Fig 2: The #4-helix within the C-segment of LBP interacts with oxidized TGs.a LBP structure solved in the PDB database and predicted by Alphafold, with the black box indicating the differential #4α-helix. b Predict secondary structures and perform homology comparisons of LBP in various mammals. c Molecular docking diagram of WT-Tridocosahexaenoin with LBP. LBP is represented as a dark blue cartoon model, while the ligand is depicted as a cyan stick model. The binding site is shown as a magenta stick structure. (yellow: Hydrogen bonds, magenta: ionic interactions, and green dashed lines: hydrophobic interactions). d Molecular docking diagram of oxidized-Tridocosahexaenoin with LBP. e Representative co-stained images captured by confocal microscope showing the expression of C-segment LBP-GFP (LBPSP+216-481-GFP, green) for 48 h, treated with Bodipy C12 (orange) or Bodipy C12+PC for 16 h. Experiments repeated three times independently with similar results. f Representative co-stained images captured by confocal microscope showing the expression of H294A mutation (green) and H294G mutation (green) for 48 h, treated with Bodipy C12 (orange) for 16 h. Experiments repeated five times independently with similar results. g Comparison of Alphafold predicted LBPWT(green) and LBPF436L(blue) protein structures. The F436L site is indicated in red (upper panel). h Representative confocal images of expression LBPF436L-GFP (green) for 48 h and co-stained after adding Bodipy C12 (orange) for 16 h. Experiments repeated five times independently with similar results. i Hepatic TG content of LBP−/− mice with overexpressing LBPWT-GFP or LBPF436L-GFP. Shown are means ± s.d., two-tailed unpaired t-test (n = 3, biologically independent). j Working model for LBP and PRDX4 interactions: 1. LBP expression upregulated response to oxidative stress; 2. LBP translocation to LD for promoting LD growth; 3. PRDX4 resolution promoting the export of LBP from the LDs; 4. Export of LBP enhancing lipolysis and phospholipid synthesis. Illustration by Mo Liu, with permission.

Fig 3: The localization of LBP to LDs drives hepatic steatosis.a Hepatic TG content of WT mice under NT, 24 h fasting, or LPS-treatment (5 mg/kg, i.p.) for 24 h. Shown are means ± s.d., one-way ANOVA (n = 3, biologically independent). b Representative bright-field microscope image showing separated LDs from the same treatment as described in (a), scale bar = 10 μm. Experiments was conducted with three biological replicates and produced similar results. c TEM and Oil red O staining of LDs in the livers of WT and LBPKI/KI mice with 16-week HFD. TEM scale bar = 2 μm, Yellow arrows indicate LDs. Oil red scale bar = 20 μm. Experiments was conducted with three biological replicates and produced similar results. d Primary hepatocytes isolated from WT, LBP−/−, LBPKI/KI mice treated with Bodipy C12 (orange) for 16 h and co-stained with PLIN2 (violet) and LBP (blue). Scale bar = 2 μm. Experiments repeated two times independently with similar results. e Hepatic TG content of LBP−/− mice overexpressing GFP/LBP-GFP. Shown are means ± s.d., two-tailed unpaired t-test (n = 6, biologically independent). f Representative confocal imaging of LDs separation from liver in (e). Scale bar = 10 μm. g Representative TEM images of control-APEX2 and LBP-APEX2 morphology, scale bar = 500 nm. Experiments repeated two times independently with similar results.

Fig 4: LBP binds with unsaturated TG to prevent against lipid peroxidation.a A pie chart showing the difference in lipid composition between WT (n = 3) and LBPKI/KI (n = 3) mice liver after 16 weeks of HFD. b Analysis of chain length of differential TGs in (a). Shown are means ± s.e.m., (n = 3, biologically independent). c Analysis of saturation of differential TGs in (a). Shown are means ± s.e.m., (n = 3, biologically independent). d Targeted quantitative lipidomics illustrating the ratio of fatty acid content in the liver tissue of LBPKI/KI and LBP−/− groups compared to the WT group. e Confocal microscopy images show co-stained lipid droplets (Bodipy 493/503, green) formed by PA and LBP (violet), with DAPI (blue) counterstaining. Line graphs represent relative intensity quantification of the area indicated by the red arrow. Scale bars: 5 µm for the main image, 1 µm for the magnified area. Experiments were independently repeated three times with consistent results. f Kinetic line graph illustrating the results of Seahorse analyses, indicating that overexpression of LBP leads to an increase in FAO rate.Shown are means ± s.d. (n = 4, biologically independent). g Analysis of basal, maximal OCR, and ATP-linked respiration data from the experiment in (f).Shown are means ± s.d., 2way ANOVA (n = 4, biologically independent). h A pie chart illustrates the changes in lipid composition bound to LBP before and after 3 h 500 uM H2O2 stimulation. i Analysis of chain length of differential TGs in (h). Shown are means ± s.e.m., (n = 3, biologically independent). j Analysis of saturation of differential TGs in (h). Shown are means ± s.e.m., (n = 3, biologically independent). k LBP lipid overlay assay using purified LBP with non-treatment or 1 h 1 mM H2O2 treatment lipid-bound membranes. l Dose-response curves for determination of Kd for the binding of LBP to tridocosahexaenoin using a labeled MST assay. Shown are means ± s.d., (n = 3, biologically independent).

Fig 5: Redox signaling regulates LBP-induced LD accumulation via phospholipid biogenesis and the LBP chaperone function of PRDX4.a The bar graph compares the hepatic TG levels in WT mice and LBPKI/KI mice, with/without NAC treatment. Means ± s.d., 2way ANOVA (n = 5–6, biologically independent). b Bright-field microscope image of LDs in LBPKI/KI mice, fasted for 24 h and treated with/without NAC treatment for 3 h. Scale bar = 10 µm Experiments repeated three times independently with similar results. c A pie chart representing the difference in lipid composition between WT mice that were fasted for 24 h (n = 3) and those treated with NAC for 3 h (n = 3). d MSEA analysis for the lipidomic data in (c). e Changes in LBP content within LDs extracted from liver tissue with/without NAC treatment for 3 h in 24 h fasting mice (n = 3, biologically independent). f Confocal microscopy images (upper panel) and quantification (lower panel) of LBP (blue) co-stained with PRDX4 (violet) at different ratios of Bodipy C12 (orange) and FL-HPC (green). Scale bar = 5 µm, magnified area scale bar = 1 µm. Experiments repeated three times independently with similar results. g Staining of LBP (green) and PRDX4 (violet) co-localized with Bodipy C12 (orange) in HepG2 cells after 16 h of Bodipy C12 treatment and 3 h of starvation. Scale bar = 5 µm. Experiments repeated two times independently with similar results. h Confocal images show that knockdown of PRDX4 does not affect the localization of LBP in LD, but increases LD size, whereas knockdown of LBP affects PRDX4 localization in LD. The shRNA scaffold plasmid with dual promoters for expressing green fluorescent protein; red represents LBP; violet denotes PRDX4, while Bodipy C12 is orange. Scale bar = 5 µm. Experiments repeated three times independently with similar results. i Effects of PRDX4-knockdown on LBP-mediated regulation of TG accumulation. Means ± s.d., one-way ANOVA (n = 3, biologically independent). j Effects of PRDX4-knockdown on LBP-mediated regulation of lipolysis. Means ± s.d., one-way ANOVA (n = 3, biologically independent). k Representative confocal image shows that shPRDX4 (green) reduces the ability of LBP (violet) to leave the LD (Bodipy C12, orange) after 3 h of PC treatment. Scale bar = 5 µm, magnified area scale bar =1 µm. Experiments repeated three times independently with similar results.