Differential RD-1-specific IFN-γ host responses to diverse Mycobacterium tuberculosis strains in HIV-uninfected persons may be explained by genotypic variation in the ESX-1 region

Short Communication

Differential RD-1-speci fi c IFN- g host responses to diverse

Mycobacterium tuberculosis strains in HIV-uninfected persons may be explained by genotypic variation in the ESX-1 region

Michele Tomasicchio

^a

, Jason Limberis

^a

, Ruben van der Merwe

^b

, Rachael Jacobson

^c

, Richard Meldau

^a

, Grant Theron

^b,a

, Mark Nicol

^c

, Rob Warren

^b

, Keertan Dheda

^a,c,d,

*

aCentreforLungInfectionandImmunity,DivisionofPulmonology,DepartmentofMedicineandUCTLungInstitute&SouthAfricanMRC/UCTCentreforthe StudyofAntimicrobialResistance,UniversityofCapeTown,CapeTown,SouthAfrica

bDepartmentofScienceandTechnology/NationalResearchFoundationCentreofExcellenceforBiomedicalTuberculosisResearch,SouthAfricanMedical ResearchCouncilCentreforTuberculosisResearch,DivisionofMolecularBiologyandHumanGenetics,FacultyofHealthSciences,StellenboschUniversity, CapeTown,SouthAfrica

cInstituteofInfectiousDiseasesandMolecularMedicine,UniversityofCapeTown,CapeTown,SouthAfrica

dFacultyofInfectiousandTropicalDiseases,DepartmentofInfectionBiology,LondonSchoolofHygieneandTropicalMedicine,London,UK

ARTICLE INFO

Articlehistory:

Received29January2020

Receivedinrevisedform19April2020 Accepted21April2020

Keywords:

Mycobacteriumtuberculosis Interferon-gammareleaseassay Hostimmuneresponse Strain

ABSTRACT

Objectives: Between-person variability in T-cell-specific interferon-gamma release assay (IGRA) responsesanddiscordancebetweenIGRAtestformatsarepoorlyunderstood.

Methods:WeevaluatedtheIFN-gresponses(QuantiFERON-TBGold-In-Tube[QFT-GIT]andTSPOT-TB) stratifiedaccordingtotheMycobacteriumtuberculosisspoligotypeofthecultureisolateobtainedfromthe samepatientswithconfirmedactivetuberculosis(n=91).Wefurtheranalyseddifferenceswithinthe RD-1-encodingESX-1regionbetweenthedifferentstraintypesusingwholegenomesequencing.

Results:InHIV-uninfectedpatients,TSPOT.TBandQFT-GITIFN-gresponseswere5-fold(p<0.01)and2- foldhigher(p<0.05)forthoseinfectedwithfamily33comparedtotheLAMstrain(additionally,TSPOT.

TBresponseswere5.6-fold[p<0.05]and2.6-foldhigher[p<0.05]forthepatientsinfectedwiththe family33versustheX strainandBeijingversustheLAMstrain,respectively).Multivariateanalysis revealed that strain type (determined by spoligotyping) was independently associated with the magnitudeoftheIGRAresponse(variedbyIGRAtesttype)andthisislikelyexplainedbyvariabilityinthe ESX-1regionofMycobacteriumtuberculosis(determinedbynext-generationsequencing).

Conclusions:Thesedatahaveimplicationsfortheunderstandingofbetween-personheterogeneityin IGRA responses, Mycobateriumtuberculosis-specific host immunity, and the discordance between differentIGRAtestformats.

©2020TheAuthors.PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.

ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc- nd/4.0/).

Introduction

ThereisincreasingevidencethatmembersoftheMycobacteri- umcomplexare differentiallyvirulent bothinvivo andinvitro

(Aguilar et al., 2010, de Jong et al., 2008). The molecular mechanisms that underpin these differences remain unclear, however, the 6-kDa early secretory antigenic target (ESAT-6) system-1 (ESX-1) is likely involved as it is essential for Mycobacterium tuberculosis (M.tb) virulence (Gey Van Pittius etal.,2001).

WeinvestigatedwhetherM.tbstrainsdifferentiallymodulate hostIFN-

g

responsesinatuberculosis(TB)endemicsetting.We furtheranalysedgeneticdifferenceswithintheESX-1regionofan independentsetofstrainsusingwholegenomesequencing(WGS).

*Corresponding authorat: Lung Infection and Immunity Unit, Divisionof PulmonologyandUCTLungInstitute,DepartmentofMedicine,UniversityofCape Town,SouthAfrica.

E-mailaddress:keertan.dheda@uct.ac.za(K.Dheda).

https://doi.org/10.1016/j.ijid.2020.04.053

1201-9712/©2020TheAuthors.PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

InternationalJournalofInfectiousDiseases96(2020)240–243

ContentslistsavailableatScienceDirect

International Journal of Infectious Diseases

j o u r n a lh o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i j i d

Thisisthefirststudytoshowthattherearemultiplestrain-specific geneticdifferencesintheESX-1regionthatdifferentiallyregulate hostIFN-

g

responses.

Methods

Studysite,populationanddiagnostictests

Werecruited645patientswithsuspectedTBfromthreestudy sitesinCapeTown,SouthAfrica. Informedwrittenconsentwas obtainedfrompatients18yearsorolderandwhopresentedtoa peri-urbanprimary-careTBclinic.Patientswereenrolledwhohad one or more symptom suggestive of pulmonary TB (n = 645) accordingtodefinedWHOcriteria(Bassettetal.,2010,Getahun etal.,2011),wereabletoexpectoratetwosputumspecimensand hadnotreceivedanti-TBtreatmentwithintheprevious60days.

Sputumwasobtainedforliquidcultureandculturepositivitywas usedasthereferencestandardforTB.EachpatientreceivedaHIV testaftercounselling.Wholebloodwascollectedbyvenepuncture andstimulatedimmediatelywiththeregionofdifference1(RD-1) antigensfor useintheTSPOT.TB(OxfordImmunotech,UK)and QFT-GIT (Qiagen, Germany) interferon-gamma release assays (IGRAs)asindicatedpreviously(Theronetal.,2012).

Straintypingandsequenceanalysis

Spoligotyping was performed according to the Kamerbeek method(Kamerbeeketal.,1997).

Statistics

Seeonlinedatasupplement.

Results

Patientsandsamples

The study overview and baseline characteristics of the 645 patientsisshowninFigureS1andTableS1,respectively.

ComparisonofIFN-

g

responsesaccordingtoM.tbspoligotype

The IFN-

g

responses for TSPOT.TB and QFT-GIT of the 91 patientsstratifiedbystraintypearedepictedinFigure1andS2.

ThereweredifferencesintheIFN-

g

responsesforTSPOT.TB(p= 0.005)andQFT-GIT(p=0.032)inHIV-uninfectedpatients(n=51) harbouringdifferentM.tbstrains(Figure1).ThedifferencesinIFN-

g

readoutsforTSPOT.TBwereindependentlystrain-specificand couldnotbeexplainedbyconfoundingfactorssuchasage,sex, smear grade, and race as determined by multivariate and univariate analysis (Table 1 and S2). Similarly, differences in IFN-

g

readouts for QFT-GIT were significantly associated with strain and not with age,sex or smear grade and werestrain- specific.However,theQFT-GITanalysisshowedthatracewasalso significantly associated withdifferences in IFN-

g

readouts. We further show that the strain-induced IFN-

g

responses did not changewithculturetime topositivity, meanXpert MTB/RIF CT

valueandchestradiologicalscores(FigureS3,S4andS5).

SequenceinvestigationoftheESX-1region

WeassessedthegeneticdiversitybetweentheBeijing,X,and LAM strains in the ESX-1 region using WGS data from a conveniencedatasetof104isolatescollectedintheCapeTown, Western Cape to potentially explain the differences in IFN-

g

readouts betweenthedifferentstrains (these isolates werenot fromthesamepatientswhereIGRAtestingwasperformed).

A discriminant analysis of principal components was per- formedtoinvestigateifwecouldcorrectlyidentifythestraintypes using mutations in these regions using a subset of 16 single nucleotidepolymorphisms(SNPs)thatoccurredinourdatasetand 429 previously described isolates from a global collection of strains; 2 principal componentswere retained (Table S4) (Coll et al., 2014). This model correctly typed over 99% (428/429;

TableS6)oftheglobalisolates.Itthusappearsthatmutationsin theESX-1regionarestrain-specificandallowfortheidentification ofBeijing,XandLAMstrains.

Discussion

Collectivelythesedatasuggestthatthestraintype,andpossibly the ESX-1 genotype, modulates the RD-1 antigen specific responses in humans. This is the first study to outline strain- specificmultiplemutationsintheESX-1geneandlinkthistoRD-1- specificTh1responsesinTBpatientsfromanendemicsetting.Our data are in keeping withanimal models that showdifferential virulenceofM.tbstrainsinvivo(Mancaetal.,2001).

Weshowthat theBeijingstrainis associatedwiththemost mutations in the ESX-1 region (compared to M.tb H37Rv) compared to the LAM and X strains (for example, we found mutationsinthemprBandespAgenes)whichmaypartlyexplain thedifferencesinthehostimmuneresponsesobserved(GeyVan Pittiusetal.,2001).

The LAM strain was associated with the most missense mutationsintheeccCb1andespBgenes,whicharepro-virulent and essential for thesecretion of ESAT-6/CFP-10 (Brodin et al., 2006).It iswellestablished thatEspA/EspC and theEspA/EspB dimersareimportantforsecretionofESX-1proteinsandvirulence (Fortuneetal.,2005).Thus,mutationsinespB,asdemonstrated herefortheLAMstrain,mayexpressdysfunctionalEspB,resulting indefectiveESX-1secretionandreducedvirulence.

Thereareseverallimitationstoourstudy.Weutilisedallofthe DNAfromtheisolatesforspoligotypingand,asaresult,nonewas available for sequencing.We therefore appliednext generation WGStoaconveniencesetofsamplesfromthesamegeographical location to further explore our hypothesis. We show that the mutationswithintheESX-1regionoftheindividualstrainsare conservedusingaglobalcollectionofisolates(Figure1C),andasa result,theimmunomodulatorycapacityoftheindividualstrains included in theWGSversus theIGRA analyses are likely tobe similar.Wecouldnotmakeanyinferenceaboutgeneticchangesin theESX-1regionoffamily33asWGSdatawasunavailableforthis strain.However,family33isquiterareinoursettingrepresenting 6.7% of 904South Africanisolates collectedbetween2006 and 2016(Togoetal.,2017).Spoligotypingwasnotperformedonallthe clinicalisolateswhereTSPOT.TBandQFT-GITwereperformeddue tounavailabilityofthecultureisolatesincluding,limitedmaterial for initial bio-banking, contamination, or failure to grow. By contrast,someavailable isolates(n=51) werenotspoligotyped because TSPOT.TB and QFT-GIT were not available for these samples(indeterminateresultsetc.).However,thesesampleswere unlikely to have influenced our findings because a sensitivity analysisrevealedthattherewerenodemographicdifferences(i.e.

race, HIV-status, smear-status) between the groups (included versusexcludedfromtheanalysis).Thestrain-specificIFN-

g

IGRA responseswereonlyobservedintheHIV-uninfectedsamplesand nottheHIV-infectedsamples.Multi-variantanalysisrevealedthat HIV positivity couldinfluence IGRA responsesand this is most likelyduetodepletionoftheIFN-

g

producingCD4+T-cellsduring HIVinfection.

M.Tomasicchioetal./InternationalJournalofInfectiousDiseases96(2020)240–243 241

Figure1.DifferentialhostIFN-gresponsesinHIV-uninfectedpatientsmaybeexplainedbygenotypicdifferencesintheESX-1region.PBMCswereisolatedfrom51HIV- uninfectedpatientsandstimulatedwiththeM.tb-specificantigens,and(A)TSPOT.TBand(B)QFT-GITwereperformedasdescribedbythemanufacturer.ThemeanTSPOT.TB IFN-gresponsewas5-foldhigherinpatientsinfectedwiththefamily33strainascomparedtotheLAMstrain(564SFCs/106PBMCsvs111SFCs/106PBMCs;p<0.01).

Similarly,themeanTSPOT.TBIFN-gresponsewas5.6-foldhigherinpatientsinfectedwiththefamily33strainascomparedtotheXstrain(564SFCs/106PBMCsvs100SFCs/

106PBMCs;p<0.05).ThemeanTSPOT.TBIFN-gresponsewas2.6-foldhigherinpatientsinfectedwiththeBeijingstraincomparedtotheLAMstrain(292SFCs/106PBMCsvs 111SFCs/106PBMCs;p<0.05).Interestingly,therewerenodifferencesinIFN-gresponsesobservedforHIV-uninfectedpatientsinfectedwithBeijingversustheLAMortheX strainforQFT-GIT.ThedifferencesinIFN-gresponsesobservedforTSPOT.TBversusQFT-GITcouldbeasaconsequenceofQFT-GITutilizinganadditionalantigen(TB7.7).We foundthatpatientsinfectedwithfamily33(8.7IU/ml)strainshavea2-foldincreaseinmeanIFN-gexpressioncomparedtopatientsinfectedwiththeLAM(3.9IU/ml;p<

242 M.Tomasicchioetal./InternationalJournalofInfectiousDiseases96(2020)240–243

Insummary,RD-1-specificIFN-

g

responsesareassociatedwith thespecificinfectingM.tb strainin patientsfroma TBendemic setting. This is likely explained by strain-specific genomic variabilityof theESX-1region.Thesedatahaveimplicationsfor understandinghostimmunitytoM.tb,interpretingdiagnostictest resultsandthevariabilityinthemagnitudeofIGRAresponses,and thediscordancebetweendifferentIGRAtestformats.

Conflictofinterest

Noconflictofinteresttodeclare.

Fundingsource

KDacknowledgesfundingfromtheEuropeanandDeveloping CountriesClinicalTrialsPartnership(EDCPT;TMA-2015SF,TMA- 1051-TESAII,TMA-CD2015), theSouth African MedicalResearch Council (SAMRC; RFA-EMU-02-2017), UK MRC (MR/S03563X/1) andtheWelcomeTrust(MR/S027777/1).

Ethicalapproval

ThestudywasapprovedbytheUniversityofCapeTownHealth SciencesEthicsCommittee(HREC#307/2014).

Acknowledgments

Computationswereperformedusingfacilitiesprovidedbythe University of Cape Town's ICTS High Performance Computing team:http://hpc.uct.ac.za.ThankstoMrMohammedFadulforhis assistancewiththestatistics.

AppendixA.Supplementarydata

Supplementarymaterialrelatedto thisarticle can befound, inthe onlineversion,atdoi:https://doi.org/10.1016/j.ijid.2020.04.053.

References

AguilarD,HanekomM,MataD,GeyvanPittiusNC,vanHeldenPD,WarrenRM,etal.

Mycobacterium tuberculosisstrainswith the Beijing genotype demonstrate variability in virulenceassociated with transmission. Tuberculosis(Edinb) 2010;90(5):319–25.

BassettIV,WangB,ChettyS,GiddyJ, LosinaE,Mazibuko M,etal. Intensive tuberculosisscreeningforHIV-infectedpatientsstartingantiretroviraltherapy inDurban,SouthAfrica.ClinInfectDis2010;51(7):823–9.

BrodinP,MajlessiL,MarsollierL,deJongeMI,BottaiD,DemangelC,etal.Dissection ofESAT-6system1ofMycobacteriumtuberculosisandimpactonimmunoge- nicityandvirulence.InfectImmun2006;74(1):88–98.

CollF,McNerneyR,Guerra-AssuncaoJA,GlynnJR,PerdigaoJ,ViveirosM,etal.A robustSNPbarcodefortypingMycobacteriumtuberculosiscomplexstrains.Nat Commun2014;5:4812.

deJongBC,HillPC,AikenA,AwineT,AntonioM,AdetifaIM,etal.Progressionto activetuberculosis,butnottransmission,variesbyMycobacteriumtuberculosis lineageinTheGambia.JInfectDis2008;198(7):1037–43.

FortuneSM,JaegerA,SarracinoDA,ChaseMR,SassettiCM,ShermanDR,etal.

Mutuallydependentsecretionofproteinsrequiredformycobacterialvirulence.

ProcNatlAcadSciUSA2005;102(30):10676–81.

Getahun H, Kittikraisak W, Heilig CM, Corbett EL, Ayles H, Cain KP, et al.

Developmentofastandardizedscreeningrulefortuberculosisinpeopleliving withHIVinresource-constrainedsettings:individualparticipantdatameta- analysisofobservationalstudies.PLoSMed2011;8(1)e1000391.

GeyVanPittiusNC,GamieldienJ,HideW,BrownGD,SiezenRJ,BeyersAD.TheESAT- 6geneclusterofMycobacteriumtuberculosisandotherhighG+CGram-positive bacteria.Genomebiology2001;2(10)RESEARCH0044.

KamerbeekJ,SchoulsL,KolkA,vanAgterveldM,vanSoolingenD,KuijperS,etal.

SimultaneousdetectionandstraindifferentiationofMycobacteriumtubercu- losisfordiagnosisandepidemiology.JClinMicrobiol1997;35(4):907–14.

MancaC,TsenovaL,BergtoldA,FreemanS,ToveyM,MusserJM,etal.Virulenceofa Mycobacteriumtuberculosisclinicalisolateinmiceisdeterminedbyfailureto induceTh1typeimmunityandisassociatedwithinductionofIFN-alpha/beta.

ProcNatlAcadSciUSA2001;98(10):5752–7.

TheronG,PeterJ,LendersL,vanZyl-SmitR,MeldauR,GovenderU,etal.Correlation ofMycobacteriumtuberculosis-specificandnon-specificquantitativeTh1T-cell responseswithbacillaryloadinahighburdensetting.PloSOne2012;7(5) e37436.

TogoACG,KodioO,DiarraB,SanogoM,CoulibalyG,BaneS,etal.Themostfrequent Mycobacteriumtuberculosiscomplexfamiliesinmali(2006-2016)basedon spoligotyping.IntJMycobacteriol2017;6(4):379–86.

0.05)orBeijing(4.2IU/ml;p<0.05)strainsforQFT-GIT.Differencesinstrain-specificIFN-gresponsesforbothTSPOT.TBandQFT-GITwereonlyobservedinHIV-uninfected patientsandnotinHIV-infectedpatients(FigureS2AandB).Thehorizontalbarshowsthemedianmagnitudeoftheresponseswhiletheerrorbarsrepresenttheinterquartile range.*,**,and***indicatep0.05,p0.01andp0.005,respectively.(C)Aneighbourjoiningtreeconstructedusingtheinter-isolateManhattandistancesintheESX-1 region.TheredscalebarisequivalenttooneSNPdifference.ThetreeisrootedtoH37Rv(AL123456.3).Thetabularbargraphtotherightshowsthepresence(red)orabsence (green)ofaparticularmutationineachstrain.

Table1

MultivariateregressionanalysesshowingstrainandracespecificassociationswithaheightenedIGRAresponseinculture-positive,HIV-uninfectedactiveTBcases(n=43) fromanendemicsetting.PleaseseeSupplementaryTable2and3foradditionalunivariateandmultivariateanalyses.

TSPOT.TB QFT-GIT

Term Variable Sensitivityestimate(95%CI) p-Value Sensitivityestimate(95%CI) p-Value

Age Years 2.52( 9.45,4.40) 0.464 0.05( 0.14,0.031) 0.209

Sex Male 109.04( 91.64,309.72 0.277 0.68( 2.88,1.52) 0.533

Race Mixed 93.23( 125.08,311.54) 0.391 2.82(0.54,5.10) 0.017

Smeargrade* Scanty 102.87( 438.88,233.15) 0.537 2.88( 7.44,1.68) 0.206

1+ 8.56( 232.04,249.17) 0.943 1.93( 4.38,0.51) 0.116

2+ 64.17( 155.89,284.22) 0.557 0.13( 2.46,2.72) 0.921

3+ 14.64( 292.38,263.09) 0.915 0.72( 2.56,3.99) 0.658

Straintype Beijing 171.11( 30.17,372.37) 0.093 0.64( 2.82,1.53) 0.552

Family33 447.34(152.50,742.18) 0.004 4.20(0.91,7.49) 0.014

X 90.902( 421.74,239.94) 0.580 0.05( 0.14,0.031) 0.918

* SmeargradewasclassifiedaccordingtostandardWHOcriteria:Scanty=1–9acidfastbacilli(AFB)/100oilimmersionfields.+1=10–99AFB/100oilimmersionfields.+2= 1–10AFB/50oilimmersionfields.+3=>10AFB/20oilimmersionfields.

M.Tomasicchioetal./InternationalJournalofInfectiousDiseases96(2020)240–243 243