Outlook GENOME ANALYSIS Archaeal and bacterial hyperthermophiles Archaeal and bacterial hyperthermophiles horizontal gene exchange or common ancestry? ravind and colleagues1 recently concluded that massive gene transfer has occurred from Archaea to the bacterial ancestors of the hyperthermophile Aquifex aeolicus. Their analyses were based primarily on similarity searches of all complete bacterial genomes against the nonredundant protein sequence database, showing that the genome of A. aeolicus2 has a much larger fraction of proteins with best hits to archaeal proteins than any other bacterium. In particular, they reported that 246 Aquifex proteins are most similar to archaeal proteins, with 26 of these proteins belonging to families found only in Archaea and Aquifex. Thus, the authors suggested that at least 10% of the Aquifex genes have been horizontally transferred from Archaea. Although we agree that gene transfer has played an important role in the history of life3,4, we do not agree with the conclusions that Aravind et al.1 reach. In particular, limitations imposed by their assumptions and flaws in the analyses and conclusions drawn will be discussed. The two most fundamental problems with the conclusions of the work cited of Aravind et al.1 are that the authors (1) ignore the evidence that Aquifex is the most deeply branching eubacterium with a complete genome sequence5 (Fig. 1) and (2) assume that hyperthermophilicity in Bacteria and Archaea are independent inventions. Firstly, ancestral genes passed vertically through the bacterial lineage could be transmitted to Aquifex (and possibly to other early diverging bacterial lineages) but are lost in the common ancestor of the more recently diverging bacterial lineages for which genome sequences are available. Neither the data nor the discussion of Aravind et al.1 deals with this simple explanation of genes that are shared by Aquifex and the Archaea but are absent in other bacterial genomes. Secondly, hyperthermophiles are represented among all of the deepest and least diverged lineages both in Bacteria and Archaea6 (Fig. 1), leading many workers6,7 but not all8 to argue that the last universal common ancestor (cenancestor) was a hyperthermophile. Aravind et al.1 ignored this possibility when they interpreted their data. Regardless of whether the root of the tree of life is placed in the bacterial branch9,10 or in the eukaryotic branch11, and regardless of whether life originated at a hot environment or started cool and later adapted to high temperatures12, it is possible – even likely – that hyperthermophilicity was invented once (prior to the last prokaryotic common ancestor stage; Fig. 1), and not independently at two or more later times (as explicitly assumed by Aravind et al.1). If the cenancestor was a hyperthermophile, it would be natural for it to pass genes contributing to A *Nikos C. Kyrpides nikos@darwin.life. uiuc.edu Gary J. Olsen gary@phylo.life. uiuc.edu Department of Microbiology, University of Illinois at UrbanaChampaign, IL 61801, USA. *Also at the Mathematics and Computer Science Division, Argonne National Laboratory, IL 60439, USA. 298 TIG August 1999, volume 15, No. 8 thermal tolerance to the archaeal hyperthermophiles and Aquifex by vertical inheritance. Consistent with this, we have identified presumptive homologs of at least a third of these genes in the incomplete genome of another deeply diverging bacterial hyperthermophile, Thermotoga maritima (see below). We also find methodological problems in the analysis of Aravind et al.1 The 246 Aquifex proteins reported as ‘reliable best hits’ with their archaeal homologs were defined by having an E-value (expected number of matches at least this good in random data) of at least 100 times lower than that obtained with any bacterial or eukaryotic protein, which is not a particularly stringent criterion. In reality there is no simple relationship between differences in expectation and being significantly more related. Nor is there any translation of this measure into relative phylogenetic distances (amino acid replacements per position), or even into a difference in percentage amino acid identity. Because they did not compare all pairs of sequences within a family, these data are not even sufficient for a cluster analysis, but the authors have drawn conclusions about the histories of genes (phylogenetic analyses). For statements about the histories of these genes, it would be more appropriate to use explicit phylogenetic analysis, supported by bootstrap analysis of confidence. Of the 60 protein families (27% of the 220 families that go beyond Archaea and Aquifex) for which Aravind et al.1 report such analyses, they find bootstrap support for an Aquifex–Archaea grouping in only 26 families. Thus, only 43% of the cases they examined (12% of these 220 ‘reliable best hits’ with the Archaea) are actually demonstrated to support their suggestion. Finally, because of our own interest in the set of proteins uniquely shared between Aquifex and Archaea, we repeated this analysis comparing our results with those of Aravind et al.1 Although there were a number of differences in the genes identified, the real importance of this analysis lies in the fact that the majority of these genes are found in only one or two of the four complete archaeal genomes. Thus, even if these genes have been horizontally transferred, we cannot possibly infer whether the transfer occurred from Archaea to Aquifex (as the authors suggested) or vice versa. In addition, we identified homologs of at least a third of these genes in the partial genome of Thermotoga maritima, another bacterial hyperthermophile, suggesting that vertical inheritance via a thermophilic lineage from the archaeal–bacterial common ancestor (Fig. 1) will be a more parsimonious explanation than independent lateral transfers as suggested by Aravind et al.1 0168-9525/99/$ – see front matter © 1999 Elsevier Science All rights reserved. PII: S0168-9525(99)01811-9 GENOME ANALYSIS Archaeal and bacterial hyperthermophiles FIGURE 1. A rooted phylogenetic tree of Bacteria, Eukarya and Archaea Escherichia Agrobacterium Planctomyces Flavobacterium Chlamydia Leptonema Synechocystis Bacillus Thermomicrobium Thermus Thermotoga Aquifex Outlook In summary, we find the ideas expressed by Aravind et al.1 to be very interesting, but we also argue that these authors have made assumptions (without offering justification) that led them to conclusions that do not follow from the data per se. In particular, alternative hypotheses on the history of extreme thermophilicity would suggest (regardless of the rooting of the tree) that substantial numbers of the genes discussed could be vertically inherited from the cenancestor. Giardia Tritrichomonas Physarum Entamoeba Dictyostelium Trypanosoma Paramecium Zea Coprinus Homo Desulfurococcus Sulfolobus Pyrococcus Thermoproteus Thermophilum Methanopyrus Methanobacterium formicicum Methanothermus Thermococcus Methanococcus vannielii 0.10 Methanococcus jannaschii Archaeoglobus Thermoplasma Haloferax Methanospirillum This Maximum Likelihood tree was inferred essentially as described in Ref. 5, and rooted as in Ref. 9. The heavy lines trace the evolution of extreme thermophilicity under the assumption that it originated only once. Scale bar: 0.10 amino acid substitutions per site. References 1 Aravind, L. et al. (1998) Evidence for massive gene exchange between archaeal and bacterial hyperthermophiles. Trends Genet. 14, 442–444 2 Deckert, G. et al. (1998) The complete genome of the hyperthermophilic bacterium Aquifex aeolicus. Nature 392, 353–358 3 Médigue, C. et al. (1991) Evidence for horizontal gene transfer in Escherichia coli speciation. J. Mol. Biol. 222, 851–856 4 Woese, C.R. (1998) The universal ancestor. Proc. Natl. Acad. Sci. U. S. A. 95, 6854–6859 5 Burggraf, S. et al. (1992) A phylogenetic analysis of Aquifex pyrophilus. Syst. Appl. Microbiol. 15, 352–356 6 Stetter, K.O. (1996) Hyperthermophilic prokaryotes. FEMS Microbiol. Rev. 18, 149–158 7 Pace, N.R. (1991) Origin of life-facing up to the physical setting. Cell 65, 531–533 8 Galtier, N. et al. (1999) A nonhyperthermophilic common ancestor to extant life forms. Science 283, 220–221 9 Iwabe, N. et al. (1989) Evolutionary relationship of archaebacteria, eubacteria and eukaryotes inferred from phylogenetic trees of duplicated genes. Proc. Natl. Acad. Sci. U. S. A. 86, 9355–9359 10 Brown, J.R. and Doolittle, W.F. (1995) Root of the universal tree of life based on ancient aminoacyl-tRNA synthetase gene duplications. Proc. Natl. Acad. Sci. U. S. A. 92, 2441–2445 11 Forterre, P. (1995) Thermoreduction, a hypothesis for the origin of prokaryotes. C. R. Acad. Sci. 318, 415–422 12 Forterre, P. (1996) A hot topic: the origin of hyperthermophiles. Cell 85, 789–792 Reply e welcome the discussion of the evolutionary mechanism(s) underlying the special relationship between archaeal and bacterial hyperthermophiles. First of all, to our satisfaction, we find that Kyrpides and Olsen1 agree with us on the critically important issue: such a special relationship does exist – something that was not at all obvious from the original publication on the Aquifex genome sequence2. In fact, this was the principal point that we tried to convey, as convincingly as we could, in the article that is discussed3. Perhaps we should have been more explicit about distinguishing between these basic observations and their evolutionary interpretation, which necessarily remains hypothetical. In that sense, it might be prudent to accept the criticism. However, we still believe that the hypothesis we preferred, namely massive horizontal gene exchange, is a better explanation for what is observed than the alternative hypothesis favored by Kyrpides and Olsen1, namely ances- W 0168-9525/99/$ – see front matter © 1999 Elsevier Science All rights reserved. PII: S0168-9525(99)01786-2 tral origin of the archaeal genes in Aquifex. The main reasons for this are simple and have little to do with the details of the phylogenetic methods used by us, or others, but rather stem directly from the nature of the special relationship. We discuss these reasons briefly below. With respect to the majority of its genes, Aquifex looks like a ‘garden-variety’ bacterium and does not show any specific affinity with the Archaea. A significant subset of the Aquifex genes, however, appears to be very different in that they show a much greater similarity to archaeal orthologs than to bacterial ones, and some are (so far) simply unique for Archaea and Aquifex (or, in several cases, shared with other thermophilic Bacteria). Is this the pattern of sequence conservation one would expect under the ‘conservation of ancestral features due to common lifestyle’ hypothesis favored by Kyrpides and Olsen1? Hardly so. The simplest form of this hypothesis would suggest that all (or perhaps most, allowing for some TIG August 1999, volume 15, No. 8 L. Aravind aravind@ncbi.nlm.nih.gov Roman L. Tatusov tatusov@ ncbi.nlm.nih.gov Yuri I. Wolf wolf@ncbi.nlm.nih.gov D. Roland Walker walker@ ncbi.nlm.nih.gov Eugene V. Koonin koonin@ncbi.nlm.nih.gov National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD 20894, USA. 299