Evolution of the Archaea - Molecular Evolution & Organelle Genomics

14 Pages·2002·0.38 MB·English

TheoreticalPopulationBiology61,409–422(2002)
doi:10.1006/tpbi.2002.1592
Evolution of the Archaea
Patrick Forterre
InstitutdeGee!nee!tique et Microbiologie,UMR 8621CNRS,Bat 409, Universitee!Paris-Sud, 91405OrsayCedex, France
E-mail: forterre@igmors:u(cid:1)psud:fr
and
Celine Brochier and Hervee! Philippe
Phylogee!nie,BioinformatiqueetGee!nome,UMR7622CNRS,Universitee!PierreetMarieCurie,9quaiStBernardBaa#t.C75005Paris,
France
ReceivedFebruary27,2002
Archaea, members of the third domain of life, are bacterial-looking prokaryotes that harbour many
uniquegenotypicandphenotypicproperties,testifyingfortheirpeculiarevolutionarystatus.Thearchaeal
ancestor was probably a hyperthermophilic anaerobe. Two archaeal phyla are presently recognized, the
Euryarchaeota and the Crenarchaeota. Methanogenesis was the main invention that occurred in the
euryarchaeal phylum and is now shared by several archaeal groups. Adaptation to aerobic conditions
occurred several times independently in both Euryarchaeota and Crenarchaeota. Recently, many new
groups of Archaea that have not yet been cultured have been detected by PCR amplification of 16S
ribosomalRNAfromenvironmentalsamples.Thephenotypicandgenotypiccharacterizationofthesenew
groupsisnowatoppriorityforfurtherstudiesonarchaealevolution. &2002ElsevierScience(USA)
1. INTRODUCTION Eubacteria). In the following decades, this discovery
promptedafloodofnewhypothesesaboutthenatureof
the Last Universal Cellular Ancestor (LUCA) and the
The discovery, by Woese and colleagues in the 1970s, phylogeneticrelationshipsbetweenthethreedomainsof
that three ribosome forms are present in living organ- life(Archaea,BacteriaandEukarya)(forrecentreviews
isms initiated a revolution in evolutionary biology and alternative viewpoints, see Gupta, 1998; Forterre
(Woese and Fox, 1977; Woese, 1987; Woese et al., and Philippe, 1999; Lopez-Garcia and Moreira, 1999;
1990). This tripartite division of the translation appara- Penny and Poole, 1999; Glansdorff, 2000; Woese, 2000;
tus had been previously obscured by early observation Cavalier-Smith, 2002; Gribaldo and Philippe, 2002, this
of ribosome structure that fitted well with the binary issue).
classification of organisms into eukaryotes (80S ribo- The evolution of Archaea per se is an important field
somes) and prokaryotes (70S ribosomes). However, of evolutionary biology. Since Archaea exhibit a wide
when comparative sequence analysis of the ribosome varietyofphenotypicandgenotypiccharacters,itwould
components became available, it turned out that ribo- be interesting to have a clear idea of the history of the
somal RNAs (rRNA) and ribosomal proteins of some archaeal domain, in order to understand the origin and
prokaryotes (the Archaea, formerly Archaebacteria) evolution of these characters. In this review, we will try
weredrastically different from those ofbotheukaryotes to sum up the state of the art regarding archaeal
and those of‘‘classical’’ bacteria(the Bacteria, formerly evolution. This is timely, as data from comparative
409 0040-5809/02$35.00
#2002ElsevierScience(USA)
Allrightsreserved.
410 Forterre,Brochier andPhilippe
genomics can now be combined with more traditional
phylogenetic and taxonomic approaches. At the time of
writing this review, 12 genomes of Archaea have been
completely sequenced and are available in public
databases (http://www.ncbi.nlm.nih.gov/PMGifs/Gen-
omes/a g.html).Furthermore,anup-to-datedescription
of all identified archaeal species has been recently
published in Volume I of the new Bergey’s Manual
edition (Booneand Castenholz,2001).Finally,archaeal
rRNA probes have been widely used by molecular
ecologists to investigate the worldwide distribution of
theorganismsofthisdomain,aswellasitsphylogenetic
depth(forrecentpapersseeMassanaetal.,2000;Lopez-
Garcia et al., 2001; Takai et al., 2001a, b).
2. BRIEF DESCRIPTION OF THE
DOMAIN ARCHAEA
The Archaea are prokaryotes (cell without nucleus)
thatcannotbeeasilydistinguishedfromBacteriabysize
or shape. However, although most Archaea look like
typical Bacteria, some have morphologies that are not
found in Bacteria, such as polygonal in halophilic
Archaea or very irregular cocci in particular hyperther-
mophiles (Fig. 1). This could reflect the absence of a
rigidcellwallinmostArchaea(see below) and/or novel
(still unknown) mechanisms for morphogenesis. Ar-
chaea exhibit a wide diversity of phenotypes, as is the
case for Bacteria. The first three phenotypes to be
recognized were the methanogens (strict anaerobes and
methaneproducers),thehalophiles(strictaerobesliving
in high-salt environments) and the thermoacidophiles
(aerobes living in hot and acidic environments) (Woese
and Fox, 1977). Organisms with such disparate pheno- FIG. 1. (A)ElectronmicrographoftheeuryarchaeonHaloarcula
types were first unified based on the similarities of their (formerly‘‘squarebacteria’’),fromDr.Rodriguez-Valera,(B)electron
micrograph of the crenarchaeon Pyrodictium abyssi (a hyperthermo-
rRNA sequences, and later on also by the unique
phile that can grow up to 1108C), the cells are inserted into a
structure of their membrane phospholipids (see below).
reticulated network whose function is still unknown and (C)
Many additional phenotypes were discovered among transmission electron micrograph of P. abyssi. EM pictures of P.
Archaeainthefollowingdecades,suchashyperthermo- abyssiarefromReinhardRachelandKarlStetter.
philic or psychrophilic methanogens, halophilic and/or
alkaliphilic methanogens, anaerobic, alkaliphilic and
neutrophilic hyperthermophiles (Mathrani et al., 1988;
Franzmannetal.,1997;Ollivieretal.,1998;Huberetal., Crenarchaeota account for nearly 20% of the total
2000). Finally, the explosion of environmental studies marine picoplankton oceanic biomass worldwide
based on PCR amplification of 16S rRNA has revealed (Karner et al., 2001). Many of these not yet cultivated
the widespread occurrence in water and soils of Archaea probably exhibit novel phenotypes. For
mesophilic Archaea with otherwise unknown pheno- instance, previously unsuspected archaeal methano-
types (Fig. 2). For example, a recent study has trophs were predicted recently from geomicrobiological
estimated that mesophilic Archaea of the phylum data (Orphan et al., 2001). Archaea are often viewed as
Evolutionof the Archaea 411
FIG. 2. Phylogenyofthearchaealdomainbasedon16SrRNAofbothculturedandunculturedspecies.Aneighbour-joiningphylogenetictree
wasconstructedfrom342archaealnearlycompletesequencesofSSUrRNA(1235positions).ThedistanceswerecalculatedwiththeTamuraand
Neimethod(TamuraandNei,1993)andthea-parameterwascomputedwithPAUP4b8.Scalebarscorrespondto10substitutionsper100positions
foraunitbranchlength.Thetreewasarbitrarilyrooted.Thesizeofthetrianglesisproportionaltothenumberofsequencesanalysed.Greytriangles
include SSU rRNA from both cultured and uncultured species, whereas the white triangle includes only environmental sequences (uncultured
species).
predominantoverBacteria inallextremeenvironments. in all other situations (high salt, low or high pH, low
This is indeed true for high-temperature environments, temperature,highpressure),Bacteriaarefoundtogether
since only Archaea can thrive at temperatures above with Archaeaand Eukarya (Rothschild and Mancinelli,
958C (and up to 1138C) (Huber et al., 2000). However, 2001).
412 Forterre,Brochier andPhilippe
ThemetabolicdiversityofArchaeaisalsoreminiscent withotherarchaealproteinswhentheirhomologuesare
of Bacteria. Apart from methanogenesis (presently searched in public databases. This is especially true for
unknowninBacteria)allmetabolicpathwaysdiscovered informational proteins (those involved in DNA replica-
inArchaeaalsoexistamongBacteria.Theycanbeeither tion, transcription and translation) that are usually
heterotrophs or autotrophs (chemio- or photo-lithoau- presentinmostarchaealgenomesandarenearlyalways
totroph) and use a large variety of electron donors and more similar between one archaeon and another than
acceptors (Huber et al., 2000). Photosynthesis based on between one archaeon and any bacterium or eukaryote.
chlorophyll has not been found in Archaea, whereas These archaeal informational proteins are also usually
photosynthesis based on bacteriorhodopsin, once be- muchmoresimilartothoseofEukaryathantothoseof
lieved unique to halophilic Archaea, has been recently Bacteria.
detected in planktonic Bacteria as well (Beja et al., In addition,archaeal genomes encodemany informa-
2001). Spore formers or organisms with complex cell tional proteins that have homologues in Eukarya but
cycle and multicellular stages are unknown among not in Bacteria. This is especially striking in the case of
Archaea,butthiscouldbesimplyduetoourincomplete DNA replication, since archaeal genomes encode
sampling of this domain. The only way to discriminate homologues of nearly all eukaryal DNA replication
between Archaea and Bacteria is thus by looking at the proteins but only one homologue of a bacterial DNA
molecular level. replicationprotein(EdgellandDoolittle,1997;Forterre,
Beside their specific rRNA, Archaea can be distin- 1999; Leipe et al., 1999). The eukaryotic-like putative
guished from Bacteria by the nature of their membrane replication proteins identified in archaeal genomes are
glycerolipids that are ethers of glycerol and isoprenol, mostlikelyinvolvedinactualarchaealDNAreplication,
whereas bacterial and eukaryal lipids are esters of as indicated by recent observation such as direct
glycerol and fatty acids (Kates, 1993). Archaeal interaction in vivo between the chromosomal origin
glycerolipids are also ‘‘reverse lipids’’, since the en- (oriC) and the eukaryotic-like initiator protein Cdc6
antiomeric configuration of their glycerophosphate (Matsunagaetal.,2001)andthesimilarsizeofOkazaki
backbone is the mirror image of the configuration fragments in Archaea and Eukarya (Matsunaga and
found in bacteria and eukaryal lipids. Another differ- Myllykallio, pers. comm.).
ence between Archaea and Bacteria is the absence of It is often considered that informational proteins
mureininArchaea,whereasthiscompoundispresentin constitute the ‘‘core’’ of the genome of any organism,
the cell wall of most Bacteria. Archaea exhibit a great because they are less subject to lateral gene transfer
diversity of cell envelopes (Kandler and Konig, 1998), (LGT), and are therefore more representative of the
most Archaea have a simple S-layer of glycoproteins actual history of the organisms (Jain et al., 1999).
covering the cytoplasmic membrane, whereas a few of Indeed, except for well-documented LGT between
them (Thermoplasmatales) only have a cytoplasmic Archaea and Bacteria in the case of amino-acyl tRNA
membrane containing glycoproteins. Some Archaea synthetases (Wolf et al., 1999; Woese et al., 2000), few
havearigidcellwallbasedeitheronheteropolysacchar- LGTs of informational proteins have been identified
ides (Halococcus) or pseudomurein (Methanobacter- between the two prokaryotic domains. For example, an
iales), the latter being Gram positive. The difference exhaustive study has shown that no LGT had occurred
between Archaea and Bacteria at the molecular level is between Archaea and Bacteria in the case of ribosomal
exemplified by the resistance of Archaea to most proteins (Brochier et al., 2002) even if LGTs were
antibiotics active on Bacteria. Early studies on the discovered within Bacteria (Brochier et al., 2000, 2002)
molecular biology of Archaea have shown that this and Archaea (Matte-Tailliez et al., 2002).
resistance was due indeed to critical differences in the The close similarity between eukaryal and archaeal
antibiotic targets (e.g., RNA polymerase or ribosomal informationalproteinscanbeexplainedbytwohypoth-
proteins) (for an early review on archaeal molecular eses: either Archaea and Eukarya are sister groups, and
biology, see Zillig, 1991). the‘‘eukaryotic’’charactersofArchaeaaresynapomor-
The coherence of Archaea at the molecular level is phies, or these characters are primitive (already present
now also well documented by comparative genomics intheLUCA,thussymplesiomorphies)anddivergentin
(Olsen and Woese, 1997; Forterre, 1997; Forterre Bacteria. We will not discuss this problem here (for a
and Philippe, 1999; Fitz-Gibbon and House, 1999; review see Forterre and Philippe, 1999; Poole et al.,
Makarova et al., 1999; Snel et al., 1999; Tekaia et al., 1999; Woese, 2000; Cavalier-Smith, 2002; Gribaldo and
1999; Wolf et al., 2001). In all archaeal genomes Philippe, this issue) since we will focus on the evolution
sequenced so far, most encoded proteins give first hits of the archaeal domain per se.
Evolutionof the Archaea 413
Interestingly, in contrast to informational proteins, bacterial genomes exhibit no synteny and very little
most operational proteins of Archaea are ‘‘bacterial- conservation of gene order, in agreement with rapid
like’’. This category includes enzymes of the primary genomic evolution by recombination (Huynen and
and secondary metabolisms, membrane receptors, Bork, 1998). Comparison of the two closely related
transporters and so on, but also some cell division ArchaeaPyrococcusabyssiandPyrococcushorikoshihas
proteins (Koonin et al., 1997; Jain et al., 1999). This is shown that the terminus of DNA replication is a hot
often interpreted as testifying for extensive LGT of spot of recombination, as in the case of Bacteria
operational proteins between the two prokaryotic (Myllykallio et al., 2000). Furthermore, the main
domains (Faguy and Doolittle, 1999; Makarova et al., rearrangement between the two Pyrococcus genomes
1999). However, although LGTs between hyperthermo- has occurred symmetrically to the axis formed by the
philicBacteriaandArchaeahavebeenwelldocumented origin and the terminus of replication (Makino and
by phylogenetic and/or genome context analyses (Gri- Suzuki,2001;Zivanovicetal.,2002).Thischaracteristic
baldo et al., 1999; Forterre et al., 2000; Nesbo et al., feature has been also observed in Bacteria, suggesting
2001), a supertree inferred only on operational proteins that major recombination events occurred between bi-
is very similar to the one inferred from informational directional replication forks (Tillier and Collins, 2000).
proteins (Daubin and Gouy, 2001), suggesting that These similarities in the mechanisms of genome evolu-
LGTs are not so frequent for operational genes. More tion of the two prokaryotic domains could reflect a
analysesarethusrequiredtotestthehypothesisthatthe common structural organization of the replication
similaritybetweenArchaeaandBacteriaforoperational apparatus (for example, association of the two replica-
genes is only due to LGTs. Alternative hypotheses that tion forks into a single replication factory) and of the
have been proposed to explain this similarity are the chromosomesegregationmachinery(inrelationwiththe
fusion of ancient bacterial and eukaryotic lineages to presence in Archaea of homologues of the bacterial
produce extant Archaea (Koonin et al., 1997) or the proteins Xer C/D), which are involved in the resolution
monophyly of Archaea and Bacteria, if the universal of recombinant chromosomes at the end of replication,
tree of life is rooted in the eukaryotic branch (Forterre andofthebacterialproteinsMinDandFtsZ,whichare
and Philippe, 1999). involved in the formation of septum for cell division.
However, it should be stressed that study of the
mechanisms of chromosome structure and evolution
in Archaea is still in its infancy, and that the diversity
of Archaea in this respect remains to be explored.
3. GENOME EVOLUTION IN ARCHAEA
For instance, two possible (instead of one) replica-
tion origins have been predicted in silico in
Comparative genomics suggests that the mechanisms Halobacterium (Kennedy et al., 2001) and the bacterial
of genome evolution in Archaea are quite similar to cell division proteins FtsZ and MinD are lacking in
thoseidentifiedinBacteria.Thiscanbeexplainedbythe all Crenarchaeota.
resemblance between the genome organizations in the Inadditiontochromosomalrearrangementslinkedto
two domains. For instance, despite eukaryotic-like DNA replication, recombination events have been
replication proteins, hyperthermophilic Archaea of the associated to IS elements in several archaeal genomes
genus Pyrococcus replicate their chromosome at high (Kennedy etal., 2001; Bruu.gger etal., 2002). ISelements
speed, bi-directionally and using a single replication are especially abundant in S. solfataricus and Halobac-
origin, as in Bacteria (Lopez et al., 1999; Myllykallio terium NRC1. Comparison of the three completely
et al., 2000; Matsunaga et al., 2001). Most highly sequenced Pyrococcus genomes suggested that a family
transcribed genes are transcribed co-directionally with of IS elements, only present in P. furiosus, has been
DNA replication, as in Bacteria, probably to avoid involved in most chromosomal rearrangements ob-
frequent head-to-head collision between the replication served in the latter species (Maeder et al., 1999;
and transcription apparatuses (Zivanovic et al., 2002). Lecompte et al., 2001). In Pyrococcus and Sulfolobus,
Shine–Dalgarno sequences are found in some Archaea chromosomal rearrangements linked to IS elements
as in Bacteria, and archaeal genes are often organized occur preferentially in the limit of one replichore (the
in operon. Finally, mechanisms of gene regulation in half chromosome comprised between the origin and
Archaea involve bacterial-like regulator proteins terminus of replication), suggesting a principle of
(Aravind and Koonin, 1999; Bell and Jackson, 2001). genome organization that is not yet understood
Except for closely related organisms, archaeal and (Zivanovic et al., 2001, 2002; Bruu.gger et al., 2002).
414 Forterre,Brochier andPhilippe
An important difference of the genome organization Prangishvili et al., 2001). It has been shown that IS
betweenthetwoprokaryoticdomainsisthatallBacteria elements present in Sulfolobus species have indeed
contain a gyrase and a negatively supercoiled genome, triggered active recombination between chromosomes,
whereas most Archaea have no gyrase and relaxed or plasmids and viruses (for review, see Bruu.gger et al.,
positively supercoiled genome (Charbonnier and For- 2002), in agreement with the above hypothesis on the
terre, 1994). Some Euryarchaeota, including the hyper- role of viruses in LGT. The recent analysis of the
thermophile Archaeoglobus fulgidus, have acquired complete genome of the halobacteriophage HF2 is
bacterial gyrases by LGT and have negatively super- especially relevant to the idea that viruses have been a
coiled genomes (Lopez-GarciaandForterre, 1999). Itis major vector for LGT between domains, since it
not known presently if differences in genome topology contains a mixture of archaeal, bacterial and bacter-
have an effect on the mechanisms of genome evolution. iophage-like genes (Tang et al., 2002). In particular, its
LGTs between different archaeal species, and also DNA polymerase seems more closely related to bacter-
between Bacteria and Archaea, have been involved as a iophage T4 DNA polymerase than to archaeal DNA
major mechanism of genome evolution in Archaea. polymerases (Filee!e et al., in press). Accordingly, LGTs
LGTs appear to be predominant between organisms areprobablynotlimitedbyenvironmentalbarrierssince
livinginsimilarenvironment.Thishasbeenshownboth mechanisms for genetic exchanges (e.g., virus or
by genome wide comparison (Aravind et al., 1998; transposableelements)shouldhaveevolvedandadapted
Nelson et al., 1999; Ruepp et al., 2000) and in the case themselves to various environments together with their
of some ribosomal proteins inside the archaeal hosts.
domain(Matte-Tailliezetal.,2002).Forexample,many
LGTs have been detected between the thermoacido-
philic Archaea Thermoplasma and Sulfolobus that
are otherwise phylogenetically very distant (Ruepp 4. ARCHAEAL PHYLOGENY, THE TWO
et al., 2000).
MAJOR PHYLA
IS elements are likely often involved in these LGTs.
For instance, a16 kb DNA fragmentflankedby two IS
has been transferred between Thermococcus littoralis Phylogenetic trees of the archaeal domain based on
and Pyrococcus furiosus (Diruggiero et al., 2000). IS 16S rRNA have led to split this domain into two phyla
elements could have also played a critical role in LGT (according to Bergey’s Manual definition), the Eur-
between Bacteria and Archaea since many of the IS yarchaeota and the Crenarchaeota (Woese et al., 1990).
detected in Archaea belong to families previously The Euryarchaeota have been named from the Greek
detected in Bacteria (Bruu.gger et al., 2002). The precise word euruB; meaning wide, because they encompass the
mechanism of LGT between the two prokaryotic greatest phenotypic diversity among known cultivable
domains remains mysterious. In particular, trans- species, with the halophiles, the methanogens, some
formation by naked DNA appears highly unlikely thermoacidophiles and some hyperthermophiles.
in the case of hyperthermophiles considering the In contrast, the phenotypic diversity of cultivable
high instability of DNA fragments at high tempera- Crenarchaeota is much more limited, with only
ture (Marguet and Forterre, 1994). LGT could hyperthermophilic species. This phylum has thus been
possibly involve transfection by viruses, especially named from the Greek word krZnZ meaning spring,
if one considers that operational proteins similar fromthepopularhypothesisthathyperthermophilesare
between Bacteria and Archaea include some outer attheoriginofallpresent-dayorganisms.Thisnamehas
membrane proteins (e.g., sugar transporters) known beenconserved,despitethedetectionbyPCRinvarious
to be potential viral receptors. One can thus imagine environments of many groups of mesophilic and
that bacteriophages sometimes interact with archaeal probably psychrophilic Crenoarchaeota that have not
homologues of their usual bacterial receptors, and yet been cultivated (Fig. 2).
then inject by mistake their DNA into archaeal cells Some data of comparative genomics support the
(and vice versa). Such rare events could readily division of Archaea into two distinct phyla, since many
lead to LGT if the injected DNA fragment contains proteinspresentinalleuryarchaealgenomesaremissing
mobile IS elements that can function in both prokar- in Crenarchaeota and vice versa. In particular, several
yotic domains. striking differences have been noticed between the two
Viruses and plasmids seem to be as common in phyla for some key proteins involved in DNA replica-
Archaea as in Bacteria (Benbouzid-Rollet et al., 1997; tion,chromosomestructureandcelldivision,suchasthe
Evolutionof the Archaea 415
absence of eukaryal-like histones, eukaryal RPA-like undergone LGT events during archaeal evolution, with
proteins and bacterial cell division proteins in all a bias for LGTs between Thermoplasmatales and
Crenarchaea (Bernander, 2000; Myllykallio and For- Crenarchaeota, and were thus not suitable for inferring
terre, 2000). However, it remains to be known if these species phylogeny. The phylogeny based on the remain-
differences are specific to cultivable hyperthermophilic ing 45 genes was very similar to the rRNA phylogeny
Crenarchaeota, or if they are a general characteristic of and strongly supported the partitions between Cre-
this phylum. Furthermore, some global genomic ana- narchaeota and Euryarchaeota. We have now carried
lyses have failed to recover the division of Archaea into out the same analysis, but including the genomes of
twophyla.Forexample, using fivedifferent approaches Sulfolobus tokodai and of Thermoplasma volcanium that
(e.g., gene content, gene order, concatenated ribosomal were published after our initial work was completed,
proteins) on ten archaeal genomes, Wolf et al. (2001) and we obtained similar results (Fig. 3). It should be
failed to recover the monophyly of the two phyla. noted that the Halobacteriales and the Thermoplasma-
Nevertheless, their phylogeny based on gene content talesdisplayverylongbranches.Therefore,whenavery
is very likely biased by frequent LGTs between distant outgroup is used to root the archaeal phylogeny
Thermoplasmatales (Euryarchaeota) and Sulfolobales (i.e., the Bacteria), the long-branch attraction artefact
(Crenarchaeota) (Ruepp et al., 2000). If Thermo- (Felsenstein, 1978) is a major problem (Philippe and
plasmatales are not taken into account, the two phyla Laurent, 1998). Accordingly, the long branches of
are monophyletic, in agreement with a previous study Halobacteriales and Thermoplasmatales likely emerge
based on gene content (Huynen et al., 1999). Similarly, early in ribosomal proteins’ phylogeny because they are
the use of a much more refined phylogenetic method attracted by the bacterial branch, as previously sug-
(through maximum likelihood inference of individual gested (Wolf et al., 2001).
genes and supertree reconstruction), performed on 459 Inconclusion,thedivisionofArchaeaintoCrenarch-
genes from seven Archaea, also recovered the mono- aeotaandEuryarchaeotacanbeviewedassupportedby
phyly of the two phyla (Daubin and Gouy, 2001; see phylogenetic analyses, at least for cultivable species
also Brown et al., 2001). (Fig. 2).Indeed,theanalysisofarchaealsequencesfrom
The lack of monophyly in the phylogeny based on the environment has somehow complicated the picture
ribosomal proteins obtained by Wolf and coworkers is of the archaeal domain from the 16S rRNA viewpoint.
more surprising (Wolf et al., 2001). We recently Although many of the new archaeal lineages that have
performed a detailed analysis of 53 ribosomal proteins been detected by PCR analysis of environmental
present in most of 14 archaeal species (Matte-Tailliez samples branch among Crenarchaeota and Euryarch-
et al., 2002). We found that eight genes have likely aeota, some of them branch in between. A tentative
FIG. 3. Phylogenyofthearchaealdomainbasedonconcatenatedribosomalproteinaminoacidsequences.Ribosomalproteinsequenceshave
been retrieved from completely sequenced genomes (with the exception of Haloarcula marismortui). Ribosomal proteins for which LGTs were
suspected were removed from the analysis, with our recently developed method (Brochier et al., 2002). The only modification was that all the
likelihoodwascomputedwithaG-lawmodel.Amaximumlikelihoodphylogenetictreewasconstructedfromtheconcatenated47ribosomalproteins
sequence (6220 positions). The calculations of the best tree and the branch lengths were conducted using the program PUZZLE with a G-law
correction.NumbersclosetonodesareMLbootstrapsupportscomputedwiththeRELLmethodupon2000top-rankingtreesusingtheMOLPHY
programwithoutcorrectionforamong-sitevariation.Scalebarscorrespondto10substitutionsper100positionsforaunitbranchlength.Thetree
wasarbitrarilyrootedbetweenCrenarchaeotaandEuryarchaeota.
416 Forterre,Brochier andPhilippe
third phylum, the Korarchaeota (from the Greek word rRNAtree(Fig. 2),togetherwiththeiranalysiswiththe
koroB;meaningyoungman),wassuggested(Barnsetal., refined method, will be critical to confirm or invalidate
1996). Sequences that branch even deeper than Kor- the monophyly of Archaea.
archaeota in the archaeal 16S rRNA tree were recently
reported(Takaietal.,2001a,b).Wethusconstructedan
archaeal phylogeny using all the Archaea for which
rRNAsequencesarealmostcomplete(Fig. 2).Thistree 5. STRUCTURE AND EVOLUTION OF
identifies a huge group of environmental sequences that
THE TWO ARCHAEAL PHYLA
cannot be attributed to either Euryarchaeota or
Crenarchaeota.TheKorarchaeotaandotheruncultured
lineages form a clade with cultured Crenarchaeota, but The Euryarchaeota have been divided into nine
with low bootstrap support (below 50%). It is thus not orders, five orders for the methanogens (Methanobac-
clear if the dichotomy of the archaeal domain will teriales,Methanomicrobiales,Methanococcales,Metha-
survive future studies. More phyla will very likely have nosarcinales and Methanopyrales), one for the
tobedefinedinArchaea.Thiscouldeventuallyproduce halophiles (Halobacteriales), one for the thermoacido-
an archaeal classification more similar to the bacterial philes (Thermoplasmatales) and two for the hyperther-
one (23 phyla are presently recognized in the Bacterial mophiles (Thermococcales and Archaeoglobales)
domain, Boone and Castenholz, 2001). Nevertheless, (Fig. 2).
onehasalsotobecarefulininterpretingthediversityof Phylogenetic trees based on 16S rRNA place the
environmental sequences, since divergent copies of Methanopyrales at the base of the euryarchaeal tree,
rRNA can exist within a single organism (Mylvaganam followed by Thermococcales, but with very weak
and Dennis, 1992; Amann et al., 2000). statistical support (the bootstrap supports for the four
For a long time, some authors have advocated the basal nodes are between 4% and 8% in Fig. 2). Since
paraphyly of Archaea, suggesting that Crenarchaeota Methanopyralesarehyperthermophiles(theonlyspecies
(called eocytes by Lake and colleagues) form a clade of this order, Methanopyrus kandleri, can grow up to
with Eukarya (Lake, 1988), based for example on an 1108C), this has suggested that ancestral Euryarchaeota
insertion in elongation factor 1a (Rivera and Lake, were both methanogens and hyperthermophiles. The
1992). A phylogenetic analysis of rRNA taking among- idea that methanogenesis (formation of methane from
site rate variability into account supports this claim H and CO ) is an ancestral phenotype was first
2 2
(Tourasse and Gouy, 1999). It is reasonable to assume advocated to support the name ‘‘Archaebacteria’’ itself
that the monophyly of Archaea generally observed in (WoeseandFox,1977).Thediscoveryofasubterranean
rRNAtrees(Woeseetal.,1990)isduetoalong-branch community of methanogens thriving on H has recently
2
attraction artefact, because Bacteria and Eukarya dis- prompted speculation about the possibility that metha-
play much longer branches than Archaea for rRNA. nogens are directly connected to a hot origin of life on
The use of methods less sensitive to this artefact Earthandpossiblyelsewhere(Chapelleetal.,2002).The
(Tourasse and Gouy, 1999) provided a good evidence exact position of Methanopyrales in the euryarchaeal
in favour of the paraphyly of Archaea. However, tree is thus of critical importance.
phylogenies based on genome content (Fitz-Gibbon The recent sequencing of the M. kandleri genome is
and House, 1999; Huynen et al., 1999) as well as on thusanimportantstepinourunderstandingofarchaeal
multiple markers (Brown et al., 2001; Daubin and evolution(Slezarevetal.,2002).Phylogeneticanalysisof
Gouy, 2001) recovered the monophyly of Archaea. In concatenated ribosomal proteins suggests that Metha-
addition, recent analyses of two crenarchaeal genomes nopyrales do not branch first in the euryarchaeal tree
(Aeropyrum pernix and Sulfolobus solfataricus) failed to but branch with other thermophilic and hyperthermo-
identify specific eukaryal-crenarchaeal proteins that philicmethanogens(Slesarevetal.,2002).Furthermore,
could support the hypothesis of paraphyly (Faguy and our phylogenetic analysis of ribosomal proteins limited
Doolittle,2000;Nataleetal.,2000).Onthecontrary,the to the archaeal domain (thus using more positions and
Crenarchaea lack several eukaryal proteins of the limiting the impact of long-branch attraction artefact)
informationalapparatusthatarepresentinEuryarchaea locates Thermococcales firmly at the base of the
(such as histones or some DNA replication proteins, euryarchaeal tree (bootstrap support of 86%, Matte-
Bernader, 2000; Myllykallio and Forterre, 2000). The Tailliez et al., 2002). This was confirmed by our new
completion of several genomes from Crenarchaeota, analysis with two additional species (bootstrap support
Korarchaeota and their relatives at the base of the 16S of 89%, Fig. 3). The position of Thermococcales at the
Evolutionof the Archaea 417
base of the euryarchaeal tree and the grouping of M. sincegenomesfromDesulfurococcaleshavenotyetbeen
kandleri with other methanogens thus suggest that sequenced. As in the case of Euryarchaeota, oxygen
methanogenesis (a very complex metabolic pathway) respiration is most likely a secondary adaptation in
hasoriginatedintheeuryarchaealdomainonlyafterthe Crenoarchaeota. However, we should be aware that
divergence between Thermococcales and other Eur- hyperthermophilic Crenoarchaeota that have been dis-
yarchaea. This would be in agreement with the absence covered in the last two decades represent only a small
of relics of methanogenesis in the genomes of Thermo- fraction of this phylum. From examination of Fig. 2, it
coccales, whereas such relics have been found in is striking that most of the archaeal groups in both
Archaeoglobus(Klenketal.,1997;Slesarev etal.,2002). Euryarchaeota and Crenarchaeota are in fact presently
BothrRNAandribosomalproteinbasedphylogenies uncultured. There is therefore much research to be
havesuggestedthatHalobacterialesareasistergroupto done both in classical microbiology and environ-
Methanomicrobiales (Fig. 3). This grouping makes mental genomics before a complete picture of the
sense since Methanomicrobiales are the only order of archaeal domain structure and evolution can be
methanogensthatdoesnotincludethermophilicspecies, obtained. A recent development has been the discovery
and contains halophilic species. This suggests that of parasitic archaea of reduced size (Nanoarchaeum
Halobacteriales originated from halophilic and meso- equitans) and genome (less than 0:5 Mb) that live in
philic methanomicrobiales that have lost their ability to parasitic association with hyperthermophilic
produce methane and have acquired both oxygen Crenoarchaeota of the genus Ignicoccus and branch
tolerance and respiratory enzyme complexes. Analysis very deeply in the crenoarchaeotal 16S rRNA tree
of the genes involved in aerobic respiration that can be (Huber et al., 2002).
detected in the genome of Halobacterium sp. NRCl has
indicated that these genes have been borrowed from
aerobicBacteriabyLGT(Kennedyetal.,2001).Indeed,
inoneoftheseclusters,thegenesforaerobicrespiration 6. NATURE OF THE ARCHAEAL
present in Halobacterium NRCl have no archaeal
ANCESTOR
counterpart,whereasintheothers,theyaremoreclosely
related to their bacterial than to their archaeal homo-
logues. Phylogenetic and genomic analyses of cultivable
BothrRNAandribosomalproteinbasedphylogenies Archaea suggest that the ancestor of all present-day
also suggest the existence of a large clade grouping Archaea was an anaerobe that probably lived when
Methanobacteriales, Thermoplasmatales, Archaeoglo- oxygen was still absent from the terrestrial atmosphere.
bales, Methanomicrobiales and Halobacteriales. How- Complete adaptation to an aerobic lifestyle is a
ever, phylogenies based on other genes or on gene secondary adaptation in the archaeal domain that
content failed to retrieve it (Brown et al., 2001; Daubin occurred several times independently: at least once in
and Gouy, 2001; Wolf et al., 2001). Furthermore, this Crenarchaeota(Sulfolobales)andtwiceinEuryarchaeo-
groupisnotrecoveredwhensequencesfromuncultured ta (Halobacteriales and Thermoplasmatales). However,
species are included in the 16s RNA tree (Fig. 2). The several Crenarchaeota are also microaerobes (Aeropyr-
analysis of additional genomes will thus be required to um pernix and Pyrobaculum aerophilum), so we cannot
sort out the phylogenetic relationships among these exclude that the archaeal ancestor itself had some
different lineages. oxygen tolerance. As previously noticed, this ancestor
The cultured Crenoarchaeota have been divided into was probably not a methanogen, at least if we consider
three orders according to 16S rRNA: the Thermopro- indeed that the rooting of the archaeal tree between
teales, the Desulfurococcales and the Sulfolobales Thermococcales and Crenarchaeota (Fig. 3) is correct.
(Fig. 2). The first two orders include only anaerobic Itisalsooftenconsideredthatthearchaealancestorwas
hyperthermophiles that live at neutral pH and include a chemiolithoautotroph, since chemiolithoautotrophic
the most hyperthermophilic organisms known today thermophiles are present in both archaeal phyla. For
(with maximal growth temperature up to 1138C), several authors, this is consistent with the theory of an
whereas Sulfolobales, which live at low pH (2–3), can autotrophic origin of life (Wachtershauser, 1990).
be either aerobes or anaerobes and have lower However, since heterotrophic hyperthermophiles are
temperature maxima (up to 858C). Thermoproteales present in both archaeal phyla (e.g., Thermococcales
are a sister group to Sulfolobales in 16S rRNA trees. in Euryarchaeota and Pyrobaculum, Aeropyrum and
Genomic data cannot yet be used to test this phylogeny Desulfurococcus in Crenarchaeota), one cannot exclude
418 Forterre,Brochier andPhilippe
that the archaeal ancestor was a heterotroph. In fact, evolved more slowly (Kollman and Doolittle, 2000;
considering the extent of LGTs in prokaryotes, it is Matte-Tailliez et al., 2002). The phylogeny of reverse
difficult to raise firm conclusions about the origin and gyrase, a unique DNA topoisomerase that introduces
evolution of such metabolic pathways. Moreover, the positive superturns in DNA, also argues for a
presence of many uncultivated lineages at strategic hyperthermophilicarchaealancestor,sinceitiscoherent
points of the phylogeny renders inference of the with the rRNA phylogeny (e.g., the separation of
ancestral metabolism very hypothetical until their Euryarchaeota and Crenarchaeota) (Forterre et al.,
metabolism will be characterized. 2000).
A fascinating aspect of the archaeal domain is its Reverse gyrase is present in all hyperthermophiles
connection with hyperthermophily. We have already (both Archaea and Bacteria) and absent in all genomes
mentionedthatonlyArchaeahaveapparentlybeenable from mesophiles or moderate thermophiles. From
to colonize biotopes with temperatures above 958C: comparative genomics analysis performed using the
Possibly, this can be explained by the unique structure ‘‘Phylogenetic Patterns Search Program’’ of the COG
of archaeal lipids that maintain the impermeability of database(Tatusovetal., 2001),extendedtothegenome
the cytoplasmic membrane to ions at such high of Sulfolobus solfataricus (She et al., 2001), it even
temperatures. When the membrane lipid layer becomes turned out that reverse gyrase is the only protein that
unable to maintain an efficient barrier to passive ion exhibitsthisstrikingphylogeneticdistribution(Forterre,
diffusion,thecellcannotbuilduptheiongradientsthat 2002). Phylogenetic and genome context analyses have
are required to produce ATP. The membrane of indicated that reverse gyrase has been transferred from
hyperthermophilic Archaea contains tetraether lipids ArchaeatohyperthermophilicBacteriabyLGT(atleast
that can form a monolayer highly resistant to passive twice independently), suggesting that the common
iontransfer.Thishasbeenexperimentallydemonstrated ancestor of all Bacteria was not a hyperthermophile
in vitro by the analysis of ion transport in reconstituted and that thehyperthermophilic phenotype originated in
vesicles formed with archaeal lipids. Hyperthermophilic Archaea (Forterre et al., 2000). This is consistent with
Bacteria also have unusual lipids that ‘‘try’’ to resemble the idea that LUCA was not a hyperthermophile, a
archaeal ones (tetraester or diether), but they have been viewpoint that is also supported by recent analysis
apparently unable to produce membrane that can suggestingthattheGCcontentofLUCArRNAwastoo
prevent ion leakage at temperatures above 958C (for a low to be compatible with a hyperthermophilic lifestyle
recentreviewontheimportanceoflipidsandmembrane (Galtier et al., 1999).
permeability in thermoadaptation, see Albers et al.,
2001).
Interestingly,mesophilicandevensomepsychrophilic
Archaea also have tetraether lipids (Schouten et al.,
7. CONCLUSION AND PERSPECTIVE
2000). Indeed, these unique lipids have the ability to
maintain a correct membrane fluidity in a wide
temperaturerange.Theappearanceofsuchlipidsmight The study of Archaea has confirmed the two initial
have thus predated (and anticipated) the appearance of predictionsbyWoeseandFox(1977),i.e.,thatArchaea
hyperthermophiles in the archaeal domain. Alterna- willexhibitaphenotypicdiversityatleastcomparableto
tively, these lipids might have appeared in a hyperther- that of Bacteria and that Archaea will be characterized
mophilic archaeal ancestor and their presence today in by unique features at the molecular level. In addition,
mesophilic and psychrophilic Archaea can be viewed as Archaeaturnedouttoexhibitamosaicoffeaturesfrom
a heritage from their hyperthermophilic ancestor. the two other domains that continue to stimulate
The later hypothesis is supported by the presence of discussions among evolutionists. In the course of their
many hyperthermophic lineages that emerge at the base history, Archaea have in particular retained (or
of both the euryarchaeal and the crenarchaeal phyla (at acquired) in their informational apparatus many char-
least for the cultivable species, the picture being much acters that they only share now with Eukarya. Despite
morecomplexwithunculturedspecies).Furthermore,in their differences in this respect, Archaea and Bacteria
both rRNA trees and ribosomal protein trees (Figs. 1 have been engaged in a continuous ‘‘dialogue’’ via
and 2), the hyperthermophilic Archaea have shorter LGTs, due to their similar prokaryotic lifestyles.
branches than their mesophilic counterparts, suggesting Whereas some Bacteria learned from Archaea how to
that hyperthermophiles might have retained more bypass the hyperthermophilic barrier (borrowing
ancestral characters, since their macromolecules have reverse gyrase), some Archaea learned from Bacteria