Identification of tuberculosis-associated proteins in whole blood supernatant

Identification of tuberculosis-associated proteins in whole blood supernatant Tanakaetal.BMCInfectiousDiseases2011,11:71 Identificationoftuberculosis-associatedproteins inwholebloodsupernatant TakahiroTanaka1,2,ShinsakuSakurada2,KeikoKano3,EriTakahashi3,5,KazukiYasuda4,HisashiHirano5, YasushiKaburagi3,NobuyukiKobayashi6,NguyenThiLeHang7,LuuThiLien8,IkumiMatsushita2,MinakoHijikata2, TakafumiUchida1andNaotoKeicho2* Abstract Background:Biologicalparametersareusefultoolsforunderstandingandmonitoringcomplicateddisease processes.Inthisstudy,weattemptedtoidentifyproteinsassociatedwithactivepulmonarytuberculosis(TB)using aproteomicapproach. Methods:ToassessTB-associatedchangesinthecompositionofhumanproteins,wholebloodsupernatantswere collectedfrompatientswithactiveTBandhealthycontrolsubjects.Two-dimensionaldifferencegelelectrophoresis (2D-DIGE)wasperformedtoanalyzeproteinswithhighmolecularweights(approximately>20kDa).Baseline proteinlevelswereinitiallycomparedbetweenpatientswithactiveTBandcontrolsubjects.Possiblechangesof proteinpatternsinactiveTBwerealsocomparedexvivobetweenwholebloodsamplesincubatedwith Mycobacteriumtuberculosis(Mtb)-specificantigens(stimulatedcondition)andunderunstimulatedconditions. Immunoblotandenzyme-linkedimmunosorbentassays(ELISA)wereperformedtoconfirmdifferencesinidentified proteins. Results:Underthebaselinecondition,wefoundthatthelevelsofretinol-bindingprotein4(RBP4),fetuin-A(also calleda-HS-glycoprotein),andvitaminD-bindingproteindifferedbetweenpatientswithactiveTBandcontrol subjectson2Dgels.ImmunoblottingresultsconfirmeddifferentialexpressionofRBP4andfetuin-A.ELISAresults furtherconfirmedsignificantlylowerlevelsofthesetwoproteinsinsamplesfrompatientswithactiveTBthanin controlsubjects(P<0.0001).Mtb-specificantigenstimulationexvivoalteredclusterinexpressioninwholeblood samplescollectedfrompatientswithactiveTB. Conclusions:WeidentifiedTB-associatedproteinsinwholebloodsupernatants.Thedynamicsofprotein expressionduringdiseaseprogressionmayimproveourunderstandingofthepathogenesisofTB. Background Recent advances in comprehensive analytical techni- Tuberculosis (TB) is one of the most important infec- ques, such as transcriptomics and proteomics, have tiouscausesofdeathworldwide[1].Despiteitslonghis- enabledustoidentifyproteinsassociatedwithactiveTB torical interaction with humans, our understanding of in humans. As a pioneering approach, Jacobsen et al. host response totheTB pathogen remains incomplete. compared the gene expression profiles of peripheral Investigationofthemolecularbasisofdifferencesinthe blood mononuclear cells from patients with TB and host immune status and metabolism between patients Mtb-infectedhealthydonorsbymicroarrayanalysis[2], withactiveTBandcontrolsubjectsmayprovideaclue and Mistry et al. analyzed gene expression patterns in tounderstand the disease process, and thus contribute whole blood inanattempt tofind acandidate biomar- tofuturestrategiesforTBpreventionandtreatment. kerfordiscriminating curedpatientsfromthosewitha riskofrelapse[3]. Agranoffetal.[4]identifiedamyloidAandtransthyre- *Correspondence:nkeicho-tky@umin.ac.jp 2DepartmentofRespiratoryDiseases,ResearchInstitute,NationalCenterfor tin in human serum as potential indicators for distin- GlobalHealthandMedicine,1-21-1Toyama,Shinjuku-ku,Tokyo162-8655, guishing patients with TB from those with non-TB Japan Fulllistofauthorinformationisavailableattheendofthearticle ?2011Tanakaetal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommons AttributionLicense(),whichpermitsunrestricteduse,distribution,andreproductionin anymedium,providedtheoriginalworkisproperlycited. Page2of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 inflammatoryconditions.Theyalsoreportedthatacom- Writteninformedconsentwasobtainedfromeachparti- bination of four protein markers, including amyloid A cipant.Bloodsampleswerecollectedfrompatientswith andtransthyretin, achieved adiagnostic accuracy ofup activeTBimmediatelybefore(Vietnamesepatientsam- to 78%. Chegou et al. [5] reported that EGF, VEGF, ples) or within 7 days (Japanese patient samples) of TGF-a, and sCD40L in supernatants obtained from treatment initiation. Patients with potential complica- interferon-gamma (IFN-g)-release assays (IGRAs) are tionsattributabletomalignancies,autoimmunediseases, informative markers for differentiating active disease orHIVcoinfectionwereexcludedfromthestudy. from latent infection. Although the above studies are Attheinitialscreeningandconfirmationstage,blood promising,suchcomprehensiveanalyticaltechniquesare samples were collected from 14 Japanese patients with still in the developmental stages and further investiga- bacteriologicallyconfirmedactivepulmonaryTB(9men tionsarerequiredbeforetheycanbeappliedclinically. and5women;medianage50years, range 22-75years) IGRA detects TB infection by measuring the Mtb- and 13 age- and gender-matched healthy Japanese specificimmuneresponsewithhighspecificity[6].IFN-g patients (8 men and 5 women; median age 48 years, isreleasedbyreactivationofMtb-specificeffectormem- range 24-64 years). We could not completely rule out oryTcellsinwholeblood.Despiteitsadvantages,IGRA thepossibilityoflatentTBinfectionin2ofthe13con- isnotaperfecttoolforuseinmostdevelopingcountries. trolsubjects,accordingtotheresultsofacommercially IncountrieswithahighTBburden,patientswithactive availableIGRA(QuantiFERON?-TBGoldinTube;Cel- TB,andnotthose with latent TBinfection, needtobe lestis, Victoria, Australia). However, we analyzed all immediately identified andtreated inorder toprevent samplestogetherattheinitial stagetoidentifyproteins disease transmission. However, IGRAisnotcapable of associated with active TB disease. The tuberculin skin distinguishingactiveTBfromlatentinfection.Also,cyto- test was not useful for detecting latent TB infection in kinemeasurementstobeperformedforIGRAarerather our study since most individuals in the tested popula- expensive inaresource-limited setting anddifficult to tions had received BCG vaccination after birth. Blood distribute. Thus, from apractical aswell asaresearch samples from 4 patients with active TB and 4 healthy standpoint, development of new markers for TB is individuals werechosenforscreening by2D-DIGEand desired. immunoblotting. The stability of proteins measured by In the present study, by high-resolution two-dimen- the enzyme-linked immunosorbent assay (ELISA) was sionaldifferencegelelectrophoresis(2D-DIGE)followed investigated by comparing a set of plasma samples byliquidchromatography-mass spectrometry (LC-MS), directly separated from EDTA-containing peripheral we analyzed the expression profiles of high molecular bloodandanothersetofplasmasupernatants obtained weight proteins (approximately >20kDa)that havenot fromheparinizedbloodafter18hofincubation(under beenstudiedfullyamongcomponentsofresidualwhole thesameconditionsastheIGRAnegativecontrol). bloodsupernatantsafterperformingIGRA. At the next verification stage, we utilized samples We used two comparative frameworks. One was the from 25 Vietnamese patients with sputum smear- direct comparison of plasma supernatants collected positive active pulmonary TB(13menand12women; from patients with active TB and healthy control sub- medianage35years,range20-55years)and50age-and jects. This comparison aimed to identify proteins that gender-matched Vietnamese healthy control subjects are markedly upregulated ordownregulated in the dis- (26 men and 24 women; median age 36 years, range easestate;evenifsuchproteinsarenotdiseasespecific, 21-54years)ofwhich25wereIGRApositiveand25were theymightactasusefulmarkersformonitoringthedis- IGRAnegative. NoneoftheIGRA-positive individuals easebefore,during,andaftertreatment.Theothercom- hadanysignsorsymptomsofactiveTBbutatleastsome parative framework was more TB specific since whole werereasonably suspected tohavelatent TBinfections bloodsamplesfrompatientswerestimulatedwithMtb- becausetheprevalence ofTBinthepopulation ishigh. specific antigens or left unstimulated, and the results Following IGRA, the remaining unstimulated plasma werecompared. supernatantswereusedforELISA. Methods Sample collection and preparation Patients and control subjects Whole blood was separately collected in heparin- In this study, whole blood samples collected from containing tubes precoated with mitogen as a positive Japanese and Vietnamese individuals were used. The controlorcocktailsofESAT-6,CFP-10,andTB7.7(p4) studywasapprovedbytheethicalreviewcommitteesof peptides as Mtb-specific antigens (QuantiFERON?-TB the National Center for Global Health and Medicine GoldinTube;Cellestis);thenegativecontroltubeshad (formerly the International Medical Center of Japan), noprecoat.After18hofincubationat37?C,eachsam- Tokyo, Japan, and the Ministry of Health, Vietnam. ple was centrifuged and the plasma supernatants were Page3of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 harvested and stored at -80?C until use in subsequent Molecular Imager FX system with Quantity One soft- assays. For proteomic analysis, four sample sets that ware. Mass spectrometric analysis was performed were either unstimulated or stimulated with Mtb- according to the method reported by Toda et al. [7], specific antigens or mitogen from patients with active with slight modification. Briefly, each protein spot on pulmonaryTBandfourcorrespondingsetsfromcontrol SYPRORubystainedgelswaspickedusingaspotpicker subjects were used to screen for candidate proteins by (Amersham Biosciences). In-gel digestion of proteins 2D-DIGE. To increase resolution, 14 human major was performed according to the method reported by plasmaproteins(albumin,IgG,antitrypsin,IgA,transfer- Saekietal.[8]. rin, haptoglobin, fibrinogen, alpha 2-macroglobulin, alpha1-acidglycoprotein,IgM,apolipoproteinAl,apoli- Mass spectrometric analysis poproteinAll,complementC3,andtransthyretin) were AnESIion-trapmassspectrometer(LCQDecaXPPlus, removedpriortoelectrophoresis usingaMultipleAffi- ThermoElectron)wasusedforpeptidedetection.Mass nityRemovalLCColumn-Human14(AgilentTechnolo- spectrometric analysis wasperformedasdescribedpre- gies, Santa Clara, CA, USA). The samples were then viously [8]. Protein identification was performed using concentrated by ultrafiltration (Agilent Technologies, the Mascot server (Matrix Science, Boston, MA, USA) Concentrators Spin 5 kDa MWCO, 4 ml) followed by andProteinProspector(UCSFMassSpectrometryFacil- acetone precipitation in preparation for subsequent ity, San Francisco, CA, USA). We selected the SWISS- electrophoresis. PROT Homo sapiens database and used the following parameters: peptide tolerance 1.0 Da and one missed cleavage. Carbamidomethyl modification of cysteine, Quantitative analyses by2D-DIGE Protein samples werelabeled with Cy3 andCy5(DIGE acetylation of the NH2-terminal ends of lysine, and Fluors Minimal Labeling Dyes; GEHealthcare, Buckin- phosphorylation of serine, threonine, or tyrosine were ghamshire,UK)accordingtothemanufacturer’sinstruc- consideredinthisanalysis. tions.Thesamples(50μgoftotalproteinpergel)were applied to Immobiline DryStrips (18 cm long, pH 4-7 Immunoblotting linear; Amersham Biosciences, Pittsburgh, PA, USA), Immunoblotting to detect the proteins identified as and isoelectric focusing (IEF) was performed using an described above was performed using anti-human Ettan IPGphor IEF system (Amersham Biosciences) retinol-binding protein 4 (RBP4) rabbit polyclonal IgG accordingtothemanufacturer’sinstructions.Next,SDS- (A-0040;Dako;Glostrup,Denmark),anti-humanfetuin- PAGEwasperformedusinga10-18%lineargradientgel A (AHSG) goat polyclonal IgG (G-20; Santa Cruz from DRC Co., Ltd. (Tokyo, Japan). The fluorescence Biotechnology;SantaCruz,CA,USA),anti-humanvita- intensity of each protein spot was digitally recorded min D-binding protein (VDBP) (Gc-Globulin) rabbit using a Molecular Imager FXsystem (Bio-Rad Labora- polyclonal IgG (Dako), anti-human clusterin-a mouse tories,Hercules,CA,USA)withQuantityOnesoftware monoclonal IgG1 (B-5; Santa Cruz Biotechnology), or (Bio-Rad Laboratories), anddifferential protein expres- anti-human clusterin-b rabbit polyclonal IgG (N-18; sionwasquantitativelyanalyzedusingthePDQuestsoft- SantaCruzBiotechnology). ware (Bio-Rad Laboratories). The same gel included a Total protein concentrations were determined using reference sample that had been labeled with Cy2 and theBio-Radproteinassaykit(Bio-RadLaboratories).To wasusedforspotmatching,imageanalysis,andvolume detect clusterin- a and -b, mixed protein samples normalization. Initially, allspotswereroughly matched (20 μg) were applied to 2D PAGE with 1D IEF using usinganautomatedtoolinthePDQuestsoftwaresuite. theImmobilineDryStrip(pH3-5.6nonlinear).Proteins This estimate was followed by a more detailed manual were then transferred to PVDF membranes. The curationtocorrectanyinappropriatelymatchedpairsof membranes were probed with polyclonal antibodies, proteinspots. anti-clusterin-a, and anti-clusterin-b. To detect other proteins,eachsample(10μg)wassubjectedtoconven- tional SDS-PAGE. Membranes were probed with anti- Sample preparation formass spectrometry A mixture of all samples (400 μg of total protein per VDBP,anti-fetuin-A,oranti-RBP4polyclonalantibodies. gel) was subjected to 2D-DIGE under the same condi- Anti-mouse and anti-rabbit (GE Healthcare) as well as tions as described above to isolate selected spots. To anti-goat (Santa Cruz Biotechnology) HRP-conjugated visualize individual protein spots, thegels werestained secondaryantibodieswereprepared.Proteinbandswere with SYPRORuby protein gel stain (Molecular Probes, detected using the ECL plus detection reagent (GE Eugene,OR,USA)for3h.Thefluorescenceintensityof Healthcare). Band intensities were calculated using the each protein spot was digitally measured using the QuantityOnesoftware. Page4of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 ELISA A competitive ELISA for quantitative determination of RBP4inhumanplasmawasperformedaccordingtothe manufacturer ’s instructions (AdipoGen Inc.; Seoul, Korea). The detection limit was 1 ng/ml. An AHSG ELISA kit was used to detect fetuin-A in plasma (Bio- Vender Laboratory Medicine Inc.; Modrice, Czech Republic).Thedetectionlimitwas0.35ng/ml.AQuan- tikine? HumanVitamin D-Binding Protein Immunoas- say kit was used to detect VDBP in plasma (R&D Systems,Inc.;Minneapolis,MN,USA).Themeanmini- mum detectable VDBP level was 0.65 ng/ml. Distribu- tion of levels was represented using the median and interquartilerange(IQR). Statistical analysis Proteins showing differential expression between two conditionswerefirstdeterminedwithPvaluesusingthe Student’s t-test preinstalled in the PDQuest software suite.Toselectcandidateproteinswithexpressionlevels that differed between unstimulated samples from patients with active TB and healthy control subjects, a significancelevelofP<0.05wasselected.Toselectcan- Figure 1 Schematic representation of protein expression didate proteins showingdifferential expression in Mtb- profiles.Comparisonbetweensubjects,inparticular,spotpatterns specific antigen-stimulated and unstimulated plasma showingdifferencesinplasmaproteomesbetweenpatientsand samples, a less stringent cut-off value of P < 0.10 was controlsubjects.Comparisonbetweenthepresenceandabsenceof applied. Assuming analpha error of0.1 anda standar- stimuli,especiallyexvivospotpatternsofMtb-specificantigen- dizedeffectsizeof2.0,thepowertodetectadifference stimulatedplasmasamplesandunstimulatedsamplesfrompatients. Twomajorcomparativeframeworksinourstudyareshown(bold was calculated as 0.8 given our sample size. When a arrows). normaldistributionofmeasurementswasnotpredicted, the Wilcoxon rank sum test (Mann-Whitney U test) was applied for confirmation using the JMP software (version7.0.1;SASInstitute,Cary,NC,USA). fromthepatientsandgreenindicatesproteinsdecreased in the patients compared with the control subjects. Results Yellow indicates nosignificant differences (Figure 2A). Quantitative analyses by2D-DIGE In2Dgelprofilescomparingtheantigen-stimulatedand In a preliminary experiment, we used an immobilized unstimulatedsamplescollectedfrompatientswithactive linearpHgelstripwithabroadpHrange(pH3-10lin- TB,redindicatesproteinsincreasedinthesupernatants ear)for1DIEF.Althoughmorethan500protein spots after Mtb-specific antigen stimulation, and green indi- werevisualizedinfreshplasmawithSYPRORubystain- cates proteins decreased after stimulation. Yellow indi- ing, the number of spots after incubation of whole catesnosignificantchanges(Figure2B).From367spots bloodwithstimulidecreased,anddetectablespotswere compared between patients withactive TBandcontrol primarilylocatedinthepHrange4-7(datanotshown). subjects,and293spotsgeneratedwithsamplescollected Therefore, we performed subsequent analyses using an frompatientswithactiveTBthatwereeitherstimulated immobilized linear pH gel strip with a narrower range with Mtb-specific antigens or left unstimulated, we (pH 4-7 linear) to obtain a finer resolution. We used selectedseveralcandidatesforsubsequentmassspectro- two comparative frameworks in our analyses, and the metric analysis (Table 1) according to the criteria corresponding spot patterns are schematically depicted describedintheMaterialsandMethodssection. inFigure1. Differential gel images were acquired and displayed Mass spectrometric analysis usingthePDQuest2Dgelanalysissoftware(Figure2A, Following the above criteria for selecting candidates of B).Inourcomparisonoftheproteinexpressionprofiles differentiallyexpressedproteinsbetweentwoconditions, ofpatientswithactiveTBandcontrolsubjects,redindi- atotalof41spotswereisolatedfromthecorresponding cates proteins increased in the supernatants collected 2Dgelsonthebasisthattheyshowedsufficientlystrong Page5of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 Figure22D-DIGE-basedproteinexpressionanalysis.Pseudo-coloredimagesweregeneratedusingtheMulti-ChannelViewerfunctionofthe PDQuestsoftwaresuite.FoursamplesetsthatwereeitherunstimulatedorstimulatedwiththeMtb-specificantigenormitogenfrompatients withactivepulmonaryTBandfourcorrespondingsetsfromcontrolsubjectswereusedtoscreenforcandidateproteinsby2D-DIGE.A:Analysis basedondifferencesinlevelsinsamplesfrompatientsandcontrolsubjects.Redindicatesproteinswhoselevelsincreasedandgreenindicates proteinswhoselevelsdecreasedinunstimulatedsupernatantsfrompatientswithactiveTBcomparedwithcontrolsubjects.Spot(a):HT6102, Spot(b):HT2406,andSpot(c):HT5401correspondtothePDQuestidentificationnumbersshowninTable2.B:Analysisbasedondifferencesin levelsafterstimulation.RedindicatesproteinswhoselevelsincreasedinwholebloodsupernatantsfrompatientswithactiveTBafterincubation withMtb-specificantigens.Spots(d):T3107,T4203,andT4208,correspondtoTable2. signals. Trypsin digestion of each isolated spot was active TB; numbers 12 to 14 (T4203, T3107, and followedbyLC-MSanalysis.Theproteinscorresponding T4208) were all identified as clusterin. In Table 2, P to 14 of these spots were successfully identified values indicating a significant difference between the (Figure3). means of the two conditions examined, the SWISS- Ofthe14proteins inTable 2,7(serial numbers 1to PROT accession numbers of the identified proteins as 7) were obtained as a result of comparisons between wellastheirmolecularweightsandtheoreticalpIvalues patients withactive TBandcontrol subjects; number1 areindicated. WealsousedtheHomosapiensdatabase (spot HT6102) was identified as RBP4, number 2 of expressed sequence tags (ESTs) to identify clusterin (HT2406) as fetuin-A, and number 3 (HT5401) as inspotT4208. VDBP.Four(numbers8to11)wereobtainedasaresult Mascot search scores (indices of protein matches) ofcomparisonsbetweennonspecificmitogen-stimulated were47,50,98,75,and72forspotsT4203(clusterin), and unstimulated samples collected from patients with T3107(clusterin), HT5401 (VDBP), HT2406 (fetuin-A), active TB (not analyzed in this study). The last 3 pro- andHT6102(RBP4), respectively, (Table 2),suggesting teins (numbers 12 to 14) were obtained as a result of that identification of these proteins using peak lists of comparisons between Mtb-specific antigen-stimulated MS/MS spectra obtained from the LC-MS/MS system and unstimulated samples collected from patients with arefairly reliable since all these scores were significant Table1Thenumberofspotsthatmayshowdifferentialexpression A:ComparisonbetweenpatientswithactiveTBandcontrolsubjects P<0.02 0.02?P<0.05 0.05?P<0.10 Patientsversuscontrolsubjects 18 12 24 B:Comparisonbetweenstimulatedandunstimulatedconditions P<0.02 0.02?P<0.05 0.05?P<0.10 Patients Mtbantigensversusnostimuli 0 2 2 Mitogenversusnostimuli 3 5 11 Mtbantigensversusnostimuli 0 13 Controlsubjects 1 Mitogenversusnostimuli 2 83 8 Page6of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 which is similar to that observed above; however, the protein levels were widely distributed and the differ- ences in these levels did not reach significance in the controlsubjectscomparedwithpatientswithactiveTB (33,251 ? 2,572 versus 38,971 ? 11,001). Because the three clusterin spots altered after Mtb-specific antigen stimulation were notclearly distinguished byimmuno- blotting, wedidnotattemptanyfurtherdemonstration ofchangesinthesesignalsinourstudy.Instead,pooled sampleswererunona2Dgelandfollowedbyimmuno- blotting withanti-clusterin-aandanti-clusterin-banti- bodies(because clusterin consists ofclusterin-aand-b subunits) (Figure 4C). Based on immunoreactivity and pI values, the spots detected were confirmed to be clusterin-a.Morespecifically,thethreespotscomprised a subset of possible modified forms of clusterin-a that maybedetected. Detection ofdifferentially expressed proteins byELISA BecauseRBP4andfetuin-Alevelsdeterminedbyimmu- noblotting weresignificantly different between samples from patients with active TB and control subjects, we performed further quantitative ELISA to extend the measurements to plasma samples from 14 Japanese patientswithactiveTBand13age-,gender-,andethni- city-matched control subjects. Plasma RBP4 levels in patients with active TB (median = 23.6 μg/ml; IQR = Figure 3 Representative 2D gel image obtained from pooled proteinsamples.Differentiallyexpressedproteinssuccessfully 18.4-37.9)weresignificantlylowerthanthosefromcon- identifiedbyLC-MSwererepresentedonthegelusingthePDQuest trolsubjects(median=44.6μg/ml;IQR=34.6-53.8;P= softwaresuite.SeeTable2foradditionalinformation.Numbersin 0.0033; Figure 5A). Plasma fetuin-A levels in patients thefigurecorrespondtoserialnumbersinTable2. (median = 147.9 μg/ml; IQR = 115.8-159.6) were also significantly lower than those in control subjects above the 5% confidence threshold and no other pro- (median = 211.0 μg/ml;IQR = 186.7-264.6; P= 0.0002; teinswithcomparablescoresweredetectedforeachgel Figure 5B). No significant difference were observed spot(SeeAdditionalfile1:forsupportinginformation). in plasma VDBP levels between patients (median = Theseproteinswereinterestingbecauseoftheirpoten- 110.0 μg/ml; IQR = 85.2-151.3) and control subjects tial biological significance, and we therefore analyzed (median = 105.0 μg/ml; IQR = 88.1-215.6; P = 0.5441; themfurther. Figurenotshown). We simultaneously compared the protein levels in plasma immediately separated from EDTA-containing Confirmation ofdifferentially expressed proteins by immunoblotting bloodwiththoseinplasmasupernatantsobtainedfrom Immunoblot analysis was used to confirm differential heparinized blood as a negative control for IGRAafter expression of three proteins identified in patients with 18 h of incubation without stimulants. We found that active TB compared with control subjects (Figure 4A). thedifferencesbetweenthetwotypesofplasmasamples We measured band densities using the same samples were small (coefficient of variance (CV) = 10.5% for preparedforproteinconfirmation(Figure4B).Theband RBP4; CV = 5.0% for fetuin-A; CV = 6.6% for VDBP) densityofRBP4inpatientswithactiveTB(64,283arbi- and was in a range of variation generally accepted in traryunits?3,861)waslowerthanthatincontrolsub- ELISA (CV < 15%), indicating that the measurements jects(445,894?16,590),andfetuin-Aexpressioninthe obtained under the latter condition can be substituted patients was also lower (42,710 ? 7,580) than that in forthose obtained undertheformercondition. Indeed, control subjects (343,617 ? 58,923). These results are plasma RBP4 and fetuin-A levels in samples from consistent with those of2Dgelanalysis. Moreover, the JapanesepatientswithactiveTBweresignificantlylower band density of VDBP tended to be higher in samples thanthosefromcontrolsubjects,irrespectiveofplasma frompatientswithactiveTBthanfromcontrolsubjects, conditions(datanotshown). Page7of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 Table2Characteristicsofproteinsidentifiedinthisstudy A:ComparisonbetweenpatientswithactiveTBandcontrolsubjects 2D-DIGE LC-MS/MS b +/-P Condition Serial PDQ Swiss-Plot Mascotsearch Proteinname Da pI Number value SSP#a scoree Patientsversuscontrol 1 HT6102 0.0064 - RET4_HUMAN 72 Retinolbinding 23010 5.76 subjects protein4 2 HT2406 0.0097 - FETUA_HUMAN 75 a-2-HS-glycoprotein 39325 5.43 3 HT5401 0.0331 VTDB_HUMAN 52964 5.40 + 98 VitaminDbinding protein 4 HT2303 0.0419 + CO4A_HUMAN 86 ComplementC4A 192771 6.66 5 HT1012 0.0271 - APOC3_HUMAN 105 ApolipoproteinC-III 10852 5.23 6 HT5303 <0.001 APOA4_HUMAN 190 ApolipoproteinA-IV 45399 5.28 - 7 HT1016 0.0024 APOC2_HUMAN 61 ApolipoproteinC-II 11284 4.72 - B:ComparisonbetweenstimulatedandunstimulatedconditionsinactiveTB 2D-DIGE LC-MS/MS Condition +/-c Swiss-Plot Da pI Serial PDQ P Mascotsearch Proteinname value Number SSP# score Mitogen(versusnostimuli) 8 T3601 0.0917 - C1S_HUMAN 169 Complement-C1S 76684 4.86 T3403 0.0156 + 139 Kininogen-1 71957 6.34 9 KNG1_HUMAN T3105 0.0866 - ZA2G_HUMAN 45 33872 5.57 10 Zinc-a-2-glycoprotein 11 T4512 0.0061 A1BG_HUMAN 54273 5.58 - 76 a-1B-glycoprotein 0.0640d Mtbantigens(versusno 12 T4203 + CLUS_HUMAN 47 Clusterin 52495 5.89 stimuli) 0.0687d 13 T3107 + CLUS_HUMAN 50 Clusterin 52495 5.89 0.0732d 14 T4208 + EST 52495 5.89 – Clusterin aPDQSSP#isaPDQuestsoftwarestandardspotnumberindicatingtheuniquelocationofeachspotautomaticallyassignedona2Dgelandisessentialfor comparingthesamespotsondifferentgels. bTheaveragedensityofaspotfrom2D-DIGEishigher(+)orlower(-)inpatientswithactiveTBthanincontrolsubjects. cTheaveragedensityofthespotfrom2D-DIGEishigher(+)orlower(-)underthestimulatedconditionthanthatundertheunstimulatedcondition. dTheaveragedensityofthe3spots,T4203,T3107,andT4208,whichcorrespondtoasubsetofclusterin,wassignificantlyhigherinMtb-specificantigen- stimulatedthaninunstimulatedsamples(P=0.0014). eTheMascotsearchscoreindicatesthedegreeofcompatibilitybetweenmassspectrageneratedbythesampleandaminoacidsequenceswithintheproteinof interest. Wefurtherattemptedtoverifythedifferencesobserved Discussion withsamplesfromadifferentethnicandregionalpopula- Inthisstudy,weidentifiedTB-associatedproteinsfrom tion, i.e., samples collected from Vietnamese patients. whole blood supernatants. After the removal of 14 The two proteins identified above were measured in majorplasmaproteins,RBP4,fetuin-A,andVDBPwere plasma supernatants from Vietnamese patients with initiallyidentifiedasplasmaproteinsfromunstimulated activeTBandage-,gender-,andethnicity-matchedcon- samples for which the baseline levels differed between trol subjects. The samples from these Vietnamese the patients and control subjects. Immunoblotting patientswereobtainedfromanegativecontrolofIGRA results confirmed the differential expression of RBP4 after incubation without stimulants. RBP4 levels in andfetuin-A between the twogroups. Although VDBP patients with active TB (median =17.5 μg/ml; IQR = haspreviouslybeenidentifiedasabiomarkerformyco- 14.4-23.9)weresignificantlylowerthanthoseincontrol bacterialinfectionsincattle[9],thelevelofthisprotein subjects (median = 30.5 μg/ml; IQR = 25.9-40.8; P < didnotdiffersignificantlyinourstudybecauseoflarge 0.0001;Figure5A).Fetuin-Alevelsinpatientswithactive individual variations. The changes in VDBP levels may TB(median=210.7μg/ml;IQR=178.1-235.7)werealso not have been accurately immunologically assayed in significantlylowerthanthoseincontrolsubjects(median= ourstudy. 299.4μg/ml;IQR=265.1-363.2;P<0.0001;Figure5B). Clusterinisasecretedglycoproteininvolvedinapop- Moreover,bothproteinlevelswerenotsignificantlydiffer- tosis, inflammation, and tissue injury. It was differen- entbetweenIGRA-negativeandIGRA-positivesubgroups tially expressed in patients with active TB after ofthecontrolsubjects(datanotshown). stimulation and the intensities of the three spots Page8of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 RBP4 and fetuin-A levels were significantly lower in samples collected from patients with active TBthan in controlsubjects, indicating thatourfindings arerepro- ducible instudies using well-matched control subjects. However, as shown in Figure 5, the average plasma levels of these proteins differed between Japanese and Vietnamese control subjects. This suggests that unknown factors may systemically influence tested populations or the measurement of these markers. Because this variance is crucial in a clinical setting, ' furtherbasicaswellasclinicalinvestigationsareneces- sarytoaccuratelyassessthesemarkers. NosignificantdifferenceswereobservedinRBP4and fetuin-A levels in samples from IGRA-positive and -negativecontrolsubjects.Thissuggeststhatthesepro- teins levels are not affected by latent TB infection, but thattheypresumablychangeduringdiseaseprogression viaanunknownmechanism. Intriguingly,theliteraturesupportstheideathatRBP4 andfetuin-Aarefunctionallysignificantsincetheymay beinvolved in macrophage activation [10-12]. Retinoic acidhasbeenshowntostimulateandinducemonocyte - % % differentiation, leading toinhibition ofMtbmultiplica- & & tion in human macrophages [13]. RBP4 is the specific carrier protein for retinol (vitamin A)andhas recently !" # !" $ beendescribedasanadipokinethatcontributestoinsu- linresistance[13].Thisproteinisbelievedtomodulate Figure 4 Comparison of candidate differential proteins using pathophysiological processes during bacterial infection. immunoblotting.Sampleswerefirsttreatedtoremove14major Fetuin-Awasoriginally identified asafetalproteinand proteinsandthenanalyzedbyimmunoblottingtofacilitate has been shown to affect the development of many comparisonofRBP4,fetuin-A,andVDBPlevels.(A)Representative mammalian tissues. Moreover, the results of in situ immunoblotofsamplesfrompatientwithactiveTB(Patient-1) mRNA hybridization and immunocytochemical studies versusacontrolsubject(Control-1).(B)Banddensitieswere quantifiedinpatients(n=4)andcontrolsubjects(n=4).ThePMT in adult sheep have revealed that the main sites of responsesuggestedbanddensities.(C)Two-dimensional fetuin-A expression are hepatocytes and monocytes or immunoblotofpooledsampleswithdetectionofclusterin-aand macrophages in the spleen and bone marrow [14]. -b.CLUisclusterin. Fetuin-A is known to modulate various immune and metabolic responses. Previous reports have shown that corresponding to clusterin-a were elevated in whole fetuin-A deactivates macrophages, acts as an opsonin blood supernatant samples after incubation with Mtb- for cationic-deactivating molecules including spermine specific antigens. Thesespots appeartohaveshifted in [15], reduces TNF-a production and inflammatory both the dimensions on the gel, which suggests small responses [16], andenhancesphagocytosis ofapoptotic changes intheir molecular weights andIEPs. Itiscon- cellsandmacropinocytosisbyhumanmacrophages[17]. ceivable that post-translational modifications, such as Ontheotherhand,thisproteinisknowntobeapotent degradation and/or deglycosylation, occur via an enzy- inhibitorofsystemiccalcification [18]andisassociated maticreactionthataccompaniesimmunecellactivation. withtheincidenceofdiabetesmellitus[19]. However, wehavenotdemonstrated thatthis response Our study is the first to highlight the relationship is observed only when Mtb-specific antigens are co- between these twomarkers andTB,eventhough these incubated.Todeterminewhetherclusterinhasaroleas markerlevelsmaybeaffectedbyendogenousorexogen- a marker of TB or indicates more general response to ousfactorsandarepresumablynonspecifictoTBgiven antigenstimulation,wearecurrentlyattemptingtofind theirrelative abundanceinplasmaandthebroadspec- clearandsimplemethodsindetectingthesealterations trum of functional significance proposed in the above formassscreening. references. Subsequent ELISA results for samples from Japanese Nevertheless, performing a prospective cohort study and Vietnamese subjects confirmed that both plasma may help clarify the role of these proteins in TB. Page9of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 Figure 5Comparison ofRBP4, fetuin-A, andVDBP levels byELISA.Samples were analyzed byELISA tocompare RBP4and fetuin-A levels in14JapanesepatientswithactiveTBand13age-,gender-,andethnicity-matchedcontrolsubjects,aswellas25Vietnamesepatientswith activeTB,and50age-,gender-,andethnicity-matchedcontrolsubjects.Pvaluesshowingresultsofcomparisonsaredescribedinthetext.(A) RBP4levelsinpatientsversuscontrolsubjects.(B)Fetuin-Alevelsinpatientsversuscontrolsubjects. gelelectrophoresis;LC-MS:liquidchromatography-massspectrometry;ELISA: If within-individual variation in baseline levels is rela- enzyme-linkedimmunosolventassay;IEF:isoelectricfocusing;RBP-4:retinol tivelysmall,itcanbeusedtomonitorthecourseofdis- bindingprotein-4;VDBP:vitaminDbindingprotein;IQR:interquartilerange; easebefore,during,andaftertreatment.Furtherclinical ESTs:expressedsequencetags; studies on various conditions may better characterize Acknowledgements theseproteins.Singleuseofthesemarkersortheircom- TheauthorsthankDr.HaradaoftheResearchInstituteofTuberculosis;Drs. bined use with other promising biomarkers may be a NakamichiandHanadaoftheDepartmentofRespiratorymedicine,National CenterforGlobalHealthandMedicine;andT.TotsuandDr.Kanazawaof useful tool to aid the development of new effective theDepartmentofRespiratoryDiseases,ResearchInstitute,NationalCenter therapiesandvaccines. forGlobalHealthandMedicinefortheirassistancewiththeproject.This workwaspartlysupportedbyagrantofNationalCenterforGlobalHealth andMedicineandbyagrantfromtheProgramofJapanInitiativeforGlobal Conclusions ResearchNetworkonInfectiousDiseases(J-GRID),MEXT,Japan. We identified three TB-associated proteins, RBP4, fetuin-A, and clusterin, in whole blood supernatants Authordetails using a proteomic approach. We subsequently showed 1MolecularEnzymology,GraduateSchoolofAgriculturalScience,Tohoku that both plasma RBP4 and fetuin-A levels are signifi- University,1-1Amamiya-machi,Tsutsumidori,Aoba-ku,Sendai,Miyagi981- 8555,Japan. 2DepartmentofRespiratoryDiseases,ResearchInstitute, cantlyandreproduciblylowerinpatientswithactiveTB NationalCenterforGlobalHealthandMedicine,1-21-1Toyama,Shinjuku-ku, than in control subjects. These findings may help us Institute,NationalCenterforGlobalHealthandMedicine,1-21-1Toyama, Tokyo162-8655,Japan. 3DepartmentofDiabeticComplications,Research understandandmonitorthediseaseprocessinTB. Shinjuku-ku,Tokyo162-8655,Japan. 4DepartmentofMetabolicDisorder, ResearchInstitute,NationalCenterforGlobalHealthandMedicine,1-21-1 Toyama,Shinjuku-ku,Tokyo162-8655,Japan. 5SupramolecularBiology, Additional material InternationalGraduateSchoolofArtandSciences,YokohamaCityUniversity, 1-7-29Suehirocho,Tsurumi-ku,Yokohama,Kanagawa230-0045,Japan. Additionalfile1:FigureS1-Mascotsearchresults–Information 6DepartmentofRespiratoryMedicine,NationalCenterforGlobalHealthand Medicine,1-21-1Toyama,Shinjuku-ku,Tokyo162-8655,Japan. 7National abouttheidentifiedproteinsobtainedusingtheMascotserver .(A) MascotsearchresultforT2116(clusterin)(B)MascotSearchResultfor CenterforGlobalHealthandMedicine-BachMaiHospital(NCGM-BMH) T2103(clusterin)EST(C)MascotsearchresultforT1486(clusterin)(D) MedicalCollaborationCenter,78GiaiPhongSt.,Hanoi,Vietnam. 8Hanoi MascotsearchresultforHT2482(RET4=RBP4)(E)Mascotsearchresult TuberculosisandLungDiseaseHospital,44ThanhNhanRoad,Hanoi, forHT1248(fetuin-A)(F)MascotsearchresultforHT1240(VDBP). Vietnam. Authors’contributions TTcarriedouttheplasmaproteomestudies,participatedin2D-DIGEstudies, apartofimmunoassaysanddraftedthemanuscript.SSconceivedofthe Abbreviations study,andparticipatedinthestudyplanningandcoordinationandhelped TB:Tuberculosis;Mtb:Mycobacteriumtuberculosis;IFN-γ :interferon-gamma; todraftthemanuscript.KKcarriedouttheLC-MS/MSanalysis.ET,KYand IRGA:interferon-gamma-releaseassay;2D-DIGE:two-dimensionaldifference Page10of10 Tanakaetal.BMCInfectiousDiseases2011,11:71 19. IxJH,WasselCL,KanayaAM,VittinghoffE,JohnsonKC,KosterA,CauleyJA, HHhelpedtodesignthestudy.YKparticipatedinthestudydesignand overallsupervision.NK,NTLHandLTLparticipatedinmanagementand HarrisTB,CummingsSR,ShilipakMG:Fetuin-Aandincidentdiabetis analysisofdata.IMandMHparticipatedintheacquisitionofdata.TU mellitusinolderpersons.JAMA2008,300:182-188. helpedtodraftthemanuscript.NKparticipatedinthedesignofthestudy, performedstatisticalanalysisandhavegivenfinalapprovaloftheversionto Pre-publicationhistory bepublished.Allauthorsreadandapprovedthefinalmanuscript. Thepre-publicationhistoryforthispapercanbeaccessedhere: :10.1186/1471-2334-11-71 Theauthorsdeclarethattheyhavenocompetinginterests. Citethisarticleas:Tanakaetal.:Identificationoftuberculosis-associated proteinsinwholebloodsupernatant.BMCInfectiousDiseases201111:71. Received:7October2010 Accepted:22March2011 Published:22March2011 References 1. WHO:GlobalTuberculosisControl:Epidemiology,Strategy,Financing. Geneva.WHOPress;2009,7-13. 2. JacobsenM,RepsilberD,GuschmidtA,etal:Candidatebiomarkersfor discriminationbetweeninfectionanddiseasecausedbyMycobacterium tuberculosis.JMolMed2007,85:613-621. 3. MistryR,CliffJM,ClaytonCL,etal:Geneexpressionpatternsinwhole bloodidentifysubjectsatriskforrecurrenttuberculosis.JInfectDis2007, 195:357-365. 4. AgranoffD,Fernandez-ReyesD,PapadopoulosMC,RojasSA,HerbsterM, LoosemoreA,TarelliE,SheldonJ,SchwenkA,PollokR,RaynerCF, KrishnaS:Identificationofdiagnosticmarkersfortuberculosisby proteomicfingerprintingofserum.Lancet2006,368:1012-1021. ChegouNN,BlackGF,KiddM,vanHeldenPD,WalzlG:Hostmarkersin 5. QuantiferonsupernatantsdifferentiateactiveTBfromlatentTB infection:preliminaryreport.BMCPulmMed2009,9:21. 6. WallisRS,PaiM,MenziesD,DohertyTM,WalzlG,PerkinsMD,ZumlaA: Biomarkersanddiagnosticsfortuberculosis:progress,needs,and translationintopractice.Lancet2010,375:1920-1937. 7. TodaT,SugimotoM,OmoriS,MatsuzakiT,FuruichiY,KimuraN:Proteomic analysisofEpstein-Barrvirus-transformedhumanB-lymphoblastoidcell linesbeforeandafterimmortalization.Electrophoresis2000,21:1814-1822. 8. SaekiK,YasugiE,OkumaE,BreitSN,NakamuraM,TodaT,KaburagiY, YuoA:Proteomicanalysisoninsulinsignalinginhumanhematopoietic cells:identificationofCLIC1andSRp20asnoveldownstreameffectors ofinsulin.AmJPhysiolEndocrinolMetab2005,289:E419-428. 9. SethM,LamontEA,JanagamaHK,WiddelA,VulchanovaL,StabelJR, WatersWR,PalmerMV,SreevatsanS:Biomarkerdiscoveryinsubclinical mycobacterialinfectionsofcattle.PLoSOne2009,4:e5478. 10. CrowleAJ,RossEJ:Inhibitionbyretinoicacidofmultiplicationofvirulent tuberclebacilliinculturedhumanmacrophages.InfectImmun1989, 57:840-844. 11. JersmannHP,DransfieldI,HartSP:Fetuin/alpha2-HSglycoprotein enhancesphagocytosisofapoptoticcellsandmacropinocytosisby humanmacrophages.ClinSci2003,105:273-278. 12. WangH,ZhangM,BianchiM,SherryB,SamaA,TraceyKJ:Fetuin(alpha2- HS-glycoprotein)opsonizescationicmacrophagedeactivatingmolecules. ProcNatlAcadSciUSA1998,95:14429-14434. 13. YangQ,GrahamTE,ModyN,PreitnerF,PeroniOD,ZabolotnyJM,KotaniK, QuadroL,KahnBB:Serumretinolbindingprotein4contributestoinsulin resistanceinobesityandtype2diabetes.Nature2005,436:356-362. 14. DziegielewskaKM,BrownWM,DealA,FosterKA,FryEJ,SaundersNR:The expressionoffetuininthedevelopmentandmaturationofthe hemopoieticandimmunesystems.HistochemCellBiol1996,106:319-330. 15. WangH,ZhangM,BianchiM,SherryB,SamaA,TraceyKJ:Fetuin(α2-HS- Submit your next manuscript to BioMed Central glycoprotein)opsonizescationicmacrophagedeactivatingmolecules. ProcNatlAcadSciUSA1998,95:14429-14434. and take full advantage of: 16. OmbrellinoM,WangH,YangH,ZhangM,VishnubhakatJ,FrazierA, ScherLA,FriedmanSG,TraceyKJ:Fetuin,anegativeacutephaseprotein, • Convenient online submission attenuatesTNFsynthesisandtheinnateinflammatoryresponseto • Thorough peer review carrageenan.Shock2001,15:181-185. 17. JersmannHP,DransfieldI,HartSP:Fetuin/α2-HSglycoproteinenhances • No space constraints or color ,gure charges phagocytosisofapoptoticcellsandmacropinocytosisbyhuman • Immediate publication on acceptance macrophages.ClinSci(Lond)2003,105:273-278. • Inclusion in PubMed, CAS, Scopus and Google Scholar 18. Jahnen-DechentW,SchaferC,KettelerM,McKeeMD:Mineralchaperones: aroleforfetuin-Aandosteopontinintheinhibitionandregressionof • Research which is freely available for redistribution pathologiccalcification.JMolMed2008,86:379-389. Submit your manuscript at www.biomedcentral.com/submit