Is S Phase Important for Transcriptional Silencing?

See allHide authors and affiliations

Science  26 Jan 2001:
Vol. 291, Issue 5504, pp. 608-609
DOI: 10.1126/science.291.5504.608

Transcriptional silencing is a mysterious phenomenon that renders large sections of chromosomal DNA inert. This process alters the structure of chromatin, thus preventing the transcriptional machinery from reaching genes, which consequently remain switched off. Well-known examples of silencing include the inactivation of one X chromosome during development of the female human embryo, and position effect variegation in the fruit fly. The common denominator of these forms of silencing and many others (including silencing of the mating type loci HMRa and HMLα in the yeast Saccharomyces cerevisiae) is the formation and maintenance of a repressive chromatin structure called heterochromatin. For more than 15 years, researchers have known that silencing of chromatin depends on progression of cells through S phase and they have assumed that DNA replication must be the key event required before silencing can commence.

Heterochromatin is replicated during the last portion of S phase, whereas the rest of the DNA is replicated much earlier, suggesting that the way in which heterochromatin is replicated may be quite different. Classic experiments by Miller and Nasmyth in the early 1980s (1) suggested that cells must pass through S phase of the cell cycle before silencing can be established (that is, before unsilenced chromatin can be converted to the silenced state). Moreover, several genes involved in DNA replication also participate in silencing (see the table, below). Using clever genetic tricks in yeast, Kirchmaier and Rine (page 646) and Li et al. (page 650) now challenge the assumption that DNA replication during S phase is a prerequisite for chromatin silencing (2, 3).

View this table:

Silencing of yeast mating type-specificity genes in the HMRa and HMLα loci depends on sequences called silencers that flank these genes. These silencer sequences contain binding sites for proteins such as Rap1p, Abf1p, and proteins of the origin recognition complex (ORC) (4). Together with the Sir (silent information regulator) proteins, silencers and the proteins that bind them establish a heterochromatin-like structure in yeast. This structure is characterized by H3 and H4 histone proteins with fewer acetyl groups in their amino termini and by alterations in the topology of the DNA that is wound around the histones (57). Silencing of genes in the HMRa and HMLα loci is required for sexual competence in yeast.

Besides initiating DNA replication by binding to replication origins in the DNA, the ORC recruits Sir1p to the silencer sequence, which in turn recruits other Sir proteins (8). Artificially targeting a Sir1 fusion protein to a silencer engineered to lack an ORC binding site has been shown to result in the establishing of chromatin silencing (9). In the new studies, the two groups deleted the ORC binding site from the HMR-E silencer of the yeast HMRa locus and replaced it with either LexA or Gal4 DNA binding sites. This elegant genetic trick rendered the modified HMRa locus incapable of DNA replication. In a remarkable convergence of strategy, both groups then used the same genetic approach for their next trick—they flanked HMRa with recombination sites for FLP (or the related R) recombinase. This allowed the controlled excision of HMRa from the genome as a nonreplicating DNA ring (see the figure). After excision of the ring, the investigators expressed the Sir1 fusion protein (composed of Sir 1 and either the Gal4 or LexA DNA binding domain) and examined whether silencing had been established. Both groups went to heroic lengths to detect very small amounts of ring replication, yet they saw none. Efficacy of silencing was then evaluated by RNA blot analysis of the HMRa1 gene. In both studies, silencing was established as efficiently on the excised, nonreplicating HMRa ring as on the replicated chromosome. Thus, DNA that had not been replicated during the previous S phase could still be silenced. Importantly, the S-phase requirement for establishing silencing was still retained in this somewhat contrived system, even though DNA replication per se was not required.

Zipping up unreplicated chromatin.

Yeast cells arrested in the pre-S-phase state were engineered to contain the HMRa1 gene of the HMRa mating-type locus on an excised ring lacking a replication origin (bottom panel). The HMRa1 gene in its normal chromosomal location is shown in the upper panel. Triangles represent recombination sites in the excised ring; the arrow represents the unsilenced HMRa1 transcript. Expression of a Sir1 fusion protein (zipper tab), which binds to HMRa at the E silencer (SIL, orange dot) sequence, established silencing of the HMRa1 gene after release of the cells into S phase. The HMRa1 DNA in its native chromosomal location is replicated through the action of an adjacent origin of replication (ORI, blue square), whereas the excised DNA circle cannot replicate because it does not have an ORI. Subsequently, silenced chromatin (closed zipper) was established on both the replicated DNA and unreplicated ring, resulting in equivalent amounts of HMRa1 gene silencing in both native and ring DNA.

It could be argued that the silenced chromatin established in these experiments was not authentic because silencing was ORC independent and required artificial recruitment of a Sir1 fusion protein to a single modified silencer. This argument, however, is largely refuted by the experiments of Li et al. (3), who found that HMRa rings had physical characteristics indistinguishable from those of native silent chromatin. These characteristics include hypoacetylation of histones H3 and H4, excessive negative supercoiling, and dependence on the SIR2, SIR3, and SIR4 genes. The silent chromatin observed appears, therefore, to be authentic. It remains formally possible that some aspect of the ORC-Sir1p interaction might be affected by replication, given that the interaction was by necessity not evaluated in these experiments.

It is important to emphasize that establishment, maintenance, and inheritance of silencing are three different processes. The current reports do not address the possible importance of DNA replication in the maintenance or heritability of silent chromatin. There is still compelling evidence that DNA replication could be involved in the maintenance or heritability of silencing at the HMRa and HMLα loci and probably at other loci, too; it may also be required for silencing genes in telomeres and in ribosomal DNA that lack clearly defined silencer elements. Mutations in subunits of yeast chromatin assembly factor-I (CAF-I) reduce the stable inheritance of silencing at telomeres and impair the maintenance of silencing at HMRa, HMLα, and other loci (1012). The connection with DNA replication is that CAF-I deposits newly synthesized histones onto newly replicated DNA. Furthermore, mutations in proliferating cell nuclear antigen (PCNA) that disrupt the association of this replication protein with CAF-I also impair the inheritance of silencing (13). The implication is that DNA replication may well prove crucial for “persistence” of the silent state.

This returns us to the original question: What makes S phase important for transcriptional silencing? Strictly speaking, the phase of the cell cycle required for silencing to be established is somewhere between early S phase (the point where hydroxyurea blocks cell cycle progression) and mitosis (where nocodazole has its inhibitory effect) (1). This suggests that the S-phase requirement for silencing may in reality be a point somewhere in late S phase, in G2, or possibly even in early mitosis.

The cell's DNA replication machinery is fully capable of replicating conventional chromatin or previously silenced chromatin. However, it may be challenged by the drastic alterations in chromatin structure that accompany the establishment of silencing, during which the chromatin changes from the relatively open unsilenced state to the closed heterochromatic state. To prevent such molecular conflicts, the silencing machinery may be activated by an intracellular signal sent when DNA replication has been completed. In fact, Sir proteins can move to new locations inside the cell in response to various forms of DNA damage (14, 15), indicating that they can respond to signals sent by changes in DNA state. Needless to say, there are many other possible events that could control the establishment of silencing through the generation of specific cell cycle signals or changes in nuclear structure. But what was once considered the most likely event—replication of the DNA prior to silencing—does not appear to be among them.


Navigate This Article