The Genom of Homo sapiens.pdf

RECENT SEGMENTAL DUPLICATIONS 121Comparative AnalysesAre recent interspersed duplications of genomic sequencea common property of all genomes, or is their occurrencelargely restricted to the primate genome? Ouranalysis indicates two types of recent interspersed duplicationsexist in the human genome, the chromosome-specificand interchromosomal repeats. Similar duplications,at least in apparent size or frequency, have not been reportedas of yet for any other organism. Differences inmethods of genomic and genetic characterization in thesespecies, however, could largely have explained this effect.With the advent of whole-genome sequencing, wesought to address this question in an unbiased fashion, bydirect examination of nucleotide sequence within otherspecies. An identical analysis was performed for the recentlypublished genomes of Caenorhabditis elegans(Consortium 1998), Drosophila melanogaster (Adams etal. 2000), Takifugu rubripes (Aparicio et al. 2002), andMus musculus (Waterston et al. 2002). Very little evidencefor large (≥20 kb), highly homologous (≥90%) duplicationscould be found within these species (Table 3).Although retroposon accumulation biases have been documentedfor Drosophila, no subtelomeric or pericentromericclustering of duplicated segments could be ascertained.Furthermore, these comparisons indicate thatthe human genome is enriched 5- to 100-fold for such duplicationswhen compared to genomes of model organisms(Table 3). Although there may be several explanationsfor this effect (differences in genome size,differential rates of recombination, methodological differencesin sequence assembly, etc.), the structure of thehuman genome appears structurally distinct with respectto recent large-scale interspersed duplications.Table 3. Segmental Duplications in Other Sequenced OrganismsSize Fly Worm Fugu a Mouse Human b≥1 kb 1.20% 4.25% 2.18% ND 5.23%≥5 kb 0.37% 1.50% 0.03% 1.95% 4.78%≥10 kb 0.08% 0.66% 0.00% 0.70% 4.52%≥20 kb 0.00% ND 0.00% 0.11% 4.06%a Takifugu rubripes build 3.0.b Build 31, Nov. 2002.CONCLUSIONSBased on our current analysis of genomes, the genomeof Homo sapiens is unique in the abundance and distributionof large (≥20 kb), highly homologous (≥95%) segmentalduplications. This unusual architecture of the humangenome has important practical and biologicalimplications. In the late 1990s, there was considerable debatebetween advocates of the whole-genome shotgunand those of the clone-ordered approaches for humangenome sequence and assembly (Green 1997; Weber andMyers 1997). Our analysis of the initial private and publicgenome assemblies (Lander et al. 2001; Venter et al.2001) indicated that neither effectively resolved the organizationand sequence of regions containing large segmentalduplications. Ironically, combining both of theseapproaches did provide the most effective means for theidentification, characterization, and subsequent resolutionof many of these regions. These data suggest that acombined whole-genome shotgun and clone-ordered approachmay be the best strategy for the completion ofcomplex genomes that are laden with large segmental duplications.Given the fact that the human genome harborsnearly 100 such sites, each greater than 300 kb, high-sequence-identityduplications continue to impede gene annotationand SNP characterization of the human genome.Emerging evidence that such regions vary structurally dependingon the human haplotype further complicatestheir sequence and assembly. Taken in this light, it is perhapsnot surprising that the majority of the large gaps thatremain within the human genome project as of April 2003are flanked by such duplications.From the biological perspective, this architecture hastwo important implications—one functional and the otherstructural. The ability to juxtapose segments that wouldhave never shared proximity in the genome of an ancestralspecies offers tremendous potential for exon shufflingand domain accretion of the proteome. Many suchchimeric transcripts have now been documented(Courseaux and Nahon 2001; Bailey et al. 2002a;Stankiewicz and Lupski 2002; Bridgland et al. 2003)where different portions of the transcript originate fromdiverse regions. Few of these “fusions” appear to producefunctional proteins (Hillier et al. 2003). Rare exceptionshave been noted, such as the emergence of the TRE2 (alsoknown as USP6) gene specifically within the hominoidlineage. In this case, approximately half of its 30 exonsarose from a segmental duplication of the USP32 ancestralgene, whereas the amino-terminal portion of thisoncogene originated as a duplication of the TBC1D3 ancestralgene. The fused transcript emerged during the radiationof the great apes, producing a gene with tissuespecificity different from either of the progenitor genes(Paulding et al. 2003). In addition to gene innovationthrough exon shuffling, segmental duplications have thepotential to lead to the emergence of novel genes throughadaptive evolution. The remarkable positive selection ofthe morpheus gene family on chromosome 16, where thegenes show accelerated amino acid replacement an orderof magnitude above the neutral expectation, may be anexample of such an effect (Johnson et al. 2001).From the structural perspective, the current architectureof the human genome suggests that subtle remodulationof many specific chromosomal regions has occurredover short periods of primate evolution. This observationchallenges the rather static notion of genome evolutionthat has emerged from early karyotype and chromosomepaintingstudies. Whereas the majority of the humangenome seems to fit well with this model of conservation,the duplicated regions, in contrast, have been particularlyprone to multiple, independent occurrences of rearrangementthrough duplication. In humans, the majority of intrachromosomallyduplicated copies are separated bymore than a megabase of intervening sequence. Such architecturehas been rarely observed in other model organisms.The mechanism responsible for this event is un-