The authors found that glycogen content in diethylnitrosamine (DEN)-induced liver tumours in mice was higher in smaller size tumours and lower in larger size tumours compared with adjacent tissues. In addition, glycogen content was increased in sections of early stage human hepatocellular carcinoma (HCC) and reduced in those of late stage HCC, compared with normal adjacent tissue. RNA sequencing (RNA-seq) analysis of DEN-induced, pre-malignant liver lesions in mice compared with normal adjacent tissue showed that among glycogen metabolism genes, the gene encoding glucose-6-phosphatase (G6PC) was significantly downregulated in pre-malignant lesions. G6PC catalyses the de-phosphorylation and thus release of glucose from glycogen. In line with this, mice with liver-specific G6PC knockout (G6pcΔAlb mice) showed higher liver glycogen content than control mice, confirming that G6PC reduction can mediate glycogen accumulation. The authors observed that hepatocytes in livers of G6pc-deficient mice had a larger size and increased proliferation, resembling the phenotype of Hippo-signalling deficient mice. Indeed, G6pc-deficient hepatocytes had reduced activity of MST1 and MST2 and increased activity of their downstream effector YAP (which was de-inhibited). Of note, RNA-seq analysis of small tumour lesions in DEN-treated mice showed activation of YAP target genes. Of interest, fluorescence microscopy showed that MST1 and MST2 aggregates formed multiple foci in glycogen-containing tumour nodules in DEN-treated mice as well as human HCC compared with normal tissue. Because the activity of Hippo signalling negatively correlated with glycogen content, the authors wondered if the sequestering of MST1 and MST2 aggregates within glycogen mediated the inhibition of Hippo signalling. The authors observed that glycogen underwent LLPS in solution and in cells, showing that fluorescently labelled glycogen formed droplets, which tended to coalesce over time into larger droplets or split into smaller ones. Importantly, disruption of glycogen LLPS with 1,6-hexanediol led to MST1 and MST2 dissociating from the glycogen compartment and also restored Hippo signalling activity. Further analysis showed that MST1 and MST2 were recruited into glycogen droplets by interaction with Laforin, which is a phosphatase that contains a carbohydrate domain. Laforin expression was required for the blockade of Hippo signalling and this was glycogen-LLPS dependent. In addition, Laforin competed with WW45, a Hippo pathway component that forms a complex with and activates MST1 and MST2, and that was not observed to colocalize with glycogen. This complex was disrupted by Laforin in the presence of glycogen, leading to de-inhibition of YAP signalling.
This is a preview of subscription content, access via your institution
