In collaboration with Fellous and colleagues,1 we recently described regions of healthy human liver with a deficiency of cytochrome c oxidase (COX) containing somatic mitochondrial DNA (mtDNA) mutations.2 These findings confirmed the clonal origin of the COX-deficient hepatocytes, and their distribution suggested a periportal origin. Here, we show for the first time that the hepatocytes within the COX-deficient patches originate from the periductular region (canals of Hering), where a hepatocyte stem cell population must reside.
We used sequential COX–succinate dehydrogenase (SDH) histochemistry to identify regions of human liver containing cells potentially derived from a single precursor cell (Fig. 1A,B). We subsequently sequenced the entire mtDNA from different regions of the same COX-deficient region and adjacent liver. In each case, we indentified the same clonally expanded mtDNA mutation throughout the COX-deficient zone, and the mutation was not detected in adjacent normal liver (Fig. 1C). As in our previous work,1 the presence of only mutated mtDNA throughout the COX-deficient region (mtDNA homoplasmy) indicates that the mutation arose in a common precursor cell. Some of the COX-deficient regions were large, extending up to 600 μm in approximate diameter, demonstrating the extent of the region repopulated by a single precursor. All of the COX-deficient regions were contained within a single lobule, with no evidence of extension from one lobule to another.
Several COX-deficient zones contained at least two somatic mtDNA mutations (Fig. 1D). Cells more distant from the portal tract had additional rare mutations not present in cells closer to the tract, and never the other way around. The most likely explanation is that, as the daughter cells migrate away from the bile duct, some acquire additional somatic mtDNA mutations. This is consistent with “the streaming liver” hypothesis, where regenerating hepatocytes arise from a stem cell population in the canal of Hering and move outward into the liver parenchyma.2 As previously observed,1 the COX-deficient zones were morphologically normal, and in some healthy aged subjects, up to 5% of the liver showed a COX defect despite having normal biochemical indices of liver function. A strong selection bias, either for or against the mtDNA mutations, therefore seems unlikely. This is supported by our observation of clonally expanded synonymous mtDNA substitutions in regions with normal COX activity, which have presumably accumulated through random genetic drift.3 The mtDNA mutations therefore appear to be a reliable marker of hepatocyte lineages.
The prevailing view is that mature hepatocytes divide every year, gradually replacing adjacent liver cells. On the other hand, stem cell activation is only thought to occur after an acute insult, or after chronic liver damage when the capacity for mature hepatocyte division has been overwhelmed. Our observations indicate that the stem cell population is active in healthy liver and contributes to hepatocyte turnover. Although we found no evidence that the mtDNA mutations compromise liver cell function, there is emerging evidence that stem cell proliferation could be compromised by somatic mtDNA mutations. If correct, then this could explain why the prognosis following acute liver insult is worse in older subjects.4, 5