The Genom of Homo sapiens.pdf

EVOLUTION OF ZNF GENES 135Figure 4. Tissue-specific gene expression levels for a set ofclosely related primate-specific paralogs including ZNF224.(Based on data in Shannon et al. 2003). Darker colors representhigher expression levels based on northern blot data. Genes arelisted in groups of closest relatives. Asterisks indicate genes inwhich there is evidence of alternate splicing products.due to the evolution of the promoter and enhancer sequencesthat are duplicated along with the genes. Diversificationof mRNA splicing patterns also appears to playa role in paralog divergence; for instance, ZNF284 is alternativelyspliced to yield multiple mRNA isoforms,whereas ZNF225 produces a single mRNA species(Shannon et al. 2003). Alternative splicing has beenshown to generate KK protein isoforms with truncated oreven missing KRAB A and KRAB B domains (Odeberget al. 1998; Takashima et al. 2001), and may in somecases alter finger-domain structure through the use ofcryptic splice sites within the KZNF-encoding exon(Peng et al. 2002; Hennemann et al. 2003). Both types ofalternative splicing events have potential to produce KKprotein isoforms with altered functional properties.Recent History: Comparing KK Geneand Cluster Structure in Different RodentsTo gain additional insights into the evolutionary historyof the Hsa19q13.2/Mmu7 KK gene cluster, we analyzedsequence of the homologous cluster in rat. The ratand mouse gene clusters are structurally similar, butchanges have clearly taken place since the split of the tworodent lineages 15–20 million years ago (Mya) (Fig. 3).For instance, the rat genome carries only four genes relatedto human gene ZNF235, whereas mouse carries sixparalogous copies. In another example of change, the ratcluster includes two genes related to human ZNF45 andits single mouse homolog, Zfp94 (Fig. 5). This duplicationleaves only two of the five sets of predicted primate–rodenthomologs within this cluster as conserved1:1:1 orthologs in human, rat, and mouse. In both cases(the two extra mouse ZNF235 duplicates and the extra ratZNF45-like locus), the duplication events probably predatedthe mouse/rat split and lineage-specific losses followed,as indicated by a preliminary analysis of sequencedivergence in the KZNF exons and duplicated repetitiveelements (A. Hamilton, unpubl.). The case for a mousespecificloss of the ZNF45-like gene is supported by anisolated pseudo-KRAB A sequence that is similar to the(also disrupted) KRAB A in the rat locus and in the samerelative position (Fig. 3).Besides the change in gene number, structural changesin finger repeat arrangements have also taken place in orthologouspairs of mouse and rat genes. For example, ratZfp111 carries two additional fingers inserted in the middleof the KZNF domain as compared to the same gene inthe mouse (Fig. 5), a type of change that potentially generatesa major change in the DNA-binding “code” of theprotein (see below) (Fig. 6). Mouse and rat Zfp93,Zfp108, and Zfp112 differ by loss or gain of one finger perpair, but in each case these changes have occurred at theends of the arrays (not shown). The gene-structure differencesbetween mouse and rat genes suggest that KZNFarray deletions and duplications are relatively frequentevolutionary events.Implications of the Modes of Evolutionfor Zinc-finger ArraysKZNF domains can therefore be altered in severalways that can potentially affect their ability to bind to aspecific DNA target sequence. Although typically eachmotif acts as a discrete DNA-binding element (Pabo et al.2001), adjacent finger repeats cooperate in determiningtarget site recognition and binding stability (Isalan et al.Figure 5. Examples of changes in the number of finger motifs in sets of related ZNF orthologs and paralogs. (A) Human ZNF45 withmouse and rat copies of Zfp94 (GenBank accession XM_218438 for the predicted rat sequence) and the second rat locus (fingers representedin XM_218439) (B) Mouse Zfp111 and rat Zfp111 (also known as rkr2; NM_133323) showing the addition of two fingers inthe rat array. The Zfp235 diagram represents mouse Zfp235, rat Zfp235, and human ZNF235, all of which have the same array structure;this arrangement represents the closest human relative of the rodent Zfp111 protein. Rat and mouse Zfp111 copies carry duplicationsof a block of fingers relative to Zfp235 (designated in boxes).