In this study we were able to show with two-hybrid experiments that the histone H3 homologue Cse4 interacted with the acetyltransferase Sas2. We further narrowed the Sas2-interacting region of Cse4 down a region within the N-terminus (amino acids 11-139). Therefore, we propose that Sas2 interacted with a portion of Cse4 that lies outside of the centromeric nucleosome core, whereas the essential C-terminal histone-fold domain was dispensable for the interaction.

As Sas2 is a member of the MYST-family of HATs, it contains an acetyl-CoA-binding site (HAT) and an atypical zink finger (C2HC motif). Both motifs are essential for the acetylation activity of Sas2. We tested if the interaction between Cse4 and Sas2 was still intact when the HAT-domain or the atypical zink finger, respectively, was mutated. Clearly, the interaction between Cse4 and the point-mutated Sas2 was disturbed, so that we concluded that both motifs were important for the binding of Sas2 to Cse4. As the zink-finger is absolutely required for substrate recognition in the Drosophila Sas2-homologue MOF (Akhtar and Becker, 2001), one way of explanation could be that Cse4 was an acetylation target for Sas2.

Since Sas2 functions in a complex together with Sas4 and Sas5 as a histone acetyltransferase, we hypothesized that Cse4 also binds to Sas4 and Sas5. Indeed we could verify an interaction over the N-terminal tail of Cse4 and Sas4, whereas no binding to Sas5 was detected. From our two-hybrid data we concluded that Cse4 binds via its N-terminus to two subunits of the SAS-I complex, Sas2 and Sas4.

All two-hybrid interactions were further confirmed with co-immunoprecipitation experiments in vivo. Interestingly, we were also able to precipitate Sas2-HAT^-, Sas2-Zn^- and Sas5 with Cse4, although no direct interaction was detected with the two-hybrid assay. However, as Cse4 interacted with two subunits from the SAS-I complex, Sas2 and Sas4, the interaction with the point mutated Sas2 as well as with Sas5 could be bridged by them. In this case a positive precipitation could be found even if the proteins did not bind directly to each other.

Two other proteins could also be able to bridge these interactions – Cac1, the largest subunit of the chromatin assembly factor CAF-I, and Asf1. CAF-I and Asf1 are histone chaperones that deliver histone H3 and histone H4 to DNA after replication or repair and function in partially overlapping pathways of nucleosome assembly. Furthermore, Asf1 and Cac2 interact with each other and increase CAF-I activity in nucleosome assembly (Mello, et al., 2002);(Tyler, et al., 2001).

Both proteins, Cac1 and Asf1, have been shown to interact with the SAS-I complex (Meijsing and Ehrenhofer-Murray, 2001);(Osada, et al., 2001), and we found with two-hybrid and co-immunoprecipitation experiments that they also bind to Cse4. The interaction with Cac1 could further be narrowed down to the amino acids 94-137. Although Cac1 exists in a complex, no interaction could be detected with Cac2 or Cac3, which argues for a function of Cac1 separate of CAF-I. Recently, (Sharp, et al., 2002) reported that CAF-I and Hir1 have a role in maintaining the centromeric chromatin structure. They are not absolutely required for Cse4 deposition, but their absence leads to additional deposition outside of the centromere.

Since Cac1 as well as Asf1 are both chromatin assembly factors that bind histones H3 and H4, our findings together with the results from (Sharp, et al., 2002) support a model where Cac1 and Asf1 were also responsible for delivering the histone H3 variant Cse4 to the centromere. Since CAF-I and Hir1 are not essential for recruiting Cse4 to the centromere (Sharp, et al., 2002), our results could point out an important role for the nucleosome assembly factor Asf1. In Figure 14 all interactions between Cse4, SAS-I and the chromatin assembly factors are summarized.

In our experiments we further found that neither deletion of sas2, sas4 or cac1 affected the association of Cse4 with components of the SAS-I complex or Cac1. One exception was found in a sas4Δ strain, where Cac1 and Cse4 were unable to bind to each other. If sas4 is deleted, the whole SAS-I complex is disturbed, because Sas5 binds only to Sas4 and not to Sas2. In light of our results, one interpretation could be that Sas4 and/or Sas5 are important for recruiting Cac1 to the SAS-I complex, and therefore to Cse4.

Fig. 14: A model of the histone H3 variant Cse4 and its interactions with the SAS-I complex as well as with a component of the chromatin assembly factors CAF-I and Asf1. A direct interaction is proposed with Sas2, Sas4, Cac1 and Asf1, whereas Sas5 does not bind directly to Cse4.

This model would include that Cac1 is only able to bind to the complex in the presence of the active SAS-I complex. Sas4 has been shown to be essential for histone acetyltransferase activity for Sas2 (Sutton, et al., 2003). If SAS4 is deleted, the SAS-I complex is both incomplete and inactive. In this case, Sas2 would be unable to acetylate its target, probably the histone H3 homologue Cse4 or histone H4, which would then prevent binding of Cac1 to the complex. This hypothesis would further include that Cac1 is also unable to bind to Sas2, which remains to be tested.

(Meijsing and Ehrenhofer-Murray, 2001) have reported that Sas2, Sas4 and Sas5 coeluted in gel filtration experiments in a peak of ~220 kD, even if the calculated molecular masses of myc-Sas2, myc-Sas4 and Sas5 add up to ~140 kD. Additionally, (Osada, et al., 2001) detected by Superose 6-size exclusion chromatography and a subsequent western blot a 450 kD Sas2 containing complex. This implied that additional subunits of the complex might exist. One possibility could be that Cse4 interacted with the SAS-I complex and could therefore be a missing component. Consequently, it would be interesting to investigate if a tagged Cse4 could also be found in the elution peak of the SAS-I complex. We tried to solve this question with a TAP-tagged SAS-I complex and mass spectrometry analysis (data not shown), but we were unable to detect Cse4. One explanation could be that the interaction with Cse4 takes place in only one phase of the cell cycle, although Cse4 and the SAS-I components are continuously expressed at low levels. Another possibility could be that the association of Cse4 to the SAS-I complex is not so strong. In consequence, Cse4 could be lost during the purification procedure so that no Cse4 could be detected although the interaction exists in the cell. Furthermore, the SAS-I bound Cse4 amounts could have been under the detection limit for mass spectrometry analysis, so that in consequence we were unable to find Cse4 in the assay.

The histone acetyltransferase Sas2 functions in the SAS-I complex together with Sas4 and Sas5 and acetylates histone H3 K14 and histone H4 K16 (Meijsing and Ehrenhofer-Murray, 2001);(Sutton, et al., 2003). Additionally, Sas2 has a role in heterochromatic gene silencing at the HM loci and at the telomeres (Ehrenhofer-Murray, et al., 1997);(Reifsnyder, et al., 1996).

Does Sas2 also have a role at the centromere ? A direct function for Sas2 at the centromere has not been described so far, but (Sharp, et al., 2003) discovered that the silencing protein Sir1 is a functional component of centromeric chromatin and helps to maintain CAF-I at the centromere. As Sas2 genetically interacts with Sir1 in silencing at HML (Ehrenhofer-Murray, et al., 1997), an additional function at the centromere might be possible. Furthermore, our two-hybrid and co-immunoprecipitation results that Sas2 interacts with Cse4 provide further evidence that Sas2 has additional functions in non-heterochromatic regions of the genome.

In addition to our findings that the SAS-I complex was associated with Cse4, we found that a SAS2 deletion leads to better growth in the temperature-sensitive cse4-103 strain at 34°C. The cse4-103 strain has a point mutation in two amino acids within the histone-fold-domain (I156V, L193Q). This mutation is thought to affect the interaction with the globular domain of histone H4. In consequence, the (cse4-103/H4)₂ tetrameres are destabilized at elevated temperatures and the cells are unable to survive.

How can a SAS2 deletion have a positive influence on the stability of (cse4-103/H4)₂ tetrameres ? It is known that Sas2 acetylates histone H4 at lysine 16, which is also present at the centromere in contrast to histone H3. Histone acetylation has mainly been found in the neighborhood of transcriptionally active promotors and enhancers (Kuo and Allis, 1998), where they are thought to restrict the folding of nucleosomes into the condensed 30 nm fiber (Garcia-Ramirez, et al., 1995);(Tse, et al., 1998). As the yeast centromere is not packed into heterochromatin like in other organisms, one might hypothezise that the histone tails exist in an acetylated state. In SAS2 wildtype cells, this acetylation may further destabilize the already weak interaction between cse4-103 and histone H4, so that in consequence the cells are unable to grow at elevated temperatures. If the histone acetyltransferase Sas2 is absent, the acetylation of histone H4 K16 is missing and the nucleosomes would be able to form a more compact structure that in turn helps to tighten the (cse4-103/H4)₂ tetrameres. Another possibility could be that Sas2 also acetylates Cse4 next to histone H4. A missing acetylation on Cse4 would therefore have the same consequences on the stability as on histone H4 in a sas2Δ strain.

Here, we were able to show that Sas2 has an additional role that contributes to centromere stability. The centromeric protein Ctf19 functions in a complex together with Okp19 and Mcm21 in centromere stability. This complex mediates the connection between CDEI, CDEII and CDEIII by binding to different centromeric proteins, e.g. via the interaction between Ctf19 and Cse4. We showed by two-hybrid experiments that a deletion of sas2 disrupted the binding of Cse4 to Ctf19.

How can Sas2 influence the interaction between Cse4 and Ctf19 ? Post-translational modifications of histones have already been shown to be critical for protein-protein interactions. One example are the silent information regulator (SIR) proteins Sir3 and Sir4 that interact with deacetylated tails of histone H3 and histone H4 in order to build up a repressive chromatin state. An explanation for our observation could be that Sas2 acetylates one or more lysine residues on the N-terminus of Cse4. Thereby, the acetylation may provoke a conformational change in the (Cse4-H4)₂ tetrameres that are now able to interact with other proteins, e.g. Ctf19. In a sas2Δ cell, the N-terminus of Cse4 may remain more unflexible due to the missing acetylation so that the accessibility for other interacting proteins is decreased. In that case, the interaction between Cse4 and Ctf19 would be disrupted as can be seen in a two-hybrid assay.

Further analysis with co-immunoprecipitation experiments need to be carried out to confirm this hypothesis. Additionally, chromatin immunoprecipitation or indirect immunofluorescence experiments with Sas2 could be used to test whether it can be found at some point of the cell cycle at the centromere.

We were interested in exploring the possibility that Cse4 exists in an acetylated state in the cell and if the histone acetyltransferase Sas2 is a Cse4-modifying enzyme. Indeed, we have evidence that Cse4 is acetylated, but whether Sas2 is involved in this process remains unclear.

The N-termini of the core histones H2A, H2B, H3 and H4 are post-translationally modified by methylation, acetylation and phosphorylation. These modifications are essential for modulating nucleosome structure and therefore changing gene activity. Little is known about modification of histones and histone variants at the centromere, but it has been reported that histone H3 phosphorylation in mammals precedes the phosphorylation of the Cse4-homologue CENP-A in prophase (Zeitlin, et al., 2001). This phosphorylation is necessary for directing the subcellular localization of enzymes required for completion of cytokinesis.

So far no modifications have been shown for the histone H3 variant Cse4, although they may occur because its homologue in mammals is phosphorylated. We were now able to show that Cse4 is acetylated at least at one stage in the cell cycle. This acetylation could be a specific mark for chromatin assembly factors like CAF-I and Asf1. In consequence, Cse4 is exclusively delivered to the centromere, where it replaces histone H3. One putative acetyltransferase involved in Cse4 modification could be Sas2, although no direct activity was detected. An explanation for the missing acetylation activity in our in vitro assay could be that the Cse4 concentration was too low. In that case, the radioactive signal of [¹⁴C] may be too weak to be detected. Another possibility could be that additional cofactors or pre-existing modifications at the N-terminus are needed for Cse4 acetylation. It has already been described for the histone H3 N-terminus that modifications are able to influence each other. One example is that histone H3 K4 methylation by Set7 inhibits methylation of lysine 9 by Su(var)3-9, but promotes acetylation of histone H3 by p300 (Wang, et al., 2001).

We propose the following model from our data: during replication, newly synthesized as well as pre-existing histones need to be incorporated into nucleosomes. The histone H3 variant Cse4 replaces histone H3 at the centromere by forming centromere specific (Cse4-H4)₂ tetrameres. But how do chromatin assembly factors distinguish between (Cse4-H4)₂ and (H3-H4)₂ tetrameres and their site of incorporation ?

First, Cse4 and histone H3 have a highly homologous C-terminal histone-fold domain, but their N-terminus is distinctly different. The 135 aa N-terminus of Cse4 shows no homology to known proteins and contains at least one essential function, since the deletion of the END domain (amino acids 28-60) is lethal to the cell (Chen, et al., 2000). The N-terminus of Cse4 may extend from the core and is able to interact with other proteins, so that in consequence chromatin assembly factors could distinguish between the standard core (H3-H4)₂ tetramere and centromere specific (Cse4-H4)₂ tetrameres. Another mechanism for helping the chromatin assembly factors to separate H3-H4 assembly from Cse4-H4 assembly is to divide their assembly in time. Whereas the standard core histones are exclusively expressed in early S-phase, Cse4 mRNA can be found at low levels throughout the whole cell cycle. Thus, the equilibrium is shifted towards Cse4-H4 after early S-phase, which may help to deposit them after H3-H4 deposition took place.

In this model, the histone acetyltransferase complex SAS-I may first bind to free Cse4 and specifically modifies one or more lysine residues on its N-terminus. The nucleosome assembly factor Asf1 is a histone chaperone that delivers histones H3 and H4 to the replication fork. Asf1 binds to the SAS-I complex via Sas4 (Meijsing and Ehrenhofer-Murray, 2001);(Sutton, et al., 2001) and we were able to show that Asf1 also interacted with the histone H3 variant Cse4. Thus, Asf1 might bind to the (Cse4-H4)₂ tetrameres after identification via the N-terminus of Cse4. Here, the specific Sas2-dependent acetylation may already have an important role for recognizing the specificity for centromere deposition. (Sharp, et al., 2002) reported that Asf1 didn’t play a role at the centromere, because cacΔasf1Δ cells segregated a CFIII(D8B.d) reporter minichromosome at wildtype frequencies and displayed a G2-M delay independent of spindle assembly checkpoint activation. In contrast to their observation, we found that a single asf1Δ already resulted in a significant increase of loss of an additional chromosome III (data not shown). Thus, the conclusion that the centromeres did not require Asf1 requires re-investigation.

In a subsequent step, the whole complex may be delivered to the centromere. Here, the chromatin assembly factor CAF-I may still be bound to the DNA after replication took place. Asf1 interacts with Cac2 (Tyler, et al., 2001) and may deliver the (Cse4-H4)₂ tetrameres to CAF-I that binds the histone H3 variant Cse4 via the Cac1 subunit, and histone H4 via Cac3 (Verreault, et al., 1996).

In this model, CAF-I releases Asf1 as well as SAS-I and now deposits Cse4 and histone H4 into centromere-specific nucleosomes and therefore replaces pre-assembled (H3-H4)₂ tetrameres. The acetylation of Cse4 may further stabilize the centromere by mobilizing the N-terminus that can now interact with other centromeric protein, e.g. Ctf19. Of course, further studies are necessary to prove this model and also to elucidate the role of the nucleosome assembly factor Asf1 as well as the dependence of the Cse4-Ctf19 interaction on the histone acetyltransferase Sas2.