|Home | About | Journals | Submit | Contact Us | Français|
Sugar efflux transporters are essential for the maintenance of animal blood glucose levels, plant nectar production, and plant seed and pollen development. Despite broad biological importance, the identity of sugar efflux transporters has remained elusive. Using optical glucose sensors, we identified a new class of sugar transporters, named SWEETs, and show that at least six out of seventeen Arabidopsis, two out of over twenty rice and two out of seven homologues in Caenorhabditis elegans, and the single copy human protein, mediate glucose transport. Arabidopsis SWEET8 is essential for pollen viability, and the rice homologues SWEET11 and SWEET14 are specifically exploited by bacterial pathogens for virulence by means of direct binding of a bacterial effector to the SWEET promoter. Bacterial symbionts and fungal and bacterial pathogens induce the expression of different SWEET genes, indicating that the sugar efflux function of SWEET transporters is probably targeted by pathogens and symbionts for nutritional gain. The metazoan homologues may be involved in sugar efflux from intestinal, liver, epididymis and mammary cells.
The molecular nature of cellular sugar efflux in both plants and animals is unknown despite the fact that sugar efflux is an essential component for cellular exchange of carbon and energy in multicellular organisms1–4. Sugar efflux from the tapetum or transmitting tract of the style, for example, fuels pollen development and pollen tube growth5. Flowers secrete sugars for nectar production to attract pollinators, and plants secrete carbohydrates into the rhizosphere, potentially to feed beneficial microorganisms6. Sugar efflux carriers are required at other sites, including mesophyll in leaves and the seed coat7. In mammals, glucose efflux from liver is crucial for the maintenance of blood glucose levels2.
The primary goal of pathogens is to access nutrients from their hosts for reproduction. Phytopathogenic bacteria in the genera Pseudomonas and Xanthomonas can live in the intercellular space (apoplasm) of plants, where they acquire carbohydrates for energy and carbon8. Successful pathogens probably co-opt nutrient efflux mechanisms of the host to redirect nutrient flux9. Plants and pathogens engage in an evolutionary tug-of-war, in which the plant limits pathogen access to nutrients and initiates immune responses, whereas the pathogen evolves adaptive strategies to gain access to nutrients and suppress host immunity. Insight into mechanisms used by pathogens to alter plant immunity is emerging. However, the mechanisms that pathogens use to alter host physiology, notably efflux of sugars to support growth, are poorly understood. We hypothesize that sugar efflux transporters are co-opted by pathogens to supply nutrients9. This hypothesis is supported by studies of sugar transfer from wheat leaves to powdery mildew10–12. Pathogen glucose/H+ uptake transporters have been identified13; by contrast, plant sugar efflux mechanisms have remained elusive.
To identify new transporters potentially involved in glucose efflux, we screened genes encoding uncharacterized polytopic membrane proteins from the Arabidopsis membrane protein database Aramemnon14 using a new mammalian expression system15. Candidate genes were co-expressed with the high-sensitivity fluorescence resonance energy transfer (FRET) glucose sensor, FLIPglu600μΔ13V, in human HEK293T cells, which have low endogenous glucose uptake activity15,16. Among the genes tested, AtSWEET1 (AT1G21460) expression enabled HEK293T cells to accumulate glucose as detected by a glucose-induced negative FRET ratio change (Fig. 1a). To determine whether AtSWEET1 also mediates efflux from the cytosol, the FRET glucose sensor FLIPglu600μΔ13VER was expressed in the lumen of the endoplasmic reticulum (ER; Fig. 1b). Topologically, efflux across the plasma membrane from the cytoplasmic side is equivalent to efflux into the ER that is also initiated from the cytoplasmic side (Fig. 1c). The glucose-dependent response of the ER sensor demonstrates that AtSWEET1 can mediate both uptake across the plasma membrane and efflux into the ER. SWEET1 thus seems to function as a bidirectional uniporter/facilitator. The observed sugar uptake and efflux were not due to subcellular re-localization of the FRET sensors (Supplementary Fig. 1). A carboxy-terminal green fluorescent protein (GFP) fusion of AtSWEET1 was functional in cellular uptake and localized to the plasma membrane of HEK293T cells (Supplementary Fig. 2). AtSWEET1 carrying a premature stop codon at position 198 was non-functional (Supplementary Fig. 3). Induction of endogenous GLUT glucose transporters in AtSWEET1-expressing HEK293T cells was excluded based on insensitivity of uptake to the GLUT inhibitor cytochalasin B (Supplementary Fig. 4a, b) and lack of detectable changes of messenger RNA levels of known human GLUT and SGLT glucose transporters (Supplementary Fig. 4c). The transport function of AtSWEET1 was independently demonstrated by expression in a yeast mutant lacking all 18 hexose transporters17. AtSWEET1 enabled the yeast mutant to grow on glucose (Fig. 1d) and to accumulate intracellular glucose as determined with the FRET glucose sensor FLII12Pglu700μδ6 (ref. 18) (Fig. 1e). AtSWEET1 functions as a low-affinity glucose transporter (Michaelis constant (Km) ~9 mM; Fig. 1f). Consistent with a uniport transport mechanism, uptake was largely pH-independent (Supplementary Fig. 5). AtSWEET1 did not efficiently complement mannose, fructose and galactose uptake deficiencies of the mutant (Supplementary Fig. 6). Radiotracer experiments in Xenopus oocytes were used as another measure for AtSWEET1 sugar uptake activity (Fig. 1g). Direct proof for efflux activity was obtained by monitoring time-dependent release of [14C]glucose from oocytes after injection of radiotracer (Fig. 1h). In support of the function in cellular uptake and efflux, a constitutively expressed AtSWEET1–yellow fluorescent protein (YFP) fusion localized to the plasma membrane in Arabidopsis leaves (Fig. 1i). On the basis of expression studies, AtSWEET1 is highly expressed in Arabidopsis flowers, where the protein may supply nutrients to the gametophyte or nectaries (Supplementary Fig. 7). The biochemical properties of AtSWEET1 are markedly similar to an unidentified transport activity characterized in roots using FRET sensors19. However, AtSWEET1 expression in roots was low, implicating other AtSWEET paralogues for this function.
SWEET1 belongs to a novel transporter family (PFAM PF03083) with 17 members in Arabidopsis and ~21 in rice (Supplementary Fig. 8). SWEETs fall into four subclades (Supplementary Fig. 8a) with 27–80% identity (Supplementary Fig. 8b). SWEETs are small proteins predicted to form a pore from seven transmembrane helices (Supplementary Fig. 9). Modelling indicates that the structure results from an ancient duplication of a 3-transmembrane-helix-domain polypeptide (1–3 and 5–7) fused via transmembrane helix 4 in a 3+ 1+3 configuration (Fig. 1j).
The phenotypes of several sweet mutants have been described. AtSWEET1 (clade 1) is 42% identical to its paralogue AtSWEET8 (clade 2) (Supplementary Fig. 8b). AtSWEET8 (also called RPG1) is expressed in the tapetum20, and mutation of AtSWEET8 causes male sterility, compatible with a role in glucose efflux for pollen nutrition20.
AtSWEET1 and AtSWEET8 share 31% and 34% amino acid sequence identity with rice OsSWEET11 (also called Os8N3 or Xa13; here named OsSWEET11 based on phylogeny)21. Similar to AtSWEET8, OsSWEET11 contributes to pollen viability, as RNA interference to OsSWEET11 reduced starch content in pollen and caused male sterility in rice21 (Supplementary Fig. 8b). Expression studies indicate that the import function of SWEETs may also contribute to the nutrition of growing pollen tubes. Specifically, AtSWEET1 is expressed in hydrated pollen and both AtSWEET1 and AtSWEET8 are expressed highly in pollen tubes22. AtSWEET5 (also called VEX1) is expressed in mature, hydrated and germinating pollen and is found specifically in the vegetative cell of pollen grains, which may supply the generative cell with sugars23. Silencing of the clade 3 SWEET homologue NEC1 from petunia also triggered male sterility24. NEC1 is expressed in nectaries, and developmental regulation of NEC1 correlates inversely with nectarial starch content, indicating a second function for NEC1 in sugar secretion in nectaries25. Hexoses, in particular galactose, fructose and glucose, accumulate in senescent leaves, and SWEET members may also function in mobilization of carbohydrate during senescence26. The AtSWEET15 (also called SAG29) gene is induced ~22-fold in leaves during senescence27. Taken together, the SWEET sugar transporters probably supply carbohydrates to a variety of tissues in both monocotyledonous and dicotyledonous plants. Other SWEET members also function in glucose transport. For example, co-expression of AtSWEET8 with FRET sensors FLIPglu600μΔ13V or FLIPglu600μΔ13VER in HEK293T cells leads to glucose transport across both plasma and ER membranes (Supplementary Fig. 10), and AtSWEET8 complements the yeast glucose transport mutant (Fig. 1d and Supplementary Figs 5 and 11). At least four additional Arabidopsis SWEET genes (AtSWEET4, AtSWEET5, AtSWEET7, AtSWEET13) also function in glucose transport when expressed in yeast or HEK293T cells (Supplementary Figs 5, 11 and 12).
Many pathogens acquire glucose from their hosts9,11,13, thus pathogens may highjack host sugar efflux systems dedicated for plant development. We tested whether mRNA levels of Arabidopsis SWEET family members were altered by challenge with bacterial and fungal pathogens. Pseudomonas syringae pv. tomato strain DC3000 infection highly induced mRNA levels of AtSWEET4, AtSWEET5, AtSWEET7, AtSWEET8, AtSWEET10, AtSWEET12 and AtSWEET15 in Arabidopsis leaves (Fig. 2). In contrast, the DC3000 type III secretion mutant (ΔhrcU), which cannot inject type III effector proteins into the host and is compromised in pathogenicity, did not induce three of the seven AtSWEET genes, demonstrating that SWEET mRNA abundance is modulated in a type-III-dependent manner (Fig. 2b). The fungal powdery mildew pathogen Golovinomyces cichoracearum induced a different set of AtSWEET mRNAs, most prominently AtSWEET12 (Fig. 2a, c), and previous expression data have shown that infection with the fungal pathogen Botrytis cinerea induces expression of AtSWEET4, AtSWEET15 and AtSWEET17 (ref. 28). Pathogen-specific modulation of SWEET mRNA levels, therefore, probably alters sugar efflux at the site of infection, having an impact on pathogen growth and plant immunity.
OsSWEET11 underlies the dominant allele (Xa13) of the recessive resistance gene xa13 (refs 21, 29, 30). Susceptibility alleles of xa13 confer disease resistance against bacterial blight and have been isolated from geographically diverse rice accessions30. All alleles tested carry mutations in the promoter region of the OsSWEET11 gene and interfere with pathogen-specific induction of the gene21,31. RNA interference of rice OsSWEET11 confers resistance to the Xanthomonas oryzae pathovar oryzae (Xoo) strain PXO99A, which otherwise grows in the apoplasm and xylem of the host. OsSWEET11 may, therefore, supply sugars to the pathogen by a uniport mechanism as demonstrated for the Arabidopsis homologues (Fig. 1a–h and Supplementary Figs 10–12). Consistent with cellular import/efflux functions, OsSWEET11 localizes to the plasma membrane in rice callus29. OsSWEET11 is less efficiently targeted to the plasma membrane of HEK293T cells compared to AtSWEET1 (Fig. 3a). Nevertheless, weak uptake activity was observed in HEK293T cells and oocytes (Fig. 3b, c).
Infection of rice by Xoo PXO99A requires the bacterial type III effector gene pthXo1 (ref. 21). PthXo1 is a TAL (transcriptional activator-like) effector, which directly interacts with the OsSWEET11 promoter as shown by chromatin immune precipitation (Fig. 3d), as well as transient co-expression in Nicotiana benthamiana leaves (Supplementary Fig. 13)32,33. PthXo1 secreted by Xoo PXO99A specifically activates transcription of OsSWEET11 (ref. 21), presumably to induce sugar efflux to feed bacteria in the xylem and/or apoplasm (Fig. 4a). When pthXo1 is mutated (as in strain PXO99AME2), transcription of OsSWEET11 and pathogenicity are reduced21, consistent with a model of sugar supply limiting growth of the pathogen (Fig. 4b). If OsSWEET11 becomes unavailable owing to mutations in the TAL effector binding element of the OsSWEET11 promoter, or through RNA interference21,31, the sugar supply becomes limiting and the pathogen cannot grow efficiently (Fig. 4c). Indeed, ossweet11 (xa13) mutants are resistant to PXO99A (ref. 21). xa13-mediated resistance can be defeated by PXO99A expressing the alternative TAL effector gene avrXa7 (Fig. 4d), compatible with the most parsimonious hypothesis that another OsSWEET gene is co-opted by the pathogen to support bacterial growth21. Indeed, AvrXa7 activates the paralogue OsSWEET14 (ref. 33). OsSWEET14 is targeted more efficiently to the plasma membrane in HEK293T cells (Fig. 3e) and mediates glucose import in HEK293T cells and oocytes (Fig. 3c, f, h and Supplementary Fig. 14). OsSWEET14 also functions as a low-affinity transporter(Fig. 3h), mediating efflux in both HEK293T cells and oocytes (Fig. 3g, i). Our findings support a model that, besides inhibition of plant immunity, type III effectors and some TAL effectors can function specifically in diverting nutritional resources from the host34,35.
SWEET homologues (SLC50) are also widespread in metazoan genomes and predicted to consist of seven transmembrane helices in a 3+1+3 configuration (Fig. 1j and Supplementary Figs 8 and 9). The C. elegans genome contains seven SWEET genes (CeSWEET), whereas the human genome contains a single homologue, which we name HsSWEET1 (also called RAG1AP1). CeSWEET1 mediated glucose accumulation in HEK293T cells when co-expressed with the sensor FLII12Pglu700μδ6 (ref. 16), as well as efflux from the cytosol to the ER (Fig. 5a, b). Both N- and C-terminal GFP fusions of CeSWEET1 were functional in cellular glucose uptake (Supplementary Fig. 15) and localized primarily to the Golgi, with lower levels at the plasma membrane of HEK293T cells (Fig. 5c, d). CeSWEET1 mediated [14C]glucose and [14C]galactose uptake when expressed in oocytes (Fig. 5e, f). Similar to OsSWEET14, CeSWEET1 glucose uptake did not saturate up to 50 mM, indicating that it is a low-affinity transporter (Fig. 5g). CeSWEET1 expression in oocytes can also increase glucose efflux (Fig. 5h). RNAi inhibition of CeSWEET1 affected fat accumulation in worms, compatible with a defect in cellular glucose efflux leading to lipid accumulation36. Mutations in the homologue CiSWEET1/Ci-RGA from the sea squirt Ciona leads to early developmental defects, underlining the importance of SWEETs in metazoa37.
The human homologue HsSWEET1 did not show significant glucose uptake in yeast and oocytes (Fig. 5e and Supplementary Figs 6 and 11). However, HsSWEET1 mediated weak efflux activity in oocytes (Fig. 5i). The efflux activity was not caused by unspecific leakiness of oocytes as efflux of other sugars was not increased in cells expressing HsSWEET1 or CeSWEET1 (Supplementary Fig. 16). Thus, HsSWEET1 either rectifies or, alternatively, might be involved in exocytosis. In contrast to the plant homologues, and compatible with vesicular efflux, HsSWEET1 localized to the Golgi of HEK293T cells with minimal presence at the plasma membrane (Fig. 5j and Supplementary Figs 17–19). Mutation of the potential di-leucine internalization motif at the C terminus38 did not lead to increased plasma membrane localization or increased transporter activity (Fig. 5e and Supplementary Fig. 18). Expression data indicate ubiquitous expression throughout human tissues and cell lines, with highest expression in oviduct, epididymis and intestine (Supplementary Fig. 20). Immunolocalization data from the Human Protein Atlas are consistent with a localization in absorptive enterocytes39,40. Moreover, mouse MmSWEET1 expression was induced in the mammary gland during lactation (Supplementary Fig. 21). Localization is compatible with a function in supplying glucose to the Golgi for lactose synthesis and secretion41.
Our findings provide new insights into processes that involve sugar efflux from human cells. The human genome contains two additional classes of glucose transporters42. GLUTs are uniporters, whereas SGLTs are Na+-coupled co-transporters. GLUTs and SGLTs probably handle most of the uptake activities found in human cells. GLUT2 had originally been thought to be responsible for both import and efflux of glucose in liver and intestine. However, glucose efflux from GLUT2-null hepatocytes and GLUT2 knockout mice appeared unaffected2–4. Oral glucose load of GLUT2 knockout mice resulted in normal rates of glucose appearance in the blood2. Similarly, people affected with Fanconi–Bickel syndrome, caused by GLUT2 mutations43, do not show abnormal carbohydrate ingestion, a process that requires efflux from intestinal cells44. These findings led to the hypothesis for alternative efflux routes3,4. HsSWEET1 is thus a candidate for the postulated alternative vesicular glucose efflux from the intestine and liver cells (Supplementary Fig. 22).
A new class of sugar transporters is described, members of which have been shown to function as uniporters and are thus able to support import and efflux of sugars from cells. SWEETs undoubtedly have many important native functions, including the supply of carbon skeletons and energy to the gametophyte in plants and cellular glucose efflux in animals. Our findings also support the model that in addition to the inhibition of plant immunity, type III effectors are involved in accessing nutritional resources of host plants34,35. Notably, the founding member of the SWEET family, MtN3, was identified as a nodulin-specific EST in the legume Medicago truncatula and may have a role in symbiotic Rhizobia45 nutrition. Knowledge of the full spectrum of pathogen effector molecules, and how they manipulate plant transport and metabolism to favour pathogen growth, will improve our understanding of host–pathogen interactions and may lead to new strategies for combating pathogen infections, which at the global scale lead to crop losses of over 10% annually46. Moreover, analysis of the complete SWEET family may help to solve some of the mysteries of pollen nutrition, nectar production and carbon sequestration.
Cell culture, transfection, image acquisition and FRET analysis were performed as described previously15. Yeast complementation and uptake assays were performed in EBY4000 (ref. 17). Tracer uptake and efflux assays were performed in Xenopus oocytes47. Arabidopsis Col-0 plants were grown in growth chambers under 8 h light/16 h dark at 22 °C. qPCR was performed with gene-specific primers (Supplementary Table 1). See Methods for details.
Total RNA was extracted from HepG2 or HEK293T cells using an RNeasy MINI kit (Qiagen), first strand cDNA was produced (New England Biolabs) and fragments of the predicted length were obtained by RT–PCR using a set of GLUT and SGLT primers published previously48. Samples were separated on a 2% agarose gel. For samples inoculated by Pseudomonas syringae pv. tomato DC3000, total RNA was extracted from the leaves using Trizol reagent (Invitrogen). Real-time quantitative PCR (qPCR) was performed using HotStart-IT SYBR Green qPCR Master Mix (USB) according to the manufacturer’s instructions on a 7300 PCR system (Applied Biosystems). Actin (ACT8) expression was used to normalize expression values in each sample; expression values were determined relative to the value of the sample infiltrated with 1 mM MgCl2 buffer at each time point using the comparative 2−ΔΔCt method49. For samples infected by G. cichoracearum or X. oryzae pv. oryzae, qPCR assays were performed using a LightCycler 480 (Roche). For quantification, relative transcript levels for each gene were normalized to ACT8 following the 2−ΔΔCt method49. Fold change was calculated relative to the untreated sample. Analysis was repeated twice independently. The observed induction is confirmed by microarray data (Genevestigator)50.
AtSWEET1, AtSWEET8, OsSWEET11 and OsSWEET14 ORFs were amplified by RT–PCR using specific primers from Arabidopsis and rice, respectively. First-strand cDNA from rice was provided by P. Ronald. SWEET ORFs were cloned into pDONR221 (Invitrogen) or pDONR221-f1 (ref. 51). Truncated versions of AtSWEET1-L198*, OsSWEET11-F205* and OsSWEET14-F203* were generated by introducing stop codons in transmembrane helix 7 by site-directed mutagenesis. All entry constructs were transferred to pDRf1-GW (ref. 52) and pOO2-GW (D. Loqué, unpublished results) by Gateway LR recombination reactions (a recombination reaction between an entry clone (containing attL) and a destination vector (containing attR), mediated by a host of recombination proteins to generate an expression clone (Entry Clone + Destination Vector Expression Clone)) (Invitrogen). AtSWEET1 was cloned into p112-A1NE-GW for yeast co-transformation with FLII12Pglu700μδ6 in pDRf1-GW (ref. 18). Plasmid p112-A1NE-GW was generated by inserting a Gateway cassette into the SmaI restriction site of p112-A1NE53. For radiotracer experiments, ORFs with stop codons for AtSWEET1, OsSWEET11 and OsSWEET14 were cloned into the pOO2-GW by Gateway LR recombination reactions.
The full-length splice variant HsSWEET1-1 in pDNR-LIB was obtained from Open Biosystems (Clone ID 4076256). The truncated form HsSWEET1-G194* was generated by introducing a stop codon at leucine 194 by site-directed muta-genesis54. Site-directed mutagenesis was used to mutate the putative internalization motif (HsSWEET1m Y216A, L218A, L219A). Products were cloned by in vitro BP recombination (Invitrogen) into pDONR221-f1, then mobilized into pOO2-GW by LR reactions. The shorter splice variant HsSWEET1-2 from Open Biosystems (Clone ID 3896154) in pCMV-SPORT6 was transferred into pOO2-GW by in vitro LR recombination. CeSWEET1 (K02D7.5, Open Biosystems) was cloned into pOO2-GW using an LR reaction. The ORF of AtSWEET1 without stop codon was cloned into the binary vector pX-YFP-GW by an LR reaction. AtSWEET1, OsSWEET11, OsSWEET14, CeSWEET1 and HsSWEET1 without stop codons were cloned into C-terminal GFP fusion vector pcDNA-DEST47 (Invitrogen) for localization studies. AtSWEET1, CeSWEET1 and HsSWEET1with stop codons were cloned into the N-terminal GFP fusion vector pcDNA-DEST53 (Invitrogen) for localization studies. For yeast growth assays, ORFs were expressed from pDRf1-GW. For GFP localization in yeast, AtSWEET1 and the truncated AtSWEETΔ198 were cloned in vector pDR-GW-eGFP52.
The strain EBY4000 (hxt1 through -17Δ::loxP gal2Δ::loxP stl1Δ::loxP agt1Δ::loxP ydl247wΔ::loxP yjr160cΔ::loxP)17 was transformed with AtSWEET1, AtSWEET8 and HXT5 and grown on SD (synthetic deficient) medium supplemented with 2% maltose and auxotrophic requirements. For complementation growth assays, cells were grown overnight in liquid minimum medium to an optical density at 600 nm (OD600) of ~0.6, then OD600 was adjusted to ~0.2 with water. Serial dilutions (×1, ×5, ×25 and ×125) were plated on SD media containing either 2% maltose (as control) or 2% glucose plus respective auxotrophic requirements. Growth was documented by scanning (CanoScan, Canon) after 2–5 days at 30 °C.
Yeast cells were grown in SD medium supplemented with 2% maltose and auxotrophic markers. Cells were harvested at OD600 0.5–0.7 by centrifugation, and washed twice in ice-cold distilled water. Cell pellets were weighed after supernatant had been removed. Cells were re-suspended 5–10% (w/v) in 40 mM potassium phosphate buffer, pH 6.0. Cells were pre-incubated in potassium phosphate buffer for 5 min at 30 °C. For each reaction, 330 μl pre-warmed buffer containing 20 mM glucose (0.55 μCi D-[U-14C] glucose; 590 kBq μmol−1, Amersham) was added to an equal volume of cells. 120 μl aliquot were withdrawn and transferred to ice-cold water. Cells were harvested by vacuum filtration onto a glassfibre filters (GF/C, Whatman), and washed twice in 10 ml ice-cold water. Filters were transferred to scintillation vials containing 5 ml Ultima Gold XR Scintillation liquid (Perkin Elmer). Radioactivity taken up by the cells was measured by liquid scintillation spectrometry. To determine the pH-dependence of AtSWEET1 activity, 40 mM potassium phosphate uptake buffer at specified pH was used. Three independent transformants were used for each uptake experiment.
After linearization of the pOO2 plasmids with MluI, capped cRNAs were synthesized in vitro by SP6 RNA poly-merase using mMESSAGE mMACHINE kit (Ambion, Inc.). Xenopus laevis oocytes were provided by M. Goodman. Microinjection was carried out as described47,56. 25–50 ng cRNA was injected into healthy looking oocytes (RNase-free water was used as control). Oocytes expressing AtSWEET1 were maintained at 18 °C in modified Barth’s saline (MBS, in mM: 88 NaCl, 1 KCl, 2.4 NaHCO3, 0.82 MgSO4, 0.33 Ca(NO3)2, 0.41 CaC12, 20 HEPES-Tris, pH 7.5) with 100 μM gentamycin, 100 U ml−1 penicillin and 100 μM streptomycin solution for 2–3 d. Incubation buffer was changed every 24 h. For all other SWEETs, injected oocytes were maintained in L-15 oocyte medium (7.4 g l−1 Leibovitz’s L-15 medium (Sigma), 3.57 g l−1 HEPES pH 7.5) with 100 mg l−1 gentamycin.
The assay was performed with modification as described previously57. Two days after injection, groups of 7–16 oocytes were transferred into tubes containing 200 μl Na-Ringer (in mM: 115 NaCl, 2 KCl, 1 MgCl2, 1.8 CaC12, 10 HEPES-Tris, pH 7.5) 100 mg l−1 gentamycin and D-glucose (4 μCi ml−1 D-[14C(U)]-glucose; PerkinElmer), galactose (4 μCi ml−1 D[1-14C]-galactose; American Radiolabelled Chemicals) or sucrose (4 μCi ml−1 D-[1-14C(U)]-sucrose; PerkinElmer). After incubation at 20 °C, cells were transferred to ice-cold Na-Ringer, washed three times, solubilized with 100 μl 1% (w/v) SDS, and measured individually.
Efflux was measured as described58. Two days after cRNA injection, oocytes were injected with 50 nl solution containing 5, 10 or 50 mM -glucose or sucrose with 0.18 μCi μl−1 D-[14C(U)]glucose. Cells were immediately washed once in Na-Ringer (except AtSWEET1 in MBS). At defined time points, reaction buffer (450 μl; except AtSWEET1 in 950 μl) was removed for scintillation counting. Oocytes were solubilized with 1% SDS and analysed for retained radioactivity.
FRET measurements in yeast cells were performed as described18.
Arabidopsis Col-0 plants were grown in growth chambers under 8 h light/14 h dark at 22 °C. Five-week-old leaves were infiltrated with a 1 mM MgCl2 buffer, 2 × 108 c.f.u. ml−1 Pseudomonas syringae pv. tomato DC3000 or Pseudomonas syringae pv. tomato DC3000 Δ hrcU suspensions in 1 mM MgCl2 using needleless syringes. Leaf samples were collected after 6, 12 and 24-h incubation in the light. G. cichoracearum inoculation was performed as described59. Plants were placed in a ‘settling tower’ (cardboard box) and inoculated with G. cichoracearum spores by holding infected squash leaves over the settling tower and using compressed air (duster cans) to blow spores off of the squash leaves for settling onto Arabidopsis plants. Inoculum density was ~25–35 conidiospores per mm2. After inoculation, plants were incubated for 1 h in a dark dew chamber, then transferred to a growth chamber at 16 h day length, 70% relative humidity.
Two-week-old rice seedlings (cultivar IR24) were infected with Flag-tagged effector X. oryzae pv. oryzae strains ME2(avrXa7-2F) or ME2(pthXo1-2F) at OD600 0.5. At 20 h after inoculation, ChIP complexes were prepared from 3.0 g of inoculated leaf tissue for each treatment. Immune complexes were prepared as described60 with minor modifications. Effector-associated DNA complexes were immunoprecipitated using monoclonal Flag antibody (Sigma, 12 μg ml−1). The same amount of mouse nonspecific IgG antibody was added in the control. Enriched DNA obtained was analysed by real-time qPCR using promoter and 3′ UTR specific primers (provided Supplementary Table 3). Two microlitres of eluted DNA was used in each reaction. qPCR and analysis was performed as described above. Values are expressed as a ratio of the 2−ΔΔCt value from Flag-tagged antibody precipitate complexes over the 2−ΔΔCt value of the nonspecific IgG complexes. PXO99AME2 (avrXa7-F2) served as control for effector specificity.
Multiple alignment of SWEET amino acid sequences was performed with CLUSTALW61 using default parameters, and a phylogenetic analysis was performed using the software Mega V3.1. Bootstrapping was performed 1,000 times to obtain support values for each branch. For pair-wise comparison, multiple alignments of complete amino acid sequences were conducted using the Vector NTI advance 11.0.
Fluorescence imaging of plants and mammalian cells expressing AtSWEET1–YFP, AtSWEET8–YFP, CeSWEET1–GFP, GFP–CeSWEET1, OsSWEET11–GFP and OsSWEET14–GFP was performed on a Leica TCS SP5 microscope. YFP was visualized by excitation with an argon laser at 514 nm and spectral detector set between 525 and 560 nm for the emission. GFP was visualized by excitation with an argon laser at 488 nm and spectral detector set between 500 and 545 nm for the emission. Specimens were observed with 40/0.75-1.25NA HCX PL APO CS objective.
This work was made possible by grants from the Department of Energy (DE-FG02-04ER15542) and NIH (NIDDK; 1RO1DK079109) to W.B.F., X.-Q.Q. was supported by The Carnegie Institution and the National Natural Science Foundation of China (NSFC; 30771288). NSF (IOS-0821801) and NIH (ZRO1GM06886-06A1) to M.B.M. and J.-G.K. was supported 50% by NIH and 50% by NSF. W.U. was supported in part by an NIH postdoctoral fellowship (F32GM083439-02). G.A. and F.F.W. were supported by grants from USDA NIFA (2007-35319-18103) and NSF Plant Genome (DBI-0820831).
Full Methods and any associated references are available in the online version of the paper at www.nature.com/nature.
Author Contributions W.B.F., S.L., M.B.M. and S.C.S. conceived and designed the experiments. L.-Q.C., B.-H.H., H.T., M.L.H., J.-G.K., X.-Q.Q., W.-J.G., W.U., B.C., G.A. and D.C. performed the experiments. W.B.F., S.L., M.B.M., G.A., F.F.W. and S.S. analysed the data. L.-Q.C. and W.B.F. wrote the manuscript.
Author Information Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of this article at www.nature.com/nature.