Caspases find a new place to hide

Article metrics

The philosophical muse proclaiming that “death is the essential condition of life”1 has a firm footing in biology, for programmed cell death (apoptosis) is obligatory for normal development of multicellular organisms2. But apoptosis is a double-edged sword, and deregulated cell death is implicated in a growing number of clinical disorders3. So, the apoptotic process needs to be tightly regulated. One way in which this is done is by physically segregating the different components of the apoptotic machinery — only when the death switch is tripped are the tools of execution brought together in the cytosol and the suicide programme activated.

The two main compartments known to be involved in such segregation are the plasma membrane, where both death and survival receptors reside, and the mitochondrion, which is home to several proteins that regulate apoptosis. Now, reporting on page 98 of this issue, Yuan and colleagues4 point to the endoplasmic reticulum as a third subcellular compartment implicated in apoptotic execution. Furthermore, they provide evidence linking activation of a hitherto obscure apoptotic enzyme, caspase-12, to Alzheimer's disease.

Famous mainly for the synthesis and processing of secreted proteins and the storage of intracellular calcium, the endoplasmic reticulum has already provided us with clues as to its alter ego. First, both pro- and anti-apoptotic members of the Bcl-2 family are located in intracellular membranes including the nuclear envelope, the outer mitochondrial membrane and the endoplasmic reticulum5. And second, cytosolic Ca2+ has been implicated as a pro-apoptotic second messenger6 involved in both triggering apoptosis and in regulating death-specific enzymes.

Among the most prominent of these death-specific enzymes is a family of cysteine-dependent aspartate-specific proteases known as the caspases7. These enzymes can be broadly divided into two groups: initiator caspases (such as caspase-8 and caspase-9) whose main function is to activate downstream caspases, and executor caspases (such as caspases-3, -6 and -7), which are responsible for dismantling cellular proteins. The two main apoptotic pathways — the death receptor and mitochondrial pathways — are activated by caspase-8 and caspase-9, respectively, both of which are found in the cytoplasm (Fig. 1). Caspase-8 is recruited to a death-inducing signalling complex only when death receptors such as Fas or the tumour-necrosis factor receptor are oligomerized after binding of specific ligands. In contrast, caspase-9 is activated when cytochrome c is released into the cytoplasm from the space between the inner and outer mitochondrial membranes.

Figure 1: Three distinct apoptotic signalling pathways.

a, When the mitochondrion receives appropriate apoptotic cues, or is irreversibly damaged, pro-apoptotic molecules such as cytochrome c are released into the cytosol. Together with ATP, cytochrome c forms a complex with Apaf-1 and procaspase-9, which is released in an active form. b, Oligomerization of death receptors (by specific death ligands) recruits adaptor molecules involved in activation of caspase-8. The active caspase is formed from two procaspase-8 molecules. c, After stress to the endoplasmic reticulum, including the release of Ca2+ from intracellular stores, caspase-12 is activated. Activated initiator caspases, such as caspase-8 and caspase-9, activate executioner caspases, including caspase-3. Active caspase-3 cleaves the β-amyloid precursor protein (APP), resulting in increased production of amyloid β-peptide (Aβ) which can feed back into caspase-3 activation and execute apoptosis in a caspase-12-dependent manner. The question marks denote possible, but unconfirmed, pathways.

Yuan and co-workers4 now show that another caspase, caspase-12, localizes not to the cytosol but to the endoplasmic reticulum. Caspase-12 is specifically involved in the apoptosis that results from stress in the endoplasmic reticulum. Treatment with compounds such as brefeldin A (which inhibits transport from the endoplasmic reticulum to the Golgi body) or tunicamycin (which inhibits N-glycosylation in the endoplasmic reticulum) triggers activation of caspase-12. But the strongest activation is seen in response to thapsigargin, which disrupts intracellular Ca2+ homeostasis, or to the Ca2+ ionophore A23187. Apoptosis triggered through pathways that do not involve the endoplasmic reticulum, such as serum deprivation or Fas activation, do not result in activation of caspase-12.

To study the localization of caspase-12, Yuan and colleagues used a specific caspase-12 antibody in immunoblotting experiments and detected bands in a fraction of brain extract that also contained an endoplasmic-reticulum-specific protein called TRAPα. Immunocytochemistry then revealed the perinuclear distribution of caspase-12, suggesting that it is situated in either the mitochondria or the endoplasmic reticulum. The co-localization of caspase-12 with several proteins found in the endoplasmic reticulum — grp78, presenilin-2, the β-amyloid precursor protein (APP) and green fluorescent protein targeted to the endoplasmic reticulum — confirmed that caspase-12 resides in the endoplasmic reticulum.

It seems, then, that the main function of caspase-12 is to facilitate apoptosis in cells irreversibly damaged by stress signals from the endoplasmic reticulum. Further evidence of this comes from Yuan and colleagues' studies of transgenic mice that lack intact caspase-12 protein. Thymocytes from these mice die in response to the mitochondria-dependent death signal dexamethasone, and hepatocytes die after injection of Fas-agonist antibodies, indicating that caspase-12 is not required for apoptosis initiated in the mitochondria or plasma membrane. In contrast, apoptosis of the renal tubular epithelium, which is mediated by endoplasmic reticulum stress, depends entirely on caspase-12 activity.

But this is not the end of the story. The endoplasmic reticulum, along with the associated presenilins and APP and the accumulation of Ca2+, are all implicated in Alzheimer's disease8. Apoptosis has also been linked to Alzheimer's disease, with APP identified as a specific substrate for caspase-3 (ref. 9), and Yuan and colleagues' work now supports this idea. They found that primary cortical neurons isolated from their caspase-12-deficient mice are resistant to apoptosis induced by the cleavage product of APP, the amyloid-β peptide (Aβ). To exclude the possibility that the mutant mice lack a sub-population of Aβ-sensitive neurons, the authors used an antisense approach to switch off caspase-12 in normal cells, and found that these cells were also protected against Aβ toxicity.

Putting these data together we can envisage a model in which APP is cleaved by caspase-3, releasing Aβ which, in turn, activates downstream caspases including caspase-3 (ref. 9) and, possibly, caspase-12 to amplify the effects. Although this needs further investigation, it does explain why, despite the high speed of apoptosis, Alzheimer's disease develops slowly over a period of years. Perhaps the presenilins and APP are innocent bystanders in neuronal cells that are triggered to undergo apoptosis after irreparable stress damage. The gradual accumulation of Aβ could eventually reach critical levels, triggering neuronal death through a self-perpetuating circle of APP cleavage, Aβ production and caspase-12-dependent apoptosis.

Like many important insights, Yuan and colleagues' experiments raise more questions than they answer (Fig. 1). For example, what are the physiological regulators of caspase-12? Although increases in the cytosolic concentration of Ca2+ are enough to activate caspase-12, the protein intermediates in this process are not known. What are the downstream targets of caspase-12? Because caspase-3 is activated by Aβ, caspase-12 may be the missing link between APP cleavage, activation of caspase-3 and neuronal apoptosis. Where do Bcl-2 family proteins fit in? Their function is closely linked with caspase activity, and their localization in the endoplasmic reticulum suggests that they are involved in regulating caspase-12-mediated death.

If the identification of caspase-12 in the endoplasmic-reticulum stress pathway follows recent trends, we can expect there to be redundancy in the system. Indeed, the observations4 that such apoptosis is not completely inhibited in embryonic fibroblasts from the caspase-12-deficient mice suggests that there are caspase-12-independent pathways in the endoplasmic reticulum. But what sets caspase-12 apart from other apoptotic proteases is that it really does seem to be restricted to stress responses in the endoplasmic reticulum.

This observation has clear clinical ramifications. Caspases are central to both normal programmed cell death and injury-dependent apoptosis, so any therapy that manipulates caspase activity must take into account the possible effects on tissue homeostasis. In this regard, though, caspase-12 seems to have a strong advantage as a target over other caspases. In contrast to other caspase knockouts10, caspase-12-deficient mice have no noticeable developmental or behavioural defects, and have a normal incidence of tumours. So, caspase-12 is probably not essential in normal developmental death or tumorigenesis. And if activation of caspase-12 does turn out to be confined to only a narrow band of cellular stress signals, it will be a promising potential target for treating neurodegenerative diseases and cancer with few side effects.


  1. 1

    Gilman, C. P. The Living of Charlotte Perkins Gilman: An Autobiography (Appleton-Century, New York, 1935).

  2. 2

    Jacobson, M. D., Weil, M. & Raff, M. C. Cell 88, 347–354 (1997).

  3. 3

    Thompson, C. B. Science 267, 1456–1462 (1995).

  4. 4

    Nakagawa, T. et al. Nature 403, 98–103 (2000).

  5. 5

    Krajewski, S. et al. Cancer Res. 53, 4701–4714 (1993).

  6. 6

    Nicotera, P. & Orrenius, S. Cell Calcium 23, 173–180 (1998).

  7. 7

    Wolf, B. B. & Green, D. R. J. Biol. Chem. 274, 20049–20052 (1999).

  8. 8

    Mattson, M. P., Guo, Q., Furukawa, K. & Pedersen,W. A. J. Neurochem. 70, 1–14 (1998).

  9. 9

    Gervais, F. G. et al. Cell 97, 395–406 (1999).

  10. 10

    Zheng, T. S., Hunot, S., Kuida, K. & Flavell, R. A. Cell Death Differ. 6, 1043–1053 (1999).

Download references

Author information

Correspondence to Huseyin Mehmet.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.