MCLR-elicited hepatic fibrosis and carcinogenic gene expression changes persist in rats with diet-induced nonalcoholic steatohepatitis through a 4-week recovery period

Abstract

Nonalcoholic steatohepatitis (NASH) causes liver extracellular matrix (ECM) remodeling and is a risk factor for fibrosis and hepatocellular carcinoma (HCC). Microcystin-LR (MCLR) is a hepatotoxin produced by fresh-water cyanobacteria that causes a NASH-like phenotype, liver fibrosis, and is also a risk factor for HCC. The focus of the current study was to investigate and compare hepatic recovery after cessation of MCLR exposure in healthy versus NASH animals. Male Sprague-Dawley rats were fed either a control or a high fat/high cholesterol (HFHC) diet for eight weeks. Animals received either vehicle or 30 μg/kg MCLR (i.p: 2 weeks, alternate days). Animals were euthanized at one of three time points: at the completion of the MCLR exposure period and after 2 and 4 weeks of recovery. Histological staining suggested that after four weeks of recovery the MCLR-exposed HFHC group had less steatosis and more fibrosis compared to the vehicle-exposed HFHC group and MCLR-exposed control group. RNA-Seq analysis revealed dysregulation of ECM genes after MCLR exposure in both control and HFHC groups that persisted only in the HFHC groups during recovery. After 4 weeks of recovery, MCLR hepatotoxicity in pre-existing NASH persistently dysregulated genes related to cellular differentiation and HCC. These data demonstrate impaired hepatic recovery and persistent carcinogenic changes after MCLR toxicity in pre-existing NASH.

1. INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is a progressive disease with a global prevalence of ~25% (Younossi et al., 2016). Fibrogenesis in nonalcoholic steatohepatitis (NASH), an advanced form of NAFLD, is particularly concerning because fibrosis stage is associated with development of both hepatocellular carcinoma (HCC) and cirrhosis, and is one of the most reliable predictors of liver-related mortality (Anstee et al., 2019; Dulai et al., 2017). Identification of exogenous factors that drive NAFLD progression to more malignant stages is of utmost importance. Key factors that contribute to NAFLD progression include chronic hepatic stress from a poor diet/life-style and exposure to exogenous hepatotoxins (Arman et al., 2019; Buzzetti et al., 2016; He et al., 2017; Wahlang et al., 2014, 2013). Continual or repeated hepatic stress from these factors causes fibrosis through cycles of extracellular matrix (ECM) remodeling (Bataller and Brenner, 2005; Hynes, 2009; Lotersztajn et al., 2005; Poole and Arteel, 2016). Emerging evidence indicates the combination of diet-induced NAFLD and environmental hepatotoxin exposure can exacerbate NAFLD through inflammation and ECM remodeling (Arman et al., 2019; Cichocki et al., 2017; Wahlang et al., 2014). Further characterization of the interactions between poor diet and hepatotoxins is needed to guide toxin exposure limits and risk assessment.

Microcystins are a group of more than 80 potent hepatotoxins produced by freshwater cyanobacteria (Svirčev et al., 2019). Common microcystin exposure routes in humans are through consumption of contaminated water, fish and vegetables, and recreational activities (Codd et al., 1999). The most well documented mechanism of microcystin toxicity is inhibition of the serine-threonine phosphatases, protein phosphatase 1/2A (PP1/2A) (Runnegar et al., 1993). Microcystin toxicity also causes cell death through oxidative stress and cytoskeletal disruption (Ding et al., 2001). Microcystin-LR (MCLR) is the most common and possibly one of the most potent microcystin variants known to elicit fibrosis and carcinogenesis. Recently, MCLR has been investigated for its role in NAFLD pathogenesis and progression (Arman et al., 2019; Frangež et al., 2000; Guzman and Solter, 1999; Solter et al., 1998). MCLR alone is reported to cause bridging fibrosis after 48 μg/kg intraperitoneal injection for 28 days in rats (Guzman and Solter, 1999) and a NASH-like phenotype after 40 μg/kg oral exposure for 90 days in mice (He et al., 2017). The role of MCLR as a tumor promoter is also evident: multiple epidemiological studies reported associations between HCC incidence and microcystin concentrations in serum or in surface drinking water (Fleming et al., 2002; Zheng et al., 2017). Notably, these epidemiological studies did not consider NAFLD in their analyses. In addition, several preclinical models found MCLR acts as a tumor promoter after diethylnitrosamine initiation, as evident by glutathione S-transferase Pi (Gstp) positive foci in the liver (Nishiwaki-Matsushima et al., 1992; Sekijima et al., 1999). These epidemiological and preclinical data led to the classification of MCLR as a possible human carcinogen by the International Agency for Research on Cancer (Svirčev et al., 2009; Ueno Y et al., 1996; Zheng et al., 2017). To date, very few studies have investigated the combined effect of a NAFLD-inducing diet with MCLR exposure. In those that have, the combination of MCLR toxicity along with a predisposed NASH condition in rodents causes a burnt-out NASH phenotype (i.e., decreased steatosis and increased inflammation and fibrosis) (Arman et al., 2019; Clarke et al., 2019; Lad et al., 2019). As indicated above, ECM remodeling and fibrosis are important liver repair mechanisms, but are also risk factors for NAFLD progression if unresolved. Whether the burnt-out NASH phenotype observed after MCLR toxicity will repair or persist after a recovery period has not previously been examined, and this is the primary focus of this study.

Cyanobacterial blooms and MCLR exposure occur on a seasonal basis and, therefore, to determine the long-term pathological effects of MCLR, hepatic recovery after short-term MCLR exposure must be assessed. Notably, liver damage caused by 20 μg/kg MCLR (i.p.) every 48 hours for 1 month recovered 30 days after MCLR exposure in healthy animals (Andrinolo et al., 2008; Sedan et al., 2010). The current study modeled a probable diet and MCLR exposure scenario by administering the HFHC diet before MCLR exposure to induce pre-existing NASH, administering MCLR for a short duration, and then allowing for a recovery period after MCLR toxicity, with the diet continuing during and after MCLR exposure. Using this design, the current research discovered that the fibrotic and carcinogenic effects of MCLR-elicited liver injury persisted in pre-existing NASH. These data highlight that MCLR toxicity in pre-existing NASH may contribute to the overall burden of NAFLD and promote progression to end-stage liver disease and HCC.

2. MATERIALS AND METHODS:

2.1. Ethics statement

Handling, care, and maintenance of the animals was done in the Association for the Assessment of Laboratory Animal Care International accredited program of the Laboratory Animal Resources facility of Washington State University (WSU), Spokane. All aspects of the animal research were approved by the Institutional Animal Care and Use Committee at Washington State University (approval code 04937) and all the methods are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org/arrive-guidelines).

2.2. Animals and treatments:

Eight-week-old male Sprague-Dawley rats (n=72) were purchased from Envigo (Huntingdon, Cambridgeshire, UK). All animals were maintained in 12 h light and dark cycles for the duration of the study. The study design is shown in supplemental figure 1 and described here. Animals were divided into two groups (n=36 per diet group) and fed (ad-libitum) either a control diet (Catalog# 518754, Dyets Inc., Bethlehem, PA, USA) or an HFHC diet (Catalog# D06061401, Research Diets, New Brunswick, NJ, USA) for 8 weeks. The HFHC diet is an established diet to replicate a NASH-liver pathology (Hebbard and George, 2011). Following the 8-week diet period, animals were administered vehicle (0.9% saline, 0.09% ethanol; 5 mL/kg) or MCLR (30 μg/kg; 5 mL/kg; Catalog# 10007188 Cayman Chemicals, Ann Arbor, MI, USA) via intraperitoneal injections (n=18 per diet and treatment group) every 48 h for 2 weeks. The exposure scenario to MCLR is a commonly implemented route to investigate liver damage and tumor promotion in the liver (Arman et al., 2019; Clark et al., 2007; Nishiwaki-Matsushima et al., 1992; Zhao et al., 2012). It also helps in isolating the hepatic exposure from the gastrointestinal toxicities. In addition, there is an apparent disparity in oral absorption of microcystins between preclinical species and humans, because ~0.04 μg/kg oral exposure in humans compared to 20 μg/kg intravenous exposure in rats (500-fold difference) only produced 5-fold lower plasma concentrations in humans (0.39 ng/mL versus 2 ng/mL) (Chen et al., 2009; Clarke et al., 2019). Animals were euthanized at three recovery time points: 24 hours (0 weeks), 2 weeks (Rec. 2 weeks), and 4 weeks (Rec. 4 weeks) (n=6 per diet and treatment group) after the last MCLR administration. Animals continued their respective diets throughout the study. Rats were euthanized by carbon-dioxide asphyxiation and blood was collected from the posterior vena cava into heparinized tubes. Plasma was collected by centrifugation at 10,000 × g for 5 min at 4°C and stored at −80°C until further analysis. Liver tissues were harvested, and portions were formalin fixed or snap frozen.

2.3. Histopathology Analysis:

Formalin fixed liver tissues were processed, paraffin embedded, and sectioned at 4 μm. Sections were stained with hematoxylin and eosin (H&E) or Masson’s-trichrome. The H&E-stained slides were analyzed by a board-certified veterinary pathologist, and based on the severity of the pathologies, the slides were scored from 0–5, with 0 being normal/no change and 5 being the most severe. A detailed description of the grading schemes is provided in the supplemental table 1. Immunohistochemistry (IHC) staining was also done in the formalin fixed tissues to detect the presence of Gstp (Catalog# ab53943, Abcam, Cambridge, UK). Antigen retrieval for the IHC staining was performed by boiling the deparaffinized and hydrated tissue sections in citrate buffer (10mM; pH 6.0) for 20 minutes in a rice cooker. This step was followed by an endogenous peroxidase block incubation to reduce non-specific staining. The tissues were then incubated with primary Gstp antibody (1:200) overnight at 4°C. A Goat-on-Rodent Horseradish Peroxidase (HRP) polymer system (Catalog# GHP516, Biocare Medical, Pacheco, CA, USA) was then used to probe for the protein of interest. 3,3’-Diaminobenzidene (DAB) (Catalog# 76180–670, VWR, Radnor, PA, USA) was used as the chromogen that reacts with HRP and gives a colored precipitate. Slides were counterstained with hematoxylin. Gstp foci area was determined using Image J software version 1.53.

2.4. Plasma Biochemistries:

Plasma biochemistries were analyzed using commercially available kits according to the manufacturer’s protocols. Triglyceride (Catalog# 10010303), glucose (Catalog# 10009582), and alanine transaminase (ALT) (Catalog# 700260) quantification was performed with colorimetric assays, and cholesterol (Catalog# 10007640) quantification was performed with a fluorometric assay (Cayman Chemicals, Ann Arbor, MI, USA). Insulin quantification was performed with an ELISA kit (Catalog# EZRMI-13K, Millipore, Burlington, MA, USA).

2.5. Protein preparations:

Frozen liver tissue samples were pulverized into a fine powder using mortar and pestle and stored at −80°C. Approximately 100 mg of the pulverized tissue was lysed with NP40 lysis buffer containing protease inhibitors (Catalog# PI88665, Thermo Fisher Scientific, Waltham, MA, USA) in a TissuelyzerII (Qiagen, Hilden, Germany). Protein lysate was separated by centrifugation at 10,000 x g at 4°C for 30 minutes and stored at −80°C. Protein concentrations were determined using the Pierce BCA Protein Quantification Assay kit (Thermo Fisher Scientific, Waltham, MA, USA).

2.6. Immunoblot assays:

Twenty micrograms of total protein from liver tissue lysates were prepared in Laemmli buffer with 2.5% β- mercaptoethanol (BME) and heated at 37°C for 30 min. Samples were loaded into 7.5% SDS-PAGE gels and probed individually with the following antibodies: protein phosphatase-2A (PP2A) (1: 2,000 dilution; Catalog# 05–421, Millipore), MCLR (1:2,000 dilution; Catalog# 89154–022, Enzo, Farmingdale, NY, USA), OATP1B2 (1:1,000 dilution; Catalog# 376904, Santa Cruz Biotechnology, Santa Cruz, CA, USA), Bcl-2 (1:1,000 dilution; Catalog# NB100–56101, Novus Biologicals, Littleton, CO, USA), Bid (1:1,000 dilution, Catalog# PA5–29159, Invitrogen, Carlsbad, CA, USA), Caspase-3 (1:1,000 dilution; Catalog# 9664S, Cell Signaling, Danvers, MA, USA) Caspase-9 (1:1,000 dilution, Catalog# 9508S, Cell Signaling), and alpha-smooth muscle actin (1:1,000 dilution, Catalog# ab7817, Abcam) (Arman et al., 2021, 2019; Clarke et al., 2019; Kang et al., 2021; Krajewska et al., 2004; Li et al., 2021). The blots were incubated with respective anti-mouse (1:10,000 dilution; Catalog# 20–304, Genesee Scientific, San Diego, CA, USA) or anti-rabbit (1:10,000 dilution; Catalog# 20–303, Genesee Scientific) secondary antibodies in 5% non-fat dry milk in TBS-T for 1 h at room temperature. Densitometry was performed using Image Lab (Bio-Rad, Standard Edition, Version 6.0.0 build 25). Proteins of interest were normalized to total protein stained with amido black. Total protein normalization is a commonly accepted technique for protein densitometry analysis instead of the single-protein loading control (Aldridge et al., 2008; Arman et al., 2021, 2019).

2.7. Rat liver cytokine quantification:

Total liver homogenates were diluted to 5 mg/mL and analyzed for the following cytokines by Eve Technologies (Calgary, Canada) in a multiplex magnetic bead assay: interleukin (IL)-2, IL-4, IL-6, IL-10, IL-17A, Chemokine (C-C motif) ligand (CCL)11, chemokine (C-X-C motif) ligand 1 (CXCL)1, CXCL2, interferon gamma (IFNγ), VEGF, IL-18, CX3CL1, CCL5, CXCL5 and IL-1β. All sample concentrations were within the standard range and exhibit >85% recovery.

2.8. RNA sequencing (RNA-Seq), data processing, and statistical analysis:

RNA was extracted with TRIzol reagent (Thermo Fisher Scientific) from frozen liver tissue according to the manufacturer’s protocol and submitted to the Washington State University (WSU), Spokane Genomics Core for further processing and sequencing. mRNA libraries were generated from samples with RNA quality numbers (RQN) ranging from 5.4–10 (median 6.9) using the TruSeq Stranded mRNA Library Prep Kit (Illumina) following standard procedures. Libraries (18pM) were clustered in high-output flow cells and sequenced on a HiSeq 2500 (Illumina) for 100bp from a single end to depths of 33.7–51.1 million (median 40.6) total reads per sample. Initial adapter trimming and fastq formatting was completed with bcl2fastq (v2.17.1.14).

Fastqc (v0.11.6) (Andrews, 2019) and MultiQC (v.1.8) (Ewels et al., 2016) were used for further quality analysis and two issues were noted: incomplete removal of adapter sequences and a fixed signal across all samples at the 100^th base position. After merging files by sample, Trimmomatic (v0.38) (Bolger et al., 2014) was used to remove the 100^th base and any remaining adapters in single-end mode with a custom adapter file and the options -phred33 ILLUMINACLIP:TruSeq_i7adapt_SE-MCLR.fa:2:30:10 CROP:99 was used.

Processed reads were then aligned to genes of the rat genome (Rnor6.0; Ensembl release 98) with STAR (v. 2.7.3a) (Dobin et al., 2013), including the options --outMultimapperOrder Random --outSAMtype BAM, resulting in 82.7–91.2% (median 89.4%) of sample reads uniquely mapped. The reads were quantified with featureCounts (Subread 2.0.0) (Liao et al., 2014) using the arguments -s 2 -t exon -g gene_id, resulting in a final coverage of 70.0–80.1% (median 76.9%) of total reads per sample. Differential gene expression was calculated with DESeq2 (Bioconductor v. 3.10)(Love et al., 2014) with each combination of dose, diet, and the recovery times as a single “treatment” variable using the Wald test, normal log-fold change (LFC) shrinkage, a LFC threshold of 0.585 (1.5-fold change), and the Benjamini-Hochberg p-value correction. Adjusted p-values less than 0.1 were considered significant. Median-normalized gene counts were derived by applying a variance stabilizing transformation from DESeq2 with blind = FALSE and then centering within each recovery period to the median of control diet, vehicle-exposed samples. All gene names reported in the text are the Entrez Gene accession equivalents of the Ensembl gene IDs.

Gene set analysis was performed using the consensus scoring approach and directionality classes of the piano (Bioconductor, v3.10) (Väremo et al., 2013) R package with KEGG pathways (Kanehisa, 2019; Kanehisa et al., 2020; Kanehisa and Goto, 2000) obtained from DAVID 6.8(Huang et al., 2009a, 2009b). Ten statistical tests were originally used to calculate gene set significance but three performed poorly (Wilcoxon rank-sum, mean, and sum tests). Consensus scores and median adjusted p-values were calculated using the remaining seven: Gene set enrichment analysis (GSEA), parametric analysis of gene (PAGE), maxmean statistic, Fisher’s combined probability, reporter features, tail strength, and median tests. DESeq2 Wald test statistics were used as input for GSEA, PAGE, and maxmean statistic tests. All other tests used DESeq2 adjusted p-values as input along with the LFC for directionality. Each individual test was configured with the following options in runGSA(): signifMethod = “geneSampling”, adjMethod = “fdr”, gsSizeLim = c(1, Inf), nPerm = 1e4.

For matrisome analysis, the mouse homologs of all rat genes were identified using the R package biomaRt (Durinck et al., 2009, 2005)along with (Mouse Genome Informatics) MGI identifiers. Those rat genes with MGI identifiers matching mouse matrisome genes, as defined by the Matrisome Project (vAug-2014) (Naba et al., 2016), were considered part of the rat matrisome.

2.9. Additional statistical analysis:

All data, except for the differential expression data, are represented as mean ± SD and were analyzed by two-way ANOVA or one-way ANOVA. The two-way ANOVA was followed by a Sidak multiple comparison post-test to determine the statistical differences due to MCLR exposure within each diet group in the same treatment period. The one-way ANOVA was followed by the Dunnett multiple comparison post-test to compare the significance in recovery after MCLR exposure within each diet group. All analyses were performed using GraphPad Prism 7 software (GraphPad software, INC., La Jolla, CA, USA).

3. RESULTS:

3.1. Body weight, organ weight and clinical chemistry

The final body and liver weights and liver-to-body weight ratios are shown in Table 1. MCLR did not alter body weights in either the control or the HFHC groups. MCLR had no effect on liver weights of the control groups, when compared to their vehicle counterparts. In contrast, liver weights in the HFHC groups increased after MCLR exposure at 0 weeks in comparison to the vehicle counterparts but returned to vehicle levels by the 2-week recovery time point. Liver-to-body weight ratios followed the same pattern as liver weight.

Table 1:

Body and liver weights and liver-to-body weight ratios after MCLR exposure.

Group	Body (g)	Liver (g)	Liver: Body
Control Vehicle
0 weeks	408.16 ± 16.23	10.86 ± 0.28	0.03 ± 0.001
Rec. 2 weeks	422.03 ± 24.75	11.03 ± 0.58	0.03 ± 0.001
Rec. 4 weeks	452.10 ± 27.85 #	11.04 ± 0.98	0.02 ± 0.001 #
Control MCLR
0 weeks	394.38 ± 18.40	9.31 ± 0.60	0.02 ± 0.001
Rec. 2 weeks	424.06 ± 19.74	9.65 ± 0.58	0.02 ± 0.001
Rec. 4 weeks	463.21 ± 28.23 #	10.48 ± 1.04 #	0.02 ± 0.001
HFHC Vehicle
0 weeks	445.53 ± 22.52	19.17 ± 2.58	0.04 ± 0.004
Rec. 2 weeks	451.80 ± 40.64	18.79 ± 2.60	0.04 ± 0.003
Rec. 4 weeks	482.48 ± 30.03	20.07 ± 3.43	0.04 ± 0.004
HFHC MCLR
0 weeks	416.32 ± 32.52	25.18 ± 2.74 *	0.06 ± 0.004 *
Rec. 2 weeks	458.19 ± 25.63	19.54 ± 2.77 #	0.04 ± 0.005 #
Rec. 4 weeks	459.63 ± 38.68	17.33 ± 2.28 #	0.04 ± 0.004 #

Data represent mean ± SD, n = 6 for each group. Two-way ANOVA with Sidak multiple comparison post-test:

^*p-value < 0.05 versus vehicle treatment group within each diet group and recovery period. One-way ANOVA with Dunnett multiple comparison post-test:

^#p-value < 0.05 versus 0 weeks within each diet and treatment group.

The effect of MCLR on markers of liver damage and metabolic dysregulation was analyzed for each diet and recovery period. MCLR elevated plasma ALT at 0 weeks only in the control diet group, which returned to basal levels by the 2-week recovery time point (Suppl. Fig. 2). In the HFHC group, MCLR did not alter plasma ALT at 0 weeks; however, ALT decreased in the 2- and 4- week recovery groups compared to the 0 weeks group (Suppl. Fig. 2). No change in plasma insulin levels were observed (Suppl. Fig. 3A). MCLR decreased plasma glucose levels in both the control and the HFHC groups at 0 weeks (Suppl. Fig. 3B). MCLR increased plasma triglycerides and cholesterol levels at 0 weeks only in the HFHC group (Suppl. Fig. 3C–D). All MCLR-elicited changes in plasma glucose, triglycerides, and cholesterol concentrations returned to basal levels by the 2-week recovery time point.

3.2. Liver pathology

Liver samples were stained with H&E to assess pathology and were scored by a certified veterinary pathologist (Fig. 1 and 2). The HFHC diet increased steatosis and random granulomas (Fig. 1G–I; 2A and C) compared to control diet. At 0 weeks, MCLR exposure decreased the random granulomas in the HFHC group but increased centrilobular mononuclear infiltrates in both the control and the HFHC groups (Fig. 1A vs D and G vs J; 2C–D). In addition, MCLR exposure decreased steatosis in the HFHC group but increased steatosis in the control group at 0 weeks (Fig. 1A vs D and G vs J; 2A). At the 2- and 4-week recovery time points the steatosis and centrilobular inflammation in the MCLR-treated control groups was similar to their vehicle-treated counterparts (Fig. 1B vs E and C vs F; 2A and D). In contrast, the MCLR-mediated changes in inflammatory infiltrates and steatosis in the HFHC groups persisted at the 2- and 4-week recovery time points (Fig. 1H vs K and I vs L; 2A and D). Centrilobular necrosis increased in the control group after MCLR exposure at 0 weeks but resolved at the 2- and 4-week recovery time points (Fig. 1D–F and 2F).

Figure 1: Representative H&E-stained liver samples.

Panels A-F represent control liver samples and G-L represent HFHC liver samples. Arrowheads indicate inflammation, arrows indicate necrosis and (*) represents steatosis. Animals were euthanized at specific times after the final vehicle or MCLR administration (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Original magnification was 100X.

Figure 2: Histology severity scores.

(A) Steatosis, (B) Fibrosis, (C) Random inflammation, (D) Centrilobular inflammation, (E) Random necrosis, (F) Centrilobular necrosis, scores provided by the pathologist. Each point represents an individual animal in the group, n=6 per group. The horizontal lines represent the median for each group. CV= Control vehicle; CM= Control MCLR; HV= HFHC vehicle; HM= HFHC MCLR.

3.3. Markers of MCLR exposure and toxicity

Two important mechanisms for MCLR toxicity are hepatic uptake through OATP1B2, which is the main rodent hepatic transporter responsible for MCLR liver accumulation (Fischer et al., 2005), and covalent binding of MCLR to the catalytic subunit of PP2A (Xing et al., 2006). MCLR decreased OATP1B2 protein expression in both the control and the HFHC groups at 0 weeks but returned to basal levels at the 2- and 4-week recovery time points (Suppl. Fig. 4A). MCLR did not change PP2A protein expression (37 kDa) at 0 weeks in either diet group (Suppl. Fig. 4B). However, a lower molecular weight band (25 kDa) appeared after MCLR exposure at 0 weeks in both control and HFHC groups (Suppl. Fig. 4C) and gradually decreased at the 2- and 4-week recovery time points (Suppl. Fig. 4C and F). PP2A-bound MCLR was detected at the same molecular weights as the PP2A bands (Suppl Fig. 4D–E) but, in contrast to the 25 kDa PP2A band, the 25 kDa PP2A-bound MCLR band disappeared by the 2-week recovery period (Suppl. Fig. 4E–F).

Microcystin induced oxidative stress involves decreased antioxidant enzyme expression and increased lipid peroxidation (Clark et al., 2007; Moreno et al., 2005; Guo et al., 2015; Li et al., 2015). Genes involved in the glutathione pathway and lipid peroxidation were compiled from the RNA-seq dataset. MCLR altered the expression of many glutathione pathway genes at 0 weeks, which returned to basal levels at the 2- and 4- week recovery time points (Suppl. Fig. 5).

MCLR-elicited apoptosis was analyzed by measuring the level of protein expression for several apoptotic factors (Suppl. Fig. 6). MCLR significantly decreased expression of the anti-apoptotic protein Bcl-2 only in the HFHC group at 0 weeks (Suppl. Fig. 6A). MCLR increased expression of the pro-apoptotic protein Bid in both the control and HFHC groups at 0 weeks (Suppl. Fig. 6B). MCLR increased expression of caspases 3 and 9 in the control group at 0 weeks, whereas only caspase 9 increased in the HFHC group at 0 weeks (Suppl. Fig. 6C–D). All MCLR-elicited changes in apoptotic proteins returned to basal levels at the 2- and 4- week recovery time points.

MCLR toxicity also altered liver cytokine levels (Fig. 3). MCLR decreased many cytokines at 0 weeks in both the control and HFHC groups, which returned to basal levels at the 2- and 4- week recovery time points (Fig. 3A–J). IL-18 and CX3CL1 concentrations increased due to MCLR toxicity in both the control and HFHC groups at 0 weeks and returned to basal levels at the 2- and 4- week recovery time points (Fig. 3K). CCL5 concentration was increased by MCLR only in the control group at week 0 (Fig. 3M). CXCL5 and IL-1β expression was altered by MCLR only in the HFHC group at 0 weeks (Fig. 3N–O). The MCLR-elicited changes in IFNγ and VEGF-A persisted till the 2-week recovery time point, while CCL5 and IL-1β persisted in the HFHC group till the 4-week recovery time point.

Figure 3: Liver cytokine levels.

(A) IL-2, (B) IL-4, (C) IL-6, (D) IL-10, (E) IL-17A, (F) CCL11, (G) CXCL1, (H) CXCL2, (I) IFNγ, (J) VEGF-A, (K) IL-18, (L) CX3CL1, (M) CCL5, (N) CXCL5 and (O) IL-1β Data represent mean ± SD, n = 6 for each group. Two-way ANOVA with Sidak multiple comparison post-test: *p-value < 0.05 versus vehicle treatment group within each diet group and recovery period. One-way ANOVA with Dunnett multiple comparison post-test: #p-value < 0.05 versus 0 weeks within each diet and treatment group.

3.4. Differential gene expression

Comparative liver transcriptomic analysis was performed to identify MCLR-induced differentially expressed genes in each diet group relative to their respective vehicle treated groups. MCLR significantly altered expression of 4,198 genes in control animals and 2,962 genes in HFHC animals at 0 weeks, of which most genes (2,311) overlapped between the two diet groups (55% and 78%, respectively) (Fig. 4). The number of differentially expressed genes diminished dramatically after 2 weeks of recovery but only minimally thereafter for both the control and HFHC groups (Fig. 4). At 4 weeks of recovery, the HFHC group had 12 times more differentially expressed genes (139 vs. 11), despite having ~1,200 fewer at 0 weeks of recovery (Fig. 4).

Figure 4: Total differentially expressed genes.

Circles represent the number of differentially expressed genes (control MCLR versus control vehicle; blue) (HFHC MCLR versus HFHC vehicle; red). Overlapping portions represent the genes differentially expressed by MCLR in both control and HFHC groups.

Gene set analysis with the R package ‘piano’ was used to identify KEGG pathways distinctly up- or down-regulated by MCLR exposure. A total of 36 pathways were upregulated by MCLR at 0 weeks (14 in both diets, 15 in control only, and 7 in HFHC only), many of which are associated with cancer or ECM remodeling (Suppl. Table 2). The renin-angiotensin system KEGG pathway was the only pathway persistently upregulated in the HFHC group even at the 4-week recovery time point (Suppl. Table 2). MCLR downregulated 39 KEGG pathways at 0 weeks (18 in both diets, 6 in control only, and 15 in HFHC only), with the majority comprising metabolic pathways (Suppl. Table 3). By the 4-week recovery time point, all downregulated pathways returned to basal levels (Suppl. Table 3).

3.5. Liver fibrosis, hepatic stellate cell (HSC) activation, and matrisome

Liver fibrosis, HSC activation markers, and ECM pathways were explored further. MCLR exposure caused fibrosis in both the control and HFHC groups at 0 weeks as indicated by Masson’s trichrome staining of liver samples and pathological scoring of the H&E-stained slides (Fig. 5D and J and 2B). Fibrosis resolved to a greater degree in the control group compared to the HFHC groups at the 2- and 4-week recovery time points, as evident in scoring (Fig. 2B) and the persistence of bridging fibrosis in the HFHC group (Fig 5E–F, K–L).

Figure 5: Representative liver Masson’s trichrome staining.

Panels A-F represent Control diet liver samples and G-L represent HFHC diet liver samples. The blue stain represents the collagen deposition depicting fibrosis. Animals were euthanized at specific times after the final vehicle or MCLR dose (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Original magnification was 100X.

Expression of genes responsible for HSC activation and fibrosis from the RNA-Seq dataset are shown in Figure 6. MCLR increased actin alpha 2 (Acta2), vascular endothelial growth factor D (Vegfd), transforming growth factor beta-3 (Tgfb3), and platelet derived growth factor D (Pdgfd) expression at 0 weeks in both the control and the HFHC groups when compared to their respective vehicle counterparts (Fig. 6A–D). Vegfd, Tgfb, and Pdgfd remained differentially expressed at the 2- and 4-week recovery time points compared to vehicle treated counterparts only in the HFHC group, whereas Acta2 expression returned to basal level at the 2-week recovery time point. Interferon gamma (Ifng) and interleukin 10 (Il-10) exhibited decreased expression in the HFHC MCLR groups, although there was a high degree of variability in these samples (Fig. 6E and F). Consistent with Acta2 mRNA levels, MCLR increased alpha-smooth muscle actin (α-SMA) protein levels at 0 weeks then returned to basal levels by the 2-week recovery time point (Fig. 6G–H).

Figure 6: HSC activation and fibrosis markers.

(A) Actin alpha 2 (Acta2), (B) vascular endothelial growth factor D (Vegfd), (C) transforming growth factor beta-3 (Tgfb3), (D) platelet derived growth factor D (Pdgfd), (E) interferon gamma (Ifng), and (F) interleukin 10 (Il-10). Gene expression changes were analyzed at each time point the control or HFHC diet fed animals were euthanized after the final vehicle or MCLR dose (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Gene expression was determined by RNAseq and normalized to the median of the control vehicle group for each gene. *Adjusted p-value < 0.05 versus respective vehicle treated group for each gene. (G-H) Western blot for α-SMA. Data represent mean ± SD, n = 6 for each group. Two-way ANOVA with Sidak multiple comparison post-test: *p-value < 0.05 versus vehicle treatment group within each diet group and recovery period. One-way ANOVA with Dunnett multiple comparison post-test: #p-value < 0.05 versus 0 weeks within each diet and treatment group.

The matrisome is comprised of core genes (collagens, proteoglycans, and ECM glycoproteins) and ECM-affiliated genes (ECM regulators, ECM affiliated proteins and secreted factors) (http://matrisomeproject.mit.edu/) and MCLR dysregulated many of these genes in the liver. A comprehensive analysis of the matrisome from the RNA-Seq dataset revealed similar patterns of differential gene expression at 0 weeks after MCLR exposure in the control and HFHC groups (Fig. 7A–D). MCLR dysregulated 926 matrisome genes at 0 weeks in both the control (340 genes, 37%) and HFHC (313 genes, 34%) groups (Fig. 7 and suppl. Table 4). By the 4-week recovery period, 99% of genes in the control group and only 88% of genes in the HFHC group came back to their basal expression levels. Of the 139 genes that were persistently dysregulated in the HFHC-MCLR group (Fig. 4), 39 were matrisome genes (18 core and 21 associated) (Suppl. Fig. 7). Consistent with the Masson’s trichrome staining (Fig. 5), more matrisome genes continued to exhibit differential expression in the HFHC group at the 2- and 4-week recovery time points compared to the control group (Suppl. Fig. 7).

Figure 7: Matrisome differential gene expression.

Control MCLR is compared to control vehicle and HFHC MCLR is compared to HFHC vehicle. The groups labelled are: C= Collagens; Gp= ECM glycoproteins; Pg= Proteoglycans; AP= ECM affiliated proteins; R= ECM regulators; SF= Secreted factors. Log fold change represents the Log₂ Fold Change (A and C) and p-adjusted is the DESeq2 adjusted (Benjamini Hochberg) p-value for differential expression (B and D).

3.6. Carcinogenesis, cellular differentiation, and Gstp

In addition to matrisome genes, many of the persistently dysregulated genes in Fig. 4 are involved in carcinogenesis (e.g., differentiation and metastasis). MCLR exposure increased expression of Sushi repeat-containing protein X-linked 2 (Srpx2), semaphorin 3A (Sema3), protein tyrosine kinase 7 (Ptk7), bradykinin receptor B2 (Bdkrb2), mesothelin (Msln), and gamma-synuclein (Sncg) in both control and HFHC groups at 0 weeks (Fig. 8A–D, G–H). MCLR increased Podoplanin (Pdpn) expression and decreased dual specificity tyrosine phosphorylation kinase 4 (Dyrk4) expression only in the HFHC group at 0 weeks (Fig. 8E–F). All genes remained dysregulated in the HFHC groups during the 2- and 4-weeks recovery time points, whereas all genes, except Srpx2, returned to basal levels in the control groups (Fig. 8).

Figure 8: Differential gene expression of genes involved in carcinogenesis.

(A) Sushi repeat-containing protein X-linked 2 (Srpx2), (B) semaphorin 3A (Sema3), (C) protein tyrosine kinase 7 (Ptk7), (D) bradykinin receptor B2 (Bdkrb2), (E) dual specificity tyrosine phosphorylation kinase 4 (Dyrk4), (F) podoplanin (Pdpn), (G) mesothelin (Msln), and (H) gamma-synuclein (Sncg). Gene expression changes were analyzed at each time point the control or HFHC diet fed animals were euthanized after the final vehicle or MCLR dose (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Gene expression was determined by RNAseq and normalized to the median of the control vehicle group for each gene. *Adjusted p-value < 0.05 versus respective vehicle treated group for each gene.

Cellular dedifferentiation can give rise to sub-populations of cells that exhibit stem cell-like properties (Sell, 1993). Multiple key genes involved in cellular differentiation were analyzed from the RNA-seq data (Fig. 9). Similar to carcinogenesis-related genes, MCLR exposure increased expression of epithelial cell adhesion molecule (Epcam), neurotensin (Nts), brain-expressed X-linked 1 (Bex1), fermitin family member 1 (Fermt1) interleukin 22 receptor subunit alpha 1 (Il22ra1), carboxypeptidase A (Cpa), cluster of differentiation 24 (Cd24), and cluster of differentiation 133 (Cd133) at 0 weeks in the control and HFHC groups (Fig. 9A–H). KIT ligand (Kitlg) expression increased only in the HFHC group at 0 weeks (Fig. 9I). All genes were persistently dysregulated in the HFHC group at the 2- and 4-week recovery time points, whereas all genes returned to basal levels in the control group by week 4 (Fig 9A–I).

Figure 9: Differential gene expression of genes involved in cellular differentiation.

(A) Epithelial cell adhesion molecule (Epcam), (B) neurotensin (Nts), (C) brain-expressed X-linked 1 (Bex1), (D) fermitin family member 1 (Fermt), (E) interleukin 22 receptor subunit alpha 1 (Il22ra1), (F) carboxypeptidase A (Cpa), (G) cluster of differentiation 24 (Cd24), (H) cluster of differentiation 133 (Cd133), and (I) KIT ligand (Kitlg). Gene expression changes were analyzed at each time point the control or HFHC diet fed animals were euthanized after the final vehicle or MCLR dose (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Gene expression was determined by RNAseq and normalized to the median of the control vehicle group for each gene. *Adjusted p-value < 0.05 versus respective vehicle treated group for each gene.

Gstp is a biomarker for neoplastic lesions and is linked to HCC (Andersen et al., 2010). Gstp IHC staining and scoring for the positively stained foci area revealed no significant MCLR-induced development of neoplastic lesions in both the control and HFHC groups at 0 weeks (Fig. 10D and J and Fig. 11). MCLR increased Gstp positive foci area in the HFHC group at the 4-week recovery time point (Fig. 11).

Figure 10: Representative IHC images of Gstp foci.

Panels A-F represent control diet liver samples and G-L represent HFHC diet liver samples. Animals were euthanized at specific times after the final vehicle or MCLR dose (24 hrs [0 weeks], recovery 2 weeks, or recovery 4 weeks). Original magnification was 200X.

Figure 11: Gstp foci area.

Gstp foci area was quantified for each treatment group and time. Data represent mean ± SD, n = 6 for each group. Two-way ANOVA with Sidak multiple comparison post-test: *p-value < 0.05 versus vehicle treatment group within each diet group and recovery period.

4. DISCUSSION:

To understand the factors that drive progression of NASH to fibrosis and/or HCC, it is important to differentiate the persistent versus recoverable pathological mechanisms and to identify exogenous stressors that may modulate these mechanisms. Identification of persistent pathological mechanisms related to diet and toxin exposure could facilitate lifestyle interventions and exposure mitigation strategies to decrease disease progression and burden. The current study investigated hepatic recovery after cessation of MCLR exposure in healthy versus NASH animals. An HFHC diet was provided to the animals to induce the NASH condition, which continued through the end of the study. MCLR was repeatedly administered for a short duration in both the groups: healthy (control diet) and the pre-existing NASH (HFHC diet), followed by a recovery period. The repeated MCLR exposure altered multiple biochemical, histological, and potential carcinogenic endpoints in both the control and the HFHC groups, but most of these changes persisted only in the HFHC groups at the recovery time points. Each of these endpoints and their implications are discussed herein.

Pathological endpoints are critical for assessment of NAFLD progression because liver histopathological examination remains the most reliable diagnostic and staging technique (Drescher et al., 2019). Centrilobular inflammation and fibrosis, which are characteristic of MCLR toxicity (Guzman and Solter, 1999), occurred in both the control and HFHC groups. Centrilobular inflammation persisted only in the HFHC groups, and the severity of fibrosis remained higher in the HFHC groups during the recovery period. A previous study has shown that expression of key genes involved in lipid homeostasis pathways altered after MCLR toxicity, that can lead to decreased steatosis observed in the HFHC group (Arman et al., 2019). In the current study, repeated MCLR exposure also decreased steatosis in the HFHC animals, and it persisted in the recovery time points. These changes in inflammation, fibrosis, and steatosis indicate that the burnt-out NASH phenotype induced by short-term MCLR exposure persisted through the 4-week recovery period. This phenotype is particularly pathogenic and is a growing cause for orthotopic liver transplantation and associated with HCC development (Van der Poorten et al., 2013; Yoshioka et al., 2004). The unresolved MCLR toxicity in the HFHC groups after 4 weeks of recovery indicates a potential exacerbation of pre-existing NASH and the continual stress of the HFHC diet may have impaired liver repair mechanisms (Jun and Lau, 2018), although the exact mechanism of action is difficult to ascertain at this point. These data indicate short-term MCLR can drive NASH to an irreversible burnt-out phenotype and may increase the risk of disease related morbidity and mortality.

Fibrosis is a feature of burnt-out NASH, contributes to cirrhosis and HCC development in NAFLD, and is caused by continuous ECM remodeling of matrisome components (Baiocchini et al., 2016; Lee and Friedman, 2011; Olczyk et al., 2014; Schultz and Wysocki, 2009). Regression of fibrosis can occur after removal of the causative agent (Kisseleva and Brenner, 2020). The persistent dysregulation of multiple matrisome genes in the HFHC group at 4-weeks of recovery further supports that continuous stress from the HFHC diet impairs liver repair after MCLR exposure. A major event that triggers the onset of fibrosis through ECM remodeling is the trans-differentiation of quiescent HSCs into activated myofibroblasts (Lotersztajn et al., 2005; Schwabe et al., 2020; Tsukada et al., 2006). The current data indicate MCLR toxicity in the HFHC-diet group shifts gene expression towards HSC activation. The genes Acta2 and Vegfd are critical for HSC activation, and Vegfd also promotes inflammation in the damaged tissues (Claveria-Cabello et al., 2020; Yoshiji, 2003). The genes Tgfb3 and Pdgfd encode for two profibrogenic cytokines, which contribute to fibrosis development by increasing Acta2 synthesis and by stimulating the proliferation and migration of activated HSCs, respectively (Dewidar et al., 2015; Ying et al., 2017). In addition, the anti-inflammatory cytokines IFNγ and IL-10, which participate in the prevention of fibrosis by suppressing HSC activation and promoting their apoptosis (Luo et al., 2013; Zhang and Wang, 2006), were decreased after MCLR exposure. Although Acta2 regressed to the basal levels in both control and HFHC, Vegfd, Tgfb3, and Pdgfd were persistently dysregulated in the HFHC diet at the 4-week recovery period. These changes in HSC activation markers confirm that MCLR toxicity activates HSCs and that they become deactivated during recovery.

Inflammation is another important feature of NASH and plays important roles in liver fibrosis and HCC development. In the current study, MCLR clearly caused persistent histopathological inflammation in the HFHC group, but closer analysis of cytokine levels in the liver revealed a surprising trend. Many of the proinflammatory cytokines decreased at week 0, then returned to basal level during recovery for both the control and HFHC groups. Although this may be the first report of liver cytokines decreasing after MCLR toxicity, this phenomenon has been observed in lymphocytes, spleen, serum, and intestine (Cao et al., 2019; Shi et al., 2004; Yea et al., 2001; Yuan et al., 2012). For example, IL-2, IL-4, and IL-10 decreased in the spleen of mice treated with cyanobacterial extract (Shi et al., 2004). Another study reported decreased TNF-α, IL-8, and IL-1β levels in the intestine of mice treated orally with MCLR (Cao et al., 2019). Importantly, these cytokines tended to increase at lower MCLR exposures but decrease at higher exposures and/or they exhibit variable levels at different time points. The precise mechanism for these counter-intuitive results is not clear, but it has been postulated that MCLR may decrease cytokine mRNA stability (Yea et al., 2001). The only two cytokines that exhibited persistent changes through the 4-week recovery time point were CCL5 and IL-1β. Surprisingly, although both of these proinflammatory cytokines were increased by the HFHC diet, MCLR persistently decreased their levels below the respective vehicle groups during the recovery time points. These cytokine data provide further evidence that, although MCLR toxicity causes histological inflammation, it may produce a transient immunosuppressive cytokine environment at some point after MCLR exposure. More research is needed to determine if the decreased cytokine levels in the current study was due to the specific MCLR dose used or the time point when the livers were collected (24 hours after the last exposure).

In contrast to the histopathological endpoints that persisted through the recovery periods in the HFHC group, KEGG pathway analysis revealed many upregulated pathways related to disease progression and many downregulated pathways associated with metabolism returned to baseline levels in both groups. Although the KEGG pathways associated with cancer were no longer significantly upregulated during the recovery period, many of the 139 genes that remain dysregulated have been associated with HCC (see next paragraph for further discussion). MCLR is also known to dysregulate liver metabolic pathways both in vitro and in vivo (Biales et al., 2020; He et al., 2012; Zhao et al., 2015). A previous publication reported that short-term MCLR exposure decreased hepatic valine, tyrosine, phenylalanine, leucine, isoleucine, and arginine levels and disrupted glutathione homeostasis (He et al., 2012). The present study demonstrated transient downregulation of KEGG pathways involving lipid metabolism, amino acid metabolism, bile acid biosynthesis, and amino acid metabolism that returned to baseline levels during the recovery period. Thus, these data indicate that gene expression for important disease and metabolism pathways elicited by MCLR did not persist beyond the MCLR exposure period and are not involved in the impaired recovery in the HFHC group.

NASH is a risk factor for HCC development, with the annual incidence of HCC among NASH patients ranging from 2.4–12.8% (Anstee et al., 2019). In addition, chronic MCLR exposure affects gene expression and may contribute to carcinogenesis (Clark et al., 2008, 2007; Dias et al., 2014), but the carcinogenic potential of short-term MCLR exposure in the context of pre-existing NASH has not been investigated before. Closer analysis of the 139 genes that remain dysregulated in the HFHC group reveal many associated with HCC. Sema3 and Ptk7, which are hallmarks or prognostic markers of HCC development (Li et al., 2017; Zou et al., 2019), were persistently upregulated in the HFHC group after MCLR exposure. In addition, Sema3, Srpx2, Bdkrb2, Pdpn, and Msln, which have tumorigenic potential and are associated with HCC malignancy (Cioca et al., 2017; Li et al., 2020, 2017; Lin et al., 2017; Quintanilla et al., 2019; Yu et al., 2010; Zhao et al., 2017; Zhou et al., 2019), were persistently upregulated only in the HFHC group after MCLR exposure. Sncg, which has been linked to onset of cirrhosis and the risk for HCC (Zhao et al., 2006), was also upregulated only in the HFHC group after MCLR exposure. In contrast, Dyrk4 was persistently downregulated by repeated MCLR exposure in HFHC group, which is consistent with its downregulation in HCC (Boni et al., 2020). A key feature of malignant cancer is cellular dedifferentiation and acquisition of stem cell-like properties that facilitate tumor proliferation and recurrence (Niu et al., 2017). Epithelial-mesenchymal transition (EMT) is an important process in dedifferentiation where polarized epithelial cells assume a mesenchymal phenotype and acquire enhanced migratory and invasive properties (Kalluri and Weinberg, 2009). Epcam, Nts, Fermt1 and Cd24 can promote EMT in several cancer types(Liu et al., 2014, 2017; Wang et al., 2018; Ye et al., 2016) and Bex1, Il22ra1, Cpa, Cd13, and Kitlg upregulation can produce stem cell-like properties in cancer cells (He et al., 2018; Lan et al., 2013; Maaninka et al., 2013; Mavila and Thundimadathil, 2019; Yang et al., 2020; Zhang et al., 2019). All these genes were persistently upregulated by MCLR only in the HFHC group. These data suggest that, despite a similar initial response to repeated MCLR exposure, only the HFHC group exhibited unresolved carcinogenic gene expression changes at the 4-week recovery time point that may contribute to HCC development in NASH.

Neoplastic lesions, which are abnormal masses of cells with aberrant growth patterns, faster cell division rates, and longer life spans, are also involved in carcinogenesis (Pérez-Carreón et al., 2006). Preneoplastic lesions arise early in carcinogenesis, are linked to HCC development, and can be identified using Gstp IHC staining (Andersen et al., 2010). Increased Gstp foci area has been reported with MCLR toxicity after diethylnitrosamine initiation (Nishiwaki-Matsushima et al., 1992). In the current study repeated MCLR exposure did not cause significant increase in the Gstp positive foci area in either of the diet groups. Interestingly, Gstp positive foci area continued to increase in the HFHC group through the 4-week recovery time point. This suggests that, despite no diethylnitrosamine initiation and a recovery time-period after MCLR exposure, the HFHC group still developed neoplastic lesions that may lead to HCC.

Endpoints associated with MCLR toxicity did not always affect both the diet groups (control and HFHC) or always recovered in the control group alone. For example, changes in organ weight and the organ-to-body weight ratio are common features of organ toxicity (Michael et al., 2007), but liver weights increased only in the HFHC group after repeated MCLR exposure despite clear toxicity in both groups. These data suggest transient hepatomegaly, which is commonly associated with xenobiotic hepatotoxicity, may occur after MCLR toxicity in pre-existing NASH (Guha Mazumder et al., 1988; Huang et al., 2005; Malaguarnera, 2012). OATP1B2 protein expression decreases after MCLR exposure (Arman et al., 2019), but the current study demonstrated that it returned to basal levels in both diet groups by the 2-week recovery period. Endpoints such as ALT, glucose, insulin, triglycerides, and cholesterol were altered by repeated MCLR exposure in the HFHC group but returned to basal levels at the recovery time points, indicating that biomarkers of hepatotoxicity and metabolic dysregulation were only transiently affected by MCLR. In addition, pro-apoptotic protein expression patterns were observed in both the control and HFHC groups after repeated MCLR exposure but not during the recovery time points, indicating that apoptosis was primarily related to MCLR toxicity rather than the HFHC diet. Finally, although more genes remained differentially expressed in the HFHC group compared to the control at the recovery time points, a majority of differentially expressed genes for both diet groups returned to basal levels during recovery. The distinct effects of MCLR toxicity in healthy versus NASH animals suggest a disparity in the health risks associated with MCLR exposure.

5. CONCLUSION:

These data indicate that, although markers of liver toxicity (e.g., ALT, plasma biochemistries) were more readily reversed, detrimental chronic responses (e.g., fibrosis, carcinogenicity) persisted during recovery from MCLR toxicity in the HFHC groups. Therefore, MCLR may act in conjunction with other hepatic stressors, such as poor diet/lifestyle, and contribute to the HCC development in chronic disease conditions, such as NASH. More research is needed to determine whether the fibrotic and carcinogenic changes that occurs in NASH after MCLR toxicity will accelerate HCC development.

Highlights.

• Recovery from MCLR toxicity did not resolve burnt-out NASH phenotype in NASH.

• MCLR-elicited fibrosis persisted through 4 weeks of recovery in NASH.

• MCLR persistently dysregulated cancer-related genes through 4 weeks of recovery in NASH.

ABBREVIATIONS:

Acta2

Actin alpha 2

AKT

Protein Kinase B

ALT

Alanine aminotransferase

Bcl2

B-cell lymphoma 2

Bdkrb2

Bradykinin receptor B2

Bex1

Brain expressed X-linked 1

Bex4

Brain expressed X-linked 4

Bid

BH3 interacting domain

Chemokine (C-C motif) ligand 11

CCL11

1Cd133

Cluster of differentiation 133

Cd24

Cluster of differentiation 24

Cpa

Carboxypeptidase A

Chemokine (C-X-C motif) ligand 1

CXCL1

DAB: 3,3’

Diaminobenzidene

Dyrk4

Dual specificity tyrosine phosphorylation kinase 4

ECM

Extracellular matrix

Epcam

Epithelial cell adhesion molecule

Fermt1

Fermitin family member 1

Gstp

glutathione S-transferase Pi

H&E

Hematoxylin and eosin

HCC

Hepatocellular carcinoma

HFHC

High fat/high cholesterol

HRP

Horseradish peroxidase

HSC

Hepatic stellate cells

Ifng

Interferon gamma

IHC

Immunohistochemistry

IL-2

Interleukin 2, IL-4: Interleukin 4

IL-6

Interleukin 6

IL-10

Interleukin 10

IL-17A

Interleukin 17A

Il22ra1

Interleukin 22 receptor subunit alpha 1

KEGG

Kyoto Encyclopedia of Genes and Genomes

itlg

Kit ligand

MCLR

Microcystin-LR

Msln

Mesothelin

NAFLD

Nonalcoholic fatty liver disease

NASH

Nonalcoholic steatohepatitis

Nts

Neurotensin

OATP1B2

Organic anion transporting polypeptides 1B2

Pdgfd

Platelet derived growth factor D

Pdpn

Podoplanin

PI3

Phosphatidylinositol 3-kinase

PP2A

Protein phosphatase

Ptk7

Protein tyrosine kinase 7

Sema3

Semaphorin 3A

Sncg

Gamma-synuclein

Srpx2

Sushi repeat-containing protein X-linked 2

Tgfb3

Transforming growth factor beta 3

VEGF-A

Vascular endothelial growth factor A

Vegfd

Vascular endothelial growth factor D