G. P. Amarante-Mendes, Departamento de Imunologia, Instituto de Ciências Biomédicas, Universidade de São Paulo, Av. Prof. Lineu Prestes, 1730 Cidade Universitária, São Paulo 05508-900, Brazil. E-mail: email@example.com
Based primarily on morphological features, apoptosis was described as the cell death that occurs during physiological situations, whereas necrosis was observed during acute harmful conditions. Apoptosis was, therefore, associated with a programme of cell death, as opposed to necrosis, considered an accidental, uncontrolled, pathological cell death. The apoptotic machinery was first unravelled in the nematode Caenorhabditis elegans, where a protease called CED-3 was central to the execution of cells destined to die. Inactivation of ced-3 gene prevents developmental cell death in the worm, an observation that reinforces the notion that apoptosis holds the switch between life and death. In mammals, proteins homologous to CED-3, members of the family collectively called caspases, are considered the executioner proteases responsible for generating fundamentally all aspects of apoptosis. However, inhibition of the so-called executioner caspases (i.e. inhibition of apoptosis) does not prevent cell death to occur. Consequently, in mammals, the decision switch between life and death resides upstream of the activation of caspases and the ensuing apoptotic cell death. Therefore, apoptosis is not a programme of cell death but purely a termination step of a cell death programme, responsible for proper disposal of the already-committed, dying cells.
Apoptosis, as a distinctive morphological process of cell death, was brought to us by John Kerr, Andrew Wyllie and Alastair Currie in 1972 . The term was suggested by James Cormack, a professor of Greek language, and refers to the ‘dropping off’ or ‘falling off’ of petals from flowers, or leaves from trees. Kerr, Wyllie and Currie’s seminal work described in detail the ultrastructural aspects of cell death that occurs during normal tissue turnover, atrophy and involution, but that can also be triggered by noxious agents. From this work and others, we know that apoptosis is morphologically characterized by a reduction in cellular and nuclear volume, chromatin condensation, DNA degradation into oligonucleosomal fragments, preservation of organelle structure and plasma membrane integrity and finally cellular fragmentation generating the so-called apoptotic bodies. Importantly, Kerr, Wyllie and Currie already suggested that apoptosis is an endogenous, phylogenetically conserved programme of cell death that acts in concert – although playing an opposite role – with mitosis to control the size of a tissue or cell population in multicellular organisms.
The definitive evidence that apoptosis is indeed a molecularly controlled biological process came to a great extent from studies with the nematode Caenorhabditis elegans. From the total of 1090 somatic cells that each individual worm produces during its development, 131 cells always undergo an invariable process of programmed cell death. It was Horvitz and colleagues who identified the four crucial genes involved in the control of cell death in C. elegans . Ced-3 is the executioner molecule but it needs to be activated by Ced-4. In healthy cells, Ced-9 inhibits the activation of Ced-3 by Ced-4. During the development of the nematode, Egl-1 expression is activated on those 131 cells. Egl-1 binds and neutralizes Ced-9, thereby allowing Ced-4 to activate Ced-3, which results in the death of those 131 cells . Interestingly enough, preventing apoptosis by interfering with the cell death programme resulted in worms with 131 extra cells.
These and many other results indicate that apoptosis exists to determine whether cells should live or die. For this reason, apoptosis has been long defined as a programme of cell death. But is that always so?
Molecular control of apoptosis
Apoptosis can be initiated by a variety of stimuli through two distinct pathways. The extrinsic pathway is triggered after the engagement of death receptors (DR) present on the cell surface by their specific ligands. In comparison, the intrinsic pathway is initiated by a multitude of intracellular triggers, collectively called ‘stress signals’, such as DNA damage, cytoskeleton disruption, loss of adhesion or growth factor withdrawal, hypoxia, macromolecular synthesis inhibition, endoplasmic reticulum stress, etc., that target the mitochondria and induce the release of pro-apoptotic factors to the cytosol. Downstream of both pathways are the executioner caspases (cysteine–aspartic acid proteases), the enzymes responsible for producing the well-known characteristics of apoptosis .
Caspases are a special family of proteases characterized by a cysteine in the active site and an unusual specificity for aspartic acid residues . Interestingly, not all caspases are involved in apoptosis. Caspase 1, for instance, is an inflammatory protease responsible for the maturation of interleukin-1β. Other inflammatory caspases are 4, 5, 11, 12 and 13 . Structurally, caspase zymogens are single-chain proteins that possess an amino terminus pro-domain, a large subunit in the middle of the molecule, which contains the active site, and a carboxy-terminal small subunit. During apoptosis, the initiator caspases (caspases 2, 8, 9 and 10) are activated by dimerization that occurs in the context of specific multimeric complexes, such as the apoptosome and the death-inducing signalling complex (DISC). Executioner caspases (caspases 3, 6 and 7), which are already expressed as dimers, are activated by cleavage that separates the pro-domain, the large and small subunits. Importantly, the enzymes responsible for in vivo activation of executioner caspases are the initiator caspases and granzyme B, a serine protease present on the granules of cytotoxic cells, such as natural killers and cytotoxic T lymphocytes, and responsible for the induction of apoptosis in target cells [7–9].
Apoptosis initiated via the intrinsic pathway involves a mitochondrial outer membrane permeabilization (MOMP) that is regulated by members of the Bcl-2 family (Fig. 1). Members of this family share one or more Bcl-2 homology (BH) domains. The anti-apoptotic members A1, BCL-2, BCL-XL, MCL-1 and BCL-W contain all four BH1, BH2, BH3 and BH4 domains. The pro-apoptotic members BAX, BAK and BOK do not have the BH4 domain, whereas a special subset of regulatory pro-apoptotic proteins displays only the BH3 domain and, therefore, are known as the BH3-only proteins (BID, BIM, BAD, BIK, BMF, NOXA and PUMA). Interestingly, these proteins are capable of binding each other in a somewhat peculiar pattern. BIM, BID and PUMA can bind all anti-apoptotic members, whereas BAD interacts with BCL-2, BCL-W and BCL-XL and NOXA is rather specific for MCL-1 [10, 11].
The pro-apoptotic proteins BAX and BAK are implicated in the formation of pores on the mitochondrial outer membrane, allowing the release of cytochrome c from the intermembrane space . Cytochrome c activates a multimeric complex, called apoptosome, where initiator caspase 9 proteins are activated and become capable of activating the executioner caspases 3, 6 and 7. Two distinct models have been suggested to explain the regulation of MOMP by BCL-2 members. The indirect activator model proposes that anti-apoptotic BCL-2 proteins are constantly inhibiting an active BAX/BAK channel and apoptosis go on when the BH3-only pro-apoptotic members displace anti-apoptotic BCL-2 proteins. On the other hand, the direct activator/de-repressor model suggests the existence of two subsets of BH3-only proteins, the direct activators and the sensitizers, which should act together to ensure apoptosis to occur. The direct activators would interact and activate BAX and BAK, while the sensitizers neutralize the anti-apoptotic proteins .
The extrinsic pathway is initiated by engagement of one of the DR: tumour necrosis factor (TNF)-R1, CD95/Fas, TNF-related apoptosis-inducing ligand (TRAIL)-R1/DR4 and TRAIL-R2/DR5. DR are transmembrane proteins that belong to the superfamily of the TNF-R and characterized by the presence of an intracellular 80-amino acid-long domain called death domain (DD). DD of the DR interact with the DD of adaptor proteins, such as FADD and TRADD. FADD has another domain called death effector domain (DED) capable of interacting with the initiator caspases 8 or 10 via its own DED, allowing the assembling of the multimeric complex known as DISC (Fig. 2) . In the so-called Fas type I cells, caspase 8 activated at the DISC processes and activates the major executioner caspase 3 to fully induce apoptosis. In contrast, Fas type II cells need an amplification loop mediated by the cleavage of the BH3-only protein BID by caspase 8 . The truncated form of BID translocates to the mitochondria where it induces MOMP and consequent release of cytochrome c [16, 17], apoptosome formation and activation of caspase 9. Caspase 9, in turn, activates the executioner caspases 3, 6 and 7 .
Death is not always an unwanted surprise inflicted to cells but is quite often a decision that they are equipped to make. It is well accepted that the biochemical and morphological changes associated with apoptosis are dependent on the activation of the executioner caspases [5, 18]. Therefore, this event can be considered a decision switch between life and death that cells can set off. This is indeed true in the case of the programmed cell death observed in C. elegans. Mutations that inactivate the worm caspase genes ced-3 prevent the loss of 131 cells destined to die during development [19, 20]. Similarly in mammals, inactivation of APAF-1, caspases 9 or 3 leads to accumulation of extra cells during development, in particular, in the central nervous system [3, 21, 22]. However, we and others have shown early on that in a majority of experimental as well as physiological situations where the intrinsic pathway is set in motion, preventing the activation of caspases, and therefore apoptosis, does not result in increased survival [23–25]. Instead, in such conditions, cells simply undergo a different form of cell death, indicating that the decision switch is not always governed by the apoptotic central machinery and apoptosis is sometimes merely a subprogramme responsible to imprint particular morphological and biochemical characteristics to the cell death that will nevertheless take place (Fig. 3).
In this regard, it is well accepted that the irreversible decision point that controls cell death in mammals is predominantly the MOMP, an event that will ensure cell death even in the absence of caspase activation (and therefore apoptosis) [10, 11, 13, 26]. Besides the release of cytochrome c, activation of caspases and consequent induction of apoptosis (Fig. 3, lilac line), MOMP allows the translocation of other non-apoptotic cell death-inducing molecules from the mitochondria to the cytosol. Importantly, MOMP also triggers the loss of mitochondrial function, the production of reactive oxygen species (ROS) and a significant reduction in ATP levels, setting up a life-incompatible condition (Fig. 3, red line) [11, 26, 27].
Apoptotic and non-apoptotic cell disposal
So, what is the relevance of apoptosis? In multicellular organisms where millions of cells physiologically die every day, the swift and silent elimination of these dying cells is essential to avoid additional tissue damage and to maintain homoeostasis . On the other hand, in response to cell damage/death that follows infection, it is important to rouse inflammatory reactions that can bring about humoral and cellular defence mechanisms able to handle the infective entity and at the same time restore the proper architecture and function of the damaged tissue. In fact, if we take an immunological perspective, we appreciate the recognition of different forms of cell death by the immune cells, which can be silent, tolerogenic or immunogenic .
Phagocytes need signals to distinguish living cells from necrotic or apoptotic cells . Living cells express on their surface molecules such as CD31, CD47 and CD46, which are negative signals for phagocytosis [31–33]. Downregulation or loss of these molecules contribute to reveal the apoptotic cells to macrophages. Perhaps, the most important apoptosis-related ‘eat me’ signal is the translocation of phosphatidylserine (PS) residues from the inner to the outer leaflet of the plasma membrane [34, 35]. Macrophages seem to recognize PS via its cell surface receptor Tim4 (T-cell immunoglobulin domains and mucin domain) , although other PS receptors may exist. Another ‘eat me’ signal is Ptx3 (pentraxin 3). During neutrophil apoptosis, endogenous Ptx3 translocates to plasma membrane where it can interact with Fc-gamma receptor on macrophages to mediate phagocytosis [30, 37]. In addition, macrophages can also interact with apoptotic cells via certain opsonins, such as Gas6 (growth arrest-specific 6), thrombospondin-1, protein S, milk fat globule-epidermal growth factor-8 (Mfge8), mannose-binding lectin and C1q [30, 38, 39].
Importantly, the recognition and uptake of apoptotic cells contribute to the maintenance of tissue homoeostasis by stimulating the production of anti-inflammatory substances and inhibiting the release of pro-inflammatory cytokines. For instance, recognition of apoptotic cells by macrophages causes the release of IL-10, TGF-β and prostaglandin E2, which negatively modulate inflammation [40, 41]. Furthermore, apoptotic cells can actively inhibit the production of the pro-inflammatory cytokines by macrophages and dendritic cells, such as IL-1β, IL-8, TNF-α and leukotriene B4 . Finally, it has been shown that apoptotic cells themselves can release IL-10 and TGF-β immediately following caspase activation, contributing to its strong anti-inflammatory response [43, 44].
In contrast, interactions between necrotic cell and phagocytes result in a strong pro-inflammatory reaction, possibly mediated by the activation of nuclear factor-kappa-light-chain-enhancer of activated B cells (NF-κB) in macrophages in a toll-like receptor (TLR)-dependent way . This is likely to be due to the leakage of certain intracellular components during necrosis, which does not occur during apoptosis, because the maintenance of membrane integrity until very late stages of cell death is one important aspect of apoptosis. These pro-inflammatory components are collectively called alarmins or damage-associated molecular pattern (DAMP) [29, 30]. One example is the nuclear DNA-binding protein HMGB1 (high-mobility group box 1) , which interacts with TLR2, TLR4 and TLR9, resulting in the upregulation of pro-inflammatory and antigen-presenting properties of macrophages and DC [47, 48, 29]. Interestingly, HMGB1 released from cells that underwent apoptosis before necrosis (secondary necrosis) has its stimulatory activity inactivated by a caspase-dependent activation of ROS . Indeed, blocking the oxidation sites in HMGB1 prevented the induction of tolerance by apoptotic cells.
Taking together, one can consider apoptosis an efficient programme of cell disposal (rather than cell death), because the caspase-dependent cellular and molecular modifications, such as the appearance of ‘eat me signals’, DNA degradation, and inactivation of DAMP, create a tolerogenic environment essential to avoid autoimmune diseases. Deficient execution of apoptosis, or inefficient engulfment and digestion of apoptotic cells may produce autoimmunity, as we will discuss next.
Cell death/disposal and autoimmunity
A number of autoimmune diseases are caused by problems in the regulation of cell disposal. For instance, if instead of undergoing an apoptotic death, cells die by necrosis and therefore release DAMP and other intracellular contents, the result is likely to be the activation of autoreactive T- and B lymphocytes. In addition, malfunction in a single crucial step of apoptosis can also impair the tolerogenic potential of apoptosis. This is illustrated in caspase-activated DNAse (CAD)-deficient mice. CAD is the endonuclease responsible for the oligonucleosomal DNA degradation observed in apoptosis. Effector caspases cleave the inhibitor of CAD (iCAD), releasing CAD to translocate to the nucleus, where it performs the oligonucleosomal DNA cleavage. Although CAD-deficient cells undergo caspase-mediated cell death, they present defective DNA fragmentation, chromatin condensation and apoptotic body formation . Importantly, CAD null mice exhibit an increased susceptibility to a systemic lupus erythematosus (SLE)-like disease, with high levels of autoantibodies .
Systemic lupus erythematosus is a chronic autoimmune disease associated with several clinical manifestations affecting skin, kidney and blood vessels . Patients with SLE present autoantibodies against intracellular molecules such as DNA, RNA and nuclear proteins . SLE is the most well-studied autoimmunity associated with nucleic acid-related autoantigens, but other related diseases have also been described, such as systemic sclerosis (scleroderma), mixed connective tissue disease, poly-dermatomyositis and primary Sjogren’s syndrome . Nucleic acid-associated antigens are the most frequent autoantigens in autoimmunity. Although it is still unclear, there are increasing evidence that nucleic acids may act as co-stimulators of macrophages and dendritic cells via TLR stimulation, which results in breaking down apoptosis-induced tolerance . In this situation, T cells can be stimulated and cooperate with autoreactive B cells to produce antibodies against self-antigens .
Autoimmune response can also occur when the rate of apoptosis is excessive and the clearance of apoptotic cells or bodies is not efficient enough. C1q is a classical complement system protein that also works as an opsonin for dying cells. C1q is a PS-interacting protein that can bind apoptotic cells, thereby mediating a quick clearance of the dying cells . Importantly, in both humans and mice, deficiency in C1q is associated with the appearance of SLE or SLE-like phenotype . Another example is Mfge8, a protein capable of bridging PS on apoptotic cell and integrins on phagocytes. The absence of Mfge8 impairs the uptake of apoptotic cells, which undergo secondary necrosis and release their intracellular content, activating autoreactive B cells . Indeed, Mfge8-null female mice suffer from a SLE-type disease with the production of anti-DNA and anti-phospholipid antibodies [57, 58].
Finally, after recognition and uptake of apoptotic cells, phagocytes need to properly digest the residual apoptotic bodies. In fact, even when apoptosis is fully processed, its tolerogenic property may be lost if the very final event – digestion – do not occur properly. Although CAD seems to be a very important endonuclease, acting on the apoptotic cell itself, breaking down the DNA into oligonucleosomal fragments, DNAse II is also critical for the digestion of apoptotic cell DNA. DNAse II takes action inside the phagocytes, where it further digests apoptotic cell DNA inside the phagocytic lysosomal organelles. Deficiency in DNAse II results in the accumulation of DNA in lysosomes, with consequent initiation of phagocyte pro-inflammatory activity, likely to contribute to immune response against autoantigens .
Apoptosis may have appeared during evolution as a molecular switch that controls some forms of developmental cell death, as it is clearly the case of the nematode C. elegans, for instance. However, much more than a switch that triggers cell death, apoptosis seems to be a programme that coordinates biochemical and morphological changes that prevents unnecessary and potentially harmful responses, thereby maintaining tissue homoeostasis. Executioner caspases orchestrate the exposure of ‘eat me’ signals and consequent induction of anti-inflammatory cytokines by phagocytes; the DNA degradation and the formation of apoptotic bodies; and the conversion of potential pro-inflammatory proteins into tolerogenic molecules.
The work in the Amarante-Mendes laboratory is supported by grants from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP-Brazil), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES-Brazil) and the Brazilian Research Council (CNPq-Brazil).