FEBS Letters 586 (2012) 3279–3286

journal homepage: www.FEBSLetters .org

The C-terminal domains of ADA2 proteins determine selective incorporationinto GCN5-containing complexes that target histone H3 or H4 for acetylation

Edith E. Vamos a, Imre M. Boros a,b,⇑a Department of Biochemistry and Molecular Biology, University of Szeged, Közép fasor 52, H-6726 Szeged, Hungaryb Institute of Biochemistry, Biological Research Center, Temesvári krt. 62, H-6726 Szeged, Hungary

a r t i c l e i n f o a b s t r a c t

Article history:Received 2 March 2012Revised 28 June 2012Accepted 29 June 2012Available online 13 July 2012

Edited by Ned Mantei

Keywords:Histone acetylationADA2SAGA complexATAC complexGCN5 KAT2

0014-5793/$36.00 � 2012 Federation of European Biohttp://dx.doi.org/10.1016/j.febslet.2012.06.051

⇑ Corresponding author. Address: Közép fasor 52, H+36 62 544651.

E-mail address: borosi@bio.u-szeged.hu (I.M. Boro

ADA2 adaptor proteins are essential subunits of GCN5-containing histone acetyltransferase (HAT)complexes. In metazoa ADA2a is present in the histone H4-specific ATAC, and ADA2b in the histoneH3-specific SAGA complex. Using domain-swapped ADA2 chimeras, we determined that the in vivofunction of Drosophila melanogaster SAGA and ATAC HAT complexes depend on the C-terminalregion of the ADA2 subunit they contain. Our findings demonstrate that the ADA2 C-terminalregions play an important role in the specific incorporation of ADA2 into SAGA- or ATAC-type com-plexes, which in turn determines H3- or H4-specific histone targeting.� 2012 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction

Histone acetyltransferases (HATs) are important chromatin reg-ulators acting frequently as components of large multiproteincomplexes. The specificity of the catalytic subunits of HAT com-plexes is often modified by adaptor proteins, and depends onwhether the substrate is an isolated histone or a component ofnucleosomes [1,2]. In Saccharomyces cerevisiae, GCN5 (general con-trol nonderepressed 5), together with ADA2- and ADA3-type adap-tor proteins forms the acetyltransferase catalytic module of severalHAT complexes. Recent data suggest a similarly organized HATmodule in mammalian HAT complexes as well [3]. In metazoa,however, two distinct ADA2 adaptor proteins, ADA2a and ADA2b,are subunits of two functionally very distinct HAT complexes.

We reported previously that in D. melanogaster dAda2bd842 thirdinstar larvae the level of K9 and K14 acetylated histone H3 is se-verely decreased [4,5]. Independently obtained results corrobo-rated these data, and complex purification studies revealed thatdADA2b is a subunit of the SAGA-type Drosophila complex [6,7].In contrast with the above, dAda2a mutations caused a decreasein the levels of K5 and K12 acetylated histone H4 [5], and dADA2awas identified to be a component of a smaller complex designatedas dATAC [6,8]. Subsequently, detailed analysis of the smaller com-

chemical Societies. Published by E

-6726 Szeged, Hungary. Fax:

s).

plex revealed that in addition to dGCN5, it contains another HATcatalytic subunit, dATAC2 (KAT14), and that it plays a role inH4K16 acetylation in embryos [9]. These findings led us to askwhether the histone H4 acetyltransferase activity of dATAC couldbe attributed to dGCN5 or to another HAT present in the complex.In other words, do dADA2 proteins determine GCN5 histone spec-ificity? The two dADA2 proteins are similar in that they contain ZZand SANT domains in their N terminal regions followed by threeless conserved ADA boxes [6,10]. Despite their similarity, however,dADA2a and dADA2b participate strictly in their cognate com-plexes, and are not able to functionally substitute each other [4].

In an attempt to gain information on the interactions determin-ing ADA2 complex specificity we generated ADA2a/ADA2b hybridsby domain swapping. We tested the function of the hybrid proteinsin vivo, and found that for both dADA2 proteins the C-terminal do-main was important to partially restore function. We discuss herewhether the decrease in H4 acetylation in Ada2a mutants is a director indirect result of lost ATAC function.

2. Materials and methods

2.1. Plasmid constructs

For the generation of plasmid constructs directing theexpression of dADA2a, dADA2b and their chimeras in S2 cellsand in D. melanogaster, coding sequences (FlyBaseIDFBgn0037555,

lsevier B.V. All rights reserved.

3280 E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286

FBgn0263738) were extracted from previously constructed pBTM-ADA2 plasmids [10,11]. To generate recombinant constructs theGateway cloning system and pENTR3C and pAFW vectors wereused (Invitrogen, Carlsbad, USA). Detailed description of cloningsteps and primer sequences are given in the Supplementary data.Chimera constructs were sequenced and tested for protein expres-sion in the Drosophila S2 cell line.

2.2. Drosophila S2 cell transfection and Western blot

Hybrid ADA2 protein expression was verified by transienttransfection of pAFW-dAda2bS2a, pAFW-dAda2bM2a, pAFW-dA-da2bL2a, pAFW-dAda2a2b, pAFW-dAda2a and pAFW-dAda2b plas-mids into Drosophila S2 cells using the Effectene TransfectionReagent (Qiagen, Hilden, Germany). Production of the hybrid pro-teins was detected by Western blots using M2 anti-FLAG antibody(SIGMA-Aldrich, St. Louis, MO, USA) in 1:10.000 dilution. The samemembrane was reprobed with anti-Tubulin monoclonal antibody(SIGMA-Aldrich). To compare histone H3, H3K14ac and H4K12aclevels, total protein extracts were prepared from third instar larvaeof the dGcn5E333st and w1118 genotype and immunoblotted withcommercially available acetylated histone-specific antibodies (Ab-cam, Cambridge, MA, UK; Upstate, Lake Placid, NY, USA) at dilu-tions specified by the manufacturer. Membranes were developedwith anti-rabbit-HRP and anti-mouse-HRP (DACO, Copenhagen,Denmark) secondary antibodies and ECL (Millipore, Wemecula,CA) reagents as recommended by the manufacturer.

2.3. Fly stocks, generation of transgenic lines, genetic crosses andrescue experiments

Drosophila flies were raised and crosses were performed at 25 �Con standard medium. The dAda2ad189, dAda2bd842 and Gcn5E333st al-leles have been described previously [4]. As control the w1118 strainwas used [12]. Mutant alleles were balanced in heterozygotes withTM6c, Tb Sb, chromosomes. The phiX51D line, containing a phiC31integrase source and an attP docking site, was kindly provided byKonrad Basler (University of Zurich, Zurich, Switzerland) [13].The P[act-GAL4] strain was obtained from the Bloomington Dro-sophila Stock Center (Indiana University, Bloomington, USA). Forthe generation of transgenic lines the corresponding DNA frag-ments in the pTWF vector were injected into phiX51D embryos.In rescue experiments transgenes were expressed in dAda2ad189

or dAda2bd842 homozygous background with the help of the [act-Gal4] driver. L3 animals identified as non-Tubby were transferredto new vials, allowed to develop at 25 �C, and scored for pupa for-mation or hatching rate. A detailed description of crosses and res-cue experiments is given in the Supplementary data.

2.4. Immunofluorescence

Salivary gland polytene chromosome spreads of wandering lar-vae were prepared and immunostained as described [14]. Primaryantibodies specific for H3K9ac, H3K14ac, H4K5ac, and H4K12acwere from Abcam or UPSTATE and were used at 1:200 or 1:300dilution. Mouse anti-Pol II antibodies (7G5) were used for stainingcontrol [15]. Secondary antibodies were Alexa-Fluor-488-conju-gated goat anti-mouse IgG and Alexa-Fluor-555-conjugated goatanti-rabbit IgG (Molecular Probes, Eugene, OR, USA) (used at1:500 dilutions).

2.5. Quantitative RT-PCR

For quantitative determination of specified transcripts, totalRNA was isolated from white pupa synchronized for spiracle ever-sion, using the RNeasyPlus Mini Kit (Qiagen, Hilden, Germany).

First-strand cDNA was synthetized using TaqMan Reverse Tran-scription Reagent (Applied Biosystems, Foster City, USA). Quantita-tive Real-Time PCR (Q-RT-PCR) reactions were performed in an ABI7500 RT-PCR system using Maxima� SYBR Green/ROX qPCR MasterMix (Fermentas, Burlington, Canada). mRNA levels were quantifiedby setting the CT values against a calibration curve and normalizedto the expression level of the Rp49 gene. The DDCT was used forthe calculation of the relative abundances [16]. The primer se-quences have been previously described [17,18].

3. Results

3.1. Verification of the role of dADA2 proteins in histone H4 and H3acetylation, and construction of dADA2 hybrids

In an attempt to find an explanation for the complex-specificnature of dADA2 proteins, we first reexamined ADA2 domainstructure comparisons and our previous findings on the changesof acetylated histone levels in dAda2 mutants. In accord with ourearlier data, we consistently found that dAda2a and dAda2b nullthird instar larvae (L3) had decreased histone H4K5ac/K12ac andhistone H3K9ac/K14ac levels, respectively. We also tested the ef-fect of dGcn5E333st mutation on the acetylation of histones at L3stage, and found that both acetylated histone H3 (K9 and K14)and histone H4 levels (K5 and K12) were reduced (Fig. 1A andnot shown). A comparison of the dADA2 protein sequences re-vealed important differences in their C-terminal regions. Similarlyto most metazoan ADA2 proteins, at the C-terminal region of dA-DA2a are four amino acid stretches forming alpha-helixes(Fig. 1C). This structure of yeast ADA2 was defined recently as aSWIRM domain. In contrast with this, there are no structural mo-tifs forming recognizable SWIRM domains in either the short orthe long isoform of dADA2b. In light of the above we constructedplasmids encoding hybrid dADA2 proteins as depicted in Fig. 1C.For these constructs we used the short isoform of dADA2b. We be-lieve this is a bona fide SAGA subunit, as indicated by its co-elutionwith other SAGA subunits during complex purification [10]. Fur-thermore, a transgene expressing the ADA2b short form alone par-tially rescues Ada2b null lethality and restores H3 acetylation(manuscript in preparation). To confirm that improper folding ofthe hybrid proteins did not cause rapid degradation in vivo, theirsynthesis was tested in S2 cells (Fig. 1B). Next we establishedtransgenic Drosophila lines with the hybrid constructs, usingADA2 coding regions without any epitope to avoid possible inter-ference with function and/or unwanted interactions. In order to ex-clude differential expression levels we made chromosomalinsertions into targeted landing sites [13] , and constructed similartransgenic lines expressing full length dADA2a and dADA2b ascontrols.

3.2. Expression of dADA2 chimeric proteins partially rescues dAda2a ordAda2b mutants

We tested the chimeric transgenes for their ability to rescue dA-da2a and dAda2b mutants. Earlier analysis had shown that the dA-da2bd842 null allele is pupal (P)5 lethal and that only a smallpercentage reaches pharate adult (pA) stage [4]. Expression of aUAS-dADA2b transgene under the control of the act-GAL4 driverpartially rescued dAda2bd842 P5 lethality and supported develop-ment to P14 stage in 80% of the animals (Fig. 2B). The phenotypicfeatures characteristic of P5 and P14 stages make the scoring unbi-ased. A full rescue with this transgene is not expected, since it ex-presses the short isoform of dADA2b. As mentioned above, ouranalysis has indicated that this isoform of dADA2b alone is insuffi-cient for complete rescue (manuscript in preparation). In contrast,

Fig. 1. (A) Western blot indicates reduced histone H3 and H4 acetylation level in Gcn5E333st late-third-instar larvae. (B) Transient expression of hybrid ADA2 constructs in theDrosophila S2 cell line. (C) Domain structure of dADA2a, dADA2b proteins, and overview of ADA2a/ADA2b chimeras. Numbers indicate the relative amino acid position at thesites of fusions. (D) Sequence alignment of SWIRM domains. The secondary structure prediction was performed using PRALINE Multiple Sequence Alignment. The positions ofthe first and last amino acids of the sequence segments are indicated. Highlighted residues indicate predicted secondary structure motifs: gray, alpha helix; box, beta strand.Hs: Homo sapiens, Dm: Drosophila melanogaster; Sc: Saccharomyces cerevisiae. NCBI Reference sequences are: NP_001479.3, NP_001014636.1, NP_010736.1, NP_689506.2,NP_001027151.1, NP_649773.1.

E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286 3281

UAS-dADA2a transgene expression resulted in development onlyup to P12 stage with phenotype characteristics similar to thoseof escaper dAda2b mutants. Out of the four dADA2 chimeras testedin the dAda2bd842 background, only dADA2a2b resulted in partialrescue; over 50% of the animals reached pharate adult (P14) stage.A small percentage of the animals completed development evenfurther and emerged as adults, but died within 12 hours, withthe wings still folded (Fig. 2A). However, transgene carriersreached P14 stage with two days delay compared to their hetero-zygous siblings. dADA2b2a chimeric transgenes, which containeddifferent lengths of the dADA2b N-terminal region fused to theC-terminal half of dADA2a failed to show any rescue of the dA-da2bd842 phenotype (Fig. 2B).

The hybrid transgenes were next investigated for their potentialto rescue dAda2ad189 defects. dAda2ad189 is a null allele, and homo-zygote carriers are late L3 lethal. Act-GAL4 forced expression of theUAS-dADA2a transgene results in complete rescue, while the UAS-

dADA2b transgene had no effect on the phenotype (Fig. 3B). HybriddADA2 producer transgenes containing N-terminal dADA2b frag-ments fused to a 311 aa C-terminal segment of dADA2a rescuethe L3 lethality of the animals: among 197 larvae which expresseda short (1–145 aa) dADA2b-N terminal region in front of the dA-DA2a C-terminal region (dADA2bS2a), 176 formed prepupa (P1),38 reached pupal stage (P5), and four animals reached pharateadult stage. Similarly, among 214 dADA2bM2a transgene carrierlarvae 226 formed normal brownish P1 prepupa, 86 reached pupastage (P5) and 4 pA (P12) stage (Fig. 3B). dADA2bS2a and dADA2b-M2a pupae are more similar to heterozygous siblings in shape, col-or and texture than dAda2a mutants (Fig. 3A). While dAda2ad189

mutants spent up to two weeks in L3 stage, animals expressingthe dADA2bS2a or dADA2bM2a transgene reached P5 stage in 8days after egg laying. No UAS-dAda2bL2a or UAS-dAda2a2b larvae(out of 228 and 175 scored, respectively) presented significant res-cue of the dAda2ad189 phenotype (Fig. 3B).

Fig. 2. ADA2a2b chimera transgene partially rescues the dAda2bd842 mutant phenotype. (A) Phenotypic characteristics of the indicated genotypes at specific developmentalstages. (B) Percentages of animals reaching specific stages of development.

3282 E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286

Taken together, these results suggested that the C-terminal re-gion determined the rescue abilities of hybrid dADA2 proteins. TheC-terminal half of dADA2b fused to the N-terminal half of dADA2adirects a partial rescue of dAda2bd842, and vice versa concerning dA-da2ad189 mutation and dADA2b2a fusions.

3.3. Expression of dADA2 hybrid proteins restores histone H3 or H4acetylation in dAda2ad189 or dAda2bd842 mutants

After establishing that dADA2a/dADA2b chimeric transgenes re-sulted in partial phenotypic rescue, we performed polytene chro-mosome staining to determine the effects of the same transgeneson histone acetylation in dAda2ad189 and dAda2bd842 backgrounds.Images of immunostained salivary gland polytene chromosomesof transgene carrier larvae underscored the results of rescue exper-iments. Expression of the dADA2bS2a and dADA2bM2a transgenesin dAda2ad189 mutants recovered the intensity of banding patternfor H4K5ac- and H4K12ac-specific antibodies (Fig. 4A). Westernblots developed with antibodies specific for H4K5ac andH4K12ac corroborated the results of polytene staining (Fig. 4C).Moreover, the distorted banding pattern of polytene chromosomescharacteristic of dAda2ad189 mutants was also substantially im-proved by ectopic expression of chimeric dADA2bS2a or dADA2b-M2a protein. In dAda2bd842 mutant larvae the UAS-dAda2a2btransgene was capable of restoring the H3 acetylation at lysine 9and 14 to some extent (Fig. 4B and C). In contrast, the UAS-dAda2b-S2a, UAS-dAda2bM2a or UAS-dAda2bL2a transgenes had no detect-

able effect on either H3K9 or H3K14 acetylation, in accord withtheir failure in rescue.

Thus, the changes of in vivo acetylation levels of H3 and H4 ly-sines indicate that the C-terminal regions of dADA2a or dADA2bproteins are able to direct hybrid ADA2 proteins into functionallydistinct HAT complexes.

3.4. Chimeric dADA2 proteins did not improve expression of selectedgenes affected by dAda2a or dAda2b mutations

We reported recently that dAda2b and dAda2a mutations differstrikingly in the number of RNAs whose level they alter [17,18]. Inan attempt to determine whether chimeric transgene expressionrestored mutant RNA levels similar to that characteristic of wildtype, we compared selected RNA levels in mutants and hybridtransgene expressers by Q-RT-PCR. Among the genes we found ear-lier to be affected by dAda2b mutation, we choose sugarbabe (sug),cap and collar (cnc), Frost (Fst) and Hus 1-like (Hus-1), which werefound earlier to be either down- or up-regulated in dAda2b mu-tants [18]. From dAda2ad189 -dependent genes we chose somebelonging to the Halloween group (phantom (ptm), spookier (spok),disembodied (dib) and shadow (sad)) which were found earlier tobe underrepresented by RNA in dAda2a mutants [17]. Surprisingly,on comparing RNA levels of dAda2 mutants and hybrid transgenecarrier larvae we did not detect a significant change in mRNA levelcorresponding to any of the above genes in transgene-expressinglarvae relative to mutants (Fig. 5 and not shown).

Fig. 3. ADA2b2a chimera transgenes partially rescue the dAda2ad189 mutant phenotype. (A) Phenotypic characteristics of indicated genotypes at specific developmentalstages. (B) Percentages of animals reaching specific stages of development.

E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286 3283

4. Discussion

The D. melanogaster ATAC and SAGA complexes share GCN5 andADA3 subunits and differ, in addition to several other subunits, incontaining one of two complex-specific ADA2 type-adaptors. Ge-netic analysis revealed that mutations removing dADA2b fromdSAGA result in a decrease in histone H3K9ac and H3K14ac levels[4,19]. Mutations affecting dADA2a, a dATAC subunit by definition,result in decrease in H4K5ac and H4K12ac levels [4,5,8]. Thus, thepresence of the dADA2a or dADA2b adaptor protein in a complexcorrelates with the histone specificity of that complex. However,in addition to GCN5 (KAT2), dATAC harbors at least one additionalsubunit with HAT activity [9]. Mutations of Atac2 resulted in a dropin H4K16ac level in late embryos. The L2 lethality of Atac2 mutantsprevented study of the effect of this HAT on the acetylation of poly-tene chromosomal histones. Both dSAGA and dATAC seem to be in-volved in global as well as promoter-specific histone modifications.Furthermore, in addition to its HAT activities, SAGA plays a role inchromatin regulation via a deubiquitination module as well[20,21]. All the above combined explain a high level of interest inexploring molecular interactions determining SAGA and ATACfunctions.

We have shown earlier that distinct phenotypes result from theloss of Drosophila Ada2a and Ada2b functions. Our results, com-bined with data from others, established the complex-specific nat-ure of dADA2a and dADA2b. The experiments reported hereindicate that hybrid ADA2 proteins integrate into distinct HAT

complexes depending on their C-terminal region. Based on theirin vivo rescue ability and their corresponding histone acetylatingactivity, these complexes can be assumed to be SAGA and ATAC.Sequence comparisons and functional assays of ADA2 proteinsidentified three regions that are believed to mediate importantfunctions. The ZZ finger was implicated in interaction with GCN5[22]. The yADA2 SANT domain potentiated the catalytic functionof yGCN5 and was important for acetylating nucleosomal histones[1,23]. The region following the SANT domain was shown to inter-act with yADA3 in vitro [22,24,25]. A less conserved region amongADA2 proteins is the SWIRM domain. Studies on mammalianADA2b revealed a conditional role of the SWIRM domain in chro-matin acetylation by STAGA [2]. Our structure comparison revealedno SWIRM domain in dADA2b, while it is present in dADA2a(Fig. 1D). In this respect the two ADA2 factors of D. melanogasterdiffer more than ADA2 proteins of other metazoa.

SAGA complexes are thought to be built from structural andfunctional modules (for a review see: [26]). The HAT module con-sists of GCN5, ADA3, ADA2 and SGF29. Detected physical interac-tions place ADA2 between GCN5 and ADA3. SGF29 makes contactwith ADA3. Human ADA2 interacts with ADA3 through its N-ter-minal region following the SANT domain and its C-terminal SWIRMdomain. Contacts with GCN5 are made through the SANT domain[2]. By yeast two-hybrid analysis we detected physical interactionsof dADA2 both with dADA3 and dGCN5 [5]. It is important to notehere that physical interactions of ADA2b and ADA3 have also beendetected with several non-SAGA subunit transcription regulators,

Fig. 4. Images of polytene chromosomes immunostained with antibodies specific for histone H3K9ac, K14ac, H4K5ac, K12ac and Pol II (Ab 7G5) as indicated on the left.Genotypes are indicated at the top. The images were obtained using identical data-recording settings (A and B). (C) Western blots of protein extracts of L3 larvae. Genotypesare indicated at the top, antibodies used to develop the blots are shown on the left. H4K12ac and H3K14ac blots were stripped and developed with H3-specific antibodies todemonstrate equal loading.

3284 E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286

such as p53, among others. Information on the structural organiza-tion of ATAC is more scarce. Since both SAGA and ATAC containGCN5, ADA3 and SGF29 subunits [9], one would expect that com-bined with a complex-specific ADA2 protein, they would both formsimilar GCN5-containing HAT modules. The detected interactionsof dADA2a with both dGCN5 and dADA3 might give support to thisassumption. This leads to the question of whether the histoneH4K5 and K12 acetyltransferase activity can be attributed toGCN5 or to ATAC2. A third possibility could be that it is neitherof these, but instead other HAT(s) affected by mutations of ATACsubunits that are the underlying cause of H4 acetylation decrease.Reduced H4ac levels in Ada2a mutants might result from knockingout the activity of the GCN5 module. Several observations can beput forward to argue for a direct role of GCN5 in H4K5 and K12modifications. First, ATAC2 was found to be specific for H4K16.Second, Gcn5 mutants and Ada2a mutants display practically iden-tical phenotypes in respect of both developmental failures and re-duced H4K5ac/K12ac levels. Importantly, Gcn5 alleles differingonly in the HAT or the ADA2 interacting region of the protein alsoshow phenotypes identical to those of Ada2a null mutants, and ge-netic interactions between Ada2a and Gcn5 mutants have beendemonstrated in different settings [5]. The effects of Ada2a andGcn5 mutations on the transcriptomes of larvae are also very sim-ilar (manuscript in preparation). Taken together, the possibilitythat dGCN5 acts on H4 K5 and K12 side chains as a catalytic unitof the ATAC complex can hardly be excluded, though neither canit be proven by existing direct evidences. The observed loose spec-ificity of GCN5 and its reduced, but still existent affinity towardsH4 supports this possibility. Analyses of the functional conse-

quences of yeast and human ADA2–ADA3–GCN5 interactions con-cluded that GCN5 activity and significantly, its affinity towardshistones, depended on the context in which the enzyme was posi-tioned and also the context of its substrate [1,2,23,27]. GCN5 canaccept histone H4 side chains as substrate [28] and yGcn5 muta-tions cause reduced H4 acetylation [29]. Therefore, within ATAC,a GCN5-containing HAT module could be in a suitable conforma-tion for H4 K5 and K12 acetylation. Alternatively, the loss of dA-DA2a might disrupt the ATAC complex or distort it in such a waythat ATAC2 became inactivated. However, the different lethalityphase of Ada2a and Atac2 mutants does not support this assump-tion. Atac2 mutants are L2 lethal, while Ada2a lethal phase is lateL3. If indeed the product of Ada2a is required for ATAC2 function,than one would expect identical phenotypes. A counter argumentto this can be based, however, on different penetrance of thematernal gene products. Interplay between units of ATAC couldtarget the complex to chromatin via GCN5 bromodomain interac-tions, while H4 acetylation is performed by ATAC2. In this respectthe difference between the two dADA2 proteins with regard to thepresence of a SWIRM domain, which has chromatin binding affin-ity, might be significant.

H4-specific acetylation is partially restored in dAda2ad189 mu-tants by transgenes expressing the dADA2a C-terminal region.The acetylation extends all over the polytene chromosomes. Simi-larly, the partial rescue of SAGA-specific H3 acetylation by dADA2bC-terminal-containing transgene expression seems to be more of aglobal than a gene-specific modification change. The restored acet-ylation levels did not correlate in either case with restored mRNAlevels of the limited number of genes we tested by Q-PCR. This can

Fig. 5. The mRNA levels of spookier and Frost in dAda2 mutant L3 larvae carrying wild type or hybrid ADA2 transgene as indicated. The changes in the mRNA levels weredetermined by real-time PCR and compared to wild type (w1118). Ada2b⁄ indicates transgene expressing dADA2b short isoform. Error bars represent standard deviationscalculated from three independent experiments.

E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286 3285

be the result of the partial rescue, or it could reflect a less directlink between these histone acetylation and gene expressionchanges. However, we should point out here that the dADA2b shortform, which we used for the construction of hybrid ADA2 proteins,when expressed from a transgene was unable to restore the mRNAlevels of the studied genes. Acetylation of K5 and K12 is linked tohistone deposition into newly synthesized chromatin during Sphase [30]. Among the identified Drosophila HATs (Chameau,Tip60, Enoki, CBP (nejire), ELP3), several have been shown to targetthese side chains of H4 (for review see [31]). Our laboratory re-cently reported microarray data on the transcriptome of dAda2amutants [17]. In mutants we detected down-regulated mRNA lev-els corresponding to HAT, HDAC and histone chaperone genes atlate L3 stage, when a global decrease in K5 and K12 acetylatedH4 on polytene chromosomes is clearly observable (Supplemen-tary Table 1). The decreased mRNA levels are unlikely to resultfrom nonspecific RNA degradation since other genes, among themseveral involved in acetyl-CoA metabolism, are represented by ele-vated mRNA levels as compared to wild type. Moreover, strong cor-relation exists between expression changes in dAda2a and dAda3mutants. Strikingly, GCN5 stands out being represented by an in-creased mRNA level in ATAC subunit mutants (Supplementary Ta-ble 1). Based on the microarray data one can thus hypothesize thata perturbed balance of HATs and HDACs leads to the global H4acetylation changes in ATAC mutants. These indirect effects ofATAC mutations might cover gene specific alterations caused di-rectly either by the GCN5 or ATAC2 catalytic unit of the ATACcomplex.

The results we report here are in accord with previously col-lected genetic data and underline the different histone specificityof the two Drosophila GCN5-containing HAT complexes. The dem-onstration of the important contribution of the ADA2 C-terminalregions to complex formation poses questions and opens avenuesto follow in the exploration of subunit interactions in order to elu-

cidate the roles of these GCN5-containing complexes in chromatinregulation.

Acknowledgements

We are grateful to Lászlo Bodai and Zoltán Villányi for help withDrosophila crosses and discussions. We also thank Ágota Tüzesi,Nóra Zsindely, Adrienn Csebella-Bakota and Katalin Ökrös for theirhelp with experiments. This work was supported by the HungarianState Science Fund (OTKA 77443) to I.M.B.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.febslet.2012.06.051.

References

[1] Sterner, D.E., Wang, X., Bloom, M.H., Simon, G.M. and Berger, S.L. (2002) TheSANT domain of Ada2 is required for normal acetylation of histones by theyeast SAGA complex. J. Biol. Chem. 277, 8178–8186.

[2] Gamper, A.M., Kim, J. and Roeder, R.G. (2009) The STAGA subunit ADA2b is animportant regulator of human GCN5 catalysis. Mol. Cell. Biol. 29, 266–280.

[3] Koutelou, E., Hirsch, C.L. and Dent, S.Y.R. (2010) Multiple faces of the SAGAcomplex. Curr. Opin. Cell Biol. 22, 374–382.

[4] Pankotai, T. et al. (2005) The homologous Drosophila transcriptional adaptorsADA2a and ADA2b are both required for normal development but havedifferent functions. Mol. Cell. Biol. 25, 8215–8227.

[5] Ciurciu, A., Komonyi, O., Pankotai, T. and Boros, I.M. (2006) The Drosophilahistone acetyltransferase Gcn5 and transcriptional adaptor Ada2a are involvedin nucleosomal histone H4 acetylation. Mol. Cell. Biol. 26, 9413–9423.

[6] Kusch, T., Guelman, S., Abmayr, S.M. and Workman, J.L. (2003) Two DrosophilaAda2 homologues function in different multiprotein complexes. Mol. Cell. Biol.23, 3305–3319.

[7] Guelman, S., Suganuma, T., Florens, L., Weake, V., Swanson, S.K., Washburn,M.P., Abmayr, S.M. and Workman, J.L. (2006) The essential gene wda encodes a

3286 E.E. Vamos, I.M. Boros / FEBS Letters 586 (2012) 3279–3286

WD40 repeat subunit of Drosophila SAGA required for histone H3 acetylation.Mol. Cell. Biol. 26, 7178–7189.

[8] Guelman, S. et al. (2006) Host cell factor and an uncharacterized SANT domainprotein are stable components of ATAC, a novel dAda2A/dGcn5-containinghistone acetyltransferase complex in Drosophila. Mol. Cell. Biol. 26, 871–882.

[9] Suganuma, T., Gutierrez, J.L., Li, B., Florens, L., Swanson, S.K., Washburn, M.P.,Abmayr, S.M. and Workman, J.L. (2008) ATAC is a double histoneacetyltransferase complex that stimulates nucleosome sliding. Nat. Struct.Mol. Biol. 15, 364–372.

[10] Muratoglu, S. et al. (2003) Two different Drosophila ADA2 homologues arepresent in distinct GCN5 histone acetyltransferase-containing complexes. Mol.Cell. Biol. 23, 306–321.

[11] Le Douarin, B., Heery, D.M., Gaudon, C., vom Baur, E. and Losson, R. (2001)Yeast two-hybrid screening for proteins that interact with nuclear hormonereceptors. Methods Mol. Biol. 176, 227–248.

[12] Ryder, E. et al. (2004) The DrosDel collection: a set of P-element insertions forgenerating custom chromosomal aberrations in Drosophila melanogaster.Genetics 167, 797–813.

[13] Bischof, J., Maeda, R.K., Hediger, M., Karch, F. and Basler, K. (2007) Anoptimized transgenesis system for Drosophila using germ-line-specific phi C31integrases. Proc. Natl. Acad. Sci. USA 104, 3312–3317.

[14] Pile, L.A. and Wassarman, D.A. (2002) Localizing transcription factors onchromatin by immunofluorescence. Methods 26, 3–9.

[15] Besse, S., Vigneron, M., Pichard, E. and Puvion-Dutilleul, F. (1995) Synthesisand maturation of viral transcripts in herpes simplex virus type 1 infectedHeLa cells: the role of interchromatin granules. Gene Expr. 4, 143–161.

[16] Winer, J., Jung, C.K., Shackel, I. and Williams, P.M. (1999) Development andvalidation of real-time quantitative reverse transcriptase-polymerase chainreaction for monitoring gene expression in cardiac myocytes in vitro. Anal.Biochem. 270, 41–49.

[17] Pankotai, T. et al. (2010) Genes of the ecdysone biosynthesis pathway areregulated by the dATAC histone acetyltransferase complex in Drosophila. Mol.Cell. Biol. 30, 4254–4266.

[18] Zsindely, N., Pankotai, T., Ujfaludi, Z., Lakatos, D., Komonyi, O., Bodai, L., Tora, L.and Boros, I.M. (2009) The loss of histone H3 lysine 9 acetylation due todSAGA-specific dAda2b mutation influences the expression of only a smallsubset of genes. Nucleic Acids Res. 37, 6665–6680.

[19] Qi, D., Larsson, J. and Mannervik, M. (2004) Drosophila Ada2b is required forviability and normal histone H3 acetylation. Mol. Cell. Biol. 24, 8080–8089.

[20] Weake, V.M., Lee, K.K., Guelman, S., Lin, C.H., Seidel, C., Abmayr, S.M. andWorkman, J.L. (2008) SAGA-mediated H2B deubiquitination controls thedevelopment of neuronal connectivity in the Drosophila visual system. EMBO J.27, 394–405.

[21] Zhao, Y. et al. (2008) A TFTC/STAGA module mediates histone H2A and H2Bdeubiquitination, coactivates nuclear receptors, and counteracts heterochro-matin silencing. Mol. Cell 29, 92–101.

[22] Candau, R. and Berger, S.L. (1996) Structural and functional analysis of yeastputative adaptors. Evidence for an adaptor complex in vivo. J. Biol. Chem. 271,5237–5245.

[23] Balasubramanian, R., Pray-Grant, M.G., Selleck, W., Grant, P.A. and Tan, S.(2002) Role of the Ada2 and Ada3 transcriptional coactivators in histoneacetylation. J. Biol. Chem. 277, 7989–7995.

[24] Candau, R., Moore, P.A., Wang, L., Barlev, N., Ying, C.Y., Rosen, C.A. and Berger,S.L. (1996) Identification of human proteins functionally conserved with theyeast putative adaptors ADA2 and GCN5. Mol. Cell. Biol. 16, 593–602.

[25] Berger, S.L., Pina, B., Silverman, N., Marcus, G.A., Agapite, J., Regier, J.L.,Triezenberg, S.J. and Guarente, L. (1992) Genetic isolation of ADA2: a potentialtranscriptional adaptor required for function of certain acidic activationdomains. Cell 70, 251–265.

[26] Samara, N.L. and Wolberger, C. (2011) A new chapter in the transcriptionSAGA. Curr. Opin. Struct. Biol. 21, 767–774.

[27] Grant, P.A., Eberharter, A., John, S., Cook, R.G., Turner, B.M. and Workman, J.L.(1999) Expanded lysine acetylation specificity of Gcn5 in native complexes. J.Biol. Chem. 274, 5895–5900.

[28] Poux, A.N. and Marmorstein, R. (2003) Molecular basis for Gcn5/PCAF histoneacetyltransferase selectivity for histone and nonhistone substrates.Biochemistry 42, 14366–14374.

[29] Zhang, W., Bone, J.R., Edmondson, D.G., Turner, B.M. and Roth, S.Y. (1998)Essential and redundant functions of histone acetylation revealed by mutationof target lysines and loss of the Gcn5p acetyltransferase. EMBO J. 17, 3155–3167.

[30] Sobel, R.E., Cook, R.G., Perry, C.A., Annunziato, A.T. and Allis, C.D. (1995)Conservation of deposition-related acetylation sites in newly synthesizedhistones H3 and H4. Proc. Natl. Acad. Sci. USA 92, 1237–1241.

[31] Allis, C.D. et al. (2007) New nomenclature for chromatin-modifying enzymes.Cell 131, 633–636.