Characterization of Mammalian Cellular Memory.

Cellular differentiation is accompanied by the coordinated activation and silencing of specific subsets of genes and associated alterations in chromatin structure. This process, which is under the control of extrinsic and intrinsic cellular signals, results in the establishment of lineage-specific domain structures that must be maintained during dynamic nuclear events, such as cell division. The propagation of specific patterns of gene expression is termed “cellular memory”. Our goal is to identify the elements that are involved in the establishment and propagation of transcription states and to determine the molecular basis of memory establishment in mammals. To this end, K562 cells were transfected with a γ-globin promoter-GFP-metallothionein response element (MRE) cassette (GGM cassettes) and, using a procedure to derive stably transfected cells in the absence of selection, multiple stable transformants harboring a single copy of the cassette were obtained. In most clones, zinc (Zn) induction resulted in transition of the transgene from the silent to the active state, with the maintenance of the active state dependent on continuous exposure to Zn. However, at one genomic site (clone 6177), a high level of GFP expression was maintained even after Zn removal. Characterization of epigenetic marks of this transgene, including CpG methylation, histone modifications and nuclear localization, at various stages of GFP expression may provide insight into the molecular mechanisms of mammalian cellular memory. In initial experiments, Na-bisulfite conversion/sequencing revealed that, prior to the establishment of memory, up to 8% of CpGs in the GFP gene are methylated, and after memory establishment, no CpG methylation is detectable. However, pretreatment of 6177 cells with a DNA methylation inhibitor, 5-dAzaC for 48hs before Zn induction resulted in no change in the memory establishment program, suggestiong that reduction of CpG methylation is not a cause of memory establishment in 6177 cells. We also have examined the relationship between cellular memory and histone modifications. Comparison of the histone modification status of the active GFP gene in cells before and after memory establishment revealed that H4 K20 dimethylation is present in the both the inactive and active GFP gene prior to memory establishment, but is significantly reduced upon memory establishment. Interestingly, this mark is known to be associated with silenced regions of euchromatin and the chromocenter, but not with active genes, in Drosophila. Thus, these results have revealed a unique nature of histone modification at the 6177 site: a marker of silenced chromatin co-exists with active transcription prior to the establishment of cellular memory. This co-existence may be the mechanism by which the establishment of memory of the active state is prevented at stage II. Currently, we are cloning the 6177 integration site to identify cellular memory elements, the function of which will be further analyzed at defined genomic sites.