SWEET sugar transporters for phloem transport and pathogen nutrition



Many intercellular solute transport processes require an apoplasmic step, that is, efflux from one cell and subsequent uptake by an adjacent cell. Cellular uptake transporters have been identified for many solutes, including sucrose; however, efflux transporters have remained elusive for a long time. Cellular efflux of sugars plays essential roles in many processes, such as sugar efflux as the first step in phloem loading, sugar efflux for nectar secretion, and sugar efflux for supplying symbionts such as mycorrhiza, and maternal efflux for filial tissue development. Furthermore, sugar efflux systems can be hijacked by pathogens for access to nutrition from hosts. Mutations that block recruitment of the efflux mechanism by the pathogen thus cause pathogen resistance. Until recently, little was known regarding the underlying mechanism of sugar efflux. The identification of sugar efflux carriers, SWEETs (Sugars Will Eventually be Exported Transporters), has shed light on cellular sugar efflux. SWEETs appear to function as uniporters, facilitating diffusion of sugars across cell membranes. Indeed, SWEETs probably mediate sucrose efflux from putative phloem parenchyma into the phloem apoplasm, a key step proceeding phloem loading. Engineering of SWEET mutants using transcriptional activator-like effector nuclease (TALEN)-based genomic editing allowed the engineering of pathogen resistance. The widespread expression of the SWEET family promises to provide insights into many other cellular efflux mechanisms.


The primary products of photosynthesis are sugars. Sugars are thought to move symplasmically between cells, yet at several tissue/organ interfaces intercellular transport occurs apoplasmically, requiring at least two transporters – one secreting sugars from cells, and the other mediating reuptake into the adjacent cells (Giaquinta, 1983; Lalonde et al., 2004). Examples include phloem loading in so-called apoplasmic phloem-loading species, sugar secretion from the seed coat for filial tissue nutrition, and nutrition of the gametophyte. In addition, plants secrete sugars from the nectary for attracting pollinators, as well as from roots to support the growth of beneficial microorganisms. Identifying and characterizing the secretion process as well as proteins involved in facilitating this transfer had presented a major challenge in the study of whole-plant photoassimilate translocation and allocation. Photoassimilate is synthesized in mesophyll cells and needs to be translocated to heterotrophic tissues to enable their growth and development. In apoplasmic phloem loaders, sucrose is the major translocation form of photoassimilate and its long-distance translocation is driven by hydrostatic pressure generated in the phloem. The initial step in the transport pathway, referred to as phloem loading, is the transfer of photoassimilate from the photosynthetic mesophyll cells into the sieve element/companion cell (SE/CC) complex of the collection phloem. Two major phloem-loading mechanisms have been proposed: apoplasmic loading and symplasmic loading (with polymer trapping combined with apoplasmic loading or without polymer trapping; Turgeon, 2010). In the case of apoplasmic loading, sucrose transporters (SUTs) have been identified as H+-coupled sucrose symporters and shown to be essential for sucrose translocation (Lalonde et al., 2004). More recently, a new class of sugar transporters, the SWEETs, have been implicated in the secretion of sucrose as a prerequisite for SUT1-mediated phloem loading (Chen et al., 2012). Interestingly, pathogens such as Xanthomonas oryzae, which causes rice blight, directly induce SWEET gene expression, probably in order to access the plant's sucrose pools (Chen et al., 2010). In this review, both the role of SWEET proteins in efflux of sugars for apoplasmic phloem loading and their role in the efflux of sugars for pathogen nutrition when ectopically induced by pathogens are discussed.

The mechanism of phloem loading

Apoplasmic loading

In the apoplasmic mechanism of phloem loading, sucrose is released from mesophyll cells into the mesophyll apoplasm and then diffuses towards the phloem, or potentially from phloem parenchyma cells into the phloem apoplasm after symplasmic transfer from mesophyll cells through the plasmodesmata via the bundle sheath towards the phloem (Giaquinta, 1983; Ayre, 2011). However, for decades there has been a lack of direct evidence to support this hypothesis, until the recent discovery of SWEET efflux carriers probably localized in the phloem parenchyma cells (Chen et al., 2012). After efflux into the phloem apoplasm, sucrose is taken up into the SE/CC complex by the SUT sucrose transporter, which can accumulate sucrose in the SE/CC complex against an outward-directed sucrose concentration gradient (Giaquinta, 1983; Riesmeier et al., 1992, 1994; Bush, 1993; Sauer & Stolz, 1994; Ayre, 2011). The SUTs play an essential role in phloem loading of sucrose (Riesmeier et al., 1994; Gottwald et al., 2000; Lalonde et al., 2004), and in sucrose partitioning from the vacuole to the cytoplasm (Reinders et al., 2008; Eom et al., 2011). Recent publications show that all members of the SUT4 phylogenetic clade are targeted to at least the tonoplast and the plasma membrane, which may be correlated with the diverse physiological functions of the SUT4 sucrose transporters, such as potato (Solanum tuberosum) StSUT4, which can affect the expression of ethylene biosynthesis-related genes and clock-regulated genes (Chincinska et al., 2013). Most SUTs have been characterized as secondary active sucrose importers, with the exception of a group of specialized sucrose facilitators (SUFs, close relatives of SUT1), PsSUF1, PsSUF4 and PvSUF1, found in seed coats of developing pea (Pisum sativum) and bean (Phaseolus vulgaris). PsSUF1, PsSUF4 and PvSUF1 show pH- and energy-independent and bidirectional sucrose transport activity in a sucrose uptake-deficient yeast mutant (Zhou et al., 2007). However, there is, as yet, no physiological characterization of these transporters, and no additional SUFs have been identified from pea or bean, nor orthologous SUFs from other species.

Symplasmic loading with a polymer trap combined with apoplasmic loading

For some species, a polymer trap mechanism has been proposed in which sucrose diffuses symplasmically from the photosynthetic sites into the CCs of minor veins through narrow plasmodesmata. In CCs, sucrose is converted to raffinose and stachyose, which are too large to diffuse back, but are able to proceed into the SEs through wider plasmodesmata. The polymer trap hypothesis is supported by the discovery that down-regulation of the two galactinol synthase (GAS) genes responsible for raffinose family oligosaccharide (RFO) synthesis was shown to inhibit RFO synthesis and long-distance transport of photoassimilate (McCaskill & Turgeon, 2007). There is a growing body of evidence showing that polymer trap and apoplasmic phloem-loading mechanisms co-exist. Cucumber mosaic virus (CMV)-infected plants of melon (Cucumis melo), previously characterized as a symplasmic loader, accumulated more sucrose in the phloem of source leaves, which was related to the elevated expression of the melon sucrose transporter CmSUT1 in vascular bundles of minor veins. These results indicate a loading machinery shift from symplasmic loading to apoplasmic loading triggered by CMV infection (Gil et al., 2011). The co-existence of two types of carbohydrate-loading strategy is also structurally indicated by the expression patterns of the Alonsoa meridionalis stachyose synthase 1 (AmSTS1) gene, which is expressed in intermediary cells, and the sucrose transporter AmSUT1 protein, which is present in ordinary companion cells (Voitsekhovskaja et al., 2009).

Symplasmic pathway without polymer trapping

In contrast to the active polymer trap mechanism, this type of symplasmic phloem loading is passive. Sugar concentrations are higher in the mesophyll than in the phloem, therefore, sugars are able to diffuse from mesophyll cells to the SEs of minor veins through plasmodesmata down a concentration gradient. There is no concentrating step being involved in sugar transport. The species with this loading strategy have elevated concentrations of translocatable sugars (e.g. sucrose) in the mesophyll cells to ensure a concentration gradient to the phloem, have a high plasmodesmatal frequency to accommodate flux and do not transport raffinose (Rennie & Turgeon, 2009). Many trees and other woody species preferentially utilize this passive loading mechanism.

SWEETs for phloem loading

SWEETs, a novel class of sugar transporters, were identified by screening for glucose and sucrose transporters using Förster resonance energy transfer (FRET)-based sugar sensors expressed in human embryonic kidney HEK293T cells characterized by low or negligible endogenous sugar transport activity (Chen et al., 2010, 2012). Phylogenetic analysis reveals that SWEETs are prevalent in plants and broadly conserved in eukaryotes. In eukaryotes, SWEET proteins are typically predicted to have seven transmembrane domains (TMDs) with two direct repeated units of 3 TMDs connected by TMD4, a topology distinct from that of the SUT/SUF sucrose transporters with 12 TMDs. In prokaryotes, the homologs of SWEETs contain only a single three-TMD repeat, but are capable of transporting sucrose (Chen et al., 2010; Yuan & Wang, 2013; M. Yuan et al., unpublished data). There are 17 SWEET members in Arabidopsis and 21 in rice (Oryza sativa), falling into four clades. Some members of clade I were found to mainly mediate glucose import and export in the HEK293T, yeast or oocyte expression system (Chen et al., 2010). More recently, using similar assays, members of clade III were found to preferentially transport sucrose across the plasma membrane. Some SWEETs were characterized as bidirectional, pH-independent and low-affinity sucrose transporters (for AtSWEET12, the Michaelis Constant (Km) of sucrose uptake was c. 70 mM and the Km of efflux was > 10 mM) via radio-tracer uptake or efflux assays in the Xenopus oocyte expression system. These data support the hypothesis that SWEETs operate via a uniporter transport mechanism and are compatible with the unidentified nonsaturable, linear components of biphasic kinetics of sucrose uptake in vivo (Ayre, 2011; Chen et al., 2012). The close paralogs AtSWEET11 and AtSWEET12 of seven members in clade III share 88% similarity and are probably expressed in leaf phloem parenchyma cells, which are adjacent to the SE/CC complex where phloem loading occurs. Thus, their involvement in the process of apoplasmic phloem loading is strongly indicated (Chen et al., 2012). The enhanced green fluorescent protein (eGFP)-tagged AtSWEET11 driven by its native promoter was localized to the plasma membrane of cells which probably correspond to the phloem parenchyma cells, further suggesting a role in sucrose transport across the plasma membrane in vivo. Physiological function studies in mutant plants support a key role of AtSWEET11 and AtSWEET12 in phloem loading (Chen et al., 2012). Although the single mutant plants of atsweet11 or atsweet12 did not display strong phenotypes, double mutant plants showed slower growth and excess accumulation of carbohydrate in the leaves (Chen et al., 2012). A reduction in the efflux of sugars from leaves of double mutant plants was directly demonstrated in a 14CO2 feeding experiment (Chen et al., 2012). However, the defect was moderate in comparison to that of the H+-sucrose symporter mutant atsuc2 (Gottwald et al., 2000), and this may be a result of functional redundancy with other SWEET paralogs or compensation by other AtSWEETs. The possibility of compensation is supported by the finding that expression of AtSWEET13 is up-regulated in the atsweet11;12 double mutant background (Chen et al., 2012), but confirmation of functional redundancy awaits phenotypic analysis of higher order mutants (e.g. the triple mutant atsweet11;12;13). Interestingly, although AtSWEET14 has the highest similarity to AtSWEET13 among the 17 AtSWEETs, the expression of AtSWEET14 is not affected in atsweet11;12. This suggests that we could differentiate the functions of AtSWEET13 and AtSWEET14 at this point. Taken together, these data indicate that AtSWEET11 and AtSWEET12 play an important role in the efflux of sucrose into the phloem apoplasm and are responsible for the step before apoplasmic phloem loading mediated by H+-sucrose symporters such as SUT1/SUC2 (sucrose transporter 1 or sucrose carrier 2) (Fig. 1a; Baker et al., 2012; Braun, 2012; Chen et al., 2012). These findings represent a highly significant breakthrough in our conceptual understanding of phloem transport and open up exciting new opportunities to discover the regulatory mechanism of sucrose transport from leaf sources to sites of utilization in sinks.

Figure 1.

The roles of SWEET proteins in the phloem-loading process and access of patho-zgens to nutrition. (a) Schematic representation of the steps involved in phloem loading in a leaf. Sucrose produced during photosynthesis in the mesophyll cells of apoplasmic loaders is transported from cell to cell through the plasmodesmata to the phloem parenchyma cells and then is exported into the phloem apoplasm by the SWEET sucrose effluxer, probably localized in the plasma membrane of phloem parenchyma cells. Subsequently sucrose is taken up and concentrated in the sieve element/companion cell complex by an H+-coupled sucrose symporter SUT/SUC energized by proton pump ATPase. (b) A model of the mechanism by which pathogenic bacteria access nutrition via the induction of SWEETs. The expression of SWEET genes may be elevated in mesophyll cells in addition to phloem parenchyma cells upon bacterial infection. As a consequence, more sucrose may be released and easily accessed by pathogens.

The rice ortholog of AtSWEET11 and AtSWEET12, OsSWEET11 (also named Os8N3/Xa13), was also characterized as a sucrose transporter (Chen et al., 2012). It was found to be localized to the plasma membrane of transformed calli and preferentially expressed in the phloem of rice leaves; however, resolution was not sufficient to discern parenchyma cells (Chu et al., 2006b). The localization and function of OsSWEET11 indicate a role in preparation for phloem loading.

SWEETs for access by pathogens to carbon

The expression of various AtSWEET genes can be induced by biotrophic bacteria or fungi, which indicates that many pathogens depend on AtSWEET activity to a certain extent (Chen et al., 2010). In rice, the SWEET gene OsSWEET11/Xa13/Os8N3 is co-opted during infection by Xanthomonas oryzae pv. oryzae (Xoo), one of the most devastating rice diseases world-wide (Chu et al., 2006a; Yang et al., 2006). The dominant gene OsSWEET11 is transcriptionally up-regulated by a specific PthXo1 (a pathogenicity gene identified from Xoo strain PXO99A) transcriptional activator-like (TAL) effector secreted by Xoo upon infection. The TAL effector specifically binds to the promoter of OsSWEET11 and activates transcription. OsSWEET11 is required for bacterial growth (Yuan et al., 2009). RNAi lines of OsSWEET11 confer resistance to Xoo (Yang et al., 2006). Nature mutations of the PthXo1 TAL docking site in the OsSWEET11 promoter abolish pathogen-specific induction and pathogenicity (Chu et al., 2006b), which is consistent with the function of OsSWEET11 required for supplying sucrose to the pathogens through a uniport mechanism. The expression of dominant OsSWEET11 may be activated and induced upon Xoo infection in the mesophyll tissues around infection sites such as wounds in addition to phloem parenchyma cells, which presumably leads to more sucrose secreted through OsSWEET11 into the intercellular spaces of nonvascular mesophyll tissues from infected leaves relative to noninfected leaves, and a better environment for bacterial reproduction, invasion and colonization of the xylem (Fig. 1b). When the availability of OsSWEET11 becomes limited (e.g. as a result of mutations in the TAL effector binding element of the OsSWEET11 promoter occurring naturally or as a result of application of the genomic editing strategy, or through RNA interference), not enough sucrose is presumably available to support pathogen growth efficiently and plants show resistance to specific Xoo infection.

OsSWEET11 is not the only member of the SWEET family that plays a role in the access of pathogens to nutrition. OsSWEET14/Os11N3, another member of the SWEET family, can be hijacked by Xoo using the TAL effector AvrXa7 (initially identified as a avirulence gene, corresponding to resistance gene Xa7 against Xoo in rice) or PthXo3 and can be activated to support pathogen growth (Antony et al., 2010; Chen et al., 2010). The dominant OsSWEET13/Xa25, but not recessive ossweet13/xa25 is specifically induced by Xoo strain PXO339. The recessive ossweet13/xa25 confers race-specific resistance to PXO339 (Liu et al., 2011). In conclusion, SWEETs have crucial roles in pathogenesis.

TAL effectors are so essential and specific for Xoo infection of plants that scientists have been developing new strategies to control pathogen infection based on the properties of TAL effectors. TAL effector nucleases (TALENs) are artificial restriction enzymes generated by fusing a native or customized TAL effector DNA-binding domain to a DNA cleavage domain of Fok I (a restriction endonuclease, naturally found in Flavobacterium okeanokoites) (Christian et al., 2010), which has been used to create site-specific genome modification in various organs (Sun & Zhao, 2013). TALEN technology has also been successfully applied to produce disease-resistant rice (Li et al., 2012, 2013). For example, a series of ossweet14 alleles carrying disrupted TAL effector-binding elements in the promoter were generated by TALEN-based promoter editing strategies. These mutants confer resistance to the AvrXa7- or PthXo3-dependent Xoo strains and do not show any developmental defect relative to wild type (Li et al., 2012). A major advantage of TALEN technology is that the TALEN gene and the selective marker used to introduce TALEN can be removed by segregation. As TAL-DNA-binding specificity is predictable and TAL effectors from Xoo bacteria tend to activate the cognate host genes, the potential binding elements of unknown SWEETs bound by TAL effectors can be predicted and easily modified through TALEN-based technology to facilitate research into and crop improvement against different types of pathogen. Designer TAL effectors (dTALEs) were used to study the function of OsSWEET12, an as yet uncharacterized SWEET gene with no corresponding naturally occurring TAL effector identified. The expression of OsSWEET12 was induced upon infection of the corresponding Xoo strain transformed with dTALEs which were designed to bind to the putative TATA boxes of OsSWEET12, consequently, OsSWEET12 conferred susceptibility to this dTALE expressed Xoo strain (Li et al., 2013). This example strongly supports the view that dTALEs represent a powerful technology that can be used to analyze unknown gene functions before the TALEN strategy is applied to plants.

Roles of SWEETs in other efflux processes

AtSWEET8/RPG1 (Ruptured Pollen Grain1) was found to be expressed in the tapetum during male meiosis (Guan et al., 2008). AtSWEET8 was characterized as a bidirectional glucose transporter when co-expressed with a cytosolic localized glucose sensor or endoplasmic reticulum (ER)-targeted glucose sensor in HEK293T cells (Chen et al., 2010). The atsweet8/rpg1 mutant exhibits severely reduced male fertility (Guan et al., 2008), which is compatible with a role in glucose efflux from the tapetum for pollen growth. Similar to AtSWEET8, OsSWEET11-silenced plants had lower pollen viability (Yang et al., 2006). The knockout mutant of AtSWEET14/Os11N3 produced smaller seeds and had delayed growth, suggesting a role in seed filling (Antony et al., 2010). AtSWEET5/VEX1 (Vegetative Cell Expressed 1), a gene encoding a glucose transporter, is specifically expressed in the vegetative cells of pollen grains, indicative of a role in supplying sugar for generative cell development (Engel et al., 2005). The homolog of AtSWEET9 from Petunia hybrida, NEC1 (a nectary tissue expressed gene), is specifically expressed in nectaries (Ge et al., 2000), suggesting a role in nectar secretion. AtSWEET17, expressed in the parenchyma and the vascular tissue, functions as a vacuolar fructose transporter, exporting fructose out of vacuoles and controlling fructose content in leaves (Chardon et al., 2013). Mutants of AtSWEET17 show stunted growth and lower seed yield, which indicates a role in carbohydrate allocation in plants (Chardon et al., 2013).

Future directions

The identification of SWEET proteins as sugar facilitators has shed light on questions raised decades ago. For example, SWEETs may function in exporting sucrose into the phloem apoplasm from the seed coat to support filial tissue growth. SWEET may also facilitate sucrose efflux along the phloem path to nourish lateral tissues which also act as transient storage reserves in the stem and root. Furthermore, the identification of SWEET proteins also raises many new questions. How is sucrose efflux regulated? The analysis of the regulation of SWEETs at the transcriptional and translational levels may help to gain new insights into this process. Is SWEET-mediated sucrose efflux from putative phloem parenchyma cells coordinated with SUT1-mediated sucrose uptake into the SE/CC complex, and if so, how? Does this coordination maximize efficient phloem translocation between source and sink tissues? Although the mechanism of phloem loading in rice is still unclear, rice is believed to primarily utilize the passive symplasmic phloem-loading pathway, with possible contributions from the apoplasmic pathway (Eom et al., 2012). Thus, the questions of ‘if’ and ‘when’ OsSWEET11, the sucrose efflux carrier most likely to be responsible for exporting sucrose into the phloem apoplasm, needs to be active for phloem loading in rice remain to be answered.

What is the mechanism by which the various SWEETs mediate the transport of various sugars? Is this mechanism related to the structure of SWEET proteins? What relationship is there between the structure and function of SWEETs? To address these questions, crystal structures of SWEET proteins are needed. Many pathogens other than Xoo strains can alter the expression of various SWEET genes in Arabidopsis. How do these interactions occur between hosts and pathogens even without the type III secretion system? Attempting to answer these questions will enrich our knowledge of the physiological function of SWEET proteins in plants.

Increasing the partitioning of photoassimilate from photosynthesizing leaves to harvestable organs may be important in the context of the development of engineering strategies that aim to increase crop yield potential, as there is evidence that maize (Zea mays) displays a dramatic reduction in seed dry weight, which contributes significantly to seed yield, when assimilate availability is strongly reduced during seed filling (Borras et al., 2004). Increasing the availability of sucrose in the collection phloem of source leaves for translocation to harvestable tissues by genetic manipulation of carbon partitioning could be a major future strategy. However, this availability of sucrose is determined by several factors, among which translocating assimilates to the phloem region and transporting sucrose into the SE/CC complex are key. Ectopic expression of the potato sucrose transporter StSUT1 specifically in cotyledon storage parenchyma cells leads to a substantial increase in the sucrose uptake and growth rates of developing pea cotyledons, but, unexpectedly, no changes in seed dry weight have been detected. The underlying reason for this is still unclear; however, strategically manipulating the expression of SWEET or together with SUT expression in a tissue- or cell-specific manner to improve carbon partitioning and increase crop yield is an exciting prospect.


I am very grateful to Wolf B. Frommer for support, advice and encouragement during my postdoctoral research and for comments on this manuscript. I am also very grateful to Alexander Jones and Davide Sosso for critical reading and editing of the manuscript. This work was made possible by grants from the Division of Chemical Sciences, Geosciences and Biosciences, Office of Basic Energy Sciences at the US Department of Energy (DOE) under grant number DE-FG02-04ER15542 (to W. B. Frommer).