Báo cáo khoa học: Site specificity of yeast histone acetyltransferase B complex in vivo pot

Here we demonstrate that both Lys12 and Lys5 of solu-ble, non-chromatin-bound histone H4 are in vivo targets of acetylation for the yeast HAT-B enzyme.. In yeast, the substitution mutati

complex in vivo

Ana Poveda* and Ramon Sendra

Departament de Bioquı´mica i Biologı´a Molecular, Universitat de Vale`ncia, Spain

Histone acetylation is a highly dynamic

post-transla-tional modification involved in the regulation of

chro-matin activity in eukaryotic organisms [1,2] Although

the mechanism is not completely understood, the

long-known link between histone acetylation and gene

expression was definitively settled by the identification

of a number of transcriptional regulators as histone

acetyltransferases (HATs) and histone deacetylases

Acetylation influences transcription by facilitating the

access of the transcriptional machinery to the DNA

sequence and by creating specific recognition sites for

regulatory proteins that promote transcription [1]

Histone acetylation has also been proposed to be

involved in chromatin assembly during replication

[1,3] This notion emerged from the finding in different

eukaryotic organisms that newly synthesized histones are acetylated [4,5], and deacetylated shortly after their incorporation into chromatin [6] In many eukaryotes, newly synthesized histone H4 assembled onto nascent DNA is diacetylated on Lys5 and Lys12 [5,7] The N-terminus of newly synthesized histone H3 is also acetylated, but in a more heterogeneous and less con-served manner [5,8,9] It is considered that the acetyla-tion of histones may somehow favor their deposiacetyla-tion onto DNA mediated through specific interactions with histone chaperones [1]

The enzyme that is assumed to catalyze the specific acetylation of newly synthesized histone H4 on its N-terminal tail is the type B HAT, HAT-B complex Enzymes operationally classified as type B, in contrast

Keywords

acetylation; acetyltransferase; chromatin;

histones; yeast

Correspondence

R Sendra, Departament de Bioquı´mica i

Biologia Molecular, Facultat de Cie`ncies

Biolo`giques, C ⁄ Dr Moliner 50,

46100-Burjassot, Vale`ncia, Spain

Fax: +34 96 354 4635

Tel: +34 96 354 3015

E-mail: ramon.sendra@uv.es

*Present address

IGH-Institute of Human Genetics, CNRS

Montpellier, France

(Received 25 January 2008, revised 25

Feb-ruary 2008, accepted 28 FebFeb-ruary 2008)

doi:10.1111/j.1742-4658.2008.06367.x

Saccharomyces cerevisiaeHat1, together with Hat2 and Hif1, forms the his-tone acetyltransferase B (HAT-B) complex Previous studies performed with synthetic N-terminal histone H4 peptides found that whereas the HAT-B complex acetylates only Lys12, recombinant Hat1 is able to modify Lys12 and Lys5 Here we demonstrate that both Lys12 and Lys5 of solu-ble, non-chromatin-bound histone H4 are in vivo targets of acetylation for the yeast HAT-B enzyme Moreover, coimmunoprecipitation assays revealed that Lys12⁄ Lys5-acetylated histone H4 is bound to the HAT-B complex in the soluble cell fraction Both Hat1 and Hat2, but not Hif1, are required for the Lys12⁄ Lys5-specific acetylation and for histone H4 bind-ing HAT-B-dependent acetylation of histone H4 was detected in the solu-ble fraction of cells at distinct cell cycle stages, and increased when cells accumulated excess histones Strikingly, histone H3 was not found in any

of the immunoprecipitates obtained with the different components of the HAT-B enzyme, indicating the possibility that histone H3 is not together with histone H4 in this complex Finally, the exchange of Lys for Arg at position 12 of histone H4 did not interfere with histone H4 association with the complex, but prevented acetylation on Lys5 by the HAT-B enzyme, in vivo as well as in vitro

Abbreviations

FACS, fluorescence-activated cell sorting; H4K12ac, histone H4 isoform with acetylated Lys12; H4K12R, K12R substitution mutant of histone H4; H4K5ac, histone H4 isoform with acetylated Lys5; HA, hemagglutinin; HAT, histone acetyltransferase; HAT-B, histone

acetyltransferase B; HU, hydroxyurea; WCE, whole cell extract.

to type A, only acetylate histones not associated with

DNA, and are not involved in transcriptional

regu-lation HAT-B enzymes were originally isolated from

cytosolic extracts [10–15], but several

immunotion analyses have indicated a mainly nuclear

localiza-tion [16–20] In vitro, native HAT-B enzymes from

a wide variety of species establish the specific

Lys5⁄ Lys12 acetylation pattern characteristic of newly

synthesized histone H4 [10,13,15–17,21–23] In the

yeast Saccharomyces cerevisiae, the HAT-B complex

consists of at least three protein subunits [19,20]: the

catalytic subunit, Hat1; the enzymatic activity

stimula-tory protein, Hat2 [13]; and Hif1, which, in vitro, has

histone chaperone and chromatin assembly activity

[20] Recently, Hat1 and Hat2 have been found to

interact with the origin recognition complex,

suggest-ing a novel role for the Hat1–Hat2 subcomplex at the

replication fork [24] It has been reported that

although recombinant Hat1 is able to modify Lys5

and Lys12 [13,25], the isolated HAT-B complex

exclu-sively acetylates Lys12 of histone H4 [13,22] Deletions

of HAT1, HAT2 or HIF1 produce no apparent

pheno-type [13,19,20,22], but combined with specific

muta-tions in the N-terminus of histone H3, cause defects in

both telomeric gene silencing [19,20,26] and resistance

to DNA-damaging agents [20,27] Such defects are

reproduced by the substitution of Lys for Arg at

posi-tion 12 of histone H4, but not at posiposi-tion 5 [26,27]

Moreover, Hat1 is recruited to the sites of DNA

double-strand breaks, where it is specifically required

for the histone H4 acetylation on Lys12, but

appar-ently not on Lys5 [28]

Despite many correlations linking the acetylation of

histone H4 with chromatin assembly, direct evidence

actually indicates that the specific histone H4

Lys5⁄ Lys12 diacetylation pattern, and also the HAT-B

enzymes that generate it, are dispensable for this

pro-cess In yeast, the substitution mutation of Lys5 and

Lys12 of histone H4, in combination with deletion of

the histone H3 N-terminus, does not result in defective

chromatin assembly, either in vitro or in vivo [29],

although the acetylation state of newly synthesized

yeast histone H4 is not known Likewise, in chicken

DT40 cells, it has been shown that HAT1 is not

neces-sary for replication-coupled chromatin assembly [30]

Thus, the biological role of the conserved Lys5⁄ Lys12

acetylation of histone H4 and hence the function of

the HAT-B enzymes found in all eukaryotes are

elu-sive

Many reports have described the characterization

and the site specificity of type B enzymes from

differ-ent species in vitro [10,13,15–17,21–23,31], but analyses

of their in vivo specificity are few and not at all

conclu-sive [19] Only recently has it been demonstrated in chicken DT40 cells that the homozygous HAT1 dele-tion results in a reduced diacetyladele-tion level on Lys5 and Lys12 of histone H4 in a cytosolic extract [30]

S cerevisiae Hat1 was the first HAT to be identified [22], and moreover its biochemical properties, both as

an isolated subunit and as part of the HAT-B complex [13,19,20,25,31–33], have been studied Despite all these studies, its in vivo site specificity has not been directly ascertained

In this article, we demonstrate that both Lys12 and Lys5 of non-chromatin-bound histone H4 are authen-tic targets of acetylation for the S cerevisiae HAT-B complex in vivo Moreover, these positions are acety-lated in histone H4 associated with the HAT-B enzyme from the yeast soluble fraction The requirements for the distinct components of the complex for the acetyla-tion and the associaacetyla-tion of histone H4 have also been analyzed

Results

Direct identification of Lys12 and Lys5 of soluble, non-chromatin-bound histone H4 as in vivo targets of acetylation by yeast Hat1

Previous work in our laboratory failed to detect any defect in the in vivo steady-state level of acetylation on Lys12 of histone H4 in hat1, hat2 or hif1 null mutant strains as compared to the wild-type under normal growth conditions [19] The apparent independence of histone H4 Lys12 acetylation from the HAT-B enzyme

in vivowas actually interpreted as a consequence of the very short half-life of this modification, which would make detection difficult

We persisted in investigating the in vivo specificity of the yeast HAT-B complex, and found that incubation

of cells with hydroxyurea (HU) resulted in an increase

of the histone H4 isoform with acetylated Lys12 (H4K12ac) in a HAT1-dependent manner (Fig 1A)

HU is a ribonucleotide reductase inhibitor that causes

a depletion of deoxynucleotides, and thereby slows down DNA synthesis The acetylation analysis was carried out by immunoblotting with a specific antibody

to H4K12ac Cells were incubated in the presence of

200 mm HU (a concentration commonly used to syn-chronize yeast cultures) for the indicated time periods

In wild-type cells, HU promoted acetylation of his-tone H4 Lys12, which is reflected by an increase in the H4K12ac level 2 h after HU addition In contrast, the H4K12ac amount did not significantly change in hat1D mutant cells, even after 12 h of incubation (Fig 1A) Importantly, an antibody against the C-terminus of

histone H3 (anti-H3Ct), used as a control for histone

loading, did not detect differences in the amount of

histone H3 between the two strains, indicating that

cells lacking Hat1 display normal levels of histones

during the course of HU treatment

To investigate whether HU induces an increase of

the Hat1 protein, we used a yeast strain expressing a

hemagglutinin (HA)-tagged version of Hat1 The

Hat1–HA protein level did not increase with HU

incu-bation time, but actually slightly diminished (Fig 1B)

Apparently, HU treatment does not alters the

enzy-matic activity of the HAT-B complex, as the

chro-matographic HAT profiles and activity levels, in

particular that corresponding to the HAT-B peaks,

were very similar in HU-treated and untreated cells

(supplementary Fig S1)

We investigated whether HAT1-dependent

his-tone H4 Lys12 acetylation was also increased in

response to other genotoxic agents, such as

methyl-methanesulfonate, phleomycin, and 4-nitroquinoline

n-oxide (4NQO) Like HU, these other agents increased the amount of H4K12ac in wild-type but not

in hat1D cells (supplementary Fig S2) Fluorescence-activated cell sorting (FACS) analysis (supplementary Fig S2) revealed a certain degree of qualitative corre-lation between the H4K12ac level and the enrichment

of the culture in S-phase cells The most potent effect

on both was generated by HU

In order to further examine the effect of HAT1 dele-tion on acetyladele-tion of histone H4 Lys12, we purified histones from wild-type and hat1D mutant yeast chro-matin, before and after incubation with 200 mm HU for 3 h In agreement with our previous results [19], immunoblotting analysis revealed no difference in his-tone H4 Lys12 acetylation between purified hishis-tones from wild-type and mutant cells left without HU treat-ment However, in striking contrast to the results obtained with whole cell extract (WCE), we did not observe a significant difference in histone H4 Lys12 acetylation between the two strains after HU incuba-tion (Fig 2A) As histones were obtained from isolated chromatin, these results show that HU-induced, Hat1-dependent histone H4 acetylation (Fig 1A) is restricted to non-chromatin-bound, soluble, ‘free’ his-tone H4 To investigate this further in yeast, sphero-plasts of wild-type and hat1D cells (HU-treated and untreated) were lysed and fractionated by centrifuga-tion into soluble and chromatin pellet fraccentrifuga-tions, as shown in Fig 2B A significant amount of H4K12ac was found in the soluble fraction of wild-type cells after incubation with HU, but not in hat1D mutant cells (Fig 2C) H4K12ac was even detected in the solu-ble fraction of untreated wild-type cells, although its level increased substantially after treatment with HU

In addition, antibodies against the recombinant yeast full-length histone H4 (anti-ryH4) and the C-terminus

of histone H3 (anti-H3Ct), which recognize the corre-sponding histones independently of the modification state, revealed that HU treatment increased the amount of soluble histone H4 and histone H3, as had been previously described [4,15,34] Such an accumula-tion of histones was identical in wild-type and hat1D mutant cells With respect to the chromatin fractions, histone H4 Lys12 acetylation was not significantly dif-ferent between wild-type and hat1D cells, supporting the results obtained with purified histones

We investigated the requirement for Hat1 for acety-lation of other acetylatable positions on histone H4 and histone H3 in the soluble fraction (Fig 3) The results clearly indicate that histone H4 Lys5 is an authentic target for Hat1 in vivo As shown in the immunoblot in Fig 3, like H4K12ac, the histone H4 isoform with acetylated Lys5 (H4K5ac) was detected

Fig 1 Hydroxyurea (HU) treatment of yeast cells reveals the

involvement of Hat1 in the acetylation of histone H4 on Lys12

in vivo (A) Solid HU was added to exponential-phase cultures of

strains W303-1a (wild-type; +) and RS1263 (hat14; )) to a final

concentration of 200 m M and, at the indicated time points, equal

amounts of cells were collected, and used for preparation of

WCEs Proteins were resolved by 15% SDS ⁄ PAGE, and transferred

to a nitrocellulose membrane The membrane was stained with

Ponceau S (upper panel), and probed with the antiserum against

histone H4 acetylated on Lys12 (middle panel) As a specific

load-ing control for histones, a second immunoblot with an antibody

against the C-terminus of histone H3 (a-H3Ct, lower panel) was

carried out (B) Strain BQS1154, expressing HA-tagged Hat1, was

incubated with 200 m M HU, and, at different time points, cells from

identical volumes were processed to obtain WCEs Hat1 protein

levels were revealed by immunoblotting with mouse 12CA5

anti-body against the HA epitope Mr, molecular mass markers.

in the soluble fraction of wild-type cells, but not of

hat1D mutant cells In addition, HU treatment also

increased the amount of Hat1-dependent H4K5ac In

contrast, acetylation at the other potentially

acetylat-able sites within the histone H4 N-terminus, Lys8 and

Lys16, was hardly visible on soluble histone H4,

although strong bands on histone H4 in purified

con-trol histones were observed In any case, their

acetyla-tion levels were independent of Hat1

In budding yeast, there is evidence that the

N-termi-nal tail of newly synthesized histone H3 is

monoacety-lated preferentially on Lys9, but also on Lys14, Lys23,

or Lys27 [8] Except for Gcn5, which is responsible for

histone H3 Lys9 acetylation [9], the acetylation

enzymes for the other positions are unknown We

detected histone H3 acetylated at these positions in the

soluble fraction, although with varying degrees of

intensity Except for the histone H3 isoform with

acet-ylated Lys14, which apparently did not change, the

other three histone H3 isoforms increased after HU

treatment In neither case did loss of Hat1 have any

effect on acetylation at these Lys residues (Fig 3)

Recent evidence also indicates acetylation in the

glob-ular domains of histone H3 and histone H4 In yeast,

acetylation of histone H3 Lys56 and histone H4 Lys91

has been described, and both seem to be linked to

nucleosome assembly [35,36] As shown in Fig 3, the

histone H3 isoform with acetylated Lys56 and the his-tone H4 isoform with acetylated Lys91 were detected in the soluble fraction, and the levels of both were signifi-cantly increased by HU treatment, but the amount of neither of them was dependent on the presence of Hat1 Although some caution must accompany the interpretation of the immunoblotting assays, due to a possible lack of reactivity or specificity of the anti-bodies, our results indicate that Hat1 is apparently not involved in the acetylation of any site on soluble histone H4 and histone H3 except for Lys12 and Lys5

of histone H4

Involvement of different components of the yeast HAT-B complex in the acetylation of soluble histone H4

We examined the presence of histone H4 acetylation

on Lys12 and Lys5 in soluble fractions obtained from wild-type and hat1D, hat2D and hif1D deletion strains Deletion of the HAT2 gene caused a considerable reduction in the acetylation of Lys12 and Lys5 of solu-ble histone H4 (Fig 4) Consistently, very low immu-nosignals were also obtained in soluble fractions of HU-treated hat1D and hat2D cells In contrast, the levels of Lys12 and Lys5 acetylation were equal in wild-type and hif1D soluble fractions On the other

Fig 2 Hat1 acetylates Lys12 of soluble,

non-chromatin-bound histone H4 in vivo (A)

Histones purified from wild-type (W303-1a)

and hat1D mutant (RS1263) cells, before

and after incubation with HU, were

sepa-rated by SDS⁄ PAGE and immunoblotted

with antibody to H4K12ac (lower panel)

His-tone species are indicated on the

Pon-ceau S-stained membrane (upper panel) In

order to facilitate the detection of even

small differences in acetylation degree, two

distinct amounts of histones were loaded

(1.0 and 0.25 lg) (B) Schematic

representa-tion of the experimental procedure used to

fractionate yeast cells (C) Equal numbers of

wild-type and hat1D mutant cells were

har-vested at 0 or 4 h after incubation with HU.

Soluble and chromatin-associated proteins

(pellet) were fractionated as illustrated in

(B), and histones in both fractions were

ana-lyzed by immunoblotting with antibody to

H4K12ac, anti-ryH4, and anti-H3Ct Owing

to the low level of histones in the soluble

fraction, approximately 10 times more was

loaded as compared to the pellet fractions.

hand, the amount of total soluble histone H4 was

similar in all four strains, and was equally increased by

HU treatment (as revealed with anti-ryH4) These data

indicate that Hat2, but not Hif1, participates in the

catalytic function of the HAT-B complex in vivo

Hat1-dependent acetylation of soluble histone H4

throughout the cell cycle and in Rad53-deficient

cells

We previously observed fairly constant levels of yeast

Hat1 protein throughout the cell cycle [19] We

there-fore checked the presence of soluble H4K12ac at

distinct cell cycle stages Wild-type and hat1D mutant cells growing asynchronously were left without treat-ment or incubated with either a-factor, which arrests cells in G1phase, or hydroxyurea or nocodazole, which prevent the G2⁄ M transition, or transferred to minimal medium without a nitrogen source, which arrests cells in G0phase The cell cycle phases of the arrested cells were confirmed by DNA flow cytometry Soluble histone H4 Hat1-dependently acetylated on Lys12 was present in cells arrested at all cell cycle stages, G1, S, G2⁄ M and also G0 (Fig 5A) Similar results were obtained with cells at different cell cycle stages from synchronized cultures by release from a a-factor block [19] (results not shown)

In S cerevisiae, the checkpoint protein kinase Rad53 regulates histone protein levels, and thus Rad53-defi-cient yeast cells exhibit abnormally high amounts of soluble histones [34] We therefore investigated whether such an excess of soluble histone H4 is also acetylated

by Hat1 For this purpose, we deleted the HAT1 gene

in wild-type and rad53D mutant strains, and examined the levels of H4K12ac in the corresponding soluble fractions (Fig 5B) As expected, asynchronously grow-ing rad53D mutant cells displayed a higher amount of soluble histone H4 than wild-type cells, but only the excess soluble histone H4 from HAT1 cells was acety-lated on Lys12 The accumulation of HAT-B-dependent acetylation of Lys12 in Rad53-deficient cells was further confirmed on yeast strains harboring the chromosomal

Fig 3 Histone H4 Lys12 and histone H4 Lys5 are the only

acetyla-tion sites in soluble histone H4 and histone H3 that are dependent

on Hat1 Soluble histones from wild-type (W303-1a) or hat1D

(RS1263) cells, in the presence or absence of HU, were analyzed

by immunoblotting using antibodies against different acetylated

iso-forms of histone H4 and histone H3 A Ponceau S-stained

mem-brane (top panel) and an immunoblot with anti-H3Ct (lowest panel)

are shown as a loading control Purified yeast histones (yhis) were

included to check antibody reactivity.

Fig 4 Acetylation on Lys12 and Lys5 of soluble histone H4 by the HAT-B complex in vivo Soluble fractions were prepared from wild-type (W303-1a), hat1D (RS1263), hat2D (YSTT11) and hif1D (YSTT49) yeast cells, before and after incubation with 200 m M HU for 3 h Histones in these extracts were analyzed by immunoblot-ting using antibody to H4K12ac, antibody to H4K5ac, and anti-ryH4.

RAD53gene under the glucose-switched off GAL1

pro-moter in wild-type, hat1D, hat2D or hif1D strains

(supplementary Fig S3) Results showed that Lys12

acetylation of excess soluble histone H4 present in

Rad53-deficient cells was absolutely dependent on Hat1

and Hat2, but not on Hif1

Histone H4 acetylated on Lys12 and Lys5 is

associated with the HAT-B complex in the yeast

soluble fraction

To gain further insights into the organization and the

molecular determinants of the yeast HAT-B complex,

we attempted to determine: (a) whether HAT-B

enzyme, present in the soluble fraction, contains

asso-ciated histone H4; (b) the acetylated sites; and (c) the

involvement of the different HAT-B components in

histone H4 binding To address these questions, we

performed immunoprecipitation experiments with

solu-ble extracts from yeast strains that express tagged

forms of each of the three components of the HAT-B

complex (Hat1–HA, Hat2–HA, and Hif1–Myc)

Im-munoprecipitates (bound fractions), input materials and unbound materials were examined by immuno-blotting with antibodies against histone H4, his-tone H3, and acetylated isoforms All three HAT-B components, Hat1, Hat2 (Fig 6A, lanes 6 and 15, respectively) and Hif1 (Fig 6B, lane 27) coimmunopre-cipitated H4K12ac Furthermore, histone H4 present

in the soluble extracts from yeast cells lacking Hat1 or Hat2 was not coprecipitated with any of the other complex components (Fig 6A, lanes 9 and 21; and Fig 6B, lanes 30 and 33), indicating that both Hat1 and Hat2 are necessary for histone H4 binding In contrast, both Hat1 and Hat2 were still able to copre-cipitate H4K12ac in the absence of Hif1 (Fig 6A, lanes 12 and 18), indicating that Hif1 is dispensable for the interaction of histone H4 with Hat1⁄ Hat2 Results corresponding to Fig 6A,B were entirely reproduced when blots were probed with the antibody

to H4K5ac (results not shown)

When the blots were probed with anti-H3Ct, an im-munosignal was not obtained in any of the immuno-precipitates of Hat1, Hat2, or Hif1 (Fig 6A, lanes 6,

Fig 5 Soluble histone H4 at different cell

cycle stages and excess histone H4 in

Rad53-deficient cells is acetylated by the

HAT-B complex (A) Wild-type and hat1D

mutant cells were grown asynchronously to

exponential phase, and either harvested

(asy.) or arrested in G1phase (by incubation

with 4.5 lgÆmL)1a-factor for 3 h; a), in

S phase (200 m M HU for 3 h), in

G2⁄ M phase (15 lgÆmL)1nocodazole, 3 h);

NZ and in G 0 phase by nitrogen deprivation

for 14 h (DN) Cells were fractionated, and

the soluble fractions were analyzed by

immunoblotting with antibody to H4K12ac

and anti-ryH4 Cell cycle stages were

monitored by FACS (bottom) (B) Soluble

fractions from the yeast strains YAV49

(sml1D), BQS1386 (hat1D, sml1D), YAG101

(rad53D, sml1-1) and BQS1358 (hat1D,

rad53D, sml1-1) were analyzed with

antibody to H4K12ac and anti-ryH4 These

strains also bear an sml1 mutation to

suppress the lethality due to rad53 deletion

[34].

12, 15 and 18; and Fig 6B, lane 27) These results are

disturbing, because it is assumed that histone H3 and

histone H4 form tetramers [37] or dimers [38], with an

equal stoichiometry It is well known that histone H3

is particularly susceptible to proteolytic degradation

We cannot completely rule out the possibility that

his-tone H3 proteolysis is also occurring in yeast soluble

extracts in our experiments, but its presence in input

and unbound fractions argues against this possibility

Moreover, intact histone H3, and also H4K12ac, were

detected in immunoprecipitates from soluble extracts

of cells expressing Flag-tagged Cac1 or Asf1, two

his-tone H3⁄ H4 chaperones (Fig 6C) These additional

controls also indicate the absence of specific

his-tone H3 degradation under our immunoprecipitation

assay conditions Likewise, none of the specific

anti-bodies to acetylated histone H3 used in Fig 3

gener-ated immunosignals corresponding to histone H3 on the HAT-B complex immunoprecipitates (results not shown) Altogether, our data suggest that histone H3

is not part of the HAT-B complex in the soluble frac-tion of yeast cells

In vivo, the HAT-B complex requires an acetylatable Lys at position 12 for acetylation

on Lys5, but not for binding histone H4 Recombinant yeast Hat1, as well as native HAT-B enzymes from various species, modify histone H4

in vitro on Lys12 preferentially over Lys5 [13,17,22,23,25,31] Thus, HAT-B complex acetyltrans-ferase activity results in an ordered acetylation, with Lys12 being acetylated before Lys5 [23,31] To investi-gate the in vivo requirement for histone H4 Lys12 on

Fig 6 Histone H4 acetylated on Lys12 and Lys5 is bound to the HAT-B complex in the soluble cell fraction Soluble extracts of yeast cells expressing Hat1–HA, Hat2–HA (A) or Hif1–Myc (B) were used for immunoprecipitation with rat monoclonal antibody to HA (3F10) and mouse monoclonal antibody to Myc (9E10), respectively Yeast strains expressing a tagged HAT-B component, but with a deletion of any other companion protein (hat1D, hat2D, or hif1D), were also examined The input (I), unbound (U) and bound (B) fractions were analyzed by immu-noblotting with antibody to H4K12ac, anti-ryH4, and anti-H3Ct Untagged wild-type strain (wt, no tag) W303-1a was used as a negative con-trol (C) Extracts of exponentially growing cells expressing Cac1–Flag or Asf1–Flag were used for immunoprecipitation with Flag M2 antibody agarose beads I, U and B fractions were probed by immunoblotting with anti-H3Ct or antibody to H4K12ac Purified yeast histones (yhis) were used as a control Yeast strains (with the relevant gene modifications in parentheses) used in these experiments were: (A) BQS1154 (HAT1-HA); BQS1189 (HAT2-HA); BQS1172 (HAT1-HA, hat2D); BQS1184 (HAT1-HA, hif1D); BQS1309 (HAT2-HA, hat1D) and BQS1304 (HAT2-HA, hif1D); (B) BQS1187 (HIF1-MYC); BQS1202 (HIF1-MYC, hat1D); and BQS1225 (HIF1-MYC, hat2D); (C) YAV49 (CAC1-FLAG); and YAV52 (ASF1-FLAG).

the acetylation of Lys5 and also on the histone

H4–HAT-B complex association, we made use of yeast

strains expressing wild-type or a K12R substitution

mutant histone H4 (H4K12R) from a centromeric

plasmid as the only source of histone H4 In addition,

these strains contained Hat1 or Hif1 tagged with the

HA epitope First, we checked that K12R substitution

does not interfere with the recognition of H4K5ac by

the antibody to H4K5ac (supplementary Fig S4) In

agreement with this, antibody to H4K5ac yielded

bands with similar intensity on WCEs prepared from

cells containing wild-type histone H4 or H4K12R

(Fig 7A) However, when soluble fractions were

ana-lyzed with the same antibody, cells expressing

wild-type histone H4 were characterized by a well-defined

band, whereas in cells expressing H4K12R, only a very weak signal was detected (Fig 7A) These results imply that an acetylatable Lys at position 12 is essential for the efficient acetylation of Lys5 of soluble histone H4

Analyses of histone H4 association with the HAT-B complex were performed by coimmunoprecipitation and subsequent immunoblotting As expected, the antibody to H4K12ac (control) generated a signal in immunoprecipitates from soluble extracts containing wild-type histone H4 (Fig 7B, lanes 3 and 9) but not

in those containing H4K12R (Fig 7B, lanes 6 and 12) Remarkably, anti-ryH4 revealed that as much Hat1–

HA as Hif1–HA coimmunoprecipitated both wild-type and K12R mutant histone H4 (Fig 7B, lanes 3, 6, 9 and 12) This finding demonstrates that Lys12 and its acetylation are not involved in the binding of his-tone H4 to the HAT-B complex In addition, the anti-body to H4K5ac revealed that H4K12R coprecipitated with Hat1 was acetylated very weakly on Lys5 (Fig 7B, lane 6)

Finally, we investigated the in vitro activity of yeast Hat1 and Hat1-dependent type B complex towards wild-type or K12R mutant histone H4 in purified yeast core histones A HAT-B complex, partially purified by anion exchange chromatography of soluble extracts from wild-type cells, was used in the enzymatic assays (Fig 8A) As a control, equivalent chromatographic fractions from a hat1D mutant strain were also assayed A recombinant yeast Hat1 (ryHat1) was also included in the assays (Fig 8C) Whereas wild-type histone H4 was efficiently acetylated by native HAT-B enzyme, H4K12R was modified very weakly, if at all (Fig 8A) Thus, in vitro as well as in vivo, Lys5 in H4K12R represents only a very poor substrate for the yeast HAT-B complex Importantly, this finding sup-ports the in vivo results that an acetylatable Lys at position 12 of soluble histone H4 is required for fur-ther modification on Lys5 Moreover, immunoblotting with antibody to H4K5ac, after HAT assays, revealed that the yeast HAT-B complex indeed acetylates Lys5

on wild-type histone H4 Figure 8B shows that incuba-tion of yeast or chicken histones with acetyl-CoA and chromatographic fractions containing the HAT-B enzyme increased the H4K5ac immunosignal, which was not the case when hat1D fractions (or buffer solu-tion) were used Thus, these results demonstrate that the yeast HAT-B complex acetylates Lys5 in the con-text of intact histone H4 in vitro

In contrast to Hat1 as part of the HAT-B complex, recombinant Hat1 was able to acetylate H4K12R, although to a lesser extent than wild-type histone H4 (Fig 8C)

Fig 7 Exchange of Lys for Arg at position 12 of histone H4

pre-vents acetylation of Lys5 in the soluble cell fraction (A) H4K5ac in

WCEs and soluble fractions from yeast strains expressing either

wild-type (PKY501, wt H4) or the K12R mutated version (LDY105,

K12R H4) of histone H4 was analyzed by immunoblotting (B)

Hat1–HA and Hif1–HA were immunoprecipitated from soluble

extracts of cells that express wild-type histone H4 or H4K12R.

Input (I), unbound (U) and bound (B) fractions were analyzed for the

presence of associated histone H4 Yeast strains: BQS1399

(expressing Hat1–HA, wild-type histone H4); BQS1401 (Hat1–HA,

K12RH4); BQS1403 (Hif1–HA, wild-type histone H4); and BQS1405

(Hif1–HA, K12RH4) Blots were probed with antibody to H4K12ac,

anti-ryH4, and antibody to H4K5ac.

The main result of this study has been the

demonstra-tion that the S cerevisiae HAT-B complex is involved

in the acetylation of both Lys12 and Lys5 of soluble

histone H4 in vivo We showed that both Hat1 and Hat2 are essential for this specific histone H4 Lys12⁄ Lys5 acetylation, whereas the third component

of the complex, Hif1, is not These results are in agree-ment with in vitro data indicating that the absence of Hif1 alters neither the activity nor the specificity of the rest of the HAT-B enzyme [19,20], whereas Hat2 has the ability to enhance the catalytic potential of the Hat1 subunit [13] Therefore, the functional role of Hif1 in the HAT-B complex is downstream of the acet-ylation of histone H4

It had been previously determined that, in vitro, the yeast HAT-B complex exclusively acetylates Lys12

on histone H4 N-terminal synthetic peptides [13,22], whereas recombinant Hat1 modifies Lys12 and Lys5 [13,25] Moreover, in yeast cells, indirect evidence has shown the involvement of Hat1 in the modification of Lys12, although not of Lys5, of histone H4 [26–28] However, we have demonstrated that Lys5 is a bona fide target for acetylation by the yeast HAT-B complex

in vivo Remarkably, the inability of the HAT-B complex to acetylate histone H4 containing the K12R substitution, both in vivo and in vitro, indicates a sequential order of acetylation, with Lys12 being modi-fied before Lys5 As Arg mimics unacetylated Lys, this inability to use Lys5 as a target on H4K12R strongly suggests that acetylation of Lys12 is a prerequisite for the subsequent acetylation of Lys5 An identical sequential order of site usage has been found for HAT-B enzymes isolated from maize and rat liver [23], and also from human cells [31], in vitro Yeast recom-binant Hat1 is able to modify H4K12R, which comple-ments previous findings showing less stringent site specificity for Hat1 alone [13,25], and suggests the involvement of the other complex components in the site selection mechanism In contrast to previous stud-ies [13,22], we have found that, even in vitro, the yeast HAT-B complex modifies Lys5 as well as Lys12, just like other type B enzymes from diverse species [10,15,16,23,31] The reason for this discrepancy may

be the different substrates used Earlier experiments, indicating Lys12 as the only acetylation site, were car-ried out with histone H4 N-terminal peptides [13,22]

We have used whole histone H4 of yeast or chicken erythrocytes, as in numerous other studies [10,15,16,23,31] We suggest that an interaction of his-tone H4, beyond its N-terminus, with the HAT-B com-plex is needed in order to establish a physiological acetylation pattern Interestingly, in line with this idea,

in vitro, the human HAT-B enzyme acetylates the his-tone H4(1–41) N-terminal fragment more efficiently than the shorter histone H4(1–34) fragment [16] It seems reasonable that at least part of the differential

Fig 8 Effect of K12R substitution on histone H4 acetylation by

the yeast HAT-B complex and ryHat1 in vitro (A) Soluble extracts

from wild-type and hat1D strains were subjected to Q-Sepharose

HP chromatography, and fractions containing HAT-B enzyme

(HAT1) were used for HAT activity assays Equivalent fractions

from the deletion mutant strain were also tested as negative

con-trols (hat1D) These acetyltransferase assays were carried out by

mixing 2 lg of purified whole yeast histones, containing wild-type

histone H4 (wt H4) or K12R mutant histone H4 (K12R H4),

[1- 14 C]acetyl-CoA, and aliquots of the chromatographic fractions (2

and 8 lL) After incubation, histones were separated by 15%

SDS ⁄ PAGE, and gels were stained with Coomassie brilliant blue

(upper panel), and subsequently subjected to fluorography (lower

panel) Histones are indicated on the left (B) Yeast histones with

wild-type histone H4 and chicken erythrocyte core histones were

assayed with chromatographic fractions containing (HAT1) or

lack-ing (hat1D, or buffer) HAT-B enzyme, and subsequently analyzed by

immunoblotting with antibody to H4K5ac (C) Recombinant yeast

Hat1 (ryHat1; approximately 0.01 and 0.04 lg) was used in HAT

assays carried out as in (A).

potential as substrates of the two peptides is due to

different positions being modified by the enzyme Our

data indicate a site specificity of the yeast HAT-B

complex that exactly matches the specificity of other

type B HATs [10,15,16,23,31], thus pinpointing a much

higher degree of conservation of these enzymes than

previously assumed

The levels of acetylation on Lys16 and Lys8 are

extremely low in the soluble histone H4 of wild-type

cells, and also nearly undetectable on Lys12 and Lys5

in hat1D cells These observations strongly suggest

that, in yeast, Lys12 and Lys5 are the only N-terminal

positions that are acetylated in soluble histone H4, and

that the HAT-B complex must be the only enzyme

involved in this specific modification However, in

con-trast with these results, histone H4 acetylated on

Lys16 and, to a greater degree, on Lys8 has been

detected in the cytoplasmic fraction of chicken DT40

cells In addition, chicken cells lacking Hat1 retain a

significant level of histone H4 Lys12 and Lys5

tion in the soluble fraction [30] Although the

acetyla-tion pattern of the soluble H4 histones of yeast and

chicken could be different, contamination with

chro-matin could also explain the presence of

Hat1-indepen-dent histone H4 Lys12 and Lys5 acetylations and

other acetylated positions in the soluble fraction

Hat1-dependent acetylated histone H4 is present in

the soluble fraction in different cell cycle stages, which

shows that, in addition to DNA replication, it may

also participate in other processes outside of S phase

A dynamic nucleosome disassembly⁄ reassembly

pro-cess is a well-established feature of sites undergoing

transcription [39,40], but a global histone H4 exchange

independent of replication and transcription has also

been described in yeast [41] Reassembly makes use of

histones from the soluble pool [40,41], in which

his-tone H4, as our results indicate, must be acetylated by

the HAT-B complex

In addition, we have found that excess histone H4

accumulating in the soluble fraction in cells treated

with HU or in cells deficient in the Rad53-dependent

histone degradation pathway [34] is acetylated by

Hat1 It therefore seems that all new histone H4

mole-cules appearing in the soluble fraction contain the

spe-cific Lys12⁄ Lys5 acetylation pattern generated by the

type B enzyme

In contrast to studies on different species where

newly synthesized histone H4 in a diacetylated form

was obtained from chromatin [4,5,42,43], we did not

detect HAT-B-dependent acetylation on chromatin

histone H4 In yeast, the Hat1-dependent acetylation

could be eliminated immediately upon the deposition

of histone H4 into chromatin It cannot be ruled

out that this deacetylation occurs either during, or even prior to, histone H4 deposition Mutational analysis has shown that specific Lys residues in the N-termini of histone H3 and histone H4 play critical roles in nuclear import, suggesting that acetylation could serve to release histones from nuclear transport factors [44] Formally, for such a role, the deacety-lation would not necessarily have to be post-deposition

Although, in vitro, Hat1 and Hat2 [44] and also Hif1 [20] bind H4⁄ H3 histones, in vivo, both Hat1 and Hat2, together, are involved in the physical interaction with histone H4, whereas Hif1 is not Furthermore, both targets of acetylation, Lys12 and Lys5, are found to be acetylated in the histone H4 bound to the HAT-B complex Current models propose that the acetylated state at the his-tone H4 N-terminus is involved in the stable binding

of histone H4 to the HAT-B complex [1,20,24] How-ever, this is not consistent with the ability of H4K12R, which also lacks acetylation at Lys5, to be bound by the HAT-B complex Even Hif1, in the context of the HAT-B complex, is associated with histone H4 that lacks acetylation at the N-terminal tail As Hif1 exhibits chromatin assembly activity

in vitro [20], we must not rule out completely its par-ticipation in chromatin assembly independently of histone H4 acetylation Our data also suggest that the N-terminus (at least segment 1–12) is not involved in the stable association of histone H4 with the complex Our previous two-hybrid assays indi-cated an in vivo interaction between Hif1 and frag-ment 1–59 of histone H4 that was dependent on Hat2 [19]; thus, the portion of histone H4 involved

in the interaction with the HAT-B complex must be located between residues 13 and 59 Verreault et al [16] found that helix 1 (residues 31–40) of his-tone H4, situated in the hishis-tone-fold domain, is criti-cal for binding to the Hat2 human homolog p46 Reasonably, the yeast HAT-B complex could use the same determinants to bind histone H4, although additional contacts with Hat1 seem to be necessary for efficient and stable binding of histone H4 It is possible that all or some of these interactions are also responsible for the acetylation specificity of the HAT complex discussed above

Although histone H3 has always been found with histone H4 [37], and they are usually obtained from the cells in a 1 : 1 ratio, we have not detected his-tone H3 in the HAT-B complex from the yeast soluble fraction The controls carried out argue against the specific proteolytic degradation of histone H3 in the yeast fractions obtained and processed by our