One attractive strategy to prevent or control allograft rejection is to genetically modify the donor tissue before transplantation. In this study, we have examined the feasibility of gene transfer to human corneal endothelium, using a number of recombinant adenovirus constructs. Ex vivo infection of human corneas with adenoviral vectors containing lacZ, under transcriptional control of either cytomegalovirus (CMV) or Rous sarcoma virus (RSV) promoters, provided high-level gene expression, which was largely restricted to endothelium. Expression of the reporter gene persisted at relatively high levels for up to 7 days, followed by a decline to indetectable levels by 28 days. RT-PCR analysis of lacZ transcription showed a similar picture with a short period (3-7 days) of RNA transcription after infection. In contrast, adenoviral DNA persisted for at least 56 days. Subsequently, we examined the expression of a potential therapeutic gene, CTLA-4 Ig fusion protein. Following infection of human corneas with adenoviral vectors encoding CTLA-4 Ig protein, high levels of the fusion protein were detected in corneal culture supernatants for up to 28 days. This protein was functionally active, as determined by binding to B7.1 (CD80)-expressing transfectants. This study suggests that genetic alteration of donor cornea before transplantation is a feasible approach for preventing or controlling allograft rejection. Similar gene-based strategies might also be feasible to prevent rejection of other transplanted tissues or organs.