We read with great interest the article by De Alwis and colleagues that proposes that liver stem cells originate from the canals of Hering.1 These authors have confirmed our previous observations,2 namely that clonally-derived patches of hepatocytes are invariably abutting portal tracts, and then by sequencing the entire mitochondrial genome in cells at three locations along the portal tract:hepatic vein axis, they have gone on to suggest that cells must be migrating in that direction. Akin to constructing a phylogenetic tree, cells in all three zones had two common mutations, whereas the outermost two zones shared an additional C7794T mutation and the very outermost group of hepatocytes had a further unique T9540V mutation. Although we broadly agree with the conclusion of De Alwis et al., we have some reservations and also suggest their results throw out the possibility of a hitherto unrecognized property of stem cells.
First, the canals of Hering, the proposed location of hepatic stem cells, are arborizing biliary conduits that can extend beyond the limiting plate.3 Thus, in theory, clonal populations could have origin from even a midzonal location; in our study, these were never observed.2 More crucially, the De Alwis study has not reported their common mutations in associated cytokeratin 19–positive biliary cells; thus, we believe their conclusion is premature and not warranted by their data. We suggest their, and our, data can be explained by a hepatic stem cell found in or very close to the limiting plate. Indeed, serial hepatocyte transplantation studies in the Fah null mouse can only be explained by the presence of highly clonogenic hepatocytes,4 and in the simple pulse-chase labeling experiments designed by Gershom Zajicek, that also suggested that hepatocytes migrated toward the hepatic vein, the cells that immediately labeled with3H-thymidine were hepatocytes (not biliary cells) located approximately 70 mm from the portal rim.5
Second, we believe this study may have unearthed an unsuspected stem cell property: the maintenance of mitochondrial DNA (mtDNA) integrity. Because the study by De Alwis et al. sequenced the entire mitochondrial genome of each sample, would it not be logical to assume that the most long-lived cells would accumulate the most mtDNA mutations? However, the most long-lived cells are not those in the cytochrome c oxidase (COX)-deficient patch furthest from the portal tract, but the portally located stem cells, and yet the De Alwis study (in their figure 1D) found that their periportal hepatocytes contained the fewest mutations! Perhaps then, stem cells in their niche have a mechanism for protecting/repairing their mitochondrial genome, as has been proposed for stem cell genomic DNA.6 Our previous studies in the human intestine have suggested that detectable deficiencies in COX only become apparent after the age of 40, when up to 80% of a stem cell's mitochondria possess the appropriate mutation,7 but the De Alwis study would suggest that once the hepatocyte leaves the niche, it can acquire further mutations (in this case, two) within a matter of months! So, can we conclude that either stem cells have an inherent mechanism in place that efficiently repairs most mtDNA damage or are they simply not subjected to the normal degree of oxidative damage, and once cells leave the niche, perhaps more oxidative stress and/or less efficient repair conspire to increase the mtDNA mutation load? The study by De Alwis et al. certainly suggests we should enquire further into this aspect of stem cell biology in the hunt for the elusive hepatic stem cell.