The SWEET gene family in Hevea brasiliensis – its evolution and expression compared with four other plant species

SWEET proteins play an indispensable role as a sugar efflux transporter in plant development and stress responses. The SWEET genes have previously been characterized in several plants. Here, we present a comprehensive analysis of this gene family in the rubber tree, Hevea brasiliensis. There are 36 members of the SWEET gene family in this species, making it one of the largest families in plant genomes sequenced so far. Structure and phylogeny analyses of these genes in Hevea and in other species demonstrated broad evolutionary conservation. RNA‐seq analyses revealed that SWEET2, 16, and 17 might represent the main evolutionary direction of SWEET genes in plants. Our results in Hevea suggested the involvement of HbSWEET1a, 2e, 2f, and 3b in phloem loading, HbSWEET10a and 16b in laticifer sugar transport, and HbSWEET9a in nectary‐specific sugar transport. Parallel studies of RNA‐seq analyses extended to three other plant species (Manihot esculenta, Populus trichocarpa, and Arabidopsis thaliana) produced findings which implicated MeSWEET10a, 3a, and 15b in M. esculenta storage root development, and the involvement of PtSWEET16b and PtSWEET16d in P. trichocarpa xylem development. RT‐qPCR results further revealed that HbSWEET10a, 16b, and 1a play important roles in phloem sugar transport. The results from this study provide a foundation not only for further investigation into the functionality of the SWEET gene family in Hevea, especially in its sugar transport for latex production, but also for related studies of this gene family in the plant kingdom.

SWEET proteins play an indispensable role as a sugar efflux transporter in plant development and stress responses. The SWEET genes have previously been characterized in several plants. Here, we present a comprehensive analysis of this gene family in the rubber tree, Hevea brasiliensis. There are 36 members of the SWEET gene family in this species, making it one of the largest families in plant genomes sequenced so far. Structure and phylogeny analyses of these genes in Hevea and in other species demonstrated broad evolutionary conservation. RNA-seq analyses revealed that SWEET2, 16, and 17 might represent the main evolutionary direction of SWEET genes in plants. Our results in Hevea suggested the involvement of HbSWEET1a, 2e, 2f, and 3b in phloem loading, HbSWEET10a and 16b in laticifer sugar transport, and HbSWEET9a in nectary-specific sugar transport. Parallel studies of RNA-seq analyses extended to three other plant species (Manihot esculenta, Populus trichocarpa, and Arabidopsis thaliana) produced findings which implicated MeSWEET10a, 3a, and 15b in M. esculenta storage root development, and the involvement of PtSWEET16b and PtSWEET16d in P. trichocarpa xylem development. RT-qPCR results further revealed that HbSWEET10a, 16b, and 1a play important roles in phloem sugar transport. The results from this study provide a foundation not only for further investigation into the functionality of the SWEET gene family in Hevea, especially in its sugar transport for latex production, but also for related studies of this gene family in the plant kingdom.
SWEET (Sugars Will Eventually be Exported Transporter) proteins, which feature up to seven transmembrane TM helix domains, selectively transport different kinds of sugar substrates, including sucrose, fructose, and glucose [1]. As a sugar efflux transporter, SWEET proteins play important roles in plant growth and development, stress responses, and plant-microbe interactions. Cellular sugar efflux is an essential function in many processes, such as phloem loading, nectar secretion, nourishing symbionts such as mycorrhiza, and in maternal efflux for filial tissue development [1]. Sugar efflux systems can be hijacked by pathogens for access to nutrition from hosts [2], and accordingly, mutations that block recruitment of the efflux mechanism by the pathogen facilitate plant resistance to their attack [3]. Previous studies on SWEETs have been focused mainly on two model plant species, namely Arabidopsis thaliana and Oryza sativa [4][5][6]. In A. thaliana, 17 SWEET family members have been characterized, and they fall into four phylogenetic clades, in which AtSWEET1-3 are in Clade I, AtS-WEET4-8 in Clade II, AtSWEET9-15 in Clade III, and AtSWEET16-17 in Clade IV. The SWEETs from most of the other plants are named following the nomenclature adopted in A. thaliana. SWEET genes, in their many isoforms, are versatile and their functions in plants are widely encompassing. For example, AtSWEET1 acts as a glucose transporter [7], with AtS-WEET9 being a nectary-specific sugar transporter which is essential for nectar production [6]. AtS-WEET11 and AtSWEET12 catalyze sucrose export from phloem parenchyma cells in source leaves and play a critical role in phloem loading [5]. AtSWEET16 and AtSWEET17, as vacuolar SWEET proteins, function as fructose-specific exporters, connecting the vacuolar lumen to the cytosol [8,9]. In O. sativa, OsSWEET11, located on the plasma membrane and expressed in the phloem of leaves, is presumably involved in phloem loading, as is the case with its Arabidopsis homologues, AtSWEET11 and AtSWEET12 [5]. OsSWEET11 and OsSWEET14 are specifically exploited by bacterial pathogens for virulence by means of direct binding of a bacterial effector to the SWEET promoter [4,10]. Recently, genome-wide expression patterns of SWEET genes have been characterized in Brassica napus, Pyrus bretschneideri, Sorghum bicolor, and Glycine max [11][12][13][14], all pointing to important roles of SWEETs in plant growth, development, and stress responses.
Natural rubber (cis-1,4-polyisoprene, NR) is an important industrial and strategic raw material, the sole commercial source of which is Hevea brasiliensis (the Para rubber tree, Hevea hereafter), a perennial tropical tree species [15]. As sucrose is the precursor molecule for rubber biosynthesis and latex regeneration [16], understanding the mechanisms of its transport and metabolism in the rubber tree is of fundamental importance to improving rubber productivity [17]. Significant progress was made in the understanding of Hevea sucrose transport and metabolism with the cloning of six sucrose transporter (SUT) genes, among which HbSUT3 (HbSUT1B) was found to be the key member responsible for sucrose loading into laticifers [18,19]. HbNIN2 has also been identified as the key gene responsible for sucrose catabolism in rubber-producing laticifers [20]. Moreover, two Hevea sucrose synthase genes, HbSUS2 and HbSUS3, were found to exert negative control over sucrose catabolism in the laticifers [21]. Hevea SWEET genes have not hitherto been investigated in detail, but their characterization has recently been facilitated by the Hevea genome and transcriptome having been independently sequenced by research groups from China [22], Malaysia [23,24], and Thailand [25], and by the availability of the proteome of Hevea latex [26].
We report here a genome-wide analysis of the SWEET gene family in Hevea where we compare the results with those from four other plant species, viz. Manihot esculenta and Ricinus communis belonging to the same family (Euphorbiaceae) as Hevea, and two model plants, A. thaliana and Populus trichocarpa. The study encompassed a total of 127 SWEET genes, the expression patterns of which were analyzed in different tissues in response to various treatments, and at several phases of tissue development. In addition, the gene structure and phylogeny of these genes were compared to help further understanding of the roles of SWEET genes in Hevea sugar transport. As a further objective in this investigation, data on SWEET genes and their expression in four other plant species were examined, along with the results from Hevea, to compare the structure of their respective gene families and appraise the functions of their members in the plant kingdom.

Results and Discussion
Genome-wide identification of SWEET gene families in Hevea and four other plant species We identified all SWEET gene family members in five plant species (Hevea, A. thaliana, P. trichocarpa, M. esculenta, and R. communis) from their published genome sequences. In this exercise, the SWEET genes from the three Euphorbiaceae plants (Hevea, M. esculenta, and R. Communis) were characterized for the first time. The most recent genome and protein sequences of these species were downloaded from Phytozome v10. Local BLAST searches of the genomes were performed using the published SWEET sequences of three model plants of A. thaliana, O. sativa, and P. trichocarpa as queries [1,5]. This analysis identified a total of 127 SWEET genes in the five selected plant species, comprising 36 SWEET genes in Hevea (Table 1a) [22], 28 in M. esculenta (Table 1b), 18 in R. communis (Table 1c), 17 in A. thaliana [10], and 28 in P. trichocarpa [6]. All the SWEET gene members newly identified this study were named according to the nomenclature of the A. thaliana SWEET gene family. The gene numbers of SWEET families identified here for the two model plants (A. thaliana and P. trichocarpa) matched those previously reported [5].
The lengths of SWEET-coding regions (CDS) were similar among the three Euphorbiaceae plants examined, ranging from 504 to 915 bp in Hevea, 513 to 906 bp in M. esculenta, and 504 to 891 bp in R. communis (Table 1). The molecular weights of the SWEET proteins in three Euphorbiaceae species ranged from 18.7 to 33.9 kDa, while their isoelectric points (pIs) fell between 6.24 and 10.20 (Table 1).

Phylogenetic analysis of the SWEET gene families
In order to establish the phylogenetic relationships in the SWEET gene families among Hevea and the four other plant species, we aligned the 127 SWEET protein sequences in plants and constructed a phylogenetic tree as shown in Fig. 1 ( Table S1). The plant SWEET proteins were clustered into four major groups with high bootstrap values, designated Clades I to IV. The 36 Hevea SWEET genes were dispersed among the four groups with 11, 7, 12, and 6 isoforms, respectively, in Clades I, II, III, and IV. Similarly, the SWEET family of genes in the other four species were also clustered into the above four groups, with 3, 5, 7, and 2 isoforms, respectively, in A. thaliana, 11, 3, 8, and 6 in P. trichocarpa, 6, 4, 12, and 6 in M. esculenta, and 3, 6, 6, and 3 in R. communis (Table 1, Fig. 1). Phylogenetic analysis as well as amino acid sequence comparison revealed universal existence of paralogous SWEET gene pairs and clusters in the five species. In Hevea, nine such SWEET gene pairs (HbSWEET2a/2b in Clade I, HbSWEET2c/2d in Clade I, HbSWEET2e/2f in Clade I, HbSWEET4a/4b in Clade II, HbSWEET5a/5b in Clade II, HbSWEET10e/10f in Clade III, HbSWEET15a/15b in Clade III, HbSWEET16b/16c in Clade IV, and HbSWEET17a/17c in Clade IV) and one gene cluster (HbSWEET1a, 1b, and 1c in Clade I) were identified. Except for the pairs of HbSWEET2c/2d, HbSWEET2e/2f, and HbSWEET10e/10f, the Ka/Ks values of the other paralogous gene pairs were less than 0.5, showing that these genes had undergone a purifying selection ( Table 2). The different expression patterns exhibited by the two genes in most of the gene pairs suggested that a functional divergence had occurred after gene duplication (Fig. 4). In A. thaliana, there were two SWEET gene pairs (AtS-WEET6/7 and AtSWEET16/17) and one paralogous gene cluster (AtSWEET11, 12,13,14). In P. trichocarpa, there were two SWEET gene pairs (PtSWEET15a/15b and PtSWEET17a/17b) and four paralogous gene clusters (PtSWEET1a, 1b, 1c, 1d; PtSWEET2a, 2b, 2c; PtSWEET3a, 3b, 3c; and PtSWEET10a, 10b, 10c, 10d). In M. esculenta and Upon further examining the genomic locations, we found that some SWEET genes in the same clade were located adjacent to each other. For example, in Hevea, HbSWEET4b and HbSWEET4c were located on scaf-fold0371, HbSWEET10a and HbSWEET10c on scaf-fold1273, HbSWEET10b, HbSWEET10d, and HbSW EET10f on scaffold0491, HbSWEET11 and HbSWEE T12 on scaffold0807, and HbSWEET17a and HbSWEE T17b on scaffold0340. In P. trichocarpa, PtSWEET1b, PtSWEET1c, and PtSWEET1d were located adjacent to each other on chromosome 2, PtSWEET3a and PtS WEET3c on chromosome 15, PtSWEET10a, PtSWE ET10b, PtSWEET10c, and PtSWEET11 on chromosome 15, PtSWEET16b, PtSWEET16c, PtSWEET17a, and PtSWEET17b on chromosome 13. In R. communis, RcSWEET4a and RcSWEET4d were located on scaf-fold39822, and RcSWEET10a, RcSWEET10b, RcSWE ET11, and RcSWEET12 on scaffold30147. These adjacent gene pairs and clusters had apparently been derived from tandem duplication events.

Structural organization of SWEET genes
The exon-intron structures of the 127 SWEET genes in five plant species were determined based on the predicted sequences. As shown in Fig. 2A, most Hevea SWEET members within the same groups share similar gene structures in terms of intron number, domain localization, and exon length. Although the lengths vary, introns are inserted into nearly the same locations of the gene ORFs. Most SWEET members contain 3-5 introns. Of the 36 members in Hevea, for example, 24 have 5 introns, 7 have 4 introns, and 5 have 3 introns (Fig. 2A, Table 1a). In the total of 127 SWEET genes among the five plant species, there were only three SWEET members with no introns, namely RcSWEET4b, RcSWEET4c, and AtSWEET6, all of which were clustered in Clade II ( Fig. 2A-E). Some SWEET members lacked exons at the 5 0 end, such as HbSEET2f, 4c, 5b, 15a, 10e, and 10c, MeSWEET4, 3b, 2b, 12a, and 10b, RcSWEET2, 5, and 4a, and PtSWEET1c, 16c, and 15a ( Fig. 2A-E). Most SWEET members contain 4-7 TM helix domains, and 25 of the 127 members lost one to three of the seven TM helix typical of plant SWEETs (Fig. 3). In addition, the lengths of most AtSWEET and RcSWEET genes are shorter than those of the other plant SWEET genes, perhaps reflecting a relationship between gene length and genome size of a given species.

Tissue expression of SWEET genes
To investigate the functions of SWEET genes, gene expression profiles in different tissues were analyzed by using Solexa sequencing data in Hevea, M. esculenta, P. trichocarpa, and A. thaliana (Tables S2, S3). Analysis of gene expression from the Sequence Read Archive (SRA), adopted in the present study, has limitations as the data were compiled from different sources where genetic differences in the tested tissues and dissimilarities in the experimental conditions can make comparisons difficult. Nonetheless, such an analysis provides a broad overview of the functionalities of the various Hevea SWEETs relative to their counterparts in other plant species. The results provide useful indicators as to which SWEET genes are most commonly expressed from among the numerous isoforms. These results would serve as a guide for future follow-up research where more exacting methodologies can be employed.
As shown in Fig. 4A, the expression levels of four SWEET genes (HbSWEET1a, 2e, 2f, and 3b) in Hevea  source leaves were significantly higher than those of the other members, while three SWEET genes (HbSWEET1c, 10a, and 10b) were mainly expressed in sink leaves. In the other three plant species, MeS-WEET17, MeSWEET2a, MeSWEET15b, PtSWEET2a, and PtSWEET16d were highly expressed in leaves and AtSWEET11 was mainly expressed in seedling plants (11 days old) (Fig. 4B-D). Some of the above-mentioned SWEET members might be involved in phloem loading and leaf development. In Hevea bark where rubber-producing laticifers reside, HbSWEET1a and HbSWEET16a showed a predominance expression, while in latex, the cytoplasm of laticifers, HbSWEET2a, HbSWEET10a, HbSWEET10b and HbSWEET16b were the predominant isoforms. These SWEET genes might play an important role in sugar transport between the laticifers and their neighboring bark tissues, and contribute to the regulation of sucrose concentrations in laticifers together with the sucrose transporter responsible for apoplasmic sucrose uptake of laticifers, HbSUT3 [18,19]. In A. thaliana, AtSWEET9 has been identified as a nectary-specific sugar transporter [6]. In Hevea, HbSWEET9a exhibited a male flower-specific abundant expression and might have a similar function in nectary production as its A. thaliana orthologue, AtSWEET9. In addition, 14 other Hevea SWEET genes, viz . HbSWEET1a, 1c, 2a,  2e, 2f, 3a, 3b, 7, 10a, 10b, 10e, 11, 16b, and 17b, were also expressed at high levels in flowers (Fig. 4A). No SRA expression data in flowers were found in the other three plant species. In Hevea roots, seven SWEET genes (HbSWEET1a, 2a, 2f, 3b, 4c, 10e, and  17c)  MeSWEET15b were expressed at high levels in the storage roots of M. esculenta; their activities may be related to starch formation. Six P. trichocarpa SWEET genes (PtSWEET2a, 2c, 3a, 3c, 16b, and 16c) were expressed at high levels in the roots. On the other hand, most of A. thaliana SWEET genes showed low or no expression in the roots. PtSWEET16b and PtSWEET16d exhibited high expression in xylem fiber cells that may be related to xylem formation. There were many SWEET genes showing universal expressions in most tissues examined. These included HbSWEET1c, 10e, 2c, 3b, 17c, 2d, 2e, 2f, 1a, 16a, 6,  2a, 16b, 10b, and 10a, MeSWEET17, 1a, 1b, 16b, 17c,  10b, 10a, 2a, 2b, 17b, 15b, 10d, 16a, 3a, and 9a,  PtSWEET2a, 16d, 15b, 16b, 2b, 17a, 2c, 16c, 3a, and 10c, and AtSWEET1, 2, 17, 11, 12, and 16. Interestingly, isoforms of SWEET2, 16, and 17 were observed among the universally expressed SWEET genes in all plant species examined, which might represent the main evolutionary direction of SWEET genes in plants. As shown in Fig. 4A, transcripts of 11 HbSWEET genes (HbSWEET4a, 4b, 5a, 5b, 9b, 10c, 10f, 12, 15a, 15b, and 17a) were barely detectable in almost all the tissues and all the treatments examined. Such genes comprise a large portion (~1/3) of the total HbSWEET gene family. This character seems to be shared by the SWEET gene families in other plant species. For example, similar expression patterns were observed for seven of 28 SWEET genes in M. esculenta (Fig. 4B), 12 of 28 in P. trichocarpa (Fig. 4C), and 4 of 17 in A. thaliana (Fig. 4D). This result suggests that the SWEET gene families in higher plants might have experienced an event of gene expansion followed by nonfunctionalization in the course of evolution. A similar phenomenon has been reported in our studies for the CDPK and CDPK-related kinase gene families in Hevea [27]. In addition, we found that most genes in Clade II have low or no expansion in all tissues examined in the four plant species.
Expression profile of SWEET genes in response to hormone and stress treatments Expression levels of SWEET genes in Hevea were also examined under various kinds of hormone and stress treatments. Ethephon, an ethylene generator, is widely used to stimulate latex production of the rubber tree, but the yield-stimulating mechanisms are still poorly understood [15,22]. As shown in Fig. 4A, expressions of HbSWEET10a were obviously upregulated by ethephon treatment in latex. In addition, HbSWEET10a was the predominant SWEET isoform in latex, the expression of which was higher than any of the other SWEET members, suggesting its important role in sugar transport of laticifers. Expressions of HbSWEET10a and HbSWEET2a appeared to be regulated by methyl jasmonate (MeJA) although in differing manners. Expressions of HbSWEET2c, HbSWEET2d, and HbSWEET3 were downregulated under drought treatment. Under low temperature treatment, expressions of HbSWEET1c were upregulated, whereas HbSWEET2c, 2d, 16a, and 17c were downregulated. Expressions of Hevea SWEET genes were also regulated by other kinds of stress treatments. For example, the expressions of HbSWEET1b, HbSWEET1c, and HbSWEET10e were affected by tapping panel dryness, a complex physiological disorder affecting latex production severely [28]; HbSWEET17b expressions were downregulated under the infection of Corynespora cassiicola, a fungal pathogen causing a leaf fall disease in Hevea [29].
The expression levels of SWEET genes in M. esculenta, P. trichocarpa, and A. thaliana were also examined when the plants were subjected to treatments of hormones and different stresses, including fungus infection, drought, and cold ( Fig. 4B-D). The expressions of six MeSWEETs (MeSWEET1a, 10a, 10b, 15b, 17, and 17c) were affected by fungus infection. Expressions of PtSWEET2b and PtSWEET16d were induced by MeJA in roots. Expressions of PtSWEET15b were induced by drought and ABA (abscisic acid) treatments. In the model plant A. thaliana (Fig. 4D), the expressions of ATSWEET16 and ATSWEET17 were upregulated by MeJA in roots. In seedlings, expressions of ATSWEET12 were upregulated by MeJA, while those of ATSWEET16 were downregulated. Under cold treatment, expressions of ATSWEET1 and ATSWEET2 were upregulated, while those of ATS-WEET16 and ATSWEET17 were downregulated.

Expression analyses of HbSWEET10a, HbSWEET16b, and HbSWEET1a based on qPCR
Rubber is synthesized and stored in the cytoplasm (latex) of highly specialized cells called laticifers that are differentiated from the cambium and arranged in rings. To further examine the expression of HbSWEET genes in latex and bark, quantitative RT-PCR (qPCR) analyses of HbSWEET10a, HbSWEET16b, and HbSWEET1a were performed. As shown in Fig. 5, the results from qPCR were in broad agreement with the sequencing-based expression analyses. HbSWEET10a and HbSWEET16b were mainly expressed in latex; HbSWEET1a was mainly expressed in bark and flower (Fig. 5A). was obviously downregulated after 24 hours of ethephon treatment in latex, which agrees well with the results based on RNA-seq (Fig. 4A, 5B, Table S3-1). We also further examined the expression of HbSWEET1a under ethephon treatment in bark, while HbSWEET1a was obviously upregulated (Fig. 5B).
The process of rubber harvesting, namely tapping, produces a conspicuous stimulating effect on latex production in virgin Hevea trees, and it has been partially ascribed to an enhanced sucrose uptake and sucrose catabolism in the laticifers [18,20]. As shown in Fig. 5C, HbSWEET10a and HbSWEET16b were obviously upregulated by tapping in three different clones PR107, Reyan7-33-97, and Reyan8-79. All above results revealed that HbSWEET10a, 16b, and 1a might play an important role in sugar transport in laticifer and bark.

Conclusion
In this study, a genome-wide analysis of SWEET gene families was undertaken for the first time in Hevea, M. esculenta, and R. communis. In silico analysis of the Hevea genome database facilitated the identification of 36 SWEET genes. The phylogenetic analysis of 127 SWEETs from Hevea and four other plant species (A. thaliana, P. trichocarpa, M. esculenta, and R. communis) classified all these SWEETs into four major groups. Members within each group might have had common evolutionary origins as seen from the sharing of similar protein motifs, exon-intron structures, and basic molecular functions. Solexa sequencing analyses revealed that SWEET2, 16, and 17 were universally expressed in different tissues of all the plant species examined, possibly representing the main evolutionary direction of plant SWEET gene families. Extensive expressional analyses in different tissues and in response to various experimental treatments, including hormones, and biotic and abiotic stresses, identified multiple tissue-specific SWEET isoforms and isoforms showing striking responses to some of the treatments in Hevea and three other plant species (A. thaliana, P. trichocarpa, and M. esculenta). These results indicate versatile roles of SWEETs in plant growth, development, and stress responses and provide a foundation for further functional investigation of the SWEET gene families in the plant kingdom.

Materials and methods
Database search for SWEET genes in H. brasiliensis and four other plant species Sequences of A. thaliana and P. trichocarpa SWEET genes were downloaded from the A. thaliana Information Resource (http://www.Arabidopsis.org/) and GenBank (http://www.ncbi.nlm.nih.gov/genbank). The genome and protein sequences of A. thaliana [30], P. trichocarpa [31], M. esculenta [32], and R. communis [33] were downloaded from Phytozome v10 (http://www.phytozome.net/). The H. brasiliensis genome and transcriptome data were obtained from GenBank (http://www.ncbi.nlm.nih.gov/nucc ore/448814761) [22]. Local BLAST alignment was performed using published SWEET sequences from A. thaliana and P. trichocarpa as queries to search against the deduced proteome of each species for the candidate SWEETs from H. brasiliensis, A. thaliana, P. trichocarpa, M. esculenta, and R. communis. All putative candidates were manually verified with the InterProScan server (http://www.ebi.ac.uk/ Tools/pfa/iprscan/) to confirm the presence of protein kinase and TM helix domains.

Phylogenetic and gene structure analyses
Multiple alignments of the amino acid sequences of SWEETs from five species were performed using the Clustal X (version 1.83) program. The phylogenetic tree was constructed with MEGA6.0 [34] by employing the neighbor-joining (NJ) method with a bootstrap test for 1000 replicates. Exon-intron structures of the six species SWEET genes were analyzed by comparing the cDNA and their genomic DNA sequences through the web server GSDS 2.0 (http://gsds.cbi.pku.edu.cn/). The KaKs-Calculator program (https://sourceforge.net/projects/kakscalcu lator2/) was used to calculate the nonsynonymous (Ka) and synonymous (Ks) substitutions in coding regions.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Table S1. SWEET Accessions. Table S2. Basic information for Solexa sequencing data of Hevea brasiliensis and three other plant species. Table S3. RNA-seq analysis of the expressions of SWEET genes in Hevea brasiliensis and three other plant species.