Deconvolution of Bulk RNAseq Data from Human Liver Samples Reveals a Metabolic Switch in Periportal Hepatocytes in Alcoholic Hepatitis

Abstract

The prevalence of alcoholic liver disease (ALD) is rapidly increasing, however, few if any therapies are successful in the treatment of this disease as the development of such therapies is hampered by the lack of animal models. In this work, we aim to identify points of metabolic divergence driving the progression of ALD, which may provide therapeutic potential in the treatment of the disease. To determine the differences in metabolic landscapes across these progressing stages of ALD, bulk RNA‐seq was performed on alcoholic hepatitis (AH), progressing ALD, non‐alcoholic liver disease (NALD), and healthy control human liver samples. Hierarchical clustering of metabolic genes revealed N‐glycan metabolism, nucleotide metabolism and protein modification associated with AH samples, while oxidative phosphorylation, drug metabolism and retinol metabolism associated with healthy control and NALD samples. To further investigate which hepatic cell types may be involved in this shift in metabolic activity, we performed deconvolution of the bulk RNA‐seq data using a previously published single‐cell RNA‐seq dataset from healthy human liver tissue. Two imputed populations of hepatocytes, the periportal Hep 4 population and the mid‐lobular Hep 6 population, showed a decrease in cell proportions with increasing disease severity from healthy control to AH. Oppositely, there was an increase in cholangiocyte and hepatic stellate cell (HSCs) proportions with AH progression. Hierarchical clustering of metabolic genes from each cell type showed similar clustering to the bulk data, with groups of genes partitioned into (1) those upregulated with AH and (2) those upregulated in healthy control and NALD. Clusters of upregulated metabolic genes within the AH states from the Hep 4, cholangiocyte, and HSC populations all showed overrepresentation of metabolic pathways common to those upregulated in the bulk data, which differed from the Hep 6 population, whose enriched processes were downregulated in the bulk data. However, both hepatocyte populations displayed overrepresented metabolic pathways similar to the bulk data when analyzing the cluster of upregulated metabolic genes in the NALD and healthy states. These results indicate that the Hep 4, HSC and cholangiocyte populations act synergistically to upregulate specific metabolic genes in the AH state, while only the hepatocyte populations are responsible for the upregulation of metabolic genes in the healthy control and NALD states. It is crucial to further analyze the Hep 4 cluster of cells as they seem to be driving the metabolic switch activity separating the AH states from the NALD and healthy control disease states.