An endonuclease at last

Article metrics

Cell death is seldom chaotic. Rather, it is a subtly orchestrated disassembly of critical molecular structures, permitting the doomed cells to vanish with minimum disruption to the surrounding tissue. This idea was suggested by the uniform sequence of structural changes observed in cells dying in many different circumstances — the process of apoptosis1. That such death might be effected by a dedicated set of gene products became clear when developmental cell death in the nematode Caenorhabditis elegans was shown to require the activity of several genes for which, to this day, no other major function has been identified2.

On pages 43 and 96 of this issue, two papers from Shigekazu Nagata's group describe a mechanism for disassembly during apoptosis of the most complex of all biological molecules, chromatin3,4. Despite earlier premonitions to the contrary, they find that chromatin is cleaved by a nuclease that is activated exclusively in apoptosis. And it is already clear that at least some elements of this mechanism are highly conserved between species.

Cleavage of chromatin at internucleosomal sites was first reported in dying cells by the Czech radiobiologist Miroslav Skalka5. Soon after, the ‘chromatin ladder’ pattern of regularly sized DNA fragments was recognized as characteristic of apoptosis6. Several candidates for the nucleases responsible were reported, but their properties were diverse enough to preclude identity with each other. Indeed, the view was expressed that DNA cleavage, appearing after the irreversible commitment to death, might be effected by recruitment of any available nuclease, rather than requiring an apoptosis-specific enzyme.

Meanwhile, another family of effector molecules attracted more attention — the cysteine-containing aspartate-specific proteases, or caspases. The ced3 gene product, which is necessary for all developmentally regulated cell death in the nematode, is the prototype of a large family of mammalian caspases at least some of which act in an autocatalytic cascade. Caspases cut their substrate proteins at tetrapeptide sites with characteristic motifs7. Caspase 3, for example, recognizes DEVD sequences and, during apoptosis, it cleaves and inactivates several significant cellular proteins in the cytosol, nucleus and cytoskeleton. Caspase 3 itself is activated by signals initiated by lethal stimuli of many different types and is thus a key player in turning on the pleiotropic effector events of apoptosis.

Working with cytosolic extracts, Nagata's group had already identified a factor that can cleave chromatin at internucleosomal sites8. Now they show that this activity depends on two interacting molecules3. The first is a hitherto unknown nuclease (Mr, 40,000) — a 343-amino-acid, basic protein with a tell-tale nuclear localization signal at the carboxy terminus. The nuclease is expressed in many types of cell, but it is usually located in the cytoplasm in an inactive, latent form. In fractionated cytoplasmic extracts, it can be activated by caspase-3 digestion, earning it the name caspase-activated DNase (CAD).

The second molecule of the partnership is a strongly acidic protein (Mr, 29,000-30,000; there are two isoforms of 265 and 331 amino acids). This protein binds to, stabilizes and inactivates CAD, so it is called inhibitor of CAD (ICAD). ICAD releases CAD after caspase-3 digestion, and it is ICAD, not CAD, that possesses caspase cleavage sites. Using site-directed mutagenesis, Sakahira et al.4 show that one of these cleavage sites is essential for CAD release and activation. The authors also find that hyperexpression of ICAD in human Jurkat cells (a well-established test system for apoptotic stimuli) blocks the chromatin changes of apoptosis. Significantly, however, ICAD hyperexpression does not abrogate cell death, as shown by endogenous activation of caspase-3 activity, the characteristic surface-membrane expression of phosphatidylserine and eventual loss of mitochondrial oxidative function.

In these experiments, CAD and ICAD were purified from mouse cells, but the sequence of murine ICAD is closely homologous to a human protein, DFF45, discovered last year9 which, like ICAD, contains caspase 3 sites and is cleaved in apoptosis. Cleavage of DFF45 also permits nuclear endonuclease activation. The new data thus unequivocally show an apoptosis-specific pathway, initiated by caspase cleavage, which terminates in activation of a nuclease responsible for internucleosomal digestion of DNA. Interestingly, although the authors do not comment on this, their gels also show that active CAD cleaves chromatin to generate transient large fragments of uniform size before the appearance of the ‘nucleosomal ladder’: previous studies had indicated that the enzymes effecting these two functions might be different.

Why should there be a mechanism that specifically destroys DNA during apoptosis, and what pressures promote its development and expression? In apoptosis, caspase cleavage of DNA protein kinase and poly(ADP-ribose) polymerase unhooks DNA repair from DNA damage10,11. Lamin cleavage unpins the nuclear envelope12. Cleavage of gelsolin, actin and intermediate filaments probably effect the hugely altered cell shape and movement of apoptosis13,15. So what does CAD do, and why should it be there?

There is little evidence that apoptosis-associated nuclease activity is sequence specific. But it can be argued that unrestricted DNA cleavage is essential for the successful completion of apoptosis, because unpackaged DNA from even one cell, uncoiling over some 1.5 m, could create an unmanageable extracellular environment for its neighbours. Further, DNA is dangerous stuff. Bacteria have restriction nucleases that destroy undesirable alien genomes, but a somewhat different strategy is needed to cope with extraneous DNA in eukaryotic tissues. The enormous size of the eukaryotic genome, and even the relative complexity of most viral DNA, make it improbable that cutting-site motifs could be found that satisfactorily discriminate viral from host DNA. Moreover, it is part of the strategy of apoptosis that the entire nuclear DNA of the dying cell finds itself within neighbouring viable cells, and, in theory, this could colonize and scramble their genomes. What is needed is a mechanism that destroys DNA effectively (host as well as viral) in the process of removing virally infected or otherwise unwanted cells: in other words an endonuclease restricted to the process of apoptosis but not to any particular target sequence. And this is precisely what the CAD/ICAD pathway provides. The strategy of maintaining metazoan life seems to involve investment in a superbly designed, highly efficient and tightly controlled removal kit that is competent not only to kill unwanted cells on cue, but to bury the evidence, and fast.


  1. 1

    Kerr, J. F. R.et al. Br. J. Cancer 26, 239–257 (1972).

  2. 2

    Horvitz, H. R., Ellis, H. M. & Sternberg, P. N. Neurosci. Comm. 1, 56–65 (1982).

  3. 3

    Enari, M.et al. Nature 391, 43–50 (1998).

  4. 4

    Sakahira, H., Enari, M. & Nagata, S. Nature 391, 96–99 (1998).

  5. 5

    Skalka, M.et al. FEBS Lett. 72, 271–274 (1976).

  6. 6

    Wyllie, A. H.et al. Nature 284, 555–556 (1980).

  7. 7

    Thornberry, N. A.et al. J. Biol. Chem. 272, 17907–17911 (1997).

  8. 8

    Enari, M.et al. Nature 380, 723–726 (1996).

  9. 9

    Liu, X.et al. Cell 89, 175–184 (1997).

  10. 10

    Kaufmann, S. H.et al. Cancer Res. 53, 3976–3985 (1993).

  11. 11

    Casciola-Rosen, L.et al. J. Exp. Med. 183, 1957–1964 (1996).

  12. 12

    Takahashi, A.et al. Proc. Natl Acad. Sci. USA 93, 8395–8400 (1996).

  13. 13

    Kothakota, S.et al. Science 278, 294–298 (1997).

  14. 14

    Caulin, C.et al. J. Cell Biol. 138, 1379–1394 (1997).

  15. 15

    Na, S.et al. J. Biol. Chem. 271, 11209–11213 (1996).

Download references

Author information

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.