Hematopoietic cells as hepatocyte stem cells: A critical review of the evidence


  • Snorri S. Thorgeirsson,

    Corresponding author
    1. Laboratory of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute/NIH, Bethesda, MD
    • National Institutes of Health – National Cancer Institute NCI Bldg. 37, Room 4146A, 37 Convent Drive MSC 4262, Bethesda, MD 20892-4262
    Search for more papers by this author
    • fax: 301-496-0734

  • Joe W. Grisham

    1. Laboratory of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute/NIH, Bethesda, MD
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


The authors reviewed 77 published reports available before August 1, 2005 that examined the ability of hematopoietic cells to generate hepatocytes in the liver. A list of these publications and a synopsis of each are available on-line. We interpret the evidence provided by this data set to suggest that one or more types of hematopoietic cells may rarely acquire the hepatocyte phenotype in the liver (frequency ≤10−4), although the nature of the hematopoietic cells involved and the mechanisms responsible for acquisition of a hepatocyte phenotype are still controversial. Hematopoietic stem cells do not appear to be direct precursors of hepatocytes, which, instead, can be generated from cells of the macrophage–monocyte lineage. Fusion between hepatocytes and transplanted hematopoietic cells has been substantiated as a mechanism by which hepatocytes that carry a bone marrow tag are generated, but direct transdifferentiation of hematopoietic cells has not been demonstrated. In conclusion, hematopoietic cells contribute little to hepatocyte formation under either physiological or pathological conditions, although they may provide cytokines and growth factors that promote hepatocyte functions by paracrine mechanisms. Cells of the endodermal hepatocyte lineage are far more potent generators of hepatocytes than are hematopoietic cells. (HEPATOLOGY 2006;43:2–8.)

Hepatic and hematopoietic tissues maintain a close association throughout the lifespan of mammals. The embryonic liver is the major site of blood cell formation from mesodermal cells that migrate from the yolk sac and/or the aortic-gonadal-mesonephros region of the embryo.1 Subsequently, endodermal precursor cells emerge from the foregut and differentiate into hepatocytes and cholangiocytes in the fetal liver.2 Through the production of specific cytokines and growth factors, embryonic hepatic epithelial cells and intrahepatic hematopoietic cells mutually promote the development and differentiation of the other type of cell.3 Although fetal liver epithelial cells and hematopoietic progenitor cells express a few genes in common (see Suskind and Muench,4 and references therein; and Petersen et al.5), a common endo-mesodermal stem cell has not been identified in the fetal liver.4, 6

After the major site of blood cell formation moves from the liver to the bone marrow toward the end of liver development, mesodermal cells originating from hematopoietic cells, including Kupffer cells (liver macrophages),7 hepatic lymphocytes [natural killer (NK) and T cells],8 sinusoidal endothelial cells,9 stellate cells,10, 11 and bone marrow stem cells (see Kotton et al.12 and references therein) remain in the liver and continue to be partially replenished from the bone marrow in adult animals.


NTBC, 2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione.

Online Presentation of the Data Set

Recent reports suggest that hematopoietic cells also generate hepatocytes in adult animals. We assessed 77 reports (published before August 1, 2005) that have examined this possibility to determine whether the weight of evidence they present suggests that hematopoietic cells are a significant source for replacement of hepatocytes in adult animals. The list of publications (References S1–S77) that constitute the data set and tabulated synopses of the studies (Tables S1–S7) are available on-line as supplementary material (at the HEPATOLOGYwebsite: http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html).

Discussion of the Data Set

Our discussion of the data set emphasizes important questions concerning hematopoietic cells and hepatocytes: (A) Can hematopoietic cells generate hepatocytes? (B) If so, do hematopoietic cell–derived hepatocytes contribute significantly to hepatocyte replacement under physiological conditions and after pathological loss? (C) If hematopoietic cells can generate hepatocytes, which cells do so? (D) What are the mechanisms by which this is accomplished? (E) Do intrahepatic hematopoietic cells contribute to hepatocyte function and renewal by mechanisms other than transformation into hepatocytes?

Can Hematopoietic Cells Generate Hepatocytes?

Transplanted hematopoietic cells can generate hepatocyte-like cells in the liver at very low frequency (≤10−4). The strongest support for this opinion comes from experimental studies that analyzed the generation of hepatocytes from hematopoietic cells when allogeneic bone marrow cells were transplanted into lethally irradiated mice (S1–S30,S54). These experimental studies, which are based on the standard method used to identify hematopoietic cells reconstituting ablated hematopoietic lineages, can be better controlled than retrospective studies in humans. Also, a variety of methods to identify hepatocytes derived from hematopoietic cells used in experimental studies reduces the chance of systematic errors in detecting such cells.

In more than 80% of the experimental groups the yield of presumed hematopoietic cell–derived hepatocytes was less than 0.05% of the recipient hepatocytes when selective procedures designed to increase the yield of hematopoietic cell–derived hepatocytes were not used, and in only 6% of groups did the yield exceed 1.5%. When various selective conditions, designed to impede proliferation of endogenous hepatocytes, were imposed, 50% of the groups showed yields greater than 1.5%, whereas the fraction of yields less than 0.05% decreased to 45%. Our analysis suggests several possible explanations for high yields, including differences in post-engraftment amplification of hepatocytes derived from hematopoietic cells, use of culture-derived hematopoietic cells that may have been “reprogrammed” in vitro, and imprecision in the detection of hepatocytes derived from hematopoietic cells.

Engraftment Versus Amplification.

Few studies have attempted to distinguish initial engraftment of bone marrow–derived hepatocytes from subsequent amplification of their progeny. Quantitative studies of the engraftment of transplanted hematopoietic cells as hepatocyte-like cells in livers of lethally irradiated Fah−/− mice receiving 2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione (NTBC), which blocks the formation of hepatotoxic fumarylacetoacetate, show that engraftment occurs at a frequency between 1/104 and 1/106 native hepatocytes, which represents 50 to 500 bone marrow–derived hepatocytes per liver when 1 to 2 × 106 bone marrow cells were transplanted (S6,S17). Thus, only 0.0025% to 0.025% of the transplanted cells engraft, constituting 0.01% to 0.0001% of the recipient's hepatocytes. Similar engraftment levels can be deduced from other studies in mice (S2,S5–S7,S16,S19,S23).

Initial engraftment of bone marrow–derived hepatocytes is followed by amplification of the engrafted cells when the endogenous hepatocytes are killed in Fah−/− mice by removing NTBC (S6,S17) and in mice receiving transplants of Bcl2tg bone marrow by exposure to Jo2 (anti-Fas) antibody (S5). Endogenous hepatocytes are selectively killed by these regimens; however, progeny of transplanted hematopoietic cells are less susceptible. After removal of NTBC from Fah−/− mice, the 50 to 500 bone marrow–derived hepatocytes that engrafted rapidly expanded to yield a population of hematopoietic cell–derived hepatocytes that constituted more than 30% of the total hepatocyte population (S6,S17). Less dramatic expansion of engrafted hepatocytes derived from Bcl2tg bone marrow cells occurred in mice exposed to Jo2 antibody (S8).

Amplification of engrafted cells was also sometimes found after bone marrow transplantation in which the liver was damaged by CCl4 (S18,S30); however, in most studies post-engraftment amplification was slight. Little hepatocyte proliferation occurs under physiological conditions in the liver of adult animals, and selective proliferation of hepatocyte-like progeny of transplanted bone marrow is not expected if these cells respond to regulatory signals in a manner similar to host hepatocytes. Even 70% partial hepatectomy (or equivalent cell loss from hepatotoxins such as CCl4) elicits less than two consecutive cycles of proliferation by each residual hepatocyte. These results support different levels of amplification of engrafted hematopoietic cells as a major source of variation in the yields of bone marrow–derived hepatocytes.

Yields of hepatocytes assumed to be derived from transplanted hematopoietic cells of greater than 1.5% were described in 3 studies in unirradiated mice (S18,S26,S27) and in one study in unirradiated rats (S33). Because chimeric bone marrow is not produced by transplantation of allogeneic hematopoietic cells into unirradiated adult recipients, hematopoietic cell–derived hepatocytes would have to be generated by direct transformation of transplanted cells as they passed through the liver. This explanation was proposed in two studies that used this experimental format: in one study, 2% to 7% yields of presumed bone marrow–derived hepatocytes were found in recipient livers within 2 days after transplantation of allogeneic Fra25Lin “homed-recovered” bone marrow cells into unirradiated mice either exposed or not to CCl4 (S18); however, this study has not been independently replicated; in the other study, greater than 12% of the recipient hepatocytes were assumed to be derived from hematopoietic cells when allogeneic whole bone marrow or Liv8 bone marrow cells were infused directly into the portal vein of unirradiated mice whose livers were damaged by CCl4 (S26,S27); however, a study of similar design in rats failed to replicate this finding (S34,S40). Further studies are needed.

Unirradiated pre-immune fetal animals develop partial hematopoietic chimerism after receiving xenogeneic transplants of human cord blood cells (S56–S60). In one study, greater than 10% of the hepatocyte population was presumed to be derived from human CD34+ cord blood cells that were transplanted into the peritoneal cavity of fetal sheep (S56), but only rare hematopoietic cell–derived hepatocytes were detected in the other study, in which CD34+ human cord blood cells were transplanted into the coelomic cavity of fetal sheep (S57). Similar levels of hematopoietic cell chimerism occurred in both studies, however. Low levels of hepatocytes presumed to be derived from hematopoietic cells were also apparently found in fetal goats (S59) and mice (S60) that received transplants of human cord blood cells. Further studies are needed.

Cells Derived Ex Vivo From Cultured Bone Marrow May Be Unusually Adept at Generating Hepatocytes.

In 5 of 6 studies in which culture-derived hematopoietic cells were transplanted, the yield of hepatocytes putatively derived from them was greater than 1.5%. Culture-derived cells included β2MThy1+ bone marrow cells from cholestatic rats that were cocultured with primary hepatocytes (separated by a semipermeable membrane) (S33), and several cells that were derived from the adherent fraction of bone marrow, including multipotential adult progenitor cells (S10), macrophages produced from cultured bone marrow under the influence of macrophage colony-stimulating factor (S19), Flk+CD31CD34 cells from fetal bone marrow (S50), mesenchymal stem cells (S55), and unrestricted somatic stem cells (S58). The basis for the high yields of hepatocytes from culture-generated bone marrow is not certain; however, prolonged cell culture necessary to derive cell lines may lead to nuclear reprogramming13–15 and allow the resulting cell lines to be more plastic. Notably, stromal cells isolated directly from bone marrow did not engraft as hepatocytes when transplanted (S19). Further studies are needed.

Studies in Humans.

Retrospective studies of putative bone marrow–derived hepatocytes in livers of humans receiving transplants of allogeneic bone marrow (S61–S64) and liver (S3,S61,S62,S66–S71), and of naturally occurring bone marrow chimerism in bovines and humans (S73–S77) have shown yields that range from 0% to 8%. These results may be skewed because most yields are either very low or very high. Among 4 studies of allogeneic bone marrow transplantation in humans, no potential bone marrow–derived hepatocytes were found in one study of 5 patients (S64), whereas in two other studies of 8 patients, assumed bone marrow–derived hepatocytes constituted 3.1% to 12.3% of the hepatocyte populations (S61,S63). In the other study of 9 patients, yields ranged from 0.5% to 2.0%, but data were not provided for individual patients (S62). Similarly, among 9 studies of allogeneic liver transplants in humans (S3,S61,S62,S66–S71), which included 127 patients, no putative bone marrow–derived hepatocytes were found in 79 patients (62%), whereas the yields in the remaining 48 patients ranged up to 3.3%. Although few possible bone marrow–derived hepatocytes were detected in studies of women who developed XY chimeric bone marrow after pregnancies with male fetuses (S74–S77), 4% of the hepatocyte population was putatively derived from bone marrow in one patient (S75).

Possible Erroneous Identification of Bone Marrow–Derived Hepatocytes.

Errors in identifying hepatocytes derived from hematopoietic cells may be a significant problem. Detection of putative bone marrow–derived hepatocytes in female subjects with XY chimeric bone marrow depends on the identification of Y chromosomes in cells expressing hepatocyte phenotypic properties. This method may result in erroneous assignment of some of the numerous Y+ nonparenchymal and inflammatory cells as hepatocytes. After sex-mismatched liver or bone marrow transplants, the recipient liver contains more Y+ nonparenchymal cells than Y+ hepatocytes (S8,S68), even at the very low levels of bone marrow microchimerism in women who carried male fetuses (S77). Precise identification of Y+ hepatocytes in a liver containing large numbers of Y+ nonparenchymal and inflammatory cells, as occurs in these patients, poses a notoriously difficult problem,16 especially because the studies were conducted with sections of liver tissue and often used nonspecific methods to identify hepatocytes. In this situation Y+ nonparenchymal cells may be falsely identified as Y+ hepatocytes, because nuclei of the small Y+ nonparenchymal cells may insensibly overlap the larger hepatocyte nuclei. Low yields of putative bone marrow–derived hepatocytes detected in studies that used isotype analysis in HLA-mismatched liver transplants (S70) further suggest that the identification of hepatocytes generated from bone marrow cells by detecting a Y chromosome is highly error-prone.

Erroneous identification of hepatocytes derived from hematopoietic cells also may affect studies that use other methods of tagging and detection, such as fluorescent transgene products and immunochemically detected antigens.17 For example, the HepPar1 (OCH1E5) antibody is often used to detect hepatocytes derived from xenogeneic transplants of human hematopoietic cells (S46–S48,S51,S52,S54,S56,S57,S59,S60), but this antibody may lack species specificity. HepPar1, which reacts with an unknown epitope,18 does not distinguish human hepatocytes from those of some other species.19 In one of the studies reviewed here, murine hepatocytes were decorated by the antibody (S48). The specificity of the methods used to identify bone marrow–derived hepatocytes must be rigorously confirmed.

Do Hematopoietic Cell–Derived–Hepatocytes Contribute Significantly to Liver Maintenance?

The results of these studies indicate that hematopoietic cells contribute little to hepatocyte replacement in adult animals under physiological and most pathological conditions, as concluded independently.20, 21 As indicated by low DNA labeling indices in healthy adult hepatocytes,22 the physiological turnover of hepatocytes in adult mice and rats is approximately 400 days, with a daily turnover rate that averages approximately 0.25% of the hepatocyte population. Generation of hepatocytes from bone marrow cells at a frequency of 10−4 to 10−6 is far too low to replace host hepatocytes lost to physiological turnover. This conclusion is supported by published studies that appear to be most analogous to hepatocyte formation during postnatal growth and physiological turnover in adults. For example, no evidence of significant contribution to hepatocyte formation by circulating cells from chimeric bone marrow was detected in parabiotic mice (S7) or bovine freemartins (females who share placental circulation with male twins) (S73). Results from xenogeneic human cord blood transplants in fetal animals have not been reproducible (S56–S60).

Replacement by hematopoietic cell–derived hepatocytes of larger numbers of endogenous hepatocytes lost to trauma or disease depends on the capacity of the latter to expand rapidly and selectively. Studies in Fah−/− mice that received transplants of Fah+/+ bone marrow cells (S6,S17) suggest that amplification of such rapidity and selectivity may be possible; however, the selective advantage that hepatocytes derived from Fah+/+ bone marrow cells have over the lethally damaged endogenous hepatocytes in livers of Fah−/− recipients appears to be unique. Progeny of transplanted hematopoietic cells have no selective advantage over endogenous hepatocytes under most conditions. This was clearly demonstrated in a study in which the proliferation of both endogenous and bone marrow–derived hepatocytes was assessed in livers of mice exposed to CCl4 (S30); although proliferation of the small number of bone marrow–derived hepatocytes was increased by approximately 6-fold, proliferation of the much more numerous endogenous hepatocytes was amplified by approximately 10-fold (S30), and most of the new hepatocytes were derived from the host hepatocytes (S30).

Residual endogenous hepatocytes rapidly reactivate quiescent cell cycling and quickly replace large numbers of cells that are lost to disease and toxicity.22–24 Furthermore, when residual hepatocytes incur damage that prevents them from proliferating in response to cell loss, they can be replaced from endodermal hepatic epithelial stem cells through the oval cell reaction.25 When isolated and transplanted, fully differentiated hepatocytes engraft efficiently and then replicate at the rate of pre-existing host hepatocytes,26 and Fah+/+ hepatocytes amplify consecutively through hundreds of cycles when transplanted serially into Fah−/− mice.27 [Unlike hematopoietic cells, transplanted Fah+/+ hepatocytes do not fuse with endogenous Fah−/− hepatocytes (S9).] Approximately 10% to 20% of transplanted hepatocytes engraft in the liver, and approximately 6% of the engrafted cells proliferate in response to hepatocyte deficits (S6).28, 29 Direct comparison of engraftment and expansion of transplanted bone marrow cells and hepatocytes under comparable conditions (S6)30 indicates that authentic hepatocytes engraft much more efficiently than do bone marrow cells, and the proliferation capacity of authentic hepatocytes is greater than that of bone marrow–derived hepatocytes (Fig. 1).

Figure 1.

Relative engraftment efficiencies of hematopoietic cells and mature hepatocytes.

Which Hematopoietic Cells Generate Hepatocytes?

Evidence shows that cells of macrophage–monocyte lineage can generate hepatocytes. Although the allogeneic transplantation of whole bone marrow into irradiated recipients cannot identify the specific hematopoietic lineages that generate hepatocytes, these studies indicate that stem cells do not directly transition into hepatocytes, because populations enriched in stem cells (KSL or KSTL [S2,S6,S7] and SP [S12,S13,S17] populations) are not qualitatively more efficient than whole bone marrow. A possible exception is the Fra25Lin “homed-recovered” preparation of bone marrow stem cells, which were proposed to transform directly into hepatocytes (S18). Either the bone marrow stem cells in the Fra25Lin “homed-recovered” population have properties that differ radically from the stem cells in KSL/KSTL and SP populations or bone marrow stem cells do not transform directly into hepatocytes. Further studies are needed.

Lymphocyte lineages were excluded as the source of hematopoietic cell–derived hepatocytes by the demonstration that bone marrow from either Rag1−/− (S19) or Rag2−/−γc−/− (S17) transgenic mice, which lack lymphocyte lineages, were fully competent to generate hematopoietic cell–derived hepatocytes when transplanted into irradiated recipients. Cells of monocyte–macrophage lineage were first suggested to be the source of hematopoietic cell–derived hepatocytes by transplantation of bone marrow from mice transgenically tagged in the LysM locus (S16,S17,S23). Hepatocytes were also generated from infused macrophage populations produced ex vivo from cultured bone marrow (S19); however, the ex vivo generated macrophages may have undergone nuclear reprogramming in culture, resulting in increased plasticity.13–15

What Are the Mechanisms by Which Hematopoietic Cells Generate Hepatocytes?

Posited mechanisms of hepatocyte generation from hematopoietic cells include both transdifferentiation of bone marrow cells and fusion between a host hepatocyte and a hematopoietic cell.31 Transdifferentiation is hypothesized to involve a simultaneous shift in the transcriptional activity of many genes to change directly the phenotype of a hematopoietic cell into that of a hepatocyte. In contrast, fusion between a hepatocyte and a hematopoietic cell produces a heterokaryotic hybrid cell that initially contains the genetic elements and organelles of both cell types. Expression of the hepatocyte phenotype by such a hybrid cell requires either that the nucleus of the hematopoietic cell be extruded from the heterokaryon or that the hematopoietic cell–specific genes be reprogrammed to generate hepatocyte-specific proteins (Fig. 2).

Figure 2.

Hepatocyte generation from transplanted hematopoietic cells.

Direct transdifferentiation of hematopoietic cells into hepatocytes has not been conclusively validated as a mechanism for generation of hepatocytes in any study. Demonstration of transdifferentiation requires that changes in the activity of multiple hematopoietic cell- and hepatocyte-specific genes be shown to occur in individual cells in “real time” (a clonal change), and that the affected cell generates hepatocytes when transplanted. A shift in the expression of a few genes characteristic of bone marrow cells and hepatocytes was shown in Fra25Lin “homed-recovered” cells examined in vitro and in vivo (S18); however, in addition to examining only a few genes, this study did not show that the gene alterations were clonal or that the same hematopoietic cell that expressed hepatocyte properties generated hepatocytes in vivo. The only evidence for the formation of hepatocytes by direct transdifferentiation of hematopoietic cells is the lack of demonstration of an alternate mechanism.

Transdifferentiation of mesodermal hematopoietic cells into endodermal hepatocytes is conventionally thought to be developmentally prohibited after germ layers are formed at gastrulation and, accordingly, should not occur in adult animals. However, processes that resemble transdifferentiation, such as mesenchymal–epithelial transformation32 and metaplasia,33 occur under pathological circumstances in adult animals and may be responsible. Mesenchymal–epithelial transformation might explain the acquisition of hepatocyte phenotypes and functions by some hematopoietic cells that emerge in culture from adherent cells that are thought to originate from bone marrow stromal cells.15 The nuclei of hematopoietic cells may be reprogrammed during prolonged cell culture,13–15 thereby increasing their plasticity to generate hepatocytes. Further studies are needed.

In contrast to the lack of evidence for generation of hepatocytes by direct transdifferentiation of hematopoietic cells, derivation of hepatocyte-like cells by the fusion of a hematopoietic cell with a host hepatocyte has received experimental validation in several studies in both normal and pathological livers of hematopoietic cell transplant recipients, and shown to occur at a frequency of 10−4 to 10−6. Fusion has been demonstrated in Fah−/− mice in 3 independent studies that used different methods to detect hybrid cells (S14,S15,S17). Furthermore, the hematopoietic nuclei contained in the heterokaryotic hybrid cells were shown to be reprogrammed to downregulate hematopoietic genes and upregulate hepatocytic genes (S15), thereby correcting the metabolic defect characteristic of Fah−/− hepatocytes. (Phenotypic correction would not occur if the hematopoietic cell nucleus, which contains an intact Fah gene, were simply extruded.) Fusion of hematopoietic cells and hepatocytes at a similar low frequency was also demonstrated in livers of healthy mice by means of Cre-lox recombination (S16). However, another study using Cre-lox recombination did not detect fusion of bone marrow cells and healthy hepatocytes but did detect such fusion in one animal exposed to a hepatotoxic chemical (S22). This result raises the possibility that damaged hepatocytes may be more “fusogenic” than are healthy hepatocytes.

Cells of the monocyte–macrophage lineage appear to be fusion partners of hepatocytes in both healthy mice (S16,S23) and in Fah−/− mice receiving NTBC (S17,S19). Of the few xenogeneic transplant studies that have attempted to recognize fusion between cells of two different species by detecting cells that contain mixtures of species-specific proteins or DNA sequences (S46,S48,S52,S55,S58), most examined an insufficient number of hepatocytes in recipient livers to exclude its occurrence at the expected low frequency. In a study in which human umbilical cord blood cells were transplanted into SCID mice, hepatocytes that contained either a mixture of murine and human chromosomes or only the latter were detected (S51), suggesting the simultaneous occurrence of fusion and transdifferentiation. However, eukaryotic hepatocytes were found to be derived from hybrid hepatocytes that were fusion products of Fah−/− hepatocytes and Fah+/+ bone marrow cells, indicating that some chromosome sets were eliminated from heterokaryotic cells (S14). This is a matter of considerable importance if fused hybrid cells are to be used in the treatment of liver diseases in humans, because in other circumstances aneuploidy is associated with neoplastic transformation.

Do Intrahepatic Hematopoietic Cells Contribute to Hepatocyte Function and Renewal by Mechanisms Other Than the Direct Formation of Hepatocytes?

Hematopoietic cells and the nonparenchymal liver cells derived from hematopoietic cells contribute to the maintenance and replacement of hepatocytes by indirect mechanisms that do not involve their direct transformation into hepatocytes. Hematopoietic cells resident in the liver produce cytokines and growth factors that support the differentiation and growth of hepatocytes during embryogenesis.3 In adult animals, cytokines and growth factors produced by nonparenchymal liver cells derived from bone marrow regulate proliferation and maintain function of hepatocytes.23, 24, 34

Several studies suggest that cytokines and growth factors produced by infused hematopoietic cells may support liver function and repair in adult animals without forming new hepatocytes from the infused cells. Notably, infusion of granulocyte colony-stimulating factor prevented the death of mice from CCl4 toxicity in the absence of hematopoietic cell transplants (S30). Infusion of autologous bone marrow cells into the left hepatic lobe of patients undergoing major resection of the tumor-containing right lobe was associated with augmented growth of the residual left lobes (S65). Although the direct transformation of the infused bone marrow cells into hepatocytes was not examined, the observed effect on liver growth appears to be greater than would be suggested from most experimental studies we have reviewed (S65). Other situations in which the outcomes were possibly not mediated by the direct transformation of hematopoietic cells into hepatocytes include reduction in CCl4-induced fibrosis (S29,S42) and the reduction of plasma copper levels in mice with a mutated WND gene (S25). Further studies are needed to separate the effects of hematopoietic cell transplantation that are mediated by growth factors and cytokines secreted by the transplanted cells and their nonparenchymal liver cell progeny and from the effects that depend on the transformation of hematopoietic cells into hepatocytes.

A Final Comment on the Generation of Hepatocytes From Hematopoietic Cells

Although we conclude that the data are sufficient to indicate that mesodermal hematopoietic cells can generate hepatocytes at a very low frequency, this is not an effective pathway under most conditions. One therefore might view the generation of endodermal hepatocytes from mesodermal hematopoietic cells as a biological curiosity, reflecting a low level of “leakiness” of an otherwise precisely controlled process of germ layer segregation. However, that nuclei of differentiated cell types can be experimentally reprogrammed to generate new cell lineages is well established.13, 14 Therefore, a fundamental question is this: Why do the multi-potential hematopoietic stem cells not differentiate into hepatocytes more efficiently? The answer to this question is not known. We suggest that the answer may relate to the importance of the tissue-specific stem cell niches as determinants of the specificity of differentiation of stem cells.35 Transplantation of hematopoietic stem cells either directly into an organ (e.g., the liver) or via the vascular system is unlikely to deposit them into organ-specific stem cell niches (with the exception of bone marrow transplants into bone marrow stem cell–ablated recipients). The fact that partially and/or fully differentiated hepatocytes are the most efficient cells in rebuilding the liver after transplantation of putative stem cells, either hematopoietic or hepatic, supports this hypothesis. Future challenges for improving the usefulness of easily accessible stem cells in regenerative medicine therefore may include deciphering the non-cell autonomous signals in local stem cell niches.