Liver progenitor/oval cells differentiate into hepatocytes and biliary epithelial cells, repopulating the liver when the regenerative capacity of hepatocytes is impaired. Recent studies have shown that hematopoietic bone marrow (BM) stem/progenitor cells can give rise to hepatocytes in diseased/damaged liver. One study has reported that BM cells can transdifferentiate into liver progenitor/oval cells, but it has not been proven that the latter can repopulate the liver. To answer this question, we have lethally irradiated female DPP4− mutant F344 rats and transplanted them with 50 million wild-type male F344 BM cells. One month after transplantation, the recipient BM was reconstituted with male hematopoietic cells, determined by quantitative polymerase chain reaction using primers for Y chromosome–specific sry gene. In addition, DPP4+ cells, single or in clusters and predominantly in the periportal region, were detected in all liver sections of recipient rats. Animals were subjected to the following three different liver injury protocols for activation and expansion of oval cells: D-galactosamine, retrorsine/partial hepatectomy (Rs/PH), and 2-acetylaminofluorene/partial hepatectomy (2-AAF/PH). In all three models, prominent expansion and accumulation of cytokeratin 19–positive (CK-19+) oval cells was observed. However, most of the DPP4+ clusters dispersed over time, and their total number decreased. Very few oval cells (less than 1%) showed double DPP4/CK-19 labeling. None of the small hepatocytic clusters in the Rs/PH or 2-AAF/PH model were comprised of DPP4+ cells. These data demonstrate that the sources of oval cells and small hepatocytes in the injured liver are endogenous liver progenitors and that they do not arise through transdifferentiation from BM cells.
Over the years, substantial evidence has accumulated suggesting the existence of potential liver stem cells (LSCs) in the adult liver. In all cases, the putative LSCs were activated to proliferate and differentiate when the regenerative capacity of terminally differentiated hepatocytes was compromised. The progeny of potential LSCs, referred to as oval cells, behave like bipotential progenitors capable of differentiation into mature hepatocytes and biliary epithelial cells, thus recapitulating hepatoblast differentiation during fetal development [1–5]. Oval cells also reveal some phenotypic characteristics of hematopoietic progenitor cells; they express c-kit and its ligand stem cell factor  and the related flt-3 and flt-3 ligand , CD34 , Thy-1 (CD90) , and Sca1 . In addition, oval cells are phenotypically heterogeneous [11,12], and some investigators identified them in the periductular/intraportal zone, consistent with the interpretation that some may originate from an extrahepatic (bone marrow [BM]) source [13,14].
Extensive research during the past few years has changed our view concerning stem cells. It has become evident that stem cells are very pliable and that they can change their phenotype when taken from the stem cell niche and transplanted into a new residence [15–18]. These studies raised the possibility that BM stem cells may be useful for liver cell transplantation and prompted several investigators to study whether BM progenitors brought into the liver can transdifferentiate into hepatocytes. In general, the protocols for these studies used female animals, lethally irradiated and rescued with male genetically labeled male BM cells or purified hematopoietic stem cells (HSCs). In most cases, the animals were subjected to liver injury, and after different periods of time (ranging from several days to months), the appearance of Y chromsosome–positive hepatocytes in their livers was observed [19–23]. Similar results were obtained with humans: appearance of Y chromosome–positive hepatocytes in livers of female patients that had received male BM transplants and also female livers grafted into male patients [24–27].
However, recent studies carefully analyzing the fate of transplanted BM cells or purified hematopoietic cell fractions showed that a single green fluorescent protein–positive HSC can reconstitute the BM of lethally irradiated nontransgenic recipients but did not contribute appreciably to the non-hematopoietic tissues, including the liver . Convincing evidence that cell fusion between donor HSC and endogenous hepatocytes accounts for the high liver repopulation by HSC in the fumaryl acetoacetate hydrolase deficient (FAH−/−) model of massive liver injury has been reported from the laboratories of Russell and colleagues  and Grompe and colleagues .
If BM stem cells can engraft and differentiate into oval cells in the liver, they would be a valuable source for gene therapy. Petersen et al.  transplanted syngeneic BM cells into lethally irradiated female animals treated with 2-acetyl-aminofluorene (2-AAF) and CCl4, which causes hepatic necrosis and impairment of hepatocyte proliferation, and reported the appearance of BM-derived oval cells in the liver of these animals.
To determine unambiguously whether BM cells can differentiate into oval cells and repopulate the injured liver, we have substituted the BM of lethally irradiated female DPP4-deficient F344 rats with BM cells from syngeneic normal male F344 rats and then subjected the recipients to three models of activation and expansion of oval cells. We found that endogenous liver progenitors, and not BM progenitors, are the overwhelming source of expanding oval cells in the injured liver.
Materials and Methods
Male DPP4+ F344 rats, 8–10 weeks of age, purchased from Taconic Farms, German Town, NY, were used for isolation of donor BM cells. Female mutant DPP4− F344 rats (cell transplantation recipients) were provided by the Special Animal Core of the Liver Research Center at the Albert Einstein College of Medicine. Animals were maintained according to approved institutional animal care protocols.
Total Body Irradiation
DPP4− F344 rats, weighing between 180 and 200 g, received total body irradiation (TBI; 10.5 Gy [LD100] at 78.7 cGy per minute) in a Lucite chamber using a Cesium-137 gamma-ray irradiator (Mark I irradiator Model 68 with a 6,000-Ci source [J.L. Shephard, San Fernando, CA] with a rotating plate at 5 rpm (dose variation between the isocenter of the rotating plate edge of the lucite chamber at ≤1%). A uniform dose (± 1%) was obtained with either instrument, as determined by thermal luminescent dosimetry. Animals were turned over (anterior to posterior) in the middle of each γ-irradiation for whole-body dose uniformity.
The hind limbs of DPP4+ F344 male rats were used for isolation of BM cells. The cell suspension was passed through a 70-mm cell strainer, and the cells were collected by centrifugation at ×400 g for 10 minutes at 4°C. The cell pellet was suspended in 5 ml of lysis buffer (150 mM NH4Cl, 10 mM KHCO3, 0.1 mM EDTA) and incubated for 5 minutes at 4°C. The cells were washed two times in RPMI 1640 medium, resuspended in the same medium, and counted.
The rats were transplanted through the tail vein with 5 × 107 freshly isolated BM cells 24 hours after TBI.
Experimental animals were divided into three groups. To the first group of animals (n = 10), one dose of 950 mg/kg body weight of D-galactosamine (D-gal) was given i.p. 6 weeks after BM transplantation (BMT). Animals were euthanized on days 4 and 5 thereafter. Two control animals did not receive D-gal (Fig. 1).
In the second animal group (n = 11), 4 weeks after BMT, 10 rats received two doses of retrorsine (Rs) 2 weeks apart, each of 30 mg/kg body weight i.p. Four weeks after the second dose, two-third partial hepatectomy (PH) was performed on all animals. Animals were euthanized on days 1, 2, 3, 5, 7, 10, 14, 20, and 30 after PH. One control rat did not receive Rs and was euthanized 4 weeks after BMT.
In the third set of animals (n = 26), 2-AAF pellets (2 × 35 mg pellets/animal, 14-day release [Innovative Research of America, Sarasota, FL]) were implanted subcutaneously 6 weeks after BMT. PH was performed on day 7 after 2-AAF treatment. Animals were euthanized on days 6, 8, 10, 12, 14, and 16 after PH. Control animals did not receive 2-AAF.
Peripheral blood was collected from all animals 1 month after BMT, at the time of PH and euthanasia.
BM Reconstitution: Isolation of DNA and Real Time–Polymerase Chain Reaction
Rat blood DNA from the recipient animals at different time points was isolated with a standard phenol/chloroform procedure and purified using the DNEasy kit (Qiagen Inc., Valencia, CA). Serial dilutions of control male DNA were prepared beginning with 50 ng per reaction and used as a standard.
Rat-specific Sry primers were as follows: primer 1, (5′-catcgaagggttaaagtgcca-3′); primer 2, (5′-atagtgtgtaggttgttg tcc-3′). These primers amplified a 104-bp fragment, as previously described . Rat-specific GAPDH primers were as follows: primer 1, (5′-ggcattgctctcaatgacaa-3′); primer 2, (5′-atgtaggccatgaggtccac-3′). GAPDH primers amplified a stretch of 94 bp of the rat GAPDH gene (accession number NM_017008). Quantitative polymerase chain reaction (PCR) was carried out using SYBR Green PCR Master Mix (Applied Biosystems, Warrington, U.K.), 0.5 μM of each primer, and 10 to 100 ng of template DNA on ABI Prism sequence detection system 7000 (Applied Biosystems). The relative ratio and standard deviation between Sry and GAPDH (reconstitution ratio) were calculated using the comparative CT method (ΔΔ CT value), as recommended by the ABI Prism Sequence detection system 7700.
Histochemistry and Immunohistochemistry
DPP4 expression was determined by enzyme histochemistry on 5-μm liver cryosections, as previously reported . For all other procedures, the slides were fixed in ice-cold 4% paraformaldehyde.
Cytokeratin 19, CD45, and CD90 Immunohistochemical Detection
Endogenous peroxidase and alkaline phosphatase were blocked with KPL blocking solution (Kirkegaard & Perry Lab. Inc., Gaithersburg, MD). Sections processed for cytokeratin 19 (CK-19) were permeabilized in 0.3% Triton X-100 in phosphate-buffered saline, pH 7.4. Blocking was with 2% goat serum, 1% bovine serum albumin (BSA), and 0.05% Tween-20. The primary antibodies were anti-rat CK-19 monoclonal antibody (Novocastra Laboratories Ltd, Newcastle, U.K.) and anti-rat CD45 and anti-rat CD90 monoclonal antibodies (both from BD Biosciences, San Jose, CA). Secondary antibodies were goat anti-mouse immunoglobulin G (IgG) horseradish peroxidase–conjugated or alkaline phosphatase–conjugated (Sigma). Peroxidase was developed with diamin-obenzidine (DAB) as substrate and alkaline phosphatase with Histomark (Kirkegaard & Perry) as substrate.
CD26/CK-19 Double-Immunochemical Staining
Endogenous peroxidase and alkaline phosphatase were blocked as above, followed by treatment with 0.3% Triton X-100 and blocking in 2% rabbit serum containing 0.05% Tween-20. Endogenous biotin was blocked using a commercial avidin/biotin blocking kit (Vector Laboratories, Burlingame, CA). Sections were stained with mouse monoclonal anti-CD26 (anti-DPP4, Biotrend Chemikalien GmbH, Koln, Germany) and mouse anti-CK-19 (Novocastra) as primary antibodies. Secondary antibodies were biotin-conjugated rabbit anti-mouse IgG2a and peroxidase-conjugated rat anti-mouse IgG1 (Zymed Laboratories Inc., San Francisco). The biotynlated antibody was used in combination with the Vectastain ABC-AP kit (Vector Laboratories). Alkaline phosphatase activity was developed using Histomark red (Kirkegaard & Perry) and peroxidase activity with DAB as substrates.
CD26/CK-19, CD26/CD45, and CK-19/CD45 Immunofluorescent Double Labeling
After fixation, the sections were incubated in sodium borohydride followed by blocking in a solution containing 2% goat serum, 1% BSA, and 0.05% Tween-20 for CD26/CK-19 labeling or 2% goat serum, 2% rabbit serum, 1% BSA, and 0.05% Tween-20 for CD26/CD45 and CK-19/CD45 detection. Endogenous biotin was blocked with the avidin/biotin blocking kit (Vector Laboratories).
For CD26/CK-19 and CD26/CD45 simultaneous detection, mouse anti-CD26 (anti-DPP4, Biotrend, Cologne, Germany) and mouse anti-CK-19 (Novocastra) or mouse anti-CD45 (BD Biosciences) were used as primary antibodies. The secondary antibodies were biotin-conjugated goat anti-mouse IgG2 and fluorescein isothiocyanate (FITC)–conjugated goat anti-mouse IgG1 (Nordic Immunological Laboratories, Tilburg, Netherlands) or FITC-conjugated rabbit anti-mouse IgG1 (Rockland Inc., Gilbertsville, PA). For CD26 detection, sections were stained with Cy3-conjugated goat anti-biotin (Rockland Inc.) as a third antibody.
For CK-19/CD45 double staining, sections were exposed to a mixture of mouse monoclonal anti-CK-19 (Progen Biotechnik GmbH, Heidelberg, Germany) and mouse monoclonal anti-CD45 (BD Biosciences). The secondary antibodies were biotin-conjugated goat anti-mouse IgG2 (Nordic) and FITC-conjugated rabbit anti-mouse IgG1 (Rockland). For CK-19 detection, Cy3-conjugated goat anti-biotin (Rock-land Inc) was used as a third antibody.
To be able to determine whether BM cells differentiate into liver progenitor/oval cells, we ablated the BM of female DPP4-deficient F344 rats and reconstituted it by transplanting 5 × 107 normal male syngeneic F344 rat BM cells 24 hours after TBI (Fig. 1). After BM reconstitution was complete (4–6 weeks), recipient animals were treated with chemical agents that cause liver injury and impaired hepatocyte proliferation, resulting in activation, proliferation, and accumulation of oval cells. The recipient animals were divided into three experimental groups, and each group was subjected to one of the following liver injury protocols: D-gal injection, which causes hepatocyte necrosis and proliferation of oval cells with a peak on days 4 to 5 [34,35]; Rs/PH, which causes maximum expansion of oval cells between days 7 and 14 after PH [36,37]; and 2-AAF/PH, which results in strong proliferation of oval cells with maximum accumulation in the liver on days 10 through 12 [38,39].
In the liver of wild-type F344 rats, all epithelial cells express the enzyme DPP4, including oval cells. As shown in Figure 2, in the livers of wild-type, DPP4+ animals subjected to 2-AAF/PH (Figs. 2A–2C), Rs/PH (Figs. 2D–2F), and D-gal (Figs. 2G–2H) treatment, both oval cells (Figs. 2B, 2E, 2H) and small basophilic hepatocytes (SBHs), considered by some investigators to represent a unique liver progenitor cell population , reveal a positive histochemical reaction for DPP4 (Figs. 2C, 2F). If BM progenitors were the source of oval cells, we would expect to see numerous DPP4+ oval cells in all three models of liver injury in which hepatocyte proliferation is compromised.
To be able to determine accurately whether BM progenitor cells transdifferentiate into oval cells, we used male BM cells of normal F344 rats and transplanted them into female recipients deficient in the exopeptidase DPP4. Because we did not know in advance which fraction of BM cells might possess the capability to transdifferentiate into oval cells, we transplanted 50 × 106 unfractionated BM cells into lethally irradiated female rats. Reconstitution of the female BM with male hematopoietic cells was determined using the Y chromosome–specific sry gene. BM substitution in all experimental groups was at least 80%, determined 30 days after transplantation, at the time of PH or at the time of animal euthanasia. These results ensured that if BM cells can transdifferentiate into oval cells, we should be able to detect them as DPP4+ oval cells in the liver of BM transplant recipients.
DPP4+ Cells in the Liver of Recipient Animals
The serine exopeptidase DDP4 (CD26) was described as a cell-surface molecule expressed on the surface of resting and activated T cells, activated B cells, and activated NK cells [41,42]. Upon reconstitution of the BM of DPP4-deficient animals with normal BM cells, we expected to find DPP4+ cells in the liver of the recipient rats. Indeed, 4–6 weeks after BMT, a substantial number of such cells were observed, especially as clusters in the portal region of the liver lobules of untreated rats (Fig. 3A) and at the time of partial hepatectomy (Fig. 3B). It has to be noted that after the various injury protocols for activation and proliferation of oval cells, the number of DPP4+ cells did not increase. In contrast, most cells from the DPP4+ periportal clusters dispersed throughout the liver lobule and appeared as single cells, including 2-AAF/PH (Fig. 3C), D-gal (Fig. 3D), and Rs/PH (Fig. 3E).
The 2-AAF/PH and Rs/PH models are also characterized by the appearance and expansion of foci of SBH (Figs. 3G, 3H). As could be clearly seen, the clusters of small hepatocytes that could be considered as hepatocytic progenitor or transitional cells do not contain DPP4+ small hepatocytes (compare with Figs. 2C, 2F), although scattered DPP4+ cells could be seen at the border of these clusters. Consequently, SBHs in these models have an endogenous origin and do not originate from the transplanted BM cells.
Because the DPP4+ cells by their location and appearance could be either liver progenitor/oval cells or blood cells, we compared the distribution of the oval cells and the DPP4+ cells in the liver of the three experimental models.
Expansion of Oval Cells in the Liver Injury Models
One of the phenotypic landmarks of oval cells is the expression of the intermediate filament CK-19. As shown in normal liver (Fig. 4A) or in the 2-AAF/PH liver at the time of PH (Fig. 4B), only biliary epithelial cells are positive for CK-19. Oval cells that appear in all three models are small with a large nucleus-to-cytoplasmic ratio. They are CK-19+ and form duct-like/arborizing structures that are readily distinguishable from the other small cells. In our experiments, the number of oval cells peaked in the 2-AAF model between days 9 and 13 after PH (Fig. 4C), in the D-gal model on day 5 (Fig. 4D), and in the Rs/PH model between days 7 and 14 (Fig. 4E), although some oval cells were still present 30 days after PH (Fig. 4F). In both 2-AAF/PH and Rs/PH livers, the clusters of small hepatocytes were CK-19− (Figs. 4G, 4H) as previously reported . The same results were obtained with another antibody specific for oval cells, OV-6 (data not shown). Notably, the high number of oval cells in these tissues did not coincide with the low number of scattered DPP4+ cells. These results, combined with those presented in Figure 4, show that (a) in all three models, the oval cells were activated and proliferated and accumulated, whereas the number of DPP4+ cells did not increase after the injury, and (b) oval cells divided rapidly on activation and formed groups and branching structures of CK-19+ cells, which were not evident with the DPP4+ cells, most of which appeared as single cells. Taking these data into account, we concluded that it is not likely that BM progenitors transdifferentiate into oval cells in the Rs/PH, D-gal, or 2-AAF/PH injured liver.
Nature of the DPP4+ Cells in the Liver
The DPP4+ Cells in the Liver Are Blood Cells
If the DPP4+ cells in the liver were not oval cells, then the question arises as to whether they are blood cells (mainly T cells) expressing this enzyme. All leukocytes express the surface molecule CD45. We analyzed serial liver sections from animals euthanized 6 weeks after BM cell transplantation before the liver injury, when the DPP4+ cells appear in clusters in the periportal region, and found that the DPP4+ cell in these clusters coexpressed CD45, showing that these cells were in the hematopoietic lineage (Figs. 5A, 5B). However, most of these cells did not express CD90 (Fig. 5C), which is considered a marker for oval cells . The same result was obtained using double-fluorescent immunolabeling for simultaneous detection of DPP4 and CD45 on liver sections from animals subjected to Rs/PH (Fig. 6A) or D-gal (Fig. 6B) treatment and from animals subjected to the 2-AAF/PH protocol (Fig. 6C, liver tissue taken at the time of PH; and Fig. 6D, 16 days after PH). These data provide substantial evidence that the DPP4+ cells in the liver are hematopoietic cells.
Oval Cells Are CD45−
In consecutive analyses conducted with all three models of liver injury, we found that the accumulating CK-19+ oval cells (red color) do not coexpress the leukocyte-specific marker CD45 (green color). Shown in Figure 6 is the Rs model at the time of PH (Fig. 6E), the Rs model 30 days after PH (Fig. 6F), the D-gal model 4 days after injection of the agent (Fig. 6G), and the 2-AAF model 10 days after PH (Fig. 6H). It should be noted, however, that in all three cases, substantial accumulation of CD45+ blood cells was observed. These data showed that oval cells expanding in the above models are not CD45+ and that they are distinct from the donor (DPP4+) leucocytes accumulating in the livers of injured animals.
DPP4+ Cells Are Almost Exclusively CK-19−
To obtain final proof that the DPP4+ cells in the liver of the transplanted animals are not oval cells, we performed double labeling to detect coexpression of CK-19 and DPP4. CK-19+ cells were labeled with FITS (green color), whereas the DPP4+ cells were labeled with Cy3 (red color). The results, presented in Figures 7A and 7B, show that the label for the DPP4+ cells found in the liver 6 weeks after transplantation or at the time of PH (Rs/PH model) did not overlap with that of CK-19+ cells. Very rarely we were able to detect coexpression of CK-19 and DPP4 in single cells, as shown in Figure 7C (2-AAF model 10 days after PH) and Figure 7F (Rs model 30 days after PH). In most sections examined, such coexpression of CK-19 and DPP4 was not observed, as shown in Figure 7D (2-AAF model 16 days after PH) or Figure 7E (D-gal model 4 days after injection of the agent).
To determine more precisely the frequency of transdifferentiation of BM progenitors into oval cells of the liver, we counted the number of oval cells that were positive for both CK-19 and DPP4 relative to the total number of DPP4+ cells. As shown in Table 1, the number of double-labeled cells (DPP4+ and CK-19+) was less than 1% of all DPP4+ cells in the livers of the control animals before the liver injury and in the livers of rats subjected to the three models of expansion of oval cells, D-gal, Rs/PH, and 2-AAF-PH. These results showed again that very few BM cells appeared as oval cells in the liver and that transdifferentiation of BM progenitors into oval cells is a very rare event, if it occurs at all.
Table Table 1.. Quantitation of DPP4+ cells and of double-labeled DPP4+ and CK-19+ cells in the liver of recipient animals after bone marrow transplantation in different experimental models
Liver sections from control animals (30 days after bone marrow transplantation) and from animals subjected to the three different models of liver injury were stained for DPP4 and CK-19 as described in Materials and Methods. Time after liver injury is given in parentheses. The percentage of DPP4+/CK-19+ cells was determined by counting the cells in three independent sections from each animal model.
Tissue-specific stem cells are of growing interest in the field of biomedical research. These cells can be regarded as a valuable source for organ repopulation and gene therapy. However, liver stem cells originating from the quiescent liver, which can be used in gene therapy protocols, have not yet been identified and isolated. More encouraging would be the use of their progeny, liver progenitor/oval cells that proliferate and accumulate in certain liver injury models. Because the source of oval cells in the liver was not determined unambiguously and some authors suggested that a subclass of these cells may originate in the BM [13, 14, 31], it was very important to evaluate whether BM can indeed serve as a source of oval cells in the liver.
In this study, we tried to determine the origin of oval cells and more specifically whether BM cells serve as a source of oval cells in different models of activation, proliferation, and differentiation of liver progenitor cells. To answer this question unequivocally, we substituted the BM of lethally irradiated female rats with genetically marked (DPP4+) male BM cells so that we could follow the appearance of DPP4+ cells in the liver. Hepatocytes, bile duct cells, and oval cells all express this enzyme. Using double immunohistochemical and immunofluorescent labeling, we found that the accumulating oval cells reveal the characteristic marker of oval cells, CK-19, but were DPP4− for the most part (<1% were DPP4+). On the other hand, all DPP4+ cells coexpressed CD45, proving that they remained in the hematopoietic lineage.
We studied the appearance and origin of the oval cells over a wide range of time, from 47 days (D-gal model) to 105 days after BMT (Rs/PH model). The same result was obtained with earlier and later activation of oval cells, and in almost all cases, BM progenitors that engraft into the liver do not transdifferentiate into oval cells. These data show in addition that the observed hepatocytes originating from BM cells are not a product of oval cell maturation and differentiation but appear by a different mechanism, most probably through fusion of BM progenitors with resident hepatocytes, as reported in the mouse FAH−/− model of massive and continuous liver injury [29,30]. We did not detect any DPP4+ cells in the foci of small hepatocytes, considered to be hepatocyte progenitors or transient cells that appear as foci of small basophilic hepatocytes in both the Rs/PH and 2-AAF/PH models. These data clearly demonstrated that BM cells do not transdifferentiate into small hepatocytes.
A study addressing this same question was conducted and recently published by Grompe and coworkers . Using a DDC 3′5-diethoxycarbonyl-1, 4-dihydrocollidine model to activate proliferation of oval cells in mice, the authors showed that the oval cells accumulating in the liver of DDC-treated mice have an endogenous origin and that they are not the product of transdifferentiation of BM progenitor cells.
Our finding in rats and those of Wang et al.  in mice provide substantial evidence that oval cells in the rodent liver can be considered the specific progeny of liver stem cells and not the progeny of hematopoietic stem cells. Efforts should be made to isolate a pure fraction of these cells from the liver and use them in transplantation experiments for liver repopulation and gene therapy.
This research was supported by the National Institutes of Health Grants R21 DK61145 (to M.D.D.). The authors thank Ethel Hurston for technical assistance.