Acetylation curtails nucleosome binding, not stable nucleosome remodeling, by FoxO1.

Transcriptional activity of FoxO factors is controlled through the actions of multiple growth factors signaling through protein kinase B, whereby phosphorylation of FoxO factors inhibits FoxO-mediated transactivation by promoting nuclear export. Phosphorylation of FoxO factors is enhanced by p300-mediated acetylation, which decreases their affinity for DNA. The negative effect of acetylation on FoxO DNA binding, together with nuclear FoxO mobility, is eliminated by over-expression of the de-acetylase Sirt1, suggesting that acetylation mobilizes FoxO factors in chromatin for inducible gene expression. Here, we show that acetylation significantly curtails the affinity of FoxO1 for its binding sites in nucleosomal DNA but has no effect on either stable nucleosome binding or remodeling by this factor. We suggest that, while acetylation provides a first, essential step toward mobilizing FoxO factors for inducible gene repression, additional mechanisms exist for overcoming their inherent capacity to stably bind and remodel nuclear chromatin.

An early developmental transcription factor complex that is more stable on nucleosome core particles than on free DNA.

In vivo footprinting studies have shown that transcription factor binding sites for HNF3 and GATA-4 are occupied on the albumin gene enhancer in embryonic endoderm, prior to the developmental activation of liver gene transcription. We have investigated how these factors can stably occupy silent chromatin. Remarkably, we find that HNF3, but not GATA-4 or a GAL4 control protein, binds far more stably to nucleosome core particles than to free DNA. In the presence of HNF3, GATA-4 binds stably to an HNF3-positioned nucleosome. Histone acetylation does not affect HNF3 binding. This is evidence for stable nucleosome binding by a transcription factor and shows that a winged helix protein is sufficient to initiate the assembly of an enhancer complex on nonacetylated nucleosomes.

Binding of the winged-helix transcription factor HNF3 to a linker histone site on the nucleosome.

The transcription factor HNF3 and linker histones H1 and H5 possess winged-helix DNA-binding domains, yet HNF3 and other fork head-related proteins activate genes during development whereas linker histones compact DNA in chromatin and repress gene expression. We compared how the two classes of factors interact with chromatin templates and found that HNF3 binds DNA at the side of nucleosome cores, similarly to what has been reported for linker histone. A nucleosome structural binding site for HNF3 is occupied at the albumin transcriptional enhancer in active and potentially active chromatin, but not in inactive chromatin in vivo. While wild-type HNF3 protein does not compact DNA extending from the nucleosome, as does linker histone, site-directed mutants of HNF3 can compact nucleosomal DNA if they contain basic amino acids at positions previously shown to be essential for nucleosomal DNA compaction by linker histones. The results illustrate how transcription factors can possess special nucleosome-binding activities that are not predicted from studies of factor interactions with free DNA.

Developmental regulation of alpha-fetoprotein expression in intestinal epithelial cells of transgenic mice.

The alpha-fetoprotein (AFP) gene is transcribed in most epithelial cells lining the fetal mouse small intestine, but transcription persists in only a subset of enteroendocrine cells representing less than 1% of the total intestinal epithelial cells in the adult. The decrease in AFP expression after birth is mediated in part by a repressor element lying between -838 and -250 bp of the AFP gene. Deletion of this element from AFP minigene constructs results in high-level minigene expression in the intestines of adult transgenic mice. Although high levels of AFP minigene RNA are expressed, the fetal pattern of expression is not maintained upon deletion of the repressor element. Instead, the number of cells in which the minigene is expressed increases from less than 1% to approximately 10% of the epithelial cells in the adult small intestine, and includes the majority of the goblet cells in addition to the enteroendocrine cells. In contrast, the pattern of AFP minigene expression in the enterocytes is unaffected by deletion of the repressor element and continues to decrease in the neonate. These studies indicate that the identified AFP repressor is active specifically in goblet cells. The decrease in AFP expression in the enterocytes may be mediated by a separate cis-acting element that is contained in the AFP minigene construct. Alternatively, it is possible that mature enterocytes lack some of the positive factors required for initiation and maintenance of minigene transcription in the absence of the identified negative element.

The zonal expression of alpha-fetoprotein transgenes in the livers of adult mice.

The developmental regulation of the alpha-fetoprotein (AFP) gene in liver results in high-level expression in the fetus, followed by dramatic transcriptional repression after birth. We have examined the mouse AFP gene for transcriptional control sequences that may be involved in its postnatal repression in liver. We showed previously that removal of a DNA region between positions -250 base pairs (bp) and -838 bp of the AFP gene resulted in the persistence of expression of an AFP minigene in the postpartum liver of transgenic mice (Vacher and Tilghman, Science 250:1732-1735, 1990). This study examines the distribution of these transgene transcripts in liver using in situ hybridization. We show that there is a zonal distribution of minigene transcripts in the adult livers of these animals. Hepatocytes surrounding the central veins express high levels of minigene transcripts, while hepatocytes in the intermediate and portal areas contain few, if any, transcripts. Quantitative RNAse protection analysis shows a decrease in transgene RNA levels after birth, consistent with repression in all but a small subset of hepatocytes. These results indicate that repression in the pericentral hepatocytes is dependent upon the presence of a cis-acting, negative-regulatory domain, which is located between the enhancers and the proximal promoter of the AFP gene. In contrast, this domain is not essential for complete repression of AFP transgenes in the intermediate zone and periportal hepatocytes.