The Synthesis and Evolution of a Supermodel

1890

designed to address. Examples are a newestimate of the flux of meteors entering theatmosphere (9), the idea that mixedorganic/salt particles could have had a rolein the origin of life (14), and the tracking ofsmoke from fireworks (15). Yet public pol-icy questions continue to provide a focus tothe researchers involved in developinginstrumentation to measure the composi-tion of particles in the atmosphere.

References1. International Panel on Climate Change, Climate

Change 2001, The Scientific Basis (Cambridge Univ.Press, Cambridge, UK, 2001).

2. D.W. Dockery et al., N. Engl. J. Med. 329, 1753 (1993).3. P. H. McMurry, Atmos. Environ. 34, 1959 (2000).4. J.T. Jayne et al., Aerosol Sci. Technol. 33, 49 (2000).5. C. A. Noble, K. A. Prather, Mass Spectrom. Rev. 19, 248

(2000).6. P. Liu et al., Aerosol Sci. Technol. 22, 314 (1995).7. R.Weber et al., J. Geophys. Res. 108, 8421 (2003).8. T. Novakov, D.A. Hegg, P.V. Hobbs, J. Geophys. Res. 102,

30023 (1997).

9. D. M. Murphy, D. S. Thomson, M. J. Maloney, Science282, 1664 (1998).

10. J. F. Hamilton et al., Atmos. Chem. Phys. 4, 1279 (2004).11. M. P. Tolocka et al., Environ. Sci. Technol. 38, 1428

(2004).12. M. Kalberer et al., Science 303, 1659 (2004).13. C. Marcolli, B. Luo, T. Peter, J. Phys. Chem. A. 108, 2216

(2004).14. C. M. Dobson, G. B. Ellison,A. F.Tuck,V.Vaida, Proc. Natl.

Acad. Sci. U.S.A. 97, 11864 (2000).15. D.Y. Liu, D. Rutherford, M. Kinsey, K. A. Prather, Anal.

Chem. 69, 1808 (1997).

10.1126/science.1108160

Synthesis is the merging of disparatesources of knowledge to create astronger, more compelling whole. In

the biological sciences, barriers to synthe-sis—including the specialization of subdis-ciplines and their fractionation into depart-ments and curricula—have increased overthe past few decades. Developmentalistsdissect their favorite ecologically irrelevantmodels with exquisite detail, while evolu-tionists tap away at hundreds of fascinatingspecies with genetically toothless tools.Organismal biologists can take great solacein a report on page 1928 of this issue byColosimo and colleagues (1) that showshow genomics can be used to buck thistrend and lead us to new insights into fun-damental evolutionary problems.

The threespine stickleback (Gaster-osteus aculeatus) is a finger-sized speciesof fish that exhibits multiple examples ofparallel evolution. The threespine stickle-back populations that inhabit the streamsand lakes of the northern Pacif ic andAtlantic rims show intriguing variations inmorphology and behavior compared tomarine populations (2). A particularly strik-ing instance is the reduction in body armorexhibited by freshwater populations.Whereas marine sticklebacks carry a row ofup to 36 armor plates extending from headto tail (complete morph), freshwater stick-lebacks either carry a gap in the row ofplates (partial morph) or retain only a fewplates at their anterior end (low morph).Sticklebacks reside in diverse freshwaterhabitats that include numerous glacial lakesin western Canada that were formed as thelast ice age retreated 10,000 or so years ago(3). Lindsey hypothesized in 1962 (4) that

rather than a single stickleback populationwith reduced body armor founding the pop-ulations in all of these lakes, parallel evolu-tion must have occurred. He attributed par-allel evolution to selection either on inde-pendent mutations or on a rare allele whosephenotypic effect is cryptic, that is, remainshidden in the marine population. The newstudy (1) combines quantitative genetics,genomics, population genetics, molecularevolution, f ield studies, and moleculardevelopmental biology to demonstrate thatboth new mutations and cryptic variationhave contributed to body armor reduction.In so doing, this study provides one of thefirst dissections of a skeletal polymorphismto the gene level, and thereby elevates the

stickleback to the status of supermodel forthe study of developmental evolution.

The story begins with high-resolutionlinkage mapping of a major effect locus forarmor reduction in a cross between “com-plete” and “low” body plate morphs of thethreespine stickleback. This locus wasmapped to an approximately 0.7-cM inter-val of the genome. Although expression ofthe phenotype varies in different crossesbecause of the segregation of modifier loci,loss of body plates is largely dependent on agenerally recessive allele that accounts foras much as 75% of the difference betweenmorphs (5). To positionally clone the generesponsible for this effect, Colosimo et al.(1) performed a chromosome walk acrossthe region of interest, tiling six bacterialartif icial chromosome clones coveringmore than a megabase. Half of this walkwas completely sequenced, and microsatel-lites at 12-kb intervals were typed in a set of46 of the complete-armor morphs and 45 ofthe low-plate-number morphs from aninterbreeding stream population inCalifornia. This so-called linkage disequi-

E V O L U T I O N

The Synthesis and Evolutionof a Supermodel

Greg Gibson

The author is in the Department of Genetics, NorthCarolina State University, Raleigh, NC 27695, USA. E-mail: [email protected]

Lightening the load. Marine threespine sticklebacks (blue) have a robust set of body armor (indi-cated as multiple rays in each body), whereas multiple freshwater populations on either side of thenorth Pacific rim have independently lost their body armor during the course of evolution. Each low-morph stickleback population in lakes and streams of western North America carries an Eda allele(red coloration) that resembles the rare allele found in marine populations.This finding suggests thatthe Eda allele has increased in frequency under adaptive selection. By contrast, a Japanese marinepopulation has a different Eda allele on the common background, implying that this case of armorreduction evolved through an independent mutation. C

RED

IT:P

REST

ON

HU

EY/ S

CIE

NC

E

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

25 MARCH 2005 VOL 307 SCIENCE www.sciencemag.orgPublished by AAAS

1891

librium approach—now the standard inmapping human disease genes—reducedthe peak of the candidate interval to just 16kb, centered on a marker in the secondintron of a gene called Ectodysplasin (Eda).

Eda encodes a member of the tumornecrosis factor (TNF) family of secretedsignaling proteins (6). The gene has a his-tory of involvement in abnormal develop-ment of skin. Mutations in this gene cause avariety of human syndromes (7) and theTabby phenotype in mice (8); meanwhile,absence of the EDA receptor results in lossof scales in medaka f ish (9). Similarlysevere mutations in these genes in naturalpopulations would be likely to have suchdeleterious pleiotropic effects that selectionwould preclude their reaching high frequen-cies. Sequencing of a low-morph Eda alleledetected multiple nucleotide differenceswith respect to the complete-morph allele,but only four of these led to an amino acidchange in the EDA protein, and none ofthese changes obviously affected EDA’sfunction (1). Thus, the precise polymor-phism that leads to the quantitative skeletalphenotype is still unknown. Reverse tran-scription polymerase chain reaction meth-ods detected the expected alternative spliceproducts in both morphs of stickleback, butbecause the transcript abundance in devel-oping epidermis was too low to detect bywhole-mount in situ hybridization, it is notyet clear whether there is a quantitative orspatial difference in expression that mightexplain the phenotypic effect of the low-morph allele.

It remains possible that Eda is not thecausative gene, particularly because therefined interval includes another TNF ligandand two other genes of interest (these mayalso, or alternatively, influence a couple ofcorrelated physiological traits). Kingsley’sgroup (1) has established more direct evi-dence for the involvement of EDA in platedevelopment by generating transgenic stick-lebacks that transiently overexpress a murineversion of the gene. A handful of these fishexhibit partial rescue of plate development ina homozygous low-morph background.Thus, there is little doubt that Colosimo andco-workers have nailed the source of parallelmorphological evolution in the threespinestickleback to a single gene.

The authors next performed a populationsurvey of Eda genotypes to address the ques-tion of whether adaptive evolution hasfavored multiple independent mutations orrepeated selection of an allele that is rare inmarine populations (see the figure). All of theNorth American and European low-morphpopulations share a haplotype consisting of aset of single-nucleotide polymorphisms cov-ering the interval centered on Eda, whereasthe Japanese low-morph allele is clearly dis-

tinct. Because the Japanese allele fails tocomplement the North American one, thisimplies that at least two independent muta-tions have led to a reduction in body armor oneither side of the Pacific. By contrast, detec-tion of at least 14 instances of a similar haplo-type in North America—one that is rare in themarine population from which the stream andlake populations of Canada and Californiawere founded—strongly implies a singlegenetic basis for these instances of parallelevolution.

Counting up the number of nucleotidesubstitutions between the two sequencedhigh- and low-morph alleles suggests a dateof 2 million years for their separation,which is two orders of magnitude longerthan the inferred age of the postglacial pop-ulations. This method does not actually datethe causative mutation, which could havearisen on the low-morph haplotype at anytime before the founding of the freshwaterpopulations in which it has risen to a highfrequency. The haplotype has an estimatedallele frequency of 0.6 to 3.8% in popula-tions of marine fish sampled from coastalBritish Columbia and California. This is toolow to produce an appreciable number ofhomozygotes, but large enough to create apool of alleles that would be available forselection upon introgression into freshwaterenvironments. Preliminary sequence com-parisons suggest that the causal site itself isquite ancient, as there is considerable diver-sity even within the low-morph haplotype.Had the mutation appeared relativelyrecently in an isolated stream, then found itsway back into the marine pool from whencethe other populations were founded,sequence diversity in the haplotype wouldbe much lower than that observed in theprevalent marine haplotype.

Ancient or recent, the more importantpoint is that for the f irst time we have aclear demonstration that after alteration ofenvironmental circumstances, adaptiveevolution can act independently on anallele that is present in but has little effecton the morphology of the ancestral species.Colosimo et al. (1) refer to this phenome-non as selection on cryptic variation. Amore technical definition of cryptic varia-tion allows us to expand the scope of thepotential impact of standing variation onrapid morphological evolution (10). Inmarine sticklebacks, the low-morph effectis hidden by the fact that only rare f ishcarry the relevant allele, and these fish areheterozygous rather than homozygous forthe low-morph variant. Strictly speaking,though, cryptic variation refers to the situa-tion where the phenotype of individuals ismodif ied by the genetic background orenvironment such that a previously neutralvariant becomes functional and adaptive.

Because plate reduction is modif ied byother loci (5) and may be influenced byenvironmental factors such as calcium con-centration in the water, it is likely that morethan just selection on Eda is contributing tothe uncovering of this hidden variation.

These results should provide fuel forthose who wish to emphasize the distinctionbetween soft and hard selection. Hard selec-tion—positive selection on new muta-tions—is known to lead to a substantialreduction in nucleotide diversity around thefocal site, which can be used as the basis fordetection of selective sweeps (11). By con-trast, soft selection acts on standing varia-tion that has been in the population for sometime, as a result of a change in the environ-ment or genetic background (both of whichshould occur when marine sticklebacksadmix with those inhabiting streams). It isexpected to leave a very different geneticfootprint (12, 13), and this system providesa superb opportunity to contrast these sce-narios. There is also a lovely symmetry inthe fact that the mouse Tabby mutation, nowknown to be due to a mutation in the Edagene (8), provides a classic example ofcanalization (14). Canalization refers to thebuffering of genetic variation and promotesthe maintenance of cryptic variation.

Studies such as that by Colosimo andcolleagues highlight how the disparatebranches of biology can be synthesized toprovide fresh perspectives on fundamentalevolutionary phenomena. The NationalEvolutionary Synthesis Center (15) inDurham, North Carolina, has just receivedNSF funding to promote synthetic research.Few groups have the capacity to pull offsuch an integrative accomplishment, butthere is little reason why interactive teamscannot contribute to the emergence ofnumerous other supermodel organisms.

References1. P. F. Colosimo et al., Science 307, 1928 (2005).2. M.A. Bell, S.A. Foster, Eds., The Evolutionary Biology of

the Threespine Stickleback (Oxford Univ. Press,Oxford, 1994).

3. G. Orti, M. A. Bell, T. E. Reimchen, A. Meyer, Evolution48, 608 (1994).

4. C. C. Lindsey, Can. J. Zool. 40, 271 (1962).5. P. F. Colosimo et al., PLoS Biol. 2, E109 (2004).6. A. T. Kangas, A. R. Evans, I. Thesleff, J. Jernvall, Nature432, 211 (2004).

7. K. Paakkonen et al., Hum. Mutat. 17, 349 (2001).8. A. K. Srivastava et al., Proc. Natl. Acad. Sci. U.S.A. 94,

13069 (1997).9. S. Kondo et al., Curr. Biol. 11, 1202 (2001).

10. G. Gibson, I. M. Dworkin, Nature Rev. Genet. 5, 681(2004).

11. N. L. Kaplan, R. R. Hudson, C. H. Langley, Genetics 123,887 (1989).

12. H. Innan, Y. Kim, Proc. Natl. Acad. Sci. U.S.A. 101,10667 (2004).

13. J. Hermisson, P. S. Plennings, Genetics, in press.14. R. B. Dun,A. S. Fraser, Nature 181, 1018 (1958).15. National Evolutionary Synthesis Center (www.

nescent.org).

10.1126/science.1109835

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

www.sciencemag.org SCIENCE VOL 307 25 MARCH 2005Published by AAAS