Cytokine-related Mechanisms of Apoptosis

The concept that accumulation of unwanted cells is due to uncontrolled cell proliferation has been amended. Cells also accumulate when proliferation rates are normal but death rates are lower. The realization that cell death is a normal process in development and cellular homeostasis has opened a completely new avenue for exploration of the causes and treatments of disease.

Figure 1. Signal transduction by TNF RI (A. and B.), TNF RII(C.), and FAS/APO-1/CD95 (D.)

Failure of cells to die is an integral mechanism in some cancers and autoimmune disorders. Conversely, an abnormal increase in cell death is observed in neurodegenerative disorders and ischemic injury. During development the number of cells differentiating into a cell type designed to perform specific functions often exceeds the number of cells required to perform the functions (e.g., innervation).2 The superfluous cells are eliminated through cell death.3, 4 A miscue in this process is detrimental to the development of the organism.

Pathogens have evolved molecules that affect the death pathway. HIV appears to induce cell death by attacking the molecular machinery involved in inhibiting cell death.5 Other pathogens preserve their host by inhibiting the induction of a cell death pathway.

In eukaryotic multicellular organisms a mechanism has evolved that eliminates unwanted cells with minimal disturbance to the organism. This process of "programmed cell death" or "cell suicide" is called apoptosis. The apoptotic elimination of unwanted cells is through a program engrained within the targeted cell.6 The suicide program leads to partial autodigestion of intracellular components, with the cell splitting into plasma-membrane limited vesicles called apoptotic bodies. The lipid distribution between the outer and inner leaflets of the plasma membrane enclosing the apoptotic bodies is critical for elimination of these cellular remnants. Phosphatidylserine, normally present in the inner leaflet and excluded from the outer leaflet, is exposed in the outer leaflet of apoptotic bodies. The presence of phosphatidylserine in the outer leaflet serves to mark apoptotic bodies for elimination by phagocytic cells.7 Thus, the process of apoptosis does not elicit an inflammatory response. This is in marked contrast to the inflammatory response during necrosis, where loss of plasma membrane integrity and release of cytoplasmic contents into the surroundings accompanies cell death and provokes an inflammatory response.

There are two pathways that lead to apoptosis: i) positive induction by ligand binding to a plasma membrane receptor and ii) negative induction by loss of a suppressor signal. Each leads to activation of cysteine proteases with homology to IL-1 beta-converting enzyme (ICE). Some cysteine proteases are present as zymogens (inactive pro-forms) in non-apoptotic cells and are activated by autoproteolysis or by cleavage by other cysteine proteases. The cascade of cysteine protease activation leads to proteolysis of proteins critical for cell viability.

Positive Induction of Apoptosis: The tumor necrosis factor (TNF) family of receptors (TNFR) is characterized by homology in the extracellular domains. Some of these receptors initiate apoptosis, some initiate cell proliferation and some initiate both. Signaling by this family requires clustering of the receptors by trimeric ligand and subsequent association of proteins with the cytoplasmic region of the receptors.

The TNFR family contains a sub family with homologous cytoplasmic 80 amino acid domains. This domain is referred to as a death domain (DD), so named because proteins that contain this domain are involved in apoptosis.8

The distinction between members of the TNFR family is exemplified by two TNFRs coded by distinct genes. TNF RI (55 kDa) signals both the initiation of apoptosis and the activation of the transcription factor NF kappa B (NFkB). TNF RII (75 kDa) functions to signal activation of NFkB but not the initiation of apoptosis. TNF RI contains a DD; TNF RII does not.

Other members of the TNFR family with DD are Fas/APO-11,9 and DR-3.10 Aggregation of these receptors initiates apoptosis. Aggregation of the receptors by trimeric ligand orients the DD in a conformation that recruits adaptor proteins. The adaptor proteins also contain a DD and association of adaptors with the receptor is via a homotypic DD interaction. As will become obvious below, homotypic domain interactions between proteins involved in revving up the cell for self-destruction is a common theme of apoptosis.

Aggregation of TNF RI through association with trimeric ligand induces association of the adaptor proteins TRADD (TNFR associated protein with a DD) and FADD (Fas associated protein with a DD, Fig 1A).11, 16 FADD also contains a 'death effector domain' (DED). The DED of FADD recruits the zymogen form of the cysteine protease FLICE/Mach alpha 1. FLICE/Mach alpha 1 zymogen contains a large pro-region that is homologous to the DED of FADD.12, 13 This pro-region is removed during activation. The large DED-containing pro-region appears to distinguish FLICE/Mach alpha 1 from most of the other cysteine proteases that contain small, DED-lacking pro-regions, suggesting that the proteases activated earliest in the apoptotic proteolytic cascade have large DED-containing pro-regions.

TNF can also induce activation of NFkB when bound to TNF RI (Fig. 1B). The adaptor TRADD has at least two binding domains, DD and TRAF (TNFR associated factor), of approximately 280 amino acids located near the N-terminus and C-terminus, respectively.11 DD causes TRADD to associate with TNF RI, and TRAF, through homotypic TRAF do main interactions, recruits TRAF-2. The C-terminus of TRAF-2 contains a "ring domain" that binds and activates NFkB.11, 14 NFkB is then transported into the nucleus where it promotes transcription.

The TNF RII subgroup receptors contain a cytoplasmic domain different from TNF RI. Oligomerization of TNF RII leads to the binding of two proteins, TRAF-1 and TRAF-2, which form an oligomer through homotypic interaction of their TRAF domains (Fig. 1C).15 TRAF-2 then binds and activates NFkB.15 CD40-mediated activation of NFkB proceeds through a similar pathway except that an additional protein, TRAF-3, is involved.17

Fas/APO-1/CD95-initiated cytosolic adaptor complex is similar to the complex formed on the cytoplasmic domains of aggregated TNF RI (Fig. 1D). An additional protein, RIP (receptor interacting protein), has been identified in the complex formed on Fas.18 RIP contains a DD at its N-terminus and a serine kinase-domain at its C-terminus.19 The function of the kinase is unknown. RIP also interacts with and activates NFkB. There is an adaptor protein named RAIDD (RIP-associated ICH1 homologous protein with a DD) that binds to the pro-region of ICH1 and links RIP to the cysteine protease ICH1.20

Similar to the complex formed on TNFRI, the DED of the adaptor FADD binds the DED present in the pro-region of the cysteine protease FLICE/Mach alpha 1.13 Thus, there appears to be redundant pathways by which apoptosis can be initiated by a single receptor. Although not shown in Fig. 1A, RIP has also been found associated with the TNF RI.20 The pro-region released after proteolytic cleavage of FLICE/Mach alpha 1 has been found on the DD complex associated with Fas.13 It appears that active FLICE/Mach alpha 1 is released to initiate the cysteine protease cascade, with the DED-containing pro-region remaining with the complex.

Many of the proteins in the cytoplasmic apoptotic-inducing complexes have been identified, but their regulation is unknown. Many are phosphorylated, but significance of post-translational modifications is unclear. Other proteins also associate with the cytoplasmic receptor-associated complexes. Two mammalian proteins, c-IAP-1 and c-IAP-2 (cellular inhibitor of apoptosis), are found associated with TRAF-1, TRAF-2 and TNF RII cytoplasmic complex.20 These proteins contain regions homologous to the baculoviral inhibitors of apoptosis (all contain BIRs, baculovirus IAP repeats whose functions are unknown). Clarification of the functions of these proteins and the identification of other complex-associated proteins will undoubtedly aid in dissecting the involvement and functions of proteins involved in direct activation of apoptosis and of other signaling events that depend on the TNFR family.

Loss of Cytokine-Dependent Suppression of Apoptosis: The viability of many cells is dependent on a constant or intermittent supply of cytokines or growth factors. In the absence of the factor, the cells under go apoptosis. The Bcl-2 family of proteins are central components to apoptosis resulting from the absence of incoming signals generated by cytokine binding. Over-expression of some family members (e.g., Bcl-2 and Bcl-xL) suppresses apoptosis when cytokines are withdrawn. Over-expression of others (e.g., Bad, Bax, and Bik) override the incoming signals from the cytokine-receptor and induce apoptosis. The suppressor members of the Bcl-2 family form homodimers or form heterodimers with inducer-type members. Suppressor members prevent apoptosis when homodimerized but are ineffective in protecting from apoptosis when heterodimerized with an inducer member. Thus, the dimerization state of the suppressor members of the Bcl-2 family is one determinant of cellular life or death. Recent reports have begun to unite the Bcl-2 family of proteins with other components into a single path way that, although similar in many aspects to the TNF/Fas pathway, is unique and provides an additional method to eliminate unwanted cells.

Figure 2. Activationof apoptosis on withdrawal of cytokine. Panel A shows cell with cytokine and panel B shows cell without cytokine.

The molecular mechanisms of apoptosis caused by withdrawal of cytokines (Fig. 2) may be best under stood by starting with the roles of the Bcl-2 family members. The suppressors Bcl-2 and Bcl-xL are integral membrane proteins22 found as dimers facing the cytosol. They are predominantly in the outer mitochondrial membrane, with less in the endoplasmic reticular and nuclear membranes.23 Dimerized Bcl-2 or Bcl-xL have regions that bind to a protein that, in mammalian cells, is the homolog to CED-4, an adaptor protein identified in the nematode Caenorhabditis elegans.24, 25 The mammalian homolog of CED-4 has not been identified, but there is evidence for its existence.25

CED-4 also binds to zymogens of cysteine proteases that contain large pro-regions. In this tripartite configuration, the cysteine protease is maintained as a zymogen (inactive). When the CED-4 interaction with Bcl-2 or Bcl-xL is disrupted, CED-4 interaction with the zymogen appears to be altered to induce autoproteolytic activation of the cysteine protease. FLICE/Mach alpha 1 is a member of the family of cysteine proteases that associate with CED-4. As in the TNF RI/Fas pathway, FLICE/Mach alpha 1 may be one of the earliest proteases activated during initiation of the proteolytic activation cascade.

The finding that the adaptor protein CED-4 prevents zymogen activation when bound to Bcl-2 or Bcl-xL homodimers but activates zymogens when not bound likely explains the ability of overexpressed Bcl-2 or Bcl-xL to protect cells from apoptosis caused by numerous cellular signals or insults. When Bcl-2 or Bcl-xL homodimer levels are high, CED-4 is bound to the homodimers and is, therefore, unable to activate the cysteine protease cascade. This would also explain the anti-apoptotic activity of Bcl-2 and Bcl-xL truncated forms missing the mitochondrial targeting amino acid sequence.26 As long as Bcl-2 or Bcl-xL can homodimerize, they have the potential to sequester all of CED-4.

Bcl-xL heterodimerization with the inducer family members, Bax, Bak, or Bik, causes release of the CED-4-zymogen heterodimer from Bcl-xL.25 During or after dissociation from Bcl-xL, CED-4 induces the zymogen to undergo autoproteolytic activation. Thus, activation of cysteine proteases in this pathway directly depends on the dimerization state(s) of the Bcl-2 family members, and the dimerization state appears to be controlled by the availability of the apoptosis inducer members. The availability of inducer members for heterodimerization with Bcl-2 or Bcl-xL may be regulated by signals generated by the binding of cytokines.

Bad is a proapoptotic member of the Bcl-2 family and is sequestered in the cytosol when cytokines are present.27 Sequestration occurs through the binding of Bad by a multifunctional protein, 14-3-3.27 This protein binds to phosphorylated proteins, and the Bad that is bound by 14-3-3 is phosphorylated on serine residues. Upon removal of cytokines, Bad becomes dephosphorylated at specific serine residues, dissociates from 14-3-3 and heterodimerizes with Bcl-xL. Bad phosphorylation and reassociation with 14-3-3 can be induced by addition of cytokines. Furthermore, hyper-phosphorylated Bad does not bind to Bcl-xL. These results suggest that the cytokine-engaged receptor prevents apoptosis by inducing the phosphorylation of the apoptotic-inducing members of the Bcl-2 family, thereby making them unavailable for dimerization with Bcl-2 or Bcl-xL.

The identity of the kinase pathway that couples cytokine receptor binding to phosphorylation of the pro-apoptotic Bcl-2 family members is under intense study. Activated protein kinase B (PKB) has been shown to prevent the apoptotic cell death of Rat-1 cells that are induced by c-myc.28 Growth factor-dependent activation of PKB is through phosphatidylinositol-3,4-bisphosphate generated by phosphoinositide 3-kinase (PI3K) whose activation depends on the GTP-binding protein RAS. PI3K is active when cytokines are bound to their receptors (reviewed in ref. 29). A potential roleof RAS activation of another kinase, RAF, has also been suggested.30 Similarly, other phosphorylation events have been shown to affect apoptosis under experimental con ditions.31, 32 The great advances in understanding the pathways that trigger apoptosis sets the stage for investigations into the regulation of the pathways by kinases and by levels of expression of proteins (e.g., the Bcl-2 family).

The Bcl-2 family may also be central to another type of apoptosis induction. It has long been recognized that the Bcl-2/Bcl-xL dimerization state regulates the sensitivity of cells to death induced by free radicals.33, 34 Free radical-induced cell death is accompanied by lipid peroxidation. Bcl-2 overexpression prevents free radical-induced lipid peroxidation. Studies using in vitro assays of apoptosis have shown that activation of the cysteine protease CPP32 is dependent on the release of cytochrome c from the mitochondria.35 Cytosolic cytochrome c was also found in cultured cells and over-expression of Bcl-2 decreased cytosolic cytochrome c.36, 37 Cytochrome c is a one-electron carrier in mitochondrial electron transport. It is possible that cytosolic cytochrome c propagates or initiates free radical production. Although the mechanism for activation of apoptosis by these events is unknown, the findings suggest a function for the Bcl-2 family in regulating free radical damage. Bcl-xL forms pores in artificial membranes38 and have a crystal structure similar to the B-subunit of diphtheria toxin.39 Diphtheria B-subunit translocates the A-subunit across membranes. Thus, Bcl-xL and Bcl-2 have the potential to translocate materials across membranes. Mitochondrial membrane pore formation and subsequent loss of mitochondrial transmembrane potential has been found to be one of the earliest cellular events associated with apoptosis.40 The molecular relationship between cytochrome c, the Bcl-2 family, activation of cysteine proteases, and free radicals remains to be determined.


  1. Thompson, C.B. (1995) Science 267:1456.
  2. Oppenheim, R.W. (1991) Annu. Rev. Neurosci. 14:453.
  3. Nagata, S & P. Goldstein (1995) Science 267:1449.
  4. Jacobson, M.D. et al. (1997) Cell 88:347
  5. Strack, P.R. et al. (1996) Proc. Natl. Acad. Sci. USA 93:9571.
  6. Weil, M. et al. (1996) J. Cell Biol. 133:1053.
  7. Fadok, V.A. et al. (1992) J. Immunol. 148:2207.
  8. Tartaglia, L.A. et al. (1993) Cell 74:845.
  9. Pitti, R.M. et al. (1996) J. Biol. Chem. 271(22):12687.
  10. Chinnaiyan, A.M. et al. (1996) Science 274:990.
  11. Hsu, H. et al. (1996) Cell 84:299.
  12. Muzio, M. et al. (1996) Cell 85:817.
  13. Boldin, M.P. et al. (1996) Cell 85:803.
  14. Rothe, M. et al. (1994) Cell 78:681.
  15. Rothe, M. et al. (1995) Science 269:1424.
  16. Chinnaiyan, A.M. et al. (1995) Cell 81:505.
  17. Hu, H.M. et al. (1994) J. Biol. Chem. 269(48):30069.
  18. Stanger, B.Z. et al. (1995) Cell 81:513.
  19. Hsu, H. et al. (1996) Immunity 4:387.
  20. Duan, H. & V.M. Dixit (1997) Nature 385:86.
  21. Rothe, M. et al. (1995) Cell 83:1243.
  22. Nguyen, M. et al. (1993) J. Biol. Chem. 268(34):25265.
  23. Krajewski, S. et al. (1993) Cancer Res. 53:4701.
  24. Wu, D. et al. (1997) Science 275:1126.
  25. Chinnaiyan, A.M. et al. (1997) Science 275:1122.
  26. Borner, C. et al. (1994) J. Cell Biol. 126(4):1059.
  27. Zha, J. et al. (1996) Cell 87:619.
  28. Kauffmann-Zeh, A. et al. (1997) Nature 385:544.
  29. Franke, T.F. et al. (1997) Cell 88:435.
  30. Wang, H-G. et al. (1996) Cell 87:629.
  31. Nishina, H. et al. (1997) Nature 385:350.
  32. Haldar, S. et al. (1995) Proc. Natl. Acad. Sci. USA 92:4507.
  33. Hockenberry, D.M. et al. (1993) Cell 75:241.
  34. Kane, D.J. et al. (1993) Science 262:1274.
  35. Liu, X. et al. (1996) Cell 86:147.
  36. Kluck, R.M. et al. (1997) Science 275:1132.
  37. Yang, J. et al. (1997) Science 275:1129.
  38. Minn, A.J. et al. (1997) Nature 385:353.
  39. Muchmore, S.W. (1996) Nature 381:335.
  40. Castedd, M. et al. (1996) J. Immunol. 157:512.