The Role of Lineage-Specific Gene Family Expansion in the Evolution of Eukaryotes

12 Pages·2002·0.37 MB·English

Letter
The Role of Lineage-Specific Gene Family
Expansion in the Evolution of Eukaryotes
Olivier Lespinet, Yuri I. Wolf, Eugene V. Koonin,1 and L. Aravind
NationalCenterforBiotechnologyInformation,NationalLibraryofMedicine,NationalInstitutesofHealth,
Bethesda,Maryland20894,USA
A computational procedure was developed for systematic detection of lineage-specific expansions (LSEs) of
protein families in sequenced genomes and applied to obtain a census of LSEs in five eukaryotic species, the
yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, the nematode Caenorhabditis elegans, the fruit fly
Drosophila melanogaster, and the green plant Arabidopsis thaliana. A significant fraction of the proteins encoded in
each of these genomes, up to 80% in A. thaliana, belong to LSEs. Many paralogous gene families in each of the
analyzed species are almost entirely comprised of LSEs, indicating that their diversification occurred after the
divergence of the major lineages of the eukaryotic crown group. The LSEs show readily discernible patterns of
protein functions. The functional categories most prone to LSE are structural proteins, enzymes involved in an
organism’s response to pathogens and environmental stress, and various components of signaling pathways
responsible for specificity, including ubiquitin ligase E3 subunits and transcription factors. The functions of
several previously uncharacterized, vastly expanded protein families were predicted through in-depth protein
sequence analysis, for example, small-molecule kinases and methylases that are expanded independently in the
flyandinthenematode.ThefunctionsofseveralothermajorLSEsremainmysterious;theseproteinfamiliesare
attractive targets for experimental discovery of novel, lineage-specific functions in eukaryotes. LSEs seem to be
one of the principal means of adaptation and one of the most important sources of organizational and
regulatorydiversityincrown-groupeukaryotes.
[Supplemental material is available online at ftp://ncbi.nlm.nih.gov/pub/aravind/expansions, and http://www.
genome.org.]
The eukaryotic crown group (the unresolved assemblage of eages, are largely constructed from the same set of protein
lineages in the eukaryotic tree, which includes plants, ani- domains,andarebasedonthesamearchitecturalprinciples
mals,fungi,andsomeprotists,asopposedtoearlybranching (Chervitzetal.1998;AravindandSubramanian1999;Rubin
eukaryotes,whichareallunicellularprotists),althoughonly etal.2000;Landeretal.2001).
representingtheproverbialtipoftheeukaryoticphylogenetic This unity notwithstanding, preliminary comparative
iceberg,encompassesaremarkablevarietyoforganisms(Pat- studies on the sequenced eukaryotic genomes also provided
terson1999;DacksandDoolittle2001).Thisdiversityisap- clues as to what evolutionary phenomena might underlie
parentinbothmorphologicalandbiochemicalfeaturesofthe theirdiversity.Attheleveloftheproteinsetsencodedinthe
crown group that spans the entire range from unicellular crown-group genomes, the main contributing forces appear
yeasts and chlorophytes, through facultatively multicellular to be the emergence of new domain architectures through
slimemolds,togenuinemulticellularorganisms,plants,ani- domainaccretionanddomainshuffling,lineage-specificgene
mals,andfungi(Soginetal.1996;Patterson1999).Thecom- loss,andlineage-specificexpansionofproteinfamilies(Ara-
plete,ornearlycomplete,genomesequencesfromthreema- vindandSubramanian1999;Aravindetal.2000;Rubinetal.
jorbranchesofthecrowngroup,plants,animals,andfungi 2000;Landeretal.2001).Lineage-specificexpansion(LSE)is
are starting to provide the first molecular explanations for definedinrelativeterms,astheproliferationofaproteinfam-
both their unity and diversity. From one viewpoint, the ilyinaparticularlineage,relativetothesisterlineage,with
crown-groupeukaryotesareremarkablyuniforminthatthey whichitiscompared(Jordanetal.2001).Thus,iftwosister
share a large set of conserved orthologs in the core compo- lineages, for example, Drosophila and Caenorhabditis repre-
nentsoftheiressentialfunctionalsystems,suchasthosein- sentinginsectsandnematodes,respectively,arecompared,all
volved in DNA replication and repair, most aspects of RNA protein-family proliferation events (duplications to n-
metabolism, cytoskeletal organization, protein degradation, plications)thatoccurredineitheroftheselineagesaftertheir
andsecretion(Chervitzetal.1998;Rubinetal.2000;Lander separationareconsideredLSEs.
etal.2001).Furthermore,componentsofthesignaltransduc- Preliminary analysis of proteins from the crown-group
tion pathways, structural and regulatory components of the eukaryotic genomes revealed some tangible correlations be-
nucleus, and pre-mRNA processing complexes, although tween LSE and emergence of new biological functions, re-
showingcleardifferencesbetweenthemajorcrown-grouplin- sponse to diverse environmental pressures, and organiza-
tionalcomplexity.SomeofthemoststrikingcasesofLSEare
1Correspondingauthor.
relatedtopathogenandstressresponseandinclude,among
[email protected];FAX[phone].
Articleandpublicationareathttp://www.genome.org/cgi/doi/10.1101/ otherfamilies,expansionsoftheimmunoglobulinsuperfam-
gr.174302. ilyassociatedwiththevertebrateimmunesystem,AP-ATPases
1048 GenomeResearch 12:1048–1059©2002byColdSpringHarborLaboratoryPressISSN1088-9051/01$5.00;www.genome.org
www.genome.org
Lineage-Specific Gene Expansion in Eukaryotes
involvedinplantdiseaseresistance(Hulbertetal.2001),and exclusionofhomologsfromotherspeciesandparalogsfrom
the cytochrome P450 family, which participates in detoxifi- thesamespeciesthatdonotbelongtothegivenLSC(Fig.1;
cationsystemsinbothplantsandanimals(Nelson1999;Tijet Supplementary Material available online at ftp://ncbi.nlm.
et al. 2001). Transcription factors represent another func- nih.gov/pub/aravind/expansions, and http://www.genome.
tionalcategoryofproteinsthattendtoshowwidespreadLSE: org).Thus,theclustersgeneratedbytheautomaticprocedure
the independent expansions of the POZ–C2H2 and C4DM– usedhereappearedtorepresentpredominantly,ifnotexclu-
C2H2 fusions in insects, the nuclear hormone receptors in sively,authenticLSEsand,therefore,couldbeutilizedreliably
nematodes,andtheKRAB-domain-fusedZn-fingersinverte- forquantitativeandqualitativeanalysesofthisphenomenon.
brates,apparentlymadesubstantialcontributionstotheevo- Certainlimitationsrelatedtothecurrentstateofsequencing
lutionofdevelopmentalanddifferentiationfeaturesspecific andannotationoftheeukaryoticgenomesneedtobekeptin
to each of these lineages (Sluder et al. 1999; Aravind et al. mind when interpreting these clusters. Only one genome,
2000;Riechmannetal.2000;Coulsonetal.2001;Landeret that of S. cerevisiae, should be considered truly complete,
al.2001). whereasinothers,somegenesareobviouslystillmissing,for
Despite a wealth of anecdotal information, we are un- example,thosethatresideinheterochromatinicregions.Fur-
awareofasystematiccomparativeanalysisofLSEsineukary- thermore,giventheknownproblemswithgenepredictionin
oticgenomes.Withthisobjective,wedevisedaprocedureto plant and animal genomes, we removed nearly identical se-
systematicallydetectLSEs.HavingidentifiedLSEsinfiveeu- quencespriortotheLSCanalysis(seeMethods).Thiselimi-
karyotic proteomes, those of Saccharomyces cerevisiae, Schizo- natedpotentialredundancy,butsometrue(nearlyidentical)
saccharomycespombe,Caenorhabditiselegans,Drosophilamela- paralogsresultingfromrecentduplicationscouldhavebeen
nogaster, and Arabidopsis thaliana, we predicted, wherever lostintheprocess.Giventhisprocedure,theresultspresented
feasible, the biochemical or biological functions of the lin- hereshouldbeconsideredconservativeestimatesofthenum-
eage-specificclusters(LSC)andexploredtheirpotentialroles berofgenesinLSCs.Ontheotherendofthespectrum,ex-
inthediversificationofthecrowngroup.Here,wepresenta tremely diverged members of LSCs (or even entire LSCs),
systematic analysis of the demography of LSEs and provide which retain minimal sequence conservation, could have
evidence for a major involvement of LSEs in the generation beenmissedbythisanalysis.
ofthediversityofbiologicalfunctionsinmulticellulareukary- The two ascomycete yeasts, S. pombe and S. cerevisiae,
otes. were the closest pair of sister lineages compared. The two
animals,D.melanogasterandC.elegans,representedaslightly
greater phylogenetic divergence relative to each other,
RESULTS AND DISCUSSION
whereas the plant A. thaliana represented an even deeper
branch with respect to animals and fungi. Thus, the LSCs
Identification and Validation of Candidate
fromeachofthesespeciesenabledustoexaminetheroleof
Lineage-Specific Clusters
LSEsindiversificationofeukaryotesatdifferentlevelsofevo-
UsingtheclusteringproceduredescribedintheMethodssec- lutionarydivergence.
tion, we delineated candidate LSCs for five eukaryotic ge-
nomes. The automatically generated LSCs were further sur-
Proteome-Wide Demography of Lineage-Specific
veyedforfalsepositives,thatis,proteinsthatwereunrelated
to the rest of the proteins in the cluster, by using BLAST Family Expansion
searchesandmultiplealignments.Asubsetoffalse-positives The detected LSEs encompassed between ∼20% of the pro-
arosefromcompositionallybiasedsegmentsthatescapedfil- teome(theyeasts)and∼80%(A.thaliana)(Fig.2A).Oneofthe
tering during the automatic process. The presence of some causesforthisdiverserangeofLSEsappearstobethephylo-
false-positiveswasmainlyduetooneormoreoftheproteins geneticdistancefactor;thetwoyeastspecieshaveaccruedfar
inaclustercontainingmultipledomainsorbeingartificially fewerLSEsafterdivergingfromtheircommonancestorcom-
fusedtoanotherprotein.Themajorityofsuchfalse-positives paredwithA.thaliana,whichhasnoclosesisterlineagesin
weredetectedamongA.thalianaproteins,inwhichgenepre- theanalyzedsetofgenomesandhas,accordingly,gainedthe
diction errors resulted in artificial fusions of distinct genes. greatest number of expansions after its divergence from the
Onseveraloccasions,theseartificialgenefusionsresultedin commonancestorwithfungiandanimals.Positivelinearcor-
an erroneous merger of one or more distinct clusters; these relations, with moderate-to-strong significance, were ob-
weremanuallyseparated.Additionally,afewsmallerclusters servedbetweentheproteomesizeandeachofthefollowing:
thatbelongedtoalargerLSE-specificexpansionweremerged. (1)fractionofproteinscontainedinLSCs(Fig.2A),(2)num-
On average, ∼9% of the LSCs of size greater than two were berofLSCs(Fig2B),and(3)averagenumberofproteinsper
subjectedtomanualcorrections. LSC (Fig. 2C). The majority of the clusters in each species
Theautomaticprocedureusedfordelineatingcandidate consistedoftwomembers.Ineachcase,thenumberoftwo-
LSCs included single-linkage clustering of proteins by se- memberclustersshowedanegativecorrelationwiththepro-
quencesimilarityandanultrametrictreeconstructionusing teomesize,whereasthenumberofclusterswiththreeormore
UPGMA (see Methods). These methods accurately reproduce members showed a positive correlation with the proteome
phylogenetic relationships only under the strict molecular size (Fig. 2D). Thus, larger proteomes had more proteins in
clockhypothesis.Therefore,toverifythephylogeneticcoher- largerLSCsattheexpenseoftwo-memberLSCs.Foreachspe-
enceofthecandidateclusters,10ofthecandidateLSCsfrom cies,thedistributionoftheLSCsbythenumberofmembers
each analyzed species that consisted of 4 or more members followedthenegativepowerlaw:P(k)=ck(cid:1)(cid:1)inwhichP(k)is
andhadhomologsinotherspecieswerechosenforphyloge- thefrequencyoffamilieswithexactlykmembersandcand(cid:1)
neticanalysis.Ineachtree,theproteinsfromthecandidate are constants (Fig. 3). The differences between the slopes of
LSCgroupedtogetherand,in48ofthe50cases,thisgrouping thesepowerlawdistributions(indouble-logarithmiccoordi-
was strongly supported by bootstrap analysis (>70%) to the nates)werecompatiblewiththeaforementionedcorrelations
Genome Research 1049
www.genome.org
Lespinet et al.
Figure1 Phylogeneticanalysisofselectedeukaryoticlineage-specificexpansions.Groupssupportedbyabootstrapvalue>70%arecoloredpink
forDrosophilamelanogaster,redforHomosapiens,orangeforCaenorhabditiselegans,greenforArabidopsisthaliana,andyellowforSchizosaccha-
romycespombe.(A)Prolylhydroxylases.(B)Smallmoleculekinases(Chstandsforcholinekinase).(C)Patched-likeprotein.(D)MAP-Kinases.(E)
P450familyhydroxylases.(F)MBOATmembraneacyltransferases.(At)Arabidopsisthaliana;(Bs)Bacillussubtilis;(Ce)Caenorhabditiselegans;(Dd)
Dictyosteliumdiscoideum;(Dm)(Drosophilamelanogaster;(Hs)Homosapiens;(Pbcv1)ParameciumbursariaChlorellavirus1;(Rs)Ralstoniasola-
nacearum;(Sa)Staphylococcusaureus;(Sc)Saccharomycescerevisiae;(Sm)Sinorhizobiummeliloti;(Sp)Schizosaccharomycespombe.Completetree
descriptions(fulllistsofGInumbersorgenenames,andbootstrapvalues)areavailableintheSupplementaryMaterialonlineatftp://ncbi.nlm.nih.gov/
pub/aravind/expansions,andhttp://www.genome.org..
between the degree of clustering and proteome size, that is, fromtheclosestsisterlineageincludedintheanalysis.LSCs
theyeastLSCsshowedthesteepestdecay,whereasthosefrom with EC=1 are those families that have been invented de
A. thaliana had the flattest distribution (Fig. 3). This is also novo and proliferated thereafter in a particular lineage. The
consistentwithearlierobservationsthat,ingeneral,thesize relativeabundanceofLSCsintheECrangebetween0and0.9
distributionofparalogousproteinfamiliesinproteomesfol- isroughlyconstantforalltaxaconsideredhere,withslightly
lowed the power law decay (Huynen and van Nimwegen >5%oftheLSCsineachofthebinsofsize0.1inthisrange
1998; Qian et al. 2001). These findings suggest that LSCs (Fig.4).Notably,∼40%(onaverage)oftheLSCspresentina
evolvedlargelythroughastochasticprocessofgeneduplica- givenproteomewereintheECrangeof0.9to1(Fig.4).Thus,
tionwherebytheprobabilityofduplicationwithinaclusterat nearlyone-halfoftheparalogousproteinclustersencodedin
anygiventimeisproportionaltothesizeofthecluster,rather eukaryotic genomes have been generated almost entirely
thanthroughgenome-scaleduplications. through LSE. This applied to the full range of evolutionary
TocharacterizetheroleofLSEsintheevolutionofthe distancesexploredhereandtherewasnoobviousdependence
respective classes of paralogous proteins in each lineage, we ontheevolutionarydepthatwhichLSEswereidentified;the
devisedtheexpansioncoefficient(EC),whichistheratioof fraction of paralogous classes contained in these exclusive
the number of proteins in LSCs to the total membership of LSCswasevengreaterintheyeastS.cerevisiaethanitwasin
thegivenclassofparalogsinagivenproteome.TheECisa A.thaliana(Fig.4).Thisobservation,togetherwiththecorre-
measure of the fraction of a given paralogous class that has lations between proteome size and different parameters of
evolvedthroughLSEafterthedivergenceofthegivenlineage LSEs (Fig. 2), suggests that the ancestral core set of proteins
1050 Genome Research
www.genome.org
Lineage-Specific Gene Expansion in Eukaryotes
eralimplausibilityofahighlycom-
plex common ancestor for the
crown group, this mechanism for
the evolution of apparent LSEs
would necessarily entail indepen-
dentgenelossesinthesameparalo-
gousfamilyinmultiplelineages,as
opposed to a single expansion.
Therefore, the lineage-specific du-
plicationscenarioismoreparsimo-
nious than the scenario based on
thelineage-specificlosses.
Analysis of the top 25 LSEs
with EC=1 from all proteomes
pooledtogether,indicatedthatthe
majorityofthemare(cid:2)-helicalpro-
teinsorhaveconservedpatternsof
histidines and cysteines. Typical
Figure 2 Linear correlation between the proteome size and parameters of eukaryotic lineage- examples include the (cid:2)-helical
specificexpansion(LSE)infiveeukaryoticspecies.Correlationcoefficients(r)andsignificancelevels(P)
nonspecific lipid-transfer protein
were determined using ordinary least square linear regression. (At) Arabidopsis thaliana; (Ce) Cae-
norhabditiselegans;(Dm)Drosophilamelanogaster;(Sc)Saccharomycescerevisiae;(Sp)Schizosaccharo- inplants,theC4DmdomaininD.
mycespombe.(A)Theproteomesize(X-axis)isplottedagainstthepercentageoftheproteomemade melanogaster that chelates cations
upofLSEs.(B)Theproteomesize(X-axis)isplottedagainstthenumberoflineage-specificclusters.(C) through conserved cysteines, and
Theproteomesize(X-axis)isplottedagainstthemeannumberofproteinsinlineage-specificclusters. the T20D4.15 family of disulfide-
(D)Theproteomesize(X-axis)isplottedagainstthepercentageofduplication((cid:2))andthepercentage
ofn-plication(n>=3)((cid:1))amongtheLSCs. bonded secreted proteins from C.
elegans. Thus, de novo emergence
of protein domains that substan-
inherited by the crown-group lineages from their last com- tiallycontributedtoLSEsappearstohaveinvolvedprimarily
monancestorcontainedfewparalogscomparedwiththeex- inventionofstructurallysimplefolds.Thesefoldscouldhave
tant proteomes. Subsequent to the divergence of the indi- evolvedthroughcompactionoflong(cid:2)-helicalcoiledcoilsor
viduallineages,manygenesinheritedfromthecommonan- throughdisulfide-bond-ormetal-supportedstabilizationme-
cestor as well as gene families invented de novo have diatedbyafewstrategicallyplaced,conservedcysteinesand/
undergone one or more rounds of duplication. This process or histidines. Invention of such simple domains could have
seemstohavebeenparticularlyactiveinthegenerationofthe beenmoreexpedientthanemergenceofcomplex(cid:2)/(cid:3)struc-
largeproteomesofmulticellulareukaryotesandprobablypro- turesthatrequireseveralspecificstabilizinginteractionstobe
videdthemwiththerawmaterialfortheircellulardifferen- fixed(AravindandKoonin2000).
tiation.Inprinciple,itcouldbearguedthattheancestorhad
Biological Significance
asmanyparalogousfamiliesasthemostcomplexoftheex-
of Lineage-Specific Expansions
tantgenomesorevenmore,andtheappearanceofLSEhad
beencreatedbylineage-specificgeneloss,whichiscommon Theaboveobservationsshowthat,quantitatively,LSEsarea
intheevolutionofatleastsomeeukaryoticlineages(Aravind majorcomponentofthedifferencesbetweentheproteomes
etal.2000;Braunetal.2000).However,apartfromthegen- ofvariouseukaryotictaxa.Newparalogousfamiliescouldpro-
videthematerialforspecificadap-
tations and for evolution of new
functional systems. In qualitative
terms,wesoughttoinvestigatethe
biological significance of LSEs by
identifying conserved domains,
subcellular localization signatures,
such as signal peptides and trans-
membrane regions, and other fea-
turesofproteinsinLSCsthatmight
allowpredictionoftheirfunctions
(when less than obvious). These
identificationsforthetopfiveLSCs
in each organism are shown in
Table 1. We categorized the LSCs
intobroadfunctionalclassestodis-
cern global functional trends and
alsoinvestigatedindividualLSCsin
anattempttogainamoredetailed
Figure3 Sizedistributionofthelineage-specificclustersinthreeeukaryoticspecies.(Blue)Schizo-
understanding of their actual bio-
saccharomycespombe;(pink)Caenorhabditiselegans;(green)Arabidopsisthaliana.Clustersize(X-axis)
isplottedagainstthenumberofLSCsindoublelogarithmiccoordinates.Theequationsofthepower logical roles (Table 2; Supplemen-
lawdistributionfittingthelinearpartofthedataareshownonthegraph. taryMaterialavailableonline.).
Genome Research 1051
www.genome.org
Lespinet et al.
Figure4 Distributionoflineage-specificclustersbyExpansionCoefficient(EC).TheX-axisshowsrangesofECvalues(seetext)andtheY-axis
showsthepercentageofLSCswithineachECrange.(Yellow)Schizosaccharomycespombe;(orange)Saccharomycescerevisiae;(pink)Drosophila
melanogaster; (red) Caenorhabditis elegans; (green) Arabidopsis thaliana. In each class, the average value of the five species is indicated by a
horizontalline.
Although LSEs occurred in most biological functional adaptationofthenematodes(Johnstone2000).Similarly,in
classes,LSCswithpredictedorganism-specificfunctions,such D. melanogaster and Arabidopsis, prominent LSEs are, respec-
as pathogen and stress response, transcription regulation, tively,theinsectcuticularproteins(Andersenetal.1995)and
controlledproteindegradationmediatedbytheubiquitinsys- pectin/cellulose biosynthesis enzymes (Willats et al. 2001),
tem,proteinmodification,signaltransduction,chemorecep- bothofwhicharecriticalfortheformationofmorphological
tion, and small molecule metabolism were most abundant featuresuniquetotheselineages.Typically,theseproteinsare
(Tables1and2).Atypicalexampleofanexpansionrelatedto required in large amounts as structural components of the
an organism-specific function is that of the C. elegans colla- respective organisms; hence, these lineage-specific expan-
gens,whicharerequiredforcuticleformation,acharacteristic sionscouldprincipallyhelpinincreasedproductionofthese
Table1. TheTopFiveLineage-SpecificGeneFamilyExpansionsinFiveEukaryotes
Saccharomyces Schizosaccharomyces Drosophila Caenorhabditis Arabidopsis
serevisiae pombe melanogaster elegans thalianab
LSC LSC LSC LSC LSC
Rank name/function Na name/function Na name/function Na name/function Na name/function Na
1 Uncharacterized 25 Wtffamilyof 20 Trypsin-likeserine 178 7TMOdorant 264, Plant-specific 316
Ecm34p-like non-globular proteases receptors 228 kinases
proteins proteins (twodistinct
clusters)
2 Hexose 15 MayorFacilitator 13 Insectcuticle 88 Uncharacterized 182 Plant-specific, 251
transporters Superfamily proteins proteins F-box
transporter containinga containing
nematode- proteins
specific
domain
3 Aminoacid 14 Ser/Thrrepeat- 11 Cytochrome 83 Integral 151 Extracellular 221
permeases containing P450family membrane domainsoften
non-globular hydroxylases O-acetyl/Acyl associatedwith
proteins transferases kinases
4 Cellwall 11 Alpha-amylases 11 C4DM+ 82 7TMreceptors 122 PPRmodule 194,
glycoproteins Zn-finger proteins 195
(twodistinct
dusters)
5 Cellwall 11 Gal4-likefungal 8 POZ-containing 55 C-typelectins 115 Apoptotic 150
mannoproteins C6finger transcription 55 (secreted (AP)-ATPases
factor proteins)
Odorant
receptors
aNumberofmembersintheLSC.
bAfamilyof171transposons,withmutatorelements,wasnotincluded.
1052 Genome Research
www.genome.org
Lineage-Specific Gene Expansion in Eukaryotes
Table2. FunctionsofSelectedLineage-SpecificProteinClustersinFiveEukaryotes
Speciesb
(no.of
Nameoftheclustera members) Biologicalfunctionsandothercomments
Transcriptionregulation
AP2-likeDNA-bindingproteins At(117) Plant-specifictranscriptionfactorswithmultiplerolesinstressandethylene
responseanddevelopment(Riechmannetal.2000).
MYB-likeDNA-bindingproteins At(100,48) HTH-domain-containingtranscriptionfactorswithdiverserolesin
developmentandregulationofvariousenvironmentalresponses
(Riechmannetal.2000).
WRKY-likeDNA-bindingproteins AT(68) DNA-bindingproteinsinvolvedinregulationofdevelopmentandpathogen
response.
RF-Afamilyofnucleic At(47) Anexpansioninvolvingtheconservedarchaeo-eukaryoticreplicationfactorA
acid-bindingproteins(OBfold) thatispresentinasinglecopyinothereukaryoticlineages(Wold1997).
Viv1/PVAL-liketranscription AT(41) Plant-specifictranscriptionfactorsinvolvedinabscisicacidresponse,seed
factors differentiation,anddevelopment(Riechmannetal.2000).
Nuclearhormonereceptors Ce(66,43,26, Zn-dependentDNA-bindingproteinstypifiedbyvertebratesteroidreceptors.
26,andother ManyoftheC.elegansmembersofthisfamilymayfunctionindependently
smallclusters) ofligands,andcharacterizedmemberslikeodr-7haverolesincell-type
differentiation(Sluderetal.1999).
C4DM+Zn-finger-containing Dm(82) TranscriptionfactorstypifiedbytheZeste-white5family.Consistofa
proteins DNA-bindingC2H2-fingerandC4DM,apredictedZn-dependent
protein–proteininteractiondomain(Landeretal.2001).
SAZ-typeMybdomain- Dm(40) AspecializedversionoftheMYBDNA-bindingdomaintypifiedby
containingproteins transcriptionfactors,suchasStonewall,Adf-1,andZeste.
POZ+Zn-finger Dm(55) AclassofDNA-binding,chromatin-associatedtranscriptionfactors,suchas
Broad-complex,Lola,andtrithorax-likeconsistofaspecificversionofthe
POZdomainfusedtoaC2H2-finger.
C6finger-containingproteins Sp(4) Gal4-likeC6Znfingersareamongthemostcommontranscriptionfactorsin
theascomycetefungi.
Pathogen/stressresponse
AP-ATPases AT(150,29,17) Plantdisease-resistancelociproducts,typicallyconsistofaTIRandan
AP-ATPasedomaincombinedwithleucine-richrepeats(LRRs)(Hulbertetal.
2001).
Pepsin-likeproteases At(51),Ce(16) Secretedproteasesthatcouldbeinvolvedinextracellularregulatory
proteolyticcascades.
Subtilisin-likeproteases At(57) Secretedproteasesthatcouldbeinvolvedinextracellularregulatory
proteolyticcascades.
Papain-likeproteases At(14) Thiolproteasesthatcouldbeinvolvedinstressresponsesandingermination.
MetalloproteasescontainingCUB Ce(23) Membrane-associatedmetalloproteasesthatcouldbeinvolvedinproteolytic
domains cascadesonthecellsurface.
C-typelectins Ce(115,42) Extracellularproteinscontainingadhesionmodulespotentiallyinvolvedin
Dm(28) recognitionofspecificpathogensurfacemolecules.
Chitinases Ce(33)Dm(17) Enzymespotentiallyinvolvedinhydrolysisofcellwallsoffungalpathogens.
Toll-likereceptors Dm(8) Keyreceptorsoftheanti-pathogenresponsepathways.
CUB-domainproteins Ce(40) Extracellularadhesionproteins.
P450hydroxylases At(124,34,33, Oxidoreductasesinvolvedindetoxificationofdiversexenobioticsthrough
28)Dm(83) hydroxylation(Nelson1999;Tijetetal.2001).
Ce(46,16)
PRI-domainproteins At(24)Ce(40) Secretedproteinsthatcouldfunctionasinhibitorsofenzymesoradhesion
molecules.
Cellwallmannoproteins Sc(11) Involvedincoldshockandanoxicstressresponse.
(cid:2)-helicalperoxidases At(73) Enzymesgeneratingnascentoxygenaspartoftheoxidativedefense
mechanisms.
Signaling
Concanavalin-likelectins At(43) Someoftheselectinsarefusedtokinasesasextracellularreceptordomains
andprobablyfunctionascarbohydratereceptors.
PPR-moduleproteins AT(194,195) (cid:2)-superhelicalproteinsthatcouldfunctionasprotein–proteininteraction
scaffoldsinvariouscontexts.
Calcium-dependentprotein AT(44) TheprincipaltransducersofCa++signalingthatmediatethispathwayin
kinases variouscontexts.
Plant-specificproteinkinases At(316) Involvedinvarioussignalingpathways,suchashormoneresponse,disease
resistance,anddevelopment.Oftenfusedtovariousotherdomains,
includingApple,LRRs,andbulblectins.
Octicosapeptidemoduleproteins At(72,17,14) ACa++-bindingsignalingmodule;somearefusedtoVTV1-likeDNA-binding
domainsandGAFdomains(Ponting1996).
NPH-3-like,plant-specific At(30) SpecializedPOZdomains,someofwhichareinvolvedinplantlightresponse
POZ-domainproteins signaling.
PP2Cphosphatases At(20) Phosphoserinephosphatasesthatfunctionindiversesignlaingpathways,e.g.,
abscisicacidsignaling.
(Tablecontinuedonfollowingpage.)
Genome Research 1053
www.genome.org
Lespinet et al.
Table2. (Continued)
Speciesb
(no.of
Nameoftheclustera members) Biologicalfunctionsandothercomments
Worm-specificS/Tkinases Ce(65) Adistinct,nematode-specificbranchofthecaseinkinasefamily.
Receptorguanylatecyclases Ce(13,12) Potentialreceptorsofsecretedpeptidefirstmessengersbyanalogytomating
fusedtoproteinkinases pheromonereceptorsofseaurchins.
Worm-specificdomains Ce(42) Uncharacterizeddomainprobablyinvolvedinspecificprotein–protein
interactions;somearefusedtoSET,caspase,kinase,andPHDdomains.
POZ-domainproteins Ce(26,29) OftenfusedtoMATHdomains,possiblyfunctionaschromatin-associated
adaptors.
Insulin-likepeptides Ce(11) Probablyfunctionasnematode-specificpeptidehormonesorgrowthfactors.
Sec14-domainproteins Dm(23) Probablyparticipateinregulationofproteintraffickingandvesicularcargo
loading.
SET-domainproteinswithan Dm(10) ProteinmethyltransferasescontainingadivergentSETdomainwitha
insertedmetal-chelating characteristicinsertofametal-chelatingmodule.Probableregulatorsof
module chromatindynamics.
Geko-domainproteins Dm(8,17) AlargefamilyofDrosophila-specificcysteine-richproteins,theonly
characterizedmember,Geko,isinvolvedinolfaction.TheLSCmightbe
functionallycoupledtothecorrespondinglyexpandedolfactoryreceptor
families.
Ubiquitinsignaling/proteinunfoldinganddegradation
F-boxproteins At(251,64,41, Specificity-definingE3subunitsofubiquitinligases;fusedtoseveralother
23)Ce(111, domainsthatmightactasscaffoldsfortheassemblyoftheubiquitinating
46,21) enzymecomplexes(KipreosandPagano2000).
RING-fingerproteins At(74,16,12) ThemajorityoftheRINGfingersintheLSCsareoftheRING-H2category;
probablyfunctionasspecificE3-ligases
U-boxproteins At(21,18) RING-fingerderivativesthatprobablymediatemultiubiquitinationofspecific
targets.
Ubiquitin-domainproteins At(11) Probablyutilizedsimilarlytoubiquitin,butcouldspecificallyconjugatewith
differentproteins.
Adenoviral-typeproteases At(117) ProbablyinvolvedindeubiquitinationasexemplifiedbyULP1/SMT4(Liand
Hochstrasser2000;Nishidaetal.2000).
GH3-domainproteins At(17) ShareaconserveddomainwiththeE1subunitsofubiquitinligases;mightbe
negativeregulatorsofthesignalosome.
MATH-domainproteins Ce(81)At(73) RelatedtotheMATHdomainsoftheubiquitincarboxy-terminalhydrolases
andE3-ligasesoftheTRAFfamily;couldfunctionasadaptorsinubiquitin
pathways.
Prolylhydroxylases Dm(19)At(10) Hydroxylationofprolinesbytheseenzymesmightprovidetargetsfor
ubiquitinationbyspecificE3-ligases(AravindandKoonin2001).
Cyclophilin-typepeptidyl-prolyl Dm(10) Catalyzeisomerizationofproline-containingpeptidebonds;mightfunctionin
isomerases regulatingaggregationofproteincomplexes.
Chemoreceptorsandsmallmoleculesensors
7-transmembraneolfactory Ce(264,228, Receptorsforodorants/environmentalchemicals(Dryer2000;Glusmanetal.
receptors 122) 2001).
Insect-typeodorantreceptors Dm(55) Receptorsforodorants/environmentalchemicals.
Pheromone-bindingproteins Dm(27) Probablyinvolvedinthebindinganddeliveryofodorantstochemoreceptory
cells.
Patched-typesterolbinding Ce(15) Bindlipidsandsterolsinvariouscontextsincludingstabilizationofreceptor
membraneproteins complexes.
Juvenilehormoneandother Dm(27) Probablyinvolvedinthebindinganddeliveryofsmallmoleculesintheinsect
small-molecule-binding haemolymph.
proteins
Lipid-bindproteins(NLTP) At(49,26) Cysteine-rich(cid:2)-helicalproteinsinvolvedinlipidbindinganddeliveryin
variouscontextsandwaxdeposition.
Jacalin-typelectins At(44) Mightbeinvolvedinsugarbindingandstorage.
Hemocyanins Dm(10) Copper-dependentoxygentransportproteins.
Cyaninfamilyproteins At(34) Copper-bindingproteins.
IonChannelsandTransporters
Degenerinfamilychannels Dm(24) Sodiumchannels,probablyfunctionintactilereceptionandrelated
ion-dependentsignalingpathways.
Potassiummchannels Ce(15) Potassiumchannelsofthedoubleporecategory,probablyfunctionas
pH-dependentchannels.
Innexin-typechannels Ce(20) ChannelsrelatedtotheDmShaking-Bprotein,mightbeinvolvedinthe
formationofgapjunctions.
cNMP-gatedchannels At(21) Cyclicnucleotide-gatedchannelscontaininganintracellularcNMP-binding
domain.
Aminoacidtransporters At(33) AminoacidtransportersoftheN-aminoacidtransporterfamily.
(Tablecontinuedonfollowingpage.)
1054 Genome Research
www.genome.org
Lineage-Specific Gene Expansion in Eukaryotes
Table2. (Continued)
Speciesb
(no.of
Nameoftheclustera members) Biologicalfunctionsandothercomments
Potassiumtransporters At(17) Belongtotheplanttinyroothairfamily;probablyinvolvedinpotassium
uptake.
Na-P-transporter-relatedproteins Ce(26) Probablyinvolvedinphosphateuptakebysymport.
Hexosetransporters Sc(15) Belongtothe12TMsugartransportersuperfamily.
ABCtransporters Dm(11,9,5) TransporterscontainingtwoABC-classATPasedomains.
Smallmoleculemetabolism
Lipases At(106) Afamilyofphospholipidlipasesoftheflavodoxinfold;involvedin
degradationofphosphatidylcholine.Couldbeinvolvedinmetabolizing
lipidsingerminationordegradinglipidmembranesofpathogens.
2-OG-Fedioxygenases At(67) Hydroxylasesinvolvedinthebiosynthesisofnumerousplantsecondary
metabolites,suchasgibberellins(AravindandKoonin2001).
NH2cinnamoyl/benzoyl- At(56) Transfersaromaticcarboxylicacidgroupstodiversetargetsinthebiosynthesis
transferase ofplantsecondarymetabolites.
SmallmoleculeO-methylases At(38,15) Catalyzethemethylationstepinthebiosynthesisofdiverseplantproducts,
suchascaffeicacid.
GlutathioneS-transferases At(14)Ce(28) Catalyzetheconjugationofelectrophilicsubstrates,particularxenobiotic,to
Dm(27) glutathioneaspartoftheirtransportanddetoxification;additionallyhave
peroxidaseandsmallmoleculeisomeraseactivities.
Predictedsecretedsmall Ce(32) Containspecificdisulfidebonds;probablycatalyzemethylationofextracellular
moleculemethylases smallmolecules.
Integralmembrane Ce(151) Afamilyofmembrane-associatedacyltransferasescloselyrelatedtothe
O-acyltransferases bacterialmembraneassociatedacyltransferasesthatacylatemacrolide
antibioticsandcellsurfacepolysaccharides.
Predictedsmallmoleculekinases Ce(23)Dm(45) Relatedtoaminoglycosideandlipidkinases;probablyinvolvedin
phosphorylationofsmallmolecules,suchasodorantsand/orxenobiotics.
Structural/morphologicalproteins
Cystine-richexpansions At(35) Plantcell-wallglycoproteins.
Pectinmethylesterases At(89) Involvedinthebiosynthesisofpectins,majorstructuralcomponentsofplants.
Pectin-associatedproteins At(26) Four-cysteine(cid:2)-helicaldomains,somefusedtopectinesterases.
Cuticularcollagens Ce(34,32,26, Theprincipalstructuralcomponentofthenematodecuticle(Johnstone2000).
11)
Majorspermproteinfamily Ce(32,10) Theprincipalstructuralcomponentofnematodesperms.
Insectcuticularproteins Dm(88) Theprincipalstructuralcomponentoftheinsectcuticle(Andersenetal.
1995).
Peritrophin-likeproteins Dm(40) Insect-specificextracellularmatrixproteins.
Cellwallglycoproteins Sc(11) Proteincomponentoftheyeastcellwall.
Ecm34p-likeproteins Sc(25) Proteincomponentoftheyeastcellwall.
aThemembersofeachLSCarelistedintheSupplementaryMaterialsection,inwhichtheLSCscanbeidentifiedbytheirnamesandthenumber
ofmembers.
bSpeciesabbreviations:(At)Arabidopsisthaliana;(Ce)Caenorhabditiselegans;(Dm)Drosophilamelanogaster;(Sc)Saccharomycescerevisiae;(Sp)
Schizosaccharomycespombe.ThenumberofmembersineachLSCisindicatedinparentheses;commasseparatedistinctLSCsthatbelongto
thesameclassofparalogousproteins.
proteins.Extendingthisanalogy,itispossiblethatseveralof components include the massive proliferation of apoptotic
the LSCs with no detectable homologs elsewhere could rep- (AP-)ATPasesandtheaccompanyingmoderateexpansionof
resentasyetuncharacterized,butabundant,lineage-specific metacaspasesinplants,andtheparallelexpansionofcaspases
structuralproteins(Table2). in vertebrates (Aravind et al. 2001; Holub 2001). These pro-
ManyoftheidentifiedLSCshadpredictedbiochemical teinsareeitherknownorpredictedtoparticipateinmultiple
characteristics that pointed to roles in stress and pathogen pathways associated with apoptosis or hypersensitive re-
response.Particularlystrikinginthiscategorywastheexpan- sponse.Inthiscontext,alsoofinterestaretheexpansionsof
sionofproteasesofthepepsin-likeandsubtilisin-likefamilies molecules containing modules functioning in extracellular
in A. thaliana, trypsin-like proteases in D. melanogaster, and adhesion. Prominent examples of these include the C-type
Zn-metalloproteasesinC.elegans(Table2).Allofthesepro- lectins(D.melanogaster,C.elegans),PR1proteins(C.elegans,
teases are predicted to be secreted molecules, and their re- A.thaliana),CUBdomainproteins(C.elegans),andthebulb-
peated,independentexpansionsuggeststhattheyarewidely lectindomain(A.thaliana).Aswiththeimmunoglobulindo-
utilizedeitherfordirectdegradationofpathogenproteinsor mainprotein,thatarehighlyexpandedinvertebrates,these
ascomponentsofstress-triggeredproteolyticcascadesbroadly moleculesprobablyparticipateintherecognitionandbind-
analogoustothevertebratecomplementandclottingsystems ingofspecificpathogensasapartofdefensemechanismsof
(Bouchard and Tracy 2001; Southan 2001). Alternatively, in thecorrespondingorganisms(Table2).
thecaseofplants,theycouldaidinproteindigestioninthe Earlier analysis of the LSEs involving transcription fac-
processofgermination.Better-understoodcasesofsimilarlin- tors had suggested that they included proteins regulating
eage-specific expansions related to stress/pathogen-response criticalaspectsofthedevelopmentoftheorganism(Aravind
Genome Research 1055
www.genome.org
Lespinet et al.
andKoonin1999;Riechmannetal.2000;Landeretal.2001). radationhasbeenshowntooccurthroughtherecognitionof
Forexample,theproteinsbelongingtothePOZandSAZ-type hydroxyproline by ubiquitin ligase complexes (Ivan et al.
MybdomainexpansionsinD.melanogaster(Table2)regulate 2001).Thus,theLSEof2-oxoglutarate-dependentprolylhy-
as diverse functions as maintenance of the antero-posterior droxylases(AravindandKoonin2001)detectedinD.melano-
Hox gene expression pattern, neurite outgrowth and path- gasterandA.thalianacouldrepresentanothercaseinwhich
finding, and organogenesis (Aravind and Koonin, 1999; therangeofthecoreubiquitinationpathwayisexpandedvia
Landeretal.2001).Thus,itappearsthatproliferationofnew diversificationoftheterminaleffectors.
transcriptionfactorfamilies,followedbytheirrecruitmentas The role of LSE in the diversification of proximal com-
upstreamordownstreamregulatorswithrespecttocorecon- ponents of signal transduction systems, receptors, had been
served developmental pathways, have contributed substan- noticedpreviouslyinthecasesofindependentexpansionsof
tiallytotheevolutionofmorphologicaldiversityinanimals. odorant receptors/7-transmembrane chemoreceptors seen in
Thegeneralityofthisobservationwasreinforcedbytheevi- different animal lineages (Dryer 2000; Glusman et al. 2001)
denceofmassive,lineage-specificexpansionanddiversifica- and plant receptor kinases containing extracellular leucine-
tion of various transcription-factor families in the plant A. rich repeats, bulb lectin, or EGF-like extracellular domains
thaliana (Table 2). Many of these include well-characterized (ShiuandBleecker2001).Here,wedetectedotheranalogous
DNA-bindingproteins,suchastheMADSboxandMYBdo- expansionsofupstreamsignalingproteins,suchaspotassium
main proteins, that have been shown previously to partici- channels, innexin family channels (both in C. elegans), and
pate in plant-specific functions, including development of tetraspaninsanddegenerin-typechannelsinD.melanogaster
flowersandotherstructures,meristemaldifferentiation,and (similarLSEsofK-channelsandtetraspaninsarealsoseenin
organ-specific gene expression (Riechmann et al. 2000). In humans). The proteins involved in these expansions are
this study, we detected certain unexpected expansions of linked to the organism’s responses to external as well as in-
DNA-binding proteins in plants that might point to previ- ternalhomeostaticstimuli.Thus,suchexpansionscouldserve
ously unrecognized transcription regulators. Examples in- as the raw material for the behavioral and physiological ad-
clude the proteins homologous to the mitochondrial tran- aptation of organisms to their specific environments. Lin-
scription termination factor, which, in other eukaryotes, is eage-specific expansions are also seen in a range of protein-
presentinasinglecopythatfunctionsinthemitochondrion modifyingenzymesofdifferentsignaltransductioncascade,
(Fernandez-Silva et al. 1997). The additional paralogs in suchasproteinkinasefamiliesinmostlineages,SET-domain
plantshaveprobablyacquireddifferenttranscription-related protein-methylases in D. melanogaster, and PP2C phospha-
functionsbecausetheyformatightcluster,distinctfromthe tasesinplants.Aswiththeubiquitinsystem,theseappearto
ancestralmitochondrialversion.Plantsalsoshowanexpan- beameansoflinkingwell-conservedstemsofsignalingpath-
sionoftheDNA-bindingreplicationfactorA(RF-A),with>40 waystodistinctsetsofterminaltargets.
copiesinA.thaliana,incontrasttotheone-threecopiesob- AnotheraspectoftheinvolvementofLSEsintheevolu-
servedinothereukaryotes.Theexpansionanddivergenceof tion of signal-transduction networks is the extensive prolif-
RF-A in plants suggest that the plant-specific paralogs are eration of families of proteins containing adaptor domains.
probablyutilizedastranscriptionfactorsratherthanintheir Alongwiththeirexpansion,manyadaptordomainshavealso
usual capacity in replication (Wold 1997). These and other recombinedwithavarietyofotherdomains,probablyallow-
such examples (Table 2) illustrate that transcription factors ingtheemergenceofnewnetworksofinteractions.Astriking
arerecruitedfromawidevarietyofpre-existingsourcesand example is the major expansion of proteins containing the
diversifytooccupynewfunctionalnichesviaLSE. small Ca-binding octicosapeptide (OOP) module (Ponting
WeobservedamajorroleofLSEintheelaborationofthe 1996)inA.thaliana.SomeOOPmodulesarefusedtoVIV1-
ubiquitinpathway,whichisinvolvedinthedegradationand like plant-specific DNA-binding proteins and a specialized
regulatory modifications of proteins (Hershko and Ciecha- classofGAFdomains,suggestingthattheylinktranscription
nover 1998). Evidence of LSE was obtained for several com- regulationandsmallmoleculeinteractionstoCa-dependent
ponentsoftheubiquitinsystem,inparticular,E3subunitsof signaling. Another notable case is a novel adaptor domain,
ubiquitin ligases containing the F-box domain (Kipreos and typifiedbytheamino-terminaldomainoftheCaspase-1Aiso-
Pagano2000)(A.thalianaandC.elegans)andtheRING-finger form,whichsofarwasdetectedonlyinC.elegans.Altogether,
(A.thaliana).BecausetheE3proteinsarespecificitydetermi- theC.elegansgenomeencodes>40membersofthisdomain
nantsthatareinvolvedintargetingtheconservedubiquitin- family, which, in addition to the caspase fusion, also form
ligationmachinerysystemtospecificsubstrates(Jacksonetal. multidomainproteinswithSET-domainmethylases,PHDfin-
2000), their diversification through LSE probably provides a gers,andkinases.Giventhe(cid:2)-helicalstructurepredictedfor
meansofharnessinganotherwiseconservedsystemtoregu- thisdomain,andenrichmentinchargedresidues,itprobably
latethedegradationofdiversesetsoftargets.Inasimilarvein, functionsasaprotein–proteininteractionmodule.
both nematodes and plants also show independent LSEs of Another, somewhat unexpected generalization that
theMATHdomain.Thisdomain,whichtendstoformfusions emergedfromthepresentanalysisistheprevalenceofsmall
toubiquitincarboxy-terminalhydrolasesorRING-fingerE3s molecule-modifyingenzymesamongtheLSEs.Inplants,the
(Aravindetal.1999;Polekhinaetal.2002),mightserveasan proliferationofsuchenzymes,namelymethylasesofthecaf-
additionaladaptorthatmediatesde/ubiquitinationofspecific feicacidO-methylasefamily,dioxygenasesofthegibberellin-
targets. A. thaliana has a prominent proliferation of the ad- hydroxylasefamily,andavarietyoflipasesandacyltransfer-
enovirus-likethiolproteasesuperfamilywhosemembers(e.g., ases, correlates with the plethora of secondary metabolites,
Smt4/Ulp1)inyeastandinvertebrates,removeubiquitin-like suchaspigments,volatilearomaticcompoundsalkaloids,and
proteinsfromtheirtargets(LiandHochstrasser2000;Nishida waxes that are produced by plants (Seigler 1998). However,
etal.2000).Thus,inplants,thisLSCprobablycontributesto their large numbers suggest that the entire diversity of me-
further diversification of the regulation of ubiquitin- tabolites produced even by plants such as A. thaliana with
dependentproteindegradation.Targetingofproteinsfordeg- relativelysimplegenomesisunder-appreciatedtoalargeex-
1056 Genome Research
www.genome.org
Lineage-Specific Gene Expansion in Eukaryotes
tent.Interestingly,animalsalsohaveseveralLSEsassociated METHODS
withsmallmoleculemetabolism.Someofthese,suchasgly-
The protein set for the nematode C. elegans was from the
cosyltransferasesandacyltransferases,suggesttheremightbe WormPep20 data set (http://www.sanger.ac.uk/Projects/
an as yet unexplored, lineage-specific diversity of carbohy- C_elegans/wormpep);theproteinsetsforotheranalyzedeu-
drates and lipid moieties that are associated with glycopro- karyotic species were extracted from the NCBI (NIH) nonre-
teins, lipoproteins, and other cellular metabolites. The two dundant(nr)proteinsequencedatabase.Thehumanprotein
independentexpansionsofpredictedsmall-moleculekinases setwasnotsystematicallyanalyzedbecauseofextensiveprob-
relatedtoethanolamineandaminoglycosidekinases(Honet lemswithgenepredictions,resultinginfragmentaryproteins,
al. 1997) (in D. melanogaster and, to a lesser extent, in C. artificial fusions, and inclusion of pseudogene translations
andtranslationofnoncodingDNA.
elegans)andtheexpansionofsecretedmethylasesinC.elegans
Identical or nearly identical (98% or greater) sequences
areparticularlyenigmatic.Giventheroleoftherelatedbac-
wereremovedfromthedatasetsusingtheBLASTCLUSTpro-
terialkinasesandmethylasesinxenobioticresistance(Hagg- gram. For documentation on its use, see ftp://ftp.ncbi.
blom1990),theseenzymesmightbeusedtomodifyarange nlm.nih.gov/blast/documents/README.bcl. LSCs were iden-
of xenobiotics encountered by the animals in their specific tifiedusingthefollowingprocedure:BLASTcomparisonsfor
environments.Alternatively,theycouldmodifyvariousenvi- all proteins in the analyzed set of complete eukaryotic ge-
ronmental substances to convert them to forms more easily nomeswererunagainstthedatabaseconsistingofthesame
sensedbythechemoreceptorsoftheseorganisms. set of proteins. Symmetrical relative similarity scores
(R =R =max(S /S ,S /S ),inwhichS istheBLAST
AB BA AB AA BA BB AB
bitscoreforqueryAandsubjectBwererecorded.Suchscores
range from 0 (no significant hit found) to 1 (identical pro-
teins). For each protein A in a given genome X (e.g., C. el-
Conclusions egans),asetofcandidatefamilycomembers{B}wasdefinedas
asetofproteinsfromthesamegenomeXsatisfyingthecon-
A computational procedure for systematic detection of lin- dition(R >R ;for(cid:1)C⊄ X)(i.e.,similaritybetweenthegiven
AB AC
eage-specific expansions of protein families was developed proteinAandanotherC.elegansproteinBisgreaterthanthat
andappliedtoobtainacomprehensivecensusofLSEsinfive betweenAandanyproteinCfromanyothergenome).Then,
eukaryoticgenomes.LSEsappeartohaveplayedanimportant allsuchsetsfromXweremergediftheysharedatleastone
role in the growth and differentiation of the proteomes of member (single-linkage clustering), resulting in grouping all
multicellular eukaryotes. Many paralogous gene families in proteinsfromXintoclusters{A}(manyofwhichmightcon-
tain only a single protein). This procedure leads to heavy
crown-group eukaryotes appear to have evolved almost en-
overclustering because, even if only one pair of proteins in
tirelythroughLSEafterthedivergenceoftheexaminedsister
two distinct LSCs passes the comembership condition (e.g.,
lineages from their ancestors. This fundamental process of
duetofluctuationsintheobservablesimilarity),thetwoLSCs
genefamilyexpansionwasactiveatawiderangeofphyloge- are merged by the single-linkage algorithm. This over-
netic distances, from the relatively close species of yeasts to inclusivesetofclusterswasrefinedthroughidentificationof
the much earlier separation of plants from the rest of the the most closely related proteins from other genomes. For
crown-group taxa. Generally, the fraction of proteins found each A⊄ {A}, the best alien hit C was identified as [C | max-
in LSCs and the fraction of large families among LSCs posi- (RAC); C⊄ X]. Sets {A}∪ {C} (i.e., candidate LSC members and
tivelycorrelatewiththesizeofeukaryoticproteomes. theirclosestalienrelatives)weresubjecttoUPGMAclustering
onthebasisofrelativesimilarityscores.Underthisprocedure,
Examination of the known and predicted functions of
proteins from other genomes that show high similarity to
thedetectedLSEsrevealscertaingeneralprinciples.Genesen-
some candidate LSC members may intrude into the clus-
codingproteinstypicallyrequiredinlargequantitiesascom- ter and split it apart. Subclusters {A(cid:3)} satisfying [A(cid:3)⊂ X] (i.e.,
ponentsofanorganism’smorphologicalstructuresareoften UPGMAsubtreesconsistingofproteinsexclusivelyfromthe
subjecttoLSEandappeartobefixedversionsofthecommon currentlyanalyzedgenomeX)andincludingmorethanone
phenomenonofgeneamplification,withfine-tuningadded proteinwereconsideredtorepresentLSCs.
through sequence diversification (Kondrashov et al. 2002). Proteinsequencesimilaritysearcheswereperformedus-
Another major set of LSCs consists of proteins involved in ing the gapped BLASTP program against the nonredundant
recognition and binding of pathogens and xenobiotics and protein sequence database (NCBI, NIH). Iterative profile
searchestodetectmoredistantrelationshipswereperformed
withstandingenvironmentalstress.TheseLSCsprobablypro-
usingthePSI-BLASTprogram(Altschuletal.1997),withthe
videtherawmaterialforgeneratingthediversityrequiredto
inclusion threshold typically set at E=0.01; only predicted
counterrapidlychangingpathogensandtorespondtoother
globular regions from proteins were used as seeds for
variableenvironmentalfactors.Expansionfollowedbydiver- PSI-BLASTsearches.Proteinswerepartitionedintoprobable
sificationoftheproteinsintheLSCsappearstobeacommon globular and nonglobular regions using the SEG program
meansofgeneratingnewspecificitiesinsignalingpathways. (Wootton1994).Conserveddomainsweredetectedusingdo-
Inparticular,intheubiquitinsystem,alargenumberofthe main-specific PSSMs constructed using the PSI-BLAST pro-
E3componentsoftheubiquitinligase,whichtargetittospe- gram (Chervitz et al. 1998). Multiple alignments were con-
cific proteins, are drawn from LSEs. Expansions of adaptor structed using the T_Coffee (Notredame et al. 2000) and
ClustalX (Thompson et al. 1997) programs and corrected
modulesfollowedbytheirfusiontodiversedomainsprobably
manuallyonthebasisofPSI-BLASTsearchresults,which,on
resultintheemergenceofnovelinteractionsthatcontribute
some occasions, correctly detect conserved sequence motifs
to signaling and transcription regulation. Several expanded
missed by multiple alignment methods. These alignments
enzymefamiliesalsopointtotheexistenceofan,asyet,un- were used to construct Neighbor Joining phylogenetic trees
discovereddiversityofsmallmoleculemetabolitesinvarious (Saitou and Nei 1987) using the PAUP* (Swofford 1998) and
lineages. Thus, LSE seems to be one of the most important PHYLIP (Felsenstein 1996) package (the evolutionary dis-
sourcesofstructuralandregulatorydiversityincrown-group tances were calculated using the PROTDIST program of
eukaryotes,whichwascriticalforthetremendousexploration PHYLIP),andthesupportfornodesofinterestwasevaluated
ofthemorphospaceseenintheseorganisms. by use of 1000 bootstrap replicates. Secondary structure of
Genome Research 1057
www.genome.org