At telomeres, silent mating type loci, and rDNA arrays, budding yeast presents a form of transcriptional gene silencing that shares many of the same characteristics as heterochromatin in higher eukaryotes (Grewal and Moazed 2003). Genetic approaches have identified most of the components involved in the establishment and maintenance of yeast silent chromatin, yet the biochemical analysis of its assembly is just at its beginning. This thesis work was aimed at reconstituting an in vitro system that could be used to dissect and analyze the biochemical characteristics of silent chromatin of budding yeast.
The complexes formed by the Sir proteins
Early studies suggested that at the subtelomeric regions and at the HM loci the Sir proteins interact together to form a Sir2-3-4 complex, which was thought to be the basic unit of silent chromatin in S. cerevisiae (Moretti et al. 1994; Hecht et al. 1996;
Moazed et al. 1997; Strahl-Bolsinger et al. 1997). In contrast, Sir2 is the only Sir protein directly involved in silencing at the rDNA (Fritze et al. 1997; Smith and Boeke 1997). At the rDNA, Sir2 is part of the RENT complex together with Net1 and Cdc14 (Shou et al. 1999; Straight et al. 1999; Ghidelli et al. 2001). Even though mediated by different complexes, silencing at all silent regions requires the NAD-dependent histone deacetylase activity of Sir2 (Tanny et al. 1999; Tanny et al. 2004).
In fact, the two different complexes are able to produce very similar effects on Pol II-transcribed genes with mechanisms that are different. In Sir2, separation-of-function mutants could be isolated that affect only rDNA or only telomeric repression (Cockell et al. 2000; Cuperus et al. 2000; Garcia and Pillus 2002). These mutants revealed that a specific N-terminal portion of Sir2 is required for its binding to Net1 and for its localization to the nucleolus (Cockell et al. 2000). Other mutants caused Sir4 to re-locate to the nucleolus from the telomeres and HM loci (Cuperus et al. 2000). Thus it seems that sub-nuclear localization of Sir2 helps to determine which complex it forms.
Interestingly, silencing at subtelomeric regions and at the rDNA array were sensitive to the balanced expression of Sir2, Sir3, and Sir4 (Marshall et al. 1987; Renauld et al.
1993; Maillet et al. 1996; Fritze et al. 1997; Cockell et al. 1998a; Smith et al. 1998).
For example, over-expression of Sir2 disrupts silencing at the telomeres but improves it at the rDNA (Cockell et al. 2000).
To determine the biochemical mechanism behind these differences, we examined certain separation-of-function mutants biochemically. We could show that the Sir2-3-4 complex is a heterotrimer formed by one copy of each Sir protein, and that Sir2, when alone, adopts a homotrimeric conformation. We proposed that Sir2 multimerizes by intermolecular interaction between the N-terminal portion of the catalytic core of one Sir2 molecule and the substrate binding pocket of an adjacent one. This model was supported by structural studies on Hst2, a homolog of Sir2 (Zhao et al. 2003). The importance of Sir2 multimerization for proper silencing at the rDNA array was shown by characterizing a sir2 mutant that was specifically defective for silencing at the rDNA. This mutant form of Sir2 could not build a homotrimer in vitro.
On the other hand, this mutant was functional for TPE and silencing at HM loci and could correctly form a Sir2-3-4 complex. Importantly, the same mutant was able to complement the rDNA silencing defect of a catalytically inactive sir2 allele. The fact that two mutants, unable to support silencing on their own, were fully functional when expressed together implies that they interact together as a multimeric complex.
Nevertheless, we do not know if Sir2 exists as a homotrimer when bound to silent chromatin at the rDNA. However, it is possible that the RENT complex itself is formed by a multimeric Sir2 associated with Net1 and Cdc14.
To better characterize the telomeric Sir complexes, we purified the Sir2-4 complex and showed that it is a heterodimer formed by only one molecule of Sir2 and one of Sir4. We proposed that Sir4, at telomeres and HM loci, prevents the multimerization of Sir2. This observation correlates well with the observation that the region of Sir4 that interacts with the N-terminal domain of Sir2 is the same that is responsible for Sir2 homo-trimerization (Moazed et al. 1997; Cockell et al. 2000). The dominance of Sir2-4 complex formation over Sir2 homotrimerization correlates with an increase in deacetylation activity in Sir2 when it is complexed with Sir4. In fact Sir2 has its highest histone deacetylase activity when in complex with Sir4, while it presents intermediate activity when present in the Sir2-3-4 complex and the lowest when alone.
To summarize I would like to propose a model of nucleation and spreading of silent chromatin. Based on previous ChIP data (summarized in (Luo et al. 2002)) one might speculate that the first step of the assembly of silent chromatin is the recruitment of Sir4 by Rap1 or Sir1. Considering our data and previous published work (summarized
in (Rusche et al. 2003) and in (Gasser and Cockell 2001)), the next step would consist on the formation of the Sir2-4 complex that deacetylates the nucleosomes close to the Sir nucleation sites. The strong histone deacetylase activity of Sir2-4 may be important to create a large zone of deacetylated histone tails. The deacetylated histone tails would favor the recruitment of Sir3 (Liou et al. 2005) or of the pre-formed Sir2-3-4 complex (Cubizolles et al. 2006) to the regions of silent chromatin. The histone deacetylation mediated by Sir2 promotes the binding of the Sir2-3-4 complex to the regions of silent chromatin and overcomes other activities that are believed to counteract the spreading of silencing, such as the H4K16 histone acetyl transferase Sas2, the H3K79 histone methyl transferase Dot1, that is dependent on the acetylated state of H4K16 and the chromatin remodeling activity of the SWR1 complex responsible for the exchange of H2A with Htz1 (this work and (Jackson and Denu 2002; Kimura et al. 2002; Ng et al. 2002; Suka et al. 2002; van Leeuwen et al. 2002;
Meneghini et al. 2003; Liou et al. 2005; Altaf et al. 2007)). Finally, our work argues that the spreading of the Sir2-3-4 complex inward from the chromosome end may require a crucial amount of AADPR. In support of this model, we show that O-AADPR increases the affinity of the Sir2-3-4 complex for binding to chromatin and Moazed’s laboratory has shown that O-AADPR induces a conformational change in the Sir2-3-4 complex and favors multimerization of oligonucleosomes (this work and (Liou et al. 2005; Onishi et al. 2007)). The Sir2-3-4 complex could therefore be responsible for maintaining the deacetylated state of histone tails at silent chromosomal regions and for promoting its own spreading from the telomeres along the chromosome arms.
Since Rap1, Sir3, and Ku bind to the same domain of Sir4, we imagine that different complexes, such as a Sir2-4/Rap1, a Sir2-4/Ku, a Sir2-3-4, or hybrids of those, could form at telomeres. It would be interesting to reconstitute these complexes and test their properties of assembling silent chromatin.
The expression and purification of Sir4 alone or of Sir3 and Sir4 together revealed that Sir4 tends to aggregate with itself. Our biochemical analysis, and work from others, supports the idea that cellular Sir4 is almost always in a complex with Sir2 and/or Sir3 to form the Sir2-4 and Sir2-3-4 complexes (this work and (Ghidelli et al.
2001; Hoppe et al. 2002; Tanny et al. 2004; Liou et al. 2005; Rudner et al. 2005)).
The tendency of Sir4 alone to aggregate, and thus to form non-physiological complexes, may explain why over-expression of Sir4 causes disruption of silencing at
the telomeres and HM loci (Marshall et al. 1987; Cockell et al. 1998a). In contrast, over-expression of Sir3 improves the spreading of silencing from telomeres along the chromosome arms in vivo (Renauld et al. 1993). The spreading of silencing was still dependent on Sir2 and Sir4 even though only Sir3 could be detected at the internal locations (Hecht et al. 1996; Strahl-Bolsinger et al. 1997). To see whether Sir3 can bind nucleosomes alone, I expressed and purified Sir3 and tested its ability to assemble silent chromatin. Purified recombinant Sir3 was a monomer in solution. Our data do not directly contradict previously published results showing that Sir3 formed multimers (Liou et al. 2005; McBryant et al. 2006), because oligomerization of Sir3 is sensitive to salt (even at 100mM NaCl) and is compromised at low protein concentrations (McBryant et al. 2006). It is possible that the dilution effect of our sucrose gradient, the NaCl concentration used during the elution (300mM NaCl), and the NaCl present in the sucrose gradient (150mM), compromised the oligomerization state of Sir3. Our binding experiments are done at Sir3 concentrations of between 10
-8M and 10-7M, which excludes the possibility that Sir3 forms oligomers in solution under these conditions. Because of the increased local concentration of Sir3 on chromatin, it is possible that Sir3 multimerization is favored after being recruited to silent regions, and thus after interacting with nucleosomes at those locations. This model will be discussed below.
In vitro reconstitution of yeast silent chromatin reveals the mechanism of binding of the Sir2-3-4 complex to chromatin
The availability of purified Sir proteins and Sir complexes enabled us to characterize how they bind chromatin. We first wanted to understand the contribution of each single Sir protein and of single regions of the nucleosome in the formation of silent chromatin. The tail of histone H4 has been shown to be important for silencing in yeast in vivo, since deletion of the first 19 amino acids of histone H4 was sufficient to completely disrupt silencing at the HML locus (Kayne et al. 1988). Similarly, any point mutation of lysine 16, arginine 17, histidine 18, or arginine 19 of H4 to a neutral amino acid led to complete derepression of silencing at both the HMLα (Johnson et al.
1990; Park and Szostak 1990) and sub-telomeric regions (Aparicio et al. 1991). These data led to the idea that the Sir proteins bind chromatin through the tail of histone H4.
With our reconstituted system, we could prove that the binding of the Sir2-3-4 complex to the tail of H4 is mediated by Sir3 and not by Sir2-4. We did not see a role of the histone H3 tail in the binding of either Sir2-3-4 or Sir3. The Sir2-3-4 complex, in opposition to Sir3 alone, showed a lower sensitivity to the lack of the histone H4 tail. This apparent contradiction could be explained by the fact that Sir2-3-4 has, on the nucleosome, more binding sites than just the H4 tail. In this regard, I showed that Sir2-3-4 complex binds, through Sir3, the region surrounding H3K79 and, through Sir4, the nucleosomal DNA. Regarding the binding to DNA, it has been proposed that acetylation at H3K56 by RTT109 causes an alteration of the interactions between the histone octamer and the nucleosomal DNA (Xu et al. 2007). It is possible that the distortion of the nucleosomal DNA induced by H3K56 acetylation, disrupts the binding of the Sir2-3-4 complex to the nucleosome by reducing its affinity for nucleosomal arrays. Since substitutions at H3K56 did not disrupt binding of the Sir2-3-4 complex to telomeres (Xu et al. 2007), it is also plausible that acetylation of H3K56 prevents the compaction of the chromatin fiber into higher order structures such as that mediated by Sir2-3-4. These models could be easily tested with the in vitro reconstitution system.
The lower affinity of the Sir2-3-4 for nucleosomes lacking the H4 tail could be explained differently. We entertain the possibility that the tail of histone H4 in vivo primarily serves to recruit the Sir2-3-4 complex to the silent regions. Once recruited, its binding would then be stabilized by interactions with the nucleosomal DNA and with the surface of the nucleosome core. Importantly, the mutation of H3K79 to alanine affects silencing at telomeres and HM loci (Ng et al. 2002; van Leeuwen et al.
2002), suggesting that Sir proteins bind this region of the nucleosome. Interestingly, methylation of H3K79 by Dot1 is thought to prevent the binding of Sir proteins at silent regions (van Leeuwen et al. 2002). Indeed, we showed that the affinity of Sir2-3-4 for nucleosomes is strongly decreased by H3K79 methylation. Moreover, we could prove that this effect is mediated by Sir3. The fact that Sir3, on its own, is less sensitive than the Sir2-3-4 complex may suggest that the correct functioning of Sir3 (and in general of Sir proteins) requires its presence in the Sir2-3-4 complex.
From these experiments we could conclude that, in the Sir2-3-4 complex, Sir3 is binding both the H4 tail and the region surrounding H3K79. Interestingly, the acetylated tail of histone H4 has been shown to be necessary for H3K79 methylation (Altaf et al. 2007), thus the defect in silencing seen in strains missing the histone H4
tails might be indirect and linked to the dilution effect of Sir proteins as described for Dot1 and H3K79 mutants (van Leeuwen et al. 2002).
When we monitored the binding of Sir2-3-4 and Sir2-4 to chromatin, we noticed that the Sir4 complexes had an affinity for naked DNA. Indeed, we could prove that Sir4 binds naked DNA with a high affinity and poor sequence specificity. It is possible that Sir4 binds the nucleosomal linker DNA, since both the Sir2-3-4 and the Sir2-4 complex showed a lower affinity for core mono-nucleosomes, lacking linker DNA, than for arrays of nucleosomes. We speculate that the affinity of Sir4 for naked DNA is important for stabilizing the binding of Sir2-3-4 to the nucleosome. It is also possible that Sir4, through its binding to DNA, has a role in the folding of silent chromatin into higher order structures. This model will be tested in the future using electron microscopy and sedimentation analysis.
I would like to propose a model where silencing in yeast is mediated by several interactions between the Sir2-3-4 complex and the nucleosome as proposed by Gottschling’s laboratory (van Leeuwen and Gottschling 2002). These interactions stabilize the binding of the Sir2-3-4 complex to chromatin in vivo and probably are necessary to ensure that the Sir2-3-4 complex counteracts the action of anti-silencing factors such as Sas2, SWR1 and Dot1. Once stably bound to the nucleosome, Sir2-3-4 promotes the folding of the chromatin fiber into a structure that induces transcriptional gene silencing. It is possible that all the interactions between the Sir2-3-4 complex and the nucleosome described so far are necessary for this process of folding of silent chromatin.
Analysis of Sir4 domains reveals their importance in the formation of silenced chromatin in yeast
Sir4 is essential for silencing in yeast (Rine and Herskowitz 1987) since it is involved in both the recruitment and the spreading of other Sir proteins (Hecht et al. 1996;
Strahl-Bolsinger et al. 1997; Luo et al. 2002). Because of its size and instability, the biochemical characterization of Sir4 has always been very limited. In previous studies, it has been shown that an N-terminal fragment of Sir4, representing its first 344 amino acids, and a C-terminal of Sir4, formed by its last 700 amino acids, could complement a null allele of SIR4 (Marshall et al. 1987), thus suggesting that these two portions of
Sir4 have different functions. Our reconstituted system offered us a good opportunity to investigate the function of these two domains. We could show that the N-terminal fragment of Sir4 could bind both DNA and chromatin. This observation was strengthened by the fact that the N-terminal portion of Sir4 could displace the binding of the Sir2-3-4 complex. In contrast, the coiled-coil domain of Sir4 did not bind either chromatin or DNA. Importantly, we showed that both the Sir2-3-4 and the Sir2-4 complexes do not multimerize in solution (Cubizolles et al. 2006). In contrast, the coiled-coil domain of Sir4 has been shown to dimerize when present in solution at high concentration (Murphy et al. 2003). We showed in this work that the Sir2-3-4 complex binds to chromatin in a co-operative fashion. These observations suggest that the recruitment and binding of the Sir2-3-4 complex increase the local concentration of the Sir2-3-4 complex, thus inducing its multimerization. We envision a model where the compaction of the chromatin fiber at silent regions is mediated by both the binding of the N-terminus of Sir4 to the linker DNA and the dimerization of the C-terminal domains of two Sir complexes placed on the opposing faces of two adjacent nucleosomes. In support of this model, we showed that the coiled coil domain of Sir4 disrupts the co-operative binding of the Sir2-3-4 complex to chromatin. Future electron microscopic analysis may offer support to this model.
Importance of the deacetylation reaction of Sir2
Sir2 is a NAD-dependent histone deacetylase whose activity is necessary for the establishment and maintenance of silencing (Gasser and Cockell 2001). It has been previously shown that Sir2 transfers the acetyl group from the lysines of the histone tails to the ribose (created by the breakage of the NAD into NAM and ADP-ribose) and consequently generates O-Acetyl-ADP-Ribose (O-AADPR). Here we show that the O-AADPR increases the affinity of the Sir2-3-4 complex for binding chromatin. This effect is due to Sir3, since a similar increase in affinity for binding nucleosomes is seen with Sir3 alone. We propose that the deacetylation reaction of Sir2 is involved in the spreading of the Sir2-3-4 complex inside the chromosome by increasing its affinity for binding to chromatin. If O-AADPR promotes the spreading of the Sir2-3-4 complex along the chromosome, we reason that a decrease in the acetylation state of H4 affects silencing. Indeed, point mutations of the H4 tail that maintained the amino acid charge but that prevented its acetylation disrupted silencing (Braunstein et al. 1996) and the binding of Sir proteins was decreased but
not completely abolished (Yang and Kirchmaier 2006). Moreover, mutants of Sas2 that reduced its activity also presented a decrease in silencing at the telomeres (Suka et al. 2002). The deacetylation reaction of Sir2 is thus important not only to create binding sites for the Sir2-3-4, but also to produce a molecule that is important for silencing itself.
Model of recruitment and spreading of silencing
In figure 9 I propose a model of recruitment and spreading of silencing at the telomeres. Sir complexes are recruited to the telomeric TG1-3 repeat by the interaction of Sir4 with Rap1 and Ku70/Ku80 (Boulton and Jackson 1996; Hecht et al. 1996;
Strahl-Bolsinger et al. 1997; Laroche et al. 1998; Mishra and Shore 1999). The high stability of the Sir2-4 complex, and its high histone deacetylase activity, suggests that the Sir2-4 complex is probably the first complex recruited at telomeres. The Sir2-4 complex deacetylates the histones on the nucleosomes adjacent to the telomeres and recruits Sir3 at the telomeres (Braunstein et al. 1996; Tanny et al. 1999; Hoppe et al.
2002; Kimura et al. 2002; Suka et al. 2002). The interaction between Sir3 and the Sir2-4 complex is probably stabilized by the local high Sir protein concentrations, since we and others have shown that the interaction between Sir2, Sir3, and Sir4 exists even without the deacetylation reaction of Sir2 (Rudner et al. 2005; Cubizolles et al. 2006). The NAD+-dependent deacetylation reaction mediated by Sir2 produces O-AADPR that is thought to bind the AAA+ module of Sir3 and to induce its multimerization (Liou et al. 2005). The deacetylated tail of histone H4 and the region surrounding H3K79 are bound by the Sir2-3-4 complex through the BAH domain of Sir3. High local concentrations of the Sir2-3-4 complex favor the interaction between adjacent Sir2-3-4 complexes through the coiled coil domain of Sir4 and the formation of “dimers of Sir2-3-4 complexes”.
Rap1 Rap1 Rap1
I° step: Nucleation of silent chromatin at telomeres -Ku and Rap1 recruit Sir4
-Sir4 recruits Sir2 and Sir3
-Sir2, Sir3 and Sir4 form the Sir2-3-4 complex
-Sir2, Sir3 and Sir4 form the Sir2-3-4 complex