Edit · 2026 01 11 201353 Nature S41594 025 01712 4

Saving needs a token. Log in or register to get one. Edits are public and licensed CC-BY-SA-4.0.

# Structural basis of glucose-6-phosphate transport by human SLC37A2 | Nature Structural & Molecular Biology

**URL:** https://www.nature.com/articles/s41594-025-01712-4

**Saved:** 2026-01-11T04:13:58.252Z

---

Article
Published: 12 November 2025
Structural basis of glucose-6-phosphate transport by human SLC37A2
Qinxuan Lai, Mengzhen Xu, Yongxiang Gao, Zhisen Yang, Linfeng Sun & Xin Liu

Nature Structural & Molecular Biology (2025)Cite this article

1497 Accesses

2 Altmetric

Metrics
details

Abstract

Glucose-6-phosphate (G6P) transporters are crucial for glucose metabolism by mediating G6P transport from the cytoplasm to endoplasmic reticulum (ER). However, their transport mechanisms remain poorly understood. Here, we elucidate the structural and functional basis of human solute carrier family 37 member 2 (SLC37A2), a G6P transporter implicated in metabolic regulation and macrophage inflammation. We show that SLC37A2 functions as a uniporter, facilitating G6P transport independent of inorganic phosphate gradients. Structures of SLC37A2 in the apo and G6P-bound states reveal a dimeric architecture. Both the ER luminal-open and cytosolic-open structures are captured, showing the structural dynamics during G6P transport. G6P is coordinated by SLC37A2 through interactions with its phosphate and hydroxyl groups. Furthermore, mapping mutations associated with glycogen storage disease type Ib onto SLC37A2 highlights residues essential for transport activity. Together, this work provides structural insights into G6P transport and establishes a framework for understanding related metabolic disorders.

You have full access to this article via your institution.

Download PDF
Similar content being viewed by others
Structural basis of G6P/Pi transport and inhibition in SLC37A4
Article 12 November 2025
Structures of human glucose-6-phosphate transporter reveal reciprocal antiport mechanism driving glucose-6-phosphate and inorganic phosphate exchange
Article Open access
11 December 2025
The induced-fit and catalytic mechanisms of human G6PC1
Article Open access
15 July 2025
Main

Glucose-6-phosphate (G6P) is a pivotal intermediate in cellular metabolism, acting as a central hub for multiple essential biochemical pathways, including glycolysis, gluconeogenesis, glycogen synthesis and breakdown, the pentose phosphate pathway and de novo lipogenesis1. G6P is generated and predominantly localized in the cytoplasm but can also be transported to other organelles, particularly the endoplasmic reticulum (ER)2. The final common step in both gluconeogenesis and glycogenolysis occurs within the ER lumen, where G6P is hydrolyzed into free glucose and inorganic phosphate (Pi) through ER-membrane-associated G6Pase catalytic subunit (G6PC) catalysis3. However, given its high polarity, G6P cannot freely diffuse across the lipid bilayer and specialized transport mechanisms are required4,5. G6P transporters (G6PTs) have crucial roles in this process by facilitating the transport of cytosolic G6P into the ER lumen2,6,7. Both G6P translocation and hydrolysis are essential for glucose metabolism, as deficiencies in either G6PCs or G6PTs lead to the onset of human glycogen storage disease type I (GSD-I, also known as von Gierke disease)8.

G6PT1 (also known as solute carrier family 37 member 4, SLC37A4) is the most studied G6PT family member and is widely expressed in various tissues and organs, especially the main gluconeogenic organs, including the liver, kidney and intestine9,10. G6PT1 and G6PCs have been proposed to form an ER-membrane-associated, multicomponent complex that coordinates G6P import and hydrolysis and the subsequent export of glucose and Pi11,12,13. In the liver, kidney and intestine, G6PT1 and G6PC1 form the functional complex G6Pase-α, which is essential for maintaining blood glucose homeostasis, whereas, in neutrophils and macrophages, G6PT1 associates with G6PC3 to form the G6Pase-β complex, which regulates energetic homeostasis and neutrophil function14,15. G6PT1 deficiency leads to GSD-Ib, which is characterized by fasting hypoglycemia, hepatomegaly, nephromegaly, hyperlipidemia, hyperuricemia and lactic acidemia16. To date, approximately 91 mutations have been identified in the G6PT1 gene17, most of which lead to significantly reduced G6P uptake7,18,19,20. Emerging evidence also suggests that G6PT1 is involved in autophagy21, lipid metabolism22, neutropenia, bone marrow dysfunction23 and cancer progression24.

SLC37A2 (also known as G6PT3) is a close homolog of G6PT1 (referred to hereafter as SLC37A4); along with two other transporters, SLC37A1 (also known as G6PT2) and SLC37A3, forms the SLC37 family (Extended Data Fig. 1a–c), which belongs to the organophosphate/Pi antiporter family of the major facilitator superfamily (MFS)25,26,27. SLC37A2 and SLC37A4 share 23.7% sequence identity (Extended Data Fig. 1a,b). Unlike SLC37A4, SLC37A2 is expressed at low levels in liver and kidney and is predominantly expressed in macrophages28, suggesting tissue-specific metabolic regulation of the SLC37 family14. SLC37A2 has a crucial role in the activation of inflammatory macrophages through metabolic regulation28,29. SLC37A2 deficiency has been linked to disrupted macrophage metabolism, impaired oxidized low-density lipoprotein uptake and defective apoptotic cell clearance, leading to an increased risk of atherosclerosis30. Additionally, SLC37A2 has been suggested to be involved in bone metabolism31, adipocyte differentiation28, vitamin D signaling32,33, mast cell regranulation34 and phagosome shrinkage35.

To date, the widespread view is that SLC37A1, SLC37A2 and SLC37A4 are Pi-linked G6P antiporters, capable of both Pi/Pi and G6P/Pi exchange, whereas SLC37A3 lacks these abilities36. It has been reported that SLC37A1, SLC37A2 and SLC37A4, when solubilized from COS-1 cell microsomes and reconstituted into proteoliposomes, can efficiently import G6P or Pi only in the presence of a high intraluminal Pi concentration (50 mM)36,37. SLC37A4-mediated G6P transport can be inhibited by chlorogenic acid (CHA) and vanadate, whereas SLC37A1 and SLC37A2 are insensitive to CHA36,37. However, an alternative view suggests that SLC37A4 operates as a facilitative, bidirectional G6P uniporter38. Previous data demonstrated that SLC37A4-mediated G6P import was not affected by preloading Pi in liver microsomal vesicles and could occur in the absence of Pi38. Vanadate was also shown to have little inhibitory effect on G6P import38. Thus, the transport mechanism of G6PTs remains unknown.

In this study, we characterize the G6P transport by human SLC37A2 and SLC37A4 and the results support that both SLC37A2 and SLC37A4 function as G6P uniporters. Moreover, we determine the structures of SLC37A2 in the apo state and in the presence of G6P using cryo-electron microscopy (cryo-EM). Both the ER luminal-open and the cytosolic-open conformations are captured, providing critical insights into the transport dynamics underlying G6P transport.

Results
Characterization of G6P transport by SLC37A2 and SLC37A4

G6P transport by SLC37 family transporters has been characterized through the use of isolated microsomes expressing the transporter or reconstituted proteoliposomes with proteins solubilized from the isolated microsomes36,37. To avoid possible influences from endogenous factors, we first tried to express and purify the wild-type (WT) SLC37A2 and SLC37A4 proteins using the human embryonic kidney 293F (HEK293F) cell-based transient expression system. Both SLC37A2 and SLC37A4 proteins were obtained in high yields, with approximately 1 mg L−1 of cells for SLC27A2 and 0.5 mg L−1 of cells for SLC37A4, and exhibited good homogeneity following gel filtration (Extended Data Fig. 2a). Notably, on SDS–PAGE, SLC37A2 exhibited an apparent molecular weight between 55 kDa and 95 kDa, markedly higher than its calculated mass of 54.4 kDa, and appeared as a diffuse smear, indicating possible glycosylation (Extended Data Fig. 2b). To assess the glycosylation status, we treated the purified SLC37A2 and SLC37A4 proteins with PNGase F. The deglycosylation treatment resulted in a clear downward shift of SLC37A2 to a discrete band near its theoretical molecular weight, whereas SLC37A4 displayed only a modest mobility shift (Extended Data Fig. 2b). These results indicate that SLC37A2 undergoes more extensive glycosylation than SLC37A4. With the purified proteins in detergent micelles, we established a proteoliposome-based transport assay using 3H-radiolabeled G6P (3H-G6P). We first tested G6P transport according to the assay conditions reported by Zilberstein et al.36 The results revealed that both SLC37A2 and SLC37A4 could efficiently transport 3H-G6P (Extended Data Fig. 2c). We then compared 3H-G6P import in the presence and absence of a cross-membrane Pi gradient by changing the inside buffer and outside reaction buffer. The results revealed that proteoliposomes preloaded with SLC37A2 or SLC37A4 had similar 3H-G6P accumulation, which was more than that observed for the control liposomes with no protein loaded under both conditions (Fig. 1a). No significant differences were detected between 3H-G6P transport with and without a Pi gradient for either SLC37A2 or SLC37A4 (Fig. 1a). This result is consistent with that reported by Marcolongo et al.38, suggesting that SLC37A2 and SLC37A4 can function as uniporters to mediate G6P transport. Time-course analysis of 3H-G6P transport by SLC37A2 and SLC37A4 in the absence of Pi revealed a gradual accumulation of radiolabeled substrate (Fig. 1b). We then tested the dose-dependent accumulation in proteoliposomes preloaded with SLC37A2 or SLC37A4 at different G6P concentrations (adjusted by mixing radiolabeled and nonradiolabeled G6P). The transport kinetics were determined by the Michaelis–Mention equation. The Km values were 18.64 ± 1.62 mM for SLC37A2 and 21.45 ± 1.60 mM for SLC37A4 and the Vmax values were 318.3 ± 8.52 nmol mg−1 protein min−1 for SLC37A2 and 487.6 ± 11.84 nmol mg−1 protein min−1 for SLC37A4, respectively, indicating their weak affinity for G6P and that these transporters operate efficiently only in the presence of high G6P concentrations (Fig. 1c).

Fig. 1: Characterization of G6P transport by SLC37A2 and SLC37A4.

a, Examination of the effect of the Pi gradient on G6P transport by SLC37A2 and SLC37A4, as determined by a proteoliposome-based 3H-G6P transport assay. b, Time course of G6P transport by SLC37A2 and SLC37A4. In a and b, control refers to liposomes without protein. Significant differences were evaluated using a two-tailed unpaired t-test. NS, not significant; ****P < 0.0001. The exact P values are labeled above the columns. c, Determination of the Km for SLC37A2 and SLC37A4. The Km values were obtained by fitting using the Michaelis–Menten model. d, CHA competition assay for G6P transport by SLC37A2. e, Vanadate competition assay for G6P transport by SLC37A2. f, Pi competition assay for G6P transport by SLC37A2. g, CHA competition assay for G6P transport by SLC37A4. h, Vanadate competition assay for G6P transport by SLC37A4. i, Pi competition assay for G6P transport by SLC37A4. j. Sugar and sugar phosphate competition assay for G6P transport by SLC37A2. Fru, fructose; Glc, glucose; Gal, galactose; Man, mannose; Fuc, fucose. In d–j, the results were normalized to the net 3H-G6P accumulation of SLC37A2 or SLC37A4 in the absence of chemicals other than 3H-G6P in the outer buffer. Significant differences were determined using one-way ANOVA with Dunnett’s multiple-comparisons test. ****P < 0.0001. The other exact P values are labeled on the figures. For all transport assays, data are the means ± s.d. (n = 3 technical replicates). The chemical structures of CHA, vanadate, Pi, sugars and sugar phosphates are shown in Extended Data Fig. 2e,f.

Source data

Full size image

Vanadate, a structural and chemical analog of phosphate, has been reported to efficiently suppress the transport activity of SLC37A2 and SLC37A4, while CHA only inhibits SLC37A4 (refs. 36,37,39). To evaluate the effects of CHA, vanadate and Pi on G6P transport, we applied concentration gradients of each compound to the external buffer of proteoliposomes and monitored 3H-G6P transport by SLC37A2 and SLC37A4 (Fig. 1d–i). SLC37A2 exhibited insensitivity to CHA across the tested range, whereas SLC37A4 was inhibited in a concentration-dependent manner, retaining only 60% of transport activity at 48 mM CHA compared to untreated control, with a half-maximal inhibitory concentration (IC50) value of 10.01 ± 3.02 mM (Fig. 1d,g). In contrast, vanadate suppressed the activity of both transporters, reducing transport to approximately 40% for SLC37A2 and 30% for SLC37A4 at 186 mM, with IC50 values of 54.11 ± 8.76 mM and 50.15 ± 8.16 mM, respectively (Fig. 1e,h). Similarly, Pi reduced transport activity to about 20% for SLC37A2 and 30% for SLC37A4 at 200 mM, with corresponding IC50 values of 64.45 ± 10.18 mM and 61.52 ± 12.67 mM (Fig. 1f,i). These results indicate that vanadate and Pi act as relatively weak inhibitors of both transporters, while CHA selectively inhibits SLC37A4.

We also examined the effects of different sugars and sugar phosphates on 3H-G6P transport by SLC37A2 using a proteoliposome-based competition assay (Fig. 1j). Similar to the pronounced inhibition of nonradiolabeled G6P, the other six sugar phosphates, namely fructose-6-phosphate, glucose-1-phosphate, galactose-1-phosphate, mannose-1-phosphate, fucose-1-phosphate and glucose-1,6-phosphate, significantly inhibited 3H-G6P transport, with a greater than 70% reduction compared with the control group (Fig. 1j). In contrast, the tested hexoses and pentoses weakly inhibited 3H-G6P transport. Hexoses in the pyranose form, such as glucose, galactose, mannose, fucose and rhamnose, inhibited 3H-G6P transport by 30–40% (Fig. 1j). Furthermore, we tested the transport of 3H-glucose by SLC37A2 and SLC37A4 using the proteoliposome-based transport assay (Extended Data Fig. 2d). Although the accumulation of 3H-glucose of SLC37A2 and SLC37A4 differs from empty liposomes, the transport activity of SLC37A2 and SLC37A4 to glucose is weaker than that of G6P possibly because of the lack of a phosphate group that may be critical to the substrate binding. Pentoses in the pyranose form, such as xylose and arabinose, and a hexose in the furanose form (fructose) had even weaker inhibitory effects on 3H-G6P transport, with a reduction of less than 20% compared with that of the control group (Fig. 1j). Together, these results suggest that the phosphate moiety has a critical role in SLC37A2 substrate recognition. Notably, glucose-1,6-biphosphate, which contains two phosphate groups, exhibited stronger inhibition than G6P, further highlighting the role of the phosphate group in substrate binding. The cyclic monosaccharide moiety further enhances the binding affinity, as revealed by the weaker inhibition of Pi alone (Fig. 1f), and is independent of the configurations of the hydroxyl groups of the sugar moiety. However, whether SLC37A2 and SLC37A4 transport sugar phosphates other than G6P awaits further investigation.

Structure determination of SLC37A2 in the apo state

We prepared protein samples of SLC37A2 and SLC37A4 in different detergents and attempted to determine their structures using cryo-EM single-particle analysis. Trials for SLC37A4 were unsuccessful. Although the EM images revealed well-distributed particles, the two-dimensional (2D) classification and three-dimensional (3D) reconstruction results suggest that SLC37A4 is a monomer, making structure determination quite challenging (Extended Data Fig. 3a–c). In contrast, 2D classification of SLC37A2 clearly revealed features of the protein in a possible dimeric form (Extended Data Fig. 3d,e). After 3D classification and reconstruction, we obtained an EM map with C2 symmetry at an overall resolution of 3.2 Å, as estimated by the gold-standard Fourier shell correlation (FSC) = 0.143 criterion (Fig. 2a, Extended Data Figs. 3d–h and 4a, and Table 1). The high map quality allowed model building and assignment of most of the residues of SLC37A2, except for an internal loop (residues 246–279), because of the vague density in this region (Fig. 2b and Extended Data Fig. 4b).

Fig. 2: Structures of SLC37A2 determined in the apo state and in the presence of G6P.

a, Overview of the density map of SLC37A2 in the apo state. b, Overall structure of SLC37A2apo in the ER luminal-open state. c, A section view of the electrostatic potential surface of the SLC37A2apo monomer in the luminal-open conformation. d, Superposition of the dimer interfaces in the symmetric and asymmetric SLC37A2 conformations. e, Details of the SLC37A2 dimer interface. f, Proteoliposome-based 3H-G6P transport assay by SLC37A2WT and SLC37A2ΔN. Net 3H-G6P accumulation of the mutant calculated by deleting the 3H-G6P accumulation of the control group was normalized to the net accumulation of SLC37A2WT. Data are the means ± s.d. (n = 3 technical replicates). Significant differences between SLC37A2ΔN and SLC37A2WT were determined using a two-tailed unpaired t-test. The exact P value is labeled above the column. g, Density map and structure of the symmetric G6P-unbound SLC37A2G6P dimer in the ER luminal-open conformation. h, Density map and structure of the symmetric G6P-bound SLC37A2G6P dimer in the cytosolic-open conformation. The G6P molecules are shown in spheres. i, Density map and structure of the asymmetric SLC37A2G6P dimer. The G6P molecules are shown in spheres. j, Superposition of the luminal-open monomers in the symmetric and asymmetric SLC37A2 conformations. k, Superposition of the cytosolic-open monomers in the symmetric and asymmetric SLC37A2 structures.

Source data

Full size image
Table 1 Cryo-EM data collection, refinement and validation statistics
Full size table

SLC37A2 adopts a dimeric architecture with the symmetry axis perpendicular to the membrane (Fig. 2b). Each SLC37A2 monomer contains 12 transmembrane segments (TMs), TM1–TM12 (Fig. 2c). Its N and C termini are both located in the cytoplasm, as suggested by topological analysis of SLC37A4 (ref. 40). The N-terminal region forms a short helix (residues 7–14, named H0), which attaches to the membrane and lies almost perpendicular to the bilayer (Fig. 2d). The dimer interface forms between the N-terminal region of one monomer and the cytosolic region of TM3 and TM6 and the TM6–TM7 linker of the other through hydrogen bonds and hydrophobic packing interactions (Fig. 2e). For example, S3 of the N-terminal loop interacts with E231 and D235 in the TM6–TM7 linker by hydrogen bonding (Fig. 2e) and S4 interacts with R115 in TM3 (Fig. 2e). Notably, the N-terminal sequence involved in dimer formation is conserved in SLC37A1 and SLC37A2 but is not present in SLC37A3 and SLC37A4 (Extended Data Fig. 1a). This explains the different oligomeric states of SLC37A2 and SLC37A4 observed by cryo-EM. We generated a truncation mutant of SLC37A2 by deleting the N-terminal 13 residues (referred to as SLC37A2ΔN) (Extended Data Fig. 2a,b) and examined its transport activity. No significant differences were detected between 3H-G6P transport for the WT SLC37A2 or the truncation mutant, indicating that the dimer formation has little effect on the G6P transport by SLC37A2 although the oligomeric state of SLC37A2 in the proteoliposome remains to be further validated (Fig. 2f).

SLC37A2 adopts a canonical MFS fold with 12 TMs organized into two separate domains41; TM1–TM6 form the N-domain and TM7–TM12 form the C-domain. The two domains are connected by a long linker between TM6 and TM7 and can be superposed by two-fold pseudosymmetry with an axis perpendicular to the lipid bilayer (Extended Data Fig. 4c). Within each domain, the six TMs are organized into two inverted repeats, with the first three TMs intertwining with the last three TMs through an approximately 180° rotation around an axis parallel to the membrane plane (Extended Data Fig. 4d). On the cytosolic side, the TM6–TM7 linker and TM10–TM11 linker form an intracellular domain (Fig. 2b), which has been identified in many MFS transporters, especially those in the sugar transporter family, and is suggested to have essential roles during substrate transport41. On the ER luminal side, the TM1–TM2 linker (residues 50–83) is also longer (Fig. 2b). An ER luminal-facing cavity formed between the N-domain and C-domain, which has a negative surface potential at the bottom but a positive surface potential in the upper region of the cavity (Fig. 2c). These findings suggest that the captured apo structure of SLC37A2 is in the outward, ER luminal-open conformation (relative to the inward, cytosolic-open conformation).

Structure determination of SLC37A2 in the presence of G6P

To elucidate substrate recognition by SLC37A2, we added G6P to the purified protein for incubation before cryo-EM sample preparation. During 3D classification, three distinct classes were identified, with two maps exhibiting two-fold symmetry similar to that in the apo state and the other map being asymmetric. After data processing, three EM maps with an overall resolution of 3.5–3.7 Å were obtained (Extended Data Figs. 5 and 6 and Table 1). Structural models were built for the three SLC37A2 maps (Extended Data Figs. 7 and 8). Structural analysis revealed that the two SLC37A2 monomers exhibit different conformations in the three EM maps. In one map, both monomers are in the luminal-open state (referred to as SLC37A2G6P, LO), which is identical to the apo state (Fig. 2g). In the second map, both monomers exhibited rearranged N-domains and C-domains and the transporter adopted an inward, cytosolic-open conformation (referred to as SLC37A2G6P, CO) (Fig. 2h). In the cavity formed by the two domains of each monomer, a clear density was observed, to which G6P docked well (Figs. 2h and 3a). The third EM map revealed that the two monomers adopted different conformations (referred to as SLC37A2G6P, LO+CO), with one SLC37A2 monomer in the luminal-open conformation and the other in the cytosolic-open conformation (Fig. 2i). One of each of the two monomers in the asymmetric dimeric structure appear similar to those in the symmetric luminal-open and cytosolic-open structures. Structural alignment of the luminal-open monomers from the SLC37A2G6P, LO map and SLC37A2G6P, LO+CO revealed a root-mean-square deviation (r.m.s.d.) of 0.6 Å, whereas alignment of the cytosolic-open monomers from the SLC37A2G6P, CO map and SLC37A2G6P, LO+CO revealed an r.m.s.d. of 0.5 Å (Fig. 2j,k). The density of only one G6P moiety was clearly identified in the cytosolic-open monomer (Fig. 2i). Additionally, the dimer interface remained the same in all symmetric and asymmetric structures (Fig. 2d).

Fig. 3: G6P coordination of SLC37A2.

a, EM densities and molecular coordination of G6P in both symmetric and asymmetric SLC37A2 cytosolic-open conformations. b, Structure of the symmetric G6P-bound SLC37A2 in the cytosolic-open conformation. c, Electrostatic analysis of the SLC37A2G6P monomer in the cytosolic-open conformation. d, Coordination of G6P in the symmetric cytosolic-open conformation of SLC37A2. Residues involved in G6P binding are shown as sticks. Electrostatic and hydrogen-bonding interactions are indicated by red dashed lines. e, Schematic diagram of G6P coordination in the symmetric cytosolic-open conformation. f, G6P transport of proteoliposomes based on SLC37A2WT and mutants involved in G6P interaction. The results were normalized to the net 3H-G6P accumulation of SLC37A2WT. Significant differences were evaluated using a one-way ANOVA with Dunnett’s multiple-comparisons test. ****P < 0.0001. The other exact P values are labeled above the columns. Data are the means ± s.d. (n = 3 technical replicates). g, Coordination of G6P in the asymmetric cytosolic-open conformation. h, Schematic diagram of G6P coordination in the asymmetric cytosolic-open conformation. i, Superposition of the G6P-binding pockets in the symmetric and asymmetric SLC37A2 cytosolic-open structures.

Source data

Full size image
G6P coordination by SLC37A2

Bound G6P molecules were identified in the cytosolic-open monomers in the SLC37A2CO or SLC37A2LO+CO structures, representing the step of substrate from the cytosol binding to SLC37A2 during the transport process. In the symmetric G6P-bound SLC37A2CO structure, G6P binds to a cavity formed by the N-domain and C-domain of each monomer, the bottom of which has a mainly positive electrostatic surface potential (Fig. 3b,c). Residues from both domains are involved in G6P coordination (Fig. 3d,e). The phosphate group of G6P is located at the bottom of the cavity and forms extensive hydrogen-bonding or ionic interactions with the surrounding residues, including Y33, H37 and R40 in TM1, Y98 in TM2 from the N-domain and Y312 in TM7 and Y405 in TM10 from the C-domain (Fig. 3d,e). These interactions explain why the phosphate group on the sugar is essential for substrate recognition. The glucose moiety of G6P participates in both hydrogen-bonding and hydrophobic packing interactions. The hydroxyl group at the C-3 position of G6P forms hydrogen bonds with the side chain of T409 in TM10 and the main chain carbonyl group of Y405 (Fig. 3d,e). Additionally, the pyranose moiety including both the hexose ring and the attached hydroxyl groups forms a CH‒π interaction with W161 in TM4, with a vertical distance of approximately 4.0 Å between the C-5 atom of G6P and the indole plane of W161 (Fig. 3d,e), which is a moderately strong interaction commonly observed between sugars and tryptophan42. Most of the interacting residues in SLC37A1–SLC37A4 are highly conserved, with Y33, R40, Y98 and W161 being invariant in all homologs (Extended Data Fig. 1a). H37 and Y312 are invariant in SLC37A1–SLC37A3 but are replaced by tyrosine and threonine residues, respectively, in SLC37A4 (Extended Data Fig. 1a). Y405 is invariant in SLC37A1 and SLC37A2 but is replaced by a serine and isoleucine in SLC37A3 and SLC37A4, respectively (Extended Data Fig. 1a). Lastly, T409 is invariant in SLC37A1 and SLC37A2 but is replaced by serine and glycine in SLC37A3 and SLC37A4, respectively (Extended Data Fig. 1a).

In the G6P-bound monomer of the asymmetric SLC37A2LO+CO structure, G6P also binds to the positively charged surface of the cytosolic-facing cavity, at a binding site almost identical to that in the SLC37A2CO structure. However, a slight shift of approximately 1 Å was observed in the binding positions of G6P in these two structures (Fig. 3g–i). This shift leads to subtle differences in the detailed binding pattern of the two G6P-bound structures. The phosphate ring of G6P in the SLC37A2LO+CO structure also coordinates to Y33, H37, R40, Y98, Y405 and Y312 (Fig. 3g,h). While the hydroxyl group at the C-3 position still forms a hydrogen bond with the side chain of T409, it no longer interacts with Y405, as observed in the SLC37A2CO structure. Moreover, the hydroxyl group at the C-1 position forms an additional hydrogen bond with the side chain of T188 in TM5, which is not observed in the SLC37A2CO structure. T188 is invariant in SLC37A1 and SLC37A2 but is replaced by alanine and methionine residues in SLC37A3 and SLC37A4, respectively (Extended Data Fig. 1a). G6P also closely packs to the indole ring of W161 in the SLC37A2LO+CO structure.

To verify the G6P-binding site, we generated a series of single-point mutants of SLC37A2, in which each interacting residue was replaced by alanine, and examined their 3H-G6P transport activity using a proteoliposome-based assay (Fig. 3f). The net 3H-G6P transport by most of the mutants, including Y33A, R40A, Y98A, W161A, Y312A and T409A, was less than 50% that of the WT protein and 3H-G6P import by H37A was approximately 70% of that of the WT. The transport activity of the T188A mutant was comparable to that of the WT, which is consistent with the observation that T188 is less conserved among SLC37 family members. Together, these results support the G6P-binding site observed in the SLC37A2LO+CO and SLC37A2CO structures and illustrate how SLC37A2 recruits G6P from the cytoplasm.

Sugar-bound structures have been determined for sugar transporters of the MFS superfamily, including xylose/glucose-bound bacterial XylE in the luminal-facing, partially occluded state43, glucose/maltose-bound human glucose transporter 3 (hGLUT3) in the outward-occluded state44, glucose-bound malaria parasite PfHT1 in the occluded state45 and glucose-bound plant STP10 in both the cytosolic-facing and the outward-occluded states46 (Extended Data Fig. 9a). In these sugar transporters, the sugar substrates are coordinated mainly by the C-domain in a similar position, whereas no direct polar interactions are formed with the N-domain. In contrast, as we observed in the G6P-bound SLC37A2 structure, both the N-domains and the C-domains of SLC37A2 are involved in G6P coordination. Structural alignment of the G6P-bound SLC37A2 structure and the glucose-bound hGLUT3 structure revealed a similar protein fold (Extended Data Fig. 9b). Notably, while both glucose and G6P are located more adjacent to the C-domain, their binding sites are different, with G6P located closer to the cytosolic side. The pyranose moieties of glucose and G6P also bind in different locations and adopt different conformations in these hGLUT3 and SLC37A2 structures. In hGLUT3, the pyranose moiety lies perpendicular to the membrane plane along the transport path formed between the N-domains and C-domains. In SLC37A2, the pyranose moiety is almost parallel to the membrane plane. In hGLUT3, each hydroxyl group of glucose is coordinated with the surrounding residues through hydrogen bonds, collectively inducing strong substrate specificity. In contrast, only the hydroxyl groups at the C-1 or C-3 position form direct interactions with SLC37A2, resulting in weak specificity for the sugar moiety. Collectively, these findings suggest that hGLUT3 and SLC37A2 adopt different substrate recognition mechanisms during glucose and G6P transport.

Conformational dynamics of SLC37A2 during G6P transport

MFS transporters have been characterized as using the canonical alternating access mechanism during substrate transport, in which the substrate-binding site formed by the N-domains and C-domains is alternatively exposed to both sides of the membrane41. Structural alignment of the SLC37A2 luminal-open monomer in the apo state and the cytosolic-open monomer in the G6P-bound state, both of which were taken from the symmetric dimeric structures, revealed an r.m.s.d. of 2.5 Å (370 Cα atoms aligned) (Fig. 4a). While the six TMs of the N-domain merged well, the TMs of the C-domain exhibited relative rotation around the center of the C-domain during the conformational switch. During the switch from the cytosolic-open G6P-bound state to the luminal-open apo state, on the cytosolic side, TM10 and TM11 of the C-domain move toward TM4 and TM2 of the N-domain, respectively, thus sealing the G6P-binding site from the cytoplasm. On the ER luminal side during the same process, TM7, TM8 and TM11 of the C-domain move away from TM1, TM5 and TM2 of the N-domain, respectively, leading to opening of the G6P-binding site toward the ER lumen for substrate release. The relative movements of the two domains are essentially rigid-body rotations, particularly for the N-domain, which remains nearly unaltered. Alignment of the N-domains only from the two states revealed an r.m.s.d. of 0.7 Å (Fig. 4b). While most of the TMs of the C-domain follow rigid-body rotation, local rearrangements are observed in certain TMs, particularly in TM7 (Fig. 4b). The C-terminal half of TM7 bends away from the substrate-binding cavity and TM1 in the cytosolic-open state and straightens the kink in TM7. Further analysis of TM1 and TM7, which are on opposite sides of the binding cavity, revealed that residues on the C-terminal half of TM7 undergo large shifts upon G6P release (Fig. 4c). As a rotational pivot, the side chain of K308 flips away from G6P. Y312, which forms a hydrogen bond with the phosphate group of G6P, moves away during the cytosolic-open to luminal-open transition. The critical role of Y312 in G6P binding was illustrated by the reduced transport activity of the Y312A mutant (Fig. 3f). Y316 and W317 on TM7 also rotate away from TM1 to open the gate on the ER luminal side of the G6P-binding site. Notably, the main chain carbonyl group of W317 forms a hydrogen bond with K47 in the cytosolic-open conformation (Fig. 4c). We examined the roles of these TM7 residues by substituting each residue to alanine and evaluating their transport activity. The results revealed that K47A and W317A exhibited significantly reduced net 3H-G6P transport compared with the WT, supporting essential roles of these two gating residues on the luminal side of SLC37A2 (Fig. 4d). In contrast, the transport activity of Y316A was approximately 30% greater than that of the WT (Fig. 4d). The side chain of Y316 does not form apparent interactions with surrounding residues in either the luminal-open or the cytosolic-open state. Acting as a latch toward the G6P-binding site, substituting Y316 to alanine may facilitate G6P entry into the transporter, thus leading to increased transport activity. However, this remains to be further investigated.

Fig. 4: Conformational dynamics of SLC37A2 during G6P transport.

a, Structural alignment of SLC37A2 in the symmetric luminal-open and cytosolic-open conformations. b, Structural superposition of the N-domains and C-domains from the symmetric luminal-open and cytosolic-open conformations of SLC37A2, respectively. c, Structural rearrangement of TM1 and TM7 during the transition of SLC37A2 from the cytosolic-open to luminal-open conformations. G6P and the residues involved in conformational transition are shown as sticks. d, G6P transport by SLC37A2WT and the mutants involved in the conformational transition. The results were normalized to the net 3H-G6P accumulation of SLC37A2WT. Significant differences were evaluated using a one-way ANOVA with Dunnett’s multiple-comparisons test. ****P < 0.0001. The other exact P value is labeled above the column. Data are the means ± s.d. (n = 3 technical replicates). e, The ER luminal gate of SLC37A2 in the cytosolic-open conformation. f, The cytosolic gate of SLC37A2 in the luminal-open conformation. g, Transport mechanism of SLC37A2 for G6P transport. The SLC37A2 structures are shown as surface models with the two monomers colored differently. TM1, TM7, Y312 and G6P are labeled and shown as cartoons or sticks. The structures of SLC37A2G6P, CO (left), SLC37A2 predicted by AlphaFold2 (middle) and SLC37A2apo, LO (right) are used to illustrate the G6P transport cycle.

Source data

Full size image

When MFS transporters undergo conformational changes during substrate transport, the gates for the binding pocket are alternatively sealed on one side of the transporter41. In the cytosolic-open G6P-bound state of SLC37A2, the luminal gate is closed through the close packing of TM1, TM2 and TM5 from the N-domain with the opposing TM7, TM11 and TM8, respectively (Fig. 4e). Both polar and hydrophobic interactions, including the aforementioned interaction between K47 and W317, were observed to mediate the pacing between the N-domain and the C-domain. In the luminal-open apo state, the cytosolic gate closed mainly by two three helix bundles, one formed by TM2, TM4 and TM11 and the other formed by TM5, TM8 and TM10 (Fig. 4f). The interactions among each bundle are mediated mainly through hydrogen-bonding or ionic interactions.

Mapping SLC37A4-related GSD-Ib mutations in the SLC37A2 structure

To date, multiple mutations have been identified in SLC37A4 that can cause GSD-Ib27. We summarized the missense mutations in SLC37A4 and the corresponding sites in SLC37A2 by sequence alignment. The disease-associated sites in SLC37A4 were mapped to the structure predicted by AlphaFold2 (ref. 47) and the corresponding sites in SLC37A2 were mapped to the cytosolic-open SLC37A2G6P, CO structure (Extended Data Fig. 9c). These residues are scattered throughout almost all the TMs of SLC37A4 and SLC37A2 except TM11 and TM12. Many of the mutations impact regions around the G6P-binding site in SLC37A2, including Y36 (Y24H in SLC37A4), R40 (R28C/H), W161 (W118R), N192 (G149E), W317 (W246R) and G403 (G339C/D), and are highly conserved in SLC37A4 (Extended Data Fig. 9e–g). Among these residues, R40 and W161 directly interact with G6P, as shown by the G6P-bound structure (Fig. 3d,e). As we showed for the R40A and W161A mutants, substitution of these residues disrupts the interaction between SLC37A2/A4 and G6P, thus leading to reduced transport activity. Y36 does not interact with G6P; instead, its side chain protrudes into the space between TM3 and TM4. The results of the transport assay revealed that the activity of the Y36H mutant was reduced by ~30% compared to that of the WT (Extended Data Fig. 9d). N192 is near the ER luminal gate of SLC37A2 and seals the gate with the adjacent residue L315 in the cytosolic-facing conformation. This residue is not conserved and is replaced by glycine in SLC37A4 (G149). In SLC37A2, the N192E mutant also presented reduced G6P transport activity, suggesting its role in the conformational transition or gating of SLC37A2 (Extended Data Fig. 9d). In the cytosolic-open conformation of SLC37A2, W317 is involved in the interaction between TM1 and TM7 and the formation of the ER luminal gate. The W317R (W246R in SLC37A4) substitution impaired the cation‒π interaction between K47 and W317, leading to decreased transport activity, as shown for the W317A mutant. G403 of SLC37A2 is located in the middle of TM10. Structural alignment of the luminal-open and cytosolic-open structures of SLC37A2 revealed that G403 serves as a pivot point during the rotation of TM10 (Extended Data Fig. 9g). Replacing this residue with cysteine or aspartic acid may inhibit its flexibility. The transport activity results revealed that the G403C and G403D mutants presented much lower transport activity than the WT, supporting the essential role of this glycine residue in the observed conformational changes (Extended Data Fig. 9d). Together, the structural information of SLC37A2 and these transport analyses provide insights into the pathological mechanism of this clinical disorder.

Discussion

In this study, we characterized G6P transport by ER-membrane-localized SLC37 family transporters, with a focus on the structural elucidation of SLC37A2. The proteoliposome-based 3H-G6P transport assay results provide compelling evidence that SLC37A2 and SLC37A4 (or G6PT1) function as G6P uniporters independent of Pi gradients. This contrasts with previous models36, which were based on assays using reconstituted proteoliposomes from microsomes with endogenous factors, in which SLC37 transporters acted as G6P/Pi exchangers. The weak inhibition by Pi further supports the uniporter model that Pi may act as a competitive inhibitor rather than a coupling ion. Meanwhile, the sugar competition assay further emphasized the importance of the cyclic monosaccharide moiety and phosphate group for substrate recognition of SLC37A2. Because of its structural similarity to Pi, vanadate may interact with the same substrate-binding site as the phosphate moiety of G6P in SLC37A2 and SLC37A4. CHA contains a quinic acid moiety esterified by a caffeic acid, which is distinct from the chemical structure of G6P (Extended Data Fig. 2e). Its selective inhibition mechanism toward SLC37A4 remains to be further investigated, particularly by a CHA-bound structure of SLC37A4. These results improve our understanding of the transport mechanism of SLC37 transporters and warrant a reevaluation of their roles in G6P transport. Notably, the Km values for G6P transport by SLC37A2 and SLC37A4 are much higher than those determined for sugar transporters of the MFS superfamily (0.47 mM for XylE toward xylose43, 1.28 mM for SUC1 toward sucrose48, 0.80 mM for PfHT1 toward glucose45 and 2.6 μM for STP10 toward glucose49). On the one hand, these data suggest that SLC37A2 and SLC37A4 have low substrate selectivity and binding, as shown by the broad inhibition of G6P transport by other sugar phosphates. On the other hand, these data indicate additional regulation by the G6P concentration in the cytosol or other factors under certain physiological conditions, which require further in vivo investigations.

The cryo-EM structures in both the cytosolic-open, G6P-bound state and the luminal-open, apo state provide insights into the G6P binding and conformational transition mechanisms of SLC37A2 (Fig. 4g). Both the N-domains and the C-domains are involved in G6P coordination mainly through interactions with the phosphate group of G6P, which is different from MFS sugar transporters and contributes to the low sugar‒base selectivity of SLC37A2. Unlike most sugar transporters whose structures have been determined, SLC37A2 exhibits a dimeric architecture. However, the interface-forming sequence is missing in SLC37A3 and SLC37A4, suggesting different oligomeric states of the SLC37 transporters. During the SLC37A2 structural transitions, the N-domain follows the rigid-body movements common for MFS transporters, whereas both rigid-body movements and local rearrangements are observed with the C-domain. Local alterations have also been observed for TM7 and TM10 of human GLUTs44, supporting the flexible dynamics of the C-domain in substrate binding and transport. Among MFS transporters, the transport cycle typically involves structural transitions between different conformations, including the cytosolic-open and luminal-open conformations captured in this study and the yet-to-be-determined ligand-free or ligand-bound occluded states. Further investigations are needed to capture SLC37A2 in other states to elucidate the transport mechanism.

Taken together, the results of this study advance our knowledge of SLC37 transporters and provide a structural framework for G6P transport between the cytoplasm and ER lumen, which is an essential step in gluconeogenesis and glycogenolysis. This study also paves the way for understanding the pathological mechanisms associated with SLC37A4 mutations and developing precision therapies for human GSDs.

Methods
Plasmid construction and generation

The complementary DNA (cDNA) sequence of full-length human SLC37A2 and SLC37A4 are publicly available from UniProt under accession codes Q8TED4 and Q43826, respectively. The cDNA of WT SLC37A2 was synthesized by Sangon Biotech Company, while the cDNA of WT SLC37A4 was cloned from a human cDNA library extracted from HEK293F cells. The SLC37A2 mutants were generated by site-directed mutagenesis following a standard two-step PCR. The DNA fragments were subcloned into the pCAG vector (Invitrogen) with a tandem C-terminal FLAG (DYKDDDDK) plus 8×His tag for protein expression and purification. All recombinant plasmids were verified by DNA sequencing. A list of the primers used is provided in Supplementary Table 1.

Protein expression and purification

HEK293F cells (Gibco, A14635) with a density of 2 × 106 ml−1 were transfected by recombinant plasmids through a polyethylenimine (PEI; Polysciences)-mediated method. In detail, 1.5 mg of plasmid and 4 mg of PEI were mixed in 45 ml of SMM 293T-II medium (Sino Biological) and set aside for 15 min before adding to 800 ml HEK293F cells, which were then incubated statically for another 15 min and further cultured at 37 °C, 130 rpm under 5% CO2 for 48–60 h. Transfected cells were collected through centrifugation at 2,100g for 15 min and resuspended in HEPES buffer containing 25 mM HEPES pH 7.4 (adjusted by NaOH) and 150 mM NaCl. The suspension was supplemented with 1 mM PMSF, 0.7 mg ml−1 pepstatin A, 1.3 mg ml−1 aprotinin, 5 mg ml−1 leupeptin, 5 mM ATP disodium salt (Sangon Biotech), 5 mM MgCl2 and 1.5% (w/v) dodecyl maltopyranoside (Anatrace). All the protein purification procedures were carried out at 4 °C. Membrane solubilization was performed for 2 h with constant rotation at 22 rpm, followed by centrifugation at 20,000g for 1 h to remove insoluble materials. The supernatant was subsequently incubated with anti-FLAG M2 resin (Sigma-Aldrich) equilibrated with HEPES buffer for 45 min. The resin was then successively rinsed with wash buffer A containing 25 mM HEPES pH 7.4 (adjusted by NaOH), 150 mM NaCl, 2.5 mM ATP disodium salt, 2.5 mM MgCl2, 0.02% (w/v) glyco-diosgenin (GDN; Anatrace) and wash buffer B containing 25 mM HEPES pH 7.4 (adjusted by NaOH), 150 mM NaCl and 0.02% GDN. The protein was eluted with wash buffer B plus 200 μg ml−1 FLAG peptide (Sigma-Aldrich). After concentration through a Centricon (50-kDa molecular weight cutoff; Millipore), the eluent was applied to a Superose 6 Increase column (Cytiva) with HEPES–Na buffer plus 0.01% GDN for size-exclusion chromatography. Peak fractions were then pooled together and concentrated to around 12.5 mg ml−1 or 6 mg ml−1 for cryo-EM sample preparation or proteoliposome reconstitution, respectively. The protein deglycosylation assay was carried out using the PNGase F enzyme (New England Biolabs) in the denaturing condition following the manufacturer’s protocol. Briefly, 20 μg of protein was incubated in the denaturing buffer at 100 °C for 10 min and then the protein was supplemented with 2 μl of GlycoBuffer 2, 2 μl of 10% NP-40 and 1 μl of PNGase F for 1 h at 37 °C.

Proteoliposome reconstitution

Proteoliposomes were prepared as previously described36,50 with modification. In detail, the lipid mixture containing the polar lipid extract of Escherichia coli, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-oleoyl-sn-glycero-3-phospho-L-serine and cholesterol (all lipids were purchased from Sigma-Aldrich) with a ratio of 60:17.5:10:12.5 (w/w) was dissolved by an organic solvent composed of chloroform and methanol (3:1, v/v) to a 20 mg ml−1 lipid stock. The desired amount of lipid mix was dried out through a continuous nitrogen stream and then resuspended in inner buffer (normally HEPES–Na buffer containing 25 mM HEPES pH 7.4 (adjusted by NaOH) and 150 mM NaCl) to 20 mg ml−1. To verify the function of inner phosphate gradient for G6P transport, a HEPES–Pi buffer (25 mM HEPES, 150 mM NaCl and 50 mM Na2HPO4/NaH2PO4 pH 7.4 (adjusted by NaOH)) was used as the inner buffer. To follow the proteoliposome condition Janice et al. mentioned36, the inner buffer was set to a phosphate buffer containing 50 mM KH2PO4 pH 7.0 (adjusted with KOH), 1 mM DTT and the protease cocktails. The raw lipid was repeatedly frozen and thawed ten times and extruded through 0.4-μm films (Cytiva) 41 times to generate liposomes with uniform size, which were then incubated with 1.25% (w/v) n-octyl-β-D-glucopyranoside (β-OG, Anatrace) at 4 °C for 1 h. The GDN-purified proteins were incubated with β-OG-embedded liposomes with a protein-to-lipid ratio of 1:40 (w/w) at 4 °C for 1.5 h. As for the negative control group, an equal amount of HEPES–Na buffer plus 0.01% (w/v) GDN was added to the liposomes instead. After incubation with four batches of 100 mg ml−1 Bio-Beads SM2 (Bio-Rad) for detergent removal, the proteoliposomes were further homogenized by repeatedly freezing and thawing five times and extruding through 0.4-μm films 31 times. Proteoliposome pellets were collected after ultracentrifugation at 160,000g, 4 °C for 1.5 h and resuspended in outer buffer (normally HEPES–Na buffer) to a final concentration of 100 mg ml−1. Pellets were resuspended in MOPS–K buffer (exterior buffer) containing 20 mM MOPS pH 7.5 (adjusted with KOH), 75 mM K2SO4, 2.5 mM MgSO4, 1 mM DTT and protease cocktails.

3H-G6P transport assay

The proteoliposome-based transport assay was performed using a rapid filtration method mentioned before50,51. Briefly, the reaction was performed at 37 °C by adding 2 μl of proteoliposomes (100 mg ml−1) to 48 μl of transport buffer composed of outer buffer and 1 μCi of D-6-3H-G6P disodium salt (1 μCi μl−1, 40 Ci mmol−1; American Radiolabeled Chemicals). The final concentration of external 3H-G6P before reaction initiation was 0.5 μM. When a desired time was reached (typically 1 min unless otherwise specified), the reaction solution was filtered immediately through a presoaked 0.22-μm nitrocellulose film (Sigma-Aldrich) and washed with 2 ml of ice-cold HEPES–Na buffer. The dried films were incubated with 2 ml of scintillation fluid overnight and then the radioactivity was measured by a liquid scintillation counter (Tri-Carb 2910TR, Perkin-Elmer). A standard calibration curve was plotted with counts per minute read from 3H-G6P (0.0022 pmol, 0.022 pmol, 0.22 pmol, 2.2 pmol and 22 pmol in 2 ml of scintillation fluid). For glucose transport assays, all conditions remained unchanged except that D-6-3H-G6P disodium salt was replaced with D-6-[3H]glucose (1 μCi μl−1, 60 Ci mmol−1; American Radiolabeled Chemicals). The final concentration of external [3H]glucose before reaction initiation was 0.67 μM. For the time-course measurement of G6P transport activity by 3H-G6P transport assay, imported G6P was quantified by scintillation counting at transport durations of 0 min, 1 min, 2 min, 5 min, 10 min and 20 min. For the Km measurement 3H-G6P transport assay, the external G6P concentration was adjusted by adding the desired amount of nonradioactively labeled G6P (0 mM, 1 mM, 2 mM, 5 mM, 10 mM, 20 mM, 40 mM, 60 mM, 80 mM and 100 mM). The data were fitted to the Michaelis–Menten equation, V = [Vmax(G6P)] × [Km + (G6P)]−1, in GraphPad Prism 8. For the CHA inhibition assay, transport buffer was supplemented with a CHA (MedChemExpress) gradient (0 mM, 0.5 mM, 5 mM, 10 mM, 20 mM, 30 mM, 40 mM and 48 mM). For the vanadate inhibition assay, transport buffer was supplemented with a sodium orthovanadate (MedChemExpress) gradient (0 mM, 1 mM, 10 mM, 20 mM, 40 mM, 80 mM, 130 mM and 186 mM). For the Pi inhibition assay, transport buffer was supplemented with a Na2HPO4 (dissolved by HEPES–Na buffer pH 7.4 (adjusted by NaOH)) gradient (0 mM, 1 mM, 10 mM, 20 mM, 40 mM, 80 mM, 150 mM and 200 mM). For the sugar competition assay, the reaction solution was supplemented with 30 mM sugars or sugar phosphates (purchased from Sigma-Aldrich, Macklin or Aladdin). All experiments were repeated at least three times.

Cryo-EM sample preparation and data acquisition

For SLC37A2apo, a 4-μl aliquot of purified SLC37A2 (12.5 mg ml−1) was applied to a glow-discharged holey carbon grid (Quantifoil Au R1.2/1.3, 300-mesh), blotted with a FEI Vitrobot Mark IV (Thermo Fisher Scientific) operating at 8 °C and 100% humidity using a blotting time of 3.5 s and plunged into liquid ethane cooled by liquid nitrogen. To determine the structure of SLC37A2 in the presence of G6P, purified SLC37A2 was incubated with 100 mM D-G6P disodium salt (Sigma-Aldrich) for 40 min before cryo-EM sample preparation. All datasets were obtained using the EPU (version 2.12) software in a super-resolution mode on the 300-kV Titan Krios microscope (Thermo Fisher Scientific), which was equipped with a K3 Summit direct electron detector (Gatan) and a GIF Quantum energy filter (Gatan). Micrographs were collected at a nominal magnification of ×81,000 with defocus values ranging from −1.0 to −2.0 μm and a calibrated pixel size of 1.1 Å. Each movie stack was acquired with an exposure time of 3 s and dose-fractioned into 32 frames, yielding a total accumulated dose of 50 e− per Å2.

Image processing

Both the SLC37A2apo and the SLC37A2G6P datasets were processed by cryoSPARC (version 3.2.0)52 following a similar procedure (Extended Data Figs. 4a and 6). For the dataset of SLC37A2apo, 1,903 raw images were imported into cryoSPARC. After motion correction, dose weighting and patch CTF estimation, 1,608,545 particles were automatically picked by the template picker and extracted with a fourfold-binned form for further processing. After several rounds of 2D classification, 1,178,278 particles were selected for primary ab initio reconstruction. Two classes representing the SLC37A2 dimer and monomer were then extracted with an unbinned form to 618,532 particles, which were used for further 3D reconstruction and nonuniform refinement to result in an electron cloud density map with an overall resolution of 3.3 Å. The 618,532 particles were also used for heterogeneous refinement using the 3.3-Å map as the fine reference and other volumes before as the bad references. After several rounds of further heterogeneous refinement, 3D reconstruction and nonuniform refinement, a final 3.2-Å map presenting an outward facing state, named SLC37A2apo, LO, was obtained. For the dataset of SLC37A2G6P, 3,170 raw images were imported into cryoSPARC. After motion correction, dose weighting and patch CTF estimation, 2,487,946 particles were automatically picked by the template picker and extracted with a fourfold-binned form for further processing. After several rounds of 2D classification, 1,570,707 particles were selected for primary ab initio reconstruction. Two classes representing the SLC37A2 dimer and monomer were then extracted with an unbinned form to 846,206 particles, which were used for further 3D reconstruction and nonuniform refinement to result in an asymmetric electron cloud density map with an overall resolution of 3.8 Å. A total of 1,646,539 particles were reextracted with an unbinned form for heterogeneous refinement using the asymmetric map and two symmetric raw maps from other datasets processed previously as the fine references, together with other volumes calculated previously as the bad references. After several rounds of further heterogeneous refinement, 3D reconstruction and nonuniform refinement, three distinct electron cloud density maps presenting the G6P-unbound luminal-open state, SLC37A2G6P, LO, the G6P-bound cytosolic-open state, SLC37A2G6P, CO, and the asymmetric conformation with G6P bound to the cytosolic-open moiety, SLC37A2G6P, LO+CO, were finally obtained in overall resolutions of 3.7 Å, 3.5 Å and 3.6 Å, respectively. All resolutions were estimated with the gold-standard FSC = 0.143 criterion with a high-resolution noise substitution method53,54. Local resolution variations were estimated using cryoSPARC.

Model building and refinement

The apo structure model of human SLC37A2 predicted by AlphaFold2 (ref. 47) was docked into the obtained 3.2-Å cryo-EM density map SLC37A2apo, LO using UCSF Chimera (version 1.15)55. The fitted structures were manually adjusted in Coot (version 0.8.6)56. Bulky residues such as phenylalanine, tyrosine, tryptophan, lysine and arginine were used to guide the sequence assignment, taking advantage of their clearly defined side-chain densities. Meanwhile, the chemical properties of amino acids were also considered to facilitate model building. Structure refinements were carried out by PHENIX (version 1.18rc1-3777) in real space57. Overfitting of the model was monitored by refining the model in one of the two independent maps from the gold-standard refinement approach and testing the refined model against the other map. The initial model of SLC37A2G6P, LO was built by docking the refined SLC37A2apo, LO model into the 3.7-Å map of SLC37A2G6P, LO and was further adjusted and refined using Coot and PHENIX. The initial model of SLC37A2G6P, CO was built by docking the refined SLC37A2apo, LO model into the 3.5-Å map of SLC37A2G6P, CO and was further adjusted and refined using Coot and PHENIX. The initial model of SLC37A2G6P, LO+CO was built by docking the refined SLC37A2apo, LO and SLC37A2G6P, CO models into the 3.6-Å map of SLC37A2G6P, LO+CO and was further adjusted and refined using Coot and PHENIX. The G6P molecule was built and refined using restraints generated by phenix.elbow57. Statistics of the 3D reconstruction and model refinement can be found in Table 1.

Statistical analysis

Data were analyzed with GraphPad Prism 8 (GraphPad Software). The significance of differences was determined using an unpaired two-tailed t-test or one-way analysis of variance (ANOVA) with Dunnett’s multiple-comparisons test. The sample sizes were chosen to ensure the reproducibility of the experiments and to get meaningful results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Model coordinates and cryo-EM density maps for the apo, luminal-open state SLC37A2apo, LO, the G6P-unbound, luminal-open state SLC37A2G6P, LO, the G6P-bound cytosolic-open state SLC37A2G6P, CO and the asymmetric state SLC37A2G6P, LO+CO were deposited to the PDB under accession codes 9UCM, 9UDX, 9UDW and 9UDV and the EM Data Bank under accession codes EMD-64048, EMD-64074, EMD-64073 and EMD-64072, respectively. Data and materials can be obtained from the corresponding authors upon request. Source data are provided with this paper.

References

Rajas, F., Gautier-Stein, A. & Mithieux, G. Glucose-6 phosphate, a central hub for liver carbohydrate metabolism. Metabolites 9, 282 (2019).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Lizák, B. et al. Glucose transport and transporters in the endomembranes. Int. J. Mol. Sci. 20, 5898 (2019).

Article
 
PubMed
 
PubMed Central
 
Google Scholar

Marcolongo, P. et al. Multiple roles of glucose-6-phosphatases in pathophysiology: state of the art and future trends. Biochim. Biophys. Acta 1830, 2608–2618 (2013).

Article
 
CAS
 
PubMed
 
Google Scholar

Arion, W. J., Walls, B. K., Lange, A. J. & Ballas, L. M. On the involvement of a glucose 6-phosphate transport system in the function of microsomal glucose 6-phosphatase. Mol. Cell. Biochem. 6, 75–83 (1975).

Article
 
CAS
 
PubMed
 
Google Scholar

Csala, M. et al. Transport and transporters in the endoplasmic reticulum. Biochim. Biohpys. Acta 1768, 1325–1341 (2007).

Article
 
CAS
 
Google Scholar

Gerin, I., Veiga-da-Cunha, M., Achouri, Y., Collet, J. F. & Van Schaftingen, E. Sequence of a putative glucose 6-phosphate translocase, mutated in glycogen storage disease type Ib. FEBS Lett. 419, 235–238 (1997).

Article
 
CAS
 
PubMed
 
Google Scholar

Hiraiwa, H., Pan, C. J., Lin, B. C., Moses, S. W. & Chou, J. Y. Inactivation of the glucose 6-phosphate transporter causes glycogen storage disease type 1b. J. Biol. Chem. 274, 5532–5536 (1999).

Article
 
CAS
 
PubMed
 
Google Scholar

Chou, J. Y., Jun, H. S. & Mansfield, B. C. Type I glycogen storage diseases: disorders of the glucose-6-phosphatase/glucose-6-phosphate transporter complexes. J. Inherit. Metab. Dis. 38, 511–519 (2015).

Article
 
CAS
 
PubMed
 
Google Scholar

Annabi, B. et al. The gene for glycogen-storage disease type 1b maps to chromosome 11q23. Am. J. Hum. Genet. 62, 400–405 (1998).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Ihara, K., Nomura, A., Hikino, S., Takada, H. & Hara, T. Quantitative analysis of glucose-6-phosphate translocase gene expression in various human tissues and haematopoietic progenitor cells. J. Inherit. Metab. Dis. 23, 583–592 (2000).

Article
 
CAS
 
PubMed
 
Google Scholar

Arion, W. J., Carlson, P. W., Lange, A. J. & Wallin, B. K. The specificity of glucose 6-phosphatase of intact liver microsomes. J. Biol. Chem. 247, 2558–2565 (1972).

Article
 
CAS
 
PubMed
 
Google Scholar

Arion, W. J., Ballas, L. M., Lange, A. J. & Wallin, B. K. Microsomal membrane permeability and the hepatic glucose-6-phosphatase system. Interactions of system with D-Mannose 6-phosphate and D-mannose. J. Biol. Chem. 251, 4891–4897 (1976).

Article
 
CAS
 
PubMed
 
Google Scholar

Arion, W. J., Lange, A. J., Walls, H. E. & Ballas, L. M. Evidence for the participation of independent translocases for phosphate and glucose-6-phosphate in the microsomal glucose-6-phosphatase system. Interactions of the System with orthophosphate, inorganic pyrophosphate, and carbamyl-phosphate. J. Biol. Chem. 255, 10396–10406 (1980).

Article
 
CAS
 
PubMed
 
Google Scholar

Cappello, A. R., Curcio, R., Lappano, R., Maggiolini, M. & Dolce, V. The physiopathological role of the exchangers belonging to the SLC37 family. Front. Chem. 6, 122 (2018).

Article
 
PubMed
 
PubMed Central
 
Google Scholar

Shieh, J. J., Pan, C. J., Mansfield, B. C. & Chou, J. Y. A glucose-6-phosphate hydrolase, widely expressed outside the liver, can explain age-dependent resolution of hypoglycemia in glycogen storage disease type Ia. J. Biol. Chem. 278, 47098–47103 (2003).

Article
 
CAS
 
PubMed
 
Google Scholar

Chou, J. Y., Cho, J. H., Kim, G. Y. & Mansfield, B. C. Molecular biology and gene therapy for glycogen storage disease type Ib. J. Inherit. Metab. Dis. 41, 1007–1014 (2018).

Article
 
CAS
 
PubMed
 
Google Scholar

Sim, S. W., Weinstein, D. A., Lee, Y. M. & Jun, H. S. Glycogen storage disease type Ib: role of glucose-6-phosphate transporter in cell metabolism and function. FEBS Lett. 594, 3–18 (2020).

Article
 
CAS
 
PubMed
 
Google Scholar

Chen, L. Y., Lin, B. C., Pan, C. J., Hiraiwa, H. & Chou, J. Y. Structural requirements for the stability and microsomal transport activity of the human glucose 6-phosphate transporter. J. Biol. Chem. 275, 34280–34286 (2000).

Article
 
CAS
 
PubMed
 
Google Scholar

Chen, L. Y., Pan, C. J. & Chou, J. Y. Structure-function analysis of the glucose-6-phosphate transporter deficient in glycogen storage disease type Ib. Hum. Mol. Genet. 11, 3199–3207 (2002).

Article
 
CAS
 
PubMed
 
Google Scholar

Pan, C. J. et al. The signature motif in human glucose-6-phosphate transporter is essential for microsomal transport of glucose-6-phosphate. Hum. Genet. 112, 430–433 (2003).

Article
 
PubMed
 
Google Scholar

Ahn, H. H. et al. Identification of glucose-6-phosphate transporter as a key regulator functioning at the autophagy initiation step. FEBS Lett. 589, 2100–2109 (2015).

Article
 
CAS
 
PubMed
 
Google Scholar

Jeon, Y. G., Kim, Y. Y., Lee, G. & Kim, J. B. Physiological and pathological roles of lipogenesis. Nat. Metab. 5, 735–759 (2023).

Article
 
CAS
 
PubMed
 
Google Scholar

Chou, J. Y., Jun, H. S. & Mansfield, B. C. Glycogen storage disease type I and G6Pase-β deficiency: etiology and therapy. Nat. Rev. Endocrinol. 6, 676–688 (2010).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Belkaid, A., Copland, I. B., Massillon, D. & Annabi, B. Silencing of the human microsomal glucose-6-phosphate translocase induces glioma cell death: potential new anticancer target for curcumin. FEBS Lett. 580, 3746–3752 (2006).

Article
 
CAS
 
PubMed
 
Google Scholar

Pao, S. S., Paulsen, I. T. & Saier, M. H. Major facilitator superfamily. Microbiol. Mol. Biol. Rev. 62, 1–34 (1998).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Bartoloni, L. & Antonarakis, S. E. The human sugar-phosphate/phosphate exchanger family SLC37. Pflugers Arch. 447, 780–783 (2004).

Article
 
CAS
 
PubMed
 
Google Scholar

Chou, J. Y. & Mansfield, B. C. The SLC37 family of sugar-phosphate/phosphate exchangers. Curr. Top. Membr. 73, 357–382 (2014).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Kim, J. Y., Tillison, K., Zhou, S., Wu, Y. & Smas, C. M. The major facilitator superfamily member Slc37a2 is a novel macrophage- specific gene selectively expressed in obese white adipose tissue. Am. J. Physiol. Endocrinol. Metab. 293, E110–E120 (2007).

Article
 
CAS
 
PubMed
 
Google Scholar

Wang, Z. et al. Solute carrier family 37 member 2 (SLC37A2) negatively regulates murine macrophage inflammation by controlling glycolysis. iScience 23, 101125 (2020).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Zhao, Q. et al. Hematopoietic cell-specific SLC37A2 deficiency accelerates atherosclerosis in LDL receptor-deficient mice. Front. Cardiovasc. Med. 8, 777098 (2021).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Ng, P. Y. et al. Sugar transporter Slc37a2 regulates bone metabolism in mice via a tubular lysosomal network in osteoclasts. Nat. Commun. 14, 906 (2023).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Wilfinger, J. et al. Primary vitamin D receptor target genes as biomarkers for the vitamin D3 status in the hematopoietic system. J. Nutr. Biochem. 25, 875–884 (2014).

Article
 
CAS
 
PubMed
 
Google Scholar

Saksa, N. et al. Dissecting high from low responders in a vitamin D3 intervention study. J. Steroid Biochem. Mol. Biol. 148, 275–282 (2015).

Article
 
CAS
 
PubMed
 
Google Scholar

Iskarpatyoti, J. A. et al. Mast cell regranulation requires a metabolic switch involving mTORC1 and a glucose-6-phosphate transporter. Cell Rep. 40, 111346 (2022).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Villani, A. et al. Clearance by microglia depends on packaging of phagosomes into a unique cellular compartment. Dev. Cell 49, 77–88.e7 (2019).

Article
 
CAS
 
PubMed
 
Google Scholar

Zilberstein, D. et al. SLC37A1 and SLC37A2 are phosphate-linked, glucose-6-phosphate a. PLoS ONE 6, e23157 (2011).

Article
 
Google Scholar

Chen, S.-Y. et al. The glucose-6-phosphate transporter is a phosphate-linked antiporter deficient in glycogen storage disease type Ib and Ic. FASEB J. 22, 2206–2213 (2008).

Article
 
CAS
 
PubMed
 
Google Scholar

Marcolongo, P. et al. The glucose-6-phosphate transport is not mediated by a glucose-6-hosphate/phosphate exchange in liver microsomes. FEBS Lett. 586, 3354–3359 (2012).

Article
 
CAS
 
PubMed
 
Google Scholar

Arion, W. J. et al. Chlorogenic acid and hydroxynitrobenzaldehyde: new inhibitors of hepatic glucose 6-phosphatase. Arch. Biochem. Biophys. 339, 315–322 (1997).

Article
 
CAS
 
PubMed
 
Google Scholar

Pan, C. J., Lin, B. C. & Chou, J. Y. Transmembrane topology of human glucose 6-phosphate transporter. J. Biol. Chem. 274, 13865–13869 (1999).

Article
 
CAS
 
PubMed
 
Google Scholar

Yan, N. Structural biology of the major facilitator superfamily transporters. Annu. Rev. Biophys. 44, 257–283 (2015).

Article
 
CAS
 
PubMed
 
Google Scholar

Kumari, M., Balaji, P. V. & Sunoj, R. B. Quantification of binding affinities of essential sugars with a tryptophan analogue and the ubiquitous role of C–H···π interactions. Phys. Chem. Chem. Phys. 13, 6517–6530 (2011).

Article
 
CAS
 
PubMed
 
Google Scholar

Sun, L. F. et al. Crystal structure of a bacterial homologue of glucose transporters GLUT1-4. Nature 490, 361–366 (2012).

Article
 
CAS
 
PubMed
 
Google Scholar

Deng, D. et al. Molecular basis of ligand recognition and transport by glucose transporters. Nature 526, 391–396 (2015).

Article
 
CAS
 
PubMed
 
Google Scholar

Qureshi, A. A. et al. The molecular basis for sugar import in malaria parasites. Nature 578, 321–325 (2020).

Article
 
CAS
 
PubMed
 
Google Scholar

Bavnhoj, L., Paulsen, P. A., Flores-Canales, J. C., Schiott, B. & Pedersen, B. P. Molecular mechanism of sugar transport in plants unveiled by structures of glucose/H+ symporter STP10. Nat. Plants 7, 1409–1419 (2021).

Article
 
PubMed
 
Google Scholar

Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Bavnhoj, L. et al. Structure and sucrose binding mechanism of the plant SUC1 sucrose transporter. Nat. Plants 9, 938–950 (2023).

Article
 
PubMed
 
PubMed Central
 
Google Scholar

Paulsen, P. A., Custódio, T. F. & Pedersen, B. P. Crystal structure of the plant symporter STP10 illuminates sugar uptake mechanism in monosaccharide transporter superfamily. Nat. Commun. 10, 407 (2019).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Ying, W. et al. Structure and function of the Arabidopsis ABC transporter ABCB19 in brassinosteroid export. Science 383, eadj4591 (2024).

Article
 
CAS
 
PubMed
 
Google Scholar

Yang, Z. S. et al. Structural insights into auxin recognition and efflux by PIN1. Nature 609, 611–615 (2022).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).

Article
 
CAS
 
PubMed
 
Google Scholar

Rosenthal, P. B. & Henderson, R. Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. J. Mol. Biol. 333, 721–745 (2003).

Article
 
CAS
 
PubMed
 
Google Scholar

Chen, S. X. et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy 135, 24–35 (2013).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Pettersen, E. F. et al. UCSF chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004).

Article
 
CAS
 
PubMed
 
Google Scholar

Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010).

Article
 
CAS
 
PubMed
 
PubMed Central
 
Google Scholar

Download references

Acknowledgements

We thank the Cryo-EM Center of the University of Science and Technology of China for the EM facility support. This work was supported by the National Natural Science Foundation of China (32471279 to X.L. and 32322041, 32321001 and W2412029 to L.S.), the Scientific Research Innovation Capability Support Project for Young Faculty (ZYGXQNJSKYCXNLZCXM-B8 to L.S.), Center for Advanced Interdisciplinary Science and Biomedicine of IHM (QYPY20230034 to L.S.), Natural Science Foundation of Anhui Province (2408085JX005 to L.S.), Fundamental Research Funds for the Central Universities (WK9100000031 to L.S. and WK9100000065 to Z.Y.) and USTC Research Funds of the Double First-Class Initiative (YD9100002004 to L.S. and YD9100002020 to X.L.). L.S. is supported by the Outstanding Young Scholar Award from the Qiu Shi Science and Technologies Foundation. X.L. and L.S. are supported by the Young Scholar Award from the Cyrus Tang Foundation.

Author information
Author notes

These authors contributed equally: Qinxuan Lai, Mengzhen Xu.

Authors and Affiliations

Department of Neurology, The First Affiliated Hospital of USTC, MOE Key Laboratory for Membraneless Organelles and Cellular Dynamics, Center for Advanced Interdisciplinary Science and Biomedicine of IHM, Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, China

Qinxuan Lai, Mengzhen Xu, Yongxiang Gao, Zhisen Yang, Linfeng Sun & Xin Liu

Hefei National Research Center for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, China

Qinxuan Lai, Mengzhen Xu, Yongxiang Gao, Zhisen Yang, Linfeng Sun & Xin Liu

Biomedical Sciences and Health Laboratory of Anhui Province, University of Science and Technology of China, Hefei, China

Qinxuan Lai, Mengzhen Xu, Yongxiang Gao, Linfeng Sun & Xin Liu

Contributions

X.L. and L.S. conceptualized the project. X.L. and L.S. designed the experiments and analyzed the data. Q.L. and M.X. performed the molecular cloning and protein purification work. Q.L., M.X. and Z.Y. performed the transport assays. Q.L., Z.Y. and Y.G. performed the cryo-EM data collection. Q.L. and Z.Y. performed the cryo-EM data processing and model building. Q.L., L.S. and X.L. wrote the manuscript. All authors revised the manuscript.

Corresponding authors

Correspondence to Zhisen Yang, Linfeng Sun or Xin Liu.

Ethics declarations
Competing interests

The authors declare no competing interests.

Peer review
Peer review information

Nature Structural & Molecular Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available. Primary Handling Editor: Katarzyna Ciazynska, in collaboration with the Nature Structural & Molecular Biology team.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data
Extended Data Fig. 1 Sequence alignment and phylogenetic tree plotting for SLC37 family members.

a, Sequence alignment of Homo sapiens SLC37A1, SLC37A2, SLC37A3, and SLC37A4. The transmembrane (TM) domains, H0, and H13 helixes of SLC37A2 are indicated above the sequences. The G6P interacting residues are indicated above the sequences with stars (★). The natural mutation sites in SLC37A4 associated with diseases are indicated below the sequences with asterisks (＊). b, The percent identity matrix of the human SLC37 family transporters. Sequence identities were calculated using the alignment tool in UniProt (https://www.uniprot.org/align). c, Phylogenetic tree of human SLC37 family members constructed with the program iTOL v7 (https://itol.embl.de/). Protein sequences of SLC37A1 (UniProt ID: P57057), SLC37A2 (UniProt ID: Q8TED4), SLC37A3 (UniProt ID: Q8NCC5), SLC37A4 (UniProt ID: O43826) were acquired from UniProt (https://www.uniprot.org/).

Extended Data Fig. 2 Protein purification and transport analyses of SLC37A2 and SLC37A4.

a, The size exclusion chromatography (SEC) profiles of wild-type SLC37A2 and SLC37A4 (SLC37A2WT and SLC37A4WT) purified from HEK293F cells, and the SEC profiles of SLC37A2WT and the N-terminal truncation mutant (1-13 aa) SLC37A2ΔN expressed in HEK293F cells. b, PNGase F treatment results of SLC37A2WT, SLC37A2ΔN, SLC37A4WT proteins. The representative SDS-PAGE result of the proteins prior to and after PNGase F treatment is shown here. The experiments were repeated independently at least three times with similar results. c, G6P transport activities of SLC37A2WT and SLC37A4WT assessed using a proteoliposome-based 3H-G6P transport assay with phosphate buffer (50 mM KH2PO4, pH 7.0, 1 mM DTT, and the protease cocktails) as the inner buffer and MOPS/K buffer (20 mM MOPS, pH 7.5, 75 mM K2SO4, 2.5 mM MgSO4, 1 mM DTT, and the protease cocktails) as the exterior buffer. The data are presented as mean ± SD (n = 3 biologically independent repeats). Significances of the SLC37A2 and SLC37A4 groups versus the control group were determined using a two-tailed unpaired t-test. **** P < 0.0001. The other exact P value was labeled above the column. d, Glucose transport activities of the SLC37A2WT and SLC37A4WT proteins assessed via a proteoliposome-based 3H-glucose transport assay in which HEPES/Na buffer (25 mM HEPES, pH 7.4, 150 mM NaCl) was used as the interior and exterior buffer. The data are presented as the mean ± SD (n = 3 biologically independent repeats). Significances of the SLC37A2 and SLC37A4 groups versus the control group were determined using a two-tailed unpaired t-test. The exact P values were labeled above the columns. e, Chemical structures of chlorogenic acid (CHA), vanadate and inorganic phosphate (Pi). f, Chemical structures of the sugars and sugar phosphates used in the 3H-G6P competition assay.

Source data

Extended Data Fig. 3 Cryo-EM analyses of SLC37A4apo and SLC37A2apo.

a, A typical cryo-EM micrograph of SLC37A4 in the apo state (SLC37A4apo). 4078 micrographs were obtained with similar results. b, Representative 2D classification images of SLC37A4apo. c, The low-resolution cryo-EM map resolved for SLC37A4apo. d, A typical cryo-EM micrograph of SLC37A2 in the apo state (SLC37A2apo). 1903 micrographs were obtained with similar results. e, Selected fine 2D classification images of SLC37A2apo. f, Local resolution map of SLC37A2apo. g, Gold-standard Fourier shell correlation curves for the overall EM map of SLC37A2apo. h, Euler angle distributions of SLC37A2apo.

Extended Data Fig. 4 Flowchart for the data processing of SLC37A2 in the apo state and the structural analyses of SLC37A2.

a, Flowchart for the data processing SLC37A2 in the apo state (SLC37A2apo). b, EM densities of the twelve transmembrane helices of the SLC37A2apo, LO map. Residues with notable side chains are labelled. c, Structural superposition of the N-domain and C-domain of SLC37A2 structure. d, Characteristic MFS-fold observed in the SLC37A2 structure.

Extended Data Fig. 5 Cryo-EM analyses of SLC37A2 in the presence of G6P.

a, A typical cryo-EM micrograph of SLC37A2 in the presence of G6P (SLC37A2G6P). 3170 micrographs were obtained with similar results. b, Representative 2D classification images of SLC37A2G6P. c, Local resolution map of SLC37A2G6P, LO. d, Gold-standard Fourier shell correlation curves for the overall map of SLC37A2G6P, LO. e, Euler angle distributions of SLC37A2G6P, LO. f, Local resolution map of SLC37A2G6P, CO. g, Gold-standard Fourier shell correlation curves for the overall map of SLC37A2G6P, CO. h, Euler angle distributions of SLC37A2G6P, CO. i, Local resolution map of SLC37A2G6P, LO+CO. j, Gold-standard Fourier shell correlation curves for the overall map of SLC37A2G6P, LO+CO. k, Euler angle distributions of SLC37A2G6P, LO+CO.

Extended Data Fig. 6 Flowchart for the data processing of SLC37A2 in the presence of G6P.

Details can be found in the imaging processing section in Methods.

Extended Data Fig. 7 EM densities of the transmembrane helices in the symmetric SLC37A2G6P, LO and SLC37A2G6P, CO maps.

a, EM densities of the twelve transmembrane helices of the symmetric SLC37A2G6P, LO map. Residues with notable side chains are labelled aside. b, EM densities of the twelve transmembrane helices of the symmetric SLC37A2G6P, CO map. Residues with notable side chains are labelled aside.

Extended Data Fig. 8 EM densities of the transmembrane helices in the asymmetric SLC37A2G6P, LO+CO map.

a, EM density maps of the twelve transmembrane helices of the ER luminal-open protomer in the asymmetric SLC37A2G6P, LO+CO structure. Residues with notable side chains are labelled aside. b, EM density maps of the twelve transmembrane helices of the cytosolic-open protomer in the asymmetric SLC37A2G6P, LO+CO structure. Residues with notable side chains are labelled aside.

Extended Data Fig. 9 Sugar binding site in SLC37A2 and structural mapping of the SLC37A4 mutations related to human GSD-Ib disease.

a, Sugar binding site in SLC37A2 and the sugar transporter homologues of the MFS superfamily. The N-domain and C-domain of SLC37A2 and the sugar transporter homologues are in different colours. The bound G6P and glucose (Glc) molecules are shown in sticks. The PDB accession codes are indicated for the homologue structures. b, Structural superposition of glucose-bound human GLUT3 (hGLUT3) and G6P-bound SLC37A2. c, Missense mutations in SLC37A4 that cause human GSD-Ib disease mapped to the predicted SLC37A4 structure using AlphaFold2 and the corresponding mutations in the determined G6P-bound SLC37A2 structure. Mutations are shown in slate spheres. Conserved sites within SLC37A4 and SLC37A2 are coloured white. Conserved sites involved in G6P binding in SLC37A2G6P, CO are coloured red. d, 3H-G6P transport by SLC37A2WT and selected mutants corresponding to those in SLC37A4 which are associated with GSD-Ib disease. Net 3H-G6P accumulations of the mutants calculated by deleting the 3H-G6P accumulation of the control group were normalized to the net accumulation of SLC37A2WT. Data are means ± SD (n = 3 biologically independent repeats). Significances of the mutants versus wild-type (WT) were evaluated using one-way ANOVA with Dunnett’s multiple comparisons test. **** P < 0.0001. e, Side-view of the mutations corresponding to those disease-related residues in SLC37A2 near the G6P binding site in SLC37A2G6P, CO. f, Top-view of the mutations corresponding to those disease-related residues in SLC37A2 near the G6P binding site in SLC37A2G6P, CO. g, G403 residue in the SLC37A2 luminal- and cytosolic-open conformations.

Source data

Supplementary information
Reporting Summary
Peer Review File
Supplementary Table 1

Primers used in this study.

Source data
Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Statistical source data.

Source Data Fig. 3

Statistical source data.

Source Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 2

Unprocessed gel.

Source Data Extended Data Fig. 2

Statistical source data.

Source Data Extended Data Fig. 9

Statistical source data.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article
Cite this article

Lai, Q., Xu, M., Gao, Y. et al. Structural basis of glucose-6-phosphate transport by human SLC37A2. Nat Struct Mol Biol (2025). https://doi.org/10.1038/s41594-025-01712-4

Download citation

Received
18 April 2025

Accepted
15 October 2025

Published
12 November 2025

Version of record
12 November 2025

DOI
https://doi.org/10.1038/s41594-025-01712-4

Share this article

Anyone you share the following link with will be able to read this content:

Get shareable link

Provided by the Springer Nature SharedIt content-sharing initiative

Subjects
Cryoelectron microscopy
Membrane proteins

Cancel

Preview · live