The Genom of Homo sapiens.pdf

134 HAMILTON ET AL.Figure 3. Three-species comparison of the zinc-finger gene cluster on Hsa19q13.2 and the homologous mouse and rat clusters. Blockarrowsrepresent orientations and relative positions of genes; lines connecting genes indicate close relationships based on the sequencesof the zinc-finger arrays (relationships can include 1:1 orthology or, in several cases, clades where a single gene in one speciesis related to an expanded group of lineage-specific paralogs in the other). Human ZNF45 is related to mouse Zfp94 and two rat genes;the alternate gray line between mouse and rat maps indicates the position of an apparent remnant of the second ZNF45-like gene inmouse (indicated by a triangle).Sequence alignments of the lineage-specific KK duplicateshave revealed mechanisms by which new KKduplicates may be rapidly diverging in function. Onemechanism involves selection for nonconservative single-nucleotidesubstitutions at critical DNA-recognitionsites. Most amino acid residues in KZNF domains are requiredfor zinc binding and structural integrity, and theseare typically highly conserved (Jacobs 1992; Mascle et al.2003). Therefore, as expected, general surveys of nonsynonymousvs. synonymous mutations for completeKZNF-domain sequences reveal a strong signature of purifyingselection. However, when only those amino acidspredicted to be key to DNA binding are considered, atrend toward positive selection is indicated in some comparisonsbetween paralogs, whereas orthologs often conservethese amino acid positions. Such evidence was recentlydocumented in alignments of lineage-specificduplicates in the Hsa19q13.2/Mmu7 clusters (Shannon etal. 2003) plus selected additional KK genes (Looman etal. 2002). A higher amino acid replacement rate for theDNA-interacting sites has also been seen between othertypes of closely related KZNF genes (Sander et al. 2003).These data suggest that, in some cases at least, selectionmay be operating to favor diversity in the target recognitionof paralogous proteins.Paralog alignments have also revealed a second mechanismfor divergence in KZNF arrays that may have evenmore striking impact on DNA-binding properties of theduplicated genes. Not all genes have the same number offunctional zinc-finger motifs, even if they are closely related.For example, alignment of the six mouse homologsof the singleton human gene, ZNF235, revealed a patternof finger deletions and duplications that alters the arrangementand number of DNA-binding motifs. In mostcases, the size and linker-spacing of the surrounding fingerswere maintained, indicating that the deletions andduplications are driven by recombination between thetandem finger repeats (Shannon and Stubbs 1998; Shannonet al. 2003). In the case of ZNF235 and its six mouserelatives, one mouse gene (Zfp235) has retained the samefinger number and arrangement as in the human gene,hinting at a shared functional homology, but all othermouse Zfp235 paralogs carry differently altered arrangementsof fingers.A third mode of KK gene divergence involves redirectionof expression patterns for the duplicated genes, suchthat new gene copies are active in tissues or cell types inwhich ancestral genes or other paralogs are not expressed.For example, in the Hsa19q13.2 gene cluster,some of the human genes are ubiquitously expressed, butclosely related paralogs ZNF225 and ZNF284 are expressedin limited, nonoverlapping tissue types (Fig. 4).Therefore, KK paralog diversification may also arise byacquisition of altered expression patterns, presumably