Immunology Today, vol, 7, No. 4, 1986
Death and the cell Cell death is an important component of many- perhaps allimmunological reactions. Target cells die after attack by complement, cytotoxic T lymphocytes, K cells, NK cells or lymphotoxins. In lymphoid cells themselves death occurs in parallel with proliferation in the normal thymus and in the periphery during immune responses. Celldeath, therefore, is sometimes a pathological event and sometimes a physiological process, apparently as tightly regulated as cell proliferation. In this article Edward Duvall and Andrew Wyllie develop the theme that the internal organization and metabolism of nucleated cells determines their mode of death by one or other of two relatively stereotyped patterns, necrosis or apoptosis. They discuss the general mechanisms involved in each pattern, their relevance to immunology, and their biological significance. There are two major morphologically and biochemically distinct modes of death in nucleated eukaryotic cells. Historically, the first to be recognized was necrosis, which occurs as a result of complement attack, severe hypoxia, hyperthermia, lytJc viral infection, or exposure to a variety of toxins and respiratory poisons. These diverse lethal stimuli increase the permeability of the plasma membrane, either through alterations in its structure (complement, viral lysis), or by failure of the cationic ion membrane pumps (hypoxia, respiratory poisons) 1. Under these circumstances Na +, K +, Ca 2+ and Mg 2+ move down concentration gradients with concomitant uptake and intracellular redistribution,of water. Initially, this produces a series of reversible changes in the injured cells. The endoplasmic reticulum dilates; the cell alters shape with 'blebbing' of the surface, (possibly as a result of alterations in Ca 2 + concentrations acting upon the cytoskeleton, compounded by the increase in intracellular water); the mitochondria become denser as their inner membrane shrinks away from their outer membrane; the nuclear chromatin flocculates; and protein synthesis declines. This reversible phase is followed, often with explosive rapidity, by irreversible changes. The mitochondria undergo 'high amplitude swelling' with dilation of both inner and outer compartments and the appearance within them of densely lipid-rich aggregates, probably of matrix lipoproteins (Fig. 1).. Several mechanisms have been suggested to account for the cell changes becoming irreversible: release of lysosomal enzymes; generation of toxic oxygen radicals; depletion of cellular ATP below a critical level; and the activation of calcium-dependent phospholipases. Although lysosomes swell because of the changes in ion and water fluxes, lysosomal enzymes are only released at a late stage, playing a part in the final dissolution of the cell after it has passed the point of no return 2. There is little evidence for the generation of toxic oxygen metabolites as a general mechanism of cell death, but good evidence that cellular depletion of ATP is associated with necrosis 3. A fall in ATP concentration alone, however, is ineffective in inducing death in the absence of extracellular Ca 2+ or in the presence of agents such as
Department of Pathology, University of Edinburgh Medical School, EdinburghEH89AG, UK
E. Duvall and A.H. Wyllie chlorpromazine or verapamil that block calcium movement 4. The most plausible hypothesis ~ is that movements of Ca 2+ activate endogenous, membrane bound, Ca2+-dependent phospholipases, leading to widespread and lethal disruption of cellular membranes by both direct enzyme action and the detergent action of accumulated long chain fatty acyl esters. Chlorpromazine blocks both the loss of phospholipids from, and development of increased permeability by, cell membranes damaged by ischaemia 4. Apoptosis
More recently it has become apparent that cells may die in a way differing morphologically and biochemically from necrosis, termed apoptosis s. Apoptosis is observed when death is part of organized tissue reactions as in embryogenesis (e.g., the disappearance of the inter~digital cells during the formation of the digits from the solid limb paddle); in metamorphosis (the resorption of the tadpole tail); in endocrine-dependent tissue atrophy (the thinning of the adrenal cortex of the rat after birth); and in the control of normal tissue turnover (breast during the menstrual cycle6). It is also seen in regressing tumours. A cell undergoing apoptosis rounds up, severing junctions with its neighbours and losing microvilii (Fig. 2). At the same time the cytoplasm condenses but the morphology of mitochondfia and ribosomes is preserved. The endoplasmic reticulum dilates, probably as a result of a net shift of water from the cytoplasm, and forms vesicles which tend to fuse with the surface membrane, giving a characteristic bubbling appearance on electron microscopy (Fig. 3). The chromatin rapidly forms dense crescent-shaped aggregates lining the nuclear membrane (Fig. 4) (this contrasts with the flocculatJon of the chromatin during necrosis), and the nucleolus fragments. Complex invaginations develop in the nuclear membrane segmenting the nucleus. The plasma membrane also becomes convoluted, so that the cell separates into a NORMAL
REVERSIBLE SWELLING IRREVERSIBLE SWELLING DISINTEGRATION
Iq;; t"
mitochondrial changes [ membrane breakdown chromatin pattern conserved
Fig.1. Morphologicalmanifestationsof necrosis. ~) 1986, ElsevierScience Publishers BV., Amsterdam
115 0167
4919/86/$02.00
Immunology Today, vol. 7, No. 4, 1986
-review CONDENSATION FRAGMENTATION .
PHAGOCYTOSIS & .\ S E C O NDARYNEC \@ROSIS
nuclear changes ~ / nuclearchangq . . ~ k k ~ a g c gestion e " by mitochondria conserved membranes intact ~ - macropnages or other cells
Fig.2. Morphology of apoptosis. Most of the resultant fragments of cells (apoptotic bodies) are phagocytosed although if the process occurs in an epithelium lining a cavity, some may be lost into the lumen. The apoptotic body within a phagosome undergoes enzymatic digestion which morphologically resembles necrosis and has thus been termed "secondarynecrosis'. cluster of membrane bound segments, 'apoptotic bodies', which often contain morphologically normal mitochondria and other organelles. Cells in this state are rapidly recognized, phagocytosed and digested, either by macrophages or epithelial cells or, in tumours, by adjacent tumour cells (Fig. 4). Although lysosomal activation occurs within the phagocytic cell, the lysosomal enzymes of the dying cell do not seem to participate in apoptosis 7,8. Incubation of rodent cortical thymocytes for several hours in vitro with a synthetic glucocorticoid hormone 9 induces apoptosis in a proportion of the cells which, because of their increased density, can then be separated from the remaining apparently viable cells. The chromatin of such apoptotic thymocytes is degraded into discrete fragments, all integer multiples of about 190 base pairs of DNA ~°. This pattern of cleavage is due to the vulnerability of the linker DNA running between nucleosomes. In contrast, during necrosis, DNA degradation is a late phenomenon; the chromatin is digested by pro!
116
teases and endonucleases into a continuous spectrum of sizes as the proteases destroy the histones and expose the entire length of DNA to the endonucleases. The identity of the endonuclease(s) responsible for the chromatin changes in apoptosis is not yet certain. Thymocytes contain a neutral endonuclease which is a likely candidate 11. Its requirement for Mg 2+ and Ca 2+ but inhibition by Zn 2+ matches the effects of these ions on apoptosis itself 12. As this enzyme is inhibited by zinc ions at concentrations similar to those found in the normal nucleus, an exchange of zinc for magnesium and calcium within the nucleus during apoptosis may initiate the characteristic chromatin degradation. The role of zinc is further supported by the observation that zinc deficiency leads to excessive apoptosis in the thymus (with resulting profound immunodeficiency) 13 and in the small intestine 14. Interestingly, the presence of zinc can inhibit programmed cell death in the developing Mullerian duct 15. Apoptosis is often dependent upon active metabolism and protein synthesis by the dying cell 1°'16, and although it is tempting to speculate that some of the proteins synthesized during apoptosis may be involved in a putative zinc transport mechanism, they are probably not metallothioneins, the proteins usually associated with heavy metal ion transport 17. Another problem is how apoptotic bodies in the tissue spaces are recognized by tissue macrophages, adjacent epithelial cells or tumour cells and phagocytosed without eliciting any inflammatory response. Mouse peritoneal macrophages bind apoptotic thymocytes in preference to normal ones. This preferential binding of apoptotic cells is blocked in vitro by simple sugars, particularly by N-acetyl glucosamine and its dimer N,N'diacetylchitobiose 18. This suggests that changes in the expression of sugar residues upon the cell surface are recognized by endogenous macrophage 'lectins'. Interestingly, the targets for the presumed macrophage lectin may be sugar residues normally buried within mature glycan chains 18. The= mechanisms whereby epithelial or tumour cells recognize their apoptotic neighbours is unknown. i
Fig.3. Scanning electron micrographs of thymocytes. (a) shows a normal thymocyte with occasionalmicrovilli; (b) is a thymocyte in the early stagesof apoptosis with dumping and shortening of microvilli; whilst (c) shows a cell with the characteristic appearance of swollen vacuolesof endoplasmic reticulum fusing with the cell surface.
Immunology Today, voL 7, No. 4, 1986
Immunologically mediated cell killing How do the deaths provoked by immune mechanisms fit into the classification of death given above? Ultrastructural studies have shown that the lesions induced by complement are tubes of polymerized activated C9 forming channels through the cell membrane 19. It is probable that Ca 2+ influx through these membrane lesions is responsible for the death of the cell rather than simple 'osmotic lysis' as the overall morphology of nucleated cells dying under complement attack is that of necrosis 2,2°. In contrast, when cell killing is mediated by effector cells, the mode of death is apoptosis. The process of killing of cells by specific cytotoxic T lymphocytes (CTL) in vitro has been divided into three stages21: recognition, programming for death* and disintegration. The recognition stage has a half-life of about 1 min, and an absolute requirement for divalent cations. The second stage (programming for death), which has a half-life of about 5 min, is also dependent on divalent cations and only occurs in the. presence of viable CTL. The third 'disintegration' phase, characterized by membrane damage and SlCr release, has a half-life of approximately 1O0 rain, is temperature dependent and does not require functioning CTL cells although it is accelerated in their presence 21,22. During programming for deatlq the CTL inflicts irreversible damage upon the target cell. Initially, as during complement attack 23, cell membrane permeability to 86Rb increases. This continues into the disintegration phase, and probably indicates non-lethal membrane perturbation rather than being the first manifestation of death 21. However, unlike the effects of complement attack, small DNA fragments become detachable from the nucleus within a few minutes 24. Ultimately the majority of the DNA is cleaved into the short polynucleosome chains characteristic of apoptosis. This fragmentation requires Ca 2+ and Mg 2+ and is inhibited by Zn 2+. The event determining DNA breakdown may be the same as that responsible for 'programming for death' as they both have the same half-life 2s. Thus, both the pattern of DNA degradation and its sensitivity to divalent cations are the same as those of steroid-induced apoptosis in thymocytes. However, although an endonuclease with the requisite properties has been found in thymocytes 11, no such endonuclease has yet been found in either effector CTL or their targets (P81 5 mastocytoma cells) 12. Perhaps in CTL killing, nucleases are activated by a rapid mechanism such as phosphorylation only during, or as a consequence of, 'programming for death'. CTL killing and thymocyte apoptosis also differ in that RNA and protein synthesis by the target cell is not required for CTL to kill mastocytoma cells 12, whereas inhibition of macromolecular synthesis abrogates thymocyte apoptosis ~°. There is at present no completely satisfactory explanation combining these results, but they do suggest that apoptosis can be activated at different points in different cell types. The morphological changes in CTL-mediated killing are also typical of apoptosis, both in light microscopy and ultrastructurally 25 ,26 . Moreover, the morphology of apoptosis has been observed in vivo during reactions which are presumed to be mediated by CTL including the *Usuallytermed'programmingfor lysis',but the modeof c~eathisapoptosisand not lysis.
r Vl WSrejection of pig liver allografts, graft versus host reactions, lichen planus*, fixed drug eruptions t, and active chronic hepatitis 27. The morphology of antibody-dependent cell cytotoxicity (K cell attack) is that of apoptosis 21'28, and P815 mastocytoma cells attacked by K cells release radiolabelled molecules in the same sequence as similar cells attacked by CTL, i.e., early release of 86Rb, later release of 51Cr-labelled molecules, and slow release of [12Sl]dU_labelled DNA21. Natural killer (NK) cells also induce apoptosis. Both the programming and disintegration stages have similar cation and temperature requirements to the same stages in CTL killing 29, whilst the transmission and scanning electron microscopic appearances are those of apoptosis 3°,31 . Lastly, lymphotoxins secreted by lymphocytes also induce apoptosis. Mouse fibroblasts on exposure to human lymphotoxins shrink and contract in exactly the same way as fibroblasts killed by CTL32. In addition, after lymphotoxin attack cells released DNA fragments similar to those released after CTL killing 33. Two points relevant to this discussion arise from studies on lymphotoxin induced apoptosis. Firstly, there is no early increase in 86Rb permeability 34 confirming that this change is not an essential preliminary to cell death. Secondly, although there is an increase in synthesis of mRNA during the induction of death, it is not essential, as inhibition of RNA *A skin diseasewhich can be drug or chemical-linked.A local ceil-mediated immuneresponsemaybeinvolvedin the pathogenesis. $ Swellingor lesionson the skinin reactionto someanalgesicsandantibiotics.
Fig, 4. Ultrastructureof apoptosis.A thymicmacrophage(M) has ingestedfour apoptotic thymocytes (A) which are distinguishableby the denselycondensed chromatin of their nucleL Comparethe chromatinpattern with that of adjacent normal thymocytes(N), x 4,600.
117
Immunology Today, voL 7, No. 4, 1986
-review5 synthesis by actinomycin D actually potentiates killing by lymphotoxin 3s. Recent studies have shown interesting parallels between the actions of effector cells and complement. CTL, K and NK cells all produce membrane lesions qualitatively similar in ultrastructure to those produced by complement. Both CTL and NK cells possess granules which contain molecules (polyperforins) from which complexes very similar to those formed by the later components of complement may be assembled and inserted into the membrane of the target cell 36. Why, however, if effector cells and complement produce similar membrane lesions do effector cells induce apoptosis whereas complement attack leads to necrosis? The difference may be determined by the number or distribution of the lesions upon the cell surface. Those produced by an effector cell may be fewer or distributed over a smaller area than those produced in experiments using complement where conditions are usually chosen to maximize the number of 'hits' on the target cell. There are other systems where severity of insult determines the form of death. Mild degrees of hypoxia, toxic assault or heat shock have been shown to lead to apoptosis whilst more severe hypoxia, greater concentrations of toxins or higher temperature lead to necrosis 27. Mild or strictly localized degrees of cell membrane damage might allow fluxes of ions such as Ca2+ and Mg 2+ sufficient to trigger the mechanism leading to apoptosis but insufficient to activate significant amounts of phospholipase. More severe membrane damage might lead to rapid activation of phospholipases, disrupting the function of the cell so severely that the systems leading to endonuclease activation and to apoptosis would be destroyed before their actions could be manifest. Further, the threshold between apoptosis and necrosis might vary between cells of different types and at various stages in the lifespan of a single cell. Cell death and homeostasis in the immune system
In the physiological regulation of lymphoid populations cell death may be essential for deletion of cells with inappropriate specificities, may counterbalance mitosis in lymphoid and accessory cell turnover 37, or may contribute to the contraction of previously expanded populations. Although necrosis is not observed under physiological conditions apoptosis is abundant in thymus s, lymph nodes 27 and spleen 38. In the thymus (where at least 80% of cells die in sZtu39) mitoses outnumber apoptotic cells during enlargement whilst the reverse is true during involution 41. In lymph nodes and spleen the so-called tingible body macrophages (conspicuous histological features of reactive centres) in fact contain the chromatin from large numbers of apoptotic cells. Most of these apoptotic cells are lymphocytes; some are plasma cells38. These endogenous cell deaths appear to be regulated but the mechanism is not understood. By analogy to developing and endocrine-dependent tissues, the deaths could follow production of lethal stimuli or the removal of necessary growth factors. Apoptosis of human lymphocytes has been observed on the withdrawal of support by conditioned media containing interleukin (A.S. Krajewski, unpublished). Conclusions 118
Nucleated, eukaryotic cells die not by a process of disordered dissolution but by one or other of two
different sets of stereotyped reactions; necrosis and apoptosis. Necrosis is associated with major perturbations in the cellular environment and has been regarded as a pathological response, whilst apoptosis has been observed in many instances where the death of the cell is not pathological but appears to be part of homeostatic regulation. However, as discussed above, which of the two reactions occurs may be determined not by any putative 'pathological' or 'physiological' quality of the initiating stimulus but. by its severity or distribution, or by the state of the target cell. The mechanisms of necrosis are relatively well understood. Two molecular events in apoptosis are reasonably defined: the activation of an endonuclease within the dying cell, responsible for the characteristic chromatin cleavage, and the expression of certain sugar residues on the cell membrane that permit recognition and hence phagocytosis of the dying cell. The mechanisms controlling and coordinating these reactions are obscure. Apoptosis is the mode of death induced by effector cells and lymphotoxins and is probably involved in many of the regulatory mechanisms in immunology - the clonal selection of lymphocytes and the regulation of sizes of populations of lymphocytes and cells derived from them. The authors are supported by the Cancer Research Campaign and the Wellcome Trust.
References
1 Trump, B.F., Berezesky,I.K. and Osornio-Vargas,A.R. (1981) in Cell Death in Biology and Pathology, (Bowen, I.D. and Lockshin, R.A. eds) pp. 209-242, Chapman and Hall, London and New York 2 Hawkins, H.K., Ericsson,J.L.E., Biberfeld, P. etal. (1972)Am. J. Pathol. 68, 255 287 3 Jennings, R.B.and Reimer, K.A. (1981)Am. J. Pathol. 102, 214-255 4 Farber,J.L., Chien, K.R. and MJttnacht, S. (1981)Am. J. PathoL 102, 271-281 5 Wyllie, A.H., Kerr, J.F.R.and Currie, A.R. (1980) Int. Rev. CytoL 68, 251-306 6 Anderson, T.J, Ferguson, D.J.P.and Raab, G.M (1982) Br. J. Cancer46, 376-382 7 Russell,J.H. (1983) ImmunoL Rev. 72, 97-118 8 Morris, R.G., Hargreaves,A.D., Duvall, E. etal. (1984)Am. J. Pathol. 115,426-436 9 Wyllie, A.H. and Morris, R.G. (1982) Am. J. Pathol. 109, 78-87 10 Wyllie, A.H., Morris, R.G., Smith, A.L. etal. (1984)J. PathoL 142, 67-77 11 Nakamura, M., Saraki,Y., Watanabe, N. etaL (1981) J. Biochem. (Tokyo)89, 143-152 12 Duke, R.C., Chervenak,R. and Cohen, J.J. (1983) Proc. Nail Acad. Sci. USA 80, 6361-6365 13 Fraker, PJ., Haas,S.M. and Luecke, R.W. (1977)J. Nutr. 107, 1889-1895 14 Elmes,M.E. (1977)J. Pathol. 123,219-224 15 Budzik, G.P, Hutson, J.M., Ikawa, H. etaL (1982) Endocrinology110, 1521 1525 16 Leiberman,MW., Verbin, R.S., Landay, M. etaL (1970) CancerRes. 30, 942-951 17 Maytin, E.V. and Young, D.A. (1983) J. BioL Chem. 258, 12718-12722 18 Duvall, E., Wyllie, A.H. and Morris, R.G. (1985) Immunology 56,351 358
Immunology Today, vol. 7, No. 4, 1986
r vl 19 Tschopp, J., Muller-Eberhard, H.J. and Podack, E.R. (1982) Nature (London) 298, 534-537 20 Goldberg, B. and Green, H. (1959) J. Exp. Med. 109,
505 510 21 Sanderson, C.J. (1981)Biol. Rev. 56, 153-197 22 Martz, E., Burakoff, S.J. and Benacerraf, B. (1974) Proc Natl Acad. Sd. USA 71,177-181
23 Boyle, M.D.P., Ohanian, S.H. and Borsos, T.J. (1976) J. ImmunoL 117, 1346 1350 24 Russell, J.H. and Dubos, C.B. (1980) J. Immunol. 125, 1256 I261 25 Russell, J.H., Masakowski, V., Rucinsky, T. etal. (1982) J. Immunol. 128, 2087 2094 26 Don, M.M., Ablett, G., Bishop, C.J. et aL (1977) Aust. J. Exp. Biol. Med. Sci. 55, 407-417 27 Seafle, J., Kerr, J.F.R. and Bishop, C.J. (1982) PathoL Annu. 17(2), 229-259 28 Stacey, N.H., Bishop, C.J., Halliday, J.W. etal. (1985)J. Cell Sci. 74, 169 179 29 Hiserodt, J.C., Britvan, L.J. and Targan, S.R. (1982) J. ImmunoL 129; 1782 1787
l&-
30 Bishop, C.J. and Whiting, V.A. (1983) Br. J. Cancer48, 441-444 31 Roder, J.C., Kiessling, R., Biberfeld, P. etal. (1978) J. ImmunoL 121, 2509-2517 32 Russell, S.W., Rosenau, W. and Lee, J.G. (1972)Am. J. Patho169, 103-118 33 Ruddle, N.H. (1985) Immunol. Today 6, 156-159 34 Rosenau, W., Goldberg, ML. and Burke, G.C. (1973) J. Immunol. 111, 1128-1135 35 Kunitomi, G., Rosenau, W., Burke, G.C. etaL (1975)Am. J. Pathol. 80, 249-260 36 Podack, E.R. (1985) Immunol. Today 6, 21-27 37 Fossum, S. and Ford, W.L. (1985)Histopathology, 9,469 499 38 Swartzendruber, D.C. and Congdon, C.C. (1963) J. Cell BioL 19, 641-646
39 Joel, D.D., Chanana, A.D., Cottier, H. etal. (1977) Cell Tissue Kinet. 10, 57-69 40 Hinsull, S.M, Bellamy, D. and Franklin, A. (1977) Age and Ageing 6, 72 84
119
Death and the cell Cell death is an important component of many- perhaps allimmunological reactions. Target cells die after attack by complement, cytotoxic T lymphocytes, K cells, NK cells or lymphotoxins. In lymphoid cells themselves death occurs in parallel with proliferation in the normal thymus and in the periphery during immune responses. Celldeath, therefore, is sometimes a pathological event and sometimes a physiological process, apparently as tightly regulated as cell proliferation. In this article Edward Duvall and Andrew Wyllie develop the theme that the internal organization and metabolism of nucleated cells determines their mode of death by one or other of two relatively stereotyped patterns, necrosis or apoptosis. They discuss the general mechanisms involved in each pattern, their relevance to immunology, and their biological significance. There are two major morphologically and biochemically distinct modes of death in nucleated eukaryotic cells. Historically, the first to be recognized was necrosis, which occurs as a result of complement attack, severe hypoxia, hyperthermia, lytJc viral infection, or exposure to a variety of toxins and respiratory poisons. These diverse lethal stimuli increase the permeability of the plasma membrane, either through alterations in its structure (complement, viral lysis), or by failure of the cationic ion membrane pumps (hypoxia, respiratory poisons) 1. Under these circumstances Na +, K +, Ca 2+ and Mg 2+ move down concentration gradients with concomitant uptake and intracellular redistribution,of water. Initially, this produces a series of reversible changes in the injured cells. The endoplasmic reticulum dilates; the cell alters shape with 'blebbing' of the surface, (possibly as a result of alterations in Ca 2 + concentrations acting upon the cytoskeleton, compounded by the increase in intracellular water); the mitochondria become denser as their inner membrane shrinks away from their outer membrane; the nuclear chromatin flocculates; and protein synthesis declines. This reversible phase is followed, often with explosive rapidity, by irreversible changes. The mitochondria undergo 'high amplitude swelling' with dilation of both inner and outer compartments and the appearance within them of densely lipid-rich aggregates, probably of matrix lipoproteins (Fig. 1).. Several mechanisms have been suggested to account for the cell changes becoming irreversible: release of lysosomal enzymes; generation of toxic oxygen radicals; depletion of cellular ATP below a critical level; and the activation of calcium-dependent phospholipases. Although lysosomes swell because of the changes in ion and water fluxes, lysosomal enzymes are only released at a late stage, playing a part in the final dissolution of the cell after it has passed the point of no return 2. There is little evidence for the generation of toxic oxygen metabolites as a general mechanism of cell death, but good evidence that cellular depletion of ATP is associated with necrosis 3. A fall in ATP concentration alone, however, is ineffective in inducing death in the absence of extracellular Ca 2+ or in the presence of agents such as
Department of Pathology, University of Edinburgh Medical School, EdinburghEH89AG, UK
E. Duvall and A.H. Wyllie chlorpromazine or verapamil that block calcium movement 4. The most plausible hypothesis ~ is that movements of Ca 2+ activate endogenous, membrane bound, Ca2+-dependent phospholipases, leading to widespread and lethal disruption of cellular membranes by both direct enzyme action and the detergent action of accumulated long chain fatty acyl esters. Chlorpromazine blocks both the loss of phospholipids from, and development of increased permeability by, cell membranes damaged by ischaemia 4. Apoptosis
More recently it has become apparent that cells may die in a way differing morphologically and biochemically from necrosis, termed apoptosis s. Apoptosis is observed when death is part of organized tissue reactions as in embryogenesis (e.g., the disappearance of the inter~digital cells during the formation of the digits from the solid limb paddle); in metamorphosis (the resorption of the tadpole tail); in endocrine-dependent tissue atrophy (the thinning of the adrenal cortex of the rat after birth); and in the control of normal tissue turnover (breast during the menstrual cycle6). It is also seen in regressing tumours. A cell undergoing apoptosis rounds up, severing junctions with its neighbours and losing microvilii (Fig. 2). At the same time the cytoplasm condenses but the morphology of mitochondfia and ribosomes is preserved. The endoplasmic reticulum dilates, probably as a result of a net shift of water from the cytoplasm, and forms vesicles which tend to fuse with the surface membrane, giving a characteristic bubbling appearance on electron microscopy (Fig. 3). The chromatin rapidly forms dense crescent-shaped aggregates lining the nuclear membrane (Fig. 4) (this contrasts with the flocculatJon of the chromatin during necrosis), and the nucleolus fragments. Complex invaginations develop in the nuclear membrane segmenting the nucleus. The plasma membrane also becomes convoluted, so that the cell separates into a NORMAL
REVERSIBLE SWELLING IRREVERSIBLE SWELLING DISINTEGRATION
Iq;; t"
mitochondrial changes [ membrane breakdown chromatin pattern conserved
Fig.1. Morphologicalmanifestationsof necrosis. ~) 1986, ElsevierScience Publishers BV., Amsterdam
115 0167
4919/86/$02.00
Immunology Today, vol. 7, No. 4, 1986
-review CONDENSATION FRAGMENTATION .
PHAGOCYTOSIS & .\ S E C O NDARYNEC \@ROSIS
nuclear changes ~ / nuclearchangq . . ~ k k ~ a g c gestion e " by mitochondria conserved membranes intact ~ - macropnages or other cells
Fig.2. Morphology of apoptosis. Most of the resultant fragments of cells (apoptotic bodies) are phagocytosed although if the process occurs in an epithelium lining a cavity, some may be lost into the lumen. The apoptotic body within a phagosome undergoes enzymatic digestion which morphologically resembles necrosis and has thus been termed "secondarynecrosis'. cluster of membrane bound segments, 'apoptotic bodies', which often contain morphologically normal mitochondria and other organelles. Cells in this state are rapidly recognized, phagocytosed and digested, either by macrophages or epithelial cells or, in tumours, by adjacent tumour cells (Fig. 4). Although lysosomal activation occurs within the phagocytic cell, the lysosomal enzymes of the dying cell do not seem to participate in apoptosis 7,8. Incubation of rodent cortical thymocytes for several hours in vitro with a synthetic glucocorticoid hormone 9 induces apoptosis in a proportion of the cells which, because of their increased density, can then be separated from the remaining apparently viable cells. The chromatin of such apoptotic thymocytes is degraded into discrete fragments, all integer multiples of about 190 base pairs of DNA ~°. This pattern of cleavage is due to the vulnerability of the linker DNA running between nucleosomes. In contrast, during necrosis, DNA degradation is a late phenomenon; the chromatin is digested by pro!
116
teases and endonucleases into a continuous spectrum of sizes as the proteases destroy the histones and expose the entire length of DNA to the endonucleases. The identity of the endonuclease(s) responsible for the chromatin changes in apoptosis is not yet certain. Thymocytes contain a neutral endonuclease which is a likely candidate 11. Its requirement for Mg 2+ and Ca 2+ but inhibition by Zn 2+ matches the effects of these ions on apoptosis itself 12. As this enzyme is inhibited by zinc ions at concentrations similar to those found in the normal nucleus, an exchange of zinc for magnesium and calcium within the nucleus during apoptosis may initiate the characteristic chromatin degradation. The role of zinc is further supported by the observation that zinc deficiency leads to excessive apoptosis in the thymus (with resulting profound immunodeficiency) 13 and in the small intestine 14. Interestingly, the presence of zinc can inhibit programmed cell death in the developing Mullerian duct 15. Apoptosis is often dependent upon active metabolism and protein synthesis by the dying cell 1°'16, and although it is tempting to speculate that some of the proteins synthesized during apoptosis may be involved in a putative zinc transport mechanism, they are probably not metallothioneins, the proteins usually associated with heavy metal ion transport 17. Another problem is how apoptotic bodies in the tissue spaces are recognized by tissue macrophages, adjacent epithelial cells or tumour cells and phagocytosed without eliciting any inflammatory response. Mouse peritoneal macrophages bind apoptotic thymocytes in preference to normal ones. This preferential binding of apoptotic cells is blocked in vitro by simple sugars, particularly by N-acetyl glucosamine and its dimer N,N'diacetylchitobiose 18. This suggests that changes in the expression of sugar residues upon the cell surface are recognized by endogenous macrophage 'lectins'. Interestingly, the targets for the presumed macrophage lectin may be sugar residues normally buried within mature glycan chains 18. The= mechanisms whereby epithelial or tumour cells recognize their apoptotic neighbours is unknown. i
Fig.3. Scanning electron micrographs of thymocytes. (a) shows a normal thymocyte with occasionalmicrovilli; (b) is a thymocyte in the early stagesof apoptosis with dumping and shortening of microvilli; whilst (c) shows a cell with the characteristic appearance of swollen vacuolesof endoplasmic reticulum fusing with the cell surface.
Immunology Today, voL 7, No. 4, 1986
Immunologically mediated cell killing How do the deaths provoked by immune mechanisms fit into the classification of death given above? Ultrastructural studies have shown that the lesions induced by complement are tubes of polymerized activated C9 forming channels through the cell membrane 19. It is probable that Ca 2+ influx through these membrane lesions is responsible for the death of the cell rather than simple 'osmotic lysis' as the overall morphology of nucleated cells dying under complement attack is that of necrosis 2,2°. In contrast, when cell killing is mediated by effector cells, the mode of death is apoptosis. The process of killing of cells by specific cytotoxic T lymphocytes (CTL) in vitro has been divided into three stages21: recognition, programming for death* and disintegration. The recognition stage has a half-life of about 1 min, and an absolute requirement for divalent cations. The second stage (programming for death), which has a half-life of about 5 min, is also dependent on divalent cations and only occurs in the. presence of viable CTL. The third 'disintegration' phase, characterized by membrane damage and SlCr release, has a half-life of approximately 1O0 rain, is temperature dependent and does not require functioning CTL cells although it is accelerated in their presence 21,22. During programming for deatlq the CTL inflicts irreversible damage upon the target cell. Initially, as during complement attack 23, cell membrane permeability to 86Rb increases. This continues into the disintegration phase, and probably indicates non-lethal membrane perturbation rather than being the first manifestation of death 21. However, unlike the effects of complement attack, small DNA fragments become detachable from the nucleus within a few minutes 24. Ultimately the majority of the DNA is cleaved into the short polynucleosome chains characteristic of apoptosis. This fragmentation requires Ca 2+ and Mg 2+ and is inhibited by Zn 2+. The event determining DNA breakdown may be the same as that responsible for 'programming for death' as they both have the same half-life 2s. Thus, both the pattern of DNA degradation and its sensitivity to divalent cations are the same as those of steroid-induced apoptosis in thymocytes. However, although an endonuclease with the requisite properties has been found in thymocytes 11, no such endonuclease has yet been found in either effector CTL or their targets (P81 5 mastocytoma cells) 12. Perhaps in CTL killing, nucleases are activated by a rapid mechanism such as phosphorylation only during, or as a consequence of, 'programming for death'. CTL killing and thymocyte apoptosis also differ in that RNA and protein synthesis by the target cell is not required for CTL to kill mastocytoma cells 12, whereas inhibition of macromolecular synthesis abrogates thymocyte apoptosis ~°. There is at present no completely satisfactory explanation combining these results, but they do suggest that apoptosis can be activated at different points in different cell types. The morphological changes in CTL-mediated killing are also typical of apoptosis, both in light microscopy and ultrastructurally 25 ,26 . Moreover, the morphology of apoptosis has been observed in vivo during reactions which are presumed to be mediated by CTL including the *Usuallytermed'programmingfor lysis',but the modeof c~eathisapoptosisand not lysis.
r Vl WSrejection of pig liver allografts, graft versus host reactions, lichen planus*, fixed drug eruptions t, and active chronic hepatitis 27. The morphology of antibody-dependent cell cytotoxicity (K cell attack) is that of apoptosis 21'28, and P815 mastocytoma cells attacked by K cells release radiolabelled molecules in the same sequence as similar cells attacked by CTL, i.e., early release of 86Rb, later release of 51Cr-labelled molecules, and slow release of [12Sl]dU_labelled DNA21. Natural killer (NK) cells also induce apoptosis. Both the programming and disintegration stages have similar cation and temperature requirements to the same stages in CTL killing 29, whilst the transmission and scanning electron microscopic appearances are those of apoptosis 3°,31 . Lastly, lymphotoxins secreted by lymphocytes also induce apoptosis. Mouse fibroblasts on exposure to human lymphotoxins shrink and contract in exactly the same way as fibroblasts killed by CTL32. In addition, after lymphotoxin attack cells released DNA fragments similar to those released after CTL killing 33. Two points relevant to this discussion arise from studies on lymphotoxin induced apoptosis. Firstly, there is no early increase in 86Rb permeability 34 confirming that this change is not an essential preliminary to cell death. Secondly, although there is an increase in synthesis of mRNA during the induction of death, it is not essential, as inhibition of RNA *A skin diseasewhich can be drug or chemical-linked.A local ceil-mediated immuneresponsemaybeinvolvedin the pathogenesis. $ Swellingor lesionson the skinin reactionto someanalgesicsandantibiotics.
Fig, 4. Ultrastructureof apoptosis.A thymicmacrophage(M) has ingestedfour apoptotic thymocytes (A) which are distinguishableby the denselycondensed chromatin of their nucleL Comparethe chromatinpattern with that of adjacent normal thymocytes(N), x 4,600.
117
Immunology Today, voL 7, No. 4, 1986
-review5 synthesis by actinomycin D actually potentiates killing by lymphotoxin 3s. Recent studies have shown interesting parallels between the actions of effector cells and complement. CTL, K and NK cells all produce membrane lesions qualitatively similar in ultrastructure to those produced by complement. Both CTL and NK cells possess granules which contain molecules (polyperforins) from which complexes very similar to those formed by the later components of complement may be assembled and inserted into the membrane of the target cell 36. Why, however, if effector cells and complement produce similar membrane lesions do effector cells induce apoptosis whereas complement attack leads to necrosis? The difference may be determined by the number or distribution of the lesions upon the cell surface. Those produced by an effector cell may be fewer or distributed over a smaller area than those produced in experiments using complement where conditions are usually chosen to maximize the number of 'hits' on the target cell. There are other systems where severity of insult determines the form of death. Mild degrees of hypoxia, toxic assault or heat shock have been shown to lead to apoptosis whilst more severe hypoxia, greater concentrations of toxins or higher temperature lead to necrosis 27. Mild or strictly localized degrees of cell membrane damage might allow fluxes of ions such as Ca2+ and Mg 2+ sufficient to trigger the mechanism leading to apoptosis but insufficient to activate significant amounts of phospholipase. More severe membrane damage might lead to rapid activation of phospholipases, disrupting the function of the cell so severely that the systems leading to endonuclease activation and to apoptosis would be destroyed before their actions could be manifest. Further, the threshold between apoptosis and necrosis might vary between cells of different types and at various stages in the lifespan of a single cell. Cell death and homeostasis in the immune system
In the physiological regulation of lymphoid populations cell death may be essential for deletion of cells with inappropriate specificities, may counterbalance mitosis in lymphoid and accessory cell turnover 37, or may contribute to the contraction of previously expanded populations. Although necrosis is not observed under physiological conditions apoptosis is abundant in thymus s, lymph nodes 27 and spleen 38. In the thymus (where at least 80% of cells die in sZtu39) mitoses outnumber apoptotic cells during enlargement whilst the reverse is true during involution 41. In lymph nodes and spleen the so-called tingible body macrophages (conspicuous histological features of reactive centres) in fact contain the chromatin from large numbers of apoptotic cells. Most of these apoptotic cells are lymphocytes; some are plasma cells38. These endogenous cell deaths appear to be regulated but the mechanism is not understood. By analogy to developing and endocrine-dependent tissues, the deaths could follow production of lethal stimuli or the removal of necessary growth factors. Apoptosis of human lymphocytes has been observed on the withdrawal of support by conditioned media containing interleukin (A.S. Krajewski, unpublished). Conclusions 118
Nucleated, eukaryotic cells die not by a process of disordered dissolution but by one or other of two
different sets of stereotyped reactions; necrosis and apoptosis. Necrosis is associated with major perturbations in the cellular environment and has been regarded as a pathological response, whilst apoptosis has been observed in many instances where the death of the cell is not pathological but appears to be part of homeostatic regulation. However, as discussed above, which of the two reactions occurs may be determined not by any putative 'pathological' or 'physiological' quality of the initiating stimulus but. by its severity or distribution, or by the state of the target cell. The mechanisms of necrosis are relatively well understood. Two molecular events in apoptosis are reasonably defined: the activation of an endonuclease within the dying cell, responsible for the characteristic chromatin cleavage, and the expression of certain sugar residues on the cell membrane that permit recognition and hence phagocytosis of the dying cell. The mechanisms controlling and coordinating these reactions are obscure. Apoptosis is the mode of death induced by effector cells and lymphotoxins and is probably involved in many of the regulatory mechanisms in immunology - the clonal selection of lymphocytes and the regulation of sizes of populations of lymphocytes and cells derived from them. The authors are supported by the Cancer Research Campaign and the Wellcome Trust.
References
1 Trump, B.F., Berezesky,I.K. and Osornio-Vargas,A.R. (1981) in Cell Death in Biology and Pathology, (Bowen, I.D. and Lockshin, R.A. eds) pp. 209-242, Chapman and Hall, London and New York 2 Hawkins, H.K., Ericsson,J.L.E., Biberfeld, P. etal. (1972)Am. J. Pathol. 68, 255 287 3 Jennings, R.B.and Reimer, K.A. (1981)Am. J. Pathol. 102, 214-255 4 Farber,J.L., Chien, K.R. and MJttnacht, S. (1981)Am. J. PathoL 102, 271-281 5 Wyllie, A.H., Kerr, J.F.R.and Currie, A.R. (1980) Int. Rev. CytoL 68, 251-306 6 Anderson, T.J, Ferguson, D.J.P.and Raab, G.M (1982) Br. J. Cancer46, 376-382 7 Russell,J.H. (1983) ImmunoL Rev. 72, 97-118 8 Morris, R.G., Hargreaves,A.D., Duvall, E. etal. (1984)Am. J. Pathol. 115,426-436 9 Wyllie, A.H. and Morris, R.G. (1982) Am. J. Pathol. 109, 78-87 10 Wyllie, A.H., Morris, R.G., Smith, A.L. etal. (1984)J. PathoL 142, 67-77 11 Nakamura, M., Saraki,Y., Watanabe, N. etaL (1981) J. Biochem. (Tokyo)89, 143-152 12 Duke, R.C., Chervenak,R. and Cohen, J.J. (1983) Proc. Nail Acad. Sci. USA 80, 6361-6365 13 Fraker, PJ., Haas,S.M. and Luecke, R.W. (1977)J. Nutr. 107, 1889-1895 14 Elmes,M.E. (1977)J. Pathol. 123,219-224 15 Budzik, G.P, Hutson, J.M., Ikawa, H. etaL (1982) Endocrinology110, 1521 1525 16 Leiberman,MW., Verbin, R.S., Landay, M. etaL (1970) CancerRes. 30, 942-951 17 Maytin, E.V. and Young, D.A. (1983) J. BioL Chem. 258, 12718-12722 18 Duvall, E., Wyllie, A.H. and Morris, R.G. (1985) Immunology 56,351 358
Immunology Today, vol. 7, No. 4, 1986
r vl 19 Tschopp, J., Muller-Eberhard, H.J. and Podack, E.R. (1982) Nature (London) 298, 534-537 20 Goldberg, B. and Green, H. (1959) J. Exp. Med. 109,
505 510 21 Sanderson, C.J. (1981)Biol. Rev. 56, 153-197 22 Martz, E., Burakoff, S.J. and Benacerraf, B. (1974) Proc Natl Acad. Sd. USA 71,177-181
23 Boyle, M.D.P., Ohanian, S.H. and Borsos, T.J. (1976) J. ImmunoL 117, 1346 1350 24 Russell, J.H. and Dubos, C.B. (1980) J. Immunol. 125, 1256 I261 25 Russell, J.H., Masakowski, V., Rucinsky, T. etal. (1982) J. Immunol. 128, 2087 2094 26 Don, M.M., Ablett, G., Bishop, C.J. et aL (1977) Aust. J. Exp. Biol. Med. Sci. 55, 407-417 27 Seafle, J., Kerr, J.F.R. and Bishop, C.J. (1982) PathoL Annu. 17(2), 229-259 28 Stacey, N.H., Bishop, C.J., Halliday, J.W. etal. (1985)J. Cell Sci. 74, 169 179 29 Hiserodt, J.C., Britvan, L.J. and Targan, S.R. (1982) J. ImmunoL 129; 1782 1787
l&-
30 Bishop, C.J. and Whiting, V.A. (1983) Br. J. Cancer48, 441-444 31 Roder, J.C., Kiessling, R., Biberfeld, P. etal. (1978) J. ImmunoL 121, 2509-2517 32 Russell, S.W., Rosenau, W. and Lee, J.G. (1972)Am. J. Patho169, 103-118 33 Ruddle, N.H. (1985) Immunol. Today 6, 156-159 34 Rosenau, W., Goldberg, ML. and Burke, G.C. (1973) J. Immunol. 111, 1128-1135 35 Kunitomi, G., Rosenau, W., Burke, G.C. etaL (1975)Am. J. Pathol. 80, 249-260 36 Podack, E.R. (1985) Immunol. Today 6, 21-27 37 Fossum, S. and Ford, W.L. (1985)Histopathology, 9,469 499 38 Swartzendruber, D.C. and Congdon, C.C. (1963) J. Cell BioL 19, 641-646
39 Joel, D.D., Chanana, A.D., Cottier, H. etal. (1977) Cell Tissue Kinet. 10, 57-69 40 Hinsull, S.M, Bellamy, D. and Franklin, A. (1977) Age and Ageing 6, 72 84
119
