Evolution: Mirror, Mirror in the Pond

Identification of mutations in an FGF receptor gene underlying scale loss ina zebrafish mutant, as well as the domesticated mirror carp breed, emphasisesthe role of genetic redundancy as a facilitator of evolutionary change and hasimplications for the molecular basis of morphological evolution.

Current Biology Vol 19 No 19R902

Florian Maderspacher

‘‘.the Fish was a shifting and shiningcreature that nobody had ever caughtbut that many said they had glimpsedin the depths of mirrors.’’Jorge Luis Borges, The Book ofImaginary Beings

When Benson died, in July this year,she had survived 63 near-deathexperiences — quite impressive evenfor someone living in rural England [1].Benson was a common carp (Cyprinuscarpio) of legendary size, weighing inat nearly 30 kg, and an incarnatedproof of what human domesticationcan do to an animal’s body. You mayhave never eaten a carp, yet carps —there are a handful of domestic carpforms — contribute over a third of theworld’s aquaculture production.Carps were domesticated in Europeby the Romans who had raised themin ponds known as ‘piscinae’, andthey were most likely domesticatedindependently in China wherenowadays most carps (andcarp-eaters) live. During the middleages, carp cultivation became hugelypopular in Europe — not least becausenearly every other day was a day offasting, on which a meat replacementwas needed: think of carp as a kindof medieval tofu. In the past twocenturies, carps were bred forprettiness in Japan as the so-callednishikigoi, or koi (Figure 1) [2,3]. Thewild carp used to live in rivers acrossEurasia, but is now on the verge ofextinction — a fate it shares with thewild ancestors of many domesticatedanimals. Two obvious bodily featuresdistinguish the domesticated carpfrom its wild ancestor: thedomesticated form is much stockierthan the torpedo-like wild carp (thatway, they fit more neatly on a plate);and, many domesticated carp havelost their scales (making kitchenhandling easier). A paper in this issue ofCurrent Biology [4] now identifies themolecular basis for this scale loss and,by extension, bears on fundamentalprinciples of the evolution ofmorphological diversity.

Four different morphs of thedomestic carp have been defined:apart from the ‘scaled carp’, whichlooks much like the wild carp, the‘mirror carp’, which lacks most scalesand retains only a few enlarged ones,the ‘line carp’ with a single line of scalesalong its flanks and the ‘leather carp’which is totally devoid of scales.Classic genetics have suggested thatthese phenotypes are controlled by twoloci [5]. Now, Rohner et al. [4] show thatscale loss in mirror carp (and otherscaleless variants) is caused by loss offunction mutations in the gene codingfor a receptor for the FGF growthfactor, called fgfr1a1. Mirror carps fromtwo different populations were found toharbour two different mutations —a small deletion and a missensemutation — in the coding region offgfr1.

The story of how fgfr1’s role in scaleloss was discovered accounts forsome of the appeal of the paper [4].By now, a handful of genes underlyingmorphological change — either duringnatural evolution or during humandomestication — have been identified:notable examples are the genesunderlying the domestication of maizefrom teosinte [6], genes drivingevolution of pigmentation patterns inflies [7] and mice [8], or genes alteringskeletal elements in populations ofstickleback [9]. Usually, such genescan be identified only through oftenlaborious genetic mapping, andpinpointing the actual molecularlesion can be even trickier.

Initially, Rohner et al. [4] hadidentified a zebrafish (Danio rerio)mutant, called spiegeldanio (spd), that,in its scale pattern, looks suspiciouslylike mirror carp. The mutant camefrom a genetic screen for adultmorphology — the developmentalbasis of which is poorly understood invertebrates. The fact that the mutantlooked so similar to the scalelesscarp — perhaps unsurprisingly so,seeing that carp and zebrafish are bothcyprinid fish family members — raiseshopes that such genetic screens couldprovide clues to phenotypic change

during evolution (or domestication) bygenerating phenotypes that look muchlike evolutionary variants; and, ofcourse, the molecular identification ofthe gene will be much easier in a modelorganism rather than a wild species.When the gene mutated in spiegeldaniowas identified, it was found to bea receptor for fibroblast growth factor(FGF) [4]. To a developmental biologist,that a mutation in this particularreceptor should cause such a mild,non-lethal phenotype must have comeas a surprise: FGF signals are used overand over throughout early vertebratedevelopment. And indeed, if the fgfr1gene is mutated in another fish, thekillifish medaka (Oryzias latipes),mutant embryos emerge that lackessentially all body structures exceptthe head [10].

Mirrors and DuplicatesHow come that the phenotypic effectsof fgfr1 mutation — scale loss asopposed to loss of the entire body —differ so much between zebrafish andmedaka, even though the genesthemselves are highly similar insequence? Zebrafish have two copiesof fgfr1, called fgfr1a and fgfr1b [4]. Andthese two genes appear to functionlargely in a redundant fashion: whenyou knock out either of them alone,nothing much happens. Only if fgfr1bis knocked out as well does theembryonic phenotype look verysimilar to the medaka mutant. Thisredundancy during earlyembryogenesis explains why thescaleless zebrafish fgfr1a mutants —and by extension the mirror carp — cansurvive to adulthood: the early functionin trunk development is buffered by thepresence of a second fgfr1 back-upcopy. Having this zebrafish mutantprovided Rohner et al. [4] with an assaysystem with which they could readilytest the function of the various carpversions of fgfr1 — a feat often missingfrom purely genetic evolutionarystudies; and as expected, the mutatedfgfr1a1 from the mirror carp indeedcannot substitute for zebrafish fgfr1ain the injection assay, while the versionfrom scaled carps can [4].

The history of gene duplications infish is a particularly tangly one. First,there were two rounds of wholegenome duplications in the ancestor ofall vertebrates [11]. Subsequently, theancestor of teleost fishes underwent anadditional duplication [12]. During

DispatchR903

further evolution, different duplicatesmay have lost their function in thedifferent lineages. This may explain thedifference in the fgfr1 phenotypebetween medaka and zebrafish, andindeed the different phenotypicspectrum generally observed duringgenetic screens in these two species.To complicate matters further, thecommon carp’s ancestor underwentanother round of genome duplication,some time in the Miocene, so that forthe two fgfr1 genes of zebrafishthere are now four paralogues in thecarp [13]. Thus, carp will have twocopies of the scale-loss gene fgfr1a,designated fgfr1a1 and fgfr1a2, thelatter of which is mutated andappears less functional no matterif derived from scaled or mirrorcarp [4].

The story of fgfr1 encapsulatesessentially all scenarios that have beenassociated with the fate of duplicatedgenes [14,15]: after duplication, mostfrequently one of the duplicates losesits function — as may have happened inthe medaka lineage — and theremaining one retains all the originalfunctions. In some instances, however,both paralogues are maintained,sharing the original function, as isevident in the embryonic function of thezebrafish fgfr1 genes. Occasionallywhen genes are duplicated one of thecopies may take on a different, newfunction and it appears that this isindeed what has happened withfgfr1a’s role in scale development.(‘New’ here is naturally misleading as itis formally not possible to discern whatis ancient and what is derived: it maywell be that in the piscine ancestor,when scales evolved, both fgfr1paralogues played a role in scaledevelopment and only later fgfr1b lostits scale function.) The fgfr1 duplicationthus provides a neat example for thelong-standing idea of geneticredundancy buffering the evolution ofgenes [15]. Were there not the secondfgfr1b paralogue, a change in adultmorphology, here the loss of scales,would not be possible, at least notthough this very genetic pathway.(Other pathways are known that canlead to loss of scales and indeeda medaka mutant that looks somewhatsimilar to the scaleless zebrafish wasshown to affect the ectodysplasinreceptor A gene [16]. In this case, themedaka phenotype is rather similar tothat of zebrafish mutants in the samegene [17].)

Figure 1. A mirror for evolution?

A koi carp displaying the scale reduction characteristic for domesticated carp breeds, such asthe mirror carp. Photograph by Matthew Harris.

Mirroring EvolutionThe analysis of the genetic basisfor scale loss in the — seeminglyparochial — domesticated mirror carpbears on a fundamental problem inevolutionary biology. This problemof pleiotropy has been buggingevolutionary biologists a greatdeal — particularly those interested inthe molecular basis of morphologicalevolution. In the 1980s and 90s, itbecame clear that embryonic andpost-embryonic development involvethe same sets of genes being usedover and over again during differentprocesses; much as FGF signalling isbeing used, among others, during trunkformation, in brain development andlater in the scales. So, how canevolution change the function of a genein one particular part of the body — sayin the scales — without at the same timeaffecting its other functions? If thecoding sequence of that gene changes,other aspects of its function might beaffected too. However, switching on oroff the expression of a gene in onepart of the body, by alterations in itscis-regulatory sequence, could alterthe function of a gene in a region-specific manner and leave the rest ofthe gene’s function unaffected [18].This is feasible because cis-regulatoryregions of developmental genes aremodular, in that certain enhancer

elements direct expression ratherspecifically in certain body partswithout affecting much the expressionof the gene in other areas. And indeed,there have been numerous examplesof such cis-regulatory changesunderlying the evolution ofmorphological characters, for instancein the evolution of pigmentation inDrosophila species [7] or in skeletalcharacters in sticklebacks [9].

However, studies of molecularadaptation have also found manyinstances of coding region changes[8,19]. The relevance of these twomodes of change — cis-regulatoryversus coding — for evolution hasrecently been debated with someheat: the opposing camps like torefer to each other as ‘cis-sies’ and‘exon-shmexons’, respectively. Thecarp fgfr1 story seems to please boththese camps: on the one hand, thechange underlying scale loss isa change in the coding region of thegene; on the other hand, the scale-specific expression of fgfr1a — anobvious prerequisite for a scale-specific function of that gene — is mostlikely due to an earlier cis-regulatorychange. So in a sense, the codingchange in fgfr1a underlying scale lossin the mirror carp and the zebrafishmutant was possible only on twoconditions: gene duplication with

Current Biology Vol 19 No 19R904

retention of the function of bothduplicates such that possible earlypleiotropic effects were bufferedagainst, and second a change inexpression of the two paraloguessuch that one became specificallyexpressed in scales.

Now, hardcore evolutionarybiologists can be hard to please, andone commonly heard interjection is thatdomesticated animals will only poorlymirror ‘real’ evolution in the wild. WhileDarwin used domestication as ananalogy to describe how naturalevolutionary change might occur,there are many obvious differences:domesticated species live in protectedenvironments, the population sizes andstructures of domesticated and wildanimals differ strongly, and theselective pressures applied by highlychoosy breeders are very different andgenerally much higher. But forunderstanding how morphologicalchange is being generated on themolecular level, these differences areperhaps less relevant, as thedevelopmental starting material isthe same, whether a fish evolves in thewild or in a breeding pond. So, foridentifying genes that lead tomorphological change in evolution,domesticated animals may still bea viable testing ground — apart fromthe interest in domestication itself. Andindeed, in some instances, similarmorphological changes in wild anddomesticated animals seem to involvethe same genes, such as the MC1Rlocus controlling pigmentation, even

Neural Coding: Nonand Conceptual

Recordings from single cells in humansensory processing forms explicit neuraconcepts needed for a causal model of

Peter Foldiak

The nature of the relationshipbetween brain activity and mentalrepresentations is a fundamentalquestion in neuroscience, withrelevance to disciplines ranging fromphilosophy to cognitive science.While the answer in general is distant,recording the activity of single neurons

though the exact type of mutation mayvary [20]. Sure enough, the mirror carp,far from being a mere domesticationoddity, will have something tocontribute to the study of naturalevolution as well. Scale loss orreduction is presumed to haveoccurred independently manytimes during fish evolution [4]. Itwill be illuminating to see if,genetically, these mirror the changesseen in the carp.

References1. The Economist, August 15th 2009, page 76,

Obituary.2. Balon, E.K. (1995). The common carp, Cyprinus

carpio: its wild origin, dmestication inaquaculture, and selection as colorednishikigoi. Guelph Ichtyol. Rev. 3, 1–55.

3. Hoffman, R.C. (1995). Environmental changeand the culture of common carp in medievalEurope. Guelph Ichthyol. Rev. 3, 57–85.

4. Rohner, N., Bercsenyi, M., Orban, L.,Kolanczyk, M.E., Linke, D., Brand, M., Nusslein-Volhard, C., and Harris, M.P. (2009). Duplicationof fgfr1 permits Fgf signaling to serve asa target for selection during domestication.Curr. Biol. 19, 1642–1647.

5. Kirpichnikov, V.S. (1999). Genetics andbreeding of Common Carp (Paris, France:INRA).

6. Doebley, J. (2004). The genetics of maizeevolution. Annu. Rev. Genet. 38, 37–59.

7. Jeong, S., Rebeiz, M., Andolfatto, P.,Werner, T., True, J., and Carroll, S.B. (2008).The evolution of gene regulation underliesa morphological difference between twoDrosophila sister species. Cell 132, 783–793.

8. Hoekstra, H.E., Hirschmann, R.J., Bundey, R.A.,Insel, P.A., and Crossland, J.P. (2006). A singleamino acid mutation contributes to adaptivebeach mouse color pattern. Science 313,101–104.

9. Shapiro, M.D., Marks, M.E., Peichel, C.L.,Blackman, B.K., Nereng, K.S., Jonsson, B.,Schluter, D., and Kingsley, D.M. (2004). Geneticand developmental basis of evolutionary pelvicreduction in threespine sticklebacks. Nature428, 717–723.

-Local but Explicit

medial temporal cortex confirm thatl representations of the objects andthe world.

in the sensory system has provedremarkably informative about themore specific question of how thenervous system encodes individualstimuli and stimulus features intopatterns of neural activity [1,2]. Arecent series of fascinating single-cellrecording experiments from humanmedial temporal cortex (MTL) [3–6]has revealed neurons that are highly

10. Yokoi, H., Shimada, A., Carl, M., Takashima, S.,Kobayashi, D., Narita, T., Jindo, T., et al. (2007).Mutant analyses reveal different functions offgfr1 in medaka and zebrafish despiteconserved ligand-receptor relationships. Dev.Biol. 304, 326–337.

11. Dehal, P., and Boore, J.L. (2005). Two roundsof whole genome duplication in the ancestralvertebrate. PLoS Biol. 3, e314.

12. Meyer, A., and Van de Peer, Y. (2005).From 2R to 3R: evidence for a fish-specificgenome duplication (FSGD). Bioessays 27,937–945.

13. David, L., Blum, S., Feldman, M.W., Lavi, U.,and Hillel, J. (2003). Recent duplication of the,common carp (Cyprinus carpio L.) genome asrevealed by analyses of microsatellite loci. Mol.Biol. Evol. 20, 1425–1434.

14. Zhang, J. (2003). Evolution by gene duplication:an update. Trends Ecol. Evol. 18, 292–298.

15. Lynch, M., and Force, A. (2000). The probabilityof duplicate gene preservation bysubfunctionalization. Genetics 154, 459–473.

16. Kondo, S., Kuwahara, Y., Kondo, M.,Naruse, K., Mitani, H., Wakamatsu, Y.,Ozato, K., et al. (2001). The medaka rs-3 locusrequired for scale development encodesectodysplasin-A receptor. Curr. Biol. 11,1202–1206.

17. Harris, M.P., Rohner, N., Schwarz, H.,Perathoner, S., Konstantinidis, P., andNusslein-Volhard, C. (2008). Zebrafish eda andedar mutants reveal conserved and ancestralroles of ectodysplasin signaling in vertebrates.PLoS Genet. 4, e1000206.

18. Wray, G.A. (2007). The evolutionary significanceof cis-regulatory mutations. Nat. Rev. Genet. 8,206–216.

19. Hoekstra, H.E., and Coyne, J.A. (2007). Thelocus of evolution: evo devo and the genetics ofadaptation. Evolution 61, 995–1016.

20. Fang, M., Larson, G., Ribeiro, H.S., Li, N., andAndersson, L. (2009). Contrasting modeof evolution at a coat color locus inwild and domestic pigs. PLoS Genet. 5,e1000341.

Florian Maderspacher is Current Biology’sSenior Reviews Editor.E-mail: florian.maderspacher@current-biology.com

DOI: 10.1016/j.cub.2009.09.008

selective to individual people, objectsor narrow categories, and invariantto changes irrelevant to objectidentity. MTL is at the top of thesensory processing hierarchy,offering an unprecedented insightinto the end result of sensoryprocessing.

The latest paper in this series,published recently inCurrent Biology [6],demonstrates that many of therecorded neurons respond not onlyto images of one specific item, forexample ‘‘Saddam Hussein’’, but alsoto the written and spoken name of thesame item. The auditory and visualselectivities are precisely aligned,so that the auditory, visual textualdescriptions and visual images