How epigenetic memory is stored and inherited is an on-going debate (Ptashne). In so-called 'cis memory', epigenetic information is proposed to be physically stored in local chromatin states that are associated, for example, with DNA methylation or histone modifications. By contrast, in 'trans memory', epigenetic information is stored in the concentration of a diffusible factor such as a transcriptional repressor.

An elegant natural system in which to study this issue is the cold-induced epigenetic silencing at Arabidopsis thaliana FLOWERING LOCUS C (FLC). FLC mRNA and protein decrease quantitatively with increasing cold exposure, as FLC becomes repressed in an increasing number of cells over time (Angel et al.).

PHD-Polycomb repressive complex 2 components are required for the induction and maintenance of the repressed FLC state. These factors promote cold-induced changes in the chromatin state at FLC, specifically a localized increase in trimethylated Lys27 of histone 3 (H3K27me3) covering nucleosomes +3, +4 and +5. This peak then 'spreads' upon return to warmth, and high levels of H3K27me3 accumulate to cover the whole FLC locus, which is associated with long-term epigenetic silencing. Taken together, these results suggested a cis-based epigenetic memory of cold, with inheritance of the silenced state occurring independently at each locus. However, as in other model epigenetic systems, until recently it was difficult to exclude the existence of trans factor-based memory in which the FLC protein (a MADS box transcriptional repressor) and/or non-coding RNA produced at FLC feed into a trans-regulatory network.

epigenetic memory of FLC repression is physically located in the local chromatin environment

A study stimulated by theoretical considerations of epigenetic regulation provided the definitive proof for cis memory (Berry et al.). Two distinguishable fluorescent reporters of FLC expression present in the same cells were used to investigate the cis memory storage capability of FLC chromatin. After cold treatment, expression of the two reporters was independent; in other words, one of the FLC reporters could be repressed in the same cell in which the other reporter was active. Furthermore, this 'mixed' expression state was stably inherited through multiple cell divisions, indicating that the epigenetic memory of FLC repression is physically located in the local chromatin environment (Berry et al.). The debate is over for Polycomb-mediated silencing — local chromatin states can hold memory.