Engineered XylE as a tool for mechanistic investigation and ligand discovery of the glucose transporters GLUTs

Dear Editor, Glucose is the primary energy supply for metabolism and a versatile precursor for biomolecule synthesis. Cellular uptake of glucose mainly relies on two types of glucose transporters, the sodium gradient driven glucose symporter SGLT family and the facilitative uniporter GLUT family. Glucose uptake is markedly elevated in tumor cells to compensate for the less-efficient ATP production via glycolysis, a phenomenon known as the Warburg effect. Overexpression of GLUT family members GLUT1 or GLUT3 has been observed in various types of tumors, making them potential targets for novel drug discovery against cancer. Structures of GLUT1, GLUT3, and GLUT5 were obtained in distinct conformations, which collectively recapitulate the nearly complete alternating access cycle, elucidate the molecular basis for substrate selection and transport, and shed light on the mechanistic understanding of pathogenic mutations. The structures are instrumental for mechanistic investigations and ligand discovery. However, in our attempt to establish an in vitro biochemical characterization and screening system for GLUT1/3, we found that recombinantly expressed GLUT proteins were fragile after detergent extraction, impeding mutational analysis. We thereby employed the xylose:proton symporter XylE from E. coli, which is one of the closest bacterial homologs of GLUTs, as a surrogate. In addition to the overall structural similarity, the substrate-binding site is highly conserved between the bacterial and human proteins despite that XylE can only bind to, but not transport glucose. Both GLUTs and XylE contain a central substrate binding site that accommodates the ligand in both outwardand inward-facing states. However, this primary binding site undergoes minor conformational changes owing to the local structural shift accompanying the switch between the outward and inward conformations. In addition, the structure of maltose-bound GLUT3 reveals a potential secondary glucose binding site on the extracellular side that is lacking on the intracellular side. These structural observations suggest that the sugar porter (SP) members may have asymmetric affinities for exofacial (access from the extracellular side, or outward-facing state) and endofacial (access from the intracellular side, or inward-facing) binding, a notion that is supported by previous characterizations of human GLUT1. In this study, we attempt to address two questions. First, does XylE share the same asymmetric binding affinities in outwardand inward-facing conformations like GLUT1? Second, can XylE be employed to distinguish endofacial and exofacial ligands for GLUTs? To test this proposition, we sought to engineer XylE variants that would exhibit constitutively outwardor inward-facing state. For this purpose, we first carried out inter-domain crosslinking experiment. Based on the structure, we identified two pairs of residues for Cys substitution (Supplementary Fig. S1a). The double mutations V35C/E302C (ExoCC) and A152C/S396C (EndoCC) were predicted to form disulfide bonds on the extracellular and intracellular side, respectively, under oxidative condition, and hence locking the protein in inward facing and outward facing, respectively (Fig. 1a, Supplementary Fig. S1a). To verify formation of the designed disulfide bonds, we introduced a DrICE protease-specific cleavage site (DEVDA) into the intracellular helix ICH3 (Fig. 1a,

mM imidazole and 0.02% DDM. The elution was concentrated and further purified with gel filtration (Superdex-200; GE Healthcare) in various buffer for different usage.
The peak fractions from gel filtration were collected and flash-frozen in liquid nitrogen for further experiments.

Crystallization, data collection and structure determination
XylE-WW protein was purified by gel filtration in buffer containing 25 mM Tris-HCl pH 8.0, 150 mM NaCl and 0.056% 6-Cyclohexyl-1-Hexyl-β-D-maltoside (Cymal-6; Anatrace). Crystals of XylE-WW were grown at 18 ℃ by the hanging-drop vapor diffusion method by mixing 1 μl protein with 1 μl reservoir buffer containing 0.1 M NaCl, 0.1 M Li2SO4, 0.1 M MES pH 6.5, 30% PEG400 (v/v). Crystals appeared in 2-3 days and grew to full size in about 1 week. The crystals were directly flash-frozen in a cold nitrogen stream at 100 K for further data collection.
The data sets were collected at SSRF beamline BL17U and integrated and scaled using HKL2000 2 . Further processing was carried out using CCP4 suite 3 . The phase was solved by molecular replacement using PHASER 4 with XylE (PDB code: 4GBY) as a searching model. The model was further rebuilt in COOT 5 and refined with PHENIX 6 . Data collection and structure refinement statistics are summarized in Supplementary Table S2. purified with same methods as described previously. After protein was eluted from Ni 2+ -NTA affinity chromatography column, 1.5 mM CuCl2 was added into protein elution to initiate crosslinking. The crosslinking system was shaken at room temperature for 2 h. Gel filtration with Superdex-200 column was followed to remove the Cu 2+ . After gel filtration, the crosslinked XylE protein was transferred to buffer containing 25 mM MES pH 6.5, 150 mM NaCl and 0.056% Cymal-6 for further isothermal titration experiment.
The validation of crosslinking was conducted through DrICE protease digestion. The crosslinked protein was mixed with DrICE protease with ratio 100:1 (w: w). Then the mixture was incubated at room temperature for 3 h. After digested protein was reduced by 200 mM DTT for 10 min at room temperature. These protein samples were applied to SDS-PAGE to exam whether they were crosslinked.

Preparation of liposomes and proteoliposomes.
The proteoliposomes were made by following steps. E. coli polar lipids (Avanti) were dissolved in chloroform/methanol mixture (3:1, v/v) and dried with nitrogen gas.
After that, liposomes were mixed with 200 μg/ml XylE or mutants protein and S5 incubated for another 1 h at 4 ℃. After protein was integrated into liposomes, β-OG was removed by incubation overnight with 400 mg/ml Bio-Beads SM2 (Bio-Rad).
Then the proteoliposomes were frozen and thawed for 5 times and extruded with 0.4 μm membrane filter (Millipore) for 21 time. The homogenized proteoliposomes were ultracentrifugated at 100,000 g for 1 h and rinsed with ice-cold KPM 6.5 buffer for 2 times to remove extra sugar. Finally, the proteoliposomes were resuspended in ice-cold KPM 6.5 buffer to 100 mg/ml before the counterflow assay. Control liposomes were made with the same protocol except for protein insertion step.
The specific radioactivity of D-3 H-xylose was 20 Ci/mmol and final concentration of the external D-3 H-xylose was 0.5 μM. The uptake of radiolabeled substrates was allowed for 30 s. Then proteoliposomes were rapidly filtered with 0.22 μm filters (Millipore) and washed with 2 ml ice-cold KPM 6.5 buffer. The filter was then taken for liquid scintillation counting. All experiments were repeated for three times. Error bars represent s.d.

S6
The binding affinity between wild type XylE and xylose was measured with an ITC200 micro calorimeter (MicroCal). The wild type XylE and XylE mutant proteins were purified as previous description. In gel filtration step, all the XylE protein for ITC test was transferred to buffer containing 25 mM MES pH 6.5, 150 mM NaCl and 0.056% Cymal-6. The peak fractions were pooled and concentrated by Centricon   Supplementary Table   S1.

GLUTs inhibitors and XylE variants.
Binding affinity measurement between GLUTs inhibitors and XylE variants through MicroScale Thermophoresis. The data was fitted by the NT analysis 1. 5