Hailfinger and colleagues1 put forward an interesting hypothesis to explain the mystery of hepatocyte heterogeneity. Based on the observation that tumors carrying mutations of either the H-Ras or Catnb genes express sets of genes resembling those predominating in periportal and pericentral regions of liver lobuli, they propose that hepatocyte heterogeneity is determined by two opposing gradients. The first of these would involve soluble factors from the bloodstream inducing Ras activation prevailing in the periportal zone, while the second would consist of an inverse gradient of Wnt signals inducing activation of β-catenin prevailing in the pericentral zone. Although this hypothesis is attractive for a number of reasons, the experimental evidence provided by Hailfinger et al.1 does not in itself constitute substantial support.
Recent observations have left no doubt that expression of glutamine synthetase (GS), the most pronounced marker enzyme of pericentral cells,2 is somehow dependent on β-catenin signaling.3–5 However, although activation of β-catenin seems necessary, it does not appear to be sufficient, since many liver tumors with activated β-catenin do not express GS. The experimental approach taken by Hailfinger et al.1 (to induce pericentral genes in cultured hepatocytes using Wnt factors or inhibitors of GSK3β) is not conclusive, since whole hepatocyte preparations contain both periportal and pericentral cells of which only the pericentral subpopulation might respond. Moreover, the mechanism by which β-catenin signaling might influence GS expression remains obscure, since indication of a permanent activation of β-catenin in the GS-positive zone of the lobuli in adult liver is missing. Rather, during regeneration following liver damage in zone 3 we recently observed the transient activation and cytoplasmic accumulation of β-catenin in the most distal pericentral hepatocytes immediately before GS expression is switched on.6 This suggests some kind of zonal reprogramming of otherwise similarly differentiated hepatocytes which we would like to call postdifferentiation tissue patterning consistent with the role of Wnts in many similar phenomena during embryogenesis and tissue formation.7 This is also consistent with the early observation that cultured GS-positive hepatocytes maintain their phenotype throughout their lifespan in vitro despite cultivation in media mimicking periportal conditions.8
The evidence provided by Hailfinger et al.1 concerning periportal signals is less clear. Firstly, the proposed influence of Ras activation was not demonstrated in vitro. Secondly, the finding that hepatocellular tumors develop rarely in response to β-catenin mutation alone, but frequently in the presence of an additional mutation in the H-Ras gene,9 renders it doubtful that all of the tumors exhibiting β-catenin mutations in the study by Hailfinger et al.1 were really lacking Ras mutations. Thirdly, Hailfinger et al. maintain that periportal Ras activation results from blood-borne factors.1 This assumption neglects earlier evidence that liver zonation in ectopic explants develops correctly even without portal blood or the natural orientation of blood flow.10 Thus, the underlying mechanism of liver parenchymal heterogeneity remains an intrahepatic issue controlled by as yet unknown (morphogenetic) factors. Wnts and β-catenin-dependent signaling certainly have a role to play, but this seems unlikely to be in the context proposed by Hailfinger et al.1