The cytokine LIGHT is a promising candidate for cancer therapy. However, the therapeutic effect of LIGHT as a systemic anticancer agent is currently insufficient because of its instability and its binding to nonfunctional soluble decoy receptor 3 (DcR3), which is overexpressed in various tumors. Modification of proteins with polyethylene glycol (PEGylation) can improve their in vivo stability, but PEGylation may occur randomly at all lysine residues and the NH(2)-terminus; therefore, PEGylated proteins are generally heterogeneous and have decreased bioactivity. In this study, we attempted to create a lysine-deficient LIGHT mutant that could be PEGylated site-specifically and would have lower affinity for DcR3. We prepared phage libraries expressing LIGHT mutants in which all the lysine residues were replaced with other amino acids. A lysine-deficient LIGHT mutant [mLIGHT-Lys(-)] was isolated by panning against lymphotoxin beta receptor (LTbetaR). mLIGHT-Lys(-) could be site-specifically PEGylated at its NH(2)-terminus, yielding molecular uniformity and in vitro bioactivity equal to that of non-PEGylated, wild-type LIGHT. Furthermore, mLIGHT-Lys(-) was not trapped by the nonfunctional DcR3, despite binding to its functional receptors. These results suggest that mLIGHT-Lys(-) might be a useful candidate for cancer therapy.