a protein called Cdc25A in a mammalian cellline in response to ionizing radiation. Whenactive, Cdc25A removes phosphate groupsfrom another enzyme, Cdk2, activating itand promoting progression through the cellcycle. So it was reasonable to predict that adecrease in the level of Cdc25A might lead to a failure to activate Cdk2, resulting in inhibition of the cell cycle. This hypothesis is supported by the authors’ finding3 thateither persistent expression of Cdc25A, orexpression of a Cdc25A-resistant mutant ofCdk2, blocks the irradiation-induced S-phase checkpoint.

The next question was how does ionizingradiation enhance degradation of Cdc25A?From their previous work4, the authors knewthat a different type of radiation, ultravioletradiation, causes a decrease in Cdc25A levelsthat is dependent on the enzyme Chk1. Sothey now looked at whether Cdc25A couldbe a substrate for either Chk1 or Chk2 afterionizing radiation. The authors confirm pre-vious observations5,6 that ionizing radiationactivates Chk2, and find that activated Chk2phosphorylates Cdc25A in vitro.

It has been suggested that mutations ofChk2 in the germ line (the tissue from whicheggs or sperm are produced) are responsiblefor some cases of Li–Fraumeni syndrome7,characterized by a predisposition to cancer.Falck et al. show that, when they introduceeither these naturally occurring Chk2mutants, or an artificially generated, cata-lytically inactive mutant, into a mammaliancell line, the downregulation of Cdc25A that normally occurs in response to ionizing radiation is blocked. These mutants alsoinhibit the ionizing-radiation-induced S-phase delay. Conversely, introduction ofnormal Chk2 into a tumour cell line carryingmutated Chk2 restores ionizing-radiation-induced degradation of Cdc25A and S-phasedelay. The connection between ionizingradiation, activation of Chk2 and degrada-tion of Cdc25A was firmly established. Thefinal piece in the jigsaw was the identificationof the amino-acid site in Cdc25A that isphosphorylated by Chk2, and the demon-stration of the importance of this site in theirradiation-induced phosphorylation anddegradation of Cdc25A.

The authors extended their observationsby linking the involvement of Chk2 in thispathway to the protein ATM (Fig. 1).Cellsdefective in ATM are known to lack the ioniz-ing-radiation-induced S-phase delay. More-over, ATM activates Chk2 during the G1checkpoint. Falck et al. now show that theactivation of Chk2, degradation of Cdc25Aand inhibition of Cdk2 are all defective inataxia telangiectasia cells, in which ATM ismutated (hence the name of this protein).The implication is that the phosphorylationof Chk2 by ATM is required for both G1 andS-phase delays after ionizing radiation.

Eight years ago8,9, the link between the

single year, and by excluding the variation inbiological processes, such as photosynthesisand remineralization, that affect dissolvedCO2. So oceanic studies may indeed under-estimate year-to-year variation of air–seafluxes of CO2.

Most recent atmospheric and oceanicstudies4,5,9 agree that there is a relatively large— 2 Gt C — annual uptake of CO2 by theoceans. But there are large uncertainties inthe spatial distribution of CO2 air–seaexchange, such as the size of the CO2 sink inthe Southern Ocean4,9. The location of thestrong CO2 sink in the Northern Hemi-sphere8 is of especial political and economicinterest: certain atmospheric studies place it largely in terrestrial systems, notably inNorth America10. The lack of consensus onthe distribution of CO2 sinks, whether onland or in the oceans, let alone within specificnational boundaries, poses a fundamentalproblem to the implementation of the KyotoProtocol for limiting greenhouse-gas emis-sions. The work of Loukos et al.1 contributesto this debate by suggesting that there is onlymoderate year-to-year change in oceanicCO2 uptake. The results reduce the discrep-ancy between various estimates of the oceaniccontribution to interannual variation inatmospheric CO2 storage.

Loukos and colleagues1 provide a goodexample of how improved proxy coverage of pCO2

can be obtained in a specific region.However, their method of calculating dis-

solved inorganic carbon has no physicalfoundation and is unlikely to apply in oceanicregions outside the equatorial Pacific. In thelonger term, a better quantification of theoceanic sinks for atmospheric CO2 willrequire a global network of surface-waterpCO2

observations, combined with satellitedata on temperature, algal growth and windspeed. Satellites carrying a new generation ofatmospheric infrared instruments, due forlaunch in the next five years, should take useven closer to the near-instantaneous globaldetection of atmospheric CO2 concentra-tions and, by inference, of the distribution ofCO2 sinks on land and in the oceans. �

Dorothee Bakker and Andrew Watson are in theSchool of Environmental Sciences, University of EastAnglia, Norwich NR4 7TJ, UK. e-mails: d.bakker@uea.ac.uka.j.watson@uea.ac.uk1. Loukos, H., Vivier, F., Murphy, P. P., Harrison, D. E. & Le

Quéré, C. Geophys. Res. Lett. 27, 1735–1738 (2000).

2. Francey, R. J., Tans, P. P., Allison, C. E., White, J. W. C. &

Trollier, M. Nature 373, 326–330 (1995).

3. Keeling C. D., Whorf, T. P., Wahlen, M. & Van der Plicht, J.

Nature 375, 666–670 (1995).

4. Rayner, P. J., Enting, I. G., Francey, R. J. & Langenfelds, R. Tellus

B 51, 213–232 (1999).

5. Battle, M. et al. Science 287, 2467–2470 (2000).

6. Lee, K., Wanninkhof, R., Takahashi, T., Doney, S. & Feely, R. A.

Nature 396, 155–159 (1998).

7. Conway, T. J. et al. J. Geophys. Res. 99, 22831–22855 (1994).

8. Fung, I. Y. et al. Global Biogeochem. Cycles 11, 507–533

(1997).

9. Takahashi, T. et al. in Proc. Second Int. Symp. CO2 in the Oceans

(ed. Nojiri, Y.) 9–15 (Centre for Global Environmental

Research, Tsukuba, 1999).

10.Fan, S. et al. Science 282, 442–446 (1998).

news and views

766 NATURE | VOL 410 | 12 APRIL 2001 | www.nature.com

We thought sequencing the humangenome once was daunting enough,but mammalian cells have to copy

and segregate their three billion base pairs ofDNA with fidelity every time they divide. Ifthe DNA is damaged, the accuracy of thisprocess can decrease, leading to mutations inthe DNA of daughter cells. To protect them-selves, eukaryotic organisms — from yeast tohumans — have developed a variety ofmechanisms, several of which involve tran-sient arrests in the cell’s progression throughthe cell-division cycle.

Although many of the mechanistic detailsof these ‘checkpoints’ have yet to be workedout, several of the proteins involved havebeen identified. For example, in mammaliancells, the addition of phosphate groups to the Chk2 protein (a process known as phos-phorylation) by the ATM protein after expo-sure to ionizing irradiation contributes to

arrest in the so-called G1 phase of the cellcycle1,2. Surprisingly, on page 842 of thisissue3, Falck and colleagues report that thesame phosphorylation event also seems to berequired for an irradiation-induced delay tothe S phase.

The cell cycle consists of a series of phases,starting with G1 — the resting phase that fol-lows the previous cell division. During thenext phase, S phase, the cell’s chromosomesare copied. Another resting phase, G2, fol-lows, and then eventually the cell’s chromo-somes are segregated and the cell divides(mitosis). In response to DNA damage, thecell can pause during progression from G1into the S phase (at the G1 checkpoint), pro-gression through the S phase (the S-phasecheckpoint), and progression from G2 intomitosis (the G2 checkpoint).

The observation that spurred Falck et al.’sstudies3 was a marked decrease in the level of

Cell cycle

Checking two stepsMichael B. Kastan

When their DNA is damaged, cells temporarily stop multiplying to preventthe build-up of mutations. Two types of delay triggered by ionizingradiation appear to have common molecular starting points.

© 2001 Macmillan Magazines Ltd

irradiation-induced signalling pathways andthe cell-cycle machinery in the G1 phase wasmade with the discovery that the p53 protein(a target of Chk2 in the G1-checkpoint path-way) activates the expression of an inhibitorof the cell cycle, p21. Falck et al.’s experi-ments3 linking ATM (which responds to ion-izing radiation), through Chk2 and Cdc25A,to Cdk2 (which controls the cell cycle)appear to provide the same type of cruciallink for S-phase progression.

But this won’t be the end of the story.Other proteins will certainly be involved.For example, the protein Nbs1, which is alsophosphorylated by ATM, is also required for this S-phase checkpoint10,11. It will beimportant to work out how this particularphosphorylation event is linked to thoseuncovered by Falck et al. In the ionizing-radiation-induced G1 checkpoint, ATMphosphorylates several proteins to accom-plish its goal of halting the cell cycle2. Per-haps a similar scenario occurs during thecontrol of the S-phase checkpoint (Fig. 1). Itwill also be important to determine whetherinhibition of Cdk2 alone is sufficient to haltDNA replication, or whether other mecha-nisms are involved.

It is interesting that phosphorylation ofChk2 by ATM is required for both G1 and S-phase delays following ionizing radiation,even though Chk2 appears to work in differ-

ent ways in the two pathways. (In the G1arrest, Chk2 helps to prevent the degrada-tion of one of its targets, p53; in the S-phase delay, Chk2 enhances the degradationof another target, Cdc25A.) ATM is alsorequired for the G2-phase delay in responseto ionizing radiation. If Chk2 turns out to beinvolved in this delay, too, then the differ-ences between these three checkpointswould appear to lie downstream of ATM andChk2. Finally, Falck et al.’s study serves as a model for investigations of how signal-transduction pathways (in this case, oneinduced by DNA damage) can be linked totheir functional endpoint — here, control ofthe cell cycle. �

Michael B. Kastan is in the Department ofHematology-Oncology, St Jude Children’s Hospital,Memphis, Tennessee 38105, USA.e-mail: michael.kastan@stjude.org1. Hirao, A. et al. Science 287, 1824–1827 (2000).

2. Kastan, M. B. & Lim, D.-S. Nature Rev. Mol. Cell Biol. 1,

179–186 (2000).

3. Falck, J., Mailand, N., Syjuåsen, R. G., Bartek, J. & Lukas, J.

Nature 410, 842–847 (2001).

4. Mailand, N. et al. Science 288, 1425–1429 (2000).

5. Matsuoka, S., Huang, M. & Elledge, S. J. Science 282, 1893–1897

(1998).

6. Brown, A. L. et al. Proc. Natl Acad. Sci. USA 96, 3745–3750 (1999).

7. Bell, D. W. et al. Science 286, 2528–2531 (1999).

8. Harper, J. W., Adami, G. R., Wei, N., Keyomarsi, K. & Elledge,

S. J. Cell 75, 805–816 (1993).

9. El-Deiry, W. S. et al. Cell 75, 817–825 (1993).

10.Lim, D.-S. et al. Nature 404, 613–617 (2000).

11.Zhao, S. et al. Nature 405, 473–477 (2000).

news and views

NATURE | VOL 410 | 12 APRIL 2001 | www.nature.com 767

Figure 1 Steps involved in the cell-cycle checkpoints that are induced in response to ionizingradiation. The pathway investigated by Falck et al.3 is that shown in blue, orange and green. a, Ionizing radiation activates the kinase ATM, which in turn activates the kinase Chk2. b, This stepappears to affect progression from G1 to the S phase, through phosphorylation (represented by acircled ‘P’) and stabilization of the p53 protein, which itself enhances the expression of the cell-cycleinhibitor p21. c, Activation of Chk2 by ATM also affects progression through the S phase itself, by the phosphorylation of Cdc25A. This protein is more likely to be degraded when phosphorylated.(When unphosphorylated, Cdc25A removes a phosphate group from Cdk2, enabling the initiation of DNA replication, that is, S phase.) d, Nbs1 is also phosphorylated by ATM and is also involved inthe ionizing-radiation-induced inhibition of S-phase progression, although it is not known how itties in to the pathway described by Falck et al. e, The targets of ATM that control progression from G2 into mitosis (M phase) have not been described.

Ionizingradiation

Chk2(inactive)

Chk2(active)

???

Cdk2

???

ATM(active)

ATM(inactive)

S

M

M

a

c

ed

Cdc25A

Cdk2 Degradation

Cdc25A

p53p53

p21

M

b

P

P

G1

G1

G2

G2

S

S G2

G1Nbs1

PP

Nbs1

Mdm2

Daedalus

Flattening the flatsThe art of making true optical flats — glassdiscs with surfaces polished so accuratelythat any unevenness is small comparedwith the wavelength of light —wasrevealed in the nineteenth century. (Untilthen, people had merely ground theweakest possible mirrors.) The trick is togrind three pieces of material against eachother: A against B, B against C, and Cagainst A. All three then become true flats— or at least as flat the distortion of theEarth’s gravity permits. Lens telescopeshave an advantage over mirrors in that asmall sag in the lens makes essentially nodifference to its optical performance.Daedalus has been musing on these factsbecause of the high value of true flats toastronomy. The accurate distance of manystars could be deduced if a space telescope,equipped with flats, could be used as anoptical interferometer.

So Daedalus now shows how make trueflats. The best Earthbound flats would bemade by skilled opticians, and would beflown on the space shuttle. In microgravityit should be easy to complete the ultrafinegrinding of the flats against each other,giving three flats of amazing performance.

Ideally, two of these flats should beattached to a space telescope, one in its lightpath and the other many kilometres away.The art of stabilizing an object in space hasbeen perfected over many decades, andDaedalus hopes that the distant flat can beturned and held so as to shine the light of adistant star unwaveringly upon its fellowflat at the space telescope. The resultshould be an amazingly accurateinterferometric measurement of thedistance of the distant star.

Daedalus has no interest in the distanceof stars per se. He wants to measure thedistance of stars in nearby galaxies, andthus determine the Hubble constant withaccuracy. It is rather shocking that thisimportant constant seems to vary from 60to 85 — and that astronomers can choosewhichever value they like. They can alsomeasure it in km s�1 Mpc�1, instead of s�1

as the rest of us would have to, but that'sanother matter.

An accurate Hubble constant wouldallow cosmologists to say clearly if theUniverse were ‘open’ (doomed to expand forever) or ‘closed’ (due to fall back on itself asthe galaxies do not quite have mutual escapevelocity). Daedalus likes the idea of a closedUniverse, in which the galaxies only justslow before falling back. But he recognizesthat a Hubble constant accurately related tothe time of the Big Bang will be needed tomake it work. David Jones

© 2001 Macmillan Magazines Ltd