Critical residues for histone acetylation by Gcn5, functioning in Ada

Critical residues for histone acetylation

by Gcn5, functioning in Ada

and SAGA complexes, are also required

for transcriptional function in vivo

Lian Wang,

Lin Liu,

and Shelley L. Berger

Molecular Genetics Program, The Wistar Institute, Philadelphia, Pennsylvania 19104 USA

Several previously known transcription cofactors have been demonstrated in vitro recently to be histone

acetyltransferases and deacetyltransferases, suggesting that remodeling of chromatin through histone

acetylation plays a fundamental role in gene regulation. Clear evidence has not yet been obtained, however, to

demonstrate that histone acetylation is required for gene activation in vivo. In this study we performed an

alanine-scan mutagenesis through the HAT (h

istone acetyltransferase) domain identified previously by

deletion mapping in recombinant yeast Gcn5. We identified multiple substitution mutations that eliminated

completely Gcn5’s ability to potentiate transcriptional activation in vivo. Strikingly, each of these mutations

was also critical for free and nucleosomal histone acetylation by Gcn5 functioning within the native yeast

HAT complexes, Ada, and SAGA. Moreover, the growth phenotypes of these mutations as measured by colony

size and liquid growth assay closely tracked transcription and HAT activities. In contrast, mutations that did

not affect in vivo function of Gcn5 were able to acetylate histones. These data argue strongly that acetylation

is required for gene regulation by Gcn5 in vivo, and support previous arguments that nucleosomal histones are

among the physiological substrates of acetylation by Gcn5.

Gene expression is a highly regulated process at the level

of transcriptional activation. Recruitment of the tran-

scriptional basal machinery by activators and cofactors

has been shown to be an important component of gene

regulation in vitro (Zawel and Reinberg 1995) and in vivo

(Struhl 1995). Transcriptional activator proteins facili-

tate association of components of the basal machinery

with the TATA box and start site of transcription

(Triezenberg 1995). Under physiological conditions,

DNA is complexed with histones forming nucleosomes

that assemble into higher order chromatin (Wolffe 1992).

Chromatin assembly strongly inhibits transcription

(Owen-Hughes and Workman 1994; Paranjape et al.

1994), probably by blocking binding of transcription fac-

tors to their cognate DNA-binding sites. Therefore, in

vivo, transcription factors must overcome the DNA ac-

cess problem posed by nucleosomes. Moreover, genetic

manipulations of histones confirm their important role

in regulation of transcription (Grunstein 1990).

Both transcriptional activators and cofactors have

critical roles in altering the repressive chromatin struc-

ture to promote binding of basal factors. Activation do-

mains have been shown to be involved in remodeling

chromatin (Svaren et al. 1994), and nucleosome struc-

tural changes occur in the absence of transcription sug-

gesting that activators may induce transcription by caus-

ing chromatin changes before initiation (Fascher et al.

1993). Transcriptional cofactors also activate or repress

transcription by destablizing or stablizing the repressive

nucleosome structure (Kingston et al. 1996). For ex-

ample, the SWI/SNF family of protein complexes acti-

vates transcription by disrupting the nucleosome struc-

ture in an ATP-dependent manner to increase the bind-

ing of transcription factors (Cote et al. 1994; Peterson et

al. 1994).

Nucleosomes are composed of ∼146 bp of DNA

wrapped around the histone octamer, consisting of two

molecules each of the histones H2A, H2B, H3, and H4

(Wolffe 1992). Each core histone contains a globular do-

main and extended basic amino-terminal tail that stabi-

lizes histone octamer association with DNA through

ionic interactions. Covalent modifications of nucleo-

somal histones have been correlated with modulation of

chromatin structure and function (Bradbury 1992). These

modifications include phosphorylation, ubiquitination

and, in particular, acetylation of e amino groups of ly-

sines in the histone tails. Acetylation is associated with

both gene activation (hyperacetylation) and gene silenc-

ing (hypoacetylation) (for review, see Loidl 1994; Turner

and O’Neill 1995). Acetylation of histones increases the

affinity of transcription factors for nucleosomal DNA

(Lee et al. 1993; Vettese-Dadey et al. 1996).

These authors contributed equally to this work.

Corresponding author.

E-MAIL [email protected]; FAX (215) 898-0063.

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

The recent identification of histone acetyltransferase

(HAT) and histone deacetylase (HDAC) enzymes as pre-

viously recognized transcription cofactors (Pazin and Ka-

donaga 1997; Wade and Wolffe 1997) suggests a direct

mechanistic connection between the activity of tran-

scription factors and changes in chromatin structure that

precede initiation. The first nuclear HAT enzyme, p55,

was purified and cloned from Tetrahymena, and was

found to be closely related to yeast Gcn5 (Brownell et al.

1996), a known transcription coactivator/adaptor (Guar-

ante 1995). The ability of yeast Gcn5, and its binding

partner yeast Ada2, to associate with activation domains

(Barlev et al. 1995; Chiang et al. 1996; Silverman et al.

1994) and with the TATA-binding protein (TBP; Barlev

et al. 1995), suggests that acetylation of nucleosomal his-

tones through activation domain recruitment of Gcn5

may be a fundamental and novel feature of gene regula-

tion. Supporting this notion were subsequent revelations

that several other transcription coactivators/adaptors

possess intrinsic HAT activity, including additional

Gcn5 family members [hGcn5 (Wang et al. 1997; Yang et

al. 1996) and P/CAF (Yang et al. 1996)] as well as unre-

lated CBP/p300 (Bannister and Kouzarides 1996; Ogry-

zko et al. 1996) and TAF

250 (Mizzen et al. 1996).

Conversely, the first HDAC was identified (Taunton

et al. 1996) and strikingly, it also was a previously

known transcription factor, Rpd3 (Vidal and Gaber

1991), possessing transcriptional repression activity.

Several other transcriptional repressors have been shown

to be HDACs, including Sin3 (Hassig et al. 1997; Kadosh

and Struhl 1997; Laherty et al. 1997; Zhang et al. 1997)

and the corepressor N-CoR/SMRT (Alland et al. 1997;

Heinzel et al. 1997). Despite these provocative observa-

tions, however, direct evidence has not yet been ob-

tained that acetylation or deacetylation of histones has a

direct role in gene regulation.

Although recombinant proteins have been tested for

histone acetylation and deacetylation activity, recent

studies demonstrate that, in vivo, the HATs (Grant et al.

1997; Horiuchi et al. 1997; Saleh et al. 1997) and HDACs

(Rundlett and Grunstein 1996; Zhang et al. 1997) func-

tion as components of high-molecular-weight com-

plexes. For example, yeast Gcn5 is the catalytic HAT

subunit present in two complexes, termed Ada and

SAGA (S

pt, Ada, Gcn5 acetyltransferase; Grant et al.

1997). The Ada complex also contains Ada2 and Ada3,

which were similarly isolated as suppressors of growth

toxicity caused by overexpression of the potent chimeric

activator composed of the yeast Gal4 DNA-binding do-

main fused to the herpes virus VP16 activation domain

(Berger et al. 1992; Pin˜a et al. 1993; Marcus et al. 1994).

In addition to the Ada proteins, the SAGA complex con-

tains Spt7 (Gansheroff et al. 1995) and Spt20 (Roberts and

Winston 1996) proteins, which were isolated as suppres-

sors of Ty element insertions. Interestingly, Spt20/Ada5

was identified in both Ada and Spt genetic selections

(Marcus et al. 1996; Roberts and Winston 1996) and is

present in the SAGA complex (Grant et al. 1997). The

importance of determining acetylation activity within

these native complexes is illustrated by the observation

that recombinant yeast Gcn5 acetylates only free his-

tones, whereas, in association with Ada and SAGA com-

plexes, Gcn5 acetylates nucleosomal histones (Grant et

al. 1997). In addition, genetic studies demonstrate that

SAGA contains functions involved in transcriptional ac-

tivation distinct from those mediated by the Gcn5/

Ada2/Ada3 group (Roberts and Winston 1997), under-

scoring further the importance of testing HAT activity in

the context of native complexes.

The boundaries of the HAT domain in recombinant

yeast Gcn5 were mapped and the region was shown to be

required for Gcn5’s role in transcriptional activation in

vivo (Candau et al. 1997). This region of the Gcn5 family

members Tetrahymena p55 (Brownell et al. 1996), yeast

Gcn5 (Georgakopoulos and Thireos 1992), human Gcn5

(Candau et al. 1996), and human P/CAF (Yang et al.

1996) is remarkably conserved in primary sequence

(Brownell et al. 1996). Certain sequence motifs have

been identified in these HAT domains and are found in

other acetyltransferases (Reifsnyder et al. 1996; Neuwald

and Lansman 1997). Paradoxically, the other transcrip-

tional coactivators possessing intrinsic HAT activity,

CBP/p300 and TAF

250, display no obvious sequence

homology within the putative HAT domain compared

with the Gcn5 family (Bannister and Kouzarides 1996;

Mizzen et al. 1996; Ogryzko et al. 1996). In fact, recent

data demonstrates that, in addition to histones, p300

acetylates the transcriptional activator p53, and, surpris-

ingly, with nearly equal strength (Gu and Roeder 1997).

The data described above raise important questions—

what are the critical residues for acetylation of histone

substrates in Gcn5 and are these residues required for

Gcn5’s role in transcriptional activation? If so, this

would argue that first, acetylation is a critical and direct

determinant of gene regulation in vivo, and second, that

histones are likely to be a critical physiological sub-

strate. In the current study, we have undertaken alanine-

scan mutagenesis to address these issues. The role of

specific mutations was tested on Gcn5 functioning

within the context of native yeast Ada and SAGA com-

plexes. Our results demonstrate clearly that critical resi-

dues for histone acetylation in Gcn5 are also required for

its in vivo function in growth and as a transcription co-

factor. These data provide compelling evidence that

acetylation by Gcn5 is crucial for transcriptional activa-

tion in vivo.

Results

Construction of alanine-scan mutants

within the putative HAT domain of Gcn5

The four Gcn5 family members that have been identified

are yeast Gcn5, human Gcn5, human P/CAF, and Tet-

rahymena p55. Although the overall sequence identity

between the four proteins is <40%, the amino acid se-

quence in the putative HAT domain is 67% identical and

82% similar (Fig. 1A). The primary structural similarity

is reflected in conservation of function between the hu-

man and yeast HAT domains, whereas no other domain

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 641

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

of human Gcn5 is capable of function in yeast (Wang et

al. 1997). The HAT domain of Gcn5 can be divided into

four subdomains, I, II, III, and IV (Brownell et al. 1996),

based on exceptionally strong sequence identity between

all family members (Fig. 1A).

To identify critical amino acids for HAT enzymatic

activity and to determine whether HAT activity is cru-

cial for Gcn5’s role in transcription and growth, an ala-

nine-scan mutagenesis was performed. Three or four ad-

jacent amino acids were grouped (indicated as three or

four amino acid letter code) and mutagenized to alanine

(Fig. 1B) using site-directed mutagenesis. The mutagenic

oligonucleotides (Table 1) contained a PstI restriction

site allowing initial positive identification of each mu-

tant, which was confirmed by DNA sequencing. Thirty-

one substitution mutants were created and their pheno-

types studied.

Growth complementation by Gcn5 substitution

mutants in the gcn5

−

strain

The ability of the mutant GCN5 genes to complement

phenotypes caused by GCN5 disruption in yeast was

tested. Yeast lacking GCN5 grow poorly on minimal

synthetic media, and transformation of wild-type GCN5

completely complements the growth defect (Georgako-

poulos and Thireos 1992; Marcus et al. 1994). Comple-

mentation by the Gcn5 substitution mutants was tested

by transforming plasmids bearing each of the mutant

genes into the gcn5

−

strain and assessing growth using

colony size (Fig. 2A) and doubling time (Fig. 2B). The

mutant genes were expressed using the natural promoter

and terminator of GCN5 to obtain normal levels of pro-

tein expression. The majority of Gcn5 substitution mu-

tants (19 of the 30 mutants shown, as well as GGI lo-

cated between regions I and II) complemented growth as

well as wild-type Gcn5, as the yeast transformants

yielded colony sizes that were indistinguishable from

wild type (Fig. 2A). Several mutants (YIA, RGY, DNY,

KDY242, and TLM) displayed variable intermediate

colony sizes, ranging between that achieved by yeast

bearing wild-type GCN5 or a disruption of the gene (Fig.

2A). Six mutants (KQL, PKM, FAE, KDY196, AIGY, and

FKK) completely failed to complement the colony

growth defect in the gcn5

−

strain (Fig. 2A).

Relative growth phenotypes were quantitated by mea-

suring doubling times in liquid minimal growth media.

The Gcn5 mutants that exhibited the poorest growth in

the colony assay were analyzed, as well as several repre-

sentative mutants that exhibited wild-type growth. The

results paralleled the colony assays because the most

defective mutants (KQL, PKM, FAE, KDY196, AIGY, and

FKK) grew at essentially the rate of the gcn5

−

strain, and

the wild-type mutants grew comparably with GCN5

(Fig. 2B). Furthermore, the RGY mutant that showed in-

termediate-sized colonies was also intermediate in its

rate of doubling (Fig. 2B). Based on the two growth

complementation assays, we tentatively grouped the

mutants into three classes—those that were essentially

wild type, those that exhibited significant loss of growth

complementation, and those that were completely de-

fective.

The failure of Gcn5 mutants to complement growth

may be caused by defective function of the Gcn5 substi-

tution mutants, or alternatively, by structural defects in

the protein causing either instability or inability to form

requisite protein–protein interactions in vivo. To address

the question of protein stability, the expression levels of

the mutant Gcn5 proteins in yeast were determined rela-

tive to wild-type Gcn5. Gcn5 was expressed under con-

trol of the galactose-inducible GAL1-10 promoter and

intracellular protein expression levels were determined

by Western analysis (Fig. 3, top). Although there was

variation in protein levels, all Gcn5 mutants were pre-

sent at levels comparable with wild-type Gcn5, and, in

particular, all mutants displaying partial or complete de-

fects in the growth assay were detectable in vivo (Fig. 3,

top).

To determine the structural integrity of the mutant

proteins in vivo, we tested interaction with Ada2, as our

previous results demonstrated that Ada2 interaction is

critical for Gcn5 function in vivo (Candau et al. 1997).

Figure 1. Amino acid sequence within

the putative HAT domains of the Gcn5

family. (A) Sequence comparison between

the putative HAT domains of yeast Gcn5,

human P/CAF, human Gcn5 and Tetrahy-

mena p55. The protein sequence within

the putative HAT domains are shown by

single amino acid letter-code. Numbers at

left indicate amino acid residues. The

highly conserved regions I–IV (Brownell et

al. 1996) are indicated at the top. Solid

boxes indicate amino acid identity and

shaded boxes indicate similarity. (B) Yeast

Gcn5 substitution mutants created in the

study. The substitution mutants are

named according to the group of amino ac-

ids mutated to alanine, each of which is

indicated by brackets.

Wang et al.

642 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

Yeast ADA2 was fused to glutathione-S-transferase

(GST) followed by cotransformation into yeast of expres-

sion plasmids bearing GST–ADA2 and GCN5. Yeast ex-

tracts were incubated with GST beads and the binding of

wild-type Gcn5 or Gcn5 mutants to GST–Ada2 was de-

termined by immunoblot. The Gcn5 mutants were able

to bind to Ada2 and importantly, mutants defective for

growth complementation also interacted with Ada2 (Fig.

3, bottom). Overall, the yeast expression experiments

and Ada2-interaction results indicate that the defective

phenotypes of the substitution mutants were not due to

global structural defects.

Transcriptional activation in the presence of Gcn5

mutants in the gcn5

−

strain

The ability of Gcn5 substitution mutants to support

transcriptional activation in the gcn5

−

strain was tested.

Full transcriptional potency by certain transcriptional

activators requires Gcn5 (Georgakopoulos and Thireos

1992; Marcus et al. 1994). Two chimeric activators were

tested. The first was the yeast Gal4 DNA-binding do-

main fused to the potent activation domain of herpes

simplex virus activator VP16 (Gal4–VP16), and the sec-

ond was the bacterial LexA DNA binding domain fused

to the activation domain from the yeast activator Gcn4

(LexA–Gcn4). The expression reporters were bacterial

LacZ driven by yeast Gal4 DNA-binding sites or bacte-

rial LexA DNA-binding sites, and transcriptional po-

tency was determined by quantitative b-galactosidase as-

says.

Both Gal4–VP16 and LexA–Gcn4 transcriptional activ-

ity were reduced in the GCN5 deletion strain to ∼9% of

wild type (Fig. 4). The Gcn5 substitution mutants di-

vided into three classes. Most Gcn5 mutants were ca-

pable of potentiating Gal4–VP16 or LexA–Gcn4 activity

to approximately the level observed in the GCN5

strain

(Fig. 4). Several mutants that displayed partial comple-

mentation in the growth assay also showed an interme-

diate level of activity in potentiating transcriptional ac-

tivation by both activators (Fig. 4). The six mutants

(KQL, PKM, FAE, KDY196, AIGY, and FKK) that were

unable to complement the growth defect of gcn5

−

, also

failed to complement gcn5

−

function in the transcription

assay for either activator (Fig. 4).

Therefore, the majority of the Gcn5 substitution mu-

tants had similar phenotypes exhibited by wild-type

Gcn5 in growth and transcription assays (hereafter re-

Table 1. Sequences of mutagenic oligonucleotides

Name

Amino acid

Sequence of oligonucleotides

LKN 120 GATGGTCCTAACTGGAGCTGCAGCCATTTTTCAAAAGCAATTACC

IFQ 123 CTAACTGGATTAAAAAACGCTGCAGCAAAGCAATTACCAAAAATG

KQL 126 GGATTAAAAAACATTTTTCAAGCTGCAGCACCAAAAATGCCC

PKM 129 GGATTAAAAAACATTTTTCAAAAGCAATTAGCTGCAGCGCCCAAAGAATACATTGCC

PKE 132 CAATTACCAAAAATGGCTGCAGCATACATTGCCAGGTTAG

RLV 135 ATGCCCAAAGAAGCTGCAGCCAGGTTAGTCTATG

YIA 138 CAAAGAATACATTGCCGCTGCAGCCTATGATCGAAGTCATC

YDR 141 GCCAGGTTAGTCGCTGCAGCAAGTCATCTTTCCATG

SHL 144 GTTAGTCTATGATCGAGCTGCAGCTTCCATGGCTGTCATTAGG

GGI 159 CGATAAGAGAGAAGCCGCAGCAATTGTTTTCTGTGCC

FAE 171 CCATTGACTGTCGTAGCTGCAGCAACATATCGACCTTTCG

IVF 174 GAATTCGCAGAAGCTGCAGCCTGTGCCATCAGTTCG

CAI 177 CGCAGAAATTGTTTTCGCTGCAGCCAGTTCGACGGAACAG

EQV 183 GCCATCAGTTCGACGGCTGCAGCACGCGGTTATGGTGCG

RGY 186 GGAACAGGTAGCTGCAGCTGGTGCGCATC

GAHL 189 GGTACGCGGTTATGCTGCAGCTGCAATGAATCACTTAAAAGAC

MNHL 193 GGTTATGGTGCGCATCTAGCGGCTGCAGCAAAAGACTATGTTAG

KDY196 196 GAATCACTTAGCTGCAGCTGTTAGAAATACC

NIK 205 GTTAGAAATACCTCGGCTGCAGCATATTTTTTGACATATGC

YFL 208 CCTCGAACATAAAAGCTGCAGCGACATATGCAGATAATTAC

TYA 211 CGAACATAAAATATTTTTTGGCTGCAGCAGATAATTACGC

DNY 214 TATTTTTTGACATATGCAGCTGCAGCCGCTATTGGATACTTT

AIGY 217 GCAGATAATTACGCTGCTGCAGCCTTTAAAAAGCAAGGC

FKK 221 TACGCTATTGGATACGCTGCAGCGCAAGGCTTCACTAA

QGF 224 CGCTATTGGATACTTTAAAAAGGCTGCAGCCACTAAAGAAATCACGTTGG

TKEI 227 AAAAAGCAAGGCTTCGCTGCAGCAGCCACGTTGGATAAAAG

GYI 239 GTATATGGATGGCTGCAGCTAAAGATTATGAAGG

KDY242 242 GGATGGGATATATTGCTGCAGCTGAAGGTGGT

EGG 245 GGATATATTAAAGATTATGCTGCAGCTACGCTGATGCAATG

TLM 248 GATTATGAAGGTGGTGCTGCAGCGCAATGTTCTATGTTACC

PRI 256 GCAATGTTCTATGTTAGCTGCAGCACGATATTTGGACGC

The name of each mutant represents the amino acids mutagenized.

The position of the first amino acid mutagenized in each mutant is indicated.

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 643

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

ferred to as wild-type mutants; light gray bars in Fig. 4).

The assays identified six Gcn5 mutants (KQL, PKM,

FAE, KDY196, AIGY, and FKK) that profoundly reduced

Gcn5 function in vivo (‘‘defective’’ mutants; black bars

in Fig. 4). In addition, several mutants (YIA, RGY, DNY,

KDY242, and TLM) displayed various degrees of partial

function and are likely to represent important, but not

crucial, residues for Gcn5 activity (‘‘intermediate’’ mu-

tants; dark gray bars in Fig. 4).

Dominant-negative effect of Gcn5 mutants

Mutations in catalytic site residues may cause mutant

proteins to exert dominant-negative effects when over-

expressed in vivo. Dominant-negative phenotypes are

likely caused by the mutant enzyme competing with the

wild-type counterpart for association with other proteins

or assembly into protein complexes. To determine

whether the phenotypically defective Gcn5 mutants ex-

ert a dominant defective growth effect in vivo, each was

overexpressed in yeast containing normal levels of wild-

type Gcn5.

Overexpression of wild-type Gcn5 itself caused yeast

to grow slowly compared with transformed vector (Fig.

5), perhaps attributable to sequestration of associated

proteins (Gill and Ptashne 1988). Overexpression of the

wild-type Gcn5 mutants showed similar phenotypes as

wild-type Gcn5 (data not shown) and were therefore

judged not to be dominant-negative. Overexpression

of most defective Gcn5 mutants caused even stronger

growth inhibition relative to wild-type Gcn5 (Fig. 5).

Figure 3. Stability and Ada2 interaction of Gcn5 substitution

mutants in yeast. Wild-type GCN5, mutant GCN5, or vector

alone were cotransformed with GST–ADA2, and proteins were

induced using the GAL1-10 promoter. (Top) Gcn5 protein levels

in the yeast extracts were determined by Western blot analysis

using Gcn5 antisera. (Bottom left) The yeast extracts were in-

cubated with GST beads to allow GST–Ada2 to bind. Wild-type

or mutant Gcn5 bound to GST–Ada2 was determined by West-

ern analysis using Gcn5 antiserum. (Bottom right) Co-expres-

sion of wild-type Gcn5 with GST alone was used as a negative

control.

Figure 2. Growth phenotypes of GCN5 substitution

mutants in the gcn5

−

strain. Wild-type GCN5, each of

the GCN5 mutants, or vector alone were introduced

into gcn5

−

strain. (A) Colony growth assay. Transfor-

mants were streaked onto synthetic minimal medium.

Colony growth was assessed after 2 days of growth at

30°C. (B) Doubling time of the Gcn5 substitution mu-

tants. Aliquots of cultures in liquid synthetic minimal

medium were taken every 2 hr and the A

600

of the cul-

tures was measured. The generation time was calculated

as the OD doubling time during exponential growth.

Wang et al.

644 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

This was true both of mutants that were completely

(KQL, PKM, FAE, AIGY, and FKK) or partially (YIA,

RGY, and KDY242) defective (Fig. 5). Two defective mu-

tants (completely defective KDY196 and partially defec-

tive TLM) exhibited a similar phenotype as wild-type

Gcn5 (Fig. 5) and therefore were not dominant-negative.

The mutants are likely to exert their dominant-negative

effects by incorporation into normal protein complexes

(Grant et al. 1997) because first, they are vastly overex-

pressed relative to wild-type Gcn5, and second, they in-

teract normally with Ada2 (Fig. 3) and within native

complexes (Fig. 6B,C) in the gcn5

−

strain where wild-

type Gcn5 is not present.

Therefore, two lines of evidence indicate that the

Figure 5. Dominant-negative effects of Gcn5 HAT mutants. Wild-type GCN5, the GCN5 mutants, or vector alone were transformed

into yeast strain bearing wild-type GCN5. The transformants were restreaked onto galactose plates to induce expression of trans-

formed Gcn5 proteins and grown at 25°C for 6 days.

Figure 4. Transcriptional activation in

the presence of Gcn5 substitution mu-

tants. GAL4–VP16 (left panel) and LexA–

GCN4 (right panel) were cotransformed

with wild-type GCN5, each of the GCN5

mutants or vector alone, along with the ap-

propriate LacZ reporters. b-Gal activities

of the mutants are shown as a percentage

of wild-type Gcn5, which was set at 100%.

(Solid bars) Mutants that were defective in

the growth and transcription assays; (dark

gray bars) mutants having intermediate

phenotypes; (light gray bars) mutants hav-

ing activity similar to wild-type Gcn5. Er-

ror bars represent the standard error about

the mean from three independent experi-

ments.

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 645

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

Gcn5 mutants displaying defects in vivo are functional.

First, all of the defective mutants possess structural in-

tegrity as manifested by their ability to interact with

Ada2 (Fig. 4), and second, the majority of the defective

mutants exhibit dominant negativity in vivo (Fig. 5).

Histone acetylation by Gcn5 substitution mutants

within Ada and SAGA complexes

Previous studies have tested HAT activity of recombi-

nant proteins. The HATs and HDACs are active within

high-molecular-weight complexes, however, which in-

fluences their specificity (Rundlett and Grunstein 1996;

Grant et al. 1997). To directly assay whether the defec-

tive phenotypes of the substitution mutants were caused

by loss of ability to acetylate histones, HAT activity of

the Gcn5 substitution mutants was tested in the context

of high-molecular-weight native HAT complexes (Grant

et al. 1997). Two high-molecular-weight native yeast

complexes have been identified. Both contain Gcn5 as

the HAT catalytic subunit, and specifically acetylate his-

tone H3 and, to a lesser extent, histone H2B. The smaller

complex (800 kD) was named Ada and the larger one (2

MD) SAGA, based on the identity of their constituent

proteins (Grant et al. 1997). We partially purified Ada

and SAGA complexes from yeast expressing the Gcn5

substitution mutants, to determine whether the muta-

tions affected the HAT activity within these physiologi-

cal HAT complexes.

Wild-type Gcn5 or the substitution mutants were

transformed into the gcn5

−

strain using vectors to ex-

press normal levels of protein. Whole-cell extracts were

prepared from 26 out of the 31 mutants created and were

fractionated through Ni

agarose and MonoQ ion ex-

change resins (Grant et al. 1997). HAT activities were

determined on free histone substrates for 22 out of these

26 extracts, which were determined by Western analysis

to have nearly normal Gcn5/Ada2 protein levels in both

the Ada and SAGA complexes (see below). These in-

cluded 11 mutants that were partially or completely de-

fective in growth (Fig. 2) and transcription (Fig. 4), as well

as 11 wild-type mutants.

The HAT activity of the peak SAGA fraction from

wild-type or mutant Gcn5 was then determined by liq-

uid HAT assay quantitation of [

H]acetyl group incorpo-

ration into histones using scintillation counting and vi-

sualization of histones by PAGE and fluorography.

SAGA fractions derived from strains bearing wild-type

mutant Gcn5 were able to acetylate histones, as shown

by the examples of LKN, PKE, YDR, GAHL, and GYI

(Fig. 6A). Four of these had levels as high as, or nearly as

high as, wild-type Gcn5 (LKN, PKE, YDR, and GYI) as

quantitated by scintillation counting (Fig. 6A, top) and

confirmed by visualization of [

H]acetyl incorporation

into histone H3 (Fig. 6A, middle). These mutant com-

plexes displayed Gcn5 and Ada2 protein levels that were

comparable with wild-type Gcn5 (Fig. 6B, top). The re-

maining wild-type mutant (GAHL) possessed >50% of

wild-type activity (Fig. 6A) and was found to be some-

what reduced in protein level compared with wild-type

Gcn5 and the other wild-type mutants (Fig. 6B, top; data

not shown). In addition, visualization of [

H]acetyl in-

corporation into histone H3 catalyzed by SAGA bearing

GAHL indicated clearly higher activity than the defec-

tive Gcn5 mutants (Fig. 6A, middle).

The six severely defective mutants (KQL, PKM, FAE,

KDY196, AIGY, and FKK), as well as partially defective

RGY, displayed greatly reduced ability to acetylate his-

tones, using both the quantitative assay as well as his-

tone visualization (Fig. 6A, top and middle). Five of the

extracts showed background levels of HAT activity

(∼25% of wild-type level). The level of acetylation by

SAGA prepared from the defective FKK mutant was 45%

of wild type in the quantitative assay, but was nearly as

low as background levels by [

H]-histone fluorography.

(As is obvious in Fig. 6, we consistently observe that

fluorography is a more sensitive measure of acetylation

than scintillation quantitation.) The protein levels of

both Ada2 and the defective mutant Gcn5 proteins were

variable, but the levels were similar to those in the strain

expressing wild-type Gcn5 (Fig. 6B, bottom). Therefore,

consistent with the stability assay (Fig. 3), the defects of

these mutants were not attributable to lack of protein

expression, or importantly, inability to be incorporated

into the SAGA complex.

We then determined the HAT activity of the Ada com-

plexes using the identical set of mutant Gcn5 proteins as

used for assay of the SAGA complexes. In the case of the

Ada complex only [

H]histone-H3 fluorography was used

to determine HAT activity, as a strong overlapping

Gcn5-independent histone H4 acetylation activity

(Grant et al. 1997; see Fig. 7) obscured quantitation by

scintillation counting of the Gcn5-dependent histone H3

activity. As observed for the SAGA complex, each of the

wild-type mutants displayed activities comparable with

wild-type Gcn5, whereas HAT activity by the defective

mutants was not detectable (Fig. 6A, bottom). Western

analysis of Ada2 and Gcn5 in the Ada complex (Fig. 6C)

of the wild-type (Fig. 6C, top) and defective (Fig. 6C, bot-

tom) Gcn5 mutants indicated comparable levels of pro-

tein with wild-type Gcn5. Note that the Ada complex

containing the GAHL mutant eluted primarily in frac-

tion 20 from the MonoQ column (Fig. 6C, top left),

whereas the remainder of the wild-type mutants eluted

in a broader peak covering fractions from 20 (Fig. 6C, top

left) to 24 (Fig. 6C, top right).

Overall, the results demonstrate a clear correlation be-

tween complementation of function by the Gcn5 substi-

tution mutants in the gcn5

−

strain (Figs. 2 and 4) and

acetylation of histones (Fig. 6). Importantly, HAT activ-

ity by the mutant Gcn5 proteins was similar in both the

Ada and SAGA complexes, such that mutants that pos-

sessed wild-type levels of activity in the Ada complex

were also competent in the SAGA complex and mutants

that were defective in Ada were also defective in SAGA.

Acetylation of nucleosomal histones by Gcn5

substitution mutants

Both the SAGA and Ada complexes acetylate nucleo-

Wang et al.

646 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

somal histones. Because the level of nucleosomal acety-

lation compared with free histone acetylation is mark-

edly inhibited in the SAGA complex (D. Sterner, P.

Grant, J. Workman, and S.L. Berger, unpubl.), we tested

the ability of the Gcn5 substitution mutants to acetylate

nucleosomal histones in the context of the Ada complex.

In addition, the Gcn5-dependent H3/H2B HAT activity

in the Ada complex overlaps a Gcn5-independent H4/

H2A HAT activity (Grant et al. 1997; Fig. 7), which pro-

vides an internal control for variation in extract prepa-

ration.

Fractions between 18 and 30 were tested for HAT

activity on nucleosomal substrates using a liquid

HAT assay and incorporation of the [

H]acetyl group

into histones was visualized by PAGE and fluorogra-

phy. The peak of Gcn5-dependent H3/H2B activity in

the wild-type extract eluted between fractions 20–

24, whereas the H4/H2A activity peaked at fraction

24–26 (Fig. 7, top left). As observed previously (Grant

et al. 1997), in extracts prepared from the gcn5

−

strain, the H4/H2A HAT activity remained intact,

whereas the H3/H2B activity disappeared (Fig. 7,

left).

Strikingly, five mutants that failed to complement

loss of Gcn5 in vivo (KQL, PKM, FAE, AIGY, and FKK)

and were defective in acetylation of free histones in the

SAGA and Ada complexes also were completely defec-

tive in acetylation of nucleosomal histones in the Ada

complex (Fig 7, right). The remaining defective mutant

(KDY196) had severely reduced activity, which was

barely above background levels present in the gcn5

−

ex-

tract (Fig. 7, right). In addition, one of the partially de-

fective mutants in the in vivo assays (RGY) showed se-

verely impaired activity in the nucleosome acetylation

assay (Fig. 7, right), as also observed in the free histone

assay (Fig. 6A). In contrast, the Gcn5-independent H4/

H2A acetylation activity was not affected by these Gcn5

substitution mutations (Fig. 7, right). As described in the

previous section, the presence of Gcn5 and Ada2 in the

Ada complex was confirmed for all mutants by Western

analysis (Fig 6C).

In contrast, mutants that displayed wild-type pheno-

Figure 6. Free histone HAT activity of Gcn5 substitution mutants in SAGA and Ada complexes. (A) SAGA and Ada complexes

prepared from Gcn5 mutants displaying wild-type activity in functional assays (light gray bars in Fig. 4) and Gcn5-defective mutants

(solid bars in Fig. 4) were tested for HAT activity. SAGA and Ada complexes of each mutants containing peak Gcn5 and Ada2 proteins

detected by Western blots (B,C) were used. Complexes were incubated with free core histones and [

H]acetyl CoA. Reaction mixtures

were then subjected to liquid scintillation accounting (top) as well as SDS-PAGE and fluorography (middle and bottom). The activity

from SAGA complexes (middle) and Ada complexes (bottom) are indicated by arrows. Quantitation of corresponding scintillation

assays for SAGA is shown as a percentage of wild-type Gcn5. Error bars represent the standard error about the mean from three

independent experiments. Western blots of Gcn5 and Ada2 proteins from SAGA (B) and Ada (C) complexes. Peak fractions from SAGA

and Ada complexes were detected for the presence of Gcn5 and Ada2 using immunoblot analysis using Gcn5 and Ada2 antisera.

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 647

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

types in complementation of growth and transcription in

the gcn5

−

strain (LKN, PKE, YDR, GAHL, and GYI), were

competent for acetylation of nucleosomal H3/H2B with

varying efficiencies relative to wild-type Gcn5 (Fig. 7,

left). Significantly, all wild-type Gcn5 mutants were

relatively strong in nucleosomal H3/H2B acetylation

compared with all defective mutants. Therefore, the data

demonstrate that mutations that are most deleterious for

in vivo function of Gcn5 similarly completely impair its

histone acetylation function as a component of native

SAGA and Ada complexes.

Discussion

Several transcriptional cofactors have been revealed re-

cently to possess histone acetyltransferase activity in

vitro. These include members of the Gcn5 family as well

as the unrelated coactivators p300/CBP and TAF

250.

The emergence of this new class of transcription coacti-

vators functioning as HATs has raised important ques-

tions. First, do these proteins function as acetyltransfer-

ases in vivo, second, are nucleosomal histones their

natural substrates, and, finally, is acetyltransferase ac-

tivity required for transcription function? To address

these issues, we systematically created substitution mu-

tants in the putative HAT domain of yeast Gcn5, and

compared their effects on growth and transcriptional ac-

tivation in vivo with histone acetylation in vitro. If

acetylation of histones by Gcn5 is critical, then there

should be strong correlation between HAT activity and

in vivo function.

Correlation between in vivo function of Gcn5

and in vitro histone acetylation

We identified six substitution mutations (KQL, PKM,

FAE, KDY196, AIGY, and FKK) in Gcn5 that were highly

deleterious to Gcn5 function in vivo. Strikingly, five

were inert for acetylation of histones, either as free or

nucleosomal substrates, and the sixth mutant (KDY196)

had very low activity. Importantly, other mutants that

were intermediate in function in vivo (YIA, RGY, DNY,

KDY242, and TLM) were also impaired significantly in

both free histone and nucleosomal histone acetylation

(RGY acetylation is shown in Figs. 6 and 7; others not

shown). Finally, substitution mutants that displayed

wild-type function in vivo, displayed relatively high lev-

els of acetylation in vitro (five examples are shown in

Figs. 6 and 7; complexes from six others were prepared

and were found to possess significant levels of HAT ac-

tivity). Therefore, these data (summarized in Fig. 8) dem-

onstrate a clear correlation between residues within

Gcn5 that are critical for acetylation of histone sub-

strates, and residues that are absolutely required for

growth and transcriptional activity of Gcn5.

It is important to stress that we analyzed histone

acetylation by Gcn5 within the context of native yeast

SAGA and Ada complexes. Gcn5, as well as other HATs

(Chen et al. 1997) and HDACs (Rundlett and Grunstein

1996), exists in protein complexes containing multiple

components that, in the case of Gcn5, alter substrate

specificity in two ways. Recombinant yeast Gcn5 acety-

lates only free histones and primarily histone H3 (his-

tone H4 is acetylated to a far lesser degree) (Brownell et

Figure 7. Nucleosomal HAT activities of

Gcn5 substitution mutants in Ada com-

plexes. The same wild-type mutants and

defective mutants used in the free histone

assays (Fig. 6) were subjected to nucleo-

somal HAT assays using Ada complexes.

Even-numbered Mono Q column fractions

from Ada complex of each mutant were in-

cubated with oligonucleosome cores and

[

H]acetyl CoA. Reactions were loaded

onto SDS-PAGE and the gels were analyzed

by fluorography. (Left) Wild-type Gcn5,

vector alone, and wild-type Gcn5 substitu-

tion mutants; (right) defective Gcn5 mu-

tants. Positions of core histones on the gel

are indicated by arrows. Numbers at the

top are the Mono Q column fractions. The

H3/H2B nucleosomal activity is Gcn5-de-

pendent, and the H4/H2A nucleosomal

HAT activity is Gcn5-independent (Grant

et al. 1997).

Wang et al.

648 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

al. 1996). Gcn5 in native yeast SAGA and Ada complexes

acetylates both free and nucleosomal histones and tar-

gets both H3 and H2B (Grant et al. 1997). The altered

specificity of Gcn5 in complexes, as opposed to recom-

binant form, suggests that physiologically relevant

acetylation activity is best evaluated in the context of

these native HAT complexes, as we have done in the

current study.

Mutations that impair transcription

similarly impair growth

Gcn5 mutants were tested for their effects on transcrip-

tion using two heterologous model activators, Gal4–

VP16 and LexA–Gcn4. We showed previously that Ada2,

which is present in the Ada and SAGA complexes and is

required for Gcn5 function in vivo, binds to both of these

activation domains (Barlev et al. 1995). The Gcn5 sub-

stitution mutants showed very similar phenotypes in the

transcription assays, such that the wild-type, intermedi-

ate, and deleterious mutants displayed parallel effects for

function by each activator. In addition, the natural HIS3

and PHO5 promoters also display dependency on the de-

fective mutants, and have normal function in the pres-

ence of wild-type mutants (R. Bolotserkovskaya, N. Bar-

lev, and S.L. Berger, unpubl.). Therefore, the effects of the

substitution mutations that reduce histone acetylation

have similar effects on both synthetic model promoters

and natural yeast promoters, arguing that the effects are

generally consistent for activators that require Gcn5 for

function.

Moreover, we found a striking correlation between the

effect of the substitution mutations on Gcn5 function in

growth and in transcriptional activation assays, such

that colony size and doubling time of the wild-type, in-

termediate, and deleterious mutants mirrored transcrip-

tion function. The consistency between growth and tran-

scription suggests that results obtained using the model

activators Gal4–VP16 and LexA–Gcn4 reflect natural de-

pendency of yeast genes transcription on Gcn5.

Taken together with previous data regarding function

and biochemical interactions of the Ada, Gcn5, and Spt

proteins (Grant et al. 1997; Roberts and Winston 1997),

the data support the model that Gcn5 functions within

Ada and/or SAGA complexes to acetylate nucleosomal

histones in promoter regions of dependent genes. Ada

and SAGA may be recruited to promoters via interac-

tions of Ada2 or Gcn5 with activation domains (Silver-

man et al. 1994; Barlev et al. 1995; Chiang et al. 1996),

which may position the HAT domain to acetylate

nucleosomes proximal to the TATA box. Spt proteins

may have distinct roles in the SAGA complex, either for

complex integrity (Spt7 and Spt20/Ada5; Grant et al.

1997) or recruitment of TBP and associated proteins

(Spt7 and Spt8; Eisenmann et al. 1992, 1994). Based on

the strong correlation we observe between residues

within Gcn5 required for HAT activity and function, we

conclude that acetylation by Ada and/or SAGA appears

to be a critical in vivo function.

Residues important for function within the Gen5

HAT domain

We identified critical residues for function in all four

Figure 8. Comparison of mutant Gcn5

phenotypes in acetylation vs. transcrip-

tion. Quantitative data are taken from Fig.

6A (top) and Fig. 4 (the Gal4–VP16 and

LexA–Gcn4 data were averaged). Acetyla-

tion of free histones by the SAGA complex

is shown at left and transcriptional activa-

tion by the chimeric activators is shown at

right.

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 649

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

conserved subdomains of Gcn5. Mutations in regions I

(KQL and PKM), II (FAE and RGY), and III (AIGY and

FKK) were most debilitating for histone acetylation

within the Ada and SAGA complexes. Previous evolu-

tionary comparisons have identified conserved se-

quences in the putative HAT domain between the Gcn5

subfamily of acetyltransferases and many additional pro-

teins, leading to speculation that these proteins form an

extensive superfamily (Reifsnyder et al. 1996; Neuwald

and Lansman 1997). In particular, there is very strong

conservation of conserved Gcn5 regions II and III and

members of the superfamily (Neuwald and Lansman

1997). Because the apparent commonality within the su-

perfamily is acetyl-CoA binding, this region may bind

acetyl-CoA. In particular, RGY, AIGY, and FKK show

very strong conservation across the superfamily. There-

fore, our results that these mutations are extremely del-

eterious to histone acetylation are consistent with a role

in acetyl-CoA binding. Whereas AIGY and FKK are com-

pletely debilitating to function of Gcn5 in vivo, it was

surprising that RGY mutant was not affected as dramati-

cally. Because the two glycines in RGYG show the most

strong conservation in this short sequence (Neuwald and

Lansman 1997), it may be that the alanine substitution

preserves some function. The dominant negativity of all

three of these mutants RGY, AIGY, and FKK supports

the interpretation of a defect in acetyl-CoA binding, but

retention of other critical protein interactions.

Interestingly, the FAE mutant, which occurs at the

amino-terminal border of region II, is one of the most

dysfunctional mutants, but is not strongly conserved in

the superfamily. The biochemical function of the FAE

residues will be important to determine. In addition, an-

other region of conservation in the superfamily, located

between regions I and II, contains the conserved GGI

(Neuwald and Lansman 1997). In our study, the GGI mu-

tant was not affected in vivo, but, as is true for the

RGYG mutant, the alanine substitutions may not be

profound sequence changes for these particular residues.

Overall, our data support previous speculation that re-

gions II and III are involved in acetyl-CoA binding.

Regions I and IV of the Gcn5 subfamily display no

significant homology to the other members of the super-

family (Neuwald and Lansman 1997). The KQL and PKM

mutations in region I, however, result in similar pheno-

types as the mutations in the putative acetyl-CoA-bind-

ing site in regions II and III. These phenotypes include

loss of histone acetyltransferase activity, and loss of

growth and transcription function in vivo, as well as

strong dominant negativity. One possibility is that re-

gion I is involved in binding of the particular substrates

acetylated by Gcn5. If this is true, then the lack of ho-

mology in region I within the superfamily would imply

that the other family members acetylate substrates

other than nucleosomal histones H3 and H2B. These al-

ternate substrates could include free histones, histones

H4 or H2A, or completely dissimilar acetylation targets.

The analysis of mutants KDY196 and KDY242 sug-

gests an important role for these residues in processes

other than direct nucleosome acetylation. Both mutants

had extremely debilitated function in vivo, but retained

low acetylation function (see Fig. 7 for KDY196; data not

shown for KDY242). These phenotypes suggest a pos-

sible regulatory function for these residues or interaction

with other proteins that are not involved in acetylation.

In particular, because region IV is not well conserved

with other acetyltransferases (Neuwald and Lansman

1997), and KDY242 was the only profoundly defective

mutant in this region, this is consistent with a function

of region IV specific to regulation of Gcn5.

Overall, the data demonstrate that all four highly con-

served subdomains contain important elements for func-

tion of Gcn5. Regions I, II, and III are likely to be in-

volved in enzymatic activity and substrate specificity,

whereas region IV may have a regulatory role.

Are histones bona fide physiological

substrates of Gcn5?

As noted above, the HAT domains of Gcn5 subfamily

members, although remarkably conserved with respect

to each other, display no obvious sequence conservation

with the putative HAT domains of either CBP/p300 or

TAF

250 (nor are the latter two HAT domains mutually

conserved). These observations raise the possibility that

the known HATs share yet unidentified structural mo-

tifs. A second, and not exclusive explanation is that sig-

nificant differences exist in the substrates of acetylation

by the Gcn5 subfamily compared with the other putative

acetyltransferases. Notably, p300 has been shown re-

cently to acetylate certain lysines in the carboxyl termi-

nus of the tumor suppressor and activator p53 (Gu and

Roeder 1997).

Our data are consistent with the hypothesis that

nucleosomal histones are critical physiological sub-

strates of the Gcn5 family. We scanned the conserved

residues of the entire HAT domain, and observed a

strong consistency between the effect of the most del-

eterious Gcn5 substitution mutants on function in vivo

and their effect on histone acetylation in vitro. More-

over, recent data indicate that normal remodeling of

nucleosomes at the PHO5 promoter during transcrip-

tional activation requires Gcn5, and in particular, the

remodeling mirrors the pattern of dependency on the

HAT domain mutations depicted here (P. Gregory, A.

Schmid, M. Zavari, L. Liu, S.L. Berger, and W. Ho¨rz, un-

publ.). Taken together the data strongly suggest that his-

tones are among the key substrates of Gcn5 in vivo.

Materials and methods

Yeast strains

The yeast strain PSY316 (MATa ade-101

his3-200 leu2-3,112

lys2 ura3-52) and its derivatives PSY316

gcn5 (Candau and

Berger 1996) were used for transformation of HAT domain sub-

stitution mutations in conserved regions II, III, and IV. The

TRP1 disruption of IPY3

gcn5 (Roberts and Winston 1997) was

constructed using the hisgURA3 cassette as described (Alani et

al. 1987), and was used for integration of the HAT domain sub-

stitution mutations in conserved region I.

Wang et al.

650 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

Plasmids

pRS414–GCN5 was made by amplifying GCN5 including its

natural promoter and terminator from genomic DNA using

PCR with primers containing 58-XhoI → 38-BglII sites and then

cloning GCN5 into the low-copy yeast expression vector

pRS414 at XhoI and BamHI sites. GCN5 with BglII ends was

ligated into high-copy yeast galactose-inducible expression vec-

tor pRS424(P

Gal

) to produce pRS424(P

Gal

)-GCN5. GCN5 was

mutagenized in pRSET, pRS414 or pRS424(P

Gal

)backbone and

then subcloned into other vectors by exchanging BamHI–EcoRI

(for subregion I) or NcoI (subregion II–IV) fragments. GCN5 sub-

region I mutants were subcloned into pRS306 for integration.

The yeast-inducible expression vector containing GST was

made by inserting GST fragment (amplified by PCR from pGEX

plasmid with BamHI and KpnI sites) into the yeast galactose

inducible expression vector pYES. GST–ADA2 was made by

inserting ADA2 at BglII in frame with GST.

Site-directed mutagenesis

The Chameleon double-stranded site-directed mutagenesis kit

(Stratagene) was used to construct the alanine-scan mutants of

Gcn5 using mutagenic oligonucleotides shown in Table 1.

Briefly, the GCN5 plasmid was heat-denatured and hybridized

with two oligonucleotide primers. The selection primer

changed one nonessential unique restriction site to a new re-

striction site on the plasmid backbone. The second primer en-

coded the specific mutations, in which three or four amino acids

in Gcn5 were changed into three or four alanine residues, and a

new PstI restriction site was created simultaneously. Unmuta-

genized parental plasmids were selected against by digestion

using the mutagenized restriction site in the vector backbone,

and mutagenized progeny were identified using the introduced

PstI site and were confirmed by sequencing.

Growth and dominant-negative assays

Growth phenotypes of Gcn5 substitution mutations were ana-

lyzed after transformation of IPY37

gcn5

trp1 (conserved re-

gion I mutations) PSY316

gcn5

trp1 (conserved regions II–IV

mutations). Single colonies were streaked on minimal synthetic

medium, and were incubated at 30°C for 2 days. For the dou-

bling time assay, single colonies were inoculated into liquid

fully supplemented synthetic media and rotated overnight at

30°C. The cultures were then diluted to an A

600

of 0.02, and the

OD was checked every 2 hr up to a total of 28 hr.

Dominant-negative assays were carried out by transforming

high-copy expression plasmids pRS425(P

Gal

) or pRS424(P

Gal

)

containing GCN5 mutants into IPY36 or PSY316 strains bearing

wild-type GCN5. Transformants were plated on SD (2% glu-

cose) plates, incubated at 30°C for 3 days. Single colonies were

restreaked on SD (2% galactose) medium, and incubated at

room temperature for 3 days.

Transcription assays

Transcription assays were performed in IPY37

gcn5

trp1 (con-

served region I mutations) and PSY316

gcn5

trp1 (conserved

regions II–IV mutations). Plasmids expressing Gal4–VP16

(Berger et al. 1990) and LexA–Gcn4 (Marcus et al. 1994) and the

corresponding reporters were cotransformed. b-Gal activity was

determined as units per milligram of protein. The LacZ report-

ers used were pLGSD5 (Guarente et al. 1982) for Gal4–VP16

activation, and LexA–8x (eight LexA-binding sites) (Candau et

al. 1996) for LexA–Gcn4 activation.

Yeast protein overexpression and in vivo GST-binding assay

Yeast PSY316

gcn5 strain transformed with GST–ADA2 and

pRS424(P

Gal

)–GCN5 or GCN5 substitution mutants were in-

oculated into 5 ml of synthetic glucose media and grown over-

night. The cultures were diluted in 100 ml of synthetic media

containing 2% lactate, and grown for 5–6 hr. Galactose was

added to 2% final concentration to induce protein expression.

Cells were harvested when cultures reached A

600

= 1. Cell pel-

lets were washed and resuspended in 400 µl of storage buffer (20

M HEPES at pH 8.0, 5 nM EDTA, 20% glycerol, 1 mM b-mer-

captoethanol, 1 µ

M PMSF). Cells were disrupted by vortexing

with glass beads and supernatants were collected. This extract

(200 µl) was incubated with 100 µl of GST bead slurry (50%) for

1 hr in PBS buffer–0.1% Triton-100, and beads were washed

extensively with PBS buffer. The proteins bound to GST beads

were analyzed by SDS-PAGE and immunoblotted using Gcn5

antiserum (Candau and Berger 1996).

Yeast protein complex preparation

Strains transformed with GCN5 substitution mutants were

grown selectively in 200 ml of synthetic media, and then trans-

ferred to 4 liters of YPD media and grown to A

600

= 2. Cells were

harvested and resuspended in 30 ml of extraction buffer (40 m

HEPES at pH 7.3, 350 mM NaCl, 0.1% Tween, 10% glycerol,

protease inhibitors). Cells were lysed by glass beads using a bead

beater, and extracts were clarified by centrifugation for 1 hr at

40,000g. Next, extracts were incubated with 10 ml Ni–NTA–

agarose slurry (50%) at 4°C for 2.5 hr. Resins were poured into

a column and washed subsequently with 10 ml of extraction

buffer, 10 ml of buffer A (20 m

M Imidazole at pH 7.0, 100 mM

NaCl, 0.1% Tween 20, 10% glycerol, protease inhibitors).

Bound proteins were eluted with 10 ml of buffer A with 300 m

imidazole. The 300 mM imidazole eluate was directly loaded

onto a Mono Q HR5/5 column equilibrated in buffer B (50 m

Tris at pH 8.0, 100 mM NaCl, 0.1% Tween 20, 10% glycerol).

After a 10-ml wash with buffer B, bound proteins were eluted

with a 25-ml linear gradient of 100 m

M to 500 mM NaCl in

buffer B. Fractions were collected and subjected to HAT assay.

Western blot analysis

The substitution mutations in the HAT domain of Gcn5 unpre-

dictably altered the elution position of the SAGA complex from

the Mono Q column, and therefore the peak fraction of SAGA

elution for each mutant was determined by Western blot analy-

sis using both Ada2 and Gcn5 antisera (data not shown). Then

final Western blot analyses were performed with this peak frac-

tion from the MonoQ column (Fig. 6B). For the Ada complex,

peak fractions for protein analysis were fractions 20 or 24; frac-

tion 22 could not be used because of a strong nonspecific cross-

reacting protein (data not shown). Western blot analyses were

performed with Gcn5 and Ada2 antisera (Barlev et al. 1995;

Candau and Berger 1996).

HAT assays for free and nucleosomal histones

Different amount of fractions (2 and 6 µl) were used for the

nucleosomal HAT assay and similar histone acetylation pat-

terns (H3/H2B relative to H4/H2A) were observed (data not

shown), indicating that the amount of extracts used in the HAT

assay was in the linear range of HAT enzymatic activity. For

nucleosomal HAT assays, MonoQ fractions from Ada complex

of each Gcn5 mutant were used. Three microliters of each even

numbered fraction (from 18 to 30) were incubated with 1 µg of

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 651

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

oligonucleosome cores and [

H]acetyl CoA in HAT buffer

(Grant et al. 1997) at 30°C for 30 min. The reactions were sub-

jected to SDS-PAGE, and gels were then treated with Intensify

Buffer (DuPont NEN) and fluorographed. For free histone HAT

assays, the same fractions of SAGA and Ada complexes used in

Western analysis were incubated with 1 µg core histones and

[

H]acetyl CoA in HAT buffer. Reactions and fluorograph were

carried out as indicated above.

Acknowledgments

We are grateful to R. Marmorstein for preparation of Figure 1A.

We thank C.D. Allis and M.H. Kuo for communication of un-

published results. We thank J. Workman for oligonucleosomes

and D. Sterner and F. Winston for yeast strains. We thank P.

Lieberman, G. Moore, F. Rauscher, G. Rovera, and J. Workmen

for support and valuable discussions; N. Barlev and G. Moore for

critical reading of the manuscript; and A. Kulak for expert edi-

torial help. The work was supported by grants from the Na-

tional Science Foundation and The Council for Tobacco Re-

search to S.L.B; S.L.B is the recipient of an American Cancer

Society Junior Faculty Research Award.

The publication costs of this article were defrayed in part by

payment of page charges. This article must therefore be hereby

marked ‘‘advertisement’’ in accordance with 18 USC section

1734 solely to indicate this fact.

References

Alani, E., L. Cao, and N. Kleckner. 1987. A method for gene

disruption that allows repeated use of URA3 selection in the

construction of multiply disrupted yeast strains. Genetics

116: 541–545.

Alland, L., R. Muhle, H. Hou Jr., J. Potes, L. Chin, N. Schreiber-

Agus, and R.A. DePinho. 1997. Role for N-CoR and histone

deacetylase in Sin3-mediated transcriptional repression. Na-

ture 387: 49–55.

Bannister, A. and T. Kouzarides. 1996. The CBP co-activator is

a histone acetyltransferase. Nature 384: 641–643.

Barlev, N., R. Candau, L. Wang, P. Darpino, N. Silverman, and

S. Berger. 1995. Characterization of physical interactions of

the putative transcriptional adaptor, ADA2, with acidic ac-

tivation domains and TATA-binding protein. J. Biol. Chem.

270: 19337–19344.

Berger, S.L., W.D. Cress, A. Cress, S.J. Triezenberg, and L.

Guarente. 1990. Selective inhibition of activated but not

basal transcription by the acidic activation domain of VP16:

Evidence for transcriptional adaptors. Cell 61: 1199–1208.

Berger, S.L., B. Pina, N. Silverman, G.A. Marcus, J. Agapite, J.L.

Regier, S.J. Triezenberg, and L. Guarente. 1992. Genetic iso-

lation of ADA2: A potential transcriptional adaptor required

for function of certain acidic activation domains. Cell

70: 251–265.

Bradbury, E.M. 1992. Reversible histone modifications and the

chromosome cell cycle. BioEssays 14: 9–16.

Brownell, J., J. Zhou, T. Ranalli, R. Kobayashi, D. Edmondson, S.

Roth, and C. D. Allis. 1996. Tetrahymena histone acetyl-

transferase A: A transcriptional co-activator linking gene ex-

pression to histone acetylation. Cell 84: 843–851.

Candau, R. and S.L. Berger. 1996. Structural and functional

analysis of yeast putative adaptors: Evidence for an adaptor

complex in vivo. J. Biol. Chem. 271: 5237–5345.

Candau, R., P. Moore, L. Wang, N. Barlev, C. Ying, C. Rosen,

and S. Berger. 1996. Identification of functionally conserved

human homologues of the yeast adaptors ADA2 and GCN5.

Mol. Cell. Biol. 16: 593–602.

Candau, R., J. Zhou, C.D. Allis, and S.L. Berger. 1997. Histone

acetyltransferase activity and interaction with ADA2 are

critical for GCN5 function in vivo. EMBO J. 16: 555–565.

Chen, H., R. Lin, R. Schiltz, D. Chakravarti, A. Nash, L. Nagy,

M. Privalsky, Y. Nakatani, and R. Evans. 1997. Nuclear re-

ceptor coactivator ACTR is a novel histone acetyltransferase

and forms a multimeric activation complex with PCAF and

CBP/p300. Cell 90: 569–580.

Chiang, Y., P. Komarnitsky, D. Chase, and C. Denis. 1996.

ADR1 activation domains contact the histone acetyltrans-

ferase GCN5 and the core transcriptional factor TFIIB. J.

Biol. Chem. 271: 32359–32365.

Cote, J., J. Quinn, J.L. Workman, and C.L. Peterson. 1994.

Stimulation of GAL4 derivative binding to nucleosomal

DNA by the yeast SWI/SNF complex. Science 265: 53–60.

Eisenmann, D., K. Arndt, S. Ricupero, J. Rooney, and F. Wins-

ton. 1992. SPT3 interacts with TFIID to allow normal tran-

scription in Saccharomyces cerevisiae. Genes & Dev.

6: 1319–1331.

Eisenmann, D., C. Chapon, S. Roberts, C. Dollard, and F. Win-

ston. 1994. The Saccharomyces cerevisiae SPT8 gene en-

codes a very acidic protein that is functionally related to

SPT3 and TATA-binding protein. Genetics 137: 647–657.

Fascher, K., J. Schmitz, and W. Horz. 1993. Structural and func-

tional requirements for the chromatin transition at the

PHO5 promoter in Saccharomyces cerevisiae upon PHO5

activation. J. Mol. Biol. 231: 658–667.

Gansheroff, L., C. Dollard, P. Tan, and F. Winston. 1995. The S.

cerevisiase SPT7 gene encodes a very acidic protein impor-

tant for transcription in vivo. Genetics 139: 523–536.

Georgakopoulos, T. and G. Thireos. 1992. Two distinct yeast

transcriptional activators require the function of the GCN5

protein to promote normal levels of transcription. EMBO J.

11: 4145–4152.

Gill, G. and M. Ptashne. 1988. Negative effect of the transcrip-

tional activator GAL4. Nature 334: 4721–4724.

Grant, P.A., L. Duggan, J. Cote, S.M. Roberts, J. Brownell, R.

Candau, R. Ohba, T. Owen-Hughes, C.D. Allis, F. Winston,

S.L. Berger, and J.L. Workman. 1997. yGCN5 function

within multisubunit ADA and SPT/ADA adaptor com-

plexes to acetylate nucleosomal histones. Genes & Dev.

11: 1640–1650.

Grunstein, M. 1990. Nucleosomes: regulators of transcription.

Trends Genet. 6: 395–400.

Gu, W. and R. Roeder. 1997. Activation of p53 sequence-specific

DNA binding by acetylation of the p53 C-terminal domain.

Cell 90: 595–606.

Guarante, L. 1995. Transcriptional coactivators in yeast and

beyond. Trends Biol. Sci. 20: 517–521.

Guarente, L., R.R. Yocum, and P. Gifford. 1982. A GAL10-

CYC1 hybrid yeast promoter identifies the GAL4 regulatory

region as an upstream site. Proc. Natl. Acad. Sci. 79: 7410–

7414.

Hassig, C.A., T.C. Fleischer, A.N. Billin, S.L. Schreiber, and D.E.

Ayer. 1997. Histone deacetylase activity is required for full

transcriptional repression by mSin3A. Cell 89: 341–347.

Heinzel, T., R.M. Lavinsky, T.M. Mullen, M. Soderstrom, C.D.

Laherty, J. Torchia, W.M. Yang, G. Brard, S.D. Ngo, J.R.

Davie, E. Seto, R.N. Eisenman, D.W. Rose, C.K. Glass, and

M.G. Rosenfeld. 1997. A complex containing N-CoR, mSin3

and histone deacetylase mediates transcriptional repression

[see comments]. Nature 387: 43–48.

Kadosh, D. and K. Struhl. 1997. Repression by Ume6 involves

recruitment of a complex containing Sin3 corepressor and

Rpd3 histone deacetylase to target promoters. Cell 89: 365–

371.

Wang et al.

652 GENES & DEVELOPMENT

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

Kingston, R.E., C.A. Bunker, and A.N. Imbalzano. 1996. Repres-

sion and activation by multiprotein complexes that alter

chromatin structure. Genes & Dev. 10: 905–920.

Laherty, C.D., W.M. Yang, J.M. Sun, J.R. Davie, E. Seto, and

R.N. Eisenman. 1997. Histone deacetylases associated with

the mSin3 corepressor mediate mad transcriptional repres-

sion. Cell 89: 349–356.

Lee, D., J. Hayes, D. Pruss, and A. Wolffe. 1993. A positive role

for histone acetylation in transcription factor access to

nucleosomal DNA. Cell 72: 73–84.

Loidl, P. 1994. Histone acetylation: Facts and questions. Chro-

mosoma 103: 441–449.

Marcus, G., N. Silverman, S. Berger, J. Horiuchi, and L.

Guarente. 1994. Functional similarity and physical associa-

tion between GCN5 and ADA2 - putative transcriptional

adaptors. EMBO J. 13: 4807–4815.

Marcus, G., J. Horiuchi, N. Silverman, and L. Guarente. 1996.

ADA5/SPT20 links the ADA and SPT genes, which are in-

volved in yeast transcription. Mol. Cell. Biol. 16: 3197–3205.

Mizzen, C., X.-J. Yang, T. Kokubo, J. Brownell, A. Bannister, T.

Owen-Hughes, J. Workman, L. Wang, S.L. Berger, T. Kouzari-

des, Y. Nakatani, and C.D. Allis. 1996. The TAF250 Subunit

of TFIID Has Histone Acetyltransferase Activity. Cell

87: 1261–1270.

Neuwald, A.F. and D. Lansman. 1997. GCN5-related histone

N-acetyltransferases belong to a diverse superfamily that in-

cludes the yeast SPT10 protein. Trends Biochem. Sci.

22: 154–155.

Ogryzko, V., R. Schlitz, V. Russanova, B. Howard, and Y. Na-

katani. 1996. The Transcriptional Coactivators p300 and

CBP are Histone Acetyltransferases. Cell 87: 953–959.

Owen-Hughes, T. and J.L. Workman. 1994. Experimental analy-

sis of chromatin function in transcription control. Crit. Rev.

Eukary. Gene Exp. 4: 403–441.

Paranjape, S.M., R.T. Kamakaka, and J.T. Kadonaga. 1994. Role

of chromatin structure in the regulation of transcription by

RNA polymerase II. Annu. Rev. Biochem. 63: 265–297.

Pazin, M. and J. Kadonaga. 1997. What’s up and down with

histone deacetylation and transcription? Cell 89: 325–328.

Peterson, C.L., A. Dingwall, and M.P. Scott. 1994. Five SWI/

SNF gene products are components of a large multisubunit

complex required for transcriptional enhancement [see com-

ments]. Proc. Natl. Acad. Sci. 91: 2905–2908.

Pin˜a, B., S. Berger, G.A. Marcus, N. Silverman, J. Agapite, and L.

Guarente. 1993. ADA3: A gene, identified by resistance to

GAL4-VP16, with properties similar to and different from

those of ADA2. Mol. Cell. Biol. 13: 5981–5989.

Reifsnyder, C., J. Lowell, A. Clarke, and L. Pillus. 1996. Yeast

SAS silencing genes and human genes associated with AML

and HIV-1 Tat interactions are homologous with acetyl-

transferases. Nature Genet. 14: 44–49.

Roberts, S. and F. Winston. 1996. SPT20/ADA5 encodes a novel

protein functionally related to the TATA-binding protein

and important for transcription in Saccharomyces cerevi-

siae. Mol. Cell. Biol. 16: 3206–3213.

———. 1997. Essential functional interactions of SAGA, a Sac-

caromyces cerevisiae complex of Spt, Ada and GCN5 pro-

teins, with the Snf/Swi and Srt/mediator complexes. Genet-

ics 147: 451–465.

Rundlett, S.E. and M. Grunstein. 1996. HDA1 and RPD3 are

members of distinct yeast histone deacetylase complexes

that regulate silencing and transcription. Proc. Natl. Acad.

Sci. 93: 14503–14508.

Saleh, A., V. Lang, R. Cook, and C. Brand. 1997. Identification of

native complexes containing the yeast coactivator-repressor

proteins NGG1-ADA3 and ADA2. J. Biol. Chem. 272:

5571–5578.

Silverman, N., J. Agapite, and L. Guarente. 1994. Yeast ADA2

protein binds to the VP16 protein activation domain and

activates transcription. Proc. Natl. Acad. Sci. 91:

11665–11668.

Struhl, K. 1995. Yeast transcriptional regulatory mechanisms.

Annu. Rev. Genet. 29: 651–674.

Svaren, J., J. Schmitz, and W. Horz. 1994. The transactivation

domain of Pho4 is required for nucleosome disruption at the

PHO5 promoter. EMBO J. 13: 4856–4862.

Taunton, J., C.A. Hassig, and S.L. Schreiber. 1996. A mamma-

lian histone deacetylase related to the yeast transcription

regulator Rpd3p. Science 272: 408–411.

Triezenberg, S.J. 1995. Structure and function of transcriptional

activation domains. Curr. Opin. Genet. Dev. 5: 190–196.

Turner, B.M. and L.P. O’Neill. 1995. Histone acetylation in

chromatin and chromosomes. Sem. Cell Biol. 6: 229–236.

Vettese-Dadey, M., P.A. Grant, T.R. Hebbes, C. Crane-Robin-

son, C.D. Allis, and J.L. Workman. 1996. Acetylation of his-

tone H4 plays a primary role in enhancing transcription fac-

tor binding to nucleosomal DNA in vitro. EMBO J. 15: 2508–

2518.

Vidal, M. and R. Gaber. 1991. RPD3 encodes a second factor

required to achieve maximum positive and negative tran-

scriptional states in S. cerevisiae. Mol. Cell. Biol. 11: 6317–

6327.

Wade, P. and A. Wolffe. 1997. Histone acetyltransferases in con-

trol. Curr. Biol. 7: 82–84.

Wang, L., C. Mizzen, C. Ying, R. Candau, N. Barlev, J. Brownell,

C.D. Allis, and S. Berger. 1997. Histone Acetyltransferase

activity is conserved between yeast and human GCN5 and

required for complementation of growth and transcriptional

activation. Mol. Cell. Biol. 17: 519–527.

Wolffe, A. 1992. Chromatin: Structure and function. Academic

Press, London, UK.

Yang, X., V. Ogryzko, J. Nishikawa, B. Howard, and Y. Naka-

tani. 1996. A p300/CBP-associated factor that competes

with the adenoviral E1A oncoprotein. Nature 382: 319–324.

Zawel, L. and D. Reinberg. 1995. Common themes in assembly

and function of eukaryotic transcription complexes. Annu.

Rev. Biochem. 64: 533–561.

Zhang, Y., R. Iratni, H. Erdjument-Bromage, P. Tempst, and D.

Reinberg. 1997. Histone deacetylases and SAP18, a novel

polypeptide, are components of a human Sin3 complex. Cell

89: 357–364.

Gcn5’s HAT activity required for function in vivo

GENES & DEVELOPMENT 653

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from

12:1998, Genes Dev.

Lian Wang, Lin Liu and Shelley L. Berger

vivo

and SAGA complexes, are also required for transcriptional function in

Critical residues for histone acetylation by Gcn5, functioning in Ada

References

http://genesdev.cshlp.org/content/12/5/640.full.html#ref-list-1

This article cites 57 articles, 24 of which can be accessed free at:

License

Service

Email Alerting

click here.right corner of the article or

Receive free email alerts when new articles cite this article - sign up in the box at the top

Cold Spring Harbor Laboratory Press

Cold Spring Harbor Laboratory Press on September 2, 2024 - Published by genesdev.cshlp.orgDownloaded from