The Genom of Homo sapiens.pdf

146 RUDD, SCHUELER, AND WILLARDdetailed functional annotation in terms of the impact differentsequence elements have on chromosome structureand function. Although some sequences in the vicinity ofcentromeres may indeed be without demonstrable function(and likely no different in that respect from the bulkof human genome sequences), others are plausible candidatesfor defining boundaries between euchromatin andheterochromatin, for establishing chromosomal regionswith characteristic levels of gene expression, for influencing(i.e., blocking or mediating) potential position effectson gene function, for anchoring chromosomeswithin preferred nuclear territories, and for contributingto sister chromatid cohesion during cell division, in additionto being responsible for centromere specification andfunction.As a step toward establishing an experimental approachsuitable for functional genome annotation, we andothers developed an assay based on formation of humanartificial chromosomes (Harrington et al. 1997; Ikeno etal. 1998), building on the success and impact of yeast artificialchromosome technology for understanding thefunction of components of the budding yeast genome(Murray and Szostak 1983). The development of an efficientand tractable human artificial chromosome systemwould involve assembly of required chromosomal elements(centromere, telomeres, and origins of DNA replication),together with genomic fragments whose genic orother functions one wished to examine (Larin and Mejia2002). Progress toward this goal has been made, and severaldifferent approaches have been used or are under development,based on cotransfection of candidate genomicsequences (Grimes et al. 2001), ligation of synthetic centromereand telomere components (Harrington et al.1997), or modification of human sequences isolated inyeast (Henning et al. 1999; Kouprina et al. 2003) and bacterialartificial chromosome constructs (Ebersole et al.2000; Mejia et al. 2001; Schueler et al. 2001; Grimes etal. 2002; Rudd et al. 2003). Whereas much of the early focusof this technology has been on optimization of denovo centromere formation (Harrington et al. 1997;Ohzeki et al. 2002; Rudd et al. 2003), proof-of-principleexperiments have shown that genes containing large fragmentsof chromosomal DNA from the human genomecan also be expressed and thus are amenable to study usingsuch assays (Grimes et al. 2001; Mejia et al. 2001;Ikeno et al. 2002).The most straightforward assay is illustrated in Figure6. In this approach, BACs containing ~30–100 kb of αsatellite are modified to introduce a drug-resistancemarker for selection in mammalian cells and, if desired,additional fragments from the human genome for functionaltesting. The BAC is then transfected (or microinjected)into cells in culture, and the resulting drug-resistantcolonies are screened for the presence of acytogenetically visible human artificial chromosome(Fig. 6). In ~5–50% of colonies (depending in part on theparticular α-satellite sequences used), a mitotically stablehuman artificial chromosome is detected in a high proportionof cells, containing both vector and input sequencesand colocalizing with kinetochore proteins detectedby indirect immunofluorescence. Importantly, anumber of non-α-satellite and noncentromeric controlgenomic fragments are incapable of de novo centromereformation using this assay (Ebersole et al. 2000; Grimeset al. 2002), indicating that the assay is specific for functionalcentromeric sequences.Using such an assay, we have demonstrated that bothDXZ1 and D17Z1 sequences are capable of generating denovo centromeres in human artificial chromosomes (Harringtonet al. 1997; Schueler et al. 2001; Grimes et al.2002; Rudd et al. 2003). This provides functional annotationfor at least part of the pericentromeric contigs describedearlier. However, it should be emphasized that theability of sequences adjacent to these higher-order repeatarrays to function as centromeres in this assay has not yetbeen evaluated. Furthermore, the α-satellite sequencesalone do not completely recapitulate mitotic centromerefunction, as human artificial chromosomes show a levelof chromosome nondisjunction and anaphase lag that issignificantly higher than that of intact, endogenous centromeres(Rudd et al. 2003). This suggests that other sequencesin these centromeric contigs may be necessaryfor at least some aspect of faithful chromosome segregationand stability. Thus, to completely annotate these regionswill require testing both monomeric α satellite andthe D17Z1-B sequences, using both the human artificialchromosome assay and extended chromatin studies toidentify which DNA is involved in the assembly of thespecialized chromatin that underlies centromere function(Blower et al. 2002; Cleveland et al. 2003).Figure 6. Functional centromere annotation using a human artificialchromosome assay (Grimes et al. 2002; Rudd et al. 2003).α-Satellite DNA hypothesized to play a role in centromere function(purple arrows) is cloned into a BAC vector containing adrug-resistance gene (R) and transfected into human tissue culturecells. Drug-resistant clones are screened for the presence ofan artificial chromosome. The artificial chromosome can beidentified by hybridization to a red α-satellite probe. Like normalchromosomes, artificial chromosomes bind antibodies tocentromere and kinetochore proteins (green).