Our research focuses on the role of higher-order chromatin structure in regulating gene expression. A topic of special interest is epigenetic memory, both how it is established and the determinants of its stability. The central hypothesis to be tested is that defects in epigenetic memory contribute to aging, cancer, and age-related diseases. To test that hypothesis, we are pursuing two areas of inquiry. First, we are developing techniques to assess the type and extent of chromatin remodeling that accompanies developmental programs, terminal differentiation, or senescence. Second, we perform structure-function studies to understand the roles that specific chromatin regulatory gene products may play in the establishment or maintenance of epigenetic memory.

Genome Sampling to Map Higher-Order Chromatin Domains and Regions of Chromatin Remodeling
Russanova, Howard
Several examples in the literature provide direct or indirect evidence for chromatin remodeling in conjunction with differentiation, cancer, or aging. Most such examples derive from serendipitous observations; thus, we have only a limited understanding as to the extent and types of chromatin changes that may occur in these contexts. Differential display is well established as a systematic means to screen for alterations in mRNA levels. We have combined chromatin immunoprecipitation (ChIP) with a modified form of differential display in order to screen for regions of unusual chromatin structure or domains involved in chromatin remodeling. To date, most of our ChIP experiments have involved chromatin fractionation according to histone acetylation level. Elevated levels of acetylation are typically associated with active gene transcription, whereas underacetylation correlates with repression and silencing. The human genome is organized into alternating gene-rich and gene-poor regions. The observation that markedly underacetylated loci identified in the screens reside uniformly within gene-poor regions is an important validation of the ChIP genome sampling approach. By contrast, highly acetylated loci lie predominantly within gene-rich regions. The differences are highly significant and would not be observed if screening yielded loci at random with respect to higher-order chromatin domains.

The HL-60 promyelocytic cell line has served as a useful model system in which to search for domains of chromatin remodeling. HL-60 cells differentiate into macrophage-like cells in response to treatment with phorbol esters. A screen of several thousand loci revealed several in which terminal differentiation along the macrophage lineage is accompanied by elevated histone acetylation. One such locus, located on chromosome 2 at a position 40 kilobases upstream from the EML4 gene locus, exhibits an average six-fold increase in acetylation. The success of genome sampling in the HL-60 system provides a solid basis for proceeding to searches for chromatin remodeling in the contexts of development, cancer, and aging.

Histone H3 Methylation Marks the Murine H19-Igf2-Imprinted Region
Tchernov; in collaboration with Stewart, Pfeifer
The murine H19-Igf2 imprinted region is characterized by maternal expression of the H19 locus versus paternal expression of the Igf2 gene. A differentially methylated region (DMR) is located upstream from the H19 locus. In maternal cells, the DMR is unmethylated, binds to the transcription factor CTCF, and functions as an insulator to prevent activation of the Igf2 promoter by enhancers located downstream from H19. Conversely, in paternal cells, the DMR is methylated and repressed with respect to transcriptional and insulator activities. Recent evidence indicates that histone H3 methylation at lysine position 9 (K9) can precede DNA methylation as a mark for transcriptional silencing. Such is the case in diverse systems ranging from DNA methylation in Neurospora and Arabidopsis to X-chromosome silencing in mammalian embryonic stem cells. Experiments by members of our group and collaborators revealed a high level of histone H3-K9 methylation near the H19-associated DMR and promoter region in paternal but not maternal cells. By contrast, we detected H3-K9 methylation at control regions for the Igf2 locus in paternal cells only. Histone H3 methylation at lysine position 4 (K4) is typically associated with transcriptional activation. We found highly localized H3-K4 methylation in the Igf2 promoter in paternal cells, but not in maternal cells. Each of these results is consistent with a role of histone H3 K4/K9 methylation in control of imprinted genes. Of further importance is the fact that the characterization of selective histone methylation at the Igf2 locus provides a reference, or positive control, for genome sampling to detect changes in H3 methylation in the contexts of development and age-related processes.

Association of Hox-a10 and Sirtuin2
Bae
As noted above, histone deacetylation is typically associated with transcriptional silencing. Numerous transcriptional repressors recruit RPD3-family deacetylases. Previous work by our group and others revealed that members of this family, including HDAC1, HDAC2, and HDAC3, function within multiprotein complexes to mediate repression. The role of RPD3 deacetylases in epigenetic silencing, as opposed to transcriptional repression, appears to be complex. In some cases, inhibitors of RPD3 deacetylases such as sodium butyrate and trichostatin A can disrupt silencing. On the other hand, deletion of RPD3 and RPD3-like deacetylases in budding yeast enhances rather than impairs silencing. The NAD-dependent deacetylase SIR2 is essential for silencing in budding yeast, and SIR2-like proteins constitute a highly conserved family ranging from yeast to mammals. To date, however, no evidence has been reported that Sirtuins, the mammalian SIR2 homologs, function in transcriptional regulation. Experiments performed by our group, using the yeast two-hybrid system, co-immunoprecipitation of in vitro translated products, and co-immunoprecipitation from mammalian cell extracts, revealed that Sirtuin2 interacts with the homeodomain protein Hox-a10. Further, over-expression of Sirtuin2 counteracts trans-criptional activation by Hox-a10. These results establish a link between mammalian SIR2 homologs and transcriptional control. In addition, given that Hox-a10 plays a role in genito-urinary tract formation, the results suggest the possibility that Sirtuin2 may modulate the establishment or maintenance of repressive chromatin domains during development.