Introduction

Cucumber, Cucumis sativus L., is among the most widely cultivated and consumed vegetable crops throughout the world. In 2017, cucumber was grown on 919,146 hectares with a total production of 83,753,861 tons worldwide, and China is the largest producer with 77.4%, and 54.4% total production and acreage of the world, respectively (www.fao.org/faostat/en/). Cucumber was the first among major horticulture crops with a publicly available draft genome. The small, diploid genome (~400 Mbp), annual growth habit, self-compatible mating system, and relatively short life cycle (~3 months from seed to seed) offer significant advantages for genetic studies. The development of high-quality draft genomes and high-density genetic maps, coupled with utilization of high-throughput genotyping methods have greatly accelerated genetic mapping and gene/QTL cloning in cucumber. The 2016 Cucumber Gene Catalog documented 199 simply inherited genes or major-effect QTL1. In recent years, many genes listed in the catalog as well as new ones have been molecularly characterized or fine mapped. Hundreds of QTL for horticulturally important traits have been identified. While a wealth of data has been accumulated, a systematic review and inventory of the mutants, molecularly characterized or tagged genes, and QTL for cucumber is lacking. In addition, the QTL names used in various studies are inconsistent and confusing. It is imperative to develop a community standard for assignment of QTL names. Therefore, the objectives of this article are to: (1) review cloned and fine mapped genes or major-effect QTL. (2) Develop recommendations for QTL nomenclature for future QTL mapping studies. (3) Inventory published QTL in cucumber.

Genes conferring simply inherited traits

As of July 2019, candidate genes have been identified for 51 simply inherited traits in cucumber (Table 1). Genes for additional 30 traits have been fine mapped with the target loci delimited to <2.0 Mbp (Table 2). It should be pointed out that, we used “cloned gene” in this review not in its strict term because for many mutants, identification of the candidate genes was based on genetic evidence, and their functions have not been validated or verified. Also, some major-effect QTL were counted as simply inherited genes, which often contribute to >20% observed phenotypic variance in QTL analysis. Details of the 81 genes are presented in supplementary File 1 (Table S1) including polymorphisms between the parents in the candidate gene, diagnostic markers, and primer sequences. Allelic variants for six genes (cul, gl1, gl3, m, pm, and rl) are listed separately in Table S1 (hence the total number is 88). The two variants of the CsGL3 gene exhibit different phenotypes, which are listed as two genes. Three genes have names that are duplicated with previously reported ones including glabrous2 (gl2)2, ts (tender spine)3, and sf-1(ref. 4), which were re-assigned gl4, tsp, and sf-2, respectively. The CsSEP gene was the candidate for a mutant with very long sepals5, which was assigned els-1 (extra-long sepal-1) in this work.

Table 1 Details of identified genes for simply inherited traits in cucumber (as of July 2019).
Full size table
Table 2 List of fine mapped genes or major-effect QTL in cucumber (as of July 2019).
Full size table

Among the 81 cloned or fine mapped genes or major-effect QTL, 14 are EMS-induced mutations, and the rest are spontaneous mutations identified from natural populations. Of the 51 cloned candidate genes, 42 mutants are due to SNPs; other polymorphisms include small or large deletions, and retrotransposon insertions. In most cases, the SNPs or insertions result in frame shift or amino acid substitutions, or alternate splicing (supplementary File 1). For convenience, the 81 genes/QTL were classified into six categories: Vegetative organ (23), Flower (7), Fruit (28), Disease resistance (18), Abiotic stress tolerance (1), and Miscellaneous (MISC) (4). Phenotypes of some representative plant architecture, leaf or fruit mutants are shown in Fig. 1. Distribution of the 81 genes or QTL across 7 cucumber chromosomes are illustrated in Fig. 2.

Fig. 1: Phenotypes of representative mutants in cucumber.
figure 1

a–f shows mutant and wild-type phenotypes for littleleaf (ll, A2), glabrous3 (gl3, B2), roundleaf (rl, c), super compact-1 (scp-1, D2), short hypocotyl1 (sh1, E1), yellow plant (yp, F1), respectively. g thru j show phenotypic variation in spine size and density (g), fruit flesh color (white, orange, yellow, and green), cavity size (h), fruit size, shape, and fruit epidermal features (i, j) in natural populations.

Full size image
Fig. 2: Chromosomal locations of 81 cloned (black) or fine mapped (red) genes in cucumber.
figure 2

Ruler to the left indicates locations (in Mbp) in the Gy14 V2.0 draft genome assembly (drawn to scale).

Full size image

Establishment of controlled vocabularies to describe quantitative traits and recommendations for QTL nomenclature in cucumber

Most horticulturally important traits in cucumber are controlled by QTL. With the exponential increase of QTL mapping studies in cucumber, one complicating issue is the naming of quantitative traits and corresponding QTL, which is currently very confusing. It is common that the same name was used for different traits or different names were used for the same trait. Thus, we reviewed the literature and phenotyping manuals from both public institutions and private seed companies. We also consulted colleagues in the cucumber research community and proposed the following rules for use of abbreviations to name quantitative traits in cucumber.

  1. 1.

    For disease/insect resistances: use common names except for Fusarium wilt and Fusarium crown rot, for which FOC and FCROS have been widely used, respectively.

  2. 2.

    For a trait name with one word, use first three letters.

  3. 3.

    For a trait name with two words, use the initial from each word. In a few cases, three letters (one from the initial of one word and two from another word) are used to avoid duplication with other traits, or for better understanding of its meaning.

  4. 4.

    For traits with more than two words, use the initial from each word.

Based on inputs from the community, 130 quantitative traits were identified. Their full names and recommended QTL names (abbreviations) are listed in Table 3. Considering the common practices taken by the cucurbit research community, we also recommend the following rules in assigning QTL names:

  1. 1.

    QTL name format: Trait name.chr#.QTL order on chromosome.

  2. 2.

    When multiple QTL on the same chromosome (linkage group) are reported for the same trait, the numbering order follows the order of discovery in the literature.

  3. 3.

    The use of capital or lower case letters depends on the inheritance of the trait (dominant, co-dominant, or recessive).

Table 3 Proposed nomenclature for quantitative traits in QTL mapping studies in cucumber.
Full size table

Thus, par6.2 is the second QTL of parthenocarpic fruit set on Chr6 (more parthenocarpic fruit is recessive); Pm1.1 is the first QTL of powdery mildew resistance on Chr1 (resistance is dominant); FS5.3 is the third consensus fruit size QTL on Chr5, and fsd6.2 is the second QTL for fruit spine density on Chr6. These rules will be applied in the following discussions for all QTL described but original names are also included for clarity.

For convenience, the 130 quantitative traits were classified into eight categories: Vegetative organ (22), Flower (9), Fruit (50), Seed (6), Abiotic stress tolerance (7), Disease resistance (26), Insect resistance (9), and Miscellaneous (MISC) (1). Under each category, there are also subcategories based on specific plant organs, pathogens, or abiotic stresses (Table 3). In the following sections, under each category, we will briefly discuss selected simply inherited genes and QTL for phenotypic characteristics and their potential in cucumber breeding. Many genes and QTL have a long history of research, but only the most recent literature was cited in the text to save space. The complete list of genes/QTL and references is provided in three supplemental files (1, 2 and 3). For many genes and major-effect QTL, readers can also consult the 2016 Cucumber Gene Catalog1 for complete historical references.

Genes and QTL for whole plant vegetative growth and development

Simply inherited genes for mutants of vegetative organs

Due to the ease of identification, mutants for foliage characteristics and plant architecture traits such as leaf shape, size, color, and plant height or vine length are frequently reported. Genes responsible for eight cucumber leaf mutants have been identified. The five leaf color mutants are yp (yellow plant)6, v-1 (virescent leaf-1)7, vl (variegated leaf)8, vyl (virescent yellow leaf)9, and Psm (Paternal sorting of mitochondria)10. These mutations show a range of phenotypes. The yp plant exhibits golden yellow color throughput its life. In the v-1 mutant, the cotyledons and first 2-3 true leaves are light yellow that turn to green when fully expanded; subsequent true leaves are green from the beginning. The young leaves on the vyl mutant are yellow and gradually turn green when mature, whereas all leaves of the vl mutant show a green and light yellow/white variegation which is especially obvious on younger leaves. All these mutants show some degree of retarded growth and reduced vine length, but the fertility and fruit set seem unaffected. The Yp gene (CsCHLI) is a homolog of the gene for the Mg chelatase I subunit; Mg chelatase is a rate-limiting enzyme in the chlorophyll biosynthesis pathway. The candidate gene for Vyl is predicted to encode a DnaJ-like zinc finger protein involved in regulation of chloroplast development, whereas v-1 seems to encode a cyclic-nucleotide-gated ion channel protein (CsCGNC). The nuclear pentatricopeptide repeat 336 gene (CsPPR336) is the candidate for the Psm locus underlying paternally transmitted mosaic phenotypes10.

Wild type cucumber leaves are flat and have seven lobes with toothed or smooth margin. Three non-lobe, round leaf mutants, rl-1, rl-2 and rl have been identified, which are all due to allelic mutations in the PINOID (CsPID) gene encoding a regulator for the auxin polar transporter PIN (PIN-FORMED)11,12.

The leaf margin of the two curly leaf mutants, cul-1 and cul-2 rolls upward forming a shallow cup; both mutants are due to allelic mutations in the CsPHB gene for a class III homeodomain-leucine zipper (HD-ZIP III) transcription factor13. The tendrilless (ten) mutation is caused by a SNP in the TEN gene encoding a TCP transcription factor14. Another tendrilless-1 (td-1) mutation has been mapped to a ~190 kb region in Chr6 (ref. 15). The phenotypes of the two tendrilless mutants are very different; ten is phenotypically normal except that the ‘tendril’ develops into leaves with long petioles and thin branches, whereas td-1 mutation has more widespread pleiotropic effects.

The littleleaf (ll) mutant, which produces leaves approximately one quarter of the size of standard American pickling cucumbers, was identified ~40 years ago. LL is a homolog of Arabidopsis STERILE APETALA (CsSAP) encoding a WD40 repeat domain-containing protein16. QTL analysis revealed co-localization of major-effect QTL for fruit size, fruit weight, seed weight, and multiple lateral branches with the LL locus indicating pleiotropic effects of the ll mutation. In addition, ll cucumbers often have poor internal fruit quality, which may hinder its use in pickling cucumber breeding.

Plant architecture, especially plant height or vine length, is important in cucumber breeding. So far, six mutants with reduced internode length or compact growth habit have been reported including compact (cp)17, compact-1(cp-1)18, short internode (si)19, super compact-1(scp-1)20, super compact-2 (scp-2)21, and dwarf (dw)22. The cp-1, dw, scp-1, and scp-2 mutants have extremely short internodes with little value in practical use. Both scp-1 and scp-2 are due to mutations of genes in the brassinosteroid (BR) biosynthesis pathway including CsCYP85A for the BR-C6-oxidase, and CsDET2 for the steroid 5-alpha-reductase20,21. The si mutant exhibits short internode (~50% of WT) and small fruit, which is a homolog for the gene encoding a member of the VIER F-BOX PROTEIN subfamily of the F-Box protein family (CsVFB1)19.

Hypocotyl elongation of modern commercial cucumbers is sensitive to environmental conditions. For example, high temperature or low light intensity may increase hypocotyl length resulting in poor seedling quality for transplanting. The semi-wild Xishuangbanna (C.s. var. xishuangbannesis, XIS) and wild (C.s. var. hardwickii, HARD) cucumber populations are enriched with the short hypocotl1 (sh1) allele, which renders hypocotyl elongation insensitive to UVB-free light and temperature changes23. Sh1 (CsSH1) is a homolog of the gene encoding a human SMARCA3-like chromatin remodeling factor. The sh1 mutation may be of value in use for mass seedling production in protected environments.

Four glabrous (trichome-free) mutants have been reported. The “glabrous1” (csgl1) or “micro-trichome”(mict) mutant shows no visible trichomes on all aerial organs except the hypocotyl. CsGL1 encodes a Class I HD-ZIP transcription factor24,25. The csgl2 mutant exhibits glabrous stem, petioles, and leaves, but fruit, sepals, fruit peduncles, and flower pedicel are covered with sparse and fine hairs, and the candidate gene for this mutation is unknown26. The csgl3 (tril) mutant is completely free from trichomes which encodes a Class IV HD-ZIP transcription factor; the glabrous phenotype in csgl3 is due to either SNPs or retrotransposon insertion in the coding region27,28,29. The csgl4 mutant has glabrous fruit skin but reduced size and number of trichomes on the stem and leaves; CsGL4 was thought to encode a C-type lectin receptor-like tyrosine-protein kinase2.

QTL for vegetative growth and development-related traits

Significant variation exists among cucumbers for size of vegetative organs such as hypocotyl length (HL), cotyledon area (CA), leaf area (LA), vine length (VL), internode length (IL), total number of nodes (NN), lateral branch number (LBN), and whole plant (above-ground) biomass (BIO). QTL mapping studies for these traits were conducted primarily using three RIL populations from the following crosses: S94 × S06, 9110Gt × 9930, and PI 183967 × 931 (ref. 30,31,32,33,34). Details of all detected QTL for these traits and their chromosomal locations are presented in Fig. 3, and Supplementary Files 3 (Table S3) and 4 (Fig. S1).

Fig. 3: Chromosomal locations of vegetable organ-, flower- and fruit set-related QTL on cucumber chromosomes 1, 5, and 6.
figure 3

Ruler to the left indicates locations (in Mbp) in the Gy14 V2.0 draft genome assembly (drawn to scale). Vertical black lines are chromosomes. Cloned genes and QTL are aligned to the left and right of each chromosome, respectively. Vertical bar for each QTL represents 1.5 or 2.0 LOD confidence interval on the chromosome. Dashed rectangles indicate gene/QTL hot spots or clusters. CA = cotyledon area, CsFS = consensus fruit size and shape, FFN = first flower node, FT = flowering time, FPL = fruit peduncle length, FW = fruit weight, HL = hypocotyl length, IL = internode length, LA = leaf area, LBN = lateral branch number, MPF = multiple pistillate flowers, NN = node number, SCS = seed cavity size, SDS = seed size. SGY = sub-gynoecious, PAR = Parthenocarpy, and VL = vine length.

Full size image

Six moderate-effect (PVE ~10%) and one (hl6.2) large-effect (PVE = 22.6%)32,34 QTL, were identified for HL in two RIL populations, but none are co-localized with sh1. QTL mapping on cotyledon/leaf length and width was conducted in two RIL populations32,33,34. Eight CA (cotyledon area) and five LA (leaf area) consensus QTL from these studies are listed in Supplementary File 3, of which three LA and CA QTL were co-localized suggesting possible shared mechanisms in regulation of cotyledon and leaf sizes in the two populations.

In the 9110Gt × 9930 RIL population, 7 QTL for plant architecture-related traits were detected including four for IL, one for NN, and two for VL. Given the role of node number and internode length on vine length, the 3 major-effect QTL (il1.1, nn1.1, and vl1.1) are co-localized on Chr1 (Fig. 3). The number of lateral branches (LBN) varies significantly in different cucumbers. The littleleaf (ll) mutant H19 also has multiple lateral branches, which is likely due to the pleiotropic effect at the ll locus16. In the S94 × S06 RIL population, there were 6 QTL underlying LBN variation including two major-effect QTL (lbn1.2, and lbn6.2), but none is located nearby the ll locus indicating multiple mechanisms regulating branch numbers.

The observed clustering of these size- or length-related QTL for vegetative organs (Fig. 3; Supplementary File 4) on four chromosomal locations on Chr1, Chr5, and Chr6, suggest common genetic basis for these traits.

Genes and QTL for reproductive development

Simply inherited genes for sex determination

A cucumber plant can bear male, female, or bisexual flowers, and their combinations result in five major sex morphs: monoecious (male and female flowers), andromonoecious (male and perfect flowers), gynoecious (female only), androecious (male only) and hermaphroditic (bisexual flowers only). In cucumber, sex determination depends primarily on the F (femaleness), m (andromonoecy), and a (androecy) loci, all of which are members of the aminocyclopropane-1-carboxylic acid synthase (ACS) gene families (CsACS1 for F; CsACS2 for M, and CsACS11 for A) catalyzing the rate-limiting step in ethylene biosynthesis35,36,37. The F locus is consisted of two copies of ACS1 (CsACS1 and CsACS1G). Additional genes or modifiers affecting sex expression also exist. For example, mutations in CsACO2 (a-1) for the 1-aminocyclopropane-1-carboxylate oxidase result in androecy38. A major-effect QTL, Sgy3.1, controls F locus-independent high percentage of female flowers on monoecious plants39,40(also see below). These diverse sex-determination genes provide opportunities to fine tune sex expression for cucumber production.

QTL for reproductive development-related traits

Flower- and fruit set-related QTL

Flowering time (FT) and sex expression are directly related to fruit timing and yield, respectively. The wild cucumber (C.s. var. hardwickii), semi-wild XIS cucumber, and some landraces from India and Pakistan require short-day length for flower induction. For example, it takes six or more months for the XIS cucumber accession WI7167 to flower under long-day conditions41, while most modern varieties will flower in 30–50 d after planting. In two studies, four QTL (ft1.1, ft5.1, ft6.1, and ft6.2) were found to control flowering time variation in populations derived from two XIS cucumber accessions (SWCC8 and WI7167)41,42 (Supplementary File 5 or Fig. S2). Two other studies used populations derived from crosses between cultivated cucumber lines with <5d FT difference. In each case, a single major-effect FT QTL (da1.1 and Ef1.1) was detected32,43; both are very close to ft1.1. It was suggested that ft6.2 in WI7167 is a major-effect QTL regulating day-length sensitive flowering while ft1.1 regulates flowering time within cultivated cucumbers41.

Early fruit yield is influenced by flowering time and position of the first flower node (FFN). Nine FFN QTL were identified in two monoecious (ff) × gynoecious (FF) RIL populations30,32,44. Among three major-effect FFN QTL, ffn6.2 was located near the F locus as expected, while QTL ffn1.2 and ffn3.2 also have major effects (Fig. 3). These studies revealed the complicity of genetic control of the FFN trait, which is obviously the results of the interplay among factors affecting both flowering time and sex expression.

A gynoecious plant carrying the homozygous FF gene has one or more female flowers on each node, which may not be ideal in some production systems with less optimal cultural practices or poor production conditions since not all female flowers will develop into marketable fruit. The term “sub-gynoecious” (SubG) type sex expression was used to describe the plant that starts with male flowers in the first 5–10 nodes and then has continuous female flowers on the main stem with an overall percentage of female flowers (PFF) of >80% (ref. 39). In a segregating population derived from the cross between S-2-98 (SubG) and M95 (M), 4 QTL, Sgy3.1, Sgy4.1, Sgy6.1, and Sgy6.1, were found to regulate PFF with Sgy3.1 having the strongest effect (PVE = 54.6%)39. In another study, Win et al. confirmed the major-effect QTL Sgy3.1, and identified two additional QTL, Sgy1.1 and Sgy1.2, which are able to increase, and decrease PFF, respectively40 (Supplementary File 3). A gene for the GA20-oxidase was proposed to be the candidate gene for the dominantly inherited Sgy3.1 locus40. Phenotypically, an F gene-independent SubG plant is similar to the one that is heterozygous at the F locus (Ff), which usually starts with male flowers in the first few nodes (1–10) followed by continuous female flowers. When QTL mapping for PFF was conducted using populations derived from gynoecious (FF) × monoecious (ff) crosses44,45, as expected, the major-effect QTL for PFF was consistent with the F locus (Fig. 3). Minor-effect PFF QTL were detected in these studies, which seem to co-localize with SubG QTL Sgy3.1 and Sgy6.139,40 (Supplementary File 5). These observations suggest the PFF is influenced by multiple genetic factors although the F and Sgy3.1 loci play the major roles in gynoecious and SubG plants, respectively.

Some gynoecious cucumber lines may bear multiple pistillate flowers (MPFs) at each node. Five MPF QTL have been identified with each having similar effect (PVE ~10%)46. Parthenocarpic fruit set (PAR) is critical for cucumber production in protected environments. Lietzow et al. and Wu et al. detected 12 PAR QTL in two sources, but only two (par2.1 and par7.1) are co-localized between the two studies47,48. The inconsistent results reflect the difficulties in accurate phenotyping for PAR, which is difficult to separate from yield.

Many of the FT- and sex-expression-related traits are correlated and may be regulated by common, hormone-related pathways, which can be evidenced from QTL clusters for different traits on chromosomes 1, 3, 5, 6, and 7 (Fig. 3; Supplementary File 5).

Genes and QTL for fruit–related traits

Genes for simply inherited fruit-related traits

Fruit skin and flesh color

Cucumber fruit exhibits a wide spectrum of skin colors that can vary from light green, yellow green, green, dark green, to creamy, white, yellow, brown, orange, or red (Fig. 1). The white skin color (w) is due to a mutation in the CsAPRR2 gene, which plays an important role in fruit pigment accumulation49. Mutations in the lgp (light green peel, CsARC5) and lgf (light green fruit, CsYcf54) genes cause change of dark green fruit color to light green50,51. The orange/red mature fruit color locus R is allelic to the black spine gene B, which encodes a R2R3 MYB transcription factor52,53.

Most cucumber fruits have white flesh. The semi-wild XIS cucumber has orange flesh (or) and accumulates high-level β-carotene at mature fruit stage. This is due to a mutation in CsBCH for β-carotene hydroxylase33. The yellow flesh (yf) locus from PI 200815 was fine mapped into a 150-kb region on Chr7 (ref. 54). The green flesh (gf) in immature cucumber, results of accumulation of chlorophyll, is controlled by two loci55.

Fruit epidermal features

The external appearance of cucumber fruit is important for consumer acceptance or processing. Several simply inherited genes determine fruit epidermal features, some of which are tightly linked on Chr5 (Fig. 2) including Heavy/no netting (H/h), Warty/smooth fruit (Tu/tu), Dull/glossy fruit skin (D/d), Ribbed/non-ribbing fruit (Fr/fr), Mottled/uniform immature fruit color (U/u), Large/small spines (SS/ss), and Tough/tender fruit (Te/te). Interestingly, specific allele combinations of these genes are characteristic of different market classes. For example, the European Long, Chinese Long, and US pickling cucumbers often have u-H-tu-ss-te-fr-d, u-h-Tu-ss-te-Fr-d, and U-h-Tu-SS-Te-fr-D haplotypes, respectively. This is likely the result of diversifying selection during breeding for different market classes.

The number of spines on the fruit vary widely in cucumbers of different market classes. The few spine1 (fs1) mutation identified from a dense-spined Chinese Long line is due to an 812-bp deletion in the promoter region of CsGL3 (ref. 56); but higher density spines in Chinese Long cucumber seem to require both CsGL3 and the QTL fsd6.1 (ref. 57). Some cucumbers have numerous (ns) but small spines (ss) with the ns being a homolog for the gene encoding an auxin transporter-like protein 3 (CsLAX3)58,59. Fruit spines usually are hard and prickly and may cause an itching response on the skin. A tender spine (tsp) mutant does not trigger itching, which seems due to an N-terminal deletion in Tsp for a C-type lectin receptor-like tyrosine-protein kinase3. A non-synonymous mutation within the same gene was proposed to confer glabrous trait (csgl4) with smaller and fewer trichomes2. Fruit spines often sit on a bulge structure of several layers of cells called tubercles (warts). The Tu (tuberculated) locus controls wart development, and Ts1 regulates tubercle size, which encodes a C2H2 zinc finger domain-containing transcription factor (CsTu), and an oleosin (CsTs1), respectively; CsTu can bind directly to the promoter of CsTs1 to promote its expression60,61.

QTL for fruit size/shape, external and internal fruit quality traits

Fruit size and shape

Cucumber exhibits diverse fruit size (FS) and fruit shape. Fruit shape is defined using fruit shape index (FSI) which is the ratio of fruit length (FL) to fruit diameter (FD). In some cases, simply inherited genes have been found to play important roles in fruit size control. For example, the fruitful1 (CsFUL1) gene is a key player in fruit elongation in Chinese Long cucumber62. Of two short fruit mutants (sf-1 and sf-2) recently identified4,63, sf-2 encodes a cucurbit-specific RING-type E3 ligase, which results in its enhanced self-ubiquitination and degradation, as well as increased expression CsACS2 (m locus). This may also explain the elongated fruit due to an allelic mutation of the m locus (m-1) on an andromonoecious plant (m-1m-1); an andromonoecious cucumber plant (mm) usually sets round fruit64. Fruit size variation in cucumber is also influenced by fruit carpel number (CN). CN variation (3 vs 5) is controlled by the Cn gene that is a homolog of CLAVTATA3 (CsCLV3)65. Cucumber fruit shape can be round, oval, oblong, long or very long. A spontaneous mutant bears mango-shaped fruit (mango fruit, mf) which is due to a SNP in the WUSCHEL-related homeobox1 (CsWOX1) gene66.

In most cases, fruit size and shape are controlled by QTL. A number of QTL mapping studies on fruit size/shape have been conducted in cucumber. Pan et al. reviewed the genetic architecture of fruit size variation in cucumber, and identified 19 consensus fruit size (FS) and 11 fruit shape (FSI) QTL67. Among them, the consensus FS QTL FS1.2 and FS2.1 are the homologs of tomato SUN (CsSUN2) and SlTRM5 (TONNEAU1 Recruiting Motif) (CsTRM4), respectively67,68,69. Details of these consensus FS QTL are presented in Supplementary Files 3 and 6. In addition, fruit weight (FW) is apparently correlated with fruit size, which is also an important component for fruit yield. QTL mapping have identified 19 FW QTL in three studies42,44,70(Supplemental Files 3 and 6). Almost all FW QTL are co-localized with consensus FS QTL indicating a close correlation between them.

Two other traits often correlated with fruit length are fruit neck length (FNL) and fruit peduncle length (FPL). Fruit neck is the stem-end of the fruit with reduced fruit expansion, which usually does not have spines. Long fruit neck is an undesirable trait because it gives non-uniform external appearance and often has a bitter taste due to accumulation of cucurbitacins. FNL is strongly associated with fruit length. In the only QTL mapping study for FNL44, all five QTL were co-localized with the FS consensus QTL (Supplementary File 6). Fruit peduncle connects the stem and the fruit. There is significant variation in FPL among different cucumber market classes. Seven FPL QTL were identified in two studies44,71; all of which are co-localized with FS consensus QTL.

A fruit with small seed cavity and thick flesh is preferred for both processing and fresh market uses. Structurally, fruit seed cavity size (SCS) and fruit flesh thickness (FTH) are two traits to describe the endocarp and mesocarp of the cucumber pepo fruit, respectively. Eight and six consensus QTL have been identified for SCS and FTH, respectively44,72.

As discussed earlier, most fruit epidermal feature genes are simply inherited (Tables 1 and 2), but some show quantitative variation. For example, Tian et al. found that fruit skin wax (glaucosity) (FSW) accumulation is controlled by five QTL, with fsw5.1, and fsw6.1 having moderate effects73 (Supplementary File 3). Shimomura et al. and Miao et al. examined fruit wart size (FWS) and density (FWD) and identified 3 and 2 QTL, respectively74,75. In both cases, the major-effect QTL is consistent with the Tu locus (Table 1). Fruit spine density on cucumber fruit may vary from very few large spines, many small spines (ss), to high-density spines or ultra-high-density hairs (or numerous spines, ns). The ns and ss single genes have been cloned or fine mapped (Tables 1 and 2). Bo et al. examined spine density in bi-parental and natural populations, and identified three QTL: fsd6.2, fsd6.1, and fsd4.1 that control high and ultra-high spine densities, which had major-, moderate, and minor effects, respectively57. The fsd6.2 locus, which is a variant of the CsGL3 gene (Table 1) regulates high spine density, but for ultra-high spine density, both fsd6.1, and fsd6.2 are required.

QTL for seed-related traits

Cucumber seed did not seem to be a target of selection during long-term cultivation. Cucumber seeds are white or gray in color, but seed size does show significant variation especially between the wild and cultivated cucumbers. The wild cucumber accession PI 183967 has very small seeds. In two studied, Wang et al. and Lietzow conducted QTL analysis for seed length, width, and weight76,77. Most QTL for the three traits are co-localized, and the seven consensus QTL for seed size (SDS) are summarized in Supplementary File 3. Seed size did not seem to have any obvious correlation with other size or length-related traits (Supplementary File 6).

Genes and QTL for disease resistances and abiotic stress tolerances

Genes for simply inherited disease resistances

Major cucumber diseases of worldwide importance include downy mildew (DM), powdery mildew (PM), angular leaf spot (ALS), target leaf spot (TLS), anthracnose (AR), Fusarium wilt (FOC), scab, and various viruses like cucumber mosaic virus (CMV), watermelon mosaic virus (WMV), zucchini yellow mosaic virus (ZYMV), and papaya ringspot virus (PRSV). The cucumber accession PI 197087 from India and its derivatives like Gy14 are resistant to DM, ALS and AR that is conferred by dm1, psl, and cla, respectively. It was found that the cucumber Staygreen (CsSGR) is the causal gene underlying the dm/psl/cla locus (Chr5 in Fig. 2); thus, the durable resistance against the three different pathogens (bacterial, oomycete, and fungal) in Gy14 is due to a loss-of-susceptibility mutation in CsSGR, which encodes the Mg dechelatase that plays critical regulatory roles in the chlorophyll degradation pathway78,79. The dm1-conferred DM resistance was less effective since 2004 when new DM pathogen strains emerged in the cucumber field in the US. Two major-effect QTL for resistance against the post-2004 DM strain(s) (dm4.1 and dm5.2) were identified from PI 197088 and PI 330628 (ref. 80,81).

Another well characterized loss-of-susceptibility R gene in cucumber is the mlo locus for PM resistance82,83,84. Multiple allelic variants at this locus have been identified in PM resistant accessions; all result in the loss of function of CsMLO. Additional PM resistance genes near the mlo locus are also possible85 (Table 2, Fig. 2). The Chinese Long line, Jin5-508, carries a dominantly inherited PM resistance gene Pm1.1 which has been mapped in a 41.1-kb region containing two cysteine-rich receptor-like protein kinase genes86.

Three recessively inherited TLS resistance genes, cca-1, cca-2, cca-3, have bene mapped on Chr6 (Fig. 2)87,88. Among them, cca-3 seems to belong to the CC-NB-ARC type R gene family88 which has ~73 homologs in the cucumber genome. In addition, the closely linked ccu for scab resistance and Foc for Fusarium wilt resistance were mapped to a region on Chr2 containing a cluster of several NB-LRR R gene homologs89,90.

The candidate gene for the zym locus (CsVPS4) for ZYMV resistance encodes the vacuolar protein sorting-associated protein 4 (VPS4)-like protein91. Several variants of the zym locus have been identified in different ZYMV resistance sources92. Three virus resistance genes have been mapped on Chr6 including prsv for PRSV, wmv for WMV and cmv6.1 for CMV93,94,95. Previous studies indicated tight linkage of resistances to three potyviruses (PRSV, ZYMV and WMV) in cucumber. Molecular mapping results seem to suggest that they are different loci (Fig. 2).

QTL for disease resistances and abiotic stress tolerance

QTL studies have been carried out for resistances to the following diseases: PM, DM. FOC, Gummy stem blight (GSB), Melon Yellow Spot Virus (MYSV), and the Cucurbit Yellow Stunting Disorder Virus (CYSDV). The results are summarized in Table 4, and their chromosomal locations are illustrated in Fig. 4. More details for each QTL are presented in Supplementary File 2 (Table S2).

Table 4 Summary of disease resistance QTL identified in cucumber.
Full size table
Fig. 4: Chromosomal locations of disease resistance genes and QTL in cucumber.
figure 4

Ruler to the left indicates locations (in Mbp) in the Gy14 V2.0 draft genome assembly (drawn to scale). Vertical black lines are chromosomes. Cloned genes and QTL are aligned to the left and right of each chromosome, respectively. Vertical bar for each QTL represents 1.5 or 2.0 LOD confidence interval on the chromosome. Regions delimited by blue dashed rectangles indicate resistance gene/QTL hot spots or clusters.

Full size image

QTL mapping for PM resistance (PMR) has been conducted from six resistance sources including PI 197088 (ref. 81,96,97), S06 (ref. 98), K8 and H136 (ref. 99), WI2757 (ref. 100), and IL52 (ref. 85). Diverse mapping populations, phenotyping and genotyping methods were used in these studies with varying power of QTL detection. However, based on chromosomal locations of these QTL, 19 consensus PMR QTL could be inferred (Supplementary File 2). The co-localization of QTL from different resistance sources may suggest that they belong to the same locus, or are closely linked. For example, pm5.3 was detected in PI 197088, IL52, WI 2757 and K8; both pm5.1 and pm6.3 were detected in S06, K8 and PI 197088. The pm5.3 locus (syn. pm5.1, pm-h) encodes a barley MLO homolog (CsMLO1), and multiple variants at this locus are responsible for PMR in different lines82,83,84. The pm/dm5.3 QTL has been shown to confer complete PM resistance in IL52, and the gene for a GATA transcriptional factor was proposed to be its candidate85.

QTL mapping for DM resistance (DMR) has been conducted in PI 197085, PI 197088, WI 7120 (PI 330628), WI 2757, S94, TH118FLM, IL52, and K8 (Table 4). Sixteen QTL were identified in PI 197088, and four of them are major-effect QTL contributing to DMR (dm4.1, dm5.1, dm5.2, and dm5.3)81,101,102. PI 330628 carries five DMR contributing QTL with dm4.1 and dm5.2 having the largest effect80. WI 2757 exhibits moderate resistance to post-2004 field DM strains and carries both dm1 from PI 197087 and dm5.2 with unknown origin81. Among the 17 consensus DMR QTL, 11 could be detected in at least two resistance sources (Table 4; Supplementary File 2). Interestingly, the two major-effect QTL, dm5.1 and dm5.2 were detected in five resistance sources, whereas dm1.1 and dm6.4 were each identified in four lines. These observations suggest that cucumbers from different origins may share some comment genetic basis for DMR although the magnitude of these QTL are affected by genetic backgrounds and environmental conditions.

QTL mapping studies for resistances to other pathogens are sporadic. Fusarium wilt is a soil-borne disease, which is more serious in cucumber production under protected environments. A major-effect QTL for Fusarium wilt resistance, Foc2.1 was identified, which is closely linked with the scab resistance (ccu) locus in a region with multiple members of NB-LRR resistance gene homologs89,90,103. The wild cucumber line PI 183967 is highly resistant to GSB. The adult plant and seedling GSB resistances were controlled by four and five QTL, respectively104,105, but only one minor-effect QTL (gsb6.2) is shared between the two stages. Two minor-effect GSB resistance QTL (gsb4.1, and gsb6.2) were also detected in a C. hystrix introgression line106. For virus resistances, four QTL for the resistance to isolate MYSV-FuCu05P-2 have been identified107. A major-effect QTL for CYSDV resistance (cysdv5.1) was mapped to a region close to the mlo locus for PMR108.

In cucumber breeding, it has long been observed that there is a positive correlation between resistances to different pathogens such as DMR and PMR, resistance to Fusarium wilt and scab, and resistance to different potyviruses (e.g., PRSV, and ZYMV). Indeed, several lines used in the above-mentioned QTL mapping studies possess dual resistances to PM and DM (for example, PI 197088, K8, IL52, and WI2757). The chromosomal locations of consensus resistance QTL to different pathogens are illustrated in Fig. 3. Clearly, many disease resistance QTL are co-localized, which is especially true for PM and DM. Also, it seems there are several hot spots on chromosomes 5 and 6 where resistance loci to different pathogens are highly enriched (Fig. 3). This offers potential advantages in disease resistance breeding for cucumber. However, at the molecular level, whether these resistance genes or QTL belong to the same locus, or are closely linked await further investigation.

Cucumber is of tropical origin and is sensitive to low temperature. In temperate growing regions or production areas at a high altitude, low temperature germination (LTG) ability is a trait that may allow for early planting. In two studies109,110, four LTG QTL were identified: LTG1.1, LTG1.2, LTG2.1, and LTG4.1. The two major-effect contributing QTL, LTG1.1 and LTG1.2, are 2-Mbp apart on Chr1 (Supplementary File 3). Waterlogging is a serious environmental stress in many cucumber production regions. One strategy for cucumber plants to deal with the waterlogging stress is the production of hypocotyl-derived adventitious roots (AR). In the waterlogging resistant line Zaoer-N, three QTL contribute to increasing AR numbers under waterlogging111. The gene for an AAA-ATPase domain-containing protein has been shown to be a candidate for the major-effect QTL for adventitious root numbers, ARN6.1112.

Genes for MISC horticulturally important traits

The bitter tasting cucurbitacins are tetrocylic terpenes present widely in cucurbit crops. Three bitterness related genes have been cloned including Bi (bitterfree), Bl (bitter leaf), and Bt (bitter fruit)113. Bi encodes a cucurbitadienol synthase that catalyzes the cyclization of 2,3-oxidosqualene into the tetracyclic cucurbitane skeleton, the first committed step of cucurbitacin biosynthesis. Both Bl and Bt encode two basic helix-loop-helix (bHLH) transcription factors that are expressed specifically in leaves and fruits, respectively. Bl binds to the E-box elements of the Bi promoter to activate its transcription for cucurbitacin biosynthesis in cucumber leaves; Bt has similar biochemical function as Bl but regulates cucurbitacin biosynthesis in the fruit113. Abiotic stress influences cucurbitacin biosynthesis by modulating the expression of Bl and/or Bt113.

Cucumber foliage or fruit are usually non-fragrant, but some varieties from Thailand have pandan-like fragrance from leaves and fruit, which is controlled by the fgr (fragrance) locus (CsBADH) encoding the betaine aldehyde dehydrogenase114.

Concluding remarks

New genomic technologies and resources for cucumber have allowed for a surge in research leading to QTL mapping and identification of candidate genes associated with a wide array of phenotypic traits. In this work we documented 81 simply inherited genes or major-effect QTL and 322 QTL for 42 quantitative traits, providing chromosome locations, allelic variants and associated polymorphisms, predicted functions where appropriate, and diagnostic markers that could be used for marker-assisted selection in cucumber breeding. Despite the increased effort in cucumber, the number of cloned genes and narrowly defined QTL is still quite limited, and in most cases the proposed functions have not been verified. Looking to the future, it is anticipated that studies in cucumber will be able to draw on an increasing number of genomic tools, both to identify and verify important genes. Cucumber collections in major gene banks are rich in genetic variation that could be explored to identify novel genes or alleles. Genome-wide association analysis may play an important role to accomplish this. EMS mutagenesis is also a powerful tool to generate novel mutations and development of efficient genetic transformation and gene editing systems will allow characterization of gene functions.

It is hoped that the present work will serve as starting point for the systematic inventory of cucumber genes, quantitative trait loci, genetic stocks, and mutants, to benefit the cucurbit community in the years to come. As the information about cucumber genes continues to grow, it has also become imperative for the community to adopt a standard nomenclature to describe QTL. Standardized nomenclature, as has been adopted for numerous other species, facilitates continued progress and minimizes confusion when comparing results across publications. We hope the vocabularies for quantitative traits and the QTL naming rules we recommended here will help achieve this goal.