energy resolution of the system is defined bythe quantum uncertainty principle and isdetermined by the time the UCNs spend inthe trap. If this time can be substantiallyincreased, an energy resolution as high as10118 eV could be achieved. More intensesources of UCNs (which are under develop-ment) are required for such precision, andcould lead to new studies of fundamentalphysics. For example, improved tests of theequivalence principle are needed to investi-gate the interplay of quantum mechanicsand gravity. The equivalence principle saysthat in a uniform gravitational field all parti-

cles, regardless of their mass or composition,fall with the same acceleration (9.8 m s12

near the surface of the Earth). This meansthat the inertial and gravitational masses of the neutron have to be equivalent. Untilnow, such assumptions have been hard to check systematically, but the work ofNesvizhevsky and colleagues could providephysicists with a new probe of the funda-mental properties of matter. ■

Thomas J. Bowles is at the Los Alamos NationalLaboratory, Los Alamos, New Mexico 87545, USA.e-mail: tjb@lanl.gov1. Nesvizhevsky, V. V. et al. Nature 415, 297–299 (2002).

protein kinase C, are likely suspects5. By activating this kinase, the lipid moleculesseem to reduce the activity of a moleculeknown as IRS-1, a key component of theinsulin signalling pathway.

The latest instalment of this story beginswith an apparent paradox. Although obesityis associated with insulin resistance and lipotoxicity, so too is a rare human disorder,lipodystrophy, in which fat tissue is absent.As a result, excess lipid accumulates in tissues such as the liver and skeletal muscle, and patients often suffer from a virtuallyuntreatable resistance to insulin. Geneticallyengineered experimental animals show thesame symptoms7. In both obesity and lipo-dystrophy, then, excess lipid in cell typesother than adipocytes is associated withinsulin resistance. If the lipotoxicity theoryis true, then depletion of this intracellularlipid should improve sensitivity to insulin.

Leptin is produced by adipose cells and‘reports’ nutritional information to regu-latory centres, in the hypothalamus and elsewhere, to regulate the amount of adipose tissue. Treatment of mice with agenetically engineered form of the hormoneindeed reduces the amount of adipose tissue. But it also reduces the levels of intracellular lipid in, for example, skeletalmuscle, liver and pancreatic beta cells1,8,9,and — as predicted by theory — improvessensitivity to insulin6,7. Treatment of onestrain of lipodystrophic mice with leptindepletes the massive fat deposits in the liverand elsewhere, and corrects the animals’ diabetes7. In a second lipodystrophic strain,fat-cell transplants from normal mice correct the diabetes, but transplants of fat cells that do not produce leptin (from ob/ob

news and views

268 NATURE | VOL 415 | 17 JANUARY 2002 | www.nature.com

Obesity is often associated with insulin-resistant diabetes, and the health con-sequences of excessive fat can largely

be attributed to this connection1. The causeof the association is not known. But fromstudies using the hormone leptin to treatsevere insulin-resistant (type II) diabetes in experimental animals, it seems that thelikely culprit is fat, not in specialized adiposetissue, but elsewhere. Writing on page 339 of this issue2, Minokoshi et al. add to this picture by showing that leptin can cause fatto be cleared from skeletal muscle and bydescribing the cellular mechanism involved.

Diabetes is characterized by increasedlevels of glucose in the blood stream, whichresults in organ damage. Blood glucose iscontrolled by insulin, which is secreted bybeta cells in the pancreas. Insulin stimulatesthe uptake and use of glucose by muscle andfat cells (adipocytes), and reduces its outputby the liver, thus lowering glucose levels in the blood. Insulin also stimulates lipidbiosynthesis in fat and liver cells.

Obesity is generally associated with resistance to insulin-stimulated glucosetransport in muscle and fat, and to insulin-stimulated suppression of glucose produc-tion in the liver. When insulin secretion cannot meet the increased demand in obesesubjects, a rise in blood glucose levels ensues.These events can, to some extent, be reversedby even modest weight loss, which improvesinsulin signalling and protects pancreaticbeta cells. But why is obesity associated with insulin resistance, and weight loss withimproved insulin signalling? Various reasonshave been proposed, one of which is thatexcess lipids, particularly lipids in the ‘wrongplaces’, can inhibit insulin signalling and alsoimpair beta-cell function3–6.

According to this ‘lipotoxicity hypothe-sis’, insulin resistance develops when excesslipids are deposited in insulin-sensitive celltypes other than adipocytes (which areuniquely designed to store fat for use as energy during hard times). The other celltypes include those of liver and skeletal mus-cle. The result of excess lipid is an inhibitionof insulin action. The precise identity of the lipid factor responsible is not known,although fatty acyl CoA (a biochemicallyactive fatty acid) and/or diacyl glycerol molecules (one glycerol molecule bound to two fatty acids), acting through a form of

Diabetes

Fat in all the wrong placesJeffrey Friedman

Obesity and a rare, congenital absence of fat cells are associated withdamaging levels of fat in various tissues, and diabetes. Leptin helps toremedy these problems by causing oxidation of fatty acids in mitochondria.

Figure 1 Leptin’s control of fat in skeletal muscle2. In cells, there is a balance between transport offatty acids into mitochondria and their subsequent oxidation, and storage of these compounds astriglycerides in the cytoplasm. This balance is regulated mainly by malonyl CoA, a fatty acid that isgenerated by the enzyme acetyl CoA carboxylase (ACC). Malonyl CoA inhibits transport of fatty acidsinto mitochondria, thereby preventing their oxidation12. Leptin causes the phosphorylation of AMP-activated protein kinase (AMPK), which in turn phosphorylates ACC, inactivating it13. Leptin thusinhibits malonyl CoA synthesis, leading to greater mitochondrial import and consumption of fattyacids. These events seem to result both from the direct action of leptin on skeletal muscle and from itsindirect influence that operates through the hypothalamus.

AMPK

Malonyl CoAAcetyl CoA

ACCCytoplasmicfatty acids

Mitochondria

Oxidationand energyproduction

CO2 + H2O

MuscleFatty acids

ACCp

AMPKp

Leptin

HypothalamusLeptinsignal

© 2002 Macmillan Magazines Ltd

mice) fail to do so (M. Reitman, personalcommunication).

Leptin also improves insulin sensitivity in other settings. In normal animals, itreduces cellular lipid content, increases glu-cose uptake by muscle and increases insulin sensitivity10. In leptin-deficient ob/ob mice, an improvement in the efficacy of insulin isevident at doses that do not affect weight11. In each of these cases, the effects of leptin areevidently not just a consequence of its abilityto reduce food intake1,7.

How, then, does leptin clear lipid fromnon-adipose tissue? This is where Minokoshiet al.2 come in. They first demonstrate thatleptin increases fatty-acid consumption, byoxidation, in skeletal muscle. They also show that leptin activates an enzyme —AMP-activated protein kinase, or AMPK —in skeletal muscle. In cells, a delicate balancecontrols whether fatty acids are transportedinto mitochondria and metabolized or arestored in the cytoplasm as triglycerides (Fig. 1). This balance is mainly regulated by malonyl CoA, a fatty acid that is generatedby the enzyme acetyl CoA carboxylase(ACC). Malonyl CoA inhibits transport offatty acids into mitochondria, thereby preventing them from being metabolized12.AMPK phosphorylates ACC, inactivatingit13. By activating AMPK in muscle, leptininhibits malonyl CoA synthesis and shifts the balance towards fatty-acid oxidation and away from fat storage (Fig. 1). Theseeffects are similar to those seen in mice inwhich the genes that encode ACC have beenknocked out14.

Leptin-mediated phosphorylation ofAMPK seems to operate both directlythrough skeletal muscle and indirectlythrough the hypothalamus, although themechanisms involved are not known. Wenow need to find out whether leptin’s lipid-clearing effects on liver and other tissueswork in the same way as those on skeletalmuscle — interestingly, different forms ofthe ACC gene are present in skeletal muscleand liver14.

All in all, the data suggest that we maynow have an explanation for the associationof obesity, insulin resistance and diabetes.The findings also suggest that leptin may beof benefit not only in obesity but in other settings as well. Some obese subjects areresistant to leptin and do not seem torespond to the genetically engineered formof the protein. However, some obese subjectsdo lose weight in response to leptin therapy,and the hormone could prove useful in a subset of obese patients, especially diabeticones15. Leptin might also be effective in treat-ing human lipodystrophy, a possibility that is now being tested (P. Gorden, personalcommunication). Lipodystrophy is rare as aninherited disorder, but some HIV-infectedpatients develop this condition. Indeed, itmay be that, in both lean and obese leptin-

sensitive individuals, leptin could reduceexcess fat at sites such as the liver and heart,and in beta cells, and prevent the damage it causes.

Finally, there is an evolutionary angle tothese results. Leptin’s emergence in verte-brates may have helped to prevent excessiveweight gain through lipid deposition in adipose tissue in times of surfeit. Excessiveweight can easily be imagined to be dis-advantageous if, for instance, it makes ananimal less able to evade attack. But if lipid in other cells increases the risk of diabetesand other complications of obesity, and leptin acts to reduce this, then this hormonemay also have evolved to limit accumulationof lipid in the wrong places6. ■

Jeffrey Friedman is at The Howard Hughes MedicalInstitute, Rockefeller University, 1230 York Avenue,New York, New York 10021-6399, USA.

e-mail: friedj@mail.rockefeller.edu 1. Friedman, J. M. Harvey Lecture 95, 107–136 (1999–2000).

2. Minokoshi, Y. et al. Nature 415, 339–343 (2002).

3. Randle, P. J. Diabetes Metab. Rev. 14, 263–283 (1998).

4. McGarry, J. D. Science 258, 766–770 (1992).

5. Shulman, G. I. J. Clin. Invest. 106, 171–176 (2000).

6. Unger, R. H., Zhou, Y.-T. & Orci, L. Proc. Natl Acad. Sci. USA

96, 2327–2332 (1999).

7. Shimomura, I., Hammer, R., Ikemoto, S., Brown, M. &

Goldstein, J. Nature 401, 73–76 (1999).

8. Shimabukuro, M. et al. Proc. Natl Acad. Sci. USA 94, 4637–4641

(1997).

9. Shimabukuro, M., Zhou, Y. T., Levi, M. & Unger, R. H. Proc.

Natl Acad. Sci. USA 95, 2498–2502 (1998).

10.Kamohara, S., Burcelin, R., Halaas, J. L., Friedman, J. M. &

Charron, M. J. Nature 389, 374–377 (1997).

11.Pelleymounter, M. A. et al. Science 269, 540–543 (1995).

12.McGarry, J. D., Mannaerts, G. P. & Foster, D. W. J. Clin. Invest.

60, 265–270 (1977).

13.Winder, W. W. & Hardie, D. G. Am. J. Physiol. 277, E1–E10

(1999).

14.Abu-Elheiga, L., Matzuk, M. M., Abo-Hashema, K. A. H. &

Wakil, S. J. Science 291, 2613–2616 (2001).

15.Heymsfield, S. et al. J. Am. Med. Assoc. 282, 1568–1575

(1999).

news and views

NATURE | VOL 415 | 17 JANUARY 2002 | www.nature.com 269

The irresistible cravings of addicts pro-vide an extreme example of short-termbehaviour with adverse long-term con-

sequences. The field of study known asexperimental and behavioural economicsindicates, however, that some of the princi-ples that underpin addictive behaviour arequite common and can help us to under-stand human behaviour more generally.

At a meeting* held late last year, partici-pants discussed how the tools and conceptsof this field can be used to understand, predict and influence people’s decisions incircumstances in which some of the rewardsor costs of a choice accrue in the future. In trade jargon, these are known as ‘inter-temporal choices’, and at some time or otherwe all have to make them. Examples are decisions about savings and the amount ofcredit-card debt we run up, and about eatinghabits and even marriage — all of thesechoices have positive or negative conse-quences that lie in the future.

If animals have to choose repeatedlybetween a smaller reward arriving soon (SRS)and a larger reward arriving later (LRL), theSRS is often preferred even though therepeated choice of the LRL would maximizethe overall gain1–3. One experiment2, forinstance, involved hungry pigeons, with aninitial trial phase lasting 30 seconds and anoutcome phase of 10 seconds. Irrespective

of the pigeons’ choices, each trial of theexperiment lasted 40 seconds. In each trial,the pigeons could press one of two keys 2 seconds before the end of the initial phase.One key led to 2 seconds of access to a grainhopper right at the start of the outcomephase; the other key led to 4 seconds of grain-hopper access after 4 seconds of theoutcome phase. Almost all pigeons preferred the immediate, but more limited, access tothe grain. They thus strongly ‘discounted’the LRL in favour of the SRS — in otherwords, they subjectively devalued the LRLrelative to the SRS.

The pattern of discounting also gives rise to preference reversals, because animals are very ‘impatient’ in the short term andrelatively ‘patient’ in the long run1–3. If, for example, the choice between the twokeys had to be made after 2 seconds of theinitial phase, leaving a 28-second wait untilthe beginning of the outcome phase, thepigeons preferred the LRL in more than 80%of cases. So if both the SRS and the LRL wereshifted into the future, the animals changedtheir preferences in favour of the LRL,despite the fact that the absolute time differ-ence between the two options remainedunchanged. Their behaviour was thus notconsistent over time.

One explanation for this is that, through-out evolutionary history, future rewardshave been uncertain. An animal foraging for food may be interrupted or, in the case of reproductive opportunities, die before it

Behavioural science

The economics of impatienceErnst Fehr

In experiments, animals often prefer smaller, immediate rewards overlarger rewards that are deferred — thus failing to maximize their total gain. Many people exhibit similar behaviour.

*Nobel Symposium on Experimental and Behavioural Economics,

Saltsjöbaden, Sweden, 4–6 December 2001.

© 2002 Macmillan Magazines Ltd