Hif1 is a component of yeast histone acetyltransferase B, a complex mainly localized in the nucleus.

Hat1 is the catalytic subunit of the only type B histone acetyltransferase known (HAT-B). The enzyme specifically acetylates lysine 12, and to a lesser extent lysine 5, of free, non-chromatin-bound histone H4. The complex is usually isolated with cytosolic fractions and is thought to be involved in chromatin assembly. The Saccharomyces cerevisiae HAT-B complex also contains Hat2, a protein stimulating Hat1 catalytic activity. We have now identified by two-hybrid experiments Hif1 as both a Hat1- and a histone H4-interacting protein. These interactions were dependent on HAT2, indicating a mediating role for Hat2. Biochemical fractionation and co-immunoprecipitation assays demonstrated that Hif1 is a component of a yeast heterotrimeric HAT-B complex, in which Hat2 bridges Hat1 and Hif1 proteins. In contrast to Hat2, this novel subunit does not appear to regulate Hat1 enzymatic activity. Nevertheless, similarly to Hat1, Hif1 influences telomeric silencing. In a localization analysis by immunofluorescence microscopy on yeast strains expressing tagged versions of Hat1, Hat2, and Hif1, we have found that all three HAT-B proteins are mainly localized in the nucleus. Thus, we propose that the distinction between A- and B-type enzymes should henceforth be based on their capacity to acetylate histones bound to nucleosomes and not on their location within the cell. Finally, by Western blotting assays, we have not detected differences in the in vivo acetylation of H4 lysine 12 (acK12H4) between wild-type and hat1Delta, hat2Delta, or hif1Delta mutant strains, suggesting that the level of HAT-B-dependent acK12H4 may be very low under normal growth conditions.

Acetylation of specific lysine residues in the N termini of nucleosome core histones is a reversible and dynamic process that occurs in all eukaryotic organisms studied and has been correlated with several essential processes, including modulation of gene expression and nucleosome assembly (reviewed in Refs. [1][2][3][4]. The current view of the mechanism by which acetylation of histones influences different biological processes is that it acts as a signaling system, actively participating in the recognition of and in the interaction with regulatory proteins or other complexes (5)(6)(7)(8)(9). Specific acetylation patterns on the histones are thought to act as recognition sites for specific proteins. It is well established that most HAT 1 enzymes have a marked preference with regard to target histone as well as lysine residue(s). Moreover, all of the native HATs are high molecular mass multiprotein complexes in which the different components are able to modulate substrate preferences (1,4,10).
On the basis of structural similarities, histone acetyltransferases are at present classified into distinct families found in eukaryotes from yeast to mammals (1,4). In addition, on the criteria of subcellular localization, histone preference, and ability to modify nucleosomal histones, native HATs have also been classified into two different groups. Type A HATs (HAT-A) are nuclear enzymes able to acetylate the N termini of the core histones after their assembly into nucleosomes and chromatin and are usually associated with transcription-related acetylation events, whereas type B HATs (HAT-B) are thought to be primarily cytoplasmic enzymes that acetylate only free histones, presumably prior to their deposition on DNA to form nucleosomes. In yeast nine different proteins with HAT activity have been described to date. Eight of them are the catalytic subunit of HAT-A complexes, with physical or functional connections to transcriptional regulation as follows: Gcn5 (11), Elp3 (12), and HpaII (13) are members of the GNAT superfamily (for Gcn5 related N-acetyltransferase); Esa1 (14), Sas2 (15), and Sas3 (16) are HATs of the MYST family (for MOZ, YBF2/ Sas3, Sas2, and Tip60); and Taf II 145 (17) and Nut1 (18) which are, respectively, components of the basal transcription factor TFIID and a subunit of mediator (reviewed in Ref. 1). The ninth yeast protein with HAT activity is another member of the GNAT superfamily, Hat1 (19), an enzyme that represents the prototypical catalytic subunit of HAT-B complexes and therefore is assumed to have a functional link to histone deposition and chromatin assembly (20). In fact, Hat1 and its homologs from different organisms (21)(22)(23)(24) are the only type B HATs identified at present.
The presence of a type B HAT in yeast was first described in our laboratory as an enzymatic activity obtained from a cytoplasmic fraction with a molecular mass of 130 kDa, specific for histone H4 but unable to accept nucleosomal histones as sub-strates (25,26). Subsequently, the gene for the catalytic subunit of this HAT was identified (19,20). Extensive biochemical characterization of the HAT-B enzyme from yeast cytosolic extracts showed that this complex is composed of two subunits, Hat1 as the catalytic subunit and Hat2, a yeast homolog of the mammalian proteins Rbap46 and Rbap48. The molecular mass obtained for this native enzyme (Ͼ200 kDa) suggests a subunit composition more complex than a simple heterodimer consisting of one Hat1 and one Hat2 subunit. The complex specifically acetylated lysine 12 of free histone H4, whereas recombinant Hat1 extended its specificity to lysines 5 and 12 of free H4 and, at a lesser extent, to H2A (19,20). Comparative studies of the HAT-B activity recovered from wild-type and hat2⌬ strains showed that in the absence of Hat2 the specific activity of the enzyme is ϳ10-fold lower than that from wild-type cells, whereas the specificity toward lysine 12 of H4 remains unchanged. Moreover, the presence of Hat2 was required in vitro for stable binding of Hat1 to the histone H4 tail (20).
There are several reasons to attribute a role to the HAT-B complex in the acetylation of histones prior to their assembly in nucleosomes. First, HAT-B is unable to acetylate histones when they are in nucleosomes. Second, it has been reported that in the cytoplasm Hat1 and Hat2 proteins together with acetylated forms of histones H4 and H3 interact with the karyopherin Kap123, a receptor implicated in the nuclear import of H4 and H3 (27). Finally, there is a correlation between the acetylation pattern of newly synthesized histone H4 and the in vitro site specificity of HAT-B complexes. Thus, in all eukaryotic organisms where it has been studied, newly synthesized and deposited nucleosomal histone H4 was shown to be diacetylated on lysine residues 5 and 12, and this non-random pattern of acetylation is highly coincidental to that generated by biochemically isolated native HAT-B enzymes (20, 22, 23, 28 -32). Although there is not a strict pattern of acetylation, newly synthesized histone H3 also seems to be involved in chromatin assembly (33,34). Nevertheless, at present no cytoplasmic type B HAT specific for free histone H3 has been described. Despite all these data, it has been demonstrated by specific substitution mutations at the N-terminal domain of histone H4 that lysine residues 5 and 12 are not required for nucleosome assembly in vitro or in vivo (35). Moreover, the human histone chaperone chromatin assembly factor-1, a protein complex that is thought to selectively target newly synthesized H3 and H4 histones to nascent DNA, binds H3-H4 tetramers lacking their N-terminal tails (36). Thus, although the acetylation of newly synthesized histones H3 and particularly H4 seems to be conserved, its functional significance remains unknown. In this sense, it is not surprising that the search for a direct phenotype for yeast cells lacking of HAT1 or HAT2 genes has proven elusive. Only recently it has been demonstrated that deletion of HAT1 or HAT2, in combination with specific substitution mutations in the N-terminal tail of histone H3, results in a significant defect in telomeric silencing (37). Furthermore, a similar combination of the HAT1 deletion and particular lysine residue substitution H3 mutations resulted in sensitivity to DNA-damaging agents, demonstrating that Hat1 contributes to the recombinational repair of DNA doublestrand breaks (38). How can an enzyme such as HAT-B (consisting of Hat1 and Hat2) traditionally localized in the cytoplasm play a role in nuclear events such as telomeric silencing or DNA repair? Certainly, biochemical studies on different organisms have previously found type B HATs in cytosolic fractions (20,22,26,29,31,32,39,40). However, in distinct species Hat1 has also been detected in the nucleus at specific physiological stages, such as the Xenopus oocyte maturation (23), maize germination (24), or S-phase of human cells (22).
Moreover, by using a combination of genetics and biochemical approaches, we have detected a yeast nuclear HAT complex, namely HAT-A3, that contains Hat1 and Hat2 (40). These reports indicate that at least some of Hat1 (and Hat2) is located in the nucleus, possibly associated with other modulating proteins, suggesting that HAT-B complexes may have any other functional roles in addition to marking histones for deposition, although an indirect effect cannot be ruled out. In any case, it seems that some degree of uncertainty has been generated by the absence of a more systematic analysis on the cellular compartmentalization of the type B enzyme.
In order to shed some light on these issues, we first undertook a search for proteins, other than Hat2, that bind in vivo to Hat1 and to the N-terminal tail of histone H4. By using the yeast two-hybrid system, we found a novel protein, Hif1, that is able to interact with both Hat1 and histone H4. Biochemical analyses demonstrate that Hif1 is indeed a member of HAT-B complex in yeast, and hence we have also investigated the involvement of Hif1 in telomeric silencing. In addition, our studies on the subcellular localization of Hif1, Hat1, and Hat2 proteins have shown clearly that the HAT-B complex is mainly nuclear, contrary to the current view. Finally, by Western blotting experiments, we have not found a dependence of the acetylation of histone H4 lysine 12 with HAT-B proteins in vivo.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Media-Yeast strains used in this study are listed in Table I. Gene knockouts with kanMX4, natMX4, and his5 ϩ were performed as described (41)(42)(43). Deletions of the desired loci were verified by PCR on genomic yeast DNA. The hat2::HIS3 mutation in strains YSTT11 and YSTT62 is a deletion-insertion in which 255 bp upstream through nucleotide 804 of the open reading frame of HAT2 were deleted and replaced by HIS3 (originally constructed by P. Kaufman). This deletes the sequence for the first 268 amino acids of the Hat2 protein.
Genomic ORFs corresponding to Hat1, Hat2, and Hif1 proteins were tagged at their C termini with epitope tags by targeted integration using a PCR-based strategy (44). HAT1 and HAT2 were tagged with a sequence encoding six copies of the HA epitope followed by Kluyveromyces lactis TRP1 or Schizosaccharomyces pombe HIS3 as selectable markers. HIF1 was tagged with nine tandem repeats of the Myc epitope followed by the TRP1 marker. DNA fragments, with flanking sequences homologous to the desired gene, were amplified from the appropriated template plasmids (a gift from P. M. Alepuz; see Ref. 44). Correct homologous integration was verified by PCR on isolated genomic DNA. Transformants were also assayed for expression of the desired tagged protein by Western blotting.
The bait plasmid pSTT20, encoding LexA-Hat1, was constructed by cloning a PCR fragment with the entire Hat1 coding sequence into the EcoRI-XhoI sites of a derivative of the vector, pBTM116 (45). The bait plasmid pGBD-H4-(1-59) and plasmid libraries used in the two-hybrid screening were described previously (46). The HIF1 gene was cloned from the genome into a derivative of pBR322 to create pSTT40. This was used to construct plasmid pSTT65, which contains the sequence coding for residues 8 -385 of Hif1 fused in-frame with GAD in the vector pGAD424.
Growth of yeast cells was performed on standard yeast extract/ peptone/dextrose (YPD) or synthetic complete (SC) media supplemented as required. Manipulation of cells and media preparation were performed according to standard procedures.
Two-hybrid Analyses-Two-hybrid screening with the bait plasmid pSTT20 (pLexA-Hat1) was performed in strain L40 as described previously (47). Two-hybrid screening with the bait plasmid pGBD-H4-(1-59) was carried out with the strain PJ69-4A (48) as described previously (46). DNA of pGAD plasmids from those positive clones that arose on a medium lacking adenine and histidine and having ␤-galactosidase activity were purified and amplified. The presence of an ORF fused in-frame to GAD was confirmed by DNA sequencing. Effect of a HAT2 deletion on the H4-(1-59) two-hybrid interactions was carried out with YTH2, a hat2⌬ derivative of PJ69-4A.
Preparation of Enzymatic Extracts and Fractionation of HAT Complexes-Whole-cell extracts for enzymatic determinations and for fractionation of HAT complexes were obtained by the salt dissociation/ ultracentrifugation method described previously (49), with minor modifications. Briefly, yeast cells grown to exponential phase in liquid YPD medium were harvested by centrifugation, washed twice in distilled water, and spheroplasted with Zymolyase as described (49). Spheroplasts were lysed in buffer (5 ml/g cells) containing 75 mM Tris-HCl, pH 7.9, 0.25 mM EDTA, 10 mM 2-mercaptoethanol, 0.1% (v/v) Tween 20, and the protease inhibitors 1 mM PMSF, 2 M E64, 25 M 3,4-DCI, and 2 g/ml chymostatin and homogenized with vigorous magnetic stirring. This and all subsequent steps were performed at 4°C. Solid NaCl was added to give a final concentration of 0.6 M, and the mixture was stirred for 30 min. The homogenates were ultracentrifuged for 2 h at 100,000 ϫ g, and the resulting supernatants were saved and dialyzed against three changes of buffer B (15 mM Tris-HCl, pH 7.9, 0.25 mM EDTA, 5 mM 2-mercaptoethanol, 0.05% (v/v) Tween 20, 10% (v/v) glycerol, 10 mM NaCl). The dialyzed extracts were made to 80 mM NaCl by addition of solid NaCl, centrifuged for 20 min at 27,000 ϫ g, and loaded onto Q-Sepharose FF (Amersham Biosciences) columns (0.75 ϫ 2 cm) equilibrated in buffer B containing 80 mM NaCl. After washing with the same buffer, bound proteins were eluted with 20 volumes of a linear 80 -400 mM NaCl gradient in buffer B. Fractions were collected and assayed for histone acetyltransferase activity.
Histone Acetyltransferase Assay-HAT activity on specific histones was determined essentially as described (40). Aliquots of 15 l from chromatographic or immunoprecipitation fractions were mixed with 8 g of chicken erythrocyte core histones (50) and 0.008 Ci of [1-14 C]acetyl-CoA (54 mCi/mmol, ICN) in a final volume of 20 l, and the resulting reaction mixtures were incubated for 20 min at 30°C. Reaction products were subjected to 16% PAGE in the presence of SDS to resolve protein histones. Gels were first Coomassie Blue-stained, destained, dried, and then fluorographed.
Western Blotting and Immunoprecipitation Assays-For analysis of HA-and Myc-tagged proteins, aliquots from whole-cell extracts prepared as described (51), or from chromatographic fractions, were loaded on SDS-10% polyacrylamide gels, and the resolved proteins were transferred to nitrocellulose membranes using a semi-dry transfer apparatus (Bio-Rad). The blots were blocked and probed using standard procedures (52) and visualized using the ECL chemiluminescent detection kit (Amersham Biosciences) according to the manufacturer's instructions. Antibodies were the anti-HA clone 12CA5, and the anti-Myc clone 9E10 monoclonal antisera (Roche Applied Science), and the polyclonal anti-HA Y-11 (Santa Cruz Biotechnology).
For the analysis of histone H4 acetylated on lysine 12, purified histones (see below) or whole-cell extracts, prepared as described above, were electrophoresed on SDS-16% polyacrylamide gels and transferred onto 0.2-m pore size nitrocellulose membranes following the procedure described by Thiriet and Albert (53). Blots were probed with a 1:10,000 dilution of a rabbit antiserum against acK12H4 (Upstate Biotechnology, Inc.) and processed exactly as described by the manufacturer for the ECL Advance Western blotting detection kit (Amersham Biosciences).
For immunoprecipitation experiments, 1 g of rat anti-HA (3F10) or mouse anti-Myc (9E10) monoclonal antisera were mixed with 200 l of crude extracts prepared as described above for HAT activity determinations or of Q-Sepharose fractions and incubated for 4 h at 4°C. Twenty microliters of pre-equilibrated protein G-Sepharose (Amersham Biosciences) were then added and incubated for 4 h on a rotating wheel. After centrifugation (500 ϫ g for 1 min), supernatants were saved, and the beads were washed six times with 0.5 ml of buffer B containing 150 mM NaCl. Input materials, supernatants, and a suspension of beads in 150 mM NaCl in buffer B were assayed for HAT activity using free histones as a substrate, and in some experiments were also analyzed by Western blotting.
Histone Purification-Yeast histones were purified by acid extraction of isolated chromatin. Briefly, yeast cells, grown in YPD medium and harvested by centrifugation, were resuspended in 4 ml/g cells of digestion buffer (50 mM Tris-HCl, pH 7.5, 1 M sorbitol, 5 mM MgCl 2 ) containing 75 mM 2-mercaptoethanol and incubated for 20 min at room temperature. Cells were collected, resuspended in digestion buffer (4 ml/g of cells), and after addition of 6 mg (per g of cells) of Zymolyase 20T were incubated at 35°C for 30 -45 min with gentle agitation. All subsequent steps were performed at 4°C. Following the addition of 15 ml/g cells of ice-cold wash buffer (50 mM Mes, pH 6.0, 1 M sorbitol, 5 mM MgCl 2 , and the following protease inhibitors: 1 mM PMSF, 2 M E64, and 5 g/ml chymostatin), the spheroplasts were pelleted by centrifugation at 1000 ϫ g for 5 min and lysed in 7 ml/g cells of lysis buffer (50 mM Mes, pH 6.0, 75 mM KCl, 0.5 mM CaCl 2 , 0.1% Nonidet P40, plus the following protease inhibitors: 1 mM PMSF, 20 M 3,4-DCI, 2 M E64, and 5 g/ml chymostatin). After incubation for 5 min, the lysates were centrifuged at 12,000 ϫ g for 5 min, and pellets, containing crude nuclei, were resuspended with 7 ml/g buffer HS (10 mM Mes, pH 6.0, 430 mM NaCl, 1 mM PMSF, 20 M 3,4-DCI, 2 M E64, and 5 g/ml chymostatin) containing 0.5% (v/v) Nonidet P-40 and incubated for 5 min. The samples were centrifuged (15,000 ϫ g, 5 min), and the pellets, containing crude chromatin, were washed once in buffer HS (7 ml/g of initial cells) and recovered by centrifugation at 20,000 ϫ g for 5 min. Histones were extracted from the final pellet with 0.25 M HCl and recovered after acetone precipitation as described (25).
Cell Synchronization-Yeast cell synchronization with ␣-factor at 3 g/ml and FACS analysis with an EPICS XL (Coulter Inc) flow cytometer were performed as described by Igual et al. (54).

Identification of Hif1 as Both a Hat1-and a Histone H4interacting Protein by the Yeast Two-hybrid
System-To identify proteins that interact with Hat1, we performed a yeast two-hybrid search, using as bait full-length Hat1 (amino acids 1-374) fused in-frame to LexA. Several plasmids encoding Hat2, a known Hat1-interacting protein (20), were identified. Also found were several plasmids containing fragments of a previously uncharacterized ORF, YLL022C, and here named HIF1, Hat1 interacting factor 1 ( Table II). The Hat1-Hif1 twohybrid interaction was also seen with LexA-Hat1 and a pGAD plasmid containing almost full-length Hif1 pGAD-Hif1-(8 -385) (Table II).
A parallel two-hybrid analysis was performed to identify proteins able to interact with histone H4 (amino acids 1-59) (46). Screening of 7 ϫ 10 6 transformants identified 10 positive clones. Plasmid purification and sequencing revealed that these clones corresponded to three fusion fragments of the gene HIF1 (nucleotides 22-470, 1 clone; 22-552, 6 clones; and 22-555, 3 clones). Confirmation of the H4-Hif1 interaction was done by reconstruction. Yeast strain PJ69-4A was transformed with the bait and the prey plasmids described in Table II and plated on selective media. Cell growth and ␤-galactosidase activity was only obtained when both fusion proteins were co-expressed. Because histone H4 is the main, if not unique, in vitro protein substrate target of acetylation by Hat1 (19,20,55), the detected H4-Hif1 interaction suggests that the Hat1-Hif1 association described above is meaningful.
Both Hat1-Hif1 and Histone H4-Hif1 Interactions in the Two-hybrid Assays Are Dependent on HAT2-The two-hybrid data presented above plus results published previously indicated a series of interactions between the catalytic protein Hat1, its main substrate histone H4, and the additional companion proteins Hat2 and Hif1. Therefore, we studied whether these interactions could be interrelated. We analyzed the twohybrid interactions between the same proteins in yeast strains containing deletions in the HIF1 or HAT2 genes. Whereas the Hat1-Hat2 association was still detected in hif1⌬ cells (results not shown), the Hat1-Hif1 interaction was abolished in a hat2⌬ mutant (Table II). This result suggests that Hat2 might act as a bridge between Hat1 and Hif1 proteins. Thus, it can be formally thought that in vivo Hat1, Hat2, and Hif1 proteins form a heterotrimeric association.
In order to explore a potential mediating role of Hat2 in the H4-Hif1 association, we also assayed two-hybrid interactions between these proteins in a yeast hat2⌬ strain. Activation of the reporter genes was not detected when the fusion proteins GAD-Hif1-(8 -185) and GBD-H4-(1-59) were co-expressed in the absence of HAT2 gene (Table II), indicating that Hat2, in the context of the two-hybrid assay, may also act as mediator for the H4-Hif1 association. Such a mediating role is specific because the interaction between Bdf1 and histone H4 described previously (46) was not affected by the HAT2 deletion (Table  II).
Hif1 Is an Integral Component of the Yeast HAT-B Complex-Our two-hybrid experiments showed an in vivo association between Hif1, Hat1, and Hat2 proteins. It was therefore important to test these interactions in protein extracts and to identify the multiprotein HAT complex(es) in which Hif1 takes part together with (or without) Hat1 and Hat2 proteins. To investigate this, we analyzed the presence of Hif1 in the different yeast HAT complexes resolved by ion-exchange chromatography. We used a yeast strain expressing a genomically Myc epitope-tagged version of Hif1. A whole-cell extract prepared from log-phase cells was fractionated onto a Q-Sepharose FF column (40). Identical chromatography of an extract from the isogenic wild-type, non-tagged Hif1 strain was performed to serve as a control of HAT activity elution (Fig. 1A). Beyond the random variability from experiment to experiment, the chromatographic HAT activity profiles from both strains were similar (Fig. 1, A and B), indicating that the Myc tag at the C terminus of Hif1 alters neither enzymatic activity nor chro- ϩ a a Interaction observed on transformants obtained directly from the library screening.
Hif1 Is Part of the Yeast Nuclear HAT-B Complex matographic properties of any of the recovered HAT complexes in which it takes part. Next, fractions from the Q-Sepharose elution of the tagged Hif1 strain were analyzed by Western blotting. The antiserum directed against Myc tag detected a tight co-elution of Hif1 protein with the HAT-B peak (Fig. 1C, fractions 28 -36) but not with any of the other HAT complexes, not even with the HAT-A3 (fractions 22-27, activity on H4, Fig.  1B), which also contains Hat1 and Hat2 proteins (40). Next, we tested whether Hif1 in those fractions is physically associated with a HAT activity by co-immunoprecipitation analysis. In this experiment, aliquots from the chromatographic fractions were incubated with the anti-Myc antibody, and the resulting precipitates, together with the input and unbound materials, were assayed for HAT activity using chicken erythrocyte core histones and [ 14 C]acetyl-CoA as substrates. As shown in Fig.  1D, Hif1 co-immunoprecipitates a HAT specific for histone H4. An important depletion of activity on H4 in the unbound fractions, in comparison with the inputs, was also detected in these assays, as expected (Fig. 1D, middle panel). Q-Sepharose chromatography of yeast whole-cell extracts results in the recovery of five HAT enzymes (40) (Fig. 1, A and  B). Another HAT enzyme overlaps with the HAT-B peak, HAT-A4, which preferentially acetylates histones H3 and H4 (fractions ϳ27-35 in Fig. 1, A and B). We have demonstrated previously that these enzymes are different, because the HAT-A4 complex, contrary to HAT-B, is not dependent on HAT1 or HAT2 (40). Formally Hif1 could participate in a less abundant HAT complex, not recognized in our profiles. Therefore, the unequivocal assignment of the HAT-B enzyme as the Hif1containing complex required a more thorough analysis. We addressed this in two ways: first, we prepared whole-cell extracts from wild-type, hat1⌬, or hat2⌬ yeast strains expressing Myc-tagged Hif1 protein, and subsequently subjected them to immunoprecipitation using anti-Myc. As a negative control, an extract from a non-tagged wild-type strain was processed in the same way. Input, unbound, and bound materials were assayed for HAT activity. As shown in Fig. 2A, Hif1 co-precipitated an H4-specific HAT enzyme in a wild-type background but not in hat1⌬. This result suggested that in our protein extracts Hat1 may be the catalytic subunit of the Hif1-interacting HAT complex. Moreover, Hif1 was apparently unable to co-precipitate any HAT activity in a hat2⌬ genetic background ( Fig. 2A), reaffirming the two-hybrid data indicating that Hat2 functions as a bridge between Hat1 and Hif1 in a native HAT complex (see below). Second, to confirm the association of Hif1 with the HAT-B complex resolved by anion-exchange chromatography, we generated a double-tagged yeast strain containing Hat1-HA and Hif1-Myc. Fractionation and partial purification of the HAT complexes was obtained on a Q-Sepharose column, and the fractions corresponding to the HAT-B peak were used for reciprocal co-immunoprecipitation experiments. Immunoprecipitated fractions were used subsequently to monitor HAT activity by enzymatic assay and the presence of the other companion protein by Western blotting. The results are shown in Fig. 2B. Hat1 precipitated by anti-HA pulled down both an H4-specific HAT activity and Hif1 protein (Fig. 2B, left panels). In addition, Hif1, immunoprecipitated by the anti-Myc antibody, co-precipitated a H4-HAT activity and, reciprocally, Hat1 protein (Fig. 2B, right panels). In summary, Hat1 and Hif1 proteins co-precipitate an H4-specific HAT activity and each other in chromatographic fractions corresponding to the HAT-B complex. These results together with those from the two-hybrid experiments (Table II) and those presented in Fig.  1 demonstrate that in vivo Hif1 is indeed a component of the native yeast HAT-B complex.
As part of our studies on the association of Hif1 with the HAT-B enzyme, we compared the Q-Sepharose chromatographic HAT profiles of hif1⌬ deletion mutant strains with those of the isogenic wild-type strains. The HAT-B peak was recognized by HAT assays and also by Western blotting when the strain expressing HA-tagged-Hat1 was used for these analyses. Curiously, the Q-Sepharose chromatographic HAT-B complex from the hif1⌬ strains was almost identical, both the position in the gradient elution and the level of activity against histone H4 in in vitro assays (results not shown). In contrast, the absence of Hat1 or Hat2 proteins completely abolished the chromatographic peak corresponding to the HAT-B complex (40). In addition, by using peptides corresponding to the N FIG. 1. Hif1 co-purifies with a histone H4-specific HAT enzyme. Cell extracts prepared from yeast strains expressing untagged Hif1, as a control, and Myc-tagged Hif1 were chromatographed onto Q-Sepharose FF columns. Activities of histone acetyltransferases were determined over the whole chromatographic eluates. After incubation of aliquots from the eluted fractions with chicken erythrocyte core histones and [ 14 C]acetyl-CoA, histones were resolved by SDS-16% PAGE, and the gels were subsequently fluorographed. A, fluorogram from the wild-type strain containing untagged Hif1 (W303-1a). Five HAT complexes are identified in this fractionation (40). The names of these complexes are indicated above the gradient position where they elute. Migration positions of the four core histones are indicated. B, fluorogram obtained from a strain expressing a genomic Hif1 version tagged with the Myc epitope at its C terminus (BQS1187). Note that Myc tag on Hif1 affects neither the elution position nor the apparent activity of any of the HAT complexes. C, analysis of the Hif1-Myc protein elution by Western blotting. Aliquots of the indicated fractions from the Myc-tagged Hif1 chromatography (elution shown in B) were subjected to SDS-10% PAGE and transferred onto nitrocellulose membrane, and the resulting blot was probed with monoclonal anti-Myc antibody (9E10). Note that Hif1 signal co-elutes with the HAT peak corresponding to the B-enzyme. D, indicated fractions from the Hif1-Myc chromatography shown in B were subjected to immunoprecipitation with anti-Myc antiserum. Input and the recovered unbound and bound fractions were directly assayed for HAT activity as above, and the corresponding fluorograms are shown.
terminus of histone H4, we analyzed site specificity of the HAT-B complex, with or without Hif1. Lysine residues 12 and 5 were the sole targets for acetylation by HAT-B with or without Hif1 (results not shown). In conclusion, these data indicate that Hif1 does not change the chromatographic behavior of the HAT-B complex and, more important, suggest that the function of Hif1 may not be to regulate the activity and/or specificity of the catalytic subunit Hat1.
Hat2 Mediates the Interaction between Hat1 and Hif1 Proteins-Results from the two-hybrid experiments indicated that Hat2 may act as a bridge between Hif1 and Hat1 or histone H4 (Table II). A mediating role in the H4-Hat1 interaction has been suggested previously for yeast Hat2 (20) and also for its human homolog, Rbap46 (22). Our co-immunoprecipitation data also suggest that Hat2 functions as mediator in the Hat1-Hif1 interaction. The inability of Hif1-Myc to co-precipitate the H4-specific HAT activity in a hat2⌬ extract seems to indicate such mediating function ( Fig. 2A). Moreover, a careful inspection of the fluorogram shown in Fig. 2A revealed that the input fractions contained a reduced level of HAT activity on histone H4, not only in the hat1⌬ extract as expected but also in the hat2⌬ extract ( Fig. 2A). A reasonable interpretation is that Hat2 is involved in the affinity and activity of the catalytic Hat1 protein on its substrate, histone H4 (20,22); therefore, in the absence of HAT2, Hat1 present in the protein extract has very low enzymatic activity. Besides this interpretation, we thought of two more possibilities to explain the immunoprecipitation results with the hat2⌬ extract: (i) Hat1 could be unstable in hat2⌬ cells or in our extract; and (ii) Hat1 could be coimmunoprecipitated by Hif1, even in the absence of Hat2, but due to its very low activity, it would not be detected in the precipitated fraction. Because these alternative possibilities could cast doubt on the mediating function of Hat2, we felt it was necessary to explore them.
To rule out possibility i, we analyzed the HA epitope-tagged Hat1 protein level in isogenic wild-type and hat2⌬ mutant strains by Western blotting. Hat1 was as stable and appeared as intact in extracts prepared from hat2⌬ mutant cells as from HAT2 cells (Fig. 3A). To rule out objection ii, we generated yeast strains expressing genomically HA-tagged-Hat1 and Myctagged-Hif1 proteins in both wild-type and hat2⌬ genetic backgrounds. Experiments of reciprocal co-immunoprecipitation, with subsequent Western blotting of the resulting fractions, were performed with whole-cell extracts from these doubly tagged strains. As shows Fig. 3B, Hif1 pulled down Hat1 in the wild-type extract (detected by Western blotting with HA antibody, lane 9) but not in the hat2⌬ extract (lane 10). Reciprocally, Hat1 co-precipitated Hif1 in the wild-type extract (detected with anti-Myc, lane 17) but was not pulled down in the hat2⌬ extract (lane 18). Taken together these results demonstrated unambiguously that Hat2 is essential for the Hat1-Hif1 association.

FIG. 2. Hif1 is associated with the Hat1-containing HAT-B complex.
A, cell lysates prepared from wild-type (BQS1187), hat1⌬ (BQS1202), and hat2⌬ (BQS1225) yeast strains expressing Myc-tagged Hif1 or from an untagged wild-type control strain (W303-1a) were immunoprecipitated with anti-Myc antiserum (9E10). Input, supernatants (unbound) and beads (bound) were assayed for HAT activity using chicken free core histones and [ 14 C]acetyl-CoA as substrates. After incubations, histones were resolved by SDS-16% PAGE, and the resulting fluorogram is shown. Positions of core histones are indicated. B, an extract from BQS1226 cells, co-expressing Hat1-HA and Hif1-Myc, was chromatographed onto a Q-Sepharose FF column as in Fig. 1. Indicated fractions corresponding to partially purified HAT-B enzyme were then immunoprecipitated with rat anti-HA (3F10, left panels) or mouse anti-Myc (9E10, right panels) monoclonal antibodies. The fluorograms obtained after HAT assays of the input, unbound, and bound fractions are presented. Migrations of H4, H3, and H2A histones are indicated. Immunoprecipitates were also examined by Western blotting using mouse monoclonal anti-Myc and rabbit polyclonal anti-HA antisera, as indicated (lower panels).

FIG. 3. Hat1-Hif1 association is mediated by Hat2.
A, Hat1 protein levels in extracts from wild-type (BQS1154) and hat2⌬ (BQS1172) yeast strains expressing HA epitope-tagged Hat1 were examined by Western blotting analysis using mouse anti-HA (12CA5) antibody. Control extract from the untagged wild-type strain (W303-1a) was also included. Equal amounts of proteins, estimated by a previous Coomassie Blue-stained electrophoresis gel, were loaded. B, reciprocal co-immunoprecipitation assays were performed with whole-cell extracts prepared from the wild-type (BQS1226) or hat2⌬ (BQS1306) doubly tagged strains expressing Hat1-HA and Hif1-Myc. As indicated, extracts were incubated with anti-HA (3F10) or anti-Myc (9E10) antibodies, and input material, and the resulting unbound and bound fractions were probed for both Hat1-HA-(upper panel) and Hif1-Myc (lower panel)-tagged proteins by Western blotting analyses, with rabbit anti-HA and mouse anti-Myc antibodies, respectively. IP, immunoprecipitation.
In the reciprocal co-immunoprecipitation experiments, shown in Fig. 3B, we found that anti-HA and anti-Myc antibodies pulled down Hat1-HA and Hif1-Myc very efficiently. The almost complete absence of signal in the unbound fractions (lanes 3 and 15, corresponding to the experiment with the wild-type strain, Fig. 3B) indicates the high efficiency of the immunoprecipitation. However, as can be seen, Hat1-HA was still present in the unbound fraction meaning that a portion of Hat1 was not co-precipitated with Hif1-Myc (lane 5, unbound). Similarly, Hat1 did not completely immunodeplete Hif1-Myc, because a significant amount of this protein remained in the supernatant (lane 13, unbound). Although it cannot be ruled out that the inefficient co-immunoprecipitation in our extracts was partially an artificial consequence of the experiment, these results suggest that in vivo only a fraction of Hat1 may be associated to Hif1 and vice versa. This interpretation is also consistent with the fact that Hat1 is also found in additional complexes (23,27,40).
Subcellular Localization of HAT-B Complex-As mentioned above, biochemical fractionations have indicated that type B HATs are cytoplasmic enzymes (20,22,26,29,31,32,39,40). However, in several species Hat1 has also been detected in the nucleus (22)(23)(24)40). In an effort to clarify the intracellular localization of the HAT-B proteins, we examined yeast cells expressing tagged versions of Hat1, Hat2, or Hif1 proteins by indirect immunofluorescence using the corresponding anti-tag antibodies (Fig. 4). Strikingly, the three tagged protein components of the HAT-B enzyme were mainly located in the nucleus. In log-phase cultures, cells display a Hat1-HA or Hat2-HA signal that is predominantly nuclear (Fig. 4A). A similar result was obtained with the yeast strain containing Myc epitopetagged Hif1 (Fig. 4B). When a non-tagged, wild-type strain was used as a negative control for both antibodies, no signal was detected (Fig. 4, no tag panels). Furthermore, for the three proteins, thorough inspections of the visualized cells indicated that the nuclear localization occurs in cells at all stages of the cell cycle. Our results do not rule out that B-enzyme proteins may also be in the cytoplasm, but they clearly show that, contrary to the common view, a dominant portion of the yeast HAT-B complex is located in the nucleus.
Next, to further characterize the subcellular localization of the three HAT-B components, we asked whether their nuclear residence is interdependent. To address this question, the subcellular localization of each tagged component, Hat1-HA, Hat2-HA, and Hif1-Myc, was visualized in yeast cells defective in any of the other companion proteins. Hif1-Myc nuclear signal was neither dependent on HAT1 nor on HAT2 (Fig. 5A). The nuclear residence of Hat2-HA also exhibited no difference between cells with a wild-type, hat1⌬, or hif1⌬ genetic background (results not shown). On the contrary, the nuclear localization of the Hat1-HA depends on HAT2 but not on HIF1 (Fig. 5B). In the hat2⌬ strain the Hat1-HA signal was visualized throughout the cell (Fig. 5B, lower panels). Therefore, Hat2 may participate in the nuclear localization of the catalytic subunit Hat1.
Similarly to HAT1, the Deletion of HIF1 Gene, in Combination with Specific Substitution Mutations in the N Terminus of Histone H3, Causes Mild Defects in Telomeric Silencing-Single hat1⌬ or hat2⌬ mutants lack an obvious phenotype or growth defects (19,20). However, it has been found that deletion of the HAT1 gene, in combination with specific acetylatable lysine substitution mutations in the N terminus of histone H3, impairs telomeric silencing (37). Specifically, a double substitution K9R,K14R (lysine to arginine) histone H3 mutant, combined with a deletion of HAT1, has a significant defect in telomeric silencing. Because Hif1 and Hat1 are associated in vivo, we asked whether HIF1 would also be involved in telomeric silencing. To address this question, we used a series of yeast strains containing, besides a URA3 reporter gene integrated at telomere VII-L, various combinations of deletions in HAT1 and HIF1 genes and the K9R,K14R substitution mutation in histone H3. In all strains the chromosomal copies of the (BQS1202) and HIF1-Myc hat2⌬ (BQS1225) strains were processed for indirect immunofluorescence as described in Fig. 4. B, exponential yeast cells of the hif1⌬ (BQS1184) or hat2⌬ (BQS1172) strains expressing Hat1-HA were assayed by indirect immunofluorescence with the corresponding antibodies, as described in Fig. 4. DIC, differential interference contrast; DAPI, 4,6-diamino-2-phenylindole. genes for histone H3 and H4 (HHT1-HHF1 and HHT2-HHF2) were deleted, but they carried episomal copies of wild-type (HHF2) H4 and wild type (HHT2) or K9R,K14R mutated version (HHT2 (K9,14R)) H3 genes. We estimated the level of silencing in these strains by measuring the fraction of cells that can form colonies on SC ϩ 5-fluoroorotic acid (5-FOA) plates relative to the plating control (SC plates). As shown in Fig. 6, in strains expressing wild-type H3, although the double mutant hat1⌬hif1⌬ strain seemed to have a slight defect in telomeric silencing, neither hat1⌬ nor hif1⌬ single mutations generated a significant effect. They had an efficiency of plating on 5-FOA similar to the wild-type strain. As expected, a decrease in the silencing of reporter URA3 gene was observed in yeast cells expressing K9R,K14R histone H3. Moreover, combining the K9R,K14R H3 mutation with a deletion of HIF1 gene led to a greater defect in telomeric silencing. This decreased level of telomeric silencing was similar to that resulting from a deletion of HAT1 (Fig. 6), which was known previously (37). We realize that our results do not show a defect on telomeric silencing as severe as that obtained by Kelly et al. (37) for the hat1⌬ mutation. Although we do not know the reason for this difference, perhaps variations in the media and their constituents and also in the culture conditions (56) may explain it. In any case our results are consistent because they indicate the involvement of HAT1 in telomeric silencing redundantly with histone H3. Moreover, the similar defect observed for hif1⌬ suggests that Hif1 participates together with Hat1 in the telomeric silencing function. In line with this idea, when the hat1⌬ hif1⌬ double mutant combined with the histone K9R,K14R allele was assayed, a further decrease in the level of telomeric silencing was not observed (Fig. 6).
Deletions of HAT1, HAT2, or HIF1 Genes Do Not Affect the in Vivo Acetylation Level of Histone H4 Lysine 12-During nucleosome assembly, in all species where it has been studied, newly synthesized and deposited histone H4 appears diacetylated on lysines 5 and 12 (28,33,34,39). This conserved Lys-5/Lys-12 deposition pattern of acetylation is thought to be generated by the Hat1-containing HAT-B complex. In vitro, native HAT-B enzymes from a number of widely divergent species are capable of modifying histone H4 on these positions (20, 22, 23, 29 -32).
In the yeast Saccharomyces cerevisiae, although it is not known whether newly synthesized H4 is diacetylated on Lys-5/Lys-12, recombinant Hat1 and also cytoplasmic isolated HAT-B complex acetylate these positions, with preference for lysine 12 over lysine 5 (19,20,55). Despite the abundant data on in vitro site specificity of type B enzymes from several species, analyses about their specificity in vivo are absent. We have attempted to ascertain the requirement of the HAT-B proteins in the in vivo modification of lysine residue 12 of histone H4, comparing its steady-state level of acetylation in wild-type, hat1⌬, hat2⌬, and hif1⌬ strains. We carried out this study by Western blotting analysis using a commercially available anti-acK12H4 antibody (Upstate Biotechnology, Inc.). First, we checked the antibody specificity on purified yeast histones and histones present in whole-cell extracts. As a first specificity control, we purified histones and prepared WCE from yeast cells expressing wildtype or K12R substitution mutant H4 from a centromeric plasmid as the only source of histone H4. The acK12H4 antiserum generated a strong signal on the samples from wild-type strains, but no signal, even after very long exposure, was detected on purified histones or on WCE obtained from the strain expressing K12R H4 (results not shown). As a positive control, purified chicken erythrocyte and yeast histones were incubated with recombinant yeast Hat1, in the presence or absence of the other substrate, acetyl-CoA, and the reaction products were subjected to Western blotting examination. As expected, histones incubated in the presence of acetyl-CoA generated a much stronger signal (results not shown). Together, these control tests indicated a rather high specificity for the antibody against acetylated lysine 12 of histone H4. Fig. 7 shows the results when the antibody was used for the analysis of HAT-B proteins on deletion mutants. Purified histones from isolated chromatin (Fig. 7A) or from WCE, which may contain total cellular histones, from chromatin, and also putative "free" histones ( Fig. 7B) were analyzed from log-phase asynchronous wild-type, hat1⌬, hat2⌬, or hif1⌬ cells. No significant difference in the acetylation level of Lys-12 of H4 was observed between the wild-type and mutant strains on purified histones nor on histones present in WCE. The apparent absence of a Lys-12 acetylation defect in the HAT-B mutants is striking, particularly in the hat1⌬ strain, given its demonstrated in vitro preference for this position. It is possible to find multiple explanations that could account for these results, and many of them have as a consequence that the fraction of Hat1- FIG. 6. Deletion of HIF1 or HAT1, together with the K9R,K14R substitution mutation in histone H3, causes a similar defect in telomeric silencing. Yeast cells were grown in YPD liquid medium to log-phase at 28°C, and an identical number of cells were spread on SC and SC plus 5-FOA (1 g/liter) plates and incubated at 28°C for 3 days. Telomeric silencing was quantified for each strain from at least three repetitions of the assay by counting the number of colonies that grew on SC plus 5-FOA relative on SC. The mean, together with the standard deviation, is represented. Yeast strains are as follows: BQS1228 (wildtype H3); BQS1240 (wild-type H3 hif1⌬); BQS1232 (wild-type H3 hat1⌬); BQS1244 (wild-type H3 hif1⌬ hat1⌬); BQS1230 (K9,14R H3); BQS1242 (K9,14R H3 hif1⌬); BQS1234 (K9,14R H3 hat1⌬); and BQS1230 (K9,14R H3 hif1⌬ hat1⌬). All strains contained the URA3 gene integrated at telomere VII-L, and expressed the indicated histone H3 allele from a centromeric plasmid as the sole source of histone H3. FIG. 7. Acetylation of lysine 12 of histone H4 is apparently not  dependent on HAT1, HAT2, or HIF1 in vivo. Wild-type and the hat1⌬, hat2⌬, and hif1⌬ deletion mutant yeast strains, as indicated, were asynchronically grown to exponential phase in YPD medium. A, histones purified as described under "Experimental Procedures" were resolved by electrophoresis on a 16%-polyacrylamide gel containing SDS, transferred to a nitrocellulose membrane, stained with Ponceau (upper panel), and probed with the antiserum against histone H4 acetylated on K12 (lower panel). B, whole-cell extracts (WCE) from the same yeast strains were subjected to Western blotting analysis as in A. Upper panel, Ponceau-stained membrane; lower panel, immunoblot. Only the region of the gel corresponding to the histones is shown. No other protein in the extracts was detected in the immunoblot. dependent Lys-12 H4 acetylation is always very low. With the aim to find out if Hat1 acetylates lysine 12 of H4 in vivo, we tried conditions that might increase the difference in the Hat1dependent acetylation level between wild-type and hat1⌬ strains. It is expected that a loss or inhibition of a histone deacetylase involved in vivo in Lys-12 H4 deacetylation would increase the difference between cells with or without HAT1. Therefore, we analyzed the acK12H4 level in (i) yeast cells containing a deletion of HAT1 combined with deletion mutants in any of several histone deacetylases RPD3, HDA1, HOS1, or HOS2. It is known that RPD3 and HDA1 are required for the in vivo global deacetylation of Lys-12 of H4 (57,58). Site specificity of the other histone deacetylases is not known. (ii) We analyzed the acK12H4 level in yeast cells grown and extracted in the presence of the histone deacetylase inhibitor, trichostatin A, which has a similar effect on yeast cells as the deletion of several histone deacetylases (59). Only RPD3 deletion and trichostatin A treatment generated an appreciable and significant increase in the acetylation of lysine 12 of H4. However, the deletion of HAT1 gene in these strains did not cause any reduction in the level of acK12H4 either (results not shown).
Alternatively, the difficulty in detecting the HAT1-dependent acetylation of Lys-12 of histone H4 could be a consequence of the masking effect due to other Lys-12 H4 acetylating HATs. To explore this possibility, we combined a hat1⌬ mutation with a temperature-sensitive esa1 ts mutation, esa1-⌬414 (60). Esa1p is a yeast-essential HAT that participates in the global Lys-12 acetylation of histone H4 in vivo (61). Our Western blotting experiments revealed that histone H4 from esa1 ts mutant cells, after incubation at the restrictive temperature, indeed had a much lower, although still detectable, level of Lys-12 acetylation than the wild-type cells. Again, however, we did not observe a further reduction in the esa1 ts hat1⌬ double mutant (results not shown).
By taking into account the functional role proposed for Hat1 in the acetylation of free H4 prior to its deposition onto chromatin, we reasoned that perhaps the HAT1-dependent Lys-12 H4 acetylation would be transient. Therefore, the acK12H4 isoform would only accumulate during a short period at the G 1 /S phase of the cell cycle, when the new histones are synthesized (62). In order to investigate this, we analyzed the acetylation of Lys-12 of H4 in whole-cell extracts prepared from wild-type and hat1⌬ cells proceeding synchronously through the cell cycle after release from an ␣-factor arrest. FACS analysis did not show any obvious difference between the strains (not shown). Western blotting experiments revealed neither an apparent peak of acetylation on Lys-12 of H4 along the cell cycle nor, more important, a significant difference in the acK12H4 content between wild-type and hat1⌬ cells (results not shown). If Hat1 acetylates this position in a particular state of the cell cycle, then the window of permanence of such a modification must be very narrow.
Collectively our Western blotting results indicate that in vivo, under normal conditions, the acetylation state of lysine residue 12 of histone H4 is apparently not dependent of HAT1. Lys-12 acetylation by Hat1 could be necessary only in particular physiological stages, not tested here, or perhaps the fraction of HAT1-dependent K12 acetylation is very small and undetectable with our methods.
Hat1 Protein Is Present throughout the Cell Cycle-If the Hat1 enzyme acts to acetylate nascent histone H4, then a peak of Hat1 protein might also be generated during the late G 1 and early S-phase of the cell cycle. To investigate this, we analyzed the content of Hat1 protein along the cell cycle. Yeast cells expressing Hat1-HA were synchronized by release from an ␣-factor block. Extracts of cells, obtained from identical vol-umes taken at intervals after release of ␣-factor arrest, were prepared, and the Hat1-HA level was examined by Western blotting (Fig. 8A). The budding index was determined microscopically as a control of the synchronization quality (Fig. 8B). The result showed that Hat1 was present throughout the cell cycle and that there was no peak of Hat1 protein at the G 1 /S transition, as defined by cultures in which 50% of the cells are budded. This result is consistent with HAT1 transcription data obtained during the cell cycle, indicating a constant level of HAT1 message (63).

DISCUSSION
By using the two-hybrid system, we identified Hif1 as a protein that interacts with the catalytic subunit of the HAT-B complex, Hat1, and with its substrate, histone H4 (Table II). This interaction was confirmed biochemically (Fig. 1) and by co-immunoprecipitation (Figs. 2 and 3). It is quite likely that the histone-binding subunit of the complex, Hat2, acts as a bridge between Hif1 and Hat1 and between Hif1 and H4, because the interactions are abolished in a hat2⌬ mutant. The chromatographic properties of the HAT-B complex are consistent with it being a heterotrimer, consisting of Hat1, Hat2, and Hif1.
Deletions of HAT1 or HAT2 lead to a loss of the HAT-B complex and of HAT activity (20,40). On the other hand, hif1⌬ mutation does not abolish enzymatic activity of the complex in vitro and, surprisingly, does not alter its chromatographic properties significantly. Based on these results, Hif1 appears to be a non-essential subunit of the complex. However, deletion of HIF1 does cause precisely the same minor telomeric silencing defect as is seen in hat1⌬ or hat2⌬ mutants (only observable in the presence of certain H3 lysine-to-arginine mutations; Fig. 6), suggesting that Hif1 does influence the properties of the HAT-B complex in vivo. Hif1 is fairly well conserved among various budding yeasts, including distantly related ones such as Candida glabrata and K. lactis, but no obvious homolog is found in fission yeast nor in larger eukaryotes. In contrast, FIG. 8. Hat1 protein level is approximately constant throughout the cell cycle. Yeast strain BQS1154, which has an HA epitopetagged version of Hat1, was grown to log-phase in YPD medium. Cells were arrested for 2.5 h with 3 g/ml ␣-factor. After release from the ␣-factor arrest, cells from identical volumes at the indicated time points were collected, and WCEs were prepared. The Hat1 level was examined by Western blotting assay by using the mouse 12CA5 antiserum against the HA epitope. A, immunoblot visualized with a chemiluminescent detection system (ECL Advance, Amersham Biosciences). Lane a, asynchronous cells. B, at the same time points as in A, the percentage of cells with buds (budding index), as a control of the synchronization efficiency, was determined.
Hat1 and Hat2 are highly conserved, and obvious homologs can be found in all eukaryotes including mammals. Our protein sequence analysis of Hif1 has not identified any motif that could indicate its function.
It remains to be determined whether Hif1 takes part in other macromolecular complexes. It is possible that the association of Hif1 to Hat1-Hat2 is weak, transient, or both. Such an interpretation would explain why Hif1 was not identified as a component of the HAT-B complex during its original isolation after several chromatographic steps (20). Hif1 could have been lost in that extensive purification.
A major result of this study is that all three components of the HAT-B complex are nuclear (Fig. 4). Although we cannot rule out the possibility that a small fraction of any one of these three proteins may reside in the cytoplasm, clearly the great majority is in the nucleus. At first sight this seems surprising, because HAT-B complexes have been isolated from cytoplasmic fractions from several species (20,22,26,29,31,32,39,40). This has led to the widely held view that all HAT-B enzymes such as Hat1 that acetylate free histones but not nucleosomes are cytoplasmic (1). This view has continued despite reports, prior to ours, that Hat1 can be found in the nucleus (22)(23)(24). It is well known that some nuclear enzymes leak out of the nucleus after cell lysis. For example, for many years mammalian DNA polymerase ␣ was considered a cytoplasmic enzyme even though it is now realized that it clearly resides in the nucleus and plays an essential role in DNA replication (64). We think such leakage explains why Hat1 is often found in cytoplasmic fractions.
The known role of Hat2 in increasing the in vitro activity of Hat1 has already been mentioned (20,22), and this work has shown its mediating function in the Hat1-Hif1 and H4-Hif1 associations in vivo. The immunolocalization results presented here suggest a novel and additional function for Hat2. It may be involved in the import or in the maintenance of the catalytic protein Hat1 in the nucleus. On the other hand, the subcellular localization of Hif1 independently of HAT2 seems to indicate that its nuclear location may occur in a different way than that of Hat1, and thus the assembly of the HAT-B complex must happen in the nucleus.
Hat1 is generally considered to play a role in deposition of histone H4 during nucleosome assembly for three reasons: 1) the specificity of the enzyme for acetylation of H4 lysine residues 12 and 5; 2) its ability to acetylate free H4 but not nucleosomes; and 3) its cytoplasmic location. Even though we have shown that Hat1 is clearly nuclear, that is not a reason to abandon the concept that it is important for deposition of H4. After its synthesis in the cytoplasm, H4 is transported into the nucleus by the karyopherin Kap123 (27). The histone can then be acetylated by HAT-B in the nucleus prior to its deposition into nucleosomes.
We did several experiments to look for Hat1-dependent H4 Lys-12 acetylation in yeast, using an antibody highly specific for this modified residue. We could detect acK12H4 in yeast histone preparations and whole-cell extracts, but in no case could we detect a difference in the level of acetylation between wild-type and hat1⌬, hat2⌬, or hif1⌬ mutant strains (Fig. 7). Apparently, Hat1-dependent Lys-12 acetylation is very transient. If indeed Hat1 plays a role in histone deposition, then our results are consistent with earlier results showing that the acetylation of lysines 12 and 5 of histone H4 during deposition is short-lived (28,33).