A Protein Complex (SWR1-Com) Associated with the Histone Variant H2A.Z
H2A.Z is important for specifying euchromatic regions in the genome of S. cerevisiae
(Meneghini et al. 2003
). To determine which other proteins contribute to directing H2A.Z to its chromosomal locations, we purified proteins associated with a soluble pool of H2A.Z from whole cell extracts of a yeast strain harboring an allele of HTZ1
that encodes a carboxyl-terminal fusion to the tandem affinity purification (TAP) tag (see Materials and Methods
) (Rigaut et al. 1999
). These initial purifications were performed under low salt conditions and with limited wash steps to maximize protein complex recovery, with more stringent conditions used subsequently to distinguish strong from weak and potentially artifactual interactions (see below). The protein compositions of the samples were determined using Direct Analysis of Large Protein Complexes methodology, which consists of tryptic digestion of the mixture, multidimensional microcapillary chromatography, tandem mass spectrometry, and genome-assisted analysis of the acquired spectral data (Link et al. 1999
; Sanders et al. 2002
). A protein was judged to be associated with H2A.Z if the number of corresponding peptides in the H2A.Z-TAP purified material was higher than in the material purified from strains lacking a tagged H2A.Z protein and if the protein passed additional criteria described below. Proteins established to nonspecifically copurify with TAP-tagged proteins were excluded from the analysis (Shevchenko et al. 2002
). Using these criteria and additional purifications (described below), we identified 15 proteins associated with H2A.Z (), of which 13 form a complex designated SWR1-Com ( and see below). The largest subunit corresponded to Swr1p (Swi2/Snf2-related), an uncharacterized member of the Swi2/Snf2 family of ATP-dependent chromatin remodeling enzymes (Pollard and Peterson 1998
Peptides in TAP Purifications
Subunit Architecture of SWR1-Com and Overlap with NuA4 and Ino80-C Complexes
SWR1-Com Shared Subunits with the Essential HAT NuA4 and the Ino80-C Chromatin Remodeling Complex
Four SWR1-Com subunits that are also found in the Ino80-C chromatin remodeling complex are Rvb1p, Rvb2p, Arp4, and actin (Shen et al. 2000
). Similarly, Yaf9p, as shown below and by others (Le Masson et al. 2003
), as well as Arp4p and actin, are also components of the NuA4 HAT (Galarneau et al. 2000
). To determine whether the proteins that associated with H2A.Z formed one discrete complex, multiple complexes, or were copurifying contaminants, three of these proteins were themselves tagged with the TAP domain. Complexes from the soluble fraction of whole cell extracts were purified in conditions similar to those used for the H2A.Z-TAP purification, and the composition of the purified material was evaluated by the same procedure used for the H2A.Z-TAP (summarized in and ). With the exception of the histone chaperone Nap1p and the import protein Kap114p, the proteins that copurified with TAP-tagged Swr1p were similar to the set found with H2A.Z, except that two additional proteins, designated here as Swc3p and Swc7p, were identified. Similarly, purifications from strains with TAP-tagged Swc4p and Yaf9p yielded nearly all the proteins associated with Swr1p and H2A.Z and lacked Nap1p and Kap114p. Like the Swr1-TAP material, the Swc4-TAP-associated material contained Swc3p and Swc7p, supporting the assignment of these two proteins to SWR1-Com. TAP-tagged Swc4p and Yaf9p also associated with most of the subunits of the NuA4 complex (including Tra1p, Epl1p, Eaf3p, Yng2p, and the catalytic subunit Esa1p). These data suggested that SWR1-Com and NuA4 shared the Yaf9p, Swc4p, Arp4, and actin subunits.
Representative complex purifications under high stringency conditions (see Materials and Methods
) from strains with either the Swr1-TAP, Yaf9-TAP, or an untagged control strain are shown in A. Proteins that copurified with both Swr1-TAP and Yaf9-TAP represented subunits of SWR1-Com (A, arrows), whereas proteins that only copurified with Yaf9-TAP represented specific NuA4 subunits (A, stars). A schematic representation of the domain structures of SWR1-Com subunits is presented in B. Several of the proteins in the complex contained motifs (SANT, Bromo, YEATS, and HIT) found in proteins associated with chromatin, suggesting that SWR1-Com acts directly on chromatin.
SWR1-Com Shared Subunits with NuA4 and Contained Proteins with Motifs Involved in Chromatin Biology
Since the histone chaperone Nap1p and the import factor Kap114p copurified with H2A.Z but not other members of the complex, they were likely to be part of an H2A.Z-containing protein complex distinct from SWR1-Com. Affinity purification of TAP-tagged Rvb2p, an established component of the Ino80-C chromatin remodeling complex, yielded peptides corresponding to the other known subunits of the Ino80-C complex as well as six members of SWR1-Com, three of which (Swc4p, Arp4p, and actin) are also subunits of NuA4 (Table S1
). Consistent with the assignment of SWR1-Com subunits, a percentage of the cellular pool of these proteins cosedimented with each other upon glycerol gradient centrifugations of whole cell extracts (Figure S1
The initial purifications suggested that Swc4p and Bdf1p, both of which have domains that are involved in recognition of histone tails, copurified with H2A.Z-TAP and might be part of SWR1-Com. Independent assessment of the composition of the complexes deduced by mass spectrometry was obtained by analytical-scale affinity purifications of Yaf9-TAP, Esa1-TAP, Rvb2-TAP, Swr1-TAP, and Ino80-TAP from cells containing a version of Swc4p that was fused at its carboxyl-terminus to a triple hemagglutinin (HA) tag. These analytical-scale affinity purifications were more stringent than the initial TAP purifications and therefore served to eliminate false-positive results and to provide independent tests of interactions. Anti-HA epitope antibodies and antibodies against Tra1p, the largest subunit of NuA4, were used to analyze the copurified material. Both Yaf9-TAP and Esa1-TAP associated with comparable amounts of Tra1p and Swc4p, supporting the assignment of Yaf9p and Swc4p as new subunits of NuA4. Likewise, Rvb2-TAP and Swr1-TAP copurified with a substantial amount of Swc4-HA, but Ino80-TAP did not. Rvb2-TAP, Swr1-TAP, and Ino80-TAP were not associated with Tra1p (A). These data were consistent with Swr1p and Rvb2p being components of SWR1-Com and not of NuA4. Further supporting the assignment of Swc4p as a subunit of NuA4, significant amounts of the NuA4 subunits Tra1p and Esa1p were present in material from Swc4-TAP analytical-scale purifications (B).
Swc4p and Bdf1p Were Components of SWR1-Com
The number of peptides corresponding to Bdf1p in the TAP purifications was low, and Bdf1p was found only in the H2A.Z-TAP and the Yaf9-TAP (). Bdf1p's potential presence was tested further by additional analytical-scale affinity purifications from strains carrying Yaf9-TAP, Swr1-TAP, Swc4-TAP, and Esa1-TAP and immunoblotting with an antibody against Bdf1p. Bdf1p associated with Swr1-TAP, Yaf9-TAP, and Swc4-TAP but not with Esa1-TAP or untagged control material, supporting the assignment of Bdf1p as a subunit of SWR1-Com (C).
SWR1-Com Selectively Associated with Histone H2A.Z Versus H2A
To determine whether subunits of SWR1-com associated specifically with H2A.Z or both H2A.Z and H2A, TAP-tagged versions of H2A.Z and H2A were purified from cells containing HA-tagged versions of the five different SWR1-Com components Swr1p, Swc2p, Swc3p, Swc4p, and Swc7p, and the nuclear import factor Kap114p. The composition of the copurifying material was then evaluated by immunoblotting with antibodies against the HA tag, Bdf1p, and Tra1p. Yaf9-TAP served as a positive control for recovery of SWR1-Com and NuA4.
H2A.Z associated with a substantial fraction of the SWR1-Com as judged by the comparable intensity of the signal for SWR1-Com subunits in the material copurified with H2A.Z-TAP and Yaf9-TAP, whereas no NuA4 copurified with H2A.Z-TAP based upon the absence of Tra1p in the H2A.Z-TAP sample (A). In contrast, histone H2A copurified with only trace amounts of Swc2-HA, Swc3-HA, Swc4-HA, Swc7-HA, and Bdf1p and with virtually no Swr1-HA (see ). Kap114-HA associated with both H2A.Z-TAP and H2A-TAP but not Yaf9-TAP, suggesting that it did not discriminate canonical and variant H2A. Hence, based on the apparent relative strength of the interactions, SWR1-Com (in contrast to Kap114-HA) associated primarily with H2A.Z, although weak affinity of SWR1-Com to H2A was possible. Furthermore, these experiments also supported the assignment of Swc3p and Swc7p to SWR1-Com despite peptides for these two proteins being present only in the initial Swr1-TAP and Swc4-TAP purifications.
SWR1-Com Associated Selectively with H2A.Z and Contained H2B
Canonical H2A that is not bound to chromatin is usually found in a H2A/H2B dimer (Jackson 1987
). The presence of H2B in the SWR1-Com was investigated by purifying SWR1-Com from strains containing H2A.Z-HA and H2B-HA. SWR1-Com contained H2A.Z-HA and also H2B-HA (B). These data raised the possibility that this complex used H2A.Z/H2B dimers as a substrate.
Similar Gene Expression Profiles of htz1Δ and swr1Δ Cells
To determine the extent to which the role of H2A.Z depends upon SWR1-Com function, genome-wide transcription profiles of swr1
Δ cells were compared to the profiles of htz1
Δ cells (Meneghini et al. 2003
). To permit an optimal comparison, experiments were performed under the conditions used previously to analyze htz1
Δ cells (see Materials and Methods
). Due to the role of H2A.Z in anti-silencing, H2A.Z-dependent genes tend to be located near silenced domains such as telomeres. This theme was echoed in the results from the swr1
Δ mutant. Specifically, 42 of the 94 (45%) Swr1p-dependent genes were within 20 kb of a chromosome end, which is less than 3% of the genome (A). This enrichment is highly significant, as judged by p
-values estimated from the hypergeometric distribution (B). Swr1p-dependent genes were underrepresented from regions more than 40 kb from a telomere, suggesting that, as seen earlier for H2A.Z, the telomere-proximal genes were most sensitive to loss of Swr1p function.
Chromosomal Distribution of Swr1p-Activated Genes
Comparison of the transcript profile across the genome of swr1Δ cells to that of htz1Δ cells also revealed a marked overlap (A). Ninety-four genes displayed reduced expression in the swr1Δ mutant compared to wild type. Of these 94 Swr1p-dependent genes, 64 were also reduced in expression in htz1Δ (A). This remarkably large overlap is highly significant (p = 2.9 × 10−80, calculated using the hypergeometric distribution) and even more impressive for telomere-proximal genes. These data suggested that Swr1p and H2A.Z shared a common function in regulating gene expression.
Relationship of Genes Activated by Swr1p or H2A.Z
A substantial number of H2A.Z-dependent genes (116) did not appear to require Swr1p for expression. A color representation of the swr1Δ and htz1Δ datasets grouping the genes described in A (B) revealed that a subset of these appeared to have mildly reduced expression levels in swr1Δ cells but were not reduced enough to meet the stringent significance cutoff. However, there also were clear examples of genes that required H2A.Z for expression but not Swr1p. Likewise, there were clear examples among the 94 genes that required Swr1p for expression but did not require H2A.Z ().
Swr1p Was Required for H2A.Z Deposition In Vivo
The evidence linking H2A.Z and Swr1p function and the association of both H2A.Z and H2B with SWR1-Com suggested that SWR1-Com was responsible for depositing H2A.Z onto chromatin in vivo, perhaps in the form of an H2A.Z/H2B dimer. (The Swr1p relatives in the ACF and RSF complexes perform related roles in assembling chromatin in vitro (reviewed in Haushalter and Kadonaga 2003
). If so, then cells lacking Swr1p should display reduced levels of H2A.Z in chromatin.
To test this prediction, we performed chromatin immunoprecipitation (ChIP) experiments comparing wild type to swr1
Δ strains expressing a functional amino-terminal triple-HA-tagged version of H2A.Z expressed from the HTZ1
promoter at the normal chromosomal locus (HA3-H2A.Z). Consistent with H2A.Z being in a stable complex with Swr1p that protected it from protein degradation, the level of HA3-H2A.Z in swr1
Δ strains was reduced approximately 2- to 3-fold (Figure S2
). To normalize the signals from each experimental locus assayed, we measured the levels of DNA derived from a control locus whose expression is H2A.Z independent (the middle of the PRP8
open reading frame [ORF]) in samples derived from each whole cell extract and precipitate (see Materials and Methods
). We first examined HA3-H2A.Z levels at two chromosomal regions where H2A.Z prevents the spread of Sir-dependent silencing: one flanking the silent mating locus HMR
and another near the telomere on the right arm of chromosome XIV (A). In wild type, H2A.Z was present at levels similar to those described previously (B); HA3-H2A.Z was depleted from the silenced HMR
locus and enriched in the flanking euchromatic regions (B). In addition, HA3-H2A.Z was depleted from the most telomere-proximal locus tested, AAD3,
presumably because of telomeric silencing of this gene. Likewise, HA3-H2A.Z was highly enriched at the YNR074C
gene, a telomere-proximal gene on chromosome XIV strongly protected from silencing by H2A.Z.
ChIP Analysis of HA3-H2A.Z Deposition in the HMR Region and Near the Right Telomere of Chromosome XIV
In striking contrast, in swr1Δ cells, the enrichment (relative to the PRP8 locus) of HA3-H2A.Z at every locus tested approached a ratio of one (C). These data were consistent with Swr1p being essential for the deposition of H2A.Z in the HMR region and near the right telomere of chromosome XIV. However, because the ChIP measurements were normalized to the PRP8 locus, we considered the possibility that a uniform amount of HA3-H2A.Z remained at all chromosomal loci examined in the mutant, for instance if there was a specific increase in the association of HA3-H2A.Z with the PRP8 locus rather than a decrease at all other loci. This possibility was discounted by the approximately 13-fold mean decrease in the ratio of DNA obtained from the pellet versus whole cell fractions in swr1Δ cells relative to wild type, indicating a substantial defect in the absolute chromatin association of HA3-H2A.Z in cells lacking Swr1p.
H2A.Z is also deposited at several loci that are not near silenced regions (Meneghini et al. 2003
). The function of H2A.Z at these regions is unknown. To determine if Swr1p was also required for H2A.Z deposition at such loci, we examined H2A.Z levels at 12 euchromatic regions on chromosome III that each displayed some level of deposition of HA3-H2A.Z (A). These loci were identified in a comprehensive study of H2A.Z deposition on chromosome III (M.D.M., M. Bao, H.D.M., unpublished data). As with the regions examined above, the relative ChIP enrichment of HA3-H2A.Z approached one at each of these loci in the absence of Swr1p (B), and the absolute amount of DNA precipitated from these loci showed a large decrease in the swr1
Δ mutant (data not shown). Thus, Swr1p was broadly required for the deposition of HA3-H2A.Z, even in regions distant from silenced domains. It is worth noting that for several of the loci examined in the swr1
Δ mutant, the ChIP enrichment was significantly less than one, suggesting that there may exist some residual deposition of HA3-H2A.Z at the PRP8
locus in these cells.
ChIP Analysis of H2A.Z Deposition at Nontelomeric Euchromatic Sites
NuA4 Function Was Required for SWR1-Com to Support Cell Growth
The sharing of four proteins between the SWR1-Com and NuA4 (see ) raised the possibility that SWR1-Com was functionally linked to NuA4, which is the major HAT for histones H4 and H2A. Initial purifications from Yaf9-TAP and Swc4-TAP strains suggested that the protein encoded by the nonessential gene YDR359C
was a subunit of NuA4, consistent with some earlier results from high-throughput studies (Gavin et al. 2002
). Initial efforts to fuse a triple HA-tag to the carboxy-terminus of YDR359C
were unsuccessful, but we noticed that all proteins encoded by YDR359C
orthologs from the Saccharomyces
sensu strictu strains had a carboxy-terminal extension of approximately 22 amino acids, suggesting a possible error in the S. cerevisiae
sequence. Therefore, we chose to integrate a triple HA-tag at the chromosomal location that corresponded to the second-to-last codon of the sensu strictu strains and found that this version was now successfully tagged. Recently, a revised copy of YDR359C
with the stop codon at the location we chose was deposited in GenBank and named EAF1
. Therefore, we used this name here rather than a previously assigned name for the shorter version of YDR359C.
Eaf1p contained a SANT domain as well as an HSA domain that is associated with SANT domains and found in helicases (Letunic et al. 2002
Analytical-scale affinity purifications showed that Eaf1-HA, similar to Tra1p, copurified with Yaf9-TAP, Swc4-Tap, and Esa1-TAP but not with H2A.Z-TAP and Swr1-TAP (A). In addition, global H4 acetylation defects were evident in eaf1Δ but not in htz1Δ, yaf9Δ, or swr1Δ cells (B). This was based on the reduced signal in immunoblot experiments obtained with an antibody directed against tetra-acetylated H4. Similar experiments with antibodies directed against individual acetylated residues revealed that the H4 acetylation defect of strains lacking EAF1 was most profound on K8 and K12 of H4, whereas K5 and K16 of H4 were less affected (B). Similarly, K9 of H2A did not have a strong acetylation defect in any of the mutants (B). The physical association and the H4 acetylation defects provided independent evidence that Eaf1p was a subunit of NuA4.
Eaf1p Was a Subunit of the NuA4 HAT
To explore genetic links between SWR1-Com and NuA4, phenotypic and double mutant analyses were performed. SWR1-Com mutants and the strain lacking EAF1 shared sensitivities to genotoxic and stress conditions (A). The eaf1Δ strains were also slow growing whereas the other strains were not. All mutant strains tested were sensitive to the DNA replication inhibitor hydroxyurea (HU) and the microtubule poison benomyl and to caffeine and formamide, reagents that elicit a number of cellular responses (A). Strains lacking HTZ1 and YAF9 were comparably sensitive to HU and formamide, but htz1Δ strains were more sensitive to benomyl and caffeine. Strains lacking SWR1 were less sensitive than the other strains to HU and formamide, but the sensitivity to caffeine and benomyl was comparable to that of yaf9Δ strains (A). Cells lacking the NuA4 subunit Eaf1p were most sensitive to HU and caffeine, and their sensitivity to benomyl and formamide was comparable to that of htz1Δ mutants (A). While the severity of the defects varied, the similar phenotypes of mutants in SWR1-Com and NuA4 suggested that the two complexes were broadly required for resistance to DNA damage and genotoxic stress.
NuA4 and SWR1-Com Shared Similar Phenotypes and Interacted Genetically
To test whether the sensitivity to DNA damage and genotoxic stress was a shared function of SWR1-Com and NuA4, or whether these sensitivities were caused by independent functions, double mutant analysis was performed using the EAF1 gene as an exemplary NuA4 subunit. No viable spores were obtained that had deletions of EAF1 and HTZ1, SWR1, or YAF9 (B). Thus, SWR1-Com and NuA4 interacted genetically, and the two complexes shared at least one essential function.