Macrophage potentiates the recovery of liver zonation and metabolic function after acute liver injury

The liver is an exclusive organ with tremendous regenerative capacity. Liver metabolic functions exhibit spatial heterogeneity, reflecting liver zonation. The mechanisms controlling the proliferation of hepatocytes and the accompanying matrix reconstruction during regeneration have been well explored, but the recovery potential of differentiated metabolic functions and zonation after liver injury remains unclear. We employed a mouse model of carbon tetrachloride (CCl4) induced-acute liver injury with clodronate-induced macrophage depletion to clarify the impact of liver injury on liver metabolism and recovery dynamics of metabolic function and liver zonation during regeneration. Depleting macrophages suppressed tissue remodelling and partially delayed cell proliferation during regeneration after liver injury. In addition, recovery of metabolic functions was delayed by suppressing the tissue remodelling caused by the depleted macrophages. The model revealed that drug metabolic function was resilient against the dysfunction caused by liver injury, but glutamine synthesis was not. Metabolomic analysis revealed that liver branched-chain amino acid (BCAA) and carbohydrate metabolism were suppressed by injury. The plasma BCAA concentration reflected recovery of hepatic function during regeneration. Our study reveals one aspect of the regenerative machinery for hepatic metabolism following acute liver injury.


Histological analysis
PFA perfusion-fixed liver tissues were additionally post-fixed with 4% PFA for 24 h, dehydrated, and embedded in paraffin. Paraffin-embedded tissues were sectioned by a microtome at 5 µm thickness. Tissue sections were incubated in 10 mM sodium citrate buffer (pH 6.0) at 110°C for 20 min to retrieve the antigen for Ki-67 immunostaining. The tissue sections were incubated in 3% H2O2/methanol (MeOH) for 10 min to quench endogenous peroxidase activity. The tissue sections were washed with PBS, blocked in 5% bovine serum albumin (BSA)/PBS for 40 min, and incubated with rabbit monoclonal anti-Ki67 [SP8] antibody (1:40 in 0.5% BSA/PBS; GTX16667, Gene Tex) at 4°C overnight, followed by washes with PBS and incubation with peroxidase-conjugated Affinipure goat anti-rabbit IgG (H+L) (1:1000 in 0.5% BSA/PBS; Jackson Immuno Research) for 2 h. The Ki-67-positive signal was detected with the ImmPACT DAB Peroxidase Substrate Kit (SK-4105, Vector Laboratories). Histological images were captured by the ZEISS Axio Imager A2 upright microscope.
Three different images were collected from one tissue section in each mouse to quantify the number of Ki-67 positive nuclei, the PAS positive area and the necrotic cell area. Images were quantified using the ImageJ macro in Fiji software.
The pseudo-colour images (R, G, and B) were merged to the RGB image using Fiji software. Briefly, liver slices were peremeabilised with 0.5%v/v Triton X-100 in PBS for 1 h at 37°C.
The tissue slices were incubated with a reaction enzyme mixture for 2 h at 37°C and mounted with FluorSave Reagent. Images were captured by a Leica STELLARIS 5 confocal microscope. Three different images were collected from one tissue section in each mouse to quantify the TUNEL positive nuclei. Images were quantified using the ImageJ macro in Fiji software.

GC/MS-based untargeted metabolomics
Frozen liver tissues (50 mg) were homogenised in 1 mL solvent mixture (MeOH/chloroform/water = 2.5:1:1 v/v/v). Then, 10 µL of an internal standard solution (1.2 mg/mL ribitol in water) was added to the homogenates, and the homogenates were vigorously mixed on a temperature-controlled shaker (MBR-022UP, Taitec) for 30 min at 20°C. Subsequently, the homogenates were centrifuged at 16,000 × g for 5 min at 4°C to discard insoluble matter. Water (200 µL) was added to the supernatants and mixed well. using helium as the carrier gas (constant flow of 1.120 mL/min). Column oven temperature was held at 80°C for 2 min, increased to 330°C (15°C/min), and held for 9 min. Other analytical conditions were as follows: injector temperature, 250°C; septum purge flow, 5 mL/min; solvent delay, 3.5 min; MS ion source temperature, 200°C; MS interface temperature, 250°C; and MS scan mode, 85-500 m/z scan. A saturated hydrocarbon standard mixture (1021-58321, GL Science) was used to adjust retention time drifts. A retention index (RI) was calculated from retention time of the alkanes. Raw experimental data were converted to netCDF type data using JEOL Escrime software. Then, netCDF data were converted to Abf type data on the Reifycs Abf Converter. Spectral deconvolution and peak identification were performed using MS-DIAL 3.70 software coupled with a reference database (GL-Science DB Kovats RI; for InertCap 5MS/NP column) downloaded from the RIKEN PRIMe website (http://prime.psc.riken.jp/). Normalised peak intensity of the metabolites in each sample was calculated using MS-DIAL software. Exported raw text data from MS-DIAL software were converted into a data table format using R scripts written inhouse.
The peak intensity of the metabolites was standardised to a mean of 0 and variance of 1 prior to the data analysis. Principal component analysis, Pearson's correlation coefficient calculation, hierarchical clustering, and data visualisation were performed in R software. The pathway analysis was performed at the Metaboanalyst 4.0 website (https://www.metaboanalyst.ca/).

Supplementary Figures
Supplementary Figure S1. The effect of clodronate-loaded liposome (CLO) treatment on F4/80 expression in the mouse liver. Representative images of F4/80 immunofluorescent stained liver sections (scale bar, 100 µm); magenta, F4/80; blue, cell nuclei. Image quantification was performed by using image J. Data are presented as mean ± SEM (n = 6/group). Intergroup differences between the vehicle and CLO group at each time-point were compared using Welch's t-test; *p < 0.05, ***p < 0.001. Figure S2. The effect of clodronate-loaded liposome (CLO) treatment on expression of macrophage marker genes in the mouse liver. Gene expression levels in the liver were measured by qPCR and normalised with 18S rRNA. White bars indicate vehicle-treated mice, and black bars indicate CLO-treated mice. Data are presented as mean ± standard error of the mean (SEM; n = 6/group). Intergroup differences between the vehicle and CLO group at each time-point were compared using Welch's t-test; *p < 0.05, **p < 0.01, ***p < 0.001. Adgre1, adhesion G protein-coupled receptor E1; Clec4f, C-type lectin domain family 4 member F. Figure S3. The effect of macrophage depletion on protein expression of PCNA and cyclin D1 in the mouse liver. Proliferating cell nuclear antigen (PCNA) and cyclin D1 protein expression measured using immunoblotting. β-actin was used as loading controls. Data are presented as mean ± SEM (n = 6/group). Intergroup differences between the vehicle and CLO group at each time-point were compared using Welch's t-test; *p < 0.05. Figure S4. The effect of macrophage depletion on expression of cyclins in the mouse liver. Gene expression levels of cell cycle genes were measured by qPCR and normalised with 18S rRNA. White bars indicate vehicle-treated mice and black bars indicate CLO-treated mice. Data are presented as mean ± SEM (n = 6/group). Intergroup differences between the vehicle and CLO group at each time-point were compared using Welch's t-test; *p < 0.05. Ccnd1, cyclin D1; Ccne1, cyclin E1; Ccna2, cyclin A2; Ccnb1, cyclin B1. Figure S7. The effect of macrophage depletion on apoptotic cell death in the mouse liver. Representative images of apoptotic cells in liver sections (scale bar, 100 µm); green, TUNEL positive nuclei (apoptotic cells); blue, cell nuclei. Image quantification was performed by using image J. Data are presented as mean ± SEM (n = 6/group). Intergroup differences between the vehicle and CLO group at each time-point were compared using Welch's t-test; ***p < 0.001.