identification of pivotal cellular factors involved in hpv-induced dysplastic and neoplastic...

Identification of Pivotal CellularFactors Involved in HPV-InducedDysplastic and Neoplastic CervicalPathologiesSTEFANO MATTAROCCI,1 CLAUDIA ABBRUZZESE,2 ANNA M. MILEO,2

MARIANTONIA CAROSI,3 EDOARDO PESCARMONA,3 CARMEN VICO,2

ANTONIO FEDERICO,2 ENRICO VIZZA,4 GIACOMO CORRADO,4 IVAN ARISI,5

ARMANDO FELSANI,6 AND MARCO G. PAGGI2*1Department of Molecular Biology, University of Geneva, Geneva, Switzerland2Department of Development of Therapeutic Programs, Regina Elena National Cancer Institute, Rome, Italy3Department of Pathology, Regina Elena National Cancer Institute, Rome, Italy4Department of Gynecology, Regina Elena National Cancer Institute, Rome, Italy5Genomics Facility, European Brain Research Institute (EBRI) Rita Levi-Montalcini, Rome, Italy6CNR, Istituto di Neurobiologia e Medicina Molecolare, Rome, Italy

Cervical carcinoma represents the paradigm of virus-induced cancers, where virtually all cervical cancers come from previous “high-risk”HPV infection. The persistent expression of the HPV viral oncoproteins E6 and E7 is responsible for the reprogramming of fundamentalcellular functions in the host cell, thus generating a noticeable, yet only partially explored, imbalance in protein molecular networks andcell signaling pathways. Eighty-eight cellular factors, identified as HPV direct or surrogate targets, were chosen and monitored in aretrospective analysis for their mRNA expression in HPV-induced cervical lesions, from dysplasia to cancer. Real-time quantitativePCR (qPCR) was performed by using formalin-fixed, paraffin embedded archival samples. Gene expression analysis identified 40 genessignificantly modulated in LSIL, HSIL, and squamous cervical carcinoma. Interestingly, among these, the expression level of a panel offour genes, TOP2A, CTNNB1, PFKM, and GSN, was able to distinguish between normal tissues and cervical carcinomas.Immunohistochemistry was also done to assess protein expression of two genes among those up-regulated during the transition betweendysplasia and carcinoma, namely E2F1 and CDC25A, and their correlation with clinical parameters. Besides the possibility of significantlyenhancing the use of some of these factors in diagnostic or prognostic procedures, these data clearly outline specific pathways, and thus keybiological processes, altered in cervical dysplasia and carcinoma. Deeper insight on how these molecular mechanisms workmay help widenthe spectrum of novel innovative approaches to these virus-induced cell pathologies.J. Cell. Physiol. 229: 463–470, 2014. � 2013 Wiley Periodicals, Inc.

Infection from different pathogens is recognized as a majorcause of cancer (zur Hausen, 2009; de Martel et al., 2012).Among these infectious agents, the body of knowledgeconcerning the role of “high-risk” human papillomavirus (HPV)strains, mainly HPV 16 and HPV 18, in cervical cancer appearslarge and consistent. Over the last few years, cervical cancerincidence and death rates are falling, mainly due to the access toreliable screening procedures which facilitate early diagnosisand treatment (Siegel et al., 2013). While HPV vaccinesrepresent a promising preventive approach towards HPVinfection and related neoplastic diseases, their therapeuticeffectiveness is yet to be proven. Despite ongoing massive anti-HPV vaccination campaigns, a significant number of cancers arestill expected to occur mainly in developing countries due tocultural and socio-economic reasons and to the potentialselection of high-risk HPV strains not neutralized by thevaccines.

HPV-induced tumors usually arise several years after viralinfection, displaying the integration of the viral genome intocancer cell chromosomes, which allows the stabilizedexpression of the E6 and E7 oncogenes and of the respectiveprotein products, the E6 and E7 oncoproteins. These arethe major effectors of HPV-driven cell transformation andappear also central for cancer cell clones maintenance and

proliferation by acting via several molecular mechanisms ableto weaken normal cell cycle control and modulate the cellularapoptotic response (Fehrmann and Laimins, 2003; Munger

Stefano Mattarocci and Claudia Abbruzzese contributed equally tothis work.

Contract grant sponsor: Associazione Italiana Ricerca sul CancroAIRC.Contract grant sponsor: Ministero della Salute.Contract grant sponsor: Ministero dell’Universita dell’Istruzione edella Ricerca.Contract grant sponsor: Fondazione Roma.Contract grant sponsor: CNR “Medicina Personalizzata”.

*Correspondence to: Marco G. Paggi, Regina Elena NationalCancer Institute, Via Elio Chianesi, 53, Rome 00144, Italy.E-mail: [email protected]

Manuscript Received 18 July 2013Manuscript Accepted 29 August 2013

Accepted manuscript online in Wiley Online Library(wileyonlinelibrary.com): 16 September 2013.DOI: 10.1002/jcp.24465

ORIGINAL RESEARCH ARTICLE 463J o u r n a l o fJ o u r n a l o f

CellularPhysiologyCellularPhysiology

� 2 0 1 3 W I L E Y P E R I O D I C A L S , I N C .

et al., 2004; Whiteside et al., 2008; McLaughlin-Drubin andMunger, 2009; Bellacchio and Paggi, 2013). Interestingly, E6 andE7 from high-risk HPV strains are void of any enzymatic activity,but display the peculiarity to generate specific protein–proteininteractions through which they are able to reprogramfundamental cellular functions. Nowadays, the most investi-gated cellular targets of E6 and E7 are the p53 and theretinoblastoma (RB) family proteins, respectively (Helt andGalloway, 2003; Felsani et al., 2006). In several cases, theendogenous factors targeted by the small DNA virusoncoproteins undergo dramatic changes also in their cellularamount, due to direct effects on the protein or totranscriptional interference (Munger et al., 2001; Helt andGalloway, 2003; Felsani et al., 2006; Mileo et al., 2013).

In HPV-induced pathologies of the uterine cervix, thediscovery of sensitive and specific biomarkers differentiallyexpressed in non-progressive and progressive diseases mightefficiently integrate the current procedures for risk assessmentand inform therapeutic decisions.

The purpose of our study was to identify pivotal factorsinvolved in dysplastic and neoplastic cervical pathologies, inorder to gain information on the molecular processesimplicated in these HPV-induced processes. Therefore, weselected a series of cellular factors whose amount appearsmodulated in HPV-related cervical pathologies plus otherschosen through experimental data generated in our laboratory.We then performed an extensive quantitative PCR (qPCR)analysis by monitoring a supervised set of 88 cellular genespossibly involved in cervical dysplastic and neoplastic diseases.Our samples consisted of a set of 127 formalin-fixed paraffin-embedded (FFPE) archival cervical samples, including controlcases, low-grade squamous intraepithelial lesions (LSIL), high-grade squamous intraepithelial lesions (HSIL) and carcinomas.The expression of a subset of genes was found significantlydifferent in the various categories of cervical samples, thussome of these genes have been selected as potential indicatorsfor transition from dysplasia to cancer. For two of thesegenes, specifically E2F1 and CDC25A, analysis of their proteinproduct was evaluated via immunohistochemistry (IHC) insquamous cervical carcinoma samples.

Our results can significantly promote the use of some of thesefactors in diagnostic/prognostic procedures by identifying a panelof genes/proteins whose expression can be correlated withknown clinical parameters. It is crucial to direct new investigativeefforts to improve our knowledge and understanding on themolecular mechanisms behind HPV infection/transformation inorder to widen the spectrum of novel and innovative approachesagainst virus-induced pathologies.

Materials and MethodsPatients

Tissues from 127 cervical surgical specimens (control [normalcervical epithelium from patients who underwent hysterectomyfor reasons other than cancer; 22 cases], LSIL [35 cases], HSIL [27cases], cervical cancer [43 cases]) were evaluated. All the patientswere diagnosed and treated at the Regina Elena National CancerInstitute, Rome, Italy. Patients underwent international standardtherapeutic protocols. Clinical data (patient diagnosis and staging)were obtained from the Regina Elena National Cancer Institutedatabases. Samples were collected according to the currentinstitutional ethical guidelines.

RNA extraction and quantitative gene expressionanalysis in archival samples

Total RNA extraction from FFPE cervical specimens was done asdescribed (Abbruzzese et al., 2012). Total RNA was checked forquality and quantified using the 2100 bioanalizer (Agilent, Santa

Clara, CA). RNA samples, even with a low RNA integritynumber (RIN), were retrotranscribed using the High CapacitycDNA Reverse Transcription Kit (Applied Biosystems,Branchburg, NJ). Each sample was then analyzed by qPCR usingcustom-made microfluidic cards (Applied Biosystems)containing sequence specific primer pairs for 88 genes to beexamined and for six endogenous control genes (Table S1,Supplementary Material). The cards were processed as indicatedby the manufacturer using the 7900HT thermal cycler apparatusequipped with the SDS software version 2.3 (Applied Biosystems)for data collection.

Histological examination and IHC

The histological diagnosis was re-evaluated in 2-mm FFPE sectionsafter routine laboratory haematoxylin/eosin staining.

IHC analysis was performed as described (Mattarocciet al., 2009) after an antigen retrieval step, treating slides inmicrowave oven at 800W for 15min� 2 in 300ml of 10mMcitrate buffer, pH 6.0.Immunostaining was carried out by using aprimary monoclonal antibody for Cdc25a (sc-56264, Santa CruzBiotechnology, Inc. Santa Cruz, CA), applied overnight (ON) atroom temperature (RT) at a 1:50 dilution. E2F1 was detected via amouse monoclonal antibody (sc-251, Santa Cruz Biotechnology)applied ON at RT at a 1:20 dilution. For both antibodies, optimalworking dilution was defined on the basis of titration experiments.The secondary antibody solution and streptavidin-biotin, bothfrom the AFN600-IFU kit (ScyTek, Logan, UT), were appliedaccording to the manufacturer’s instructions. Finally, 3-amino-9-ethylcarbazide (AEC substrate kit, ScyTek) was used as the finalchromogen. Mayer’s haematoxylin was used for nuclear coun-terstaining. Negative controls for each tissue section wereprepared by omitting the primary antibody.

To increase the number of samples to be evaluated, 208-sampleTissue Micro Arrays (TMA) of cervical carcinoma cases(US Biomax CR2088, Rockville, MD), containing 181 squamouscervical carcinoma samples with I–III grading, 9 cervicaladenocarcinoma samples (not evaluated) plus 16 normal cervixsamples as negative controls, were stained for CDC25A andE2F1 expression following the procedures described above.

The total number of evaluable squamous cervical carcinomasamples was 219.

Scoring and quantification of gene expression andimmunoreactivity

Gene expression. qPCR reaction data were analyzed usingthe HTqPCR Bioconductor package software (version 1.4.0) forthe R statistical computing environment, version 2.12.1 (Dvingeand Bertone, 2009). Ct values were normalized by the deltaCtmethod on the average of the endogenous controls. Thenormalization stability score of these control RNAswas confirmedby the GeNorm software. Data not acquired (N/A) were assignedwith a default Ct value of 40. Differential expression analysis wasconducted using the functions from the R limma package (Smythet al., 2005), after having removed the Ct values corresponding tothe endogenous controls and those with an Inter Quartile Range�1.5. P-values were corrected for multiple testing followingBenjamini and Hochberg (1995).Immunoreactivity. Three examiners (M.C., E.P., and M.G.P.)

independently evaluated the staining pattern of Cdc25A and E2F1in the same cervical carcinoma case series employed for mRNAexpression (42 cases) plus in an independent set of squamouscervical carcinomas on TMA slides (181 cases). The final scorederived from the independent analysis of the examiners, withsubsequent discussion for the cases in which divergent diagnoseswere given. According to the amount of specific staining, caseswere classified in tertiles as follows: (a) negative/low; (b) medium;(c) high.

JOURNAL OF CELLULAR PHYSIOLOGY

464 M A T T A R O C C I E T A L .

Cervical cancer samples staining positive for E2F1 displayed animmunoreactivity characterized by a distinct nuclear staining withor without diffuse cytoplasmic staining. IHC results for E2F1 wereexpressed as a percentage of the positive number of nuclear orcytoplasmic cells separately, counted in 10 microscope cellularfields (40�). Tumors were defined positive for E2F1 when 20% ormore tumor cells showed unequivocal nuclear or cytoplasmicstaining. Cases with faint, uncertain cytoplasmic and/or nuclearstaining were regarded as negative. In normal cervical epithelium,nuclear staining appeared confined to the basal and parabasallayers.

CDC25A displayed a diffuse or granular cytoplasmic staininghighly specific for carcinoma cells. IHC results were expressed as apercentage of the positive cells, counted in 10 microscope cellularfields (40�). Tumors were defined positive for CDC25A when20% or more tumor cells showed unequivocal staining. Cases withfaint, uncertain staining were regarded as negative. In normalcervical epithelium, nuclear staining was limited to theintermediate and superficial layers.

Statistical analysis

To analyze differential expression of DCt levels in single pairwisecomparisons (e.g., control vs. HSIL or control vs. carcinoma), thenon-parametric Mann–Whitney U-test was applied to evaluatestatistical significance. All categorical variables, such as tumor stageor International Federation of Gynecology and Obstetrics(F�ed�eration Internationale de Gyn�ecologie et d’Obst�etrique,FIGO) stage, were tested for statistical significance by using theFisher’s exact test applied to contingency tables, where expressionlevels, for both mRNA and protein, were divided into threegroups.

For all statistical tests, a two-tailed P-value< 0.05 wasconsidered as statistically significant.

Hierarchical clustering and principal component analysis (PCA)of samples were computed by the Multi Experiment Viewer ver.4.6 (Saeed et al., 2006).

ResultsChoice of the genes to be assayed and of the cervicalsamples to be analyzed

A number of genes were chosen after a thorough survey of thepublished literature was carried out. We sought for those whosemRNA and/or protein expressions that significantly correlatedwith histopathological score and other clinical parameters incervical dysplasia and cancer.Other geneswere chosen accordingto data derived from our in vitro studies on protein–proteininteractions involving HPV-16 oncoproteins and host cell factors(Mileo et al., 2006, 2009, 2013; Severino et al., 2007) orpreliminary qPCR data in HPV-16 E7-expressing HaCaTimmortalized human keratinocytes (data not shown). The finaloutput consisted of a list containing 88 genes, plus 6 housekeepinggenes chosen for normalization, whose amount was thusdetermined in a supervised analysis. The complete list of assayedgenes, respective protein name and references or data employedfor their selection, is displayed in Table S1 (SupplementaryMaterial). mRNA expression for the supervised set of these 88genes has been thus determined in FFPE archival humanspecimens for a total of 127 individuals subdivided into fourcohorts by the pathologists, that is, control (22 cases), LSIL (35cases), HSIL (27 cases), cervical cancer (43 cases).

Gene expression analysis (qPCR)

Raw data on the expression of 88 genes in all the analyzedsamples are reported in Table S2 (Supplementary Material).Results are shown in Figure 1 and expressed as log2 fold change(FC). Assuming an arbitrary cutoff value of 1.75 Log2 FC, the

expression of 40 genes appeared significantly modulated inLSIL, HSIL, and cervical carcinoma, when compared withcontrols (samples of normal cervical tissue). Genes werecategorized as follows:

� Up-regulated at all stages, from control (baseline) to LSIL, HSIL,and carcinoma: PAK6 and MKI67. Up-regulation of thesegenes was linked to a pathological state of the cervix uteri(dysplasia or carcinoma).

� Down-regulated at all stages, from control (baseline) to LSIL,HSIL, and carcinoma: MYC, SIVA1, BCL2, SNAI1, PTGS2,and FOS. Down-regulation of these genes was linked to apathological state of the cervix uteri (dysplasia or carcinoma).Among these, FOS, SNAI1, and BCL2 displayed themost evidentdown-regulation.

� Progressively up-regulated, from LSIL to HSIL and Carcinoma:E2F1, CCNE2, FOXM1, TP73, CDC25A, BIRC 5, andTOP2A. Their expression increased significantly in dysplasticlesions and even more in carcinomas. Among these genes,it is worth noting the consistent increase in expression betweenHSIL and carcinoma found in BIRC5, CDC25A, and E2F1.

� Progressively down-regulated, from LSIL to HSIL and Carcino-ma: RAD52, CAT, PKM2, IGFBP5, MAPK3, EDNA, CAV1,GSN, ABCB1, PFKM, PTGES, VIM, CTNNB1, APC, TOP2B, andLMNA. Their expression decreased significantly in dysplasticlesions and even more in carcinomas. Among these genes, it isworthy to note the consistent drop in expression betweenHSIL and carcinoma found in IGFBP5, EDNRA, and CAV1.

� Up-regulated in HSIL and Carcinoma only: CCNE1, MMP8,and CCNB1. These genes are up-modulated solely in high-gradedysplasias and carcinomas.

� Up-regulated in LSIL (or HSIL), but down-regulated inCarcinoma: EDN3, CST6, BCL2L1, KLK5, KRT10, and CDH1.Their expression was lower in carcinomas than in dysplasticlesions, even if often higher than in controls. Among thesegenes, it is important to consider the consistent drop inexpression observed in EDN3, KRT10, and KLK5.

A further analysis outlined a set of four genes (TOP2A,CTNNB1, PFKM, and GSN) whose expression, whendysplastic samples (LSIL and HSIL) were grouped (dysplasia),appeared strongly correlated with the overall patient status ordiagnosis. These genes were selected as a highly differentiallyexpressed small subset by Mann–Whitney U-test(10�11< P< 10�5) between sample groups in all possiblepairwise comparisons (control vs. dysplasia, control vs.carcinoma, carcinoma vs. dysplasia). In this analysis, weconsidered three group samples: Control cases, dysplasia andcervical carcinoma. In Figure 2A, PCA of samples, based on theexpression values of TOP2A, CTNNB1, PFKM, and GSN,showed a clear segregation between normal tissues andcervical carcinomas, while dysplastic samples gave scatteredvalues. Figure 2B shows hierarchical clustering and heat maprepresenting the level of expression of TOP2A, CTNNB1,PFKM, and GSN across the samples analyzed. TOP2Aexpression levels appeared to be clearly on the rise along withthe severity of the disease, while the opposite occurred forGSN, CTNNB1, and PFKM. Also in this case, the same 4-geneset strongly discriminated between normal and cancer samples,while dysplastic samples (LSIL plus HSIL) once again gaveheterogeneous results. Data are expressed as log2 FC incomparison with controls (normal cervical epithelium).Interestingly, the analysis of the data obtained from our cohortof squamous cervical cancer patients outlined a significantcorrelation between low levels of CTNNB1mRNA expressionand the presence of lymph node dissemination (Fisher’s exacttest P¼ 0.005).


G E N E S I N V O L V E D I N C E R V I C A L D Y S P L A S I A A N D C A N C E R 465

Fig. 1. Supervised gene expression in LSIL, HSIL, and squamous cervical carcinomas from FFPE samples. Genes significantly modulated inCervical Lesions: comparison with Controls. mRNA analysis was performed in four cohorts for a total of 127 patients, that is, control (22cases), LSIL (35 cases), HSIL (27 cases), cervical cancer (43 cases) and is expressed as log2 fold change (FC) in comparison with control(normal cervical epithelium; red color scale¼up-regulation toward control; green color scale¼down-regulation toward control). For eachgene listed, a statistical significance of P< 0.05 was achieved at least in one pathological class.



Protein expression analysis (IHC)

We focused our attention on the genes that, among the 40genes differentially regulated, showed a significantly increasedexpression between the dysplastic (LSILþHSIL) and thecancerous status, with the aim to perform a determination ofthe amount of protein products in cancer samples by means ofIHC. This choice appeared justified by the fact that the

transition between dysplasia and cancer is conceptually andclinically central, allowing to generate potentially newinformation about viral carcinogenesis processes and patientdiagnosis and prognosis as well. Another key point, in order tofacilitate a swift transition from research to clinic, was thechoice of genes which appeared up-regulated during thetransition between dysplasia and cancer. In fact, when IHC isused for determination, up-regulation toward controls is easier

Fig. 2. mRNA expression of a set of four genes (TOP2A, CTNNB1, PFKM, and GSN) is strongly correlated with the overall patient status ordiagnosis. A: principal component analysis (PCA) of expression values of patient samples. The dataset used for clustering is composed ofnormalized DCt expression values of genes, TOP2A, CTNNB1, PFKM, and GSN for all the samples. The color code shows how the samplessegregate in the plot, particularly along the PCA1 axis, according to diagnostic category: control, dysplasia, carcinoma. The color code is thesame as in the heat map plot. B: Hierarchical clustering and heat map of expression values of patient samples, using Euclidean distance metricand average linkage. The dataset used for clustering is composed of expression values of genes TOP2A, CTNNB1, PFKM, and GSN for all thesamples. Values are mean centered normalized DCt, where for each gene the mean is calculated across all the samples. On the bottom of thediagram, a color code shows how the samples segregated in the cluster, according to diagnostic category: control, dysplasia, and carcinoma. Inboth panels, genes were selected given their extremely different expression values in the three groups, as shown by the P-value< 106 in allpairwise non-parametric tests (Mann–Whitney U-test) between the groups (control vs. dysplasia, control vs. carcinoma, dysplasia vs.carcinoma). The color code, green for control, ochre for dysplasia, and purple for Carcinoma, is the same in both panels.



and safer to asses than down-regulation. On these bases, thegenes significantly up-regulated during the transition betweendysplasia and carcinoma were: TOP2A, FOXM1, BIRC5,CCNE2, CDC25A, TP73, and E2F1. Among those, we chose toassay for protein expression E2F1 and CDC25A, two geneswhose mRNA expression was significantly increased insquamous cervical carcinomas and for which IHC data are notyet available in the literature. We examined the expression ofthese two gene products in 42 cases of cervical carcinomabelonging to the samples from which mRNA was extractedand analyzed. In addition, the samples contained in a tissuemicroarray consisting in 181 cases of squamous cervicalcarcinoma were included, for a total of 223 carcinomasevaluated.

Positive staining for E2F1 significantly correlated withhistopathological grade (Fisher’s exact test P¼ 0.031), beinglower in high-grade carcinomas. No significant differenceswere found when considering tumor stage (T), lymph nodestage (N0 or N1) or tumor stage according to FIGOclassification (Stages I–IV). Analyzing the same parameters,positive staining for CDC25A reached statistical significanceonly for lymph node stage (Fisher’s exact test P¼ 0.03), whereN1 samples showed higher protein content than the N0 ones(Table 1).

Examples of IHC staining for E2F1 and CDC25A in normaland cervical squamous carcinomas are shown in Figure 3.

Discussion

This study aimed at identifying the key factors involved incervical dysplasia and carcinoma in order to recognize themolecular processes implicated in these HPV-inducedpathologies and to use the expression of specific cellularfactors to more accurately address important clinical issues.

The results reported herein allow identify a cluster of genesand gene products whose expression is modulated during thetransition from normal epithelium to dysplasia and cancer inHPV-generated cervical pathologies. Most of our findings areevidently consistent with the literature data we used to build upthe dataset of genes analyzed in our samples (see Table S1Supplementary Data). The results found on other genes arestrongly in accordance with the findings reported in ourprevious studies (Severino et al., 2007; Mileo et al., 2013).Finally, out of the 88 assayed genes, a cluster of four genes(Fig. 2), when evaluated concomitantly, appear capable todiscriminate between normal and cancer cervical tissues:TOP2A, GSN, PFKM, and CTNNB1.

TOP2A mRNA expression was significantly higher incervical carcinoma samples, in accordance with proteinexpression as evaluated by IHC (Branca et al., 2008; Guoet al., 2011; Brown et al., 2012).

GSN mRNA expression was significantly lower in cervicalcarcinoma samples. Our previous study shows a reducedgelsolin cleavage by caspase-3 in the presence of the HPV-16 E7oncoprotein (Mileo et al., 2013), which is in agreement with adecreased GSN mRNA synthesis in cervical carcinomas.

PFKM mRNA expression was significantly lower in cervicalcarcinoma samples. Phosphofructokinase M is a masterregulator enzyme in the glycolytic process. While there areseveral studies describing the effect of the interplay betweenHPV infection and pyruvate kinase M2 (PKM2) in increasingtumor glycolysis and enhancing the Warburg effect(Zwerschke et al., 1999; Mazurek et al., 2001; Tamadaet al., 2012), to our knowledge, this is the first report in whichmRNA expression of PFKM, another key glycolytic regulatorenzyme, appears modulated in HPV-expressing cervicalpathologies.

CTNNB1 mRNA expression was significantly lower incervical carcinoma samples. Data in literature outline an

inverse correlation between IHC-determined beta-cateninexpression and tissue differentiation, as well as low expressionin cervical carcinoma cases with lymph node dissemination(Cheng et al., 2012). The data herein presented are definitelyconsistent with these results.

In spite of the high statistical significance of the differences inmRNA expression in this panel of four selected genes (TOP2A,GSN, PFKM, and CTNNB1), even pooling the two dysplasticforms LSIL and HSIL was impossible to stratify the dysplasticcases, while cervical carcinomas were found to be efficientlysegregated from normal samples (Fig. 2, parts A and B). Thereason why the intermediate, dysplastic state did not segregatesuccessfully could also be explained by the intrinsic variabilityfound in these FFPE specimens due to their occasionally smallsize and consequent possible contaminationwith unpredictableamounts of adjacent peri-lesional tissue.

On the other hand, the significant variation of E2F1determination via IHC according to histopathological grade,being lower in high-grade samples, is not surprising. E2F1 is themost investigated member of the E2F family of transcriptionfactors that actively promote the cell entrance into the S phase.It binds to the “pocket” region of the Retinoblastomaoncosuppressor protein, and of its related suppressor factors

TABLE 1. Evaluation of E2F1 and CDC25A protein expression in squamouscervical carcinoma samples by immunohistochemistry: correlation with otherclinico-pathological parameters

E2F1 lowexpression(n¼ 83)

E2F1 middleexpression(n¼ 74)

E2F1 highexpression(n¼ 62) P-value

Histopathological gradeG1 10 8 5 0.031�G2 36 44 43G3 37 22 14

Tumor sizeT1 69 52 50 0.181T2 11 19 9T3 2 1 0T4 1 0 0N/A 0 2 3

Lymph node stageN0 81 67 58 0.208N1 2 4 0N/A 0 3 4

Tumor stage (FIGO)Stage I 69 50 46 0.083Stage II 11 19 9Stage III 1 3 0Stage IV 2 0 0N/A 0 2 7

Cdc25A lowexpression 0.944

(n¼ 98)

Cdc25A middleexpression(n¼ 71)

Cdc25A highexpression(n¼ 50) P-value

Histopathological gradeG1 14 6 3 0.41G2 49 42 32G3 35 23 15

Tumor sizeT1 78 56 37 0.466T2 16 13 9T3 3 0 0T4 0 0 1N/A 1 2 3

Lymph node stageN0 97 65 43 0.03�N1 0 4 2N/A 1 2 5

Tumor stage (FIGO)Stage I 78 55 35 0.944Stage II 16 13 9Stage III 2 1 1Stage IV 1 0 1N/A 1 2 4

�Statistically significant (Fisher’s exact test).



p107 and pRb2/p130, to be functionally inactivated and thusunable to stimulate the cell to enter the S phase (Ewen, 1994;Weinberg, 1995; Paggi et al., 1996; Mulligan and Jacks, 1998;Paggi and Giordano, 2001). In the presence of HPV-16 E7, E2F1is displaced from the Retinoblastoma protein, thus being ableto exert its trans-activating activity (Chellappan et al., 1992;Morris et al., 1993; Bellacchio and Paggi, 2013). However, at thesame time, it becomes substrate of the specific proteasesdevoted to its turnover (Hateboer et al., 1996).

In our cohort of patients, strong IHC staining forCDC25A appeared significantly correlated with theevidence of lymph-node dissemination in the patient. Indeed,the CDC25A tyrosine phosphatase is a key factor in cellcycle progression and its up-regulation is associated withcarcinogenesis. In particular, the E7 oncoprotein fromHPV-16 is able to increase CDC25A protein levels(Nguyen et al., 2002; Gao et al., 2009).

Now, it is important to consider that the majority of theseassays detect the expression of surrogate markers ofaberrations in cell proliferation genes and that these testscannot be swiftly employed in a clinical setting, even after astronger statistical validation, due to the intrinsic complexity ofthe non-routine procedures utilized. Nevertheless, theseresults outline genes and gene products whose expression,appearing critical in discriminating between normal and cancercervical tissues, can be of great value in recognizing key HPV-induced molecular derangements that are hallmarks of atransformed phenotype.

In the near future, we plan to increase the number ofspecimens to be assayed and to perform phosphoproteomicanalysis by reverse-phase protein microarrays (RPPM) in orderto explore the activation status of several signal transductioncomponents in fresh, surgically obtained cervical cancersamples. Activated signaling profiles will highlight potentially

relevant molecular biomarkers, thus generating novelclassification ranks and identifying druggable targets as well.

Indeed, grasping further knowledge of the molecularmechanisms employed by HPV in inducing and maintaining thecancerous state will be vital in understanding these processesand toward selecting effective therapies in defeating the HPVoncogenic machinery.

Acknowledgments

This work was supported by grants from Associazione ItalianaRicerca sul Cancro (AIRC; www.airc.it) and Ministero dellaSalute (www.ministerosalute.it) grants to M.G.P., fromMinistero dell’Universitadell’Istruzione e della Ricerca (www.miur.it; FIRB No. RBAP10L8TY) and Fondazione Roma to I.A.,and grant CNR “Medicina Personalizzata” to A.F. The authorsthank Daniela Panichi for the preparation of the FFPE archivalspecimens, Manuela Natoli for her help in qPCR mRNAdeterminations, and Tania Merlino for English language editing.

Literature Cited

Abbruzzese C, Mattarocci S, Pizzuti L, Mileo AM, Visca P, Antoniani B, Alessandrini G,Facciolo F, Amato R, D’Antona L, Rinaldi M, Felsani A, Perrotti N, Paggi MG. 2012.Determination of SGK1 mRNA in non-small cell lung cancer samples underlines highexpression in squamous cell carcinomas. J Exp Clin Cancer Res 31:4.

Bellacchio E, Paggi MG. 2013. Understanding the targeting of the RB family proteins by viraloncoproteins to defeat their oncogenic machinery. J Cell Physiol 228:285–291.

Benjamini Y, Hochberg Y. 1995. Controlling the false discovery rate—A practical andpowerful approach to multiple testing. J R Stat Soc Ser B Methodol 57:289–300.

Branca M, Ciotti M, Giorgi C, Santini D, Di Bonito L, Costa S, Benedetto A, Bonifacio D, DiBonito P, Paba P, Accardi L, Syrjanen S, Favalli C, Syrjanen K. 2008. Predicting high-riskhuman papillomavirus infection, progression of cervical intraepithelial neoplasia, andprognosis of cervical cancer with a panel of 13 biomarkers tested in multivariate modeling.Int J Gynecol Pathol 27:265–273.

Brown CA, Bogers J, Sahebali S, Depuydt CE, De PF, Malinowski DP. 2012. Role of proteinbiomarkers in the detection of high-grade disease in cervical cancer screening programs. JOncol 2012:289315.

Fig. 3. Immunohistochemical staining for E2F1 and CDC25A in cervical squamocellular carcinoma and in normal cervical tissue.Representative samples of squamous cervical carcinoma and normal cervical tissue which appear positive for E2F1 and CDC25A staining inIHC. E2F1 staining is essentially nuclear, with or without cytoplasmic staining. CDC25A staining appears as a diffuse or granular cytoplasmicstaining, highly specific for carcinoma cells. Original magnification¼ 20�.



Chellappan S, Kraus VB, Kroger B, Munger K, Howley PM, Phelps WC, Nevins JR. 1992.Adenovirus E1A, simian virus 40 tumor antigen, and human papillomavirus E7 proteinshare the capacity to disrupt the interaction between transcription factor E2F and theretinoblastoma gene product. Proc Natl Acad Sci USA 89:4549–4553.

Cheng Y, Zhou Y, JiangW, Yang X, Zhu J, Feng D,Wei Y, Li M, Yao F, HuW, XiaoW, Ling B.2012. Significance of E-cadherin, beta-catenin, and vimentin expression as postoperativeprognosis indicators in cervical squamous cell carcinoma. Hum Pathol 43:1213–1220.

de Martel C, Ferlay J, Franceschi S, Vignat J, Bray F, Forman D, Plummer M. 2012. Globalburden of cancers attributable to infections in 2008: A review and synthetic analysis.Lancet Oncol 13:607–615.

Dvinge H, Bertone P. 2009. HTqPCR: High-throughput analysis and visualization ofquantitative real-time PCR data in R. Bioinformatics 25:3325–3326.

Ewen ME. 1994. The cell cycle and the retinoblastoma protein family [review]. CancerMetastasis Rev 13:45–66.

Fehrmann F, Laimins LA. 2003. Human papillomaviruses: Targeting differentiating epithelialcells for malignant transformation. Oncogene 22:5201–5207.

Felsani A, Mileo AM, Paggi MG. 2006. Retinoblastoma family proteins as key targets of thesmall DNA virus oncoproteins. Oncogene 25:5277–5285.

GaoG, PengM, Zhu L,Wei Y,WuX. 2009. Human papillomavirus 16 variant E7 gene inducestransformation of NIH 3T3 cells via up-regulation of cdc25A and cyclin A. Int J GynecolCancer 19:494–499.

Guo M, Baruch AC, Silva EG, Jan YJ, Lin E, Sneige N, Deavers MT. 2011. Efficacy of p16 andProExC immunostaining in the detection of high-grade cervical intraepithelial neoplasiaand cervical carcinoma. Am J Clin Pathol 135:212–220.

Hateboer G, Kerkhoven RM, Shvarts A, Bernards R, Beijersbergen RL. 1996. Degradation ofE2F by the ubiquitin-proteasome pathway: Regulation by retinoblastoma family proteinsand adenovirus transforming proteins. Genes Dev 10:2960–2970.

Helt AM, Galloway DA. 2003. Mechanisms by which DNA tumor virus oncoproteins targetthe Rb family of pocket proteins. Carcinogenesis 24:159–169.

Mattarocci S, Abbruzzese C, Mileo AM, Visca P, Antoniani B, Alessandrini G, Facciolo F,Felsani A, Radulescu RT, Paggi MG. 2009. Intracellular presence of insulin and itsphosphorylated receptor in non-small cell lung cancer. J Cell Physiol 221:766–770.

Mazurek S, ZwerschkeW, Jansen-Durr P, Eigenbrodt E. 2001. Effects of the human papillomavirus HPV-16 E7 oncoprotein on glycolysis and glutaminolysis: Role of pyruvate kinasetype M2 and the glycolytic–enzyme complex. Biochem J 356:247–256.

McLaughlin-Drubin ME, Munger K. 2009. The human papillomavirus E7 oncoprotein.Virology 384:335–344.

Mileo AM, Piombino E, Severino A, Tritarelli A, Paggi MG, Lombardi D. 2006. Multipleinterference of the human papillomavirus-16 E7 oncoprotein with the functional role ofthe metastasis suppressor Nm23-H1 protein. J Bioenerg Biomembr 38:215–225.

Mileo AM, Abbruzzese C, Mattarocci S, Bellacchio E, Pisano P, Federico A, Maresca V,Picardo M, Giorgi A, Maras B, Schinina ME, Paggi MG. 2009. Human papillomavirus-16 E7interacts with glutathione S-transferase P1 and enhances its role in cell survival. PLoSONE4:e7254.

Mileo AM, Abbruzzese C, Vico C, Bellacchio E, Matarrese P, Ascione B, Federico A, Della BS,Mattarocci S, Malorni W, Paggi MG. 2013. The Human papillomavirus-16 E7 oncoprotein

exerts anti-apoptotic effects via its physical interaction with the actin-binding proteingelsolin. Carcinogenesis doi: 10.1093/carcin/bgt192

Morris JD, Crook T, Bandara LR, Davies R, LaThangue NB, Vousden KH. 1993. Humanpapillomavirus type 16 E7 regulates E2F and contributes tomitogenic signalling. Oncogene8:893–898.

Mulligan G, Jacks T. 1998. The retinoblastoma gene family: Cousins with overlappinginterests. Trends Genet 14:223–229.

Munger K, Basile JR, Duensing S, Eichten A, Gonzalez SL, GraceM, Zacny VL. 2001. Biologicalactivities and molecular targets of the human papillomavirus E7 oncoprotein. Oncogene20:7888–7898.

Munger K, Baldwin A, Edwards KM, Hayakawa H, Nguyen CL, Owens M, Grace M, Huh K.2004. Mechanisms of human papillomavirus-induced oncogenesis. J Virol 78:11451–11460.

NguyenDX,Westbrook TF, McCanceDJ. 2002. Human papillomavirus type 16 E7maintainselevated levels of the cdc25A tyrosine phosphatase during deregulation of cell cycle arrest.J Virol 76:619–632.

Paggi MG,GiordanoA. 2001.Who is the boss in the retinoblastoma family? The point of viewof Rb2/p130, the little brother. Cancer Res 61:4651–4654.

Paggi MG, Baldi A, Bonetto F, Giordano A. 1996. Retinoblastoma protein family in cell cycleand cancer: A review. J CellBiochem 62:418–430.

Saeed AI, Bhagabati NK, Braisted JC, LiangW, Sharov V, Howe EA, Li J, ThiagarajanM,WhiteJA,Quackenbush J. 2006. TM4microarray software suite. Methods Enzymol 411:134–193.

Severino A, Abbruzzese C, Manente L, Avivar Valderas A, Mattarocci S, Federico A, StaraceG, Chersi A, Mileo AM, Paggi MG. 2007. Human papillomavirus-16 E7 interacts with Siva-1and modulates apoptosis in HaCaT human immortalized keratinocytes. J Cell Physiol212:118–125.

Siegel R, Naishadham D, Jemal A. 2013. Cancer statistics, 2013. CA Cancer J Clin 63:11–30.Smyth GK, Michaud J, Scott HS. 2005. Use of within-array replicate spots for assessing

differential expression in microarray experiments. Bioinformatics 21:2067–2075.Tamada M, Suematsu M, Saya H. 2012. Pyruvate kinase m2: Multiple faces for conferring

benefits on cancer cells. Clin Cancer Res 18:5554–5561.Weinberg RA. 1995. The retinoblastoma protein and cell cycle control. Cell 81:323–330.Whiteside MA, Siegel EM, Unger ER. 2008. Human papillomavirus and molecular

considerations for cancer risk. Cancer 113:2981–2994.zur Hausen H. 2009. The search for infectious causes of human cancers: Where and why

(Nobel lecture). Angew Chem Int Ed Engl 48:5798–5808.Zwerschke W, Mazurek S, Massimi P, Banks L, Eigenbrodt E, Jansen-Durr P. 1999.

Modulation of type M2 pyruvate kinase activity by the human papillomavirus type 16 E7oncoprotein. Proc Natl Acad Sci USA 96:1291–1296.

Supporting Information

Additional supporting information may be found in the onlineversion of this article at the publisher’s web-site.



identification of pivotal cellular factors involved in hpv-induced dysplastic and neoplastic...

Documents