1911

divergence dates not falling well within thePrecambrian.

Is the conclusion drawn by Rokas et al.sound? At first glance, it appears so, butcan their conclusion stand up to closerscrutiny? Accepting that the genes ana-lyzed by the authors evolved without geneduplication and that the amino acids arealigned correctly, most phylogenetic meth-ods assume that the evolutionary dynamicsof the 12,060 amino acid sites are inde-pendently and identically distributed, andthat they evolved under the same station-ary, reversible, and homogeneous condi-tions (8). The assumptions arise from theneed to render phylogenetic methodstractable and easy to use, and they areunlikely to be realistic. To account for theobservation that the sites in a gene mayevolve at different rates, some phyloge-netic methods are able to model rate het-erogeneity across sites using a Γ distribu-tion (9). Rokas et al. used this approach forthe whole alignment but did not considerthat different parts of the alignment mayrequire different Γ distributions. Nor didthey consider that some sites may varynonindependently (10) and that the distri-bution of variable sites may vary acrosslineages and through time, an issue that isnotoriously difficult to resolve (11). A logicalextension to the work would be to partitionthe alignment and estimate the evolutionaryrates for different genes separately.

Violation of the assumed stationary,reversible, and homogeneous conditionsmay lead to compositional differences in thealigned amino acid sequences and hence toerrors in phylogenetic estimates (12). Rokaset al. recognized this potential source oferror but used a test that is known to beflawed, even though better tests are known(13). Furthermore, they chose a phyloge-netic method that, while it accounts for com-positional variation in the sequence align-ment, is unsuitable: It assumes that the sitesare independently and identically distrib-uted, which they have already shown not tobe the case. Moreover, they used a singleMarkov (probabilistic) model to analyze thealignment of amino acids, in effect using a“one size fits all” approach, where it wouldhave been better to use several Markov mod-els to capture gene-specific differences inthe evolutionary processes (14).

Rokas et al. used nonparametric boot-strap and posterior probabilities to gaugesupport for the pattern and order of specia-tion events (branches in their phylogenetictree). The former is widely recognized to bestatistically unwise. Bootstrap values areestimates of the expected frequency withwhich speciation events (internal branches)occur in the optimal tree, using data con-structed from the original alignment by sam-

pling sites with replacement (15). It is not ameasure of accuracy or confidence, but ofdata consistency. Further, the increase inbootstrap value when more genes areincluded may be misleading, because longersequences naturally tend to have higherbootstrap values (see the figure). The poste-rior probabilities of speciation events beingcorrectly identified are also prone to errorwhen the phylogenetic assumptions are vio-lated in the sense described above.

In light of these concerns, are the conclu-sions of Rokas et al. justified? Should weignore their study? Most certainly not,because they have produced a wealth of dataand have shown that it might just be possiblethat the fossil record can be reconciled withmolecular data. This, in itself, should because for celebration and an incentive toacquire sequence data from the remaining 26animal phyla. Likewise, it should encouragedevelopment of methods that assess whendata violate phylogenetic assumptions, andthat cope with such data. To achieve thesegoals, we need to know more about the struc-ture and function of gene products before wecan develop models that appropriatelyaddress the early evolution of animals.

References

1. C. Darwin, The Origin of Species by Means of NaturalSelection. (John Murray, London, ed. 6, 1888), p. 313.

2. D. E. G. Briggs, R. A. Fortey, Paleobiology 32, S94(2005).

3. S. J. Gould, Wonderful Life: The Burgess Shale and theNature of History (Hutchinson Radius, London, 1989).

4. E. J. Douzery, E. A. Snell, E. Bapteste, F. Delsuc, H.Philippe, Proc. Natl. Acad. Sci. U.S.A. 101, 15386(2004).

5. K. J. Peterson et al., Proc. Natl. Acad. Sci. U.S.A. 101,6536 (2004).

6. K. J. Peterson, N. J. Butterfield, Proc. Natl. Acad. Sci.U.S.A. 102, 9547 (2005).

7. A. Rokas, D. Krüger, S. B. Carroll, Science 310, 1933(2005).

8. V. Jayaswal, L.S. Jermiin, J.Robinson,Evol. BioinformaticsOnline 1, 62 (2005).

9. Z.Yang, Trends Ecol. Evol. 11, 367 (1996).10. D. Penny, B. J. McComish, M.A. Charleston, M. D. Hendy,

J. Mol. Evol. 53, 711 (2001).11. P. J. Lockhart et al., Mol. Biol. Evol. 23, 40 (2006).12. S.Y.W. Ho, L. S. Jermiin, Syst. Biol. 53, 623 (2004).13. L. S. Jermiin, S.Y.W. Ho, F.Ababneh, J. Robinson,A. D.W.

Larkum, Syst. Biol. 53, 638 (2004).14. M. Pagel,A. Meade, Syst. Biol. 53, 571 (2004).15. J. Felsenstein, Evolution 39, 783 (1985).16. J. W. Valentine, D. Jablonski, D. H. Erwin, Development126, 851 (1999).

17. A. Rambaut, N. C. Grassly, Comp. Appl. Biosci. 13, 235(1997).

18. D. L. Swofford, Phylogenetic Analysis Using Parsimony(*and Other Methods), 4.0 Beta (Sinauer, Sunderland,MA, 2002).

10.1126/science.1122440

www.sciencemag.org SCIENCE VOL 310 23 DECEMBER 2005

P E R S P E C T I V E S

The relationship between organismaldevelopment and aging has long beena matter of intense debate. It seems

natural to posit that developmental timingmechanisms that culminate in reproductivematurity continue to affect post-reproduc-tive biology, with consequences for totalorganism life span. On the other hand, evo-lutionary theories of aging discount regu-lated aging per se, because the force ofselection declines with age and drops pre-cipitously after reproductive potential ends.In other words, a program that actively agesthe organism is unlikely to be selected for inevolution. Instead, aging is thought to entailthe passive stochastic accumulation ofdamage to molecules, cells, and organs,leading to loss of fertility and organismaldemise. Therefore, the notion that regulatedintrinsic biological timers control agingseems superficially untenable.

Just this possibility, however, has beenraised by Boehm and Slack on page 1954 ofthis issue (1). The have found that compo-

nents of a nematode’s (Caenorhabditis ele-gans) heterochronic circuit—namely thelin-4 microRNA and its target, the nuclearprotein encoded by lin-14—not only perturbdevelopmental timing but also influenceorganismal life span. They do so by regulat-ing insulin/IGF-1 (insulin-like growth fac-tor–1) signaling, a cellular regulatory path-way whose modest decrease in activity leadsto increased longevity across taxa (2).

Just as each cell in a developing organismhas a positional identity that is determinedby gradients of morphogens and hierarchiesof transcription factor activity, cells alsohave a temporal identity dictated by regula-tory signaling cascades. Pioneering work inC. elegans led to the discovery of the hete-rochronic loci (3), which constitute a regula-tory circuit that confers temporal identity tothe various tissues. These genes determinecellular programs of division, migration, anddifferentiation that are appropriate for a spe-cific developmental stage. In addition, theinteractions among these heterochronic lociensure the proper succession of larval tem-poral fates. Normally, C. elegans develops toadulthood through four larval stages, L1 toL4. Worms with mutations in the het-erochronic loci inappropriately express cel-

P H Y S I O L O G Y

The Tick-Tock of Aging?Adam Antebi

The author is at the Huffington Center on Aging,Department of Molecular and Cellular Biology, BaylorCollege of Medicine, Houston, TX 77030, USA. E-mail:aantebi@bcm.tmc.edu

Published by AAAS

on August 2, 2020

http://science.sciencem

ag.org/D

ownloaded from

1912

lular programs at thewrong stage that couldresult, for example, inthe expression of adultfeatures in the juvenileor conversely, juvenilefeatures in the adult. Amolecular analysis hasrevealed that most het-erochronic loci are evo-lutionarily conserved.Perhaps most strikingare the examples of lin-4and let-7 microRNAs,short 21- to 24-nucleo-tide RNAs that post-transcriptionally regu-late gene expression(4–6). First discoveredin the worm, orthologsare found conservedacross species, includ-ing human (7). Thisspawned the discoveryof large families of simi-lar molecules whosediverse functions are only just beginning tobe explored. As part of an early larval timer,a rise in lin-4 microRNA expression triggerslarval stage L2 and later developmental pro-grams by decreasing the expression of amolecular target, lin-14. It does so by bind-ing with imperfect complementarity tosequences in the 3´-untranslated region ofthe messenger RNA (mRNA) that encodeslin-14, inhibiting translation and mRNA sta-bility (4, 5). A decrease in the expression oflin-14 mRNA and protein is seen in thedeveloping worm, but expression persists insome tissues in the adult.

However, the role of lin-14 and otherheterochronic loci in the adult worm hasbeen little explored, largely because cells ofthe adult are postmitotic (except for germ-line cells) and therefore do not display anyovert stage-specif ic cellular programs.Clearly, though, adult animals undergogermline maturation, growth, homeostasis,and metabolic changes, as well as seem-ingly coordinated shifts in gene expression(8). Conceivably, such changes could arisefrom intrinsic timing mechanisms at workin the adult. Hence, Boehm and Slacksought to test the hypothesis that the het-erochronic loci influence adult longevity.

Remarkably, they found that overex-pression of lin-4 microRNA results in amodest increase in adult life span (by 15%),and conversely, that loss of a functionallin-4 gene shortens life (by 53%). Theseadult phenotypes, as in larvae, depend onlin-14. Worms lacking a functional lin-14gene (loss-of-function mutants) were long-lived (by 28%), whereas those that overexpresslin-14 (gain-of-function mutants) were short-

lived (by 63%). In worms lacking both lin-4 and lin-14, the long-lived lin-14 pheno-type largely prevailed (a life span 14%longer than wild type), suggesting that lin-4 works at least in part through lin-14.Similarly, the developmental phenotypes oflin-14 mutants (precocious) prevail overlin-4 (delayed) in the lin-4 lin-14 doublemutant worms. The fact that the sameepistatic relations hold for lin-4 and lin-14in both the developing larva and agingadult supports the idea that a similar regu-latory pathway is at work.

The question then arises as to whetherlongevity is determined by developmentalevents or adult events. To address this,Boehm and Slack took advantage of condi-tional manipulation of lin-14 expression.By using a temperature-sensitive allele,they bypassed developmental defects andperformed shifts to the nonpermissive tem-perature that caused a decrease in lin-14expression in the adult. Interestingly, post-developmental temperature shifts stillcaused extended life span. Accordingly,decreasing lin-14 expression by RNA inter-ference in adults also extends life, whichsuggests a function independent of develop-ment and specific for adults. As correlatesof longevity, lin-14 loss-of-functionmutants were found to be more resistant toheat stress and slower to accumulate lipo-fuscin, a marker of aged tissues.

Molecular genetic studies first identi-fied insulin/IGF-1 signaling as a key modu-lator of nematode life span (9, 10).Remarkably, the same was later shown to betrue for flies and mice (2). Specif ically,decreased insulin/IGF-1 signaling results in

the nuclear translocation of a transcriptionfactor called DAF-16/FOXO. This tran-scription factor turns on genes for stressresistance, DNA repair, innate immunity,and heat shock, and, as a consequence, theworm’s life span is doubled.

Given the central role of insulin/IGF-1signaling in aging, Boehm and Slack askedwhether lin-4 and lin-14 impinged on thispathway. Indeed, they found that wormswith a mutation in daf-16 as well as inhsf-1, a longevity gene encoding a tran-scription factor for turning on heat shockproteins, abolished longevity mediated bylin-14. Moreover, longevity of a wormwith a mutation in daf-2, the gene encod-ing the insulin-like receptor, was not fur-ther increased by a loss-of-function muta-tion in lin-14. Finally, a short-lived lin-4mutation largely suppressed the longevityof the daf-2/insulin-like receptor mutant,placing lin-4 activity downstream or paral-lel to that of the receptor. These geneticstudies argue that lin-4 and lin-14 couldsomehow regulate daf-16 via insulin/IGF-1signal transduction or a parallel signalingpathway (see the figure).

The union of signaling pathways thatcontrol developmental timing and life spanis not without precedent, because the wormnuclear hormone receptor DAF-12 oper-ates in both (11). However, if a conservedmicroRNA and its target converge on DAF-16/FOXO to influence adult longevity, thisraises the intriguing notion that intrinsicdevelopmental clocks can modulate aging.Alternatively, lin-4 and lin-14 may haveundescribed metabolic outputs somewhatindependent of a timer. Notably, the hete-rochronic circuit is initialized by food cues,with nutrient inputs at distinct points of L1and L3 diapause, periods of arresteddevelopment entered under conditions ofstarvation. In particular, both lin-4 anddaf-16 are required for entry into the L3dauer diapause, which is a long-livedstress-resistant stage.

In either scenario, the results of Boehmand Slack raise a myriad of questions. Howdo lin-4 and lin-14 converge on daf-16?What tissues are involved? Is this a mecha-nism of regulation observed in the wild-typeworms under some conditions? What othersignaling pathway components are involved?Do microRNAs regulate longevity in higheranimals? If so, what are they and what aretheir targets?

Finally, how might the paradox of intrin-sic timers and “regulated” aging be recon-ciled with the evolutionary theories ofaging? One possibility is that the tempo ofreproductive development needs to be coor-dinated between the tissues, as well as withnutrient availability and the environment.Under conditions of adversity, regulatory

23 DECEMBER 2005 VOL 310 SCIENCE www.sciencemag.org

Environmental and physiological signals

Short life

Low

High

High

Low

Long life

DAF-2

High

Low

Low

High

lin-14 mRNA

lin-4 microRNA

DAF-16

LIN-14 DAF-2

DAF-16

LIN-14

Regulation of adult life span by the lin-4 microRNA. (Left) When lin-4 microRNA activity is high, expression of lin-14 mRNA and protein arelow. Hence, the DAF-16 transcription factor is active and promotes longlife. (Right) When lin-4 activity is low, lin-14 activity is high, and DAF-16is inhibited, resulting in short life. lin-4 and lin-14 gene products maywork downstream of, or in parallel to, DAF-2 (the insulin-like receptor) tomodulate DAF-16. Proteins are depicted as oval shapes.

P E R S P E C T I V E S

Published by AAAS

on August 2, 2020

http://science.sciencem

ag.org/D

ownloaded from

1913

signaling pathways that delay reproductionand increase somatic endurance couldadaptively retard organismal decline. Whenexercised in the adult, such mechanismscould secondarily extend life. Anothernotion is that developmental timing mecha-nisms that determine a species’ life planmay somehow influence the life span.Perhaps the great natural variation in animal

life spans is determined by such global tem-poral regulators.

References1. M. Boehm, F. Slack, Science 310, 1954 (2005).2. C. Kenyon, Cell 120, 449 (2005).3. V. Ambros, H. R. Horvitz, Science 226, 409 (1984).4. R. C. Lee, R. L. Feinbaum, V. Ambros, Cell 75, 843

(1993).5. B.Wightman, I. Ha, G. Ruvkun, Cell 75, 855 (1993).

6. B. J. Reinhart et al., Nature 403, 901 (2000).7. A. E. Pasquinelli et al., Nature 408, 86 (2000).8. S.A. McCarroll et al., Nat. Genet. 36, 197 (2004).9. C. Kenyon, J. Chang, E. Gensch, A. Rudner, R. Tabtiang,

Nature 366, 461 (1993).10. K. D. Kimura, H. A. Tissenbaum, Y. Liu, G. Ruvkun,

Science 277, 942 (1997).11. B. Gerisch, C. Weitzel, C. Kober-Eisermann, V. Rottiers,

A.Antebi, Dev. Cell 1, 841 (2001).

10.1126/science.1122816

CR

EDIT

:P.H

UEY

/SC

IEN

CE

www.sciencemag.org SCIENCE VOL 310 23 DECEMBER 2005

P E R S P E C T I V E S

Molecules with identical nuclei hav-ing nonzero spin can exist in differ-ent states called nuclear spin modi-

fications by most researchers and nuclearspin isomers by some. Once prepared in aparticular state, the arrangement of spinscan in principle convert to another arrange-ment. The concept of nuclear spin modifi-cations in molecules (1) is intriguing tochemists because interconversion betweendifferent spin states is often slow enough tobe considered negligible for many pur-poses. This in turn allows one to treat dif-ferent spin modifications as though theywere different molecules both in terms ofspectroscopy and collision dynamics. Inaddition, these spin modifications are asso-ciated one-to-one with different rotationallevels in the molecule. In fact, however, theinterconversion rates are not zero, so thescientific questions become: How can theseextremely slow rates be measured, and howcan these rates be explained by theory? Onpage 1938 of this issue, Sun et al. (2) pre-sent answers to both of these questions inthe case of the ethylene molecule.

In the case of hydrogen, it is well knownthat rotational levels with even and odd Jquantum number belong to the para (wherethe nuclear spins are aligned in oppositedirections and I = 0) and ortho (the spins areparallel and I = 1) nuclear spin modifica-tions, respectively (I is the total nuclear spinangular momentum of the two H nuclei).Furthermore, nearly pure para-H2 can bereadily prepared by cooling hydrogen tolow temperature on a paramagnetic catalyst

as initially shown by Bonhoeffer andHarteck in 1929 (3). Once prepared, a para-H2 sample can be preserved for months atroom temperature in a glass container with-out converting to ortho-H2. This remark-able stability of a single spin modificationwas ascribed by Wigner to the smallness ofthe nuclear spin interaction term that mixesortho and para states (4).

Spin modifications are relevant also forpolyatomic molecules with two or moreidentical and equivalent nuclei. Thus, forexample, H2O, H2CO, etc., have ortho (I = 1)and para (I = 0) species; NH3, CH3F, etc.,

have ortho (I = 3/2) and para (I = 1/2)species; and CH4 has ortho (I = 1), meta (I =2), and para (I = 0) species. More compli-cated molecules such as C2H4 have morethan three spin modif ications, and theirsymmetry is used to label different spinmodifications, as reported by Sun et al. (2).Polyatomic molecules have faster intercon-version rates than H2, not because theirmagnetic interactions are larger butbecause their rotational levels are closer.This is particularly true for spherical-topmolecules in which levels are clustered andsome levels with different spin modifica-tions are very close. Thus, Ozier et al.observed a spectrum between ortho and

para CH4, by shifting an ortho level close toa para level with a magnetic f ield, anddetermined the small mixing term (5). For aheavy spherical top like SF6 with very highrotational level (J = 53), the levels are soclose that Bordé and colleagues naturallyobserved transitions between different spinspecies (6).

The theory for spin conversion in poly-atomic molecules by collision was first for-mulated by Curl et al., who enumeratedmixing terms in the spirit of Wigner’s the-ory—that is, choosing those nuclear spininteraction terms that are invariant withrespect to exchange of the entire sets ofnuclear coordinates but are not invariant toexchange of spin coordinates alone (7).They identified levels belonging to differ-ent spin modifications that are accidentallyclose and surmised that interconversionoccurs through those pairs of levels. Thevalidity of the theory of Curl et al. has beendemonstrated by a series of beautiful exper-

iments by Chapovsky and his collaboratorssince 1980. They introduced the method oflight-induced drift developed in the formerSoviet Union (see the f igure) and suc-ceeded in separating ortho and para speciesof CH3F, the first such separation since H2

and D2 were separated several decades ago.[Readers are referred to an excellent reviewby Chapovsky and Hermans (8) for moredetails of history, experiments, and theoryof the field.] This is the technique used bySun et al. in their studies.

The exciting aspect of the experiment byTakagi and colleagues. (2) is that ethylene,C2H4, has symmetry with higher dimensionthan that of other molecules so far studied,

C H E M I S T RY

Nuclear Spin Conversion

in MoleculesJon T. Hougen and Takeshi Oka

J. T. Hougen is in the Optical Technology Division,National Institute of Standards and Technology,Gaithersburg , MD 20899–8441, USA. E-mail:jon.hougen@nist.gov T. Oka is in the Department ofChemistry and Department of Astronomy andAstrophysics, The Enrico Fermi Institute, Universityof Chicago, Chicago, IL 60637, USA. E-mail: t-oka@uchicago.edu

Laser

Getting the drift. Laser light tuned slightly below resonance excites a molecule moving away fromthe laser (red sphere on the right) and increases its size. The molecule is slowed down by collisionswith buffer gas because of its larger size. A molecule in the same level moving toward the laser (redsphere on the left), however, is not excited and keeps moving with the normal speed.This causes a netdrift toward the laser of molecules with selected spin modification.The increase of the molecular sizeis greatly exaggerated for clarity. [Adapted from (8)]

Published by AAAS

on August 2, 2020

http://science.sciencem

ag.org/D

ownloaded from

The Tick-Tock of Aging?Adam Antebi

DOI: 10.1126/science.1122816 (5756), 1911-1913.310Science 

ARTICLE TOOLS http://science.sciencemag.org/content/310/5756/1911

CONTENTRELATED

http://stke.sciencemag.org/content/sigtrans/2006/316/tw465.abstracthttp://science.sciencemag.org/content/sci/310/5756/1954.full

REFERENCES

http://science.sciencemag.org/content/310/5756/1911#BIBLThis article cites 11 articles, 3 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

© 2005 American Association for the Advancement of Science

on August 2, 2020

http://science.sciencem

ag.org/D

ownloaded from