In 1990 a group of scientists led by Ralph Brinster, while developing a model of neonatal bleeding disorders, generated a novel murine line carrying a urokinase-type plasminogen activator transgene controlled by an albumin promoter (Alb-uPA).1 Urokinase overproduction (targeted to the liver by the Alb promoter) yielded pups with a bleeding phenotype. While early perinatal mortality rates were high due to bleeding complications, some animals survived this early period only to succumb to subacute liver failure at 2-3 weeks of age; expression of the Alb-uPA transgene had proven to be toxic to the mouse hepatocytes. Not all animals died at this point, however, and when the investigators examined these late survivors, they found that the bleeding phenotype had been lost.2 An elegant series of experiments showed that hepatocytes within the transgenic liver had spontaneously deleted a portion of the transgene. Freed from the constraints of its expression, these nontransgenic hepatocytes began to proliferate and replace the diseased hepatic parenchyma with “normal” functioning hepatocytes, an advantage they recognized could be shared by transplanted hepatocytes also lacking the transgene.1
Demonstration of the ability to “rescue” the mice from liver failure with congenic hepatocyte transplants3 and subsequently xenografted rat hepatocytes followed.4 Petersen et al.5 and Dandri et al.6 moved forward with woodchuck and human hepatocyte transplantation respectively into mice hemizygous for the Alb/uPA transgene (on an immunodeficient RAG background). They were able to generate modest repopulation of livers, and support for infection with woodchuck hepatitis B virus and human hepatitis B virus (HBV). Our own group began working to develop an animal model of hepatitis C (HCV) infection in 1994, and successfully established HCV infections in scid/Alb-uPA mice carrying the transgene in homozygous fashion where liver cell repopulation frequently exceeded 50% of liver area on immunohistological analysis.7
The chimeric scid/Alb-uPA mouse in homozygous form, is a challenging model. Immunodeficiency and the bleeding tendency inherent in the transgene result in significant risk from infection and death due to perinatal trauma respectively. Optimal repopulation requires transplantation with high quality human hepatocytes within a short period after birth and so creates significant challenge for many laboratories due to difficulty accessing suitable tissue. Not all mice transplanted achieve high level repopulation, and not all those with high level repopulation develop high titre HCV infection after delivery of standardized inocula. Variations in transgene expression or activity of residual immune cells could be some of the answers as to why such variation occurs. Clearly more development and refinement of this model is needed to facilitate broader application in the research community. Nevertheless, we believe the model is the best small animal model of HCV infection currently available, and holds promise of useful application in many other fields.
In the current issue of this journal, Meuleman et al. present a detailed series of experiments that more fully characterizes the repopulation of the mouse liver with human hepatocytes in homozygous scid-Alb/uPA mice.8 Histological analysis of the chimeric liver reveals direct communication between biliary radicals of mouse and human origin. Clearly these are not isolated nodules of human hepatocytes in a mouse liver, but rather the human hepatocytes have become an integrated component of the mouse liver and maintain life-sustaining function. Aside from accumulation of glycogen, the human hepatocytes appear normal. Function appears to follow form, with the human proteome in the sera of mice inclusive of human albumin and 21 other human specific proteins. The paper also reports infection of chimeric mice with HBV and HCV, providing important independent confirmation of previous reports.6, 7 The authors have provided insight into critical details of the model, information of the type that should help focus study to generate a more reliable and reproducible system, a model simpler to use and yet equally as valid in application.
A particularly interesting finding is the presence of cells with markers of human hepatic progenitor cells within the parenchyma of the mouse livers; progressive differentiation into mature hepatocytes was suggested by immunohistochemical analysis. If these cells are truly progenitors, it will prompt a reevaluation of previous observations and conclusions from experiments using unpurified cell preparations from liver tissue, preparations that are not simply hepatocytes, but inclusive of a broad range of cell types resident in the liver. Progenitor cells have an important role in hepatic regeneration; indeed Braun et al. have documented an oval cell reaction in uPA mice subjected to additional proliferative stimuli.9 Nevertheless, the interesting question as to the relative contributions of hepatic progenitor cells and mature hepatocytes to repopulation in this model, as in natural hepatocyte hyperplasia after resection or injury, remains incompletely answered.
The scid-Alb-uPA model provides an excellent platform from which to study the basic biology of liver repopulation after injury. Using detailed immunohistochemical analysis with validated cell surface markers should permit investigators to examine histological changes that occur over a period of intense hepatocyte regeneration. Cell sorting approaches prior to transplantation might be used to study the relative contributions of specific cell types to repopulation as hinted at in the present study. This information will be of central importance to the future prospects of hepatocyte transplantation or tissue-engineered hepatic replacement therapies.
Mice with chimeric human livers should be useful in the evaluation of drug metabolism within human livers. Such a goal spurred an academic/industrial conglomerate in Japan to study the Alb/uPA mouse. Tateno et al. recently reported their findings in the scid-Alb/uPA mouse model,10 suggesting improved repopulation and overall animal health using adjuvant administration of an anticomplement therapy to prevent human complement-mediated attack on murine tissues. This was a logical step forward in a hepatocyte xenograft model where high circulating levels of human complement could be predicted to challenge the health of the recipient. Human complement in the mouse's circulation is one component of the alteration of the serum protein profile (proteome) to that of the donor from that of the recipient as has been previously reported in clinical whole organ xenografts of baboon livers into patients suffering from HBV and HIV.11 Tateno and colleagues proposed use of the mouse model for in vivo study of drug metabolism during preclinical evaluation of candidate therapeutics. An in vivo system with human hepatocytes in a stable environment for many weeks may provide a substantial improvement over current methods used to study drug metabolism that rely on short-term human hepatocyte cultures that can vary markedly in health and metabolic activity.
Studies of potential toxicity of drugs to the human liver are also integral to drug development. Toxicity data from other species can be notoriously poor at predicting outcomes in subsequent human clinical trials. A mouse with a chimeric human liver should prove a helpful bridge to clinical development. As an example, reports of in vitro death induced in human hepatocytes by TRAIL dramatically impacted clinical development of this promising anticancer therapeutic.12 We utilized our chimeric mouse model to demonstrate that nontagged TRAIL could reduce tumor size in vivo at dosages that had no adverse impact on human hepatocytes,13 helping to remove impediments to further evaluation of this interesting family of anticancer compounds.
The scid-Alb-uPA mouse is the best available small animal model of HCV infection. It provides for infection of human hepatocytes in vivo with wild-type virus (without the need for adaptive mutations), and has demonstrated the ability to generate infectious virus that can be transmitted to other chimeric mice — in short, the full viral life cycle. This is as close to the natural infectious process as we can get at present in a system that can generate sufficient numbers of study subjects for rigorous experimentation and statistical analysis. The model should be able to help elucidate the early extracellular and intracellular events of HCV infection and the mechanism by which the virus subverts the cells machinery for its own purposes. Studies such as these may provide both useful targets for generation of new therapeutic approaches and fertile fields for application of genomic and proteomic techniques to enhance our understanding of basic HCV biology.
At the present time, we have validated the model with both interferon α2b and small molecule antivirals for both HCV and HBV (unpublished data). Small molecule compounds with activity in replicon systems are being evaluated for in vivo antiviral efficacy in an ongoing fashion. There is also a role for pursuing nontraditional antiviral therapies against HCV. Such approaches could include gene therapies, antisense therapies, or RNA interference strategies. We have confirmed the ability of a gene therapy utilizing a modified BID molecule with an adenovirus vector to reduce HCV titers by over 2-log,14 supporting the range of study possible in the model, and generating yet another gene therapy that must await advances in the science of vector delivery for a chance at clinical evaluation. The model also has utility in the in vivo examination of potential neutralizing antibodies, which might be employed in passive immunotherapy or to help in development of prophylactic vaccines.15 We have evidence that neutralizing antibodies against HBV can prevent primary infection with HBV when administered prior to inoculation (unpublished data), in a situation analogous to the strategies used to prevent reinfection of liver allografts with the hepatitis B virus.
The study of infection facilitated by a mouse with a chimeric human liver is not limited to hepatitis viruses, and extension to other pathogens with obligate life cycles in the liver could greatly broaden the impact of the model. One such example is malaria, which remains one of the greatest killers in history.16 The most dangerous form, Plasmodium falciparum must undergo maturation in human (or chimpanzee) hepatocytes before it is capable of reinfecting human red blood cells; a vaccine targeting the liver stage of the life cycle might prevent blood infection completely.17 The scid/Alb-uPA mouse might provide an excellent model for the entire life cycle of the parasite, and help develop inexpensive therapies to supplement preventative measures in endemic areas.
While the scope of utility of the chimeric mouse model is vast, the model is not without its limitations. Immunodeficiency allows human cell engraftment but prevents use of the model for directly studying active immunization. Our ongoing studies have achieved partial reconstitution with donor-specific immunity; while imperfect, it should help alleviate this problem. Breeding animals carrying the Alb-uPA transgene can also have challenges: in the homozygous form, where the transgene is most useful in stimulating high-level human chimerism, it is also most lethal to the pups due to perinatal and perioperative bleeding complications. With careful attention to breeding protocols, large numbers of animals can be (and are) generated on a regular basis, but the experience to make these decisions can be hard won. Finally, hepatocyte supply can be limiting. Cryopreserved hepatocytes function, but at a lower efficiency than fresh hepatocytes. Investigators should continue to work towards developing stable cell lines that maintain a differentiated hepatocyte phenotype. Cell lines we have used so far (HepG2, Huh-7) have proven rapidly lethal to recipient animals, which experience accelerated tumor cell growth and death from overwhelming tumour burden.
The scid/Alb-uPA mouse carrying a chimeric human liver has been a significant step forward in studying the basic biology of human hepatitis viruses.18 This work only begins to scratch the surface of what the ultimate utility of this model can be to investigators across a wide range of fields. The detailed characterization of the uPA-SCID mouse chimera reported by Meuleman et al.8 characterizes the type of study that will help propel the model to successful development in a broad range of applications in health and science that hold significant promise for improvement of the human condition.