conditions. So the variations between birdsfrom the different populations could not be due to physiological adaptations to theirrespective local climates. Instead, the authorssuggest that the differences in metabolic rateare specified by their genes.

What selective pressures shape the geneticmake-up of different stonechat populations?Stonechats living in mild climates are year-round residents, whereas in locations withharsh winter climates, the birds are long-dis-tance migrants. So the higher metabolic rateof stonechats from the Austrian and Kazak-stan populations could be a consequence ofselective pressures associated with migration.Wikelski et al. suggest that this is not the casebecause, energetically, migration is a cheapoption compared with thermoregulation.The high metabolic rates of the Austrian andKazakstan populations are more likely to beselected for by occasional periods of freezingtemperatures during the breeding season.

Survival of warm-blooded animals in coldclimates is possible only if food consumptionis sufficient to meet the energy requirementsfor maintaining body temperature. But highmetabolism has other costs besides increas-ing appetite — lifespan correlates inverselywith heart rate across a wide range of mam-mal species. Mammals with low metabolicrates (and therefore slow heart rates) tend tolive longer. Indeed, the relationship betweenmetabolic rate and senescence is a centralissue in the study of longevity4. Higher meta-bolic rates lead to faster cell divisions and a more rapid accumulation of mutations,which might eventually impair cell functionand contribute to the ageing process. So thelower metabolism of Kenyan stonechats willallow them to survive better where food isscarce, and might also permit slower senes-cence. This indeed appears to be a general feature of tropical birds and mammals: thepace of life is slow,but survival is high.

Wikelski and colleagues have looked athow metabolic rate varied between popula-tions,but other researchers have investigated

the variation within single populations. Inthese studies, between half and two-thirds of the inter-population variation was attri-buted to individual characteristics thatremained unchanged in individuals fromyear to year5. So these studies also hint at agenetic basis for metabolic rate. And sea-sonal6 or between-individual7 variations inorgan size and other aspects of body com-position explained much of the individualvariation. It seems that the selective pres-sures that influence metabolism may becomplex, influencing metabolic rate throughaffecting the density of mitochondria — theenergy-producing organelles inside cells —or the size of the muscles or liver.

The genetic basis of metabolic rate has far-reaching implications. For example, it is nowapparent that metabolism should be consid-ered when selecting animal populations toreintroduce into particular environments.Recent reintroduction programmes broughtScandinavian red kites to Scotland but Iberian kites to England. Which is the bettersource population? If Scandinavian kiteshave a higher metabolic rate than their Iberian relatives, they will be better adapted

to surviving cold conditions, but they willneed more food to sustain them.

The new study1 raises many other ques-tions. Have island populations evolved lowermetabolic rates to cope with food scarcity?Have equivalent selection pressures shapedother features, such as insulating plumage orpelt, to match the variations in metabolism?Within a population, do individuals withlower metabolic rates have a slower pace oflife? Wikelski et al. have demonstrated thatmetabolic rate is a genetic characteristic thatcan respond to selective pressure. But, asthese questions show, there is a lot more tolearn about how the environment shapes theevolution of this characteristic. ■

Robert W. Furness is at the Institute of Biomedicaland Life Sciences, Graham Kerr Building,University of Glasgow, Glasgow G12 8QQ, UK.e-mail: r.furness@bio.gla.ac.uk1. Wikelski, M. et al. Proc. R. Soc. Lond. B

doi:10.1098/rspb.2003.2500 (2003).2. Klaassen, M. Oecologia 104, 424–432 (1995).3. Bryant, D. M. & Furness, R. W. Ibis 137, 219–226 (1995).4. Kirkwood, T. B. L. Nature 270, 301–304 (1997).5. Bech, C., Langseth, I. & Gabrielsen, G. W. Proc. R. Soc. Lond. B

266, 2161–2167 (1999).6. Bech, C. et al. Comp. Biochem. Physiol. A 133, 765–770 (2002).7. Bech, C. & Ostnes, J. E. J. Comp. Physiol. B 169, 263–270 (1999).

news and views

780 NATURE | VOL 425 | 23 OCTOBER 2003 | www.nature.com/nature

Figure 1 A little bird with a lot of variation.Different stonechat populations live in different climates and have different metabolicrates. Wikelski et al.1 show that this variation is inherent in the animal’s genes.

Cancer

A twist in a hedgehog’s taleMatthew P. Scott

The genetics of development can often explain the genesis of cancer.This now seems to be true for cancers of the gut, but the patterns ofgene expression in these tumours tell a tale with a twist.

Growth is beautifully orchestratedduring normal development to produce animals of certain sizes.

Similarly, in adult tissues that constantlyregenerate, such as blood, gut and skin, adelicate balance is established between celldeath and cell division. But genetic damagecan destroy the system of checks and balances, sometimes causing cancer. Indeed,it is becoming increasingly clear that manytumours result from normal developmentalprocesses gone awry. Several types of cancerhave been linked to the disruption of the‘hedgehog’ signalling pathway, which is crucial to the normal development of manytissues1,2. In this issue, Berman et al.3 andThayer et al.4 now implicate hedgehog signal-ling in the genesis of cancers of the digestivesystem. Unusually, however, the fault doesnot lie with the molecules that propagatethe hedgehog signal — instead, the tumoursmake too much hedgehog.

The hedgehog protein (Hh) is so calledbecause it was discovered as the trigger for asignalling pathway that organizes the patternof bristles on fruitfly larvae. But Hh isinvolved in the development of many organsand tissues in numerous animal species.

After being secreted from cells, Hh binds to a receptor protein on the surface of cellsnearby.The receptor then transmits an intra-cellular signal, often culminating in changesin gene expression. Most of the componentsthat transduce the signal within cells havebeen conserved during more than half a billion years of evolution and are recogniz-ably similar in insects and mammals.

Two components of the Hh pathway arePatched (Ptc), the cell-surface receptor pro-tein to which the secreted signal can bind,and Smoothened (Smo), an intracellularprotein that activates genes in response to theHh signal. In the absence of Hh, Ptc inhibitsthe activity of Smo. But when a burst ofHh molecules is secreted by nearby cells,the molecules bind to Ptc, unleashing Smo,which then transmits the signal through afurther chain of regulators. The result is theactivation of certain genes, some of whichencode proteins that trigger growth.

These regulatory relationships are the keyto understanding what happens in severaltypes of cancer. If Ptc is not sufficientlyactive, Smo can be overactive and cells can divide when they should not. The mostcommon cause of the deregulation is genetic

G.

HO

FMA

NN

types, which led them to hypothesize thatthis signalling molecule might be the triggerfor tumour growth. To test this idea theytried to inhibit the growth of culturedtumour cells by using a Hh-blocking anti-body. Treatment with the antibody blockedthe growth of a wide range of tumour cells,whereas the addition of Hh caused tumourgrowth to spurt.

Meanwhile, Thayer et al. examined Hhproduction in 20 human pancreatic cancerbiopsies and found that the protein was aberrantly expressed in 70% of the speci-mens. To investigate whether Hh might con-tribute to the genesis of pancreatic cancer,these authors examined transgenic mice inwhich Hh had been overproduced in thedeveloping pancreas. All of these mice con-tained abnormal pancreatic cells that boresimilarities to precancerous cells observed ina form of human pancreatic cancer.

Importantly, the link between the Hhpathway and cancer of the digestive tract isaccompanied by ideas for treatment. BothBerman et al. and Thayer et al. inhibited

news and views

NATURE | VOL 425 | 23 OCTOBER 2003 | www.nature.com/nature 781

tumour growth in mice with the drugcyclopamine. This chemical is a teratogen —a compound that can cross the placenta andcause defects in a developing fetus.It was firstdiscovered in extracts from the corn lily, abeautiful but treacherous flower often foundin alpine meadows. If a pregnant sheep eatsthe plant, the fetus develops with cyclopicfacial features — the same defect that iscaused by inadequate Hh activity in peopleand mice. It turns out that cyclopamine canblock the activity of Smo, so this, or otherdrugs that affect the Hh pathway, havepotential as anti-cancer drugs10.

Both groups used a ‘subcutaneousxenograft model’ of digestive-tract cancer to test the effect of cyclopamine. In thismodel, tumour cells are seeded under theskin of immune-deficient mice. Berman et al. waited until the tumours had grown to a certain size, before injecting them withcyclopamine each day.Remarkably,the treatedtumours regressed completely within 12days. Meanwhile, Thayer et al. found thattreatment with cyclopamine reduced the

mutation. Basal-cell carcinoma of the skin— a common result of exposure to sunlight— is most often triggered by ultraviolet dam-age to both copies of the ptc gene5,6.Mutationof ptc has also been implicated in certainbrain7 and muscle tumours8. Cancer can alsoresult from mutations in smo that convert theprotein to a permanently activated form9.

Berman et al.3 and Thayer et al.4 nowimplicate the Hh pathway in the genesis ofcancers of the digestive system. Rather littlehas been understood about the origins andgrowth of these often fatal forms of cancer,butthe role of Hh in the normal development ofthese tissues hints that defective Hh signallingmight allow growth to explode. Curiously,however, it seems that neither ptc nor smo ismutated in tumours of the digestive tract.Instead,the cancer cells make too much Hh.

Berman et al. surveyed the production of Hh in cultured cells derived from severaldifferent types of digestive-system tumour,including those of the oesophagus, stomach,biliary tract, pancreas and colon. Theydetected Hh expression in all of these cell

G.R

OW

ELL

/CO

RB

IS

The soaring, jagged peaks of theworld’s second highest mountainrange, the Andes, could owe theirstature not to the power of the Earthbut to that of the sea. So proposeSimon Lamb and Paul Davis in thisissue (Nature 425, 792–797; 2003).

Most mountain ranges arecreated by the collision of twocontinental plates, such as thegrinding action of India against Asiathat thrust up the Himalayas. But theAndes are perched on a point wherean oceanic plate slips down beneatha continental one. Great mountainsaren’t usually born of such meetings.Heavy, dense oceanic plates tend toslip underneath continental ones,causing major earthquakes but notworld-class mountains. If the push of the mid-Atlantic ridge on tectonicplates were the only factor driving up the Andes, one calculation showsthat they would be no more than two kilometres high — half theiractual height.

Lamb and Davis argue that theextra push comes from the fact thatthe forces of the plate collision arefocused on a small area — a stretchof the plate boundary where thefriction in the trench between thePacific and South America isparticularly high. The cold watercurrent that sweeps up the west

coast of Peru and Chile from highsouthern latitudes encourages littlewater evaporation, and therefore littlerain. That means there are no riversto dump worn mountain sedimentback into the ocean — sediment thatcould act as lubrication. Instead, thetrench off the coast of the Andes isrough and dry. That extra friction,say the authors, helps to prop up the mountains.

There could even be a positivefeedback loop in place, causing themountains to bulk up more and

more over time. The higher themountains grow, the drier thecoastline becomes — as any wet air coming in from the Atlantic isblocked by the towering peaks — which in turn further reduceserosion and props the mountains up still more.

Complex relationships betweenrock, air and sea have been foundbefore, although it’s usuallymountains that are thought to affectclimate, rather than the other wayaround. The Himalayas, for example,

are believed to have changed the airflow enough to spark the formationof the Indian monsoons. And as allthat rain weathered away themountains’ rocks, carbon dioxidewas taken out of the air and sentdown streams as carbonate, to beburied at the bottom of the sea; thisextraction of greenhouse gas isthought to have cooled the globalclimate. But in the Andes, rather thanthe mountains making the climate,the climate might actually have madethe mountains. Nicola Jones

Earth science

How do your mountains grow?