Evolution: Flower colour power

6 | JANUARY 2004 | VOLUME 5 www.nature.com/reviews/genetics

H I G H L I G H T S

Turning down flowers on the basis oftheir colour is unlikely to raise yourpopularity stakes. Yet, some animalpollinators are adapted for this par-ticular choosiness, which enablesthem to single out the specfic plantspecies that they help to propagateby their colour. For example, thered-flowered species of the mon-keyflower genus Mimulus is visitedexclusively by hummingbirds, whereasits pink sister species is selectivelypollinated by bumblebees. Such fas-tidiousness ensures the reproductiveisolation of the two flower speciesin overlapping habitat ranges. Byexperimentally switching the colourof the two types of Mimulus, TobyBradshaw and Doug Schemske havenow shown that the adaptation ofeach plant species to a specific polli-nator might be due, in large measure,to a mutation at a single locus thataffects flower colour.

The YUP (yellow upper) genecontrols whether yellow pigment isdeposited in flowers: the pigment

is present in the red, hummingbird-pollinated M. cardinalis, but absentin the pink, bumblebee-pollinatedM. lewisii (shown in the figure). Theauthors substituted the YUP allele ofone species by crossing it into thenear-isogenic line of the other, tocreate pale, yellow-orange M. lewisiiand dark-pink M. cardinalis. Theeffect of this colour change on polli-nator presence was striking: bumble-bees preferred the ‘mutant’ pink M.cardinalis — the species normallyselected by birds — over the red,wild-type variety by 74-fold, andhummingbirds visited the ‘mutant’yellow-orange M. lewisii — the speciesnormally favoured by bees — 68 timesmore than the pink, wild-type one.That the effect was so large and sym-metrical reinforces the idea thatmutations in YUP have a markedeffect on pollinator preference andthat this locus alone — under theright ecological conditions — couldinitiate an adaptive shift in pollinatorchoice.

Flower colour power

E V O L U T I O N

From the prokaryotic mayhem of theprimordial soup emerged the multicellulareukaryotes, which occupy more space, docleverer tasks and have bigger and morecomplex genomes. But what route didphenotypic evolution take in going fromone to the other? Did adaptivediversification depend on having a morecomplex genome or was it the other wayaround? Mike Lynch and John Coneryprovide statistical evidence that the morecomplex genomes of multicellulareukaryotes arose passively and thereforewithout much adaptive purpose. Theypropose that only once the new genomicfeatures were in place would they have beenexploited for adaptive purposes.

The genomes of multicellular eukaryotesare not only much bigger than unicellularones, but they also have more genes, moreintrons and more mobile elements.Although there are plausible advantagesthat these features would bring, it is alsopossible that more complex genomes arosebecause there was nothing to prevent themfrom arising. The ‘something’ that could

prevent this is purifying selection, whichpurges undesirable variants from apopulation.

Purifying selection is less powerful insmaller populations, in which traits have abetter chance of spreading through thepopulation owing to random forces. Theauthors have calculated that as organismsget larger, on average their population sizegets smaller. They show that the effectivepopulation size (N

e) — the number of

individuals that actually contribute to thenext generation — can vary by severalorders of magnitude between the largestand smallest organisms. So, multicellularspecies, with their much smaller N

e—

which, in turn, is probably caused by theirlarger cell and body sizes — would be freerto accumulate non-selected changes totheir genome.

The authors tested their theoreticalexpectations against the characteristicfeatures of multicellular genomes. Forexample, they show that multicellularspecies probably have more genes becausethey retain duplicated genes longer than do

unicellular species, as mutations takelonger to erode them. So, rather than onecopy of the duplicate pair degenerating outof existence, both could survive by splittingbetween them the role of the ancestrallocus (through a process known as‘subfunctionalization’). A similar sort ofreasoning can be proposed to explain theemergence of a large number of intronsand mobile elements.

Of course, the idea is not that all complextraits arose by chance — rather, it is that anon-adpative expansion in the genomeprovided the genetic raw material forselection to work on. For example, it isperfectly feasible that once a large numberof introns were present, they would be putto use in alternative splicing, therebypaving the way for more adaptiveevolutionary changes. As the authors pointout, more directed experiments are neededto prove their model and that ‘exceptionalspecies’ within each group should be goodtesting ground for their theories.

Tanita CasciReferences and links

ORIGINAL RESEARCH PAPER Lynch, M. & Conery, J. S.The origins of genome complexity. Science 302, 1401–1404(2003) WEB SITESMike Lynch’s laboratory:http://www.bio.indiana.edu/facultyresearch/faculty/Lynch.htmlJohn Conery’s home page:http://www.cs.uoregon.edu/~conery

Evolution, the passive way

P O P U L AT I O N G E N E T I C S

Image courtesy of D. Schemske and T. Bradshaw.

NATURE REVIEWS | GENETICS VOLUME 5 | JANUARY 2004 | 7

One possible explanation for the paltrynumber of extra genes that humanshave, compared with much lesscomplex animals, is that we haveevolved many different ways toregulate the same genesduring development. Justhow important evolu-tionary rewiring ofthe regulatory cir-cuitry can be is now evi-dent from a new and thor-ough study that compares thetranscriptional circuits that regulatemating type in two yeast species.

Mating behaviour in yeast is controlled bythe mating-type locus. In Saccharomyces cere-visiae, this locus (MAT ) encodes three transcrip-tional regulators (α1, α2 and a) and, until now, itwas thought that the same applied to the homolo-gous locus in its distant relative Candida albicans(MTL). However, an unintentional knockout of apreviously overlooked open reading frame at thislocus showed that MTL encodes an additionalregulator (a2).

To investigate the role of a2 and to get to thebottom of mating-type regulation in C. albicans,Annie Tsong and colleagues deleted each regulatorindividually and in all possible combinations. Theythen analysed the mating behaviour and transcrip-tional profiles of the 16 possible mutant strains.

Their complete mating-behaviour data setallowed the authors to determine that a2 and α1are positive regulators of their respective matingtypes, whereas a1 and α2 each contribute one halfof a heterodimer that negatively regulates the abil-ity to switch from the white phase to the opaquephase that is needed to mate efficiently. By contrast,in S. cerevisiae, which lacks a2, cells default to the a-mating-type when MAT does not contribute any-thing, and when α2 is present, it acts as a negativeregulator of a-type-mating.

Together with gene-expression data, these matingexperiments provide some unique insights intothe regulation of mating. Even more interesting thanthe differences in the control of individual genes in C. albicans and S. cerevisiae were the differences in transcriptional circuitry. For example, a-specificgenes are under negative control in S. cerevisiae, butin C. albicans they are under positive control. There isalso an extra level of mating control present in C. albicans that is reflected in the white-to-opaque

phenotypic shiftthat is required formating.

The authors suggest thatthese distant relatives mightretain aspects of the trancriptionalcircuitry present in their commonancestor. They postulate that negative regu-lation of a-specific genes in the S. cerevisiae lin-eage replaced the ancestral positive regulation (stillretained in C. albicans), whereas white-opaqueswitching in the latter is likely to be a recent adapta-tion to life in mammalian hosts, now coupled to amuch more ancient regulatory circuit, also foradaptive reasons.

So, mixing and matching regulatory-circuit elements seems to be a viable evolutionary means ofincreasing complexity and adapting to new environ-ments. The remaining question seems to be one ofscale: can such transcriptional rewiring explain dif-ferences in complexity as large as those that arefound between humans and yeast?

Nick CampbellReferences and links

ORIGINAL RESEARCH PAPER Tsong, A. E. et al. Evolution of acombinatorial transcriptional circuit: a case study in yeast. Cell 115,389–398 (2003) FURTHER READING Johnson, A. The biology of mating in Candidaalbicans. Nature Rev. Micro. 1, 106–116 (2003)WEB SITESandy Johnson’s laboratory:http://itsa.ucsf.edu/%7Emicro/Faculty/Johnson/johnson_index.html

Rewiring thecircuitry

G E N E R E G U L AT I O NCould Ronald A. Fisher — theinnovative twentieth-century sta-tistician and evolutionary biologist— possibly have got it wrong? Histheory that adaptation proceeds bythe gradual accumulation of an infi-nite number of infinitesimally smallsteps might now have to give way tothe increasingly popular view that thesame result can be achieved by takingjust a few, perhaps even one, big step.How close this study of pollinatorshifts has come to demonstrating sucha paradigm shift will depend on iden-tifying additional mutations in the‘adaptive walk’ from bumblebee tohummingbird pollination.

Tanita Casci References and links

ORIGINAL RESEARCH PAPER Bradshaw, H. D.Jr & Schemske, D. W. Allele substitution at a flowercolour locus produces a pollinator shift inmonkeyflowers. Nature 426, 176–178 (2003) WEB SITESToby Bradshaw’s laboratory:http://faculty.washington.edu/tobyDoug Schemske’s laboratory:http://www.plantbiology.msu.edu/schemske.shtmMimulus page:http://www.ex.ac.uk/~MRMacnai/mimulus.html