Human tumor necrosis factor mutants with preferential binding to and activity on either the R55 or R75 receptor

Previously, we reported that the cytotoxic activity of human (h) tumor necrosis factor (TNF) on murine (m) L929 cells requires the integrity of three loops (positions 30-36, 84-88 and 138-150) which cluster around the interface between each two subunits of the trimeric hTNF structure. The collection of hTNF mutants was further characterized by their activity on various human cell systems as well as by their binding to the two types of hTNF receptor (R), R55 and R75. It turned out that two amino acids (Leu29 and Arg32) were specifically involved in hR75 binding, as Leu29-->Ser (L29S) and Arg32-->Trp (R32W) mutant molecules had largely lost binding to hR75, but not to hR55. In order to screen for more highly R55-specific mutants, nine other amino acids were inserted at these two positions; only the substitutions L29G and L29Y showed an increased differential binding as compared to L29S, while no further improvement was found with mutations at position 32 compared to R32W. Biological assays mediated either by hR55 or hR75 confirmed the results obtained by physical binding to purified receptors. A similar substitution in mTNF, Arg32-->Tyr, also resulted in a preferential loss of binding to hR75 and a large decrease in mR75-mediated bioactivity. Except for the double mutant L29S-R32W, all other tested amino acid substitutions in the loops at positions 30-36 or 84-88 of hTNF led to a substantial loss of affinity for both receptors and a concomitant reduction of biological activity. In the loop at positions 138-150, the non-conservative replacement of Glu by Lys at position 146 (E146K) resulted in an even lower binding to R75 as compared to R32W, while binding on and bioactivity through R55 was only slightly reduced. Remarkably, a reversed differential binding was observed after substitution at position 143 in hTNF; replacing Asp by non-conservative residues such as Tyr, Phe or Asn resulted in a much larger decrease in binding to R55 than to R75. In conclusion, receptor-specific mutants such as R32W, E146K and D143N can be used to study the function either of R55 or R75 on different human cell types. In vivo, we presume that the R55-specific mutants will retain antitumor activity in the absence of R75-dependent, severe side effects.