Deep sequencing of blood and gut T-cell receptor β-chains reveals gluten-induced immune signatures in celiac disease

1 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 www.na u e.com/scien i ic epo s Deep sequencing o blood and gu T-cell ecep o β-chains e eals glu en-induced immune signa u es in celiac disease Dawi A. Yohannes1,2, Tobias L. F ei ag1,3, And ea de Kauwe1,2, Ka i Kaukinen4, Kalle Ku ppa5, Pi jo Wacklin6, Ma kku Mäki5, T. Pe e i A s ila1,3, Robe P. Ande son7,9, Da io G eco8,10 & Päi i Saa alainen1,2 Celiac disease (CD) pa ien s moun an abno mal immune esponse o glu en. T-cell ecep o (TCR) epe oi es di ec ed o some immunodominan glu en pep ides ha e p e iously been desc ibed, bu he global immune esponse o in i o glu en exposu e in CD has no been sys ema ically in es iga ed ye . He e, we cha ac e ized signa u es associa ed wi h glu en di ec ed immune ac i i y and iden i ied glu en-induced T-cell clono ypes om o al blood and gu TCR epe oi es in an unbiased manne using immunosequencing. CD pa ien o al TCR epe oi es showed inc eased o e lap and subs an ially al e ed TRBV-gene usage in bo h blood and gu samples, and inc eased di e si y in he gu du ing glu en exposu e. Using di e en ial abundance analysis, we iden i ied glu en-induced clono ypes in each pa ien ha we e composed o a la ge p i a e and an impo an public componen . Hie a chical clus e ing o public clono ypes associa ed wi h die a y glu en exposu e iden i ied subse s o highly simila clono ypes, he mos p oli e a i e o which showing signi ican en ichmen o he mo i ASS[LF] R[SW][TD][DT][TE][QA][YF] in PBMC epe oi es. These esul s show ha CD-associa ed clono ypes can be iden i ied and ha common glu en associa ed immune esponse ea u es can be cha ac e ized in i o om o al epe oi es, wi h po en ial use in disease s a i ica ion and moni o ing. Celiac disease (CD) is a complex diso de wi h an o e all es ima ed p e alence o 1% among people o Eu opean ances y1. I is cha ac e ized by small in es inal illous a ophy leading o nu ien malabso p ion bu may man- i es wi h a wide ange o gas oin es inal and ex a-in es inal symp oms. In pa ien s, ce eals con aining glu en, in pa icula whea , ba ley and ye, ac i a e glu en-speci ic immuni y leading o disease elapse. Consequen ly, a s ic li e-long glu en ee die (GFD) is cu en ly he only a ailable ea men o CD. The mos impo an gene ic de e minan s o suscep ibili y o CD a e Human Leukocy e An igen (HLA) alleles. Abou 90% o CD pa ien s ca y HLA-DQA1*05 and -DQB1*02 ha oge he encode HLA-DQ2.5, while he emainde ca y HLA-DQA1*0301 and -DQB1*0302 (HLA-DQ8), and/o HLA-DQA1*02 and -DQB1*02 (HLA-DQ2.2). HLA-DQ2.5, DQ2.2 and/o DQ8 a e ound in abou hal o he gene al popula ion and a e neces- sa y bu no su icien o de eloping he disease. CD4+ T-helpe cells speci ic o glu en epi opes p esen ed by HLA-DQ2.5, HLA-DQ2.2, o HLA-DQ8 a e conside ed cen al o he pa hogenesis o CD2. Sys emic adminis a ion o pep ides wi h immunodominan epi opes o glu en-speci ic CD4+ T-cells causes diges i e symp oms ha a e ypically associa ed wi h glu- en inges ion3. CD4+ T-cells speci ic o glu en a e p esen in he small in es ine2 and ci cula e a inc eased 1Resea ch P og ams Uni , Immunobiology, Uni e si y o Helsinki, Helsinki, Finland. 2Depa men o Medical and Clinical Gene ics, Uni e si y o Helsinki, Helsinki, Finland. 3Depa men o Bac e iology and Immunology, Uni e si y o Helsinki, Helsinki, Finland. 4Depa men o In e nal Medicine, Tampe e Uni e si y Hospi al and Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland. 5Cen e o Child Heal h Resea ch, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al, Tampe e, Finland. 6Finnish Red C oss Blood T ans usion Se ice, Helsinki, Finland. 7Wal e and Eliza Hall Ins i u e o Medical Resea ch, Melbou ne, Aus alia. 8Ins i u e o Bio echnology, Uni e si y o Helsinki, Helsinki, Finland. 9P esen add ess: ImmusanT, Inc., Camb idge, MA, USA. 10Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland. Co espondence and eques s o ma e ials should be add essed o P.S. (email: [email p o ec ed]) Recei ed: 7 July 2017 Accep ed: 6 Decembe 2017 Published: xx xx xxxx OPEN www.na u e.com/scien i ic epo s/ 2 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 equencies six days a e glu en ein oduc ion4. Pe iphe al blood collec ed a e sho - e m glu en challenge ha - bou s expanded popula ions o gu -homing, e ec o memo y, CD4+ T-cells speci ic o glu en. In HLA-DQ2.5+ CD pa ien s, glu en- eac i e CD4+ T-cells de ec ed in blood by IFNγ ELISpo a e sho - e m whea , ba ley o ye challenge p e e en ially a ge immunodominan epi opes in one o h ee sho pep ides5. Glu en challenge in CD pa ien s also inc eases he equencies o CD8+ αβ and γδ T-cells, bu hei an igen speci ici ies ha e no been de e mined6–8. CD4+ e ec o T-cells in blood speci ic o he mos commonly ecognized epi opes, HLA-DQ2.5-glia-α2 and HLA-DQ2.5-glia-ω2, show biased bu dis inc pai ing o T-cell ecep o (TCR)α and TCRβ genes: TRBV7- 2 wi h TRAV26-1 in T-cells speci ic o HLA-DQ2.5-glia-α2, and TRBV4 wi h TRAV4 in T cells speci ic o HLA-DQ2.5-glia-ω29–11. CD4+ T-cells speci ic o ei he o hese epi opes showed ea u es o an igen d i en selec ion such as con e gen ecombina ion and semi-public esponse9–11. The semi-public esponse sugges s a common disease mechanism ac oss pa ien s, since andom clono ype sha ing be ween indi iduals is unlikely, owing o he highly di e se T-cell epe oi e in indi iduals gene a ed ia he V(D)J ecombina ion machine y12–14. Ano he subse o glu en eac i e CD4+ T-cells speci ic o HLA-DQ2.5-glia-α1a had a biased usage o TRBV20- 1 o TRBV29-115. Simila ly, in HLA-DQ8+ CD pa ien s, HLA-DQ8- es ic ed glu en eac i e T-cells show biased usage o TRBV9-116. Despi e such de ailed cha ac e iza ions o isola ed T-cells speci ic o selec ed immunodominan glu en pep- ides, a global iew o he TCR epe oi e di ec ed o in i o glu en exposu e is lacking. Ad ances in nex gene a ion sequencing (NGS) now p o ide an oppo uni y o explo e he ull T-cell esponse induced by glu en i espec i e o whe he glu en o o he an igens a e a ge ed. This no el app oach could p o- ide ex ensi e de ail complemen a y o wha has been lea ned om gliadin- e ame based app oaches8–11,17 and sys ema ic glu en epi ope mapping s udies5,18. In his s udy, we applied deep TCRB CDR3 sequencing o cha ac- e ize CD pa ien immune epe oi es du ing glu en exposu e, and o iden i y glu en eac i e T-cell clono ypes in an unbiased manne . Resul s TCR epe oi e da a was gene a ed om subjec s in h ee expe imen s (Table1 and Supplemen a y TableS1). We ob ained an a e age o 14694 unique p oduc i e nucleo ide TCRB clono ypes om 544066 eads o each p e- o al glu en challenge (day 0) and pos -challenge (day 6) pa ien PBMC sample. Fo he gu biopsy samples, he e was an a e age o 8784 unique p oduc i e nucleo ide TCRB clono ypes om 1414060 eads pe pa ien sample. Clono ypes om he gu we e bigge in size ( eads/clone) compa ed o clono ypes in PBMC epe oi es, e lec - ing di e ences in he p opo ion o nai e and memo y T-cells in he issues (Supplemen a y TableS1). Fo so ed samples wi h CD4+ IFNg+ cells (0.5–3.5% o ga ed lymphocy es ac oss all samples), we ob ained an a e age o 739 unique p oduc i e clono ypes om 114111 eads pe indi idual. Glu en-speci ic CD4+ IFNg+ TCRB epe oi es a e as di e se as o al pa ien PBMC epe oi es while pa ien gu epe oi e di e si y is inc eased du ing ac i e CD. Signi ican changes in di e si y ollowing an igen exposu e indica e une en shi in epe oi es, such as mono-, oligo- o poly- clonal expansion, wi h possible implica ions in disease de elopmen 19,20. To e alua e TCR epe oi e di e si y and a oid possible bias due o di e ences in sequencing dep h, we downsampled all samples o 27077 eads ( he size o he second Subjec Diagnosis Sex Age HLA-DQ s a us Cu en die GFD du a ion TG IgA se ology*Glu en challenge Sample Expe imen CD005 CD F 61 DQ2.5/DQ2.5 GFD 1 y nega i e Whea 3d PBMC 1,2 CD006 CD F 61 DQ2.5/DQ5 GFD 16 y nega i e Whea 3d PBMC 1 CD0011 CD M 55 DQ2.5/DQ8 GFD 10 y nega i e Whea 3d PBMC 1 CD0039 CD F 55 DQ2.5/DQ2.2 GFD 2 y nega i e Whea 3d PBMC 1,2 HC014 HC F 30 DQ2.5/DQ6 GFD 4 wk nega i e Whea 3d PBMC 1 HC036 HC F 52 DQ2.5/DQ6 GFD 4 wk nega i e Whea 3d PBMC 1 CD037 CD F 50 DQ2.5/DQ6 GFD NA nega i e Whea 3d PBMC 2 CD025 CD F 25 DQ2.5/DQ2.5 GFD 6 y nega i e Ba ley 3d PBMC 2 CD027 CD F 40 DQ2.5/DQ2.5 GFD 1 y nega i e Ba ley 3d PBMC 2 CD034 CD F 58 DQ2.5/DQ5 GFD 5 y nega i e Rye 3d PBMC 2 CD042 CD F 65 DQ2.5/DQ5 GFD 27 y nega i e Rye 3d PBMC 2 CD044 CD F 64 DQ2.5/DQ6 GFD 7 y nega i e Rye 3d PBMC 2 CD1GB CD F 67 DQ2/DQ6 AD/GFD 1 y 90/neg NA Biopsy 3 CD2GB CD F 47 DQ2/DQ6 AD/GFD 1 y 57/neg NA Biopsy 3 CD3GB CD M 56 DQ2/DQ6 AD /GFD 1 y 21/neg NA Biopsy 3 CD4GB CD F 20 DQ2/DQ2 AD /GFD 1 y 44/neg NA Biopsy 3 CD5GB CD F 39 DQ2/DQ2 AD /GFD 1 y >100/neg NA Biopsy 3 Table 1. In o ma ion on he s udy subjec s and he ea men condi ions o he expe imen s. * TG IgA nega i e: <5 U/ml. www.na u e.com/scien i ic epo s/ 3 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 smalles so ed epe oi e). Fo each epe oi e, he median o 100 di e si y es ima es om 100 downsamples was used o analysis. The so ed CD4+ IFNg+ samples did no show any di e ence in di e si y compa ed o he highly di e se p e- and pos - challenge o al pa ien PBMC epe oi es (Fig.1), sugges ing high di e si y in he so ed glu en-speci ic TCR epe oi es. CD pa ien pos -challenge o al PBMC epe oi es showed no change in di e si y compa ed o hei p e-challenge epe oi es. In e es ingly, al hough no s a is ically signi ican , all CD pa ien gu biopsy epe oi es showed inc eased di e si y du ing ac i e disease, wi h a median inc ease in di e si y o 6% ( ange: 1.9–7.9%) in biopsies be o e commencing GFD compa ed o biopsies collec ed one yea a e commencing GFD (Wilcoxon Signed Rank Tes , P = 0.06, Fig.1). Glu en exposu e induces subs an ial TRBV gene usage pe u ba ion. We ound no s a is ically signi ican di e ence in mean TRBV gene usage be ween all pos - e sus p e- challenge PBMC epe oi es, and be ween all GFD- ea ed e sus ac i e disease gu biopsy epe oi es (measu ed ei he om unique sequences o analyze o e all TRBV gene usage o om o al eads o de ec expanding/con ac ing genes) (Supplemen a y Figu eS2). To in es iga e his u he in each indi idual, we compa ed TRBV gene usage be ween he glu en exposed and unexposed epe oi es o each subjec . When conside ing unique clono ypes,TRBV5, TRBV6, TRBV7 and TRBV18 amily genes we e mo e equen ly used in pa ien o al PBMC epe oi es a e glu en chal- lenge while only TRBV6 showed inc eased usage in one pa ien ’s biopsy epe oi e du ing ac i e CD (Fig.2A and B; o all TRBV-gene esul s see Supplemen a y Fig.S3). The e we e no V gene usage di e ences be ween unique clono ypes in he so ed and he unso ed epe oi es in he wo pa ien s CD005 and CD039, om whom we had bo h so ed and unso ed epe oi e da a. Impo an ly, howe e , subs an ial TRBV gene usage pe - u ba ion was obse ed when he analysis was pe o med using o al eads (Fig.2C and D, Supplemen a y Figu eS3). Pa icula ly o he so ed epe oi es, s ong o e -usage o genes was obse ed in he TRBV18, TRBV19, TRBV29 and TRBV30 amilies, in addi ion o hose p e iously epo ed in TRB6 and TRB79–11,15,16,21 (Supplemen a y Figu eS3E). O e all, TRBV gene usage compa ison using o al eads ( ep esen ing all T-cells) de ec ed skewed TRBV gene usage in pa ien s due o an igen induced clonal expansion om he a ailable pool o unique clono ypes. Glu en exposu e is associa ed wi h inc eased epe oi e o e lap be ween CD pa ien s. To assess he e ec o glu en exposu e on amino acid clono ype sha ing be ween indi iduals, we compa ed he in e -indi idual epe oi e o e lap be ween pai s o glu en unexposed epe oi es o ha o pai s o glu en exposed epe oi es. Pa ien PBMC samples did no show signi ican di e ence in o e lap a e glu en chal- lenge, despi e a ma ginal inc ease (Fig.3A). In con as , he e was signi ican ly inc eased o e lap among pa ien gu biopsy epe oi es du ing ac i e disease compa ed o biopsies aken a e one yea GFD (Wilcoxon Signed Rank Tes , P = 0.01, Fig.3A). We also obse ed inc eased o e lap om he combined compa ison o all pai s o pa ien PBMC and GUT epe oi es (Fig.3C), indica ing ha he glu en induced public clones in blood a e a leas pa ly he same as hose esiding in he gu e en in un ela ed pa ien s. These esul s o e all sugges ha inc eased epe oi e o e lap among pa ien s is a ea u e o he immune esponse o glu en. On he o he hand, Figu e 1. TCR epe oi e di e si y in glu en exposed CD pa ien s. (a) No di e ence in epe oi e di e si y is obse ed be ween glu en speci ic CD4+ IFNg+so ed TCRB epe oi es and p e- (day0) o pos - challenge (day6) CD pa ien PBMC epe oi es. Gu epe oi es show in e es ing inc ease in di e si y (dec ease in clonali y) du ing glu en exposu e (Wilcoxon pai ed Signed Rank es , wo- ailed, p = 0.06; pai ed - es , wo- ailed, p = 0.0077. The p- alue om he - es is shown he e since he Wilcoxon signed ank es has limi a ions when applied o small sample size e en hough i is he main es employed in his s udy). Repe oi e samples we e downsampled o 27077 eads 100 imes, each ime es ima ing di e si y. Fo each sample, median o he 100 di e si y es ima es is used o plo ing and s a is ical compa ison. Ho izon al ba s indica e median. (b) The di e si y inc ease in ac i e CD gu epe oi es anged om 1.9 o 7.9% (median: 6%) o he ea ed gu epe oi e di e si y in pa ien s. www.na u e.com/scien i ic epo s/ 4 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 his inc easing o e lap was no seen when using nucleo ide clono ypes o he analysis (Fig.3B), possibly due o he glu en-induced highe epe oi e di e si y a he nucleo ide le el, de ec able numbe o which is likely coding o he same amino acid clono ypes which a e sha ed among pa ien s. Di e en ial abundance (DA) analysis iden i ies glu en-induced and -speci ic T-cells. We applied he i s DA analysis app oach (desc ibed in he Me hods sec ion) o iden i y DA clono ypes in each indi idual. Fo he ou pa ien PBMC epe oi es, we ound a median o 448 en iched ( ange: 216–465) and 355 de-en iched ( ange: 226–808) clones ollowing glu en challenge while he wo heal hy con ols had 243 en iched ( ange: 241–245) and 393 de-en iched ( ange: 157–628) clono ypes. Fo he i e gu biopsy epe oi es, a median o 631 en iched ( ange: 384–1130) and 941 de-en iched ( ange: 466–1127) clono ypes we e obse ed du ing ac i e disease. These esul s gi e a pu a i e es ima ion o he amoun o glu en-induced T-cell clono ypes, which we de ined as clono ypes showing signi ican change in clonal size as a esul o glu en exposu e ega dless o hei speci ici y (Fig.4). Fo he wo pa ien s (CD005 and CD039) ha had so ed and unso ed epe oi es, DA clones iden i ied om hei so ed epe oi es, compa ed o hei unso ed p e-challenge and pos -challenge samples, showed low o e - lap wi h hose de ec ed om he compa ison o hei unso ed pos -challenge e sus p e-challenge epe oi es (Supplemen a y TableS2 and Figu eS4). Since DA clono ypes iden i ied om so ed CD4+ IFNg+ epe oi es a e speci ic o glu en, he low o e lap pe haps indica es ha he T-cells speci ic o he 3 immunodominan glu en pep ides among in i o s imula ed so ed clones ep esen only a ac ion o all clono ypes en iched a e in i o glu en challenge, he la e including bo h glu en speci ic and non-speci ic T-cells. We also ound li le o e lap in glu en-induced DA clones ac oss pa ien s, despi e he inc eased o e lap in pa ien o al epe oi es du ing glu en exposu e (Fig.3B and Supplemen a y Figu eS4). Gi en he obse a ion ha glu en exposu e inc eased in e -indi idual epe oi e o e lap, we applied he sec- ond DA analysis app oach (see Me hods sec ion) and e alua ed he di e en ial abundance o 11834 public clo- no ypes ha we e obse ed in a leas wo di e en indi iduals (cons i u ing 4.7% o he o al 251259 unique amino acid clono ypes a ailable in all samples). This analysis was pe o med sepa a ely o he CDPBMC (n = 4), HCPBMC (n = 2), and CDGUT (n = 5) epe oi es (Table2). F om he esul ing public DA clones, only wo o he public en iched DA clono ypes (CASSLGDTQYF and CASSFSYEQYF) om pa ien PBMCs o e lapped o hose de ec ed om heal hy PBMCs, and one (CASSLTWDTEAFF) o hose de ec ed om pa ien gu epe oi es (Supplemen a y Figu eS4). Mo e han hal o he public DA clono ypes we e among he ecen ly published 10692 ep esen a i e public TCRB clono ypes ( epPC)14, indica ing possible clinical u ili y in main aining de ailed eposi o y o public TCR clono ypes and hei speci ici ies. The lis o public en iched clono ypes is a ailable in he supplemen a y da ase 1. We also in es iga ed he p esence o p e iously epo ed glu en eac i e TCRB CDR3 clono ypes7,11,15 in ou public DA clono ypes. We ound 4 p e iously epo ed clono ypes: CASSLRSTDTQYF, CASSFRSTDTQYF, CASSIRHTDTQYF and CASSLNWDTEAFF among en iched clones in pa ien PBMCs, and CASSLGYEQYF among en iched clones in pa ien gu epe oi es. We did no ind any p e iously known clono ype in he Figu e 2. TRBV gene usage is subs an ially pe u bed upon glu en exposu e in CD pa ien s. TRBV gene usage in ea ed e sus un ea ed epe oi es is shown o selec ed gene segmen s. Hea maps show di e en ially used TRBV genes in a leas one pa ien ( ishe ’s exac es , BH adjus ed p- alues < 0.05), numbe s a e p- alues, g een shows odds a io (OR) > 1 (o e - ep esen a ion) and ed OR < 1 (unde - ep esen a ion) du ing glu en exposu e. (a) and (b) show he esul o pa ien PBMC and gu epe oi es espec i ely when TRBV gene usage is compa ed a unique clono ypes. (c) and (d) o pa ien PBMC and gu epe oi es espec i ely when TRBV gene usage is compa ed using all eads. www.na u e.com/scien i ic epo s/ 5 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 de-en iched clono ypes o pa ien PBMC epe oi es and en iched/de-en iched clones o heal hy con ol PBMC epe oi es (Table2), consis en wi h he assumed glu en-speci ici y o hese clones. Nex , we examined sequence simila i y be ween public DA clono ypes o iden i y conse ed s uc u es. Hie a chical clus e ing o he 33 en iched CDPBMC clono ypes esul ed in i e clus e s. The op clus e had a signi ican o e - ep esen a ion o he MEME egula exp ession mo i ASS[LF]R[SW][TD][DT][TE][QA][YF] (Fig.5). The same analysis o he de-en iched clono ypes esul ed in six clus e s wi h he op clus e showing no signi ican mo i . We ound no o e - ep esen ed mo i in he op clus e o he en iched HCPBMC as well as CDGUT epe oi es (Supplemen a y Figu es S5–8. We u he sough o explo e he con ibu ion o public clono ypes in he immune esponse o glu en. O he glu en-induced DA clono ypes in each indi idual as de e mined by app oach one, 10% o 15% we e also p esen in ou se o public clono ypes, accoun ing o an a e age o app oxima ely 40% and 20% o he cells ( eads) in he en iched DA clones o he CDPBMC and CDGUT espec i ely (Supplemen a y TableS2 and Figu eS9). This es ima es he amoun o public clono ypes in ol ed in glu en esponse in each indi idual, and does no neces- sa ily e lec consis ency in esponse beha io o he clono ypes ac oss pa ien s since mos o hese clono ypes a e no de ec ed om he analysis using app oach wo. In e es ingly, compa ed o he p i a e DA clono ypes o each pa ien , highe a e o con e gen selec ion (mul iple nucleo ide clono ypes coding o iden ical amino acid clono ype) appea s o be a ea u e o hese ‘public subse ’ o he DA clono ypes wi h an inc easing end du ing glu en exposu e (Supplemen a y Figu eS10). Figu e 3. In e -indi idual epe oi e o e lap compa ison. (a) Inc eased in e -indi idual amino acid epe oi e o e lap is obse ed in gu biopsy samples du ing glu en exposu e (n = 10 possible pai s a each ime poin o he 5 CD GUT samples, Wilcoxon pai ed Signed Rank es , **p = 0.01), while CD pa ien PBMC epe oi es show non-signi ican inc ease in o e lap. (b) No inc eased in e -indi idual o e lap is obse ed in he nucleo ide epe oi es o bo h PBMC and gu biopsy samples du ing glu en exposu e. (c) In e -indi idual epe oi e o e lap o combined da a (4 CD PBMC and 5 CD GUT samples, n = 36).The e idence o public componen in he esponse o glu en is s onge when epe oi e ypes a e combined o he o e lap analysis, showing e en highe in e -indi idual epe oi e o e lap be ween un ela ed indi iduals and epe oi es d awn om sepa a e issues (Wilcoxon pai ed Signed Rank es , **P = 0.0002). PBMC-gu pai s show signi ican ly inc eased o e lap du ing glu en exposu e (Wilcoxon pai ed Signed Rank es , **P = 0.0064). Ho izon al ba s indica e median. Pa ien PBMC pai s a e in iangles, gu biopsy pai s a e in ci cles, mixed pbmc-gu pai s a e in ec angles. www.na u e.com/scien i ic epo s/ 6 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 Discussion O e he las ew yea s, he T-cell epe oi e o a ious au oimmune diseases has been analyzed using non-biased a ge ed deep sequencing app oaches. In he p esen s udy, we ha e u ilized T-cell epe oi e deep sequencing and in es iga ed he immune epe oi e in glu en exposed CD pa ien s. We p o iled and cha ac e ized global immune epe oi e ea u es in pe iphe al blood in pa ien s ollowing a sho 3-day glu en exposu e, and in he gu du ing ac i e CD. We also iden i ied and cha ac e ized clono ypes ha showed clonal expansion (o con ac ion) du ing glu en exposu e. In ou s udy, he immune epe oi e du ing glu en exposu e was gene ally di e se in bo h PBMC and gu . We did no de ec an al e a ion o he high TCR di e si y obse ed in CD pa ien PBMC epe oi es. Glu en-speci ic pe iphe al CD4+ IFNg+ TCR epe oi es we e also simila in di e si y o he o al unso ed pa ien PBMC epe oi es, sugges ing high poly-clonali y in he CD4+ T-cell esponse speci ic o glu en. Ha dy e al. ha e epo ed such di e si y a he TCR le el in a smalle epe oi e o CD4+ IFNg+ T-cells speci ic o he dominan HLA-DQ2.5-glia-α1a/α2 epi ope in bo h child en and adul CD pa ien s, wi h adul pa ien s no showing any mo e na owing o he epe oi e speci ic o he epi ope compa ed o ha o child en despi e mo e exposu e o glu en18. The high TCR di e si y is also consis en wi h he c oss- eac i i y o he glu en-speci ic T-cells, in which T-cells bea ing di e se TCRs a e di ec ed o speci ic immuno-dominan glu en pep ides and hei homologs5,18. In ac i e CD pa ien gu biopsies, he e was an inc ease in di e si y (by an a e age o 6%) compa ed o he ea ed epe oi es a e one yea , con i ming epo s ha ha e ound highe epe oi e di e si y in ac i e CD pa ien biopsies22. The inc eased di e si y in he gu e lec s he massi e inc ease in mucosal in il a ing T-cells Figu e 4. En iched and de-en iched clono ypes in each subjec . Di e en ial abundance analysis be ween he glu en unexposed and glu en exposed epe oi es o each subjec was pe o med. (a) Le panel, he analysis de e mined a median o 448 en iched (i.e., subs an ially highe equency) ( ange: 216–465) and 355 de- en iched (i.e., subs an ially lowe equency) ( ange: 226–808) clones ollowing glu en challenge in CD pa ien PBMC epe oi es. In heal hy con ols, 243 en iched ( ange: 241–245) and 393 de-en iched ( ange: 157–628) clono ypes we e es ima ed. Fo gu epe oi es, o which samples we e aken one yea apa , an a e age o 631 en iched ( ange: 384–1130) and 941 de-en iched ( ange: 466–1127) clono ypes we e obse ed du ing ac i e disease. Righ panel, he numbe o bo h en iched and de-en iched unique amino acid clono ypes ela i e o he o al numbe o all unique amino acid clono ypes obse ed in he wo epe oi es o each subjec is shown. (b) The ela i e clonal size ( ela i e o o al sequencing eads) o he en iched and de-en iched clono ypes, be o e and a e glu en exposu e. Fo bo h en iched and de-en iched, he combined clonal size o he DA clono ypes in each subjec is shown. Ho izon al ba s indica e median. www.na u e.com/scien i ic epo s/ 7 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 in ac i e CD. I was p obably possible o de ec his change in he gu due o he ela i ely s able na u e o he gu epe oi es, in addi ion o he gu being he main si e o in lamma ion and/o he longe glu en exposu e (Supplemen a y Figu eS11). In gene al, he immune esponse associa ed wi h glu en exposu e appea s o in ol e a di e se se o T-cell clono ypes di ec ed o glu en, and o he possible uniden i ied an igens in ol ed in CD pa hology, such as ye o be iden i ied sel -an igens and commensal lo a, he la e po en ially a ec ed by DA clono ypes in: Repe oi e ype No. o Clono ypes (RP P alue < 0.01)+ No. o DA clono ypes (SFS Accu acy > = 0.75)*So ed To al CDPBMC CDPBMC 0CDPBMC 6To al CDGUT CDGUT GFD CDGUT ac i e s udy (Qiao e al.)11 s udy (Han e al.)7 s udy (Pe e sen e al.)15 epPC (B i ano a e al.)14 En iched Public clono ypes CDPBMC 70 33 4 33 11 33 17 14 14 3 1 1 21 HCPBMC 95 95 5 51 26 36 37 26 24 0 0 0 47 CDGUT 60 30 0 15 8 13 30 15 30 0 1 0 15 De iched public clono ypes CDPBMC 75 38 3 38 38 16 18 10 17 0 0 0 22 HCPBMC 102 101 4 56 35 36 46 29 31 0 0 0 53 CDGUT 45 0 0 0 0 0 0 0 0 0 0 0 0 Table 2. Di e en ial abundance analysis esul s on public clono ypes. +RankP oduc (RP) esul s om app oach wo in he DA analysis. *Sequen ial o wa d ea u e (SFS) selec ion applied on he RP ou pu esul ing in di e en ially abundan (DA) clono ypes. Figu e 5. Public di e en ially en iched CD pa ien PBMC clono ypes. (a) DA clono ypes a e en iched in pos - challenge CD pa ien samples (colo in ensi y indica es p e alence le el), hei p e alence in heal hy con ol p e- and pos -challenge samples is also shown o compa ison. The donu plo on he igh shows he TCRBV genes used by hese DA clono ypes in CD pa ien pos -challenge samples, om he leas o he mos used. Clono ypes we e hie a chically clus e ed using a modi ied Le ensh ein dis ance ha ook physicochemical p ope ies in o accoun ; he clus e indica ed wi h he g een ba has he highes a e age old change inc ease du ing glu en exposu e. (b) MEME mo i sea ch o he op clus e : signi ican ly o e - ep esen ed mo i shown in sequence logo and egula exp ession, ASS[LF]R[SW][TD][DT][TE][QA][YF] (E- alue = 2.4E-10); Top clus e con ains p e iously known HLA-DQ2 es ic ed, alpha gliadin epi ope speci ic CD4+ clono ypes wi h he ASSxRxTDTQY mo i (highligh ed wi h ligh g een). www.na u e.com/scien i ic epo s/ 8 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 glu en in luence on mic obio a. Chemokines sec e ed by he in lamma o y cells in ac i e CD could also a ac “by-s ande ” T-cells o many o he speci ici ies un ela ed o glu en o he si e o in lamma ion. Glu en exposu e led o subs an ial pe u ba ion o he V-gene usage in bo h he PBMC and gu epe oi es in each pa ien , mainly due o p oli e a i e changes o he a ailable unique clono ypes. We obse ed signi ican o e -usage o genes in a ious TRBV gene amilies in one o mo e pa ien s esul ing om clonal expansion. We also obse ed signi ican al e a ions in p e iously epo ed TRBV genes ha a e p e e en ially selec ed o speci ic immuno-dominan gliadin epi opes7,10,11,16, such as TRBV07-02 (o e -used in all pa ien PBMC epe oi es and wo gu epe oi es) and TRBV04-02 (o e -used in ou gu epe oi es and wo PBMC epe oi es). Simila anal- yses o he unique clono ypes sugges ed ha pa ien epe oi es main ained a V-gene usage p o ile ha emained mo e o less s able be o e and a e glu en exposu e, pa icula ly in he gu , likely an indica ion o clonal pe sis- ence in he a ailable immune epe oi e. In e -indi idual epe oi e o e lap was signi ican ly inc eased among pa ien s du ing glu en exposu e, pa - icula ly in he gu , sugges ing ha he immune esponse o glu en is cha ac e ized by a public componen . The e idence o he inc eased epe oi e o e lap was e en mo e p onounced when we combined pa ien gu and PBMC epe oi es and compa ed hem o hei ea ed coun e pa s. While semi-public esponse o speci ic immuno-dominan glu en epi opes ha e been p e iously epo ed7,11, ou da a shows ha inc eased amino acid epe oi e sha ing among CD pa ien s can be de ec ed om he o al epe oi e (ac oss issues, and a e bo h a sho glu en challenge o long- e m glu en exposu e) and is likely a majo ea u e o he immune esponse in CD. We iden i ied hund eds o di e en ially abundan clono ypes wi h signi ican ly al e ed p e alence du ing glu en exposu e om o al epe oi es o each indi idual, bo h in pa ien s and heal hy con ols. DeWi e al.23 showed ha DA clono ypes iden i ied in such manne , i.e. by compa ing mul iple o al epe oi es o an indi- idual, cap u ed app oxima ely 60% o known an igen-speci ic CD8+ cells, whose esponse magni ude is much highe compa ed o CD4+ T-cells, in a model o acu e i al in ec ion. Since ou epe oi e da a con ain esponse signals coming om bo h CD8 and CD4 T-cell popula ions, ou es ima es a e only a ough app oxima ion, possi- bly an uppe bound, o he equency o glu en-induced T-cells. F om hese esul s, we obse ed ha he immune eac ion o glu en in ol ed a la ge p i a e componen ; public clono ypes cons i u ed be ween 10–15% o he di e en ially abundan clono ypes and had an inc eased a e o con e gen selec ion du ing glu en exposu e. The ole and signi icance o he glu en-induced public clono ypes compa ed o hose ha a e p i a e in CD pa hogen- esis is unclea . Public clono ypes a e gene ally less complex and mo e abundan 24; TCR sequences ha a e mo e likely o be p oduced in he hymus also pass hymic selec ion mo e o en25, indica ing ha posi i e selec ion o some abundan public clono ypes o he pe iphe y may ha e a c ucial immunological impo ance. Public clo- no ypes in mice ha e been shown o equen ly ha bo au oimmune speci ici ies26,27. They ha e been shown o ha e capaci y o c oss- eac i i y, as well as speci ici y o sel and o eign an igens28. Sel - eac i e T-cells ha e also been associa ed wi h mo e e icien esponse o o eign an igens in e ms o esponse a es, clonal expansion size and sensi i i y o in lamma o y signals29. O e all, while public clono ypes a e likely c ucial o e icien immune esponse, dys egula ion o hei unc ions may lead o abe an sel -immuni y. By examining changes in he p e alence o a selec ed se o public clono ypes, we iden i ied en iched public clono ypes ac oss CD pa ien s in PBMC and gu epe oi es. Only one clono ype was sha ed be ween he public en iched DA clono ypes o PBMC and gu , possibly due o he di e ences in he du a ion o glu en exposu e o in he abundance o a ious T-cell subse s in each compa men ; o no e, CD pa ien gu epe oi es a e domina ed by cy o oxic glu en-independen in es inal in aepi helial lymphocy es (IELs) du ing ac i e disease30. In pa ien PBMCs, he amino acid pa e n ASS[LF]R[SW][TD][DT][TE][QA][YF] wi h conse ed usage o he a ginine a he i h posi ion, gene alizing p e iously epo ed mo i ASSxRxTDTQY, was o e - ep esen ed in he mos p oli e a i e clus e o en iched clono ypes, which also con ained clono ypes ha we e exac ly iden ical o he p e iously epo ed glu en speci ic T-cell clones10. The a ginine esidue in glu en eac i e clono ypes has been shown o be e y c ucial in he immune esponse9,11,21. The dominance o such a ginine bea ing clono ypes is educed among he public en iched clono ypes o he gu , possibly sugges ing ha he public clono ypes speci ic o he highly immunogenic epi opes play an impo an ole mainly in he ea ly esponse o glu en exposu e. This obse a ion o inc eased a ginine bea ing clono ypes ollowing glu en exposu e is also de ec able in o al pa ien PBMC epe oi es bu no in o al gu epe oi es, pa icula ly du ing ac i e disease (Supplemen a y Figu eS12). Wi h p olonged exposu e o glu en, a leas in he gu , he p e alence o a ginine bea ing glu en eac i e clono- ypes may wane, wi h pa ien gu epe oi es becoming inc easingly mo e di e se and domina ed by in il a ing T-cells wi h clono ypes di ec ed o o he an igens. I is possible ha a ginine bea ing glu en eac i e public clo- no ypes may commi o long e m-memo y mo e o en, as hey likely ha e s ong unc ional a idi y o glu en epi opes e idenced by hei s ong en ichmen 31,32. Once glu en ole ance is b oken, hese T-cells could be among hose ha igge he ea ly adap i e esponse o glu en upon e-exposu e. This s udy used small numbe o samples; he PBMC and gu samples we e also no om he same pa ien s. La ge numbe o samples would ha e allowed a mo e ep esen a i e compa ison o epe oi es, while ma ched samples would ha e p o ided he possibili y o ack glu en eac i e clono ypes in bo h compa men s. In ou expe imen al se up, we mainly wan ed o explo e he impac o glu en exposu e o he o al epe oi e in- i o, which also e lec s possible eac ion d i en by T-cell subse s o he han T-helpe cells (such as egula o y T-cells, and CD8+ T-cells o which he an igen speci ici y is unknown), al hough we did no u he s udy hei pheno- ypes. Thus, he glu en-induced clono ypes we ound we e no necessa ily speci ic o glu en, bu could ha e been due o seconda y e ec s o i s inges ion. The e o e, i would be e y in o ma i e o s udy in u u e he comple e p o ile o he in e es ing public clono ypes iden i ied, such as hei ull TCRαβ pai ing and gene exp ession pa - e ns using o ins ance single cell analysis me hods. In spi e o hese limi a ions, we we e able o iden i y majo immunological signa u es associa ed wi h glu en exposu e, iden i y glu en-induced clono ypes om o al pa ien epe oi es, and es ima e con ibu ions o p i a e and public clono ypes o he glu en-induced immune esponse in pa ien s. www.na u e.com/scien i ic epo s/ 9 SCIEnTIFIC RepoRTS | (2017) 7:17977 | DOI:10.1038/s41598-017-18137-9 In summa y, glu en exposu e appea s o elici di e se eac ion a he TCR le el pa icula ly in he gu , and in PBMC as e idenced by he high TCR di e si y obse ed in he glu en speci ic PBMC CD4+ IFNg+ epe oi es. Inc eased epe oi e o e lap was seen in un ela ed pa ien s, and he immune esponse in gene al was cha ac e - ized by a la ge p i a e and an impo an public componen . Deepe le el o epe oi e sequencing wi h la ge num- be o samples could enable iden i ica ion and acking o disease associa ed public clono ypes o immunological signa u es, wi h po en ial applica ions such as ea ly diagnosis, disease and ea men moni o ing, and pa ien s a i ica ion based on ac o s such as pa ien HLA- ypes and clinical symp oms. Me hods Subjec s and s udy design. The s udy pa icipan s who p o ided blood o in es inal biopsy ma e ial a e p esen ed in Table1. All subjec s we e posi i e o he high- isk CD-associa ed suscep ibili y alleles HLA- DQA1*05 and HLA-DQB1*02 encoding HLA-DQ2.5. In CD subjec s, he diagnos ic c i e ia we e duodenal illous a ophy accompanied by ele a ed se um an i- issue ansglu aminase IgA. Fo he i s wo expe imen s, glu en challenges, blood collec ions, isola ion o PBMC and o e nigh IFN-γ ELISpo we e pe o med acco ding o Ande son e al.4 and Tye-din e al.5. CD subjec s on long- e m GFD consumed one o whea b ead (4 slices, equi alen o app oxima ely 10 g/d glu en), ba ley lakes (120 g/d, equi alen o app oxima ely 8 g/d ho dein), o ye b ead (equi alen o app oxima ely 8 g/d secalin). In addi ion, wo heal hy subjec s who had excluded ood wi h glu en o 4 weeks comple ed he same 3-day whea b ead challenge. PBMCs we e isola ed om blood collec ed immedia ely be o e commencing glu en challenge (day 0) and 3 days a e comple ing glu en challenge (day 6). PBMCs we e ei he c yop ese ed o used o measu e esponses o glu en-de i ed pep ides s imula o y o CD4+ T cells in o e nigh IFN-γ ELISpo assays. In he i s expe imen , DNA was ex ac ed om ozen PBMC s ocks collec ed be o e and a e whea glu en challenge om 4 CD dono s wi h posi i e IFN-γ ELISpo esponses o glu en pep ides, and 2 heal hy dono s, o bulk TCR analysis. In he second expe imen , PBMC collec ed om 8 CD dono s 6 days a e commencing glu en challenge wi h whea (n = 3), ba ley (n = 2), o ye (n = 3) we e hawed, and incuba ed wi h a mix u e o h ee immunodominan glu en pep ides ( o an igen p esen a ion by mainly B-cells) o iden i y CD4+ T cells sec e ing IFN-γ (see below o de ails). TCR analysis was pe o med on CD4+ IFNg+ cells so ed by low cy ome y. In he hi d expe imen , TCR analysis was pe o med on DNA isola ed om esh 2nd pa duodenal biopsies om 5 addi ional CD pa ien s a he ime o diagnosis du ing ac i e disease and again a e one yea GFD ea men . E hics. The s udy design and ec ui ing o pa ien s we e app o ed by he E hics Commi ees o he Pi kanmaa Hospi al Dis ic , Finland. All subjec s ga e w i en in o med consen . Rele an guidelines and egula ions we e ollowed when pe o ming he expe imen s. Isola ion o glu en-speci ic IFNg-p oducing T cells. The c yo-p ese ed PBMC samples we e hawed by esuspension in o comple e RPMI, and e-s imula ed, a e a es ing pe iod o 60 min a 37°C, wi h a pool o 3 immunodominan glu en pep ides (LQPFPQPELPYPQPQ “DQ2.5-glia-α1a & α2”, QPFPQPEQPFPWQP “DQ2.5-glia-ω1 & ω2” and PEQPIPEQPQPYPQQ “DQ2.5-ho -3”)5,33, each in a inal concen a ion o 100 mic o- g am/ml. A e 16 h, cells we e esuspended in PBS, and incuba ed wi h IFNg ca ch eagen , ollowed by s aining wi h PE an i-IFNg de ec ion an ibody (IFNg sec e ion assay ki ; Mil enyi Bio ech) and FITC an i-CD4 an ibody (clone OKT4; Biolegend), acco ding o manu ac u e ’s ins uc ions. IFNg sec e ing CD4+ T cells we e sepa a ed by FACS so ing, using a BD Facs A ia II ins umen . Cells we e also s ained o gu -homing ma ke CCR9 in 6 o he 8 samples, which showed ha he selec ed CD4+ IFNg+ popula ion was en iched wi h cells exp essing CCR9 compa ed o CD4+ IFNg− cell ac ion (84% s 40%, Supplemen a y Figu eS13). DNA was ex ac ed om he so ed cell pelle s using a DNA/RNA ex ac ion ki (Qiagen). DNA ex ac ion and Sequencing. DNA was ex ac ed (AllP ep DNA/RNA Mini Ki , Qiagen) om 2-10 million pelle ed and lysed p e- and pos -challenge cell samples o 4 whea challenged and whea epi ope espond- ing pa ien s, as well as om 2 non- esponding heal hy con ols. Fo gu samples, o al DNA was ex ac ed om snap ozen biopsy specimens, ob ained a he ime o diagnosis o 5 DQ2 + CD pa ien s wi h un ea ed, ac i e disease and con i med mucosal lesions, as well as biopsies o he same 5 pa ien s a e 1 yea o glu en- ee die (GFD), using QIAamp Mini Ki (Qiagen, Valencia, CA) as desc ibed in Wacklin e al.34. DNA samples (1ug pe sample) we e sen o Adap i e Bio echnologies and we e deep sequenced o TCRB CDR3 egion (Sea le, USA, www.adap i ebio ech.com; www.immunoseq.com) wi h hei ImmunoSeq assay which employs op imized mul iplex PCR o a ge and ampli y he TCR CDR3 egion and Illumina o sequencing35. Fo each unique TCRB CDR3 sequence in a sample, he nucleo ide and p edic ed amino acid sequence, he e-a anged V D and J genes, and he numbe o sequencing eads we e de e mined. Sequence da a and s a is ical analysis. P oduc i e sequences wi hou ameshi mu a ions o s op codons, cons i u ing ~ 80% o ou epe oi es, we e used in all subsequen analyses. We employed downsampling, p e-p ocessing, and da a no maliza ion p ocedu es app op ia e o each analysis as desc ibed in he espec i e sec ions. All analyses we e done in R (h ps://www. -p ojec .o g/) and g aphs we e made using R and P ism 6.0 (G aphPad So wa e, La Jolla Cali o nia USA, www.g aphpad.com). Repe oi e di e si y was es ima ed using he Shannon en opy index no malized by o al numbe o unique amino acid clono ypes, −∑⁎⁎ CN(Clog2())/(log2( )) iii whe e Ci is he ela i e equency o unique clono ype i in he epe oi e, and N is he o al numbe o unique amino acid clono ypes. A no malized en opy alue o 1 e lec s high di e si y while 0 e lec s low di e si y. The di e si y o e e y sample was es ima ed as he median o 100 di e si y alues calcula ed om 100 downsampling o each sample o 27077 eads ( he size o he second smalles so ed epe oi e). Di e si y compa ison be ween g oups was mainly pe o med using he