Little evidence of bone marrow-derived hepatocytes in the replacement of injured liver

Abstract

We have tested the ability of bone marrow (BM) cells (BMCs) to form hepatocytes in liver injury models. We used three models: (i) carbon tetrachloride (CCl₄) treatment, (ii) albumin-urokinase transgenic mouse [TgN(Alb1Plau)], and (iii) hepatitis B transgenic mouse [TgN(Alb1HBV)]. As a nonselective liver injury model, irradiated C57BL/6 (B6) mice were transplanted with BMCs from GFP transgenic mouse [TgN(ActbEGFP)] or β-galactosidase transgenic mouse [TgN(MtnLacZ)] followed by the administration of CCl₄. Irradiated TgN(Alb1HBV) and TgN(Alb1Plau) were also transplanted with BMCs from TgN(ActbEGFP) or TgN(MtnLacZ). Approximately 1.5 × 10⁶ hepatocytes per liver were analyzed for GFP-positive cells, and the whole livers were inspected for β-galactosidase expression. No GFP-positive hepatocytes and no gross blue staining of the livers with 5-bromo-4-chloro-3-indolyl β-d-galactoside in any of the 18 recipient mice analyzed were detected. The livers from female animals with gender-mismatched BM transplantation were also tested with Y chromosome fluorescent in situ hybridization analysis to detect donor-derived cells. A total of five isolated hepatocytes were positive for Y chromosome in 4.1 × 10⁵ hepatocytes analyzed. Our results demonstrate that there is little or no contribution of BMCs to the replacement of injured livers in these models. We conclude that BM-derived cells cannot generally lead to a cure of liver damage.

There have been a number of reports that show the potential of adult rodent bone marrow (BM) cells (BMCs) to transdifferentiate into a variety of cell types (1-7), including hepatocytes (8-10). One of the best examples for the transdifferentiation has been the ability of BM-derived hapatocytes to repopulate the liver of mice with fumarylacetoacetate hydrolase knockout mouse (FAH-/-) and correct the liver disease (10), although recent reports have shown that this correction of liver disease is caused by fusion of donor BM to recipient hepatocytes rather than transdifferentiation (11, 12). However, it is not fully elucidated whether or not the significant level of the contribution of BM-derived hepatocytes in the FAH-/- model, either in the form of transdifferentiation or cell fusion, can be generalized in other liver injury models. Here we report our results with nonselective liver injury model and two selective liver injury models {carbon tetrachloride (CCl₄), albumin-urokinase transgenic mouse [TgN(Alb1Plau)] (13), and hepatitis B transgenic mouse [TgN(Alb1HBV)] (14)}, to examine how general is the phenomenon of the repopulation by BM-derived hepatocytes. It is of considerable clinical importance to know the level of contribution of BM-derived hepatocytes in a slowly progressive chronic liver disease model like TgN(Alb1HBV), because chronic viral hepatitis is a leading cause of cirrhosis and hepatocellular carcinoma worldwide (15, 16).

Materials and Methods

Animals. TgN(Alb1HBV) and GFP transgenic mice [TgN(ActbEGFP)] (17) were kindly provided by F. V. Chisari (The Scripps Research Institute) and Masaru Okabe (Osaka University, Osaka), respectively. β-Galactosidase transgenic mouse [TgN(MtnLacZ)] (13) and TgN(Alb1Plau) were purchased from The Jackson Laboratory, and the latter was back-crossed to C57BL/6J (B6) (The Jackson Laboratory). Animal study protocols were approved by The Salk Institute Animal Care and Use Committee. Donor mice used in these studies were generally 7-8 weeks old, and recipient animals were 8-12 weeks old except for neonatal transplantation. Biochemical analysis of serum from animal was performed by ANILYTICS (Anilytics, Gaithersburg, MD).

BM Transplantation. BMCs were isolated from the femurs and tibias of TgN(ActbEGFP) or TgN(MtnLacZ). Erythrolysis was done by treating the cells with buffered ammonium chloride (StemCell Technologies, Vancouver). BM mononuclear cells were obtained by Histopaque-1077 (Sigma) density gradient centrifugation. Adult recipient mice were injected via the tail vein with these cells either with preparative irradiation (1,200 cG) or without irradiation.

Neonatal BM Transplantation. After whole-body irradiation with 200-400 cG, neonatal recipient mice (1-5 days of age) were injected with BMCs through the superficial temporal vein by using hypothermic anesthesia.

CCl₄ Liver Injury. B6 mice were i.p. injected with CCl₄ (0.02-1.0 ml/kg of animal weight, twice a week, total of eight times).

Choline-Deficient, Ethionine-Supplemented Diet. Mice were administered a diet consisting of a 1:1 mixture of choline-deficient chow (ICN) and normal chow and drinking water supplemented with 0.15% (wt/vol) DL-ethionine (ICN) for 2 weeks (18).

Tissue Preparation. Transplanted animals were anesthetized and perfused intracardially with 4% paraformaldehyde/PBS. Perfused livers were dissected and further fixed in 4% paraformaldehyde at 4°C overnight. In some cases, animals were killed, and thin liver slices (2-3 mm) were removed immediately. Portions of the liver slices were snap-frozen, and the other slices were fixed in 4% paraformaldehyde at 4°C overnight. Fixed liver slices were washed with PBS, cryoprotected by incubation in increasing concentration of cold sucrose (10%, 15%, and 20%; total 12-24 h), and quick-frozen in cryo-embedding compound (Microm International, Walldorf, Germany).

Histology and Immunofluorescence Analysis. For intrinsic GFP analysis, 10-μm frozen sections were cut from several distinct lobes and mounted with VECTASHIELD with 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories). Immunohisto-chemical stainings with anti-mouse albumin (Biogenesis, Brentwood, NH) or anti-GFP (Novus Biologicals, Littleton, CO) were performed on 10-μm frozen sections by using standard protocol. Appropriate Cy3-conjugated secondary reagents (Jackson ImmunoResearch) were used for the detection and mounted with VECTASHIELD with DAPI. Immunofluorescence analysis was performed by using a DeltaVision restoration microscopy system (Applied Precision, Issaquah, WA) consisting of an Olympus IX70 microscope equipped with HBO100 epiillumination source. Excitation wavelengths were 490 nm for GFP, 555 nm for Cy3, and 360 nm for DAPI. Fluorescent emission was collected at 528, 617, and 457 nm, respectively. In the analysis of GFP-positive cells, autofluorescent cells were excluded by examining red emission (617 nm) and green emission (528 nm). For detection of β-galactosidase, the whole lobes of livers were fixed in 0.2% paraformaldehyde/0.05% glutaraldehyde at 4°C overnight and stained with 5-bromo-4-chloro-3-indolyl β-d-galactoside (Sigma). For some samples, a small portion of the livers were removed and snap-frozen for fluorescent in situ hybridization before the fixation of the rest of the whole livers.

Fluorescent in Situ Hybridization. Cryostat sections, 6 μm in thickness, were fixed three times, 10 min each, in Carnoy's fixative. Serial ethanol dehydration was done, and the slides were air-dried at room temperature. Sections were denatured at 65°C for 1.5 min in prewarmed 70% formamide and 2× SSC solution, pH 7.0 and quenched in ice-cold 70% ethanol for 4 min. Dehydration by serial ethanol washing was done again. The mouse Y chromosome probe labeled with Cy3 (STAR FISH, Cambio, Cambridge, U.K.) was denatured at 65°C for 10 min and applied on the sections at 37°C. The sections were covered with parafilm and incubated overnight in a hydrated slide box at 37°C. Then, the sections were washed according to the manufacturer's instructions and mounted with VECTASHIELD with 4′,6-diamidino-2-phenylindole (Vector Laboratories). Fluorescence analysis was performed by using a DeltaVision restoration microscopy system (Applied Precision) as described above. In the analysis of Cy3 signal, autofluorescent nonspecific signals were excluded by examining green emission (528 nm) and red emission (617 nm).

Calculations of Section Size and Cell Numbers. At least three images of liver sections were analyzed to determine the average number of hepatocytes per unit area for each animal. The surface area of sections was measured by scanning glass slides along with a size standard and by analyzing the scanned images with METHAMORPH 6.1 (Universal Imaging, Downingtown, PA) software.

Results and Discussion

The first model we tested was the CCl₄-induced liver injury model. Recently, it has been suggested to be effective in inducing hepatocytic differentiation of BMC (11). Four female B6 mice were lethally irradiated and transplanted with male TgN(ActbEGFP) or TgN(MTnLacZ) BM mononuclear cells (1 × 10⁶). TgN(ActbEGFP) is known to express GFP constitutively in most cells including hepatocytes, and TgN(MTnLacZ) allows inducible expression of β-galactosidase in hepatocytes by administration of heavy metal ions. Donor engraftment (79-92%) was confirmed 4 weeks posttransplantation, and the animals were administered CCl₄ (0.02 ml/kg animal weight, twice a week for 4 weeks). The existence of liver injury was confirmed by the elevated serum alanine transaminase levels (146-217 units/liter). Four or 8 weeks after the last administration of CCl₄, the livers of the recipients were checked for donor-derived cells (Table 1). In the liver sections (50 sections per liver for GFP fluorescence and 10 sections per liver for GFP immunohistochemistry), GFP-positive cells were located in sinusoids or associated with other larger vessels. These cells showed several different types of morphology, including globular, elongated, or star-shaped cells, and were readily distinguishable from hepatocytes that are large polyhedral cells with round nuclei (Fig. 1A). Additionally, the GFP-positive cells were negative for the hapatocyte marker albumin by immunostaining. In animals transplanted with TgN(MTnLacZ), the whole livers was tested for β-galactosidase expression after induction with cadmium; however, no gross blue staining was detected (Fig. 1 B and C), suggesting that no distinct nodules consisted of donor BM-derived hepatocytes.

Table 1. Summary of experimental data.

Liver injury	No. of mice	Type and no. of donor cells	Donor derived cells in peripheral blood, %	Weeks from BM transplantation to analysis	Weeks from the end of CCl4 to analysis	Donorderived hepatocytes* or nodules observed	Y chromosomepositive hepatocyte nuclei/hepatocytes analyzed
Transplantation → CCl4 liver injury
0.02 ml/kg × 8 times	1	GFP (1 × 106 BMMNC)	95 (12 w)	12	4	0	1/130,000
0.02 ml/kg × 8 times	1	LacZ (1 × 106 BMMNC)	92 (4 w)	12	4	0	2/60,000
0.02 ml/kg × 8 times	1	GFP (1 × 106 BMMNC)	96 (12 w)	16	8	0	1/140,000
0.02 ml/kg × 8 times	1	LacZ (1 × 106 BMMNC)	91 (4 w)	16	8	0	1/80,000
CCl4 liver injury → transplantation (without preparative irradiation)
0.02 ml/kg × 8 times	2	GFP (1 × 105 BMC)	0 (4 w)	4	4	0
0.02 ml/kg × 8 times	2	LacZ (1 × 105 BMC)	0 (4 w)	4	4	0
0.2 ml/kg × 8 times	2	GFP (1 × 107 BMC)	ND	4	4	0
1.0 ml/kg × 8 times	2	GFP (1 × 107 BMC)	0 (2 w), ND	2,4	2,4	0
TgN(Alb1Plau)	1 (neonate)	GFP (1 × 106 BMC)	17 (13 w)	15		0
TgN(Alb1HBV)	1	GFP (1 × 107 BMC)	44 (11 w)	13		0
TgN(Alb1HBV)	1	GFP (1 × 107 BMC)	97 (7 w)	23		0
TgN(Alb1HBV)	1	LacZ (5 × 106 BMC)	≈100 (7 w)	32		0
TgN(Alb1HBV) + CDE†	1 (neonate)	GFP (1 × 106 BMC)	30 (9 w)	47		0
TgN(Alb1HBV) + CDE†	1 (neonate)	GFP (1 × 106 BMC)	18 (9 w)	47		0

ND, not done; BMMNC, BM mononuclear cells.

^*Sixty sections containing 1.5 × 10⁶ hepatocytes on average were analyzed per liver.

^†CDE, choline-deficient, ethionine-supplemented diet.

Fig. 1.

(A) GFP-positive nonhepatocytes observed in a liver section from CCl₄-treated mouse. Bromo-4-chloro-3-indolyl β-d-galactoside (X-gal) staining of the whole liver from TgN(MtnLacZ) as a positive control (B) and B6 transplanted with TgN(MtnLacZ) BM mononuclear cells and administered CCl₄ (C). No X-gal positive nodule was observed in the B6 liver. (D) Y chromosome-positive hepatocyte nucleus (arrow) from a mouse in CCl₄ liver injury model. (Scale bars: 50 μm, A; 30 μm, D.)

The livers from these four animals were also tested for Y chromosome-positive hepatocytes, because Mezey et al. (19) suggested a discrepancy between transgenic expression tagging and DNA markers might occur. In 40 liver sections (10 sections for each animal), a total of five isolated hepatocytes were positive for Y chromosome (Table 1 and Fig. 1D). A nonirradiated protocol was also tested, in which eight B6 mice were first administered CCl₄ eight times, then the protocol was followed by transplantation of TgN(ActbEGFP) or TgN(MTnLacZ) BMC (Table 1). Protocols differing in the number of transplanted cells and the dosage of CCl₄ administered were tested. Not surprisingly there was no donor engraftment in the nonradiated animals. Four weeks after transplantation, the liver of each animal was analyzed for donor-derived GFP- or β-galactosidase-positive hepatocytes. No donor-derived hepatocytes were observed in any of the recipients.

The TgN(Alb1Plau), in which regeneration stimulus for hepatocytes is present for 6-8 weeks after birth (13), was tested as a selective liver injur y model. An irradiated neonate TgN(Alb1Plau) was transplanted with GFP-positive BMCs (1 × 10⁶). After having confirmed the donor cell engraftment (17%), the recipient liver was assessed for donor-derived cells at 15 weeks of age. There were no GFP-positive hepatocytes observed in 60 sections of the liver (Table 1).

Another selective liver injur y model we tested is TgN(Alb1HBV), which produces toxic quantities of hepatitis B surface antigen within the hepatocytes. Two TgN(Alb1HBV) were lethally irradiated and transplanted with GFP-marked BMCs (1 × 10⁷) from TgN(ActbEGFP). After having confirmed the donor cell engraftment (44% and 97%) and the existence of mild liver injury by slightly elevated serum alanine transaminase levels (121 and 143 units/liter), the livers of the recipient mice were analyzed for GFP-positive cells at 13 and 23 weeks posttransplantation. However, none of the liver sections contained GFP-positive donor-derived hepatocytes (Table 1).

One female TgN(Alb1HBV) was transplanted with 5 × 10⁶ BMCs derived from male TgN(MTnLacZ). Seven weeks posttransplantation, ≈100% of the recipient peripheral blood cells were positive for the Y chromosome and elevated serum alanine transaminase level (185 units/liter) was confirmed. Thirty-two weeks after transplantation, the whole liver was examined for β-galactosidase expression after induction with cadmium. However, nodules originating from donor-derived hepatocytes were not observed (Table 1).

We next transplanted GFP-positive BMCs (1 × 10⁶) into two irradiated neonatal TgN(Alb1HBV) for long-term follow-up. These mice were administered a choline-deficient, ethionine-supplemented diet for 2 weeks at 5 months posttransplantation. This diet is known to induce proliferation of an intrahepatic progenitor population, “oval cells,” of which potential source is thought to be BM (8). These mice were kept for 6 more months to allow the induced oval cells to differentiate further. At 47 weeks posttransplantation, the recipient livers were tested for donor-derived hepatocytes. In 60 sections from each mouse (50 for GFP fluorescence and 10 for GFP immunohistochemistry), there were no GFP-positive hepatocytes in either of the recipients.

It is possible that, in this model, primary hepatocyte transplantation will be as unsuccessful to replace injured liver as BM transplantation, probably because the liver injury level is not so severe and the recipient hepatocytes could repair the injury. However, the situation is different between BM transplantation and hepatocyte transplantation. In the BM reconstituted model, donor BM-derived hematological cells are continuously supplied, circulate in the recipient liver, and have a lifelong chance to compete with endogenous defective hepatocytes. Our data show that even with this continuous chance BMCs failed to repopulate the diseased liver in the TgN(Alb1HBV) model.

Thus, analysis of 18 mice in the nonselective and even in selective liver injury models revealed only five isolated hepatocytes that might have been derived from donor BMCs, although it still remains to be elucidated whether these cells arise from spontaneous fusion events (12, 20) or transdifferentiation. We conclude from these data (Table 1) that the contribution of BM-derived hepatocytes to the replacement of injured livers may be very low except for a certain limited experimental condition (10, 12, 20). We do not believe that BM-derived hepatocytes can generally lead to a cure of liver disease.