transcriptional silencing: replication redux

4
R816 Dispatch Transcriptional silencing: Replication redux David Shore Recent studies indicate that, contrary to long-held belief, DNA replication does not have a direct role in transcriptional silencing, but progression through S phase of the cell cycle is nevertheless required for the establishment of silent chromatin. Address: Department of Molecular Biology, Sciences II, University of Geneva, 30 quai Ernest-Ansermet, CH-1211, Geneva 4, Switzerland. Current Biology 2001, 11:R816–R819 0960-9822/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. Transcriptional silencing in the budding yeast Saccharomyces cerevisiae, first characterized at the silent HM mating-type loci and later adjacent to telomeric TG-repeat tracts, has served for many years as a model system for understanding the establishment and inheritance of large domains of repressed chromatin — heterochromatin — in more complex eukaryotes. Silent mating-type loci are repressed through the action of flanking cis-acting elements, called the E and I silencers (see Figure 1). Almost since their discovery, all four silencers have been known to act as replication origins when cloned on plasmids in yeast. This striking finding led naturally to the notion that the establishment of silencing would be intimately linked to the process of DNA replication, an idea that was first tested experimentally in a series of experiments reported by Miller and Nasmyth [1], in what has since become a classic reference in the field. The ‘Miller and Nasmyth’ experiment took advantage of a temperature-sensitive allele of SIR3 — which encodes an essential structural component of silent chromatin — to turn on and off repression. They found that the loss of silencing upon temperature upshift in a sir3-ts strain occurred rapidly, and in the absence of cell-cycle progres- sion, whereas the re-establishment of repression following a temperature downshift required progression through the cell cycle, and in particular passage through S phase. Miller and Nasmyth [1] thus proposed that DNA replica- tion is required in some way to re-set the silent chromatin state, and suggested that the silencer replication origins would play a direct role in this process. The subsequent identification of the origin recognition complex (ORC) — the six-protein complex that binds to all known replication origins — and the direct demonstration of its role in silencing strengthened the notion that a replication fork initiating from silencer elements would turn out to play a critical role in the establishment of a repressed chromatin structure [2–4]. Subsequent work on the silencers and their replication origins has steadily eroded the idea that DNA replication plays a direct role in establishing silencing, and two recent reports [5,6] have now brought about its apparent demise. Ironically, however, concurrent genetic and biochemical studies have continued to point to numerous connections between silencing and replication. Furthermore, although the elegant experiments to be discussed below clearly show that silent chromatin can be set up in the absence of replication fork passage, they still underscore the impor- tance of S phase progression. So it is clear that many inter- esting questions remain. A closer look at the silencer—replication connection Troubles with the DNA replication model for the estab- lishment of silencing first emerged when it was discovered that neither of the silencer elements at HML are actually functional replication origins in their chromosomal context, despite the fact that they behave as active origins on plasmids [7]. This observation made it hard to imagine that replication initiation from the silencers themselves is essential for the establishment of repression, a notion that was further supported by studies showing that the DNA replication initiation and silencing functions of ORC could be genetically separated [8,9]. Additional work from the Fox laboratory [10] uncovered multiple initiator elements at the HMR-E silencer, some of which even appear to be antagonistic to silencing. Nonetheless, these observations still left open the possibility that replication through a silent region — whether or not it initiates from the silencer elements themselves — would be required to re-assemble silent chromatin at a previously derepressed locus. The experimental tools required to address this issue emerged from studies designed to ask what the precise role of the silencer elements is in the establishment and maintenance of the repressed state. Protein targeting studies, initially using the Gal4p DNA-binding domain (GBD), demonstrated that the silencer elements could, to at least some extent, be bypassed by tethering Sir proteins to specific sites in chromatin [11]. This mode of establish- ing silent chromatin in a previously active region, referred to as ‘targeted silencing’, was first demonstrated with a GBD–Sir1 hybrid, but it works equally well with other GBD–Sir hybrids provided they are expressed at appropri- ate levels in the cell [12]. Using a similar targeting approach, Rine and colleagues [13] showed that regulated expression of a LexA–Sir1 hybrid could promote silencing at an HMR locus whose E-silencer replication function was missing, thus formally disproving the idea that a linked origin activity is required to establish silent chromatin.

Upload: david-shore

Post on 18-Sep-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Transcriptional silencing: Replication redux

R816 Dispatch

Transcriptional silencing: Replication reduxDavid Shore

Recent studies indicate that, contrary to long-heldbelief, DNA replication does not have a direct role intranscriptional silencing, but progression through Sphase of the cell cycle is nevertheless required for theestablishment of silent chromatin.

Address: Department of Molecular Biology, Sciences II, University of Geneva, 30 quai Ernest-Ansermet, CH-1211, Geneva 4, Switzerland.

Current Biology 2001, 11:R816–R819

0960-9822/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved.

Transcriptional silencing in the budding yeastSaccharomyces cerevisiae, first characterized at the silent HMmating-type loci and later adjacent to telomeric TG-repeattracts, has served for many years as a model system forunderstanding the establishment and inheritance of largedomains of repressed chromatin — heterochromatin — inmore complex eukaryotes. Silent mating-type loci arerepressed through the action of flanking cis-actingelements, called the E and I silencers (see Figure 1).Almost since their discovery, all four silencers have beenknown to act as replication origins when cloned onplasmids in yeast. This striking finding led naturally to thenotion that the establishment of silencing would beintimately linked to the process of DNA replication, anidea that was first tested experimentally in a series ofexperiments reported by Miller and Nasmyth [1], in whathas since become a classic reference in the field.

The ‘Miller and Nasmyth’ experiment took advantage of atemperature-sensitive allele of SIR3 — which encodes anessential structural component of silent chromatin — toturn on and off repression. They found that the loss ofsilencing upon temperature upshift in a sir3-ts strainoccurred rapidly, and in the absence of cell-cycle progres-sion, whereas the re-establishment of repression followinga temperature downshift required progression through thecell cycle, and in particular passage through S phase.Miller and Nasmyth [1] thus proposed that DNA replica-tion is required in some way to re-set the silent chromatinstate, and suggested that the silencer replication originswould play a direct role in this process. The subsequentidentification of the origin recognition complex (ORC) —the six-protein complex that binds to all known replicationorigins — and the direct demonstration of its role insilencing strengthened the notion that a replication forkinitiating from silencer elements would turn out to play acritical role in the establishment of a repressed chromatinstructure [2–4].

Subsequent work on the silencers and their replicationorigins has steadily eroded the idea that DNA replicationplays a direct role in establishing silencing, and two recentreports [5,6] have now brought about its apparent demise.Ironically, however, concurrent genetic and biochemicalstudies have continued to point to numerous connectionsbetween silencing and replication. Furthermore, althoughthe elegant experiments to be discussed below clearlyshow that silent chromatin can be set up in the absence ofreplication fork passage, they still underscore the impor-tance of S phase progression. So it is clear that many inter-esting questions remain.

A closer look at the silencer—replication connectionTroubles with the DNA replication model for the estab-lishment of silencing first emerged when it was discoveredthat neither of the silencer elements at HML are actuallyfunctional replication origins in their chromosomalcontext, despite the fact that they behave as active originson plasmids [7]. This observation made it hard to imaginethat replication initiation from the silencers themselves isessential for the establishment of repression, a notion thatwas further supported by studies showing that the DNAreplication initiation and silencing functions of ORC couldbe genetically separated [8,9]. Additional work from theFox laboratory [10] uncovered multiple initiator elementsat the HMR-E silencer, some of which even appear to beantagonistic to silencing. Nonetheless, these observationsstill left open the possibility that replication through asilent region — whether or not it initiates from the silencerelements themselves — would be required to re-assemblesilent chromatin at a previously derepressed locus.

The experimental tools required to address this issueemerged from studies designed to ask what the preciserole of the silencer elements is in the establishment andmaintenance of the repressed state. Protein targetingstudies, initially using the Gal4p DNA-binding domain(GBD), demonstrated that the silencer elements could, toat least some extent, be bypassed by tethering Sir proteinsto specific sites in chromatin [11]. This mode of establish-ing silent chromatin in a previously active region, referredto as ‘targeted silencing’, was first demonstrated with aGBD–Sir1 hybrid, but it works equally well with otherGBD–Sir hybrids provided they are expressed at appropri-ate levels in the cell [12]. Using a similar targetingapproach, Rine and colleagues [13] showed that regulatedexpression of a LexA–Sir1 hybrid could promote silencingat an HMR locus whose E-silencer replication function wasmissing, thus formally disproving the idea that a linkedorigin activity is required to establish silent chromatin.

Page 2: Transcriptional silencing: Replication redux

Dispatch R817

Remarkably, though, this ORC-independent form oftargeted silencing still required passage through S phase.

In a separate series of studies, the Broach [14] andGartenberg [15] laboratories used related site-specificrecombination systems (FLP or R) to assess the impact ofexcising silent chromatin domains from the chromosome,and to study their physical and biochemical properties asisolated DNA rings [14,15]. These studies showed thatsilencing could persist on chromatin rings, at least tem-porarily, but could not be inherited through cell divisionon rings lacking silencers. Interestingly, both groups alsonoted that the silenced state was associated with a topolog-ical change in the excised DNA ring.

DNA replication is not required to establish silentchromatin…In the recent work from the Rine [5] and Gartenberg [6]labs, targeted silencing and ring excision have beencombined to address the conclusion of the classic Millerand Nasmyth paper [1], namely that DNA replicationitself is the essential S phase event necessary for the estab-lishment of silencing. Both groups used a very similarapproach (illustrated in Figure 1), in which non-replicatingrings were formed by recombination, and silencing of therings was attempted by inducing expression of a DNA-binding domain hybrid (either a Gal4–Sir1 or a LexA–Sir1hybrid). In both systems, the ORC binding site within theHMR–E silencer was replaced by the appropriate hybridprotein binding site, and the I silencer element on theother side of the HMR locus was eliminated, thus assuringthat repression is strongly dependent upon the tetheringof Sir1 to the mutated E silencer, and that the wholeexcised locus contains no functional replication origin.

The results of these experiments were quite clear.Induced expression of the Sir1 hybrid protein led to strongrepression of the engineered HMR locus, regardless ofwhether that locus was present in the chromosome or inthe form of an excised DNA circle. Significantly, the HMRcircle became silenced but did not replicate (as predicted)when the cells passed through S phase (or at any otherpoint thereafter). This critical point was well documentedby both groups. Thus, both experiments show that silentchromatin can be established in the absence of DNAreplication. A skeptic might argue at this point that therepression established by targeting Sir1 to non-replicatingDNA rings is different than native silencing, or even thansilencing established on a similar, but replicating circle.However, Li et al. [6] showed that the other Sir proteinswere still required, and, more importantly, that theexpected change in histone acetylation patterns and ringtopology were also observed. Therefore, the outcome ofthe experiment met all of the established criteria forbona fide silent chromatin.

…yet establishment of silent chromatin requires passagethrough S phaseHaving shown that silent chromatin could be set up onnon-replicating rings, both groups were then in a positionto ask an additional, incisive question. Is the establish-ment of silencing on non-replicating DNA rings stilldependent upon cell cycle progression, and in particularon the passage through S phase? Remarkably, both of thegroups found that cells prevented from traversing S phase,either by a G1 block caused by the α-factor pheromone orby treatment of the cells with the DNA replicationinhibitor hydroxyurea, were unable to establish repressionon the non-replicating rings. This result is of course per-fectly consistent with the early findings of Miller and

Figure 1

Establishment of transcriptional silencing on non-replicating DNA rings.The E silencer element (top) of the HMR silent mating-type locus consistsof three elements, A, E and B, which are, respectively, binding sites forthe origin recognition complex (ORC) and for the multifunctionalregulatory proteins Rap1p and Abf1p. The I silencer, normally found tothe right of the a1 gene, consists of ORC and Abf1p binding sites. It isdeleted in the experiments described here and not shown in the figure.Replacement of the ORC binding site with multiple binding sites for a‘tethering’ protein, such as LexA or Gal4p, renders the silencer inactive.Silencing can be restored by expression of the appropriate DNA-bindingdomain–Sir1p hybrid protein. Excision of the locus from the chromosome,through the action of a site-specific recombinase at flanking recognitionsites (gene arrowheads), generates a non-replicating DNA ring (right).Remarkably, the a1 gene contained on this ring becomes silenced in theabsence of replication, but this process still requires passage through Sphase, as is also the case in the absence of ring formation (left).

a1

S phase

+recombinase+Sir1 hybrid

Current Biology

+Sir1 hybrid

S phase

Rap1ORC Abf1

n

A E B

Page 3: Transcriptional silencing: Replication redux

R818 Current Biology Vol 11 No 20

Nasmyth [1] and the later observation that the S phasedependence of silencing is not related to ORC function atthe silencer [13]. The stunning conclusion from thepresent work, then, is that there is indeed an S phasedependence to establish silent chromatin, but its mecha-nism is not in any way directly related to the passage of aDNA replication fork through the region in question.

Before considering the significance of these new findings,it is worth considering the possibility that their moststraightforward interpretation is incorrect. After all, theprotein targeting systems used are rather contrived, andwork by bypassing the normal mechanism by whichsilencer elements are thought to recruit Sir proteins andinitiate repression. Although it is difficult to exclude thispossibility, the fact that these artificial systems still displayan S phase dependence would itself seem to argue thatthey do mimic an essential feature of a native silent locus.Nonetheless, the possibility remains that a native silencerbenefits by the passage of a replication fork in some waythat cannot be measured with the present systems.

Understanding the relationship between silencing andS phaseAt the end of the day, then, the present studies, ratherthan refuting earlier work, have instead re-defined a keyquestion: if not passage of a replication fork, what is theS phase event(s) important for the establishment of silentchromatin? This brings us back to a point raised earlier,namely that concurrent studies have continued to point toa connection between DNA replication machinery andsilencing. For example, as pointed out by Kirchmaier andRine [5], an essential DNA helicase, Dna2p, and the Rfc1protein, a loading factor for proliferating cell nuclear antigen(PCNA), play important roles in telomeric and rDNAsilencing, respectively. Furthermore, mutation of either oftwo replication-coupled chromatin assembly factors, CAF-1or Asf1p, weakens silencing. Significantly, PCNA mutantsdefective in an interaction with CAF-1 display weakenedtelomeric and HM locus silencing [16,17].

The finding that PCNA is left behind on a replicatedtemplate and can provide a signal for subsequent CAF-1-dependent chromatin assembly [18] may provide a clue tohow replication and chromatin assembly-associated factorscould participate in silencing in the absence of forkpassage itself. It might be interesting, therefore, toexamine the effect of mutations in these factors on theestablishment of silencing on non-replicating DNA rings,or to examine directly their possible physical associationwith assembling, non-replicating heterochromatin. Analternative, but not mutually exclusive, possibility is thatthe S phase requirement for the establishment of silencingreflects a cell-cycle-dependent modification of one ormore important silencing factors. Apropos of this notion,

hyperphosphorylation of Sir3 in response to pheromonetreatment of cells was shown to strengthen silencing [19].

In considering the conflicting evidence for a role of DNAreplication in the establishment of silent chromatin, it isworth remembering that the mating-type silencing systemis highly redundant, both at the level of the cis-actingsilencer elements themselves and with respect to thenumerous trans-acting factors that make partial contribu-tions to repression. Even at telomeres, where silencing is a hit-or-miss proposition — repression is ‘variegated’ —and serves a still poorly understood biological role, thereappear to be multiple pathways for Sir recruitment and the process is subject to negative regulation by addi-tional factors [20]. It would not be surprising to find, then,that the cell uses multiple S-phase-specific mechanisms topromote heterochromatin assembly at particular sites. Thepresent studies may provide a key tool to elucidate thosethat are independent of DNA replication itself.

AcknowledgementsI would like to thank M. Gartenberg for helpful discussions and N. Roggli for theartwork. Research in my own laboratory is supported by the Swiss NationalScience Foundation, the Swiss Cancer League, and the Canton of Geneva.

References1. Miller AM, Nasmyth KA: Role of DNA replication in the repression of

silent mating type loci in yeast. Nature 1984, 312:247-251.2. Bell SP, Kobayashi R, Stillman B: Yeast origin recognition complex

functions in transcription silencing and DNA replication. Science1993, 262:1844-1849.

3. Foss M, McNally FJ, Laurenson P, Rine J: Origin recognition complex(ORC) in transcriptional silencing and DNA replication inS. cerevisiae. Science 1993, 262:1838-1844.

4. Micklem G, Rowley A, Harwood J, Nasmyth K, Diffley JF: Yeast originrecognition complex is involved in DNA replication andtranscriptional silencing. Nature 1993, 366:87-89.

5. Kirchmaier AL, Rine J: DNA replication-independent silencing inS. cerevisiae. Science 2001, 291:646-650.

6. Li YC, Cheng TH, Gartenberg MR: Establishment of transcriptionalsilencing in the absence of DNA replication. Science 2001,291:650-653.

7. Dubey DD, Davis LR, Greenfeder SA, Ong LY, Zhu JG, Broach JR,Newlon CS, Huberman JA: Evidence suggesting that the ARSelements associated with silencers of the yeast mating-type locusHML do not function as chromosomal DNA replication origins.Mol Cell Biol 1991, 11:5346-5355.

8. Fox CA, Loo S, Dillin A, Rine J: The origin recognition complex hasessential functions in transcriptional silencing and chromosomalreplication. Genes Dev 1995, 9:911-924.

9. Dillin A, Rine J: Separable functions of ORC5 in replicationinitiation and silencing in Saccharomyces cerevisiae. Genetics1997, 147:1053-1062.

10. Palacios DeBeer MA, Fox CA: A role for a replicator dominancemechanism in silencing. EMBO J 1999, 18:3808-3819.

11. Chien CT, Buck S, Sternglanz R, Shore D: Targeting of SIR1 proteinestablishes transcriptional silencing at HM loci and telomeres inyeast. Cell 1993, 75:531-541.

12. Marcand S, Buck SW, Moretti P, Gilson E, Shore D: Silencing ofgenes at nontelomeric sites in yeast is controlled bysequestration of silencing factors at telomeres by Rap 1 protein.Genes Dev 1996, 10:1297-1309.

13. Fox CA, Ehrenhofer-Murray AE, Loo S, Rine J: The origin recognitioncomplex, SIR1, and the S phase requirement for silencing.Science 1997, 276:1547-1551.

14. Bi X, Broach JR: DNA in transcriptionally silent chromatin assumesa distinct topology that is sensitive to cell cycle progression. MolCell Biol 1997, 17:7077-7087.

Page 4: Transcriptional silencing: Replication redux

15. Cheng TH, Li YC, Gartenberg MR: Persistence of an alternatechromatin structure at silenced loci in the absence of silencers.Proc Natl Acad Sci USA 1998, 95:5521-5526.

16. Zhang Z, Shibahara K, Stillman B: PCNA connects DNA replicationto epigenetic inheritance in yeast. Nature 2000, 408:221-225.

17. Sharp JA, Fouts ET, Krawitz DC, Kaufman PD: Yeast histonedeposition protein Asf1p requires Hir proteins and PCNA forheterochromatic silencing. Curr Biol 2001, 11:463-473.

18. Shibahara K, Stillman B: Replication-dependent marking of DNA byPCNA facilitates CAF-1-coupled inheritance of chromatin. Cell1999, 96:575-585.

19. Stone EM, Pillus L: Activation of an MAP kinase cascade leads toSir3p hyperphosphorylation and strengthens transcriptionalsilencing. J Cell Biol 1996, 135:571-583.

20. Mishra K, Shore D: Yeast Ku protein plays a direct role in telomericsilencing and counteracts inhibition by Rif proteins. Curr Biol1999, 9:1123-1126.

Dispatch R819