Structure and mechanism of the mammalian fructose transporter GLUT5

Abstract

The altered activity of the fructose transporter GLUT5, an isoform of the facilitated-diffusion glucose transporter family, has been linked to disorders such as type 2 diabetes and obesity. GLUT5 is also overexpressed in certain tumour cells, and inhibitors are potential drugs for these conditions. Here we describe the crystal structures of GLUT5 from Rattus norvegicus and Bos taurus in open outward- and open inward-facing conformations, respectively. GLUT5 has a major facilitator superfamily fold like other homologous monosaccharide transporters. On the basis of a comparison of the inward-facing structures of GLUT5 and human GLUT1, a ubiquitous glucose transporter, we show that a single point mutation is enough to switch the substrate-binding preference of GLUT5 from fructose to glucose. A comparison of the substrate-free structures of GLUT5 with occluded substrate-bound structures of Escherichia coli XylE suggests that, in addition to global rocker-switch-like re-orientation of the bundles, local asymmetric rearrangements of carboxy-terminal transmembrane bundle helices TM7 and TM10 underlie a 'gated-pore' transport mechanism in such monosaccharide transporters.

Extended Data Fig. 1. Anisotropy descriptors of bGLUT5 data reported by the UCLA-MBI Diffraction Anisotropy Server and 2Fo-Fc electron density maps for the bovine and rat GLUT5 structures

a. Degree of anisotropy of bGLUT5 data, resolution limits for the 3 principle axes (left), and panel illustrating steps along correction of bGLUT5 data for anisotropy (right). b Representative portions of the electron density map (1.5σ) for bGLUT5 overall model (left) and a close-up of the substrate binding site (right); residues highlighted are numbered based on rGLUT5 for sake of clarity c. Electron density (1.0σ) for rGLUT5 showing one of the inter-bundle salt-bridge clusters that form in the open outward-facing conformation.

Extended Data Fig. 2. Superimposition of open inward-facing bGLUT5 and hGLUT1 structures, and comparison of the substrate-binding site in bGLUT5 and inward-facing XylE

a. Ribbon representation of inward-facing bGLUT5 (colored as in Fig. 1a) and inward-facing hGLUT1 (light grey) structures, as viewed in the plane of the membrane. The D-glucopyranoside moiety of the detergent molecule bound to GLUT1 (n-nonyl-β-D-glucopyranoside (β-NG)) is shown as sticks. Density for ICH5 at the C-terminus is missing in both hGLUT1 and bGLUT5 inward-facing structures and highlighted with the dotted ellipse. The beginning of TM1 kinks further outwards in the bGLUT5 structure compared to hGLUT1 and residues 1 to 18 could not be built. The r.m.s.d. (root mean square deviation) after superposition of the two structures is 1.12Å for 364 pairs of Cα atoms (see Methods). b The substrate-binding in the inward-facing bGLUT5 structure (coloured as in Fig. 1), is very similar to that seen in inward-facing XylE (4JA4) structure (shown in light-grey). Only non-conserved residues and the equivalent glutamine to Q166 are labeled for XylE.

Extended Data Fig. 3. Structure of the rat GLUT5-Fv complex

a. Cartoon representation of the complex between rGLUT5 (grey) and 4D111Fv (heavy-chain variable region (V_H) is in blue; light-chain variable region (V_L) is in red). 4D111Fv binds to the cytoplasmic domain of GLUT5, including ICH2 (residues 226, 230, 234), the loop between ICH2 and ICH3 (residues 238, 240, 241), and ICH3 (residue 243), with ~848 Ų of buried surface area at the interface. b. Packing of the rat GLUT5-Fv complex molecules in the crystal. The unit cell is represented as green lines.

Extended Data Fig. 4. Sequence alignment of rat GLUT5 (rGLUT5), bovine GLUT5 (bGLUT5), human GLUT5 and GLUT7 (hGLUT5, hGLUT7), human GLUT1-4 (hGLUT1-4), Saccharomyces cerevisiae (HXT7), Plasmodium falciparum (PfHT1), Arabidopsis thaliana (GlcT) and E. coli XylE

Secondary structure elements of rat GLUT5 are indicated above the alignment, and coloured as in Fig. 1a. Strictly conserved residues are highlighted in black-filled boxes, and highly conserved residues are shaded in grey. Green boxes highlight central cavity residues that are specific to GLUT5 and red boxes highlight those that are conserved among GLUTs. Purples boxes highlight residues forming the salt-bridges between cytosolic TM segments. A blue box (TM5) highlights Gln166, whose mutation to glutamic acid, as present in GLUT7, weakens D-fructose binding but supports strong D-glucose binding in rGLUT5. The brown box (TM8) highlights Glu336 that is conserved across all the GLUTs and replaced with glutamic acid in XylE. Red bars underneath the alignment indicate the sugar porter (SP) family motifs^,. Note that because bGLUT5 and hGLUT5 have an additional amino acid at position 8 their numbering differs from rGLUT5 by 1 amino acid. For clarity, bGLUT5 residues are labeled using rGLUT5 numbering.

Extended Data Fig. 5. D-fructose binding monitored by tryptophan fluorescence quenching

a. Cartoon representation of the outward-facing rGLUT5 structure, as viewed from the plane of the membrane with the coloring as shown in Fig. 1a. Atoms in all tryptophan residues are shown as spheres and tryptophan W419, whose fluorescence is quenched by substrate, is labeled. b. Emission fluorescence spectra for purified deglycosylated rGLUT5 wildtype like mutant N50Y (referred to as “WT”), shown in the range of 320-360 nm with an excitation wavelength of 295 nm after the addition of 40 mM D-fructose (top), and 40 mM L-fructose (bottom). Emission fluorescence spectra for purified WT protein that had been previously incubated with the inhibitor HgCl₂ is also shown for D-fructose (middle). c. Tryptophan fluorescence quenching (excitation 295 nm; emission 338 nm) after incubation of purified rGLUT5 N50Y with either 40 mM D-fructose (black-filled bar) or L-fructose, D-glucose, D-mannose, D-xylose or D-galactose as labeled (open bars). Tryptophan fluorescence quenching for purified WT protein that had been previously incubated with the inhibitor HgCl₂ is also shown for D-fructose (non-filled bar) d. As in c., rGLUT5 with a single tryptophan residue (W419), which contains the following mutations: N50Y, W70F, W191F, W239F, W265F, W275F, W338F and W370F. No tryptophan quenching was observed for D-fructose (5 mM HgCl₂), L-fructose, D-glucose or D-galactose. In all experiments errors bars, s.e.m.; n = 3.

Extended Data Fig. 6. Substrate-specificity in GLUT5

a. Time-dependent uptake of D-[¹⁴C]-fructose by rGLUT5 wildtype (open squares and triangles) and the deglycosylated mutant N50Y (filled squares and triangles) in proteoliposomes incubated with or without the inhibitor HgCl₂ as labeled. Non-specific uptake was estimated with 0.1 mM L-[¹⁴C]-Glucose for wildtype (filled circles) and the N50Y mutant (open circles). In all experiments errors bars represent a spread of duplicates. Inset shows SDS-PAGE analysis of the purified rat GLUT5 wildtype and the deglycosylated N50Y mutant. b. Tryptophan fluorescence quenching (excitation 295 nm; emission 338 nm), after incubation of purified rat GLUT5 mutant (N50Y, W70F, W191F, W239F, W265F, W275F, W338F, W370F) that contains one single tryptophan residue, W419, with increasing concentrations of D-fructose (filled squares) and to the protein previously incubated with the inhibitor mercury chloride (open circles). c. Slab through the surface of the outward-facing rGLUT5 structure as viewed in the plane of membrane. The structure of substrate-bound XylE structure was further superimposed onto rGLUT5 and is shown here as a grey ribbon. In XylE, Trp392 (Trp388 in hGLUT1) is located at the bottom of the cavity (spheres; magneta) and coordinates D-xylose (stick form; yellow). In GLUT5, the equivalent residue is an alanine, making the cavity deeper. d. D-fructose binding as measured by tryptophan fluorescence quenching (excitation 295 nm; emission 338 nm) after incubation with 40 mM D-fructose for WT (open bar), and TM7 mutations of Ile295 (interacts with TM10 residues) and Tyr296 and Tyr297 residues. Equivalently located tyrosine residues in XylE occlude the sugar-binding site from the outside. Fluorescence quenching for the mutants are displayed as a percentage of total WT binding. In all experiments errors bars, s.e.m.; n = 3.

Extended Data Fig. 7. The intracellular helical domain (ICH)

a. Cytoplasmic view of the ICH domain after superposition of the open, outward-facing rGLUT5-Fv (grey) and outward-facing occluded E. coli XylE (teal) (4GBY) structures. b. In the outward-facing GLUT5 structure ICH1-3 are linked together by several salt-bridges (side chains are labeled and shown as sticks in yellow). In contrast, no polar interactions are formed between ICH5 and either ICH1-3 or cytoplasmic ends of N-terminal TM bundle helices. A salt-bridge forms (dotted line in magenta), however, between Glu225 in ICH3 and Arg407 in TM 11, which also forms part of the inter-bundle salt-bridge network (side chains are labeled and shown as sticks in cyan). c. In the inward-facing GLUT5 structure, this inter-bundle salt-bridge network is not formed, because the cytoplasmic ends of the N- and C-terminal bundle have moved apart; consistently, the ICH domain functional role is proposed to act as a scaffold domain that further helps to stabilize the outward-facing conformation.

Extended Data Fig. 8. Access to the central cavity and substrate-binding site is gated by TM7 on the outside and TM10 on the inside

a. Superposition of outward-facing open GLUT5 and outward-facing occluded E. coli XylE (4GBY) structures. The TM numbering for outward-facing occluded XylE has an additional asterisk “*”. The inward-facing GLUT5 structure is colored as in Fig. 1a and that of XylE in grey. The bound D-xylose in stick form in green. The r.m.s.d. is 1.38Å for 290 pairs of Cα atoms (see Methods). b. Superposition of inward-open GLUT5 and inward-occluded E. coli XylE structure (4JA3) with coloring and annotation as described in a. The r.m.s.d. is 1.80Å for 274 pairs of Cα atoms (see Methods). The bound D-xylose in 4GBY is represented in stick form in green. The ICH domain is not shown for clarity. c. Superposition of inward-facing open GLUT5 and inward-facing open XylE (4JA4) structures as viewed from the cytoplasmic side with coloring and annotation as described in a. The ICH domain is not shown for clarity. The r.m.s.d. is 1.70Å for 273 pairs of Cα atoms (see Methods).

Fig. 1. Structures of rat GLUT5 in the open outward-facing conformation and bovine GLUT5 in the open inward-facing conformation

a. Ribbon representation of open outward-facing rat GLUT5 (left) and open inward-facing bovine GLUT5 (right) structures, viewed in the plane of the membrane. TMs 1 and 4 and TMs 2, 3, 5 and 6 in the N-terminal TM bundle are colored in blue and light-blue, respectively. TMs 7 and 10 and TMs 8, 9, 11 and 12 in the C-terminal TM bundle are colored in red and yellow-brown, respectively. The intracellular domain helices ICH1 to ICH5 are shown in grey. b. Slab through the surface electrostatic potential of the open outward- (left) and open inward-facing (right) GLUT5 structures, as viewed within the plane of membrane, which highlight the accessibility of the sugar to the central cavity (shown as a dotted ellipse). c. Ribbon diagrams of GLUT5 viewed from the cytoplasm in the open outward- (left) and inward-facing (right) conformations.

Fig. 2. The fructose-binding site of GLUT5

a. The substrate-binding site in the inward-facing bGLUT5 structure (left panel; colored as in Fig. 1) is very similar to the inward-facing hGLUT1 structure (right panel; light-grey). To facilitate comparison to rGLUT5, bGLUT5 residues are labeled with rGLUT5 numbering. For hGLUT1 only Q161 and all other residues that are different in bGLUT5 are labeled. The D-glucopyranoside moiety of bound n-nonyl-β-D-glucopyranoside in hGLUT1 is shown as sticks. b. D-fructose binding to GLUT5 as measured by tryptophan (Trp) fluorescence quenching (excitation 295 nm; emission 338 nm) after addition of increasing concentrations of D-fructose to WT (black squares) and to WT protein that had been previously incubated with the GLUT inhibitor HgCl₂ (open circles). c. Trp fluorescence quenching for purified substrate-binding site mutants after addition of 40 mM D-fructose (non-filled bars) relative to WT (filled bar). d. Trp fluorescence quenching after addition of either 40 mM D-fructose or D-glucose to purified WT (black-filled bar) or Q166E (non-filled bars); pre-incubation with the inhibitor HgCl₂ is indicated. e. Trp fluorescence quenching after addition of increasing concentrations of D-glucose to either purified Q166E (black squares), Q166E previously incubated with HgCl₂ (open circles) or WT (open triangles). In all experiments errors bars, s.e.m.; n = 3.

Fig. 3. Inter-TM salt-bridges form between bundle cytoplasmic ends in the outward-facing conformation

a. Cartoon representation of GLUT5 as viewed from the cytoplasm in the outward- (left) and inward-facing (right) conformations. ICHs are not shown for clarity. TMs are colored as in Fig. 1a., and residues forming salt-bridges are shown as sticks. To facilitate comparison to rGLUT5, bGLUT5 residues are labeled with rGLUT5 numbering. Breakage of conserved salt-bridges as seen here for GLUT5 has also been predicted for XylE b. Superimposition of the N- and C-terminal 6-TM bundles. Strictly conserved and pseudo symmetry related charged residues forming the salt-bridges are labeled and are shown as sticks. c. D-fructose binding as measured by Trp fluorescence quenching after incubation with 40 mM D-fructose for purified WT GLUT5 (black-filled bar) and single alanine mutations of key acidic residues E400 and E151 that form inter-bundle salt-bridges, and G389, which is located in the hinge point of TM 10 critical for TM10 conformational change (non-filled bars). Trp fluorescence quenching by D-fructose for these mutants is displayed as a percentage of WT binding. In all experiments errors bars, s.e.m.; n = 3.

Fig. 4. Substrate-induced gates are predominantly formed by TMs 7 and 10 in the C-terminal bundle

a. Superposition of GLUT5 open outward- and inward-facing (*) structures, as viewed from the extracellular (left) and intracellular (right) side of the membrane. TMs are colored as in Fig. 1a, except inward-facing TMs 1* and 4* and TMs 7* and 10* that are coloured in orange and cyan, respectively. ICHs have been removed for clarity. b. Superimposition of the GLUT5 open outward- and inward-facing structures as viewed in the plane of the membrane. For clarity, TMs 5, 5*, 8, and 8* are not shown. Cavity-closing contacts are mostly formed by TMs 1* and 7* on the extracellular side in the inward-facing conformation and by TMs 4 and 10 on the intracellular side in the outward-facing conformation. These TMs are the first TMs in each of the four 3-TM repeats of the MFS fold^,. D-xylose, as it is in the occluded-outward-facing XylE structure (4GBY), is shown in stick-form. With the inward movement of TM7 conserved tyrosine residues are likely to occlude the substrate from exiting, as seen for the equivalently located tyrosine residues in the substrate-occluded XylE structure and as supported by D-fructose binding data (Extended Data Fig. 6d). The opening movement of TM10 to enable cytosolic substrate release has been described previously for XylE ^, and other unrelated MFS transporters^, c. Interactions between hydrophobic residues between TM7 and TM10 in the outward-facing conformation (left) are lost in the inward-facing conformation (right). To facilitate comparison to rGLUT5, bGLUT5 residues are labeled with rGLUT5 numbering.

Fig. 5. Alternating-access transport mechanism in GLUT5

Schematic representation of the “rocker-switch” type movement of the N- and C-terminal TM bundles and of the local, gating conformational changes of TMs 7 and 10 supporting a “gated-pore” type transport mechanism in GLUT5.