kemo inggris

Molecular Ecology (2008) 17, 1648–1657 doi: 10.1111/j.1365-294X.2008.03713.x

© 2008 The AuthorsJournal compilation © 2008 Blackwell Publishing Ltd

Blackwell Publishing LtdFAST-TRACK

Drosophila chemoreceptor gene evolution: selection, specialization and genome size

ANASTASIA GARDINER,* DANIEL BARKER,* ROGER K. BUTLIN,† WILLIAM C. JORDAN‡ and MICHAEL G. RITCHIE**School of Biology, University of St Andrews, St Andrews, Scotland KY16 9TH, UK, †Animal and Plant Sciences, The University of Sheffield, Sheffield S10 2TN, UK, ‡Institute of Zoology, Zoological Society of London, Regent’s Park, London NW1 4RY, UK

Abstract

Chemoperception plays a key role in adaptation and speciation in animals, and the sensesof olfaction and gustation are mediated by gene families which show large variation inrepertoire size among species. In Drosophila, there are around 60 loci of each type and it isthought that ecological specialization influences repertoire size, with increased pseudog-enization of loci. Here, we analyse the size of the gustatory and olfactory repertoires amongthe genomes of 12 species of Drosophila. We find that repertoire size varies substantiallyand the loci are evolving by duplication and pseudogenization, with striking examples oflineage-specific duplication. Selection analyses imply that the majority of loci are subjectto purifying selection, but this is less strong in gustatory loci and in loci prone to duplication.In contrast to some other studies, we find that few loci show statistically significant evidenceof positive selection. Overall genome size is strongly correlated with the proportion ofduplicated chemoreceptor loci, but genome size, specialization and endemism may beinterrelated in their influence on repertoire size.

Keywords: Drosophila, duplication, gene family evolution, gustatory receptor (GR), olfactoryreceptor (OR), pseudogenization

Received 30 September 2007; revision received 16 November 2007; accepted 13 January 2008

Introduction

Chemosensory perception plays a crucial role in ecologicaladaptation, allowing the finding of food, mating partnersand oviposition sites, kin recognition and the avoidance oftoxins and predators. Animals developed chemoreception,or the senses of smell (olfaction) and taste (gustation), as atool for interacting with the external world and changes inolfactory or gustatory sensitivities often play a key role inhabitat and mate choice. Chemical cues are recognized anddiscriminated by chemoreceptor proteins expressed inchemosensory neurones first identified in mammals (Buck& Axel 1991). Chemoreceptor repertoires have been describedin several vertebrates (Freitag et al. 1995; Nef et al. 1996;Glusman et al. 2001; Zhang & Firestein 2002; Olenderet al. 2004; Alioto & Ngai 2005; Quignon et al. 2005) andinvertebrates (Troemel et al. 1995; Robertson 1998; Clyne

et al. 1999; Gao & Chess 1999; Clyne et al. 2000; Scott et al.2001; Hill et al. 2002; Robertson et al. 2003; Robertson &Wanner 2006). In vertebrates, a large family of olfactory anda smaller set of gustatory receptors recognize environmentalchemical cues, while pheromones are discriminated byvomeronasal receptors (Niimura & Nei 2006). The olfactoryrepertoires are often very divergent between species andcomprise from 40 to 100 receptor genes in fish and up to~1000 in mammals (Gaillard et al. 2004) and this is thoughtto reflect variation in the importance of chemoperceptionduring their evolution (the functional olfactory repertoiresof humans are relatively small, but these loci are regularlyfound among genomic studies identifying loci underselection, Voight et al. 2006). The largest chemoreceptorsuperfamily (~1300 genes) was described in nematodesand represents approximately 7% of all their protein-codinggenes (Robertson & Thomas 2006). The chemoreceptors ofvertebrates and nematodes are thought to belong to aclass of G-protein-coupled seven transmembrane-domainreceptors (GPCR) (Mombaerts 1999).

Correspondence: Michael G. Ritchie, Fax: +44 1334 463366; E-mail: [email protected]

Aspire

Highlight

D R O S O P H I L A C H E M O R E C E P TO R G E N E E V O L U T I O N 1649


The complete chemoreceptor repertoires have beenidentified in several insects, including Drosophila melanogaster(Clyne et al. 1999; Gao & Chess 1999; Clyne et al. 2000; Scottet al. 2001; Robertson et al. 2003; Guo & Kim 2007; McBride2007; Nozawa & Nei 2007), mosquito (Anopheles gambiae,Fox et al. 2001; Hill et al. 2002), and honeybee (Apis mellifera,Robertson & Wanner 2006). The chemoreceptor superfamilyin Drosophila melanogaster comprises 60 olfactory (OR) and60 gustatory receptor (GR) genes that encode at least 62 ORand 68 GR proteins through alternative splicing (Robertsonet al. 2003). A chemosensory superfamily of similar sizewas identified in Anopheles gambiae with 79 OR and 76 GR(Hill et al. 2002). In contrast, expansion of the olfactoryreceptor genes (170 OR), alongside dramatic reduction ofthe gustatory receptor repertoire (10 GR) was observed inthe honeybee (Robertson & Wanner 2006). Comparison ofthese insect OR/GR families shows very strong interspecificdivergence and species–specific expansions of particulargenes or gene clusters, with only a few orthologous groupsestablished (Hill et al. 2002; Robertson & Wanner 2006).Such patterns in the evolution of species-specific subfamiliesare presumably associated with individual species’ adap-tations and responses to their environment (McBride 2007).Understanding the evolution of these gene families wouldallow us to obtain new insights into the processes of ecologicaladaptation and, in some groups, also speciation (Bray &Amrein 2003; Nakagawa et al. 2005; Kurtovic et al. 2007;van der Goes, van Naters & Carlson 2007). Adaptationto host plant phenology can influence speciation in phyto-phagous insects and recently loci influencing ecologicalspecialization to a host plant have been identified as apotentially key step during ecological speciation in Drosophilasechellia (Matsuo et al. 2007). Changes in chemoperceptionmay be key events in social interactions, including kinrecognition and mate choice. Pheromones are particularlyimportant to mate recognition in many species and phe-romone recognition is a known function of some OR andGR in Drosophila.

Drosophila is an ideal system with which to study theevolution of gene families, because the genome sequencesof 12 species, spanning a range of divergence levels, recentlybecame available (Drosophila 12 Genomes Consortium 2007).Five species (the widely distributed Drosophila simulans,African D. erecta, D. ananassae, D. yakuba and an endemicspecialist D. sechellia) belong to the melanogaster group.These species, plus Drosophila pseudoobscura and D. persimilisfrom the obscura group, and D. willistoni are classifiedwithin the subgenus Sophophora. The remaining species,the cactus-breeding Drosophila mojavensis, widely distributedD. virilis and Hawaiian endemic D. grimshawi, are withinthe subgenus Drosophila and are the most distant fromD. melanogaster, with an estimated divergence time of 40–60million years. McBride (2007) recently described extensiveloss of olfactory genes in D. sechellia and suggested this

reflects host specialization (with some retained loci beingsubject to diversifying selection). Here, we analyse andcompare the incidence of duplication and pseudogenizationin both gene families from all 12 species (including additionalspecialist species), examine patterns of sequence variationfor evidence of selection on GR and OR loci, and examinethe potential roles of specialization, endemism and genomesize in influencing the size of the superfamily of chemo-sensory receptor loci.

We have identified the OR and GR genes that areorthologous to those of D. melanogaster to compare patternsin their evolution. Extensive gene families evolve by dupli-cation, divergence and pseudogenization (Ohno 1970). Thereis considerable debate about the mode of evolution of genefunction within gene families; duplicate copies might besilenced, they might evolve new functions (neofunctional-ization) or existing functions might be subdivided amongnew copies (subfunctionalization) (Lynch & Conery 2000).Families such as OR and GR loci, where loci could specializein function by developing affinities for novel ligands, maybe particularly prone to subfunctionalization (i.e. subdividingsubstrate affinities within an overall function of chemore-ception). Here, we examine the incidence and nature ofselection acting on these loci by analysing patterns ofsequence divergence.

We describe the evolution of the chemoreceptor super-family by lineage-specific duplication, as well as pseudo-genization and gene loss, and compare patterns seen in theGR and OR families. Accelerated expansion of copy numberfor several OR and GR genes was observed in D. ananassae,D. willistoni and D. grimshawi. Our results suggest thatpurifying selection predominates in the evolution of theseloci, but positive selection has acted on around 15% of thegenes.

Materials and methods

Genomic assemblies of Drosophila melanogaster (sequencedby the Berkeley Drosophila Genome Project and Celera),D. simulans and D. yakuba (Washington University), D. erecta,D. ananassae, D. mojavensis, D. virilis and D. grimshawi(Agencourt), D. willistoni (Aranche/Celera), D. sechelliaand D. persimilis (Broad Institute) and D. pseudoobscura(Baylor) were obtained from the AAA Web Server (http://rana.lbl.gov/drosophila) in January 2006.

For each D. melanogaster GR and OR protein (Table S2,Supplementary material,), orthologues from 11 other specieswere predicted using tblastn and genewise searches(Altschul et al. 1997; Birney et al. 2004; additional details arein the Supplementary material, see also Tables S3 and S4),except that data for the OR family of D. willistoni wereadapted from Guo & Kim (2007). Reciprocal blasts wereused to check the identity of all duplicated loci. We conductedan additional analysis verifying all our orthologous groups

http://rana.lbl.gov/drosophila

1650 A . G A R D I N E R E T A L .


using the inparanoid algorithm (Remm et al. 2001), foundto perform the best from a series of orthologue identificationprocedures examined by Hulsen et al. (2006), and we alsocompared our orthologous groups with those identified bythe Consortium project (Drosophila 12 Genomes Consortium2007) where they used a fuzzy reciprocal blast algorithmand synpipe. All assignments agree, although our procedureidentified additional orthologues. The resulting sets of single-copy or (in the case of duplicated genes) most-conservedorthologues were multiply aligned at the codon levelusing clustal w (Thompson et al. 1994) and protal2dna(K. Schuerer, C. Letondal; http://bioweb.pasteur.fr).

To draw comparisons on chemoreceptor repertoire sizeamong specialist, generalist and endemic species, weclassified D. sechellia, D. erecta and D. mojavensis as specialistand D. sechellia and D. grimshawi as island endemics (Powell1997). The data on genome size were taken from Gilbert(2007). Tests for positive selection were performed usingthe codeml program in paml (Yang 1997) to calculate ω(= dN/dS) ratios of the normalized nonsynonymous substi-tution rate (dN) to the normalized synonymous substitutionrate (dS). ω > 1 is considered to be strong evidence of positiveselection for amino acid replacements (Yang & Bielawski2000) whereas ω ≈ 0 indicates purifying selection. Evidenceis assessed using likelihood-ratio tests (LRT) for hierarchicalmodels (see Supplementary material for further details).Selection analysis was performed on genes considered tobe orthologous from the gene query of D. melanogaster. Ifduplicates were present, we considered the gene copy withthe highest alignment score vs. the query as the functionallyconserved orthologue and this copy was included in thedata set for the further tests. Drosophila willistoni was omittedfrom selection analysis due its extreme codon bias (this is acontrast from a related study of OR loci, Guo & Kim 2007).

The phylogeny of Gr85a nucleotide sequences was recon-structed using mrbayes (Ronquist & Huelsenbeck 2003)with the HKY +4Γ model and one chain of 30 000 000generations, sampled every 50 000 generations. The first351 trees sampled were discarded as burn-in, leaving afinal Markov chain Monte Carlo (MCMC) sample of 250trees. The long interval between samples improved theirindependence (first-order autocorrelation in log likelihoodsr = –0.025). The majority-rule consensus tree (rooted withDmelGr47a and DmelGr58a as an outgroup) and cladeposterior probabilities are presented.

Phylogenetically corrected tests for a correlation betweenproportion duplicated, proportion pseudogenized or func-tional chemoreceptor count on the one hand and genomesize, ecology (specialist or generalist) or location (island ormainland) on the other were performed using the programcontinuous (Pagel 1997; Pagel 1999) (see also Supplementarymaterial, Table S1). This allows a statistical test of thecorrelation, beyond any contribution from the historicalinfluence of the species phylogeny. The species phylogeny

was downloaded from http://insects.eugenes.org/species/news/genome-summaries/gene-phylogeny.html. The non-Drosophila species were removed, leaving a rooted tree ofthe 12 sequenced Drosophila species with branch lengths.This was manually combined with trait data and analysedusing continuous. A null model was fitted to the phylogenyand traits, in which the contribution of the phylogeny (λ)was estimated but the covariance between the traits wasset to zero. An alternative model was fitted, in which λ andthe covariance between the traits were simultaneouslyestimated. The strength of evidence for the alternativemodel was assessed using a LRT. A P value for the correlationbetween the traits was obtained using the χ2 approximationof continuous. Where one trait is continuous and the otheris binary, the test may be regarded as a phylogeneticallycorrected t-test (Organ et al. 2007).

Results

Description of repertoires

The estimated size of the chemoreceptor superfamily in the12 species varies between 115 (Drosophila mojavensis) and166 loci (Drosophila grimshawi), although alternative splicingof a few loci means that slightly more receptors are encoded(Table 1). The GR family ranges from 52 (D. mojavensis) to83 genes (D. grimshawi), while the OR family ranges from60 genes (Drosophila erecta and D. sechellia) up to 83 genes(D. grimshawi). The sizes of the GR and OR repertoires inDrosophila species are correlated (r = 0.664, P = 0.019; Fig. 1),although this was not significant without D. grimshawi.The greatest expansions of both families are in D. ananassae,D. willistoni and D. grimshawi. These species are not closelyrelated. A binomial generalized linear model of theproportion of pseudogenes classified by species and receptortype shows that the proportion of pseudogenes varies

Fig. 1 Total number of gustatory and olfactory loci in Drosophilaspecies.

http://bioweb.pasteur.fr

http://insects.eugenes.org/species/news/genome-summaries/gene-phylogeny.html



significantly among species [deviance ratio (DR) = 6.34,d.f. = 11,23, P < 0.001] but not between the receptor types.Overall, around 8% of loci were pseudogenized. The pro-portion of pseudogenes is highest in D. sechellia, D. persimilisand D. grimshawi, and lowest in D. yakuba, D. simulans andD. mojavensis (Table 1; Fig. 2a).

Most duplications of loci seemed to be independentevents, in that linked pairs of loci were not observed tohave coduplicated. Analyses of the proportion of loci thatunderwent duplication showed that this varied betweenspecies (DR = 7.51, d.f. = 11,23, P < 0.001), being highest inD. willistoni and D. grimshawi and lowest in the obscuragroup (Table 1, Fig. 2b), but also almost varied betweenreceptor type (DR = 3.69, d.f. = 1,23, P = 0.055), with aslightly higher proportion of duplicated OR (8.1%) than GR(5.6%) loci. The proportion of pseudogenes was unrelatedacross species to the proportion of duplicated genes. Thelargest expansions in copy number of GR genes occurredon the phylogenetic branches that lead to D. willistoni,D. grimshawi, and the melanogaster group. The greatest extentof GR gene duplication was found in D. grimshawi. Forinstance, the gene Gr85a has multiple copies in D. mojavensis(6 copies), D. virilis (7 copies), D. willistoni (7 copies) andD. grimshawi (17 completely sequenced copies and 7 partialgene sequences). The expansion of the Gr85a repertoire inD. grimshawi is one of the largest expansions of the chemo-sensory genes (Fig. 3). Or42b has also duplicated extensivelyin D. grimshawi to give at least eight intact copies of this gene.

Predictors of gene family size

McBride (2007) reported that D. sechellia had lost OR andGR loci at a faster rate than D. simulans and D. melanogaster

Table 1 Size of GR and OR repertoires, the number of pseudogenes (ψ) and duplicated genes and the number of receptors encoded in eachspecies (the latter is greater than the number of loci because of alternative splicing). Here and elsewhere, data for the OR family of Drosophilawillistoni were adapted from Guo & Kim (2007) and data for the OR/GR families of Drosophila melanogaster were adapted from Robertsonet al. (2003). McBride (2007) confirmed the status of some OR and GR pseudogenes in Drosophila simulans and Drosophila sechellia genomesby direct re-sequencing. We incorporated her data into our table

SpeciesTotal GR and OR loci Total GR loci

Predicted number of encoded GRs

Estimated number of ψGRs Total OR loci

Predicted number of encoded ORs

Estimated number of ψORs

D. melanogaster 121 60 68 2 61 62 2D. simulans 126 65 73 2 61 64 0D. sechellia 125 65 71 11 60 62 6D. yakuba 125 63 71 0 62 65 1D. erecta 118 58 64 4 60 63 2D. ananassae 142 71 80 7 71 72 6D. pseudoobscura 126 55 66 3 71 74 5D. persimilis 124 55 66 9 69 72 14D. willistoni 153 73 83 5 80 81 8D. mojavensis 115 52 65 2 63 64 1D. virilis 117 53 64 3 64 65 8D. grimshawii 166 83 92 13 83 83 15

Fig. 2 The proportion of (a) pseudogenes and (b) duplicated locivaries among species.

1652 A . G A R D I N E R E T A L .


and concluded that niche specialization in D. sechelliaunderlies this process. We now have more species withwhich to test this hypothesis. D. erecta is also a specialistspecies, showing a ‘seasonal specialism’ with Pandanus(Rio et al. 1983), and Drosophila mojavensis is cactophilic.Although it can utilize a number of cactus species, it has ahistory of host shifts followed by a degree of specialism(Newby & Etges 1998; Matzkin et al. 2006). The proportionof pseudogenized loci across both families did not differbetween these specialist vs. generalist species (t = 0.30,d.f. = 3, P = 0.78) but did differ between island vs. mainland

species (t = 3.75, d.f. = 3, P = 0.033, mean 0.07 and 0.15 formainland and island species, respectively), suggesting thatthe endemism of D. sechellia may be more important thanits specialization. The proportion of duplicated loci did notdiffer between the generalist and specialist species either(t = 0.76, d.f. = 3, P = 0.50) or between island vs. mainlandspecies (t = –0.21, d.f. = 1, P = 0.87). Comparisons of thetotal functional chemoreceptor repertoire size, total genomesize, and the proportion of pseudogenes and duplicatedloci showed that the strongest relationship across specieswas between genome size and the proportion of duplicated

Fig. 3 Gr85a tree for 12 Drosophila species:Dmel, D. melanogaster; Dsec, D. sechellia;Dsim, D. simulans; Dyak, D. yakuba; Dere,D. erecta; Dana, D. ananassae; Dpse, D.pseudoobscura; Dwil, D. willistoni; Dmoj,D. mojavensis; Dvir, D. virilis; Dgri, D.grimshawi. Pseudogenes are indicated bysymbol ‘P’. The second copy of Gr85a ofD. yakuba and D. erecta were named asGr85aL2 (L ~ Like), because of relatively lowamino acid similarity of these receptorswith DmelGr85a. Partial genes (Gr85a ofD. persimilis and Gr85aL2 of D. erecta) wereexcluded. Numbers show clade posteriorprobabilities. Branch lengths show expectedsubstitutions per site (as mean of theposterior probability density).



loci (Fig. 4a–c), so chemoreceptor gene duplication can bepartly explained by larger-scale genomic changes. Incombined analyses with general linear models with ecology(generalist or specialist) and location (island or mainland)as categorical predictors and genome size as a covariate,the proportion of pseudogenes remained higher for islandspecies (6% vs. 16%, P = 0.04) but did not differ between

specialist and generalist species (6.2% vs. 8.6%, respectively,P = 0.933). Curiously, despite these effects, the total repertoiresize of chemoreceptors did not vary with overall genomesize, endemism or specialism (Fig. 3c).

However, the species’ phylogeny contributes significantlyto the relationship between genome size and the proportionof duplicated loci. We tested for an effect of phylogenyusing the λ parameter (Pagel 1999; Freckleton et al. 2002)and completed correlation analyses between traits whilecorrecting for phylogenetic effects using the generalizedleast squares approaches implemented in continuous(Pagel 1997; Pagel 1999; see Supplementary material forfurther details). The proportion of duplicated loci stillvaried significantly with genome size after correcting forphylogeny (r = 0.568, P = 0.03) and island and mainlandspecies still differed in their proportion of pseudogenizedloci (r = 0.529, P = 0.047). These associations were alsosignificant in phylogenetically corrected partial correlationanalyses with all variables included, but no other relationshipwas significant (see Supplementary material).

Selection analysis: adaptive evolution and nature of selection

Table 2 shows all the loci which had a P value of less than0.05 from the paml analysis. Overall, we tested 107 GRand OR gene alignments for evidence of selection so aBonferonni correction for the number of loci tested wouldyield a critical P value of 0.00047. This is probably extremelyconservative with a high likelihood of type II errors (onlyone of our analyses would yield a significant result, forGr21a). However, the Benjamini & Hochberg (1995) sequentialprocedure (with a false discovery rate of 0.05) only addsone more locus, suggesting that Or56a is also significant.A signature of positive selection on ORs in the Drosophilagenomes was detected by Guo & Kim (2007) for Or9a,Or10a, Or19a, Or43a, Or56a, OrN1 and OrN2. The loci Or9a,Or43a and Or56a also rank highly on our list.

The strength of selection acting on the two gene familieswas compared using general linear models, with lociclassified into the gene families and as duplicated or undu-plicated. Loci were classified as duplicated if they hadundergone duplication anywhere within the genus. Becauseω is calculated across the whole data set, our estimatesrepresent the average selection on these loci, rather thanselection specifically on duplicated copies. Figure 5 showsmean ω in duplicated and unduplicated loci of both families.ω was smaller for OR than GR loci (0.116 vs. 0.149, respec-tively, F1,92 = 12.23, P < 0.001) and in loci which have notundergone duplication (0.111 for singletons, 0.162 forduplicated loci, F1,92 = 25.94, P < 0.001). There was a slightinteraction so that ω was weakest for Gr loci which hadundergone duplication (F1,92 = 4.05, P = 0.047). As these ωvalues are all considerably less than one, they imply that

Fig. 4 The proportion of (a) duplicated (b) pseudogenized and (c)the total number of functional receptors vs. genome size (Mb). Thethree specialist species are indicated with filled circles.

1654 A . G A R D I N E R E T A L .


overall there has been weaker purifying selection on GR loci,especially those loci prone to duplication. Within duplicatedloci, there was no relationship between the number ofduplication events and ω.

The proportion of sites identified as undergoing diversi-fying selection (p1) did not vary between OR and GR loci(mean = 0.0157 vs. 0.0178, F1,85 = 0.18, NS), nor between

singleton and duplicated loci (0.0164 vs. 0.0171, F1,85 = 0.02,NS), nor did ω calculated from these sites (1.966 vs. 1.908,F1,85 = 0.01, NS, for OR vs. GR and 2.099 vs. 1.775,F1,85 = 0.40, NS, for singleton vs. duplicated loci).

Discussion

We found that both the OR and GR gene families haveevolved through gene duplication, pseudogenizationand gene loss. The observed sequence diversity implies apredominant history of purifying selection with positivediversifying selection influencing a small fraction of loci, ifany. The changes we have described among 12 speciesof Drosophila further show how the families have changedduring species divergence, and aspects of this evolutionmight reflect species’ adaptation to their chemical environ-ment, although the best predictor of the extent of geneduplication is total genome size. Genes which are prone toduplication show evidence of relaxed constraints, whichwill facilitate evolutionary divergence in receptor functionbetween these loci.

The largest chemosensory repertoires were observed inDrosophila ananassae, D. willistoni and D. grimshawi. Thesethree species are distantly diverged from each other andtheir chemosensory repertoires evolved through independentlineage-specific expansions of particular gene subfamilies.Drosophila ananassae is widespread and belongs to the

Table 2 The 20 loci with the strongest indications of positive selection by paml analysis. p1 is the proportion of positively selected sites withω1, calculated applying the M8 model. P values are uncorrected assuming a χ2 distribution with 2 d.f. (a Bonferonni adjusted critical valueis 0.00047). See text for further details

GeneSingleton (S) or duplicate (D)

Number of species analysed

Number of codons analysed 2(ln LM8 − ln LM7) p1 M8 ω1 M8 P

Gr21a S 11 447 15.46678 0.01523 1.090 0.00044Or43a S 11 376 13.28885 0.03052 1.627 0.00130Or56a D 10 419 13.16793 0.05017 1.131 0.00138Or9a S 10 382 11.90603 0.00266 8.596 0.00260Or94a S 9 387 11.6437 0.03505 1.255 0.00296Gr28a S 11 448 11.63358 0.01109 1.268 0.00298Gr39aB S 11 381 9.11125 0.01808 2.099 0.01051Gr22e D 9 387 8.970896 0.07159 1.041 0.01127Gr5a S 8 444 8.645066 0.05909 1.114 0.01327Gr32a S 11 453 8.113068 0.00461 7.503 0.01731Gr93a S 11 419 8.046178 0.05013 1.163 0.01790Or59c S 8 411 7.931448 0.00485 16.62 0.01895Or33a D 8 377 7.929124 0.01480 2.972 0.01898Gr23aA S 9 370 7.583684 0.05792 >1 0.02255Or22c S 11 397 7.390392 0.00995 3.281 0.02484Or10a S 11 404 7.154664 0.08639 >1 0.02795Or67b S 11 421 6.813414 0.2402 >1 0.03315Or19a−1 D 11 385 6.771462 0.06427 1.352 0.03385Gr68a S 10 398 6.424416 0.02204 1.163 0.04027Gr93c D 11 397 6.004922 0.01129 2.348 0.04967

Fig. 5 Variation in mean ω among the gene families.



melanogaster group (the other species examined here beingDrosophila melanogaster, D. simulans, D. sechellia and D. yakuba),but differs from the other species of this group by having ahigh spontaneous recombination rate in males (Singh &Singh 2003), perhaps reflected in both GR and OR familiesbeing expanded in this species. Drosophila willistoni has thelargest chemosensory repertoire, also showing expansionin both families. The willistoni lineage has an unusualGC content which is associated with higher nucleotidesubstitution rates (Singh et al. 2006).

The chemosensory repertoire of the Hawaiian species,D. grimshawi, underwent the most dramatic changes.Drosophila grimshawi shares the smallest number of orthol-ogous genes with D. melanogaster, having lost severalancestral GR (Gr5a, Gr22e, Gr58c) and OR (Or1a, Or2a, Or65b)genes. The expansion of the GR family in D. grimshawioccurred mostly because of rapid duplication of theGr85a gene (Fig. 3), while the OR repertoire is one of thelargest and was formed by the duplication of about 16gene subfamilies (see also Guo & Kim 2007; Nozawa &Nei 2007). Such a rapid mode of gene gain and loss islikely to correlate with the divergence and rapid radia-tion of endemic Hawaiian Drosophila from an ancestralimmigrant that arrived in the Hawaiian Islands (DeSalle1992). Although some strains of D. grimshawi are specialist(like other Hawaiian Drosophila), the strain sequencedis a generalist (Piano et al. 1997). The patterns seenhere are likely to reflect both relaxed selection duringcolonization and strong selection on particular loci in thenew environment.

Expansion of the olfactory, but not the gustatory, familyoccurred in cactus-breeding Drosophila mojavensis, widespreadD. virilis, D. pseudoobscura and D. persimilis. In contrast,a slight expansion of the gustatory gene subfamily wasobserved in the melanogaster subgroup. Generally, our datasupport the widely accepted assumption that lineage-specific gene expansions and gene loss shaped the chemo-sensory repertoires in insects, although the extent of theseprocesses is much less compared with extensive duplica-tions of the ORs in vertebrates and chemoreceptors innematodes. Pseudogenization, in general, is also low ininsects. What predicts these expansion and duplicationevents? McBride (2007) proposed that specialization increasespseudogenization because the specialist species D. sechelliahas lost functionality in more loci than its relatives. Ouranalyses do not unambiguously support a role of spe-cialization since high rates of pseudogenization are alsoseen in generalist species (Fig. 4b). Cactophilic D. mojavensisand D. erecta (specialized on Pandanus) do not show elevatedrates of pseudogenization or duplication. Drosophila mojavensisshows frequent changes in its host affinity with colonizationhistory (Etges 1992; Newby & Etges 1998), and the sequencedstrain is from Santa Catalina where it is restricted to onehost species. Drosophila erecta’s specialization is seasonal

depending on Pandanus abundance (Rio et al. 1983), so itmight be the case that this is a less specialist species thanD. sechellia and, as stated above, D. grimshawi is also variablein specialization. The strongest predictor that we see is thatgenome size predicts the extent of gene duplication (Fig. 4a).Drosophila sechellia is also an endemic so either endemismor specialization could have contributed to the patternsseen here and by McBride (2007). We find a significanteffect of endemism rather than specialization on theproportion of pseudogenes, although only two endemicspecies are available for study. It is likely that endemismwill lead to relaxed selection followed by selective expansionof particular loci during colonization, and perhaps anoverall increase in genome size (Nardon et al. 2005)(although D. sechellia does not have a relatively largegenome) so these factors are probably interrelated andresults from more species will be required to disentanglethe effects of specialization and endemism.

Overall, we find evidence that positive selection has actedon very few loci after correcting for multiple comparisons.Curiously, although individual loci are rarely significant,we found that 20 loci out of 107 (19%) yield an uncorrectedP value less than 0.05, which is a significantly higherproportion than 5% (G = 25.67, 1d.f., P < 0.001), perhapssuggesting that a fair proportion of the loci are under positiveselection but the individual tests lack power. Studies ofchemoreceptor evolution in other animals also suggeststhat the proportion of loci showing evidence for positiveevolution is low (Gimelbrant et al. 2004; Gilad et al. 2005;Nielsen et al. 2005; Thomas et al. 2005). For instance, itwas reported that only 5% of chemosensory receptors innematodes evolved through adaptive evolution (Thomaset al. 2005). The overall pattern of variation in ω is consistentwith purifying selection being the most pervasive form ofselection on these loci, with stronger selection on OR lociand weaker selection on loci which have undergoneduplication within the genus. McBride (2007) foundevidence that the spatial location of codons under selectiondiffered between GRloci in D. sechellia and D. melanogasteror D. simulans, probably because of less strong purifyingselection, although interestingly, this was not seen for ORloci. Our estimates of selection are similar between thegene families. A gustatory locus which has been shown tobe a pheromone receptor (Gr68a, Bray & Amrein 2003);plus two suggested to share this function (Gr32a and Gr39a,Amrein 2004) are included in our list of loci potentiallyunder selection. The chemoreceptor family of Drosophilaplays a role in a diverse range of behaviours contributingto adaptation and divergence between these species. Someof the genes with currently unknown functions potentiallycontribute to ecological adaptation or pheromone recognition.Comparative genomic studies such as this are essentialguides to manipulative and functional studies of thesekey loci.

1656 A . G A R D I N E R E T A L .


Acknowledgements

We thank Richard Emes for helpful comments on usage of thepaml package, Michiel Plugge, James Artherton, Sharon Fury andTanya Sneddon for contributions to project development. Thegenomic sequences of D. erecta, D. ananassae, D. mojavensis, D. virilisand D. grimshawi were provided by Agencourt Bioscience Corpora-tion. The genomes of D. simulans and D. yakuba were sequenced byWashington University (St. Louis), while D. sechellia and D. persimilisby the Broad Institute. The Baylor University provided the data forthe D. pseudoobscura genome. D. melanogaster genome was providedby the Berkeley Drosophila Genome Project and Celera. This workwas supported by the Natural Environment Research Council UK,grant NE/C003187/1. D.B. was supported by a Research CouncilsUK Academic Fellowship. This is a companion paper to Drosophila12 Genomes Consortium (2007).

References

Alioto T, Ngai J (2005) The odorant receptor repertoire of teleostfish. BMC Genomics, 6, 173.

Altschul SF, Madden TL, Schaffer AA et al. (1997) Gapped blastand psi-blast: a new generation of protein database searchprograms. Nucleic Acids Research, 25, 3389–3402.

Amrein H (2004) Pheromone perception and behavior in Drosophila.Current Opinion in Neurobiology, 14, 435–442.

Benjamini Y, Hochberg Y (1995) Controlling the false discoveryrate — a practical and powerful approach to multiple testing.Journal of the Royal Statistical Society. Series B-Methodological, 57,289–300.

Birney E, Clamp M, Durbin R (2004) genewise and genomewise.Genome Research, 14, 988–995.

Bray S, Amrein H (2003) A putative Drosophila pheromone receptorexpressed in male-specific taste neurons is required for efficientcourtship. Neuron, 39, 1019–1029.

Buck L, Axel R (1991) A novel multigene family may encodeodorant receptors: a molecular basis for odor recognition. Cell,65, 175–187.

Clyne PJ, Warr CG, Freeman MR et al. (1999) A novel family ofdivergent seven-transmembrane proteins: candidate odorantreceptors in Drosophila. Neuron, 22, 327–338.

Clyne PJ, Warr CG, Carlson JR (2000) Candidate taste receptors inDrosophila. Science, 287, 1830–1834.

DeSalle R (1992) The origin and possible time of divergence of theHawaiian Drosophilidae: evidence from DNA sequences.Molecular Biology and Evolution, 9, 905–916.

Drosophila 12 Genomes Consortium (2007) Evolution of genes andgenomes on the Drosophila phylogeny. Nature, 450, 203–218.

Etges WJ (1992) Premating isolation is determined by larvalsubstrates in cactophilic Drosophila mojavensis. Evolution, 46,1945–1950.

Fox AN, Pitts RJ, Robertson HM, Carlson JR, Zwiebel LJ (2001)Candidate odorant receptors from the malaria vector mosquitoAnopheles gambiae and evidence of down-regulation in responseto blood feeding. Proceedings of the National Academy of Sciences,USA, 98, 14693–14697.

Freckleton RP, Harvey PH, Pagel M (2002) Phylogenetic analysisand comparative data: a test and review of evidence. AmericanNaturalist, 160, 712–726.

Freitag J, Krieger J, Strotmann J, Breer H (1995) Two classes ofolfactory receptors in Xenopus laevis. Neuron, 15, 1383–1392.

Gaillard I, Rouquier S, Giorgi D (2004) Olfactory receptors. Cellularand Molecular Life Sciences, 61, 456–469.

Gao Q, Chess A (1999) Identification of candidate Drosophilaolfactory receptors from genomic DNA sequence. Genomics,60, 31–39.

Gilad Y, Man O, Glusman G (2005) A comparison of the humanand chimpanzee olfactory receptor gene repertoires. GenomeResearch, 15, 224–230.

Gilbert DG (2007) DroSpeGe: rapid access database for new Dro-sophila species genomes. Nucleic Acids Research, 35, D480–D485.

Gimelbrant AA, Skaletsky H, Chess A (2004) Selective pressureson the olfactory receptor repertoire since the human-chimpanzeedivergence. Proceedings of the National Academy of Sciences, USA,101, 9019–9022.

Glusman G, Yanai I, Rubin I, Lancet D (2001) The complete humanolfactory subgenome. Genome Research, 11, 685–702.

van der Goes van Naters W, John R, Carlson JR (2007) Receptorsand neurons for fly odors in Drosophila. Current Biology, 17,606–612.

Guo S, Kim J (2007) Molecular evolution of Drosophila odorantreceptor genes. Molecular Biology and Evolution, 24, 1198–1207.

Hill CA, Fox AN, Pitts RJ et al. (2002) G protein-coupled receptorsin Anopheles gambiae. Science, 298, 176–178.

Hulsen T, Huynen MA, de Vlieg J, Groenen PMA (2006) Bench-marking ortholog identification methods using functionalgenomics data. Genome Biology, 7, R31.

Kurtovic A, Widmer A, Dickson BJ (2007) A single class of olfactoryneurons mediates behavioural responses to a Drosophila sexpheromone. Nature, 446, 542–546.

Lynch M, Conery JS (2000) The evolutionary fate and con-sequences of duplicate genes. Science, 290, 1151–1155.

Matsuo T, Sugaya S, Yasukawa J, Aigaki T, Fuyama Y (2007)Odorant-binding proteins OBP57d and OBP57e affect tasteperception and host-plant preference in Drosophila sechellia.PLoS Biology, 5, 985–996.

Matzkin LM, Watts TD, Bitler BG, Machado CA, Markow TA (2006)Functional genomics of cactus host shifts in Drosophila mojavensis.Molecular Ecology, 15, 4635–4643.

McBride CS (2007) Rapid evolution of smell and taste receptorgenes during host specialization in Drosophila sechellia. Proceedingsof the National Academy of Sciences, USA, 104, 4996–5001.

Mombaerts P (1999) Seven-transmembrane proteins as odorantand chemosensory receptors. Science, 286, 707–711.

Nakagawa T, Sakurai T, Nishioka T, Touhara K (2005) Insectsex-pheromone signals mediated by specific combinations ofolfactory receptors. Science, 307, 1638–1642.

Nardon C, Deceliere G, Loevenbruck C et al. (2005) Is genome sizeinfluenced by colonization of new environments in dipteranspecies? Molecular Ecology, 14, 869–878.

Nef S, Allaman I, Fiumelli H, De Castro E, Nef P (1996) Olfactionin birds: differential embryonic expression of nine putative odorantreceptor genes in the avian olfactory system. Mechanisms ofDevelopment, 55, 65–77.

Newby BD, Etges WJ (1998) Host preference among populationsof Drosophila mojavensis (Diptera: Drosophilidae) that use differenthost cacti. Journal of Insect Behavior, 11, 691–712.

Nielsen R, Bustamante C, Clark AG et al. (2005) A scan for posi-tively selected genes in the genomes of humans and chimpanzees.PLoS Biology, 3, e170.

Niimura Y, Nei M (2006) Evolutionary dynamics of olfactory andother chemosensory receptor genes in vertebrates. Journal ofHuman Genetics, 51, 505–517.



Nozawa M, Nei M (2007) Evolutionary dynamics of olfactoryreceptor genes in Drosophila species. Proceedings of the NationalAcademy of Sciences, USA, 104, 7122–7127.

Ohno S (1970) Evolution by Gene Duplication. Springer-Verlag,Heidelberg, Germany.

Olender T, Fuchs T, Linhart C et al. (2004) The canine olfactorysubgenome. Genomics, 83, 361–372.

Organ CL, Shedlock AM, Meade A, Pagel M, Edwards SV (2007)Origin of avian genome size and structure in non-avian dinosaurs.Nature, 446, 180–184.

Pagel M (1997) Inferring evolutionary processes from phylogenies.Zoologica Scripta, 26, 331–348.

Pagel M (1999) Inferring the historical patterns of biologicalevolution. Nature, 401, 877–884.

Piano F, Craddock EM, Kambysellis MP (1997) Phylogeny of theisland populations of the Hawaiian Drosophila grimshawicomplex: evidence from combined data. Molecular Phylogeneticsand Evolution, 7, 173–184.

Powell JR (1997) Progress and Prospects in Evolutionary Biology:The Drosophila Model. Oxford University Press, Oxford, UK.

Quignon P, Giraud M, Rimbault M et al. (2005) The dog and ratolfactory receptor repertoires. Genome Biology, 6, R83.

Remm M, Storm CEV, Sonnhammer ELL (2001) Automaticclustering of orthologs and in-paralogs from pairwise speciescomparisons. Journal of Molecular Biology, 314, 1041–1052.

Rio B, Couturier G, Lemeunier F, Lachaise D (1983) Evolution of aseasonal specialization in Drosophila erecta (Dipt — Drosophilidae).Annales de la Societe Entomologique de France, 19, 235–248.

Robertson HM (1998) Two large families of chemoreceptorgenes in the nematodes Caenorhabditis elegans and Caenorhabditisbriggsae reveal extensive gene duplication, diversification,movement, and intron loss. Genome Research, 8, 449–463.

Robertson HM, Thomas JH (2006) The putative chemoreceptorfamilies of C. elegans. In: Wormbook (ed. Greenwald I). Doi:10.1895/wormbook.1.7.1.

Robertson HM, Wanner KW (2006) The chemoreceptor superfamilyin the honey bee Apis mellifera: expansion of the odorant, but notgustatory, receptor family. Genome Research, 16, 1395–1403.

Robertson HM, Warr CG, Carlson JR (2003) Molecular evolutionof the insect chemoreceptor gene superfamily in Drosophilamelanogaster. Proceedings of the National Academy of Sciences, USA,100, 14537–14542.

Ronquist F, Huelsenbeck J (2003) mrbayes 3: Bayesian phylogeneticinference under mixed models. Bioinformatics, 19, 1572–1574.

Scott K, Brady R, Cravchik A et al. (2001) A chemosensory genefamily encoding candidate gustatory and olfactory receptors inDrosophila. Cell, 104, 661–673.

Singh SR, Singh BN (2003) Behavioral genetics of Drosophilaananassae. Genetics and Molecular Research, 2, 394–409.

Singh ND, Arndt PF, Petrov DA (2006) Minor shift in backgroundsubstitutional patterns in the Drosophila saltans and willistonilineages is insufficient to explain GC content of coding sequences.BMC Biology, 4, 37.

Thomas JH, Kelley JL, Robertson HM, Ly K, Swanson WJ (2005)Adaptive evolution in the SRZ chemoreceptor families ofCaenorhabditis elegans and Caenorhabditis briggsae. Proceedingsof the National Academy of Sciences, USA, 102, 4476–4481.

Thompson JD, Higgins DG, Gibson TJ (1994) clustal w: improvingthe sensitivity of progressive multiple sequence alignment throughsequence weighting, position-specific gap penalties and weightmatrix choice. Nucleic Acids Research, 22, 4673–4680.

Troemel ER, Chou JH, Dwyer ND, Colbert HA, Bargmann CI(1995) Divergent seven transmembrane receptors are candidatechemosensory receptors in C. elegans. Cell, 83, 207–218.

Voight BF, Kudaravalli S, Wen XQ, Pritchard JK (2006) A map ofrecent positive selection in the human genome. PLoS Biology, 4,446–458.

Yang Z (1997) paml: a program package for phylogenetic analysisby maximum likelihood. Computer Applications in the Biosciences,13, 555–556.

Yang Z, Bielawski JP (2000) Statistical methods for detectingmolecular adaptation. Trends in Ecology & Evolution, 15, 496–503.

Zhang X, Firestein S (2002) The olfactory receptor gene superfamilyof the mouse. Nature Neuroscience, 5, 124–133.

Anastasia Gardiner is a postdoctoral researcher interested incomparative genomics and gene family evolution. Daniel Barkeris a Research Councils UK Academic Fellow whose research ismostly in bioinformatics, comparative genomics and phylogeny.Bill Jordan is interested in the role of genetics in chemical signalproduction and perception and how it influences population andspecies differentiation. Roger Butlin works on the evolution andgenetics of pre- and postzygotic reproductive isolation, whereasMike Ritchie limits himself to premating isolation and populationgenetics.

Supplementary material

The following supplementary material is available for this article:

Table S1 Phylogenetically corrected, second-order partial corre-lations

Table S2 Gustatory (GR) and olfactory (OR) receptor genes ofDrosophila melanogaster (Dmel) used in tblastn searches ofgenome sequences for other Drosophila spp.

Table S3 Gustatory receptor genes in 11 Drosophila species:Dsim, D. simulans; Dsec, D. sechellia; Dere, D. erecta; Dyak, D. yakuba;Dana, D. ananassae; Dpse, D. pseudoobscura; Dper, D. persimilis;Dwil, D. willistoni; Dmoj, D. mojavensis; Dvir, D. virilis; Dgri,D. grimshawi. Number of pseudogenes is given in parentheses andincluded in the number of loci. Partial sequences in D. grimshawigenome are indicated as nc* (not complete)

Table S4 Olfactory receptor genes in 10 Drosophila species: Dsim,D. simulans; Dsec, D. sechellia; Dere, D. erecta; Dyak, D. yakuba;Dana, D. ananassae; Dpse, D. pseudoobscura; Dper, D. persimilis;Dmoj, D. mojavensis; Dvir, D. virilis; Dgri, D. grimshawi. Number ofpseudogenes is given in parentheses and included in the numberof loci

This material is available as part of the online article from:http://www.blackwell-synergy.com/doi/abs/10.1111/j.1365-294X.2008.03713.x(This link will take you to the article abstract).

Please note: Blackwell Publishing are not responsible for the contentor functionality of any supplementary materials supplied by theauthors. Any queries (other than missing material) should bedirected to the corresponding author for the article.

http://www.blackwell-synergy.com/doi/abs/10.1111/j.1365-294X.2008.03713.x

10.1895/wormbook.1.7.1

kemo inggris

Documents