Deciding when to flower is of crucial importance to plants; every season has advantages and disadvantages, and different plant species adopt different strategies. Elsewhere in this issue, Sibum Sung and Richard M. Amasino (Nature 427, 159–164; 2004) and Caroline Dean and colleagues (Nature 427, 164–167; 2004) investigate how such decisions are made at the molecular level. They uncover a mechanism that prevents the model plant Arabidopsis thaliana (pictured) from blooming until the coming of spring.

Plants take a variety of environmental factors into account when choosing when to flower, such as the length of the day, the plant's age and the requirement for an extended cold period (a process called vernalization). All of these factors work in part through the gene FLOWERING LOCUS C (FLC), whose protein product blocks flowering by repressing numerous genes required for flower development. During a prolonged cold spell, for example, the normally high levels of expression of FLC are lowered, remaining low even after warm weather returns.

Several genes are needed for vernalization: Dean and colleagues studied two of these, VRN1 and VRN2, whereas Sung and Amasino identified another, VIN3. All three encode proteins with counterparts in animals that either bind DNA directly, or change the structure of the chromatin into which DNA is packaged.

Following this lead, the two groups found that vernalization induces changes in histone proteins (components of chromatin) in the vicinity of the FLC gene — and that VRN1, VRN2 and VIN3 mediate these changes. Specifically, cold causes the loss of acetyl groups from particular lysine amino acids in histone H3. Such patterns of deacetylation mark genes that are permanently inactivated or silenced. The researchers found that whereas VIN3 is needed to deacetylate H3 during a cold snap, VRN1 and VRN2 are required afterwards, to maintain the silenced state.

Interestingly, these changes in histone acetylation are confined to a region of the FLC gene that was recently shown to contain a binding site for the FLOWERING LOCUS D (FLD) protein (Y. He et al. Science 302, 1751–1754; 2003). FLD is related to a component of the human histone deacetylase complex, and is also involved in promoting flowering by silencing FLC. Plants lacking FLD show both high levels of histone acetylation and a considerable reluctance to flower.

Silencing is an effective means of controlling long-term gene expression, as it persists even after cells divide. In animals, switching silencing on or off is a well-known way to control development. It seems that plants share this system, using it to preserve the memory of winter's passing.