Homeobox gene expression profile indicates HOXA5 as a candidate prognostic marker in oral squamous cell carcinoma

INTERNATIONAL JOURNAL OF ONCOLOGY

+RPHRER[JHQHH[SUHVVLRQSURÀOHLQGLFDWHVHOXA5 as a FDQGLGDWHSURJQRVWLFPDUNHULQRUDOVTXDPRXVFHOOFDUFLQRPD CAMILA OLIVEIRA RODINI1, FLÁVIA CALÓ AQUINO XAVIER1.$7,Ô&,$%$7,67$6,/9$3$,9$1, MARIA FERNANDA DE SOUZA SETÚBAL DESTRO1, RAQUEL AJUB MOYSES23('520,&+$/8$57(2, MARCOS BRASILINO CARVALHO3, ERICA ERINA FUKUYAMA4+($'$1'1(&.*(120(352-(&7 *(1&$325, ELOIZA HELENA TAJARA6,7, OSWALDO KEITH OKAMOTO8 and FABIO DAUMAS NUNES1 1

/DERUDWRU\RI0ROHFXODU3DWKRORJ\'HSDUWPHQWRI2UDO3DWKRORJ\6FKRRORI'HQWLVWU\8QLYHUVLW\RI6mR3DXOR $Y3URI/LQHX3UHVWHV&LGDGH8QLYHUVLWiULD6mR3DXOR2Department of Head and Neck Surgery, 6FKRRORI0HGLFLQH8QLYHUVLW\RI6mR3DXOR$Y'U$UQDOGR&HUTXHLUD&pVDU6mR3DXOR 3 +HDGDQG1HFN6XUJHU\'LYLVLRQ+HOLySROLV+RVSLWDO&RPSOH[5XD&{QHJR;DYLHU6DFRPm6mR3DXOR 4 'HSDUWPHQWRI+HDGDQG1HFN6XUJHU\$UQDOGR9LHLUDGH&DUYDOKR&DQFHU,QVWLWXWH5XD'U&HViULR0RWWD-XQLRU 9LOD%XDUTXH6mR3DXOR5$XWKRUOLVWDQGDGGUHVVHVSUHVHQWHGLQWKH$SSHQGL[6Department of Molecular Biology, 6mR-RVpGR5LR3UHWR6FKRRORI0HGLFLQH$Y%ULJDGHLUR)DULD/LPD9LOD6mR3HGUR6mR-RVpGR5LR3UHWR 7 'HSDUWPHQWRI*HQHWLFVDQG(YROXWLRQDU\%LRORJ\,QVWLWXWHRI%LRVFLHQFHV8QLYHUVLW\RI6mR3DXOR5GR0DWmR WUDYHVVD&LGDGH8QLYHUVLWiULD6mR3DXOR8Human Genome Research Center, Department of Genetics, %LRVFLHQFHV,QVWLWXWH8QLYHUVLW\RI6mR3DXOR5XDGR0DWmRWUDY&LGDGH8QLYHUVLWiULD6mR3DXOR%UD]LO 5HFHLYHG6HSWHPEHU$FFHSWHG1RYHPEHU '2,LMR Abstract. The search for molecular markers to improve diagnosis, individualize treatment and predict behavior of tumors has been the focus of several studies. This study aimed to DQDO\]HKRPHRER[JHQHH[SUHVVLRQSURÀOHLQRUDOVTXDPRXV cell carcinoma (OSCC) as well as to investigate whether some of these genes are relevant molecular markers of prognosis and/ or tumor aggressiveness. Homeobox gene expression levels ZHUHDVVHVVHGE\PLFURDUUD\VDQGT573&5LQ26&&WLVVXHV and adjacent non-cancerous matched tissues (margin), as well as in OSCC cell lines. Analysis of microarray data revealed the expression of 147 homeobox genes, including one set of six at least 2-fold up-regulated, and another set of 34 at least 2-fold GRZQUHJXODWHGKRPHRER[JHQHVLQ26&&$IWHUT573&5 assays, the three most up-regulated homeobox genes (HOXA5, HOXD10 and HOXD11) revealed higher and statistically VLJQLÀFDQWH[SUHVVLRQOHYHOVLQ26&&VDPSOHVZKHQFRPSDUHG WRPDUJLQV3DWLHQWVSUHVHQWLQJORZHUH[SUHVVLRQRIHOXA5 had poorer prognosis compared to those with higher expression 3  $GGLWLRQDOO\WKHVWDWXVRIHOXA5, HOXD10 and HOXD11 expression levels in OSCC cell lines also showed a

Correspondence to:3URIHVVRU)DELR'DXPDV1XQHV)DFXOGDGHGH 2GRQWRORJLD8QLYHUVLGDGHGH6mR3DXOR$Y3URIHVVRU/LQHX3UHVWHV &LGDGH8QLYHUVLWiULD&(36mR3DXOR%UD]LO E-mail: [email protected]

Key words: oral cancer, homeobox genes, HOXA5, prognosis

VLJQLÀFDQWXSUHJXODWLRQZKHQFRPSDUHGWRQRUPDORUDONHUDWLQRF\WHV5HVXOWVFRQÀUPWKHSUHVHQFHRIWKUHHVLJQLÀFDQWO\ upregulated (>4-fold) homeobox genes (HOXA5, HOXD10 and HOXD11 LQ26&&WKDWPD\SOD\DVLJQLÀFDQWUROHLQWKHSDWKRgenesis of these tumors. Moreover, since lower levels of HOXA5 predict poor prognosis, this gene may be a novel candidate for development of therapeutic strategies in OSCC. Introduction 2UDO VTXDPRXV FHOO FDUFLQRPD 26&&  LV WKH VL[WK PRVW common cancer worldwide and often invades tissues locally and metastasizes to cervical lymph nodes (1-3). Oncogene overexpression or inactivation mechanisms on tumor suppressor genes through mutations, loss of heterozygosity, deletions, or epigenetic modifications have been the major factors in its development, local invasion and local metastasis (1,4). The homeobox genes encode transcription factors that acts either by activating or repressing downstream target genes essential to cell growth and differentiation. It is estimated that WKHKXPDQJHQRPHLQFOXGHVDWOHDVWKRPHRER[JHQHVRI which belong to the HOX family. These genes are functionally important during embryonic morphogenesis and also regulate the adult tissue architecture, identity and homeostasis, cell-cell interactions and cell-extracellular matrix interactions (5). In cancer, normal HOX gene expression is disrupted, affecting various pathways that promote tumorigenesis and metastasis, including the activation of anti-apoptotic pathways and suppression of differentiation (6). HOX genes have been found to be aberrantly expressed in a variety of solid tumors such DVO\PSKRPD PHODQRPD EUHDVW HQGRPHWULDO

2

RODINI et al+20(2%2;*(1((;35(66,21352),/(,125$/&$1&(5

 OLYHU OXQJ WK\URLG DQGHVRSKDJXVFDQFHU (16). Aberrant expression of HOX genes was also observed in 26&&KRZHYHUKRZWKH\FRQWULEXWHWRRUDOFDQFHUSKHQRW\SH DQGLWVWLVVXHVSHFLÀFIHDWXUHVUHPDLQVXQFOHDU  Detection of OSCC is currently based on expert clinical examination and histological analysis of suspicious areas, but it may be undetectable in hidden sites. Therefore, sensitive and specific biomarkers for OSCC may be helpful to screening KLJKULVNSDWLHQWV :KLOHVHYHUDOVWXGLHVSURSRVHGWKHLGHQWLÀFDWLRQRIJHQHH[SUHVVLRQSDWWHUQVLQKHDGDQGQHFNFDQFHU  MXVWDIHZLQYHVWLJDWHGWKHGLIIHUHQWLDOH[SUHVVLRQSURÀOH of homeobox genes family in OSCC (17,25-27) as well as their correlation to tumor behavior, clinical parameters and survival UDWHV REWDLQLQJVLJQLÀFDQWUHVXOWV Thus, the purpose of the present study was to search for distinct pattern of homeobox gene expression through a genomewide analysis. Some up- and down-regulated homeobox genes ZHUHFKRVHQIRUIXUWKHUYDOLGDWLRQE\T573&5DQGFRUUHODWHG with prognosis. Up-regulated homeobox genes were also validated on OSCC cell lines. 0DWHULDOVDQGPHWKRGV Tissue samples. Specimens were obtained during surgical UHVHFWLRQIURPSDWLHQWVDJHG•\HDUVDGPLWWHGIRUGLDJQRVLV and treatment at Arnaldo Vieira de Carvalho Cancer Institute, +RVSLWDO+HOLySROLVDQG+RVSLWDOGDV&OtQLFDVRI6mR3DXOR University Medical School. Histopathological diagnosis was performed according to the WHO classification of WXPRUVE\WKH'HSDUWPHQWRI3DWKRORJ\RIHDFK,QVWLWXWLRQ Clinicopathological staging was determined by the TNM FODVVLÀFDWLRQRIWKH,8&& 7KHVWXG\ZDVDSSURYHGE\ the Ethics Committee of each Institution and was based on the criteria of the Helsinki convention. Fresh surgical samples of primary OSCC and their corresponding non-neoplastic margin tissues were immediately VQDSIUR]HQLQOLTXLGQLWURJHQXSRQVXUJLFDOUHPRYDO$IWHU KLVWRORJLFDOFRQÀUPDWLRQDOOWLVVXHVDPSOHVZHUHFKHFNHGSULRU to RNA extraction so that each OSCC sample contained at least WXPRUFHOOVDQGWKHFRUUHVSRQGLQJVXUJLFDOPDUJLQVZHUH UHSRUWHGDV¶WXPRUIUHH·*(1&$32+HDGDQG1HFN*HQRPH 3URMHFW &RQVRUWLXPZDVUHVSRQVLEOHIRUVDPSOHFROOHFWLRQDQG initial processing, clinical data collection, providing of histopathological analysis of tissue samples, and informed consent DFTXLVLWLRQRIHDFKSDWLHQW Cell lines and cell culture. 6&&   DQG  26&& cell lines) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and kindly provided E\ 3URIHVVRU 5LFDUGR 'HOOD &ROHWWD 6FKRRO RI 'HQWLVWU\ 81,&$03 26&&DVZHOODV+D&DWFHOOOLQHVZHUHJURZQDV described previously (25). Normal oral keratinocytes (NOK) were obtained from oral epithelial fragments under enzymatic digestion method, kindly provided by Dr Maria Fatima Guarizo .OLQJEHLO  RNA extraction and cDNA synthesis. Total cellular RNA was extracted using TRIzol® Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions and the RNA integrity was evaluated based on the intensity of 28S and 18S

U51$EDQGVLQDJDURVHJHOVDQGRQ$UDWLREHWZHHQ DQG RNA obtained from tissue samples (1 µg) and cell lines (4 µg) was reverse transcribed to single-stranded cDNA using High Capacity cDNA Reverse Transcription kit (Applied Biosystems™, Foster City, CA, USA) and Superscript III™ with oligo(dT) primers and RNase OUT (Invitrogen), respectively, after incubation with DNAse I (Invitrogen). Microarray hybridization. Ten tissue samples of primary 26&&RIWRQJXHDQGÁRRURIWKHPRXWKDVZHOODVDSRRORIQRQ neoplastic surgical margins were used for microarray analysis. Experiments were carried out as described in Severino et al  XVLQJ&RGH/LQN:KROH*HQRPH%LRDUUD\V*(+HDOWKFDUH 3LVFDWDZD\1-86$ UHSUHVHQWLQJKXPDQWUDQVFULSWV DQGDUUD\VZHUHVFDQQHGRQD*HQH3L[%$UUD\6FDQQHU (Axon Instruments), according to the recommended scanning procedures and settings. The data were treated with Code-Link IHDWXUHH[WUDFWLRQVRIWZDUHY$QRUPDOL]HGVLJQDOIRUHDFK WUDQVFULSWZDVREWDLQHGWKURXJKTXDQWLOHQRUPDOL]DWLRQ )RU global homeobox gene expression visualization, a hierarchical clustering using the Euclidean distance and the average linkage algorithm was performed (MeV® MultiExperiment Viewer software version 4.1, Boston, MA, USA) (32,33). Individual homeobox gene expression profile in OSCC samples and their respective non-neoplastic oral tissues were compared with each other. Differentially expressed genes were LGHQWLÀHGE\FDOFXODWLQJWKHUDWLRRIWKHPHDQQRUPDOL]HGÁXRrescence values obtained from each sample group. Results were expressed as fold variation, and genes displaying greater than 2-fold changes in transcript abundance in all tumor samples ZHUHVHOHFWHG7KHDUUD\GHVLJQDQGUDZGDWDÀOHVDUHDYDLODEOH at the Gene Expression Omnibus database (GEO) under the DFFHVVLRQQXPEHU*6(7KHPRVWXSUHJXODWHGKRPHRER[ JHQHVZHUHVHOHFWHGDQGDQDO\]HGE\T573&5 qRT-PCR. Samples of OSCC tissues and non-neoplastic margins were assessed for the expression levels of selected homeobox genes (HOXA5Q HOXD10 and HOXD11Q  7KHVDPH was performed for all cell lines described above. Endogenous housekeeping gene coding for the hypoxanthine guanine phosphoribosyltransferase gene (HPRT 10B )FFDFFDFFFWJWWJFWJWDDQG5WFFFFWJWWJDFWJJWFDWES ZDV XVHGIRUGDWDQRUPDOL]DWLRQDQGUHODWLYHTXDQWLÀFDWLRQZDV SHUIRUPHGXVLQJUHODWLYHVWDQGDUGFXUYHDQDO\VLVZLWKD UHDOWLPH3&56\VWHP$SSOLHG%LRV\VWHPV)RVWHU&LW\&$  $PSOLÀFDWLRQRIVSHFLÀF3&5SURGXFWVZDVGHWHFWHGXVLQJWKH 6