Binding of the histone chaperone ASF1 to the CBPbromodomain promotes histone acetylationChandrima Dasa,1, Siddhartha Roya,b,2, Sarita Namjoshia, Christopher S. Malarkeyb, David N. M. Jonesb,Tatiana G. Kutateladzeb, Mair E. A. Churchillb, and Jessica K. Tylera,3

aDepartment of Biochemistry and Molecular Biology, University of Texas MD Anderson Cancer Center, Houston, TX 77030; and bDepartment ofPharmacology, University of Colorado School of Medicine, Aurora, CO 80010

Edited* by Michael Grunstein, David Geffen School of Medicine at University of California, Los Angeles, CA, and approved February 13, 2014 (received forreview October 10, 2013)

The multifunctional Creb-binding protein (CBP) protein plays a piv-otal role in many critical cellular processes. Here we demonstratethat the bromodomain of CBP binds to histone H3 acetylated onlysine 56 (K56Ac) with higher affinity than to its other monoacety-lated binding partners. We show that autoacetylation of CBP iscritical for the bromodomain–H3 K56Ac interaction, and we proposethat this interaction occurs via autoacetylation-induced conforma-tion changes in CBP. Unexpectedly, the bromodomain promotesacetylation of H3 K56 on free histones. The CBP bromodomain alsointeracts with the histone chaperone anti-silencing function 1 (ASF1)via a nearby but distinct interface. This interaction is necessary forASF1 to promote acetylation of H3 K56 by CBP, indicating that theASF1–bromodomain interaction physically delivers the histones tothe histone acetyl transferase domain of CBP. A CBP bromodomainmutation manifested in Rubinstein–Taybi syndrome has compro-mised binding to both H3 K56Ac and ASF1, suggesting that theseinteractions are important for the normal function of CBP.

Chromatin is the physiological template for all genomic pro-cesses. The histone proteins that package the DNA into

chromatin are subject to posttranslational modifications, in-cluding acetylation, methylation, phosphorylation, ubiquitina-tion, and sumoylation, that serve to regulate DNA-templatedphenomena such as transcription, replication, repair, and re-combination (1). Many histone posttranslational modificationsmediate their function by interacting specifically with and recruiting“reader” modules of multifunctional proteins, which often them-selves have activities that subsequently further modify the chro-matin structure to make the DNA either more or less accessible.For example, the bromodomain is the specific reader module foracetylated lysines on histones and nonhistone proteins (reviewedin ref. 2), where acetylation is one of the most abundant post-translational modifications in human cells.The bromodomain is found in many transcriptional coregulators

and histone-modifying complexes, including histone acetyl trans-ferases (HATs), enzymes that themselves mediate acetylation.Structural studies have revealed that bromodomains have a con-served structural fold that consists of a left-handed four-helixbundle and two interspersed ZA and BC loops which constitutethe active acetyl lysine-binding pocket (3). Despite this conservedoverall structure, different bromodomains recognize distinct acet-ylated lysines in different proteins because the specific amino acidresidues within the loops of each bromodomain are critical fordetermining the acetyl lysine-binding specificity (4, 5).The general theme for bromodomain function is that they

serve to anchor the bromodomain-containing protein to acetylatedchromatin templates or to acetylated transcriptional activators.For example, the bromodomains of the yeast ATP-dependentnucleosome remodeler Swi2 and the HAT GCN5 are requiredfor anchoring these chromatin-modifying complexes to acetylatedchromatin templates in vitro (6). In other cases, the interaction ofbromodomains with non-histone acetylated proteins is important.For example, BRD4, a member of the bromo and extra terminal(BET) subfamily of bromodomain-containing proteins, interactswith the acetylated RelA subunit of NF-κB to promote tran-scriptional activation of inflammatory genes (7). BRD3 has been

shown to interact with the acetylated transcription factor GATA1,recruiting this transcription factor to both active and repressedtarget genes in a histone acetylation-independent fashion (8).Accordingly, there much interest in generating inhibitors to blockthe interaction between bromodomains and their acetylatedbinding partners and thus to reverse changes in gene expressionin human disease states.The Creb-binding protein (CBP) is an important bromodo-

main containing a transcriptional coactivator that functions asa HAT (9). The transactivation function of CBP is mediated byZinc-finger domains CH1 and CH3, whereas the KIX and p160domains serve as platforms to interact with several transcrip-tional activators (10). The general model for the transcriptionalcoactivator function of CBP is one of physical recruitment topromoters and enhancers via interactions with transcriptionalactivators. At the DNA, CBP can acetylate the chromatin-boundhistones in the immediate vicinity, causing the chromatin toadopt a more open and accessible state that facilitates tran-scription. A PHD finger is an integral part of the HAT domain ofCBP and also is important for the acetylation function of CBP (11).Autoacetylation of the closely related protein p300 is crucial for

Significance

The Creb-binding protein (CBP) transcriptional coactivator con-tains a histone acetyl transferase domain and a bromodomain.Bromodomains bind to acetylated lysines, and their function aspreviously understood was limited to mediating recruitment tochromatin via binding to acetylated proteins. Here we show thatthe acetyl lysine-binding activity of the CBP bromodomain hasunexpected roles in CBP-mediated acetylation of nonchromatinbound histones, and we show that the interaction betweena bromodomain and acetyl lysine is stimulated by autoacety-lation. Furthermore, we find that the histone chaperone anti-silencing function 1 binds to the bromodomain of CBP to presentfree histones correctly for efficient acetylation. Through a combi-nation of structural, biochemical, and cell-based analyses, thesestudies enhance our understanding of bromodomain functionand regulation.

Author contributions: C.D. and J.K.T. designed research; C.D., S.R., S.N., and C.S.M. per-formed research; D.N.M.J., T.G.K., M.E.A.C., and J.K.T. analyzed data; and C.D. and J.K.T.wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.

Freely available online through the PNAS open access option.

Data deposition: Atomic coordinates and structure factors reported in this paper havebeen deposited in the Protein Data Bank, www.pdb.org (PDB ID code 4OUF and RCSB IDcode rcsb084961).1Present address: Biophysics and Structural Genomics Division, Saha Institute of NuclearPhysics, Kolkata, West Bengal 700064, India.

2Present address: Department of Structural Biology and Bioinformatics, Indian Institute ofChemical Biology, Kolkata, West Bengal 700032, India.

3To whom correspondence should be addressed. E-mail: jtyler@mdanderson.org.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1319122111/-/DCSupplemental.

E1072–E1081 | PNAS | Published online March 10, 2014 www.pnas.org/cgi/doi/10.1073/pnas.1319122111

its HAT function (12), but the importance of autoacetylationhas not yet been shown for CBP. Adjacent to the HAT domain,CBP also possesses a bromodomain that interacts with acety-lated histones and nonhistone proteins with varying bindingpreferences. There is strong evidence for an important func-tional role for the interaction of the CBP bromodomain withacetylated transcriptional activators [e.g., acetylated p53 (13)].CBP acetylates p53 on lysine 382, which subsequently is boundby the CBP bromodomain, leading to CBP recruitment (via thebromodomain) to promote p53-mediated gene activation thatultimately determines the cellular responses to stress in the formsof senescence, cell-growth arrest, or apoptosis (13–16). Althoughthe isolated bromodomain of CBP can bind to acetylated histo-nes, including H3 K36 and H4 K20 (5), in vitro, the physiologicalrelevance of the CBP bromodomain binding to acetylated histo-nes is not clear. The closest hint at a function for the interactionbetween acetylated histones and the CBP bromodomain was pro-vided by the finding that the CBP bromodomain can increase theability of CBP to acetylate nucleosomal histones in vitro; however,this interpretation is not straightforward, because this functionof the CBP bromodomain also was dependent on the interactionbetween CBP and the Epstein–Barr virus-encoded transcriptionaltransactivator Zta (17).

In addition to its role in acetylating chromatin-bound histones,we previously had shown that CBP also acetylates non–chro-matin-bound histones, specifically on lysine 56 of histone H3 inDrosophila melanogaster and human cells (18). At least in yeast,the histone modification H3 K56Ac (histone H3 acetylated onlysine 56) plays an important role in the delivery of free histonesto the replication-dependent histone chaperones so as to drivechromatin assembly following DNA synthesis (19, 20). Beforebeing delivered to the replication-dependent histone chaperones,the free histones are all sequestered by the ubiquitous histonechaperone anti-silencing function 1 (ASF1). We had shown pre-viously that human ASF1 is essential for the CBP-mediated acet-ylation of free histones on H3 K56 in cells, leading us to proposethat ASF1 may physically present the free histones to CBP foracetylation (18). However, the mechanism whereby ASF1 presentsfree histones to CBP for acetylation was unknown. Once in-corporated into the chromatin, the nucleosomes carrying K56Achave a looser intrinsic structure (21, 22), promoting histone ex-change during transcriptional activation (23–25).In the present study, we show that the CBP bromodomain

interacts with the H3 K56 acetylation mark with a higher affinitythan its previously characterized acetyl lysine-binding partners. Inaddition, we find that the CBP bromodomain interacts with the

H3 K56Ac

A

GS

T (In

put)

H3

K56

H

3 K

56A

c

No

pept

ide

-GST

H3

K36

Ac

H4

K20

Ac

p53

K38

2Ac

GS

T-br

omo

(Inpu

t)

H3

K56

H

3 K

56A

c

No

pept

ide

H3

K36

Ac

H4

K20

Ac

p53

K38

2Ac

Incubated with GSTIncubated with

GST-CBP bromoB

C

E

H3 K36Ac H4 K20Ac p53 K382Ac

GST-p300 bromoGST-CBP bromo

Bound to H3 K56Ac Input

CBP bromoGCN5 bromo

D

-GST

-GST

H3 K56 KD=50 M KD=500 M KD=300 M KD=100 M

M

M M

GS

T (In

put)

GS

T bo

mo

(Inpu

t)

H3

K9A

c H

3 K

14A

c H

3 K

18A

c H

3 K

23A

c H

3 K

27A

c H

3 K

36A

c H

3 K

56A

c

No

pept

ide

H4

K5A

c H

4 K

8Ac

H4

K12

Ac

H4

K16

Ac

H4

K20

Ac

No

pept

ide

+ G

ST

GST bromo

-GST

Input

Fig. 1. The CBP bromodomain binds to H3 K56Ac with higher affinity than to other monoacetylated lysines. (A) In vitro peptide pull-down interaction of theGST-CBP bromodomain with different site-specific acetylated histone peptides. (B) In vitro peptide pull-down interaction of the GST-CBP bromodomain withH3 K56 and H3 K56Ac peptides. (C) Comparison of the binding of the CBP and p300 bromodomains for H3 K56Ac and H3 K9Ac peptides. (D) Comparison ofthe H3 K56Ac-binding preferences of CBP and GCN5 bromodomains. (E) ITC shows that CBP bromodomain binds to the H3 K56Ac peptide but not to theunmodified peptide. H3 K36Ac, H4 K20Ac, and p53 K382Ac were used as positive controls. Kd values are shown.

Das et al. PNAS | Published online March 10, 2014 | E1073

BIOCH

EMISTR

YPN

ASPL

US

ASF1 histone chaperone via a nearby but distinct interface. Fromthe structural perspective, a key disease mutation that we haveidentified in the CBP bromodomain can attenuate the interactionwith both H3 K56Ac and ASF1. Functionally, we show that thebromodomain of CBP is critical for promoting the acetylation offree histones by CBP; this acetylation also is enhanced by ASF1.We find evidence that autoacetylation of CBP is critical notonly for the substrate interaction with the HAT domain but alsofor the bromodomain to bind to its acetyl lysine-binding partners.These results clearly indicate that, although individual domainsof CBP have designated functions, in the full-length context theprotein conformation plays a key role in dictating its interactionsand hence cellular function.

ResultsThe Bromodomain of CBP Shows Preferential Interaction withAcetylated H3 K56. To investigate further the interaction betweenthe CBP bromodomain and acetylated lysines within histones,a candidate-based approach was used. Escherichia coli-expressedrecombinant GST-tagged CBP bromodomain was incubated withimmobilized biotinylated acetylated peptides, and the relative

amount of interacting GST-CBP bromodomain was measuredby Western blotting against GST. In comparison with severalacetylated histone peptides (H3 K9Ac, H3 K14Ac, H3 K18Ac,H3 K23Ac, H3 K27Ac, H3 K36Ac, H4 K5Ac, H4 K8Ac, H4K12Ac, H4 K16Ac, and H4 K20Ac), H3 K56Ac showed thebest binding with the bromodomain of CBP (Fig. 1A) (3).Among the well-studied interaction partners of the CBP bro-modomain are H3 K36Ac, H4 K20Ac, and p53 K382Ac (5, 13).The interaction of the CBP bromodomain with H3 K56Acappeared to be stronger, at least as measured by peptide pull-downassays, than the interaction with these previously characterizedacetyl-binding partners (Fig. 1B). Because the bromodomains ofCBP and p300 are highly similar (differing by only four aminoacids), we asked whether the p300 bromodomain also recognizesH3 K56Ac. The CBP and p300 bromodomains showed efficientinteractions with H3 K56Ac (Fig. 1C), whereas the less closelyrelated human GCN5 bromodomain does not bind to H3 K56Ac(Fig. 1D). The binding affinity of the CBP bromodomain to theH3 K56Ac peptide subsequently was measured by isothermaltitration calorimetry (ITC) and was found to be higher than thatfor the peptides previously identified as interacting with the CBP

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

Y

1175

C

Input

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

Y

1175

C

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

Y

1175

C

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

Y

1175

C

No peptide H3 K56Ac H3 K56

α-GST

A

ED

CB

GF

WT

D11

16R

Q11

18R

F112

6A

K11

70E

Input

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

WT

D11

16R

Q

1118

R

F112

6A

K11

70E

p53 K382Ac H4 K20Ac

α-GST

CB

P br

omo:

CB

P br

omo:

CBP bromo

Fig. 2. Structural analysis of the H3 K56Ac–CBP bromodomain interaction. (A and B) Superimposed 1H,15N HSQC spectra of the CBP bromodomain (0.2 mM)collected during titration with H3K56ac peptide (A) or unmodified H3K56 peptide (B) are color-coded according to the peptide concentration (Insets). (C) Thehistogram displays normalized 1H,15N chemical shift changes observed in the corresponding (A) spectra of the bromodomain. The red line indicates a sig-nificant change (greater than 0.1). The ZA and BC loops are indicated by green arrows. (D and E) The crystal structure of the CBP bromodomain. The protein isshown as a ribbon (D) or a solid surface (E). Residues that exhibit significant H3K56ac-induced perturbations in resonance are colored orange and are labeled.(F) Test of the effect of CBP bromodomain mutations on binding to immobilized H3 K56Ac peptide. The CBP bromodomain was GST tagged. (G) Test of theeffect of CBP bromodomain mutations on binding to immobilized p53 K382Ac and H4 K20Ac.

E1074 | www.pnas.org/cgi/doi/10.1073/pnas.1319122111 Das et al.

bromodomain (H3 K36Ac, H4 K20Ac, and p53 K382Ac (Fig.1E). The binding of H3 K56Ac to the CBP bromodomain is anexothermic reaction, and the enthalpy is greater than in theinteractions of the CBP domain with other known peptides.To investigate whether the stronger interaction of the CBP

bromodomain with H3 K56Ac, as compared with other acetyllysine peptides, could be explained by additional contacts withthe CBP bromodomain, we mapped the binding interface byNMR. H,15N heteronuclear single-quantum coherence (HSQC)spectra of the uniformly 15N-labeled CBP bromodomain wererecorded while the H3 K56ac peptide (residues 49–59 of H3) orthe corresponding unmodified H3 peptide was added graduallyto the NMR samples. Substantial chemical shift changes in thebromodomain were observed during titration with H3 K56ac(Fig. 2A). A lack of resonance perturbations upon the addition ofunmodified H3 peptide suggested that the CBP bromodomaindoes not recognize this peptide and that the acetylated Lys56residue is essential (Fig. 2B). To characterize the H3 K56ac in-teraction structurally, we determined the crystal structure of theCBP bromodomain at a 1.4-Å resolution and defined the H3K56ac-binding site by NMR (Fig. 2C). The overall fold of theCBP bromodomain in the apo state is similar to the previouslydetermined fold of the protein bound to the H4 K20ac [ProteinData Bank (PDB) ID 2RNY] or p53 K382ac (PDB ID 1JSP)peptides (Fig. S1) (5, 13). The structures of the ligand-bound andunbound states superimpose over Cα atoms with rsmds of 1.5 Åand 2.4 Å, respectively, indicating that interaction with the acety-lated peptides causes small conformational changes and that thebinding site is essentially preformed.The H3 K56Ac peptide binds to ZA and BC loops (Fig. 2 D and

E), which are in the general area of the CBP bromodomain thatmediates the interaction with other acetyl lysine-binding partners(5). However, several residues in this area were uniquely perturbedby H3 K56Ac, suggesting distinct contacts within the CBP–H3K56Ac complex (Fig. S2). In the attempt to identify mutations thatspecifically disrupt binding to H3 K56Ac without compromisingother acetyl lysine interactions of the CBP bromodomain, we gen-erated D1116R, Q1118R, F1126A, and K1170Emutants. We foundthat although the residue substitution to an oppositely chargedamino acid (in the case of D1116R) or the loss of the aromatic sidechain (in the case of F1126A) reduces the interaction with H3

K56Ac (Fig. 2F), these mutations also disrupted the binding of p53K382Ac and H4 K20Ac peptides (Fig. 2G). Furthermore, theQ1118R CBP bromodomain mutation, which had a marginal affecton the H3K56Ac interaction, greatly reduced binding to both p53K382Ac and H4 K20Ac (Fig. 2G). These results are consistent withthe idea that the weaker associations can be disrupted more readilythan the interaction with H3 K56Ac (Fig. 1E).

Autoacetylation of CBP Is Required for the Bromodomain Interactionwith Acetylated Lysines. All previous analyses of the interactionsbetween the CBP bromodomain and acetylated lysines had beenperformed only with the isolated CBP bromodomain. Therefore,we examined the ability of the CBP bromodomain to recognizeacetylated lysines in the context of the full-length CBP protein.When we tested the ability of Sf9-expressed full-length CBP tobind to peptides carrying acetylated lysines, we found, to oursurprise, that recombinant full-length CBP failed to bind to anyof the acetylated peptides (Fig. 3A) to which the isolated CBPbromodomain bound (Fig. 1B). In comparison, full-length CBPthat was expressed and purified from HeLa cells was highly ef-fective in binding to both acetylated and to unacetylated histonepeptides where the interaction with the acetylated peptide wasmediated via the bromodomain and the interaction with theunacetylated peptide was mediated via the HAT domain (Fig.3B). To investigate whether something in the HeLa cells ren-dered the full-length CBP capable of binding to acetylated andunacetylated histone peptides, we incubated the Sf9-expressedrecombinant CBP with or without HeLa whole-cell extract.Strikingly, incubating the recombinant CBP with the HeLa cellextract fully restored its ability to bind to acetylated peptides viathe bromodomain and to unacetylated histone substrates via theHAT domain (Fig. 3C). Because p300 is known to be autoacet-ylated, and this autoacetylation activates the HAT activity ofp300 (12), we tested whether acetyl CoA was the component thatwas being supplied by the HeLa cell extract to activate therecombinant CBP so that it was able to bind not only to thesubstrates of the HAT domain but also to the acetyl lysine-binding partners of the bromodomain. Indeed, incubation of therecombinant full-length CBP with acetyl CoA was sufficient torender it able to bind to both the acetylated peptide and theunacetylated peptide (Fig. 3D). We confirmed that these con-

H3

K56

H3

K56

Ac

H3

K36

Ac

H4

K20

Ac

p53

K38

2Ac

Input biotinyated peptides

Recombinant CBP-FL

A

Acetyl coA+ + - - M

-Ac CBP

Anti-pan acetyl western

E

HeL

a w

hole

cel

l ext

ract

:

Input peptides

+

+

+ -

-

-

H3

K56

H3

K56

Ac

H3

K36

Ac

H4

K20

Ac

Rec

ombi

nant

CB

P

Anti-biotin dot blot

CBP-FL

CBP bromo

CBP HAT-ve

C

Anti-biotin dot blot Rec

ombi

nant

CB

P H3

K56

H3

K56

Ac

Input peptides + Acetyl coA

H3

K56

H3

K56

Ac

CBP-FL

CBP bromo

--

D

Pur

ified

from

HeL

a ce

lls

H3

K56

H3

K56

Ac

Input peptides

H3

K56

H3

K56

Ac

Bound to CBP

CBP-FL

CBP bromo

CBP HAT-ve

B

Anti-biotin dot blot Anti-biotin dot blot

Fig. 3. CBP autoacetylation is required for the CBPbromodomain to bind to acetylated lysines. (A) In-teraction of unmodified and acetylated H3 peptideswith immobilized full-length recombinant CBP (CBP-FL). (B) Interaction of unmodified and H3 K56Ac pep-tides with FLAG-purified HeLa cells expressing FLAG-CBP. (C) Interaction of unmodified and H3 K56Acpeptides with immobilized baculoviral-expressed CBPupon incubation with HeLa whole-cell extract. (D)Interaction of unmodified and H3 K56Ac peptideswith immobilized baculoviral-expressed CBP uponpreincubation with acetyl CoA. (E ) Autoacetylationof CBP can be detected by pan-acetyl antibody in thepresence of acetyl CoA.

Das et al. PNAS | Published online March 10, 2014 | E1075

BIOCH

EMISTR

YPN

ASPL

US

ditions were sufficient to lead to autoacetylation of the recombi-nant CBP (Fig. 3E). We interpret these results as indicating thatthe unacetylated CBP has an inaccessible conformation in whichneither the HAT domain nor bromodomain can recognize bindingpartners. We propose that autoacetylation of CBP promotes aconformational change that exposes the HAT and bromodo-mains so that these domains now can recognize their respectivebinding partners.

The Bromodomain of CBP Promotes Acetylation of Free Histones.Next we sought to investigate the function of the interaction be-tween the CBP bromodomain and H3 K56Ac. First, we askedwhether the interaction is important for recruiting CBP to pro-moters that have increased levels of H3 K56Ac during transcrip-tional induction, such as the promoter of the Gene regulated inbreast cancer 1 (GREB1) gene in MCF7 cells upon estradioltreatment (Fig. S3). We compared the ability of transfected FLAG-tagged full-length or bromodomain-deleted CBP to be recruited tothe GREB1 promoter upon estradiol treatment by ChIP analysisand found that the two CBP constructs were recruited to thepromoter with equal efficiency (Fig. S3). Therefore, the interactionbetween the bromodomain and acetylated histones, or any otherprotein for that matter, is not required for its recruitment tochromatin, at least at the GREB1 promoter.Next we asked whether the interaction between H3 K56Ac

and the CBP bromodomain can stimulate acetylation on freehistones, given that CBP acetylates H3 K56Ac on free histones(18). We purified full-length CBP, CBP with the bromodomaindeleted (Δbromo), or a catalytically inactive HATmutant (HAT-ve)expressed from baculoviruses in Sf9 cells (Fig. 4A) (26). Usingequal amounts of these CBP proteins in an in vitro HAT assayon recombinant H3/H4 tetramer and purified core histone sub-strates, we found that, unlike full-length CBP, Δbromo CBP

showed a compromised ability to acetylate H3 K56 in a timecourse-dependent manner (Fig. 4 A and B). These results sug-gested that the interaction between the CBP bromodomain andacetylated lysine promotes histone acetylation on free histones.To verify this result, we used a competitive inhibitor of the in-teraction between the bromodomain and acetylated lysines, calledJQ1 (27). Upon titration of increasing amounts of JQ1 into theHAT assay with a constant amount of full-length CBP, we ob-served that acetylation of H3 K56 was inhibited in vitro ina concentration-dependent manner (Fig. 4C). This result indi-cates that the interaction between the bromodomain and acety-lated lysines plays an important role in stimulating the acetylationof H3 K56Ac on free histones. To investigate whether the in-teraction of the bromodomain with acetylated lysine also was im-portant for stimulating H3 K56 acetylation in vivo, we used theCBP/p300 bromodomain-specific inhibitor CBP-112 from SGC-Oxford and found that that it inhibited acetylation of H3 K56Acbut not of H3 K27Ac (Fig. 4D). Taken together, these results in-dicate that the CBP bromodomain stimulates acetylation of H3K56 in a manner that is dependent on the interaction betweenacetylated lysine and the bromodomain.To distinguish between the possibilities that (i) the bromo-

domain further stimulates the endogenous activity of the HATdomain or (ii) the bromodomain overcomes the inhibitory effectof another domain on the HAT domain, we compared the HATactivity of equimolar amounts of full-length CBP, Δbromo CBP,the isolated HAT domain, and the isolated HAT domain withthe isolated bromodomain supplied in trans. Two interestingresults were obtained from this experiment. First, the isolatedHAT domain had higher HAT activity than full-length CBP(Fig. 4E). This result suggests that another domain of CBPinhibits the activity of the HAT domain, although it is possible thatthe recombinant HAT domain has a higher specific activity merely

Time

-H3 K56Ac

A - FL bromo HAT-veTime

B

-H3 K56Ac

E

Amido black

FLbr

omo

HAT

-ve

HAT

HAT

+ b

rom

o

-

-H3 K56Ac

- JQ1:

C

-H3 K56Ac

-H3

Amido black

Amido black

CBP:

- FL bromo HAT-veCBP:

CBP:

FL brom

oH

AT-v

e

CBP-

HAT

brom

o

CBP: - + + + + + + + -

CBP:

In vitro HAT assay

In vitro HAT assay

In vitro HAT assay In vitro HAT assay

-H3 -H4

-H3 -H4 =H2A/H2B

-H3 -H4 =H2A/H2B

-H3 K27Ac

Unt

reat

ed

DM

SO

1 μM

2 μM

5 μM

-H3 K56Ac

-H3

CBP-112D

Fig. 4. The CBP bromodomain stimulates acetylation of free histones. (A and B) In vitro HAT assays with recombinant H3–H4 (A) and purified core histones(B) with baculovirus-expressed full-length (FL), Δbromo, or HAT-ve CBP. A Coomassie-stained gel of the purified CBP proteins is shown on the left. Acetylationis detected by Western blotting for H3 K56Ac. Amido black stains of the same membranes are shown below. (C) In vitro HAT assay with full-length CBP andincreasing concentrations of the bromodomain interaction inhibitor JQ1 with core histones as substrate. Input H3 was assessed by Western blotting forunmodified H3 on the same membrane. (D) Alteration of H3 K56Ac but not of H3 K27Ac total levels upon treatment with increasing concentrations ofCBP-112. Whole-cell extracts were prepared from HeLa cells treated with a range of CBP-112 concentrations (1, 2, and 5 μM) and were probed for α-H3 K56Acand α-H3 K27Ac. (E) Comparative analysis of in vitro acetylation ability of equimolar amounts of full-length, Δbromo, and HAT-ve CBP, the isolated HATdomain, and a combination of isolated HAT domain and isolated bromodomain. Amido black stains of the same membrane are shown below. A Coomassie-stained gel of the isolated CBP domains is shown on the right.

E1076 | www.pnas.org/cgi/doi/10.1073/pnas.1319122111 Das et al.

because of its mode of expression in E. coli. It is noteworthy that theRING domain of the related p300 protein recently has been shownto inhibit its HAT activity (28). Second, supplying the bromodomainin trans to the HAT domain did not further stimulate the HATactivity of the isolated CBP HAT domain, indicating that the CBPbromodomain does not enhance the endogenous activity of theHAT domain per se. From these results, we propose that thephysical tethering of the bromodomain to the HAT domainsomehow stimulates the catalytic activity of the HAT domain,perhaps through its ability to bind to acetylated proteins.

The Histone Chaperone ASF1 Binds to the Bromodomain of CBP. Wepreviously have shown that acetylation of H3 K56Ac in humansrequires not only CBP but also the histone chaperone ASF1 (18).Furthermore, endogenous Drosophila ASF1 and CBP coimmu-noprecipitate with each other (18). Humans have two isoforms ofASF1, ASF1A and ASF1B, which differ mainly in their C-terminalregion. To investigate whether human ASF1A and/or humanASF1B interact with human CBP, we performed coimmunopre-cipitation analyses from extracts made from HeLa cell lines stablyexpressing ASF1A-FLAG or ASF1B-FLAG. Both ASF1A and

ASF1B coimmunoprecipitated with endogenous CBP (Fig. 5A).ASF1A previously has been shown to bind to the bromodomain ofthe CCG1 protein (29). Therefore, we wondered whether thebromodomain of CBP is important for the interaction with ASF1.To address this question, we immunoprecipitated FLAG-taggedfull-length CBP, bromoΔ CBP, or HAT-ve CBP that was expressedfollowing transient transfection into HeLa cells and determinedhow much ASF1A was coimmunoprecipitated. Compared with thebinding of ASF1A to full-length CBP, the interaction with thebromoΔ CBP was greatly reduced, although the two constructswere expressed at similar levels (Fig. 5B). Given that the bromo-domains of CBP and p300 differ by only four amino acids, it wasnot surprising to find that endogenous ASF1 also coimmunopre-cipitated with p300 from HeLa cells (Fig. 5C). To determinewhether ASF1 interacts directly with the bromodomain, we per-formed in vitro binding assays with ASF1A, ASF1B, or the con-served core of ASF1A (amino acids 1–167) and the bromodomainof CBP. All three versions of ASF1 interacted with the CBP bro-modomain (Fig. 5D), indicating that the interaction is direct and ismediated by the conserved N-terminal region of ASF1. Further-more, the bromodomain of p300 also binds to ASF1A (Fig. 5E). To

-ASF1:

A

-FLAG:

AS

F1A

-FLA

G

-CB

P

ASF1A-FLAG ASF1B-FLAG

ASF1B

AS

F1B

-FLA

G Input

prei

mm

une

Non

e

-CB

P

prei

mm

une

Non

e

ASF1B p ASF1A

ASF1A p :IP-antibody

-CBP -CBP:

-ASF1A X-reaction-

Input FLAG IP

-CBP -ASF1:

-FLAG:

B

D

AS

F1A

ASF1B

AS

F1A

N-te

rm

Input

ASF1A

ASF1 N-term

AS

F1B

AS

F1A

Bound to GST

AS

F1B

AS

F1A

Bound to GST-CBP bromo

AS

F1B

Amidoblack

ASF1B ASF1A

ASF1 N-term

GST GST-CBP bromo

AS

F1A

N-te

rm

AS

F1A

N-te

rm

G

Inpu

t

WT

- D37

R

L140

E

L140

A

Y10

1F

V45

R

E11

6R

T27R

I31R

-GST: -CBP bromo

M D37

R

L140

A

I31R

T27R

E11

6R

V45

R

Y10

1F

L140

E

ASF1A 6his:

-ASF1

F

Inpu

t

Non

e

prei

mm

une

-p30

0

-ASF1A

M GS

T-p3

00 b

rom

o

Inpu

t

GS

T-p3

00 b

rom

o

GS

T-p3

00 b

rom

o

-GST:

IP-antibody C

E ASF1A 6his + -

ASF1A

Chemical shifts in ASF1A residues upon CBP bromo addition

-p300

-6his:

Fig. 5. ASF1 binds to the bromodomain of CBP. (A) Coimmunoprecipitation of human ASF1A and ASF1B with CBP. Whole-cell extracts were made from stablecell lines expressing ASF1A-FLAG or ASF1B-FLAG and were tested for their ability to coimmunoprecipitate with α-CBP antibodies. The letter “p” denotesphosphorylated forms of ASF1. (B ) Transient transfection of FLAG-tagged full-length, Δbromo, and HAT-ve CBP in HeLa cells followed by M2-Agarosepulldown and probing for the presence of ASF1A in the complex by Western blot with an antibody specific to ASF1A. (C) Coimmunoprecipitation of ASF1Awith p300 from HeLa cells. Whole-cell extracts were made from HeLa cells, and ASF1A was coimmunoprecipitated with α-p300 antibodies. (D) The CBPbromodomain binds directly to the N-terminal of ASF1A/B. The GST-CBP bromodomain was bound to ASF1A, ASF1B, or ASF1 (1–167) and analyzed by im-munoblotting. (E) The p300 bromodomain binds directly to ASF1A. The recombinant GST-p300 bromodomain was bound to immobilized ASF1A and wasanalyzed by immunoblotting. (F) ASF1A interacts with the CBP bromodomain by NMR. The magnitude of the chemical shift perturbations for each amino acidof ASF1A, when CBP binds to it, and the location of these amino acids on ASF1 are shown. The SD of the chemical shift was 0.017 ppm. Residues with chemicalshift changes greater than 2 SD are indicated in blue, changes greater than 3 SD in orange, and changes greater than 4 SD in red. (G) (Upper) ASF1A mutationsshowing compromised binding to the wild-type CBP bromodomain. (Lower) The purified ASF1 proteins used for the binding assay.

Das et al. PNAS | Published online March 10, 2014 | E1077

BIOCH

EMISTR

YPN

ASPL

US

determine which surface of ASF1 was contacting the CBP bro-modomain, we performed NMR studies titrating the CBP bromo-domain into H- 15N–labeled ASF1A (1–154). The residues ofASF1A that shifted upon interaction with CBP are indicated in Fig.5F, and they all reside largely on one face of ASF1A. To confirmthe contribution of these ASF1A amino acids to the interactionwith the CBP bromodomain, we generated point mutations inASF1A and showed that the I31R, T27R, E116R, and Y101Fsubstitutions individually obliterated the interaction (Fig. 5G). TheV45R mutation reduced the interaction, whereas mutation of L140to an A or E reproducibly increased the interaction between theASF1 and the CBP bromodomain. It is possible that these side-chains, which are shorter than the original leucine sidechain, eitherbetter fit the surface of the CBP bromodomain or are able to formbetter interactions. Regardless, the mutagenesis of ASF1 confirmsthat the residues identified by NMR indeed are important for theinteraction with the CBP bromodomain. Interestingly, the interfaceof ASF1 that interacts with the CBP bromodomain is distinct fromthe interface that interacts with the CCG1 bromodomain (30).To determine whether the CBP bromodomain interacts with

ASF1 via the same interface with which it interacts with acety-

lated histones, we performed NMR analysis titrations of ASF1A(1–154) with the H-15N–labeled bromodomain. The amino acidsof the CBP bromodomain that showed chemical shift changeswith the addition of ASF1 are shown in Fig. 6 A–C. Mutationalanalysis of the CBP bromodomain confirmed that CBP muta-tions K1180E, W1165A, N1183R, N1162R, and N1162E disruptthe ASF1–CBP bromodomain interaction (Fig. 6D). The mutantCBP bromodomain proteins were expressed and purified toequivalent levels (Fig. S4). Fig. 6E indicates the residues of theCBP bromodomain whose mutation disrupts the interaction withASF1 or with the acetylated K56 peptide: ASF1 and H3 K56Acbind to adjacent regions of the CBP bromodomain, althoughthrough different specific amino acids (Fig. 6E).

A Mutation That Causes Rubinstein–Taybi Syndrome Disrupts theInteraction of the CBP Bromodomain with ASF1A and H3 K56Ac. Uponsuperimposing the H3 K56Ac- and ASF1-binding sites onto theCBP bromodomain structure, we noted that both regions lay adja-cent to the Y1175 amino acid (Fig. 6E) that is mutated to causeRubinstein–Taybi syndrome (RTS) (31). Therefore, we introducedthe Y1175 RTS mutation into the isolated CBP bromodomain and

WT

Y11

75C

Input

Y11

75C

Y11

75C

- + - +

ASF1

-GST

WT

WT

A B

G

WT

K11

80E

S11

79I

W11

65A

N11

63R

GS

T

-GST

WT

K11

80E

S11

79I

W11

65A

N11

63R

Input

E11

83R

N11

62R

N11

62E

GS

T

-GST

E11

83R

N11

62R

N11

62E

GS

T

D

CBP bromo:

Input

Bound to ASF1-6His

Bound to ASF1-6His

H3 K56Ac binding ASF1A binding

Mutation affects:

E F

C

ZA BC

loops

CBP bromo:

CBP bromo: Rubinstein Taybi

CBP bromo

CBP bromo

Chemical shifts in CBP bromo residues upon ASF1A addition

Fig. 6. Mapping the interaction of CBP with ASF1A. (A) HSQC spectra of the 15N-labeled CBP bromodomain in the absence (gray) and presence (red) of 1:1ASF1A. The chemical shifts of CBP F1185 and R1163 are enlarged in the Inset. (B) Map of the chemical shift perturbations of the CBP interaction with ASF1. Thechanges in chemical shift induced in CBP by ASF1A are shown for each residue by number. The SD of the chemical shift was 0.07 parts ppm. Residues withchemical shift changes greater than 2 SD are indicated in blue, changes greater than 3 SD in orange, and changes greater than 4 SD in red. (C) The X-raycrystal structure of CBP mapping the ASF1A-binding site. Amino acids in CBP with chemical shift perturbations greater than 2 SD are shown on the crystalstructure according to the color scheme in B. (D) Analysis of the effect of CBP bromodomain mutations on interaction with immobilized ASF1A. (E) Surfacerepresentation of the CBP bromodomain with ASF1-binding residues that were confirmed by mutagenesis in magenta and H3 K56Ac-binding residues thatwere confirmed by mutagenesis in marine blue. The Y1175 residue mutated in RTS is shown in green. (F) Cartoon representation of CBP bromodomainstructure with Y1175 (present in wild type) marked in green and the corresponding RTS point mutant Y1175C marked in red. (G) Y1175C mutant CBPbromodomain shows compromised ASF1-binding ability.

E1078 | www.pnas.org/cgi/doi/10.1073/pnas.1319122111 Das et al.

determined the effect on the interaction with ASF1A and H3K56Ac. Mutating Y1175C in the CBP bromodomain did not affectthe expression level or stability of the protein (Fig. S4) but didgreatly reduce binding to the H3 K56Ac peptide (Fig. 2F) and toASF1A (Fig. 6G). These results indicate that the inability of CBP tobind to acetylated histone H3 K56 and ASF1 could have directimplications in the pathogenesis of RTS.

The Bromodomain of CBP Is Critical for ASF1-Dependent Stimulationof the CBP HAT Activity.Given that knockdown of ASF1A reducesCBP-mediated acetylation of H3 K56Ac in cells (18), we askedwhether ASF1A also stimulates CBP-mediated acetylation invitro. These assays were performed with limiting amounts ofCBP and in conditions that favor the H3–H4 dimer, i.e., with theH3–H4 tetramer equilibrium tipped towards the H3–H4 dimerstate. Furthermore, the histones were preincubated with ASF1Ato promote formation of the ASF1A:H3–H4 dimer complex. Wefound that ASF1A enhanced acetylation of H3 K56Ac in a dose-dependent manner in vitro (Fig. 7A). However, ASF1A did notstimulate the HAT activity of CBP in general, because this sameincrease in acetylation was not observed for H2B K5Ac (Fig. 7A)or for p53 (Fig. S5).Overexpression of CBP is known to increase levels of acety-

lated histone substrates in cells above that mediated by the en-dogenous CBP (32). We also observed this increase with theacetylation of H3 K56Ac (Fig. 7B). Fractionation of the histonesinto chromatin-bound and non–chromatin-bound indicated thatthis additional H3 K56Ac that occurred upon overexpressionof CBP resided not only on the chromatin-bound but also onthe non–chromatin-bound histones (Fig. 7C), consistent withASF1- and CBP-mediated acetylation occurring on free histonesthat later are incorporated into chromatin. Next, we examinedwhether the CBP bromodomain was required for the CBP-mediated acetylation of histones in cells. We found that over-expression of Δbromo CBP did not stimulate a further increase inH3 K56Ac levels over that mediated by endogenous CBP (Fig. 7B).However, the bromodomain was not required for the additionallevel of CBP-mediated acetylation of H3 K27Ac (Fig. 7B) or H3K18Ac (Fig. S6). As such, this result shows that the CBP bromo-domain is important for the ASF1-stimulated acetylation of H3K56Ac by CBP in cells. To validate this result, we asked whetherASF1 can stimulate CBP-mediated acetylation of H3 K56Ac in the

absence of the CBP bromodomain in vitro. Indeed, we found thatASF1 stimulated acetylation of H3 K56Ac on free histones in vitroby full-length CBP but not by Δbromo CBP (Fig. 7D). This resultdemonstrates that the bromodomain is required for ASF1 tostimulate CBP-mediated acetylation of free histones on H3 K56Ac.Next we asked whether the interaction of ASF1A with the

bromodomain or with histones is required for its ability tostimulate CBP-mediated H3 K56 acetylation on free histones.We found that the ASF1 I31R, T27R, and E116R mutations thatdisrupted the interaction between ASF1 and the CBP bromo-domain (Fig. 5) had a greatly reduced ability to stimulate theHAT activity of CBP (Fig. 7E). This result demonstrates that theinteraction between ASF1 and the CBP bromodomain enhancesthe ability of CBP to acetylate the histones. We also examinedwhether the interaction between ASF1A and histones is requiredfor ASF1 to stimulate the CBP-mediated acetylation. The V94Rmutation of ASF1A destroys its ability to bind to histones (33).This V94R mutant version of ASF1A did not stimulate CBP-mediated acetylation (Fig. 7F), indicating that the ASF1:H3–H4interaction is critical for CBP-mediated acetylation of H3 K56 onfree histones. Taken together, these data suggest that ASF1physically recruits the histones to CBP for acetylation via theinteraction between the CBP bromodomain and ASF1.

DiscussionIn this study, we have uncovered two unique functions for theCBP bromodomain in promoting histone acetylation. First, thebromodomain promotes acetylation of free histones on H3 K56,presumably via the interaction between the acetylated lysine andthe ZA and BC loops of the bromodomain. Furthermore, wehave found that autoacetylation of CBP is required for the in-teraction between the bromodomain and acetylated histones tooccur. Second, the bromodomain promotes the ASF1-stimu-lated acetylation of free histones on H3 K56, by interactingspecifically with ASF1 via the side of the four-helix bundle ofthe bromodomain. We have characterized the interactions of theCBP bromodomain with H3 K56Ac and with ASF1A structurallyand have identified mutations that disrupt these interactions.Finally, we show that a mutation in the bromodomain that causesRTS is defective in both of these interactions, suggesting that theseinteractions may be important for normal CBP function.

0

0.5

1

1.5

2

2.5

1 2 3 4

H3

K56

Ac

/ H3

Amido black

-H3 K56Ac

-H3

-H2B K5Ac

CBP-FL:

H3K56Ac

H3

Tubulin

-

1 2 3 4

-H3 K56Ac

-Tubulin NC C CBP-FL -

NC C

NC = Non-chromatin C = Chromatin

A

B

C

-H3 K56Ac

D

E

H3 K27Ac

H3

Tubulin

0

0.5

1

1.5

1 2 3 4

H3

K27

Ac

/ H3

-

1 2 3 4

WT

- I31R

T27R

E11

6R

-H3

ASF1

-H3 K56Ac

-H3 K56Ac

F

Amido black

ASF1

-H4

+ - + + - - ++

+ ASF1:

In vitro HAT assay

In vitro HAT assay

In vitro HAT assay

In vitro HAT assay

CBP-FL: + - + ++ - - +

+ ASF1:

CBP bromo: - - - - - -

++ - + - + + +

-H3 -H4

Fig. 7. The interaction of ASF1 with histones andthe CBP bromodomain stimulates histone acetyla-tion. (A) HAT assays with either H3–H4 or H2A–H2Bsubstrates in the presence of full-length CBP (CBP-FL)in the presence or absence of ASF1A. (B) (Upper)Transient transfection of full-length, Δbromo, andHAT-ve CBP into HeLa cells, followed by analysis ofthe total H3 K56Ac and H3 K27Ac levels by Westernblotting. (Lower) Quantification of acetylation. (C)Chromatin fractionation to monitor the distributionof H3 K56Ac between chromatin and free histonesafter transfecting CBP. Tubulin was used as the non-chromatin marker. (D) In vitro HAT assay to monitoracetylation mediated by full-length or Δbromo CBPconstructs in the presence or absence of ASF1. (E) HATassay with CBP in the presence of wild-type or mutant(defective in interaction with bromodomain) ASF1. (F)HAT assay with CBP in the presence of increasingconcentrations of wild-type or mutant (defective ininteraction with histones) ASF1.

Das et al. PNAS | Published online March 10, 2014 | E1079

BIOCH

EMISTR

YPN

ASPL

US

While our study was in progress, the Knapp laboratory reportedthat the CBP bromodomain also binds to H3 K56Ac with an affinityhigher than that for its other previously published monoacetylatedhistone substrates (3). In their study, the affinities of the CBPbromodomain for all of the acetylated histone substrates were 5- to10-fold higher than ours. However, it is important to note that in-teraction affinities measured by ITC or any other means are notabsolute but instead are relative within an analysis that is performedunder a fixed set of conditions. The relative affinities of the CBPbromodomain for the various acetylated histone substrates weresimilar in our study (Fig. 1E) and the previous work (3). We wereunable to identify residues of the CBP bromodomain that uniquelymediate the interaction with H3 K56Ac but not other acetylatedbinding partners. This result suggests that the different acetylatedbinding partners bind to the same residues of the CBP bromodo-main, but presumably the histone residues around K56 yield a morecomplementary fit with the surface of the CBP bromodomain thanwith the surface around the other acetylated lysines. As such, we arenot able to disrupt only the interaction of H3 K56Ac with the CBPbromodomain to test the function of this interaction within cells.However, we did uncover an unexpected role for the CBP bro-modomain in promoting acetylation on free histones.A function for any bromodomain in promoting acetylation on

non–chromatin-bound histones or proteins has not been demon-strated previously. Although it is possible that the bromodomain-acetyl K56 interaction may promote acetylation of the other H3histone within theH3–H4 tetramer in vitro, this in unlikely to be thecase in vivo given that most free H3–H4 exists bound to Asf1 as anH3–H4 heterodimer (34–36). Insight into the potential mechanismof the function of the CBP bromodomain in promoting acetylationcan be obtained by looking at the related,muchmore highly studiedprotein p300. Although there is only 65% homology overall be-tween CBP and p300, with each having unique physiological func-tions, theirHATdomains are very highly conserved. The p300HATdomain appears to function via a hit-and-run or Theorell Chancemechanism in which Acetyl CoA binds first, followed by the sub-strate (37). The rate-limiting step for the acetyl transferase activityof p300 has been proposed to be the release of the product (38). Assuch, one could imagine that the bromodomain could increase therate of catalysis by binding to H3 K56Ac, thereby sequestering theproduct away from the active site of the HAT domain. Consistentwith this notion, titration of competitive inhibitors of the acetyllysine-binding pocket of the bromodomain blocked the ability of thebromodomain to stimulate acetylation of H3 K56 by the HAT do-main (Fig. 4C andD). This role for the bromodomain in promotinghistone acetylation appears to be unique for promoting acetylationon free histones, because acetylation of H3 K27Ac on chromatin,which is mainly mediated by CBP and p300 (39), was not reducedby the CBP/p300-specific bromodomain inhibitor in cells (Fig. 4D).Studies of the related p300 HAT domain have demonstrated

that autoacetylation of an autoinhibitory loop is catalytically im-portant for histone acetylation (12, 40). In this work we demonstratethat CBP also becomes autoacetylated. Furthermore, autoacetyla-tion of CBP is required not only to allow the HAT domain to bindto unacetylated substrates but also to allow the bromodomain tobind to its acetylated binding partners. This result suggests thatautoacetylation does more than enable the autoinhibitory loop andthe inhibitory RING domain (28) to move away from the HATactive site. Functionally, the autoacetylation of p300 causes a con-formational change within the p300 HAT domain that is detectedby increased accessibility to proteinase K and promotes transcrip-tional activation (41). The autoacetylation of p300 also occursbeyond the autoacetylation loop, because p300 can be autoace-tylated even when the autoacetylation loop is deleted (42). Wesuggest that, within the nonacetylated CBP, the acetyl-bindingpocket of the bromodomain is blocked by another region ofCBP. The CBP bromodomain does not bind to the HAT domain(28), indicating that the situation is more complicated than theHAT domain simply burying the acetyl-binding pocket of thebromodomain. This inaccessible conformation presumably is mademore accessible upon CBP autoacetylation, enabling the HAT

domain to bind to unacetylated substrates and the bromodomainto bind to acetylated products. It will be interesting to determinewhether the cell uses the regulation of the autoacetylation stateof CBP to control the interaction between the bromodomainsand acetylated proteins.The histone chaperone ASF1 has been shown previously to

bind to bromodomains (29), but the mode of interaction with theCBP bromodomain is quite distinct from that observed with theCCG1 bromodomain (30). We find that the interaction betweenthe CBP bromodomain and ASF1 and the interaction betweenASF1 and histones promoted histone acetylation on free histones.We propose that, by interacting with the CBP bromodomain,histone-bound ASF1 forms a ternary complex in which H3 K56 isoptimally presented to the HAT domain for acetylation by CBP.This model is reminiscent of the manner in which ASF1 promotesacetylation of H3 K56Ac on free histones by the yeast HATRtt109 (20, 23, 43). Our work shows that histone chaperones canalso promote acetylation by a metazoan HAT enzyme, and wehave uncovered the mechanism structurally. Furthermore, ourstudies extend the role of CBP/p300 from being HAT enzymesthat acetylate chromatin-bound histones to also playing an im-portant role in the acetylation of free histones in the cell.CBP mutations are found in many human diseases, including

many types of human tumors (44). Translocations involving CBPcause acute myeloid leukemia and mixed-lineage leukemia (44).RTS also is caused bymutations in a single copy of the CBP gene; itis reported that mutations leading to loss of HAT activity of CBPlead to RTS (45). These mutations also are present in the PHDfinger of CBP that is critical for its acetylation activity (46). TheRTS-causing mutation Y1175C lies immediately adjacent to thesurfaces of the CBP bromodomain that we mapped by mutagen-esis to mediate the interaction with ASF1 and H3 K56Ac. In-terestingly, this CBP mutation implicated in RTS compromisesthe binding of both the acetylated histone and the histonechaperone. Our results suggest that the loss of H3 K56Ac andASF1 binding to CBP may lead to the progression of RTS.

Materials and MethodsSite-Directed Mutagenesis and Protein Production. The point mutants weregenerated using the Phusion SDM kit (New England Biolabs) following thestandard protocol. Details of the expression and purification of therecombinant proteins are given in SI Materials and Methods.

Transient Transfection Assays. Transient transfections of mammalian ex-pression constructs of CBP (full-length, Δbromo, and HAT-ve) were per-formed using Fugene HD following the standard protocol. A transfectiontime of 48 h was necessary for the expression of the transgene. M2-agarosepull down was performed, and then the complex was eluted from the beadswith 0.1 M glycine HCl, pH 3.5. The solution subsequently was reequilibratedto neutral pH, diluted, and taken for further pull-down assays.

Peptide Pull-Down Assays. One microgram of GST-tagged protein was in-cubated with 1 μg of biotinylated histone peptides in binding buffer [50 mMTris·HCl (pH 7.5), 150 mM NaCl, 0.05% Nonidet P-40, 1mM DTT) overnight at4 °C. Each binding reaction was incubated with Streptavidin Sepharosebeads (Amersham) at 4 °C for 1 h. After binding the beads were spun downand washed with binding buffer three times at 4 °C. The beads subsequentlywere boiled with 5× SDS sample buffer and analyzed by Western blotting.Details of the peptides are provided in in SI Materials and Methods.

Immunoprecipitations. FLAG immunoprecipitation was performed as describedelsewhere (47). CBP and p300 immunoprecipitation were performed as de-scribed elsewhere (18).

Structural Analyses.NMR and X-ray crystallography used standard techniques,as described in SI Materials and Methods.

HAT Assays. HAT assays were performed following the standard protocol (48).Details are given in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Jay Bradner for the generous gift ofJQ1; Danny Reinberg for the cell lines expressing FLAG-tagged ASF1A and

E1080 | www.pnas.org/cgi/doi/10.1073/pnas.1319122111 Das et al.

ASF1B; Paul Lieberman for the kind gift of the baculovirus expressionvectors and transient transfection vectors expressing full-length CBP,Δbromo CBP, and HAT-ve CBP; Lee Kraus for the GST-p300 bromodomain;Peter E. Wright for the GST-GCN5 bromodomain constructs; MichelleBarton for purified recombinant p53 protein; Panagis Filippakopoulos forCBP-112; John Ladbury for generous assistance and advice with ITC and forassistance from the Center for Biomolecular Structure at the University ofTexas MD Anderson Cancer Center; Jay C. Nix at Advanced Light SourceBeamline 4.2.2; Catherine Musselman for helping with X-ray crystallo-graphic and NMR data collection; and Jean Scorgie and Luke Smith for

help with preparation of the ASF1 proteins. The use of core facilities wassupported in part by University of Colorado Cancer Center Support GrantP30CA046934. This work was supported by NIH Grant GM64475 (to J.K.T.),a Cancer Prevention and Research Institute of Texas Rising Star and Uni-versity of Texas Texas Stars and Senior Trust Recruitment Awards (to J.K.T.),and National Institutes of Health Grants R01GM079154 (to M.E.A.C.) andRO1GM096863 (to T.G.K.). C.D. was supported by a Susan Komen Racefor the Cure Fellowship and a Ramalingaswami Fellowship from the De-partment of Biotechnology, Ministry of Science and Technology, Govern-ment of India.

1. Gardner KE, Allis CD, Strahl BD (2011) Operating on chromatin, a colorful languagewhere context matters. J Mol Biol 409(1):36–46.

2. Filippakopoulos P, Knapp S (2012) The bromodomain interaction module. FEBS Lett586(17):2692–2704.

3. Filippakopoulos P, et al. (2012) Histone recognition and large-scale structural analysisof the human bromodomain family. Cell 149(1):214–231.

4. Singh M, Popowicz GM, Krajewski M, Holak TA (2007) Structural ramification foracetyl-lysine recognition by the bromodomain of human BRG1 protein, a centralATPase of the SWI/SNF remodeling complex. ChemBioChem 8(11):1308–1316.

5. Zeng L, Zhang Q, Gerona-Navarro G, Moshkina N, Zhou MM (2008) Structural basis ofsite-specific histone recognition by the bromodomains of human coactivators PCAFand CBP/p300. Structure 16(4):643–652.

6. Hassan AH, et al. (2002) Function and selectivity of bromodomains in anchoringchromatin-modifying complexes to promoter nucleosomes. Cell 111(3):369–379.

7. Huang B, Yang XD, Zhou MM, Ozato K, Chen LF (2009) Brd4 coactivates transcrip-tional activation of NF-kappaB via specific binding to acetylated RelA. Mol Cell Biol29(5):1375–1387.

8. Lamonica JM, et al. (2011) Bromodomain protein Brd3 associates with acetylatedGATA1 to promote its chromatin occupancy at erythroid target genes. Proc Natl AcadSci USA 108(22):E159–E168.

9. McManus KJ, Hendzel MJ (2001) CBP, a transcriptional coactivator and acetyl-transferase. Biochem Cell Biol 79(3):253–266.

10. Perissi V, et al. (1999) Factor-specific modulation of CREB-binding protein acetyl-transferase activity. Proc Natl Acad Sci USA 96(7):3652–3657.

11. Kalkhoven E, Teunissen H, Houweling A, Verrijzer CP, Zantema A (2002) The PHD typezinc finger is an integral part of the CBP acetyltransferase domain.Mol Cell Biol 22(7):1961–1970.

12. Thompson PR, et al. (2004) Regulation of the p300 HAT domain via a novel activationloop. Nat Struct Mol Biol 11(4):308–315.

13. Mujtaba S, et al. (2004) Structural mechanism of the bromodomain of the coactivatorCBP in p53 transcriptional activation. Mol Cell 13(2):251–263.

14. Barlev NA, et al. (2001) Acetylation of p53 activates transcription through recruitmentof coactivators/histone acetyltransferases. Mol Cell 8(6):1243–1254.

15. Sakaguchi K, et al. (1998) DNA damage activates p53 through a phosphorylation-acetylation cascade. Genes Dev 12(18):2831–2841.

16. Sachchidanand, et al. (2006) Target structure-based discovery of small molecules thatblock human p53 and CREB binding protein association. Chem Biol 13(1):81–90.

17. Deng Z, et al. (2003) The CBP bromodomain and nucleosome targeting are requiredfor Zta-directed nucleosome acetylation and transcription activation. Mol Cell Biol23(8):2633–2644.

18. Groth A, et al. (2007) Regulation of replication fork progression through histonesupply and demand. Science 318(5858):1928–1931.

19. Li Q, et al. (2008) Acetylation of histone H3 lysine 56 regulates replication-couplednucleosome assembly. Cell 134(2):244–255.

20. Chen CC, et al. (2008) Acetylated lysine 56 on histone H3 drives chromatin assemblyafter repair and signals for the completion of repair. Cell 134(2):231–243.

21. Shimko JC, North JA, Bruns AN, Poirier MG, Ottesen JJ (2011) Preparation of fullysynthetic histone H3 reveals that acetyl-lysine 56 facilitates protein binding withinnucleosomes. J Mol Biol 408(2):187–204.

22. Aaltonen T, et al.; CDF Collaboration (2011) Observation of the baryonic flavor-changing neutral current decay Λb(0)→Λμ+μ-. Phys Rev Lett 107(20):201802.

23. Felsenfeld G, et al. (2004) Chromatin boundaries and chromatin domains. Cold SpringHarb Symp Quant Biol 69:245–250.

24. Kaplan T, et al. (2008) Cell cycle- and chaperone-mediated regulation of H3K56acincorporation in yeast. PLoS Genet 4(11):e1000270.

25. Watanabe S, Radman-Livaja M, Rando OJ, Peterson CL (2013) A histone acetylationswitch regulates H2A.Z deposition by the SWR-C remodeling enzyme. Science340(6129):195–199.

26. Chen CJ, Deng Z, Kim AY, Blobel GA, Lieberman PM (2001) Stimulation of CREBbinding protein nucleosomal histone acetyltransferase activity by a class of tran-scriptional activators. Mol Cell Biol 21(2):476–487.

27. Filippakopoulos P, et al. (2010) Selective inhibition of BET bromodomains. Nature468(7327):1067–1073.

28. Delvecchio M, Gaucher J, Aguilar-Gurrieri C, Ortega E, Panne D (2013) Structure of thep300 catalytic core and implications for chromatin targeting and HAT regulation. NatStruct Mol Biol 20(9):1040–1046.

29. Chimura T, Kuzuhara T, Horikoshi M (2002) Identification and characterization ofCIA/ASF1 as an interactor of bromodomains associated with TFIID. Proc Natl Acad SciUSA 99(14):9334–9339.

30. Akai Y, et al. (2010) Structure of the histone chaperone CIA/ASF1-double bromodo-main complex linking histone modifications and site-specific histone eviction. ProcNatl Acad Sci USA 107(18):8153–8158.

31. Bartsch O, et al. (2002) Molecular studies in 10 cases of Rubinstein-Taybi syndrome,including a mild variant showing a missense mutation in codon 1175 of CREBBP.J Med Genet 39(7):496–501.

32. Bannister AJ, Kouzarides T (1996) The CBP co-activator is a histone acetyltransferase.Nature 384(6610):641–643.

33. Mousson F, et al. (2005) Structural basis for the interaction of Asf1 with histone H3and its functional implications. Proc Natl Acad Sci USA 102(17):5975–5980.

34. English CM, Adkins MW, Carson JJ, Churchill ME, Tyler JK (2006) Structural basis forthe histone chaperone activity of Asf1. Cell 127(3):495–508.

35. English CM, Maluf NK, Tripet B, Churchill ME, Tyler JK (2005) ASF1 binds to a heter-odimer of histones H3 and H4: A two-step mechanism for the assembly of the H3-H4heterotetramer on DNA. Biochemistry 44(42):13673–13682.

36. Tagami H, Ray-Gallet D, Almouzni G, Nakatani Y (2004) Histone H3.1 and H3.3complexes mediate nucleosome assembly pathways dependent or independent ofDNA synthesis. Cell 116(1):51–61.

37. Liu X, et al. (2008) The structural basis of protein acetylation by the p300/CBP tran-scriptional coactivator. Nature 451(7180):846–850.

38. Thompson PR, Kurooka H, Nakatani Y, Cole PA (2001) Transcriptional coactivatorprotein p300. Kinetic characterization of its histone acetyltransferase activity. J BiolChem 276(36):33721–33729.

39. Pasini D, et al. (2010) Characterization of an antagonistic switch between histone H3lysine 27 methylation and acetylation in the transcriptional regulation of Polycombgroup target genes. Nucleic Acids Res 38(15):4958–4969.

40. Karanam B, Jiang L, Wang L, Kelleher NL, Cole PA (2006) Kinetic and mass spectro-metric analysis of p300 histone acetyltransferase domain autoacetylation. J Biol Chem281(52):40292–40301.

41. Black JC, Choi JE, Lombardo SR, Carey M (2006) A mechanism for coordinatingchromatin modification and preinitiation complex assembly. Mol Cell 23(6):809–818.

42. Hansson ML, et al. (2009) The transcriptional coactivator MAML1 regulates p300autoacetylation and HAT activity. Nucleic Acids Res 37(9):2996–3006.

43. Han J, Zhou H, Li Z, Xu RM, Zhang Z (2007) Acetylation of lysine 56 of histone H3catalyzed by RTT109 and regulated by ASF1 is required for replisome integrity. J BiolChem 282(39):28587–28596.

44. Iyer NG, Ozdag H, Caldas C (2004) p300/CBP and cancer. Oncogene 23(24):4225–4231.45. Murata T, et al. (2001) Defect of histone acetyltransferase activity of the nuclear

transcriptional coactivator CBP in Rubinstein-Taybi syndrome. HumMol Genet 10(10):1071–1076.

46. Kalkhoven E, et al. (2003) Loss of CBP acetyltransferase activity by PHD finger mu-tations in Rubinstein-Taybi syndrome. Hum Mol Genet 12(4):441–450.

47. van Hagen M, Overmeer RM, Abolvardi SS, Vertegaal AC (2010) RNF4 and VHL reg-ulate the proteasomal degradation of SUMO-conjugated Hypoxia-Inducible Factor-2alpha. Nucleic Acids Res 38(6):1922–1931.

48. Ogryzko VV, Schiltz RL, Russanova V, Howard BH, Nakatani Y (1996) The transcrip-tional coactivators p300 and CBP are histone acetyltransferases. Cell 87(5):953–959.

Das et al. PNAS | Published online March 10, 2014 | E1081

BIOCH

EMISTR

YPN

ASPL

US