Morphological and biochemical characterization of a human liver in a uPA-SCID mouse chimera


  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 703


A small animal model harboring a functional human liver cell xenograft would be a useful tool to study human liver cell biology, drug metabolism, and infections with hepatotropic viruses. Here we describe the repopulation, organization, and function of human hepatocytes in a mouse recipient and the infections with hepatitis B virus (HBV) and hepatitis C virus (HCV) of the transplanted cells. Homozygous urokinase plasminogen activator (uPA)-SCID mice underwent transplantation with primary human hepatocytes, and at different times animals were bled and sacrificed to analyze plasma and liver tissue, respectively. The plasma of mice that were successfully transplanted contained albumin and an additional 21 human proteins. Liver histology showed progressive and massive replacement of diseased mouse tissue by human hepatocytes. These cells were accumulating glycogen but appeared otherwise normal and showed no signs of damage or death. They formed functional bile canaliculi that connected to mouse canaliculi. Besides mature hepatocytes, human hepatic progenitor cells that were differentiating into mature hepatocytes could be identified within liver parenchyma. Infection of chimeric mice with HBV or HCV resulted in an active infection that did not alter the liver function and architecture. Electron microscopy showed the presence of viral and subviral structures in HBV infected hepatocytes. In conclusion, human hepatocytes repopulate the uPA+/+-SCID mouse liver in a very organized fashion with preservation of normal cell function. The presence of human hepatic progenitor cells in these chimeric animals necessitates a critical review of the observations and conclusions made in experiments with isolated “mature” hepatocytes. Supplementary material for this article can be found on the HEPATOLOGY website ( (HEPATOLOGY 2005;41:847–856.)

Despite decades of research, the processes that govern liver development and regeneration are only partially understood. A good understanding of the mechanisms that play a role in these processes is important because it may lead to new treatments of human liver diseases. Several in vivo experimental conditions have been used to study liver regeneration and repair and the interaction of different cell types in these processes. These include partial hepatectomy, administration of toxic compounds, or a combination of both, and transgenic expression of certain proteins.1, 2 In contrast to in vitro experiments, these approaches raise fewer questions concerning the influence of artificial matrices on the function and behavior of hepatocytes and other liver cell types.

Although these procedures have generated useful information on rodent liver regeneration, stem cell activation, and other processes, extrapolating these findings to the human situation would be imprudent. Therefore, a suitable small animal model to study human liver regeneration is needed.

Successful transplantation of human hepatocytes into mice or rats requires that the recipient animals do not reject the graft and provide a “supportive niche” that promotes engraftment and expansion of the liver cells.3 The former condition can be attained by using recipient animals that have an inborn (genetic) or acquired (drug-induced) immune deficiency. An environment that favors hepatocyte engraftment is encountered in animals with severe, chronic liver disease caused by the overexpression of a “noxious” protein, as in the urokinase plasminogen activator (uPA)-transgenic mouse.4

uPA transgenic mice backcrossed with “severe immune deficient” mice, such as Swiss athymic nude (nu/nu) mice,5 recombination activation gene 2 (RAG-2) knockout mice,6, 7 or SCID/beige mice,8 allow the engraftment and repopulation by xenogeneic hepatocytes.

In 2 of these studies, hepatocytes of human origin were used that could be infected with hepatitis B virus (HBV)7 or hepatitis C virus (HCV)8 after colonization. These 2 studies provided ample molecular and serological proof of successful repopulation and infection of human hepatocytes. However, they paid less attention to the architecture and organization of the human hepatocytes in this xenogeneic environment.

We backcrossed the same Alb-uPA mouse on the immunodeficient SCID mouse (CBySmn.CB17-Prkdcscid) and examined with great care the repopulation, organization, and functional behavior of the transplanted cells using immunohistochemical, ultrastructural, and serological methods. We infected chimeric animals with HBV and HCV and examined the infected livers in a similar fashion.


uPA, urokinase plasminogen activator; HBV, hepatitis B virus; HCV, hepatitis C virus; HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; PAS, periodic-acid-Schiff.

Materials and Methods

Breeding of the uPA-SCID Mice

B6SJL-TgN(Alb1Plau)144Bri mice were, at least 7 times, back-crossed on CBySmn.CB17-Prkdcscid mice. Both strains were purchased from The Jackson Laboratories (Bar Harbor, ME). Screening for the SCID background was done with an in-house mouse immunoglobulin M (IgM) sandwich ELISA. Genotyping of the uPA-SCID mice was performed via a multiplex polymerase chain reaction as described before.9 Human hepatocytes were transferred only in animals homozygous for the uPA-transgene. The study protocol was approved by the Animal Ethics Committee of the Faculty of Medicine of the Ghent University.

Isolation of Human Hepatocytes and Transfer into uPA-SCID Mice

Human liver specimens were collected from patients undergoing a partial hepatectomy for the resection of metastatic disease. All patients gave written, informed consent, and all experiments were approved by the Ethical Committee of the Ghent University Hospital. Hepatocytes were isolated from tumor-free sections via a collagenase digestion. Briefly, tissue was first perfused with Liver Perfusion Medium (Invitrogen, Merelbeke, Belgium) and then with Liver Digest Medium (Invitrogen). Subsequently, the digested liver was placed in Hepatocyte Wash Medium (Invitrogen). The cell suspension released from the liver was filtered through a 70-μm cell strainer (BD Falcon, Erembodegem, Belgium) and hepatocytes were separated from non-parenchymal cells via three low-speed centrifugation steps (5 minutes, 50g). Cell viability, measured using the Trypan Blue exclusion test, generally exceeded 90%. One million hepatocytes were injected into the spleen of 6- to 14-day-old uPA+/+-SCID mice. At selected moments, mice were bled, and plasma was stored at −70°C until further analysis.

Quantification of Human Albumin

Human albumin in mouse plasma was measured using an in-house sandwich ELISA. Maxisorp Immunoplates (Nunc, Roskilde, Denmark) were coated with goat anti-human albumin antibodies (Bethyl Laboratories, Montgomery, TX). After blocking and washing, diluted samples or calibrators were added, and human albumin bound to the plate was detected with a HRP-conjugated goat-anti-human albumin antibody (Bethyl Laboratories).

HBV and HCV Infection.

Five weeks after transplantation, chimeric mice were infected with either HBV or HCV. As a source of infectious HBV, serum from a patient suffering from a chronic HBV infection (hepatitis B surface antigen [HBsAg+], hepatitis B e antigen [HBeAg+], and anti-HBs) with a high HBV DNA content (107 cop/mL) was used. One hundred microliters of this serum was injected intraperitoneally. To induce an HCV infection, serum from an acutely infected chimpanzee was used (gift from Dr. J. Bukh). Twenty microliters of this serum (strain J4, 3.16 × 104 cop/mL) was injected intraperitoneally.

Detection and Quantification of Viral DNA/RNA and Proteins

HBV DNA and HCV RNA levels were quantified using the Cobas Amplicor HBV Monitor test and the Cobas Amplicor HCV Monitor test v2.0 (Roche Diagnostics, Mannheim, Germany), respectively. HBsAg and HBeAg were determined with the Axsym HBsAg V2 and the Axsym HBe 2.0 system (Abbott, Chicago, IL), respectively. The antigen levels were expressed as Signal/Noise values.

Detection of Human Proteins in Chimeric Plasma

Fifty microliters unfractionated mouse EDTA plasma was used for a non-gel proteome analysis. The proteins were digested with trypsin, and combined fractional diagonal chromatography (COFRADIC™) was used to isolate the amino terminal peptides out of this complex peptide mixture.10 The isolated peptides were further analyzed by LC-MS/MS using a Q-TOF1 mass spectrometer (Micromass UK Limited, Cheshire, UK) and the obtained MS/MS spectra were linked to peptide sequences stored in a tailor-made database ( containing human and murine full and sequentially ragged protein sequences,10 using the MASCOT database search engine ( Peptides identified by a MASCOT score that exceeded the identity threshold score of MASCOT at the 95% confidence level were considered as positively identified.

uPA-SCID Mouse Livers Used for Histopathological Evaluation

The livers of 5 uPA-SCID mice were used for extensive histological evaluation. The liver of a 2-week-old nontransplanted, noninfected uPA-SCID mouse was used as a reference. Two mice were killed 36 and 78 days after transplantation, respectively. One HBV- and one HCV-infected chimeric animal was killed 25 days and 41 days after infection, respectively.

Histochemistry and Immunohistochemistry

A detailed description of all histochemical procedures can be consulted online at the HEPATOLOGY website (


Plasma Analysis of Chimeric Mice

The success of human liver cell engraftment and expansion in the transplanted uPA+/+ transgenic SCID mice was evaluated by quantifying human albumin in mouse plasma at regular intervals. Figure 1 shows that human albumin concentrations reached a median level of 3 mg/mL by week 4 after transplantation and increased to 7 mg/mL by week 7. Thereafter the albumin levels remained quite constant until at least 14 weeks after transplantation.

Figure 1.

Human albumin concentrations in human hepatocyte uPA-SCID mice at different times after hepatocyte transplantation. This box-and-whisker plot represents the median and interquartile ranges of human albumin levels (mg/mL) in the plasma of 21 uPA-SCID mice transplanted with human hepatocytes.

To evaluate whether the transplanted human hepatocytes were functional and behaved like mature, normal hepatocytes, we analyzed the human proteome of the mouse plasma. Seven weeks after transplantation, plasma from a successfully transplanted mouse was analyzed by COFRADICTM. We decided to use this particular proteomics method because an a priori in silico study indicated that alike plasma proteins of different origin (mouse and human) mainly had sequence differences at their N-terminal extremity. We unambiguously identified 22 different human proteins in this mouse plasma (Table 1). Besides these 22 proteins of human origin, several other plasma proteins were detected that were either specifically from murine origin or of which the identified peptides did not allow discrimination between a human or murine origin (data not shown).

Table 1. Proteome Analysis of Chimeric Mouse Plasma
 ProteinAccession Number
  1. NOTE. Overview of the 22 human proteins detected in the plasma of transplanted uPA-SCID mice using a mass spectrometric technique. Most of these proteins are produced exclusively or nonexclusively by hepatocytes. However, in this context, hepatocytes can be the only source. Some proteins are generally regarded to have a membrane or nuclear localization and are probably remnants of dying hepatocytes. Additional information on the above mentioned proteins can be retrieved via the corresponding SwissProt ( accession numbers.

1Afamin precursorP43652
2Alpha-1 antitrypsinP01009
3Alpha-1-antichymotrypsin precursorP01011
4Alpha-2-HS-glycoprotein precursorP02765
5Apolipoprotein A-II precursorP02652
6Apolipoprotein E precursorP02649
7Complement C2 precursorP06681
8Complement C4b-binding protein alpha chain precursorP04003
9Complement C3 precursorP01024
10Complement factor B precursorP00751
11Chloride channel protein 7P51798
12Eukaryotic translation initiation factor 4 gammaQ04637
13Fibrinogen alpha/alpha-E chain precursorP02671
14Fibrinogen beta chain precursorP02675
15Inter-alpha-trypsin inhibitor heavy chain H3 precursorQ06033
16Kininogen precursorP01042
17Prostaglandin F2 receptor negative regulator precursorQ9P2B2
18Protein-tyrosine phosphatase delta precursorP23468
19RNA-binding protein 5P52756
20Serum albumin precursorP02768
21Synaptobrevin-like protein 1P51809
22Zinc finger protein 70Q9UC06

Histological Studies of Uninfected Chimeric Livers

Chimeric uPA-SCID Mice 36 and 78 Days After Transplantation.

Massive numbers of mature human hepatocytes were identified in the livers of transplanted mice. These could easily be discriminated from their murine counterparts by their larger size and the peculiar appearance of their cytoplasm (Fig. 2A). The human hepatocytes were somewhat swollen and had a rarefied cytoplasm with wisps and small clumps of eosinophilic material in an otherwise empty-appearing cytoplasm. The cell membranes appeared mildly thickened. The empty-appearing areas of the cytoplasm on hematoxylin and eosin staining corresponded to areas of glycogen storage, as shown by periodic-acid-Schiff (PAS) staining (Fig. 2B). The stainings for albumin, mitochondria, pan-cytokeratin, and CK18 confirmed the human origin of these hepatocytes (Fig. 2C-F), and the cytoplasmic staining pattern was also in wisps and small clumps, indicating that the organelles were displaced and clumped together by the excessive glycogen accumulation. The morphological features of the hepatocytes are identical to the morphology of hepatocytes in all glycogen storage diseases, except type IV.11 This was confirmed by comparing our present findings with 4 archival clinical liver biopsy specimens from patients with glycogenosis and with descriptions of glycogenosis in the literature.11

Figure 2.

Histology of a chimeric liver. (A) Hematoxylin and eosin staining easily allows identification of the human hepatocytes (H) within mouse (M) parenchyma. The cytoplasm of the human hepatocytes has an unusual pale and granular appearance, resembling the morphology of hepatocytes in glycogen storage diseases. (B) PAS staining shows that, in contrast to the negative mouse hepatocytes, glycogen accumulation is restricted to the cytoplasm of the human hepatocytes. (C) Staining for human albumin unveils the human origin of these pale hepatocytes. Several ceroid macrophages (arrows) also can be seen within mouse parenchyma. Despite a low background staining of the lumen of sinusoids and larger blood vessels (arrowheads), mouse hepatocytes never stain positive for human albumin. Clusters of human hepatocytes can also be visualized with antibodies specific for human mitochondria (D), human pan-cytokeratin (E), and human CK18 (F). Human hepatic progenitor cells (black arrows) and intermediate hepatocyte-like cells (white arrow) were also observed (F). (Original magnifications: A, B, D, and F, ×400; C and E, ×200)

Apart from this massive glycogen storage, the human hepatocytes appeared quite “healthy” and showed no signs of damage or degeneration. This is in sharp contrast with the extensive damage of mouse liver parenchyma, where ceroid macrophages formed large clusters (Fig. 3A) and focal confluent necrosis of mouse hepatocytes was visible, both indicating moderate to severe hepatocyte damage and loss. Occasionally, human hepatocytes contained a mitotic figure, and steatosis was minimal.

Figure 3.

Five weeks after transplantation, approximately one fourth of liver parenchyma is repopulated by pale human hepatocytes that are organized in nodules. (A) Ceroid macrophages are restricted to the areas occupied by diseased mouse hepatocytes, as shown by PASα staining. (B) Staining with a polyclonal anti-CEA antibody clearly shows connections between mouse canaliculi (arrows) and canaliculi that are somewhat bigger and that are formed by the surrounding human hepatocytes (middle). (C) Staining of the liver with an antibody specific for human Ki-67 showed that proliferating human hepatocytes were preferentially localized at the edge of the nodules. The scarce presence of sinusoids lined by spindle-shaped human endothelial cells was visualized by staining for human CD34 (D). (Original magnifications: A, ×100; B, ×400; C and D, ×200)

In the liver sections of the 44-day-old mouse, also two red foci were seen. Hepatocytes in red foci were larger than the surrounding mouse hepatocytes. The red foci contained no ceroid macrophages and were devoid of oval cells. The foci were rather sharply demarcated and slightly compressed surrounding mouse parenchyma. These findings indicate that the loss of transgene expression in the cells of these foci leads to disappearance of cell damage and to a growth advantage compared with transgene-expressing parenchyma. Interestingly, most hepatocytes within the red foci had a cytoplasm that was partly positive on PAS staining, whereas surrounding, transgene-expressing hepatocytes were generally PAS negative.

The transversal section through the liver of the 44-day-old mouse showed that the human hepatocytes were organized into approximately 80 nodules, which were well demarcated with sometimes a singular hepatocyte infiltrating into the surrounding mouse liver parenchyma. Between 10 and 90 human hepatocytes were present in the section through the nodules. Between the nodules, there were occasionally singular human hepatocytes or small groups of, maximally, 5 hepatocytes. Overall, 25% of the liver parenchyma consisted of human cells. In contrast to mouse liver parenchyma, these nodules rarely contained ceroid macrophages (Fig. 3A). Only 5 of approximately 80 nodules contained 1 or 2 ceroid macrophages that were not located in the center of the nodule, suggesting that these ceroid macrophages are rather transiting than resident. There was no preferential localization of the human hepatocyte nodules in the liver; they were observed as well periportally as around the centrolobular vein. Areas of confluent necrosis of mouse liver parenchyma always contained human hepatocyte nodules at their periphery.

As shown in Fig. 3B, human hepatocytes formed normal canalicular structures that connected to mouse canaliculi that were smaller. Ki-67 staining showed that 18% of human hepatocytes were proliferating, and these cells were preferentially localized at the edge of the nodules (Fig. 3C).

In the liver of the 90-day-old mouse, human hepatocytes no longer formed nodules but appeared as large areas separated from each other by small areas of mouse liver parenchyma. Overall, the human hepatocytes represented 66% of the mouse liver parenchyma. Again, human areas contained very few ceroid macrophages; 6% of the human hepatocytes were Ki-67 positive and showed no preferential distribution. In both mice, areas consisting of human hepatocytes contained scarce mouse oval cells, at numbers that appeared lower than in mouse liver parenchyma. Sinusoids lined by human endothelial cells were present but rare (Fig. 3D).

Fifteen nodules in the 44-day-old mouse contained, in addition to mature hepatocytes, 1 to 5 human hepatic progenitor cells (Figs. 2F and 4A-B) that were recognized on the basis of their morphology and immunohistochemical phenotype.12 Hepatic progenitor cells were present both in the center and at the edge of the nodules. Reactive bile ductules containing human cells were not observed. Occasionally, human intermediate hepatocyte-like cells were seen neighboring human hepatic progenitor cells located at the human–mouse parenchymal interface. Similarly, singular human hepatic progenitor cells and intermediate hepatocyte-like cells were focally present at the human–mouse parenchymal interface in the 99-day-old mouse liver. In this animal, one reactive bile ductule with a dilated lumen was completely lined by human cells.

Figure 4.

Human hepatic progenitor cells (HPCs) were identified in the livers of chimeric mice at the interface between human (H) and mouse (M) hepatocytes. (A) HPCs are small cells with scanty cytoplasm and a relatively large oval nucleus and strongly immunoreactive for pan-cytokeratin (black arrows). Intermediate hepatocyte-like cells (white arrows) could be recognized as polygonal cells with a size and immunophenotype intermediate between that of HPCs and hepatocytes. Some mature hepatocytes display a decreased and granular staining pattern, possibly as a consequence of the glycogen accumulation. (B) Staining with an anti-CK7 antibody confirms the presence of both these cell types, which are surrounded by mature hepatocytes that do not stain for CK7. Staining of the liver of a non-transplanted uPA-SCID mouse with anti–pan-cytokeratin (C) and anti-CK7 (D) antibodies served as a negative control. (Original magnifications: A and B, ×400; C and D, ×100)

We never observed dysplasia or malignant transformation of human hepatocytes, nor tumor cells of the type for which the patient underwent liver surgery.

Ultrastructural Analysis.

Human hepatocytes had a normal canalicular and sinusoidal pole and showed no abnormalities, besides the glycogen accumulation. Hepatic progenitor cells of types I (most undifferentiated type) and III (having some hepatic features)13, 14 were present in the chimeric liver. Some progenitor cells showed signs of damage, indicating a mouse origin, and were forming a chimeric canalicular structure with a human hepatocyte (Fig. 5A). Healthy type III hepatocyte-like cells formed part of a canaliculus with adjacent mature human hepatocytes. The absence of any damage strongly suggests that these cells were of human origin (Fig. 5B).

Figure 5.

Ultrastructural analysis of the chimeric mouse liver. (A) Two hepatic progenitor cells (P) and a small rim of a third progenitor cell are obvious. Remark the thin basement membrane surrounding the progenitor cells (thin arrow). Two progenitor cells present signs of cytoplasmic damage (arrow), indicating a murine origin. The progenitors form part of a bile canaliculus (C) with a large mature human hepatocyte (H) showing unusual glycogen storage (asterisk). The observation of chimeric bile canaliculi supports our immunohistochemical analysis showing connections between mouse and human bile canaliculi. A stellate cell (open arrow) is apparent in the Disse space as well as collagen fibers (cf). (B) A type III progenitor cell (P) is forming a bile canalicular structure (C) with two adjacent human hepatocytes (H). The absence of any damage within the cytoplasm of this progenitor cell suggests a human origin. A small processus of a stellate cell (open arrow) can be seen in the Disse space. (Original magnifications: A, ×12,000; B, ×24,000)

In Vivo HBV Infection

Five weeks after successful transplantation with cells from 3 different donors, 9 animals were infected with HBV. Twenty days later, mouse plasma was analyzed for viral proteins and HBV DNA. All animals showed signs of infection, but in 3 mice the serum markers were rather low, namely, an HBsAg level of 52.5 S/N, HBeAg below the detection limit, and HBV-DNA levels of approximately 2 × 105 cop/mL. These values changed very little until the animals died. In the remaining 6 mice, the HBV infection was highly replicative, with HBsAg and HBeAg levels of, on average, 1,130 S/N and 663 S/N, respectively, and HBV-DNA levels exceeding 2 × 1010 cop/mL.

Twenty-five days after HBV inoculation, 1 animal was sacrificed for histological analysis of the liver. Human hepatocytes were loaded with glycogen and formed large areas occupying 87% of mouse liver parenchyma. No other signs of damage or cell loss were observed. No human hepatocytes had a “ground glass” appearance. Almost all human hepatocytes had a granular cytoplasmic staining for HBsAg, and many were also strongly positive near the cell membrane (Fig. 6A). Figure 6B shows that a large majority of the human hepatocytes displayed a strong nuclear and a moderate cytoplasmic staining for hepatitis B core antigen.

Figure 6.

Immunohistochemical and ultrastructural analysis of chimeric livers infected with HBV (A-C) or HCV (D). HBsAg-specific staining (A) shows a granular positivity of the cytoplasm with accentuation near the membrane, whereas hepatitis B core antigen–specific staining (B) shows a strong nuclear and mild cytoplasmic staining. Liver sections from uninfected chimeric mice were used as controls and were always negative (insets). Using electron microscopy (C), longitudinally cut HBsAg tubules (thin arrows) as well as complete Dane particles (thick arrows) could be observed in the cisternae of the endoplasmic reticulum. Human hepatocytes infected with HCV could be revealed with antibodies recognizing the HCV E2 protein (D). (Original magnifications: A, B, and insets, ×200; C, ×77,000; D, ×400)

At the interface between red foci and human hepatocytes, there was no sharp border, no compression, and the two cell types appeared to infiltrate each other's areas. These findings suggest that the human hepatocytes and the “transgene-free” mouse hepatocytes profit equally from the growth stimuli present.

Ultrastructural analysis of the liver of an HBV-infected mouse showed that the nuclei of numerous hepatocytes contained noncoated virus particles with a diameter of 21 to 24 nm corresponding to HBV core particles. In the cisternae of the endoplasmic reticulum, 1 or 2 longitudinally transsected tubules and cross-sectioned spheres of HBsAg were present. In some cisternae, core particles surrounded by a clear halo and a dark ring of approximately 40 nm in size, corresponding to Dane particles, were observed (Fig. 6C).

In Vivo HCV Infection

uPA+/+-SCID mice, untransplanted or successfully transplanted with human hepatocytes, were infected with HCV of genotype 1b. In untransplanted but HCV-injected animals, HCV RNA was never detected. Two weeks after infection, the plasma of transplanted animals contained ≥ 2.35 × 106 IU/mL HCV RNA. The viral load increased over time and reached levels of up to 8.1 × 107 IU/mL by week 10. Plasma from these animals was used to infect other transplanted mice. Three microliters infectious mouse plasma (containing approximately 105 IU HCV RNA) sufficed to establish a new infection. Animals infected in this way also displayed the same massive increase in viral load. Recalculating all viral titers with respect to the blood volume of the infected animals, and taking into account the serial infections, we achieved a 109-fold amplification of viral RNA. The in vivo HCV infection could be maintained for at least 4 months.

Six weeks after infection, one HCV-infected animal was sacrificed for histological examination. Seventy-seven percent of liver parenchyma consisted of healthy human glycogen accumulating hepatocytes, which formed large areas separated from each other by small areas of diseased mouse hepatocytes. Almost all human hepatocytes showed a granular cytoplasmic staining with the anti-HCV E2 antibody (Fig. 6D). This granular staining pattern has always been observed in liver biopsy specimens from chronic HCV patients.15, 16 As in the livers of non-infected and HBV infected mice, singular human hepatic progenitor cells and intermediate hepatocyte-like cells were focally present at the human–mouse parenchymal interface. Moreover, some of the ductular clusters consisted of ductules that were lined by a mixture of bile ductular cells of mouse and human origin. Despite careful ultrastructural analysis, no viral particles or viral tubular structures could be retrieved.


Recently, two groups independently created chimeric mice harboring human hepatocyte grafts and successfully induced HBV7 and HCV8 infections in these animals. Both groups provided convincing evidence for solid hepatocyte engraftment and active viral replication but paid less attention to the functional integrity and structural organization of the transplanted hepatocytes in the xenogeneic environment. We wanted to fill this gap and present here a detailed description of the functional and morphologic characteristics of the chimeric liver before and during active HBV and HCV infections.

To evaluate the human hepatocyte graft-take and expansion in uPA+/+-SCID mice, the human albumin concentration in mouse plasma was monitored at regular intervals. In successfully transplanted mice, the albumin level reached a plateau value of approximately 7 mg/mL (median value) around week 7, and this level remained unchanged in the weeks that followed. Although the human albumin level is a reliable marker of the integrity and functional status of the human hepatocytes, we performed a proteomic analysis10 of the mouse plasma and demonstrated the presence of 21 additional and less abundant human proteins. Most of these proteins are known plasma proteins synthesized by the liver. Apart from these “classical” human plasma proteins, some human proteins with a membrane or nuclear localization could also be detected. Their presence in the plasma may be due to membrane turnover or release from dying hepatocytes. The presence of these atypical plasma proteins in our chimeric mouse plasma is not unexpected because such proteins are also found in the plasma of healthy blood donors.17, 18

The integrity and spatial organization of the human hepatocytes in the chimeric liver was also examined by histology. Early after engraftment, the human hepatocytes appeared as sharp nodules that expanded and became blurred as time went on and occupied up to 87% of liver parenchyma. Although the human hepatocytes resided in a xenogeneic environment, they showed no signs of damage or degeneration and appeared to have remodeled the diseased mouse liver in a progressive and organized fashion. We also demonstrated connections between human and mouse canaliculi. These chimeric ducts must be open and functional, because no signs of cholestasis were found.

The only, but striking, abnormality of the transplanted liver cells was the abundant accumulation of glycogen. The cause of this aberrant polysaccharide accumulation remains unclear, but this phenomenon is most likely attributable to inappropriate recognition of murine signals (e.g., hormones) by human hepatocytes. This feature is independent of HBV or HCV infection, and it is probably not just a compensation for the absence of glycogen in the diseased mouse hepatocytes because mouse hepatocytes in red foci have a normal appearance and normal intracellular glycogen content. The overall appearance of the chimeric mouse and the aspect of the liver cells in histology studies suggest that this metabolic disturbance has a negative effect on the animal as a whole and the grafted hepatocytes in particular. This is confirmed by the fact that uPA+/+ mice transplanted with human hepatocytes never looked as healthy as their uPA+/+ -SCID littermates that were “rescued” with a murine hepatocyte graft. Although the latter developed normally and after a few weeks could not be discriminated from uPA−/− SCID mice (P. Meuleman, unpublished data, 2004), the former survived but suffered from severe growth retardation and failure to thrive. This can be a result of the observed glycogenosis, but it is plausible that also other biochemical pathways may not function optimally, possibly because of communication failure between mouse ligands or receptors and their human counterparts. The inconsistency between the albumin levels (approximately 20% of normal values) and tissue analysis (repopulation up to 87%), again indicates suboptimal communication between human cells and the mouse environment. These observations deserve further study considering the growing interest in transplantation of xenogeneic liver tissue or hepatocytes to treat hepatic failure in humans.

Once human hepatocytes had stably engrafted in the uPA+/+ -SCID host, these animals could indeed easily be infected with hepatotropic viruses. After an infection with HBV, we could detect high levels of HBsAg, HBeAg, and HBV DNA in the mouse plasma. The concentrations of HBsAg and HBV DNA observed in our experiments largely exceeded those reported by Dandri et al.,7 and these investigators were unable to find HBeAg. This may be linked to the inferior hepatocyte engraftment (5%-15%) obtained in heterozygous uPA animals. The immunohistochemical staining pattern of the infected mouse liver corresponds very well to that observed in chronic hepatitis B patients during the viral tolerance/replication phase.19 However, the proportion of HBV-positive hepatocytes seen in our mouse model is considerably higher than routinely observed in biopsy specimens of such patients.19 This result correlates well with the extremely high amounts of HBV DNA values found in the infected mouse plasma.20, 21 We were also able to visualize subviral HBV particles and infectious Dane particles inside the human hepatocytes of our chimeric mice.

Infections of chimeric mice with HCV were as successful as the HBV infections. Viral RNA could be detected in the plasma of all infected mice (n = 7), and HCV RNA levels were much higher than those routinely observed during chronic HCV infections in humans. High viremia of this magnitude can be observed, however, when HCV-infected patients are undergoing immune suppressive therapy after transplantation.22–24 Histological analysis of the liver of an HCV-infected mouse showed that most human hepatocytes were infected. Despite careful electron microscopic analysis, no viral or subviral particles could be identified in human hepatocytes. This is not surprising, because these structures are also rarely seen in biopsy specimens of chronically infected patients.13 However, sequential infections proved the presence and de novo production of infectious HCV particles.

Although most human hepatocytes were infected with either HBV or HCV and although the viral replication was high, these viruses had no negative effects on the function, survival, or proliferation of the infected hepatocytes. This finding is in agreement with the current view that hepatocyte damage in patients with active chronic viral hepatitis is induced by the immune system rather than the virus itself.25 Whether long-term infections with HBV or HCV have any negative or cytopathic effect on the transplanted human hepatocytes is currently being explored.

A final, but very important, observation is the fact that the chimeric mouse liver contained human hepatic progenitor cells in addition to mature human hepatocytes. In close association with the human progenitor cells we noticed human intermediate hepatocyte-like cells, which suggests that the progenitor cells are differentiating toward mature hepatocytes through this intermediate stage. Occasionally, human hepatic progenitors were observed within murine ductular structures, illustrating the bipotential characteristics of these cells. The presence of progenitors in all the analyzed hosts indicates that these cells must have been present in the injected cell suspensions. This report suggests the presence of progenitor cells in a cell suspension obtained via a widely used technique for the isolation of hepatocytes. This observation is important because certain qualities attributed to mature hepatocytes, such as the capacity to trans-differentiate into biliary cells26, 27 and the extensive potential to repopulate damaged liver,28, 29 might be related to the presence of these progenitor cells. Data obtained in previous studies using so-called hepatocyte suspensions should at least be critically reviewed. The number of progenitor cells in the hepatocyte graft and their role in the repopulation of the diseased liver is being examined.

In conclusion, the uPA+/+ -SCID mouse easily accepts human hepatocyte grafts, and the chimeric animals thus created represent a successful small animal model to study HBV and HCV infections and new therapeutic compounds to treat these infections. In addition, this model also may prove useful for the study of metabolic aspects of xenogeneic transplantation and the hepatic engraftment, maturation, and differentiation of stem cells of different origins.


The authors thank S. Couvent and F. Clinckspoor for the serologic analysis of the mouse plasma, Dr. J. Bukh for his kind donation of HCV inoculum and A. Staes and L. Martens for the isolation and analysis of the amino terminal peptides.