key: cord-0329103-njn9kw4u
authors: Li, Wenyang; Chen, Jennifer Y.; Sun, Cheng; Sparks, Robert P.; Pantano, Lorena; Rahman, Raza-Ur; Moran, Sean P.; Pondick, Joshua V.; Kirchner, Rory; Wrobel, David; Bieler, Michael; Ho Sui, Shannan J.; Doerner, Julia F.; Rippmann, Jörg F.; Mullen, Alan C.
title: Nanchangmycin regulates FYN, FAK and ERK to control the fibrotic activity of hepatic stellate cells
date: 2021-10-09
journal: bioRxiv
DOI: 10.1101/2021.10.08.463221
sha: 609d2742a102c154a56a3ac5f4626b8389d3984d
doc_id: 329103
cord_uid: njn9kw4u

Chronic liver injury causes fibrosis, characterized by the formation of scar tissue resulting from excessive accumulation of extracellular matrix (ECM) proteins. Hepatic stellate cell (HSC) myofibroblasts are the primary cell type responsible for liver fibrosis, yet there are currently no therapies directed at inhibiting the activity of HSC myofibroblasts. To search for potential anti-fibrotic drugs, we performed a high-throughput compound screen in primary human HSC myofibroblasts and identified 19 small molecules that induce HSC inactivation, including the polyether ionophore nanchangmycin (NCMC). NCMC induces lipid re-accumulation while reducing collagen expression, deposition of collagen in the extracellular matrix, cell proliferation, and migration. We find that NCMC increases cytosolic Ca2+ and reduces the phosphorylated protein levels of FYN, FAK, ERK1/2, HSP27 and STAT5B. Further, depletion of each of these kinases suppress COL1A1 expression. These studies reveal a signaling network triggered by NCMC to inactivate HSC myofibroblasts and reduce expression of proteins that compose the fibrotic scar. The identification of the antifibrotic effects of NCMC and the pathways by which NCMC inhibits fibrosis provides new tools and therapeutic targets to combat the development and progression of liver fibrosis.

we evaluated 18 housekeeping genes, consisting of seven commonly used genes as well as 139 eleven reference genes identified from the literature (Eisenberg & Levanon, 2013). We first 140 analyzed RNA sequencing data from HSCs under multiple conditions (Chen et al., 2017) . 141

Among these candidates, GUSB, POLR2A, EMC7, VCP, PSMB2 and VPS29 showed the 142 lowest standard deviation (0.15 or less). Further comparison of expression of these genes in 143 inactivated HSCs (induced by the addition of nortriptyline or ceramide (Chen et al., 2017) ) and 144 culture-activated HSCs revealed that GUSB, POLR2A, EMC7 and PSMB2 had the least fold 145 change in expression (10% or less upon HSC inactivation). Thus, we selected these four 146 reference mRNAs for further evaluation. GAPDH, which is used routinely as a reference 147 control, was also included for comparison ( Figure 1 -figure supplement 1A) . We quantified 148 expression using qRT-PCR in HSC cDNA samples reverse-transcribed from equal amounts 149 of total RNA. PSMB2, which encodes proteasome 20S subunit beta 2, showed the least 150 variation as indicated by standard deviation and was chosen as the reference mRNA for the 151 secondary qRT-PCR-based screen (Figure 1 -figure supplement 1B) . 152 153 qRT-PCR was performed to quantify ACTA2, COL1A1 and PSMB2 mRNA levels in each 154 sample. Relative fold changes were calculated compared to DMSO control. We defined the 155 following as criteria for compound advancement: 1. Fold change of COL1A1 was reduced to 156 less than 0.5 of DMSO control (FDR<0.05); 2. Fold change of ACTA2 was reduced to less 157 than 0.5 of DMSO (FDR<0.05); 3. Averaged PSMB2 expression was between 0.2-2.0 of 158 DMSO ( Figure 1C and Supplementary Table 5 ). This last criterion was added to avoid 159 selecting compounds where large changes in PSMB2 expression made it difficult to interpret 160 changes in ACTA2 and COL1A1 expression. Of the 140 compounds, a total of 44 compounds 161 met all three criteria. Five compounds were not commercially available, and 39 compounds 162 were advanced for further analysis. 163 164 Next, we evaluated dose response curves (DRCs) for each compound at eight different 165 concentrations, from 10 pM to 10 µM, using a Bodipy lipid accumulation assay similar to that 166 employed in the primary screen. Dose response curves were scored blindly by three 167 researchers (Supplementary Table 6 and Methods), and nortriptyline served as a reference. 168

Of the 39 compounds, 19 received an average score that was the same as or higher than 169 nortriptyline controls ( Figure 1D , Figure 1 -figure supplement 2, and Table 1) . 170 171 Nine compounds had a DRC score less than 2 and an EC50 less than 5 µM and were 172 considered the highest priority. Two subgroups of compounds were identified within this group 173 based on similar bioactivity -histone deacetylase inhibitors (HDACIs), including trichostatin A 174 and abexinostat, and Na/K-ATPase inhibitors, including ouabain, digitoxigenin, and digoxin. 175

Histone deacetylases are linked to a variety of fibrotic disorders, including liver fibrosis (Pang 176 & Zhuang, 2010) . HDACIs, such as MC1568 and Valproate, have been reported to reduce 177 HSC activation and alleviate liver fibrosis in animal models (Yoon et al., 2019) . The presence 178 of HDACIs in our final candidate list supports the validity of our screening approach in 179 identifying potential liver fibrosis inhibitors. Na/K-ATPase activity may play a role in non-180 alcoholic fatty liver disease (Sodhi et al., 2017) , but it is not clear how Na/K-ATPases regulate 181 HSC activity and liver fibrosis. Due to the toxicity and narrow therapeutic dose range of cardiac 182 glycosides, which limit their potential application in treatment of liver fibrosis, we decided not 183 to pursue further evaluation of this group of compounds. Nanchangmycin (NCMC), a natural 184 product of Streptomyces nanchangensis, is a polyether insecticidal antibiotic (Sun et al., 2002) 185 and is one of the most potent hits. Studies of NCMC are limited, but it has been shown to have 186 a broad spectrum of antiviral activity against diverse arboviruses (Rausch et al., 2017) As we switched to NCMC from a different source with higher purity, we found that the new 204 NCMC stock has a lower EC50 in HSCs from human donors and HSCs from C57BL/6 mice 205 We next quantified the effect of NCMC on ACTA2 and COL1A1 expression in multiple primary 211 human HSC lines. NCMC treatment reduced both ACTA2 and COL1A1 levels at 100 nM or 212 higher ( Figures 3A-B) . We also observed a dose-dependent effect of NCMC on Acta2 and 213

Col1a1 expression in murine HSCs at day 2 ( Figure 3C ). To investigate how NCMC affects 214 the level of collagen deposited into the ECM, we performed the scar-in-a-jar assay to 215 accelerate the process of ECM deposition with addition of molecular crowding reagents and 216 TGF-β (Chen et al., 2009; Good et al., 2019) . NCMC treatment significantly decreased 217 collagen staining intensity and fiber area ( Figures 3D-E) . In addition to two-dimensional (2D) 218 cell culture models, we also tested NCMC's effect on COL1A1 expression in spheroids 219 consisting of primary human HSCs and primary rat hepatocytes. Both the basal expression of 220 COL1A1 and TGF-β-induced COL1A1 expression were significantly reduced by NCMC 221 treatment ( Figure 3F ). 222 223 HSCs were next treated with NCMC and analyzed by RNA-sequencing analysis, which 224 revealed that NCMC broadly affects genes associated with fibrosis. Among the top gene sets 225 negatively enriched in the NCMC-treated group were ECM-related signatures, including ECM 226 structural constituent and collagen-containing ECM, as well as signatures relevant to migration, 227 including contractile fibers. Of note, genes associated with oligosaccharide lipid intermediate 228 biosynthetic process were positively enriched, possibly contributing to the re-accumulation of 229 lipid droplets (Figure 3G and Supplementary Table 7) . We compared the RNA sequencing 230 data with a canonical HSC gene signature (Zhang et al., 2016) , an HSC-specific signature that 231 is highly and uniquely expressed in HSCs and correlates with the extent of fibrosis (Zhang et 232 al., 2016) , and the liver cirrhosis signature from Disgenet database (Piñero et al., 2020) . We 233 observed that these signatures were significantly negatively enriched ( kinase array to define kinase signaling molecules modulated by NCMC. HSCs were treated 296 with DMSO or 1 μM NCMC for 1 and 18 hours ( Figure 5A) . Among the 45 proteins tested, 297 FYN phosphorylation at Y420 was reduced by approximately 40% at both 1 hour and 18 hours. 298

We selected HSP27 (HSPB1), ERK1/2 (MAPK3/1), STAT5A/B, and FAK (PTK2) to study 299 further in addition to FYN because 1) they also showed decreased phosphorylation at 18 hours, 300

and 2) genes encoding these products are expressed at a relatively high level in HSCs, as 301 indicated from RNA sequencing data, suggesting that these may also be potential mechanistic 302 targets of NCMC in HSCs. 303

To further investigate the role of these seven kinases in human HSCs, we depleted each 305 kinase using pooled siRNAs in human HSCs from three donors. We observed a consistent Pipeline Pilot, and the most common structure for each cluster was defined based on this 533 value. The strongest hit with the most common structure for each cluster was selected as the 534 representative for the cluster. Promiscuous bioactive compounds that contain pan assay 535 interference structures (PAINS) (Baell & Nissink, 2018) , or that we identified as frequent hits 536 in screens at ICCB-L were not included for further analysis, as the exhibited bioactivity may 537 be attributed to interference with specific assay readouts and/or nonspecific, intractable 538 mechanism of action (Matlock et al., 2018) . Frequent hits were defined as having a positive 539 hit rate of more than 20% in screens performed at ICCB-L or more than 10 total positive hits 540 in the database of ICCB-L screens. One additional compound was removed because the 541 molecular formula was the same as another selected compound, and one compound was 542 removed due to similarity in structure to nortriptyline (Supplementary Table 4 The dose response curve screen was performed using an adapted lipid accumulation assay 565 with live human primary HSCs. Briefly, cells were plated at a density of 2500 cells/well in 384-566 well plates. After 24 hours, compounds were added in duplicate at concentrations from 0.001 567 to 10 µM. Nortriptyline (10 µM) and DMSO (0.1%) served as controls. Cells were incubated 568 with compounds for 24 hours, followed by treatment with Bodipy (1 µg/ml; ThermoFisher, cat# 569 D3922) and NucLight Rapid Red (final dilution 1:4000; Essen BioSciences, cat# 4725) for an 570 additional 12 hours to stain lipid droplets and nuclei. Fluorescent signals were measured using 571 an Incucyte S3 system. 572 573 Bodipy stained area and nuclei count were determined selecting two fields per well. The 574

Bodipy-stained area per nuclei count was calculated per field and the mean was determined. An expression dataset containing gene name and log2 (fold change) was generated based on 684 the RNA sequencing results and loaded to the software as the input file. The c5.all.v7.4 gene 685 matrix was used as the database of gene sets, and gene sets smaller than 10 or larger than 686 1000 in size were excluded for the analysis. The canonical HSC gene signature and specific 687 HSC gene signature were obtained from previous publication (Zhang et al., 2016) , and the 688 liver cirrhosis signature was downloaded from Disgenet database (Piñero et al., 2020) . Among 689 the 44 genes in the canonical HSC signature, 35 were found in our differential expression list. 690

Among the 122 genes in the specific HSC signature, 97 were found in our differential 691 expression list. Among the 103 genes in the liver cirrhosis signature, 69 were found in our 692 differential expression list. These genes were listed in Supplementary Table 8 . Error bars represent mean ± SEM (n=4). One experiment was performed for each donor 1280

shown. ns indicates not significant (p > 0.05), * indicates p < 0.05, ** indicates p < 0.01, *** 1281

indicates p < 0.001, and **** indicates p < 0.0001 (one-way ANOVA test). 1282 Western blot of HSCs treated with DMSO control or 1 µM NCMC for 18 hours. β-actin is used 1307 as a loading control (representative of three experiments). 1308 1309

Targeting acid ceramidase inhibits YAP/TAZ signaling to reduce 864 fibrosis in mice

From 867 NASH to HCC: current concepts and future challenges

Seven Year Itch: Pan-Assay Interference Compounds

Liver fibrosis

Is liver fibrosis reversible?

Is liver fibrosis reversible?

Calmodulin inhibitor W13 induces sustained 879 activation of ERK2 and expression of p21(cip1)

The Scar-in-a-Jar: studying potential antifibrotic compounds from the 883 epigenetic to extracellular level in a single well

Tricyclic Antidepressants Promote Ceramide Accumulation to Regulate 887

Interactions Regulate the Subcellular Localization of Extracellular Signal-Regulated 891 Kinases (ERKs)

Regulation of mitogen-activated protein kinase phosphatase-1 expression by

Novel human hepatic organoid model enables testing of drug-952 induced liver fibrosis in vitro

Aglycone polyether ionophores as broad-spectrum agents inhibit multiple enveloped 956 viruses including SARS-CoV-2 in vitro and successfully cure JEV infected mice. 957 bioRxiv

Extracellular matrix gene expression increases 959 preferentially in rat lipocytes and sinusoidal endothelial cells during hepatic fibrosis in 960 vivo

Molecule Reactivity Identifies Promiscuous Bioactive Compounds

STAT5A and 965 STAT5B-Twins with Different Personalities in Hematopoiesis and Leukemia

Fate tracing reveals hepatic stellate cells as dominant contributors to liver 969 fibrosis independent of its aetiology

PGC-1alpha-responsive genes involved in oxidative 973 phosphorylation are coordinately downregulated in human diabetes

Ionomycin enhances Ca2+ influx by stimulating store-976 regulated cation entry and not by a direct action at the plasma membrane

Deaths: Final Data 979 for 2018

Mitogen-activated protein kinases p42mapk and p44mapk are required for 982 fibroblast proliferation

Histone deacetylase: a potential therapeutic target for fibrotic 985 disorders

& 988 Furlong, L. I. (2020). The DisGeNET knowledge platform for disease genomics: 2019 989 update

Screening Bioactives Reveals Nanchangmycin as a Broad 992 Spectrum Antiviral Active against Zika Virus

Calcium Pathways in Human Neutrophils-The Extended Effects of Thapsigargin and 996 ML-9

Calcium influx 998 induces neurite growth through a Src-Ras signaling cassette

pNaKtide Attenuates Steatohepatitis and Atherosclerosis by Blocking 1002

Amplification in C57Bl6 and ApoE Knockout Mice Fed a Western

Gene set enrichment analysis: a knowledge-based approach for 1006 interpreting genome-wide expression profiles

Streptomyces 1009 nanchangensis', a producer of the insecticidal polyether antibiotic nanchangmycin and 1010 the antiparasitic macrolide meilingmycin, contains multiple polyketide gene clusters

Regression of liver fibrosis: evidence and 1014 challenges

Ionophore antibiotic X-206 is a potent and selective inhibitor of 1018 SARS-CoV-2 infection <em>in vitro</em>. bioRxiv

The Na-K-ATPase and calcium-signaling microdomains

Deactivation of hepatic stellate cells during liver fibrosis resolution in mice

Calcium influx via the NMDA 1027 receptor induces immediate early gene transcription by a MAP kinase/ERK-dependent 1028 mechanism

Anti-bacterial and anti-1030 viral nanchangmycin displays anti-myeloma activity by targeting Otub1 and c-Maf

HDAC Inhibitors: Therapeutic Potential in Fibrosis-1033 Associated Human Diseases

Global 1036 epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of 1037 prevalence, incidence, and outcomes

A hepatic stellate cell gene expression signature associated with 1041 outcomes in hepatitis C cirrhosis and hepatocellular carcinoma after curative resection

Focal 1044

Adhesion Kinase Regulates Fibroblast Migration via Integrin beta-1 and Plays a 1045 Central Role in Fibrosis

Focal 1047

Adhesion Kinase Regulates Hepatic Stellate Cell Activation and Liver Fibrosis

The roles of nuclear focal adhesion kinase (FAK) on Cancer: 1050 a focused review

were reverse transcribed from equal amounts of total input RNA. All samples were normalized 1219to the mean value of its own control group before they were combined for each of the reference 1220 mRNAs. Each dot represents the result from one sample, and bars represent mean ± standard 1221 deviation (sd) of all the tested samples. The value of sd is indicated above each mRNA. 1222 HSCs from human donor 2 were treated with compounds at indicated concentrations. Each 1226 dot represents one well (n=2 technical replicates per concentration). Curves were generated 1227 by fitting the data into a sigmoidal equation. For digitoxigenin and digoxin, the highest two or 1228 three concentrations respectively were not used for fitting the curves due to the toxicity at 1229 these concentrations. If a compound was tested more than one time, only one representative 1230 curve is shown here, while the average score from all curves was used for ranking the 1231 compounds. Results from each individual experiment and data from all replicates are in 1232Supplementary Table 6 . 1233