*3.1. Single Cell RNA Sequencing Identifies Four Di*ff*erent Clusters of Myofibroblasts*

Chronic liver injury involves the activation of HSCs and their subsequent transformation towards collagen secreting MFB. To assess the heterogeneity of activated MFB, we isolated liver non-leukocytes non-parenchymal cells from three weeks-CCl4-treated mice and rested HSCs from untreated control mice. The presence of liver fibrosis after three weeks of CCl4 treatment was confirmed by a hematoxylin and eosin (H&E) stain as well as α smooth muscle actin (α-SMA) immunohistochemistry on FFPE tissue sections (Figure 1A). To capture all potential hepatic MFB from fibrotic livers, we excluded CD31 positive endothelial cells as well as CD45 positive leukocytes, but subjected all remaining double negative cells to scRNASeq analysis. FACS purified retinol positive HSCs, isolated from livers of untreated mice, served as a control (Figure 1B). Both HSCs and MFB were identified by their expression of platelet derived growth factor receptor-β (*Pdgfrb*), while the expression of alpha smooth muscle actin (*Acta2*), collagen, type III, alpha 1 (*Col3a1*)*,* and transforming growth factor, beta induced (*Tgfbi*) allowed the distinction of differently activated states of HSCs and MFB (Figure 1C).

**Figure 1.** Identification of four sub-populations of activated myofibroblasts (MFB) by single cell RNA sequencing (scRNASeq). (**A**) Representative images of formalin fixed and paraffin-embedded hematoxylin and eosin (H&E) or anti-α-smooth muscle (α-SMA) stained liver sections of control mice and mice subjected to repetitive carbon tetrachloride (CCl4) injections for 3 weeks. (**B**) Treatment scheme of mice during CCl4 treatment. Gating strategy for the isolation of resting hepatic stellate cells (HSCs) from healthy mice and activated MFB from CCl4 treated mice by FACS. (**C**) t-distributed stochastic neighbor embedding (t-SNE) plots mapping the identity of cells to resting HSCs (blue) and MFB from CCl4 treated liver (red), and the expression of marker genes used for identifying HSCs and MFB. (**D**) Definitive subset clustering of resting HSCs and activated MFB after exclusion of contaminating cells from scRNASeq data sets. (**E**) Log fold change (avg-logFC) gene expression of the top five marker genes for each cluster. For a better comparability, the number of cells in each cluster is aligned. (**F**) Avg-logFC gene expression of genes in the corresponding Gene Ontology (GO) categories, with overrepresented p-value in each cluster. *n* = 4 with an average of 5000 cells per condition and ~60,000 reads per cell.

In the next step, we excluded contaminating endothelial cells, leukocytes, and hepatocytes from our dataset to identify clusters of resting HSCs and activated MFB. We found that the HSCs form a highly homogenous cluster, while the MFB separated into four different sub-clusters, which we termed MFB I to IV (Figure 1D). Analysis of the most significantly expressed marker genes for each cluster allowed the further functional differentiation of these subtypes (Figure 1E). Besides common markers, such as PDGFR-β or various collagens, resting HSCs were uniquely characterized by a high expression of ficolin A (*Fcna*), which has been described to trim extracellular collagen as well as being an activator of a lectin complement pathway, and the hepatokine (*Angptl6*), an inducer of energy expenditure and regulator of the expression of fibroblast derived growth factor 21 (*Fgf21*) in white adipose tissue [8,9].

The major population of activated MFB, termed MFB I, was defined by a high expression of *Acta2*, the smooth muscle cell specific cytoskeletal protein, transgelin (*Tglna*), as well as various types of

collagens, such as *Col1a1*, *Col3a1*, or *Col6a3*. The second cluster, MFB II, expressed less extracellular matrix associated genes, but did express the inflammation associated serum leukocyte protease inhibitor (*Slpi*), complement factor *C3* (*C3*), serum amyloid A3 (*Saa3*), and cluster of differentiation 74 (*Cd74*). These data indicate towards the existence of a distinct subset of immunoregulatory MFB, characterized by a reduced capacity of modulating the extracellular matrix. The subset, MFB III, comprised proliferating fibroblasts, indicated by an expression of components of the activator protein 1 (*Ap1*) transcription factor, such as anti-apoptotic jun D (*Jund*) and its dimer forming partner *FBJ osteosarcoma oncogene B* (*FosB*). The smallest subset, MFB IV, displayed a mixed phenotype with a high expression of extracellular matrix modulators, such as the matrix gla protein (*Mgp*) and fibulin 1 (*Fbln1*), as well as growth arrest specific 6 (*Gas6*). Some of these marker genes have been described for portal fibroblasts [10]. While clusters MFB I, III, and IV showed a high expression of genes associated with the Gene Ontology (GO) category collagen fibril organization and extracellular matrix buildup (Figure 1F), cluster MFB II expressed less matrix associated genes, and more genes associated with the GO category positive regulation of leukocytes and immune regulation (Figure 1F). Due to the combination of characteristics of myeloid leukocytes, as well as of myofibroblasts, cluster MFB II most likely includes trans-differentiated myeloid myofibroblasts, which have been described before [11].
