Protective packaging for DNA

Article metrics

Histone proteins are best known for their structural role in packaging DNA into a compact form. But it seems that one such protein also works as a tumour suppressor, helping to prevent cancer developing.

Inside every human cell there is an impressive two metres or so of DNA, which must be organized into a compact form to fit into the cell nucleus. To do this, cells begin by wrapping the DNA around histone proteins. But one of these proteins, histone H2AX, has a dual role — it also helps cells to deal with potentially dangerous breaks that can occur in the DNA double helix. Writing last week in Cell, Celeste et al.1 and Bassing et al.2 highlighted the importance of this function: they showed that it helps to prevent genome instability and cancer.

The compact form of DNA in the nucleus is known as chromatin. The basic unit of chromatin, the nucleosome, is composed of DNA wrapped around two copies each of histone proteins H2A, H2B, H3 and H4. The H2AX variant makes up about 10–15% of total cellular H2A in higher organisms such as humans, frogs and mice, and is incorporated randomly into nucleosomes throughout the genome. It differs from H2A primarily in that it has an extended tail at one end — the carboxy terminus. Following DNA damage, this tail rapidly becomes labelled with phosphate groups, generating a species called γ-H2AX3. In lower organisms such as yeast, the carboxy-terminal tail of H2A itself contains a similar motif to the H2AX tail, and becomes phosphorylated when DNA is damaged.

Although it is unclear exactly what γ-H2AX does following DNA damage, microscopy studies have shown that it is generated in the chromatin flanking a DNA double-strand break, and that mammalian repair and signalling proteins are recruited to these sites, ultimately amassing in large numbers. These visible protein accumulations, which can span millions of bases of DNA, are known as foci. γ-H2AX is not required for the initial recruitment of repair factors, but is needed for later foci formation4.

Wherever possible, the cell's aim is to repair the DNA break, which can be done in various ways — two of the most common methods are non-homologous end-joining (NHEJ) and homologous recombination. Previous work5 indicates that mutating the phosphorylation site on the carboxy terminus of yeast H2A impairs, but does not prevent, the repair of double-strand breaks by NHEJ. No obvious NHEJ defect has been observed in mouse cells lacking H2AX, although their mechanism for repairing double-stranded breaks by homologous recombination seems to be defective6,7. H2AX is also required following low (but not high) levels of DNA damage in order to trigger the 'checkpoint' that halts cell division to allow time for DNA repair8. The new papers1,2 help us to integrate these seemingly disparate findings into a more coherent model (Fig. 1).

Figure 1: Model of the response to DNA double-strand breaks in mammalian cells, incorporating the new results1,2.

a, Double-strand breaks are produced by, for instance, ionizing radiation, or during the rearrangement of immune-receptor genes. b, The lesion is detected and various kinase enzymes (ATM, ATR and DNA-PK) are activated, in turn phosphorylating histone H2AX protein. ATM, ATR and DNA-PK also transduce the signal to downstream proteins (not shown), leading to the activation of checkpoints that prevent cell division when levels of DNA damage are high, and the production of various repair and signalling factors, which are recruited to the lesion (c). d, Further factors are recruited, aided by the phosphorylated H2AX, and perhaps involving intermediary proteins. H2AX may therefore enhance accurate, efficient repair, or prevent inaccurate or inefficient repair. H2AX phosphorylation may also lead to changes in chromatin, or facilitate a checkpoint that prevents cell division when there is little DNA damage.

Because defects in the detection and repair of double-strand breaks can cause cancer, Celeste et al.1 and Bassing et al.2 wanted to see whether H2AX functions as a tumour suppressor — a protein that hampers tumour development. To do this, they generated mice in which either one or both copies of the H2AX gene were inactivated, either alone or in combination with inactivation of the tumour suppressor p53. Strikingly, although mutations in H2AX alone engendered little or no increase in cancer development, they dramatically enhanced tumour formation when combined with p53 deficiency.

The authors also found evidence of genome instability in the mice that lacked both H2AX and p53. Analysis of cancers arising from B cells, a type of immune cell, revealed that segments of DNA had been moved (translocated) from one chromosome to another. The regions involved indicated that the translocations were failed rearrangements of immunoglobulin genes. These types of rearrangements generate receptor diversity among both B and T cells in the immune system, and they occur through an intermediate formed by a double-strand break that is, at least in some cases, repaired by NHEJ. Such translocations were, however, rarely found in T-cell malignancies, but these frequently harboured rearrangements indicative of inappropriate joining of spontaneously arising double-strand breaks. Significantly, chromosomal breaks and translocations, some possibly involving the immune-receptor genes, are also seen in mice that lack H2AX but have p53 (ref. 7). The lack of marked tumour formation in these circumstances presumably indicates that cells undergoing cancer-causing genome rearrangements are eliminated by p53-dependent cell death.

Some of the effects of H2AX deficiency1,2 are similar to those seen in mice deficient in NHEJ, in that these — when also deficient in p53 — develop cancers that are associated with frequent translocations involving immune-receptor genes9,10. Taken together, then, it seems that the fidelity of NHEJ is partially disrupted when H2AX is impaired. This is consistent with the recent report that some immunoglobulin rearrangements are defective in H2AX-deficient mice even when p53 is present11.

So H2AX joins the growing number of genome 'caretakers' that also function as tumour suppressors. Notably, Celeste et al. and Bassing et al. show that p53-deficient mice with one defective copy of the H2AX gene also develop tumours, albeit at lower rates than their wholly H2AX-deficient counterparts. Furthermore, the normal copy of H2AX is retained in these tumours. In genetic parlance, then, H2AX displays 'haploinsufficiency' — it functions in a dose-dependent manner. And, unlike the situation with 'classical' tumour suppressors, complete loss of H2AX function is not required for tumour development.

These are exciting findings, but the question remains: how does the phosphorylated form of H2AX influence the repair of double-strand breaks and hence genome stability? As repair still occurs in the absence of H2AX, it is improbable that this protein is required in the catalytic steps of the process. As highlighted by the new reports1,2, it seems instead to affect the efficiency or accuracy of the repair.

One clue to how it might do this comes from the fact that foci formation is impaired in its absence. Perhaps the phosphorylation of the H2AX tail produces a binding site for a protein or complex that facilitates the accumulation of checkpoint and repair proteins at sites of damage, following the initial recruitment of a few such proteins. This accumulation of factors might stabilize the DNA ends, perhaps maintaining them in proximity and so preventing inappropriate repair. Also, foci formation may be the mechanism by which H2AX mediates the cell-cycle checkpoint when levels of DNA damage are low. Other DNA-damage-induced covalent modifications, either on the H2AX tail or elsewhere in the nucleosome, might influence the interactions of γ-H2AX.

Another, not necessarily exclusive, possibility is that phosphorylation of H2AX directly affects chromatin structure. The H2AX tail is not a part of the nucleosome core, but extrudes outwards, in the region where the DNA enters the nucleosome. Phosphorylation here could easily be envisaged to affect the association of DNA with a nucleosome, the binding of linker histones, or the ability of nucleosomes to fold into higher-order structures. This could, in turn, provide a more amenable template for manipulation by repair factors, or it could stabilize the DNA ends as above. There is in vivo evidence to support this hypothesis5,12, although it is not known whether chromatin structure is being directly or indirectly altered by histone phosphorylation.

Whatever the mechanism, Celeste et al.1 and Bassing et al.2 have shown that a structural component of chromatin can also act as a tumour suppressor in mice. The same might be true in humans, as the H2AX gene maps to a region of the genome that is frequently mutated in human tumours. If so, this could have medical applications. For instance, the H2AX 'status' of a person might help to define their underlying predisposition to cancer. Moreover, a look at p53 and H2AX activity in a particular tumour could ultimately help clinicians in their choice of therapy.


  1. 1

    Celeste, A. et al. Cell 114, 371–383 (2003).

  2. 2

    Bassing, C. H. et al. Cell 114, 359–370 (2003).

  3. 3

    Redon, C. et al. Curr. Opin. Genet. Dev. 12, 162–169 (2002).

  4. 4

    Celeste, A. et al. Nature Cell Biol. 5, 675–679 (2003).

  5. 5

    Downs, J. A., Lowndes, N. F. & Jackson, S. P. Nature 408, 1001–1004 (2000).

  6. 6

    Bassing, C. H. et al. Proc. Natl Acad. Sci. USA 99, 8173–8178 (2002).

  7. 7

    Celeste, A. et al. Science 296, 922–927 (2002).

  8. 8

    Fernandez-Capetillo, O. et al. Nature Cell Biol. 4, 993–997 (2002).

  9. 9

    Difilippantonio, M. et al. Nature 404, 510–514 (2000).

  10. 10

    Gao, Y. et al. Nature 404, 897–900 (2000).

  11. 11

    Reina-San-Martin, B. et al. J. Exp. Med. 197, 1767–1778 (2003).

  12. 12

    Fernandez-Capetillo, O. et al. Dev. Cell 4, 497–508 (2003).

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.