Comparative Genomics of the Archaea (Euryarchaeota): Evolution of

22 Pages·1999·5.75 MB·English

Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Research
Comparative Genomics of the Archaea
(Euryarchaeota): Evolution of Conserved Protein
Families, the Stable Core, and the Variable Shell
Kira S. Makarova,1,2,4 L. Aravind,1,3 Michael Y. Galperin,1 Nick V. Grishin,1
Roman L. Tatusov,1 Yuri I. Wolf,1,4 and Eugene V. Koonin1,5
1NationalCenterforBiotechnologyInformation,NationalLibraryofMedicine,NationalInstitutesofHealth,
Bethesda,Maryland20894USA;2DepartmentofPathology,F.E.HebertSchoolofMedicine,UniformedServicesUniversityof
theHealthSciences,Bethesda,Maryland20814-4799USA;3DepartmentofBiology,TexasA&MUniversity,CollegeStation,
Texas70843USA
Comparative analysis of the protein sequences encoded in the four euryarchaeal species whose genomes have
been sequenced completely (Methanococcus jannaschii, Methanobacterium thermoautotrophicum, Archaeoglobus fulgidus, and
Pyrococcus horikoshii) revealed 1326 orthologous sets, of which 543 are represented in all four species. The proteins
that belong to these conserved euryarchaeal families comprise 31%–35% of the gene complement and may be
considered the evolutionarily stable core of the archaeal genomes. The core gene set includes the great majority
of genes coding for proteins involved in genome replication and expression, but only a relatively small subset of
metabolic functions. For many gene families that are conserved in all euryarchaea, previously undetected
orthologs in bacteria and eukaryotes were identified. A number of euryarchaeal synapomorphies (unique
shared characters) were identified; these are protein families that possess sequence signatures or domain
architectures that are conserved in all euryarchaea but are not found in bacteria or eukaryotes. In addition,
euryarchaea-specific expansions of several protein and domain families were detected. In terms of their apparent
phylogenetic affinities, the archaeal protein families split into bacterial and eukaryotic families. The majority of
the proteins that have only eukaryotic orthologs or show the greatest similarity to their eukaryotic counterparts
belong to the core set. The families of euryarchaeal genes that are conserved in only two or three species
constitute a relatively mobile component of the genomes whose evolution should have involved multiple events
of lineage-specific gene loss and horizontal gene transfer. Frequently these proteins have detectable orthologs
only in bacteria or show the greatest similarity to the bacterial homologs, which might suggest a significant role
ofhorizontalgenetransferfrombacteriaintheevolutionoftheeuryarchaeota.
Phylogenetic analysis of rRNA and a set of proteins 1996; Brown and Doolittle 1997; Edgell and Doolittle
involved in translation, transcription, and replication 1997). However, it has been aptly noted that archaea
hasledtotheconceptofarchaeaasathirddivisionof have a “eubacterial form and eukaryotic content”
life,distinctfromeitherbacteriaoreukaryotes(Woese (Keeling et al. 1994). Indeed, beyond the common
et al. 1978, 1990; Woese and Gupta 1981; Pace et al. “negative”trait,namelythesmallcellsizeandtheab-
1986;Zillig1991).Furthermore,rootingofparalogous sence of a nucleus, archaea and bacteria share major
trees for translation elongation factors and proton aspects of genome organization and expression strat-
ATPases suggested that archaea are a sister group of egy. The most important of these common features
eukaryotes(Gogartenetal.1989a,b;Iwabeetal.1989; includethe(typically)singlecircularchromosome,the
GribaldoandCammarano1998).Thisconceptappears absence of introns in protein-coding genes, the oper-
to be gaining further support from the generally eu- onicorganizationofmanygenes,andtheabsenceofa
karyoticlayoutofthegenomeexpressionsystems,par- 58-terminal cap and the presence of a ribosomal-
ticularly the system of DNA replication whose princi- binding (Shine-Dalgarno) site in archaeal mRNAs
palcomponentsareorthologoustotherespectiverep- (Brown and Doolittle 1997). Furthermore, several op-
lication proteins of eukaryotes but apparently do not erons,particularlythoseencodingribosomalproteins,
havecounterpartsinbacteria(MushegianandKoonin are conserved in archaea and bacteria (Brown and
Doolittle1997;KooninandGalperin1997).
4Present address: Institute of Cytology and Genetics, Russian Academy of Theanalysisofthefirsttwocompletelysequenced
Sciences,Novosibirsk630090,Russia. archaeal genomes, those of Methanococcus jannaschii
5Correspondingauthor.
[email protected];FAX[phone]. (Bult et al. 1996) and Methanobacterium thermoautotro-
608 GenomeResearch 9:608–628©1999byColdSpringHarborLaboratoryPressISSN1054-9803/99$5.00;www.genome.org
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Archaeal Genomics and Evolution
phicum (Smith et al. 1997), showed, somewhat unex- thologs that precludes their detection altogether. As a
pectedly given the already established archaeal– result,thefinalstepintheconstructionoftheoriginal
eukaryotic clade, that the bacterial form of archaea is collectionofCOGsinvolvedconsiderablemanualcor-
complemented by considerable bacterial content. It rection.Thedistancesseparatingthefourarchaealspe-
hasbecomeclearthatthemajorityofarchaealproteins cies are intermediate between those that are seen
show the greatest similarity to their bacterial ho- among close bacterial species such as Escherichia coli
mologs,whichislikelytoindicatebacterialorigin,and andHaemophilusinfluenzae(intheoriginalCOGanaly-
onlyaminoritylook“eukaryotic”(Kooninetal.1997; sis, these species were not considered independently)
Smithetal.1997).Infunctionalterms,thereisaclear and those between phylogenetically remote species
splitbetweenthebacterialandeukaryoticcomponents suchasbacteriaandeukaryotes.Inquantitativeterms,
ofthearchaealgenomes—theeukaryoticgenesarepri- the mean percent identity of the best hits in all-
marilythosecodingforcomponentsofthetranslation, against-all interspecies comparisons of protein se-
transcription, and replication machineries, whereas quences is in the range of 41%–46% for the archaea,
thebacterialonestypicallyencodemetabolicenzymes 57%forE.coliversusH.influenzae,andbetween30%–
andproteinsinvolvedincelldivisionandcellwallbio- 35% for most distant bacterial lineages and bacteria
genesis (Koonin et al. 1997; Smith et al. 1997). These versus eukaryotes or archaea (N.V. Grishin, unpubl.;
findingsraisedtheissueofpossibleextensivegeneex- ftp://ncbi.nlm.nih.gov/pub/koonin/gen2gen). It ap-
changebetweenbacteriaandarchaea(Fengetal.1997; pears that the intermediate level of sequence conser-
Kooninetal.1997;DoolittleandLogsdon1998). vationseenamongthearchaeaishighenoughtopre-
Subsequently, the complete genome sequences of ventmost,ifnotall,artificiallumpingofCOGsattrib-
two additional archaeal species, namely Archaeoglobus utable to paralogous families, but low enough for the
fulgidus (Klenk et al. 1997) and Pyrococcus horikoshii consistency criterion to be valid and useful. For these
(Kawarabayasi et al. 1998a,b), have been reported. All reasons, most of the archaeal COGs delineated by the
four available complete archaeal genomes represent automaticprocedurewerecorroboratedbysubsequent
only one of the two (or possibly three) main archaeal case-by-case evaluation. Furthermore, given the typi-
subdivisions—the Euryarchaeota (Olsen et al. 1994; callyhighlysignificantsimilaritybetweenarchaealor-
Pace1997).Nevertheless,theyshowsufficientdiversity thologs,itismostunlikelythatanysignificantnumber
toallowus,forthefirsttime,toembarkonasystematic of them have been missed as a result of low sequence
comparativeanalysisofarchaealgenomes.Wedescribe conservation.
here the results of a detailed comparative analysis of Figure1showsthebreakdownofthearchaealpro-
thefourcompleteeuryarchaealproteinsets.Ourprin- teinsetintermsoftheirconservationinthefourcom-
cipal approach included the delineation of sets of or- plete genomes. The majority of the proteins in each
thologousgenesandexaminationofphylogeneticpat- species—from 58% for P. horikoshii to 71% for M. jan-
terns in these families (Tatusov et al. 1997; Koonin et naschii—belong to the archaeal families of likely or-
al.1998). thologs(COGs),andanothersizablefraction(from7%
RESULTS AND DISCUSSION
Orthologous Families Delineated by Comparison
of Four Euryarchaeal Genomes and the Principal
Types of Events in ArchaealEvolution
The proteins encoded in the genomes of the four eu-
ryarchaealspeciescompriseaverygoodsetforthede-
lineation of families of likely orthologs [designated
clustersoforthologousgroups(ofproteins),COGs;Ta-
tusov et al. 1997)]. In the original COG analysis, we
emphasized that to use consistency between different
genomes to support the derivation of COGs, the se-
quences of the compared proteins should be maxi-
mallyindependent;therefore,thiscriterionworksbest
Figure 1 Conserved families and unique proteins encoded in
with phylogenetically distant genomes. At large phy- thefourcompletearchaealgenomes.(COGs+)Distanthomologs
logenetic distances, however, correct identification of of COGs; (NARCHOM) nonarchaeal homologs (only); (unique)
proteinswithoutdetectablehomologsinotherspecies(fordetails
COGs may be hampered by other problems, such as
see text); (Af) Archaeoglobus fulgidus; (Ph) Pyrococcus horikoshii;
difficulty in distinguishing orthologs from paralogs,
(Mt)Methanobacteriumthermoautrophicum;(Mj)Methanococcus
and in some cases, very low similarity between or- jannaschii.
Genome Research 609
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Makarova et al.
most likely that the ancestral form also had been au-
totrophic; thus, the absence of the representatives of
many COGs in P. horikoshii is best explained by lin-
eage-specificgeneelimination.Atleastsomeofthear-
chaeal COGs with two members are also likely to re-
flectgeneloss.Thus,ahigherestimateforthenumber
ofancestralgeneslostineachgenomecanbeobtained
byaddingupallCOGswiththreeortwomembersthat
donotincludethegivenspecies.Theresultvariesfrom
atotalof220genesforM.jannaschiito451genesforP.
horikoshii.
Thus,theanalysisoftheconservedarchaealfami-
lies reveals major genome plasticity, with only a mi-
nority of families represented in all genomes. These
Figure 2 Representation of the four archaeal species in the observationsmakeallthemorepertinentthequestion:
COGs.(F)Archaeoglobusfulgidus;(T)Methanobacteriumthermo-
which essential cellular functions are provided by the
autrophicum;(J)Methanococcusjannaschii;(H)Pyrococcushoriko-
shii. setof543universalarchaealCOGsandwhicharenot
represented by it, and, accordingly, are performed by
nonorthologous (unrelated or paralogous) proteins in
different species—the phenomenon described as non-
forM.jannaschiito11%forA.fulgidus)wereidentified orthologous gene displacement (Koonin et al. 1996a;
asdistanthomologsoftheCOGs.Amongtheremain- MushegianandKoonin1996).
ingproteinsthathadnoarchaealhomologs,forarela-
tivelysmallfraction(from1%inM.jannaschiito4%in
The Core Set of Conserved Euryarchaeal Genes,
A.fulgidus),homologsweredetectedinothertaxa(pri-
Lineage-Specific Gene Loss, and Nonorthologous
marilybacteria),andtherest(~ 20%)hadnodetectable
GeneDisplacement
homologs. This distribution suggests that a conserved
TheCOGsrepresentedinallfoureuryarchaealspecies
archaeal gene set does exist. This core gene set, how-
aresignificantlyenrichedinproteinsthatareinvolved
ever,includesaminorityofthearchaealgenesasindi-
in genome expression, compared to the entire collec-
cated by the fact that only 543 of the 1326 identified
COGs (40%) are represented in all four archaeal spe-
cies; the remaining COGs are roughly equally divided
betweenthosethatincludethreeandtwospecies(Fig.
2).TheuniversalarchaealCOGsencompass31%–35%
of the proteins encoded in each of the individual ge-
nomes.Thisnumberappearstobeanimportantmea-
sureoftheevolutionarystabilityofthegenomes—the
rest of the gene complement in each of the archaea
must have been subject to evolutionary events other
thanverticalinheritance,suchasduplicationwithsub-
sequentrapiddivergence,horizontalgenetransfer,and
lineage-specificgeneloss.
These results provide at least a rough estimate of
thelikelyamountofgenelossineachspecies,aswell
as the number of COGs represented in the ancestral
euryarchaeon. A conservative estimate of the number Figure3 Distributionofpredictedproteinfunctionsintheuni-
versalandnonuniversalsubsetsofthearchaealCOGs.(Bluebars)
of genes that might have been lost in each genome is
Theuniversalsubset(543COGswithfourmemberseach);(red
provided by the number of COGs that include three bars)thenonuniversalsubset(783COGswithtwoorthreemem-
archaealspeciesotherthanthegivenone.Thisnumber bers each). (Vertical axis) Number of COGs; (horizontal axis)
isintherangeof50to70forM.jannaschii,M.thermo- functionalcategories:1,translation,ribosomestructure,andbio-
genesis;2,transcription;3,DNAreplication,repair,recombina-
autotrophicum,andA.fulgidus,asopposedto206inP.
tion; 4, energy production and methanogenesis; 5, amino acid
horikoshii (Fig. 2). The greatest number of COGs that metabolism;6,nucleotidemetabolism;7,carbohydratemetabo-
arenotrepresentedinP.horikoshiiisnotsurprisingas lism;8,coenzymemetabolism;9,lipidmetabolism;10,molecu-
larchaperonesandrelatedfunctions;11,cellwallbiogenesisand
it is a heterotrophic organism that lacks a number of
celldivision;12,secretionandmotility;13,inorganiciontrans-
biosynthetic capabilities (Gonzalez et al. 1998). The port; 14, general functional prediction only; 15, no functional
majority of the archaea are autotrophs and it seems prediction.
610 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Archaeal Genomics and Evolution
tion of the archaeal COGs. In particular, most of the routeforprolinebiosynthesisinM.jannaschiiappears
basic components of the translation, transcription, to be through the deacetylation of N-acetylglutamate
and replication systems are conserved consistently in g-semialdehyde into g-glutamic semialdehyde, fol-
allfourspecies;thesameistrueofanumberofproteins lowed by its conversion into pyrroline-5-carboxylate
implicatedinrepairandrecombination(Fig.3). and then to proline as shown for bacteria and yeast
In other functional categories of genes, the ge- (AdamsandFrank1980).M.jannaschiiencodesanor-
nomeplasticityrevealedbyCOGanalysisismorepro- thologoftheN-acetylornithinedeacetylase(ArgE)that
nounced.Becauseoftheapparentlossofanumberof catalyzesthefirststepofthispathway.Thesecondstep
biosynthetic pathways in the heterotrophic P. horiko- of the pathway, conversion of g-glutamic semialde-
shii,therearerelativelyfewmetabolicenzymesamong hyde to pyrroline-5-carboxylate, occurs spontane-
the all-archaeal COGs, and in fact, it does not seem ously. However, the ortholog of the bacterial enzyme
possible to delineate even a single metabolic pathway forthelaststepofprolinebiosynthesis,namelypyrro-
that would be completely orthologous in all four ar- line-5-carboxylate reductase (ProC), is not encoded in
chaea(Table1).Amongthethreeautotrophicspecies, the M. jannaschii genome and should have been dis-
most of the steps of the central pathways are repre- placed by another dehydrogenase that remains to be
sented by orthologs; nevertheless, almost each path- identified experimentally. Remarkably, A. fulgidus en-
way has at least one step where nonorthologous dis- codesonlytheArgEorthologandM.thermoautotrophi-
placement is likely (Table 1). The biosynthesis of cumonlytheProCortholog.Itappearsthatinthiscase,
branchedchainaliphaticaminoacids(leucine,isoleu- weobservenonorthologousdisplacementofanentire
cine,valine)isanexampleofacomplexpathwaythat (albeit short) pathway whereby acquisition of the or-
is,initsentirety,representedbyorthologsinthethree nithinecyclodeaminasegenebyA.fulgidusandM.ther-
autotrophicarchaeaaswellasinmostbacteria.Thisis, moautotrophicumhasmadetheenzymesoftheoriginal
however,anexceptionratherthantheruleamongthe pathwayofprolinebiosynthesisdispensable.
archaealmetabolicpathways—fewofthemconsistex- In addition to the cases of apparent nonortholo-
clusively of orthologs of bacterial enzymes. In most gousdisplacement,thereareseveralimportantgapsin
pathways,atleastoneortworeactionsarepredictedto ourunderstandingofmetabolicpathwaysinalleuryar-
becatalyzedeitherbyknownarchaea-specificenzymes chaeota. The archaeal version of sugar metabolism is
orbyyetuncharacterizedones(Table2).Inthereadily particularly puzzling. There is no doubt that autotro-
detectablecasesofnonorthologousgenedisplacement, phic archaea possess the capabilities to synthesize ri-
oneofthealternativesolutionsisfrequentlybasedon bose, deoxyribose, and the sugar components of the
orthologsoftherespectivebacterialenzymes,whereas cellenvelope.Itisunclear,however,howtheyaccom-
theotheroneseemstobeuniqueforarchaeaandisnot plishthisintheabsenceofaldolase,fructosebisphos-
always identifiable. This is, for example, the situation phatase, transaldolase, transketolase, and pentose-5-
with a critical reaction in glycolysis, namely the for- phosphate 3-epimerase (see Table 1). Genes for all
mation of pyruvate from phosphoenolpyruvate. M. these enzymes are missing in M. thermoautotrophicum
jannaschii and P. horikoshii encode an ortholog of the andA.fulgidus,whereasM.jannaschiihasgenescoding
bacterial pyruvate kinase that is predicted to catalyze forthethreelatterenzymesbutnottheformertwo.It
this reaction. Pyruvate kinase, however, is not detect- appears that compared to bacteria, the archaeal sugar
able in the other two archaea. Given that the other metabolism shows systematic nonorthologous dis-
components of the trunk portion of the glycolytic placement of enzymes. Interestingly, one of the ar-
pathwayarepresentandthatthereactioncatalyzedby chaeal COGs includes predicted aldolases that are
pyruvatekinaseisindispensableforthecompletionof highly conserved in all four archaea and are ortholo-
glycolysis, nonorthologous displacement must be in- gous to the recently identified class I fructose-
voked.Themostlikelydisplacingenzymeisphospho- biphosphate aldolase from E. coli (Thomson et al.
enolpyruvate synthase, which is conserved in all ar- 1998).Therearetwoparalogousrepresentativesofthis
chaea and might produce pyruvate by reversing its familyofaldolasesinM.jannaschiiandA.fulgidusand
typicalreaction. only one member in M. thermoautotrophicum and
Nonorthologousgenedisplacementisnotablealso P.horikoshii(Table1).Theseenzymesarelikelytocata-
in the archaeal amino acid metabolism. For example, lyze key reactions both in pentose and in hexose bio-
different archaeal species apparently use radically dif- synthesis;theexactpathwaysremaintobestudiedex-
ferentpathwaystosynthesizeproline.InM.thermoau- perimentally.
totrophicumandA.fulgidus,prolinecanbeformedfrom Archaeal COGs that contain four or three mem-
ornithine in a single reaction catalyzed by ornithine bersaccountforthemajorityofknownhousekeeping
cyclodeaminase(Sansetal.1988).M.jannaschiiandP. functions, with several notable exceptions (e.g., those
horikoshii lack this enzyme, and while the latter is ex- inthetranslationmachinerydiscussedabove),andin
pected to be a proline auxotroph, the only possible a sense, may be considered an idealized minimal ar-
Genome Research 611
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Makarova et al.
Table1. OrthologousandNonorthologousMetabolicPathwaysandEnzymesinArchaea
Nonorthologous
Orthologs Genes genedisplacement:
ofbacterial missing orthologsof
genesfound inall bacterialgenes
inallfour four foundonly
Enzymes(genes) archaeal archaeal insomeofthe Consequencesfor
Pathway inthepathwaya genomes genomes archaealgenomes archaealmetabolism
Glycolysis hexokinase(glk),phosphogluco- dhnA,tpi, glkpfkA pgiispresentonlyin bacterial-typehexokinase
isomerase(pgi), gapA,pgk, Mj;pykAispresent andphosphofructokinase
phosphofructokinase(pfkA), pgmd,eno inMjandPh,not areapparentlydisplaced
aldolase(fba/dhnA), foundinMtandAf bynonorthologous
triosephosphateisomerase ADP-dependentenzymes;
(tpi),glyceraldehyde thelackofpyruvate
3-phosphatedehydrogenase kinaseinMtandAfis
(gapA),3-phosphoglycerate probablycompensated
kinase(pgk),phosphoglycero- byphosphoenolpyruvate
mutase(pgm/yibO),enolase synthaseworkinginthe
(eno),pyruvatekinase(pykA) reversedirection.
Gluco- phosphoenolpyruvatesynthase ppsA,eno, fbp pgiispresentonlyin otherphosphohexomutases
neogenesis (ppsA),enolase(eno),phos- pgm,pgk, Mj,notfoundin (e.g.,phosphomannomu-
phoglyceromutase(pgm), gapA,tpi, Mt,Af,andPh tase)areprobablyused
3-phosphoglyceratekinase dhnA forpolysaccharide
(pgk),glyceraldehyde3-phos- biosynthesisinMt,Af,
phatedehydrogenase(gapA), andPh.
triosephosphateisomerase
(tpi),aldolase(fba/dhnA),
fructosebisphosphatase(fbp),
phosphoglucoisomerase(pgi)
Pentose glucose-6-phosphatedehydro- rpiA zwf, tktAgeneissplitin thispathwayisnotfunc-
phosphate genase(zwf),6-phosphoglu- gnd Mj,absentinMt, tionalinanyofthese
shuntand conatedehydrogenase(gnd), Af,andPh;talAand archaea.Themechanism
pentose transketolase(tktA);trans- yhfDarepresentin ofpentosephosphate
biosynthesis aldolase(talA),pentose-5- Mj,notinMt,Af,or biosynthesisisnotclear.
phosphate-3-epimerase(yhfD), Ph;deoCispresent ApredictedDhnA-type
ribose5-phosphateisomerase onlyinMt aldolasethatishighly
(rpiA),deoxyribose-phosphate conservedinallfour
aldolase(deoC) archaea(MJ0400,
MJ1585;MTH579;
AF0108,AF0230;
PH0082)maycatalyze
theformationofribose
fromglyceraldehyde-
3-phosphateandacetal-
dehyde.Alternatively,in
Mt,thisreactionmight
becatalyzedbytheDeoC
ortholog(MTH818).
Entner– glucose-6-phosphate edd zwf, archaealacktheclassical
Doudoroff dehydrogenase(zwf), eda Entner–Doudoroff
pathwayb,c 6-phosphogluconate path-wayandinstead
dehydratase(edd), appeartopossessa
2-keto-3-deoxy-6-phospho- modified,nonphos-
gluconatealdolase(eda) phorylatedver-sion.In
Mj,Af,andMt,members
oftheabovenewfamily
ofaldolasesmayfunction
inthispathwayas2-keto-
3-de-oxygluconatealdo-
lase(anonorthologous
displace-mentofEda).
Thearchaealgluconate
dehydrataseremains
unknown,whichpre-
cludesacompleterecon-
structionofthispathway.
TCAcyclec citratesynthase(gltA),aconitase acnA,icd, sucA, gltAispresentinMt MtandAfcanreduce
(acnA),isocitrate sucC,sucD, sucB andAf,butnotin a-ketoglutaratetocitrate
dehydrogenase(icd), frdA,frdB, MjorPh;Phhas andfurthertosuccinate
612 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Archaeal Genomics and Evolution
Table1. (Continued)
Nonorthologous
Orthologs Genes genedisplacement:
ofbacterial missing orthologsof
genesfound inall bacterialgenes
inallfour four foundonly
Enzymes(genes) archaeal archaeal insomeofthe Consequencesfor
Pathway inthepathwaya genomes genomes archaealgenomes archaealmetabolism
a-ketoglutaratedehydro- fumA,mdh fumAand,possibly, orsuccinyl-CoA;inMj
genase(sucA,sucB),succinyl- acnAgenes onlythepartfrom
CoAsynthase(sucC,sucD), oxaloacetateto
fumaratereductase(frdA,frdB), succinyl-CoAisoperative.
fumarase(fumA),malate
dehydrogenase(mdh)
Purinebio- phosphosphoribosylpyro- prsA,purF, purK, purTispresentin allfourarchaeaare
synthesis phosphatesynthase(prsA), purD,purL, purH2 MjandPhbutnot probablycapableof
amidophosphoribosyl- purM, inMtandAf; purinesynthesisdenovo;
transferase(purF),GAR purE,purC, purH1ispresent carboxylationofAIR
synthase(purD),GARtrans- purB,purA, onlyinAf;guaB is probablyoccurs
formylase(purN/purT),FGAM guaA presentinMj,Mt, spontaneously.Thestill
synthase(purL),AIRsynthase andPh,butnotin unidentifiedenzymesthat
(purM),NCAIRsynthase AF catalyzeformylationof
(purK),NCAIRmutase(purE), SAICARandAICARinall
SAICARsynthase(purC), fourarcheaeaand
adenylosuccinatelyase(purB), formylationofGARinMt
AICARtransformylase(purH2), andAfapparentlyuse
IMPcyclohydrolase(purH1), formateandATPas
adenylosuccinatesynthase substrates.IMP
(purA),IMPdehydrogenase dehydrogenaseismissing
(guaB),GMPsynthase(guaA), AFandisprobably
displacedbya
nonorthologous
dehydrogenase.
Pyrimidine carbamoylphosphatesynthase pyrB,ygeZ, none carAandcarBare Mj,Mt,AfandprobablyPh
biosynthesis (carA,carB),aspartate pyrD,pyrE, missinginPh arecapableofpyrimidine
carbamoyltransferase(pyrB), pyrF,pyrH, synthesisdenovo.The
dihydroorotase(pyrC/ygeZ), ndk,pyrG identityofcarbamoyl-
dihydroorotatedehydrogenase phosphatesynthaseinPh
(pyrD),orotatephosphoribo- remainsunclear.
syl-transferase(pyrE),oroti-
dine-5’-phosphatedecarboxy-
lase(pyrF),UMPkinase(pyrH),
NDPkinase(ndk),CTP
synthase(pyrG)
Histidine phosphosphoribosylpyro- prsA,hisG, none hisI2ispresentinMj Mj,Mt,andAfareall
biosynthesisc phosphatesynthase(prsA), hisI1,hisA, andMt,butnotin capableofhistidine
ATP-phosphoribosyltransferase hisH,hisF, Af;hisB1isfound biosynthesis;phospho-
(hisG),phosphoribosyl-ATP hisB2,hisC, onlyinMj;prsA ribosyl-ATPpyrophospha-
pyrophosphatase(hisI2), hisD andhisCgenesare taseinAfisprobably
phosphoribosyl-AMPcyclo- presentinPh displacedbysomeother
hydrolase(hisI1),58-ProFAR ATP/ADPase;allarchaea
isomerase(hisA),imidazole- encodedistanthomo-
glycerolphosphatesynthase logsofyeasthistidinol
(hisH,hisF),imidazoleglycerol phosphatase(HDsuper-
phosphatedehydratase familyhydrolases;Aravind
(hisB2),histidinolphosphate andKoonin1998b),one
aminotransferase(hisC), ofwhichmightdisplace
histidinolphosphatase(hisB1), HisB1inAfandMt.
histidinoldehydrogenase(hisD)
Branched threoninedeaminase(ilvA), ilvA,ilvB, none none allenzymesofleucine,
chain acetohydroxyacidsynthase ilvN,ilvC, isoleucine,andvaline
aminoacids (ilvB,ilvN),acetohydroxyacid ilvD,leuA, biosynthesisinbacteria,
bio- isomeroreductase(ilvC), leuC,leuD, archaea,andyeastare
synthesisc dihydroxyaciddehydratase leuB,ilvE orthologous.
(ilvD),2-isopropylmalate
synthase(leuA),
isopropylmalateisomerase
(leuC,leuD),
3-isopropyl-malate
Genome Research 613
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Makarova et al.
Table1. (Continued)
Nonorthologous
Orthologs Genes genedisplacement:
ofbacterial missing orthologsof
genesfound inall bacterialgenes
inallfour four foundonly
Enzymes(genes) archaeal archaeal insomeofthe Consequencesfor
Pathway inthepathwaya genomes genomes archaealgenomes archaealmetabolism
dehydrogenase(leuB),
glutamatetransaminase(ilvE)
Aromatic 3-deoxyheptulosonate aroD,aroE, aroG, none themechanismof
aminoacids 7-phosphatesynthase(aroG/ aroA,aroC, aroB, 3-dehydroquinate
biosynthesisc kdsA),3-dehydroquinate pheA1, aroK synthesisremainsunclear;
synthase(aroB),3-dehydro- pheA2, shikimatephosphorylation
quinatedehydratase(aroD), tyrA2,tyrB, inallautotrophicarchaea
shikimatedehydrogenase trpD1, isprobablyperformedby
(aroE),shikimatekinase(aroK), trpE, anarchaea-specifickinase.
5-enolpyruvoylshikimate trpD2,
3-phosphatesynthase(aroA), trpC2,
chorismatesynthase(aroC), trpC1,
chorismatemutase(pheA1), trpA,trpB
prephenatedehydratase
(pheA2),prephenate
dehydrogenase(tyrA2),
tyrosineaminotransferase
(tyrB),antranilatesynthase
(trpD1,trpE),antranilate
phosphoribosyl-transferase
(trpD2),phosphoribosylan-
tranilateisomerase(trpC2),
indole-glycerolphosphate
synthase(trpC1),tryptophan
synthase(trpA,trpB)
Threonine aspartokinase(thrA1),aspartate thrA1,asd, none thrBispresentinMj inMtandAf,homoserine
biosynthesis semialdehydedehydrogenase thrA2,thrC andPh,butnotin kinaseisprobably
(asd),homoserinedehydroge- MtandAf displacedbyadifferent
nase(thrA2),homoserine kinase.
kinase(thrB),threonine
synthase(thrC)
Methionine aspartokinase(metL1),aspartate metL1,asd, metA, metB/metCisfound inallfourarchaea,thethree
biosynthesis semialdehydedehydrogenase metL2, onlyinPh,notin stepsleadingfrom
(asd),homoserinedehydro- metE Mj,Mt,orAf homoserineto
genase(metL2),homoserine homocysteineare
transsuccinylase(metA), probablydisplacedbya
cystathionineg-synthase singlereactioncatalyzed
(metB),b-cystathionase byasulfurtransferase.
(metC),methioninesynthase
(metE/metH)
Arginine acetylglutamatesynthase argA2,argB, none argEispresentinMj, inMt,acetylornithinaseis
biosynthesis (argA2),acetylglutamate argC, Af;andPh,butnot probablydisplacedbya
kinase(argB),acetylglutamate argD,argF, inMt;argGispres- differentacetyltransferase.
phosphatereductase(argC), argH entinMj,Mtand
acetylornithineaminotrans- AfbutnotinPh
ferase(argD),acetylornithinase
(argE),ornithinecarbamoyl-
transferase(argF),argininosuc-
cinatesynthase(argG),
argininosuccinatelyase(argH)
NAD aspartateoxidase(nadB), nadB,nadC, none nadAispresentinMj, nadAisprobablydisplaced
biosynthesis quinolinatesynthase(nadA), nadE Mt,andPh,but inAFbyadifferent
quinolinatephosphoribosyl- notinAF enzyme;nadDgene
transferase(nadC),nicotinic (predictedE.coliyneB)
acidmononucleotideadenylyl- remainsunidentifiedin
transferase(nadD),deamido- bacteriaandarchaea.
NADammonialigase(nadE)
Riboflavin GTPcyclohydrolaseII(ribA), ribD2,ribB, ribC ribAandribD1are ribCisdisplacedbyan
biosynthesisc pyrimidinedeaminase(ribD1), ribE presentinAfbut archaea-specificriboflavin
pyrimidinereductase(ribD2), notinMjorMt synthase(Eberhardtetal.,
614 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Archaeal Genomics and Evolution
Table1. (Continued)
Nonorthologous
Orthologs Genes genedisplacement:
ofbacterial missing orthologsof
genesfound inall bacterialgenes
inallfour four foundonly
Enzymes(genes) archaeal archaeal insomeofthe Consequencesfor
Pathway inthepathwaya genomes genomes archaealgenomes archaealmetabolism
3,4-dihydroxybutanone-4- 1997);themechanismof
phosphatesynthase(ribB), 2,5-diamino-6-ribosyl-
6,7-dimethyl-8-ribityllumazine amino-4-pyrimidone
synthase(ribE),riboflavin 58-phosphateformation
synthase(ribC) MjandMtremains
unclear.
Siroheme Glutamyl-tRNAreductase hemA,hemL, none none allenzymesofsiroheme
biosynthesisc (hemA),glutamate hemB, biosynthesisinarchaea
1-semialdehydeaminotrans- hemC, areorthologousto
ferase(hemL),probilinogenIII hemD, bacterialones.
synthase(hemB),hydroxy- cysG2,
methylbilanesynthase(hemC), cysG1
uroporphyrinogenIIIsynthase
(hemD),uroporphirinogen
methyltransferase(cysG2),
dimethyluroporphirinogenIII
dehydrogenase(cysG1)
Cobalamin uroporphyrinogenIIImethylase cysG2,cbiH, cobA, cbiL,cbiJ,cbiT,and inMj,Mt,andAf,cobAand
biosynthesisc (cysG2),precorrin-2methylase cbiF,cbiE, cobD cobNarepresent cobDgeneproductsare
(cbiL),precorrin-3Bmethylase cbiC,cbiA, inMjandMtbut probablydisplacedby
(cbiH),precorrin-4methylase cbiP,cbiB, notinAf;cobTis archaea-specificadenosyl-
(cbiF),precorrin-6Areductase cobS foundonlyinAf andaminotransferases,
(cbiJ),precorrin6Bmethylase respectively;itisnotclear
(cbiE),precorrin6Bdecarbox- whetherthispathwayis
ylase(cbiT),precorrin-8x functionalinAf.
isomerase(cbiC),cobyrinic
acida,c-diamidesynthase
(cbiA),cobaltinsertionprotein
(cobN),cob(I)alaminadenosyl-
transferase(cobA),cobyricacid
synthase(cbiP),cobyricacid
aminotransferase(cobD),
cobinamidesynthase(cbiB),
nicotinate-nucleotide:dimethyl-
benzimidazolephosphoribosyl-
transferase(cobT),cobalamin
synthase(cobS)
Biotin pimeloyl-CoAsynthetase(bioW)e, none none alltheenzymesof probablyonlyMjiscapable
biosynthesis 7-keto-8-aminopel- thepathwayare ofbiotinbiosynthesis.
argonatesynthetase(bioF), presentinMjbut
7,8-diaminopelargonate onlyoneortwo
aminotransferase(bioA), enzymescanbe
dethiobiotinsynthetase(bioD), foundinMt,Af,
biotinsynthetase(bioB), andPh
biotin-[acetyl-CoAcarboxylase]
holoenzymesynthetase(birA)
aThegenesandpathwaysfollowthebiochemicaldataandnomenclaturedescribedforE.coliandS.typhimurium(Neidhardtetal.
1996).Genescodingformultidomainproteinswithmorethanoneenzymaticactivityaredividedintoseparatedomains,startingfrom
theamino-terminaldomain.Knowncasesofnonorthologousgenedisplacementsareindicatedwithaslash;genesencodingdifferent
subunitsofthesameenzymeareseparatedbycommas.
b6-Phosphogluconate dehydratase gene (edd) is apparently present in Mj, Mt, and Af, but these ORFs probably function as dihy-
droxyaciddehydratases(ilvD),whichisaparalogofedd.
cMostenzymesofthesepathwaysarenotencodedinthePyrococcushorikoshiigenome;exceptionsareindicated.
dOrthologsofB.subtilispgmgene,correspondingtotheE.coliyibO.
eOrthologsofB.subtilisgenebioW,notfoundinE.coli.
Genome Research 615
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Makarova et al.
Table2. SynapomorphiesinEuryarchaeota(Examples)
Representatives
Uniquesharedcharactersnonarchaeal
COGdescription Mj Mt Af Ph homologs(seealsoFig.8)
DNApolymeraseII MJ1630 MTH1536 AF1722 PHBN021 ahighlyconservedarchaealenzymewithout
largesubunit similaritytoanyotherproteins,withthe
exceptionofaC4Znfingerresemblingthose
ineukaryoticDNApolymerased.
DNApolymeraseII MJ0702 MTH1405 AF1790 PHBN023 Predictedactivephosphohydrolase(phosphatase)
smallsubunit, PHAZ021 inarchaeaandaninactivatedformin
predicted eukaryotes(AravindandKoonin1998a).
phosphohydrolaseof
thecalcineurin-like
superfamily
PredictedATP-depen- MJ0414 MTH1221 AF0849 PHBG013 verylimitedsimilaritytootherATP-dependent
dentDNAligase DNAligasesexceptforonefromAquifex
aeolicus,probablyduetohorizontaltransfer
(AltschulandKoonin1998).
DNAexcisionrepair MJ1505 MTH1415 AF0358 PHAI012 consistsofatypicalhelicasedomainanda
enzyme nucleasedomainasopposedtotheapparent
eukaryoticorthologs(ERCC4/RAD1)inwhich
thehelicasedomainappearstobeinactivated
(Aravindetal.1999).
DnaG-typeprimase-like MJ1206 MTH891 AF1899 PHAN003 uniquedomainorganizationwithaN-terminal
proteins helicasemotifcombinedwiththeDnaG-type
(Toprim)domain(Aravindetal.1998).
DNA-directedRNA MJ0396/ MTH264/ AF1116/ PHBT008/ twosingledomain(S1andC4Znfinger
polymerasesubunit MJ0397 MTH265 AF1117 inPH744 domains)proteinsinallEuryarchaeota;afusion
(E8/E9) inSulfolobus;onlytheS1domainproteinisa
RNApolymerasesubunitineukaryotes(Fig.8).
DNA-dependentRNA MJ1042/ MTH1051/ AF1888/ PHCB020/ thesplitofthelargestRNApolymerasesubunit
polymeraseA8/A9 MJ1043 MTH1052 AF1889 PHCB021 geneintotwoadjacentgenesisuniqueto
subunits archaea.Botheukaryotesandbacteriaencode
highlyconservedorthologsofthearchaealA8
andA9subunitsasasinglepolypeptide(the
b8-subunitinbacteria).
PredictedHTH MJ0188 MTH1282 AF1259 PHCN020 uniquedomainorganization:twoCBSdomains
transcriptional fusedwithanHTHdomain.
regulators
Archaeosinesynthetase MJ1022 MTH1665 AF0587 PHBN035 two-domainarchitecture,withanadditional,
(archaea-specific predictedRNA-bindingPUAdomain,as
tRNAmodification) opposedtobacterialhomologs(queuine
synthetases)thatconsistoftheenzymatic
domainalone(Fig8;AravindandKoonin
1999a).
Translationelongation MJ0459 MTH1699 AF0574 AP000001 thearchaealEF-1bisasmallproteinof~ 120
factorEF-1b aminoacidswhereasalleukaryotichomologs
(orthologs?)containanadditionaldomain
homologoustoGSTs(Kooninetal.1994).
ATP-dependent MJ1417 MTH785 AF0364 PHBH031 onlythecarboxy-terminal,proteasedomainis
proteaseLon highlyconservedinarchaeaandbacteria;the
amino-terminalATPasedomaininthearchaeal
proteinsisdistinctfromtheATPasedomainof
Lon.
PilTfamilyATPase MJ1533 MTH246 AF1951 PHBP012 uniquedomainorganization—ATPase+
amino-terminalPINdomain.
GMPsynthetase MJ1131 MTH710 AF0253 PHAU017 ATPase(toprow)andglutamineamidotransferase
subunits—PP-family (bottomrow)moietiesoftheGMPsynthetase
ATPasesand areseparatepolypeptides.Orthologsofeach
MJ1575 MTH709 AF1320 PHAU016
glutamine subunitinbacteriaandeukaryotesaredomains
amidotransferase ofasingleprotein.
Predictedenzymewith MJ0202 MTH1744 AF1104 PHBQ042 orthologwiththesamedomainarchitectureonly
anATP-grasp inAquifex;allotherhomologsaredistantly
domainandaredox relatedandlacktheredoxcenter.
activecenter
616 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 7, 2023 - Published by Cold Spring Harbor Laboratory Press
Archaeal Genomics and Evolution
chaeal gene complement. The COGs with two mem-
bers appear to account for more specific functions
linked to the organism’s particular life style, for ex-
ample, a number of COGs that include enzymes in-
volvedinmethanogenesisinM.jannaschiiandM.ther-
moautotrophicum.
Relationships Between Euryarchaeal Protein Families
and Their Bacterial and EukaryoticHomologs
The majority of the archaeal COGs have homologs in
other taxa. In the present analysis, we attempted to
distinguish carefully between true orthology (see Figure5 Relationshipbetweenmembersoftheuniversaland
nonuniversal subsets of the euryarchaeal COGs from M. jan-
Methods) and other homologous relationships that
naschii and A. fulgidus and their bacterial and eukaryotic ho-
typicallyincludeweaksequenceconservationordiffer-
mologs(1)M.jannaschii,theuniversalsubset;(2)M.jannaschii,
ences in domain architectures. There are notable dif- thenonuniversalsubset;(3)A.fulgidus,theuniversalsubset;(4)
ferences in the distribution of the apparent phyloge- A.fulgidus,thenonuniversalsubset.(Bacterial)Reliablebesthits
tobacterialproteins;(eukaryotic)reliablebesthitstoeukaryotic
neticaffinitiesfortheCOGsrepresentedinallarchaea
proteins.Areliablebesthitwasdefinedasonewithane-valueat
(universal) and those that include only three or two least10000timeslowerthanthatfortheotherdivisions(eukary-
archaeal species. For >50% of the universal archaeal oticorbacteria,respectively).Onlythehitswithe-values<0.001
COGs, orthologs were identified in both bacteria and wereanalyzed.(Redbars)Bacterial;(bluebars)eukaryotic;(yel-
lowbars)uncertain.
eukaryotes, in a sharp contrast to the nonuniversal
COGs for which this fraction comprised of only 28%
(Fig. 4A,B). A significant majority of the COGs that difference might reflect true phylogenetic affinities,
have only bacterial orthologs are not conserved in all differenceinevolutionaryratesindifferentfunctional
archaea,whereasmostoftheCOGsthathaveonlyeu- categories of proteins, or both. However, the finding
karyotic orthologs belong to the universal subset (Fig. that COGs consisting of two to three euryarchaeal
4A,B).Furthermore,thoseCOGsthatdonothaveany memberstypicallyshowagreatersimilaritytobacterial
homologs outside the archaea are poorly represented homologs,mightsuggestasignificantcontributionof
intheuniversalsubset. horizontaltransferofbacterialgenesintoarchaea.
Acomplementary,quantitativeanalysisofthedis- The functional distinction between bacterial and
tributionofsequencesimilaritiessupportstheseobser- eukaryotic COGs in archaea is clear-cut and is related
vations.ArchaealproteinsfromtheCOGsthatinclude tothefunctionaldifferencebetweentheuniversaland
only two or three species typically show the greatest specialized subsets discussed above (see Fig. 3). The
similaritytobacterialhomologs,incontrasttotheuni- bacterial COGs within the universal subset comprise
versalCOGsthataresignificantlyenrichedinproteins primarilyproteinsinvolvedinenergyproduction(e.g.,
mostsimilartotheeukaryotichomologs(Fig.5).This ferredoxins and numerous components of hydrog-
enasecomplexes),certainmetabolic
functions, such as coenzyme bio-
synthesis, and transport system
components.Interestingly,thisbac-
terial set also includes enzymes in-
volved in protein degradation and
potentially in chaperone-like func-
tions, such as three families of pre-
viously undetected predicted zinc-
dependent proteases (K.S. Maka-
rova, L. Aravind, and E.V. Koonin,
unpubl.).Furthermore,thebacterial
component of the universal COG
subset includes several repair en-
zymes, proteins involved in cell di-
vision, for example, chromosome
partitioning ATPases and stress re-
sponse proteins, such as the ho-
Figure4 Taxonomicdistributionofnonarchaelhomologsforuniversalandnonuniversal
mologs of the bacterial universal
subsetsofthearchaelCOGs.(A)Theuniversalsubset(543COGswithfourmemberseach);
stressproteinUspA.
(B)thenonuniversalsubset(783COGswithtwoorthreememberseach).
Genome Research 617
www.genome.org