Lipid-associated macrophages’ promotion of fibrosis resolution during MASH regression requires TREM2

Significance Since metabolic dysfunction-associated steatohepatitis (MASH) is the most common liver disease in the United States, the mechanisms underlying its regression are key to developing therapies. We elucidate the macrophage heterogeneity in livers undergoing MASH regression and demonstrate that lipid-associated macrophages (LAM) that emerge during MASH progression become the dominant macrophage subpopulation during regression and drive the restorative phenotype. LAMs express TREM2, and TREM2 is required for both the emergence of LAMs and their reparative functions. We propose efficient collagen degradation as a key protective mechanism mediated by TREM2 signaling. Absence of TREM2+ macrophages not only diminishes collagen resorption but also disrupts the metabolic coordination with other cell-types, leading to ineffective hepatic stellate cell inactivation and elimination of hepatic steatosis during regression.

background by intraperitoneal injection of CCl4 twice a week for 6 weeks (2 ml/kg; diluted in corn oil 1:4).The regression group (Reg) underwent a recovery period of 10 days following the final CCL4 administration.

Quantitative Polymerase Chain Reaction (qRT-PCR)
Total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA) followed by Rneasy column (Qiagen, Valencia, CA).The cDNAs were synthesized with MultiScribe™ Reverse Transcriptase (ThermoFischer Scientific, Waltham, MA) according to the manufacturer's instructions.qRT-PCR was performed using a QuantStudio 3 Real-Time PCR system (Applied Biosystems, Carlsbad, CA).The expression levels of genes were calculated and normalized to housekeeping gene Hprt using the ΔΔCT method (Invitrogen).Primer sequences are shown in Table S1.

Immunoblotting
Immunoblots (Western blots) were carried out on total protein extracts from liver tissues.
Frozen tissues were homogenized in RIPA buffer containing protease and phosphatase inhibitors (ThermoFischer Scientific, Waltham, MA), followed by centrifugation at 13000rpm.Protein concentrations were determined by BCA assay (Bio-Rad, Hercules, CA).Equal amounts of protein lysates were separated by SDS-PAGE, transferred to polyvinyl difluoride membrane, and subjected to immunoblot analysis for the indicated primary antibodies (Table S2).Proteins were visualized using the ECL detection system (Pierce, Rockford, IL) with the appropriate secondary antibodies.

Histology
The left lateral lobe of mice livers was fixed in 10% neutral-buffered formalin for 48 hours, embedded in paraffin (Formalin-fixed paraffin-embedded [FFPE]), sectioned, and processed for hematoxylin and eosin (H&E), picrosirius red stain (SR), and immunostaining with the indicated antibodies (Table S2).All histology quantifications were performed by ImageJ software (NIH).The SR positive area was normalized to the nonsteatosis area and plotted as a bar graph.For the linear regression analysis (Figure 2E) and to quantify fibrosis resolution (Figure 2F, 6H) in Foz/Foz and Foz::Trem2 -/-mice during MASH regression, the regression SR scores calculated as above were adjusted to their respective MASH progression-SR scores (0w Regression) and plotted using GraphPad prism.This normalization allowed for a cleaner visualization of the changes in SR-deposits over the course of MASH regression for both mouse models.For quantification of Trem2immunofluorescence images (Figure 1B, C), Trem2-positively stained cells in five randomly imaged nonoverlapping high-power fields in each sample were manually counted and plotted.For quantifications of GPNMB staining (Figure 5A, D and S5A), the number of GPNMB + hepatic crown-like structures (hCLS) in five randomly imaged

Primary liver cell isolation (Perfusion method):
Primary liver cells were isolated as described before (2).Briefly, mice livers were perfused by cannulation through the inferior vena cava with 0.33mg/ml pronase (Roche, Indianapolis, IN), followed by 0.67mg/ml collagenase D (Roche, Indianapolis, IN), for 5-10 minutes before ex vivo digestion in a solution containing pronase, collagenase D and DNase1 (Roche, Indianapolis, IN) for 10min in a rotating incubator at 37 0 C. The dissociated cells were strained through a 100µm mesh.The cell suspension was centrifuged at 50g for 1 minute to remove the hepatocytes, supernatant was collected and centrifuged at 800g for 7 minutes.Cell pellet washed with GBSS-B buffer at 700g for 7 minutes.Cells were finally isolated by Nycodenz gradient centrifugation at 2000g for 20 minutes.The Hepatic stellate cells (HSC) layer (upper layer) and the other nonparenchymal cells (NPC) layer (lower layer) were then collected and washed with GBSS-B at 800g for 7 minutes and processed as described below.

Hepatic Stellate Cells (HSC):
For in vitro experiments, HSC were plated at a density of 400,000 cells per well in a 24 well plate in DMEM supplemented with 10% FBS and 1% Antibiotic-Antimycotic (AA).Following overnight culture, cells were washed with PBS and maintained in DMEM with 2% FBS, and 1% AA.

Hepatic Macrophages:
The NPC layer was further processed for macrophage enrichment.The hepatic macrophages were enriched from the NPC fraction by their ability to readily attach and spread on tissue culture plates (in RPMI media supplemented with 10% FBS and 1% AA for 20min), and subsequently washing off the non-adherent cells (3)(4)(5).
To generate conditioned media (CM) from liver macrophages, the macrophageenriched population was cultured in RPMI with 2% FBS, 1% AA for 24 hrs.Conditioned media (CM) was collected, centrifuged, and was either used immediately or stored at -80°C.Live macrophage cell count was recorded at the end of the experiment for normalization.
Stimulation of HSC with macrophage-conditioned media: 250μl of the macrophage CM generated as described above was added to the isolated HSC for 24 hours.HSCs were harvested for total RNA isolation and subsequent qRT-PCR of fibrotic genes.Fibrotic gene expression was normalized to Hprt as well as the live macrophage cell count of the CM.

Collagenase activity assay
Collagenase activity was measured in isolated macrophages using a fluorescein-labeled substrate, DQ collagen type I, from bovine skin (Molecular Probes, Eugene, OR).Briefly, the isolated cells were lysed in a reaction buffer (0.05 M Tris•HCl, 0.15 M NaCl, 5 mM CaCl2, 0.2 mM sodium azide, pH 7.6) supplemented with 0.1% TritonX and incubated with 25ug/ml DQ Collagen in microplate wells at room temperature for 30min-3hrs.Fluorescence was measured using λ excitation = 485 nm and λ emission = 535 nm and normalized to the total protein content of the cell lysates.

Single cell RNAseq (scRNAseq):
Total non-parenchymal cells (NPC) were isolated as described above.The total NPC fraction was primarily dominated by endothelial cells.To take a deeper look into the various TREM2+ and TREM2-macrophage sub-populations, in the context of MASH and MASH regression, we enriched the immune cell fraction from the total NPC population (from healthy, MASH and MASH regression Foz mice) by sorting (FACS) out the CD31+ endothelial cells and subjected the immune cell fraction to 10X Genomics 3' scRNA-seq (Chromium Next GEM Single Cell 3′ GEM, Library & Gel Bead Kit v3.1).scRNA-Seq data was aligned with Cell Ranger version 3.0 and, each library was mapped to the mm10 genome.The Seurat package (version 3.0) was used for processing and merging raw count data, normalization, and clustering.Gene expression of each cell was normalized by total transcript counts and multiplied by 10,000.Cells of suspected low quality were removed, which had too few genes expressed (<500), which had high mitochondrial gene expression (>5%).Visualization was performed with Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction and the Louvain algorithm with a resolution of 0.5 was used to cluster the cells.
Cell Identity: Cell identity of the 18 immune cell clusters was determined from Panglao datasets (Dataset S1).Monocyte macrophage sub-clusters were identified based on previously reported cell identity markers (Figure 3C and S3C).
Identification of Trem2 Correlated Pathways: First, the Pearson correlation of all genes with the Trem2-gene expression was computed within the entire MASH (and/or MASH regression) macrophage populations.All macrophage clusters (clusters 0, 1, 2, 5, 8, and 15) were included in the analysis.This vector of correlation values was then input to Gene Set Enrichment Analysis (GSEA) to identify pathways enriched in genes positively or negatively correlated (p<0.05) with Trem2 expression (6).

Transcriptomic Analysis of Foz/Foz MASH and human MASH:
To determine the similarity in gene expression of Foz/Foz MASH to human MASH, the hepatic transcription profile of 12-week Foz+WD (MASH+Fibrosis) vs 12-week WT+WD (Steatosis only) was compared with a publicly available human MASH vs healthy obese gene expression dataset GSE48452 as before (7).Genes whose expressions are significantly altered between Foz+WD 12w (MASH) and WT+WD 12w (simple steatosis) are initially plotted.Genes that are also significantly upregulated and downregulated in the human dataset GSE48452 comparing MASH to healthy obese are determined and highlighted in orange and green respectively.A total of 257 genes are significantly dysregulated in both human and Foz mice MASH.Of these, 234 genes are dysregulated in the same direction in both species, indicating these genes are consistently upregulated or downregulated in both mice and humans.Among these commonly dysregulated genes,

Luminex
To identify differences in cytokine and chemokine profiles in NASH livers that develop in the presence or absence of TREM2, we subjected liver lysates of 16 weeks fructose+WD (FrWD) fed WT and Trem2 -/mice to luminex analysis.Liver tissues were homogenized in 1 mL of the lysis buffer tissue extraction reagent 1 (ThermoFischer Scientific, Waltham, MA) containing EDTA-free protease inhibitor cocktail (Roche, Indianapolis, IN).Liver lysates or cell culture supernatants were analyzed with the Luminex mouse cytokine/chemokine magnetic bead-custom made (R&Dsystems, Minneapolis, MN).
Samples were read on a Luminex MAGPIX Instrument (Luminex, Austin, TX) and MFIs (Mean Fluorescent Intensity) were normalized to absolute values with standard curves generated using the best-fit feature in the Masterplex software (Hitachi Solutions, Irvine, CA).Data normalization was performed using total liver protein levels and concentrations graphed in GraphPad Prism (San Diego, CA) software.

nCounter analysis
To generate bone marrow derived macrophages (BMDM), bone marrows from the tibia and femora of WT and Trem2 -/-littermates were flushed out and cultured in DMEM containing 10% FBS and 100ng/ml M-CSF.A homogeneous population of adherent BMDM cells were obtained after 6-days culture.On day 6, the cells were collected and re-plated for 24hrs.The cells were then treated with 50ng/ml IL-4, and a combination of IL-10 and TGFb for 48h.The cells were subsequently harvested, and RNA was extracted using Direct-zol RNA Miniprep kits (Zymo Research).Comprehensive RNA expression analysis was conducted by nCounter using Myeloid Innate Immunity Panel (NanoString).
Extracted total RNA (50ng) was mixed with the master mix and capture ProbeSet in the kit, and the mixture was hybridized for 19 h at 65C.The hybridized samples were immediately loaded into Prep Station, and immobilized RNA samples on the cartridge was analyzed with Digital Analyzer (NanoString).Data analysis was performed using nSolver v4.0 (NanoString).

Flow Cytometry
Non-parenchymal cells (NPC) enriched in monocyte/macrophages were isolated based on previous reports (8)(9)(10).Briefly, freshly dissected livers were cut in small pieces and digested in RPMI media containing Collagenase D (2mg/ml) and DNase (0.1-0.2mg/ml) for 20 minutes in orbital shaker (150rpm) at 37C.The content was then presshomogenized through the mesh screen of 100uM Falcon cell strainer with the plunger from a 3-5CC syringe.After centrifugation, the pellet was resuspended in 33% Percoll and centrifuged for 20 min at 500G without brake.After RBC lysis, the pellet was used for flow cytometric analysis.The cells were first incubated with CD16/32 in order to block the Fc receptor, to reduce nonspecific binding.They were then stained with monoclonal antibodies (Table S2) for 30 min to 1 h on ice.The gating strategy for liver macrophages is outlined in Figure S5C-K (as previously reported ( 11)).Sorting and analysis were performed using a SONY MA900 Cell Sorter (Sony Biotechnology), SONY MA software (v.3.1.1)and FlowJo software (v.10.6.2,BD).The populations of F4/80+Cd9+, F4/80+Clec4f+, and Cd11b+Ly6c+ cells within each mouse group (Foz and Foz Trem2-/during MASH progression and regression) were quantified as the proportion of live CD45+ cells and were expressed as fold change relative to the Foz+WD 12w (Foz+WD MASH) group.
the 20 with the highest absolute fold change are annotated.Dotted vertical lines indicate log fold change of 0.5 or -0.5.Dotted horizontal line indicates p=0.05.

Fig. S1. 1 .
Fig. S1.1.TREM2 in MASH.(A) Violin plot representing the expression levels of Trem2 in various liver cell types from a proteogenomic atlas of the murine liver (12) (left panel) and an

Fig. S2 .Fig. S3 .
Fig. S2.Absence of TREM2 prevents effective MASH and fibrosis resolution.12w WD-fed Foz/Foz and Foz::Trem2 -/-mice, were either switched to a chow diet for an additional 4-8w to model MASH regression or continued on WD as age-matched controls (MASH progression 20w) (A) Changes in body weight, liver weight, and liver-to-body weight ratio were plotted.(B) FFPE liver sections from indicated mice were subjected to aSMA staining (Scale Bar 100µm).The tissue staining was quantified by ImageJ and

Fig. S4 .Fig. S5 .
Fig. S4.Changes in gene signature of various macrophage sub-population duringMASH and regression.(A) Pathway enrichment analysis of all macrophage clusters

Fig. S6 .
Fig. S6.Comparison of gene signatures in various macrophage sub-populations inMASH vs regression.(A-D)scRNAseq data from MASH and regression macrophage subpopulations (clusters 1, 2, 5 and 15) was analyzed as described in Figures3 and 4. Trem2correlated genes that are significantly altered are represented as heat maps showing relative file).Single cell RNAseq of Liver Non-Parenchymal cells (NPC) -Cluster Identity based on Panglao database Dataset S2 (Separate file).Trem2+ve and Trem2-ve correlated meta-pathways and individual pathways during MASH progression.Each meta-pathway is depicted in individual tabs.Only significant pathways with FDR<0.3 are shown.Dataset S3 (Separate file).Trem2+ve and Trem2-ve correlated meta-pathways and individual pathways during MASH regression.Each meta-pathway is depicted in individual tabs.Only significant pathways with FDR<0.3 are shown.