Length of the linking domain of human pro-tumor necrosis factor determines the cleavage processing

Several studies have indicated that only one cleavage site (Ala-1/Val+1) is involved in the release of mature TNF from human pro-TNF, whereas others have suggested that the linking sequence (residues -20 to -1) may be important. We previously demonstrated that a pro-TNF deletion mutant, delta -20- -1, was able to form a trimeric structure and mediate TNF cytotoxicity in a juxtacrine fashion without releasing mature TNF. We constructed seven mutants with smaller deletions within this region. Three 15-residue deletion mutants, delta -20- -6, delta -15- -1 and delta -20- -16, -10- -1, were noncleavable, although able to form a trimer and to mediate cytotoxicity through cell-to-cell contact. Three five- or ten-residue deletion mutants, delta -20- -16, delta -10- -1, and delta -5-, -1, behaved like the wild-type TNF; all formed a trimer and released mature TNF. These results suggested that in pro-TNF (1) the number of residues between the base of the trimer and the plasma membrane determines accessibility of the cleavage site to the pro-TNF processing enzyme(s) since small deletions did not block cleavage whereas large ones did regardless of the presence of the native cleavage site (-1/+1), (2) the native cleavage site is not sufficient for releasing mature TNF because mutant delta -20- -6, in which the native cleavage site was intact, was noncleavable, and (3) alternative cleavage site(s) may exist since mutants delta -10- -1 and delta -5- -1, which lack the native cleavage site, were cleavable.