Isolation of a novel retinoic acid-responsive gene by selection of genomic fragments derived from CpG-island-enriched DNA.

One of the primary goals in transcription factor research is the elucidation of the genetic networks controlled by a factor or by members of a family of closely related factors. The pleiotropic effects of retinoic acid (PA) in the developing and adult animal are mediated by ligand-inducible transcription factors (RA receptors [RARs] and retinoid X receptors [RXRs]) that belong to the superfamily of nuclear receptors. Regulatory regions of PA effector genes contain RAR and RXR binding sites (RAR elements [RAREs] and RXR elements [RXREs]) that generally consist of direct or everted repeats of the core half-site motif, (A/G)G(G/T)TCA. In order to identify novel genes regulated by RA, we devised a selection strategy based on the premise that regulatory regions of a large number of housekeeping and tissue-specific genes are embodied within CpG island DNA. In this method, referred to as CpG-selected and amplified binding, fragments derived from the CpG island fraction of the murine genome are selected by a gel mobility shift assay using in vitro-transcribed and -translated RXR-RAR. Multiple rounds of selection coupled with amplification of the fragments by PCR enabled us to clone a population of CG-rich fragments of which approximately one-fifth contained consensus RAREs or RXREs. Twelve genomic fragments containing novel response elements are described, and the transcription unit associated with one of them, NN-84AG, was characterized in detail. The mouse NN-84AG transcript is upregulated by RA in F9 embryonal carcinoma cells and is homologous to an expressed sequence tag (EST41159) derived from a human infant brain cDNA library. Cloning of the murine NN8-4AG genomic sequence places the RXRE in the proximity of the transcription initiation sites of the gene. Although sequence analysis indicates that the EST41159 gene product is novel, a region of amino acid identity with sequences of a yeast polypeptide of, as yet, unknown function and the Drosophila trithorax protein suggests the presence of an evolutionarily and functionally conserved domain. Our study demonstrates that transcription factor binding sites and corresponding regulated genes can be identified by selecting fragments derived from the CpG island fraction of the genome.