Investigating the role of lipid genes in liver disease using models of steatotic liver disease in zebrafish (Danio rerio).

Steatotic liver disease (SLD)1 is a heritable trait2 with a global estimated prevalence of 32.4%.3 It often remains undiagnosed until manifesting as steatohepatitis due to inflammation, which can progress to fibrosis, cirrhosis, liver failure and hepatocellular carcinoma. At early stages, weight loss is advised to prevent and treat SLD, while at later stages, liver transplantation is required for survival. No drugs have been approved for SLD treatment or prevention of disease progression. As its prevalence is increasing globally—due to the obesity and diabetes epidemics—SLD is projected to become the leading cause of liver transplantation within the next 10 years.4 Moreover, SLD increases the risk of cardiovascular outcomes, so efficient medication for its treatment and prevention of disease progression are desperately needed. Genome-wide,5, 6 exome-wide7 and proteomic6 association studies have identified at least 26 DNA regions (loci) that are associated with liver fat or the odds of SLD.5-7 A subset of these are also associated with cirrhosis (flagged by DNA variants in/near PNPLA2, PNPLA3, MBOAT7, TM6SF2, MARC1, GPAM, APOE)6, 7 and hepatocellular carcinoma (flagged by DNA variants in/near PNPLA2, PNPLA3, TM6SF2, APOE), with effect sizes for hepatocellular carcinoma risk proportional to those for liver fat.6 In line with this, Mendelian randomisation analyses show causal effects of SLD on hepatocellular carcinoma.8 Half of this causal effect is mediated by severe fibrosis, but SLD also causes hepatocellular carcinoma in patients without fibrosis through insulin resistance and inflammation.8, 9 Examples from other cardiometabolic diseases illustrate that some GWAS-identified loci harbour genes encoding the molecular targets of existing medications, for example PPARG (thiazolidinediones) and ABCC8 and KCNJ11 (sulphonylureas) for diabetes; and HMGCR (statins) and PCSK9 (PCSK9 inhibitors) for LDLc and coronary artery disease. Hence, by functionally characterising candidate genes for SLD, we will likely not just (1) increase our understanding of disease aetiology at a molecular level but also (2) identify targets that can be translated into efficient medication to treat SLD and prevent disease progression. So far, only a handful of genes have functional evidence of being robustly implicated in SLD. This in part reflects the lack of model systems that: (1) capture the complex interactions that occur between relevant cell types, tissues and organs in a live, intact vertebrate organism; (2) allow quantification of relevant phenotypes at high resolution; (3) facilitate the throughput required to systematically characterise the role of many candidate genes; and (4) yield results that are relevant for the disease in humans. Thanks to their well-annotated genome, with orthologues of ~82% of disease-related genes10 and external fertilisation of eggs, zebrafish (Danio rerio) may provide an opportunity to tackle these challenges. Earlier this year, Li et al. showed that compared with zebrafish fed on a control diet, adult male zebrafish challenged with a high-fat diet for 30 days had significantly higher BMI, hepatic triglyceride and total cholesterol levels, liver fat area, hepatic oxidative stress, inflammatory lesions, ballooning, apoptosis, serum aspartate and alanine aminotransferase (AST and ALT) levels.11 These results provide important proof of concept evidence in favour of using zebrafish as a model organism for functional studies in SLD. However, experiments in adult zebrafish are not an obvious choice for systematic genetic screens since they do not support the throughput required to study the role of a large number of genes in a large enough number of individuals per gene to ensure adequate statistical power. Since most organs—including the liver—are functional by Day 5 post-fertilisation in zebrafish, and since zebrafish larvae are optically transparent during early development, efforts to develop and validate SLD model systems for systematic image-based genetic screens in zebrafish larvae seem worthwhile. In this edition of Liver International, Shihana et al.12 used bright field and fluorescence imaging, biochemistry and qPCR to examine the role of CRISPR/Cas9-induced mutations in the zebrafish orthologues of three GWAS-identified candidate genes for alcohol and SLD-related cirrhosis (FAF2, TM6SF2 and PNPLA3) on SLD-related traits in zebrafish larvae. The authors show positive, non-linear statistical interactions of 2-day challenges—that is a medium of 2% ethanol or a high-fat diet—with CRISPR/Cas9-induced mutations in faf2, tm6sf2 and pnpla3, for hepatic accumulation of lipids and neutrophils, as well as for the expression of several genes involved in lipid metabolism (hmgcra, fabp10a, notch1, serpb1, echs1). Negative interactions were observed for the expression of genes involved in triglyceride hydrolysis (acox1) and fatty acid oxidation (atgl). The results of Shihana et al. illustrate the potential of gene perturbation experiments in zebrafish larvae to improve our understanding of the role of genes in SLD-related traits. Moreover, the study provides a good starting point for future efforts aiming to systematically characterise the role of candidate genes in vivo using zebrafish larvae. Such future efforts should ideally also benefit from several (recent) advances that facilitate further upscaling and more robust conclusions. First, advances in automated positioning, orienting and imaging of non-embedded, live, intact zebrafish larvae13 facilitate: (i) more standardised acquisition of images; (ii) a higher throughput; (iii) a larger sample size; and (iv) the possibility to perform longitudinal studies with images obtained before, during and after metabolic challenges and treatments. Second, a large number of images acquired in a standardised manner enables the training of deep learning-based neural networks for robust, unbiased and automated image analysis.14, 15 Third, the use of additional fluorescently labelled transgenes16-18 aids in the visualisation and quantification of further traits related to SLD progression, like hepatic infiltration by macrophages, stellate cell activation and potentially even fibrosis. In terms of gene editing, future efforts should ideally ensure that crispants and sibling controls both undergo gene editing at a control gene,14 since (1) sham injections with just Cas9 result in differential expression of >1000 genes by Day 5 post-fertilisation, including genes involved in wound healing19; and (2) DNA cleavage per se induces an innate immune response.20 Moreover, as a result of a whole-genome duplication in the evolution of teleost fish,21 zebrafish have multiple orthologues for a substantial proportion of human genes, including TM6SF2. It has been shown that CRISPR/Cas9-induced mutations can upregulate the expression of genes with sequence similarity to the CRISPR/Cas9-edited gene.22 When aiming to assess the role of a human candidate gene in disease-related traits, it is therefore arguably advisable to perturb all zebrafish orthologues of a human candidate gene simultaneously in future gene perturbation studies. In summary, systematically characterising candidate genes for a role in SLD using CRISPR/Cas9 and live fluorescence imaging is anticipated to increase our understanding of SLD aetiology at a molecular level and identify targets that can be translated into efficient medication for SLD treatment and prevention of disease progression. Once a larger number of genes have been functionally characterised in a systematic manner, this data can likely be used to also quantify the extent to which disease-related pathway(s) and mechanisms are conserved across species. Marcel den Hoed is the co-founder (2022) of a contract research organization (Veyviser A/B) that provides target validation as a service. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.