Compared with the life of a multicellular organism, cells are cheap. Regulated cell death - apoptosis - is used for a variety of processes by such organisms, from the sculpting of limbs during development to the destruction of infected cells to prevent the spread of infection. It is clearly an extremely important process and yet it is relatively poorly understood; only in the last decade or so has significant progress been made in understanding the signalling cascades involved, and even now many of the details remain unclear. It does seem, however, that the default state of many cells is suicidal, and that this fate is only averted by the detection of extracellular 'survival signals'.

As our knowledge increases, however, it seems that our understanding of the scale and influence of apoptosis can only increase. One particularly interesting speculation is that caspases - the family of proteases responsible for cell execution in apoptosis - may have several other important functions, including a role in the regulation of cell-cycle progression. In this essay, I will outline the basic mechanisms of apoptosis, before considering how it is or may be involved in the larger regulation of the cell life cycle.

Apoptosis: Step-by-Step Death

The term 'apoptosis' was originally coined to describe the morphological characteristics of a certain type of cell death, in contrast to the uncontrolled process of necrosis. In the latter, cells swell and burst, spilling their contents across neighbouring cells with the potential to induce a damaging inflammatory response; in the former, cells shrink and their contents condense, leading to membrane blebbing. Their nuclear contents also condense, and DNA is degraded to give rise to a 'ladder' structure. Cells undergoing apoptosis identify themselves to their neighbours, most noticeably by exposing phosphatidylserine (PS) on the outer leaflet of the plasma membrane, when normally it is maintained in the inner leaflet. This acts as a signal for neighbouring cells to remove the remnants of the dying cell by phagocytosis to be recycled.

In mammalian cells, there are two main pathways for apoptosis: Signalling via mitochondria, and signalling via 'death-receptors' such as CD95 or Fas (figure one). There is a certain amount of cross-talk between the pathways, and both result in the destruction of the same apoptotic substrates. They are both discussed below; however, first it would be prudent to introduce the major players in apoptosis: Caspases.

Caspases are so named because all known examples possess an active-site cysteine residue, and cleave protein substrates after aspartic acid residues. This can have two main effects, as shown in figure two; it may lead to loss of function, or it may lead to gain of function. Over a dozen caspases are known in mammals, and each has distinct effects. A given caspase's specificity is determined by the four residues amino-terminal of the cleavage site, and such cleavage has been used to explain many of the characteristic features of apoptotic cells. For example, caspase-3 activates CAD (caspase-activated DNase) by cleavage of an inhibitory subunit (ICAD). CAD subsequently acts on genomic DNA to produce the characteristic DNA 'ladder' gel electrophoresis pattern by cleaving between nucleosomes to produce multiple fragments that are all integer multiples of ~180bp in length. Caspases also act on such cytoskeletal proteins as fodrin and gelsolin, helping to explain the shrivelling and blebbing seen in apoptotic cells.

Caspase Activation

Caspases themselves are synthesised (like most proteases) as inert zymogens. All contain three domains; an N-terminal prodomain, a p20 domain, and a p10 domain. Three mechanisms of activation have been described so far: Processing by an upstream caspase, induced proximity to other procaspases, and association with a regulatory subunit (figure three). The type of activation applied to a given caspase depends on the apoptotic pathway of which it is a part, and the class to which the caspase belongs.

Members of this family have been placed in one of two classes on the basis of the size of their prodomain. Type I 'initiator' caspases contain an extended prodomain that is capable of serving as an interaction domain for assembly with other proteins. Caspase-9 is an example of this type of caspase. By contrast, type II 'effector' caspases have short prodomains and are in general activated by cleavage by upstream caspases. Caspase-3 is a good example of a type II caspase. This type of activation leads to the concept of 'caspase cascades' - the sequential activation of a series of caspases - which are extremely important in the apoptotic response.

The Death Receptor Pathway

One important role of apoptosis is in the functioning of the immune system. It is involved in the development of lymphocyte repertoires, ensuring that self-recognising cells are destroyed before they can become activated, and is also a pathway by which activated lymphocytes induce the destruction of infected cells.

Lymphocytes can activate the apoptotic process in two ways. Firstly, they secrete a range of proteins on to the surface of the infected cell. One of these, perforin, can assemble to form transmembrane channels that allow other proteins to enter the cell; one of these is the protease granzyme B, which cleaves and activates specific procaspases to initiate a caspase cascade.

The second mechanism is probably the more significant, and is the one outlined in figure one. Lymphocytes express proteins such as Fas ligand and CD95 ligand on their cell surface. These can interact with specific receptors on target cells, causing the receptors to aggregate (this is trimerisation in the case of the Fas/FasL interaction). The aggregated receptors recruit adaptor proteins such as FADD (Fas-associated death-domain protein) to form what has been termed a DISC (death-induced signalling complex), which in turn recruits procaspase-8 molecules.

Several types of domain mediate interactions between proteins in these complexes. Caspase-8 is a type I caspase; its prodomain contains a 'death-effector domain (DED). By contrast, caspase-9, which forms a part of the so-called apoptosome, contains a 'caspase-activation and recruitment domain' (CARD). What is striking about these domains is that despite showing very little sequence homology, their tertiary structures are extremely similar, consisting of six antiparallel a-helices arranged in a Greek key formation. Upstream regulators such as CD95 and Fas contain a third type of interaction domain, known as a 'death domain' (DD); it has a similar structure to the other two. It seems likely that all three are derived from a common evolutionary ancestor.

Despite being uncleaved, the recruited procaspase-8 molecules have a very low intrinsic activity; holding them together for any length of time will result in activation by proteolytic cleavage (figure three b). This does seem to be a rather crude mechanism by which to control the fate of a cell, however; it seems likely that there are additional levels of regulation in vivo to modulate the process.

Apoptotic Signalling Via Mitochondria

It is becoming very clear that the mitochondrion is not just the powerhouse of the cell - it is also its arsenal. Mitochondria sequester a potent cocktail of pro-apoptotic proteins, of which it appears that the most important is the humble electron carrier cytochrome c, which is required for activation of caspase-9 in the cytosol (figure one). Numerous trigger signals for apoptosis can induce the release of these proteins from the intermembrane space, including DNA damage, direct damage to the mitochondrion by toxic drugs and (in some cases) signalling via death receptors.

Mitochondria undergo a number of morphological changes in apoptotic cells. They are normally dispersed throughout the cell; however, one early event that is triggered at least by tumour necrosis factor (TNF) is clustering of mitochondria around the nucleus. This may result from defective kinesin-mediated transport of the organelle. It is not clear whether this movement plays a role in the release of mitochondrial proteins.

In fact, it is not clear how mitochondrial proteins are released at all. What is known is that there are a number of proteins involved in the regulation of this process, most notably members of the Bcl-2 family. There are two theories with respect to the molecular mechanism of cytochrome c release; the non-specific rupture of the outer mitochondrial membrane, and the formation of cytochrome c conducting channels (figure four). Briefly, the former could result from hyperpolarisation caused by the closure of the voltage-dependent anion channel (VDAC), or from the opening of an unselective mitochondrial megachannel called the 'permeability transition pore'. Both of these would be predicted to lead to osmotic swelling of the matrix; because the surface area of the inner mitochondrial membrane (IMM) is significantly greater than that of the outer mitochondrial membrane (OMM), this will result in rupture of the OMM. Cytochrome c-specific channels could be formed in three ways; by the Bcl-family protein Bax alone, by Bax in association with VDAC, or by the formation of a lipidic pore resulting from destabilisation of a region of the OMM by Bax.

The jury is still out as to which of these mechanisms is correct; the channel theory has the advantage of maintaining mitochondrial function during the time necessary for the activation of caspases, which is important because caspase activation is ATP-dependent. However, there is no evidence that Bcl-family proteins can form channels in vivo; it has only been demonstrated in artificial membranes. Even if such channels can form, it is by no means certain that they would be large enough to allow the translocation of cytochrome c to the cytosol.

In any case, once release has happened, events proceed rapidly - although not necessarily inevitably - to cell death. Cytochrome c associates with apoptotic protease activating factor-1 (Apaf-1) and procaspase-9 to form the apoptosome, which can act on caspase-3 to activate it. This leads to subsequent degradation of apoptotic targets.

Regulation Of Apoptosis

Unsurprisingly, cell death is an extremely well-regulated process. There are numerous steps that either shut apoptosis down completely, or serve to accelerate it, with the result that when the conditions are correct, apoptosis is rapid and efficient, but if there is any doubt then total degradation is prevented.

There are several families implicated in the regulation of apoptosis. Chief amongst these is the Bcl-2 family of proteins, of which around twenty members have been identified in mammals so far. There are three functional groups of Bcl-2 proteins (figure five), depending on their domain structure. Members of the first group are characterised by four short, conserved, Bcl-2 homology (BH) domains, and a single transmembrane domain that localises the proteins to the outer surface of mitochondria (and occasionally the ER). All members of this group are anti-apoptotic proteins, and act to prevent cell death. Group II proteins, on the other hand, all possess pro-apoptotic activity, and all lack the N-terminal BH4 domain. Group III is large and diverse; the only common features of proteins in this group are the presence of the BH3 domain, and the fact that they also have pro-apoptotic activity. Many examples of this domain have only very low sequence homology, suggesting that it has arisen through convergent evolution.

The roles of some Bcl-2 family proteins are illustrated in figure one; however, as a generalisation, it seems that these proteins spend most of their time simply blocking each other's moves. A large number of family members can homodimerise, but more importantly, many pro- and anti-apoptotic members can heterodimerise. When this occurs, the net result is mutual inactivation; as a result, the ratio of pro- to anti-apoptotic proteins is extremely important in determining the fate of a cell.

Bcl-2 proteins do have other roles, as well; Bax, for example, is strongly implicated in the release of cytochrome c from mitochondria, as described above, and several other Bcl proteins have inhibitory effects on this process (figure one). In C.elegans, the Bcl-homologue CED-9 acts as an apoptosis suppressor in two ways: By binding directly to the CED-3 caspase and inhibiting it, and by binding to CED-4 and inhibiting its ability to activate the CED-3 procaspase. However, no mammalian Bcl-2 proteins have been observed to operate by this mechanism.

Mammals also carry a family of genes encoding inhibitor-of-apoptosis proteins (IAPs). Whereas Bcl-2 proteins can block the mitochondrial branch of apoptosis by preventing the release of cytochrome c, IAPs block both the mitochondrion and receptor-mediated pathways by binding directly to and inhibiting both initiator and effector caspases. Recently, the structural basis of this inhibition has been determined for the IAP family member XIAP and caspase-7. The crystal structure of caspase-7 complexed with the inhibitory region of XIAP shows that the peptide binds the catalytic groove of the caspase, completely filling the active site and thus blocking substrate entry. Binding was expected to be at the BIR2 (baculoviral IAP repeat, named for its homology to baculoviral IAPs) domain, but instead it is at a preceding short linker region. The BIR2 domain is thus proposed to stabilise the interaction between the caspase and XIAP. The interaction of the IAP inhibitor Smac/DIABLO with XIAP is proposed to destabilise this interaction and liberate the caspase activity (figure one).

Extracellular Signals

Apoptosis is also regulated by extracellular controls, in the form of signals from other cells. A classic example of this is the apoptosis that causes the tadpole's tail to disappear, which is triggered by a surge of thyroid hormone in the bloodstream. Other normal cell deaths occur because the cells fail to receive enough apoptosis suppressing 'survival signals'. Most of our cells seem to require such signals; if a cell is experimentally isolated from its neighbours, it will undergo apoptosis unless it is supplied with the appropriate signals. This is thought to ensure that cells survive only when and where they are needed. The serine/threonine kinase PKB has been identified as an important component of survival signal transduction (figure six).

PKB is activated via the PI3K signalling pathway. It is known to be overexpressed in a number of tumours; it has also been shown to phosphorylate Bad, a pro-apoptotic Bcl-2 protein, preventing Bad from binding and inhibiting Bcl-Xl, and caspase-9, inhibiting it. Both of these reduce the level of apoptosis. As shown in figure six, PKB can also influence the level of transcription by phosphorylating IkB and releasing NF-kB to translocate to the nucleus, where it activates transcription of anti-apoptotic genes. Crucially, PI3K is known to be sensitive to survival signals; these can include such hormones as insulin. They activate it, which in turn results in the activation of PKB. In the absence of survival signals, PKB becomes less active, and apoptosis becomes less strongly suppressed.

Caspases And The Cell Cycle

Somewhat surprisingly, there is a growing body of evidence linking caspases (and, indeed, other apoptosis regulators) to control of the cell cycle (figure seven). To ensure genomic stability, a series of checkpoints exist to disrupt cell-cycle progression when damage to the genome is detected. Apoptosis can be viewed as the ultimate checkpoint, to be used to eliminate cells with irreparable DNA damage; it is also induced when cell division has become unregulated.

Evidence linking caspases to non-apoptotic roles in development and differentiation has come from studies on knockout mice. For example, caspase-8-negative and FADD-negative mice show impaired heart muscle development. By contrast, expression of a dominant-negative FADD can lead to cell proliferation. In addition, FADD has recently been shown to become phosphorylated by an unknown serine protein kinase during the G2/M transition. Also, where signalling through death receptors alone triggers apoptosis, signalling through T-cell receptors (TCRs) and death receptors together has been observed to trigger cell survival and proliferation.

Together, this evidence can be used to support the model shown in figure seven, in which caspase activation via FADD is necessary either to prototypically remove a block of entry into mitosis (a hypothetical mitosis entry blocker, MEB), or to activate other cell-cycle regulatory proteins. The cell-cycle regulatory kinases CDK2 and CDC27 are known to be caspase-8 targets. This can be explained evolutionarily; if a cell were about to stop functioning anyway following caspase activation, there would be little selection against the presence of fortuitous caspase cleavage sites on irrelevant proteins. However, the fact that caspase-8 has been observed to be active in some non-apoptotic cells suggests that this explanation is, at the least, incomplete.

Critically, however, it is not clear how caspase cleavage could be restricted to cell-cycle regulators alone, leaving other vital proteins intact. One possible answer is the specific subcellular compartmentalisation of caspases; in some cell types, certain caspases have been observed to translocate to organelles - including the nucleus - upon activation.

Other components of the apoptotic machinery may be linked to cell-cycle control. For example, Bcl-2 has been observed to delay the re-entry of resting cells into the cell cycle. However, as all mutations that suppress the anti-apoptotic effect of Bcl-2 also abolish this effect on cell cycle, the two activities may not be independent. Meanwhile, survivin, an IAP, is specifically induced in the G2/M phase and associates with microtubules of the spindle apparatus at the start of mitosis. It was believed to bind to and inhibit caspase-3, but following the discovery of survivin homologues in yeast and C.elegans, it is not certain that this protein is involved in apoptosis at all; in such organisms, they are primarily required for efficient mitotic cell division. It may be that there are two classes of IAP: Evolutionarily ancient forms that control mitotic division, of which survivin may be one, and more recent forms such as XIAP that do control caspase activation, but have no influence on cell-cycle control. Alternatively, it may be that survivin represses a default apoptotic pathway that is required for a certain step of the cell cycle, or spindle assembly in mammals.

Conclusions

Apoptosis can be stimulated in three main ways: By cell damage, in particular by damage to the cell DNA or to the mitochondria; by extracellular molecules, or by interactions with other cells - with the most important examples of this taking place in the immune system - or by the absence of survival signals normally transduced through such proteins as PKB. However, it appears that it is not independent of normal cell function. Rather, it is intertwined with such processes as DNA replication, cell cycle regulation and respiration at multiple levels, a fact that becomes increasingly clear when its potential involvement in disease - for example, neurodegeneration in Parkinson's disease - is considered. It is a safeguard against cancer, and an extremely useful tool for the sculpting of organs and tissues during development. Overall, the precarious balance between life and death endured by individual cells is a small price to pay for the benefits and privileges that multicellularity brings with it.