retracted -...

Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199

Molecular and Cellular Pathobiology
Cancer
Research

High-Mobility Group A1 Proteins Regulate p53-MediatedTranscription of Bcl-2 Gene

Francesco Esposito1, Mara Tornincasa2, Paolo Chieffi3, Ivana De Martino2,Giovanna Maria Pierantoni2, and Alfredo Fusco1,4

Abstract

Authors'SperimentCellulare eUniversitàSperimentaCenter-Cenand the Eu

Corresponcologia SpRicerche, vFax: 39-08

doi: 10.115

©2010 Am

www.aacr

DowDowDow

R

8

DowDowDow

We have previously described a mechanism through which the high-mobility group A1 (HMGA1) proteinsinhibit p53-mediated apoptosis by delocalizing the p53 proapoptotic activator homeodomain-interacting pro-tein kinase 2 from the nucleus to the cytoplasm. By this mechanism, HMGA1 modulates the transcription ofp53 target genes such as Mdm2, p21waf1, and Bax, inhibiting apoptosis. Here, we report that HMGA1 antag-onizes the p53-mediated transcriptional repression of another apoptosis-related gene, Bcl-2, suggesting a nov-el mechanism by which HMGA1 counteracts apoptosis. Moreover, HMGA1 overexpression promotes thereduction of Brn-3a binding to the Bcl-2 promoter, thereby blocking the Brn-3a corepressor function onBcl-2 expression following p53 activation. Consistently, a significant direct correlation between HMGA1and Bcl-2 overexpression has been observed in human breast carcinomas harboring wild-type p53. Therefore,this study suggests a novel mechanism, based on Bcl-2 induction, by which HMGA1 overexpression contri-butes to the escape from apoptosis leading to neoplastic transformation. Cancer Res; 70(13); 5379–88. ©2010 AACR.13

, 201

r

e
Introduction
The high-mobility group A (HMGA) family is comprisedof three proteins: HMGA1a, HMGA1b, and HMGA2. Theyare encoded by two distinct genes, HMGA1a and HMGA1bproteins being products of the same gene generatedthrough alternative splicing (1). These proteins bind the mi-nor groove of AT-rich DNA sequences through their DNA-binding domain, the so-called “AT-hooks.” HMGA proteinsdo not have transcriptional activity per se; however, by in-teracting with the transcription machinery, they alter thechromatin structure and thereby regulate the transcription-al activity of several genes (2, 3). In normal cells and adulttissues, the levels of HMGA proteins are low or absent (4).In contrast, in neoplastically transformed cells as well as inembryonic cells, the constitutive expression of HMGA pro-teins is exceptionally high (5). Their overexpression is main-ly associated with a highly malignant phenotype, alsorepresenting a poor prognostic index because HMGA over-expression often correlates with the presence of metastasisas well as reduced survival (6). Moreover, both in vitro andac

ted D

ec

Affiliations: 1Istituto di Endocrinologia ed Oncologiaale del CNR c/o 2Dipartimento di Biologia e PatologiaMolecolare, Facoltà di Medicina e Chirurgia di Napoli,

degli Studi di Napoli “Federico II”, 3Dipartimento di Medicinale, Seconda Università di Napoli, and 4Naples Oncogenomictro di Ingegneria Genetica, Biotecnologie Avanzate-Napoli,ropean School of Molecular Medicine, Naples, Italy

ding Author: Alfredo Fusco, Istituto di Endocrinologia ed On-erimentale “Gaetano Salvatore” Del Consiglio Nazionale delleia Pansini 5, 80131 Naples, Italy. Phone: 39-081-746-3602;1-229-6674; E-mail: [email protected].

8/0008-5472.CAN-09-4199

erican Association for Cancer Research.

journals.org

Researcon Decembecancerres.aacrjournals.org nloaded from



etr

Researcon Februarycancerres.aacrjournals.org nloaded from



in vivo studies have established the oncogenic role of HMGAgenes (7–11).Recently, we have shown a novel mechanism, based on

HMGA1-p53 interaction, by which HMGA1 proteins havea role in the process of carcinogenesis. In fact, HMGA1binds p53 and interferes with the p53-mediated transcrip-tion of Bax, p21waf1, and Mdm2, leading to a reduction ofp53-dependent apoptosis (12). HMGA1 is also able to inter-fere with the apoptotic function of p53 by another mecha-nism that involves the p53 proapoptotic activatorhomeodomain-interacting protein kinase 2 (HIPK2), whichbinds to and phosphorylates p53 on Ser46 leading to apo-ptosis (13, 14). Indeed, HMGA1 overexpression promotesHIPK2 relocalization from the nucleus to the cytoplasm, in-hibiting p53 apoptotic function, whereas HIPK2 overexpres-sion re-established HIPK2 nuclear localization andsensitivity to apoptosis (15). Consistently, a strong correla-tion among HMGA1 overexpression, HIPK2 cytoplasmic lo-calization, and a low spontaneous apoptosis index wasobserved in wild-type p53-expressing human breast carcino-mas (15).On the basis of these data, we looked for other genes reg-

ulated by HMGA1 proteins among p53 target genes involvedin cell apoptosis. In particular, we have focused our attentionon B-cell lymphoma gene 2 (Bcl-2). Bcl-2 represents a goodcandidate as the target for HMGA1 proteins because its pro-moter region contains some AT-rich DNA sequences and itstranscription is regulated by p53 (16, 17). The Bcl-2 proteinexerts an antiapoptotic function: it inhibits the release of cy-tochrome c into the cytosol which, in turn, activates caspase-9 and caspase-3, leading to apoptosis (18). The Bcl-2 gene hasbeen found to be overexpressed in different cancers, includ-ing B-cell lymphoma (19), melanoma (20), neuroblastoma

mbe

5379

h. r 17, 2018. © 2010 American Association for Cancerh. r 17, 2018. © 2010 American Association for Cancerh. r 17, 2018. © 2010 American Association for Cancerh. 16, 2019. © 2010 American Association for Cancerh. 16, 2019. © 2010 American Association for Cancerh. 16, 2019. © 2010 American Association for Cancer

http://cancerres.aacrjournals.org/






Esposito et al.

5380

Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199 Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199

(21), breast (22), prostate (23), ovarian (24), and lung carcino-mas (25). Bcl-2 and its homologues might intervene in onco-genesis at different levels by favoring the persistence ofmalignant cells that, in normal circumstances, would beeliminated by apoptosis, and also by promoting resistanceto conventional cancer treatments (26).Here, we show that Bcl-2 is a gene regulated by the

HMGA1-p53-HIPK2 complex and that HMGA1 is able toabolish the repression promoted by p53 on Bcl-2 expres-sion. Subsequently, we looked for other components of thismultiprotein complex, finding that Brn-3a belongs to thiscomplex. The POU transcription factor Brn-3a strongly ac-tivates the expression of the Bcl-2 proto-oncogene, protect-ing neuronal cells from programmed cell death (16).Recent works show that Brn-3a expression is not restrictedto neuronal cells, as its activity was also detected in can-cer cells of nonneuronal origin (27). Latchman and collea-gues showed that the transcriptional repression exerted byp53 on the Bcl-2 promoter is achieved through Brn-3a, sug-gesting that the two proteins could cooperate in the neg-ative control of the expression of this gene (16). Finally, weshow that HMGA1 overexpression reduces Brn-3a bindingto the Bcl-2 promoter, removing Brn-3a from its role ascorepressor following p53 activation. In addition, validatingour data by immunohistochemical analyses, we found asignificant association between HMGA1 and Bcl-2 overex-pression in 11 of 14 bioptic tissues of human breast car-cinomas carrying wild-type p53.

e
Materials and Methods
Cell culture, transfections, and transactivation assaysND7, MCF7, mouse embryonic fibroblasts, and SAOS-2

cells were maintained in DMEM with 10% FCS (Life Technol-ogies; Invitrogen), glutamine, and antibiotics. FRTL5, FRTL5-KiMSV, and FRTL5-KiMSV αs HMGA1 cells were maintainedin F12 with 5% calf serum (Life Technologies; Invitrogen),glutamine, and antibiotics. Cells were transfected with plas-mids by LipofectAMINE plus reagent (Invitrogen) as sug-gested by the manufacturer. Cells were transientlytransfected with previously described reporter vectors (12,13, 16) and normalized with the use of a cotransfected β-gal construct. Luciferase activity was analyzed by Dual-LightSystem (Applied Biosystems). For inhibition of HMGA1 ex-pression, small interfering RNAs and corresponding scramblesmall interfering RNAs (Santa Cruz Biotechnology, Inc.) wereused as suggested by the manufacturer.

Expression constructsThe pCAG-p53, pCMV-Hmga1b, pCEFL-HA-Hmga1b,

pFLAG-HIPK2, pEGFP-HIPK2, pLTR-Brn-3a, and the Bcl-2promoter-reporter constructs have been previously described(12, 13, 16).

Western blotting and coimmunoprecipitationProtein extraction, Western blotting, and coimmunopreci-

pitation procedures were carried out as reported elsewhere

Retrac

ted D

ec

Cancer Res; 70(13) July 1, 2010

Researcon Decembecancerres.aacrjournals.org Downloaded from

Researcon Februarycancerres.aacrjournals.org Downloaded from

(12). Differential nuclear/cytoplasmic cell lysates were ob-tained as previously reported (15). The antibodies used forWestern blotting were as follows: anti-FLAG M5 (Sigma-Aldrich), anti-HA 12CA5 (Roche), anti-p53 DO1, anti–Brn-3a14A6, anti-Sp1 H225, and anti–γ-tubulin C11 (Santa CruzBiotechnology), anti–Bcl-2 50E3 (Cell Signaling), anti-HMGA1are polyclonal antibodies raised against a synthetic peptidelocated in the NH2-terminal region and recognize bothHMGA1a and HMGA1b isoforms (12), and anti-HIPK2 M01(Abnova).

In vitro protein translation and electrophoreticmobility shift assayDNA-binding assays with the recombinant protein were

performed as previously described (12). The radiolabeleddouble-strand oligonucleotide corresponds to a region span-ning from −526 to −477 nucleotides of the human Bcl-2 pro-moter region (16).

Cell viability and terminal nucleotidyl transferase–mediated nick end labeling assaysCell viability and terminal nucleotidyl transferase–mediated

nick end labeling (TUNEL) assays were carried out as reportedelsewhere (12).

Reverse transcriptase-PCRTotal RNA was isolated using TRI-reagent solution (Sigma)

according to the protocols of the manufacturer and treatedwith DNase (Invitrogen). Reverse transcription was per-formed according to standard procedures (Applied Biosys-tems). cDNA was amplified by PCR using the followingprimers: Mu-Bcl-2-up 5′-ggTCATgTgTgTggAgCg-3′, Mu-Bcl-2-dw 5′-gCAgAgTCTTCAgAgAC AgC-3′, Mu-Gapdh-up 5′-TggTgAAggTCggTgTgAAC-3′, and Mu-Gapdh-dw 5′-ggAA-gATggTgATg ggCTTC-3′.

Indirect immunofluorescenceIndirect immunofluorescence was carried out as reported

elsewhere (15).

Chromatin immunoprecipitation and reprecipitationAfter transfection, chromatin samples were processed

for chromatin immunoprecipitation (ChIP) and re-ChIP ex-periments as reported elsewhere (12). Samples were sub-jected to immunoprecipitation with the following specificantibodies: anti-p53 (Ab-7; Calbiochem), anti–Brn-3a(14A6; Santa Cruz Biotechnology), anti-HIPK2 (M01; Abno-va), and anti-HMGA1 (12). Immunoprecipitated chromatinwas amplified by PCR using the following primers: Hu-Bcl-2-pr-up 5′-gAAgCAgAAgTCTgggAATC-3′, Hu-Bcl-2-pr-dw5′-gCgTTTCCCTgTACACACTg-3′ , Hu-GAPDH-pr-up5′-gTATTCCCCCAggTTTACATg-3′ , Hu-GAPDH-pr-dw5′-TTCTCCATggTggTgAAgAC-3′, Mu-Bcl-2-pr-up 5′-ACA-CACAgAgCgAC TgTgTA-3′, Mu-Bcl-2-pr-dw 5′-gAgg-CACCTggATCTTTTCT-3′, Mu-GAPDH-pr-up 5′-TACTCgCggCTTTACgg g-3′, and GAPDH-pr-dw 5′-TggAACAgggAg-gAgCAg-3′.

mber 1

3, 20

18

Cancer Research

h. r 17, 2018. © 2010 American Association for Cancerh. 16, 2019. © 2010 American Association for Cancer



Role of HMGA1 on Bcl-2 Transcription


ImmunohistochemistryImmunohistochemistry was carried out as reported else-

where (15). HMGA1 and Bcl-2 proteins were consideredoverexpressed when more than 10% of the neoplastic cellspresented a strong immunoreaction. p53 was consideredpositive (mutated) only when a distinct nuclear stainingwas observed in more than 10% of cancer cells.

Results

HMGA1 correlates with Bcl-2 protein expressionLooking for a possible correlation between HMGA1 and

Bcl-2 protein expression, we analyzed their expression levelin the MCF-7 breast cancer cell line stably transfected withHMGA1b. All the HMGA1-transfected MCF7 clones showedhigher Bcl-2 protein levels than the untransfected cells(Fig. 1A). This correlation was then confirmed in other cellsystems. Indeed, mouse embryonic fibroblasts null for thehmga1 gene express lower Bcl-2 mRNA and protein levelsin comparison with the wild-type mouse embryonic fibro-blasts (data not shown). Moreover, a rat thyroid cell line,transformed with the Kirsten sarcoma virus (FRTL5-KiMSV), expressing HMGA1 shows higher Bcl-2 protein

www.aacrjournals.org


R

e


levels in comparison with normal uninfected cells. Con-versely, the suppression of HMGA1 protein expression inthe FRTL5-KiMSV–infected cells, by an antisense constructfor HMGA1, leads to a drastic reduction in Bcl-2 mRNAand protein levels (data not shown). To further supporta causal correlation between HMGA1 and Bcl-2 expression,a neuronal cell line (ND7) transiently interfered for the ex-pression of HMGA1 proteins showed a reduction in Bcl-2protein levels compared with cells transfected with ascrambled oligonucleotide (Fig. 1B).Because the interaction between HMGA1 and p53 mod-

ulates the transcription of p53 target genes, we investigat-ed whether Bcl-2 was regulated by HMGA1-p53 interaction.Therefore, ND7 cells, endogenously expressing HMGA1 andp53, were cotransfected with expression vectors encodingp53 and/or HMGA1b. According to previous data (17),p53 reduces Bcl-2 transcription, whereas HMGA1 overex-pression reverts the effects of p53 on Bcl-2 transcription,as shown by reverse transcription-PCR and Western blotanalyses (Fig. 1C and D). In addition, preliminary data alsosuggest that HMGA1a isoform and HMGA2 overexpressionare able to abolish p53 repression exerted on Bcl-2 genetranscription (data not shown). 13

, 201

8

er

d Dec

emb

Figure 1. HMGA1 correlates with Bcl-2protein expression and reverts theeffects of p53 on Bcl-2 transcription.Immunoblot analysis of Bcl-2 andHMGA1 expression in MCF7overexpressing HMGA1b protein (A)and in ND7 cells transiently interferedfor HMGA1 by small interferingRNAs (B). C, immunoblot analysisof Bcl-2 expression in ND7 cellscotransfected with expression vectorsencoding p53 and/or HMGA1bproteins. γ-Tubulin was used toequalize protein loading. D, RNA fromND7 cells transiently transfected asdescribed in C was analyzed byreverse transcription-PCR for Bcl-2expression. All cDNAs were coamplifiedwith GAPDH as an internal control.

etrac

t

Cancer Res; 70(13) July 1, 2010 5381




Esposito et al.

5382


HMGA1 proteins bind Bcl-2 promoter in vitro andin vivoSubsequently, we investigated whether HMGA1 is able to

bind the Bcl-2 gene-regulatory regions. As shown in Fig. 2A,using an electrophoretic mobility shift assay, we showedthat His-HMGA1b recombinant protein was able to bindthe region spanning from −526 to −477 nucleotides rela-ted to the transcription start site of the human Bcl-2gene (16), which contains AT-rich putative HMGA1 bind-ing sites and two p53 consensus sequences. Binding spe-cificity was demonstrated by competition experimentsshowing loss of binding with the addition of a 200-foldmolar excess of unlabeled Bcl-2 promoter oligonucleotide.Next, we evaluated whether HMGA1 proteins bind the hu-man Bcl-2 promoter in vivo by performing ChIP assays.MCF7 cells were transfected with HMGA1b, crosslinked,and immunoprecipitated with anti-HMGA1 or anti-IgGantibodies as control. Immunoprecipitation of chromatinwas then analyzed by semiquantitative PCR, using primersspanning the −676/−320 region of the Bcl-2 promoter. TheBcl-2 promoter region was amplified in both the MCF7cells transfected with HMGA1b or the backbone vectorbecause HMGA1 proteins are weakly expressed in MCF7(Fig. 2B). However, the Bcl-2 promoter amplification was

Retrac

ted D

e




higher in HMGA1-transfected cells in comparison with thecells transfected with the empty vector. Conversely, noamplification was observed with anti-IgG precipitatesand when primers for the control promoter GAPDH wereused (Fig. 2B), demonstrating that the binding is specificfor the Bcl-2 promoter. Analogous results were obtainedby ChIP performed in untransfected and HMGA1-transfectedmurine ND7 cell lines (Fig. 2C). These results indicatethat HMGA1 proteins are able to bind the Bcl-2 promoterin vitro and in vivo.

HMGA1 modulates p53-mediated transcriptionalactivity on Bcl-2 promoterTo verify the functional effects of HMGA1 on Bcl-2 pro-

moter, we evaluated the activity of a reporter vector car-rying the luciferase gene under the control of Bcl-2promoter in ND7 cells transfected with HMGA1 or anempty vector. As expected, Bcl-2 expression was drastical-ly reduced in cells overexpressing p53 compared withcontrol cells (Fig. 2D). HMGA1 overexpression alone hadno effect on Bcl-2 promoter activity, whereas the overex-pression of both HMGA1 and p53 resulted in a drasticreduction of Bcl-2 promoter repression compared withcells transfected with p53 alone. Furthermore, when

Figure 2. HMGA1b protein binds to Bcl-2 promoter. A, electrophoretic mobility shift assay performed with a radiolabeled oligonucleotide spanning from−526 to −477 of the human Bcl-2 promoter incubated with 5 and 10 ng of recombinant HMGA1b protein. To assess the specificity of the binding,His-HMGA1b protein was incubated in the presence of a 200-fold excess of unlabeled oligonucleotide. B, soluble chromatin from MCF7 cells transfectedwith HMGA1b was immunoprecipitated with anti-HMGA1. The DNAs were then amplified by semiquantitative PCR using primers that cover a region ofhuman Bcl-2 promoter (−676/−320). As an immunoprecipitation control, IgG was used. Amplification of the immunoprecipitated DNA using primers for theGAPDH gene promoter was used as control of specificity. C, ChIP was performed in a murine cell line (ND7), as described in B. HMGA1 revertsp53-mediated transcription of Bcl-2 gene. D, analysis of HMGA1 effects on p53 transcriptional activity using the Bcl-2 luciferase-reporter vector in ND7cells. All transfections were performed in duplicate and the data are means ± SD of five independent experiments. Backbone vectors were used as control.Western blot analysis of p53 and HMGA1 proteins from one representative experiment (bottom). γ-Tubulin was used to equalize protein loading.ce

mber 1

3, 20

18

Cancer Research





Published OnlineFirst June 8, 2010; DOI: 10.1158/0008-5472.CAN-09-4199

e


HMGA1 protein expression was transiently blocked by smallinterfering RNA, Bcl-2 promoter activity was significantlyreduced compared with untransfected cells. These resultsindicate that HMGA1 suppresses the inhibitory effect of p53on Bcl-2 transcription.

HMGA1 represses p53 apoptotic function mediated byBcl-2 downregulationTo evaluate the biological effects of HMGA1 on the regu-

lation of Bcl-2 expression by p53, SAOS-2 cells, whichexpress HMGA1 but are null for p53 expression, were trans-fected with a vector encoding p53, in the presence or ab-sence of HMGA1 overexpression. As expected, exogenousp53 expression induced cell death as shown by trypan bluetest (Fig. 3A). However, when HMGA1 was overexpressed to-gether with p53, the percentage of dead cells was drasticallyreduced (Fig. 3A).To verify the antiapoptotic activity of HMGA1 in a more

physiologic context, we incubated the wtp53-carrying ND7cells with H2O2 to activate the endogenous p53 in the pres-ence or absence of HMGA1 protein overexpression. As shownin Fig. 3B and C, HMGA1 overexpression significantly re-duced the number of trypan blue or TUNEL-positive cells in-duced by H2O2. Taken together, these data indicate thatHMGA1 could functionally interfere with the apoptotic func-tion of p53 mediated by Bcl-2 downregulation.

HMGA1 proteins bind to Brn-3a transcription factorThe data shown above strongly suggest the existence of a

multiprotein complex that regulates the transcription levelsof Bcl-2 supported from our previous studies, which alsoshowed that HMGA1 is able to constitute a complex withp53 and HIPK2 that regulates p53 apoptotic function (15). Ithas been previously shown that Brn-3a is a potent tran-scriptional regulator of the Bcl-2 promoter, able to bindboth p53 and HIPK2 (28). Therefore, we investigated wheth-er HMGA1 interacts with Brn-3a. ND7 cells were transientlycotransfected with hemagglutinin (HA)-tagged HMGA1band Brn-3a expression vectors. Protein lysates were immu-noprecipitated with anti–Brn-3a or anti-HA antibodies andimmunoblotted with reciprocal antibodies (Fig. 3D and E).Coexpression of Brn-3a and HMGA1 resulted in the coim-munoprecipitation of the two proteins. Conversely, therewas no coimmunoprecipitation when ND7 cells were trans-fected with HA-HMGA1b or Brn-3a expression vectorsalone. Western blot analysis showed that the transfectedcells expressed adequate levels of the Brn-3a and HA-HMGA1b proteins (Fig. 3F).

The HMGA1-p53-Brn-3a-HIPK2 complex regulates Bcl-2promoter activityThe Brn-3a transcription factor has been shown to strongly

activate the expression of Bcl-2 protecting cells from apopto-sis. However, when p53 is overexpressed, Brn-3a acts as a core-pressor sustaining the p53-repressive action on Bcl-2 promoter(16). Moreover, HIPK2 interacts with Brn-3a, promoting itsbinding to DNA and suppressing Brn-3a–dependent transcrip-tion of some Bcl-2 family members (28). In addition, our previ-

Retrac

ted D

ec




ous results indicate that HIPK2 reverts the inhibitory activityof HMGA1 on the p53-dependent promoters and leads to acomplex interplay among p53, HIPK2, and HMGA1 in the reg-ulation of p53 target gene expression (15).Therefore, to better define the role of the HMGA1-p53-

Brn-3a-HIPK2 multicomplex in the regulation of Bcl-2 tran-scription, we analyzed Bcl-2 promoter activity using a vec-tor carrying the luciferase gene under the control of Bcl-2promoter in the SAOS-2 cells transiently transfected withdifferent combinations of plasmids expressing the proteinsparticipating in the complex. As expected from previousdata (16), p53 represses Bcl-2 transcription and this effectis increased by Brn-3a (Fig. 4A). When HMGA1 was co-transfected with p53 and Brn-3a, the repression on Bcl-2promoter was antagonized. In particular, the repressive ac-tivity exerted by Brn-3a was abolished (Fig. 4A). To verifywhether HIPK2 has a role in the transcriptional regulationof the Bcl-2 gene, HIPK2 was coexpressed with p53,HMGA1, and Brn-3a in ND7 cells. HIPK2 restores the tran-scriptional repression of Bcl-2 promoter exerted by p53 andBrn-3a counteracting HMGA1 effects (Fig. 4A).

HMGA1 represses p53-Brn-3a–mediated apoptosisWe have shown above that HMGA1 proteins are able

to abolish the repression exerted by p53-Brn-3a coopera-tion on Bcl-2 promoter activity and that HIPK2 overexpres-sion reverts the effects of HMGA1 on Bcl-2 transcription.Consequently, we analyzed the biological effects of theseobservations on cell death. To this aim, ND7 cells weretransfected with a vector encoding Brn-3a alone or incombination with HMGA1, and incubated with H2O2 to ac-tivate the endogenous p53. As expected, the expression ofboth p53 and Brn-3a induced an increased apoptotic rate(Fig. 4B and C) likely due to repression of Bcl-2 promoteractivity (16). HMGA1 overexpression impaired the cooper-ation between Brn-3a and p53, and an apoptotic rate com-parable to that exerted by p53 alone was observed. Inaddition, HIPK2 overexpression inhibited the antiapoptoticactivity of HMGA1 and induced a percentage of cell deathcomparable to that promoted by p53-Brn-3a cooperation(Fig. 4B and C).

HMGA1 exerts its antiapoptotic function by promotingHIPK2 cytoplasmic delocalization in ND7 cellsWe have previously shown that HMGA1 inhibits p53-

induced apoptosis by acting on HIPK2 delocalization fromthe nucleus to the cytoplasm (15). Therefore, we askedwhether this mechanism also occurs following HMGA1 over-expression in ND7 cells. To this aim, we transfected ND7 cellswith a vector encoding EGFP-HIPK2 protein to follow its lo-calization by immunofluorescence. We observed a nuclear lo-calization of HIPK2 in EGFP-HIPK2 transfected cells.Conversely, when HMGA1 was overexpressed, we observeda cytoplasmic localization of HMGA1 that was strongly asso-ciated with cytoplasmic relocalization of EGFP-HIPK2 (Fig.5A). HIPK2 delocalization was confirmed by Western blottinganalysis on nuclear/cytoplasmic cellular extracts. Indeed, asshown in Fig. 5B, the amount of cytoplasmic HIPK2 was

mber 1

3, 20

18

Cancer Res; 70(13) July 1, 2010 5383




Esposito et al.

5384


3


higher in HMGA1-overexpressing ND7 cells compared withcontrol cells. Sp1 and γ-tubulin were used as markers of nu-clear/cytoplasmic separation as well as loading controls (Fig.5B). HIPK2 and HMGA1 protein levels from ND7 total cellextracts are shown in Fig. 5C.




HMGA1 displaces HIPK2 andBrn-3a fromBcl-2promoterWe next investigated whether HMGA1 is part of the com-

plex, including p53, Brn-3a, and HIPK2, that forms at the p53binding sites of the Bcl-2 promoter in vivo by a combinationof ChIP and re-ChIP analyses. ND7 cells were transfected

Retrac

ted D

ecem

ber 1

h. r 17, 2018. © 2010 American h. 16, 2019. © 2010 American A

Figure 3. HMGA1 interferes withthe apoptotic activity of p53. A,p53-null SAOS-2 cells weretransfected with the p53 vector incombination or not with theHMGA1b vector, and analyzed bytrypan blue exclusion test.Expression of the indicatedproteins was analyzed by Westernblotting. The result of one ofthree experiments performed isreported. B, wtp53-carrying ND7cells were transfected withHMGA1b and then incubated ornot with H2O2. The percentage ofdead cells and the protein levelswere analyzed as in A. C, ND7 cellswere treated as in B and analyzedby TUNEL assay. The percentageof TUNEL-positive cells fromone out of four representativeexperiments is reported. Theexpression of the indicatedproteins was analyzed by Westernblotting. HMGA1 interacts withBrn-3a in vivo. D, ND7 cells weretransfected with HA-HMGA1band Brn-3a vectors. Cellularextracts were prepared, andequal amounts of proteins wereimmunoprecipitated withanti–Brn-3a (D) or anti-HA(E) antibodies, and theimmunocomplexes were analyzedby Western blotting using thereciprocal antibodies. F,cellular extracts used forimmunoprecipitation experimentswere analyzed by Western blotting.

, 201

8

Cancer Research

Association for Cancerssociation for Cancer





cancerreDownloaded from


Retrac

t

cancerreDownloaded from


ed D

ecem

ber 1

3, 20

18
Figure 4. HMGA1, p53, Brn-3a,and HIPK2 coexpression regulatesBcl-2 gene transcription andapoptosis. A, the pCMV-p53vector was transfected alone orwith the indicated plasmids inSAOS-2 cells. All transfectionswere performed in duplicate; dataare mean ± SD of five independentexperiments. Empty vectors wereused as a control. Western blotanalyses of p53, HMGA1,Brn-3a, and FLAG-HIPK2protein expression from onerepresentative experiment(bottom). B, ND7 cells weretransfected with HMGA1b, Brn-3a,HIPK2 vectors, or control vectorsand then incubated or not withH2O2. Cells were collected andanalyzed by trypan blue exclusiontest. The results of one indicativeexperiment are reported. C, thesame cells reported in B wereanalyzed by TUNEL assay. Onerepresentative experiment isreported. D, expression of theindicated proteins was analyzed byWestern blot. Expression ofγ-tubulin shows equal loading ofproteins.
Cancer Res; 70(13) July 1, 2010 5385

Research. on December 17, 2018. © 2010 American Association for Cancers.aacrjournals.org

Research. on February 16, 2019. © 2010 American Association for Cancers.aacrjournals.org



Esposito et al.

5386


with vectors encoding p53 and Brn-3a plus HMGA1 aloneor in combination with HIPK2 (Fig. 6A). The crosslinked ge-nomic DNAs were immunoprecipitated in two rounds withtwo specific antibodies directed against HMGA1, p53, Brn-3a, and HIPK2 (Fig. 6B). Re-ChIPped DNA was analyzed byPCR using Bcl-2 promoter–specific primers that encompassthe p53 binding sites. As shown in Fig. 6B, HMGA1 is acomponent of the promoter-bound multimeric complexcontaining p53, Brn-3a, and HIPK2. The analysis of the sub-complexes following HMGA1 overexpression revealed themechanisms whereby HMGA1 could abolish Bcl-2 repres-sion exerted by p53 and Brn-3a. Indeed, HMGA1 overex-pression leads to a decrease in the amount of HIPK2 onBcl-2 promoter likely following HIPK2 delocalization fromthe nucleus to the cytoplasm. Moreover, the binding ofBrn-3a to its DNA consensus was strongly reduced. Thesetwo effects were reverted by HIPK2 overexpression thatreestablishes its nuclear localization, impaired by HMGA1,and its biological functions, i.e., the activation of p53 bySer46 phosphorylation (13) and the increase of Brn-3a bind-ing activity to its DNA consensus sequences (28). Therefore,these results show that HMGA1 impairs p53 proapoptoticactivity by a significant reduction of HIPK2 and Brn-3abinding to the Bcl-2 promoter.




Suppression of the p53-mediated transcription of Bcl-2by HMGA1 may have a role in human carcinogenesisTo verify whether HMGA1 also has a role in regulating Bcl-

2 promoter activity in human cancer, we analyzed, by immu-nohistochemistry, Bcl-2 and HMGA1 protein expression in apanel of human breast carcinoma samples carrying a wild-type p53. As shown in Fig. 6C, 14 of 52 (27%) breast carcino-mas analyzed carried a wild-type p53 and, interestingly, therewas an increased expression of both HMGA1 and Bcl-2 genesin 11 of these 14 samples (78%). Therefore, in these cases, theapoptotic activity of p53 is likely impaired by HMGA1, whichblocks its negative effect on Bcl-2 expression, thereby pro-moting the Bcl-2 antiapoptotic activity.

Discussion

Defective apoptosis represents a major causal factor in thedevelopment and progression of cancer. Moreover, the abilityof tumor cells to evade the engagement of apoptosis couldplay a significant role in their resistance to conventionaltherapeutic regimens.Recently, we reported that HMGA1, the aberrant expres-

sion of which is implicated in the process of carcinogenesis,binds to the p53 oncosuppressor and inhibits its apoptoticr 1

3, 20

18

e

Retrac

ted D

ecem

b

h. r 17, 2018. © 2010 American h. 16, 2019. © 2010 American A

Figure 5. HMGA1 proteinspromote HIPK2 cytoplasmicrelocalization in ND7 cells. A,EGFP-HIPK2 subcellularlocalization in ND7 cellstransfected with the indicatedvectors. Nuclei were stained withHoechst. B, nuclear (N) andcytoplasmic (C) extracts wereanalyzed by Western blotting forthe indicated proteins from ND7cells overexpressing or notHMGA1. Sp1 and γ-tubulinwere used as markers ofnuclear/cytoplasmic separation aswell as loading controls. C, totalcell extracts, from the same cellsas in B, were analyzed by Westernblotting for endogenous HIPK2and exogenous HMGA1protein expression. γ-Tubulin wasused to equalize protein loading.

Cancer Research

Association for Cancerssociation for Cancer





activity. This inhibition is associated with increased tran-scription of the p53 inhibitor Mdm2 and repression of thep53 effectors Bax and p21waf1 (12). We have also shown thatHMGA1 represses p53 apoptotic activity by promoting thecytoplasmic relocalization of the p53 proapoptotic activatorHIPK2 (15). Here, we report that the HMGA1 antiapoptoticeffect also occurs by deregulation of Bcl-2 transcriptionfollowing a mechanism quite similar to those previously de-scribed (15). First, we have found a strong positive correla-tion between HMGA1 and Bcl-2 protein expression inseveral cell systems. Then, we investigated the molecularmechanisms which could account for this correlation. Wehave found that HMGA1 proteins are a component of amultiprotein complex including p53, HIPK2, and Brn-3a,which modulates Bcl-2 promoter activity. Moreover, wehave shown that HMGA1 has a critical role in this complex.Indeed, HMGA1 overexpression impaired the transcriptionalrepression exerted by Brn-3a and p53 on the Bcl-2 promoter.Consistently, the transcriptional regulation of Bcl-2 supportsthe antiapoptotic effect of HMGA1 observed in the TUNELassays. Finally, we have shown, using ChIP and re-ChIP assays,that HMGA1 overexpression drastically reduces the binding ofHIPK2 and Brn-3a to the Bcl-2 promoter impairing p53 proa-poptotic activity. This conclusion is supported by the rever-

Retrac

ted D

e




sion of the HMGA1-induced effect on the Bcl-2 promoteractivity by the overexpression of HIPK2.Therefore, HMGA1overexpression counteracts p53-mediated

downregulation of Bcl-2 transcription that results in a drasticinhibition of p53 apoptotic activity, which then contributes tothe process of carcinogenesis. This effect has been previouslydescribed by us for other p53 target genes such as Bax andp21waf1 (12).On the basis of the data presented, it is reasonable to hy-

pothesize that HMGA1 overexpression contributes to cancerprogression through this novel mechanism, impairing p53activity notwithstanding the absence of p53 gene mutationsand/or deletions. This hypothesis seems validated by thefinding of a significant correlation between HMGA1 and Bcl-2overexpression in 11 out of 14 breast carcinoma cases harbor-ing a p53 wild-type gene, indicating that in these tumors, thereis an inactivation of the p53 apoptotic function mediated byHMGA1-dependent regulation of Bcl-2 promoter activity.In conclusion, taken together, all the data reported here allow

us to indicate a new mechanism by which HMGA1 overexpres-sion, displacing HIPK2 and Brn-3a from the Bcl-2 promoter, im-pairs the apoptotic activity exerted by p53 on Bcl-2 expression,contributing to the progression step of carcinogenesis. Thismechanism could also have a role in the resistance ofmalignant

Figure 6. HMGA1 displaces HIPK2 and Brn-3a from Bcl-2 promoter. A, lysates from ND7 cells transiently transfected as indicated were subjected toWestern blot analysis to verify protein expression levels. γ-Tubulin expression served as a control of equal protein loading. B, lysates from cells transfectedwith plasmids as in A were subjected to ChIP (right) and re-ChIP (top) using specific antibodies. Immunoprecipitates from each sample were analyzedby semiquantitative PCR using primers that cover a region of murine Bcl-2 promoter. A sample representing the total input chromatin (Input) was includedin the PCRs as a control. HMGA1 correlates with Bcl-2 expression in p53 wild-type human breast carcinomas. C, graphic distribution of 52 bioptictissues of human breast carcinomas analyzed for p53, HMGA1, and Bcl-2 protein expression by immunohistochemistry.ce

mber 1

3, 20

18

Cancer Res; 70(13) July 1, 2010 5387




Esposito et al.

5388


cells to chemotherapy and radiotherapy treatments because itis often dependent on Bcl-2 overexpression (24).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We are grateful to Professor David Latchman for the gift of the plasmidspLTR-Brn3a and pGL2-Bcl2 promoter.



Re

e


Grant Support

Associazione Italiana Ricerca sul Cancro and Ministero dell'Università edella Ricerca Scientifica e Tecnologica. This work was supported by the Na-ples Oncogenomic Center. We thank the Associazione Partenopea per le Ri-cerche Oncologiche for its support. F. Esposito is a recipient of a FIRCfellowship.

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 11/17/2009; revised 03/22/2010; accepted 04/14/2010; publishedOnlineFirst 06/08/2010.
8
References 1
1. Johnson KR, Lehn DA, Reeves R. Alternative processing of mRNAs
encoding mammalian chromosomal high-mobility-group proteinsHMG-I and HMG-Y. Mol Cell Biol 1989;9:2114–23.

2. Grosschedl R, Giese K, Pagel J. HMG domain proteins: architecturalelements in the assembly of nucleoprotein structures. Trends Genet1994;10:94–100.

3. Thanos D, Maniatis T. The high mobility group protein HMG I(Y) isrequired for NF-κB-dependent virus induction of the human IFN-βgene. Cell 1992;71:777–89.

4. Zhou X, Benson KF, Ashar HR, Chada K. Mutation responsible forthe mouse pygmy phenotype in the developmentally regulated factorHMGI-C. Nature 1995;376:771–4.

5. Chiappetta G, Avantaggiato V, Visconti R, et al. High level expressionof the HMGA1 gene during embryonic development. Oncogene1996;13:2439–46.

6. Fusco A, Fedele M. Roles of the HMGA proteins in cancer. Nat RevCancer 2007;7:899–910.

7. Berlingieri MT, Manfioletti G, Santoro M, et al. Inhibition of HMGI-Cprotein synthesis suppresses retrovirally induced neoplastic transfor-mation of rat thyroid cells. Mol Cell Biol 1995;15:1545–53.

8. Berlingieri MT, Pierantoni GM, Giancotti V, Santoro M, Fusco A. Thy-roid cell transformation requires the expression of the HMG1 pro-teins. Oncogene 2002;21:2971–80.

9. Scala S, Portella G, Fedele M, Chiappetta G, Fusco A. Adenovirus-mediated suppression of the HMGI(Y) protein synthesis as a poten-tial therapy of human malignant neoplasias. Proc Natl Acad Sci USA2000;97:4256–61.

10. Wood LJ, Maher JF, Bunton TE, Resar LM. The oncogenic propertiesof the HMG-I gene family. Cancer Res 2000;60:4256–61.

11. Reeves R, Edberg DD, Li Y. Architectural transcription factor HMGI(Y) promotes tumor progression and mesenchymal transition of hu-man epithelial cells. Mol Cell Biol 2001;21:575–94.

12. Pierantoni GM, Rinaldo C, Esposito F, Mottolese M, Soddu S, FuscoA. High mobility group A1 (HMGA1) proteins interact with p53 andinhibit its apoptotic activity. Cell Death Differ 2006;13:1554–63.

13. D'Orazi G, Cecchinelli B, Bruno T, et al. Homeodomain-interactingprotein kinase-2 phosphorylates p53 at Ser 46 and mediates apopto-sis. Nat Cell Biol 2002;4:11–9.

14. Hofmann TG, Möller A, Sirma H, et al. Regulation of p53 activity byits interaction with homeodomain-interacting protein kinase-2. NatCell Biol 2002;4:1–10.tra

cted D

ec

15. Pierantoni GM, Rinaldo C, Mottolese M, et al. High-mobility group A1inhibits p53 by cytoplasmic relocalization of its proapoptotic activa-tor HIPK2. J Clin Invest 2007;117:693–702.

16. Budhram-Mahadeo V, Morris PJ, Smith MD, Midgley CA, Boxer LM,Latchman DS. p53 suppresses the activation of the Bcl-2 promoterby the Brn-3a POU family transcription factor. J Biol Chem 1999;274:15237–44.

17. Vousden KH, Lu X. Live or let die: the cell's response to p53 [review].Nat Rev Cancer 2002;2:594–604.

18. Cory S, Adams JM. The Bcl2 family: regulators of the cellular life-or-death switch [review]. Nat Rev Cancer 2002;2:647–56.

19. Monni O, Franssila K, Joensuu H, Knuutila S. BCL2 overexpressionin diffuse large B-cell lymphoma [review]. Leuk Lymphoma 1999;34:45–52.

20. Wacheck V, Losert D, Günsberg P, et al. Small interfering RNA tar-geting bcl-2 sensitizes malignant melanoma. Oligonucleotides 2003;13:393–400.

21. Reed JC, Meister L, Tanaka S, et al. Differential expression of bcl2protooncogene in neuroblastoma and other human tumor cell linesof neural origin. Cancer Res 1991;51:6529–38.

22. Nathan B, Gusterson B, Jadayel D, et al, For the International (Lud-wig) Breast Cancer Study Group. Expression of BCL-2 in primarybreast cancer and its correlation with tumour phenotype. Ann Oncol1994;5:409–14.

23. Catz SD, Johnson JL. BCL-2 in prostate cancer: a minireview [re-view]. Apoptosis 2003;8:29–37.

24. Kassim SK, Ali HS, Sallam MM, et al. Increased bcl-2 expression isassociated with primary resistance to chemotherapy in human epi-thelial ovarian cancer. Clin Biochem 1999;32:333–8.

25. Ben-Ezra JM, Kornstein MJ, Grimes MM, Krystal G. Small cell carci-nomas of the lung express the Bcl-2 protein. Am J Pathol 1994;145:1036–40.

26. Ciardiello F, Tortora G. Inhibition of bcl-2 as cancer therapy. Ann On-col 2002;13:501–2.

27. Chiarugi V, Del Rosso M, Magnelli L. Brn-3a, a neuronal tran-scription factor of the POU gene family: indications for its involve-ment in cancer and angiogenesis [review]. Mol Biotechnol 2002;22:123–7.

28. Wiggins AK, Wei G, Doxakis E, et al. Interaction of Brn3a and HIPK2mediates transcriptional repression of sensory neuron survival. J CellBiol 2004;167:257–67.

mber 1

3, 20

Cancer Research




Retraction

Retraction: High-Mobility Group A1 ProteinsRegulate p53-Mediated Transcription ofBcl-2 GeneFrancesco Esposito, Mara Tornincasa, Paolo Chieffi,Ivana De Martino, Giovanna Maria Pierantoni, andAlfredo Fusco

This article (1) has been retracted at the request of the editors. The editors were madeaware of concerns regarding potential manipulation of data in the article. An internalreview by the editors determined that multiple g-tubulin Western blot bands appearto be duplicated in Figs. 1A, 1C, and 4A. Similarly, multiple Bcl-2Western blot bandsappear to be duplicated in Fig. 2C and 2B. Lastly, the sameWestern blot bands appearto have been used to represent Bcl-2 and a-Brn-3a CHIP in Fig. 6B. The authors werenot able to provide the original data to dispel these concerns.

A copy of this Retraction Notice was sent to the last known e-mail addresses forall 6 authors. One author (A. Fusco) did not agree to the retraction; one author(G.M. Pierantoni) takes no position; the four remaining authors (F. Esposito,M. Tornincasa, P. Chieffi, and I. De Martino) did not respond.

Reference1. Esposito F, TornincasaM, Chieffi P, DeMartino I, Pierantoni GM, Fusco A. High-mobility group A1

proteins regulate p53-mediated transcription of Bcl-2 gene. Cancer Res 2010;70:5379–88.

Published online December 15, 2018.doi: 10.1158/0008-5472.CAN-18-3454�2018 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 6905

Retrac

ted D

ecem

ber 1

3, 20

18

Research. on February 16, 2019. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-18-3454&domain=pdf&date_stamp=2018-11-30


2010;70:5379-5388. Published OnlineFirst June 8, 2010.Cancer Res Francesco Esposito, Mara Tornincasa, Paolo Chieffi, et al.

GeneBcl-2Transcription of High-Mobility Group A1 Proteins Regulate p53-Mediated

Updated version

10.1158/0008-5472.CAN-09-4199doi:

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/70/13/5379.full#ref-list-1

This article cites 28 articles, 8 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/70/13/5379.full#related-urls

This article has been cited by 6 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/70/13/5379To request permission to re-use all or part of this article, use this link

Research. on December 17, 2018. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


Retrac

ted D

ecem

ber 1

3, 20

18



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-09-4199



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/70/13/5379



Retraction

Retraction: High-Mobility Group A1 ProteinsRegulate p53-Mediated Transcription ofBcl-2 GeneFrancesco Esposito, Mara Tornincasa, Paolo Chieffi,Ivana De Martino, Giovanna Maria Pierantoni, andAlfredo Fusco

This article (1) has been retracted at the request of the editors. The editors were madeaware of concerns regarding potential manipulation of data in the article. An internalreview by the editors determined that multiple g-tubulin Western blot bands appearto be duplicated in Figs. 1A, 1C, and 4A. Similarly, multiple Bcl-2Western blot bandsappear to be duplicated in Fig. 2C and 2B. Lastly, the sameWestern blot bands appearto have been used to represent Bcl-2 and a-Brn-3a CHIP in Fig. 6B. The authors werenot able to provide the original data to dispel these concerns.

A copy of this Retraction Notice was sent to the last known e-mail addresses forall 6 authors. One author (A. Fusco) did not agree to the retraction; one author(G.M. Pierantoni) takes no position; the four remaining authors (F. Esposito,M. Tornincasa, P. Chieffi, and I. De Martino) did not respond.

Reference1. Esposito F, TornincasaM, Chieffi P, DeMartino I, Pierantoni GM, Fusco A. High-mobility group A1

proteins regulate p53-mediated transcription of Bcl-2 gene. Cancer Res 2010;70:5379–88.

Published online December 15, 2018.doi: 10.1158/0008-5472.CAN-18-3454�2018 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 6905

http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-18-3454&domain=pdf&date_stamp=2018-11-30

2010;70:5379-5388. Published OnlineFirst June 8, 2010.Cancer Res Francesco Esposito, Mara Tornincasa, Paolo Chieffi, et al.

GeneBcl-2Transcription of High-Mobility Group A1 Proteins Regulate p53-Mediated

Updated version

10.1158/0008-5472.CAN-09-4199doi:

Access the most recent version of this article at:

Cited articles


This article cites 28 articles, 8 of which you can access for free at:

Citing articles


This article has been cited by 6 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/70/13/5379To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-09-4199



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/70/13/5379


retracted -...

Documents