Overall structure of SLC37A4

To elucidate the structural basis of G6P transport, we expressed SLC37A4 in HEK293 cells, purified the protein in the detergent lauryl maltose neopentyl glycol (LMNG) and determined its structure using single particle cryo-EM (S1 Fig). Although 2D classification revealed the presence of both monomeric and dimeric species, iterative rounds of heterogeneous refinement enabled isolation of the monomeric particles and yielded a reconstruction of the apo state at 2.8 Å resolution (S2 Fig and Table 1). The SLC37A4 structure exhibits the canonical fold of the Major Facilitator Superfamily (MFS), comprising 12 transmembrane helices (TMs). These helices are organized into two distinct domains, an N-terminal domain (NTD; TM1–6) and a C-terminal domain (CTD; TM7–12), which are connected by a highly flexible cytosolic loop. Because of this flexibility, the residues 202–211 were not resolved in the cryo-EM map and were omitted from the final model. Together, the NTD and CTD display the characteristic pseudo 2-fold symmetry of the MFS family (Fig 1A–1C).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Cryo-EM data collection, refinement, and validation statistics.

https://doi.org/10.1371/journal.pbio.3003833.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Overall structure of human SLC37A4 in the apo state.

(A) Cryo-EM density map of SLC37A4-apo viewed along the membrane plane. The protein density is shown in green and the detergent micelle shown in transparent gray. (B) Cartoon representation of SLC37A4 structure in the same membrane plane orientation as in a. N-terminal domain NTD is shown in yellow and C-terminal domain CTD is shown in cyan, individual transmembrane helices are numbered. The dashed line indicates an unresolved cytosolic loop between residues 201 and 212 that connects TM6 and TM7, linking the NTD and CTD. (C) View of SLC37A4 from the ER-lumen side along membrane normal, obtained by a 90° rotation of the model in b, showing the arrangement of the NTD and CTD. (D) Overall structure of SLC37A4 in the orientation in a, highlighting 4 positively charged residues in the putative substrate-binding pocket. (E) Cutaway view in the orientation of a, showing a slice through the putative substrate binding pocket. (F) Electrostatic surface representation of SLC37A4 viewed from the same orientation in c, ER-lumen side along the membrane normal, showing the putative substrate-binding pocket and the location of R28, K29, K64 and K240.

https://doi.org/10.1371/journal.pbio.3003833.g001

In the apo state, SLC37A4 adopts a wide outward-open conformation facing the ER lumen. This conformation likely represents the state for either G6P release into the lumen or subsequent loading of Pi, corresponding to the stage of the transport cycle prior to resetting to the cytosol-facing state (Fig 1B and 1C). Electrostatic potential surface analysis reveals a positively charged cavity in SLC37A4, primarily generated by four key basic residues: R28 and K29 on TM1, K64 on TM2, and K240 on TM7(Fig 1C). Given that both G6P and Pi are negatively charged, we speculate that this positively charged pocket is responsible for substrate binding (Fig 1D–1F).

Structure of SLC37A4 bound to G6P substrate

To elucidate the molecular basis of substrate recognition, we determined the structure of SLC37A4 in the presence of G6P at 3.2 Å resolution (S3 Fig and Table 1). Superposition of the apo and G6P-bound structures shows that the overall fold is nearly identical, with a root-mean-square deviation (RMSD) of 0.29 Å between Cα atoms, indicating that substrate binding does not induce a global conformational change (Fig 2A and 2B). Both structures were captured in an outward-facing conformation, which appears to be stabilized by extensive NTD-CTD interactions on the cytosolic side (Fig 2D). Hydrogen bonds and salt bridges are formed between D72 on TM2 and T364 on TM11, R126 on TM4 and T354 on TM10, as well as T136 on TM5 and D285 on TM8. These contacts likely stabilize the NTD and CTD in an arrangement that keeps the transporter open to the ER lumen (Fig 2D).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Structural basis of G6P recognition by SLC37A4.

(A) Superposition of the SLC37A4 apo state (gray) and the G6P-bound state (NTD in yellow, CTD in cyan). The structures are shown as cylindrical cartoons viewed along the membrane plane. The bound G6P molecule is shown as red sticks in the central cavity. The dashed line indicates an unresolved cytosolic loop between residues 201 and 212 that connects TM6 and TM7, linking the NTD and CTD. (B) View of the superposed apo and G6P-bound structures from the ER-lumen side along the membrane normal, highlighting G6P and nearby residues R28, K29, Y60, K64, F237 and K240. (C) Cryo-EM density of the bound G6P molecule, with a contour level of 3σ. (D) View of the interface between the N- and C-terminal domains from the cytosolic side along the membrane normal, showing gating residues at the domain interface, including D72, R126, T136, D285, T354 and T364. (E) Electrostatic surface representation of SLC37A4 viewed from the same orientation in b. (F) Two-dimensional schematic of the G6P-binding sites, depicting hydrogen-bond interactions between G6P and K64, as well as hydrophobic contacts with F237 and Y60. (G) GFP-based thermostability assay showing G6P-induced changes in melting temperature (ΔTm) for wild type (WT) SLC37A4 and G6P-binding–site mutants (R28A, R28C, R28H, K29A, Y60A, K64A, K240A). Bars represent mean ± s.d. of measurements performed at least in triplicate. The underlying data for this figure can be found in S1 Data. (H) Relative [¹³C₆]G6P uptake for WT and mutant SLC37A4. Bars represent mean ± s.d. of measurements performed at least in triplicate. The underlying data for this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003833.g002

Closer inspection of the positively charged cavity revealed additional density not present in the apo map (Figs 2C and S4). This density is located within the central cavity below the NTD-CTD interface (Fig 2A and 2B). This density lies within about 3 Å of K64 on TM2, and is also closely surrounded by R28 and K29 on TM1, Y60 on TM2 and F237 and K240 on TM7 (Fig 2B). Based on the size, shape, and the local chemical environment, we tentatively assigned this feature as the bound G6P substrate. The phosphate group of G6P forms hydrogen bonds with the side chain of K64, whereas the sugar moiety forms hydrophobic interactions with bulky residues Y60 and F237 (Fig 2B and 2F). Collectively, these residues define a positively charged, polar-aromatic pocket (Fig 2E) that accommodates the phosphate and glucose moieties of G6P. We generated alanine mutants and assessed their thermostability to validate the roles of these potential substrate-binding residues. The observed reduction in thermal shift upon substrate addition confirmed that these residues are indeed responsible for substrate binding (Figs 2G and S5). Furthermore, despite comparable protein expression levels across all mutants (S6 Fig), these mutations significantly impaired G6P transport activity, as demonstrated by a relative [¹³C₆]G6P uptake assay (Fig 2H). Notably, the naturally occurring disease-related mutations R28C and R28H also exhibited severe defects in both substrate binding and transport (Fig 2G and 2H).

Structure of SLC37A4 bound to inhibitor CHA

Previous studies identified CHA and its analog as potent inhibitors of SLC37A4 [13–16]. Hypothesizing that inhibitor binding might trap the transporter in a distinct conformational state, allowing us to capture an alternative stage of the transport cycle and elucidate the structural basis of G6P transport, we sought to determine the structure of the CHA-bound complex. To address the heterogeneity observed in the LMNG detergent-solubilized protein, specifically the co-existence of monomeric and dimeric species, we reconstituted purified SLC37A4 into lipid nanodiscs. This approach yielded a highly monodisperse sample, which was then incubated with the inhibitor (S1 Fig). Subsequent 2D classification and heterogeneous refinement confirmed that the SLC37A4-CHA complex exists as a homogeneous monomer. Consequently, we were able to determine the structure of the inhibitor-bound complex to a resolution of 3.3 Å (S7 Fig and Table 1).

In contrast to the apo and G6P-bound structures, the CHA-bound structure adopts an inward-facing conformation. Notably, the cytosolic loop between TM6 and TM7, which was highly flexible and unresolved in the apo state, is well-resolved in this inhibitor-bound structure (Fig 3A). This state is defined by the tight closure of the gate on the ER lumen side, which prevents access from the luminal compartment (Fig 3A–3C). This closure is maintained by specific interactions between the NTD and CTD. Key stabilizing features include a salt bridge between D47 on TM2 and K389 on TM11 (Fig 3B and 3C). In addition, the bottom of the transporter is sealed by hydrogen bonds between K46 on TM2 and A388 on TM11, and between T157 on TM5 and S263 on TM8. Surface representation confirms that these residues create a tightly sealed interface, effectively locking the ER luminal gate.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Structural basis of inhibition of SLC37A4 by CHA.

(A) Overall structure of CHA-bound SLC37A4 shown as a cylindrical cartoon in the cytosol facing conformation. The N-terminal domain NTD is in yellow and the C-terminal domain CTD is in cyan. The cytosolic loop connecting the NTD and CTD is resolved in this structure. The bound inhibitor CHA is shown in magenta. (B) View of the NTD–CTD interface from the ER-lumen side along the membrane normal, highlighting polar and charged residues at the interdomain interface, including K46, D47, T157 in the NTD and S263, A388 and K389 in the CTD. (C) Surface representation of SLC37A4 in the same orientation as in b, showing the interdomain interface patch formed by T157, K46 and D47 in the NTD and S263, A388 and K389 in the CTD. (D) View of SLC37A4 from the cytosolic side along the membrane normal, showing the CHA was trapped at the interface of NTD and CTD. (E) Close-up view of the CHA-binding pocket, highlighting residues that form hydrophobic interactions with CHA, including W118, F134, W138 and G135 from the NTD and N354, G350, V351, H366 and V369 from the CTD. (F) Cryo-EM density of the bound CHA molecule, with a contour level of 8σ. (G) Two-dimensional schematic of the CHA-binding site, illustrating hydrophobic contacts and van der Waals interactions between CHA and surrounding residues G135, F134, W138, W118, N354, G350, V351, H366, V369 and G370. (H) Structural comparison of the apo (gray) and CHA-bound (yellow/cyan) SLC37A4 structures, viewed from the membrane plane (left), from the cytosol (middle), and separated into NTD and CTD (right), illustrating that CHA-binding induces a rigid-body rotation between the two domains without major intra-domain rearrangement.

https://doi.org/10.1371/journal.pbio.3003833.g003

In this inward-facing conformation, a clear ligand-like density is observed on the cytosolic side at the NTD–CTD interface, which can be well fitted with a CHA molecule. CHA occupies a cavity near the cytosolic entrance at this interface, wedged between TM4–TM5 of the NTD and TM8–TM10–TM11 of the CTD (Fig 3D–3G). The inhibitor is surrounded by residues that engage in predominantly hydrophobic, van der Waals, and aromatic interactions, including W118, F134, W138, and G135 from the NTD, and V351, N354, H366, V369, and G370 from the CTD. Together, these residues form an extensive hydrophobic and aromatic interaction network with CHA, creating a non-polar pocket that matches the shape of CHA and explaining its high affinity and ability to stabilize the inward-facing conformation.

Compared with the apo structure, which is in an outward-open conformation facing the ER lumen, the CHA-bound structure is cytosol-facing (Fig 3H, left). Overlay of the CHA-bound and apo structures shows that the NTD and CTD themselves undergo no significant conformational changes within each domain (Fig 3H, right). Instead, the two domains move largely as rigid bodies relative to one another during the transition between ER lumen-facing and cytosol-facing conformations (Fig 3H). This is consistent with the classic rocker switch mechanism of MFS transporters [17–19].

Conformational comparison reveals a putative G6P translocation tunnel in SLC37A4

Although we lack an experimental structure of inward-facing SLC37A4 bound to G6P, the competitive inhibitor CHA captures the transporter in a distinct inward-facing state. We reasoned that the structure obtained by manually removing CHA serves as a plausible candidate representing the conformation primed for G6P loading from the cytosol. Consequently, we utilized this in silico CHA-removed model to characterize the potential G6P transport pathway using the MOLEonline server [20]. Notably, with the ER luminal gate sealed in the inward-open state, K64 sits at the cytosolic entrance of the cavity rather than at its base, whereas R28, K29 and K240 occupy the base of the cavity. These three residues lie within 5–6 Å of one another and form a triangular, positively charged trap for G6P engagement (Fig 4A and 4B). We propose that G6P binding to this deep pocket initiates the transport cycle by neutralizing the local positive charge. Engagement of the phosphate by R28 and K29 likely stabilizes tighter NTD–CTD contacts on the cytosolic side, favoring the formation of an occluded intermediate in which both gates are transiently closed. This ligand-stabilized closure is then coupled to the canonical MFS rocker-switch motion, driving a rigid-body rearrangement of the two helical bundles that seals the cytosolic gate and remodels the luminal side to generate an outward-facing state.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Conformational transitions and mechanistic model of substrate transport and inhibition.

(A) Structural models of SLC37A4 in inward-open (left) and outward-open (right) conformations viewed in the membrane plane. The N-terminal domain (NTD) and C-terminal domain (CTD) are shown as yellow and cyan cartoons, respectively; horizontal black lines indicate the approximate membrane boundaries. The cluster of basic residues in the NTD (K29, R28 and K64) and K240 in the CTD are shown as green sticks. The putative translocation pathway is depicted as a transparent surface. In the inward-open state, G6P (red sticks) binds from the cytosolic side to the NTD basic cluster, while K240 in the CTD faces the cytosolic side. In the outward-open state, reorientation of the NTD and CTD exposes the binding site to the lumen, allowing G6P release and Pi binding, as indicated by arrows. (B) Close-up views of the basic residues in the inward-open (left) and outward-open (right) conformations. Distances (Å) between side-chain nitrogen atoms are indicated by dashed lines, highlighting the compact arrangement of K29, R28 and K64 in the inward-open state and their rearrangement relative to K240 upon transition to the outward-open state. (C) The schematic illustrates the conformational transitions between the inward-open, occluded, and outward-open states. The inhibition scenario (top) shows the binding of CHA to the inward-open state, which sterically hinders the transition to the occluded state (indicated by the red cross), thereby locking the transporter. In contrast, the functional antiport cycle (bottom) depicts the exchange of substrates: G6P binds to the inward-open state, facilitating the transition to the occluded state. Subsequently, the transporter shifts to the outward-open conformation to release G6P and accommodate the binding of inorganic phosphate (Pi) for transport in the reverse direction. Key residues (R28, K29, K64 on TM 1/2 and K240 on TM 7) involved in substrate/inhibitor recognition are highlighted.

https://doi.org/10.1371/journal.pbio.3003833.g004

In the outward-facing conformation represented by the apo and G6P-bound structures, the geometry of the basic cluster shifts and R28, K29 and K240 move apart, resulting in decreased affinity for G6P. The weakened affinity promotes G6P release into the ER lumen for subsequent hydrolysis by glucose-6-phosphatase. Meanwhile, the electropositive cavity is poised to bind Pi in the ER lumen, likely engaging the same basic residues, which in turn stabilizes the return transition back toward an inward-open state. Pi release to the cytosol would complete the exchange and reset the pocket for another round of G6P/Pi exchange.

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% CO₂. 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 [¹³C₆]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 [¹³C₆]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). [¹³C₆]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 [¹³C₆]G6P by LC–MS

Lyophilized [¹³C₆]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 [¹³C₆]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.