Cucumber (C. sativus L. cv. Xintaimici) plants were grown under glasshouse conditions with 28/18°C (day/night) temperature and 12/12 h (light/dark) photoperiod in Beijing, China. Tissue including root, stem, young leaf (sink leaf in apical growing point), mature leaf (functional leaf), flowers, and fruit were collected from 3-month-old plants and used for qRT-PCR analyses. The DH5a Escherichia coli strain was used for cloning and Saccharomyces cerevisiae strains EBY.VW4000 and SUSY7/ura were used for heterologous protein expression.

CsSWEET genes were identified via BLASTN searches in the National Center for Biotechnology Information (NCBI¹) database and the Cucurbit Genomics Database (CuGenDB²). The resulting 17 cucumber putative SWEET genes were named CsSWEET1-17, with the numbering based on their homologous genes in A. thaliana (Chen et al., 2010). The letter corresponds to the location on chromosome and Hu et al. (2017).

Phylogenetic analyses were conducted using ClustalW in Mega 5.0 software and an unrooted phylogenetic tree of the CsSWEET, AtSWEET, and SlSWEET families were constructed using Mega 5.0 software according to Chen et al. (2010). The evolutionary history was inferred using the neighbor joining method with 1,000 replicates. The evolutionary distances were computed using the Poisson correction distance model and are given as the number of amino acid substitutions per site.

Information on the SWEET genes was acquired from NCBI, including locus tags, gene symbols, accession numbers, the chromosomal position and the MtN3 saliva family (PFAM database code PF03083³) analysis. Structure of each CsSWEET gene was analyzed using NCBI. The TMs were analyzed using TMHMM⁴.

The CsSWEET1, CsSWEET2, CsSWEET5a, CsSWEET7b, CsSWEET10, CsSWEET12c, CsSWEET17a, and CsSWEET17c open reading frames (ORFs) were amplified by PCR and cloned into the S. cerevisiae/E. coli shuttle vector pDR196 (Fan and Zhang, 2009). PCR primers listed in Supplementary Table S3.

The recombinant vectors or the empty pDR196 vector (control) were separately transferred into the hexose uptake-deficient yeast strain EBY.VW4000 and sucrose uptake-deficient yeast strain SUSY7/ura, as previously described (Cheng et al., 2015). The transformed cells of the hexose uptake-deficient strain were grown on SD (synthetic deficient) medium supplemented with 2% maltose (glucose for sucrose uptake-deficient strain) and auxotrophic requirements. Serial dilutions (×10, ×10², × 10³) of yeast cell suspensions (CsSWEET1, CsSWEET2, CsSWEET5a, CsSWEET7b, CsSWEET17a, and CsSWEET17c) were dropped on solid SD media containing either 2% maltose (control) or 2% glucose/fructose/galactose/mannose/xylose/arabinose plus respective auxotrophic requirements (Cheng et al., 2015). Serial dilutions (×10, ×10², ×10³) of CsSWEET10, CsSWEET12c were plated on SD media containing either 2% glucose (control) or 2% sucrose plus respective auxotrophic requirements. Growth was documented by pictures after 2–5 days growth at 30°C.

Gene expression analyses were performed using qRT-PCR analysis with the SYBR green detection protocol (TaKaRa, Japan) and the ABI 7500 system (Bio-Rad, United States). Total RNA was extracted from specified tissues (root, stem, young leaf, mature leaf, male flower, female flower, fruit on the day of anthesis and fruit on the ninth day after anthesis) using RNAprep pure Plant Kit (Tiangen, Beijing, China), and then reverse-transcribed using FastQuant RT Kit (with gDNase) (Tiangen, Beijing, China). The cDNA samples were then used as a template for qRT-PCR analysis. Primers are listed in Supplementary Table S3.

The CsSWEET1, CsSWEET7b, CsSWEET12c, and CsSWEET17a ORFs without stop codons were amplified by RT-PCR using gene specific primers and the amplicons were cloned into the pCAMBIA super 1300 vector to generate C-terminal fusions with the GFP reporter. Primers are listed in Supplementary Table S3. Transient expression of the CsSWEET1-GFP, CsSWEET7b-GFP, CsSWEET12c-GFP, and CsSWEET17a-GFP fusion proteins in A. thaliana protoplasts and onion epidermal cells were performed according to Ma et al. (2008) and Fan and Zhang (2009), respectively. The empty vector expressing untargeted GFP was used as a control. GFP fluorescence was visualized using an Olympus Confocal Laser Scanning Microscope (FV1000, Japan).

Seventeen candidate cucumber SWEET genes (CsSWEET) were identified and named according to their homologs in A. thaliana (Chen et al., 2010) (Table 1). Sequences of the CsSWEETs were based on BLAST searches against the NCBI (NCBI⁵) database and Cucurbit Genomics Database (CuGenDB⁶). A phylogenetic tree based on the 17 CsSWEET protein sequences, together with 17 AtSWEET sequences and 29 SlSWEET sequences was constructed using MEGA5.0 software. The cucumber sequences clustered into the four clades that have been reported in the A. thaliana and O. sativa families (Chen et al., 2010) (Figure 1). Clade I contained CsSWEET1-3, besides, three AtSWEETs and nine SlSWEETs; Clade II contained CsSWEET5a-c, 7a-b, besides, five AtSWEETs and five SlSWEETs; Clade III contained CsSWEET9-10, 12a-c, 15, besides, seven AtSWEETs and thirteen SlSWEETs; Clade IV contained CsSWEET17a-c, besides, two AtSWEETs and two SlSWEETs. Sequence identity comparisons among the four CsSWEET protein subclasses were 28–77%, and the average identity was 42%. CsSWEET12b and CsSWEET12c showed the highest homology (77%), CsSWEET5b and CsSWEET5c, CsSWEET7a, and CsSWEET7b had 64 and 62% homology, respectively (Supplementary Table S1).

To analyze the structural characteristics of the CsSWEET genes, we examined their conserved regions, and aligned the putative protein sequences. Their exon numbers varied among the genes, CsSWEET7a, 7b, CsSWEET10, and CsSWEET12a containing five exons, and the other gene family members all containing six exons (Table 1).

SWEET proteins have seven TMs in eukaryotes, as well as two MtN3/saliva domains. They contain a basic 3-TM unit and a functional transporter containing at least four TMs (tetramer) (Xuan et al., 2013). There are 80–90 amino acids in the two cucumber MtN3/saliva domains, and they are present at almost the same positions in all the proteins, as shown in Table 1.

To confirm the presence of the TMs, the amino acid sequences were submitted to the TMHMM Server v.2.0⁷. The results confirmed that all, except 2 of the 17 CsSWEETs have seven TMs, CsSWEET5b and CsSWEET17a, which have six TMs (Figure 2). The N-terminal of CsSWEET1, 2, 3, 7a, 9, 10, 12a, 12b, 12c, 15, 17b, 17c is outside of the membrane, while it is on the inside for CsSWEET5a, 5b, 5c, 7b, 17a (Figure 2).

In order to identify the substrates being transported by the CsSWEET proteins, the hexose uptake-deficient yeast mutant EBY.VW4000 and the sucrose uptake-deficient yeast mutant SUSY7/ura3 (Riesmeier et al., 1992; Wieczorke et al., 1999) were used. We isolated six full-length cDNA clones, which encoded putative hexose transporters and two full-length cDNA clones, which encoded putative sucrose transporters, from cucumber male flowers. We then determined that CsSWEET1 and CsSWEET2 from clade I allowed glucose uptake, or glucose and fructose uptake, in the yeast EBY.VW4000 mutant (Figure 3), but did not show detectable complementation of uptake of other hexoses (Supplementary Figure S2). From clade II, CsSWEET5a supported uptake of glucose and fructose, while CsSWEET7b conferred glucose, galactose and mannose uptake (Figure 3), but not of other hexoses. From clade III, both CsSWEET10 and CsSWEET12c conferred sucrose uptake in the yeast SUSY7/ura3 mutant but did not allow hexose uptake in the EBY.VW4000 mutant (Figure 3 and Supplementary Figure S2). Finally, from clade IV, CsSWEET17a allowed glucose uptake in the yeast EBY.VW4000 mutant, while CsSWEET17c enabled glucose, galactose and fructose uptake (Figure 3), but not of other hexoses (Figure 3 and Supplementary Figure S2).

In conclusion, CsSWEET proteins from clades I, II, and IV transport hexoses, such as fructose, galactose, mannose, and particularly glucose. CsSWEET proteins from clade III transport sucrose, rather than hexoses, which is similar to results from A. thaliana and rice (Figure 3).

In order to determine the spatiotemporal expression pattern of each CsSWEET gene, the transcriptional profile in different cucumber organs was analyzed using qRT-PCR. Samples of fruit at the marketable fruit maturation stage 9 DAA (days after anthesis) (Hu et al., 2011) and other organs, including a source organ (mature leaf), sink organs (root, male flower, female flower, fruit, and young leaf) and a transporting organ (stem), were isolated. The eight analyzed CsSWEET genes from the four clades were expressed in almost all organs (Figure 4I), and at very high levels in male and female flowers, especially in the case of CsSWEET7b, -10, -12c, and -17c, where the expression was approximately 10–100 times greater than in other organs (Figure 4). But beyond that, the genes from different clades showed distinct expression patterns. In clade I, CsSWEET1 and CsSWEET2 were highly expressed in mature leaves, with CsSWEET1 also showing high expression in roots and CsSWEET2 in 9 DAA fruit, while they both showed low expression in the other organs (Figures 4A,B). In clade II, CsSWEET5a showed high expression in mature leaves, and low expression in roots and stems. CsSWEET7b was mainly expressed in roots, with little expression in stems, source leaves and fruit (Figures 4C,D). In clade III, CsSWEET10 was specifically expressed in fruit, while CsSWEET12c was expressed only in the stems and flowers (Figures 4E,F). In clade IV, CsSWEET17a was expressed in almost all organs, but with a high level in young leaves and mature leaves, while CsSWEET17c was highly expressed in stems besides flowers, and less in mature leaves and fruits (Figures 4G,H).

Previous studies have reported that different AtSWEET proteins have different subcellular localizations (Chen et al., 2015a; Eom et al., 2015), which in turn is suggestive of distinct functions. Here, the subcellular localization of the cucumber CsSWEET proteins was determined by expressing CsSWEETs-GFP fusion proteins in A. thaliana protoplasts. We found that CsSWEET1-GFP from clade I, CsSWEET7b-GFP from clade II and CsSWEET12c-GFP from clade III were targeted to the plasma membrane, while CsSWEET17a-GFP from clade IV localized to the tonoplast (Figure 5). These results were also verified in chloroplast-free onion (Allium cepa) epidermal cells (Supplementary Figure S1), suggesting that proteins from clades I, II, and III are plasma membrane proteins that may be involved in transmembrane transportation between cells, while clade IV proteins are tonoplast transporters, and may be involved in transportation and accumulation inside cells and in balancing intracellular hexose homeostasis.

In this study, 17 putative cucumber SWEET genes were identified, classified into four clades (I–IV), and named according to their homologs in A. thaliana and earlier work (Hu et al., 2017). The overall sequence identity of all cucumber SWEET proteins is 42%, while the equivalent identity for the A. thaliana SWEET proteins is 43%, and the degree of identity among all SWEET proteins shown in the phylogenetic tree was 33.03% (Figure 1). We identified six cucumber SWEET genes in clade III, five genes in clade II, and three genes in each of clades I and IV. This is similar to the numbers in A. thaliana and other plant species. In clade III, A. thaliana has 7 genes, tomato (Solanum lycopersicum) has 13, potato (Solanum tuberosum) has 15, and Medicago truncatula has 10 (Feng et al., 2015; Manck-Götzenberger and Requena, 2016). This relatively high number may be related to the important role of sucrose in long distance transport. In contrast there are only three clade IV genes in cucumber and potato, one in M. truncatula and rice, and two in A. thaliana and tomato.

We determined that CsSWEET proteins from clades I and II could mediate uptake of hexose, especially glucose, in the yeast EBY.VW4000 mutant, which has also been reported for their homologs from A. thaliana (Chen et al., 2010, 2015a). In our study, we detected some differences between the uptake preferences of the different CsSWEET proteins: CsSWEET2 allowed fructose uptake, CsSWEET5a allowed the uptake of fructose and glucose, while CsSWEET7b conferred the uptake of glucose, and to a degree mannose, as well as galactose (Figure 3). In other species, less research has been done on substrate specificity. ZmSWEET4c and OsSWEET4 function as glucose and fructose transporters, VvSWEET4 encodes a glucose transporter and OsSWEET5 encodes a galactose transporter in yeast (Guo et al., 2014; Zhou et al., 2014; Sosso et al., 2015). Clade III was reported to be a sucrose-specific clade in A. thaliana (Chen et al., 2012), CsSWEET10 and CsSWEET 12c, also belong to clade III and mediated the uptake of sucrose, but not hexoses in the EBY.VW4000 mutant. In clade IV, the A. thaliana protein AtSWEET17 was found to be a fructose-specific transporter (Guo et al., 2014), while CsSWEET17a rescued glucose uptake and CsSWEET17c rescued glucose, galactose, and fructose uptake in the yeast mutant.

We therefore conclude that there may exist some differences in the functions of the SWEET proteins among species, and that the SWEET proteins generally have a broader substrate not only uptake glucose but other hexoses. Recently, it has been found that SWEETs have the functions of transporting plant hormone, such as gibberellin (Kanno et al., 2016).

Hu et al. (2017) predicted the following subcellular localizations of CsSWEETs by WoLF PSORT (a localization prediction tool); CsSWEET1 in the chloroplast/vacuole, CsSWEET12c and CsSWEET17a in the chloroplast. In our study, using GFP fusions, we discovered that CsSWEET1, CsSWEET7b, and CsSWEET12c are plasma membrane proteins, while CsSWEET17a is a vacuole membrane protein (Figure 5). These results, obtained through experimental tests differ from those of Hu et al. (2017) predicted by bioinformatics software/tools.

In this study, all of the eight tested CsSWEET genes were most highly expressed in male and female flowers, which are strong carbon sinks. SWEET proteins have been reported as having roles in nectar secretion and also tapetum and pollen formation (Guan et al., 2008; Sun et al., 2013; Lin et al., 2014), and in flowers, a lack of sugar availability may lead to flower abortion (Lebon et al., 2008; Chong et al., 2014). CsSWEET1 and CsSWEET7b showed high expression in roots and flowers, and may be involved in the uptake and release of carbohydrates in these organs, or alternatively function in an interaction between roots and soil microorganisms. They may also play important roles in sugar partitioning during flower and root growth development (Figures 4, 5). This is similar to AtSWEET4, which is a hexose transporter that mediates sugar transport to axial sinks (Liu et al., 2016). ZmSWEET4c and OsSWEET4 also appeared to be responsible for transferring hexoses across the BETL to sustain development of the large starch-storing endosperm of cereal grains and contribute to sink strength (Sosso et al., 2015).

Most studies of SWEET proteins in sink organs have focused on flowers, seeds or roots (Guan et al., 2008; Chardon et al., 2013; Sun et al., 2013; Guo et al., 2014; Chen et al., 2015b). The roles in fleshy sink organs, such as fruits, is a new and important area of research. In cucumber fruits, phloem unloading follows an extensive apoplastic pathway and needs transporter participation (Hu et al., 2011). In this study, CsSWEET2 and CsSWEET10 showed high levels of expression in fruit and their substrates were found to be hexoses and sucrose, respectively. Accordingly, sucrose may be released into the phloem parenchymal cells by CsSWEET10 transmembrane transport and glucose and fructose derived from sucrose degradation may be released into the phloem parenchymal cells by CsSWEET2. Their high expression levels in fruit is consistent with a possible role in apoplastic sugar unloading and fruit development.

In contrast, CsSWEET17a was found to be a vacuolar transporter (Figure 5) and showed 47.1% similarity with CsSWEET17c. They are both in clade IV and show 41.55 and 51.8% similarity with AtSWEET17, respectively, which is known to be a vacuolar transporter (Chardon et al., 2013; Guo et al., 2014). They were expressed in almost all the sampled organs as they are tonoplast proteins. We speculate that these genes, from clade IV, play a role in facilitating hexose transport across the tonoplast and in balancing intracellular hexose homeostasis.