The Genom of Homo sapiens.pdf

294 ROGOZIN ET AL.BeTs are combined into orthologous clusters representedin all or a subset of the analyzed genomes (Tatusov et al.1997; Montague and Hutchison 2000). This approach,amended with procedures for detecting co-orthologousprotein sets and for treating multidomain proteins, wasimplemented in the database of clusters of orthologousgroups (COGs) of proteins (Tatusov et al. 1997, 2001).The current COG set includes ~70% of the proteins encodedin 69 genomes of prokaryotes and unicellular eukaryotes(Tatusov et al. 2003). The COGs have been extensivelyemployed for genome-wide evolutionarystudies, functional annotation of new genomes, and targetselection in structural genomics (Koonin and Galperin2002 and references therein).A simple but critically important concept that was introducedin the context of the COG analysis is a phyletic(phylogenetic) pattern, which is the pattern of representation(presence–absence) of the analyzed species in eachCOG (Tatusov et al. 1997; Koonin and Galperin 2002).Similar notions have been independently developed andapplied by others (Gaasterland and Ragan 1998; Pellegriniet al. 1999). The COGs show a wide scatter ofphyletic patterns, with only a small minority (~1%) representedin all included genomes. Similarity and complementarityamong the phyletic patterns of COGs havebeen successfully employed for prediction of gene functions(Galperin and Koonin 2000; Koonin and Galperin2002; Myllykallio et al. 2002; Levesque et al. 2003).Phyletic patterns can be formally represented as strings of“1”s (for presence of a species) and “0”s (for absence ofa species), which can be easily input to a variety of algorithms.The evolutionary parsimony methods are amongthose that naturally apply to these types of data. We recentlyshowed that parsimonious evolutionary scenariosfor most COGs involve multiple events of gene lossand/or HGT (Mirkin et al. 2003).Recently, we extended the system of orthologous proteinclusters to complex, multicellular eukaryotes by constructingclusters of eukaryotic orthologous groups(KOGs) for seven sequenced genomes of animals, fungi,microsporidia, and plants (Tatusov et al. 2003). Here, weanalyze the phyletic patterns of KOGs to extract the hiddenevolutionary signals. In particular, we reconstruct theparsimonious scenario of evolution of the crown-groupeukaryotes by assigning the loss of genes (KOGs) andemergence of new genes to the branches of the phylogenetictree, and delineate the minimal gene sets for variousancestral forms. We then shift the study from the level ofgene sets to the level of gene structure, construct the evolutionaryscenario for intron positions in highly conservedgenes, and compare the dynamics of evolution ofthe gene repertoire and gene structure.KOGS FOR SEVEN SEQUENCEDEUKARYOTIC GENOMES: MAJORTRENDS IN GENOME EVOLUTIONEukaryotic KOGs were constructed by comparing thesequences of the (predicted) proteins encoded in thegenomes of three animals (Homo sapiens, the fruit flyDrosophila melanogaster, and the nematode Caenorhabditiselegans), the flowering plant Arabidopsis thaliana,two fungi (budding yeast S. cerevisiae and fission yeast S.pombe), and the microsporidian Encephalitozoon cuniculi.The procedure for KOG construction was a modificationof the one previously used for COGs (Tatusov etal. 1997, 2001) and is described in greater detail elsewhere(Tatusov et al. 2003). Unlike in the previous COGanalyses, we strived to produce a complete evolutionaryclassification of eukaryotic genes. The original COGsconsisted, at a minimum, of proteins from three species,which enhanced the power of the analysis and allowed incorporationof even those orthologs that showed low sequencesimilarity to each other. In the present analysis,we also identified clusters of putative orthologs from twospecies (TWOGs) and lineage-specific expansions of paralogsfrom each of the analyzed genomes (Lespinet et al.2002; Tatusov et al. 2003). Thus, at least in principle,each gene in the seven analyzed eukaryotic genomes isaccounted for in the emerging evolutionary classification.Of the 112,920 analyzed proteins, 65,170 belonged toKOGs (including TWOGs), 23,436 belonged to LSEs,and 24,314 remain singletons (“orfans”). Both the considerablelevel of evolutionary conservation—given thewide phylogenetic span of the analyzed genomes, eachKOG should be considered a highly conserved family—and the major contribution of LSEs are notable. Furthermore,the number of orfans is likely to be inflated as someof these undoubtedly are gene prediction artifacts. Figure1 shows the assignment of the proteins from each of theanalyzed eukaryotes to KOGs with different numbers ofspecies (including TWOGs) and LSEs. The fraction ofproteins assigned to KOGs generally decreases with theincreasing genome size, from 81% for the fission yeast S.pombe to 51% for the largest, the human genome. Thecontribution of LSEs shows the opposite trend, being thegreatest in the largest genomes, i.e., human and Ara-# proteins14000120001000080006000400020000sce0 1 ecu2 3 4 5 6 7# species in KOGsFigure 1. Assignment of proteins from each of the seven analyzedeukaryotic genomes to KOGs of different size and toLSEs. “0” indicates proteins without detectable homologs (orfans)and “1” indicates LSEs. Species abbreviations: (ath) Arabidopsisthaliana, (cel) Caenorhabditis elegans, (dme)Drosophila melanogaster, (ecu) Encephalitozoon cuniculi,(has) Homo sapiens, (sce) Saccharomyces cerevisiae, (spo,)Schizosaccharomyces pombe.spoathhsadmecel