Fructose directly remodels the translocase of the outer membrane to impair oxidative phosphorylation in podocytes

Abstract

Excessive dietary fructose consumption contributes to the rapidly increasing prevalence of obesity, metabolic syndrome, and chronic kidney disease worldwide, and accumulating preclinical evidence has confirmed that excess fructose exposure provokes severe mitochondrial dysfunction, which serves as a critical upstream driver of progressive metabolic disturbance and renal tissue injury. Conventionally, fructose-induced mitochondrial damage is thought to originate from harmful intermediate metabolites produced during intracellular fructose catabolism, while the potential direct pathogenic effect of intact unmetabolized fructose is largely overlooked. It remains unclear whether free fructose can directly target core mitochondrial complexes to initiate functional defects independent of its metabolic breakdown. Here, we report a fructose metabolism-independent mechanism in which fructose structurally remodels the translocase of the outer membrane (TOM) complex, obstructing the import of nuclear-encoded mitochondrial proteins and inhibiting mitochondrial ribosome biogenesis as well as oxidative phosphorylation. In vitro biochemical assays confirm that fructose non-covalently binds to TOM22 and induces subtle but functionally critical conformational changes in the TOM complex, thereby blocking the transmembrane translocation of mitochondrial ribosome subunits. Notably, disrupting the fructose-TOM22 binding efficiently recovers abnormal ribosome trafficking, restores compromised oxidative phosphorylation, and ameliorates mitochondrial dysfunction and glomerular pathological lesions in fructose-treated podocytes and mouse injury models. Our findings establish an innovative mechanistic paradigm that fructose acts as a direct allosteric modulator of mitochondrial membrane complexes, identifying TOM structural remodeling as a previously unrecognized molecular trigger of fructose-associated mitochondrial and metabolic disorders.

Introduction

As an essential dietary component, fructose participates in energy metabolism by facilitating fat and glycogen storage – evolutionary adaptations critical for human and animal survival.¹ Intriguingly, endogenous fructose production occurs under both physiological and pathological conditions, where it shifts energy metabolism from mitochondrial oxidative phosphorylation to glycolysis.^2,3 During human evolution, sporadic seasonal fruit intake provided limited natural fructose, and the metabolic preference toward lipid deposition helped primitive mammals survive food shortages through efficient energy stockpile. However, modern food processing and widespread addition of high-fructose corn syrup in soft drinks, baked goods, and processed condiments have drastically lifted daily human fructose intake far beyond evolutionary adaptive thresholds. The dramatic increase in dietary fructose consumption has become a major contributor to the rising global incidence of metabolic disorders, including obesity,⁴ nonalcoholic fatty liver disease (NAFLD),⁵ chronic kidney disease (CKD),⁶ and even cancer progression.⁷ Large-scale epidemiological surveys across North America, Europe and East Asia have verified positive correlations between excessive added-fructose consumption and progressive cardiometabolic and nephritic complications, imposing heavy socioeconomic burdens on global public health systems.^8–10 Accumulating preclinical evidence has confirmed that excess fructose exposure provokes severe mitochondrial dysfunction, which serves as a critical upstream driver of progressive metabolic disturbance and renal tissue injury.

For decades, existing studies have universally attributed fructose-induced tissue damage and metabolic disorders exclusively to canonical fructose catabolism and toxic downstream intermediate metabolites, which are believed to trigger mitochondrial defects and cellular dysfunction.^3,11 Conventionally, fructose-induced mitochondrial damage is thought to originate from these harmful metabolic byproducts,^1,12 while the potential direct pathogenic effect of intact, unmetabolized fructose is largely overlooked. Strikingly, no prior research has clarified whether free fructose molecules can exert biological functions independent of metabolic processing, nor has any study identified a specific mitochondrial sensor that directly responds to fructose stimulation to initiate mitochondrial dysfunction and podocyte injury. This long-standing knowledge gap severely restricts our comprehensive understanding of fructose-driven metabolic pathogenesis and hinders the development of precise targeted therapies for fructose-related diseases.

Our previous studies showed that excessive fructose intake induces glomerular podocyte injury and obvious mitochondrial functional defects in renal tissues, especially podocytes of high fructose-fed animal models.^12,13 As terminally differentiated, post-mitotic epithelial cells lining glomerular capillaries, podocytes maintain the integrity of the glomerular filtration barrier via elaborate foot process structures, which require sustained high ATP supply supported by abundant intracellular mitochondria;^14,15 such high energy dependence makes podocytes extremely vulnerable to any mitochondrial structural or functional damage triggered by nutritional stressors like surplus fructose. Current mechanistic models almost exclusively focus on fructose metabolic breakdown and subsequent toxic metabolite accumulation, leaving a critical unresolved question: whether unmetabolized fructose can directly act as a signaling molecule to regulate podocyte mitochondrial oxidative phosphorylation independent of its metabolic processing.

The mitochondrial outer membrane’s translocase (TOM) complex represents a crucial gateway for importing over 1000 nuclear-encoded mitochondrial proteins.^16–18 More than 99% of the mitochondrial proteome is nuclear-encoded and synthesized in cytoplasmic ribosomes, whose correct compartmental translocation entirely relies on properly assembled TOM machinery before maturation inside mitochondria. This evolutionarily conserved machinery plays a particularly vital role in transporting mitochondrial ribosome subunits essential for translating the 13 proteins encoded by mitochondrial DNA, all of which are critical components of the oxidative phosphorylation system.^19–21 As established in recent high-resolution cryo-EM structural and biochemical studies, the TOM complex serves as the universal entry gate for nearly all nuclear-encoded mitochondrial proteins.^17,22 Importantly, TOM is not a passive, uniform pore; distinct precursor classes follow different physicochemical routes and exit sites within the TOM40 channel.^16,23 Presequence-containing matrix-directed precursors ( ~ 60% of the mitochondrial proteome) exit TOM at a TOM22-TOM7-TOM40-defined site and transfer to the inner mitochondrial TIM23 complex, forming a dynamic TOM-TIM23 supercomplex that governs efficient matrix protein import.²⁴ Notably, TOM22 resides at the critical TOM-TIM23 handover interface, serving as a core regulatory subunit for presequence recognition, substrate sorting, and supercomplex stabilization.^25,26 Thus, perturbations to TOM22-dependent interface integrity may disrupt mitochondrial protein import, impair mitochondrial ribosome subunit trafficking, and ultimately compromise oxidative phosphorylation capacity.

Our previous work showed that high fructose exposure downregulates electron transport chain components and ATP synthase activity while impairing mitochondrial function in podocytes.¹² These cellular and in vivo observations cannot be fully explained by canonical fructose catabolism alone, strongly suggesting an unidentified direct molecular interaction between free fructose and mitochondrial TOM machinery. Building on these unresolved questions, the present study aims to define a novel metabolism-independent fructose-sensing mechanism. In this study, we identify TOM22 as a direct fructose sensor and delineate a novel metabolism-independent pathway through which unmetabolized fructose: (1) induces structural remodeling of the TOM complex, (2) inhibits general mitochondrial protein import, especially mitochondrial ribosomal subunits, and (3) consequently disrupts mitochondrial protein translation and oxidative phosphorylation. Disrupting fructose-TOM22 interaction effectively rescues mitochondrial dysfunction and glomerular injury in both cellular and animal models. These findings establish a paradigm-shifting fructose-TOM22 axis as an unrecognized pathogenic driver of fructose-associated metabolic and renal disorders, providing a novel mechanistic target for future therapeutic intervention.

Results

Fructose directly binds to TOM22

Fructose is widely reported to induce metabolic disorders, particularly podocyte injury and kidney disease, through traditional metabolic pathways and derived intermediate metabolites.^12,13 However, direct functional protein targets of fructose remain unidentified. To systematically identify protein targets that directly interact with fructose, we first designed and synthesized a fructose probe (P-fructose) containing a diazirine UV-crosslinking group and an alkyne reporter group (Fig. 1a and Supplementary Fig. 1). The acetyl protecting group in P-fructose was strategically incorporated to balance probe stability during preparation while maintaining activity during assays, with subsequent hydrolysis in cellular environments confirmed by high-resolution mass spectrometry (Supplementary Fig. 2). When evaluating biological activity, P-fructose effectively recapitulated native fructose’s pathogenic effects, including decreased mitochondrial membrane potential (Supplementary Fig. 3a, b), increased mitochondrial ROS production (Supplementary Fig. 3c, d), inhibited mitochondrial respiration (Supplementary Fig. 3e–i), and downregulated podocin expression (Supplementary Fig. 3j–l). Meanwhile, through high-performance liquid chromatography (HPLC) analysis coupled with release experiments in cultured podocytes, we showed that P-fructose underwent time-dependent metabolic clearance when cells were transferred to P-fructose-free medium ([P-fructose] = −4% at 1 h, −12.8% at 2 h; Supplementary Fig. 4a–d), indicating that podocytes actively metabolize P-fructose, being consistent with fructose’s endogenous metabolic pathway. In 3 h stimulation paradigm (5 mM P-fructose in culture medium), HPLC analysis verified sufficient intracellular retention of intact P-fructose for target labeling (Supplementary Fig. 4e–l). These results suggested that P-fructose is a suitable chemical probe for cell-based protein target identification of fructose.

Fig. 1.

Fructose directly binds to TOM22. a P-fructose structure with a fructose molecule (shown in pink), a diazo group (shown in blue) for UV crosslinking, and an alkynyl group (shown in blue) for click chemistry. b Workflow of the compound-centric chemical proteomics experiment to identify fructose direct binding proteins in podocytes. c Silver staining photo showing click chemistry-enriched fructose binding proteins. d Venn diagram showing potential fructose binding proteins. e Top 20 enriched KEGG pathways of potential fructose binding proteins. * represents mitochondria-related pathways. MS/MS spectra of two representative TOM22 peptides: LWGLTEMFPER (MH₂⁺², 689.84) (f) and QILLGPNTGLSGGMPGALPSLPGKI (MH₂⁺², 1202.67) (g). h Representative images of Western blot analysis for P-fructose-enriched TOM22 protein with or without native fructose competition in podocytes. β-ACTIN was used as an internal control. n = 3 independent experiments. i–k MST analysis showing that fructose had high binding capacity with TOM22-eGFP (i, using eGFP as negative control) and TOM complex (j, using bovine serum albumin (BSA) as negative control). MST assay showing that fructose had lower binding capacity with TOM22-eGFP in the presence of KHK (i, using eGFP as negative control), and F-1-P had low binding capacity with TOM22-eGFP (k, using eGFP as negative control). n = 3 independent experiments. N.A. means not available. ∆F_Norm [‰] decreased the normalized fluorescence signal compared to the original state. Data are represented as the mean ± SD. l Cellular imaging of P-fructose (red with rhodamine labeling) and TOM22 (green) colocalization in podocytes with or without native fructose competition. Both P-fructose and TOM22 were observed by immunofluorescence. Scale bar: 10 μm. See also Supplementary Figs. 1–7 and Supplementary Table 1

Building on this foundation, we conducted compound-centric chemical proteomics²⁷ in P-fructose-stimulated podocytes, where UV irradiation covalently crosslinked fructose-bound proteins to the probe. In podocytes incubated with P-fructose, the cellular proteins that directly bound to fructose were covalently crosslinked to P-fructose by UV irradiation. After orthogonal conjugation was achieved with azide-biotin via click chemistry and affinity purification was performed, P-fructose-enriched proteins were analyzed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) (Fig. 1b). As shown in Fig. 1c, many proteins were significantly enriched in the P-fructose group compared with the control and competition groups. When the threshold as peptide count was ≥ 2 and the enrichment ratio was ≥ 2-fold, we collectively identified 72 potential fructose-binding proteins (Fig. 1d and Supplementary Table 1), which were involved in mitochondria-located events, including citrate cycle, pyruvate metabolism, fatty acid metabolism, and ketone body synthesis and degradation (Fig. 1e).

Our previous work demonstrated that high fructose induced podocyte injury through mitochondrial metabolic reprogramming,¹² implying fructose potentially may regulate mitochondrial function via direct interaction with mitochondrial proteins. Since fructose entry into mitochondria has not been experimentally documented, we hypothesized that its effects might be mediated through mitochondrial outer membrane proteins. We therefore focused on TOM22, a core component of the mitochondrial import machinery (TOM complex) that orchestrates protein translocation and regulates mitochondrial biogenesis and bioenergetics.¹⁷ Notably, TOM22 emerged as: (1) the most significantly enriched protein among putative fructose targets (Fig. 1f, g), and (2) the top hit in our mass spectrometry analysis (Supplementary Fig. 5), strongly suggesting its pivotal role in fructose-mediated mitochondrial regulation. We next verified that TOM22 was a direct binding protein of fructose (Fig. 1h–l). We analyzed P-fructose probe-enriched protein complex and found that specific binding of TOM22 to P-fructose was strongly outcompeted by native fructose (Fig. 1h). The data from microscale thermophoresis (MST) measurement using recombinant TOM22-enhanced green fluorescent protein (eGFP) fusion protein or recombinant TOM complex protein (labeled by amine-reactive probe) showed that TOM22 readily bound to fructose, and the dissociation constant (K_D) of fructose-TOM22/TOM complex was 30.91 ± 3.28 μM/28.08 ± 5.82 μM, whereas the eGFP control for TOM22-eGFP or bovine serum albumin (BSA) control for TOM complex showed no detectable interaction with fructose (Fig. 1i, j). However, glucose and other related sugar such as ribose, did not bind to TOM22 (Supplementary Fig. 6). Ketohexokinase (fructokinase, KHK) catalyzes the conversion of fructose into fructose-1-phosphate (F-1-P). In fact, very low expression level of fructose transporters (such as Slc2a2 and Slc2a5) and metabolizing enzymes (such as KHK-a, KHK-c, and aldolase B (AldoB)) was observed in podocytes compared with liver cell line AML12 cells and gut cell line MODE-K cells (Supplementary Fig. 7a–g). Moreover, public single-nucleus RNA-seq (snRNA-seq) datasets showed that key fructose transporters (e.g., Slc2a2 and Slc2a5) and fructose-metabolizing enzymes (e.g., KHK, AldoB) were obviously low-expressed in podocytes compared with other types of kidney cells (Supplementary Fig. 7h, i), indicating that the conversion of fructose to F-1-P may be slow or relatively little in podocytes, and therefore had more fructose to interact with TOM22. Competition/affinity assay between KHK or F-1-P and TOM22 showed that fructose binding to TOM22 was strongly competed by KHK, with K_D increased from 30.91 ± 3.28 μΜ to 5 ± 1 mM (Fig. 1i), whereas, K_D of F-1-P binding to TOM22 was 226.76 ± 99.27 μΜ (Fig. 1k), indicating a lower binding capability of F-1-P to TOM22 than that of fructose. Furthermore, we performed fluorescence microscopy to visualize the distribution of P-fructose in podocytes and observed strong colocalization of endogenous TOM22 and P-fructose (Fig. 1l). However, upon the stimulation with a high dose of native fructose, the colocalization fluorescence signals of P-fructose and TOM22 were decreased sharply. Together, these data conclusively establish TOM22 as a direct fructose-binding target within the TOM complex.

Fructose remodels the TOM complex by direct binding to TOM22 at its K105 site

After TOM22 was identified as a direct target of fructose, we mapped the potential sites at which fructose binds to TOM22. We analyzed trypsin-digested peptides of P-fructose-labeled recombinant TOM22 by high-resolution mass spectrometry (Fig. 2a). Among the identified peptides, only one tryptic peptide, ¹⁰⁶LQME^(P-Fruc)QQQQLQQR¹¹⁷, showed a precise +472.22 Da mass shift, representing the P-fructose modification on its precursor ion, and the site of P-fructose labeling was predicted with a high probability to be on E109 through the MS/MS fragmentation pattern (Fig. 2b).

Fig. 2.

Panorama of the TOM complex with fructose. a Schematic workflow for identifying the P-fructose-modified TOM22 peptide by high-resolution mass spectrometry. b Representative MS/MS spectrum of the P-fructose-modified TOM22 peptide LQME^(P-Fruc)QQQQLQQR (MH₂²⁺, 1015.99). c Alignment of the TOM complex in the apo state (yellow with 20% transparency) and the TOM complex with ligand (forest green). FUD, fructose. d Interactions between the fructose molecule and the TOM complex. The amino acid residues involved in the interactions are displayed in stick form and colored according to their respective subunits. e The interaction between TOM22-E102 and TOM40-K330 exhibits high evolutionary conservation. TOM22-E102 and TOM40-K330 are shown in stick form. In the left panel, the residues are colored according to their respective subunits, while in the right panel, the residues are colored based on their level of conservation. f The binding of the fructose molecule disrupts the interaction between TOM22-E102 and TOM40-K330 on one side of the complex. TOM22-E102 and TOM40-K330 are shown in stick form. See also Supplementary Figs. 8, 9 and Supplementary Table 2

To further explore the details of the interaction between the fructose molecule and the TOM complex, we expressed human TOM complex proteins in HEK293F cells and incubated them with the fructose molecule. Using single-particle cryo-EM, we successfully determined the structures of the TOM complex in the apo state and the TOM complex bound to the substrate, with resolutions of 2.89 Å and 2.93 Å, respectively (Supplementary Fig. 8 and Supplementary Table 2). The structures of the TOM complex in both states were aligned and compared. As previously reported, TOM22 is inserted between two TOM40 molecules along the interface of the TOM40 dimer, and is responsible for the early recognition of precursors to be transported. Subsequently, TOM22 transfers the precursors to the TOM40 core channel.^18,25,28 Our structural data show that fructose binds to one side of the TOM complex at the center of the intermembrane space (IMS) side in an open-chain form, interacting extensively with the TOM22 subunit, the TOM40 subunit, and a phospholipid molecule, forming a highly integrated binding network. While the overall structures of the two states show a high degree of consistency, significant differences are observed at the substrate-binding site, which provides a valuable opportunity to gain deeper insights into the regulation of the TOM complex by fructose (Fig. 2c–f and Supplementary Fig. 9a–h). The binding pocket of the TOM complex, characterized by a positive electrostatic potential, is indeed enriched with multiple leucine residues (L312, L328, L339, and L341). The combined hydrophobic and electrostatic environment of the TOM complex’s binding pocket not only ensures a strong interaction with the open-chain form of fructose but also appears to actively prevent cyclization, thus stabilizing the metabolically active form of the fructose (Fig. 2d and Supplementary Fig. 9h). Notably, TOM22-K105 is located at the edge of the binding pocket near the IMS side, acting as a gatekeeper for the entry and exit of fructose molecule, and likely playing a crucial role in their recruitment and stabilization (Fig. 2d).

Of note, in addition to TOM22, TIM44 (Mitochondrial import inner membrane translocase subunit 44), TIM17-A (Mitochondrial import inner membrane translocase subunit 17 A) and TOM40 were also significantly enriched among P-fructose binding proteins (Supplementary Fig. 5). Here, among these interactions, the binding of fructose significantly disrupts the highly conserved interaction between TOM22-E102 and TOM40-K330 (Fig. 2e, f and Supplementary Fig. 9a–f). In the apo state of the TOM complex, the interaction between TOM22-E102 and TOM40-K330 exhibits high evolutionary conservation (Fig. 2e). Surprisingly, the fructose molecule ‘occupies’ the position of TOM40-K330 on one side, disturbing this highly conserved interaction (Fig. 2f). Interestingly, the fructose molecule only occupies one of the two similar binding sites. This seems reasonable, as simultaneous disruption of both key sites could have drastic effects on the function of the TOM complex. Structurally, the fructose molecule’s occupation of one site is likely due to interference from the phospholipid molecule. When fructose binds to one site, steric hindrance causes the phospholipid molecule to be pushed to the other side, blocking the entrance with its head groups and preventing further binding of the fructose molecule (Fig. 2e, f and Supplementary Fig. 9e, f).

Additionally, the binding of the fructose molecule induces minor conformational changes in the TOM40 segments L307-L312 and L328-L337 (Supplementary Fig. 9g). Since TOM40 serves as the primary channel for transport through the TOM complex, these subtle conformational changes could still impact the transport function of the TOM complex.

Fructose-TOM22 binding suppresses mitochondrial ribosome transport

Given that TOM22 is essential for the biogenesis of the TOM complex and facilitates the import of mitochondrial precursor proteins,²⁹ the binding of fructose to TOM22 and the TOM complex is likely to disrupt mitochondrial protein transport. To investigate this possibility, we conducted data-independent acquisition (DIA)-based quantitative proteomics on mitochondrial fractions isolated from fructose-treated podocytes (0, 24, 48, and 72 h), enabling a comprehensive analysis of mitochondrial proteome dynamics (Supplementary Fig. 10). From the four time points, we identified 4,633 proteins, with 3,500 reliably quantified (Supplementary Fig. 10a). Notably, 78.61% (893 proteins) overlapped with the MitoCarta3.0 mitochondrial protein database (Supplementary Fig. 10b), confirming high mitochondrial enrichment. Following bioinformatics analysis (Supplementary Fig. 10c) and soft clustering (Supplementary Fig. 10d), these quantified proteins with similar expression changing trends were divided into six distinct clusters. Among them, two clusters (clusters 4 and 5 in Supplementary Fig. 10d) exhibited a progressive decline in protein abundance over time, suggesting impaired mitochondrial import under fructose stimulation. These two clusters were prioritized for further analysis to identify fructose-sensitive mitochondrial proteins whose transport was selectively disrupted (Fig. 3a). Subsequent functional annotation of these downregulated clusters revealed enrichment of several important mitochondrial pathways including mitochondrial protein import, regulation of mitochondrial gene expression, TCA cycle and pyruvate metabolism, fatty acid beta-oxidation, respiratory chain complex, together with two top-ranking categories: ‘Organellar ribosome’ and ‘Mitochondrial translation initiation’ (Fig. 3a-c, Supplementary Fig. 10e, f and Supplementary Table 3). These two categories encompassed a significant number of mitochondrial ribosome subunits (Fig. 3b). Consistent with the DIA quantitative proteomic data, the protein levels of the mitochondrial ribosome subunits MRPS10, MRPS23, MRPL22, MRPL37, MRPL41 and MRPL43 were continuously decreased in fructose-exposed podocytes (Fig. 3d, e, i). Interestingly, fructose did not downregulate the protein level of the TOM complex, including TOM22, TOM20, TOM40 and TOM70, compared with the untreated podocytes (Fig. 3d, f, i). Given that mitochondrial ribosome is indispensable for mitochondrial protein translation, we subsequently determined the protein level of ND1-5, MTCYB, MTCO2-3, and ATP6, whose translation depends exclusively on the mitochondrial ribosome, and found that their levels were notably inhibited by fructose in podocytes (Fig. 3g–i). Consistent with these observations, fructose altered mitochondrial morphology, as evidenced by the disappearance of mitochondrial cristae and disruption of mitochondrial membranes (Fig. 3j). These findings indicated that fructose does not affect stability of the TOM complex, but only suppresses the import of mitochondrial ribosome subunits and therefore inhibits mitochondrial protein synthesis rate in podocytes.

Fig. 3.

Fructose-TOM complex signaling downregulates the protein level of mitochondrial ribosome subunits and inhibits mitochondrial protein synthesis rate. a Bubble chart showing annotation enrichment analysis of String clusters of mitochondrial proteins with continuously decreased expression identified by soft clustering analysis (clusters 4 & 5 in Supplementary Fig. 10d) in DIA-based quantitative proteomics of mitochondria isolated from fructose-treated podocytes (0, 24, 48, 72 h). b Heatmap of expression profiles for mitochondrial ribosome subunits (MRPs) and TOM complex subunits from DIA proteomics. c Heatmap of expression profiles for fatty acid beta-oxidation, mitochondrial DNA (mtDNA)-encoded proteins, respiratory electron transport and TCA cycle from DIA proteomics. d–f Western blot analysis of mitochondrial ribosome subunits (MRPS10, MRPS23, MRPL11, MRPL14, MRPL22, MRPL37, MRPL41 and MRPL43) and TOM complex subunits (TOM22, TOM20, TOM40 and TOM70) in fructose-cultured podocytes (d) and the related quantification results (e, f). g, h Western blot analysis of the mtDNA-encoded proteins ND1-5, MTCYB, MTCO2-3 and ATP6, which are translated by mitochondrial ribosomes in fructose-cultured podocytes (g), and the related quantification results (h). In panels (d–h), protein expression levels were normalized to voltage-dependent anion channel 1 (VDAC1) as a mitochondrial internal control, with grayscale analysis conducted across six independent experiments. i Volcano plot showing differentially expressed mitochondrial proteins in podocytes treated with fructose for 72 h vs. 0 h from western blot analysis. j Transmission electron microscopy images showing mitochondrial ultrastructure of podocytes treated with high fructose modeling. Scale bars: 500 nm. k MST analysis for fructose binding to TOM22 wild-type (WT) and mutants E102A, K105A and E109A. eGFP was used as a negative control. n = 3 independent experiments. l Bubble charts showing the top 10 enriched gene ontology terms for the downregulated proteins identified by soft clustering analysis (clusters 9 in Supplementary Fig. 10g) in DIA-based quantitative proteomics of mitochondria isolated from fructose-treated podocytes with the TOM22^K105A mutation. m Volcano plot showing differentially expressed mitochondrial proteins in fructose-treated podocytes expressing wild-type TOM22 from DIA proteomics. n Bubble charts showing the top 10 enriched gene ontology terms for the downregulated proteins in DIA-based quantitative proteomics of mitochondria isolated from fructose-treated podocytes expressing wild-type TOM22. o, p Western blot analysis and the related quantification results for mitochondrial ribosome subunits MRPL46 (o) and MRPS21 (p) in podocytes with siRNA knockdown of endogenous TOM22 followed by overexpression of wild-type or mutant (E102A, K105A and E109A) TOM22 and fructose exposure. n = 3 independent experiments. q HPG-click chemistry assay showing newly synthesized proteins (green with HPG-Azide labeling) and MitoTracker (red) colocalization in podocytes with siRNA knockdown of endogenous TOM22 followed by overexpression of wild-type or K105A mutant TOM22 and fructose exposure. Scale bar: 15 μm. r–u PET/CT scan (r) and gamma-counter counting (s–u) showing import of mitoribosomal protein MRPL37 into isolated mitochondria with overexpression of TOM22 wild-type or mutant (E102A and K105A) with or without fructose. For all graph panels, data are represented as the mean ± SD, P value denotes one-way ANOVA, N.S., no significance, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 vs. control group. See also Supplementary Fig. 10, 11 and Supplementary Table 3

To further validate the binding sites of fructose and TOM22 and investigate the role of TOM22 in fructose-induced downregulation of mitochondrial ribosome subunits, we produced recombinant mutant TOM22 (E102A, K105A and E109A)-eGFP fusion proteins and performed MST to assess their binding capacity to fructose, respectively. Unlike the binding affinity of mutant TOM22 (E102A and E109A) to fructose was slightly inhibited compared with that of wild-type TOM22, the TOM22 K105A mutation completely lost the ability to bind fructose (Fig. 3k). In addition, we overexpressed wild-type TOM22 or its mutants (E102A, K105A, E109A) in podocytes with endogenous TOM22 knockdown, and confirmed their comparable expression at both mRNA and protein levels without fructose treatment (Supplementary Fig. 11a, b). We then applied DIA-based quantitative proteomics to profile global mitochondrial protein dynamics across four key conditions (WT/K105A ± fructose) (Fig. 3l–n; Supplementary Fig. 10g, h). Consistent with our earlier observations (Fig. 3a–i; Supplementary Fig. 10a–f), the TOM22 K105A mutation fully restored fructose-induced changes in mitochondrial ribosomal subunit abundance. Specifically, fructose exposure significantly reduced the expression of mitochondrial ribosomal subunits in cells expressing wild-type TOM22 or the E102A/E109A mutants, whereas the K105A substitution effectively prevented this downregulation of mitochondrial ribosomal subunits in fructose-treated podocytes (Fig. 3o, p; Supplementary Fig. 11c–g).

To verify the decrease of mitochondrial ribosome subunits was directly regulated by fructose-TOM22 binding, we directly measured mitochondrial translation activity by HPG-click chemistry assay and found that fructose stimulation significantly inhibited mitochondrial protein synthesis in cultured podocytes expressing wild-type TOM22, however, this inhibition was markedly rescued by TOM22 K105A mutation (Fig. 3q). We further assessed the efficiency of mitochondrial protein transport in vitro using recombinant MRPL37 and isolated wild-type or mutant TOM22 (E102A and K105A) mitochondria in the presence or absence of fructose (Supplementary Fig. 11h). We found that, similar to the results from cellular experiments, fructose notably inhibited MRPL37 transport into wild-type TOM22 mitochondria at all tested time points (15, 30, 60, and 120 min), and partially inhibited its import into TOM22^E102A mitochondria at 15 min, but failed to inhibit its import into TOM22^K105A mitochondria (Fig. 3r–u and Supplementary Fig. 11i, j). Taken together, these findings suggest that TOM22-K105 is the key binding site of fructose, and TOM complex senses fructose by fructose-TOM22 binding to suppress the import of mitochondrial ribosome subunits from the cytoplasm to mitochondria and subsequent mitochondrial protein translation in podocytes.

TOM complex senses fructose to impair mitochondrial oxidative phosphorylation in podocytes

Since mitochondrial ribosomes are essential for translating all 13 mitochondrial DNA (mtDNA)-encoded proteins, which are key subunits of the electron transport chain (ETC) and oxidative phosphorylation system,^19–21 we further investigated whether fructose binding to TOM22 could impair mitochondrial oxidative phosphorylation. Cultured endogenous TOM22 knockdown podocytes were transfected with wild-type or K105A mutant TOM22 vector and then incubated with fructose. Of note, podocytes expressing the K105A mutant TOM22 were less sensitive to fructose treatment. Fructose increased mitochondrial ROS and decreased mitochondrial membrane potential in podocytes expressing wild-type TOM22, while the TOM22 K105A mutation reversed the alteration in mitochondrial ROS (Fig. 4a) and mitochondrial membrane potential induced by fructose (Fig. 4b, c). Mitochondrial respiration in wild-type or K105A mutant TOM22-expressing podocytes treated with control or fructose was confirmed by oxygen consumption rate (OCR) analysis. Compared to the control group, fructose had no significant effect on basal respiration, ATP production, maximal respiration or spare respiratory capacity in podocytes harboring K105A mutant TOM22, while fructose caused low mitochondrial respiration in podocytes expressing wild-type TOM22 (Fig. 4d–h). mtDNA copy number analysis showed that fructose treatment impaired the maintenance of mitochondrial genome integrity in cultured podocytes, which was recovered by the TOM22 K105A mutation (Fig. 4i). Furthermore, the TOM22 K105A mutation rescued the fragmented morphology of podocyte mitochondria induced by fructose (Fig. 4j, k). Therefore, TOM complex senses fructose by fructose-TOM22 binding may impair oxidative phosphorylation and induce mitochondrial dysfunction by remodeling the TOM complex in podocytes.

Fig. 4.

TOM complex senses fructose to impair mitochondrial oxidative phosphorylation in podocytes. a Mitochondrial ROS levels were analyzed by MitoSOX staining and measured by flow cytometry in podocytes. Podocytes were transfected with siRNA to knock down endogenous TOM22 followed by overexpression of wild-type or K105A mutant TOM22 vector and stimulation with fructose. The fluorescence intensity was quantified. n = 6 independent experiments. b, c Mitochondrial membrane potential was analyzed by JC-1 staining and measured by flow cytometry in podocytes (b). The ratio of aggregate/monomer fluorescence intensity was calculated (c). n = 6 independent experiments. d–h OCR was measured with a Seahorse extracellular flux analyzer in podocytes (d). Basal respiration (e), ATP production (f), maximal respiration (g), and spare respiratory capacity (h) were calculated from OCR traces. n = 6 independent experiments. i Real-time qPCR analysis for mtDNA copy number in podocytes. n = 6 independent experiments. j, k Representative images of mitochondrial morphology in podocytes (scale bar: 10 μm) (j). Bar chart showing quantification of the mitochondrial shape of podocytes (k). For all graph panels, data are represented as the mean ± SD, P value denotes one-way ANOVA, N.S., no significance, ^*P < 0.05, ^**P < 0.01^{, ***}P < 0.001 vs. control group

Tom22^K105A mutation attenuates high fructose-induced pathophysiological injury in mouse kidney

In addition to elucidating its role in cultured podocytes, it is critical to investigate whether the fructose-TOM signaling complex contributes to essential physiological and pathological processes. To address this question, we disrupted the fructose-Tom22 interaction and generated a floxed mouse strain C57BL/6JGpt-Tomm22^{em1Cin(LSL-p.K105A)}/Gpt (Supplementary Fig. 12a) carrying a K105A mutation in the Tom22 fructose-binding site. This strain was then crossed with a renal glomerulus-specific Cre strain, C57BL/6JGpt-H11^{em1Cin(Nphs1-iCre)}/Gpt, to produce renal glomerulus-specific Tomm22^K105A knock-in mice (K105A) and their wild-type (WT) siblings (Supplementary Fig. 12b, c).

Both K105A and WT mice were fed either standard or high-fructose diet for 12 weeks as previously described (Fig. 5a).^30–32 Histologically, K105A mutation alleviated glomerular mesangial expansion in mice induced by high fructose diet (Fig. 5b, c). Additionally, the K105A mutation reversed high fructose-induced foot process fusion, effacement, and flattening in glomerular podocytes (Fig. 5d, e). To assess kidney physiological function in K105A mice compared with WT mice, with or without high fructose intake, we also measured urinary albumin and creatinine levels (Fig. 5f–h), as well as serum blood urea nitrogen (BUN) and creatinine levels (Fig. 5i, j). As expected, high-fructose diet increased urinary albumin levels and the urinary albumin to creatinine ratio (urinary albumin/creatinine) while reducing urinary creatinine levels in WT mice, however, these changes were reversed by the K105A mutation (Fig. 5f–h). Consistently, the elevated serum levels of BUN and creatinine induced by high fructose diet were also mitigated by the K105A mutation (Fig. 5i, j). These observations indicated that Tom22 K105A mutation attenuated high fructose-induced kidney injury.

Fig. 5.

Tom22^K105A mutation attenuates high fructose-induced pathophysiological injury in mouse kidney. a Strategy of high fructose modeling in WT and K105A mice. b, c Periodic Acid-Schiff (PAS) staining showing renal glomerulus morphology of WT and K105A mice with or without high fructose modeling (scale bars: 50 μm) (b) and quantitative analysis of glomerular mesangial matrix expansion (c). n = 7 independent experiments. d, e Transmission electron microscopy images showing ultrastructure of glomerular podocyte of WT and K105A mice with or without high fructose modeling (scale bars: 500 nm) (d) and quantitative analysis of foot process width (e). n = 7 independent experiments. f–h Urinary albumin (f), urinary creatinine (g), and urinary albumin to creatinine ratio (h) of WT and K105A mice with or without high fructose modeling. n = 7 independent experiments. i, j Serum BUN (i) and serum creatinine (j) of WT and K105A mice with or without high fructose modeling. n = 7 independent experiments. For all graph panels, data are represented as the mean ± SD, P value denotes one-way ANOVA, N.S., no significance, ^**P < 0.01, ^***P < 0.001 vs. control group. See also Supplementary Fig. 12

Tom22^K105A mutation disrupts fructose-Tom22 interaction to rescue the decrease of mitochondrial ribosomal subunits and impaired mitochondrial oxidative phosphorylation in mouse glomeruli

To further investigate the molecular mechanisms underlying the attenuation of high fructose-induced kidney injury in Tomm22^K105A knock-in mice, we compared the expression levels of mitochondrial ribosomal proteins in glomeruli and assessed downstream mitochondrial oxidative phosphorylation, with wild-type (WT) mice serving as controls. As shown in Fig. 6, K105A mice almost completely rescued the decrease of mitochondrial ribosomal subunits and the expression level of proteins translated by mitochondrial ribosome under high fructose diet compared with WT mice (Fig. 6a–d). Consequently, the impaired oxidative phosphorylation and mitochondrial morphology of glomerular podocytes induced by high fructose was also rescued in K105A mice (Fig. 6e–j). These data suggested that Tom22 K105A mutation disrupts fructose-Tom22 interaction to rescue the decrease of mitochondrial ribosomal subunits and impaired mitochondrial oxidative phosphorylation in mouse glomeruli.

Fig. 6.

Tom22^K105A mutation disrupts fructose-Tom22 interaction to rescue the decrease of mitochondrial ribosomal subunits and impaired mitochondrial oxidative phosphorylation in mouse glomeruli. a, b Representative images of Western blot (a) and gray analysis (b) showing the expression level of mitochondrial ribosome subunits (Mrps10, Mrps23, Mrpl11, Mrpl14, Mrpl22, Mrpl37, Mrpl41 and Mrpl43) in mouse glomeruli and the related quantification results. c, d Representative images of Western blot (c) and gray analysis (d) showing the expression level of mitochondrial DNA-encoded proteins (Nd1-5, Mtcyb, Mtco2-3 and Atp6), which are translated by mitochondrial ribosome in mouse glomeruli and the related quantification results. In panels (a-d), protein expression levels were normalized to Vdac1 as an internal control, with grayscale analysis performed on five independent experiments. e-i OCR was measured with a Seahorse extracellular flux analyzer in mouse glomeruli (e). Basal respiration (f), ATP production (g), maximal respiration (h), and spare respiratory capacity (i) were calculated from OCR traces. n = 6 independent experiments. j The mitochondria ultrastructure was detected by transmission electron microscopy in glomerular podocytes of mice (scale bars: 500 nm). k Proposed working model of fructose-TOM complex axis in oxidative phosphorylation dysfunction. This mechanistic diagram was manually drawn using Microsoft PowerPoint, and no online tools were utilized for image generation. For all graph panels, data are represented as the mean ± SD, P value denotes one-way ANOVA, N.S., no significance, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 vs. control group

In summary, we reveal a novel fructose-TOM complex axis that regulates fructose-related mitochondrial dysfunction (Fig. 6k), in which TOM22 senses nutrient fructose to directly induce a subtle but essential conformational change of the TOM complex. The fructose-driven remodeling of the TOM complex globally inhibits the import of nuclear-encoded mitochondrial proteins into the matrix, with mitochondrial ribosomal subunits serving as a representative example. This generalized impairment of mitochondrial protein import subsequently results in oxidative phosphorylation dysfunction.

Discussion

For decades, fructose has been causally linked to diverse metabolic disorders,^4–7 particularly podocyte injury in chronic kidney disease,^12,13 through its distinctive metabolic pathways and intermediate metabolites. Strikingly, however, no prior evidence has indicated that the free fructose molecule regulates mitochondrial oxidative phosphorylation and promotes podocyte pathology through mechanisms independent of canonical fructose metabolism. In this study, we have made the seminal discovery that fructose directly interacts with TOM22, a pivotal subunit of the TOM complex that orchestrates mitochondrial protein import.¹⁸ Our structural and biochemical analyses demonstrate that E109 within the ¹⁰⁶LQMEQQQQLQQR¹¹⁷ peptide of TOM22 can be modified by P-fructose, and K105 of TOM22 is indispensable for mediating fructose-TOM22 interaction, while K105 emerges as essential for mediating fructose-TOM22 binding. Most crucially, we establish that fructose binding at the K105 site fundamentally reprograms both the conformational landscape and transport functionality of the TOM complex.

The TOM complex serves as the universal entry gate for nearly all nuclear-encoded mitochondrial proteins.¹⁶ Building on our previously proposed surface potential model of human TOM-mediated translocation,¹⁸ our study reveals that fructose directly targets the outer-membrane TOM complex (TOM22/TOM40), rather than inner-membrane TIM components. While our chemical proteomics identified TIM subunits alongside TOM proteins, this likely reflects their tight physical association within the TOM-TIM23 supercomplex, not direct fructose binding. As documented in structural studies, TOM is not a passive pore but a sophisticated hub: presequence-containing matrix proteins (~60% of the mitochondrial proteome) exit TOM at a TOM22-TOM7-TOM40 site and transfer to TIM23 via a functional supercomplex, with TOM22 acting as the critical handover regulator.^{16,22–24,26,33} Fructose binding at the TOM22-TOM40 interface reorganizes the electrostatic landscape of the central channel, destabilizing the TOM-TIM23 supercomplex and broadly inhibiting presequence-dependent import.³⁴ Our DIA proteomics across four conditions (WT/K105A ± fructose) confirm this effect: while most mitochondrial proteins, including a great number of matrix proteins, are affected, mitochondrial ribosomal proteins (MRPs) show marked impairment in mitochondrial import due to their high-flux, assembly-dependent import requirements.^35–37 Thus, the phenotype stems from a general kinetic defect, possibly not specific targeting. We acknowledge that the dynamic regulation of the TOM-TIM23 supercomplex by fructose remains incompletely understood. Future work will combine advanced structural and functional assays to visualize fructose-induced conformational changes and further define the substrate-specific import mechanism. Additionally, while fructose-bound TOM complexes maintain overall structural integrity, we observe subtle but mechanistically important perturbations in phospholipid-protein interactions that may influence complex assembly dynamics. Given the established roles of phospholipids in TOM complex biogenesis and the fact that fructose targets the fully assembled complex,³⁸ future studies should explore the potential regulatory effects of fructose on TOM complex assembly pathways.

Our cryo-EM structural elucidation reveals remarkably selective recognition of fructose’s open-chain conformation (representing merely ~0.5% of aqueous equilibrium states) by the TOM complex. This ephemeral but chemically reactive form is also essential for ketohexokinase activity through its aldehyde group, and exhibits dramatically reduced binding when KHK competes for substrate (Fig. 1i). The binding pocket’s hydrophobic leucine residues critically exclude water molecules that would otherwise destabilize open-chain fructose by solvating its polar hydroxyl groups. This exquisite molecular recognition mechanism highlights the TOM complex’s capacity for conformation-specific substrate discrimination.

Consistent with our structural model, fructose sensing by the TOM complex profoundly inhibits mitochondrial ribosome subunit import in both cultured podocytes and mouse glomeruli (Figs. 3d, 6a). Given that mitochondrial ribosomes are indispensable for intramitochondrial protein synthesis, their dysfunction, frequently associated with mutations in ribosomal components, underlies pathologies including cardiomyopathy, sensorineural hearing loss, and renal failure.^39–41 Importantly, we demonstrate that fructose stimulation reduces levels of key mitochondrial ribosomal subunits known to suppress mitochondrial translation,^42,43 thereby establishing a direct mechanistic link to mitochondrial dysfunction in podocytes. Subsequent metabolic perturbations include downregulation of electron transport chain components (cytochrome c oxidase and NADH CoQ reductase complexes) and ATP synthase, accompanied by impaired oxidative phosphorylation, elevated mitochondrial ROS, and eventual podocyte injury. These molecular changes manifest as classical hallmarks of kidney pathology: glomerular hypertrophy, mesangial expansion, foot process effacement, and elevated urinary albumin/creatinine ratio alongside increased serum BUN and creatinine (Figs. 4, 5 and 6e–j). Remarkably, the TOM22 K105A mutation – which abolishes fructose binding – completely restores mitochondrial ribosome import and function while preventing all histological and physiological damage in both cellular and Tomm22^K105A knock-in mouse models (Figs. 4, 5 and 6e–j). These findings conclusively establish the fructose-TOM complex axis as a novel pathogenic mechanism impairing oxidative phosphorylation through selective inhibition of mitochondrial ribosome biogenesis, thereby identifying TOM22 K105 as a promising therapeutic target for fructose-induced metabolic disorders.

Given the TOM complex’s ubiquitous expression and central role in mitochondrial proteostasis,⁴⁴ we propose that fructose-TOM signaling represents a fundamental physiological regulatory mechanism with broad pathophysiological implications. This dual-action system may synergize with canonical fructose metabolism to: (1) downregulate oxidative phosphorylation while enhancing glycolysis, which promotes energy storage as fat/glycogen for long-term survival; or (2) reduce oxygen demand to mitigate hypoxic damage in disease states.^2,3,45 Conversely, chronic overactivation of this pathway could explain fructose’s established role in promoting metabolic disorder pathogenesis, including chronic kidney disease,⁶ NAFLD,⁵ and certain cancers.^46–49 Although most dietary fructose undergoes hepatic KHK-C-mediated conversion to F-1-P, our data reveal that even in high fructokinase tissues (such as liver, intestine), fructose directly inhibits mitochondrial ribosome import and oxidative phosphorylation through TOM22-dependent mechanisms (Supplementary Fig. 13, 14). While F-1-P shows detectable TOM22 binding (K_D ~ 6-fold higher than fructose), neither glucose nor ribose exhibits specific interaction (Fig. 1i, k and Supplementary Fig. 6), highlighting remarkable ligand specificity. Notably, many cancers downregulate KHK-C expression,^7,50,51 indicating that unmetabolized fructose may drive oncogenesis through direct TOM22-mediated mitochondrial dysfunction. Thus, in this study, although we focused on renal-specific K105A-knock-in mice, our findings unify diverse fructose-associated metabolic disorder pathogenesis under a common mechanistic framework centered on TOM complex regulation,^6,7,49 indicating that TOM22 may represent a critical molecular target underlying the pathogenesis of high fructose-induced metabolic disorders. Future investigations using whole-body K105A-KI mice and other tissue-specific models (e.g., liver or intestine) will further clarify the broader pathophysiological relevance of the frustose-TOM complex axis.

In summary, this work fundamentally advances our understanding of: (1) a previously unrecognized nutrient-sensing mechanism mediated by mitochondrial protein import machinery; (2) fructose-induced structural and functional reprogramming of the TOM complex; (3) a direct, metabolism-independent role of fructose in metabolic disorders through its interaction with TOM22. By demonstrating that fructose impairs mitochondrial ribosome biogenesis through structural remodeling of the TOM complex, this work provides mechanistic insights into fructose-induced metabolic disorders. The identification of TOM22 K105 as a critical interaction site may open new avenues for therapeutic intervention in metabolic disorders, though further validation is needed to assess its clinical potential.

Materials and methods

Synthesis of P-fructose

3-Hydroxybenzoic acid (1.38 g), t-butyldimethylsilyl chloride (TBSCI, 4.52 g), and imidazole (2.08 g) were dissolved in 20 mL of DMF. The solution was stirred at 25 °C for 24 h, and poured into a separatory funnel containing Et₂O and H₂O. The organic phase was washed with H₂O, dried over MgSO₄, filtered, and evaporated. The crude product was subjected to flash silica gel column chromatography (DCM). The mixture obtained above was dissolved in 12 mL of AcOH, 15 mL of THF, and 3 mL of water. After stirring at 25 °C for 21 h, the solution was poured into a separatory funnel containing Et₂O and H₂O. The organic phase was isolated, and the aqueous phase was extracted with Et₂O. The organic layers were dried with MgSO₄. The product was subjected to flash silica gel chromatography (DCM), and further purified by recrystallization from MeCN to give compound 2 (1.84 g, 73% over two steps).

Compound 2 (1.26 g), diacetonefructose (1.3 g), and 4-(dimethylamino)-pyridine (DMAP, 0.61 g) were dissolved in 20 mL of DCM. To this solution, N, N’-dicyclohexylcarbodiimide (DCC, 1.3 g) was added dropwise at 0 °C and the solution was stirred at 25 °C for 12 h. The mixture was filtered to remove dicyclohexylurea, washed with Na₂CO₃, HCl (1 M), and NaCl solution, and dried for MgSO₄. The residue was purified by silica gel chromatography (PE/EA) to give compound 3 (742 mg, 30%).

Compound 3 (742 mg) was dissolved in 10 mL of THF, cooled to 0 °C. tetrabutylammonium fluoride solution (TBAF, 1.5 mL) was added dropwise and stirred at 0 °C for 15 min. The reaction mixture was poured into a mixture of EA and H₂O. The organic phase was washed with NaHCO₃ and dried using Na₂SO₄. The residue was purified by silica gel column chromatography (PE/EA) to give compound 4 (588 mg, 98%).

Compound 4 (380 mg) was dissolved in 5 mL of DMF, 3-(But-3-yn-1-yl)-3-(2-iodoethyl)-3H-diazirine (273 mg; Ark Pharm, Cat#AK544198) and K₂CO₃ (277 mg) were added, and the solution was stirred at 50 °C for 12 h. The reaction mixture was quenched with H₂O and subsequently transferred onto a mixture of EA and H₂O. The organic solution was dried with Na₂SO₄. The residue was subjected to silica gel column chromatography (PE/EA) to give compound 5 (210 mg, 42%). ¹H NMR (400 MHz, CDCl3; δ) 8.02 (d, J = 9.2 Hz, 2H), 6.89 (d, J = 9.2 Hz, 2H), 4.69-4.60 (m, 2H), 4.46 (d, J = 2.8 Hz, 1H), 4.35-4.22 (m, 2H), 3.97-3.93 (m, 1H), 3.88 (t, J = 6.4 Hz, 2H), 3.84–3.76 (m, 1H), 2.09–2.04 (m, 2H), 1.99 (t, J = 2.4 Hz, 1H), 1.92 (t, J = 6.4 Hz, 2H), 1.74 (t, J = 7.2 Hz, 2H), 1.54 (s, 3H), 1.47 (s, 3H), 1.37 (s, 3H), 1.35 (s, 3H); ¹³C NMR (101 MHz, CDCl3; δ) 165.6, 162.3, 131.8, 122.7, 114.1, 109.2, 108.8, 101.8, 82.6, 70.8, 70.5, 70.1, 69.3, 65.0, 62.7, 61.4, 32.8, 32.6, 29.7, 26.5, 25.9, 25.5, 24.0, 13.3.

Cell culture and fructose treatment

No commonly misidentified cell lines were utilized in these studies. Podocytes were cultured in RPMI culture medium (Invitrogen, Cat#11875093), containing 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, and 10 U/mL recombinant IFN-γ at 33 °C. Then, podocytes were differentiated in RPMI culture medium with 10% FBS and 1% penicillin-streptomycin at 37 °C. Liver cell line AML12 and gut cell line MODE-K were cultured in Dulbecco’s modified Eagle Medium with 10% FBS and 1% penicillin-streptomycin at 37 °C. When indicated, cells were exposed to 5 mM fructose (Sigma-Aldrich, Cat#F0127) for 0, 24, 48, or 72 h. The 5 mM concentration of fructose for cell treatment was selected by summarizing literature^52–55 and the experience of our group. Many studies found that fructose concentrations in the portal vein are ~1 to 2 mM following oral administration, which are 10-fold higher than fructose concentrations from peripheral arterial blood.^52–55

High-resolution mass spectrometry analysis of P-fructose hydrolysis in podocytes

The cultured podocytes were treated with P-fructose (5 mM) for 1.5 h at 37 °C/ 5% CO₂, then collected, centrifuged at 500 × g for 5 min at 4 °C, and washed twice with phosphate-buffered saline (PBS). These podocytes were suspended by 80% methanol solution and lysed by sonication. After centrifuging at 12,000 × g for 5 min at 4 °C, the supernatant was transferred to a centrifuge tube and speedvac dried. The sample was resuspended in acetonitrile. High-resolution mass spectrometry analysis was performed with Q Exactive HF-X Orbitrap mass spectrometer (Thermo Fisher Scientific).

Mitochondrial ROS detection

Mitochondrial ROS level was detected by MitoSOX red mitochondrial superoxide indicator (Yeasen, Cat#40778ES50). Podocytes or AML12 cells were collected, centrifuged at 500 × g for 5 min and washed twice with PBS. Then, cells were stained with 5 μM MitoSOX in the dark at 37 °C for 30 min. Cells were washed twice with PBS and analyzed by the Attune NxT flow cytometer. Mitochondrial ROS level was calculated as MitoSOX fluorescence intensity.

Mitochondrial membrane potential detection

Mitochondrial membrane potential was measured by tetraethylbenzimidazolylcarbocyanine iodide (JC-1) staining (Beyotime, Cat#C2006). Podocytes or AML12 cells were collected and centrifuged at 500 × g for 5 min. After washing twice with PBS, cells were stained with 10 μg/mL JC-1 in the dark at 37 °C for 20 min. Cells were washed twice with PBS and analyzed by Attune NxT flow cytometer (Invitrogen). Mitochondrial membrane potential was analyzed by the ratio of aggregates (indicative of normal membrane potential)/monomers (indicative of loss of membrane potential) fluorescence intensity.

Mouse strain generation and modeling

All animal experiments were approved by the Institutional Animal Care and Use Committee of Nanjing University (permit: IACUC-2305001). Mice had access to the food and water freely in the environment with the constant temperature (23 ± 2) °C under a normal 12-h light/dark cycle. For the generation of renal glomerulus-specific Tomm22 ^K105A KI mice, mice with Tom22-fructose binding site K105A mutation LSL (LoxP-stop-LoxP) mouse strain C57BL/6JGpt-Tomm22^{em1Cin(LSL-p.K105A)}/Gpt were crossed with renal glomerulus-specific Cre strain C57BL/6JGpt-H11^{em1Cin(Nphs1-iCre)}/Gpt to breed renal glomerulus-specific Tomm22^K105A knock-in mice (K105A) and wild-type (WT) siblings. Then WT or K105A mutant mice were fed with a standard or 60% fructose diet (Jiangsu Xietong Pharmaceutical Bio-engineering Co., Ltd., Cat#XT704) for 12 weeks. The concentration of fructose and modeling time were based on fructose-induced metabolic disorders-related studies.^29,31,32,56 The major dietary source of fructose was high-fructose corn syrup, containing 55% fructose in the Western diet.^8,57 At the 11th week, the mice were placed in a metabolic cage to collect urine for 24 h. After mice were anesthetized, the kidney cortex, liver, and intestine tissues of mice were cut into pieces and stored at -80 ^◦C for the assays. The glomeruli of mice were isolated by the graded sieving technique.

OCR analysis

OCR was measured using an XF96 Analyzer (Seahorse Bioscience, Inc.). For cultured cells, podocyte or AML12 cells were plated at 5 × 10⁴ cells/ mL in XF 96-well microplates (Seahorse Bioscience, Inc.). Before oxygen consumption analysis, podocytes were incubated in pre-warmed assay medium (XF base medium with 10 mM glucose, 1 mM sodium pyruvate, and 2 mM L-glutamine) for 1 h without CO₂. For isolated mitochondria from tissues, the mitochondria were extracted using Qproteome mitochondrial isolation kit (QIAGEN, Cat#37612) according to the manufacturer’s instructions. Isolated mitochondria were resuspended in mitochondrial assay solution, and added at 10 μg mitochondrial protein per well in XF 96-well microplates. After centrifuging at 2,000 × g for 20 min at 4 ^◦C, pre-warmed mitochondrial assay solution including 10 mM succinate and 2 mM malate was added to each well (180 μL final). Mitochondrial stress test was measured by injecting the ATP synthase inhibitor oligomycin (1 μM), the mitochondrial uncoupler carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP, 1 μM), and the electron transport inhibitor rotenone and antimycin A (0.5 μM) successively. Basal respiration, ATP production, maximal respiratory, and spare respiration capacity were normalized by protein amount.

Antibodies

Antibodies against Podocin (NPHS2, Cat#20384-1-AP), TOM40 (Cat#18409-1-AP), TOM70 (Cat#14528-1-AP), MRPS10 (Cat#16030-1-AP), MRPS23 (Cat#18345-1-AP), MRPL11 (Cat#15543-1-AP), MRPL14 (Cat#15040-1-AP), MRPL22 (Cat#16299-1-AP), MRPL37 (Cat#15190-1-AP), ND1 (Cat#19703-1-AP), ND2 (Cat#19704-1-AP), ND4 (Cat#26736-1-AP), ND5 (Cat#55410-1-AP), MTCO2 (Cat#55070-1-AP), MTCO3 (Cat#55082-1-AP), AldoB (Cat#18065-1-AP), MRPS21 (Cat#16359-1-AP) and MRPL46 (Cat# 16611-1-AP) proteins were purchased from Proteintech. Antibodies against TOM22 (Cat#ab179826), VDAC1 (Cat#ab14734), and cytochrome C oxidase subunit IV isoform 1 (COX4 | 1, Cat#ab16056) proteins were purchased from Abcam. Antibody against TOM20 (Cat#D8T4N) protein was purchased from Cell Signaling Technology. Antibody against the MRPL41 (Cat#orb318797) protein was purchased from Biorbyt. Antibody against the MRPL43 (Cat#PA5-103527) protein was purchased from Invitrogen. Antibodies against ND3 (Cat#DF8747), MTCYB (Cat#DF2372), and ATP6 (Cat#DF9237) proteins were purchased from Affinity. Antibodies against KHK-a (Cat#21708) and KHK-c (Cat#21709) proteins were purchased from Signalway Antibody. Antibody against β-actin (Cat#ABM-0001) protein was purchased from Zoonbio Biotechnology. For Western blot, all primary antibodies were used at 1:1000. For immunofluorescence, all primary antibodies were used at 1:200. HRP-conjugated Affinipure Goat anti-Rabbit IgG (Proteintech, Cat#SA00001-2) and HRP-conjugated Affinipure Goat anti-Mouse IgG (Proteintech, Cat#SA00001-1) were used at 1:5000, whereas Goat anti-Rabbit IgG Cross-Absorbed Secondary Antibody, Alexa Fluor 488 (Invitrogen, Cat#A-11008) was used at 1:500.

Protein extraction and Western blot

Total proteins from cultured cells or animal tissues were extracted in RIPA buffer supplemented with PMSF (0.1 mM) for 10 min on ice and then centrifuged at 10,000 × g for 10 min. Protein concentration was measured using the BCA protein assay kit (Thermo Fisher Scientific, Cat#23225). Protein samples were subjected to electrophoretic separation on SDS-PAGE and transferred to the PVDF membrane. The membrane was blocked with 5% milk in Tris-buffered saline with 0.1% Tween-20 for 2 h, then incubated with primary antibodies overnight and secondary antibodies for 1 h at RT. Antigens were detected using an enhanced chemiluminescence detection kit (Tanon, Cat#180-5001). Band intensities were quantified using ImageJ (National Institutes of Health).

Immunofluorescence

Podocytes on coverslips were fixed with 4% paraformaldehyde for 30 min at RT. Podocytes were washed twice with PBS and blocked with 10% BSA and 0.2% Triton X-100 in PBS for 1 h. Then, podocytes were incubated with primary antibodies at 4 °C overnight. The next day, podocytes were washed three times in PBS and incubated with fluorescently conjugated secondary antibodies for 1 h at RT. Next, podocytes were washed and subjected to 0.5 μg/mL Hoechst for 10 min. After washing with PBS, the slides were sealed and imaged using a confocal microscope (Leica).

HPLC analysis of P-fructose content in cultured podocytes

For the release experiment, podocytes were incubated with P-fructose (5 mM), followed by culture in P-fructose-free medium for 0, 1, and 2 h. Podocytes were then centrifuged at 500 × g for 5 min at 4 °C, and washed twice with ice-cold PBS. The cell pellet was resuspended in 80% methanol and lysed via sonication. After centrifugation at 12,000 × g for 5 min at 4 °C, the supernatant was transferred to a new tube and dried in a speedvac concentrator, followed by reconstitution in methanol for P-fructose analysis using an HPLC instrument (Agilent Technologies 1200 series, USA). To evaluate cellular uptake and metabolism, podocytes were treated with P-fructose (5 mM) for 0, 1, 2, and 3 h. Then, the culture supernatant and cell pellets were collected separately. Cell pellets were processed using the same method as described above, while supernatant was first dried in a speedvac concentrator, reconstituted in 80% methanol, and incubated on ice for 5 min. The methanol concentration was then adjusted to 53% by dilution, followed by centrifugation at 15,000 × g for 5 min at 4 °C. The resulting supernatant was dried, reconstituted in methanol, and analyzed for P-fructose concentration. Chromatographic separation was performed on a C18 column (100 mm × 4.6 mm, 5 μm), maintained at 35 °C. The mobile phase consisted of (A) water with 0.1% trifluoroacetic acid (v/v) and (B) acetonitrile. A gradient elution was applied at a flow rate of 0.6 mL/min as follows: 80% B (0-5 min) and 100% B (15 min). The injection volume was 5 μL, and the detection wavelength was 257 nm.

Compound-centric chemical proteomics

Podocytes were labeled with P-fructose (5 mM), treated with fructose (10 mM) or DMSO for 1.5 h at 37 °C/ 5% CO₂, and then put on ice under UV radiation at 365 nm for 20 min. In the control group, podocytes were labeled with P-fructose and treated with fructose but without UV radiation. In the P-fructose group, podocytes were labeled with P-fructose and treated with DMSO with UV radiation. In the competition group, podocytes were labeled with P-fructose and treated with fructose with UV radiation. Podocytes were collected and centrifuged at 500 × g for 5 min. Podocytes were then resuspended in PBS, homogenized by sonication, and diluted to 2 mg/mL with PBS. The cell homogenate was combined with Azide-PEG3-biotin (100 μM; Jena Bioscience, Cat#CLK-AZ104P4), tris(2-carboxyethyl)phosphine hydrochloride (TCEP, 1 mM; Promega, Cat#VB1000), tris((1-benzyl-1H-1,2,3-triazol-4-yl)methyl)amine (TBTA, 100 μM; TargetMol, Cat#T7086) and CuSO₄·5H₂O (1 mM; Sigma-Aldrich, Cat#12849) at RT for 2 h with vortex. After the reaction, the proteomes were extracted with chloroform-methanol to remove redundant reagents. The protein interphase was washed with methanol, solubilized with 2% SDS/ PBS, and diluted 5x with PBS. The solubilized proteins were incubated with streptavidin beads (100 μL/mg protein; New England Biolabs, Cat#S1420S) overnight at 4 °C with rotation. The beads were then washed with 1% SDS three times and PBS five times. Then the beads were transferred to a screw-top Eppendorf tube with PBS and centrifuged at 1,400 × g for 3 min. The supernatant was removed. Then the beads were boiled in elution buffer (200 mM Tris, 50 mM dithiothreitol (DTT), 8% SDS, pH 6.8) for 15 min. The pull-down proteins were separated in 12.5% gel by SDS-PAGE, visualized by silver staining, and followed by in-gel trypsin digestion. After digestion, the digested peptides were then extracted from the gel with 50% acetonitrile and 5% formic acid. After desalting and vacuum drying, the samples were resuspended in 3% acetonitrile (V/V) and 2% formic acid (V/V) for LC-MS/MS analysis. The original MS/MS file data were submitted to ProteinPilot software (AB SCIEX) for data analysis, and ProteinProspector v6.4.9 was used for search compare and protein quantification. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium⁵⁸ via the PRIDE⁵⁹ partner repository with the project accession PXD047310. Reviewer account details: username: reviewer_pxd047310@ebi.ac.uk; password: LBs4rRW4.

Microscale thermophoresis analysis

Recombinant human TOM22-eGFP, mutant TOM22 (E102A, K105A, and E109A)-eGFP, and eGFP (negative control) were expressed and purified (Genscript). Recombinant human TOM complex and KHK (OriGene Technologies, Cat#TP323488) proteins were labeled by amine-reactive probe (Invitrogen, Cat#A37573) and purified by Antibody conjugate purification kit (Invitrogen, Cat#A33087) according to the manufacturer’s instructions. For the competition/affinity assay, the recombinant TOM22 and KHK proteins at the same concentration were first mixed. Then fructose and fructose-1-phosphate (F-1-P; New Research Biosciences, Cat#Y-THX-3860) were titrated in different concentrations to the purified recombinant proteins. The reaction was performed in PBS with 0.05% Tween-20. After incubating at RT for 5 min, the mixed samples were transferred in glass capillaries and analyzed by the Monolith NT.115 instrument (NanoTemper Technologies). The Hill curves and K_D were generated and fit using MO. Affinity Analysis software (NanoTemper Technologies).

Cellular imaging

Podocytes were labeled with P-fructose (5 mM) with or without competition of fructose (10 mM). After washing with PBS, podocytes were put on ice under UV radiation at 365 nm for 20 min. Then podocytes were added with NVOC2-Q-rhodamine-5-PEG3-azide (100 μM; Sigma-Aldrich, Cat#768693), TCEP (1 mM), TBTA (100 μM), and CuSO₄·5H₂O (1 mM) at RT. After 2 h, podocytes were washed twice with PBS. The image of rhodamine fluorescence was acquired using a confocal microscope (Leica).

Quantitative real-time PCR

Total RNA was isolated from cultured podocytes, AML12 cells, and MODE-K cells with TRIzol. First-strand cDNA was synthesized using HiScript® II Q RT SuperMix. qRT-PCR was performed with ChamQ^™ SYBR® qPCR Master Mix in a CFX Connect Real-Time PCR Detection System (Bio-Rad).

The primers used in this study were as follows:

Homo sapiens Slc2a2 forward primer: GCTGCTCAACTAATCACCATGC;

Homo sapiens Slc2a2 reverse primer: TGGTCCCAATTTTGAAAACCCC;

Homo sapiens Slc2a5 forward primer: GAGGCTGACGCTTGTGCTT;

Homo sapiens Slc2a5 reverse primer: CCACGTTGTACCCATACTGGA;

Homo sapiens KHK-a forward primer: GCTATTCTGTGGACCTACGCT;

Homo sapiens KHK-a reverse primer: CAATGTGGATCCACTTGAACTG;

Homo sapiens KHK-c forward primer: CTCCTGCTGCATCATCAAC;

Homo sapiens KHK-c reverse primer: CAATGTGGATCCACTTGAACTG;

Homo sapiens AldoB forward primer: GGCAGTTCCGAGAAATCCTCT;

Homo sapiens AldoB reverse primer: CTCCTTGGTCTAACTTGATTCCC;

Homo sapiens β-ACTIN forward primer: CATGTACGTTGCTATCCAGGC;

Homo sapiens β-ACTIN reverse primer: CTCCTTAATGTCACGCACGAT;

Mus musculus Slc2a2 forward primer: TCAGAAGACAAGATCACCGGA;

Mus musculus Slc2a2 reverse primer: GCTGGTGTGACTGTAAGTGGG;

Mus musculus Slc2a5 forward primer: TTCCAATATGGGTACAACGTAGC;

Mus musculus Slc2a5 reverse primer: GCGTCAAGGTGAAGGACTCAA;

Mus musculus KHK-a forward primer: CCAACATTCTGTGGACTTACG;

Mus musculus KHK-a reverse primer: CCTTCTCAAAGTCCTTAGCAG;

Mus musculus KHK-c forward primer: GCTGACTTCAGGCAGAGG;

Mus musculus KHK-c reverse primer: CCTTCTCAAAGTCCTTAGCAG;

Mus musculus AldoB forward primer: GAAACCGCCTGCAAAGGATAA;

Mus musculus AldoB reverse primer: GAGGGTCTCGTGGAAAAGGAT;

Mus musculus β-Actin forward primer: GTGACGTTGACATCCGTAAAGA;

Mus musculus β-Actin reverse primer: GCCGGACTCATCGTACTCC.

snRNA-seq reanalysis

We reanalyzed snRNA-seq data of renal tissue from the Gene Expression Omnibus (GSE183277), focusing on the expression of key fructose-metabolizing enzymes and transporters in healthy volunteer groups. Using the Seurat (v5.0) R package, we processed 21 healthy control libraries (including KC, KM, PA series) with strict quality control (nFeature_RNA: 500-6,000; mitochondrial transcripts <5%). After library-layered normalization (LogNormalize), variable feature selection, and canonical correlation analysis integration for batch correction, we performed a shared nearest neighbor graph-based algorithm (Louvain; resolution = 0.5) and UMAP visualization, annotating cell types using the original ‘subclass.l1’ metadata. We then characterized the expression of fructose metabolism and transport-related genes (KHK, AldoB, Slc2a2, and Slc2a5) across kidney cell types and computed pseudobulk averages to define the healthy kidney fructose metabolic signature.

High-resolution mass spectrometry analysis of P-fructose-labeled TOM22 peptides

10 μg of TOM22 protein was incubated either with P-fructose (20 μM, 1% DMSO) or DMSO. After 30 min incubation in the dark at 25 °C, the mixture samples were UV cross-linked on ice for 20 minutes. Samples were diluted with 1% SDS to obtain a volume of 20 μL. Samples were added DTT (10 mM) and vortexed. After heating for 15 min at 95 °C, samples were alkylated with iodoacetamide (25 mM) in the dark for 30 min at RT. Then samples were reduced with DTT (25 mM) for 15 min at RT. Samples were added with 100 μg carboxylate-modified magnetic beads (Sera-Mag-to-SpeedBeads volume ratio 1:1) and diluted to a final volume of 50 μL. Next, beads were added to 50 μL of ethanol and slowly mixed for 10 min at RT. After collecting in a magnetic rack, the beads were added to 180 μL 80% ethanol, which was repeated three times. Then beads were resuspended with 50 μL NH₄HCO₃ (25 mM) and trypsin (trypsin-to-protein mass ratio 1:50) for 16 h-digestion at 37 °C. Beads were added 1 mL acetonitrile and mixed for 10 min. After washing with 1 mL acetonitrile three times, 100 μL 2% DMSO was added to the beads. After incubation for 30 min at 37 °C, beads were centrifuged at 20,000 × g for 1 min. The supernatant was collected, added 200 μL 80% acetonitrile and centrifuged at 2000 × g for 10 min. Then the supernatant was transferred to a centrifuge tube and speedvac dried. The samples were resuspended in 0.1% formic acid (V/V). High-resolution mass spectrometry analysis was performed on an Easy-nLC 1200 system coupled with a Q Exactive HF-X Orbitrap mass spectrometer (Thermo Fisher Scientific). Under the positive-ion mode, full-scan mass spectra are acquired over the range 350-1,800 m/z. The 20 most intense precursors with charge state 2⁺ were selected for fragmentation, and MS/MS spectra were collected for 50 ms. Precursor ions were excluded from reselection for 20 s. LC-MS/MS data were analyzed by ProLuCID with static modification of carboxyamidomethylation of cysteine (+57.0215 Da), variable oxidation of methionine (+15.9949 Da), and variable P-Fructose modification of all amino acids (+472.2200 Da). Peptides were searched with a static modification for acetylation of the protein N-terminus ( +28.0313 Da). The searching results were filtered by DTASelect to achieve a defined peptide false positive rate below 1%. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the project accession PXD047275. Reviewer account details: username: reviewer_pxd047275@ebi.ac.uk; password: Drtpl2UN.

Transient co-expression of the human TOM complex and protein purification

The optimized coding DNAs for human TOMM40 (Uniprot: O96008), TOMM22 (Uniprot: Q9NS69), TOMM7 (Uniprot: Q9P0U1), TOMM6 (Uniprot: Q96B49), and TOMM5 (Uniprot: Q8N4H5) were cloned into the pcDNA3.1(-) vector, with tandem twin Strep-tag at the C terminus of TOMM22. The HEK293F cells (Invitrogen) were cultured in SMM 293T-I medium (Sino Biological Inc., Cat#M293TI) supplemented with 8% CO₂ in a ZCZY-CS8 shaker (120 rpm, Zhichu) at 37 °C. When density reached 2 × 10⁶ cells per mL, the cells were transfected with the plasmids encoding TOMM70, TOMM40, TOMM22, TOMM20, TOMM7, TOMM6, and TOMM5, incubated with Polyethylenimine (PEI) for 15 min at a mass ratio of 1:3. Transfected cells were cultured for 48 h before harvesting.

All procedures are carried out at 4 °C. Four liters of transfected cells were harvested, washed with 1 × PBS, and resuspended in 25 mM Tris pH 7.8, 150 mM NaCl, and 1 mM PMSF. The suspension was lysed by sonication for 5 min, then was further centrifuged at 20,000 × g for 45 min to obtain the membrane. The pellet was suspended and extracted in 25 mM Tris, pH 7.8, 150 mM NaCl, with 1% GND. After incubation for 2 h, the extraction was centrifuged at 20,000 × g for 25 min at 4 °C, and the supernatant was applied to Strep-Tactin Sepharose (IBA Lifesciences) by gravity at 4 °C. The resin was washed three times with the W buffer, which contains 25 mM Tris (pH 7.8), 150 mM NaCl, with 0.1% GDN. The target proteins were eluted with W buffer plus 5 mM desthiobiotin (Sigma-Aldrich, Cat#D1411), concentrated to 100 μL by 100 kDa cut-off centrifugal filter (Millipore), and further purified by Superose6 5/150 GL (GE Healthcare) also in the W buffer. The peak fractions were collected for EM sample preparation.

Cryo-EM sample preparation and data acquisition

4 μL aliquots of TOM complex in apo state/TOM complex with ligand (incubated with fructose solution at a molar ratio of 1:50) protein at a concentration of 4.5 mg/mL were respectively applied to discharged 300-mesh Quantifoil R1.2/1.3 grids (Quantifoil). Grids were blotted for 4.5 s and plunged into liquid ethane using an FEI Mark IV Vitrobot operated at 8 °C and 100% humidity.

Cryo-EM data were collected on a 300 kV Titan Krios G3 equipped with a Gatan K3 detector and a GIF Quantum energy filter (slit width of 20 eV). The defocus values ranged from −1.3 µm to −1.8 µm. Each stack of 32 frames was exposed for 2.56 s, and the exposure time of each frame was 0.08 s. The micrographs were automatically collected with the AutoEMation program⁶⁰ in super-resolution counting mode with a binned pixel size of 0.8374 Å. The total dose of each stack was about 50 e^- Å^-2. The alignment and summation of all 32 frames in each stack were performed using the whole-image motion correction program MotionCor2.⁶¹

Cryo-EM data processing

Our cryo-EM data processing workflow is shown in Extended Data Fig. 6. All cryo-EM data were processed using cryoSPARC.⁶² Contrast transfer function (CTF) parameters were estimated using patch-CTF. Blob picker was used for initial templates generation via 2D classification. Template picker was then used for all particle-picking tasks. For the TOM complex in the apo state, 4,402,241 particles were extracted from 4,981 micrographs. Subsequent two-dimensional (2D) classification, three-dimensional (3D) classification, multi-class ab-initio reconstruction, and nonuniform refinement of the best class were performed, resulting in a resolution of 2.89 Å resolution. Following the same workflow, 3,706,692 particles were extracted from 5300 micrographs. Subsequently, 146,897 particles were used to resolve the structure of the TOM complex with ligand at 2.93 Å resolution. All structures were determined with C1 symmetry. Resolution is reported using the gold-standard Fourier shell correlation with a 0.143 cutoff. See Supplementary Table 2 for collection, refinement, and validation statistics.

Cryo-EM model building and refinement

The initial models were derived using AlphaFold.⁶³ We assigned predicted models to our density maps using UCSF ChimeraX.⁶⁴ The models were further optimized by Coot.⁶⁵ The final models of all datasets were refined against the corresponding maps using PHENIX in real space with secondary structure and geometry restraints.⁶⁶ The structures were validated through examination of the Clash scores, Molprobity scores, and statistics of the Ramachandran plots by PHENIX (Supplementary Table 2). All the figures were created in PyMOL⁶⁷ and UCSF ChimeraX.

DIA-based quantitative proteomic analysis

Mitochondria from human podocytes were extracted using the Qproteome mitochondrial isolation kit according to the manufacturer’s instructions. Mitochondria pellets were lysed in RIPA lysis buffer (50 mM Tris-HCl, 1% NP-40, 0.25% Na-deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 mg/mL each of Aprotinin, leupeptin and pepstatin, 1 mM Na₃VO₄, 1 mM NaF, pH 7.4), ultrasonicated for 10 min at 4 °C using Bioruptor Plus (Diagenode), and centrifuged at 13,000 × g for 15 min to remove debris. Protein concentration in each sample was determined using a Pierce™ BCA Protein Assay Kit according to the manufacturer’s protocol, and the total protein concentrations were adjusted to be equal across samples. The mitochondrial protein was washed by 8 M urea for 3 times and 0.5 M TEAB for 6 times on a 3 kDa filter, then supplemented with trypsin for 4 h pre-digestion (enzyme-to-substrate mass ratio 1:50; Promega, Cat#V5280) at 37 °C, following an additional 8 h-digestion with trypsin/Lys-C (enzyme-to-substrate mass ratio 1:100; Promega, Cat#V5073) at 37 °C. After desalting and drying, the digested peptides were resuspended in 3% acetonitrile (V/V) and 2% formic acid (V/V) for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. The samples were analyzed on a ZenoTOF 7600 mass spectrometer (AB SCIEX). The DIA data analysis was conducted using DIA-NN⁶⁸ for quantitative proteomic analysis. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the project accession PXD047317. Reviewer account details: username: reviewer_pxd047317@ebi.ac.uk; password: A1BrkAYw. Raw data were processed using R, R Studio, and other requisite software.

Overexpression of wild-type and mutant TOM22

Short interfering RNAs (siRNA) targeting 3’ untranslated region of Homo sapiens TOMM22 (sense: 5’-GGGUCUAUGUACUUCUUUATT-3’, antisense: 5’-UAAAGAAGUACAUAGACCCTT-3’), Mus musculus Tomm22 (sense: 5’- GGCUCUUAGACCUGCUUAUTT-3’, antisense: 5’- AUAAGCAGGUCUAAGAGCCTT-3’), and negative control (sense: 5’-UUCUCCGAACGUGUCACGUTT-3’, antisense: 5’-ACGUGACACGUUCGGAGAATT-3’) were purchased from GenePharma. pcDNA3.1-EGFP-TOM22 (TOMM22 Human), pcDNA3.1-EGFP-TOM22^E102A, pcDNA3.1-EGFP-TOM22^K105A and pcDNA3.1-EGFP-TOM22^E109A plasmids for overexpression of TOM22 wild type and mutants and were generated by Abiocenter. pFastBac-His-EGFP-TOM22, pFastBac-His-EGFP-TOM22^E102A, pFastBac-His-EGFP-TOM22^K105A, and pFastBac-His-EGFP-TOM22^E109A plasmids for protein expression and purification were generated by Genscript. Podocytes or AML12 cells were transiently transfected with the siRNA pool using Lipofectamine 2000 (Invitrogen, Cat#11668027) for 24 h. Then, the endogenous TOM22 knockdown podocytes or AML12 cells were transiently transfected with wild-type TOM22 or mutant TOM22 (E102A, K105A, and E109A) plasmids using Lipofectamine 2000 and then cultured with fructose for 48–72 hours. Wild-type and mutant TOM22-containing mitochondria were then isolated from the above TOM22 and its mutants-transfected cells.

In vitro mitochondrial protein import assay

Preparation of MRP subunits

The MRPL37 protein with an N-terminal mitochondrial targeting sequence (MTS) and a C-terminal Flag affinity tag was expressed and purified using the pPIC9K eukaryotic expression system. Briefly, recombinant MRPL37 protein was expressed and secreted into the fermentation culture supernatant of Pichia pastoris, and was then purified using DEAE-fiber column chromatography system.

Import of mitoribosomal proteins into isolated mitochondria

MRP proteins (1 mg/mL) were mixed with DFO (10 mg/mL) with a molar ratio of 1:20, adjusted to pH 8-9, following shaking at room temperature for 2 h. After the reaction, PD-10 desalting columns were used for purification; the eluent used was 0.25 M sodium acetate buffer at pH 5.4-5.6, and protein-DFO conjugate was obtained.

To prepare the Zirconium-89 solution, 500 μCi ⁸⁹Zr oxalate solution was taken, and 1 M oxalic acid was used to supplement the solution to a volume of 200 μL. Subsequently, 90 μL of 2 M sodium carbonate was added to the solution, and the pH was adjusted to 7.0. Following this addition and adjustment, the solution was allowed to stand at room temperature for 3 minutes. After adding 200 μL of 0.5 M HEPES buffer and 100 μg of protein, the reaction system was supplemented with 0.5 M HEPES until the total volume of the system reached 1 mL. After shaking and reaction for 1 h at room temperature, PD-10 columns were used for purification after the reaction, and the eluent was 0.25 M sodium acetate buffer (pH 5.4-5.6) to obtain the ⁸⁹Zr-labeled protein product.

The mitochondria of each group were incubated twice with 6 μCi ⁸⁹Zr-protein complex per serving and 5 mM fructose at a final concentration, with a total system of 6 mL per serving. After incubation at 37 °C, 1 mL of mitochondrial solution was taken at each time point and centrifuged at 16,000 g for 15 min. Then it was re-suspended by PBS, centrifuged, and re-suspended three times. The precipitate was re-suspended in PBS. PET/CT scan and gamma-counter counting were performed. The gamma-counting results were adjusted by western blot and grayscale analysis of mitochondrial marker protein COX4 | 1.

HPG-click chemistry assay

Protein translation was measured using BeyoClick™ HPG-488 Nascent Protein Assay Kit (Beyotime, Cat#P1202) according to the manufacturer’s instructions. Podocytes were incubated with cycloheximide (100 μg/ml) for 12 h and then incubated with methionine-free medium for 30 min. Cells were labeled with HPG at 37 °C for 30 min, followed by fixation, washing, and permeabilization. Then, the cells were incubated with the click reaction mixture, including Azide 488 for 30 minutes in the dark. Cells were stained by MitoTracker® Deep Red FM (Invitrogen, Cat#M22426) at 37 °C for 20 min, and nuclei were stained with Hoechst for 10 min. Azide 488, MitoTracker, and DAPI have maximum excitation/emission wavelengths of 495/519 nm, 644/665 nm, and 346/460 nm, respectively. The colocalization of newly synthesized proteins and mitochondria was visualized by a confocal microscope.

Mitochondrial DNA copy number

Genomic DNA was extracted from podocytes using a cell/tissue DNA isolation kit (Vazyme, Cat#DC102-01). mtDNA copy number was measured using the quantitative real-time PCR. The primers used in this study were as follows: h-mtDNA F: CCCTAACACCAGCCTAACCA; h-mtDNA R: AAAGTGCATACCGCCAAAAG; h-HBB F: CTATGGGACGCTTGATGT; h-HBB R: GCAATCATTCGTCTGTTT. The mtDNA copy number was quantified relative to the nuclear HBB gene.

Mitochondrial morphology

A transmission electron microscope (TEM) was used to observe the mitochondrial morphology of podocytes. Podocytes were stained by MitoTracker® Deep Red FM to detect the changes of mitochondrial morphology. Mitochondria were divided into three categories to quantify morphological changes. Healthy podocytes containing long tubule mitochondria were defined as category I. Podocytes having large round, dotted mitochondria distributed throughout the cytosol were assigned to category II. Podocytes with totally fragmented mitochondria close to the nucleus represented category III. Images were acquired using a confocal laser scanning microscope (Carl Zeiss).

Renal function assessment

The urinary levels of albumin (Jiancheng Bioengineering Institute, Cat#C035-2-1) and creatinine (Jiancheng Bioengineering Institute, Cat#C011-2-1), and the serum levels of creatinine and urea nitrogen (Jiancheng Bioengineering Institute, Cat#C013-2-1) in mice were measured by Coomassie brilliant blue, sarcosine oxidase, and urease assays, respectively, according to the manufacturers.

Renal morphology assessment

The renal tissues were fixed in 4% formaldehyde solution and embedded in paraffin. The tissues were then sliced into 4 μm-thick sections for Periodic acid-Schiff (PAS) staining to observe glomerular morphology. Then, the glomerular podocytes of mice were observed by TEM. In brief, fresh renal tissues (1 mm³) were placed in a prepared electron microscope fixed solution at 4 °C overnight. After tissues were sliced into ultrathin sections, the specimens were stained with uranyl acetate and lead citrate before being examined by TEM.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 with statistical tests as indicated in the figure legends. Data were presented as mean ± SD from at least 3 biologically independent replicates unless otherwise described. Statistical significance (P values) was calculated by One-Way ANOVA with Dunnett’s post hoc test. The difference was considered significant at P < 0.05.

Author contributions

Conceptualization, L.D.K. and L.F.; Methodology, L.F., X.C., M.J.Y., W.Y.W., W.H.W., J.Z.Z., Q.D.L, N.N.W., L.C., H.H., and J.H.L.; Software: W.Y.W., W.H.W., J.Z.Z., Q.D.L, N.N.W., and J.H.L.; Formal Analysis: W.Y.W., W.H.W., J.Z.Z., Q.D.L, N.N.W., C.Z.W. and J.H.L.; Investigation, W.Y.W., W.H.W., J.Z.Z., Q.D.L, N.N.W., C.Z.W., Y.Y.C., J.H.L. and X.C.; Writing – Original Draft, L.F., W.Y.W. and W.H.W.; Writing – Review & Editing, L.D.K., L.F., M.J.Y., X.C., W.Y.W., W.H.W. and J.Z.Z.; Funding Acquisition, L.D.K., L.F. and M.J.Y.; Supervision, L.D.K., L.F. and M.J.Y. All authors have read and approved the final manuscript.

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the project accession PXD047275, PXD047310, and PXD047317. The DIA-based quantitative proteomics data for four key conditions (WT/K105A ± fructose) have been deposited to the ProteomeXchange Consortium via the iProX partner repository^69,70 with the dataset identifier PXD079442, and are also accessible in the iProX database under accession IPX0017657000. Human TOM complex bound to fructose has been deposited in the PDB with accession number 8XDN, and in EMDB with code EMD-38282. All data are available in the main text or the supplementary materials.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41392-026-02896-x.