LeukemiaResearch36 (2012) 607–618
ContentslistsavailableatSciVerseScienceDirect
Leukemia
Research
j ou rn a l h o m e pa g e : w w w . e l s e v i e r . c o m / l o c a t e / l e u k r e s
Proteomic
analysis
identifies
differentially
expressed
proteins
in
AML1/ETO
acute
myeloid
leukemia
cells
treated
with
DNMT
inhibitors
azacitidine
and
decitabine
Francesca
Buchi,
Elena
Spinelli,
Erico
Masala,
Antonella
Gozzini,
Alessandro
Sanna,
Alberto
Bosi,
Germano
Ferrari,
Valeria
Santini
∗FunctionalUnitofHematology,UniversityofFlorence,Florence,Italy
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received5August2011
Receivedinrevisedform28October2011 Accepted29November2011
Available online 9 January 2012
Keywords: AML Proteomics Azacitidine Decitabine
a
b
s
t
r
a
c
t
AzacitidineanddecitabineareDNAmethyltransferaseinhibitorsusedtotreatmyelodysplastic
syn-dromesandacutemyeloidleukemias.Tofurthercharacterizedifferentmechanismsbetweenthesetwo
agents,cellularextractsfromleukemiccellsuntreatedortreatedwitheitherdrugwereanalyzedusing
2Delectrophoresis.Numerousdifferentiallyexpressedproteinswereidentifiedwith
MALDI-TOF/TOF-MS.CyclophilinA,Catalase,NucleophosminandPCNAweredecreasedexclusivelybyazacitidine,TCP1
andhnRNPA2/B1bybothdrugs;alpha-EnolaseandPeroxiredoxin-1bydecitabine.Interestingly,the
expressionoftheproinflammatoryproteinCyclophilinA,alsosuggestedasmarkerofcellnecrosis,was
stimulatedbydecitabine.Finally,acomprehensivepathwayanalysisofdatahighlightedarelationship
betweentheidentifiedproteinsandpotentialeffectors.
© 2011 Elsevier Ltd. All rights reserved.
1. Introduction
5-Azacitidineand 5-aza-deoxycytidine–decitabinearering analogsofcytosineanddeoxy-cytosine,differingfromthe natu-ralnucleosidebecauseofnitrogeninlieuofcarboninposition5of thepyrimidine.Theefficacyofazacitidineanddecitabineas anti-neoplasticagents, widelyusedfor treatmentofmyelodysplastic syndromes(MDS)andacutemyeloidleukemia(AML)oftheelderly, isattributedtotwodistinctmechanisms:cytotoxicityand induc-tion of DNA hypomethylation [1,2]. Azacitidine is transformed intotheactivenucleotide,5-aza-2-deoxycytidine-5-triphosphate, through ATP-dependent phosphorylation by uridine-cytidine kinaseandsubsequentactivityofribonucleotidereductase. Previ-ousandrecentincorporationstudiesinleukemiccellshaveshown thatthedistributionofAZA innucleicacidsis65:35,RNA:DNA [3,4].Decitabine,quitedifferently,incorporatesprimarilyintoDNA after deoxycytidine kinase mediated phosphorylation. Regional hypomethylationisthoughttobetheprincipalandshared respon-siblefor theclinicalefficacyof DNA methyltransferase(DNMT) inhibitors. In fact, once incorporated into DNA, by covalently trapping DNMT, both these drugs deplete the treated cells of functionallyactiveDNMT1,resultinginreversiblebutprofound
∗ Correspondingauthorat:FunctionalUnitofHematology,UniversityofFlorence, AOUCareggi,LargoBrambilla3,50141Florence,Italy.Tel.:+390557947296; fax:+390557947343.
E-mailaddress:santini@unifi.it(V.Santini).
hypomethylation and re-expression of silenced oncosuppressor genes.
Therearefewdirectcomparisonsofthetwoagentsinvivoor invitro,buttheirdifferenceinmetabolismand DNA incorpora-tionsuggestspossiblediversemechanismsofactionandbiological effects[3,5,6].
WeandothershavepreviouslydemonstratedthatAML1/ETO positive leukemiccells arepreferentiallysensitivetoepigenetic drugs [7,8]and specificallytoDNMTinhibitors[9–11]. We uti-lized a two-dimensional electrophoresis system, followed by MALDI-TOF/MStoanalyzetheregulationofproteinexpressionin AML1/ETOpositivecells,tofurtherdefinethemechanismofaction ofazacitidineanddecitabine,aswellaspossibledifferencesintheir biologicalactivityinthisparticularlysuitablecellularmodel. 2. Materialsandmethods
2.1. Cellsandcultureconditions
Kasumi-1cells(ahumanAML1/ETO-positivecelllinederivedfromanacute myeloblasticleukemia)wereplatedat3×105cell/mlinRPMI1640medium
sup-plementedwith10%FetalBovineSerum(FBS)(Gibco),4mMglutamine,50units/ml penicillinand50g/mlstreptomycinat37◦Cinahumidifiedatmosphere
contain-ing5%CO2.Cellsfromhumanbonemarrowaspirateswereobtainedafterinformed
consent(previouslyapprovedbythelocalEthicalCommittee),providedby2AML patientswitht(8;21),undergoingroutinediagnosticandinvestigationalprocedures. Inalltheexperiments,mononuclearcellsisolatedbydensitygradient centrifuga-tionwithLympholyte(CEDERLANE-Canada),andKasumi-1cellswereincubatedin thepresenceorabsenceoffreshlyprepared5-azacitidine(Sigma–Aldrich)1.0M anddecitabine(Sigma–Aldrich)0.5Mfor24h.Equipotentconcentrationsofdrugs werechosenonthebasisoftheirhypomethylatingactivityaspreviouslypublished
[3,12].
0145-2126/$–seefrontmatter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.leukres.2011.11.024
608 F.Buchietal./LeukemiaResearch36 (2012) 607–618
2.2. Proteinextractionand2Delectrophoresis
Cellsafter24hincubationwerewashedtwiceincoldPBSplus100Msodium orthovanadateandProteaseInhibitorCocktail(Sigma).About20×106cellswere
lysed,andafterultracentrifugation(100,000×g,1h,4◦C),proteinconcentration
determinedbyBradfordassay(Roche).Two-dimensional(2D)electrophoresiswas performedaspreviouslydescribed[13].
2.3. Imageanalysisand2Dgelcomparison
Two-dimensionalgelimageswereacquiredbyamodifiedUMAXUtah.1100 ImageScanner(AmershamBiosciences,Uppsala,Sweden),at1200dpiof resolu-tion.GelswereanalyzedandcomparedusingthesoftwareImageMasterPlatinum Ed.(Version5.0,AmershamBiosciences,Uppsala,Sweden).Proteinabundancewas evaluatedbycomparingtheSpotStainingVolumeascalculatedbyImageMaster Platinumed.Gelswereruninduplicateforeachsampleandscatterplotv/vforthe matchingspotpoolswereobtainedtoverifythedegreeofreproducibilityamong tworeplicatesofeachsample.
2.4. Ingeldigestionandmassspectrometry
Spotstobeidentifiedwereexcisedfromthegels,de-stainedandsubjectedto trypticin-geldigestionwithSequencingGradeModifiedTrypsin(Promega,WI) at37◦Cfor2h.TheresultingpeptidesweretreatedbyZip-TipC18(Millipore),
directlyelutedby2lofMatrixsolution(5g/l2,5-dihydroxybenzoicacidin50% ACN,0.1%TFA),on384position-400(m∅AnchorChipTarget(BrukerDaltonics, Bre-men,Germany).MassspectraforPeptideMassFingerprinting(PMF)wereacquired inreflectronpositive ionmodeonUltraflexMALDI-TOF/TOFMass Spectrome-ter(matrixassistedlaserdesorption/ionizationtime-of-flight,BrukerDaltonics, Bremen,Germany),withanaverageof100lasershots/spectrum.Spectrawere externallycalibratedusingacombinationoffourstandardpeptides:angiotensinII (1046.54Da),substanceP(1347.74Da),bombesin(1619.82Da)andACTH18-39Clip human(2465.20Da),spottedontopositionsadjacenttothesamples.Experimental PeptideMassFingerprintingintherangeof600–3500DawascomparedwithNCBI proteindatabasebythesoftwareMASCOT(www.matrixscience.org).Confirmatory fragmentationanalyses(MS/MS)wereperformedwhenneeded.
2.5. Westernblotting
Kasumi-1andhumanprimaryt(8;21)AMLcellswerewashedincoldPBS plus100MsodiumorthovanadateandProteaseInhibitorsCocktail(Sigma)and lysedbyincubatingat95◦CinLaemmlibuffer(62.5mMTris/HClpH6.8,10% glyc-erol,0.005%bluebromophenoland2%SDS).Proteinconcentrationwasdetermined byBCAassay(Pierce),30galiquotofeachsamplewasseparatedbySDS-PAGE (9,12.5or15%polyacrylamidegels)andthentransferredontoPVDFmembranes (Hybond-ECL;AmershamBiosciences)byelectroblotting.Membraneswere incu-batedwithprimaryantibodies(anti-calreticulin,anti-alpha-Enolase,anti-hnRNP A2/B1,anti-HSP60,anti-HSP70,anti-HSP90␣/,antiPRXI,antiTCP1,bySantaCruz Biotechnology;anti-cyclophilinA,anti-GAPDH,anti-PCNA,anti-␣-tubulin,byCell SignalingTechnology,anti--tubulinbyUpstate,1:100016–18hat4◦CinPBS
con-taining0.1%Tween20and5%BSA(T-PBS/5%BSA).Toverifyequalloadingofsample perlane,membraneswereincubatedinstrippingbuffer(62.5mMTris–HClpH6.7, 2%SDS,100mM2-mercaptoethanol;30minat50◦C)andextensivelywashedwith
T-PBS;thenthesamemembraneswereincubatedina1:1000dilutionofanti-p38 (CellSignalingTechnology)inT-PBS/%5BSA(16–18hat4◦C).Horseradish perox-idaseconjugatesecondaryantibodiesanti-rabbitIgG(Sigma)(1:5000inT-PBS/2% BSA;45minatroomtemperature)wereemployed.Immunoblotwerewashedin T-PBSandantibodycoatedproteinbandswerevisualizedbyECLchemiluminescence detection(AmershamBiosciences).
2.6. Dataanalysis
ScannedimageswereanalyzedbyImageMasterPlatinumEd.Version5.0 (AmershamBiosciences).Inordertoanalyzereproducibility,gelsimilaritiesor experimentalvariations,ScatterplotsanalysiswasperformedbysoftwareImage MasterPlatinumEd.whichcalculatesthelineardependencebetweenspotvaluesof onegelincomparisontoanothergel.Foreachproteinwecalculatedratiobetween spotvolumeinCTRLgelandspotvolumeintreatedgel.Aftermassspectrometry analysisandidentificationbysoftwareMASCOT(www.matrixscience.org),proteins wereclassifiedbyExPASyProteomicServer(http://expasy.org/)withUniProtKB database(http://www.uniprot.org).Functionalcategorieswereperformedusing DavidBioinformaticsResources6.7database(http://david.abcc.ncifcrf.gov/)and cutoffcriteriausedwereap-valueoflessthan0.001usingBenjamini–Hochberg correction.
Proteinnetworkanalyses– TheproteinsidentifiedbyMSwereincludedina pathwayanalysisusingtheMetaCoreTMnetworkbuilding(toolGeneGoInc.,St.
Joseph,MI–http://www.genego.com/training.php).MetaCoreTMisformedbyan
annotateddatabaseofproteininteractionsandmetabolicreactions.Genenamesof theallidentifiedproteinsandthecorrespondingmeanfold-changevaluesofsuch proteinswereimportedintoMetaCoreTMandprocessedusingthenetworkanalysis
algorithm.Likethat,somenetworkswerebuilt.Pathwaysanddirectandindirect
associationsamongproteinswerealsoinvestigatedusingStringdatabases(Version 8.3,http://string-db.org/).
3. Results
3.1. ProteomiccomparisonaftertreatmentwithDNMTinhibitors Protein gels were reproducible, as verified by scatter plot v/v among the replicates of each condition of culture (Fig. S1, a and b – Supplementary data). Changes of protein expressionafter24hazacitidine anddecitabinetreatmentwere consideredsignificantwhentheirquantitydecreasedorincreased byatleast1.3-fold.
3.2. Treatmentwith5-azacitidine
Accordingtoimageanalysis,620spotswereconstantlydetected incontrolgeland486spotsintreatedcellgelandwerequantified, normalized,intergelmatched,yielding400pairsofproteins. Sixty-fourprotein spots had significantvolume differences occurring betweenuntreatedcontrolandazacitidinetreated,andweexcised 51proteinspotsfromcontrolgeland13spotsfromtreated-cellgel. MassspectrometryanalysisbyMALDI-TOFallowedusto success-fullyidentify43proteinspotswhoselevelschanged(Fig.1A,aand b)correspondingto30characteristicproteins,classifiedin func-tionalcategoriesbyDavidBioinformaticsdatabase.Resultswere reportedin Table1: Spot numberobtainedfrom ImageMaster correlatestocorrespondingspotsinFig.2A.
As shown in Fig. 1B, nine proteins (30%) identified by MS included molecules functionally classified as Heat Shock Pro-teins/Chaperons,nineproteins(30%)wereenzymes,mostofwhich metabolic,oneproteinwasinvolvedinstress/redoxhomeostasis, five(17%) werestructuralproteinsand sixproteins(20%)were involvedinsignaltransduction(Table1andSupplementarydata). Themajorityofthedetectedproteinsdecreasedafterazacitidine treatment,asshowedbytheratioofthevolumeofspotsintreated anduntreatedcells(Table1)and,afterStringanalysis,weobserved thatallproteinswerelinkedbydifferentcharacteristics,andwere involvedin severalneoplasticdiseases(Fig.S2,a– Supplemen-tarydata).
Differential expression of PCNA, TCP1, hnRNP A2/B1 and CyclophilinAwasconfirmedbywesternblot(Fig.1C, aandb). CyclophilinA,inparticular,wasidentifiedbymorethanonespotin 2Dmaps,becauseofitsdifferentisoformswhichcouldbepresents, anditsglobalexpressioninazacitidinetreatedcellswaslowerthan that inuntreated cells. Regardingbiological variation ofhnRNP A2/B1,westernblotanalysisshowedaninhibitionofexpression ofribonucleoproteinB1andanincreaseofA2,butglobalprotein expressiondecreased(ratioofdensitometryCTRL/AZA:1.3,respect toloadingcontrol)(Fig.1C,aandb).
We observed in 2D gels different spots of alpha-Tubulin, beta-Tubulin,alpha-Enolase,ACTB,HSP90,HSP70andCalreticulin (Table1–markedwithanasterisk),duetothepresenceofseveral isoforms.Forthisreason,globalexpressionoftheseproteinswas unmodifiedafterazacitidine(Fig.1C,c).Isoformsoftheseproteins, wellseparatedby2Delectrophoresis,in1Dwesternblotclustered inthesameband,precludingthevalidationofdifferencein expres-sion.
ThedecreaseofCyclophilinA,PCNAandTCP1betawas con-firmedbywesternblotwithspecificantibodiesalsoinprimary culturesof2AMLt(8;21)casesafterexposureofleukemiccells toazacitidine(Fig.1D,aandb).
3.3. Treatmentwithdecitabine
Afterimageanalysis,576spotsweredetectedincontrolcells and596spotsintreatedcells;twogelswerematchedand416pairs
F. Buchi et al. / Leukemia Research 36 (2012) 607– 618 609 Table1
Differentiallyexpressedproteinsidentifiedbymassspectrometryafterazacitidinetreatment.
SpotCTRL SpotAZA Accession numbera
HeatShockProteins/Chaperons MWb pIc SCOREd Vol.C/Vol.Ae Inducedor
overexpressedbyAZA
Inhibitedor reducedbyAZA
C98 A69 P14625 Endoplasmin* 92,637 4.73 155 1.2 X
C96 A61 P14625 Endoplasmin* 90,351 4.73 32 0.4 X
C9 P07900 Heatshockprotein90kDaalpha,classBmember1* 83,638 4.94 75 CTRL X
C121 A82 P07900 Heatshockprotein90kDaalpha,classBmember1* 83,638 4.94 173 2.1 X
C139 A94 P07900 Heatshockprotein90kDaalpha,classBmember1* 83,638 4.94 69 0.5 X
C113 A83 P07900 Heatshockprotein90kDaalpha(cytosolic),classBmember1* 83,638 4.94 46 0.5 X
C181 A125 P11142 Heatshock70kDaprotein8* 71,138 5.37 186 2.5 X
C178 A123 P11142 Heatshock70kDaprotein8* 71,138 5.37 49 0.8 X
C216 A151 P10809 Heatshockprotein60 61,388 5.24 219 1.0
C249 A184 P78371 ChaperonincontainingTCP1,subunit2(beta) 57,878 6.02 153 1.8 X
C173 A119 P38646 Heatshock70kDaprotein9 74,089 5.97 117 1.2 X
C862 A528 P62937 Cypa(CyclophilinA)* 18,154 7.82 146 0.8 X
C710 A524 P62937 Cypa(CyclophilinA)* 18,154 7.82 98 1.4 X
C705 A523 P62937 Cypa(CyclophilinA)* 18,154 7.82 103 1.2 X
C263 A173 P27797 Calreticulin* 48,325 4.28 75 2.2 X
A306 P27797 Calreticulin* 48,325 4.28 46 AZA X
C235 P30101 Protien-disulfideisomeraseA3 56,761 5.61 83 CTRL X
Enzymes
C295 A208 P06733 Alpha-Enolase* 47,538 6.99 61 0.4 X
C292 A213 P06733 Alpha-Enolase* 47,538 6.99 94 1.5 X
C268 A199 P07237 Prolyl4-hydroxylasebeta-subunit 57,578 4.69 211 1.4 X
C236 A163 P12268 Inosinemonophosphatedehydrogenase2 56,338 6.44 115 1.5 X
C470 A355 P04406 Glyceraldehyde-3-phosphatedehydrogenase 36,244 8.58 91 1.5 X
C546 A415 P60174 ChainB,TriosephosphateIsomerase(Tim) 26,877 6.51 191 1.2 X
C538 A406 P28066 Proteasomesubunit,alphatype,5 26,621 4.74 46 1.4 X
C815 P55072 Valosin-containingprotein 71,660 5.14 49 CTRL X
C665 A487 P15531 Non-metastaticcells1,protein(NM23A)isoforma 19,925 5.84 64 2.7 X
C226 A157 P04040 Catalase 59,947 6.95 93 1.8 X
Cellredoxhomeostasis
C596 A446 P09211 GlutathioneS-transferaseP 23,651 5.44 86 1.4 X
Structuralproteins
C253 A191 P68363 Alphatubulin* 50,972 4.94 149 1.9 X
C260 A194 P68363 Alphatubulin* 50,972 4.94 106 0.5 X
C280 A204 P68363 Alphatubulin* 50,972 4.94 60 0.9
C445 A334 P07437 Betatubulin* 48,233 4.78 182 1.6 X
C421 P07437 Betatubulin* 48,233 4.78 127 CTRL X
C284 P07437 Betatubulin* 48,233 4.78 70 CTRL X
C365 A274 P60709 ACTBprotein(Beta-actin)* 40,620 5.29 142 1.8 X
C330 A251 P60709 ACTBprotein(Beta-actin)* 40,620 5.29 48 0.8 X
C413 A314 P52907 F-actincappingproteinalpha1 33,115 5.45 118 2.7 X
C475 A362 Q8TCG3 Tropomysin,TPMsk3 28,934 4.72 74 0.9
DNAbinding/signaltransduction
C278 A201 Q09028 Retinoblastomabindingprotein4variant 47,910 4.74 67 1.1
C400 A299 P06748 Nucleophosmin 32,653 4.64 76 1.3 X
C463 A350 P12004 Pcna 29,156 4.57 154 2.5 X
C424 A321 P22626 Heterogeneousnuclearribonucleoproteins(hnRNP)A2/B1 37,464 8.97 95 2.2 X
A541 P62158 Calmodulin 17,152 4.09 48 AZA X
aAccessionnumberbyUniProtKB.
b MW–theoreticalmolecularweighobtainedbyExPASyProteomicServer. c pI–theoreticalisoelectricpointobtainedbyExPASyProteomicServer. d Score–indexcalculatedbyMASCOT-significantwhengreaterthan56(p<0.05). eRatiobetweenspotvolumeinCTRLandspotvolumeintreatedgels.
610 F. Buchi et al. / Leukemia Research 36 (2012) 607– 618 Table2
Differentiallyexpressedproteinsidentifiedbymassspectrometryafterdecitabinetreatment.
SpotCTRL SpotDAC Accession numbera
HeatShockProteins/Chaperons MWb pIc Scored Vol.C/Vol.De Inducedor
overexpressed by DAC
Inhibitedor reduced by DAC
C1591 D43 P07900 Heatshockprotein90kDaalpha(cytosolic),classBmember1* 83,638 4.94 89 1.4 X
C1609 D66 P07900 Heatshockprotein90kDaalpha(cytosolic),classBmember1* 83,638 4.94 85 2.3 X
C1614 D73 P07900 Heat shock protein 90 kDa alpha (cytosolic), class B member 1* 83,638 4.94 93 0.2 X
C1689 D157 P10809 Heatshockprotein60 61,388 5.24 147 1.6 X
C1651 D119 P11142 Heatshock70kDaprotein8* 71,138 5.37 161 1.3 X
C2315 D71 P11142 Heat shock 70 kDa protein 8* 71,138 5.37 101 0.5 X
C1674 D155 P27797 Calreticulin 48,283 4.28 90 4.6 X
C1713 D184 P30101 Protein-disulfideisomeraseA3* 56,747 5.61 82 1.5 X
C1768 D251 P30101 Protein-disulfide isomerase A3* 57,146 5.61 72 2.2 X
C1752 D225 P30101 Protein-disulfide isomerase A3* 57,146 5.61 88 1.2 X
C2007 D660 P30040 EndoplasmicreticulumproteinERp29 29,032 6.08 70 1.8 X
C2119 D565 P23284 Cyclophilin B 22,785 9.25 122 1.3 X
C2178 P62937 Cyclophilin A* 18,229 7.82 61 CTRL X
C2183 D606 P62937 CyclophilinA* 18,229 7.82 75 0.4 X
C2184 D612 P62937 Cyclophilin A* 18,229 7.82 91 1.4 X
C1732 D204 P78371 Chaperonin containing TCP1, subunit 2 (beta) 57,878 6.02 125 1.3 X
Enzymes
C1824 D283 P00558 Phosphoglyceratekinase1 44,985 8.30 100 3.9 X
C1875 D347 P06733 Alpha-Enolase* 47,481 6.99 121 2.4 X
C1778 D237 P06733 Alpha-Enolase* 47,481 6.99 195 2.1 X
D256 P07954 Fumaratehydratase,mitochondrial 54,773 6.99 55 DAC X
C1707 D160 P04040 Catalase 59,947 6.95 62 0.8 X
C1925 D403 Q15181 Inorganic pyrophosphatase 33,095 5.54 120 1.6 X
C1927 D402 Q15181 Inorganicpyrophosphatase 33,095 5.54 108 1.4 X
C1970 D424 P14207 Folatereceptorbeta 30,286 7.47 32 0.6 X
C2142 D593 Q8TAE6 Protein phosphatase 1 regulatory subunit 14 C 17,946 5.09 31 0.8 X
C1900 D363 P04406 Glyceraldehyde-3-phosphatedehydrogenase 36,201 8.58 59 1.5 X
C2037 D489 P60174 ChainB,TriosephosphateIsomerase(Tim) 26,877 6.51 155 1.2
C1693 D137 Q02318 Cytocrome P450 27, mitochondrial 60,595 8.43 27 0.8 X
Cellredoxhomeostasis
C2060 D522 P09211 Glutathione S-transferase P 23,569 5.44 58 1.4 X
C2072 D531 Q06830 Peroxiredoxin-1 22,324 8.27 81 2.3 X
C1848 D331 O76003 Glutaredoxin-3 37,693 5.31 83 2.0 X
Structuralproteins
C2323 D304 P60709 ACTBprotein(Beta-actin) 40,620 5.29 163 1.9 X
C1906 D375 P07437 Beta tubulin* 48,233 4.78 154 1.7 X
C1915 D391 P07437 Betatubulin* 48,233 4.78 58 1.1 X
C2269 D205 P68363 Alphatubulin 50,972 4.94 89 0.3 X
DNAbinding/signaltrasduction
C1995 D459 P35232 Prohibitin 29,843 5.57 142 1.6 X
C1933 D412 P12004 Pcna 29,156 4.57 127 0.8 X
C1888 D345 Q13347 Eukaryotictranslationinitiationfactor3subunitI 36,878 5.38 92 1.8 X
C1926 D394 P22626 Heterogeneous nuclear ribonucleoproteins A2/B1 37,464 8.97 112 1.4 X
C1940 D388 P09651 HeterogeneousnuclearribonucleoproteinA1 38,936 9.17 132 1.4 X
C1834 D307 P29992 Guaninenucleotide-bindingproteinsubunitalpha-11 42,382 5.51 54 2.0 X
C2035 D491 P52565 Rho GDP-dissociation inhibitor 1 23,250 5.03 80 1.4 X
C1862 D336 P06748 Nucleophosmin 32,653 4.64 64 0.9 X
C1841 D320 Q9Y3F4 Serine-threoninekinasereceptor-associatedprotein 38,756 4.98 51 0.6 X
C2160 P62158 Calmodulin 17,152 4.09 49 CTRL X
aAccessionnumberbyUniProtKB.
b MW–theoreticalmolecularweighobtainedbyExPASyProteomicServer. c pI–theoreticalisoelectricpointobtainedbyExPASyProteomicServer. d Score–indexcalculatedbyMASCOT-significantwhengreaterthan56(p<0.05). eRatiobetweenspotvolumeinCTRLandspotvolumeintreatedgels.
F.Buchietal./LeukemiaResearch36 (2012) 607–618 611
Fig.1.Effectofazacitidine24honproteomeofKasumi-1cells.(A)Two-dimensionalgelsbefore(a)andafter(b)treatmentwereanalyzed.Spotswerethenpickedand identifiedbyMSanalysis.(B)Functionaldistributionofidentifiedproteins.(CandD)ConfirmatoryWesternblot;anti-p38isproteinloadingcontrol.CTRL,controluntreated cells;AZA,azacitidine;DAC,decitabine.
werefound;50proteinspotshadsignificantdifferencesbetween untreatedcontrolanddecitabinetreatedcells.Weexcised36 pro-teinspotsfromcontrolgeland14spotsfromtreatedgel;mass spectrometryanalysisMALDI-TOFallowedustoidentify45spots correspondingto35characteristicproteins(Fig.2A,aandb), clas-sifiedinfunctionalcategoriesbyDavidBioinformaticsdatabase.
ResultswerereportedinTable2(spotnumberobtainedfromImage Master,correlatetocorrespondingspotsinFig.2A).
As shown in Fig. 2B, nine proteins (26%) identified by MS included molecules functionally classified as Heat Shock Pro-teins/Chaperons,tenproteins(29%)wereenzymes,mostofwhich metabolic, three proteins (9%) were involved in stress/redox
612 F.Buchietal./LeukemiaResearch36 (2012) 607–618
Fig.2.Effectofdecitabine24honproteomeofKasumi-1cells.(A)Two-dimensionalgelsbefore(a)andafter(b)treatmentwereanalyzed.Spotswerepickedandidentified byMSanalysis.(B)Functionaldistributionofidentifiedproteins.(C)ConfirmatoryWesternblot;anti-p38isshownasproteinloadingcontrol.CTRL,controluntreatedcells; DAC,decitabine.
homeostasis, three (9%) were structural proteins and ten pro-teins (29%) were involved in signal transduction (Table 1 and Supplementarydata).Afterdecitabinetreatment,themajorityof identifiedproteins(thoseselectedonthebasisofasignificant mod-ulationinexpression)weredown-regulated,althoughproteinspot numberdidnotdecrease(Table2).
String analysisshowedthat the majority ofmodulated pro-teinswerecorrelated(FigS2,b–Supplementarydata).Decrease of hnRNPA2/B1, alpha-Enolase and theincrease of Cyclophilin A,HSP60 (ratioof densitometryCTRL/AZA: 0.4, respectto con-trolloadingp38) and alpha-Tubulinwereconfirmed (Fig.2C, a and b). Several proteins (HSP90, HSP70, PDIA3, Cyclophilin A,
beta-Tubulin, alpha-Enolase) had different migration sites into both2Dgels(Table2–markedwithanasterisk),becauseoftheir isoforms(seeabove),sosomeofthesedidnotchangetheglobal expression(Fig.2C,c).
CyclophilinAexpressionwasincreasedinprimaryt(8;21)AML cellsafter24hexposuretodecitabineinvitro(Fig.1D,b).
3.4. ComparisonbetweenAZAandDACtreatments
We matched gels after treatment with azacitidine and decitabine, in order to identify commonly modulated proteins and toevaluate possibledifferent effects of thetwo agents on
F.Buchietal./LeukemiaResearch36 (2012) 607–618 613
Fig.3. ComparisonbetweenproteomeofKasumi-1afterazacitidineanddecitabinetreatments.(A)Two-dimensionalComassiestainedgelsafterazacitidine(a)anddecitabine (b)werecomparedandcommonproteinsidentified.(B)Tableofcommonproteins.AZA,azacitidine;DAC,decitabine.
614 F.Buchietal./LeukemiaResearch36 (2012) 607–618
Fig.4.ProteomenetworksbyMetacoreanalysis.Metacoreanalysiswasperformedtocomparecommonproteinsidentifiedafterazacitidineanddecitabinetreatments.(A) Proteinsequallymodulatedbyeitherdrug:(a)proteinfoldingnetwork,(b)cellcyclenetwork.(B)Proteinsdifferentlymodulatedbytwodrugs:(a)gene-specifictranscription network,(b)responsetostress-network.(C)CyclophilinApathwaynetwork.
F.Buchietal./LeukemiaResearch36 (2012) 607–618 615
616 F.Buchietal./LeukemiaResearch36 (2012) 607–618
Fig.4.(continued)
proteinexpression(Fig.3A,a andb).Considering onlyproteins withasignificantMASCOTscore(Tables1and2;score>56),and thoseproteinswhoseglobalexpression(summingisoforms)was modulated,wefound14proteinsidentifiedbothinazacitidineand decitabinetreatedgels(Fig.3B).
3.5. Proteinnetworkanalysis
A pattern analysis was performed using the software MetaCoreTMinordertoevaluatethepossiblecorrelationbetween
the2D-electrophoresis/MSidentifiedmodulatedproteinsand pos-sible cellular pathways affected by azacitidine and decitabine (Table3).Thisanalysisallowedalsotorelatethefunctionofsingle proteinstothecellularcontext.Allproteinsanalyzedwere pro-cessedbythe“networkanalysis”algorithmwithfiftynodes,which generatessub-networksrankedbyp-values.
Firstofall,weinvestigatedthecorrelationbetweenproteins that were equally modulated by either drug, particularlyTCP1 beta,PDIA3,Calreticulin,Glyceraldehyde-3-phosphate dehydroge-nase(GAPDH),GlutathioneS-transferaseP,ACTBandhnRNPA2/B1 (Fig.3B).Theanalysisclusteredproteinsintwonetworks:thefirst, whichincludeTCP1,Calreticulin,ACTB,PDIA3andGSTP1regulates proteinfolding(Fig.4A,a)andthesecond,whichincludesGAPDH, Tcp1andhnRNPA2/B1,modulatescellcycle(Fig.4A,b).
Amongproteinsthathaddifferentmodulationaftertwodrugs treatments,wefocusedouranalysisonproteinsdecreased only
byazacitidine(Fig.3B),particularlyCyclophilinA,Catalase, Nucle-ophosminandPCNA.Thisanalysisleadtoclusterproteinsinthree network:two,includingPCNA,NucleophosminandCyclophilinA, areinvolvedingene-specifictranscriptionandinresponsetostress (Fig.4B,aandb),thethirdoneisidentifiedviaCatalasepresence, involvedintriglyceridemetabolism.
SinceCyclophilinAshowedadecreaseofexpressionafter azac-itidineandanincreaseafterdecitabine,weanalyzeditspathway networkasaninterestingdifferencebetweenthetwo agents.It seemsparticularlyrelevantthatCyclophilinAisinvolvedinAML proliferationandmaintenance(Fig.4C–linksmarkedinblue). 4. Discussion
AzacitidineanddecitabineareusedfortreatmentofMDSand AML.Althoughthesetwoagentsshareasimilarchemical struc-ture and are often considered equivalent,they exertdivergent biologicaleffectsandhavesomedifferentclinicalactivity,asonly azacitidine hasbeen showntosignificantly improvesurvival in MDS[14]whiledecitabineseemstohavepreferentialactivityon CMML[15]andAML[11].Partofthedifferencesmaybedueto prevalentincorporationintoRNAofazacitidineandintoDNAof decitabine,but itappearsthatothereffects,beside hypomethy-lation,mayexplaintheirclinicalactivity[16].Characterizationof azacitidineanddecitabineeffects onproteinexpression,protein networksandpathwaysmayhelpinunderstandingtheirmolecular Table3
SummaryofproteomemodificationsafterazacitidineanddecitabinetreatmentofAML1/ETOpositiveleukemiccells.
Drug Numberofproteinspots Proteinsmodulated
AZA Decrease ↓5involvedinproteinfolding ↓3involvedingene-specifictranscription
↓3involvedincellcycle ↓3involvedinresponsetostress ↓Catalase
DAC Nochange ↓5involvedinproteinfolding ↑CyclophilinA ↓alpha-Enolase
F.Buchietal./LeukemiaResearch36 (2012) 607–618 617
mechanisms,whosediscrepancieshavestartedtobehighlighted inscientificliterature[3,5,6].AML1/ETOpositiveleukemiashave a particular good prognosisand are sensitive tochemotherapy [7–10].Mostinterestingly,theAML1/ETOcomplexbindingDNMTs andHDACformsaputativeidealtargetforepigeneticdrugs[7–10]. Weemployedasensitive,2D-electrophoresisapproachfollowed byMALDI-TOF/MStoinvestigateproteinsmodificationsaftershort terminvitrotreatmentwithazacitidineanddecitabine.This pro-teomic approach is increasingly used to implement studies of mechanismsofactionofdrugs[17,18].
We could demonstrate distinct expression pattern between proteinextractsfromuntreatedleukemiccellsandextractsfrom azacitidineanddecitabinetreatedcells.Specificproteinswere sig-nificantlydown-regulated,althoughnotcompletely,bybothdrugs. Wefocusedourattentionbothonthecommonlyrepressed pro-teins,andonthosedecreasedexclusivelybyoneagent,toverify whetherthis differencecouldjustifythediverseclinical effects of thetwo DNMTinhibitors. Quite differentlyfrom azacitidine, decitabinedidnotdecreasetheglobalnumberofproteinspots, respecttountreatedsamples.Thisisconsistentwiththelackof inhibitionof proteinsynthesisbydecitabine,observedbyother authorswithdifferentanalyticalmethods[3].
Thetwopublishedreportsongeneexpressionmodificationby DNMTinhibitors[5,6]donotidentifythesamesetofgenes,and thereisnosignificantoverlapwithourproteinfindingseither.
Althoughthetwoagentsaresupposedtoactandbeefficient inclinicsviahypomethylationofCpGislandsofgenepromoters, themajorityofmodulatedproteinsbyinvitroDNMTinhibitors treatment were repressed, similarly to what observed in gene expressionstudies[6].Moreover,wecouldnotdetectsignificant increase of tumor suppressorgenes, indicating, again similarly to what already observed by other authors, that although the genes of these proteins may be up regulated via hypomethy-lationto levelsdetectableby PCR-based methods,theyare not partofthegeneraldominantexpressionpatterninducedbythese drugs,andproteomicmethodsarenotsensitiveenoughtodetect them[5].
Inordertobetterunderstandtherelationshipsamongthe pro-teinsidentified,weusedStringdatabase,aswellasMetaCoreTM.
Mostoftheproteinswhose expressionwasmodifiedby DNMT inhibitorswereinter-relatedinthepathwaymap,confirmingthe biologicalreliability ofouranalysis.Inaddition,themajorityof modulatedproteinswereknowntobeinvolvedinneoplastic pro-gression,sothattheirdecreasedexpressionbyDNMTinhibitors mayhelptounderstandthemechanismsbywhichtheseagents exerttheiranticanceractivity.
IntheMetaCoreTM producednetwork,theidentifiedproteins
clusteredin twonetworks, onemodulating proteinfolding and polymerization(TCP1,PBAF,Calreticulin,PDIA3,GSTP1)andthe secondinvolvedinregulationofcellcycleandmitosisandRNA transport(hnRNPA2,BAF,G3P2,TCP1).
The Chaperonin Containing TCP1 in the eukaryotic cytosol ensuresthecorrectfoldingandassemblyofavarietyofproteins. Neoplastic cells show an overexpression of TCP1 beta [19]. In ourmodel,azacitidineanddecitabineprovokeddecreaseinTCP1 expression,andalsoadecreaseinitsbest-characterizedsubstrate beta-actin.Wecouldindeedconfirmthisfindingalsoinprimary AMLcells.Calreticulinisincreasedinseveralneoplasticdiseases [20],anditsexpressiondecreasedafterbothDNMTitreatments. The same inhibition is observed for Glutathione S-transferases (GSTs),whichareafamilyofenzymesthatcatalyzethe conjuga-tionofglutathione(GSH)tovariousxenobioticsandparticipatein cellularprotectionagainstoxidativestressorresistancetodrugs. Theseproteinsexhibitgeneticpolymorphismsintheirpopulation distribution,andthesevariantsarefrequentlyexpressedinAML patientandmayinfluencetheirprognosis[21,22].
Heterogeneousribonucleoproteins(hnRNPs)arecomplexesof RNA and protein present in thecell nucleus during gene tran-scriptionandsubsequentpost-transcriptionalmodificationofthe newlysynthesizedRNA(pre-mRNA).Theyareinvolvedinsplicing ofpre-mRNAandinitstransportoutofthenucleus.Overexpression ofhnRNPA2/B1iscorrelatedtocancerprogression[23], particu-larlyhnRNPB1isover-expressedinMDSandAML-MDS[24]while hnRNPA1isoverexpressedgenerallyinsolid[25]and hemato-logicalmalignances[26].Fromouranalysis,itemergedthatboth agents,inourexperimentalconditions,havethesameinhibiting activityoncellcycle/RNAmetabolismandproteinfolding.Effects ongeneexpressionandmiRNAexpressionhavebeendemonstrated tobetimedependent,butweprivilegedtheanalysisofearly modu-lationbythesetwodrugs,topossiblyexaminenoncytotoxiceffects [5].
Regardingproteinsselectivelymodulatedbyazacitidine,itlooks quite important that one cluster of decreased proteins (PCNA, CyclophilinA,NPM-1)isinvolvedinresponsetostress, suggest-ingaroleofazacitidineinimpairingreactionofleukemiccellsto DNAdamage.Thesameproteinsareregulatinggene-specific tran-scription,and azacitidinecouldinfluenceit, butnot necessarily byCpGislandhypomethylation.Besides,onlyazacitidineinhibits Catalase,which notonly protectscellsfromthetoxic effectsof hydrogenperoxide,butalsopromotesgrowthofmyeloidleukemia cells,thankstoERK-1activation[27].Inconclusions,azacitidine, butnotdecitabine,seeminglyexertsanimportantdownregulation ofanti-stressproteins.
CyclophilinA(CypA)modulationisquitepeculiar,becauseits expressionisreducedbyazacitidine,butincreasedbydecitabine, andtheseresultswereconfirmedalsoinprimaryhumant(8;21) AMLcells.RecentstudiesindicateddifferentpossiblerolesofCypA intumordevelopment,giventhatitsexpressionresultedincreased and inducedby c-Myc and ithasinhibitory activityonp53,as alsoshownbyouranalysiswithMetaCore.Thisdiscrepancyafter drug treatment could be consistent witha reported early and transientinductionof pro-carcinogenicgenesbydecitabine[5]. Althoughaspecificroleinnormalandmalignanthemopoiesishas not beendemonstrated for Cyclophilin A,it is possible thatits effectsoninflammation,neutrophilchemo-attractionand induc-tionofreactiveoxygenspeciesmaybecriticalforleukemiccells [28,29].Anotherinterestingfeatureofouranalysisisthatdecitabine isselectivelyabletodownregulatealpha-Enolase.Itcouldbethat thisoccursbecauseofdecitabinesite-specificDNAincorporation. Certainly,bydecreasingalpha-Enolaseactivity,decitabinecould actreducingleukemicproliferation.Moreover,onlytreatmentwith decitabinedown-regulatesHSP60,which indeedplaysa pivotal roleinleukemogenesis[30].
Differences between the mechanisms of azacitidine and decitabinewereobservedintheiractivitiesoncellviability, pro-teinsynthesis,cellcycle,miRNAandgeneexpression[3,5,6].The hypothesis that AZA and DAC have finely diversebiomolecular mechanismsissupportedbytheobservationthatthesetwoagents have a specific and selectivemodulation of thetarget proteins whichweidentified.Theireffectscouldbecausednotexclusively byDNAhypomethylation.Ourobservationsfurtherindicatethat azacitidineanddecitabinearenotinterchangeableandthattheir alternateorcombineduseinclinicsmaybeconceivable.
Conflictofinterest
Allauthorshavenoconflictofinteresttodeclare. Acknowledgments
Contributions.V.S.designedthestudyandwrotethemanuscript; F.B.performed experiments and wrote theresultssection; G.F.
618 F.Buchietal./LeukemiaResearch36 (2012) 607–618
performedmassspectrometryanalysis;E.S.and E.M.performed Western blot experiments and A.S., A.G. and A.B. took part in designingthestudyandrevisedthemanuscript.Fundings.Regione Toscana,BandoSalute 2009;EnteCassa diRisparmiodiFirenze (ECR); and Ministero per l’Istruzione, l’Università e la Ricerca (MIUR).
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,atdoi:10.1016/j.leukres.2011.11.024.
References
[1] Santini V, Kantarjian HM,Issa JP. Changes in DNA methylation in neo-plasia: pathophysiology and therapeutic implications. Ann Intern Med 2001;134:573–86.
[2] SantiniV.Azacitidine:activityandefficacyasan epigenetictreatmentof myelodysplasticsyndromes.ExpertRevHematol2009;2:121–7.
[3]HollenbachPW,NguyenAN,BradyH,WilliamsM,NingY,RichardN,etal.A comparisonofazacitidineanddecitabineactivitiesinacutemyeloidleukemia celllines.PLoSOne2010;5:e9001.
[4]LiLH,OlinEJ,BuskirkHH,ReinekeLM.Cytotoxicityandmodeofactionof 5-azacytidineonL1210leukemia.CancerRes1970;30:2760–9.
[5]QiuX,HotherC,RalfkiaerUM,SogaardA,LuQ,WorkmanCT,etal.Equitoxic doses of 5-azacytidine and 5-aza-2deoxycytidine induce diverse imme-diate and overlappingheritable changesin the transcriptome. PLoS One 2010;5:e12994.
[6]FlothoC,ClausR,BatzC,SchneiderM,SandrockI,IhdeS,etal.TheDNA methyl-transferaseinhibitorsazacitidine,decitabineandzebularineexertdifferential effectsoncancergeneexpressioninacutemyeloidleukemiacells.Leukemia 2009;23:1019–28.
[7]GozziniA,RovidaE,DelloSbarbaP,GalimbertiS,SantiniV.Butyrates,asa singledrug,inducehistoneacetylationandgranulocyticmaturation:possible selectivityoncorebindingfactor-acutemyeloidleukemiablasts.CancerRes 2003;63:8955–61.
[8]BarbettiV,GozziniA,RovidaE,MorandiA,SpinelliE,FossatiG,etal.Selective anti-leukaemicactivityoflow-dosehistonedeacetylaseinhibitorITF2357on AML1/ETO-positivecells.Oncogene2008;27:1767–78.
[9]SantiniV.AselectiveactivityofDNMTidecitabineonAML1ETOpositivecells? LeukRes2007;31:741–2.
[10] GozziniA,SantiniV.Butyratesanddecitabinecooperatetoinducehistone acetylationandgranulocyticmaturationoft(8;21)acutemyeloidleukemia blasts.AnnHematol2005;84(Suppl1):54–60.
[11]BergT,GuoY,AbdelkarimM,FliegaufM,LubbertM.Reversalofp15/INK4b hypermethylationinAML1/ETO-positiveand-negativemyeloidleukemiacell lines.LeukRes2007;31:497–506.
[12]StresemannC,BruecknerB,MuschT,StopperH,LykoF.Functionaldiversity ofDNAmethyltransferaseinhibitorsinhumancancercelllines.CancerRes 2006;66:2794–800.
[13]FerrariG,PastorelliR,BuchiF,SpinelliE,GozziniA,BosiA,etal. Compar-ativeproteomicanalysisofchronicmyelogenousleukemiacells:insidethe mechanismofimatinibresistance.JProteomeRes2007;6:367–75.
[14]FenauxP,MuftiGJ,Hellstrom-LindbergE,SantiniV,FinelliC,GiagounidisA, etal.Efficacyofazacitidinecomparedwiththatofconventionalcareregimens inthetreatmentofhigher-riskmyelodysplasticsyndromes:arandomised, open-label,phaseIIIstudy.LancetOncol2009;10:223–32.
[15]WijermansPW,RuterB,BaerMR,SlackJL,SabaHI,LubbertM.Efficacyof decitabineinthetreatmentofpatientswithchronicmyelomonocyticleukemia (CMML).LeukRes2008;32:587–91.
[16] FandyTE,HermanJG,KernsP,JiemjitA,SugarEA,ChoiSH,etal.Earlyepigenetic changesandDNAdamagedonotpredictclinicalresponseinanoverlapping scheduleof5-azacytidineandentinostatinpatientswithmyeloid malignan-cies.Blood2009;114:2764–73.
[17] BianchiL,BruzzeseF,LeoneA,GagliardiA,PugliaM,DiGennaroF,etal. Proteomic analysisidentifiesdifferentially expressedproteinsafter HDAC vorinostatandEGFRinhibitorgefitinibtreatmentsinHep-2cancercells. Pro-teomics2011;11:3725–42.
[18] Buchi F, PastorelliR, FerrariG, Spinelli E, Gozzini A, SassoliniF, et al. Acetylomeandphosphoproteomemodificationsinimatinibresistantchronic myeloid leukaemia cells treated with valproic acid. Leuk Res 2011;35: 921–31.
[19] CoghlinC,CarpenterB,DundasSR,LawrieLC,TelferC,MurrayGI. Characteriza-tionandover-expressionofchaperonint-complexproteinsincolorectalcancer. JPathol2006;210:351–7.
[20]GoldLI,EggletonP,SweetwyneMT,VanDuynLB,GreivesMR,NaylorSM,etal. Calreticulin:non-endoplasmicreticulumfunctionsinphysiologyanddisease. FASEBJ2010;24:665–83.
[21]LeeHS,LeeJH,HurEH,LeeMJ,LeeJH,KimDY,etal.Clinicalsignificanceof GSTM1andGSTT1polymorphismsinyoungerpatientswithacutemyeloid leukemiaofintermediate-riskcytogenetics.LeukRes2009;33:426–33. [22]VosoMT,HohausS,GuidiF,FabianiE,D’AloF,GronerS,etal.Prognostic
roleofglutathioneS-transferasepolymorphismsinacutemyeloidleukemia. Leukemia2008;22:1685–91.
[23] ChenZY,CaiL,ZhuJ,ChenM,ChenJ,LiZH,etal.FynrequiresHnRNPA2B1and Sam68tosynergisticallyregulateapoptosisinpancreaticcancer. Carcinogen-esis2011;32:1419–26.
[24]OhshimaK,KarubeK,ShimazakiK,KammaH,SuzumiyaJ,Hamasaki M, etal. Imbalancebetweenapoptosisandtelomeraseactivityin myelodys-plasticsyndromes:possibleroleinineffectivehemopoiesis.LeukLymphoma 2003;44:1339–46.
[25]Ma YL,PengJY,ZhangP,HuangL,LiuWJ,ShenTY,etal.Heterogeneous nuclearribonucleoproteinA1isidentifiedasapotentialbiomarkerfor colorec-talcancerbasedondifferentialproteomicstechnology.JProteomeRes2009;8: 4525–35.
[26]ShiY,FrostPJ,HoangBQ,BenavidesA,SharmaS,GeraJF,etal.IL-6-induced stimulationofc-myctranslationinmultiplemyelomacellsismediatedbymyc internalribosomeentrysitefunctionandtheRNA-bindingprotein,hnRNPA1. CancerRes2008;68:10215–22.
[27]Hachiya M, Akashi M. Catalase regulates cell growth in HL60 human promyelocyticcells:evidenceforgrowthregulationbyH(2)O(2).RadiatRes 2005;163:271–82.
[28]SatohK,BerkBC,ShimokawaH.Vascular-derivedreactiveoxygenspeciesfor homeostasisanddiseases.NitricOxide2011;25:211–5.
[29] SongF,ZhangX,RenXB,ZhuP,XuJ,WangL,etal.CyclophilinA(CyPA)induces chemotaxisindependentofitspeptidylprolyl cis-transisomerase activity: directbindingbetweenCyPAandtheectodomain ofCD147.JBiolChem 2011;286:8197–203.
[30]SedlackovaL,SpacekM,HollerE,ImryskovaZ,HromadnikovaI.Heat-shock proteinexpressioninleukemia.TumourBiol2011;32:33–44.