SWEET17, a facilitative transporter, mediates fructose transport across the tonoplast of Arabidopsis roots and leaves.

Fructose (Fru) is a major storage form of sugars found in vacuoles, yet the molecular regulation of vacuolar Fru transport is poorly studied. Although SWEET17 (for SUGARS WILL EVENTUALLY BE EXPORTED TRANSPORTERS17) has been characterized as a vacuolar Fru exporter in leaves, its expression in leaves is low. Here, RNA analysis and SWEET17-β-glucuronidase/-GREEN FLUORESCENT PROTEIN fusions expressed in Arabidopsis (Arabidopsis thaliana) reveal that SWEET17 is highly expressed in the cortex of roots and localizes to the tonoplast of root cells. Expression of SWEET17 in roots was inducible by Fru and darkness, treatments that activate accumulation and release of vacuolar Fru, respectively. Mutation and ectopic expression of SWEET17 led to increased and decreased root growth in the presence of Fru, respectively. Overexpression of SWEET17 specifically reduced the Fru content in leaves by 80% during cold stress. These results intimate that SWEET17 functions as a Fru-specific uniporter on the root tonoplast. Vacuoles overexpressing SWEET17 showed increased [14C]Fru uptake compared with the wild type. SWEET17-mediated Fru uptake was insensitive to ATP or treatment with NH4Cl or carbonyl cyanide m-chlorophenyl hydrazone, indicating that SWEET17 functions as an energy-independent facilitative carrier. The Arabidopsis genome contains a close paralog of SWEET17 in clade IV, SWEET16. The predominant expression of SWEET16 in root vacuoles and reduced root growth of mutants under Fru excess indicate that SWEET16 also functions as a vacuolar transporter in roots. We propose that in addition to a role in leaves, SWEET17 plays a key role in facilitating bidirectional Fru transport across the tonoplast of roots in response to metabolic demand to maintain cytosolic Fru homeostasis.

and Valich, 1988) as well as for the production of other compounds (e.g. 23 osmoprotectants). Sugars are known to regulate photosynthesis; therefore the release of 24 sugars from vacuoles could be important for modulating photosynthesis (Kaiser and 25 Heber, 1984). Moreover, vacuole-derived sugars are commercially used to produce 26 biofuels, such as ethanol, from sugarcane. Knowledge of the key transporters involved in 27 sugar exchange between the vacuole and cytoplasm is thus relevant in the context of 28 bioenergy (Grennan and Gragg, 2009). 29 To facilitate the exchange of sugars across the tonoplast, plant vacuoles are 30 equipped with a multitude of transporters (Neuhaus, 2007;Etxeberria et al., 2012;31 Martinoia et. al., 2012) comprising both facilitated diffusion and active transport systems 1 of vacuolar sugars (Martinoia et al., 2000). Typically, Suc is actively imported into 2 vacuoles by the tonoplast monosaccharide transporter (AtTMT1/2) (Schulz et al., 2011) 3 and exported by the SUT4 family (AtSUC4, OsSUT2) (Eom et al., 2011;Payyavula et al., 4 2011;Schulz et al., 2011). Two H + -dependent sugar antiporters, the vacuolar Glc 5 transporter (AtVGT1) (Aluri and Buttner, 2007) and AtTMT1 (Wormit et al., 2006), 6 mediate Glc uptake across the tonoplast to promote carbohydrate accumulation in 7 Arabidopsis. The Early Responsive to Dehydration-Like 6 (AtERDL6) protein has been 8 shown to export vacuolar Glc into the cytosol (Poschet et al., 2011), likely via an energy-9 independent diffusion mechanism (Yamada et al., 2010). Defects in these vacuolar sugar 10 transporters alter carbohydrate allocation and inhibit plant growth and seed yield (Aluri 11 and Buttner, 2007;Wingenter et al., 2010;Eom et al., 2011;Poschet et al., 2011). 12 In contrast to numerous studies on vacuolar transport of Suc and Glc, limited efforts 13 had been devoted to the molecular mechanism of vacuolar Frc transport even though Frc 14 is predominantly located in vacuoles (Martinoia et al., 1987;Voitsekhovskaja et al., 2006;15 Tohge et al., 2011). Vacuolar Frc is important in turgor pressure regulation (Pontis, 1989), 16 anti-oxidative defense (Bogdanovi et al., 2008), and signal transduction during early 17 seedling development (Cho and Yoo, 2011;Li et al., 2011). Thus, control over Frc 18 transport across the tonoplast is thought to be important for plant growth and 19 development. One vacuolar Glc transporter from the Arabidopsis MST family, AtVGT1,20 has been reported to mediate low affinity Frc uptake when expressed in yeast vacuoles 21 (Aluri and Buttner, 2007). Yet, the high vacuolar uptake activity to Frc intimates the 22 existence of additional high capacity Frc-specific vacuolar transporters (Thom et al., 23 1982). Recently, quantitative mapping of a quantitative trait locus for Frc content led to 24 the identification of a Frc-specific vacuolar transporter, SWEET17 (Chardon et al., 2013). 25 SWEET17 belongs to the recently identified SWEET (PFAM:PF03083) super 26 family, which contains 17 members in Arabidopsis and about 21 homologs in rice (Chen 27 et al., 2010;Frommer et al., 2013;Xuan et al., 2013). Based on amino acid similarity, 28 plant SWEET proteins were grouped into four subclades exhibiting 27 to 80% identity 29 (Chen et al., 2010). Transport assays using radiotracers in Xenopus oocytes or sugar 30 nanosensors in mammalian cells showed that most SWEET transporters are plasma 31 transport in vivo were still elusive. SWEET17 and 16 from Arabidopsis belong to the 23 clade IV SWEETs. Whether clade IV proteins both transport vacuolar sugars in planta 24 deserves further studies. 25 Here, we used GUS/GFP fusions to reveal the root-dominant expression and 26 vacuolar localization of the SWEET17 protein in vivo and its regulation in response to 27 intracellular Frc. Phenotypes of mutants and overexpressors were consistent with a role 28 of SWEET17 in bi-directional Frc transport across root vacuoles. The uniport feature of 29 SWEET17 transport in vivo was further confirmed using isolated mesophyll vacuoles. 30 Using the same approaches, SWEET16 was also shown to function in vacuolar sugar 31 transport in roots. Our work, performed in parallel, provide direct evidences to show the 1 Frc-specific uniport activity of SWEET17 in planta proposed by a recent study (Chardon 2 et al., 2013) and presents functional analyses to uncover important roles of these vacuolar 3 transporters in maintain intracellular Frc homeostasis in sink cells, roots. 4 Results 5 SWEET17 proteins are highly expressed in roots 6 A very recent report had indicated that SWEET17 (At4g15920) functions as a Frc 7 exporter in leaf vacuoles. However, SWEET17 expression appeared to be very low in 8 leaves (Chardon et al., 2013), indicating that SWEET17 may predominantly function in 9 other sink organs than leaves under specific environmental conditions. A quantitative 10 reverse transcription (qRT)-PCR analysis revealed that SWEET17 mRNA was expressed 11 to high levels in roots of 2-week-old seedlings (Fig. 1). In soil-grown mature plants, some 12 aerial organs, i.e. stems, flowers and siliques also accumulated high levels of SWEET17 13 transcripts. By contrast, expression of SWEET17 was comparatively low in both young 14 and mature leaves (Fig. 1). The high levels of SWEET17 transcripts in roots observed 15 here correlated well with the steady state expression profile from the Arabidopsis eFP 16 Browser (Supplemental Fig. S1A) (Winter et al., 2007)  translational fusions were analyzed. We generated transgenic Arabidopsis plants 20 expressing a C-terminal translational GUS gene fusion of SWEET17 driven by the native 21 SWEET17 promoter (SWEET17-GUS). In particular, the full length of SWEET17 gene 22 containing all introns was used to observe the genuine expression of the protein in planta. 23 In 7-d-old transgenic seedlings, SWEET17-GUS fusion proteins were mainly found in 24 cotyledons and roots ( Fig. 2A). A similar expression pattern was also observed in 2-25 week-old seedlings (Fig. 2B), where, however, much lower GUS staining was seen in 26 aerial tissues. The expression pattern of SWEET17 proteins was also consistent with the 27 expression pattern analyzed by a GUS reporter driven by the SWEET17 promoter 28 (Supplemental Fig. S1C). In roots, SWEET17 was predominantly expressed in root tips 29 ( Fig. 2C) and mature regions of roots (Fig. 2D). Only low expression was observed in the 30 elongation zone of roots (Fig. 2C). Three independent reporter lines showed comparable 1 patterns of GUS staining (data not shown). Handsections of mature roots histochemically 2 stained for GUS activity further demonstrated that SWEET17 predominantly 3 accumulated in the root cortex (Fig. 2E). The cell-type specific expression was 4 comparable with that of root array data from the Arabidopsis eFP Browser (Supplemental 5 Fig. S2, A and B) and the Translatome database (Supplemental Fig. S2C). In soil-grown 6 mature plants, expression of SWEET17-GUS was consistently observed to be high in 7 roots and low in aerial tissues, such as leaves, stems and flowers (Supplemental Fig. 8 S3A). When the reaction time of GUS staining was doubled (to 4 h), low and patchy 9 expression of the SWEET17-GUS fusion proteins was observed in mature leaves (Fig.  10 2F). After extended staining, some GUS activity was also observed in the vascular tissues 11 of flowers (Fig. 2G) and the bottom of siliques (Fig. 2H), but not in seeds (Fig. 2I). 12

Expression of SWEET17 is regulated by Frc in roots 13
Since expression of some sugar transporters is modulated by altering sugar contents 14 (Williams et al., 2000), we investigated if altered sugar levels would affect the spatial 15 pattern of expression of SWEET17. Transgenic Arabidopsis seedlings expressing 16 SWEET17-GUS fusion proteins were grown on media supplemented with 1% Suc for 5 d 17 then transferred to media without or with 1% Suc, Glc, or Frc for 2 d. Histochemical 18 staining for GUS activity showed that SWEET17 accumulation was similar in the 19 presence of Suc or Glc (Supplemental Fig. S3B), but highly induced by 1% Frc in 20 elongation regions of roots compared to control conditions (Fig. 2,J and K). To address 21 whether SWEET17 expression responds to a low intracellular sugar status, seedlings 22 expressing SWEET17-GUS fusion proteins were grown on media supplemented with 1% 23 Suc under 16 h daylength for 5 d then transferred to sugar-free media in the dark for 24 additional 2 d. The extended dark period leads to sugar starvation in plant cells (Usadel et 25 al., 2008). Interestingly, accumulation of SWEET17-GUS fusion proteins in elongation 26 regions of roots was also highly induced by darkness (Fig. 2, J and L). The same Frc and 27 sugar-starvation inducible patterns in roots were observed at the transcriptional level 28 using a SWEET17 promoter:GUS fusion (Supplemental Fig. S3C). Weak induction of 29 SWEET17-GUS expression in leaves was also observed upon exposure to darkness or by 30 combining darkness with cold stress (Supplemental Fig. S4). These observations were 31 consistent with a predicted role of SWEET17 as a Frc uniporter for uptake of excess 1 cytosolic Frc or for release the stored Frc across the vacuolar membrane to maintain 2 homeostasis. 3

SWEET17 proteins in root vacuoles 4
To investigate whether SWEET17 is targeted to the tonoplast also in roots, we 5 generated transgenic Arabidopsis plants expressing a C-terminal translational GFP fusion 6 of SWEET17 (SWEET17-GFP) driven by the native SWEET17 promoter. To allow 7 possible transcriptional regulation, the full length genomic SWEET17 gene was used. 8 Confocal images of intact roots from homozygous transformants revealed that the 9 fluorescence derived from SWEET17-GFP fusion proteins was predominantly present at 10 the tonoplast of root tips (Fig. 3A). The vacuolar localization was evidenced by 11 fluorescence surrounding small pre-mature vacuoles located inside the plasma membrane 12 labeled by FM4-64 (Fig. 3, B and C) (Wayne, 2009). The vacuolar pattern was also 13 observed in mature regions of roots as shown by fluorescence lining the inner side of the 14 nucleus and plasma membrane (Fig. 3, D-F) (Wayne, 2009). The fluorescence derived 15 from SWEET17-GFP fusion proteins in the cytosol or at the plasma membrane was not 16 significant. Tonoplast-specific localization of SWEET17-GFP was also observed in 17 mesophyll protoplasts (Supplemental Fig. S5). Despite overall low levels of fluorescence 18 in leaf tissues, SWEET17-GFP fluorescence in Arabidopsis protoplasts was clearly 19 concaved by chloroplast autofluorescence (Supplemental Fig. S5A) and not detected at 20 the plasma membrane (Supplemental Fig. S5B). These observations indicate that 21 SWEET17 locates to and functions predominantly at the tonoplast of roots. 22

Import activity of SWEET17 confers tolerance to Frc inhibition 23
High levels of cytosolic Frc inhibit root growth and arrest seedling development 24 (Bhagyalakshmi et al., 2004;Cho and Yoo, 2011). If SWEET17 could act as a vacuolar 25 importer to store excess Frc in roots, as suggested by the expression pattern ( Fig. 2 and 3), 26 altered expression of SWEET17 may affect intracellular Frc allocation and thereby affect 27 sensitivity of roots to Frc. To test this hypothesis, Arabidopsis knockout mutants, 28 sweet17-1 and sweet17-2 with T-DNAs inserted in the SWEET17 gene were obtained 29 (Chardon et al., 2013). While the relative root growth of the segregating wildtype (Col-30 TDNA) and mutants (sweet17-1 and sweet17-2) was similar when grown on media with 1 or without 1% Suc, root growth of sweet17-1 and sweet17-2 seedlings was significantly 2 more sensitive to 1 to 2% of Frc compared to the wildtype (Fig. 4A). SWEET17 thus 3 appears to be, at least partially, necessary for Frc tolerance. We therefore tested whether 4 SWEET17 is also sufficient for Frc tolerance in roots. We generated transgenic 5 Arabidopsis plants overexpressing the full length genomic SWEET17 gene driven by the 6 constitutive cauliflower mosaic virus 35S promoter. In 7-d-old seedlings,  analysis showed that levels of SWEET17 transcripts in homozygous transgenic plants 8 were highly increased in overexpressing lines, 35S:SWEET17-1, -6 and -2, compared to 9 plants transformed with the empty vector (Col-Vector; Supplemental Fig. S6A). In 10 contrast to the increased sensitivity of root growth in sweet17 mutants (Fig. 4A), three 11 independent overexpressor lines (35S:SWEET17-1, -6 and -2) showed enhanced tolerance 12 compared to wildtype plants (Col-Vector-1 and -2) in the presence of 0.1 to 1% Frc (Fig. 13 4B). Excess sugars inhibit seed germination (Dekkers et al., 2004). To examine if 14 SWEET17 also contributes to Frc translocation during early seedling development, we 15 compared germination rates of mutants and overexpressors subjected to excess Frc. 16 However, after 2 to 4 d of incubation, we did not observe significant differences between 17 all lines tested (Supplemental Fig. S6, B and C; data not shown). 18

Overexpressing SWEET17 decreases Frc accumulation in leaves 19
The induction of SWEET17 expression by energy starvation (Fig. 2, J and L) 20 indicates that SWEET17 may be able not only to import, but also export Frc stored in the 21 vacuole into the cytosol when there is a metabolic demand. To examine how the 22 increased export activity of SWEET17 proteins affects the capacity of vacuoles to store 23 sugars, we determined sugar contents of leaves of SWEET17 overexpressors under 24 standard growth conditions and in response to cold stress (4 o C) for 1 week. Cold stress 25 has been shown to induce Frc accumulation in vacuoles by 2 to 10 fold within 24 h 26 (Wormit et al., 2006). Under standard conditions, no dramatic differences in leaf sugar 27 levels were observed when comparing wildtypes and overexpressors (Supplemental compared to those isolated from control leaves (Col-Vector, Fig. 6A). By comparison, the 13 uptake activities for Glc or Suc were not significantly increased (Supplemental Fig. S8). 14 These results show that SWEET17 can import Frc specifically into vacuoles along a 15 concentration gradient in vivo. 16 To confirm if SWEET17 acts as a uniporter as observed from phenotypes ( vacuoles from plants overexpressing SWEET17 had taken up 2.5 times more Frc uptake 22 than those expressing the empty vector (Fig. 6B). The absence of ATP as well as 23 treatments of NH 4 Cl or CCCP, which disrupt the preexisting proton gradient, had no 24 impact on the transport activity (Fig. 6B). 25 To investigate the affinity of the vacuolar Frc transporter, we performed a 26 concentration-dependent uptake experiment. Saturation of Frc uptake was not observed at 27 a concentration even up to 20 mM (Fig. 6C). The deduced K m and V max values of 28 SWEET17 activity were 25.6 mM and 479.8 pmol Frc µl vacuole -1 min -1 , respectively, 29 demonstrating that SWEET17 is a low affinity Frc transporter. 30 1

Both clade IV SWEETs function as vacuolar transporters in roots 2
The Arabidopsis genome contains a close homolog of SWEET17 in clade IV, named 3 SWEET16. In a recent study, SWEET16 (At3g16690) was also shown to function as a 4 vacuolar sugar facilitator in vascular parenchyma cells (Klemens et al., 2013). However, 5 the expression of SWEET16, measured as GUS activity driven by the SWEET16 promoter 6 is relatively low in most tissues, including roots (Klemens et al., 2013). Since SWEET16 7 and 17 proteins share 70% amino acid identity (Supplemental Fig. S9A), we suspected 8 that SWEET16 may play a partially redundant function with SWEET17 in roots under 9 certain environmental stimuli. To address this hypothesis, we performed similar 10 experiments as for SWEET17. Interestingly, the (qRT)-PCR analysis showed 11 predominant expression of SWEET16 mRNA in roots compared to all aerial organs, such 12 as leaves, stems and flowers in both young seedlings and mature plants ( SWEET genes; a few other SWEET genes were expressed at low levels in roots (Fig. 7). 20 The pattern was consistent with the expression profile derived from the AtGeneExpress 21 database (Supplemental Fig. S9B). 22 To analyze the tissue-specific expression of SWEET16 proteins, we generated 23 transgenic Arabidopsis expressing C-terminal translational fusions to GUS and GFP 24 under the control of the native SWEET16 promoter (SWEET16-GUS/GFP), again using 25 the full length SWEET16 genomic gene. Similar to SWEET17, both SWEET16-GUS 26 fusion proteins and promoter activity were mainly found in roots (Supplemental To functionally characterize SWEET16 in planta, we obtained Arabidopsis sweet16 T-4 DNA insertion knockout mutants (Supplemental Fig. S11A) and generated a sweet17-5 1/sweet16-1 double mutant (Supplemental Fig. S11B) as well as SWEET16 6 overexpressor lines (Supplemental Fig. S11C). Vacuoles isolated from sweet17-7 1/sweet16-1 showed a markedly decrease in Frc uptake activity (Supplemental Fig. 8 S12A). Although overexpression of SWEET16 in leaves did not increase the Frc transport 9 activity (Supplemental Fig. S12B), Frc content in leaves was significantly decreased in 10 response to cold stress (Supplemental Fig. S7B) compared to standard growth conditions 11 (Supplemental Fig. S7A). These results imply that SWEET16 is probably involved in Frc 12 transport in planta, although at a much lower degree than SWEET17. Consistent with this 13 observation, root growth of sweet16-1 and sweet16-2 seedlings was significantly more 14 sensitive to 0.5 to 2% of Frc than the wildtype (Fig. 8A). However, compared to sweet16 15 single mutants, the sweet17-1/sweet16-1 double mutant did not show enhanced sensitivity 16 to excess Frc. In addition, overexpressing SWEET16 significantly enhanced root growth 17 at 0.5 to 2% of Frc (Fig. 8B). Organ-specific post-translational regulation may modulate SWEET17 function as been 4 observed for other sugar transporters (Krügel and Kühn, 2013). The discrepancy between 5 our observations using a whole-gene construct and the previous report using a promoter-6 GUS construct (Chardon et al., 2013) may be due to intragenic regulatory elements 7 within introns or exons that regulate gene expression. That has been observed in the 8 Arabidopsis sucrose transporters, AtSUC1 and AtSUC9 (Sivitz et al., 2007) and well 9 described in some nutrient transporters, such as AtAMT1 (Yuan et al., 2007) and 10 AtNRT2 (Laugier et al., 2012). Further studies will be required to discover the regulatory 11 mechanism. Nevertheless, these observations point out that expression of SWEET17 is 12 tightly regulated to cope with the dynamics of cellular sugar homeostasis ( Frc uptake activity of leaf-derived vacuoles from sweet17-1/sweet16-1 demonstrates that 6 despite the low levels of protein, SWEET17 is likely function in leaf vacuoles 7 (Supplemental Fig. S12A). Alternatively, the reduced import activity of Frc in roots of 8 sweet17 mutants may reduce their storage capacity and sink strength, which, in turn, can 9 reduce Suc unloading and ultimately increase sugar or carbohydrate accumulation in leaf 10 mesophyll cells (Ayre, 2011). 11

SWEET17 acts as a bi-directional Frc-specific facilitator 12
Transport assays using isolated leaf mesophyll vacuoles verified that SWEET17 is 13 an important vacuolar Frc-specific transporter in vivo ( Fig. 5; Supplemental Fig. S12A). 14 Although we cannot exclude the possibility that SWEET17 exhibits a minor transport 15 activity for other hexoses or disaccharides, such activities would be very low compared to 16 From their similarities in amino acids, expression patterns, and localizations we 1 expected that these two SWEETs would either function in a complementary way, e.g. one 2 transporting Glc, the other Frc or redundantly. Indeed, mutations in both SWEET17 and 3 SWEET16 dramatically reduced the Frc uptake activity of vacuoles to Frc in vivo 4 (Supplemental Fig. S12A). A reduced accumulation of Frc was also observed in a 5 SWEET16 overexpressor line under cold stress (Supplemental Fig. S7B). However, in 6 contrast to the previous study performed in an oocyte expressing system (Klemens et al.,7 2013), we did not detect any transport activity of SWEET16 to sugars (Supplemental Fig. 8 S12B; data not shown). It is possible that the uptake mediated by SWEET16 on the 9 tonoplast was below the detection limit due to a high background ( Fig. 6B; data not 10 shown) (Klemens et al., 2013) than SWEET17 (K m lower than 10 mM measured using 11 an oocyte system) (Chardon et al., 2013). Nevertheless, similar to SWEET17, altered 12 sensitivity of the root growth to excess Frc was observed in sweet16 mutants and 13 overexpressors (Fig. 8). However, a loss-of-function of both sweet16 and sweet17 double 14 mutant did not further reduce the root tolerance to excess Frc compared to sweet16 single 15 mutant (Fig. 8A). These results indicate that SWEET17 and 16 may function in 16 independent pathways by exchanging sugars across the tonoplast of root cells. Given that 17 SWEET17 is expressed to a much higher level in roots ( Fig. 1; Supplemental Fig. S1), it 18 is likely that SWEET17 is the dominant Frc transporter on the root tonoplast. 19 In summary, our work provides a functional characterization of the vacuolar Frc-20 specific uniporter SWEET17. Based on these results, we propose that when 21 photosynthesis is very active, Suc is allocated from the leaf to the root where Suc is 22 hydrolyzed to Glc and Frc that are imported into vacuoles for storage by hexose 23 transporters and SWEET17, respectively. Depending on the metabolic stage of the root 24 cells, defect of SWEET17 will either reduce the vacuolar loading or unloading. In this 25 case, excess of Frc in the cytosol will lead to growth retardation due to toxic effects of 26 Frc, while under energy-limiting conditions the plant is not able to use all the 27 carbohydrate reserves. During these processes, SWEET16 may also contribute to sugar 28 compartmentation across the root vacuole independently from SWEET17. 29 to 2% Suc and cultured for 16 d before analysis. The MS or 1/2MS media used in this 20 study included full or half strength of MS salt, respectively, 0.05 % (w/v) MES, and 1.5% 21 agar for solid medium (adjust to pH 5.7 with KOH). 22

Generation of plants expressing GUS and GFP reporter genes 23
For P SWEET17 :GUS and P SWEET16 :GUS transcriptional constructs, SWEET17 and 16 24 promoter fragments containing the 5' UTR region were amplified from Arabidopsis 25 genomic DNA with Phusion polymerase (New England Biolabs, MA, USA) and specific 26 primers (5-PSWT17-BP and 3-PSWT17-BP for P SWEET17 ; 5-PSWT16-BP and 3-27 PSWT16-BP for P SWEET16 ). The resulting 2746 bp and 1521 fragments, respectively, were 28 first cloned into pDONR221-f1 and then transferred to a binary vector, pWUGW that 29 contains a GUS reporter gene, using Gateway technology (Invitrogen, CA, USA). For 30 SWEET17-GUS/GFP and SWEET16-GUS/GFP translational constructs, the promoter 1 fragments were first amplified from the pDONR221-f1 clones mentioned above and 2 cloned into pGEM-T Easy (Promega, WI, USA). The promoter fragments were then 3 cloned via PstI and KpnI sites into pMDC32-SWEET17 and pMDC32-SWEET16 4 constructs (refer to the section of "Overexpression of SWEET17 and 16 in Arabidopsis") 5 to replace the 35S promoter in those plasmids. The full P SWEET17 -SWEET17 and P SWEET16 -6 SWEET16 genomic fragments were then amplified using Phusion polymerase, cloned 7 into pGEM-T Easy, digested, and subcloned into SacII and PstI sites of pUTKan or 8 pGTkan, harboring the GUS or GFP reporter gene, respectively. All constructs were 9 transformed into wildtype Arabidopsis plants using Agrobacterium tumefaciens strain 10 C58 PGV3850 and flower dipping method (Clough and Bent, 1998). Transformants were 11 identified on 1/2MS medium containing 50 µg mL -1 Kanamycin. For analysis of reporter 12 gene expression, plants from six independent transgenic lines were examined for each 13 construct. Patterns of gene expression were consistent within a construct and 14 representative 3 homozygous lines were further analyzed in other experiments. 15 Sequences of primers were listed in Supplemental Table S1. for 5 min before imaging in order to visualize the plasma membrane. GFP was visualized 29 by excitation with an argon laser at 488 nm and spectral detector set between 500 and 545 30 nm for the emission. The red fluorescence of FM4-64 was visualized by excitation with 1 an argon laser at 561 nm and spectral detector set between 566 and 585 nm for the 2 emission. 3

Overexpression of SWEET17 and 16 in Arabidopsis 21
To express SWEET17 and SWEET16 under the control of the 35S promoter, full 22 length SWEET17/16 genomic sequences including all introns were amplified using 23 Phusion polymerase with gene specific primers (SWT17-5-UTR-BP and SWT17-3-UTR-24 BP for SWEET17; SWT16-5-UTR-BP and SWT16-3-UTR-BP for SWEET16). The 25 resulting 2856 bp and 2119 bp fragments were cloned into pDONR221-f1 and then 26 transferred to a binary vector pMDC32 (Curtis and Grossniklaus, 2003) via Gateway 27 technology. The resulting plasmids, pMDC32-SWEET17, pMDC32-SWEET16, and the 28 corresponding empty vector, pMDC32 without the Gateway cassette, were transformed 29 into wildtype Arabidopsis plants using the Agrobacterium tumefaciens strain C58 30 PGV3850 and flower dipping method (Clough and Bent, 1998). Transformants were 1 identified on 1/2MS medium containing 25 µg mL -1 hygromycin B. Six single-copy T2 2 lines were obtained and three homozygous lines of highest expression were used in this 3 study. Sequences of primers were listed in Supplemental Table S1. 4 qPCR and RT-PCR analysis 5 For aerial tissues in mature plants, total mRNA was isolated from leaves, stems, 6 flowers and siliques of 7 to 8-week-old soil grown plants. For mRNA from young 7 seedlings, shoots and roots of 2-week-old plants or 7-d-old whole seedlings grown on 8 1/2MS agar were harvested for extraction. To analyze expression in mature roots, total 9 mRNA was isolated from roots of 23-d-old liquid cultured seedlings. Total mRNA was 10 isolated using TRIsure (Bioline, London, UK) reagent or RNeasy mini kit (Qiagen,11 Hilden, Germany) as instructed by the manufacture. The resulting cDNA produced by 12 MMLV (Qiagen, Hilden, Germany) was diluted and used as the template and subjected to 13 25 to 30 cycles of PCR reaction (94 o C for 30 s, 55 o C for 30 s, and 72 o C for 40 s) using 14 Taq DNA polymerase and a pair of gene-specific primers as indicated. Amplification of 15 an Actin cDNA (Actin2, At3G18780) using Act2-F and Act2-R primers (Supplemental 16 instructions on a 7300 PCR system (Applied Biosystems, CA, USA). The relative 21 expression level was determined by comparing with the expression of Actin 2 (1000*2^-22 (Ct SWEET -Ct Actin2 ). The gene-specific primers used in qRT-PCR for 17 Arabidopsis 23 SWEET genes were listed in Supplemental Table S1. 24

Extraction and assay of soluble sugars 25
Ground, freeze-dried Arabidopsis rosette leaf material ( ~ 0.5 g FW) was extracted 26 in 800 µl ice-cold 0.7 M perchloric acid for 5 min with intermittent mixing, using a 27 Mixer Mill (Retsch, Haan, Germany). All the subsequent steps were carried out between NaOH; eluent B, 150 mM NaOH and 500 mM sodium acetate. At a flow rate of 0.5 mL 9 min -1 , the gradient was as follows: 0 to 7 min, 100% A and 0% B; 7 to 26.5 min, a 10 concave gradient to 20% A and 80% B (malto-oligosaccharide elution); 26.5 to 32 min, 11 hold at 20% A and 80% B (column wash step); 32 to 40 min, step to 100% A and 0% B 12 (column re-equilibration). Peaks were identified by co-elution with known malto-13 oligosaccharide standards. Peak areas were determined using Chromeleon software 14 (Dionex, Thermo Scientific, MA, USA). 15 HEPES, pH 7.2 adjusted with imidazole, 1 mg mL -1 BSA, 1 mM DTT) and 1 ml medium 6 C. After centrifugation for 8 min at 1300g (4°C), the vacuoles were recovered from the 7 interface between the middle and upper layer. Microscopic analysis indicated that 8 contamination with intact protoplasts was less than 3%. 9

Transport analysis with Arabidopsis vacuole 10
Transport experiments were performed using silicone oil centrifugation technique as 11 described previously (Song et al., 2010). The Michaelis-Menten nonlinear least-square 12 regression fits were calculated using the SSmicmen function without initial parameters 13 within the nls function of R 2.14.0 (www.R-project.org). Heim-Voegtlin (grant_PMPDP3_139645) to D.S. We thank Dr. Bo Burla for performing 20 the curve fitting. We also thank Dr. Tong Seung Tseng and Dr. Kate Dreher at Carnegie 21 Institution for Science for constructive comments on the manuscript. 22

SUPPLEMENTAL MATERIALS 1
Supplemental Table S1. Specific primers used in this study.  Arabidopsis seeds were germinated and grown on media supplemented with 1% sucrose 5 for 5 d, then seedlings were transferred to media containing various concentrations of Frc. 6 Relative growth of primary roots was measured after 6 d of treatments. A, Comparison of 7 root growth in two independent sweet17 mutant lines and the wildtype that was identified 8 from the segregating mutant population (Col-TDNA). B, Comparison of root growth in 9 three independent SWEET17 overexpressor lines (35S:SWEET17-1, -6 and -2) and the 10 wildtype that was transformed with the empty vector (Col-Vector-1 and -2). The data 11 presented are means ± SE (n = 8). Significant differences from the wildtype were 12 determined by Student's t test indicated by asterisks: * P < 0.05, ** P < 0.01. 13 Results are means ± SD (n = 4). In B-C, the uptake level was determined after 3 to 20 2 min of incubation in 0.2 mM (B) or indicated (C) 14 C-labeled Frc. Significant differences 3 were determined by Student's t test indicated by asterisks: ** P < 0.01. 4 Arabidopsis seeds were germinated and grown on media supplemented with 1% Suc for 5 d, then seedlings were transferred to media containing various concentrations of Frc.
Relative growth of primary roots was measured after 6 d of treatments. A, Comparison of root growth in two independent sweet17 mutant lines and the wildtype that was identified from the segregating mutant population (Col-TDNA). B, Comparison of root growth in three independent SWEET17 overexpressor lines (35S:SWEET17-1, -6 and -2) and the wildtype that was transformed with the empty vector (Col-Vector-1 and -2). The data presented are means ± SE (n = 8). Significant differences from the wildtype were determined by Student's t test indicated by asterisks: * P < 0.05, ** P < 0.01. Arabidopsis seeds were germinated and grown on media supplemented with 1% Suc for 5 d, then seedlings were transferred to media containing various concentrations of Frc.
Relative growth of primary roots were measured after 6 d of treatments. A, Comparison of root growth in two independent sweet16 mutant lines and the wildtype that was identified from the segregating mutant population (Col-TDNA). B, Comparison of root growth in two independent SWEET16 overexpressor lines (35S:SWEET16-2, -3) and the wildtype that was transformed with the empty vector (Col-Vector-1, -2). The data presented are means ± SE (n = 8). Significant differences from the wildtype were determined by Student's t test indicated by asterisks: * P < 0.05, ** P < 0.01.  Seeds were germinated on media supplemented with various concentrations of Frc. Germination rates were calculated after 2 d between two independent sweet17 mutants, the double mutant sweet17-1/sweet16-1 and the wildtype that was identified from the segregating mutant population (Col-TDNA). The same experiment was also performed using two independent SWEET17 overexpressors (35S:SWEET17-1 and -6) and the wildtype that was transformed with the empty vector (Col-Vector-1). The data presented are means ± SE of four independent experiments. Significant differences from the wild type were determined by Student's t test. Results are means ± SE (n = 4). Significant differences from the wildtype (Col-Vector) were determined by Student's t test indicated by asterisks: * P < 0.05, ** P < 0.01. In (B) and (C), total mRNA was isolated from 7-d-old seedlings and the resulted cDNA products were used for amplification with primers for SWEET17 and 16 as indicated. Expression of Actin 2 was used as a loading control.