3.  Results

Sas2 belongs to the MYST family of HATs and acetylates lysine 16 on histone H4 (Meijsing and Ehrenhofer-Murray, 2001);(Suka, et al., 2002) and lysine 14 on histone H3 (Sutton, et al., 2003). To identify proteins that interact with Sas2, we performed a two-hybrid screen and identified a fragment of the histone H3 variant Cse4 (Table 1).

Table 1 : Two-hybrid interactions between the histone H3 variant Cse4, the SAS-I complex and chromatin assembly factors. The reporter strain L40c was cotransformed with a plasmid expressing Cse4 and a plasmid containing components of the SAS-I complex or chromatin assembly factors. The genes were fused either to a Gal4 activation domain (GAD) or to a DNA binding domain (LexA). Interactions between proteins were measured as expression of lacZ and HIS3. Positive interactions (+) became blue in a β-galactosidase filter assay within 2 h and were able to grow on minimal media lacking histidine, negative interactions (-) did not become blue and were unable to grow on media lacking histidine; n.d. = not determined.

The interaction between Cse4 and Sas2 was further narrowed down to the N-terminus (amino acids 11-139), whereas the C-terminal region (amino acids 137-229) showed no interaction with Sas2. Since the C-terminal region is highly similar to the globular domain of H3, this suggested that Sas2 interacted with a portion of Cse4 that lies outside of the centromeric nucleosome core.

In order to test whether Cse4 and Sas2 also interact in vivo, we performed co-immunoprecipitation experiments (Fig. 7A). We constructed a 3x-HA-tagged version of CSE4 that was fully functional in that it rescued the lethality of a CSE4 deletion (data not shown). In an immunoprecipitation with a Sas2-antibody, we were able to detect HA-Cse4, whereas the antibody did not precipitate HA-Cse4 in the absence of Sas2.

Since Sas2 exists in the SAS-I complex together with Sas4 and Sas5 (Meijsing and Ehrenhofer-Murray, 2001), we asked whether Cse4 was able to interact with Sas4 and Sas5. Whereas Cse4 interacted via its N-terminus also with Sas4, we were unable to find two-hybrid interactions between Cse4 and Sas5. The binding of Cse4 to Sas4 in vivo was confirmed by co-immunoprecipitation. Interestingly, Sas5 could be precipitated in this assay (Fig. 7A). Thus, we suggest that Cse4 and Sas5 did not interact directly with each other, but that they can be found in the same complex in vivo. Possible mediators between these two proteins are Sas2 or Sas4.

As a member of the MYST family of HATs, Sas2 contains an acetyl-CoA-binding site (HAT) and an atypical zink-finger (C2HC motif), which is necessary for substrate recognition in the Sas2 homologue MOF in D. melanogaster (Clarke, et al., 1999). We tested the relevance of these motifs for the interaction with Cse4 (Table 1, Figure 7B). Both a mutation in the atypical zink-finger (C106L) and in the acetyl-CoA-binding site (P213A/P214V) resulted in a loss of the Cse4-Sas2 two-hybrid interaction, showing that both motifs were essential for Sas2 binding to Cse4.

	Fig. 7: Interactions between Cse4 and the SAS-I complex. The antibodies for precipitation are indicated above the panel. (A) Cse4 immunoprecipitated with Sas2, myc-Sas4 and myc-Sas5. (Top) The protease deficient yeast strain AEY1558 was transformed with plasmids containing 3x-HA-CSE4 (pAE977) and SAS2 (pAE90). Precipitates were detected with α-HA-antibody. (Middle) For immunoprecipitation between 3x-HA-Cse4 (pAE977) and 6x-myc-Sas4 (pAE613) AEY2461 was used. Precipitates were analyzed with α-HA-antibody. (Bottom) Sas5 immunoprecipitated with Cse4. AEY1558 was cotransformed with plasmids containing myc-CSE4 (pAE975) and HA-SAS5 (pAE625). Detection of proteins was performed with α-HA-antibody. (B) Cse4 immunoprecipitated with Sas2 variants, carrying single point mutations in the acetyl-CoA-binding site (HAT-, P213A/P214V) or in the zink-finger (Zn-, C106L). For this purpose, 3x-HA-Cse4 (pAE976/pAE977) and Sas2-Zn- (pAE388) or Sas2-HAT- (pAE249), respectively, were introduced into AEY1559. For detection in precipitates α-HA-antibody was used.

However, the Sas2 mutants still co-immunoprecipitated with Cse4 (Fig. 7B), suggesting that the Cse4-Sas2 interaction was stabilized in vivo by other proteins. Alternatively, the folding of Sas2 in the two-hybrid may be partially compromised.

The SAS-I complex interacts with Cac1, the largest subunit of CAF-I, as well as with the nucleosome assembly factor Asf1 (Meijsing and Ehrenhofer-Murray, 2001). Since SAS-I interacted with Cse4, we next sought to determine whether Cac1 and Asf1 interacted with Cse4. With both two-hybrid and co-immunoprecipitation, we could demonstrate that Cse4 interacted with Cac1 (Table 1, Fig. 8A), where the interacting region of Cse4 could further be narrowed down to the N-terminus, and Asf1 (Table 1, Fig. 8B).

	Fig. 8: Interactions between Cse4 and chromatin assembly factors. The antibodies for precipitation are indicated above the panel. (A) Cse4 immunoprecipitated with Cac1, but not with the other two subunits of the CAF-I complex, Cac2 and Cac3. (Top Left). AEY1808 was cotransformed with plasmids containing 3x-HA-CSE4 (pAE976) and 6x-myc-CAC1 (pAE614). Precipitation was carried out with α-HA-antibody. (Top Right) No interaction was detected between 3x-HA-Cse4 (pAE977) and 6x-myc-Cac2 in AEY1558. Detection of precipitates was performed with α-HA-antibody. (Bottom Left) 3x-HA-Cse4 (pAE977) did not interact with 6x-myc-Cac3 in AEY1558 cac1Δ. Precipitates were analyzed with α-HA-antibody. (B) Cse4 interacted with Asf1 in vivo. Therefore AEY2493 with genomic 3xHA-tagged Asf1 was transformed with 6x-myc-CSE4 (pAE901). For detection of precipitated Cse4 α-myc-antibody was used.

As Cac1 exists in the CAF-I complex together with Cac2 and Cac3 (Kaufman, et al., 1997), we asked whether Cse4 interacted with the whole CAF-I complex or [page 44↓]with Cac1 alone. By co-immunoprecipitation, we could exclude the possibility that Cse4 existed in a complex with Cac2 or Cac3 as no precipitation could be detected. Recently, (Sharp, et al., 2002) reported a function for CAF-I and Hir1 outside of S-phase at the centromere. Additionally, they found a genetic interaction between Cse4 and Cac1, as cac1Δhir1Δcse4-107 cells were unable to grow at semipermissive temperatures. Besides, Cac1 and Cac2 colocalized at the centromere, which would support our finding that the centromeric H3 variant Cse4 and Cac1 directly interact with each other.

We found that the chromatin assembly factor Asf1 was also able to interact with Cse4. This interaction might be important at earlier steps in the assembly pathway, e.g. by delivering Cse4 to CAF-I, which in turn incorporates Cse4 at the centromere. As Cse4 is able to interact with subunits of the Asf1 and CAF-I histone deposition factors, there may exist different steps or mechanisms in assembling Cse4 into chromatin. In contrast to cacΔhirΔ cells, cacΔasf1Δ cells showed a G2/M delay independent of spindle assembly checkpoint and segregated a reporter minichromosome at wildtype frequencies. This result indicates, that the centromere needs Hir1, but not Asf1 in the absence of CAF-I function, so that the HIR/ASF1 chromatin assembly pathway might be bifurcated at some point.

Since Cse4 interacted with Cac1 as well as with SAS-I, we next sought to determine the dependence of these interactions on SAS-I components and Cac1, respectively. As can be seen in Figure 9A, a deletion of SAS2 did not disturb the interaction between Cse4 and Sas4 or Cac1, respectively. This suggested that the interaction between Cse4 and Sas4 or Cac1 was stable and that the presence of Sas2 was not necessary for maintaining the association between Cse4, Sas4 and Cac1.

	Fig. 9: Consequences of different mutations on the interaction with Cse4. The antibodies for precipitation are indicated above the panel. (A) The interaction between HA-Cse4 (pAE977) and myc-Sas4 (pAE613) and myc-Cac1 (pAE614), respectively, in a sas2Δ strain (AEY1559). Cse4 was precipitated with α-myc-antibody and precipitates were immunoblotted using α-HA-antibody. (B) HA-Cse4 (pAE977) and Sas2 (pAE90) were still able to interact with each other in a sas4Δ strain (AEY2461), whereas the interaction between HA-Cse4 and myc-Cac1 (pAE614) was disrupted. The detection was carried out with α-HA-antibody. (C) The chromatin assembly factor is not necessary for the interaction between HA-Cse4 and Sas2 (pAE90) or myc-Sas4 (pAE613), respectively. The interaction was tested in a cac1Δ strain (AEY1808). HA-Cse4 was precipitated with α-Sas2 or α-myc, immunoblotting was performed with α-HA-antibody.

In contrast to a sas2Δ, a deletion of SAS4 had consequences on the integrity of the complex (Fig. 8B). Whereas Cse4 and Sas2 were still able to interact with each other in a sas4Δ strain, the association of Cse4 with Cac1 was disrupted. A deletion of SAS4 leads to a disruption of the SAS-I complex, because Sas2 and Sas5 are unable to interact with each other (Meijsing and Ehrenhofer-Murray, 2001). Thus, we hypothesize that either Sas4 or Sas5 were essential for maintaining the interaction between Cse4 and Cac1.

In contrast, the chromatin assembly factor Cac1 was not essential to maintain the interaction between Cse4 and the SAS-I complex (Fig. 8C). Sas2 as well as Sas4 were still associated with Cse4 in a cac1Δ strain, showing that the interaction between Cse4 and SAS-I was not affected by the presence or absence of Cac1.

The results from the two-hybrid assay and the co-immunoprecipitation experiments indicated, that Sas2 might indeed have a role at the centromere. As Sas2 is a histone acetyltransferase that acetylates histone H3 K14 and histone H4 K16, we hypothesized that the histone acetylation might be important for the centromere as well. One possibility is that Sas2 interacts with the histone H3 variant Cse4 because it is an additional acetylation target. We next asked, if a sas2 deletion, and therefore a missing acetylation, had any effect in combination with cse4 mutations at the centromere. As a CSE4 deletion is lethal, we used a temperature sensitive mutant for the subsequent experiment.

The cse4-103 strain, which has two amino acid exchanges in the histone-fold-domain (I156V, L193Q), was generated by random mutagenesis. This temperature sensitivity is due to the weakened interaction between cse4-103 and histone H4, because the L193Q substitution is predicted to disrupt the Cse4-H4 interface (Glowczewski, et al., 2000). We inserted both cac1 and sas2 single deletions as well as cac1 sas2 double deletions into the cse4-103-strain and investigated the temperature sensitivity at 23°C, 34°C and 37°C. At the permissive temperature all strains grew equally well, whereas at 37°C all cse4-103 mutants were unable to survive. At 34°C, the cse4-103 mutant strain grew poorly compared to wildtype (Fig. 10).

When SAS2 was deleted, the cse4-103 strain grew better at 34°C as compared to the single cse4-103 mutation. Perhaps a missing acetylation on histones H3, H4 or Cse4 bySas2 stabilizes the structure and function of the centromere. As Cse4 together with histone H4 forms stable (Cse4-H4)₂ tetrameres, one possible explanation is that the target of acetylation was histone H4, because the acetylation of free histone H4 at lysine 16 has previously been described (Sutton, et al., 2003). One hypothesis is that the missing acetylation on histone H4 lysine [page 48↓]16 in sas2Δ cells forms a more compact chromatin structure at the centromere, comparable to heterochromatin at silenced regions.

	Fig. 10: Partial suppression of the cse4-103 temperature sensitivity by sas2Δ. The strains were spotted in serial dilutions onto selective media and grown for 2 days at 34°C. Strains used in this assay were AEY1194 (wildtype, wt), AEY1781 (cse4-103 sas2Δ), AEY1162 (cse4-103), AEY2373 (cse4-103 cac1Δ) and AEY2374 (cse4-103 sas2Δ cac1Δ).

Thus, the cse4-103 sas2Δ strain is still temperature sensitive, but not to the same extent as the single cse4-103 mutation. Another possibility with a similar result would be if Cse4 also was a target for acetylation by Sas2.

A deletion of CAC1 in cse4-103 or cse4-103 sas2Δ strains did not lead to a change in temperature sensitivity. Interestingly, in cse4-103 sas2Δcac1Δ strains, growth at 34°C was not improved as in cse4-103 sas2Δ strains, but was comparable to the single cse4-103 mutation. (Sharp, et al., 2002) described a function for the chromatin assembly factor Cac1 and Hir1 at the centromere. In cac1Δ cells, the distribution of Cse4 was disturbed: additionally to the centromere, extra-centromeric localization of Cse4 could be observed. This could also be the reason for our observation that the temperature-sensitivity was not improved when CAC1 as well as SAS2 were deleted in the cse4-103 strain. Taken together, Cac1 might be important for the integrity and stability of the centromere in cse4 mutated strains.

Drugs like benomyl or nocodazole affect the correct formation of microtubuli. When a yeast strain has an additional mutation in a gene that is important for the correct formation of the centromere-kinetochore complex, it becomes sensitive and is unable to grow on plates with benomyl or nocodazole. We investigated if a deletion of ASF1 or components of the SAS-I and CAF-I complex were benomyl or nocodazole sensitive, which might point to a role during centromere-kinetochore formation. None of the investigated deletions resulted in an increased or decreased benomyl sensitivity. We deduced from our data that all examined genes were dispensable for the correct formation of the centromere-kinetochore complex.

(Sharp, et al., 2003) found that a deletion of SIR1 lead to a greater resistance to the microtubule-depolymerizing drug benomyl in a point mutated cse4-107 background. cse4-107 contains a single point mutation at L175F, and is predicted to be in close association with the histone fold domain of histone H4 (Glowczewski, et al., 2000). We were now interested whether a SAS2 deletion also had an effect on benomyl sensitivity in combination with a cse4-103 mutation, which is also thought to disrupt binding to histone H4. However, benomyl sensitivity was neither increased nor decreased. In light of our results [page 50↓]we concluded that the deletion of SAS2 did not cause such a dramatic effect on centromere-kinetochore assembly such that it could be monitored with microtubule-depolymerizing drugs.

Next to components of the SAS-I complex, Cse4 additionally interacted in co-immunoprecipitation experiments with chromatin assembly factors. Chromatin assembly factors like CAF-I and Asf1, but also associated proteins like Hir1 have distinct roles in the cell cycle. Apart from their functions in heterochromatic gene silencing, they also display various cell cycle dependent phenotypes.

A single deletion of cac1 or hir1 does not have any effects on cell cycle progression as can be seen in their FACS profiles or in assays monitoring chromosome loss (Sharp, et al., 2002). However, if yeast cells have an asf1 deletion or a cac1 hir1 double deletion, respectively, are delayed in G2-M phase in the cell cycle (Sharp, et al., 2002);(Tyler, et al., 1999) and have an increased rate of chromosome loss compared to wildtype cells (Le, et al., 1997);(Sharp, et al., 2002).

We were interested in examining the effect of a sas2Δ on cell cycle progression. In order to investigate if a deletion causes a delay in the cell cycle, we performed FACS analysis with wildtype, cac1Δ, hir1Δ, sas2Δ, asf1Δ, cac1Δhir1Δ, hir1Δsas2Δ, and asf1Δsas2Δ cells. As expected, cac1Δhir1Δ as well as asf1Δ cells displayed a G2-M arrest in the cell cycle. In contrast, neither a single sas2 deletion nor a sas2 hir1 double deletion caused any effect. When Sas2 and Asf1 were both missing, the cells displayed the same phenotypes as an asf1 deletion alone (data not shown).

Another method to investigate defects within the cell cycle is to monitor loss of plasmids or chromosomes. (Stoler, et al., 1995) constructed a strain where chromosome loss could be monitored. This strain is disomic for chromosome III and contains several markers on the additional cen 130-3 mutated chromosome. Loss of the original or the additional chromosome III could be followed by [page 51↓]mating assay (MAT a/MATα), FOA resistance (URA3/ura3-52) and the colour of cells (SUP11/ade2-101). We inserted several deletions in SB230 in order to analyze if single (sas2Δ, cac1Δ, hir1Δ, asf1Δ) or double deletions (sas2Δhir1Δ, sas2Δasf1Δ, sas2Δcac1Δ) of genes involved in histone acetylation and chromatin assembly lead to elevated chromosome missegregation and chromosome loss. None of the analyzed mutated strains resulted in elevated chromosome loss compared to wildtype, except for asf1Δ and asf1Δsas2Δ where the loss rate of the additional chromosome III was comparable.

Another method to investigate correct segregation is to monitor plasmid loss. In order to do so, we transformed sas2Δ, hir1Δ, cac1Δ single, double and triple mutated strains with a plasmid carrying SUP11 as a marker (pUN90). We were able to verify the elevated chromosomal instability in cac1Δhir1Δ cells that has already been reported by (Sharp, et al., 2002). A single SAS2 deletion or sas2Δ in combination with cac1Δ and hir1Δ did not cause a further increase in plasmid loss. In summary we concluded that a single or an additional SAS2 deletion had no detectable influence on cell cycle progression or chromosome and plasmid stability under the conditions tested here.

Ctf19 is a Cse4-interacting protein at the centromere that mediates the connection between CDEI and CDEIII in a complex with Okp1 and Mcm21 (Ortiz, et al., 1999). Ctf19 is nonessential, and mutations result in chromosome missegregation, increased benomyl sensitivity and accumulation in the G2/M phase of the cell cycle (Grienenberger, et al., 2002). Since Ctf19 and Cse4 interact by two-hybrid analysis, we asked whether a SAS2-deletion had an effect on their association.

	Fig. 11: Two-Hybrid interaction between Cse4 and Ctf19 in wt (L40c) and sas2Δ (AEY1695) cells. Yeast strains were plated on YM media and grown at 30°C for 1-2 days. The interaction between Cse4 and Ctf19 was tested with the β-galactosidase filter assay, where positive interactions became blue after incubation at 30°C over night. Picture courtesy of Uta Marchfelder.

In contrast to the wildtype situation, where Ctf19 and Cse4 interacted with each other, no association was found when SAS2 was deleted (Fig.11). This suggested that Sas2 had a role at the centromere, e.g. by stabilizing the cohesion between the two centromeric proteins Ctf19 and Cse4 via putative acetylation of Cse4.

As Sas2 is a histone acetyltransferase for histone H4 K16 and histone H3 K14 (Meijsing and Ehrenhofer-Murray, 2001);(Osada, et al., 2001);(Sutton, et al., 2003) and additionally interacts with the histone H3 variant Cse4, we asked whether Sas2 was also able to acetylate Cse4 in vivo. To determine the putative acetylation, we analyzed denatured protein extracts from wildtype and sas2Δ strains with α-acetyl-lysine-antibodies. If Sas2 acetylated Cse4, we expected to be able to precipitate Cse4 in the wildtype protein extract but less or none in the sas2Δ protein extract. It was important to denature the proteins, because otherwise Cse4 might be precipitated in a complex together with histone H4, which would lead to a signal due to histone H4’s acetylation pattern.

	Fig. 12: Immunoprecipitation of HA-Cse4 with α-acetyl-lysine-antibody. HA-Cse4 was isolated with α-HA-antibody from whole cell protein extracts from wildtype (AEY2661) and sas2Δ (AEY2666) cells. The concentrated HA-Cse4 was then incubated with α-acetyl-lysine-antibody and detected via immunoblotting with α-HA-antibody.

Furthermore, to reduce the complexity of the α-acetyl-lysine precipitation, we isolated the protein extracts and performed a first immunoprecipitation with α-HA-antibody to isolate Cse4 from the extract. This additional immunoprecipitation step was necessary to partially purify Cse4 for the second immunoprecipitation with α-acetyl-lysine-antibody. After washing off the other proteins, we removed the bound Cse4 with Bead Buffer containing 1 % SDS that was then diluted to 0.1 % SDS for the subsequent second immunoprecipitation [page 54↓]step with α-acetyl-lysine-antibody (Upstate). The precipitated Cse4 was detected in an immunoblot with α-HA-antibody.

As shown in Figure 13, the amount of HA-Cse4 applied was comparable in both wildtype and sas2Δ protein extracts (Input). After immunoprecipitation with α-acetyl-lysine-antibody, HA-Cse4 could be detected in wildtype and sas2Δ extracts, whereas the control without antibody (-Ab) did not show a HA-Cse4 specific signal. We concluded that an acetylated Cse4 form existed in the cell, although differences in the intensity of the signal from wildtype and sas2Δ extracts could not be detected. As the signal varied in independent experiments, we cannot rule out the possibility that Cse4 might only be temporarily acetylated during the cell cycle, so that the acetyl group is removed at some point. Furthermore, an associated histone deacetylase might be co-purified in the assay. In this case, the acetyl-group from acetylated Cse4 could be removed during the purification, so that in consequence we were unable to detect Cse4. Additionally, it might still be possible that after removing the bound HA-Cse4 with 1% SDS from the α-HA-antibody, protein complexes containing Cse4 and histone H4 are restored upon dilution of SDS to 0,1%. In this scenario, the precipitation of HA-Cse4 with the α-acetyl-lysine-antibody may be via acetylated histone H4.

Because it was difficult to determine if Cse4 is acetylated in the cell with immunoprecipitation experiments, we also took advantage of another independent method to determine a putative Cse4 acetylation via Sas2: an in vitro acetylation assay. For this purpose, we used the bacterially expressed and purified SAS-I-complex (pAS134, plasmid and purification as described by (Sutton, et al., 2003) and Cse4. The advantage of this method was that large quantities of purified proteins could be purified from bacteria. On the other hand, co-factors or other modifications that are important for the reaction might be missing.

For the in vitro acetylation assay we used the whole purified recombinant SAS-I complex, because (Sutton, et al., 2003) reported that Sas4 and Sas5 are essential and important for Sas2’s acetylation activity. The same experiments were performed with the TAP-purified SAS-I complex from yeast (data not shown). As a positive control we applied histone H4 alone or in the same reaction with His-Cse4, so that the functionality of the assay could be monitored.

	Fig. 13: Acetylation assay with the purified and recombinant His-Cse4 and the SAS-I complex. SAS-I and His-Cse4 were purified from bacteria, whereas histone H4 and PCAF were commercially available. In this assay 2 μg substrate (histone H4, Cse4) was added to 200 ng SAS-I and 500 ng PCAF, respectively, and mixed together with 0,5 μg [14C] acetyl-CoA. After 1 h incubation the samples were loaded onto a 15 % SDS-gel, which was analyzed after the run in a phosphoimager. Figure courtesy of Jacqueline Franke.

As can be seen in Figure 13, PCAF as well as the SAS-I complex were functional and able to acetylate recombinant histone H4 (lane 1+2). Additionally, self-acetylation of components from the SAS-I complex could be detected in lane 2. When His-Cse4 was added to PCAF in the assay (lane3), no Cse4-specific acetylation signal could be detected. The signals obtained in lanes 3 and 4 could [page 56↓]not be assigned to Cse4, because they didn’t match His-Cse4 in Coomassie blue stained SDS-PAGE gels and immunoblots using an α-His antibody, respectively. To be sure that the enzymes in this assay were still functional in the presence of His-Cse4, we added His-Cse4 as well as histone H4 in the same reaction and could indeed find acetylated histone H4, although the acetylation signal was weaker than in lane 2 (lane 4). Next to the identified signals from histone H4, additional unidentified signals were obtained (lane 3+4). These signals may result from co-purified proteins that bound to His-Cse4, and were also an acetylation target of PCAF.

The main question of this assay was, whether the SAS-I complex was able to acetylate the histone H3 variant Cse4. Thus, we incubated SAS-I together with purified His-Cse4, but we were unable to detect acetylated His-Cse4 with the applied concentrations, even when self-acetylation of the SAS-I complex could be seen (lane5). The addition of histone H4 to the sample lead to its acetylation, but acetylated His-Cse4 was still not detectable (lane 6).

Additionally to the purified His-Cse4 from bacteria, we used a Cse4 peptide for our in vitro acetylation assays. As Cse4 contains several putative acetylation sites (9 lysines in the N-terminus, 7 lysines in the C-terminal histone fold domain), we chose a lysine rich region (aa 112-134 with K115, 119, 122, 126, 130, 131) within the N-terminus for this purpose. As Cse4 interacted via its N-terminus with the acetyltransferase Sas2, we hypothesized to find a putative acetylation site in this region, but we were unable to find a Cse4 specific acetylation signal with SAS-I in the in vitro assay (data not shown).