Diverse potential of secretome from natural killer cells and monocyte-derived macrophages in activating stellate cells

Chronic liver diseases, such as non-alcoholic steatohepatitis (NASH)-induced cirrhosis, are characterized by an increasing accumulation of stressed, damaged, or dying hepatocytes. Hepatocyte damage triggers the activation of resident immune cells, such as Kupffer cells (KC), as well as the recruitment of immune cells from the circulation toward areas of inflammation. After infiltration, monocytes differentiate into monocyte-derived macrophages (MoMF) which are functionally distinct from resident KC. We herein aim to compare the in vitro signatures of polarized macrophages and activated hepatic stellate cells (HSC) with ex vivo-derived disease signatures from human NASH. Furthermore, to shed more light on HSC activation and liver fibrosis progression, we investigate the effects of the secretome from primary human monocytes, macrophages, and NK cells on HSC activation. Interleukin (IL)-4 and IL-13 treatment induced transforming growth factor beta 1 (TGF-β1) secretion by macrophages. However, the supernatant transfer did not induce HSC activation. Interestingly, PMA-activated macrophages showed strong induction of the fibrosis response genes COL10A1 and CTGF, while the supernatant of IL-4/IL-13-treated monocytes induced the upregulation of COL3A1 in HSC. The supernatant of PMA-activated NK cells had the strongest effect on COL10A1 induction in HSC, while IL-15-stimulated NK cells reduced the expression of COL1A1 and CTGF. These data indicate that other factors, aside from the well-known cytokines and chemokines, might potentially be stronger contributors to the activation of HSCs and induction of a fibrotic response, indicating a more diverse and complex role of monocytes, macrophages, and NK cells in liver fibrosis progression.


Introduction
The recruitment of inflammatory immune cells from the circulation to the liver is considered as one of the hallmark characteristics of chronic fibrotic liver diseases such as nonalcoholic steatohepatitis (NASH) or viral hepatitis (1)(2)(3).Following the initial hepatocyte damage, local immune cells become activated and secrete cytokines and chemokines, which then result in the recruitment of immune cells including T cells, monocytes, and natural killer (NK) cells from the circulation toward areas of inflammation (4).Macrophages have been considered as key players in liver fibrosis progression, controlling the activation of hepatic stellate cells (HSC) and other liver fibroblasts, which, in turn, leads to increased collagen deposition and extracellular matrix buildup (5).Transforming growth factor beta (TGF-b1) plays a key role in activating HSCs via the TGFR-SMAD pathways, which ultimately leads to an upregulation of collagens, a-SMA, or CTGF upon activation (6,7).
A growing body of evidence highlights the emergence of triggering receptors expressed on myeloid cells 2 (TREM2) + lipidassociated macrophages (LAM) during the course of metabolicassociated liver disease including NASH.These macrophages are characterized by limited inflammatory and potentially protective capacity compared to other macrophage subsets (15,16).During chronic liver diseases, the constant influx of monocytes toward areas of inflammation leads to an imbalance between resident KC and MoMF (17).Controversially, scar-associated TREM2 + macrophages with an increased fibrotic potential have also been described (18).
Transcriptome analysis by next-generation sequencing (RNA-Seq) has been established as the state-of-the-art genomics assay to quantify gene expression at the bulk sample level or even at singlecell resolution.Accordingly, we have applied RNA-Seq to systematically analyze and compare gene signatures observed in human NASH livers versus gene signatures observed in vitro in cell culture models.To shed some light on the functional role of MoMF in the pathophysiology of liver inflammation and NASH, we analyzed the enrichment of in vitro-derived signatures from polarized MoMF, resting HSC, and activated fibroblasts in human NASH liver biopsies across the full spectrum of fibrosis.Furthermore, we validated our findings by assessing the potential of monocytes, MoMF, and NK cells to induce HSC to fibroblast transformation in vitro.
We found that human NASH signatures significantly overlap with TGF-b1 response genes in HSC.Moreover, human NASH liver cells are characterized by a lack of genes highly expressed in macrophage colony-stimulating factor 1 (M-CSF)-, interleukin (IL)-4-, and IL-13-stimulated MoMF in contrast to the gene signatures of monocytes, which are highly enriched, indicating a lack of tissue-repair macrophages during NASH progression.In addition, the functional assessment of the fibrotic potential of monocytes, MoMF, and NK cells showed that the secretion of TGF-b1 does not correlate with their capacity to induce HSC activation.Furthermore, IL-4/IL-13-stimulated MoMF do not induce HSC activation in vitro.In conclusion, our data indicate that IL-4/IL-13-stimulated MoMF have restorative functions, but their quantity could be insufficient to cope with the worsening of chronic inflammation in late-stage NASH.

RNA isolation and analysis
To extract RNA, cells were harvested, washed with phosphatebuffered saline (PBS), and lysed in RLT Plus Buffer (Qiagen, #1053393), followed by extraction of RNA using the RNeasy ® Mini Kit (Qiagen, #74106) according to the manufacturers' protocol.The extracted RNA samples were quantified by using NanoDrop ™ 8000 Spectrophotometer (Thermo Fisher Scientific) or NanoPhotometer ® N120 (Implen).For gene expression analysis, complementary DNA (cDNA) was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems ™ , #4368813) followed by qPCR (quantitative reverse transcription polymerase chain reaction) using TaqMan ® real-time PCR assay (Thermo Fisher, #4331182) for the following genes: COL1A1, COL3A1, COL10A1, ACTA2, and CTGF.The DDCt method was used to calculate the relative expression levels using glyceraldehyde-3-phosphate dehydrogenase as the housekeeping gene.In brief, the Ct values were normalized to the housekeeping control gene GAPDH (delta CT value).The fold change was calculated with the 2-DCt/DCt equation, which means that each delta Ct value was normalized to an untreated control (medium control).

RNA sequencing analysis of in vitrostimulated HSC and macrophages
For sequencing library preparation, 100 ng of total RNA was used as input for the TrueSeq RNA Sample Prep Kit v2-Set B (RS-122-2002, Illumina Inc., San Diego, CA, USA), producing a 275-bp fragment including adapters in average size.Eight libraries were normalized and pooled together on a single lane and sequenced as 52-bp single reads and seven bases index read on an Illumina HiSeq2000 instrument using the TruSeq SBS Kit HS-v3 (50-cycle) (FC-401-3002, Illumina Inc., San Diego, CA, USA).
The human NASH data set was previously published (20) and downloaded from GEO under accession number GSE135251.

RNA-Seq data analysis
A detailed description of the RNA-Seq analysis pipeline has been previously published (21).Briefly, reads passing the quality control filter were mapped against GRCh38 (human data).The gene expression levels were quantified based on genome annotation files from Ensembl version 86.For hierarchical clustering of the HSC and macrophage data, all genes with a significant differential expression (adjusted p-value ≤ 0.05, |log 2 foldchange| ≥0.5) in at least one of the conditions (treated vs. control) were selected.For the human NASH data set (20), we included all genes differentially expressed in at least one condition of F1-F4 vs. F0 (adjusted p-value ≤ 0.05, |log2FoldChange| ≥0.2).
Clustering was performed using DEGreport2 (v1.30.0) as described in (22).In all cases, we kept only clusters with more than 20 genes and made a pairwise comparison of MP and HSC clusters against NASH clusters by applying hypergeometric testing.We used EnhancedVolcano (23) (v1.12.0) to plot the differential expression results and ComplexHeatmap (24) (v2.10.0) to create heat maps.
We annotated all clusters by functional pathways from Reactome (25) database using clusterProfiler (26) (v4.2.2) and utilized enrichplot (27) (v1.14.1) to group similar pathways.Pairwise enrichment of human NASH clusters in gene clusters from the primary cell assays has been calculated using hypergeometric testing with R/Bioconductor.

Supernatant transfer experiment
Primary HSCs were seeded in 24-well plates at a density of 60,000 cells per well.After 2 h, the medium was replaced for starvation (w/o FBS, SteCGS) for an additional 22 h.The supernatant of stimulated immune cells was added in 1:6 dilution and incubated for 48 h.As positive control rhTGF-b1 was used at the indicated concentrations.

BrdU incorporation assay
BrdU incorporation assay for analysis of proliferation (Roche, #11647229001; Roche) was performed according to the manufacturer's guidelines with an incorporation time of 16 h and a total stimulation time with the conditioned medium of 24 h.Four independent HSCs were used for the assay.

Scratch (cell migration) assay
Approximately 30,000 HSCs (three donors) were seeded on a special 96-well plate (nr) and starved in 0.1 FCS media overnight.To set the scratch, a 96-pin wound-making tool (Incucyte ® WoundMaker, Essen bioscience) was used, and the cells were monitored by life cell imaging over 48 h in IncuCyte S3 system.Analysis of relative wound density over time was conducted by using the Integrated Cell Migration analysis module (Satorius, catalogue no.9600-0012).

Animals
Female C57BL/6J mice (8-13 weeks) were obtained from Charles River (Sulzfeld, Germany or US).Groups of two to five mice were housed under specific pathogen-free conditions in individually isolated ventilated cages with chow and unrestricted access to water ad libitum.The animals were kept under constant temperature (22°C), humidity (55%), and light conditions (12-h day/night cycle).After an acclimatization period of at least 4 days after their arrival in the facility, the animals were used for the experiments.Cervical dislocation euthanasia of the mice was performed [euthanized under isoflurane-O2 (CP-Pharma, Burgdorf, Germany; Nicholas Piramal, London, UK) anesthesia].After sacrifice, the livers were collected in cold University of Wisconsin (UW) preservation solution (DuPont Critical Care, Waukegab, IL, USA) and kept on ice until precision-cut liver slices were prepared.All experiments were approved by the animal welfare officers in Boehringer Ingelheim Pharma GmbH und Co KG as well as the Animal Review Committee of the German government and were performed according to the German Animal Protection Law.
The viability of PCLS was assessed by quantification of adenosine triphosphate (ATP) using a bioluminescence kit (Roche #11699695001).The obtained ATP value (pmol) was normalized to the total protein content (mg, measured with the Lowry method).

Analysis of mRNA expression using qRT-PCR
Quantitative reverse transcription polymerase chain reaction (qPCR) was performed in order to assess the gene expression of several markers related to fibrosis, inflammation, and fat metabolism in in vitro-generated macrophages and PCLS.QRT-PCR is a quantitative method based on the amplification reaction of gene sequences and the increase in fluorescence signal.

Histology and immunohistochemistry of PCLS
Tissue was fixed for up to 24 h at 4°C before samples were processed with an automated tissue processor (Tissue-Tek ® VIP ® 6, Sakura), embedded in paraffin, and cut in 3-µm sections.Hematoxylin and eosin (H&E) staining was performed according to standard protocols using Leica ST5020 Multistainer (Leica Biosystems GmbH, Nussloch, Germany).

Statistical analysis
GraphPad Prism 9.0 (GraphPad Software) was used to perform the statistical analyses.Statistical significance in the experimental data was calculated using one-way ANOVA with Benjamini-Krieger FDR correction, followed by paired Student's t-test.p < 0.05 was considered significant.The levels of significance are indicated by *p < 0.05, **p < 0.01, *** p < 0.001, and **** p < <0.0001.Statistics are presented as mean ± SEM as indicated.

TGF-b1 and TNF-a induce unique fibroblast activation
Liver fibrosis development is characterized by the transformation of HSC into myofibroblasts (MFB).To investigate the gradual increase of MFB activation during NASH, we performed a whole transcriptome analysis of primary human HSC treated with IL-13, TGFb-1, or TNF-a for 24 h (Figure 1A).Next, we performed a gene cluster analysis to identify the coregulated genes among the different activation conditions.Accordingly, 10 clusters of co-regulated genes were identified and functionally characterized by gene set enrichment analysis of the six main gene clusters (Figures 1B, C).As expected, TGF-b1 treatment induced a fibrotic response in HSC, indicated by upregulation of gene sets related to smooth muscle cell contraction and interaction with the extracellular matrix (Figure 1C).An overlap analysis with previously published human liver scRNA-Seq data (18) showed that the TGF-b1 response genes in HSC cluster 07 were expressed in the mesenchyme as well as the parenchymal cell types (Supplementary Figure S1A-C HSC in inflammation (Figure 1C).The IL-13 treatment induced only minor changes in gene expression compared to the control and did induce neither a fibrotic nor inflammatory response in HSC (Figure 1A).

M-CSF differentiation induces a unique macrophage phenotype
Another characteristic of NASH is the expansion of MoMF in areas of inflammation in the liver.Transcriptomic signatures of monocytes and M-CSF-differentiated MoMF polarized with IFN-g/ LPS or IL-4/IL-13 were generated and subjected to hierarchical clustering.As expected, M-CSF treatment induced the upregulation of well-known macrophage markers such as SPP1, APOE, or GPNMB (Figure 2A) (28).Hierarchical clustering identified 10 clusters of co-regulated genes across the four different conditions (Figure 2B).Clusters 01, 06, and 08 contain interferon gammadependent genes as confirmed by gene set enrichment analysis (Figure 2C), while clusters 02 and 03 contain genes that are dependent on IL-4 and IL-13 polarization.Those genes of MoMF cluster 02 were found to be expressed by resident liver KC and Frontiers in Immunology frontiersin.orgmature TREM2 + MoMF according to previously published scRNA-Seq data from liver cirrhosis and controls (18).On the contrary, genes enriched in MoMF cluster 04, which are specific for unpolarized monocytes, were also highly expressed in CD14 + and CD16 + monocytes according to the human scRNA-Seq data (Supplementary Figures 2A, B).Other enriched gene sets indicate changes in citric acid cycle, ATP synthesis, or the translational state of macrophages (Figure 2C), which is in line with previous reports (29).

NASH is characterized by gene signatures of myofibroblast and monocytes
Furthermore, we systematically compared the identified gene signatures of in vitro-polarized HSC and MoMF with liver gene signatures from human NASH samples across the full spectrum of fibrosis from F0 to F4. Accordingly, we have re-analyzed the previously published RNA-Seq data (20) by hierarchical clustering, which ultimately led to the identification of two large clusters of genes that show a strong log-linear increase or decrease of expression with fibrosis stage (Figure 3A).Pairwise enrichment analysis of the fibrosis-dependent gene clusters with our MoMFderived gene clusters revealed a significant loss of M-CSF + IL-4/IL-13 response genes (Figure 3B), indicating a continuous loss of this macrophage phenotype during NASH disease progression from F0 to F4.In contrast, the TGF-b1-dependent genes identified in our HSC activation assay were found to be strongly enriched in NASH cluster 01, which is increased with fibrosis (Figure 3C), reflecting the increasing population of TGF-b1-dependent MFB with disease worsening.Taken together, these data indicate that liver fibrosis progression, as a consequence of NASH, is characterized by loss of restorative M-CSF-dependent liver macrophages, in line with recent observations in NASH mouse models (30).

Limited contribution of M-CSF-polarized macrophages to HSC activation
We aimed to investigate the potential of cell culture supernatant from differentiated and polarized macrophages in the activation of HSCs.To assess the fibrotic potential of MoMF, we isolated primary CD14+ monocytes from human PBMCs via MACS and differentiated for 7 days with 50 ng/mL recombinant human M-CSF, followed by 6 h of stimulation with either IL-4, IL-13, or TNFa to induce unique polarization patterns, as depicted in Figure 4A.
MoMF differentiation and polarization were confirmed by evaluating cell surface markers as well as the production and release of cytokines as IL-6, IL-8, TNF-a, and active TGF-b1 (Figures 4B-E).Furthermore, we confirmed macrophage differentiations by checking the mRNA expression of M1/M2polarized secreted markers.The mRNA levels of IL-1b and MMP14, respectively, were significantly upregulated in LPS+IFNg-polarized MoMF, whereas higher mRNA levels of CCL18 and MMP12 were detected in IL-4/IL-13-treated MoMF (Figure 4D).In our experimental setting, in contrast to the classical inflammatory cytokines, the secretion of the anti-inflammatory and pro-fibrotic cytokine TGF-b1 was only moderately regulated following IL-4 or TNF-a stimulation (Figure 4E).The supernatant of the activated macrophages was then transferred to the primary human HSC.The gene expression analysis revealed that the supernatant from macrophages stimulated with TNF-a showed no induction of fibroblast activation; however, IL-4-IL-13 had a small effect on COL1A1 and COL3A1 expression (Figure 5A).The supernatant from PMA-hyperactivated MoMF induced a significant increase in CTGF, CCL2, VEGFA, and COL10A1 expression and a significant decrease in COL1A1 and COL3A1 expression in HSC (Figure 4A).
COL10A1 is an extracellular matrix-related protein and has been associated with cancer-associated fibroblasts (31).These data demonstrate that, with our experimental setting, IL-4/IL-13polarized MoMF lack the capacity to induce MFB activation of HSC and that the secretion of TGF-b1 alone is not robust enough to induce the pro-fibrotic capacity of macrophages.
To further evaluate the effect of supernatant from stimulated MoMF supernatant transfer on HSC activation, we performed scratch and proliferation assays after the supernatant transfer.No significant induction of HSC proliferation or migration, as markers of HSCs trans-differentiation, was observed (Figures 5B, C).

Limited role of undifferentiated monocytes during HSC activation
Next, we investigated undifferentiated monocytes in a similar setup as for the MoMF.The cytokine measurement of IL-6, IL-8, and TNF-a in the supernatant showed limited inflammatory polarization of freshly isolated monocytes upon the different stimuli (Figure 6A).Accordingly, we found no correlation between TGF-b1 secretion and HSC trans-differentiation with the supernatant of IL-4-, IL-13-, or TNF-a-stimulated monocytes (Figure 6B).Only COL3A1 gene expression was found to be increased by the supernatant of IL-4-and IL-13-stimulated monocytes, opposing the direction of TGF-b1 (Figure 5B).Differently from MoMF, the supernatant of TNF-a-or PMAtreated CD14 + monocytes did not induce a significant change of fibrotic response genes in HSC (Figure 6B).In summary, our data indicate a limited contribution of activated monocytes to profibrotic HSC activation.

A B C
Limited fibrotic capacity of macrophage colony-stimulating factor 1 (M-CSF)-dependent macrophages.

Activated NK cells contribute to HSC activation
Next, we investigated the impact of NK cells on HSC activation.The experimental procedure is depicted in Figure 4A.CD56 + NK cells were enriched from PBMCs by MACS, their purity assessed by using FACS (Figure 7A), and stimulated with PMA, IL-15, or IFN-g for 6 h.The medium was changed and the cells were incubated for another 16 h, and the supernatants were transferred to HSC.For quality control, the expression of some activation markers including CD69, NCR3 (natural cytotoxicity triggering receptor 3, also known as NKp30), and GZMB was analyzed by FACS and qPCR (Figures 7B, C).Furthermore, cytokine release by NK cells was quantified in NK cells following various stimulations As shown in Figure 7D, in contrast to the low secretion of TGF-b1 in MoMF and monocytes, the NK cells showed a higher background level of secreted TGF-b1, but no significant regulation upon release of stimuli.The supernatant of NK cells which were activated with IL-4, IL-13, and IL-15 reduced the HSC activation level compared to TGF-b1-stimulated cells (Figure 8A).Only PMA-hyperactivated NK cells induced a significant increase of COL10A1 in HSCs (Figure 8A), like the observations in MoMF.Nevertheless, the PMA-stimulated NK cell supernatant led to a general decrease of pro-fibrotic activation in other marker genes measured (COL1A1 and COL3A1) (Figure 8A).In line with the collagen mRNA expression data, we observed no significant induction of HSC proliferation and migration in the supernatant from stimulated NK cells (Figures 8B-D).
These data indicate that other immune cell populations besides MoMF could also modulate HSC to MFB differentiation during liver fibrosis progression.

PMA induces HSC activation ex vivo
Finally, to translate our in vitro findings to a more relevant in vivo condition, we assessed HSC activation in PCLS as a surrogate model.To prolong macrophage and monocyte viability, all PCLS were treated with M-CSF.The tissue morphology and viability of the PCLS samples were evaluated through H&E and TUNEL staining as well as measurement of total protein content and ATP release.H&E morphology was characterized by partial loss of cellular architecture with lysis of nuclear structures beginning on day 1; however, viable hepatic tissue was evident up to day 2 (Figure 9A).The morphological findings were confirmed by TUNEL staining distribution and extent.Moreover, no significant loss of total protein content and viability was pronounced between different treatments of PCLS at day 1 (Figure 9B).Various stimuli were added to each PCLS for 24 h, and HSC activation markers were quantified by qPCR (Figure 9C).The TNF-a treatment of PCLS induced a significant increase in almost all HSC activation markers.However, this observation might be due to the broad range of pro-fibrotic TNF-a effects observed in PCLS.In contrast to TNFa, IL-4 + IL-13 stimulation only increased Col1a1, but not of other markers.The PMA-treated PCLS showed a significant increase in the mRNA expression of Col1a1, Col3a1, and Ctgf (Figure 9C), in line with our in vitro data.
Our ex vivo data confirmed a strong effect of PMA on activating HSCs, which could be due to the essential role of the PKC pathway in activating HSCs.

Discussion
Our study aims to delineate the pro-fibrotic potential of the secretome of myeloid immune cells in the context of HSC activation and liver fibrosis progression.Macrophages are known for their high phenotypic plasticity and being involved in both acute and chronic liver diseases, including liver fibrosis (5,32).It has been reported that macrophages promote fibrogenesis by secreting cytokines which consequently drive the trans-differentiation of HSCs (33,34).Surprisingly, we found that, in our experimental conditions, IL-4/IL-13-stimulated MoMF failed to activate fibroblasts in vitro in contrast to PMA hyper-activated MoMF or NK cells.In addition, we did not observe an association between secreted TGF-b1 and fibroblast activation in supernatant transfer assays.This indicates that, in the presence of other soluble mediators, TGF-b1 could be less relevant than if used alone (35).These findings are of particular interest as previous studies have described scar-associated macrophages in human cirrhotic liver, with a high capacity to induce HSC activation in vitro (18).The comparison of RNA-Seq-derived gene signatures indicates that NASH progression is characterized by a lack of restorative IL-4/ IL-13-stimulated MoMF.In line with that, Stutchfield et al. showed that restorative M-CSF-differentiated monocytes and macrophages could provide therapeutic opportunities to improve recovery from acute liver injury (36).In addition, MoMF are described to play a key role in liver fibrosis reversal (37).Moreover, previous studies also suggested that liver cirrhosis could be treated by increasing the amount of liver macrophages via recombinant M-CSF treatment (10).Although our data indicate no direct fibrotic potential of M-CSF-polarized MoMF, which is of great interest in light of the recent discussions regarding the heterogenous role of MoMF in liver fibrosis (28).A recent study disclosed a pro-fibrogenic role for MoMF population expressing CCR2 (38).
Liver macrophages have often been described as the main drivers of HSC activation and thus as key players in liver fibrosis progression.This correlation has been linked to reduced inflammation and fibrosis in the absence of infiltrating monocytes in Ccr2 -/-mice or by using inhibitors of CCR2 or the ligand CCL2 (39-41).However, this concept has failed to translate into a beneficial clinical outcome (42).In line with our findings, the effect of various immune cells in activating HSCs in co-culture models has been studied by Zimmermann and Pastore (43,44).Additional functional studies might be needed to better understand the limited fibrotic potential of MoMF compared to that of tissue-resident KC or the recently described LAM (15, 45).One limitation prior to such investigations is the current lack of differentiation protocols for KC-or LAM-like macrophages in vitro.Future studies could also focus on identifying the key drivers of HSC activation, aside from TGF-b1, by in-depth secretome analysis.Taken together, our data highlight that TGF-b1 expression or secretion does not necessarily indicate a pro-fibrotic role of macrophages and that future studies will be needed to further dissect the interplay between myeloid immune cells and HSC activation in liver fibrosis.

1
FIGURE 1Transcriptome analysis of activated hepatic stellate cells (HSCs).(A) Volcano plot of differentially expressed genes in TNF-a-, TGF-b1-, and IL-13activated HSC versus the untreated control.(B) Heat map of Z-score normalized gene expression after hierarchical clustering for gene clusters 01-07 and 10, as identified for control and IL-13-, TGF-b1-, and TNF-a-stimulated HSC.(C) Gene set enrichment analysis for the main gene clusters 01 to 07.The color code indicates the significance of the enrichment for the specified gene set in the corresponding cluster (adjusted p-value for multiple testing).

2 RNA
FIGURE 2 RNA sequencing analysis of macrophage activation patterns.(A) Volcano plot of differentially expressed genes in macrophage colony-stimulating factor 1 (M-CSF) and M-CSF with either IL-4/IL-13-or LPS/IFN-g-activated MF versus untreated MoMF control.(B) Heat map of Z-score-normalized gene expression after hierarchical clustering for clusters 01 to 10. (C) Gene set enrichment analysis for the main gene clusters MF 01 to 07, 08, and 10.The color code indicates the significance of the enrichment for the specified gene set in the corresponding cluster (adjusted p-value for multiple testing).

3
FIGURE 3 Enrichment of hepatic stellate cells (HSC) and macrophage gene signatures in human non-alcoholic steatohepatitis (NASH) liver biopsies.(A) Zscore-normalized gene expression by fibrosis stage F0 to F4 in the two main gene clusters-identified by hierarchical clustering from the RNA-Seq of human NASH liver biopsies.(B) Pairwise enrichment of NASH gene cluster with gene clusters from the activated macrophages.the color code indicates -log 10 p-value.(C) Pairwise enrichment of NASH gene cluster with gene clusters from activated HSC. the color code indicates -log 10 p-value of the enrichment score.
(A) Myofibroblast (MFB) activation after macrophage supernatant transfer as quantified by qPCR.(B, C) Bar plots presenting scratch assay and proliferation experiment.(B) The supernatant from stimulated monocyte-derived macrophages (MoMF) cells [from four independent peripheral blood mononuclear cells (PBMCs)] was used to activate hepatic stellate cells (HSCs) in the scratch assay.The area under the curve for all the stimulation conditions was calculated and used for the statistics.(C) The proliferation of HSCs following supernatant transfer assay (from three independent PBMCs) was qualified in the BrdU incorporation experiment in four independent HSCs.All data are displayed as mean ± SEM, n ≥ 8; data from three individual experiments.One-way ANOVA with Benjamini-Krieger correction; *p < 0.05, ***p < 0.001, ****p < 0.0001.Sauer et al. 10.3389/fimmu.2024.1232070Frontiers in Immunology frontiersin.org

8
FIGURE 8 Variable response of NK cells to phorbol 12-myristate 13-acetate (PMA) hyperactivation.(A) Myofibroblast (MFB) activation after NK cell supernatant transfer as quantified by qPCR.COL1A1, COL3A1, COL10A1, CCL2, PDGFA, VEGFA, and CTGF were measured in NK cells stimulated with various stimuli.(B) Representative microscopy pictures of the scratch assay in response to phosphate-buffered saline (PBS), phorbol 12-myristate 13-acetate (PMA), or PDGF at two time points.Scale bar: 400 µm.(C, D) Bar plots presenting the scratch assay and the proliferation experiment.(C) Supernatant from stimulated NK cells [from four independent peripheral blood mononuclear cells (PBMCs)] was used to activate hepatic stellate cells (HSCs) in the scratch assay.The area under the curve (AUC) for all the stimulation conditions was calculated and used for the statistics.(D) The proliferation of HSCs following supernatant transfer assay (from three independent PBMCs) was qualified in the BrdU incorporation experiment in four independent HSCs.All data are displayed as mean ± SEM, n ≥ 8; data from three individual experiments.One-way ANOVA with Benjamini-Krieger correction; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.