Analysis of Osteoblast-like MG63 Cells’ Response to a Rough Implant Surface by Means of DNA Microarray

RESEARCHANALYSIS OF OSTEOBLAST-LIKE MG63 CELLS'RESPONSE TO A ROUGH IMPLANT SURFACE BYMEANS OF DNA MICROARRAYFrancesco Carinci, MDFurio Pezzetti, PhDStefano Volinia, PhDFrancesca Francioso, PhDDiego Arcelli, PhDJlenia Marchesini, PhDLuca Scapoli, PhDAdriano Piattelli, MDKEY WORDSDNA microarrayExpression profilingImplant surfaceTitaniumFrancesco Carinci, MD, is an associateprofessor of maxillofacial surgery at theUniversity of Ferrara, Italy.Furio Pezzetti, PhD, is an associate professorof histology, Institute of Histology, Universityof Bologna, and Centre of Molecular Genetics,CARISBO Foundation, Italy.Stefano Volinia, PhD, is an assistantprofessor of histology; Francesca Francioso,PhD, is a student; Diego Arcelli, PhD, is apostdoctoral fellow; and Jlenia Marchesini,PhD, is a student in the Department ofMorphology and Embryology, University ofFerrara, Italy.Luca Scapoli, PhD, is a postdoctoral fellow atthe Institute of Histology, University ofFerrara, Italy.Adriano Piattelli, MD, is a full professor anddean of the Dental School, Via F. Sciucchi 63,66100 Chieti, Italy (e-mail: [email protected]).Address correspondence to Dr Piattelli.Several features of the implant surface, such as composition, topography,roughness, and energy, play a relevant role in implant integration with bone.Little is known about the structural and chemical surface properties that mayinfluence biological responses. Expression profiling by DNA microarray is amolecular technology that allows the analysis of gene expression in a cell system.By using DNA microarrays containing 19 200 genes, we identified several geneswhose expression was significantly down-regulated in osteoblast-like cell lineMG63 on a new implant surface (titanium pull spray superficial [TPSS] surface,Oralplant, Cordenons, PN, Italy). The differentially expressed genes cover a broadrange of functional activities: (1) signaling transduction, (2) translation, (3) cellcycle regulation, (4) structural and metabolic functions, and (5) apoptosis. It wasalso possible to detect some genes whose functions are unknown. The datareported can be relevant to better understand the role of the type of surface onthe molecular mechanism of implant osseointegration and as a model forcomparing other materials.S n h g u o , y h tosurfpograpace suchras compositess, anion,d ably result in a higher bone implanto m m 1.5 e th in es irregularitiNTRODUCTION Ieveral features of the implant mined, but it has been suggestedrange prthab- tenergy play a relevant role inimplant integration withbone.1-21 Little is knownabout the structural and chemical surfaceproperties that may influence biologicalresponses.1 Surface roughnesshas been demonstrated to have positiveeffects on adsorption of molecules,local factor production, and proliferationand differentiation of cells.13,14,22The optimal value for the implant surfaceroughness is still to be detercontactpercentage.4 Titanium has beenwidely used in the biomedical field,but the factors and mechanisms underlyingthe biological response to titaniumare not yet well understood,1 andit is necessary to look for correlationsbetween surface characteristics and responseof biologic tissues at differentlevels of resolution and sophistication.1DNA microarray is a moleculartechnology that enables the analysis ofgene expression in parallel with a veryJournal of Oral Implantology 215OSTEOBLAST CELL RESPONSE TO ROUGH IMPLANT SURFACElarge number of genes, spanning a significantfraction of the human genome.It is a qualitative analysis (eg, it candifferentiate each single gene) as wellas quantitative, since it has the sensitivityto detect a change of expressionlevel in the investigated cells whencompared with normal samples. Technically,reference RNA (eg, cells culturedon machined titanium) and investigatedRNA (eg, cells cultured ona specific titanium surface) are labeledafter reverse transcription with differentfluorescent dyes (Cy3 for the referencecells and Cy5 for the investigatedcells) and hybridized to a cDNA microarraycontaining robotically printedcDNA fragments. The slides are thenscanned with a laser scanning system,and 2 false color images are generatedfor each hybridization with RNA fromthe investigated and reference cells.Genes up-regulated in the investigatedcells are conventionally designated red,whereas those with decreased expressionappear green, since more are expressedin the normal sample. Geneswith similar levels of expression in the2 samples appear yellow. The overallresult is the generation of a so-calledgenetic portrait.23,24In the present study we defined theeffects of a new implant surface ongenes by using an osteoblast-like cellline (MG63) and microarray slides containing19 200 different oligonucleotides.MATERIALS AND METHODSCell cultureOsteoblast-like cells (MG63) were culturedin sterile Falcon wells (BectonDickinson, NJ) containing Eagle's minimumessential medium (MEM) supplementedwith 10% fetal calf serum(FCS; Sigma Chemical Co, St Louis,Mo) and antibiotics (Penicillin 100 U/mL and Streptomycin 100 mg/mL; Sigma).Cultures were maintained in a 5%CO2 humidified atmosphere at 378C.MG63 cells were collected andseeded at a density of 1 3 105 cells/mL into 9 cm2 (3 mL) wells by using216 Vol. XXIX/No. Five/20030.1% trypsin, 0.02% EDTA in Ca11,and Mg-free Eagle's buffer for cell release.One set of wells contained sterilemetal disks of machined grade 3 titanium(diameter 3 cm, control implants),whereas another contained titaniumimplants with a new surface(titanium pull spray superficial [TPSS]surface; Oralplant, Cordenons, PN, Italy)covering the same area (35 cm2).This new surface was producedthrough micromechanical removal ofparts of the superficial oxide layer withthe use of aluminum oxide 0.5 mm/micropoints.The Ra value was 0.30 forthe machined surface and 2.74 for theTPSS surface. After 24 hours, the medium(3 mL of MEM with 10% FCS)was changed. Finally, after 24 hoursthe cultures reached subconfluenceand the cells were processed for RNAextraction.DNA microarrays screeningand analysisRNA was extracted from the cells byusing RNAzol. Ten micrograms of totalRNA were used for each sample. ComplementaryDNA was synthesized byusing Superscript II (Life Technologies,Invitrogen, Milano, Italy) and aminoallyldUTP (Sigma).Mono-reactive Cy3and Cy5 esters (Amersham Pharmacia)were used for indirect cDNA labeling.RNA extracted from cells grown onmachined titanium disks was labeledwith Cy3 and used as control againstthe Cy5-labeled, treated (TPSS titaniumsurface) cDNA in the first experimentand then switched. Human 19.2K DNA microarrays were used (OntarioCancer Institute). For 19.2 K slides,100 mL of the sample and controlcDNAs in DIG Easy hybridization solution(Roche) were used in a sandwichhybridization of the 2 slides, whichconstituted the 19.2. K set at 378C overnight.Washing was performed 3 timesfor 10 minutes with 1 3 sodium dodecylsulfate (SSC), 0.1% SSC at 428C,and 3 times for 5 minutes with 0.1 3SSC at room temperature. Slides weredried by centrifugation for 2 minutesat 2000 rpm. The experiment was repeatedtwice and the dyes switched. AGenePix 4000a DNA microarrays scanner(Axon, Union City, Calif) was usedto scan the slides, and data were extractedwith GenePix Pro. After removinglocal background, genes with expressionlevels of less than 1000 werenot included in the analysis, since ratiosare not reliable at that detectionlevel.23,24RESULTSDNA microarraysAfter scanning the 2 slides containingthe 19 200 human genes in duplicate,local background was calculated foreach target location. A normalizationfactor was estimated from ratios of median.Normalization was performed byadding the log2 of the normalizationfactor to the log2 of the ratio of medians.The log2 ratios for all the targetson the array were then calibrated usingthe normalization factor, and log2 ratiosoutside the 99.7% confidence interval(the median 63 times the SD 5 0.52)were determined as significantlychanged in the treated cells. Thusgenes are significantly modulated inexpression when the absolute value oftheir log2 expression level is higherthan 1.56, or else there is a threefolddifference in expression between treatedcells and the reference. GenePix Prosoftware was used to report genesabove the threshold and with less than10% difference in 3 different statisticalevaluations of the intensity ratio, thuseffectively enabling an automatedquality control check of the hybridizedspots. Furthermore, all of the positivelypassed spots were finally visually inspected.Generally, no further exclusionwas needed at this final step, sincethe previously applied automated selectionfilters were highly stringent.23,24The significance analysis of microarray(SAM) program was then performed,and a SAM score was obtained(t-statistic value). The down-regulatedgenes differentially expressed in themachined and the TPSS implant sur-UG ClusterCloneIDHs. 850Hs. 154583Hs. 25882Hs. 24049Hs. 79946Hs. 239176398842063404030624500213303747197Hs. 28346Hs. 282847Hs. 279609Hs. 7731Hs. 75511Hs. 82587149046229878339804007626826037166Hs. 343696Hs. 61271Hs. 50651Hs. 1901Hs. 240112Hs. 187958342113231010488877245662222237235958Hs. 112255Hs. 283771Hs. 105749Hs. 246857Hs. 8024Hs. 8966620634512494111335742170382694279488Hs. 10760Hs. 308026Hs. 37636Hs. 154145Hs. 57301Hs. 172870489344240655246497233177665047488246Hs. 20935Hs. 60389Hs. 20799Hs. 408754Hs. 269084Hs. 8241525396430255239524230507244896277063Hs. 196008Hs. 302965Hs. 381543231461241587161788234697 Hs. 407926239534 Hs. 37560113275 Hs. 409343200272 Hs. 347143212331 Hs. 205980face are reported in Tables 1 and 2 andin the Figure.DISCUSSIONHybridization of mRNA-derived probesto cDNA microarrays allows us to performsystemic analysis of expressionprofiles for thousands of genes simultaneouslyand to provide primary informationon transcriptional changesrelated to an implant surface (TPSSsurface; Oralplant). We identified severalgenes whose expression was defi-nitely down-regulated (Table 1).TABLE 1NameIMP (inosine monophosphate) dehydrogenase 1RNA binding motif protein 10Gemin 5Golgi autoantigen, golgin subfamily a, 2Cytochrome P450, subfamily XIX (aromatization of androgens)Insulin-like growth factor 1 receptorGlial cells missing homolog 1 (Drosophila)Pregnancy specific b-1-glycoprotein 1Mitochondrial carrier homolog 2Uncharacterized bone marrow protein BM036Connective tissue growth factorPhospholipase D1, phophatidylcholine-specificTrigger transposable element derived 7Hypothetical protein FLJ21159Janus kinase 1 (a protein tyrosine kinase)Kallikrein B, plasma (Fletcher factor) 1KIAA0276 proteinSolute carrier family 6 (neurotransmitter transporter, creatine),member 8Nucleoporin 98kDaChromosome 21 open reading frame 66KIAA0553 proteinMitogen-activated protein kinase 9IK cytokine, down-regulator of HLA IIA kinase (PRKA) anchor protein 6Asporin (LRR class 1)Major histocompatibility complex, class II, DR b 3Hypothetical protein FLJ25162Metallo phosphoesteraseMutL Homolog 1, colon cancer, nonpolyposis type 2 (E coli)KIAA1913 proteinHypothetical protein DKFZp761D221ESTsESTsESTsESTsESTsESTsESTsEsts, Moderately similar to PRO0478 protein (Homo sapiens, H sapiens)ESTsESTsESTsESTs, Weakly similar to hypothetical protein FLJ20294 (Homo sapiens, H sapiens)ESTsTransductionThe insulin-like growth factor 1 receptor(IGF1R) binds insulin-like growthfactor with a high affinity. It has tyrosinekinase activity and plays a criticalrole in transformation events. Cleavageof the precursor generates a and b subunits.It is highly overexpressed inmost malignant tissues, where it functionsas an antiapoptotic agent by enhancingcell survival. It is down-regulatedby this implant surface.25 Also,the connective tissue growth factorSymbolIMPDH1RBM10GEMIN5GOLGA2CYP19IGF1RGCM1PSG1MTCH2BM036CTGFPLD1TIGD7FLJ21159JAK1KLKB1KIAA0276SLC6A8NUP98C21orf66KIAA0553MAPK9IKAKAP6ASPNHLA-DRB3FLJ2516MPPE1MLH1KIAA1913DKFZp761D221(CTGF) is down-regulated by the TPSSsurface. CTGF binds IGF and may havea role in regulating normal and neoplasticcell growth.26Janus kinase 1 (JAK1), is a memberof a new class of protein-tyrosine kinases(PTK), is characterized by thepresence of a second phosphotransferase-related domain immediately N-terminalto the PTK domain. The secondphosphotransferase domain bears allthe hallmarks of a protein kinase, althoughits structure differs significantlyfrom that of the PTK and threonine/Journal of Oral Implantology 217Francesco Carinci et alScoreChromosome7q31.3-q32Xp11.235q349q34.1315q21.124.270623.4021523.0211122.8953722.8340322.66806 15q25-q266p21-p1219q13.211q12.114q126q23.122.596122.5692322.4911422.4524922.2113322.2003 3q2616p13.114q31.31p32.3-p31.34q34-q3522.1526222.1307822.1165422.0881522.0796622.070244p12Xq2811p15.521q21.317q21.315q355q31.314q1222.0261122.0114722.0015822.0006821.9883521.9803321.9587921.9353121.930949q226p21.310q22.218p11.21-p11. 21.922613p21.3 21.9212921.9094321.880866q23.11p31.223.2472722.6386922.4405222.3607122.3486122.16122.0304822.0221721.9400721.9392921.8964421.8805721.82693OSTEOBLAST CELL RESPONSE TO ROUGH IMPLANT SURFACE2 scale. A normalization factor2 of2 of the ratio of medians. The log2 ratios for all the targetsFIGURE. Result of 1 of the 2 slides containing 19.200 genes: 23 different genes are signifi-cantly down-regulated (in the lower left square). The y-axis displays the observed cases,whereas the x-axis depicts the expected. The graphic has a logwas estimated from ratios of median. Normalization was performed by adding the logthe normalization factor to the logon the array were then calibrated using the normalization factor, and log2 ratios outsidethe 99.7% confidence interval (the median 63 times the SD 5 0.52) were determined assignificantly changed.serine kinase family members. JAK1 isa large, widely expressed membraneassociatedphosphoprotein. JAK1 is involvedin the interferon-a, -b, and -gsignal transduction pathways. The reciprocalinterdependence betweenJAK1 and TYK2 activities in the interferon-a pathway, and between JAK1and JAK2 in the interferon-g pathway,may reflect a requirement for these kinasesin the correct assembly of interferonreceptor complexes. These kinasescouple cytokine ligand binding totyrosine phosphorylation of variousknown signaling proteins and of aunique family of transcription factors,which are termed the signal transducersand activators of transcription, orSTATs.27TranslationSignal-mediated nuclear import andexport proceeds through the nuclearpore complex, which is comprised ofapproximately 50 unique proteins collectivelyknown as nucleoporins. The98 kD nucleoporin (NUP98) functionsas 1 of several docking site nucleopor-218 Vol. XXIX/No. Five/2003ins of transport substrates and isdown-regulated by the new implantsurface.28Cell cycleInosine-59-monophosphate dehydrogenase(IMPDH1) catalyzes the formationof xanthine monophosphate (XMP)from IMP. In the purine de novo syntheticpathway, IMP dehydrogenase ispositioned at the branch point in thesynthesis of adenine and guanine nucleotidesand is thus the rate-limitingenzyme in the de novo synthesis ofguanine nucleotides. Inhibition of cellularIMP dehydrogenase activity resultsin an abrupt cessation of DNAsynthesis and a cell cycle block at theG1-S interface.29Enzyme and cytoskeletonPhosphatidylcholine (PC)-specific phospholipasesD (PLDs) catalyze the hydrolysisof PC to produce phosphatidicacid and choline. A range of agonistsacting through G protein-coupled receptorsand receptor tyrosine kinasesstimulate this hydrolysis. PC-specificPLD activity has been implicated innumerous cellular pathways, includingsignal transduction, membrane traf-ficking, and the regulation of mitosis.Hammond et al30 showed that recombinantPLD1 activity is located both inthe cytoplasm and in association withthe membrane; they suggested thatPLD1 can exist as a stable soluble proteinand that controlled interactionwith substrate-containing phospholipidsurfaces may be a physiologicallyimportant mode of regulation. Thesame authors found that ADP-ribosylationfactor-1 (ARF1) activates PLD1,suggesting that PLD1 is involved in intravesicularmembrane trafficking.The A-kinase anchor proteins(AKAPs) are a group of structurally diverseproteins, which have the commonfunction of binding to the regulatorysubunit of protein kinase A(PKA) and confining the holoenzymeto discrete locations within the cell.AKAP6 encodes a member of theAKAP family. The encoded protein ishighly expressed in various brain regionsand cardiac and skeletal muscle.It is specifically localized to the sarcoplasmicreticulum and nuclear membraneand is involved in anchoringPKA to the nuclear membrane or sarcoplasmicreticulum.31ApoptosisMAPK9 is a member of the mitogenactivatedprotein (MAP) kinase family.MAP kinases act as an integrationpoint for multiple biochemical signalsand are involved in a wide variety ofcellular processes such as proliferation,differentiation, transcription regulation,and development. This kinase targetsspecific transcription factors, andthus mediates immediate-early geneexpression in response to various cellstimuli. It is most closely related toMAPK8, both of which are involved inultraviolet (UV) radiation-induced apoptosis,which is thought to be relatedto the cytochrome c-mediated celldeath pathway. This gene and MAPK8are also known as c-Jun N-terminal kinases.This kinase blocks the ubiquitin-ation of tumor suppressor p53, andthus it increases the stability of p53 innonstressed cells.32The genes discussed are only a limitednumber among those differentiallyexpressed and reported in Table 1. Webriefly analyzed some of those with abetter-known function. There are alsoa number of expressed sequence tags(EST) whose function is unknown (Table2). However, the recognition of alteredexpression of specific EST canhelp give it a role.In conclusion, we have shown thatthe new implant surface is able tomodulate the expression of some genesthat cover a broad range of functionalactivities: (1) signaling transduction,(2) translation, (3) cell cycle regulation,(4) enzyme and cytoskeleton development,and (5) apoptosis.Recent studies on animal modelshave demonstrated that titanium surfacesare of paramount importance in in-fluencing the timing of bone healing,and rougher surfaces have been demonstratedto present a higher quantityof bone-implant contact percentage andhigher removal torque values.6,7,10,11,22The surface roughness has been demonstratedto alter the responsiveness ofdifferent types of cells.8,13,14 However,because the in vitro system differs fromthe in vivo system, more investigationsare needed. Indeed, MG63 are osteoblast-like cells and not normal osteoblast,and a monolayer cell stratum differssignificantly from bone tissue,where osteoblasts are resident in a bonematrix. Moreover, some differences betweenthe genetic portrait reported andgenes described by different authorsmay be related to the chosen time pointof the experiment. We chose to performthe experiment after 24 hours of stimulationin order to get information onthe early stages of stimulation. It is ourbelief, however, that more investigationswith different osteoblast-like cell lines,primary cultures, and time points areneeded in order to get a better comprehensionof the molecular events relatedto the interaction between the surfaceand the osseointegration process. Finally,we believe that the reported data canbe a model to compare different implantsurfaces.ACKNOWLEDGMENTSThis work was supported by grantsfrom Unife 60% (F.C.), Guya-Bioscience,and the CARISBO Foundation(F.P.).REFERENCES1. Larsson C, Thomsen P, AronssonBO, Rodahl M, Lausmaa J, KasemoB, Ericsson LE. Bone response to surface-modified titanium implants: studieson the early tissue response to machinedand electropolished implantswith different oxide thicknesses. Biomaterials.1996;17:605-610.2. Puleo DA, Nanci A. Understandingand controlling the bone-implantinterface. Biomaterials. 1999;20:2311-2321.3. Bowers KT, Keller J, RandolphBA, Wick DG, Michaels CM. Optimizationof surface micromorphology forenhanced osteoblast responses in vitro.Int J Oral Maxillofac Implants. 1992;7:302-310.4. Han CH, Johansson CB, WennenbergA, Albrektsson T. Quantitativeand qualitative investigations of surfaceenlarged titanium and titaniumalloys implants. Clin Oral Implant Res.1998;9:1-10.5. Cooper LF, Masuda T, WhitsonW, Yliheikkila P, Felton DA. Formationof mineralizing osteoblast cultures onmachined, titanium oxide grit-blasted,and plasma-sprayed titanium surfaces.Int J Oral Maxillofac Implants. 1999;14:37-47.6. Klokkevold P, Nishimura RD,Adachi M, Caputo A. Osseointegrationenhanced by chemical etching of the titaniumsurface. A torque removalstudy in the rabbit. Clin Oral ImplantRes. 1997;8:442-447.7. Gotfredsen K, Wennerberg A,Johansson C, Skovgaard LT, Hjorting-Hansen E. Anchorage of TiO2-blasted, crogrooves and surface energy on cellHA-coated and machined implants: anexperimental study with rabbits. J Bio- 511-518.med Mater Res. 1995;29:1223-1231.Francesco Carinci et al8. Park JY, Davies JE. Red bloodcell and platelet interactions with titaniumimplant surfaces. Clin Oral ImplantRes. 2000;11:530-539.9. Mustafa K, Wroblewski J, HultenbyK, Silva Loprez B, Arvidson K.Effects of titanium surfaces blastedwith TiO2 particles on the initial attachmentof cells derived from mandibularbone. A scanning electron microscopicand histomorphometric analysis.Clin Oral Implant Res. 2000;11:116-128.10. Wennerberg A, Albrektsson T,Lausmaa J. Torque and histomorphometricevaluation of c.p. titaniumscrews blasted with 25- and 75 m-sizedparticles of Al2O3. J Biomed Mater Res.1996;30:251-260.11. Wennerberg A, Albrektsson T,Johansson C, Andersson B. Experimentalstudy of turned and grit-blastedscrew-shaped implants with specialemphasis on effects of blasting materialand surface topography. Biomaterials.1996;17:15-22.12. Mustafa K, Lopez BS, HultenbyK, Wennenberg A, Arvidson K. Attachmentand proliferation of human oralfibroblasts to titanium surfaces blastedwith TiO2 particles. A scanning electronmicroscopic and histomorphometricanalysis. Clin Oral Implant Res. 1998;9:195-207.13. Schwartz Z, Martin JY, DeanDD, Simpson J, Cochran DL, Boyan BD.Effect of titanium surface roughness onchondrocyte proliferation, matrix production,and differentiation dependson the state of cell maturation. J BiomedMater Res. 1996;30:145-155.14. Martin JY, Schwartz Z, HummertTW, et al. Effect of titanium surfaceroughness on proliferation, differentiation,and protein synthesis of humanosteoblast-like cells (MG 63). JBiomed Mater Res. 1995;29:389-401.15. Den Braber ET, De Ruijter JE,Smits HTJ, Ginsel LA, Von Recum AF,Jansen JA. Effect of parallel surface migrowth.J Biomed Mater Res. 1995;29:16. Wennenberg A, Hallgren C, Jo-Journal of Oral Implantology 219OSTEOBLAST CELL RESPONSE TO ROUGH IMPLANT SURFACEhansson C, Danelli S. A histomorphometricevaluation of screw-shaped implantseach prepared with two surfaceroughnesses. Clin Oral Implant Res.1998;9:11-19.17. Feighan JE, Goldberg VM,Davy D, Parr JA, Stevenson S. The in-fluence of surface-blasting on the incorporationof titanium-alloy implantsin a rabbit intramedullary model. JBone Joint Surg. 1995;77-A:1380-1395.18. Chehroudi B, McDonnel D,Brunette DM. The effects of micromachinedsurfaces on formation of boneliketissue on subcutaneous implantsas assessed by radiography and computerimage processing. J Biomed MaterRes. 1997;34:279-290.19. Piattelli A, Manzon L, ScaranoA, Paolantonio M, Piattelli M. Histologicand morphologic analysis of thebone response to machined and sandblastedtitanium implants: an experimentalstudy in rabbit. Int J Oral MaxillofacImplants. 1998;13:805-810.20. Piattelli M, Scarano A, PaolantonioM, Iezzi G, Petrone G, Piattelli A.Bone response to RBM sandblasted titaniumimplants: an experimentalstudy in rabbit. J Oral Implantol. 2002;28:2-8.220 Vol. XXIX/No. Five/200321. Wennenberg A, Albrektsson T,Andersson B. An animal study of c.p.titanium screws with different surfacetopographies. J Mater Sci: Mater M.1995;6:302-309.22. Gotfredsen K, Berglundh T,Lindhe J. Anchorage of titanium implantswith different surface characteristics:an experimental study in rabbits.Clin Implant Dent Relat Res. 2000;2:120-129.23. Francioso F, Carinci F, Tosi L, etal. Identification of differentially expressedgenes in human salivary glandtumors by DNA microarrays. Mol CancerTher. 2002;1:533-538.24. Carinci F, Francioso F, RubiniC, et al. Genetic portrait of malignantgranular cell odontogenic tumour. OralOncol. 2003;39:69-77.25. Werner H, Karnieli E, RauscherFJ III, LeRoth D. Wild-type and mutantp53 differentially regulate transcriptionof the insulin-like growth factor Ireceptor gene. Proc Nat Acad Sci. 1996;93:8318-8323.26. Kim HS, Nagalla SR, Oh Y,Wilson E, Roberts CT Jr, Rosenfeld RG.Identification of a family of low-affinityinsulin-like growth factor binding proteins(IGFBPs): characterization of connectivetissue growth factor as a memberof the IGFBP superfamily. Proc NatlAcad Sci. 1997;94:12981-12986.27. Ihle JN. Cytokine receptor signaling.Nature. 1995;377:591-594.28. Enninga J, Levy DE, Blobel G,Fontoura BMA. Role of nucleoporin inductionin releasing an mRNA nuclearexport block. Science. 2002;295:1523-1525.29. Natsumeda Y, Ohno S, KawasakiH, Konno Y, Weber G, Suzuki K.Two distinct cDNAs for human IMPdehydrogenase. J Biol Chem. 1990;265:5292-5295.30. Hammond SM, Altshuller YM,Sung TC, et al. Human ADP-ribosylationfactor-activated phosphatidylcholine-specific phospholipase D defines anew and highly conserved gene family.J Biol Chem. 1995;270:29640-29643.31. Kapiloff MS, Schillace RV,Westphal AM, Scott JD. mAKAP: an Akinaseanchoring protein targeted tothe nuclear membrane of differentiatedmyocytes. J Cell Sci. 1999;112:2725-2736.32. Tournier C, Hess P, Yang DD, etal. Requirement of JNK for stress-inducedactivation of the cytochrome cmediateddeath pathway. Science. 2000;288:870-874.