Hepatotoxic and cholestatic injury generate distinct time-resolved and compartment-specific protein signatures. (A) Schematic overview of the experimental setup. Six animals were used in each model at each time point. (B) Numbers of quantified proteins at the indicated time points and experimental conditions; n = 6. (C) Box plots show the distribution of protein sequence coverage (coverage of possible tryptic peptides per protein in %) for the indicated matrisome categories (as defined by Naba et al.13) and all detected proteins in experimental conditions indicated. (D) Principal component analysis (PCA) of Total proteome separates time-dependent fibrogenesis and healing in CCl4- (closed symbols) and DDC- (open symbols) induced fibrosis. The first two components of data variability of 3,624 proteins identified in Total liver fractions in CCl4 and 3,521 proteins in DDC are shown; n = 6. (E) PCA shows the separation of Total (closed symbols) and E-fraction (open symbols) proteomes in time; the first two components of data variability are shown; n = 6. (F) Line plots show time-dependent changes in mass spectrometry intensities in Total (solid line) and E-fraction (broken line) proteomes for indicated selected proteins in CCl4 and DDC models; n = 4–6.

Time-resolved analysis of Total proteomes shows limited healing in matrisome-enriched protein clusters in the cholestatic model. (A) Venn diagram shows relative proportion of 1,276 and 1,836 proteins differentially expressed (t-test; Benjamini-Hochberg FDR < 5%) in Total CCl4 and DDC proteomes. (B) Hierarchical cluster analysis of the activity score of the upstream regulators at the indicated time points predicted by Ingenuity Pathway Analysis (IPA) from unique CCl4 and DDC protein signatures shown in A. (C,D) Hierarchical clustering of mean z-scored mass spectrometry (MS) intensities of proteins of Total CCl4 (C) or DDC (D) proteomes; n = 6. Profiles of z-scored MS intensities of proteins from matrisome-annotated clusters for CCl4 (I–III) and DDC (I and II) models are shown. (E) Venn diagram compares proteins from matrisome- annotated clusters shown in C and D. UniProt keyword enrichment annotation for each group within the diagram is indicated (Fisher’s test, Benjamini-Hochberg FDR < 4%).

Comparison of matrisome proteins differentially expressed between the CCl4- and DDC-derived Total proteomes with immunofluorescence (IF) localization of the main core matrisome proteins within the injured livers. (A) Venn diagram shows a comparison of matrisome proteins differentially enriched in Total CCl4 and DDC proteomes. Color coding indicates identified matrisome categories. (B–D) Representative IF images of fibronectin (B), collagen I (C), and collagen IV (D), all in green in liver sections from untreated controls (Ctrl), CCl4-, and DDC-treated mice at time points of fibrosis development (T2) and resolution (T4). Bile ducts were visualized with antibodies to keratin 19 (K19; red); nuclei were stained with DAPI (blue). CV, central vein; PV, portal vein. Scale bar = 100 μm. Line plots show time-dependent change in respective mass spectrometry intensities in Total (solid line) and E-fraction (broken line) proteomes in CCl4 and DDC models; n = 4–6.

Liver cell type dynamics and cell type-specific integrin expression during fibrogenesis and healing. (A–D) Box plots show mean z-scored mass spectrometry (MS) intensities of the indicated cell type-specific protein signatures in time. Hepatocytes (A; n = 18), HSCs, and activated portal fibroblasts (PFs) (B; n = 3 and 4), Kupffer cells (C; n = 6), granulocytes, and macrophages (D; n = 16 and 33). One-way ANOVA with Bonferroni’s post-test; *p < 0.05; **p < 0.01; †p < 0.001. (E) The line plots show the time-dependent change in MS intensities of indicated selected integrins in Total (solid line) and E-fraction (broken line) proteomes in CCl4 and DDC models; n = 4–6. (F) Representative immunofluorescence images of liver sections from untreated controls (Ctrl), CCl4-, and DDC-treated mice at time points of fibrosis development (T2) and resolution (T4) immunolabeled for integrin αv (green), K19 (red), and αSMA (blue). Arrowheads, integrin αv-positive injured hepatocytes; arrows, integrin αv-positive biliary epithelial cells of reactive ductuli. CV, central vein; PV, portal vein. Boxed areas, ×2 images. Scale bar = 50 μm.

Solubility profiling provides in-depth analysis of model-specific matrisome composition. (A) Venn diagram shows relative proportion of proteins from E-fraction proteome identified as proteins with increasing insolubility over the course of fibrosis in each model (see Figure S6 and Supporting Materials and Methods). Matrisome proteins are highlighted with color coding to indicate identified matrisome categories. (B) Line plots show time-dependent change in mass spectrometry (MS) intensities of indicated selected matrisome proteins uniquely identified in E-fraction (broken line) proteomes in CCl4 and DDC models; n = 4–6. (Solid line shows MS intensity profile in Total proteome.) (C) Representative immunofluorescence images of liver sections from untreated controls (Ctrl), CCl4-, and DDC-treated mice at indicated time points of fibrosis development (T1 and T2) and resolution (T4) immunolabeled for ECM1 (green) and collagen 1 (red). Nuclei were stained with DAPI (blue). Arrows, ECM1-positive reactive biliary epithelial cells. CV, central vein; PV, portal vein. Boxed areas, ×2 images. Scale bar = 50 μm.

Correlation analysis of protein abundance and changes in fibrotic deposits in the hepatotoxic model associates clusterin with fibrosis resolution. (A) Schematic illustrates the correlation of protein abundance changes with changes in sirius red-positive (SR+) areas of fibrous ECM deposits in CCl4-treated animals at the indicated time points. (B) The regulator of the actin cytoskeleton, coronin 1a (CORO1A), serves as an example of a protein with a positive slope of the correlation fit. The methionine cycle enzyme, adenosylhomocysteinase (AHCY), serves as an example of a protein with a negative slope of the correlation fit. (C,D) The scatter plots show the linear regression slope and the Pearson correlation coefficient for proteins of CCl4 Total (C), and E- fraction (D) proteomes. Statistical significance of the correlation is color-coded as indicated. The line plots show time-dependent change in mass spectrometry intensities of indicated representative proteins with significant correlation in Total (solid line) and E-fraction (broken line) proteomes in CCl4 and DDC models; n = 4–6. (E) Representative immunofluorescence (IF) images of liver sections from untreated controls (Ctrl), CCl4-, and DDC-treated mice at indicated time points of fibrosis development (T2) and resolution (T3 and T4) immunolabeled for clusterin (green), keratin 19 (K19, red), and collagen 1 (Col1, blue). Nuclei were stained with DAPI (blue). Arrowheads, clusterin staining signal delineating collagen deposits; arrows, clusterin-positive injured hepatocytes; yellow arrows, clusterin-positive biliary epithelial cells. CV, central vein; PV, portal vein. Boxed areas, ×2 images. Scale bar = 50 μm. (F) Representative IF images of human liver sections from different stages of chronic liver diseases of various etiologies (biliary-type, steatotic liver disease, and chronic HCV infection) immunolabeled for clusterin (Clu, green) and collagen 1 (Col1, magenta). Nuclei were stained with DAPI (blue). Top row shows increase in clusterin expression along collagen fibrils in biliary-type and metabolic syndrome-related cirrhosis compared to the stage of mild fibrosis. Bottom row documents change in clusterin staining pattern with chronic hepatitis C progression from fibrosis stage F1 to stage F4 (METAVIR grading system: F1, portal fibrosis; F2, periportal fibrosis; F3, bridging septal fibrosis; F4, cirrhosis). Arrowheads, clusterin staining delineating collagen deposits; arrows, clusterin-positive capillarized sinusoids; yellow arrows, clusterin-positive bile canaliculi (stage F1 only). PV, portal vein. Boxed areas, ×4 images. Scale bar = 50μm.

Atomic force microscopy (AFM) stiffness mapping reveals changes in the local mechanics of liver tissue upon hepatotoxic injury. (A) The line plot shows the dynamics of 32 myocardin-related transcription factor (MRTF) targets (highlighted in black) significantly enriched in ECM-enriched cluster I of CCl4 Total proteome (Figure 2C) identified by Fisher’s exact test (p = 0.01; Benjamini-Hochberg FDR = 0.03). (B,C) Representative polarized microscopy images of CCl4-treated (T2, B) and untreated control (Ctrl, C) liver sections with indicated regions (1–3) selected for AFM measurements. Note white areas corresponding to collagen fibers visualized by polarized light. Pink rectangle, region 1 (collagen-rich fiber); green rectangle, region 2 (injured hepatocytes); yellow rectangle, region 3 (interface hepatocytes). CV, central vein; PV, portal vein. Scale bar = 100 μm. Histogram shows Young’s moduli for the measured regions; n = 7 regions in three mice. (D–F) Representative polarized microscopy images of CCl4-treated (T2) liver sections with rectangle indicating the regions of AFM measurements (30 × 100 μm2) and corresponding pseudocolor Young’s modulus maps determined by AFM. Scale bar = 25 μm. Histograms show Young’s moduli for indicated regions of collagen-rich fibers (D), injured hepatocytes (E), and interface hepatocytes (F) at indicated time points of fibrosis development (T1 and T2) and spontaneous resolution (T4). Inset scatter plots show Young’s modulus values above 10 kPa for each time point.