Chromatin opening and closing.
Nucleosomes package DNA into units with about 150 bp DNA and eight histones, two each of H2A, H2B, H3, and H4. The positively charged histone proteins bind tightly to the negatively charged phosphate groups of the DNA. This binding gets in the way of protein binding. RNA polymerase II is a big, complex protein and it has to bind to promoters to start up transcription. How can it get in? Do you have to get rid of nucleosomes to turn on a gene?
Several studies have shown that in the test tube, chromatin can’t be transcribed when it is in the usual tight structure. Experiments catching chromatin in different states in the cell have shown that when genes are being expressed, the nucleosome structure is different. Sometimes the nucleosome slides along to free the promoter where RNA polymerase II needs to bind to start the transcription of a gene. Sometimes, the only thing we know is that DNase I, added to isolated chromatin, cuts more in the promoter and/or enhancer region, implying some undefined looseness (DNase I hypersensitivity). It can be shown using ChIP (chromatin immunoprecipitation) that sites of DNase I hypersensitivity are sites where RNA polymerase complex and transcription factors bind to turn on genes.
Chromatin remodeling complexes such as Swi/Snf and SAGA can rearrange the chromatin components in vitro so that transcription factors can bind and transcription can begin, or if they contain histone deacetylases, they can compact and silence the chromatin. These complexes interact with non-coding RNAs in some systems. How do these complexes know where to act to open or close up the structure?
Recently, the histone code has been discovered. In long regions of chromatin called domains, when the genes are expressed, the histone tails are acetylated to relieve the tight binding to DNA. The acetylation of lysines in the histones is position specific (only lysine 9 of histone H3 is acetylated sometimes, for example). Methylation of some of the lysines can apparently signal for tight or loose packaging. Elements called silencers and enhancers can exist in different states that provide signals for domain open or closed structure. CTCF has a role in the transitions. Although the position in the nucleus is thought to be important (near nuclear pores for expressed genes, other parts of the periphery for silent genes), how that may interact with the histone code and the domains is unclear.
An interesting aspect of epigenetic regulation is imprinting: certain packaging signals are inherited from the father or the mother, from marks placed during spermatogenesis or oogenesis. The genes affected have mono-allelic expression from the maternal or the paternal allele only. Imprinted genes are associated with some human diseases.
X-chromosome inactivation is a major epigenetic event. A non-coding RNA from Xist gene coats the inactive X. Transcription is necessary for the pairing that occurs, which enables the cell to ‘count’ the X’s and inactivate all but one. DNA methylation of the inactive X happens later to lock in the silence.
Ideas for new experiments
Eleanor Cameron: examine patterns of acetylation and methylation of histones and DNA methylation at many loci during different stages of development. The early studies by Litt et al., 2001, may or may not present the general case. An aspect worth exploring is when/where oncogenes are expressed and how that’s controlled by domain chromatin structure. They are needed at some stages; if expressed later, cancer can result. The patterns in development could be compared with those in tumor cells.
Jessica Jerrit: Do a large-scale comparison of genes in heterochromatin that must be expressed (like the Drosophila gene rolled) to see if they were associated with specific proteins and/or interact with the miRNA systems. For X inactivation, it would be interesting to see if Xist interacts specifically with one or more proteins. How does it know where to go? Could protein help direct its binding?
Brad Kamitaki: SATB1 apparently regulates gene expression via nuclear matrix association. It would be useful to know if there are other proteins that have similar roles and are part of the nuclear matrix. Perhaps one could search for proteins that associate with SATB1 or other matrix components using Two Hybrid screens. Then cells with and without a newly discovered matrix component could be compared for gene expression using microarrays. I’d also like to uncover additional mechanisms for imprinting beyond DNA methylation. This could be discovered by looking at mutants unable to imprint properly, yet had effective Dnmt enzymes. So, I’d search among mutant strains for biallelic expression, then select strains that didn’t have defects in DNA methylation and study them.
Paul Fields: What targets the complexes that activate and inactivate the genes specifically to these genes and sequences? Mutate the protein sequence and also the DNA target sequence: see how much influence of sequence specificity there may be. Also, engineer complexes to target different DNA sequences with possible therapeutic effects.