A consensus DNA recognition motif for two KDWK transcription factors identifies flexible-length, CpG-methylation sensitive cognate binding sites in the majority of human promoters.

Parvovirus initiation factor (PIF), identified in HeLa cells as a host factor essential for parvoviral DNA replication, is a ubiquitous heterodimeric cellular transcription factor. This protein complex was simultaneously identified as glucocorticoid modulatory element binding protein (GMEB) by its ability to bind to the glucocorticoid modulating element (GME) upstream of the tyrosine transaminase promoter. Here, we show that the two PIF/GMEB subunits form site-specific DNA-binding heterodimers when co-expressed from recombinant baculoviruses and homodimers when expressed separately. Degenerate oligonucleotide selection experiments, combined with analysis of dissociation rates, established that the three complexes bind to flexibly spaced tetranucleotide half-sites that conform to the consensus ACGPy N(1-9) PuCGPy, with an optimum of N=6. Binding of all three complexes is extremely sensitive to methylation of the cytosine residues in the invariant CpG half-site core, suggesting a means by which PIF/GMEB binding could be regulated in vivo. Because CpG dinucleotides are suppressed in eukaryotic genomes, such binding sites would be expected to be very rare. However, analysis of 100 human promoters showed that over half of them contained at least one site conforming to the consensus, a significant deviation from the expected random distribution. In many of these, the binding site is within 100 nucleotides of the transcriptional start site, indicating that PIF/GMEB may be involved in regulation of these genes. Oligonucleotides corresponding to five of these sequences, chosen to represent the range of half-site separations identified by the consensus, were tested for PIF/GMEB binding by mobility shift assay. All five probes bound the heterodimer efficiently and, in each case, binding was completely abrogated by 5-methylation of the C residues in the CpGs of the putative half-sites. Copyright 2001 Academic Press.