In this study, we demonstrated that fusion-derived hepatocytes could undergo ploidy reductions. Initially, serial transplantation experiments were performed. Cytogenetic analysis showed that 14% of fusion-derived hepatocytes were nearly diploid. Surprisingly, fusion-derived hepatocytes were highly aneuploid. While most regenerating nodules expressed all markers of cell fusion, a fraction of FAH positive nodules (2–5%) were Y-chromosome negative or ß-gal negative. This frequency suggests that 4–10% of the nodules were initiated by cells that had undergone ploidy reduction events. Next, we utilized a ß-gal reporter system to track fusion products in discrete ploidy populations. Polyploid hepatocytes expressed FAH (indicating that fusion-derived cells were reprogrammed to express donor genes) and ß-gal (demonstrating that these cells were derived by cell fusion). Significantly, diploid hepatocytes also expressed ß-gal and FAH, establishing that these cells originated from polyploid fusion-derived hepatocytes. Finally, we carefully tracked donor/host markers in hepatocytes that had undergone ploidy reductions by single cell genotyping. The majority of diploid daughter hepatocytes (87%) were negative for one or more markers, giving rise to a heterogeneous population of cells. These results suggest that markers/chromosomes segregate independently during ploidy reduction events.
Hepatocyte polyploidization has been documented in many species (reviewed by Gupta 
), but ploidy reversal has not been rigorously characterized. Our experiments provide proof-of-concept that ploidy reversal does occur in fusion-derived hepatocytes. Several reports also suggest that normal hepatocytes may undergo ploidy reduction. For instance, treatment of rodents with hepatotoxins thioacetamide 
and carbon tetrachloride 
led to a dramatic increase in diploid hepatocytes and concomitant decrease in polyploid hepatocytes over 72 hr. Differential proliferation and/or cell death was not seen among diploid or polyploid hepatocytes 
. Thus, it is possible that normal polyploid hepatocytes undergo ploidy reductions, but this hypothesis remains to be tested.
The high degree of aneuploidy displayed by fusion-derived hepatocytes is surprising. It is unclear whether aneuploidy resulted directly from the fusion and/or ploidy reduction events or indirectly as a consequence of the Fah
repopulation model 
. Furthermore, we cannot exclude the possibility of stochastic chromosome loss during mitosis 
. Thus, aneuploid hepatocytes could arise from the random loss of chromosomes by fusion-derived hepatocytes undergoing extensive proliferation.
A number of possibilities could explain how diploid hepatocytes are generated from polyploid fusion-derived hepatocytes. First, it is theoretically possible that binucleated fusion-derived hepatocytes could simply complete cytokinesis (). Normal binucleated polyploid hepatocytes are formed through failed cytokinesis 
. For example, a mononucleated diploid hepatocyte undergoes a regular mitosis, but then separation of the two daughter cells fails, generating a binucleated tetraploid cell with two diploid nuclei 
. Whether binucleated hepatocytes could resume cytokinesis is unclear, but it remains a possibility. In the context of fusion-derived hepatocytes, the completion of cytokinesis would generate two mononucleated diploid daughter cells, each with the same genotype as the original fusion partners. As seen in , subsets of diploid hepatocytes contained a donor marker (Fah
) and a recipient marker (Cre and/or Y-chromosome), proving that these cells were genetically distinct from the original fusion partners. Therefore, a cytokinesis-type mechanism can be excluded.
Potential mechanisms for diploid hepatocyte formation from polyploid fusion-derived hepatocytes.
The second possibility is chromosome loss via multipolar mitosis, which can lead to the random segregation of chromosome content among two or more daughter cells 
. Fusion-derived hepatocytes have increased numbers of centrosomes, which could result in the formation of multiple spindle poles during prophase. Thus, multipolar mitotic events could enrich for daughter cells with diploid chromosome content (). However, multipolar mitosis cannot adequately explain the clustering of fusion-derived hepatocytes with atypical chromosome counts. For example, triploid hepatocytes with ~60 chromosomes (XXY or XYY) comprised 17% of the metaphases analyzed (). These daughter cells likely originated from hexaploid fusion-derived hepatocytes. It is difficult to imagine how multipolar mitosis would enrich for cells with such abnormal chromosome counts. Furthermore, if ploidy reduction were achieved by multipolar mitosis, then each chromosome should be lost with the same low frequency (i.e., 1/19 for autosomes). Single cell genotyping of diploid daughter hepatocytes showed loss of R26R (located on chromosome 6 
) and the Y-chromosome at 50% or greater (). This high degree of marker segregation strongly suggests that chromosome/marker loss occurs in a non-random fashion.
Another possibility to explain the emergence of daughter hepatocytes with one-half DNA content is cell division without DNA replication (). This type of ploidy reduction was first described in the mosquito Culex pipiens 
but has never been described in mammalian cells. In this model, fusion-derived hepatocytes could proceed through G1 phase of the cell cycle, skip S-phase and progress to G2/mitosis. Pairing between homologous chromosomes would ensure proper chromosome segregation. This type of mechanism accounts for the generation of diploid daughter cells (, ) as well as enrichment for atypical triploid daughter cells (). Moreover, the high degree of marker loss seen in diploid daughter cells () is possible through a chromosome pairing interaction. Rigorous testing of all potential mechanisms must be performed to elucidate the cellular processes governing ploidy reductions in fusion-derived hepatocytes.
Finally, direct transmission of DNA via horizontal gene transfer (HGT) into diploid hepatocytes must be considered. HGT among somatic cells involves phagocytosis of apoptotic cells followed by nuclear uptake/integration of whole chromosomes or chromosome fragments by the engulfing cell 
. HGT was hypothesized to induce hepatocyte reprogramming in xenotransplantation experiments 
. In our studies, diploid host hepatocytes could acquire genes from apoptotic donor blood cells, resulting in hepatocyte reprogramming while maintaining a nearly diploid chromosome count (). However, the presence of multiple donor markers on different chromosomes by HGT is expected to be rare, and we found that nearly half of diploid fusion-derived hepatocytes harbored at least two donor markers (). Therefore, our data strongly argue against an HGT-type of mechanism.
Regardless of the mechanism, ploidy reduction events have significant implications. In the context of fusion-derived hepatocytes, ploidy reductions can be a confounding factor when tracing markers during stem cell transplantation. Donor markers can be lost during ploidy reductions, thus leading to an underestimate of engraftment. Similarly, host markers can be lost from hybrids, obscuring the existence of fusion and giving the false impression of transdifferentiation. Because cell fusion between transplanted cell types and target organs has been described in many experimental systems (reviewed in 
), the possibility of ploidy reductions needs to be considered when interpreting cell transplantation experiments. Furthermore, we propose that ploidy reduction events may contribute to tumorigenesis. The independent segregation of chromosomes from polyploid cells results in genetically heterogeneous diploid daughter hepatocytes. Individual daughter cells could lose tumor suppressors, generating a subset of hepatocytes with oncogenic potential.