A comprehensive catalog of human KRAB-associated zinc finger genes: insights into the evolutionary history of a large family of transcriptional repressors

A profile hidden Markov model search of the human genome — using models built from KRAB, SCAN, and Krüppel-type zinc finger motif sequences and refined by manual gene-model curation — identifies 423 human KRAB-ZNF protein-coding loci, of which 89 were not previously described in public gene databases. Alternative splicing at these loci generates 818 distinct transcripts that together encode at least 742 structurally distinct proteins.

Human KRAB-ZNF proteins carry between 2 and 40 tandem C2H2 zinc fingers, with a median of 12 fingers per protein. SCAN-containing members carry fewer — a median of 5.5 fingers for SCAN-ZNF loci and 8 for SCAN-KRAB-A loci — consistent with recognition of shorter DNA target sequences, which fits the fact that many SCAN-ZNF proteins bind DNA as homodimers.

A subset of KRAB-ZNF genes encode a KRAB-B variant designated KRAB-BL, whose exons are 30 base pairs larger than canonical KRAB-B exons due to a 3′ extension. Genes carrying the KRAB-B, KRAB-BL, KRAB-b, or KRAB-C modulator motifs each fall into their own distinct clade of the KRAB-A phylogenetic tree, indicating each modulator type expanded within a specific subfamily, whereas SCAN-domain-encoding genes are scattered across multiple clades rather than confined to one — a pattern consistent with an ancient SCAN-KRAB combination subject to recurrent domain loss, or with independent origins of the combination more than once.

The human genome contains 341 KRAB-ZNF pseudogene sequences in addition to the 423 protein-coding loci: 227 gene fragments, 39 full-length pseudogenes, and 75 processed (retrotransposed) pseudogenes. Three of the processed pseudogenes — LLNL1071, LLNL1040, and LLNL973 — retain an intact open reading frame and remain potentially capable of encoding a functional protein, while every other processed copy is degraded and nonfunctional.

Most human KRAB-ZNF genes are physically clustered: 384 of the 423 protein-coding loci reside within 65 discrete gene clusters, concentrated in 25 major clusters with the largest and most numerous located on chromosome 19; the remaining 39 loci occur as isolated singletons. Tandem in situ duplication, together with segmental and physically distributed duplication events, accounts for the genome-wide expansion and clustering pattern of the family.

Of the 423 human KRAB-ZNF loci, 399 have a detectable chimpanzee ortholog, 112 are conserved as 1:1 orthologs in mouse, and 103 are conserved as unambiguous 1:1 reciprocal orthologs across human, chimpanzee, dog, and mouse. At least 136 human KRAB-ZNF genes — nearly one-third of the family — are primate-specific, having arisen since the divergence of Old World and New World monkeys, and this primate-specific subset includes many very recently duplicated genes.

Seven KRAB-ZNF genes and one pseudogene distributed across human chromosomes 4, 8, 11, and 12 (the ZNF705A paralogs) share greater than 96% nucleotide identity over duplicated segments longer than 70 kb, placing this dispersed duplication event within roughly the last 15 million years. Two adjacent paralog pairs on chromosome 8 — ZNF705C with LLNL1103 (98.5% identity) and LLNL1035 with ZNF705B (>99% identity) — arose from tandem duplications recent enough that the chimpanzee genome retains only a single corresponding gene at each of the two loci.

Among 211 KRAB-ZNF genes profiled by microarray, hierarchical clustering of expression pattern groups 51 genes together with highest expression in lymphoid tissue, and a separate 50 genes into a cluster showing the converse pattern of reduced expression in immune-related tissues; genes within either expression cluster are generally drawn from different chromosomal regions.

KRAB-ZNF genes located within the same physical gene cluster are not generally coexpressed: pairwise comparisons of expression pattern across the 211 profiled genes have a median Pearson's correlation coefficient of only 0.02, indicating no global coregulation among clustered paralogs. Even very recent, highly similar paralogs can diverge sharply in expression — ZNF585A and ZNF585B, which share 85% noncoding sequence identity, display markedly different tissue expression patterns — showing that tandem clustering in this gene family reflects shared duplicative ancestry rather than shared transcriptional regulation.

The KRAB-ZNF cluster at human 5q35.3 contains six tandem genes arranged in the same order as their orthologs in chimpanzee, dog, and mouse, so some KRAB-ZNF clusters have stayed intact across mammalian evolution. Most conserved KRAB-ZNF clusters, however, show lineage-specific expansion, with different gene counts in each species and relatively few orthologous groups.

Alternative splicing of human KRAB-ZNF genes frequently produces proteins lacking one of the two functional halves — 66 loci yield transcripts encoding only the effector domains (KRAB and/or SCAN) and 79 yield zinc-finger-only transcripts. Because inclusion of KRAB-B enhances KRAB-A repressor activity and inclusion of SCAN enables dimerization, these isoforms predict functionally distinct regulatory proteins from a single KRAB-ZNF locus.