Introduction

Cucumbers (Cucumis sativas L.), the Cucurbitaceae, are fruits and vegetables with a total annual production of 75 million tons (FAO, 2014). Total production of domestic cucumber in Korea is 271,040tons per year (KOSIS, 2015), of which 28.4 tons is exported to Japan (UNI-PASS, 2015). For marketability in Japan, cucumber fruit without bloom (bloomless cucumber) is inevitable because it displays a clean and shiny appearance on the surface of the fruit (Choi et al., 2013; Mitani et al., 2011). In major Asian countries including Japan and Korea, oriental pumpkin (C. moschata Duch.) is used as a stock in grafting of cucumber for prevention of Fursarium wilt disease, maintenance of vegetative growth, and improvement of cold tolerance. For producing cucumbers exported to Japan, therefore, a specific type of pumpkin stock (bloomless stock) that prevents the cucumber fruit from blooming must be used (Seo et al., 2004; Choi et al., 2013). The seeds for bloomless stock are imported mainly from Japan, while no domestic stock cultivars are commercially available to date.

For production of winter greenhouse cucumber, a wild type pumpkin cultivar ‘Heukjong’, (C. ficifolia Bouche.) is widely used as a stock due to its nature of cold-tolerance over other non-tolerant pumpkin stocks (Tachibana, 1988; Seo et al., 2004). However, ‘Heukjong’ produces bloom in grafting of the cucumber and is inadequate for cultivating bloomless cucumbers in the winter season. Furthermore, despite the necessity for cold-tolerant bloomless stock cultivars, introgression of the bloomless trait into C. ficifolia from C. moschata is not straightforward due to the infeasibility of cross-fertilization between these two species (Robinson and Decker-Walters, 1997).

The bloom of cucumber fruit is generated when absorbed silicon (SiO2) is exuded with water onto the skin and then dried (Yamamoto et al., 1989). A gene that facilitates the uptake of silicon, Low silicon rice 1 (Lsi1), was first identified in rice (Ma et al., 2006), and its homologous genes were also identified in many other plants including maize and barley (Chiba et al., 2009; Mitani et al., 2009). There are two types of transporters that mediate silicon absorption, a channel-type transporter (or an influx transporter) and an efflux transporter (Ma and Yamaji, 2015). In rice, the Lsi1 gene encodes a channel-type transporter that facilitates passive transport of silicon across the plasma membrane between apoplast and the plant cell (Yamaji et al., 2008), while an efflux transport is mediated by the Lsi2 gene that belongs to an uncharacterized anion transporter family (Ma and Yamaji, 2015). In response to silicon, expression of Lsi1 was down-regulated in rice and cucumber (Ma et al., 2006; Sun et al 2017), but was unaltered for its homologous genes in maize and barley (Chiba et al., 2009; Mitani et al., 2009). Recently, it was reported that pumpkin homologs, CmLsi1 and CmLsi2, are associated with the transport of silicon (Mitani et al., 2011; Mitani- Uneo et al., 2011). Furthermore, a bloomless pumpkin cultivar ‘Super-unryu’ possesses a missense mutation (P>L) at position 242 amino acid that is linked to the transport activity of silicon (Mitani et al., 2011).

DNA markers have been employed to identify DNA polymorphisms and analyze genotypes and genetic linkage between genes and traits (Lee and Chung, 2011; Kwon and Choi, 2013). In modern breeding strategies, Marker-assisted selection (MAS) is the most popular application of the DNA marker. MAS shortens the time required for breeding, is not influenced by environmental factors, and can greatly increase reliability and precision in comparison to phenotype-based selections (Tanksley, 1983; Collard and Mackill, 2008; Jonah et al. 2011). Although it has been reported for mutational effect of CmLsi1 related to the bloomless trait (Mitani et al., 2011), a DNA marker that can be used to analyze allelic or mutational variations of the CmLsi1 gene is not publicly available.

Here, we developed a molecular marker for selection of a bloomless trait based on nucleotide polymorphisms of CmLsi1. We cloned the homologous gene of Lsi1 in C. ficifolia to establish the molecular basis for breeding of cold-tolerant bloomless stock cultivars of C. ficifolia..

Materials and Methods

Plant Materials

Gene-Based DNA Marker

Extraction of DNA: Genomic DNA was extracted from young leaves of the two-true leaf stage of the plant. Collected leaf tissues were placed into a 1.5 mL micro-centrifuge tube with beads and 600 μL of DNA extraction buffer and subjected to Tissuelyser (TissueLyser II, QIAGEN, Venlo, Netherlands) for homogenization. Samples were lysed in a 65°C waterbath for 45 min and further incubated by adding 200 μL of 7.5 M ammonium acetate on ice for 15 - 20 min. Incubated samples were separated by centrifugation for 10 min at 14,240 x g. The supernatant of centrifuged samples was placed in a 1.5 mL tube containing 5 μL (5 mg ∙ mL^-1) of glycogen solution and 600 μL of isopropanol in a 1.5 mL tube and centrifuged for 10 min at 14,240 x g. After decanting the supernatant, remaining pellet was washed using 300 μL of 70% EtOH. The washed pellet was resuspended in 200 μL of 0.1 M Tris. Concentration of DNA was measured using a Nanodrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA) and 20 ng ∙ μL^-1 was used for PCR reactions.

Development of CAPS Marker: Coding DNA Sequences (CDS) of wild-type allele CmLsi1(B⁺) (AB551949) and mutanttype allele CmLsi1 (B^-) (AB551950) of the C. moschata Lsi1 were obtained from NCBI and aligned for comparison using CLUSTALW (www.genome.jp/tools/clustalw) and confirmed for the SNP of the mutant allele (Fig. 1). Identified SNPs were searched for restriction sites (NEBcutter V 2.0, NEB^®, Ipswich, MA, USA) to generate a Cleaved Amplified Polymorphic Sequence (CAPS) marker using Primer premier 5 (PREMIER Biosoft, Palo Alto, CA, USA) (Fig. 1). PCR conditions for genotyping consisted of 1 μL of template DNA (20 ng ∙ μL^-1), 0.5 μL of forward and reverse primer (10 pmol), 0.1 μL of Taq polymerase (5 U ∙ μL^-1, eTaq Solg ™, SolGent, Daejeon, Korea), 0.2 μL of dNTPs (Solg ™, SolGent, Daejeon, Korea), 1 μL of 10X buffer (Solg ™, SolGent, Daejeon, Korea) and 6.7 μL of ddH2O in a total volume of 10 μL. Reactions were incubated at 95°C for 2 min, and cycled 35 times as follows: denaturation at 94°C for 15 sec, annealing at 58°C for 30 sec, extension at 72°C for 1 min using a PCR cycler (T100 ™, BIO-RAD, Hercules, CA, USA). After the last cycle, reactions were incubated at 72°C for 3 min. Restriction digestion of PCR products was performed by adding 0.3 μL of Hae III (10,000 U ∙ mL^-1, Time-Saver ™, NEB^®, Ipswich, USA), 1.5 μL of 10X buffer (CutSmart ™, NEB^®, Ipswich, Massachusetts, USA), 3.2 μL of ddH₂O to the PCR product and incubated at 37°C for 1 h. Digested products were run on a 2% agarose gel at 70V for 40 min and stained with EtBr, and checked for the results using Gel Image Analysis System (CoreBio i-MAXTM, Davinch-K, Seoul, Korea).

Analysis of CmLsi1 Gene Expression

RNA Extraction and cDNA Synthesis: Young leaves and roots from at least three plants per cultivar were collected and immediately frozen in liquid nitrogen. Total RNA was extracted using a Seed/Fruit Kit (Ribospin™, GeneAll^®, Seoul, Korea) according to the manufacturer’s instructions. Quantitated RNA using Nanodrop 1000 (Thermo Scientific, Waltham, MA, USA) was diluted to 10 ng ∙ μL^-1 and used for cDNA synthesis with an RT Premixture kit (HyperScript ™, GeneAll^®, Seoul, Korea) following the manufacturer’s instructions.

Quantitative PCR (qPCR): Gene-specific primers (Q- F and R) for CmLsi1 (B⁺) (AB551949) were designed using Primer premier 5 (Table 1). Quantitative real-time RT-PCR (qPCR) was performed in a Real-time PCR cycler (LightCycler^® 480, Roche, Basel, Switzerland) for ‘Heukjong’,‘Odaetojwa’,‘Ohmai Stock’, and ‘Ohmai Summer Stock’. Reactions were performed by adding 1 μL of cDNA (10 ng ∙ μL^-1), 0.5 μL of each forward and reverse primer (10 pmol), 5 μL of 2X SYBR Green I Master (Roche, Basel, Switzerland), and 3 μL of ddH2O in a total volume of 10 μL. CmACTIN was used as a reference gene as described by Obrero et al. (2011) (Table 1). All reactions were incubated at 95°C for 5 min, and cycled for 45 times as follows: 95°C for 10 sec, 55°C for 20 sec, 72°C for 1 min. Relative quantitation of the Lsi1 transcript to CmACTIN gene was performed using the 2^-ΔΔCT method (Livak and Schmittgen, 2011).

Table 1. PCR primers used in this study.

Cloning of Lsi1 Homolog in C. ficifolia

5’ and 3’ RACE: The full-length cDNA sequence of CmLsi1 homologous gene from C. ficifolia was obtained using RACE 5'/3' Kit (SMARTer®, Takara, Kusatsu, Japan). Total RNA was extracted as outlined above and RACE reactions were performed according to the manufacturer’s instructions. Gene-specific primers were designed from CmLsi1 (B⁺) CDS (AB551949) with primer lengths of 23 - 28 bp, GC contents of 50 - 70%, Tm>70°C and 15 bp 5'- end primer overlap sequence (GATTACGCCAAGCTT).

Cloning and Sequencing of CfLsi1: RACE reactions were checked on an agarose gel and purified the product using a gel extraction kit (ExpinTM, GeneAll®, Seoul, Korea). Purified RACE product was cloned into T-Easy Vector System I (pGEM, Progma, Madison, WI, USA) and verified the sequence by the dye terminator method (Genotech, Daejeon, Korea).

Sequence Alignment and Phylogenetic Analysis

The nucleotide sequence of CfLsi1 was used as a query sequence to conduct BLAST searches to identify homologs of the CfLsi1 from NCBI (www.ncbi.nlm.nih.gov). To determine the relationship between searched sequences, a phylogenetic tree (dendrogram) of homologs was generated with MEGA 5 (Tamura et al., 2011) using the Neighbor-Joining (NJ) algorithm and Tamura-Nei parameter model with 1,000 bootstrap replicates.

Measurement of Silicon Uptake

Seeds of eight commercial cultivars (see above) were sown in Baroker potting soil (Seoul Bio, Eumseong, Korea) in a 50 - holed tray. After 25 days of sowing, seedlings were transplanted in a plastic pot (15 cm × 13 cm × 19 cm) and grown for 30 days in a greenhouse at Pusan National University (Miryang, Korea). While transplanting, the soil collected from a paddy field of the Agriculture Research Station at Pusan National University was air- dried and sifted, and 2.4 kg of the soil was filled in the plastic pot. Three plants of eight cultivars were grown in a completely randomized design with three biological replicates. Bulk density of soil was 1.37 g ∙ cm^-1, and 576 mL of distilled water was added so that the water content in soil was 60% of the pore volume with a particle density of 2.65 g ∙ cm-1. After transplanting, the silicate fertilizer (Nonepong Eco, Nousbo, Suwon, Korea) was dissolved in 100 mL of tap water and added at the rate of 0 (untreated control), 100 and 200 mg Si ∙ kg^-1 in each pot. Throughout the whole growth, nitrogen (12.8 mg ∙ kg^-1), phosphoric acid (8.75 mg ∙ kg^-1) and potassium (12.7 mg ∙ kg^-1) and compost (1 g ∙ kg^-1) were supplemented with the same amount, and water was added to compensate the decreased amount by weighing the pot once every four days. After 30 days of transplanting, whole plants were collected from soil and dried in a dry oven at 70°C for 72 h and then pulverized. One gram of the pulverized sample was lysed in the lysis buffer (H₂SO₄:HClO:H₂O = 5: 9:1, Samchun, Pohang, Korea) and then the content of silicon was analyzed with ICP-AES (GBC model X-100, GBC, Melbourne, Australia) .

Results and Discussion

Development of CmLsi1 Marker

Coding DNA Sequences (CDS) of wild-type allele (B⁺) and mutant- type allele (B^-) of the CmLsi1 were used for alignment to confirm the non-synonymous SNP (C>T) missense mutation (P>L). To design a CAPS marker for the identified SNPs, restriction enzyme sites within the sequence were searched and recognition sites of Hae III was selected for the study. A PCR primer set CM1-CAPS (Table 1) was designed spanning the Hae III restriction site to cut the amplified product (201 bp) into 140 bp and 61 bp in the mutant-type allele (Fig. 1). Genotyping analysis of the eight stock cultivars using the CM1-CAPS marker revealed that bloom and bloomless traits were correlated with each genotype (Fig. 2). For a bloom F1 cultivar ‘Arirang’, marker analysis indicated that it is heterozygous genotype for CmLsi1 (Yamaji and Ma, 2007), which suggests that one of the parental lines of ‘Arirang’ was a bloomless cultivar homozygous for the mutant allele of CmLsi1. From this, we confirmed use of the CM1- CAPS marker for MAS-assisted breeding of bloomless trait.

Transcripts of CmLsi1 in Bloom and Bloomless Stock Cultivars

Verification of the pKHi Vector System by Silencing DFR

Expression levels of endogenous CmLsi1 gene were examined in leaf and root tissues from bloom (‘Heukjong’ and ‘Odaetojwa’) and bloomless (‘Ohmai Stock’ and ‘Ohmai Summer Stock’) cultivars using Quantitative real time RT- PCR (qRTPCR) (Fig. 3). From the tissues tested for representative cultivars, relative expression of CmLsi1 was higher in root than leaf of these cultivars (Fig. 3). The expression pattern of Lsi1 was consistent with the result from rice (Oryza sativa) and cucumber (Cucumis sativus), which showed higher expression of Lsi1 in root than leaf (Ma et al., 2006; Sun et al., 2017). Among cultivars of pumpkins (C. moschata), it has been shown that expression of Lsi1 gene in a bloomless cultivar ‘Super-unryu’ was higher than a bloom cultivar ‘Sintosa’ (Mitani et al., 2011). Curiously, it was also reported that expression of CmLsi1 in shoot apex was higher than that in root of pumpkin (Mitani et al., 2011). The findings suggest that expression of Lsi1 is species- and tissuespecific, and the relation of Lsi1 genotypes with the bloom / bloomless trait may not involve the transcriptional regulation of Lsi1 genes. However, exact mechanisms underlying these transcriptional regulations and their possible relation with functions in silicon uptake remain to be determined.

Taken together, expression pattern of the Lsi1 gene does not appear to be associated with variations of alleles in pumpkin (this study), and the expression level may differ depending on the species, cultivars and tissues examined. A link between bloomless trait of pumpkin resulting from the missense mutation caused by a non-synonymous SNP in the protein and suppression of Lsi1 expression in a bloomless cultivar needs further examination.

Cloning of CfLsi1

We identified a full-length cDNA of CmLsi1(B⁺) homologous gene in C. ficifolia through 5’ and 3’ RACE and named CfLsi1(B⁺) (Fig. 4). A sequence comparison between CmLsi1(B⁺) (accession AB551949) and CfLsi1(B⁺) revealed that there exists total 24 SNPs, although the length of CDSs was identical. Of these, only three SNPs appeared to substitute (nonsynonymous) amino acids at positions 14 (S>A), 75 (V>A), and 162 (L>F) (Fig. 5). Considering the ability of silicon exerted likely by Lsi1, these missense mutations do not appear to affect the function of the CfLsi1 gene.

It seems plausible to introgress a bloomless allele from C. moschata to ‘Heukjong’(C. ficifolia Bouche.) by hybridization to breed a cold-tolerant bloomless stock cultivar for cucumber. However, given the crossing barriers between these two species, other approaches like gene silencing technologies (RNAi) specifically targeting CfLsi1(B⁺) expression in ‘Heukjong’ cultivar that may alter the fuction of CfLsi1 protein might be useful as an alternative. Nonetheless, we consider that a full-length cDNA sequence of CfLsi1(B⁺) identified in this study would provide valuable molecular platform that can be used in future molecular breeding.

Phylogenetic Analysis of CfLsi1 and Lsi Homologs

Nucleotide sequences that showed high sequence similarity to CfLsi1(B+) of C. ficifolia from different plant species were used to generate a phylogenetic tree (Fig. 6). As expected, the retrieved sequences with high similarity revealed that CfLsi1(B+) is closely related to Nodulin 26-like intrinsic proteins (NIPs) that belong to a larger family of Major intrinsic proteins (MIPs) (Wallace et al., 2006). It has been shown that among channel-type silicon transporters (or influx transporters) all Lsi1 belong to the NIP III group of aquaporin proteins (Ma and Yamaji, 2015).

Sequence comparisons using CfLsi1(B+) showed that CfLsi1 is most similar to CpNIP1 of C. pepo than C. moschata. Also, CfLsi1 was closely related to homologs in cucumber and melon. Cucumber, melon (Cucumis melo), and watermelon (Citrullus vulgaris) are in the same Benincaseae tribe whereas C. ficifolia, C. moschata and C. pepo belong to the Cucurbiteae tribe (Schaefer et al., 2009). It is worth noting that phylogenetic analysis of CfLsi1(B+) and its select homologs (the current study) is also supported by the previous finding that within Cucurbiteae tribe C. ficifolia Bouche. was closer to C. pepo than either to oriental and western pumpkin based on analysis of relationship using SSR markers (Kim et al., 2015).

Measurement of Silicon Content

The accumulation of silicon for eight stock cultivars was measured as described above. Overall, the silicon content measured in all the pumpkin stock cultivars ranged from 0.2% to 0.5% of the total dry mass. The most significant difference between bloom and bloomless cultivars was found to be the ones supplied with 0.1 g∙ kg^-1 and 0.2 g∙ kg^-1of silicon (Table 2). The content was higher in all the treated bloom cultivars (‘Heukjong’, ‘Odaetojwa’, and ‘Arirang’) than most of the bloomless cultivars (‘Ohmai Stock’, ‘Ohmai Summer Stock’, ‘Union’, ‘Nunbusyeo’ and ‘Newtype’). Interestingly, the highest amount of silicon content was found in the bloom cultivars at 0.2 g∙ kg^-1 of silicon treatment particularly in ‘Heukjong’ cultivar (0.5% of total dry mass) (Table 2). Above findings agree with the previous results that the silicon content in shoots of bloom stock in pumpkin was higher than that of bloomless stock cultivar in pumpkin (Mitani et al., 2011), and the higher content of silicon in the cucumber grafted to the bloom stock than that grafted to the bloomless stock (Choi et al., 2013).

It was reported that the shoot of bloom cultivar in pumpkin contained 1% of silicon whereas the shoot of bloomless cultivar had 0.1% (Mitani et al 2011). In our study, however, bloom cultivars of pumpkin stocks contained approximately 0.3 - 0.5% of silicon which was less than expected, which may be attributed to the different experimental settings. In addition, a silicon content in plants may be affected by various factors other than genetic or allelic variations. For instance, silicon could be fixed to soil surface and changed into unavailable forms due to various environments such as pH, redox potential, temperature, and electrical conductivity in soil (Patrick Jr et al., 1987; Szulc et al., 2015; Carmo et al., 2016). Therefore, the difference in growth conditions and organs measured as well as experimental settings may have contributed to the differential contents of silicon. Furthermore, it has not been determined for general amount of silicon uptake in pumpkin, which makes it difficult to compare the amount of silicon uptake in the current study with others.

It is not clear for the observed silicon contents bloomless stock cultivars. It may have been affected by other genetic factors other than Lsi1, such as Lsi6 that might otherwise have affected the uptake. The presence and function of Lsi6 homologs has not been determined in pumpkin. The Lsi6 gene in rice is presumed to play a role in transferring silicon from xylem to leaf, but exact mechanisms of action are still unclear. According to Misson et al. (2004), it was reported that the mutation in Pht1;4-1 gene caused about 40% of reduction in total absorption of Pi under low inorganic phosphate (PI) conditions, but not complete inhibition. Therefore, it is thought that the missense mutation (P < L) of the CmLsi1 reduces the silicon absorption rate rather than complete inhibition in bloomless pumpkin stocks. Although we provide molecular basis for breeding of bloomless stock cultivars, functional relevance of the CmLsi1 gene that can inhibit silicon uptake and its bloomless effect on the cucumber silicon still await further investigation..