‘accretion disk’ that radiates ferociously.This process is believed to power activegalactic nuclei, such as quasars. In ourGalaxy, fuel is plentiful: hot, young stars nearthe Galactic centre blow a wind of gas pastthe black hole. Why, then, is it not brighter?

The radiation that we see at the centreof the Milky Way is believed to be the faintsqueaking of a black hole that is hiding a sub-stantial diet by advecting and swallowing theenergy in its accretion disk before it has timeto be radiated away7 So, although the sus-pected black hole at the centre of our Galaxyhas become a cornerstone of the theory8,9

that violent activity in galactic nuclei is pow-ered by accretion onto black holes, the evi-dence that the central dark mass really is ablack hole is very indirect. Accordingly, sub-stantial effort has gone into testing alterna-tive interpretations of the observations.

Black holes are extraordinarily slippery.The stars that Ghez et al. use in their analysisare 50,000 Schwarzschild radii from the cen-tre. This is closer to the centre than in anyother candidate black hole, but it is far out-side the region of strong gravity. That is, thelargest stellar velocity, 1,350 km s11, is muchsmaller than the speed of light. So the argu-ments for a black hole are still indirect. Themost plausible alternative in astrophysicalterms is a clump of dark stars — stars that aretoo low in mass to ignite nuclear reactions orstars that have died and faded from sight.Past measurements have determined a smallenough radius for the dark mass to exclude

these alternatives3,4,10. Ghez et al. strengthenthese arguments by finding a new minimumdensity of the dark mass, 821012M(

parsec13, that is almost an order of magni-tude larger than previously found. Squeez-ing dark stars to such a high density has con-sequences that are excluded by the observa-tions.

Other more exotic possibilities remain.One example is a hypothetical ball of neutri-nos11,12. But a neutrino ball of 2.62106M( is

expected to have a radius of about 0.02 par-secs. Two of the three stars studied by Ghezand colleagues are well inside this radius,only 0.005 parsecs from Sagittarius A*. Thefuture paths of these stars depend sensitivelyon whether there is a black hole or a neutrinoball at the centre of the Galaxy. So we can testsome of the more exotic black-hole alterna-tives by following the stars as they movethrough substantial orbit arcs. This shouldbe possible within a few more years.

Those of us who study galactic dynamicsgenerally observe a snapshot of objects thatchange only on timescales of millions or bil-lions of years. The orbital periods of the starsmeasured by Ghez et al. can be as short as afew decades. There is something quite grandin the realization that we can expect, withgood health and a little luck, to see theGalactic centre rotate at least once in ourlifetimes. ■

John Kormendy is in the Department of Astronomy,University of Texas at Austin, RLM 15.308, Austin,Texas 78712, USA.e-mail: kormendy@astro.as.utexas.edu1. Ghez, A. M., Morris, M., Becklin, E. E., Tanner, A. & Kremenek,

T. Nature 407, 349–351 (2000).

2. Eckart, A. & Genzel, R. Mon. Not. R. Astron. Soc. 284, 576–598

(1997).

3. Genzel, R., Eckart, A., Ott, T. & Eisenhauer, F. Mon. Not. R.

Astron. Soc. 291, 219–234 (1997).

4. Genzel, R., Pichon, C., Eckart, A., Gerhard, O. E. & Ott, T.

http://xxx.lanl.gov/abs/astro-ph/0001428

5. Ghez, A. M., Klein, B. L., Morris, M. & Becklin, E. E. Astrophys.

J. 509, 678–686 (1998).

6. Lo, K. Y., Shen, Z.-Q., Zhao, J.-H. & Ho, P. T. P. Astrophys. J.

508, L61–L64 (1998).

7. Narayan, R., Mahadevan, R. & Quataert, E. in The Theory of

Black Hole Accretion Discs (eds Abramowicz, M. A., Björnsson,

G. & Pringle, J. E.) 148–182 (Cambridge Univ. Press, 1998).

8. Rees, M. J. Annu. Rev. Astr. Astrophys. 22, 471–506 (1984).

9. Blandford, R. D. in Active Galactic Nuclei: Saas-Fee Course 20

(eds Courvoisier, T. J.-L. & Mayor, M.) 161–275 (Springer,

Berlin, 1990).

10.Maoz, E. Astrophys. J. 494, L181–L184 (1998).

11.Tsiklauri, D. & Viollier, R. D. Astrophys. J. 500, 591–595 (1998).

12.Munyaneza, F., Tsiklauri, D. & Viollier, R. D. Astrophys. J. 526,744–751 (1999).

news and views

NATURE | VOL 407 | 21 SEPTEMBER 2000 | www.nature.com 309

Reactive oxygen species are potentiallydangerous by-products of cellularmetabolism that have direct effects on

cell development, growth and survival, onageing, and on the development of cancer1.They are generated by all aerobic organisms,but their production is a double-edgedsword. On the one hand, they seem to beneeded for signal-transduction pathwaysthat regulate cell growth2 and reduction–oxidation (redox) status1. But on the other,excessive amounts of these metabolites canstart lethal chain reactions, which oxidizeand disable structures that are required forcellular integrity and survival1. Many

tumour cells seem to have increased rates ofmetabolism compared with normal cells,which would typically lead to increasednumbers of reactive oxygen species. So oneway of treating cancer might be to designdrugs to target the enzymes that regulate thelevels of reactive oxygen species. On page 390of this issue3, Huang and colleagues providesupport for this idea: they find that a com-pound that kills human leukaemia cells,but spares normal cells, may achieve thisby inhibiting such an enzyme.

Reactive oxygen species are generatedduring the production of ATP by aerobicmetabolism in mitochondria. The leakage of

Figure 1 Infrared image of the Sagittarius (Sgr) A* star cluster taken in May 1999. Filled circles showthe positions of six stars at different epochs. Ghez et al.1 have measured stellar accelerations for threestars close to the Galactic centre (S0-1, S0-2 and S0-4). The orbits of S0-1 and S0-2 (yellow ellipses)were calculated by assuming that Sagittarius A* is a supermassive black hole with a mass of 2.62106

solar masses. The orbital periods are 63 and 17 years, respectively; they are near the short end of therange of orbits allowed by the acceleration measurements. (Image kindly provided by A. Ghez.)

Cancer

A radical approach to treatmentJohn L. Cleveland and Michael B. Kastan

© 2000 Macmillan Magazines Ltd

electrons from mitochondria during theelectron-transport steps of ATP productiongenerates the reactive oxygen species super-oxide (O2

1) and hydroxyl (OH•) radicals.These species can lead to the production ofhydrogen peroxide (H2O2), from which fur-ther hydroxyl radicals are generated in areaction that either depends on, or is catal-ysed by, Fe2& ions1 (Fig. 1). Cells have evolveda series of antioxidant systems to handlethese dangerous natural by-products. Thesedefence systems include intracellular super-oxide dismutases (SODs), which convert O2

1

into H2O2; enzymes that inactivate H2O2 orhydroxyl radicals; and enzymes that trap freeradicals or transition metals (such as Fe2&)that are a reservoir for electrons.

A higher level of aerobic metabolism —and hence of reactive oxygen species — intumour cells compared with normal cellswould make enzymes in these pathways goodcandidate drug targets. However, previousattempts to inhibit SODs have not provedclinically useful. For example, ions that suc-cessfully compete with O2

1 to bind to SODswere not specific enough to cancer cells andproduced unacceptable levels of damage tonormal cells. But Huang et al.3 now show thata group of oestrogen derivatives, of all things,seem to bind to and inhibit SODs, and that— by damaging mitochondria — these com-pounds induce the suicide of a variety ofhuman leukaemia cells in vitro.

Huang et al.’s insights are their reward fortheir persistent investigation of the observa-tion that the oestrogen derivative in question— 2-methoxyoestradiol — induces thedeath of cancer cells. This oestrogen deriva-tive is a normal product of the metabolism ofoestradiol (a type of oestrogen), and has arather rich history in the literature. Althoughit is an oestrogen derivative, 2-methoxy-oestradiol cannot bind to the oestrogenreceptor4. But it does induce the suicide ofcancer cells in vitro and in vivo5–7. This resulthas been variously attributed to the effects of2-methoxyoestradiol on the formation ofblood vessels to feed the tumour, the poly-merization of tubulin (a component of thecell’s cytoskeleton), and/or the expression ofcell-suicide and cell-division regulators5–7.The findings of Huang et al.3 indicate thatthese different effects may in fact result fromthe inhibition of SODs.

There are two intracellular SODs: themitochondrial-associated manganese-dependent SOD (MnSOD), and the cyto-plasmic/nuclear copper/zinc-dependentSOD (CuZnSOD). The authors use a varietyof techniques to show that the expression ofCuZnSOD is induced by 2-methoxyoestra-diol in leukaemia cells. They reason that thiscould reflect one of two scenarios. First, 2-methoxyoestradiol might cause an increasein the amount of dangerous O2

1 in thetumour cells; more SODs would therefore beneeded to mop up the radicals. If this were

true, then more H2O2 should be produced inthe leukaemia cells in response to 2-methoxyoestradiol (Fig. 1). But instead,Huang et al. find that 2-methoxyoestradiolcauses a dose-dependent decrease in H2O2

levels. The second possibility, which fits withthis observation, is that 2-methoxyoestradi-ol suppresses SOD activity. This mightsound counterintuitive, but if this were tohappen, the cells would sense this changeand respond by increasing the level of CuZn-SOD.

Indeed, the authors show that 2-methoxyoestradiol effectively inhibits theactivity of both CuZnSOD and MnSOD invitro, and binds to SODs in intact cells.Huang et al. also find that if they manipulateSOD levels, they can control the sensitivity ofovarian cancer cells to 2-methoxyoestradiol.And O2

1 scavengers or antioxidants protectleukaemia cells from the lethal effects of thiscompound.

These are provocative findings, but theyraise several questions. First, is SOD the onlyrelevant target of 2-methoxyoestradiol thatis required for the suicide of leukaemia cells?Second, we need to know more about howthis compound inhibits SOD. For example,

where does it bind, and does it prevent thebinding of O2

1 to SODs? Structural studiesshould help in answering these questions.Third, these findings imply that this andsimilar oestrogen derivatives (Fig. 1) — orother compounds that inhibit SODs —could be used broadly in cancer treatment.But, as Huang et al. point out, there could becell-specific circumstances in which inhibit-ing SODs might compromise tumour treat-ment. Finally, will 2-methoxyoestradiol orsimilar oestrogen analogues truly be usefulfor treatment purposes? For example, thesecompounds might have toxic side effects:2-methoxyoestradiol compromises thegrowth of new blood cells5,7 (which is neededfor normal as well as tumour cells) and candamage DNA in vitro8.

Whatever the answers, the results3 raisehope that inhibiting SODs at the same timeas treating cells with agents that increase thelevels of reactive oxygen species is a promis-ing way of treating at least some cancers. Atthe very least, even if SODs turn out not to beideal drug targets, we can now be confidentthat it is reasonable to try to exploit the dif-ferences in the ways that normal and tumourcells control their redox status. ■

news and views

310 NATURE | VOL 407 | 21 SEPTEMBER 2000 | www.nature.com

Cell survival

H3CO

HO

CH3 OH

1

3

2

17

4

O2-

O2

Aerobic metabolism

2-Methoxyoestradiol

H2O2

Glutathioneperoxidase

OH.

Cell suicide

MnSOD

Catalase

CuZnSOD

H2O O2

Fe2+

Figure 1 Inhibition of superoxide dismutases triggers the suicide of leukaemia cells. Superoxide (O21),

a toxic free radical, is a cellular by-product of aerobic metabolism. Superoxide is mopped up by twosuperoxide dismutases (SODs) — copper/zinc-dependent SOD (CuZnSOD) and manganese-dependent SOD (MnSOD) — and then by other enzymes such as catalase and glutathione peroxidase.This allows the survival of the cell. Cell suicide can also occur if hydrogen peroxide (H2O2) levels risetoo much or if H2O2 is instead converted to hydroxyl radicals (OH

•) in an Fe2&-dependent reaction1.

Cancer cells have high levels of metabolism that generate reactive oxygen species, and may sometimeshave low levels of SOD activity. Huang et al.3 show that treating leukaemia cells with the oestradiolderivative 2-methoxyoestradiol leads to increases in the expression of CuZnSOD, and that thisreflects the cell’s response to inhibition of SOD activity by 2-methoxyoestradiol. The result is a netincrease in superoxide levels and the suicide of leukaemia cells. A derivative of this protein with ahydroxyl substitution at the 2-carbon (red) can also kill leukaemia cells.

© 2000 Macmillan Magazines Ltd

John L. Cleveland and Michael B. Kastan are in theDepartments of Biochemistry and Hematology/Oncology, St Jude Children’s Research Hospital, 332N. Lauderdale, Memphis, Tennessee 38018, USA.e-mails: john.cleveland@stjude.orgmichael.kastan@stjude.org1. Davies, K. J. Biochem. Soc. Symp. 61, 1–31 (1995).

2. Sundaresan, M., Yu, Z.-X., Ferrans, V. J., Irani, K. & Finkel, T.

Science 270, 296–299 (1995).

ate biogeochemistry, and possibly the ecolo-gy, of the entire world ocean will be altered byrising CO2 levels (Fig. 1).

Atmospheric CO2 equilibrates rapidlywith the surface layer of the ocean, wheremost additional CO2 combines with carbon-ate ions:

CO2&CO321&H2O → 2HCO3

1

This leads to a decrease in the concentra-tion of CO3

21, one of the building blocksof calcium carbonate, and in the saturationstate of calcium carbonate, V(V4[Ca2+]2[CO3

21]/Ksp, where Ksp is theequilibrium constant of CaCO3). Ratherthan CO2, V seems to be the controlling fac-tor, because increasing V by adding bicar-bonate promotes calcification even though itinduces higher concentrations of dissolvedCO2 (ref. 2).

Decreased calcification in response toincreased concentrations of CO2 has nowbeen reported across phylogenetically dis-tant groups that precipitate different types ofCaCO3 crystals (low-magnesium calcite,

high-magnesium calcite and aragonite),either through internal or external calcifica-tion. But the unifying theme is photosyn-thesis: to date all of the calcifying organismsshown to have a strong saturation-stateresponse are plants or animals that dependon photosynthetic symbionts.

Coccolithophorids are widely used as amodel organism to investigate photosyn-thesis and calcification. Both processes, andtheir interaction, are relatively well under-stood in these algae. We know little, however,of how they might respond to environmentalchange. Riebesell and colleagues5 have notonly begun to fill that gap, but their resultsalso draw attention to the likelihood that thecritical factor in responses to higher CO2 isnot as much the mineralogy as the link tophotosynthesis.

Rising CO2 levels have two antagonisticeffects on the CO2 fluxes mediated by calcifi-cation. The amount of CO2 generated by cal-cification will fall as a result of the decreasedrate of CaCO3 precipitation1,5. However,increased CO2 concentration also shifts theseawater carbonate equilibria with the resultthat more CO2 is released per mole of CaCO3

precipitated7. These two effects could bebalanced in coral reefs1 but the response ofpelagic calcification could lead to increasedCO2 storage capacity in the upper ocean5.CaCO3 compensation is the CaCO3 dissolu-tion that maintains a balance between inputsfrom land and burial in deep-ocean sedi-ments; this process is a potential regulator ofatmospheric CO2 during glacial–interglacialcycles8, and could also be affected. Dissolu-tion is favoured by decreasing V and increas-ing coccolithophorid primary productionrelative to calcification5, the so-called rainratio.

What are the main issues to be resolved?First, the role of photosynthetically coupledcalcification in the global carbonate cycle,

news and views

NATURE | VOL 407 | 21 SEPTEMBER 2000 | www.nature.com 311

3. Huang, P., Feng, L., Oldham, E. A., Keating, M. J. & Plunkett,

W. Nature 407, 390–395 (2000).

4. Cushman, M., He, H. M., Katzenellenbogen, J. A., Lin, C. M. &

Hamel, E. J. Med. Chem. 38, 2041–2049 (1995).

5. Fotsis, T. et al. Nature 368, 237–239 (1994).

6. Mukhopadhyay, T. & Roth, J. A. Oncogene 14, 379–384

(1997).

7. Klauber, N., Parangi, S., Flynn, E., Hamel, E. & D’Amato, R. J.

Cancer Res. 57, 81–86 (1997).

8. Tsutsui, T. et al. Carcinogenesis 21, 735–740 (2000).

Most concerns about rising concentra-tions of atmospheric carbon dioxidecentre on how climate may change.

But there may also be direct biologicaleffects. In terrestrial ecosystems, extraatmospheric CO2 may have a fertilizingeffect, resulting in increased photosynthesis.Except for a few organisms, such as seagrass-es, such an increase seems unlikely in marinesystems because most algae use bicarbonateions (HCO3

1) rather than CO2 as a photosyn-thetic substrate. There have, however, beenindications of a negative biological responsein the marine environment. Several stud-ies1–4 have shown that calcification rates ofreef-building corals and coralline algae aredepressed by increased levels of CO2. Onpage 364 of this issue, Riebesell et al.5 consid-erably extend the significance of these earlierresults by describing the influence of CO2 onthe calcification of coccolithophorids — animportant group of algae widely distributedin the surface layer of coastal waters and theopen ocean.

The authors provide experimental evi-dence that the calcification rate of twospecies of coccolithophorids maintained inculture is lower at CO2 levels predicted for2100 than at preindustrial levels. Naturalcommunities that the authors sampled in theNorth Pacific showed a similar response.Coccolithophorids are distributed through-out the world’s oceans (Fig. 1), and may bethe world’s most important producer ofcalcium carbonate (CaCO3). They formlow-magnesium calcite shells and their dis-tribution extends into subpolar waters.

The reef-building corals and corallinealgae known to be sensitive to high CO2 lev-els1–4 are bottom-dwelling. They occur most-ly in tropical and subtropical regions,although some of the algae extend to highlatitudes, and they typically form skeletonsof aragonite or high-magnesium calcite.Together, the coccolithophorids, reef-build-ing corals and coralline algae are responsiblefor well over half of the world’s CaCO3 pro-duction6. Given their widespread combineddistribution, it seems likely that the carbon-

Ocean biogeochemistry

Calcification and CO2Jean-Pierre Gattuso and Robert W. Buddemeier

Figure 1 Distribution of coccolithophorids and coral reefs, the major photosynthetic and calcifyingsystems in the marine environment. Coccolithophorid blooms (blue dots) are transient and annuallycover 142105 km2, mostly in the subpolar regions11. Coral reefs (red dots) are permanent, long-livedecosystems that cover 62105 km2 in the tropics12. Taken together, several studies1–5, including thatnow reported by Riebesell et al.5, show that the carbonate biogeochemistry of the world’s oceans issubject to alteration by rising levels of CO2. (Distribution of coccolithophorid blooms is from ref. 11.)

© 2000 Macmillan Magazines Ltd