a, Identification of glucose transport activity for AtSWEET1 by co-expression with cytosolic FRET glucose sensor FLIPglu600μΔ 13V in HEK293T cells. Individual cells were analysed by quantitative ratio imaging of CFP and Venus emission (acquisition interval 5 s; Fc/D corresponds to normalized emission intensity ratio). HEK293T/FLIPglu600μΔ 13V cells were perfused with medium, followed by square pulses of increasing glucose (Gluc.) concentrations. Orange line indicates cells expressing sensor alone; blue line indicates cells co-expressing sensor and AtSWEET1; accumulation of glucose is shown by a negative FRET ratio change (blue line; mean – s.d.; n > 10). b, FRET imaging of glucose efflux from cytosol into ER. FLIPglu600μΔ13V^ER was targeted to the ER lumen (compare with panel a; acquisition interval 10 s; mean – s.d.; n > 10). c, Cartoon for SWEET influx across plasma membrane and efflux from cytosol to ER. Cytosolic FLIPglu600μΔ 13V identifies glucose import from the extracellular face (extracellular N terminus). FLIPglu600μΔ13V^ER measures transport from the intracellular side (cytosolic C terminus). d, Complementation of yeast EBY4000 (ref. ) lacking 18 hexose transporter genes with AtSWEET1, AtSWEET8, or yeast HXT5; control, empty vector. e, Glucose accumulation in EBY4000 co-expressing AtSWEET1 and FLII¹²Pglu700μδ6 before and after glucose addition (two cycles before glucose addition; mean ± s.d.; n = 3). f, Kinetics of [¹⁴C]glucose accumulation by AtSWEET1 in EBY4000 (mean ± s.d., n = 3). g, AtSWEET1-mediated uptake of 1 mM [¹⁴C]glucose, [¹⁴C]galactose or [¹⁴C]sucrose into oocytes (mean ± s.d.; n = 8 cells). h, [¹⁴C]glucose efflux from oocytes expressing AtSWEET1 (mean ± s.e.; n ≥ 10 cells; P < 0.0005). i, Confocal imaging of AtSWEET1–YFP in leaves of stably transformed Arabidopsis (panel width = 176 μm). j, Structural model of SWEETs based on hydrophobicity plots (duplication of three transmembrane helices; red/blue triangles).

a, Induction of AtSWEET mRNAs by either the bacterium Pseudomonas syringae pv. tomato DC3000 (2 ×10⁸ c.f.u. ml⁻¹, 8 h after inoculation, measured by qPCR, normalized by MgCl₂ buffer treatment), powdery mildew fungus G. cichoracearum (~25–35 conidiospores per mm², 48 h after inoculation, measured by qPCR; normalized to 0 h values), or by Botrytis cinerea (from microarray study) in Arabidopsis leaves. ND, not detectable. b, Induction of AtSWEET4, AtSWEET5 and AtSWEET15 by P. syringae DC3000 depends on a functional type III secretion system (T3S). Samples were collected at 6, 12 and 24 h after infiltration with 2 × 10⁸ c.f.u. ml⁻¹ of DC3000 or DC3000 ΔhrcU, a T3S mutant. c, Infection by G. cichoracearum leads to induction of AtSWEET11 and AtSWEET12 but downregulation of AtSWEET15.

a, Targeting of OsSWEET11 to the plasma membrane of HEK293T cells (panel width = 61 μm). b, Co-expression of OsSWEET11 with cytosolic FRET glucose sensor FLIPglu600μΔ13V in HEK293T cells (compare to ; significant above control in 4 of 6 experiments). Truncated OsSWEET11 with premature stop codon at phenylalanine 205 (F205*) served as negative control (mean – s.d., n > 10). c, Glucose uptake (1 mM) mediated by OsSWEET11 and OsSWEET14 in oocytes (mean ± s.d., n ≥ 7). d, Enrichment of 5′-upstream OsSWEET11 promoter DNA upon infection of rice with PXO99^A containing Flag-tagged PthXo1-2F. e, Efficient targeting of OsSWEET11 to the plasma membrane of HEK293T cells (panel width = 61 μm; mean ± s.d.; n = 2). f, Co-expression of OsSWEET14 with the cytosolic FRET glucose sensor FLIPglu600μΔ13V in HEK293T cells (compare to ). Truncated OsSWEET14 with a premature stop codon at phenylalanine 203 (F203*) served as negative control (mean – s.d., n > 10). g, Co-expression of OsSWEET14 with ER-targeted FRET glucose sensor FLIPglu600μΔ13V^ER in HEK293T cells (compare with ) (mean – s.d., n > 10). h, Concentration-dependent glucose uptake by OsSWEET14 in oocytes (mean ± s.d., n ≥ 7). i, [¹⁴C]-efflux from glucose-injected oocytes mediated by OsSWEET14 (50 nl of 50 mM glucose injected; mean ± s.e., n ≥ 7).

a, The pathogenic Xoo strain PXO99^A injects TAL effector PthXo1 via the type III secretion system into rice cells. PthXo1 directly induces OsSWEET11 leading to sugar efflux. Bacteria take up glucose and multiply. b, PXO99^A pthXo1 mutant (PXO99^AME2) leads to loss of OsSWEET11 induction and reduced bacterial growth (indicated as reduced size of bacterium). c, Mutation of the OsSWEET11 effector binding element (EBE) leads to loss of PthXo1-mediated induction and reduced bacterial growth. d, Bacteria expressing effector AvrXa7 can circumvent loss of OsSWEET11 induction by inducing OsSWEET14.

a, Detection of CeSWEET1 glucose uptake in HEK293T cells by co-expressing FLII¹²Pglu700μδ6 (see legend under ) (mean + s.d., n ≥ 10). b, Detection of CeSWEET1 glucose efflux into the ER of HEK293T cells by co-expressing FLII¹²Pglu700μδ6 (see legend under ) (mean + s.d., n ≥ 10). c, d, Localization of GFP–CeSWEET1 (c) and CeSWEET1–GFP (d) fusions in HEK293T cells. e, [¹⁴C]sugar uptake (5 mM) in Xenopus oocytes by CeSWEET1, HsSWEET1 (full length HsSWEET1; splice variant HsSWEET1–2; mean + s.d., n ≥ 8), di-leucine motif mutant (HsSWEET1m) and deletion mutant G194*. f, Time-dependent [¹⁴C]glucose uptake by CeSWEET1 in oocytes (mean + s.d., n ≥ 8). g, Concentration-dependent [¹⁴C]glucose uptake by CeSWEET1 in oocytes (mean ± s.d., n ≥ 8). h, Efflux measurements from oocytes expressing CeSWEET1 (mean ± s.e.; n ≥ 8 cells; P < 0.01 after 5 min; 50 nl of 50 mM sugar injected). i, Efflux measurements from oocytes expressing HsSWEET1 (mean ± s.e.; n ≥ 8 cells; P < 0.05). j, Immunolocalization of HsSWEET1 in HEK293T cells (merged channels: immuno-labelled HsSWEET1, red; Golgi marker, green; nuclei, blue).