Fig 2: Confirmation of mRNA gene expression changes in the regression experiment using the porcine model of dietary NAFLD development. Analyses by RT-qPCR of previously observed differentially expressed genes (A). Data (mean ± SD) represent arbitrary units normalized to UBA52 expression. Statistical analyses were carried out by Mann–Whitney U test. *p < 0.05, **p < 0.01 and ***p < 0.005. Venn diagram analysis showing the transcripts displaying increased expression in the steatotic condition (yellow area) of progression and regression experiments (B). Significant (P < 0.001) association between hepatic lipid droplet content and LGALS3 expression (C), hepatic triglycerides and gene expressions (D) and hepatic inflammation (CD68 immunostaining) and gene expressions (E). Red colour denotes a positive association while blue a negative one. Correlations were calculated according to the Spearman’s rho test.

Fig 3: The Kaplan–Meier curve of LGALS genes in TCGA Glioma dataset. (A) The Kaplan–Meier curve of LGALS1 in TCGA Glioma dataset. (B) The Kaplan–Meier curve of LGALS2 in TCGA Glioma dataset. (C) The Kaplan–Meier curve of LGALS3 in TCGA Glioma dataset. (D) The Kaplan–Meier curve of LGALS3BP in TCGA Glioma dataset. (E) The Kaplan–Meier curve of LGALS4 in TCGA Glioma dataset. (F) The Kaplan–Meier curve of LGALS8 in TCGA Glioma dataset. (G) The Kaplan–Meier curve of LGALS9 in TCGA Glioma dataset. (H) The Kaplan–Meier curve of LGALS12 in TCGA Glioma dataset. (I) The Kaplan–Meier curve of LGALS9C in TCGA Glioma dataset. TCGA, The Cancer Genome Atlas **p<0.01, ***p<0.0001.

Fig 4: LGALS expression correlates with glioma grades, GBM subtypes, and IDH mutation status in TCGA Glioma dataset. (A) Relative LGALS1 expression level of LGG and HGG patients in TCGA Glioma dataset. (B) Relative LGALS1 expression level of IDH-mutant and IDH-wild-type patients in TCGA Glioma dataset. (C) Relative LGALS1 expression level of different molecular subtype patients in TCGA Glioma dataset. (D–F) Relative LGALS3 expression level of LGG and HGG patients (D), IDH-mutant and IDH-wild-type patients (E), and different molecular subtype patients (F) in TCGA Glioma dataset. (G–I) Relative LGALS3BP expression level of LGG and HGG patients (G), IDH-mutant and IDH-wild-type patients (H), and different molecular subtype patients (I) in TCGA Glioma dataset. (J–L) Relative LGALS8 expression level of LGG and HGG patients (J), IDH-mutant and IDH-wild-type patients (K), and different molecular subtype patients (L) in TCGA Glioma dataset. (M–O) Relative LGALS9 expression level of LGG and HGG patients (M), IDH-mutant and IDH-wild-type patients (N), and different molecular subtype patients (O) in TCGA Glioma dataset. IDH, isocitrate dehydrogenase; GBM, glioblastoma; TCGA, The Cancer Genome Atlas **p<0.01, ***p<0.001, ****p<0.0001, ns, No significance.

Fig 5: Hgs ablation impaired lysosome homeostasis. (A) Enrichment plot showing the lysosome pathway enriched in Hgs-cKO mice. The heatmap of genes encoding key proteins in the lysosome pathway is shown in the lower panel. (B) Western blotting of the lysosomal marker LAMP1 and CD63 in ventricular extracts from 3-month-old Hgs-cKO and Hgsfl/fl hearts. Quantification is shown below. **p < 0.01 (means ± SEM, n = 4). (C) Representative immunofluorescence images of ventricular sections from 3-month-old Hgsfl/fl and Hgs-cKO mice costained for laminin (green) and LAMP1 or CD63 (red). Scale bars, 50 µm. (D) Western blotting of CTSD and CTSB in ventricular extracts from 3-month-old Hgs-cKO and Hgsfl/fl hearts. Quantification is shown on the right. **p < 0.01 (means ± SEM, n = 4). (E) Western blotting of LGALS3 in ventricular extracts from 3-month-old Hgs-cKO and Hgsfl/fl hearts. Quantification is shown on the right. **p < 0.01 (means ± SEM, n = 4). (F) Representative confocal fluorescent images of ventricular sections from 3-month-old Hgsfl/fl and Hgs-cKO mice immunostained for LGALS3 (red) and LAMP1 (green). Scale bars, 50 µm.