The Genom of Homo sapiens.pdf

EVOLUTION OF ZNF GENES 133Figure 2. (A) Generic polydactyl zinc-finger protein, in this example depicted with subsets of its zinc-finger motifs having two separatetargets or functions. (B) Schematic of a set of zinc fingers bound to their target, with a “code” simulated by the interacting shapes(protein on top, DNA below) to show the colinearity of the amino acid sequence in the fingers and the target stretch of DNA.DNA-binding domains during the course of eukaryoticevolution, and each confers a particular type of activity tothe DNA-binding KZNF domains. The effectors functionby mediating interactions between the KZNF proteinsand other protein cofactors, including some involved directlyor indirectly in chromatin remodeling (Collins et al.2001; Schultz et al. 2002). The most prevalent effectormotif in mammalian KZNF proteins, the KRAB domain,is predicted to include charged amphipathic helices thatbind to the RBCC region of the transcriptional corepressorTRIM28 (a.k.a. KAP-1, TIF1-β) (Collins et al. 2001).TRIM28, in turn, recruits other proteins such as HP1 heterochromatinproteins, histone methyltransferases, andhistone deacetylases. The complex of proteins attached tothe TRIM28 “scaffold” is thought to alter chromatinstructure via histone modification to repress the transcriptionof the targeted gene (Schultz et al. 2002). TheKRAB can be divided into subregions called A and B,which are typically encoded on separate exons; severallines of evidence have indicated that the A box is necessaryfor repression, whereas the B box contributes but isless critical for TF function, and many KK proteins do notcontain the KRAB B motif (Mark et al. 1999).The KRAB A motifs of KK proteins are highly conserved,both within and between species; this strict conservationis likely to reflect constraints imposed by the requirementfor TRIM28 binding. In contrast, theDNA-binding KZNF domains can be radically differentin length, due to the variable number of zinc fingers theycan include. Although the individual finger motifs have acommon structure, diversity is allowed both in the sequenceof the DNA contact region and in the number ofrepeats, as discussed below.DUPLICATION AND DIVERGENCE OF THEKRAB-KZNF GENE FAMILY IN MAMMALSCluster EvolutionWith only three mammalian genomes sequenced (human,mouse, rat), our picture of KK gene evolution in thislineage is still incomplete. Nonetheless, many new insightshave arisen from comparisons between thosegenomes. First, unlike other types of vertebrate KZNFloci, most human and rodent KK genes are found in largetandem clusters containing up to 40 related genes. Althoughthe majority of human KK gene clusters are representedby related families in syntenically homologous regionsof the mouse genome, mouse and human KKclusters contain strikingly different numbers of genes. Indeed,sequence comparisons between homologous clusterspoint to active gene gain and loss since divergence ofthe primate and rodent lineages (Dehal et al. 2001; Shannonet al. 2003). Second, most mouse and human KK locicontain open reading frames capable of encoding fullyfunctional proteins (Dehal et a. 2001; E. Branscomb et al.,unpubl.), suggesting that the differential duplications haveyielded substantial numbers of novel lineage-specific proteins.Finally, KK clusters contain few pseudogenes evencompared to other types of familial gene clusters (see, e.g.,Gaudieri et al. 1999; Young and Trask 2002). Therefore,tandemly clustered ZNF genes appear to be subject to unusualselective pressures that actively favor the maintenanceof duplicated copies as functional genes.Lessons from a Differentially ExpandedCluster in Humans and MiceWe have recently described the structure and evolutionaryhistory of a pair of homologous KK gene clusterslocated in human chromosome 19q13.2 (Hsa19q13.2)and mouse chromosome 7 (Mmu7), respectively (Shannonet al. 2003). Twenty-one human genes and 10 mousegenes are found in these homologous gene clusters, althoughonly three sets of 1:1 orthologous pairs exist. Theclusters also include two cases in which a single gene inone species is related to multiple genes in the other(Fig.3). The relative orders of genes comprising thesefive sets of homologs are maintained in human andmouse, suggesting that this arrangement of genes waspresent in a common ancestor of primates and rodents.These findings are consistent with the idea that the differencesbetween mouse and human clusters reflect independenthistories of duplication and loss starting from abasic set of as few as five ancestral genes (Shannon et al.2003).