The spreading influence of chromatin modification

Specific post-translational histone modifications correlate with distinct patterns of gene activity. A new study of the mouse Dntt promoter shows that changes in gene activity are associated with local changes in histone modification that then spread outward in both directions, locking in the altered pattern of transcription.

The mouse genome consists of about 3 billion bp of DNA divided into 20 chromosomes. Stretched out, this DNA would measure nearly a meter in length, but it all fits into a spherical nucleus only several microns in diameter. It is a challenge to understand how the genome can be wrapped so tightly yet remain accessible for precisely regulated gene expression. Meeting this challenge will require a detailed appreciation of the dynamics of chromatin structure. On page 502 of this issue, a new study by Ruey-Chyi Su and colleagues¹ illustrates how localized changes in chromatin structure influence the regulated expression of a specific gene during mouse lymphocyte development.

Chromatin represses transcription

The most basic element of chromatin structure, the nucleosome, consists of an octamer of histones forming a central core and 146 bp of DNA wrapped nearly twice around it². Successive nucleosomes are separated by 20–60 bp of DNA, often associated with a linker histone called H1. This 11-nM nucleosome fiber is then wrapped into a 30-nM solenoidal structure. Our understanding of this, and subsequent higher orders of chromatin structure, is rudimentary.

Chromatin structure is a substantial barrier to the transcription machinery. Basal transcription factors, including the essential TFIID complex and RNA polymerase II, cannot recognize promoter sequences packaged in a nucleosome array³. Two processes cooperate to clear the way for gene activation⁴. The first, known as chromatin remodeling, slides nucleosomes out of the way to expose promoter elements. The second consists of the covalent modification of histones through the placement or removal of acetyl, methyl, ADP-ribosyl or phosphate groups on or from specific amino acids. Some of these histone modifications correlate with regulated gene activation or constitutive inactivity (the 'histone code')⁵. The study by Su et al.¹ characterizes the changes in histone modifications that occur when an active gene undergoes developmentally regulated inactivation.

Dntt encodes an enzyme involved in the diversification of antigen receptor genes in developing lymphocytes⁶. It is active early in B- and T-cell development but is inactivated as lymphocytes mature. Previous work identified the Dntt promoter and defined an array of transcription factors regulating its activity. Su et al. compared two model systems in which an artificial stimulus promotes the differentiation of immature T cells in vitro, resulting in inactivation of Dntt expression.

They found that nucleosomes containing histone H3 modified by acetylation of lysine at position 9 (acetylated H3-Lys9) were enriched throughout a domain extending at least 10 kb 5′ and 3′ of the promoter in cultured immature thymocytes and in a thymic lymphoma cell line, VL3-3M2. After inducing differentiation, they observed the highly localized loss of this modification over the promoter region itself, associated temporally with inactivation of Dntt transcription. In primary thymocytes, this was followed over the ensuing 18 h by the gradual spread of this region of deacetylated chromatin in either direction away from the promoter. During this interval, they detected loss of another activation-associated modification, acetylation of H3 at Lys4, followed by the acquisition of methylation at H3-Lys9, again spreading from the promoter outwards. After 6–8 h of induction, the inactivation of transcription became irreversible. Notably, VL3-3M2 cells showed the same focal loss of acetylated H3-Lys9 at the promoter, but this modification did not spread outward and transcriptional inactivation remained reversible. Thus, Su et al.¹ found that localized changes in histone modifications are associated with transient changes in gene activity, whereas permanent changes are associated with regional spread of a domain of chromatin modification.

Who's calling the shots?

Some transcription factors have, or recruit proteins that have, histone modification and remodeling activities (Fig. 1). Presumably, gene activation requires at least one such factor that can bind its recognition sequence within 'inactive' chromatin and recruit other factors that collaborate in altering local chromatin structure. These altered regions of chromatin would then expose binding sites for other factors, including the basal transcription machinery³. Histone modifications may also be necessary to allow RNA polymerase to transit across nucleosomal DNA sequences⁷. When cells progress through development, the levels of various transcription factors might change and key target sites might be vacated. Alternatively, as might be the case with Dntt, negative transcription factors might compete with and displace activating factors, altering associated chromatin remodeling and modifying activities⁸ (Fig. 1). The tendency of such alterations to occur locally and then spread globally is poorly understood. Perhaps local histone modification itself recruits additional modifying enzymes, leading to spread along the DNA. This idea, which gains support from studies in yeast and fruit flies^9,10, implies that boundary elements limit the spread of such modified chromatin domains².

Figure 1: Transcription factors recruit chromatin modification enzymes, which in turn regulate chromatin structure in the vicinity of the promoter.

Dashed lines indicate spread of chromatin changes outward from the promoter region. In this model, changes are reversible until nucleosomes are modified by histone methylation. Ac, acetylated; Me, methylated.

The study by Su et al.¹ also addresses the difference between transformed and primary immature T cells. Dntt inactivation was fully reversible in VL3-3M2 but not primary thymocytes, a difference that correlated with a failure of spreading in VL3-3M2 cells. Primary thymocytes did not undergo cell division in this culture system whereas transformed cells did. Perhaps cell cycle regulation or chromatin reassembly after replication is involved in the spread and reversibility of chromatin modification patterns. Alternatively, transformed cells might have different levels of the relevant modifying enzymes, perhaps as a consequence of the transformed phenotype. Also, whereas acetylation can be reversed by various histone deacetylases, there are no known histone demethylases. Therefore, once a genomic region is methylated, modified nucleosomes must be replaced rather than altered to remove this epigenetic mark. Further work will be required to understand the mechanisms responsible for the spread of local histone modifications and the impact of these modifications on chromatin structure and transcriptional regulation.