TNF


In response to inflammatory stimuli such as lipopolysaccharide, macrophages produce TNF. TNF binds to receptors present on virtually all cells throughout the body, causing a variety of reactions.

Panel A shows the trimerization hypothesis. In this model, the juxtaposition of three receptors results from their binding of a single TNF trimer. The resultant complex generates an activatingsignal. Panel B shows the expanding-network hypothesis. A growing hexagonal array of TNF trimers bound to TNF-receptor dimers takes account of the dimeric structure of TNF receptors free of ligands and the capacity of each ligand molecule to engage three receptor subunits. “Capping” of receptors would trigger a biologic response. Panel C shows the molecular-switch hypothesis, the most favored model. Inthis model each receptor dimer is an activatable unit. Receptor activation occurs in response to two events. First, the ligand binds to subunit B of the receptor. Second, subunit A disengages from subunit B, which permits binding of the receptor to a second available site on the ligand surface. These events cause conformational changes within the cytoplasmic domain of the receptor, leading to signaltransduction. Specifically, the death domains of the 55-kd TNF receptor or Fas receptor might undergo homodimerization. The graph shows an imaginary profile of the free energy associated with conformational changes between the “off” and “on” states of the receptor. It is supposed that a large activation-energy barrier prevents transition from the off to the on state in the absence of ligand andthat TNF effectively catalyzes this transition, thereby “throwing the switch.” Moreover, as depicted here, the free energy of the on state, which presumes a stable association between TNF and the receptor, may be substantially lower than the free energy of the off state, and this may be irreversible.

Lymphotoxin-α and TNF bind to the same 55-kd and 75-kd TNF receptors. Proteins and domains withsimilar structures or functions, such as TNF-receptor–associated death domain (TRADD), receptor-interacting protein (RIP), Fas-associated death domain (FADD, or MORT-1) molecules, and the death domains of the Fas and 55-kd TNF receptors, are generally portrayed by identical shapes but in different colors. Zn is used to denote proteins known to contain ring-finger and zinc-finger motifs that bind tothe 75-kd TNF receptor, lymphotoxin-β receptor, and CD40-ligand receptor. TNF-receptor–associated factor type 1 (TRAF-1) lacks a zinc-finger motif but binds to TRAF-2, which has both ring-finger and zinc-finger motifs. CD40-associated protein 1 (CAP-1), which is structurally very similar to TRAF-3, is thought to be capable of replacing TRAF-3 as a binding partner for the lymphotoxin-β receptor andCD40-ligand receptor. LMP-1 is a plasma-membrane protein that spans multiple domains and is encoded by the genome of the Epstein–Barr virus. It has no homology to members of the TNF family. Not all possible combinations and interactions are shown. In addition, certain binding proteins that were not discussed in the text belong neither to the zinc- and ring-finger family nor to the death-domainfamily of transducers. These include TNF-receptor–associated protein 1 (TRAP-1), a heat–shock protein analogue15; the 55.11 protein, a proteasome regulatory subunit 16; and Fas-associated protein 1 (FAP-1), a protein tyrosine phosphatase that binds to the Fas receptor near its carboxy terminal and is thought to decrease the intensity of signals generated by this receptor. 23

Proposed Catalysis ofthe Homodimerization of TRADD, RIP, and FADD by the Cytoplasmic Domain of Ligand-Activated Receptor. According to the molecular-switch model presented in Figure 2C, ligand binding causes a conformational change in the cytoplasmic domain of a dimeric receptor. The relation between the death domains (red rectangles) then favors the formation of TRADD, RIP, or FADD (or MORT-1) homodimers,...