inherited genes underlying each trait.

Table 2 Segregation of seven horticultural characteristics in 9110Gt × 9930 F9 RILs population
Full size table

The phenotypic character of F1 was normal green leaf color, bitter foliage, gynoecious sex expression, non-uniform immature fruit color, heavy netting, ribbing and dull fruit skin. Among seven morphological traits, virescent leaf (v-1), uniform immature fruit color (u), bitterfree foliage (bi) were each controlled by a single recessive gene, whereas single dominant genes were conditioning gynoecy (F), dull fruit skin (D), fruit netting (H) and fruit ribbing(Fr) (Table 1).

Ribbing fruits are common in many Northern China ecotype cucumber lines like 9930. This is the first report that ribbing fruit was controlled by a single dominant gene. We propose this gene be designated as Fr (fruit ribbing).

Linkage map construction with microsatellite and morphological markers

Screening of the two parental lines with 2,416 SSR primer pairs resulted in the detection of 320 polymorphic markers between 9110Gt and 9930 with a polymorphic rate of 13.2%. All 320 polymorphic SSRs were applied to the RIL population. However, data from 69 SSRs were not used in linkage analysis for various reasons (for example, too many missing data points, difficulty in scoring), and 3 failed to be assigned to any linkage group. Thus, data from 248 SSRs and seven morphological markers (bi, F, v-1, u, d, H, fr) were employed for linkage analysis and map construction.

The resulting cucumber genetic map (Fig. 1) has seven linkage groups (LG, corresponding to seven cucumber chromosomes, see below) with 248 SSR loci and 7 morphological trait loci spanning 711.9 cM. The average marker interval was 2.8 cM with the largest interval between two adjacent markers was 20.4 cM (SSR23474 and SSR05723) in LG1. Each LG on average contained 36 loci with mean genetic distance of 101.7 cM, but they varied significantly among different linkage groups (Table 3).

Fig. 1
figure 1

A SSR linkage map of cucumber developed with 148 recombination inbred lines from two cucumber inbred lines 9110Gt and 9930. Map distance was given in centimorgans (cM). LG = linkage group which was shown on top of each map together with corresponding chromosome assignment

Full size image
Table 3 Distribution of SSR and morphological markers among seven cucumber chromosomes mapped with RIL population from 9110Gt × 9930
Full size table

Of the 255 markers mapped, Chi-square tests indicated that 46 (18.0%) showed segregation distortion with the RIL population (see loci with asterisks in Fig. 1). In linkage groups 1–7, there were respectively 2, 8, 11, 0, 9, 15 and 1 markers with deviation from the expected 1:1 segregation among the 148 RILs. Clustering of loci with distorted segregation along the LG was obvious, which was especially true in linkage groups 3, 5 and 6 (Fig. 1). Among the 46 markers with segregation distortion, 17 biased toward 9110Gt and 29 favored 9930.

Linkage analysis placed the genes for seven horticultural traits into two linkage groups: LG5 and LG6. Interestingly, the four genes, u (uniform immature fruit color), fr (no fruit ribbing), d (glossy fruit skin) and H (heavy fruit netting) were clustered in a 2.4 cM region in LG5 (Fig. 1) flanked with microsatellite markers SSR15818 and SSR06003. The genes for bitterfree foliage (bi), gynoecy (F) and virescent leaf (v-1) were mapped in different locations of LG6. The loci SSR01331 and SSR18405 were found to flank v-1 locus 1.6 cM and 0.3 cM, respectively. The bi locus was mapped to an interval between markers SSR02309 (3.3 cM) and SSR00004 (1.9 cM), and the F locus was flanked by markers SSR13251 and CSWCT28 at 1.2 cM and 1.7 cM, respectively.

Comparison of 9930 × 9110Gt intra-varietal cross with Gy14 × PI 183967 inter-varietal cross maps

Ren et al. (2009) developed a high-resolution genetic map of cucumber using a RIL mapping population from cross between cultivated Gy14 and wild cucumber line PI 183967. The map developed from the present study shared 149 SSR loci with the Gy14 × PI 183967 RIL map, which enabled assignment of the linkage groups into seven corresponding cucumber chromosomes (Chromosomes 1–7, Fig. 1). Of the 149 common markers on the two maps, 12, 20, 30, 9, 31, 33 and 14 were located in Chromosomes 1–7, respectively (Table 3). These common markers also made it possible to compare the order of loci between the two maps. The order of SSR markers in Chromosomes 1, 2, 3 and 6 were largely consistent. For Chromosomes 4, 5 and 7, strong suppression of recombination was observed in the Gy14 × PI 183967 mapping population due to possible chromosome structural changes between cultivated and wild cucumbers (Ren et al. 2009). Comparisons between each pair of the three chromosomes revealed inconsistent orders of SSR marker loci in the three chromosomes. The comparison of Chromosomes 4, 5 and 7 maps is shown in Fig. 2A–C, respectively. Unlike in the inter-varietal mapping population, no significant recombination suppression was observed in the 9110Gt × 9930 intra-varietal mapping population in this study. For example, in Chromosome 5, six markers clustered on the wild cross map at 6.0 cM location spanned 53.0 cM (from 9.3 to 62.3 cM) on the narrow cross linkage map (Fig. 2B). The total genetic distance of LG5 was 118.8 cM in the present study which was in sharp contrast to the 59.9 cM genetic distance of the same chromosome in Ren et al. (2009) although the mapped loci in this chromosome were only one third of those by Ren et al. (2009) (54 vs. 160) (Fig. 1). These observations supported the postulation of possible chromosomal structural changes between cultivated and wild cucumbers (Ren et al. 2009; Huang et al. 2009). However, due to limited number of shared markers between the two maps, the nature and location of the structural changes are not known.

Fig. 2
figure 2

Comparison of marker orders in cucumber linkage groups 4 (A), 5 (B) and 7 (C) constructed from RIL populations of intra-varietal cross 9110Gt × 9930 (this study) (right) and inter-varietal cross Gy14 × PI 183967 (Ren et al. 2009) (left). Only common markers were shown on the wide-cross map. Vertical bars designated chromosomes. Solid or dashed lines between two maps connected common markers

Full size image

Discussion

The new intra-varietal genetic map of cucumber

Ren et al. (2009) constructed a wide cross-based high-density linkage map placing 995 SSR loci on the map. Weng et al. (2010) developed an intra-varietal genetic map in cucumber placing 174 SSR or SCAR marker loci and two genes (ll for little leaf and de for determinate growth habit) on the map with RILs developed from two cucumber inbred line Gy7 and H-19. In the present study, we constructed a second cucumber intra-varietal linkage map with 148 RILs and 248 microsatellite marker loci, which has the most mapped SSR markers on any intra-varietal cucumber genetic maps developed so far. The total length of this map was 711.9 cM which was close to the expected 750–1,000 cM as estimated from cytological evidence (Ramachandran and Seshadri 1986; Staub and Meglic 1993). While the genetic distances of Chromosomes 2, 3, 5, 6 and 7 were largely consistent to their physical lengths (Koo et al. 2005), the number of markers seems to be under-represented for Chromosomes 1 and 4, in which only 17 and 15 markers were mapped, respectively. The same trend was also found in Weng et al. (2010). The reason why so few SSRs were mapped in the two chromosomes is not known. As compared with the inter-subspecific cross genetic map (Ren et al. 2009), the 248 mapped marker loci seem to distribute evenly among seven chromosomes, and no obvious genetic recombination suppression, thus clustering of molecular markers was observed. The average marker interval across seven linkage groups was 2.8 cM (Table 3). Although this marker density is far from being high resolution, it provides a good reference for genetic mapping of target genes and is also useful for marker-assisted selection.

Segregation distortion (SD) is the deviation of genetic segregation ratios from their expected Mendelian fraction. In the present study, we found 18.0% (46 of 255) of markers with derivation from expected 1:1 segregation (Fig. 1). It was also obvious that segregation distorted loci often clustered. Among the 46 markers with SD, more than 67.4% (31/46) were concentrated in five chromosome regions, and almost one third (15 of 46) were in the long arm of Chromosome 6 (Fig. 1). SD and clustering of SD loci seem to be a common phenomenon found in cucumber genetic mapping studies (Yuan et al. 2008; Ren et al. 2009; Weng et al. 2010; this study). For example, in a QTL mapping study with 224 RILs derived from S94 (Northern China type) × S06 (Northern European type), Yuan et al. (2008) found 75 out of 258 mapped markers (29%) with segregation distortion. On the high-resolution cucumber map by Ren et al. (2009), three segregation distortion regions (SDRs) were detected in Chromosomes 1, 4 and 6. The segregation of 62% (71/114) of SSRs mapped to Chromosome 4 was distorted forming an SDR that spanned the entire short arm and the proximal portion of the long arm. The reason causing segregation distortion is unknown. Ren et al. (2009) found that all SSR marker loci within three SDRs were associated with the C. sativus var. hardwickii parent (PI 183967) and they suggested that possibly interacting allele pairs with strong effects on pollen or embryo viability or germination are located on the SDRs and that the ‘wild’ alleles confer stronger viability than the ‘domesticated’ ones. In the present study, among the 46 markers with segregation distortion, 17 biased toward 9110Gt (37.0%) and 29 (63.0%) favored 9930. Interestingly, in Yuan et al. (2008), of the 75 SD markers, 51 loci (68%) favored the female parent (S94) and 24 (32%) favored the male parent (S06). It is not known if the bias toward one parental line is due to differences between two cucumber ecotypes (or market classes, that is, Northern China fresh vs. European glass house).

Whole genome sequencing of the cucumber genome enabled large-scale development of molecular markers and thus linkage maps in cucumber (Ren et al. 2009; Cavangnaro et al. 2010). So far three SSR-based cucumber maps (Ren et al. 2009; Weng et al. 2010, and the present study) have been developed. The current map shared 39 common markers with Weng et al. (2010) and 149 loci with Ren et al. (2009). Seventy eight SSR loci were newly mapped. While the two intra-varietal SSR maps by Weng et al. (2010) and the present study used the same system as Ren et al. (2009) in assignment of the seven linkage groups into corresponding seven cucumber chromosomes. It is difficult to establish the correspondence of the seven linkage groups of our map with historical maps such as those by Bradeen et al. (2001) and Fazio et al. (2003) due to limited number of common markers. For example, we found only three common microsatellite markers (CSWCT28, CMGA165 and CSWCT13) between our map and that of Fazio et al. (2003). Since the high-resolution map by Ren et al. (2009) was backed up with cytological data (Koo et al. 2005; Han et al. 2008), we propose that the chromosome assignment system by Ren et al. (2009) as the new standard for cucumber genetic maps. It is expected that SSR markers will become more popular in various studies, thus, this system should be more convenient to use. In addition, based on the chromosomal locations of a number of already characterized cucumber genes, the locations of many of these genes could be placed onto different chromosomes (see below).

Inheritance and linkage of epidermal features of cucumber fruits

The immature fruits of cucumber is consumed fresh or cooked, or as pickles. Fruit-related characteristics are considered the most important traits for cucumber varietal improvement. However, fruit-related traits in cucumber are poorly understood genetically and only a few studies have been reported. The two parental lines of the RIL population, 9110Gt and 9930 used in this study belong to two different market classes; 9110Gt is a European greenhouse background cucumber and 9930 is a typical Northern China fresh market cucumber. Among the seven morphological traits (Table 1) segregating in this population, four fruit quality-related traits are associated with the two market class cucumbers: the European greenhouse cucumbers have uniform immature fruit color (u), smooth fruit (no ribbing, fr), glossy (d) fruits and heavy fruit netting (H), whereas typical Northern China cucumbers have non-netting (h), dull fruit (D) with mottled (U) and ribbed (Fr) skin. Segregation analysis indicated that all the four cucumber fruit epidermis characteristics were simply inherited. Interestingly, all the four genes were tightly linked in Chromosome 5 (Fig. 1), which is consistent with previous studies (Fanourakis and Simon 1987a, b; Meglic and Staub 1996; Walters et al. 2001; Heang et al. 2008; Yuan et al. 2008). In addition, this is the first study on the inheritance of fruit ribbing which was shown to be controlled by a single, dominant gene, Fr.

In addition to H, d, u and fr that were shown to cluster in cucumber Chromosome 5 in the present study, previous studies also indicated several other cucumber fruit epidermal structure-related genes were linked with the four genes which include Pe (palisade epidermis), ns (numerous spines), ss (small spines), and Tu (tuberculate fruit) and te (tender fruit) (Fanourakis and Simon 1987a, b; Wehner 2005). The Tu gene was mapped in Chromosome 5 linked with ss (small spines) by Zhang et al. (2009) with molecular markers. Thus, it seems that all nine genes controlling cucumber fruit epidermal features are located in Chromosome 5. On the other hand, for the first time, the virescent leaf locus v-1 and the locus for bitterfree foliage (bi) were mapped in cucumber Chromosome 6 (Fig. 1). It is known that Chromosome 6 also harbors several other genes like gynoecious sex expression (F), little leaf (ll) and determinate (de) growth habit (Weng et al. 2010) which were not linked with the scab resistance gene Ccu gene as suggested by Meglic and Staub (1996). Rather, Ccu has been mapped in cucumber Chromosome 2 (Zhang et al. 2010; Kang et al. 2010).

Toward map-based cloning of horticulturally important genes in cucumber

Linkage analysis placed the seven genes into two chromosomes anchored with molecular markers (Fig. 1). The recently assembled draft genome of the parental line 9930 (Huang et al. 2009) makes it more feasible to perform map-based cloning of these horticulturally important genes. Take the F gene for sex determination for example. Cucumber has long been a model system for studying sex expression in plants (Tanurdzic and Banks 2004). Ethylene plays a critical role in promoting femaleness and is considered the sex hormone of cucumber (Rudich et al. 1972). F gene was found to encode l-aminocyclopropane-1-carboxylate (ACC) synthase (CsACS1G) which is a key enzyme in the ethylene biosynthesis pathway (Mibus and Tatlioglu 2004). In the present study, two markers flanking the F locus, SSR13251 and CSWCT28, were identified which were 1.2 and 1.7 cM away from the F locus, respectively. Sequence analysis of the 9930 whole genome scaffolds flanked by the two SSR markers indeed identified coding sequences homologous to CsACS1G (data not shown), which is probably a candidate gene of the F gene in the cucumber line 9110Gt.

A second example is the bitterfree gene (bi). Cucurbitacins, a group of highly oxygenated tetracyclic triterpenes, occur widely in Cucurbitaceae species (Shibuya et al. 2004). Previous studies suggested that bitterness in cucumber was controlled by two genes, Bt (bitter fruit) and Bi (bitter foliage) (Pierce and Wehner 1990). Based on the flanking marker information of bi gene in this study (Fig. 1), we were able to pin down a 9930 scaffold that harbors a triterpenoids synthesize gene (Huang et al. 2009) which is an ortholog of CPQ gene in squash (Cucurbita pepo L.). We are doing fine mapping and function validation of the candidate gene(s) aiming to cloning of the bi gene from cucumber.