Differential effects on p53-mediated cellcycle arrest vs. apoptosis by p90Chao Daia,b, Yi Tanga, Sung Yun Jungc, Jun Qinc, Stuart A. Aaronsond, and Wei Gua,1

aInstitute for Cancer Genetics, and Department of Pathology and Cell Biology, College of Physicians and Surgeons, Columbia University, 1130 St. NicholasAvenue, New York, NY 10032; bDepartment of Biological Sciences, Columbia University, New York, NY 10027; cDepartments of Biochemistry and CellBiology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030; and dDepartment of Oncological Sciences, Mount Sinai School of Medicine,New York, NY 10029

Edited by Robert G. Roeder, The Rockefeller University, New York, NY, and approved September 26, 2011 (received for review July 11, 2011)

p53 functions as a central node for organizing whether the cellresponds to stress with apoptosis or cell cycle arrest; however, themolecular events that lead to apoptotic responses are not comple-tely understood. Here, we identified p90 (also called Coiled-CoilDomain Containing 8) as a unique regulator for p53. p90 has noobvious effects on either the levels of p53 or p53-mediated cellcycle arrest but is specifically required for p53-mediated apoptosisupon DNA damage. Notably, p90 is crucial for Tip60-dependentp53 acetylation at Lys120, therefore facilitating activation of theproapoptotic targets. These studies indicate that p90 is a criticalcofactor for p53-mediated apoptosis through promoting Tip60-mediated p53 acetylation.

The p53 tumor suppressor acts as the major sensor for a reg-ulatory circuit that monitors signaling pathways from diverse

sources, including DNA damage, oncogenic events, and otherabnormal cellular processes (1, 2). p53 monitors and responds toa multitude of stress signals by coordinating cell growth arrest orapoptosis (3–5). Central to p53 regulation of these cellular pro-cesses is its activity as a transcription factor, although transcrip-tion-independent functions of p53 are also critical under somebiological settings. The activation of p53 transcription activity re-quires multiple steps, including sequence-specific DNA binding,antirepression and acquirement of combinations of posttransla-tional modifications, and recruitment of corepressors/coactivatorsin a promoter-specific manner (6). To suppress tumor growth, p53induces either cell growth arrest or apoptosis depending on thecellular context. The molecular mechanisms that govern the choicebetween growth arrest and apoptosis are extremely important butnot well understood (4, 7). As a key player in the stress response,p53 demands an exquisitely complicated network of control andfine-tuning mechanisms to ensure correct, differentiated responsesto the various stress signals encountered by cells (2, 5, 8–10).

p53 was the first nonhistone protein known to be regulatedby acetylation and deacetylation (11, 12). There is accumulatingevidence indicating that acetylation of p53 plays a major role inactivating p53 function during stress responses (2, 13, 14). Follow-ing our early findings of C terminus p53 acetylation, we andothers recently showed that p53 is also acetylated by Tip60 (alsoknown as KAT5)/MOF (human ortholog of males absent on thefirst) at residue Lys120 (K120) within the DNA-binding domain(15–17). K120 acetylation is crucial for p53-mediated apoptosisbut has no obvious effect on p21 expression, an essential targetof p53-mediated growth arrest. Notably, although Tip60 is re-quired for K120 acetylation of p53 in vivo, the levels of K120acetylation are dynamically regulated in vivo and the interactionbetween p53 and Tip60 is not very stable, indicating that addi-tional regulators may play a role in controlling K120 acetylationand subsequent p53-mediated apoptotic response (18–20).Through biochemical purification, we identified p90 as a uniqueregulator for p53. p90, also called CCDC8 (coiled-coil domaincontaining 8), which was previously found down-regulated inhuman cancer cells (21, 22), interacts with p53 both in vitro andin vivo. Knockdown of p90 has no obvious effect on p53-mediated

activation of p21 but specifically abrogates its effect on p53 up-regulated modulator of apoptosis, also known as Bbc3 (PUMA)activation. Moreover, p90 also interacts with Tip60 and promotesTip60-dependent Lys120 acetylation of p53, therefore enhancingthe apoptotic response of p53. These data reveal p90 as an up-stream regulator of the Tip60-p53 interaction and demonstratethat p90 is specifically required for p53-mediated apoptosis uponDNA damage.

ResultsIdentification of p90 as a Unique Component of p53-Associated Com-plexes. To further elucidate the mechanisms of p53-mediatedpromoter-specific activation in vivo, we isolated p53-associatedprotein complexes from human cells. Attempts to purify p53-containing protein complexes were hindered in the past becausecells cannot tolerate expressing even low levels of wild-type p53.Interestingly, our recent studies indicate that p538KR, in which alleight p53 acetylation sites are mutated to arginine (Fig. 1A), iscompletely inactive in inducing cell cycle arrest or apoptosis (23).Moreover, p538KR retains the capacity to bind target gene promo-ters as well as to activate the p53-Mdm2 (Mouse double minute2) feedback loop, suggesting that p538KR, unlike the hot spottumor mutant p53H175R, may retain a similar conformation aswild-type p53 in human cells. Therefore we have utilized an H1299p53-null lung carcinoma cell line that stably expresses a double-tagged human p538KR mutant protein with N-terminal FLAG andC-terminal HA epitopes (FLAG-p538KR-HA). To ensure physiolo-gical interactions, we used H1299 derivatives that express theectopic p538KR protein at levels comparable to those of endogen-ous p53 in HCT116 colon cancer cells upon DNA damage treat-ment. As expected, Mdm2 is activated in the p538KR stable line toa similar level compared to that induced by DNA damage inHCT116 cells. Consistent with previous findings, proapoptotic andgrowth-arrest targets such as PUMA and p21 are not activatedin the p538KR stable line (Fig. 1B). To isolate p538KR-containingcomplexes, cell extracts from the stable line were subjected to atwo-step affinity chromatography previously described (24). Thetandem affinity-purified p53-associated proteins were analyzedby liquid chromatography (LC)MS/MS. As expected, we identifiedknown p53 binding proteins such as Mdm2, tumor protein 53binding protein 1, ubiquitin specific peptidase 7, and the CREB-binding protein as specific components of the p53 complex. Inaddition, MS analysis of a protein band p90 (with the apparent sizeat approximately 90 kDa molecular mass) revealed six peptidesequences matched with a signal cDNA sequence in the database,

Author contributions: C.D. and W.G. designed research; C.D. and Y.T. performed research;S.Y.J. and J.Q. contributed new reagents/analytic tools; C.D., S.A.A., and W.G. analyzeddata; and C.D. and W.G. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase [accession no. JN703457 (p90)].1To whom correspondence should be addressed. E-mail: wg8@columbia.edu.

www.pnas.org/cgi/doi/10.1073/pnas.1110988108 PNAS ∣ November 22, 2011 ∣ vol. 108 ∣ no. 47 ∣ 18937–18942

BIOCH

EMISTR

Y

which is also namedCCDC8 (Fig. 1C). Because none of the peptidesequences of p90 were identified from the control complexes pur-ified in parental H1299 cells, p90 is likely a unique binding partnerof p53. The cDNA of p90/CCDC8 encodes a 538 amino acid pro-tein which possesses no known functional domains other than twosmall coiled-coil regions that are likely to mediate protein–proteininteractions (Fig. 1C). Although p90/CCDC8 has been reported asa candidate tumor suppressor gene in renal cell carcinoma (RCC),the molecular function of this protein is unclear (25).

p90 is a Bona Fide p53 Interacting Protein. To investigate a role forp90 in regulating p53 function in vivo, we first tested the interac-tion between p90 and p53. Thus, we first transfected H1299 cellswith expression vectors for FLAG-tagged p53 and HA-taggedp90. Western blot analysis revealed that p90 is readily detected inp53-associated immunoprecipitates (Fig. 2A). Because p90 wasidentified in the p53 complex purified from cell extracts, weassessed the cellular localization of the p90–p53 interaction. To thisend, we established a U2OS cell line stably expressing FLAGand HA double-tagged p90. Using parental U2OS cells as control,we fractionated the cell extracts from the p90 stable line andimmunoprecipitated both nuclear and cytoplasmic fractions withM2/FLAG agarose beads. Western blot analysis showed that p53,as expected, localizes mainly in the nuclear fraction, whereas p90 ispresent in both fractions but more so in the cytoplasmic fraction(Fig. 2B). Furthermore, in the M2 immunoprecipitate, p53 is pri-marily found in the nuclear fraction (Fig. 2B). The difference inp90/p53 ratio in the cytoplasmic and nuclear fractions as well asthe p53 abundance in the nuclear M2 immunoprecipitate indicatethat, although p90 is present in both the cytoplasm and the nucleus,it interacts with p53 predominantly in the nucleus.

To further elucidate this interaction under physiological set-tings, we then raised an affinity-purified polyclonal antiserumagainst the full-length p90 protein. Upon Western blot analysis,

this antibody specifically detected in human cells an approxi-mately 90 kDa polypeptide (Fig. 2C, lane 1), the level of whichdecreases significantly after treatment with p90-specific siRNAoligos (Fig. 3A). To investigate the interaction between endogen-ous p90 and p53 proteins, extracts fromU2OS osteosarcoma cellswere immunoprecipitated with α-p53 or with the control IgG. Asexpected, the α-p53 antibody immunoprecipitated endogenousp53; more importantly, p90 is easily detected in the immunopre-cipitates obtained with the α-p53 antibody but not the controlIgG (Fig. 2C, lanes 2 and 3), confirming that p90 and p53 interactendogenously. An in vitro GST-pulldown assay was performed tofurther assess direct interaction. p53 can be divided into an N-terminal (NT) fragment containing the transactivation domain,a middle fragment (M) containing the DNA-binding domain,and a C-terminal (CT) fragment containing the tetramerizationdomain as well as the regulatory domain. Purified recombinantGST-tagged p53 full-length and fragment proteins were incu-bated with in vitro translated 35S-methione-labeled HA-p90. Fol-lowing immobilization with GST resins and recovery of capturedcomplexes using reduced glutathione, the eluted complexeswere resolved by SDS-PAGE and analyzed by autoradiography.35S-labeled HA-p90 strongly bound immobilized GST-tagged full-length and CT fragment of p53 (Fig. 2D, lanes 2 and 5), but notthe NTand middle fragments of p53 or GSTalone (Fig. 2D, lanes3, 4, and 6). These data demonstrate that p90 interacts with p53 invitro through binding directly to the C-terminal portion of p53.

Inactivation of p90 Attenuates p53-Mediated Activation of PUMA butNot p21.To understand the physiological role of p90, we examinedwhether inactivation of endogenous p90 has any effect on thestability and functions of p53. To this end, U2OS cells were trans-fected with a p90-specific (p90-RNAi#1) siRNA oligo or a con-trol (control-RNAi) siRNA oligo. As shown in Fig. 3A, lanes 1and 2, the level of endogenous p90 polypeptides was severely re-

A

B

C

Fig. 1. Identification of p90 as a component of ap53-containing protein complex. (A) Schematic repre-sentation of the p538KR protein used for protein com-plex purification. Mutations of acetylation sites areindicated. TAD, transcription activation domain; PRD,proline rich domain; DBD, DNA-binding domain; TD,tetramerization domain; CTD, C-terminal regulatorydomain. (B) Characterization of H1299 cells stably ex-pressing p538KR. Total cell extracts from FLAG-p538KR-HA/H1299 stable cell line and HCT116 cells with orwithout 8-h treatment with 20 μM etoposide wereassayed by Western blot analysis using antibodiesagainst p53, β-actin, Mdm2, PUMA, and p21. (C) Sche-matic representation of the p90 protein. p90 containstwo coiled-coil regions and two potential ATM/ATRphosphorylation sites. Peptide sequences identifiedfrom the mass spectrometric analysis are presented.

18938 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110988108 Dai et al.

duced after transfection with p90-RNAi. p53 protein level wasunaffected by p90 ablation, suggesting that p90 does not regulatep53 stability. We then assessed the effect of p90 inactivation onthe level of two important p53 downstream targets: the growth-arrest target p21 and the apoptotic target PUMA. Surprisingly,p90 ablation displayed differential effects on the two differentendogenous targets: The level of PUMA was significantly re-duced, whereas p21 expression remained unchanged. To excludeoff-target effects, we also treated cells with three additional p90siRNAs (p90-RNAi#2, p90-RNAi#3; p90-RNAi#4) that recog-nize different regions of the p90 mRNA. Again, the levels ofPUMA were decreased by p90 knockdown, although there wasno significant change for the levels of p53 and p21 (Fig. 3A, lanes3–5). Because p53 is strongly activated upon DNA damage andregulates downstream targets, we wanted to assess whether p90

affects p53 and downstream target activation upon DNA damage.U2OS cells were transfected with p90-specific siRNA oligosfollowed by treatment with the DNA damage reagent etoposide.As expected, p53 levels increased drastically upon DNA damage(Fig. 3B, lane 2 vs. lane 1), and notably, RNAi-mediated ablationof p90 displays no effect on p53 accumulation following etoposidetreatment (Fig. 3B, lane 4, 6, 8, and 10 vs. lane 2). p21 was stronglyinduced upon treatment, however, damage-induced PUMA ex-pression was severely attenuated in the cells treated with p90-RNAi (Fig. 3B, lanes 4, 6, 8, 10 vs. lane 2).

To validate that the differential effect of p90 on PUMA andp21 is p53 dependent, we inactivated both p53 and p90 in U2OSusing RNAi prior to etoposide treatment. In cells transfected withthe control siRNA, p53 accumulates and both PUMA and p21are activated significantly upon treatment (Fig. 3C, lanes 1 and

A

C

D

B

Fig. 2. p90 is a bona fide p53 interacting protein. (A) p90 coimmunopreci-pitates with p53 in an overexpression system. H1299 cells were transientlytransfected with plasmid DNA expressing HA-p90 or/and FLAG-p53. The cellextracts and the M2 immunoprecipitates (IP) were analyzed by Western blotanalysis using antibodies against HA and p53. (B) p90 localizes in both thecytoplasm and nucleus but interacts with p53 predominantly in the nucleus.Parental U2OS or FLAG-HA-p90/U2OS stable cell line were fractionated (seeMaterials and Methods). Extracts and M2 immunoprecipitates from the cyto-plasmic and nuclear fractions were assayed by Western blot analysis usingantibodies against HA and p53. β-tubulin and proliferating cell nuclear anti-gen (PCNA) were used as cytoplasmic and nuclear protein markers, respec-tively. (C) p90 interacts with p53 endogenously. Total cells extracts from U2OScells were immunoprecipitated with a p53-specific antibody (DO-1) or acontrol IgG . Extracts and immunoprecipitates were assayed by Western blotanalysis using antibodies against p90 and p53. (D) p90 interacts with p53 invitro. The full-length and fragmented GST-p53 fusion proteins or GST alonewere used in the GST-pulldown assay with in vitro translated 35S-methione-labeled HA-p90 protein. The immobilized complexes were resolved by SDS-PAGE and analyzed by autoradiography. The levels of the GST fusion proteinsare shown in the bottom panel stained by Coomassie blue.

p21

PUMA

Actin

p53

p90

#1 #2 #3 #4

A

1 2 3 4 5

p90 RNAi

ctrl

RN

Ai

p21

PUMA

Actin

p53

p90

- +- +- +- +- +

#1 #2 #3 #4

p90 RNAi

Etop

1 2 3 4 5 6 7 8 9 10

ctrl

RN

Ai

B

ctrl RNAip53 RNAip90 RNAietoposide - +

- -

+ +

p21

PUMA

Actin

p53

p90

+ +

- +- +

+ ++ + - -

- -- -- -

1 2 3 4 5 6

C

Fig. 3. Inactivation of p90 reduces basal PUMA level and differentiallyaffects p53-mediated PUMA and p21 induction upon DNA damage. (A) p90RNAi does not affect p53 stability or p21 basal level but reduces basal PUMAexpression. U2OS cells were transiently transfected with either control siRNAor four different p90-specific siRNA oligos. Cell extracts were assayed byWes-tern blot analysis using antibodies against p90, p53, PUMA, p21, and β-actin.(B) p90 RNAi attenuates PUMA but not p21 activation upon DNA damage.U2OS cells transfected with the indicated siRNAs were treated with or with-out 20 μM etoposide for 8 h. Total cell extracts were assayed by Western blotanalysis. (C) Differential regulation of PUMA and p21 activation by p90 isdependent on p53. p53 alone, or both p53 and p90 were inactivated inU2OS cells using RNAi. Subsequently, cells were treated with or without20 μM etoposide for 8 h before extraction and Western blot analysis.

Dai et al. PNAS ∣ November 22, 2011 ∣ vol. 108 ∣ no. 47 ∣ 18939

BIOCH

EMISTR

Y

2). In a p53-deficient background, DNA damage fails to activatePUMA and p21 (lanes 3 and 4), and more importantly p90 abla-tion displayed no effect on PUMA and p21 in the absence of p53(lanes 5 and 6). Taken together, these data demonstrate that p90inactivation differentially affects PUMA and p21 induction in ap53-dependent manner.

p90 is Required for p53-Mediated Apoptosis upon DNA Damage. Tofurther confirm the differential effects on p53-mediated activa-tion of p21 versus PUMA, we collected the cells at different timepoints following treatment with etoposide. At all time points, p53accumulation and p21 activation were unaffected by p90 ablationbut PUMA induction was severely attenuated (Fig. 4A, lanes 5and 6 vs. lanes 2 and 3). We furthered confirmed that p90 ablationaffected p53-dependent activation of p21 and PUMA at the tran-scription level by examining the mRNA levels of these targets.Indeed, basal PUMAmRNAwas reduced in samples treated withp90-RNAi, consistent with our finding that p90 ablation reducesbasal PUMA protein level (Fig. 3A). PUMA activation wasattenuated at the mRNA level following p90 ablation, whereasp21 mRNA level increased upon etoposide treatment at all timepoints and remained unaffected in samples treated with p90-RNAi. To further confirm these differential effects of p90 in p53responses, we repeated these experiments in the cells treated withanother DNA damage reagent doxirubicin. Again, we observedthe differential effects of p90 on p53-dependent p21 and PUMAactivation upon doxirubicin treatment (Fig. 4B, lanes 6–8 vs. lanes2–4). These results suggest that p90 is crucial for p53-dependentactivation of PUMA, which is a very important mediator of p53-mediated apoptosis (26–28). We therefore speculated whetherthe attenuation of PUMA activation by loss of p90 can be trans-lated into a phenotypic effect on apoptosis. To this end, we trans-fected U2OS cells with either control siRNA or p90 siRNA priorto etoposide treatment. Cells were collected at different time-points, stained with propidium iodide (PI), and analyzed by flow

cytometry for apoptotic cells according to DNA content. Asshown in Fig. 4C, basal level sub-G1 content is minimally affectedby inactivation of p90. However, following 18 or 24 h of etoposidetreatment, an average of 18.67% or 26.84% of cells transfectedwith control siRNA were apoptotic, whereas only 8.04% or11.49% of cells transfected with p90 siRNA were apoptotic(Fig. 4D). These data demonstrate that p90 is crucial for p53-mediated apoptosis.

Mechanistic Insights into p90-Mediated Effect on p53-DependentApoptotic Responses. Previous studies demonstrated that p53acetylation at Lys120 (p53 AcK120) by Tip60 is indispensable forapoptosis but not required for growth arrest (16, 17), leading toour speculation that p90 may regulate p53-mediated PUMAactivation through promoting Tip60-dependent acetylation ofp53 at K120. To investigate the mechanism underlying the effectof p90 on the apoptotic target PUMA, we first assessed the inter-action between p90 and Tip60. To this end, H1299 cells weretransfected with expression vectors for Tip60 and FLAG/HAdouble-tagged p90. Western blot analysis revealed that Tip60 isreadily detected in p90 associated immunoprecipitates (Fig. 5A).Using a GST-pulldown assay, we further tested the in vitro inter-action of Tip60 and p90. As shown in Fig. 5B, Tip60 bound toimmobilized GST-tagged p90 but not GST alone, demonstratingthat p90 and Tip60 interacts directly. To investigate the role ofp90 in p53 K120 acetylation by Tip60, we examined whetherTip60-mediated p53 acetylation is modulated by p90 status. Asexpected, Western blot analysis revealed that p53 was readilyacetylated by Tip60 (Fig. 5C, lane 2 vs. lane 1). Notably, althoughp90 itself does not acetylate p53 (Fig. 5C, lane 3), p53 acetylationby Tip60 was significantly enhanced upon p90 expression (Fig. 5C,lane 4 vs. lane 2). These data demonstrate that p90 promotes theaceylation of p53 by Tip60.

In order to confirm the effect of p90 on Tip60-mediated p53K120 acetylation under physiological settings, we inactivated p90

p<0.01p<0.05

Apo

ptot

icce

lls(%

)

D

p21

PUMA

Actin

p53

p90

2480 2480ctr

l RNAi

p90

RNAi

Etop (h)

A

puma

p21

HPRT

1 2 3 4 5 6

p21

PUMA

Actin

p53

2460 4 2460 4

p90 RNAi

Dox (h)B ctrl RNAi

1 2 3 4 5 6 7 8

p90

C

Etop(h)

18 h

24 h

ctrl RNAi p90 RNAi

0 h

FL2-H FL2-H

FL2-H FL2-H

FL2-H FL2-H

RT-PCR Fig. 4. Inactivation of p90 attenuates p53-depen-dent PUMA activation in time point experimentsand impairs apoptosis upon damage. (A and B) p90RNAi reduces PUMA but not p21 activation uponDNA damage. U2OS cells transiently transfected witheither control siRNA or p90-specific siRNA were trea-ted with 20 μM etoposide or 0.34 μM doxirubicin forthe indicated time. Cell extracts were analyzed byWestern blot analysis. Total RNA were isolated andRT-PCR was performed. (C) FACS analysis of p90-inac-tivated U2OS cells treated with etoposide. U2OS cellstransiently transfected with either control siRNA orp90 siRNA were treated with 20 μM etoposide forthe indicated time. Cells were fixed in cold methanol,stained with propidium iodide and subjected to DNAcontent analysis by flow cytometry. (D) p90 RNAiattenuates apoptosis. Percentages of apoptotic cellsfrom C are presented. Values are an average of threeindependent experiments. Error bars, �1 standarddeviation.

18940 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110988108 Dai et al.

in U2OS cells via RNAi and assessed endogenous acetylation ofp53 at K120. Because the steady-state levels of K120 acetylationare dynamically regulated by both acetylases and deacetylases,in order to exclude the potential effect on p53 acetylation levelsby deacetylases, cells were treated with deacetylase inhibitorstrichostatin A (for inhibiting histone deacetylase 1/histone deace-tylase 2-mediated deacetylation of p53) and nicotinamide (forinhibiting Sirt1-mediated deacetylation of p53) prior to harvest-ing (16, 18, 19). Cell extracts were immunoprecipitated with α-Ac-p53K120 or control IgG. As shown in Fig. 5D, p53 acetylationat K120 was easily detected in the cells with the deacetylase in-hibitor treatment; however, the levels of p53 acetylation at K120were significantly reduced upon p90 knockdown (Fig. 5D, lane 4vs. lane 2). Taken together, these data indicate that p90 is a cri-tical cofactor for p53-mediated apoptosis through promotingK120 acetylation of p53.

DiscussionOur findings reveal that p90 is a p53 interacting protein with dif-ferential effects on p53-mediated activation of target genes.Here, we have demonstrated that p90 is a bona fide p53 inter-acting protein and that this interaction primarily occurs in thenucleus. Inactivation of p90 attenuates apoptosis due to down-regulation of p53-mediated PUMA activation upon DNAdamage. However, p90 does not appear to affect growth-arresttargets such as p21. To dissect the molecular mechanism under-lying this differential regulation, we found that p90 interacts withthe Tip60 acetyltransferase and promotes Tip60-mediated acet-ylation of p53 at K120, a posttranslational modification that haspreviously been reported to modulate the decision between cellcycle arrest and apoptosis (15–17). Thus, p90 likely serves as anupstream regulator of the p53-Tip60 interplay that is required forapoptotic signaling and allows for transcription induction ofPUMA in cells at risk of DNA damage.

K120 is located within the p53 DNA-binding domain and isrecurrently mutated in cancer (UMD_TP53 mutation databasehttp://p53.free.fr/). Acetylation at K120 is indispensable for acti-vation of proapoptotic targets but is not required for activationof growth-arrest targets (16, 17). Although the mechanism under-lying this target specificity remains to be elucidated, it is possiblethat acetylation at K120 may impose specificity through alteringthe p53 quaternary structure and thus endowing p53 binding tolow-affinity response elements that are found on proapoptoticpromoters (29, 30). We also noticed a small amount of cytoplas-mic p53–p90 interaction. Cytoplasmic localization of p53 wasoriginally thought to passively block transactivation in the nu-cleus. However increasing evidence suggests cytoplasmic p53 hasimportant roles in regulating apoptosis and autophagy. Cytoplas-mic p53 promotes apoptosis through increasing mitochondrialouter-membrane permeabilization and release of cytochrome c(31–33). Basal levels of wild-type p53 in the cytoplasm also inhi-bits autophagy, although the exact mechanism remains to beunderstood (34). It will be interesting to explore the possibilitiesof p90 regulating transcription-independent functions of p53 inthe cytoplasm.

It is noteworthy that p90 itself is underexpressed in humantumors, including kidney cancer and myeloma, based on the can-cer gene expression profile database from Oncomine Research(21, 22). In this regard, p90 has also been identified as a candidatetumor suppressor gene as hypermethylation and transcriptionalsilencing of the p90 promoter was found in 35% of primaryRCC tumor samples (25). It will be interesting to test whetherp53-mediated apoptosis is abrogated in the human tumors lack-ing p90 expression and whether reactivation of silenced p90promotes apoptosis thereby contributing to tumor suppression.Finally, protein modifications of the components in the p53 path-way are well accepted as the key mechanisms for controlling p53function during stress responses (2, 35). Interestingly, p90 con-tains two potential ataxia telangiectasia mutated/ataxia telangiec-tasia and Rad3 related (ATM/ATR) phosphorylation sites at Ser-199 and Ser-302 (Fig. 1C). Indeed, in a screen assay performed bythe Elledge Group (36), a phosphorylated peptide derived fromp90 was identified as an ATM/ATR substrate. Future investiga-tions are required to dissect whether p90 phosphorylation uponDNA damage modulates its interaction with p53 and Tip60 aswell as p53-mediated apoptotic responses. It is possible that p90is functionally regulated by ATM/ATRmediated phosphorylationduring the DNA damage response to control the decision be-tween cell cycle arrest and apoptosis mediated by p53.

Materials and MethodsPlasmids, Antibodies and Cell Culture. The full-length p90 cDNA was amplifiedby PCR from Human MGC Verified FL cDNA (Open Biosystems) and subclonedinto pcDNA3.1/V5-His-Topo vector (Invitrogen), pCIN4-FLAG-HA, or pCIN4-HAexpression vector (37). To construct the GST-p90 plasmid, cDNA sequencescorresponding to the full-length p90 were amplified by PCR from other ex-

+ ++-

Tip60

FH-p90

Tip60

p90

IP/M2

input

Tip60

p90

1 2

A

p90

p53

Tip60

Ac-p53K120

p53

HA-p90

Tip60

F-p53 + + + ++ +

+ +-

---

input

IP/M2

1 2 3 4

C

ctrl siRNA

p90 siRNA

TSA/NTA- -

- -+ ++ +

+- +-p90

p53

Actin

Ac-p53K120

input

IP/Ac-p53K120

IP/ctrl IgG

1 2 3 4

D

B

Tip60

10%

inpu

t

GS

T

GS

T-p9

0

1 2 3

Fig. 5. p90 interacts with Tip60 and promotes Tip60-mediated p53 acetyla-tion at K120. (A) Tip60 coimmunoprecipitates with p90 in an overexpressionsystem. H1299 cells were transiently transfected with the plasmid DNAexpressing Tip60 or/and FLAG-HA-p90. Cell extracts and M2 immunoprecipi-tates were assayed by Western blot analysis using α-Tip60 and α-HA antibo-dies. (B) p90 interacts with Tip60 in vitro. The GST-p90 fusion protein orGST alone was used in the GST-pulldown assay with in vitro translated 35S-methione-labeled FLAG-HA-Tip60 protein. The immobilized complexes wereresolved by SDS-PAGE and analyzed by autoradiography. (C) p90 promotesp53 acetylation by Tip60 at K120. H1299 cells were transiently transfectedwith plasmid DNA expressing FLAG-p53, Tip60, and/or HA-p90. Cell extractsand M2 immunoprecipitates were assayed by Western blot analysis usingantibodies against HA, p53, Tip60, and p53-AcK120. (D) Inactivation of p90significantly reduces p53 acetylation at K120. U2OS cells were transientlytransfected with either control siRNA or p90 siRNA, and treated with1 μM trichostatin A (TSA) and 5 mM nicotinamide (NTA) 6 h prior to harvest-ing. Cell extracts and immunoprecipitates obtained with α-Acp53K120 orcontrol IgG were analyzed by Western blot analysis using antibodies againstp90, p53, and β-actin.

Dai et al. PNAS ∣ November 22, 2011 ∣ vol. 108 ∣ no. 47 ∣ 18941

BIOCH

EMISTR

Y

pression vectors and subcloned into pGEX (GST) vectors for expression inbacteria.

The polyclonal antibody specific for p90 was generated by Covance. Rab-bits were immunized with purified full-length GST-p90 protein. Antiserafrom the immunized rabbits were first depleted with a GST-affinity column,then affinity purified by use of a GST-p90 affinity column using the AminolinkPlus Immobilization kit (Thermo Scientific). Antibodies used for Western blotanalysis are p53 (DO-1), β-tubulin (D-10), p21 (C-19 and SX118), and prolifer-ating cell nuclear antigen (PC10) from Santa Cruz, β-actin (AC-15), PUMA(NT), and FLAG M2 from Sigma, Mdm2 (Ab-5) from EMD Biosciences, HA(3F10) from Roche Applied Science, and α-Acp53K120 antibody (16). Anti-Tip60 (CLHF) was a gift from Chiara Gorrini and Bruno Amati (European In-stitute of Oncology, Milan, Italy).

H1299 and U2OS cells were maintained in DMEM (Cellgro) and HCT116cells in McCoy’s 5A medium (Cellgro). All media were supplemented with10% fetal bovine serum (Gibco). The stable cell lines were established bytransfecting H1299 or U2OS cells with the plasmids pCIN4-FLAG-p538KR-HAand pCIN4-FLAG-HA-p90, respectively, followed by selection with 1 mg∕mLor 0.5 mg∕mL G418 (EMD Biosciences). Transfections with plasmid DNA wereperformed using the calcium phosphate method and siRNA transfections byLipofectamine2000 (Invitrogen) according to the manufacturer’s protocol.

Western Blot Analysis and Immunoprecipitation. ForWestern blot analysis, cellswere lysed in cold FLAG lysis buffer [50 mM Tris·HCl (pH 7.9), 137 mM NaCl,10 mM NaF, 1 mM EDTA, 1% Triton X-100, 0.2% Sarkosyl, 10% glycerol, andfreshly supplemented protease inhibitor cocktail]. For immunoprecipitations,cells were lysed in cold BC100 buffer [20 mM Tris, (pH 7.9), 100 mM NaCl,10% glycerol, 0.2 mM EDTA, 0.2% Triton X-100, and freshly supplementedprotease inhibitor]. For immunoprecipitations of ectopically expressedFLAG-tagged proteins or from the U2OS FH-p90 stable line, extracts were in-cubated with the monoclonal M2/FLAG agarose beads (Sigma) at 4 °C over-night. After five washes with the lysis buffer, the bound proteins were elutedusing FLAG-peptide (Sigma) in BC100 for 2 h at 4 °C. The eluted material wasresolved by SDS-PAGE and immunoblotted with antibodies as indicated.

Preparation of Cytoplasmic and Nuclear Fractions. Cytoplasmic extracts wereprepared by resuspension of pelleted cells in hypotonic buffer [10 mM Tris(pH 7.9), 10 mM KCl, 1.5 mM MgCl2, supplemented with fresh protease in-

hibitor] followed by Dounce homogenization (six strokes with Type A pestle)and subsequent low-speed pelleting of nuclei (600 × g for 10 min). The super-natant was removed for use as cytoplasmic extract. The pellet from the low-speed spin was washed once with hypotonic buffer containing 0.1% NonidetP-40, and further extracted with BC200 [20 mM Tris (pH 7.9), 200 mM NaCl,10% glycerol, 0.2 mM EDTA, 0.4% Triton X-100], supplemented with freshprotease inhibitor. The nuclear extract was then clarified by high-speed cen-trifugation (21;885 × g for 15 min). For subsequent immunoprecipitation,both fractions were adjusted to a final concentration of 150 mM NaCl and0.2% Triton X-100.

GST-Pulldown Assay. GST and GST-tagged protein fragments were purifiedas described previously (38). 35S-methione-labeled proteins were preparedby in vitro translation using the TNT Coupled Reticulocyte Lysate System (Pro-mega). GSTor GST-tagged fusion proteins were incubated with in vitro trans-lated 35S-methione-labeled proteins overnight at 4 °C in BC100 containing0.2% Triton X-100 and 0.2% BSA. GST resins (Novagen) were then added,and the solution was incubated at 4 °C for 3 h. After five washes, the boundproteins were eluted for 1.5 h at 4 °C in BC100 containing 0.2% Triton X-100and 20 mM reduced glutathione (Sigma), and resolved by SDS-PAGE. The pre-sence of 35S-labeled protein was detected by autoradiography.

siRNA-Mediated Ablation of p90 and p53. Ablation of p90 was performed bytransfection of U2OS cells with siRNA duplex oligonucleotides [p90-RNAi-1(5′-GGACUUGACAACUGACGAA-3′), p90-RNAi-2 (5′-GGCAAGAAGGUGCG-CAAAA-3′), p90-RNAi-3 (5′-GCAGAUAAUCAGAGGGCGG-3′), and p90-RNAi-4 (5′-ACACAAUGGGGUUGCGUCA-3′)] synthesized by Dharmacon. Ablationof p53 was performed by transfection of U2OS cells with siRNA duplexoligoset (On-Target-Plus Smartpool L00332900, Dharmacon). Control RNAi(On-Target-Plus siControl nontargeting pool D00181010, Dharmacon) wasalso used for transfection. RNAi transfections were performed two times withLipofectamine 2000 according to the manufacturer’s protocol (Invitrogen).

ACKNOWLEDGMENTS. This work was supported in part by grants fromNational Institutes of Health, National Cancer Institute, the Leukemiaand Lymphoma Society and The Ellison Medical Foundation. We thankE. McIntush from Bethyl, Inc. for developing the antibodies for p90/CCDC8.W.G. is an Ellison Medical Foundation Senior Scholar in Aging.

1. Vousden KH, Prives C (2009) Blinded by the light: The growing complexity of p53. Cell137:413–431.

2. Kruse JP, Gu W (2009) Modes of p53 regulation. Cell 137:609–622.3. Trigiante G, Lu X (2006) ASPP [corrected] and cancer. Nat Rev Cancer 6:217–226.4. Das S, Boswell SA, Aaronson SA, Lee SW (2008) P53 promoter selection: Choosing

between life and death. Cell Cycle 7:154–157.5. Liu G, Chen X (2006) Regulation of the p53 transcriptional activity. J Cell Biochem

97:448–458.6. Dai C, GuW (2010) p53 post-translational modification: Deregulated in tumorigenesis.

Trends Mol Med 16:528–536.7. Vousden KH, Lu X (2002) Live or let die: The cell’s response to p53. Nat Rev Cancer

2:594–604.8. Carter S, Vousden KH (2009) Modifications of p53: Competing for the lysines. Curr

Opin Genet Dev 19:18–24.9. Vives V, et al. (2006) ASPP2 is a haploinsufficient tumor suppressor that cooperates

with p53 to suppress tumor growth. Genes Dev 20:1262–1267.10. Brown L, et al. (2007) CDIP, a novel pro-apoptotic gene, regulates TNFalpha-mediated

apoptosis in a p53-dependent manner. EMBO J 26:3410–3422.11. Gu W, Roeder RG (1997) Activation of p53 sequence-specific DNA binding by acetyla-

tion of the p53 C-terminal domain. Cell 90:595–606.12. Luo J, Su F, Chen D, Shiloh A, GuW (2000) Deacetylation of p53 modulates its effect on

cell growth and apoptosis. Nature 408:377–381.13. Loewer A, Batchelor E, Gaglia G, Lahav G (2010) Basal dynamics of p53 reveal tran-

scriptionally attenuated pulses in cycling cells. Cell 142:89–100.14. Berger SL (2010) Keeping p53 in check: A high-stakes balancing act. Cell 142:17–19.15. Li X, Wu L, Corsa CA, Kunkel S, Dou Y (2009) TwomammalianMOF complexes regulate

transcription activation by distinct mechanisms. Mol Cell 36:290–301.16. Tang Y, Luo J, Zhang W, Gu W (2006) Tip60-dependent acetylation of p53 modulates

the decision between cell-cycle arrest and apoptosis. Mol Cell 24:827–839.17. Sykes SM, et al. (2006) Acetylation of the p53 DNA-binding domain regulates

apoptosis induction. Mol Cell 24:841–851.18. Mellert HS, et al. (2011) Deacetylation of the DNA-binding domain regulates

p53-mediated apoptosis. J Biol Chem 286:4264–4270.19. Wang J, Chen J (2010) SIRT1 regulates autoacetylation and histone acetyltransferase

activity of TIP60. J Biol Chem 285:11458–11464.20. Jha S, et al. (2010) Destabilization of TIP60 by human papillomavirus E6 results in

attenuation of TIP60-dependent transcriptional regulation and apoptotic pathway.Mol Cell 38:700–711.

21. Yusenko MV, et al. (2009) High-resolution DNA copy number and gene expressionanalyses distinguish chromophobe renal cell carcinomas and renal oncocytomas.BMC Cancer 9:152.

22. Zhan F, et al. (2007) Gene-expression signature of benign monoclonal gammopathyevident in multiple myeloma is linked to good prognosis. Blood 109:1692–1700.

23. Tang Y, Zhao W, Chen Y, Zhao Y, Gu W (2008) Acetylation is indispensable for p53activation. Cell 133:612–626.

24. Chen D, et al. (2005) ARF-BP1/Mule is a critical mediator of the ARF tumor suppressor.Cell 121:1071–1083.

25. Morris MR, et al. (2011) Genome-wide methylation analysis identifies epigeneticallyinactivated candidate tumour suppressor genes in renal cell carcinoma. Oncogene30:1390–1401.

26. Yu J, Wang Z, Kinzler KW, Vogelstein B, Zhang L (2003) PUMA mediates the apoptoticresponse to p53 in colorectal cancer cells. Proc Natl Acad Sci USA 100:1931–1936.

27. Villunger A, et al. (2003) p53- and drug-induced apoptotic responses mediated byBH3-only proteins puma and noxa. Science 302:1036–1038.

28. Nakano K, Vousden KH (2001) PUMA, a novel proapoptotic gene, is induced by p53.Mol Cell 7:683–694.

29. Weinberg RL, Veprintsev DB, Bycroft M, Fersht AR (2005) Comparative binding of p53to its promoter and DNA recognition elements. J Mol Biol 348:589–596.

30. Arbely E, et al. (2011) Acetylation of lysine 120 of p53 endows DNA-binding specificityat effective physiological salt concentration. Proc Natl Acad Sci USA 108:8251–8256.

31. Green DR, et al. (2004) Direct activation of Bax by p53 mediates mitochondrial mem-brane permeabilization and apoptosis. Science 303:1010–1014.

32. George DL, Leu JIJ, Dumont P, HafeyM,MurphyME (2004)Mitochondrial p53 activatesBak and causes disruption of a Bak-Mcl1 complex. Nat Cell Biol 6:443–450.

33. Mihara M, et al. (2003) p53 has a direct apoptogenic role at the mitochondria.Mol Cell11:577–590.

34. Tasdemir E, et al. (2008) Regulation of autophagy by cytoplasmic p53. Nat Cell Biol10:676–687.

35. Carter S, Vousden KH (2008) p53-Ubl fusions as models of ubiquitination, sumoylationand neddylation of p53. Cell Cycle 7:2519–2528.

36. Matsuoka S, et al. (2007) ATM and ATR substrate analysis reveals extensive proteinnetworks responsive to DNA damage. Science 316:1160–1166.

37. Abida WM, Gu W (2008) p53-Dependent and p53-independent activation of autop-hagy by ARF. Cancer Res 68:352–357.

38. Li M, et al. (2002) Deubiquitination of p53 by HAUSP is an important pathway for p53stabilization. Nature 416:648–653.

18942 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110988108 Dai et al.