High Histone Deacetylase 7 (HDAC7) ExpressionIs Significantly Associated with Adenocarcinomas

of the Pancreas

Mehdi Ouaıssi, MD,1,2,3,4 Igor Sielezneff, MD,1,2 Ricardo Silvestre, PhD,4,5

Bernard Sastre, MD,1,2 Jean-Paul Bernard, MD, PhD,2,3

Joelle Simony Lafontaine, MD, PhD,7 Marie Jose Payan, MD, PhD,8 Laetitia Dahan, MD,2,9

Nicolas Pirro, MD,1,2 Jean Francois Seitz, MD, PhD,2,9 Eric Mas, PhD,3

Dominique Lombardo, PhD,3 and Ali Ouaissi, PhD4,5,6

1Service de chirurgie digestive et oncologique, pole d’oncologie et specialite medico-chirurgicales, Assistance Pubique-Hopitaux deMarseille, Hopital Timone, 264 Rue Saint Pierre, Marseille 13000, France2Faculte de Medecine, Aix Marseille Universite, Marseille 13000, France

3INSERM, UMR911, Marseille 13000, France4IRD, UR008, Montpellier 34394, France

5IBMC, Universidade do Porto, Porto 4150-180, Portugal6INSERM, CNRS UMR 5235, Universite Montpellier 2, Montpellier, France

7Service d’Anathomopathologie, CRLC, Hopital Val d’Aurel, Montpellier 34298, France8Service d’anathomopathologie, pole d’oncologie et specialite medico-chirurgicales, Assistance Pubique-Hopitaux de Marseille,

Hopital Timone, Marseille 13000, France9Service d’oncologie digestive, pole d’oncologie et specialite medico-chirurgicales, Assistance Pubique-Hopitaux de Marseille,

Hopital Timone, Marseille 13000, France

Background: Alterations in HDACs gene expression have been reported in a number ofhuman cancers. No information is available concerning the status of HDACs in pancreaticcancer tumors. The aim of the present study was to evaluate the expression levels of membersof class I (HDAC1, 2,, 3), class II (HDAC4, 5, 6, and 7), and class III (SIRT1, 2, 3, 4, 5, and 6)in a set of surgically resected pancreatic tissues.Methods: Total RNA was isolated from 11 pancreatic adenocarcinomas (PA): stage 0

(n = 1), IB (n = 1), IIB (n = 6), III (n = 1), IV (n = 2), one serous cystadenoma (SC), oneintraductal papillary mucinous tumor of the pancreas (IMPN), one complicating chronicpancreatitis (CP), and normal pancreas (NP) obtained during donor liver transplantation.Moreover, six other control pancreatic were included. HDACs gene expression was conductedusing quantitative real-time polymerase chain reaction (qPCR). Protein expression levels wereanalyzed by Western blot and their localization by immunohistochemistry analyses of cancertissues sections.Results: Remarkably, 9 of the 11 PA (approximately 81%) showed significant increase of

HDAC7 mRNA levels. In contrast to PA samples, message for HDAC7 was reduced in CP, SC,and IMPN specimens. TheWestern blot analysis showed increased expression ofHDAC7 proteinin 9 out of 11 PA samples, in agreement with the qPCR data. Most of the PA tissue sectionsexamined showed intense labeling in the cytoplasm when reacted against antibodies to HDAC7.

Published online May 28, 2008.Address correspondence and reprint requests to: Mehdi Ouaıssi,

MD; E-mail: mehdi.ouaissi@mail.ap-hm.fr

Published by Springer Science+Business Media, LLC � 2008 The Society ofSurgical Oncology, Inc.

Annals of Surgical Oncology 15(8):2318–2328

DOI: 10.1245/s10434-008-9940-z

2318

Conclusion: The data showed alteration of HDACs gene expression in pancreatic cancer.Increased expression ofHDAC7 discriminates PA from other pancreatic tumors.Key Words: Pancreatic tumors—HDACs—SIRTs—RT-PCR—qPCR—Gene expression.

Pancreatic cancer is a major health problemworldwide. Despite serious efforts toward the identi-fication of markers which could improve the overallprognosis of patients, the pancreatic adenocarcinomaremains one of the most deadly human cancers.1 Infact, the existing tumor markers do not allow todifferentiate, with sufficient accuracy, benign frommalignant diseases of the pancreas.2 Most pancreaticcancers are still not diagnosed until after they havemetastasized. Depending on the extent of the tumorat the time of diagnosis, the prognosis is often poor,with 5-year survival of less than 5%.3 Thus, theidentification of sensitive and specific tools for pan-creatic cancer diagnosis is needed.It is well known from a large body of data that the

molecular status of each neoplasm is related to itsbiological behavior. The development from prema-lignant pancreatic lesions to invasive metastasizingcancer might be related to a stepwise acquisition ofsequential mutations and/or a differentially regulatedgene expression. Indeed, this possibility is supported,for example, by data showing that significantly moretumors in the shortest-surviving patients hadp16INK4a alterations compared with tumors of thelongest-surviving patients.4 Therefore, the identifica-tion of biomarkers will help to select patients whomight benefit from surgery, which remains the mainoption for the curative treatment of solid cancers, andadjuvant therapeutic approaches.Specific events-driven changes that led to cancer

development and progression are interconnectedcomplex molecular modifications including DNAmethylation, histone acetylation, phosphorylation,ubiquitylation, and ADP ribosylation. These epige-netic mechanisms play key roles in normal life pro-cesses and are aberrantly regulated in several humandiseases such as cancer.5–7 The steady-state level ofacetylation of core histones is controlled by theopposing actions of histone acetyltransferases(HATs) and histone deacetylases (HDACs),8 whoseactivities are correlated with gene activation and generepression or silencing, respectively. Several majorclasses of HDACs have been defined and generallyclassified into class I and class II based on similaritiesto yeast deacetylases RPD3 and HDA1, respectively,and class III with homology to yeast silent informa-tion regulatory protein (SIR2p). Class I includes

HDAC1, 2, 3, and 8; class II includes HDAC4, 5, 6,7, and 9. Given that HDAC6 and HDAC10 carry twocatalytic sites, they may be grouped in a class IIB.HDAC11 shares with class I and II enzymes con-served residues in its catalytic site and is placed in aclass IV.9

Based on their primary structure, the SIR2 familynamed Hst proteins (Homologous of Sir two) orsirtuins could be divided into five classes.10 HumanSIRT1, 2 and 3 belong to class I, SIRT4 is in class II,SIRT5 is in class III, and SIRT6 and SIRT7 are inclass IV sirtuins. SIRT8, recently identified in thyroidcarcinoma cell lines and tissues,11 shared 85% iden-tity in the core sirtuin domain with SIRT7 and thuscould be included in class IV. Evidence accumulatedover the last few years supports the notion that manynonhistone proteins could be targets of HDACs andpoints to the role of these enzymes in diverse cellularprocesses including, cell proliferation, migration, anddeath pathways.Since limited information is available concerning

the status of HDACs in pancreatic cancer, attemptswere made for the first time to explore the level ofexpression of members of histone deacetylaseencoding genes (HDACs and SIRTs) in a set of sur-gically resected pancreatic cancer tissues and anormal pancreas tissue.

PATIENTS AND METHODS

Subject Population

From May 2006 through August 2007, eight pan-creaticoduodenectomy (PD), four splenopancreatec-tomy (SP), and two surgical double derivations wereconducted in the department of Surgery at TimoneHospital (Marseille, France). All patients underwentcontrast-enhanced thoracic and abdominal computedtomography, abdominal ultrasonography, magneticresonance imaging, and blood testing. There was nopreoperative or postoperative adjuvant therapy. Twoendoscopic biliary derivations were performed forangiocholitis symptoms before surgery. Eight PD,one SP, and two surgical biliary derivations wereconducted for pancreatic adenocarcinoma, respec-tively, and were designed PA. These two double

PANCREATIC CANCER AND HISTONE DEACETYLASE 2319

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

derivations were performed for carcinosis that wasnot diagnosed by different preoperative examina-tions. Three SP were performed for serous cystade-noma (SC), one intraductal papillary mucinoustumor of the pancreas (IMPN), and one complicatingchronic pancreatitis (CP). One normal pancreaticbiopsy (NP) was obtained during donor liver trans-plantation. In addition, four samples of control tis-sues taken from the surgical specimens from differentpatients with PA [two in close proximity termednormal adjacent, NA (patients in stage IIB and IB,respectively), two as far away from the tumor aspossible termed normal distant, ND (patients in stageIIB)], and two samples from patients with adeno-carcinomas of biliary duct (BD). Data were pro-spectively collected and a standardized questionnairewas completed at the time of follow-up and of studyassessment.

Surgery

All surgical procedures were performed by twoexperienced pancreatic surgeons. All patients hadsingle-shot intravenous antibiotic prophylaxisadministered at anaesthetic induction.Octreotide sandostatine (Novartis Pharma�, Rueil

Malmaison, France) administration was started dur-ing the operation and continued for 7 days (0.1 mgsubcutaneously three times a day) in an uneventfulpostoperative course. Experienced senior surgeonscarried out all pancreatic head resections. PD wasperformed using a pylorus preserving or Whippleprocedure, and an end-to-end pancreaticojejunosto-my (PJA) was constructed with a single-layer anas-tomosis of interrupted 5-0 PDS (Ethicon suture�,

Issy les Moulinaux, France) absorbable sutures.Choice of surgical technique was decided per-opera-tively bound to surgeon’s decision. Lymphadenec-tomy was carried out along the hepatoduodenalligament and the common hepatic artery. Pancreaticand biliary ducts were never drained by transanas-tomotic catheter. Two soft drains were routinelyplaced near the pancreaticojejunal anastomose.Operative time was recorded. In the absence of afistula, drains were removed after 7 days.

Tissue Samples

All surgical specimens (11 PA, 1 CP, 1 IMNP, 1SC, 1 NP, 2NA, 2 ND, and 2 BD) were reviewed by asenior pathologist. Clinical and pathologic staging(Table 1) were reassessed according to AmericanJoint Committee on Cancer tumor–node–metastasis(TNM) staging of pancreatic cancer.12 The tumorshad been snap-frozen in liquid nitrogen during 15 s,and immediately stored at �80�C. Microscopicexamination was performed on all frozen tissuesamples before RNA isolation. Routine hematoxylinand eosin coloration was used. Features useful inestablishing a diagnosis of adenocarcinoma on frozensections were: 1–haphazard growth pattern, 2–agland immediately adjacent to a muscular vesselwithout intervening stroma or acini, 3–perineural orvascular invasion, 4–incomplete lumens of infiltratingadenocarcinoma, 5–area of one nucleus is four ormore times larger than the area of another nucleuswithin a single gland, 6–irregular nucleoli, 7–necroticglandular debris, or 8–atypical mitoses. Informedconsent for the use of the specimens was obtainedfrom all patients.

TABLE 1. TNM clinical classification and staging of tumours

Patient T* N* M* Differentiation Lymphatic embolus Perinervous invasion R* Stage Other pancreatic tumor

1 4 1 0 Moderate No Yes 1 III2 3 1 0 Moderate Yes Yes 0 IIB3 1 0 0 Well differentiated No No 0 04 3 1 0 Moderate No No 1 IIB5 2 0 0 Well differentiated No No 0 IB6 3 1 0 Moderate No Yes 0 IIB7 2 1 0 Well differentiated Yes Yes 0 IIB8 3 1 0 Well differentiated Yes Yes 0 IIB9 Any T Any N 1 – – – – IV10 Any T Any N 1 – – – – IV11 3 1 0 Well differentiated Yes Yes 1 IIB12 – – – – – – – – CP13 – – – – – – – – SC14 – – – – – – – – IPMN

* The TNM staging system followed the American Joint Committee on Cancer (AJCC). The staging of carcinoma of the pancreas is basedon the size and extent of the primary neoplasm (T), the presence or absence of regional lymph node metastases (N), and the presence orabsence of metastatic disease (M). R0 designates complete resection with no residual microscopic or macroscopic tumour; R1 designatesgrossly negative but microscopically positive margins of resection.

M. OUAISSI ET AL.2320

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

RNA Isolation and Reverse Transcription

Tissues (approximately 50 mg) were disrupted in1 ml QIAzol reagent (Qiagen, Courtaboeuf, France)using an adapted sized vessel for disruption andhomogenization with a Tissue Ruptor. Total RNAwas isolated using the RNeasy mini kit (Qiagen)according to the manufacturer’s instructions. RNAsamples were treated with DNase Set (Qiagen kit) forconvenient on column DNase digestion during RNApurification. RNA integrity was checked on agarosegel and concentration was determined by OD260/OD280 nm absorption[ 1.95. Reverse transcription(RT) reactions were performed on 1 lg of total RNAusing random hexamers and the superscipt III reversetranscriptase (Invitrogen, Cergy-Pontoise, France)according to the manufacturer’s instructions. Theabsence of contaminant DNA in the samples wasassessed by including for each cDNA a controllacking the reverse transcriptase.

PCRPrimers were designed to amplify an approximately

200 bp fragment in the coding sequence of 11 genesbelonging to the HDAC/SIR families. Nucleotidesequences were aligned using the Multalin software13

in order to design gene-specific primers. The 28SRNA gene was chosen as control. GenBank accessionnumbers and primers are summarized in Table 2.PCR was performed on 2 ll of template cDNA

(500 ng/ml) in a 20 ll final volume containing0.5 lM of each forward and reverse primer and thePCR Master mix (Promega, Charbonnieres, France)according to the manufacturer’s instructions. Thenumber of cycles was 26–30 according to the abun-dance of the transcript, in order to stay in the expo-nential amplification step. Amplification productswere analyzed on agarose gels stained with ethidiumbromide. Some specific PCR products were purifiedfrom agarose gel using the Minelute gel extractionkit (Qiagen) and sequenced by Genome Express(Meylan, France).

Quantitative Real-Time PCR (qPCR)qPCR reactions were run in triplicate on two inde-

pendent cDNA preparations from pancreatic tissuesusing the LightCycler FastStart DNA Master SYBRGreen I kit and the LightCycler� real-time PCRinstrument (Roche Molecular Biochemicals, Mann-heim, Germany). The reaction was carried out in 1·Fast-Start SYBR Green reaction mix (including TaqDNA polymerase, dNTPs and 2 mM MgCl2) con-taining 0.5 lMof each forward and reverse primer and3 ll of template cDNA, using the following cyclingconditions: 95�C for 5 min followed by 40 cycles of95�C for 10 s, 58�C for 10 s, 72�C for 15 sec. Specificityof the PCR products was assessed by performing amelting curve analysis of the amplification productsaccording to the manufacturer instructions. The

TABLE 2. Specific primers used in RT-PCR

Transcript Strand Sequence (50-30) Reference

HDAC1 F GTC CAG ATA ACA TGT CGG AGT ACA GC gi:13128859HDAC1 R CGA TGT CCG TCT GCT GCT TAT TAA GHDAC2 F CCT CAT AGA ATC CGC ATG ACC CAT AAC gi:4557640HDAC2 R AGA CAT GTT ATC TGG TCT TAT TGA CCG TAGHDAC3 F CAA GCC ATA CCA GGC CTC CCA GC gi:13128861HDAC3 R GAG ATG CGC CTG TGT AAC GCG AGHDAC4 F CCT GCA CAG ACA CGG GGA AGG TG gi:13259519HDAC4 R GAG CTG CTC TTC AGA CAG CAA GCHDAC6 F TGCT GGG CCA GAC CAC CTC AGA G gi:13128863HDAC6 R TGA CGT AGT CTG GTC CAG TGT GGC TCHDAC7 F CGG GCC AGG TGG TGG ACG ATG G gi:13259523HDAC7 R GCC AGA GGA AGC AGC ACA GTG TCSIRT1 F TGCGGGAATCCAAAGGATAATTCAGTGTC gi:7555470SIRT1 R CTT CAT CTT TGT CAT ACT TCA TGG CTC TAT GSIRT2 F CAG AAC ATA GAT ACC CTG GAG CGA A gi:13775601SIRT2 R AAG GTC CTC CAG CTC CTT CTT CSIRT3 F GTC GGG CAT CCC TGC CTC AAA GC gi:13259626SIRT3 R GGA ACC CTG TCT GCC ATC ACG TCA GSIRT5 F CGAGT CGT GGT CAT CAC CCA GAA CAT C gi:5225325SIRT5 R AC TCT TGT AAT TCT CAG CCA CAA CTC CACSIRT6 F GAG GAG CTG GAG CGG AAG GTG TG gi:20381385SIRT6 R GGC CAG ACC TCG CTC CTC CAT GG28 S F AGCCGATCCATCATCCGCAATG gi:47683010

R CAGCCAAGCTCAGCGCAAC

PANCREATIC CANCER AND HISTONE DEACETYLASE 2321

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

crossing point (Ct), defined as the point at which thefluorescence rises appreciably above the backgroundfluorescence, was determined for each transcript. Real-time PCR efficiencies calculated from the given slopesin Lightcycler software showed efficiencies of the tar-gets and reference 28S RNA gene to be very similar.Therefore, the 2�DDCt method was used to analyze therelative gene expression.14 The average Ct was calcu-lated for both the target and the 28S RNA gene andDCt = Ct,target � C t,28S was determined. Normalpancreas (NP) was used as the calibrator [for calcula-tion of DDCt = (Ct,target � Ct,28S) � (Ct,target NP �Ct,28S NP)].

14 For the normal pancreas, DDCt is zeroand 20 equals one, so that the fold change in geneexpression relative to theNP equals one. Evaluation of2�DDCt indicates the fold change in gene expressionrelative to the NP.

Western Blot Analysis

Tissue lysates were prepared by disruption andhomogenization with a Tissue Ruptor in a buffercontaining 50 mM HEPES, pH 7.0, 250 mM NaCl,0.15% Nonidet P-40, 1 mM DTT, 1 mM EDTA,0.01% PMSF, 1 mM aprotinin, and 1 mM leupeptinfor 5 min on ice. Cell membranes were further dis-rupted by sonication on ice at the highest output(twice, 30 s each; cells were kept on ice for 1 minbetween each pulse). The soluble phase was recoveredby centrifugation (12,000 g for 20 min, 4�C) andprotein concentration was determined using a Brad-ford protein assay kit (Bio-Rad Laboratories proteinassay, Hercules, CA). Equal amounts (50 lg) wereloaded on SDS-PAGE for separation and electro-blotted into nitrocellulose membranes which werethen saturated in PBS-5% milk (PBS-M) for 1 h atroom temperature (RT) and incubated at RT witheither of the following rabbit antibodies to humanHDAC/SIR proteins: HDAC1, HDAC4, HDAC7,SIRT1, SIRT2 (Santa Cruz Biotechnology, Inc.,Bergheimer Str., Heidelberg, Germany) diluted 1:500in PBS-M for 2 h at RT. After three washes in PBS–Tween 0.05% (PBS-T) and a final wash in PBS, themembranes were then incubated for 1 h at RT with agoat anti-rabbit peroxidase-conjugated IgG (Sigma-Aldrich, Saint-Louis, MO) followed by three washesin PBS-T and a final wash in PBS. The immunore-active bands were detected by ECL Plus WesternBlotting Detection System (Amersham Biosciences,Saclay, France) and ECL Hyperfilms. Equal proteinloading was guaranteed by probing the blots with arabbit antibody against b-tubulin diluted 1:500 inPBS-M (Santa Cruz Biotechnology, Inc.). All the

tumor tissues were treated as above except for onepatient (no. 9, PA stage IV, see Table 2) for whom nosufficient tumor material was available for proteinextraction.

Immunohistochemistry

Tissue samples were embedded in Tissue-Tek(Miles, Inc., Torrance, CA) and shock-frozen in li-quid nitrogen. Serial cross-sections were performedusing a cryotome and mounted on preheated (37�C)HistoBond� slides (Marienfeld, Hoerdt, France).Before specific staining, all cryosections were air-dried for 30 min at 37�C. Incubation with primaryand secondary antibodies followed the technicalinstructions of rabbit ABC staining system (sc-2018,Santa Cruz Biotechnology, Inc.). Briefly, tissue sec-tions were fixed with methanol for 5 min at RT toquench endogenous peroxidase activity and weresuccessively incubated in (a) blocking normal goatserum diluted 1.5% (v/v) in PBS (NGS-PBS) for 1 h;(b) with either of the following rabbit antibodies tohuman HDAC/SIR proteins: HDAC1, HDAC7, andSIRT2 (Santa Cruz Biotechnology, Inc.), diluted at2 lg/ml in 1.5% (v/v) of NGS-PBS for 30 min fol-lowed by three washes with PBS for 5 min each; (c)biotinylated secondary goat anti-rabbit immuno-globulins diluted at 1 lg/ml in 1.5% (v/v) NGS-PBSfollowed by three washes with PBS for 5 min each;(d) a mixture of avidin and biotinylated HRP dilutedwith PBS for 30 min followed by three washes in PBSfor 5 min each; (e) peroxidase substrate and 3,30-di-aminobenzidine chromogen for 1 min. Sections werethen washed in deionized water for 5 min, counter-stained in hematoxylin for 10 s, and immediatelywashed with several changes of deionized water. Twodrops of permanent mounting medium were added tothe sections which were covered with a glass cover-slip. Stained sections were visualized using a LeritzDMRB microscope equipped with a Leica DFC 300FX camera and the images were recorded with aLeica IM50 software.

RESULTS

Expression of HDACs Proteins

Given that a limited number of specific antibodiesagainst HDAC family members could be obtained,we conducted Western bolt analysis using commer-cially available antibodies against members of class I(HDAC1), class II (HDAC4, 7) and class III (SIRT1,

M. OUAISSI ET AL.2322

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

2) proteins reacted with tissue lysates from resectedtumors prepared as described in the ‘‘Materials andMethods’’ section. The results shown in Fig. 1A

revealed the presence of all the HDAC isoformsexamined in PA as well as in CP, SC, IMPN, and NPsamples. However, while an internal check consistingof a polyclonal antibody directed against b-tubulinshowed comparable signal intensities for tissue b-tubulin polypeptide in all samples examined, dem-onstrating therefore, that comparable quantities oftotal proteins were analyzed, significant variation inHDACs expression levels could be seen among indi-vidual tissues specimens.Moreover, complex band patterns with the presence

of multiple polypeptides could be seen in the case ofHDAC1, HDAC4, SIRT1, and SIRT2. Similarobservations have been reported by other investiga-tors.15,16 This might possibly result from post-trans-lational modifications of these molecules. Therefore,quantification of the changes in protein expressionlevels was limited. Nevertheless, examination ofHDAC7 reactive bands showed strong positive inten-sities in most of the PA specimens compared to NP.Interestingly, theHDAC7 isoform reactive bands werebarely visible in the case of CP, SC, and IMPN tissuesamples. Quantification of the bands by image analysisallowed to estimate HDAC7 protein increase byapproximately 1.4- to 3.2-fold in PA samples (Fig. 1B)whereas the ratio of tubulin expression was approxi-mately 0.7–1.2when compared toNP.Together, thesesdata suggest that, although members of class I, II, andIII HDACs are present in pancreatic tumors, theydiffer in their level of expression.

qPCR Analysis of HDACs Expression

qPCR provides a sensitive and reproducible tool tomeasure relative gene expression by comparing oneor more genes of interest to a known internal controlfor normalization. Given the changes in protein lev-els, we performed qPCR analysis of HDACs genes ontissues samples to determine whether the increaseexpression of HDAC7 is observable in the mRNAlevel compared to normal pancreas. Primers used toamplify 28S RNA allowed to normalize the amountof cDNA used for the quantitative measurements ofgene transcripts and to confirm that the mRNAsisolated from tissues were intact (Fig. 2A). As shownin Table 3, the Ct values for all the transcripts ana-lyzed when using control tissue samples: NA1-2,ND1-2 and BD1-2, were comparable to thoseobtained with NP sample. Therefore, it was reason-able to use the NP as the calibrator to determine therelative gene expression levels by the 2�DDCt method.qPCR analysis showed the expression of various

sirtuin mRNAs in tissues from PA, CP, SC, and

FIG. 1. (A) The protein levels of distinct HDAC isoforms wereassessed by Western blot analysis on the same tissue extracts. Blotswere probed with antibodies against members of class I (HDAC1),class II (HDAC4 and 7), class III (SIRT1 and 2), and b-tubulin asloading control. Normal pancreas (NP), pancreatic adenocarcino-mas cancer specimens (PA): 1 (stage III), 2 (IIB), 3 (0), 4 (IIB), 5(IB), 6 (IIB), 7 (IIB), 8 (IIB), 10 (IV), 11 (IIB); CP (chronic pan-creatitis), SC (serous cystadenoma), and IMPN (intraductal pap-illary mucinous tumor of the pancreas). For one patient (no. 9 PAstage IV, see Table 2), insufficient tumor material was available forprotein extraction. (B) Quantification of Western blots corre-sponding to HDAC7 and b-tubulin shown in (A) for the PA, CP,SC, and IMPN samples. The intensity of the protein bands wasanalyzed by densitometry using the Bioprofil BIO1D Windowsapplication V99-04 software. The relative protein levels representthe ratio of sample to normal pancreas.

PANCREATIC CANCER AND HISTONE DEACETYLASE 2323

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

IMPN (Fig. 2B). However, slight variations couldbe observed among tissues. In the case of PAsamples, increase of SIRT1 (1/11), SIRT2 (6/11),SIRT3 (4/11), and SIRT6 (4/11) transcripts couldbe seen. Furthermore, increased levels of SIRT5transcripts could be evidenced in most of thesamples analyzed [10 of the 11 PA (90%), CP, SC,and IMPN samples]. Moreover, when the data wereexamined taking into account the staging of PAtumors, all the patients in stage 0, IB, IIB, and IVhad increased levels of SIRT5 mRNA (Fig. 2B). Incontrast, remarkably, the patient in stage IIIshowed downregulation of all SIRT gene tran-scripts.

The expression of class I and II members of HDACsshowed that all the samples from PA, CP, SC, andIMPN had decreased levels of HDAC1, 2, 3, and 4transcripts. Three of the 11 PA had upregulation ofHDAC6 transcripts (more than threefold increase).Themost remarkable observation was that 9 out of the11 PA (81%) showed significant increase (1.2 to morethan 3 times) of HDAC7 mRNA levels (Fig. 3). Incontrast to PA samples, message for HDAC7 wasreduced in CP, SC, and IMPN specimens.Moreover, when the data were examined taking

into account the staging of PA tumors, the patients instages 0, IB, and IV, and four of the six in stage IIB,had increased levels of HDAC7 mRNA (Fig. 3).

FIG. 2. (A) RT-PCR of totalRNA amplified with primers for28S RNA gene. Normal pan-creas (NP), pancreatic adeno-carcinomas cancer specimens(PA): 1 (stage III), 2 (IIB), 3 (0),4 (IIB), 5 (IB), 6 (IIB), 7 (IIB), 8(IIB), 10 (IV), 11 (IIB); CP(chronic pancreatitis), SC(serous cystadenoma), andIMPN (intraductal papillarymucinous tumor of the pan-creas). (B) Sirtuin gene expres-sion in pancreatic tumor tissuesand in relation to staging ofcarcinomas. Relative mRNASIRT1, 2, 3, 5, and -6 tran-scripts were determined asdescribed in the ‘‘Materials andMethods’’ section.

TABLE 3. Quantitative real-time PCR. Mean Ct of target gene (HDAC1, 2, 3, 4, 6, 7, and SIRT1,2,3,5,and 6) and 28Stranscript

Tissue samples

Genes transcripts (mean Ct*)

28S HDAC1 HDAC2 HDAC3 HDAC4 HDAC6 HDAC7 SIRT1 SIRT2 SIRT3 SIRT5 SIRT6

NP 25,13 27,78 31,41 32,35 31,80 23,34 32,15 36,40 30,89 28,66 31,96 25,85NA1 26,65 32,52 34,43 30,91 33,65 ND 34,41 40,44 ND 30,56 31,15 25,54NA2 25,75 30,83 33,96 32,98 34,21 22,93 35,04 31,26 30,29 31,21 34,26 27,22ND1 22,69 25,75 32,75 30,63 30,04 ND 34,42 34,48 ND 27,24 30,00 26,03ND2 25,38 28,01 36,11 29,80 33,38 ND 33,76 37,73 ND 29,96 30,79 25,69BD3 22,33 26,45 33,80 27,83 30,14 ND 31,95 33,01 ND 27,19 31,64 26,05BD4 20,77 36,33 29,90 29,13 30,23 ND 31,67 33,94 ND 26,29 31,43 26,52

* qPCR were run in triplicate on two independent cDNA preparations from pancreatic tissues as described in the ‘‘Materials and Methods’’section. The mean crossing point values (Ct) were determined for the following samples: NP, normal pancreas; NA1-2, normal adjacent tissuessamples from two patients with PA; ND1-2, normal distant tissue samples from four patients with PA; BD, normal pancreas sample frompatients carrying biliary duct tumors. ND, not done.

M. OUAISSI ET AL.2324

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

Interestingly, even the patient in stage III, whichshowed low levels of all sirtuin and HDAC1, 2, 3, 4,and 6 transcripts exhibit elevated HDAC7 mRNAlevels. Together, these data suggest that upregulationof HDAC7 protein in PA tissue samples is at leastpartially due to the increase of HDAC7 mRNAlevels.

Immunohistochemical Localization of Protein

Expression

The antibodies against HDAC/SIR2 proteins weretested by immunohistochemical labeling of pancreatictissues samples. Figure 4 shows representative exam-ples of each type of staining. When compared to NP

FIG. 3. HDACs gene expres-sion in pancreatic tumor tissuesand in relation to staging ofcarcinomas. Relative mRNAHDAC1, 2, 3, 4, 6 and 7 tran-scripts were determined follow-ing the method described inmaterials and methods section.

FIG. 4. Immunohistochemical verification and localization of protein expression in pancreatic cancer tissues. Normal pancreas tissue section(A) and PA tissue (B) reacted with antibodies to HDAC7 showing strongly stained cell cytoplasm (arrow) and to a lesser extent nuclei (arrowhead) in PA specimen. (C) Positive immunolabeling of ductal cells in the nuclei revealed with anti-HDAC1 antibodies. (D) Immunolabelingfor SIRT2 protein shows intense labeling within the cytoplasm of acinar epithelial cells. (E) negative control reaction of pancreatic tissue usingcontrol rabbit serum

PANCREATIC CANCER AND HISTONE DEACETYLASE 2325

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

(Fig. 4A), most of the PA tissue sections examinedshowed intense labeling in the cytoplasm when reactedagainst antibodies toHDAC7 (Fig. 4B). Some labelingcould also be seen in the nucleus of the cells, this is notsurprising since it is well known that the class II com-prising subfamily members HDAC4, 5, 7, and 9, arecharacterized by their nucleocytoplasmic shuttling.HDAC1 showed positive immunostaining in the nu-cleus (Fig. 4C), whereas SIRT2 exhibited immunore-activity in the cytoplasmic compartment (Fig. 4D).Nuclear immunostaining was observed in the case ofantibodies to SIRT1 and HDAC4 (data not shown).The labeling was specific, as only background signalcould be seen when using non-immune rabbit serum inplace of the primary antibody (Fig. 4E).

DISCUSSION

In order to provide more insight into the biologicalbehavior of pancreatic cancer and identify new po-tential biomarkers, a number of studies have beenconducted to identify global gene expression profilesof the disease.17,18 The use of serial gene expression,oligonucleotide, and cDNA arrays, allowed to showthat a large set of genes were expressed at high levelsin pancreatic cancer tissues compared to normalpancreas.19–21 For example, biocomputational toolsused for principal component analysis showed thatamong the most differentially expressed genes inpancreatic cancer were Mesothelin, Muc4, Muc5A/C,Kallikrein 10, Transglutaminase 2, Fascin,TMPRSS3, and Stratifin.22 However, a number ofother genetic alterations for several specific genes,including K-ras, p53, p16INK4a, and Smad4, have beendocumented.23 Moreover, other markers such as theS100A6, a low-molecular-weight calcium bindingprotein that belongs to a family of proteins carryingan helix-turn-helix structural domain (EF-hand)known as S100 family, have been shown to be over-expressed in pancreatic carcinoma.24,25 Detailedanalysis of cytoplasmic and nuclear expression ofS100A6 in benign, malignant, and premalignantpancreatic ductal cells showed that high nuclearlocalization of S100A6 is significantly associated withpoor survival in pancreatic cancer patients.26

In this study, we analyzed differentially expressedmembers of class I, II, and III deacetylases encodinggenes in human pancreatic cancer tissues using theqPCR approach. To our knowledge, this is the firstreport of the use of qPCR-based approach for thestudy of human pancreatic cancer. Our study revealedthat most of the PA tumors analyzed (10/11, 90%) as

well as CP, SC, and IMPN samples showed increasedexpression of SIRT5 gene. Remarkably, 9 of the 11 PAtumors (81%) showed upregulation of HDAC7 tran-scripts compared with CP, SC, and IMPN samples,which showed reduced expression ofHDAC7message.The upregulation of HDAC7 in the PA tumors issupported by the Western blot data, showing strongpositive signals in PA tissues extracts compared tonormal pancreas or CP, SC, and IMPN specimens.The majority of PA contains activating point

mutations of the K-ras gene ([90% in most studies),and a large number of these neoplasms also exhibitalterations in genes controlling the G1/S-phase cellcycle transition such as p16INK4a.27 Moreover, thefrequency of p16INK4a genetic alteration is signifi-cantly higher in the shortest-surviving patients com-pared with longest-surviving patients.4 Othermolecular alterations may contribute to carcinogen-esis of the pancreas such as those related to growthfactors and/or their receptors.28,29 That high HDAC7transcript levels associate with most of the PA tumorswas not anticipated, as this relationship has not, toour knowledge, been shown for PA tumors nor in thecase of any other cancer type. Other HDAC andSIRT family members, in contrast, have been linkedto the outcome in certain cancers.Indeed, there are several recent reports showing

either up- or downregulation of HDAC encodinggenes. Overexpression of HDAC1 was shown ingastric cancer30 and increased expression of HDAC2and HDAC3 was reported in human colorectaltumors upon loss of APC tumor suppressor31 andcolon cancer,32 respectively, whereas HDAC5 mRNAtranscript was shown to be downregulated in coloncancer.33 Furthermore, attempts to define antigenictargets recognized by serum antibodies from coloncancer patients showed that HDAC5 is among 13defined antigens that reacted exclusively with pa-tient’s sera and not with sera from normal blooddonors.33 Moreover, in the case of class III deace-tylase members, SIRT8 was shown to be overex-pressed in thyroid carcinoma and cell lines but not inadenomas,11 whereas the SIRT2 gene was shown tobe downregulated in the case of human gliomas.34

Therefore, HDACs and SIRTs proteins may beclinically beneficial antigenic biomarkers for diagno-sis and/or prognosis of human cancer.Substantial progress has been made in deciphering

the importance ofHDACs/SIRTs as critical regulatorsof cell growth, differentiation, and death. The dereg-ulation of HDAC-mediated transcriptional repressionmay lead to tumorigenesis. In fact, a number of studieshave shown that HDAC inhibitors (HDACi) such as

M. OUAISSI ET AL.2326

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

sodium butyrate, trichostatin, suberioylanilide, andvalproic acid induced cell cycle arrest, differentiation,and apoptosis of certain tumor cells, and based onthese observations, several of these compounds arecurrently undergoing clinical trials.35 Interestingly, arecent study has reported that the vorinostat, the firstHDACi approved for clinical use in the treatment ofthe cancer cutaneous T-cell lymphoma,36 selectivelysuppresses expression of HDAC7.37 Moreover, usingsmall interfering RNAs, it has been demonstrated thatHDAC7 is a key modulator of endothelial cell migra-tion and hence angiogenesis, at least in part, by regu-lating PDGF-Bb/PDGFR-beta gene expression.38

Given that angiogenesis is required for tumor pro-gression, the data support the notion that moleculesable to interfere with HDAC7 may be of particulartherapeutic benefit. Furthermore, it has been shownthat HDAC7 is highly expressed in the nucleus ofresting CD4+CD8+ double-positive thymocytes,39

where it acts as a transcriptional repressor for the pro-apoptotic orphan receptor Nur77 and other cellulargenes involved in T-lymphocyte differentiation. UponT-cell receptor activation, phosphorylation ofHDAC7 resulted in its nuclear export leading toNur77expression and hence the facilitation of apoptosis.39–41

The critical challenge in pancreatic cancer is thedetection of early lesions at an asymptomatic stagethat may allow curative resection and offer a greatchance of cure. In the present study, we have shownthat HDAC7 upregulation is associated with adeno-carcinoma of the pancreas. This observation couldnot clearly associate with any other patient variable.Whether upregulation of HDAC7 in PA is a cause ora consequence of malignant progression is at presentunclear. Due to the aggressive nature of pancreaticcancer, almost 90% of patients are unsuitable forresection of their cancer on presentation.42 As aconsequence, relatively small number of specimenswere analyzed in this study. It is possible that with alarge patient set, relationship between HDACs/SIRTsgene expression and other patient variables such asstaging of PA may emerge. The characterization ofthe gene expression profiles in pancreatic tissuescould not only provide novel diagnostic biomarkersbut could also lead to the identification of potentialor therapeutic drug targets. Moreover, the approachwill shed light on additional molecular pathways thatmay play a role in the biology of this deadly disease.

ACKNOWLEDGMENTS

This work received support from INSERM. Theauthors wish to thank Ms Sylvie PICHET from the

NOVARTIS laboratory for her support, and MsDeborah Garcia for her efficient assistance in theorganization of the reagents supply. The authorsthank Dr. Laurent LAPLAZE for access to theDMRB microscope and images recording and MissJoana TAVARES for the Western blot imageanalysis. We are indebted to Ms M. RADAL forher technical assistance in the preparation ofcryosections of tissue samples. Dr. Leal Silvestrereceived a fellowship from the Fundacao para aCiencia e Tecnologia (FCT) Portugal. Dr. MehdiOuaıssi was supported by Assistance Pubique-Hopitaux de Marseille and University of AixMarseille and has received a grant from theNOVARTIS Laboratory.

REFERENCES

1. Greenlee RT, Murray T, Bolden S, et al. Cancer statistics.Cancer J Clin 2000; 50:7–33.

2. Goggins M, Canto M, Hruban R. Can we screen high-riskindividuals to detect early pancreatic carcinoma? J Surg Oncol2000; 74:243–8.

3. Rosenberg L. Pancreatic cancer: a review of emerging thera-pies. Drugs 2000; 59:1071–89.

4. Gerdes B, Ramaswamy A, Ziegler A, et al. p16INK4a is aprognostic marker in resected ductal pancreatic cancer. AnnSurg 2002; 235:51–9.

5. Cress WD, Seto E. Histone deacetylase, transcriptional controland cancer. J Cell Physiol 2000; 184:1–16.

6. Marks PA. Histone deacetylase inhibitors as a new drugs. CurrOpin Oncol 2001; 13:477–83.

7. Timmermann S. Histone acetylation and disease. Cell Mol LifeSci 2001; 58:728–36.

8. Amunziato AT, Hansen JC. Role of histone acetylation in theassembly and modulation of chromatin structures. Gene Expr2000; 9:37–61.

9. Bolden JE, Peart MJ, Johnstone RW. Anticancer activities ofhistone deacetylase inhibitors. Nat Rev Drug Discov 2006;5:769–84.

10. Frye RA. Phylogenetic classification of prokaryotic andeukaryotic Sir2-like proteins. Biochem Biophys Res Commun2000; 273:793–8.

11. De Nigris F, Cerutti J, Morelli C, et al. Isolation of a SIR-likegene, SIR-T8, that is overexpressed in thyroid carcinoma celllines and tissues. Br J Cancer 2002; 86:917–23.

12. Greene FL, Page DL, Fleming ID, et al., eds. AJCC CancerStaging Handbook, 6th edition. American Joint Committee onCancer. New York: Springer Verlag, 2002.

13. Corpet F. Multiple sequence alignment with hierarchicalclustering. Nucl Acids Res 1988; 22:10881–90.

14. Livak KJ, Schmittgen TD. Analysis of relative gene expressiondata using real-time quantitative PCR and the 2�DDCt method.Methods 2001; 25:402–8.

15. Avila AM, Burnett BG, Taye AA, et al. Trichostatin A in-creases SMN expression and survival in a mouse model ofspinal muscular atrophy. J Clin Inv 2007; 117:659–71.

16. Shen S, Li J, Casaccia-Bonnefil P. Histone modifications affecttiming of oligodendrocyte progenitor differentiation in thedeveloping brain. J Cell Biol 2005; 169:577–88.

17. Gress TM, Muller-Pillasch F, Geng M, et al. A pancreaticcancer-specific expression profile. Oncogene 1996; 13:1819–30.

PANCREATIC CANCER AND HISTONE DEACETYLASE 2327

Ann. Surg. Oncol. Vol. 15, No. 8, 2008

18. Crnogorac-Jurcevic T, Efthimiou E, Capelli P, et al. Geneexpression profiles of pancreatic cancer and stromal desmo-plasia. Oncogene 2001; 20:7437–46.

19. Sidransky D. Emerging molecular markers of cancer. Nat RevCancer 2002; 2:210–9.

20. Rosty C, Goggind M. Early detection of pancreatic carcinoma.Hematol Oncol Clin North Am 2002; 16:37–52.

21. Suzuki H, Gabrielson E, Chen W, et al. A genomic screen forgenes upregulated by demethylation and histone deacetylaseinhibition in human colorectal cancer.NatGenet 2002; 31:141–9.

22. Iacobuzio-Donahue CA, Ashfaq A, Maitra A, et al. Highlyexpressed genes in pancreatic ductal adenocarcinomas: acomprehensive characterization and comparison of the tran-scription profiles obtained from three major technologies.Cancer Res 2003; 63:8614–22.

23. Bardeey N, DePinho RA. Pancreatic cancer biology andgenetics. Nat Rev Cancer 2002; 2:897–909.

24. Crnogorac-Jurcevic T, Missiaglia E, Blaveri E, et al. Molecularalterations in pancreatic carcinoma expression profiling showsthat dysregulated expression of S100 genes is highly prevalent.J Pathol 2003; 201:63–74.

25. Shekouh A, Thompson CC, Prime W, et al. Development andapplication of laser capture microdissection combined withtwo-dimensional electrophoresis for the discovery of differen-tially regulated proteins in pancreatic ductal adenocarcinoma.Proteomics 2003; 3:1988–2001.

26. Vimalachandran D, Greenhalf W, Thompson C, et al. Highnuclear S100A6 (Calcyclin) is significantly associated with poorsurvival in pancreatic cancer patients. Cancer Res 2005;65:3218–25.

27. Hruban RH, Iacobuzio-Donahue C, Wilentz RE, et al. Molec-ular pathology of pancreatic cancer. Cancer J 2001; 7:251–8.

28. Ebert M, Yokoyama M, Friess H, et al. Induction of platelet-derived growth factor A and B chains and over-expression oftheir receptors in human pancreatic cancer. Int J Cancer 1995;62:529–35.

29. Friess H, Yamanaka Y, Buchler M. Enhanced expression ofthe type II transforming growth factor beta receptor in humanpancreatic cancer cells without alteration of type III receptorexpression. Cancer Res 1993; 53:2704–7.

30. Choi JH, Kwon HJ, Yoon BI, et al. Expression profile ofhistone deacetylase 1 in gastric cancer tissues. Jpn J Cancer Res2001; 92:1300–4.

31. Zhu P, Martin E, Mengwasser J, Schlag P, et al. Induction ofHDAC2 expression upon loss of APC in colorectal tumoro-genesis. Cancer Cell 2004; 5:455–63.

32. Wilson AJ, Byun DS, Popova N, et al. Histone deacetylase 3(HDAC3) and other class I HDACs regulate colon cell matu-ration and p21 expression and are deregulated in human coloncancer. J Biol Chem 2006; 281:13548–58.

33. Scanlan MJ, Welt S, Gordon CM, et al. Cancer-related sero-logical recognition of human colon cancer: identification ofpotential diagnostic and immunotherapeutic targets. CancerRes 2002; 62:4041–7.

34. Hiratsuka M, Inoue T, Toda T, et al. Proteomics-based iden-tification of differentially expressed genes in human gliomas:down-regulation of SIRT2 gene. Biochem Biophys Res Com-mun 2003; 309:558–66.

35. Ouaissi M, Ouaissi A. Histone deacetylase enzymes as poten-tial drug targets in cancer and parasitic diseases. J BiomedBiotechnol 2006; 13474:1–10.

36. Marks PA, Breslow R. Dimethyl sulfoxide to vorinostat:development of the histone deacetylase inhibitor as an anti-cancer drug. Nat Biotechnol 2007; 25:84–90.

37. Dokmanovic M, Perez G, Xu W, et al. Histone deacetylaseinhibitors selectively suppress expression of HDAC7. MolCancer Ther 2007; 6:2525–34.

38. Mottet D, Bellahcene A, Pirotte S. HDAC7 silencing altersendothelial cell migration, a key step in angiogenesis. Cir Res2007; 18:1794–801.

39. Dequiedt F, Kasler H, Fischle W, et al. HDAC7, a thymus-specific class II histone deacetylase, regulates Nur77 tran-scription and TCR-mediated apoptosis. Immunity 2003;18:687–98.

40. Kao HY, Verdel A, Tsai CC, et al. Mechanism for nucleocy-toplasmic shuttling of histone deacetylase 7. J Biol Chem 2001;276:47496–507.

41. Parra M, Kasler H, McKinsey TA, et al. Protein kinase D1phosphorylates HDAC7 and induces its nuclear export after T-cell receptor activation. J Biol Chem 2005; 280:13762–70.

42. Sener S, Frengen A, Menck HR, et al. Pancreatic cancer: areport of treatment and survival trends for 100313 patientsdiagnosed from 1985–1995, using the National Cancer Data-base. J Am Coll Surg 1999; 189:1–7.

M. OUAISSI ET AL.2328

Ann. Surg. Oncol. Vol. 15, No. 8, 2008