Activating transcription factor 6 is necessary and sufficient for alcoholic fatty liver disease in zebrafish

Abstract

Fatty liver disease (FLD) is characterized by lipid accumulation in hepatocytes and is accompanied by secretory pathway dysfunction, resulting in induction of the unfolded protein response (UPR). Activating transcription factor 6 (ATF6), one of three main UPR sensors, functions to both promote FLD during acute stress and reduce FLD during chronic stress. There is little mechanistic understanding of how ATF6, or any other UPR factor, regulates hepatic lipid metabolism to cause disease. We addressed this using zebrafish genetics and biochemical analyses and demonstrate that Atf6 is necessary and sufficient for FLD. atf6 transcription is significantly upregulated in the liver of zebrafish with alcoholic FLD and morpholino-mediated atf6 depletion significantly reduced steatosis incidence caused by alcohol. Moreover, overexpression of active, nuclear Atf6 (nAtf6) in hepatocytes caused FLD in the absence of stress. mRNA-Seq and qPCR analyses of livers from five day old nAtf6 transgenic larvae revealed upregulation of genes promoting glyceroneogenesis and fatty acid elongation, including fatty acid synthase (fasn), and nAtf6 overexpression in both zebrafish larvae and human hepatoma cells increased the incorporation of 14C-acetate into lipids. Srebp transcription factors are key regulators of lipogenic enzymes, but reducing Srebp activation by scap morpholino injection neither prevented FLD in nAtf6 transgenics nor synergized with atf6 knockdown to reduce alcohol-induced FLD. In contrast, fasn morpholino injection reduced FLD in nAtf6 transgenic larvae and synergistically interacted with atf6 to reduce alcoholic FLD. Thus, our data demonstrate that Atf6 is required for alcoholic FLD and epistatically interacts with fasn to cause this disease, suggesting triglyceride biogenesis as the mechanism of UPR induced FLD.

Figure 1. Atf6 is required for alcoholic steatosis.

A: qPCR analysis of UPR sensors atf6, perk, ire1a, and atf4 in the livers of control (EtOH-) and 350 mM ethanol-treated (EtOH+) larvae. dCt values were calculated by normalization to rpp0. Fold changes were calculated based on median dCt values. Statistics: paired t-test. *, p<0.05. B: Expression of atf6 in the livers of 350 mM ethanol treated larvae from 0–32 hours. Fold changes were determined by normalizing to control (0 mM EtOH) dCt values. Statistics: one-sample t-test. *, p<0.05. C: Images of oil red O stained whole larvae injected with two different morpholinos targeting atf6 (ATG-targeting, atf6 ATG, and splice-blocking, atf6 SPL) and exposed to either 0 or 350 mM ethanol. Livers are circled in the enlarged boxes. D: Steatosis incidence based on scoring oil red O stained larvae at 5.5 dpf. Statistics: chi-square with Fisher's Exact Test. *, p<0.05. “Total n” and “clutch n” corresponds to the number of larvae and number of clutches scored.

Figure 2. nAtf6 overexpression is sufficient to drive steatosis.

A: Heatmap of upregulated UPR effector genes in nAtf6 transgenic larvae. The log values based on mRNA-Seq analysis and the median fold change in 3–6 liver samples assessed by qPCR at 5 dpf and 14 dpf. B: Most Atf6 targets are upregulated in the liver in response to ethanol. Genes identified as part of the UPR are significantly induced in the liver of nAtf6 TG and ethanol treated larvae (log value ≥ 0.2; see Table S2). C: Oil red O stained WT and nAtf6 transgenic larvae. Livers are circled in the enlarged boxes. D: Steatosis incidence based on scoring of whole mount oil red O stained larvae at 4, 5, and 5.5 dpf. Statistics: chi-square with Fisher's Exact Test. *, p<0.05. “Total n” and “clutch n” corresponds to the number of larvae and number of clutches scored. E: Hepatic triglyceride levels in extracts from pooled livers from WT and nAtf6 transgenic larvae at 5 dpf and 14 dpf, and from single adult livers (∼9 months). Statistics: unpaired t-test. *, p<0.05. Median fold changes are noted.

Figure 3. Glyceroneogenesis and fatty acid elongation pathways are dysregulated by nAtf6 overexpression.

Schematic of the glycolysis, gluconeogenesis and glyceroneogenesis pathway (A, adapted from WikiPathways) and heatmap of associated genes (B). Upregulated genes are colored in shades of orange, downregulated genes are colored in shades of green, genes that are unchanged are white and those genes that did not appear in mRNA-Seq analyses are colored gray. Log values and median fold changes from mRNA-Seq analysis and qPCR, respectively, are noted.

Figure 4. De novo lipogenesis is enhanced in vitro and in vivo by nAtf6 overexpression.

A: ¹⁴C-acetate is preferentially incorporated into lipids in 5 dpf nAtf6 TG larvae compared to WT. The percent of ¹⁴C in lipid fraction was divided by ¹⁴C measured in the unextracted lysate from whole fish. The median fold change is noted; n =  number of samples of pooled larvae. B: Oil red O staining of HepG2 cells transfected with GFP or nATF6. Bar  = 10 µm. C: Quantification of lipid droplet number and area (square pixels). Statistics: unpaired t-test. *, p<0.05; n indicates the number of cells quantified over 20 fields from 2 independent experiments. D: Incorporation of ¹⁴C-acetate into lipids by HepG2 cells transfected with GFP or nATF6 normalized to protein. The median fold change is noted; n =  number of separate batches of cells analyzed in 2 independent experiments.

Figure 5. Fasn functions downstream of Atf6 to cause steatosis independent of Srebp activation.

A: qPCR analysis of fasn expression in atf6 morphants and uninjected larvae treated with EtOH for 32 hours. B: qPCR analysis of fasn expression in nAtf6 TG larvae and HepG2 cells transfected with nATF6. Statistics: unpaired t-test. *, p<0.05. C: Quantification of whole mount oil red O staining of nAtf6 TG larvae injected with scap MO. “Total n” and “clutch n” corresponds to the number of larvae and number of clutches scored, respectively. Statistics: chi-square with Fisher's Exact Test. *, p<0.05. D: Quantification of oil red O staining in nAtf6 TG larvae injected with fasn MO. E: Quantification of oil red O staining in atf6/fasn double morphants treated with 350 mM EtOH for 24 hours. Statistics: chi-square with Fisher's Exact Test. *, p<0.05.

Figure 6. Working model by which Atf6 functions as a positive regulator of alcoholic steatosis.

Genetic tools used in this study are illustrated in red.