(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Structural basis for substrate recognition and inhibition of human glucose-6-phosphate transporter SLC37A4 [1]

['Hui Li', 'Gmu-Gibh Joint School Of Life Sciences', 'Guangdong Provincial Key Laboratory Of Protein Modification', 'Disease', 'The Guangdong-Hong Kong-Macao Joint Laboratory For Cell Fate Regulation', 'Diseases', 'Guangzhou Medical University', 'Guangzhou', 'Guangdong', 'Guangzhou National Laboratory']

Date: 2026-08 

Abstract Glucose 6 phosphate (G6P) homeostasis is essential for maintaining blood glucose levels and coordinating anabolic and catabolic pathways. A key step in this process is the delivery of G6P into the endoplasmic reticulum (ER), where it is hydrolyzed by glucose 6 phosphatase to glucose and inorganic phosphate (Pi). This transport step is carried out by the ER carrier SLC37A4 (also known as the G6P transporter, G6PT), which imports G6P into the ER lumen while exporting Pi to the cytosol, and loss-of-function mutations in SLC37A4 cause glycogen storage disease type Ib. Despite its central role in G6P homeostasis, how SLC37A4 recognizes G6P and couples its transport to Pi antiport has remained unclear. Here we report cryo-electron microscopy structures of human SLC37A4 in three states: the apo form at 2.8 Å resolution, a G6P-bound state at 3.2 Å resolution and a chlorogenic acid (CHA) bound state at 3.3 Å resolution. SLC37A4 adopts the canonical Major Facilitator Superfamily fold and harbors a central, positively charged cavity that accommodates anionic substrates. In the G6P-bound structure, SLC37A4 adopts an outward-open conformation facing the ER lumen, in which G6P binds to the electropositive pocket. In the CHA-bound structure, SLC37A4 adopts an inward-facing conformation, with CHA bound at a cytosolic site that locks the transporter in an arrested state and prevents the conformational transitions required for G6P/Pi exchange. Combined with thermostability and transport-based analyses of G6P binding and disease variants, these structures support a rocker switch mechanism in which electrostatic neutralization of the central positively charged cavity by anionic substrate drives the conformational changes underlying G6P/Pi exchange. Together, these findings define the structural basis of G6P/Pi exchange by SLC37A4, provide a molecular rationale for pathogenic mutations in glycogen storage disease type Ib, and provide a framework for targeting SLC37A4 to modulate G6P homeostasis.

Citation: Li H, Peng X, Huang Y, Wu W, Li N, Cheng Z, et al. (2026) Structural basis for substrate recognition and inhibition of human glucose-6-phosphate transporter SLC37A4. PLoS Biol 24(7): e3003833. https://doi.org/10.1371/journal.pbio.3003833 Academic Editor: Raimund Dutzler, University of Zurich, SWITZERLAND Received: December 26, 2025; Accepted: May 18, 2026; Published: July 28, 2026 Copyright: © 2026 Li et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The cryo-EM maps of human SLC37A4-apo, SLC37A4-G6P, and SLC37A4-CHA have been deposited into the Electron Microscopy Data Bank (EMDB) under accession codes EMD-67847, EMD-67869, and EMD-67846, respectively. The atomic coordinates of human SLC37A4-apo, SLC37A4-G6P, and SLC37A4-CHA have been deposited into the Protein Data Bank (PDB) under accession codes 21NQ, 21OR, and 21NP, respectively. All other data underlying the graphs and charts in this manuscript, including source data for Fig 2 and S5 Fig, are available in the Supporting information file S1 Data.xlsx. Funding: This work was supported by the National Natural Science Foundation of China (grant 32501083 to X.W.), start-up funds from the Guangzhou National Laboratory (grant GZNL2025C01025 to X.W.), and the Pearl River Talent Recruitment Program (grant 2023QN10Y326 to X.W.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Abbreviations: CHA, chlorogenic acid; CHS, cholesteryl hemisuccinate; cryo-EM, cryo-electron microscopy; CTD, C-terminal domain; CV, column volumes; ER, endoplasmic reticulum; ESI, electrospray ionization mode; FSC, Fourier shell correlation; G6P, glucose-6-phosphate; G6PT, glucose-6-phosphate transporter; GSD Ib, glycogen storage disease type Ib; HBSS, Hanks’ Balanced Salt Solution; LC–MS, liquid chromatography–mass spectrometry; LMNG, lauryl maltose neopentyl glycol; MFS, Major Facilitator Superfamily; MRM, multiple reaction monitoring; NTD, N-terminal domain; Pi, inorganic phosphate; RMSD, root-mean-square deviation; TMs, transmembrane helices

Introduction Glucose-6-phosphate (G6P) serves as a critical metabolic hub at the intersection of several major pathways, seamlessly linking glycolysis, the pentose phosphate pathway, and glycogen metabolism to maintain cellular energy balance [1–3]. Following cellular uptake, free glucose is immediately phosphorylated to G6P by the rate-limiting enzymes hexokinase and glucokinase, a phosphorylation event that effectively traps glucose within the cell [4–6]. While cytosolic metabolism allows G6P to direct carbon flux toward immediate energy production, redox balance, or biosynthesis, G6P homeostasis crucially depends on its transport into the endoplasmic reticulum (ER) lumen to support systemic glucose release. This translocation is mediated by SLC37A4 (also known as the glucose-6-phosphate transporter, or G6PT), which functions as an obligate antiporter, moving cytosolic G6P into the ER while simultaneously exporting inorganic phosphate (Pi) back to the cytosol [7,8]. Once inside the ER, G6P is hydrolyzed by glucose-6-phosphatase into glucose and Pi, which are then returned to the cytosol to enable glucose efflux into the circulation. The physiological necessity of SLC37A4/G6PT is illustrated by the consequences of its failure, particularly in the context of glycogen storage disease type Ib (GSD Ib). Dysfunction of SLC37A4 disrupts the delicate transport process, leading directly to this rare, autosomal recessive disorder [9–12]. Despite the critical role of SLC37A4 in human physiology and disease, the molecular mechanisms by which it selectively recognizes G6P, coordinates the precise exchange of G6P and Pi, and undergoes the necessary conformational transitions for transport have historically remained poorly understood. In this study, we report high-resolution cryo-electron microscopy (cryo-EM) structures of human SLC37A4 in three distinct functional states: the apo state, the G6P-bound state, and an inhibitor-bound state complexed with chlorogenic acid (CHA). By integrating these structural insights with thermostability-shift and transport assays characterizing G6P binding and function, we elucidate the molecular mechanism of substrate recognition and conformational cycle underlying G6P transport, demonstrating how the transporter navigates between inward- and outward-facing conformations to facilitate the exchange of G6P and Pi. The CHA-bound structure further reveals a cytosolic inhibitory site and defines how CHA prevents the conformational transitions required for transport. Ultimately, this work provides a comprehensive structural framework for understanding SLC37A4 function, offers a molecular rationale for pathogenic mutations in GSD Ib, and lays the foundation for therapeutic development targeting SLC37A4.

Discussion In this study, we determined high-resolution cryo-EM structures of human SLC37A4 in apo, G6P-bound and CHA-bound states. Based on these structural snapshots together with our thermostability data, we propose a stepwise model for the G6P/Pi exchange mechanism (Fig 4C). The cycle likely begins from an inward-open conformation in which the basic cluster is arranged to favor high-affinity G6P binding from the cytosol. Substrate engagement is expected to promote compaction of the cytosolic cavity and formation of an occluded intermediate, in which both gates are transiently closed. Transition from this occluded state to an outward-facing conformation then proceeds through the canonical rocker-switch motion characteristic of Major Facilitator Superfamily transporters, reorienting the same electropositive cavity toward the ER lumen. In the outward-facing state, subtle rearrangements of the basic residues would lower G6P affinity, promoting its release for hydrolysis by glucose-6-phosphatase, while leaving the cavity primed for luminal Pi binding. Engagement of Pi would favor the reverse transition back toward the inward-open state, completing the exchange and resetting the transporter for another cycle. Our inhibitor-bound structure illustrates how these conformational changes can be selectively interrupted. CHA stabilizes an inward-facing state that remains competent for cytosolic access yet is conformationally trapped and unable to progress toward the occluded or outward-facing configurations required for transport (Fig 4C). Rather than simply competing with G6P at the same coordination site, CHA engages an adjacent hydrophobic–aromatic pocket at the cytosolic entrance and locks inter-domain contacts. This mode of inhibition reveals a druggable site that could be exploited to modulate G6P flux. To date, 54 pathogenic mutations have been reported in SLC37A4 [9]. By leveraging the structural roles identified here, these known pathogenic missense mutations in GSD-Ib patients can be categorized into three distinct classes. First, substrate recognition is impaired by variants such as R28C/H, K29Q, K64Q, and K240Q, which compromise the positively charged binding cavity (Fig 2E), or by Y60C and F237S, which abolish essential hydrophobic interactions stabilizing the sugar moiety (Fig 2B and 2F). This interpretation is supported by thermostability assays showing that mutations at these key coordinating residues (including R28, K29, Y60, K64, and K240) significantly reduce G6P-induced thermal shifts relative to the wild type (Fig 2G). Furthermore, these structural and thermostability defects directly translate to a severe loss of transport function, as demonstrated by the marked reduction in [13C 6 ]G6P uptake for these mutants (Fig 2H). Second, the outward-open conformation is destabilized by mutations including D72N, T364M, R126C, T354K, T136M, and D285N, which break critical salt bridges and hydrogen bonds locking the NTD and CTD (Fig 2D). Third, the inward-facing state is compromised by variants such as D47N, K46E, K389E, A388P, T157I, and S263P, which prevent tight sealing of the ER luminal gate (Fig 3B and 3C). The remaining unmapped mutations likely impair overall structural integrity and folding and will require further investigation; together, these mechanistic classes provide a clear molecular rationale for the loss of function observed in patients. During the preparation of this manuscript, independent studies describing SLC37A4 structures in various conformational states were published [21,22]. Our models exhibit remarkable convergence with these external datasets; notably, the overall structure of our SLC37A4-G6P complex is highly similar to those from the two previously published studies (S9A and S9B Fig), and the global architectures and the specific binding mode of CHA are virtually identical across all studies (S8 Fig). In contrast to this conserved inhibitor binding, the precise coordination of G6P varies considerably among the models (S9C Fig). Interestingly, while the position of the H366 side chain remains largely unchanged across all three structures, its involvement in substrate binding differs dramatically. In our structure (SLC37A4-G6P), the phosphate group interacts with K64 and is close to Y60, but the glucose moiety points away, leaving H366 at a distance of 10.7 Å with no direct interaction (S9C Fig, left). Conversely, in PDB 9W1P, the substrate appears inverted, with the glucose ring forming close contacts (2.5–2.8 Å) with Y60, K64, and H366 (S9C Fig, middle). Meanwhile, in PDB 9KVV, the phosphate group remains near Y60 and K64 (similar to our model), but the glucose moiety is angled upwards to form a 3.3 Å interaction with H366 (S9C Fig, right). Because G6P in our structure lacks this additional coordination from H366, it exhibits a looser binding mode involving fewer overall protein contacts. We interpret this loosely bound G6P in the outward-open conformation as representing a distinct intermediate state along the transport cycle, specifically capturing a transient moment of G6P release. Although the exact cause is unclear, subtle differences in sample preparation or lipid environments likely trapped these distinct conformational snapshots. Ultimately, despite these local variations in substrate coordination, the high consistency in overall architecture across independent investigations serves as a powerful cross-validation of our findings. This collective agreement establishes a definitive structural framework for SLC37A4, providing a highly reliable foundation for interpreting disease-associated mutations and guiding rational drug design for glycogen storage disease type Ib.

Materials and methods Cloning, protein expression and purification The cDNA encoding full length SLC37A4 was cloned into the pCAG vector with a C-terminal 3× Flag tag; all mutants were generated via standard PCR-based mutagenesis. For protein expression, HEK293F cells were maintained in suspension culture using SMM293-TII medium (Sino Biological) at 37°C under 8% CO 2 . When cell density reached 2 × 106 cells/mL, the cultures were transfected with plasmid premixed with PEI at a 1:3 mass ratio. Cells were harvested by centrifugation at 2,000g 48 h post-transfection. The cell pellets were flash-frozen in liquid nitrogen and stored at −80 °C until further use. All purification steps were performed at 4 ℃ or on ice. Cell pellets were thawed on ice-water bath and resuspended in Buffer A (25 mM HEPES pH 7.5, 150 mM NaCl, protease inhibitor cocktail containing 0.8 μM pepstatin A, 2 μM leupeptin, 2 μM aprotinin) and lysed by sonication on ice (10 s pulses, 20 s intervals, 30% amplitude). Membrane fractions were isolated by ultracentrifugation at 158,600g for 1 h. The membrane pellets were solubilized in Buffer B (25 mM HEPES pH 7.5, 150 mM NaCl, 1% (w/v) LMNG and 0.1% (w/v) cholesteryl hemisuccinate (CHS)). The insoluble material was removed by ultracentrifugation at 158,600g for 1 h. The clarified supernatant was incubated with anti-FLAG affinity resin (Smart-Lifesciences) for 1 hr. The resin was washed with 10 column volumes (CV) of Buffer C (25 mM HEPES pH 7.5, 150 mM NaCl, 0.005% [w/v] LMNG, 0.0005% [w/v] CHS). The proteins were eluted by Buffer C supplemented with 0.25 mg/mL 3× FLAG peptide (Sangon). The eluate was concentrated and further purified by gel filtration using Superdex 200 Increase 10/300 GL column (Cytiva) pre-equilibrated with Buffer D (25 mM HEPES pH 7.5, 150 mM NaCl, 0.005% LMNG, 0.0005% CHS). Peak fractions were collected and concentrated to 10–15 mg/mL for cryo-EM grid preparation. For SLC37A4 nanodisc, MSP protein and lipid extract mixture were added to the FLAG resin and incubated overnight at 4 ℃. Detergent was removed by biobeads, protein was then eluted and further purified with gel filtration using Superdex 200 pre-equilibrated with Buffer E (25 mM HEPES pH 7.5, 150 mM NaCl). In parallel, an SLC37A4–GFP fusion construct was cloned and purified using the same expression and purification pipeline, and used for thermostability assays. Cryo-EM grid preparation For cryo-EM sample preparation, freshly purified SLC37A4 at a concentration of 10–15 mg/mL was applied to glow-discharged UltrAufoil R1.2/1.3 Au 300-mesh holey carbon grids. For the ligand-bound samples, the protein was incubated with 100 mM G6P (for SLC37A4–G6P) or 10 mM CHA (for SLC37A4–CHA) for 1 hour on ice before grid preparation. Aliquots of 3 μl of sample were applied to the grids in a Vitrobot Mark IV (Thermo Fisher Scientific) operating at 4 °C and 100% relative humidity, blotted for 3.5 s with a blot force of 1, and rapidly plunge frozen into liquid ethane cooled by liquid nitrogen. Frozen grids were stored in liquid nitrogen until data collection. Cryo-EM data processing and structure refinement For SLC37A4 apo data set, the Cryo-EM data were collected on Titan Krios microscope equipped with Falcon 4 camera. Motion correction was performed with patch motion correction in CryoSPARC to correct beam-induced motion [23]. Following motion correction, the contrast transfer function of micrographs was estimated. Micrographs with CTF fit resolution worse than 4 Å were discarded. Initial particle picking was performed using blob pick with particle diameter of 80–120 Å. These particles were subjected to multiple rounds of 2D classification to generate high-quality 2D templates with visible secondary structure feature. These 2D class averages were used for template-based particle picking. To maximize the recovery of high-quality particles, the selected particles were used to train a convolutional neural network model using Topaz, which then picked additional particles missed by blob pick and template pick [24]. Particles from blob pick, template pick and Topaz were merged and duplicates were removed. The merged particles set underwent multiple rounds of 2D classification, only classes with clear secondary structure were kept for ab initio reconstruction and heterogeneous refinement. Given that SLC37A4 is a small membrane protein lacking a large soluble domain, a seeding strategy was employed in the heterogeneous refinement to enlarge particle pool [25]. Good particles were selected for NU-refinement, yielding a map with a resolution of 3.0 Å. The particle coordinates were imported into Relion and the particles were re-extracted [26,27]. A final round of 3D auto-refinement was performed, followed by Bayesian polishing and CTF refinement, which improved the overall resolution to 2.8 Å. The same data processing strategy was applied to both the SLC37A-G6P and SLC37A-CHA datasets. For model building, an initial structural model of SLC37A4 was obtained via AlphaFold prediction [28]. The predicted model was rigid-body fitted into the cryo-EM reconstruction map using UCSF chimera [29]. Regions of the predicted model that did not agree with the experimental density were manually rebuilt using COOT [30]. The model was finalized through iterative cycles of manual adjustment and automatic real-space refinement using Phenix (phenix.real_space_refine, applying secondary structure and geometry restraints) [31]. All representations of cryo-EM density and structural models were prepared with Chimera [29] and PyMOL [32]. A summary of data collection, processing, and model-refinement statistics for all three data sets is provided in Table 1. Thermal shift analysis For thermal shift analysis of SLC37A4, a GFP-based assay was employed in which GFP was fused to the C-terminus of SLC37A4. The SLC37A4–GFP fusion proteins (wild-type and mutants) were purified using the same procedure described above. The GFP thermal shift assays were carried out according to previous studies [33,34]. Purified GFP-tagged SLC37A4 was diluted to 0.6 µM in assay buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 1% (w/v) LMNG) in a final volume of 90 µl per well and incubated in the presence or absence of 10 mM G6P for 10 min. After incubation, 10 µL of 10% (w/v) β-n-octyl-β-d-glucopyranoside was added to each well. Samples were then heated to 10, 20, 30, 40, 50, 60, 70, or 80 ℃ for 10 min, centrifuged to remove aggregated protein, and the supernatant was transferred to a black 96-well plate. Soluble GFP fluorescence was measured on a plate reader (excitation 488 nm, emission 512 nm), and values were normalized to the signal at the lowest temperature condition. To generate thermal transition curves, the fluorescence data were normalized by setting the highest and lowest data points to 100% and 0%, respectively. The normalized fluorescence values were plotted against temperature, and apparent melting temperatures (Tm) for each condition were determined by fitting the data to a sigmoidal four-parameter logistic function using GraphPad Prism. To evaluate whether the mutated residues affected G6P binding, the shift in melting temperature (ΔTm) was calculated for each protein by subtracting the Tm of the apo state from the Tm of the G6P-incubated state. [¹³C]G6P uptake assay in G6PT-expressing cells For the cellular G6P uptake assay, the assay was preformed based on previously described method [22] with minor modification. HEK293F cell were transfected with SLC37A4 plasmids (WT or mutants) premixed with PEI at 1:3 mass ratio as described in the protein purification section. Forty-eight hours post-transfection, cells were pelleted by centrifugation at 1,000g for 5 min, washed once with Wash Buffer A1 (1× Hanks’ Balanced Salt Solution [HBSS, Gibco] supplemented with 2% FBS), and permeabilized with Permeabilization Buffer A2 (1× HBSS supplemented with 10 µM digitonin) at 37 °C for 10 min. This low concentration of digitonin was utilized to selectively permeabilize the cholesterol-rich plasma membrane while maintaining the integrity of the ER membrane. Following permeabilization, cells were washed with 1× HBSS twice to remove residual buffer. Uptake was initiated by adding a 1:1 molar mixture of [¹³C 6 ]G6P and unlabeled G6P (MedChemExpress) to a final concentration of 1 mM. After a 10-min incubation at 37 °C, the reaction was quenched by washing the cells twice with 500 µL of ice-cold 1× HBSS. Intracellular [¹³C 6 ]G6P was extracted using 200 µL of pre-chilled methanol: acetonitrile (2:1, v/v), followed by ultrasonication at 4 °C for 10 min. The extract was centrifuged at 12,000 rpm for 20 min at 4 °C, and 150 µL of the resulting supernatant was transferred to a fresh tube and lyophilized for subsequent quantitative LC–MS analysis. The remaining cell pellet was lysed in 400 µL of 0.1 M KOH, and the protein concentration was determined using a BCA protein assay kit (Yeasen). [¹³C 6 ]G6P uptake levels were normalized to the corresponding total protein content. All experiments were performed at least in triplicate, and data were analyzed using nonlinear regression in Prism 8.0.1 (GraphPad Software). Quantification of [¹³C 6 ]G6P by LC–MS Lyophilized [¹³C 6 ]G6P samples were quantified using liquid chromatography–mass spectrometry (LC–MS). Samples were resolved on an Agilent 1290 Infinity II-6495B UPLC–MS system equipped with an ACQUITY UPLC BEH Amide column (2.1 mm × 100 mm, 1.7 μm). The mobile phase consisted of acetonitrile:water (9:1, v/v) supplemented with 15 mM ammonium acetate and 0.3% ammonium hydroxide. Detection was performed in negative electrospray ionization mode (ESI−), and [¹³C 6 ]G6P was monitored via multiple reaction monitoring (MRM) at m/z 264.9. Data acquisition and analysis were performed using MassHunter Workstation Data Acquisition (version 10.0 SR1) and Quantitative Analysis (version 10.0) software (Agilent Technologies), respectively. Compound identity was confirmed by matching retention times and MRM transitions to a reference standard. Quantification was achieved using an external calibration curve, with final concentrations calculated based on sample volume. Western blot HEK293F cells (1 × 106 cells/mL) expressing empty vector, wild-type G6PT, or mutants were harvested, washed with PBS, and lysed in 400 μL buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 1% [w/v] DDM, 0.1% [w/v] CHS, 2 µM pepstatin A, 4.2 µM leupeptin, and 0.8 µM aprotinin). After 1.5 h at 4 °C, lysates were cleared by ultracentrifugation. Solubilized proteins were resolved by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked for 1 h at room temperature in TBST containing 5% nonfat milk and then incubated overnight at 4 °C with rabbit monoclonal anti-Flag antibody (1:3,000; Cell Signaling Technology, Cat#14793) and anti-calnexin antibody (1:1,000; Cell Signaling Technology, Cat#2679S), which served as an internal loading control. After washing, membranes were incubated with HRP-conjugated secondary antibodies for 45 min at room temperature. Protein bands were visualized by chemiluminescence using an iBright FL1500 imaging system.

Acknowledgments We thank the staff at the Guangzhou National Laboratory Center for Biological Imaging for their technical support during cryo-EM grid screening and data acquisition. We are particularly grateful to M. Z., M. L., and W. L. for their assistance with sample preparation and data collection. We also thank C. Q. and Z. M. for their assistance with mass spectrometry experiments.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3003833

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/