The Genom of Homo sapiens.pdf

296 ROGOZIN ET AL.Dm Hs Ce Sc Sp Ec At349152053611368816239837135819345035411679299842 586 3711202 1969 -3260267555000 30483835422-15802Figure 3. Parsimonious scenario of loss and emergence of genes(KOGs) for the most likely topology of the eukaryotic phylogenetictree. Numbers in boxes indicate the inferred number ofKOGs in the respective ancestral forms. Numbers next tobranches indicate the number of gene gains (emergence ofKOGs) (top, black) and gene (KOG) losses (bottom, red) associatedwith the respective branches; a dash indicates that thenumber of losses for a given branch could not be determined.Proteins from each genome that did not belong to KOGs as wellas LSEs were counted as gains on the terminal branches. Thespecies abbreviations are as in Fig. 1.elimination of numerous genes in the microsporidian;emergence of a large set of new genes at the onset of theanimal clade; and subsequent substantial gene loss ineach of the animal lineages, particularly in the nematodesand arthropods (Fig. 3). The estimated number of geneslost in S. cerevisiae after the divergence from the commonancestor with the other yeast species, S. pombe,closely agreed with a previous estimate produced using adifferent approach (Aravind et al. 2000).The parsimony analysis described here includes reconstructionof the gene sets of ancestral eukaryoticgenomes. Under the Dollo parsimony model, an ancestralgene (KOG) set for a given clade is the union of theKOGs that are shared by the respective outgroup and eachof the remaining species in the clade. Thus, the gene setfor the common ancestor of the crown group includes allthe KOGs in which Arabidopsis co-occurs with any of theother analyzed species. Similarly, the reconstructed geneset for the common ancestor of fungi and animals consistsof all KOGs in which at least one fungal species co-occurswith at least one animal species. These are conservativereconstructions of ancestral gene sets because, as indicatedabove, gene losses in the lineages branching offthe deepest bifurcation could not be detected. Under thisconservative approach, 3,365 genes (KOGs) were assignedto the last common ancestor of the crown group(Fig. 3). Most likely, a certain number of ancestral geneshave been lost in all, or all but one, of the analyzed lineagesduring subsequent evolution such that the gene setof the eukaryotic crown group ancestor might have beenclose in size to those of modern yeasts.EVOLUTION OF EUKARYOTICGENE STRUCTUREMost of the eukaryotic protein-coding genes containmultiple introns that are spliced out of the pre-mRNA bya distinct, large RNA–protein complex, the spliceosome,which is conserved in all eukaryotes (Dacks and Doolittle2001). The positions of some spliceosomal introns are3413conserved in orthologous genes from plants and animals(Marchionni and Gilbert 1986; Logsdon et al. 1995;Boudet et al. 2001). A recent systematic analysis of pairwisealignments of homologous proteins from animals,fungi, and plants suggested that 10–15% of the introns areancient (Fedorov et al. 2002). However, intron densitiesin different eukaryotic species differ widely, the locationof introns in orthologous genes does not always coincideeven in closely related species (Logsdon 1998), likelycases of intron insertion and loss have been described(see, e.g., Rzhetsky et al. 1997; Logsdon et al. 1998), andindications of a high intron turnover rate have been obtained(Lynch and Richardson 2002). It has been suggestedthat the proportion of shared intron positions decreasedwith increasing evolutionary distance and,accordingly, intron conservation could be a useful phylogeneticmarker (Stoltzfus et al. 1997). However, the evolutionaryhistory of introns and the selective forces thatshape intron evolution remain mysterious. Although recentcomparisons reveal the existence of many ancient intronsshared by animals, plants, and fungi (Fedorov et al.2002), the point(s) of origin of these introns in eukaryoticevolution and the relative contributions of intron loss andintron insertion in the evolution of eukaryotic genes remainunknown.We used the KOG data set for analysis of evolution ofintron–exon structure of eukaryotic genes on the scale ofcomplete genomes. KOGs that are represented in all analyzedspecies, with a possible exception of E. cuniculi,were selected, and orthologs from two more eukaryoticspecies, the mosquito Anopheles gambiae and the apicomplexanmalarial parasite Plasmodium falciparum,were added to these KOGs using the COGNITORmethod (Tatusov et al. 1997). Many of the KOGs includemultiple paralogs from one or more of the constituentspecies, due to lineage-specific duplications (see above);among these paralogs, the one showing the greatest evolutionaryconservation (defined as the mean similarity toKOG members from other species) was selected. For apair of introns to be considered orthologous, they were requiredto occur in exactly the same position in the alignedsequences of KOG members. Given the well-knownproblems in the annotation of gene structure and difficultiesin aligning poorly conserved regions of protein sequences,we used two approaches to the analysis of evolutionaryconservation of intron positions. Under the firstschema, all intron positions were extracted from automaticallyproduced alignments, whereas under the secondschema, only positions surrounded by well-conserved,unambiguous portions of the alignment wereanalyzed. Altogether, 684 KOGs were examined for intronconservation; these comprised the great majority, ifnot the entirety, of highly conserved eukaryotic genesthat are amenable for an analysis of the exon–intron structureover the entire span of crown group evolution. Theanalyzed KOGs contained 21,434 introns in 16,577unique positions (10,066 introns in 7,236 positions whenonly the conserved portions of alignments were analyzed);5,981 introns were conserved in two or moregenomes (4,619 in conserved regions). Most of the con-