BACKGROUND: Chitosan has been shown to be a non-toxic and efficient vector for in vitro gene transfection and in vivo gene delivery through pulmonary and oral administrations. Recently, we have shown that chitosan/DNA nanoparticles could mediate high levels of gene expression following intrabiliary infusion 1. In this study, we have examined the possibility of using polyethylene glycol (PEG)-grafted chitosan/DNA complexes to deliver genes to the liver through bile duct and portal vein infusions. METHODS: PEG (Mw: 5 kDa) was grafted onto chitosan (Mw: 47 kDa, deacetylation degree: 94%) with grafting degrees of 3.6% and 9.6% (molar percentage of chitosan monosaccharide units grafted with PEG). The stability of chitosan-g-PEG/DNA complexes was studied by measuring the change in particle size and by agarose gel electrophoresis against bile or serum challenge. The influence of PEG grafting on gene transfection efficiency was evaluated in HepG2 cells using luciferase reporter gene. Chitosan and chitosan-g-PEG/DNA complexes were delivered to the liver through bile duct and portal vein infusions with a syringe pump. Gene expression in the liver and the distribution of gene expression in other organs were evaluated. The acute liver toxicity of chitosan and chitosan-g-PEG/DNA complexes was examined by measuring serum alanine aminotranferase (ALT) and aspartate aminotransferase (AST) activities as a function of time. RESULTS: Both chitosan and chitosan-g-PEG displayed comparable gene transfection efficiency in HepG2 cells. After challenge with serum and bile, chitosan-g-PEG/DNA complexes, especially those prepared with chitosan-g-PEG (GD = 9.6%), did not form large aggregates like chitosan/DNA complexes but remained stable for up to 30 min. In addition, chitosan-g-PEG prevented the degradation of DNA in the presence of serum and bile. On day 3 after bile duct infusion, chitosan-g-PEG (GD = 9.6%)/DNA complexes mediated three times higher gene expression in the liver than chitosan/DNA complexes and yielded background levels of gene expression in other organs. On day 1 following portal vein infusion, gene expression level induced by chitosan/DNA complexes was hardly detectable but chitosan-g-PEG (GD = 9.6%) mediated significant transgene expression. Interestingly, transgene expression by chitosan-g-PEG/DNA complexes in other organs after portal vein infusion increased with increasing grafting degree of PEG. The ALT and AST assays indicated that grafting of PEG to chitosan reduced the acute liver toxicity towards the complexes. CONCLUSION: This study demonstrated the potential of chitosan-g-PEG as a safe and more stable gene carrier to the liver.