An efficient urine peptidomics workflow identifies chemically defined dietary gluten peptides from patients with celiac disease

ARTICLE An e ficien u ine pep idomics wo kflow iden ifies chemically defined die a y glu en pep ides om pa ien s wi h celiac disease B ad A. Palanski1,13,14, Nielson Weng1,2,3,14, Lichao Zhang 4, And ew J. Hilme 1, Lalla A. Fall5, Ka ya Swamina han 6, Bana Jab i7,8,9, Ca olina Sousa 10, Nielsen Q. Fe nandez-Becke 11, Chai an Khosla1,5,12,15✉& Joshua E. Elias 4,15✉ Celiac disease (CeD) is an au oimmune diso de induced by consuming glu en p o eins om whea , ba ley, and ye. Glu ens esis gas oin es inal p o eolysis, esul ing in pep ides ha elici inflamma ion in pa ien s wi h CeD. Despi e well-es ablished connec ions be ween glu ens and CeD, chemically defined, bioa ailable pep ides p oduced om die a y p o eins ha e ne e been iden ified om humans in an unbiased manne . This is la gely a ibu able o echnical challenges, impeding ou knowledge o po en ially di e se pep ide species ha encoun e he immune sys em. He e, we de elop a liquid ch oma og aphic-mass spec o- me ic wo kflow o un a ge ed sequence analysis o he u ina y pep idome. We de ec o e 600 dis inc die a y pep ides, o which ~35% ha e a CeD- ele an T cell epi ope and ~5% a e known o s imula e inna e immune esponses. Rema kably, glu en pep ides om pa ien s wi h CeD quali a i ely and quan i a i ely di e om con ols. Ou esul s p o ide a new ounda ion o unde s anding glu en immunogenici y, imp o ing CeD managemen , and cha ac e izing he die a y and u ina y pep idomes. h ps://doi.o g/10.1038/s41467-022-28353-1 OPEN 1Depa men o Chemis y, S an o d Uni e si y, S an o d, CA, USA. 2School o Medicine, S an o d Uni e si y, S an o d, CA, USA. 3Medical Scien is T aining P og am, S an o d Uni e si y, S an o d, CA, USA. 4Chan Zucke be g Biohub, San F ancisco, CA, USA. 5S an o d ChEM-H, S an o d Uni e si y, S an o d, CA, USA. 6Di ision o Blood and Bone Ma ow T ansplan a ion, S an o d Uni e si y, S an o d, CA, USA. 7Depa men o Medicine, Uni e si y o Chicago, Chicago, IL, USA. 8Commi ee on Immunology, Uni e si y o Chicago, Chicago, IL, USA. 9Depa men o Pa hology and Pedia ics, Uni e si y o Chicago, Chicago, IL, USA. 10 Facul ad de Fa macia, Depa amen o de Mic obiología y Pa asi ología, Uni e sidad de Se illa, Se illa, Spain. 11 Di ision o Gas oen e ology and Hepa ology, Depa men o Medicine, S an o d Uni e si y, S an o d, CA, USA. 12 Depa men o Chemical Enginee ing, S an o d Uni e si y, S an o d, CA, USA. 13 P esen add ess: Depa men o Medicine, B igham and Women’s Hospi al, and Depa men o Biological Chemis y and Molecula Pha macology, Ha a d Medical School, Bos on, MA, USA. 14 These au ho s con ibu ed equally: B ad A. Palanski, Nielson Weng. 15 These au ho s join ly supe ised his wo k: Chai an Khosla, Joshua E. Elias. ✉email: [email p o ec ed];[email protected] g NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions 1 1234567890():,; In humans, he p e ailing unde s anding o physiological diges ion is ha p o eins a e b oken down in o single amino acids o di- o i-pep ides be o e abso p ion as nu ien s1,2. Glu en p o eins ound in whea , ba ley, and ye a e excep ions o his ene , which esul s in no able heal h consequences. Unusual biochemical p ope ies o hese p o eins, such as a high abun- dance o glu amine (Gln, Q) and p oline (P o, P) esidues, ende hem esis an o deg ada ion by gas oin es inal p o eases3. Consequen ly, ela i ely long glu en pep ides wi h in ac immu- no oxic epi opes accumula e in he lumen o he small in es ine4 and c oss he epi helial ba ie , al hough he mechanism(s) o anspo emain con o e sial5. In app oxima ely 1 in 100 indi iduals, an abe an immune esponse o hese pep ides causes celiac disease (CeD), an au oimmune diso de ha causes small in es inal mucosal inju y cha ac e ized by illous a ophy. Common symp oms include abdominal pain, bloa ing, nausea, omi ing, and/o dia hea6. Ex ain es inal CeD mani es a ions also occu , including blis e ing skin ashes and a axia7. A p e- sen , he only e ec i e CeD ea men is a s ic , li elong adhe - ence o a glu en- ee die (GFD)8. Assays o p obe glu en immunogenici y in CeD ha e ypically elied on ex ac ed o ecombinan glu ens ha a e diges ed in i o wi h gas oin es inal p o eases. The esul ing pep ides can be used in bioassays o cha ac e ize mechanisms o immuno oxici y3. In mos cases, hese pep ides elici adap i e immune esponses media ed by HLA-DQ2 o -DQ8 an igen p esen a ion o Th1 cells in he small in es inal mucosa o pa ien s wi h CeD4,9–14. Glu en pep ides wi h al e na i e (i.e., non-T cell-dependen ) modes o ac ion ha e also been epo ed15–21. S uc u e- unc ion analyses using syn he ic glu en pep ides ha e e ealed exquisi e sequence specifici y o bo h he HLA and T-cell ecep o ; single amino acid al e a ions can esul in d ama ically al e ed a fini ies, he eby al e ing he s eng h o he immune esponse22–24. Al hough i is well es ablished ha chemical a ia ion in glu en pep ides can influence CeD immune esponses, measu ing hei na u ally p ocessed o ms and connec ing hem wi h CeD pa ien s’heal h s a us is an unde explo ed esea ch a ea. O e 20 yea s ago, ch oma og aphic analysis coupled o UV de ec ion implied he exis ence o glu en-de i ed pep ides in he u ine o pa ien s wi h CeD25. This was confi med mo e ecen ly by an ibody-based me hods26–29. Indeed, mos cu en glu en de ec ion me hods ely on monoclonal an ibodies, which ecog- nize amino acid mo i s p esen in a subse o glu en p o eins30. No wi hs anding he aluable knowledge ha has been gained om analyzing biospecimens wi h hese immuno eagen s, hey a e nei he capable o e ealing he exac glu en pep ide sequences ha a e p esen , no a e hey comp ehensi e in ha some CeD- ele an pep ides may lack he epi opes hese an i- bodies ecognize. To add ess moun ing e idence ha many disease- ele an glu en pep ides emain o be disco e ed31, new me hods a e needed o eco e and p ecisely cha ac e ize in i o glu en diges ion p oduc s in an un a ge ed ashion. In ac , no a single chemically defined pep ide om whea (o , o ou knowledge, om any die a y p o ein) has e e been iden ified om he human ci cula o y o exc e o y sys ems. Consequen ly, e y li le is known abou he physiological abso p ion, dis ibu ion, me a- bolism, and exc e ion (ADME) o glu en. Specifically, we lack insigh in o he chemical s uc u es o pep ides o med om in i o diges ion ha may s imula e he immune sys em. As a s ep owa d filling his knowledge gap, we sough o analyze human u ine by liquid ch oma og aphy coupled o an- dem mass spec ome y (LC–MS/MS). Cu en ly, LC–MS/MS is he mos widely used echnique o pep ide sequencing in com- plex biological samples32. Howe e , es ablished LC–MS/MS me hods su e om echnical limi a ions when applied o u in- a y pep idome analysis. High concen a ions o u ina y sal s and me aboli es, which a e no e ficien ly emo ed by s anda d e e sed-phase o liquid-liquid ex ac ion p ocedu es, can o e - whelm ch oma og aphy sys ems and in e e e wi h pep ide de ec ion33,34. He e, we de elop a sample p epa a ion and LC–MS/MS me hod ha u ilizes mixed ca ion exchange solid- phase ex ac ion o exclude hese in e e ing molecules in a single s ep. This wo kflow o e comes p oblems we ini ially encoun e ed wi h adap ing es ablished me hods o u ina y pep idomics, such as he need o ime-consuming s ong ca ion exchange pu - ifica ion and/o limi ed dep h o pep ide sampling. Wi h i , we can now e ficien ly iden i y die a y glu en pep ides and epo he p ecise sequences o such pep ides in he u ine o human olun ee s. We also unde ake an explo a o y clinical s udy, which e ealed whea -de i ed pep ides ha a e subs an ially di e en in hei chemical and biological p ope ies and a e di - e en ially ound in pa ien s wi h CeD e sus heal hy con ols. These pep ides a e a ac i e candida es o imp o ing CeD diagnosis and o moni o ing pa ien compliance o GFDs. They also se he s age o elucida ing mechanisms unde lying he anomalous ADME cha ac e is ics o glu en and o he die a y p o eins. Mo e gene ally, he success ul applica ion o ou u ina y pep idomic wo kflow o CeD sugges s i should be b oadly applicable o he di ec measu emen o bo h endogenous and exogenous pep ides p esen in u ine. Resul s LC–MS/MS me hod de elopmen . P e ious ch oma og aphic and an ibody-based de ec ion me hods sugges ed he p esence o glu en-de i ed pep ides in human u ine bu we e incapable o di ec ly elucida ing hei sequences25–29. To de e mine i an unbiased LC–MS/MS app oach could be used o his pu pose, we ini ially analyzed u ine samples om olun ee s who had con- sumed a meal ich in die a y glu en by ollowing an ex ac ion and analysis p o ocol ypically used in LC–MS/MS-based p o- eomics s udies (Supplemen a y Me hods). These explo a o y e o s yielded jus one glu en-de i ed pep ide iden ifica ion. Fu he mo e, e y ew endogenous human pep ides we e de ec ed (Supplemen a y Fig. 1). In acco dance wi h challenges p e iously epo ed wi h u ina y pep idome expe imen s33,34, we no ed se e e deg ada ion o he LC column and MS elec osp ay sou ce o e consecu i e analyses, p e en ing us om analyzing mul iple samples wi hou unaccep able in e up ions in ins umen ope a ion. We, he e o e, endea o ed o sys ema ically op imize each s ep in ou pep idome ex ac ion and LC–MS/MS analy ical p o ocols in o de o eliably de ec pep ides om u ine. S a ing wi h u ine om an indi idual challenged wi h die a y glu en, we fi s confi med i had measu able glu en le els by ELISA. The comme cially a ailable R5 monoclonal an ibody we used a ge s he pen ame ic mo i QXP(W/F)P ound in many glu en p o eins35 and is he e o e expec ed o eac wi h glu en pep ides p oduced by pa ien s wi h CeD and heal hy con ols. Using his e e ence specimen, we es ed a a ie y o me hods o en ich and pu i y u ina y pep ides. In addi ion, we op imized he LC g adien and MS acquisi ion me hod (Supplemen a y Me hods). These e o s esul ed in 29- and 13- old inc eases in he numbe o iden ifiable whea - and human-de i ed pep ides, espec i ely, while elimina ing longs anding p oblems wi h ins umen con amina ion by backg ound me aboli es (Supplemen a y Me hods, Supplemen a y Figs. 2 and 3, see also Me hod Specifici y sec ion below). Impo an ly, by combining solid-phaseex ac iononamixedca ionexchangecolumn wi h online e e sed-phase LC and high- esolu ion MS/MS ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 2NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions analysis, his wo kflow sho ened ou sample p epa a ion ime om 2–3days o6h(Fig.1). This allowed su ficien h oughpu o us o unde ake a compa a i e analysis o clinical u ine samples. Pilo iden ifica ion o chemically defined u ina y whea pep- ides. To e alua e ou op imized u ine pep idome ex ac ion and LC–MS/MS me hod, ou heal hy pa icipan s we e ec ui ed in o a pilo s udy ou lined as ollows (Fig. 2a): On he fi s day, pa - icipan s ini ia ed a GFD. The GFD was main ained on he sec- ond day, and a pooled u ine sample was collec ed o e 8 h. On he hi d day, pa icipan s unde wen a glu en challenge com- p ised o wo whea bagels (~18 g glu en), and subsequen ly collec ed a second pooled u ine specimen o e he nex 8 h. LC–MS/MS analysis o hese u ine samples yielded an a e age o 24 unique pep ide sequences pe pa icipan mapping o he whea p o eome, all o which we e iden ified only in he pos - glu en challenge samples (Supplemen a y Da ase 3). The numbe o unique whea pep ides a ied widely be ween indi idual olun ee s (Fig. 2b). In con as , he numbe o de ec ed human pep ides was essen ially unchanged in he GFD e sus pos - challenge samples (Supplemen a y Fig. 4). In an icipa ion o ec ui ing la ge coho s o human pa icipan s, we sough o confi m ou ini ial findings while also simpli ying he die a y glu en challenge p o ocol. The ou o iginal pa icipan s, along wi h ou addi ional heal hy pa ici- pan s, we e asked o as o e nigh in place o he 48 h GFD employed in he pilo s udy (Fig. 2c). Then, a single u ine sample was collec ed be o e consump ion o wo whea bagels o b eak as and pooled u ine collec ion o e he nex 8 h. Whea pep ides we e eadily de ec ed in all pos -challenge samples (Fig. 2d and Supplemen a y Da ase 4), while he numbe o human pep ides was simila in bo h sample g oups (Supplemen- a y Fig. 5). These esul s suppo ou ini ial finding ha a die a y glu en challenge consis en ly leads o u ina y glu en pep ides ha a e measu able by LC–MS/MS. They also indica e ha le els o hese pep ides all o unde ec able le els a e o e nigh as ing, jus i ying he use o he abb e ia ed p o ocol o subsequen s udies. Conside ing all sequences o he whea -de i ed pep ides iden ified om heal hy pa icipan s (Supplemen a y Da ase s 3 and 4), ou no ewo hy cha ac e is ics we e appa en (Table 1). Fi s , al hough 123 unique pep ide sequences we e iden ified, only one pep ide was ound in common in all 8 pa icipan s: GQQQPFPPQQPYPQPQPFPS. I s C- e minally unca ed de i- a i e, GQQQPFPPQQPYPQPQPFP, was iden ified in 7/8 pa icipan s; o he o e lapping a ian s o hese pep ides we e also de ec ed in mo e han one u ine sample. While hese sequences om α-gliadin p o eins a e no known o be ecognized by pa ien -de i ed T cells, mul iple s udies ha e demons a ed ha hey can s imula e an inna e immune esponse15–21. S ikingly, hese p e alen u ina y pep ides do no ha e a mo i ha is ecognized by monoclonal an ibodies commonly used o glu en de ec ion35,36. Thus, hey would e ade de ec ion by adi ional immunoassays. Second, we iden ified many pep ides ha bo ing known CeD- ele an T-cell epi opes37. The DQ2.5-glia-γ4c/DQ8-glia-γ1a epi- ope was mos common, wi h 32 dis inc pep ides ha bo ing he co esponding QQPQQPFPQ sequence. The DQ2.5-glia-γ5epi ope (QQPFPQQPQ) also appea ed equen ly, as i was iden ified in 14 unique pep ides. O he known T-cell epi opes, including DQ2.5- glia-γ1/DQ8.5-glia-γ1/DQ8-glia-γ2 (PQQSFPQQQ), DQ2.5-glia- γ3/DQ8-glia-γ1b (QQPQQPYPQ), and DQ2.5-glu -L1/DQ2.2- glu -L1 (PFSQQQQPV) we e iden ified less equen ly (Supple- men a y Table 1). Some T-cell s imula o y pep ides, such as FLQPQQPFPQQPQQPYPQQPQQPFPQ and SQQPQQPFPQPQ QQFPQPQQPQ, ha bo mo e han one epi ope and we e p e iously iden ified as being bo h p o eoly ically esis an and no malize loading 30 µmol c ea inine ≈1-10 mL u ine p o eins, pep ides, me aboli es, sal s block Cys educ ion (DTT) alkyla ion (IAM) deple e p o eins 10 kDa MWCO 2% o mic acid disca d p o eins mixed ca ion exchange solid phase ex ac ion disca d me aboli es/sal s d y, econs i u e pep ides (25 µL) LC-MS/MS on O bi ap Fusion Lumos 70 min g adien (C18-RP) DDA OT MS/MS housands o pep ides iden i ied in a single un u ine sample 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Time (min) 0 10 20 30 40 50 60 70 80 90 100 Rela i e Abundance 64.28 27.09 64.33 62.29 60.97 60.91 27.18 60.88 56.80 64.43 60.37 69.60 27.82 69.64 32.87 50.96 64.50 64.62 51.01 46.63 33.06 68.20 51.04 69.73 39.09 27.92 43.62 33.69 20.72 24.45 28.57 55.30 71.40 75.05 12.23 20.12 10.41 75.29 83.61 6.23 6.83 2.45 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 m/z 0 10 20 30 40 50 60 70 80 90 100 Rela i e Abundance 913.4781 z=1 743.3941 z=1 755.4108 z=2 1180.6373 z=1 1381.7540 z=1 1016.5050 z=1 686.3726 z=1 1267.6703 z=1 808.4351 z=1 1096.4949 z=2 971.5003 z=1 1459.7382 z=1 312.1554 z=? 512.2729 z=1 1344.7213 z=1 425.2390 z=? 869.4445 z=? 596.3391 z=? Fig. 1 An e ficien chemical ex ac ion and LC–MS/MS wo kflow o u ina y pep idomic analysis. U ine sample olumes a e fi s no malized based on c ea inine measu emen . To p e en he o ma ion o in a- o in e molecula disulfide bonds ha would con ound downs eam da a analysis, cys eine (Cys) esidues a e educed wi h di hio h ei ol (DTT) and alkyla ed wi h iodoace amide (IAM). P o eins a e deple ed using a cen i ugal fil a ion de ice wi h a molecula weigh cu o (MWCO) o 10 kDa unde acidic dena u ing condi ions. The fil a e is p ocessed using a mixed ca ion exchange solid-phase ex ac ion column o emo e me aboli es and sal s ha in e e e wi h LC–MS/MS analysis. A e ex ensi e washing, pep ides a e elu ed om he column wi h me hanol con aining 5% ammonium hyd oxide. Elu ed pep ides a e d ied in a cen i ugal acuum concen a o and econs i u ed in wa e . Pep ides a e sepa a ed by nano-liquid ch oma og aphy (LC) on a e e sed-phase column (RP-C18) and analyzed by andem mass spec ome y (MS/MS) on an O bi ap Fusion Lumos mass spec ome e ope a ed in he da a-dependen acquisi ion (DDA) mode, wi h p ecu so and agmen ions analyzed in high esolu ion in he O bi ap (OT). The esul ing LC–MS/MS da a a e sea ched agains he desi ed da abases (e.g., he human and whea p o eomes). The expe imen al wo kflow akes ~6 h o comple e and esul s in he iden ifica ion o housands o pep ides. NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 ARTICLE NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions 3 highly inflamma o y o CeD pa ien s h ough in silico and in i o analysis38. Thi d, some pep ides such as Py QTFPHQPQQQVPQPQQ PQQP had unde gone pos - ansla ional modifica ion ia cycliza- ion o an N- e minal Gln in o a py oglu amic acid esidue. The possibili y ha his modifica ion occu s du ing sample p ocessing canno be excluded. Howe e , N- e minal py oglu amina ion is known o p o ec pep ides om p o eoly ic deg ada ion39–41 and hus may con ibu e o hei s abili y in ci cula ion. Las , we ound glu en pep ides om human u ine ha ha e no been desc ibed by any p io s udy as ei he being esis an o gas oin es inal diges ion o as ha ing any pa hophysiological cha ac e is ics. One such pep ide, SCHVMQQQCC, is de i ed om γ-gliadin and om he low-molecula -weigh subuni s o glu enins B and C. I was iden ified in 6/8 pa icipan s, while i s C- e minally ex ended o m was de ec ed in 5/8 pa icipan s. Longe and sho e e sions we e also de ec ed in mul iple samples. This esul illus a es he powe ou un a ge ed app oach has o iden i ying whea -de i ed pep ides ha ha e eluded p io desc ip ion despi e hei seemingly common ep esen a ion. Me hod specifici y. To ou knowledge, un a ge ed de ec ion o die -de i ed pep ides in humans had no been epo ed p io o ou p oo -o -concep s udies abo e. We, he e o e, sough o confi m ha ou LC–MS/MS me hod eliably de ec s hese pep- ides and accu a ely iden ifies hei sequences in he complex backg ound o he human u ina y pep idome. To do so, we ook ad an age o he ac ha ye and ba ley also con ain p o eoly- ically esis an glu en p o eins42. Fo example, whea gliadins (one o wo majo p o ein amilies ha comp ise glu en) a e homologous o secalin p o eins in ye and ho deins in ba ley (Fig. 3a). We hus an icipa ed ha u ine analysis should dis in- guish he pa icula g ain consumed by olun ee s who a e meals ich in ei he whea , ba ley, o ye. To es his hypo hesis, wo heal hy pa icipan s as ed o e nigh , and hen, on di e en days, main ained a g ain- ee die , o a e b eak as s p epa ed wi h app oxima ely 1.5 cups o whea , ba ley, o ye flou . U ine was hen collec ed o 8 h, analyzed by ou op imized LC–MS/MS p o ocol, and he da a we e sea ched agains a cus omized da a- base ha included he human, ye, whea , and ba ley p o eomes (see “Me hods”sec ion). When pa icipan s main ained a g ain- ee die , only a single g ain pep ide (likely a alse disco e y) was iden ified. In con as , subsequen o whea , ye, and ba ley die a y challenges, we espec i ely de ec ed 51, 43, and 37 unique g ain pep ide sequences (Supplemen a y Da ase 5). The human pep ide epe oi e was simila in all specimens (Supplemen a y Fig. 6). No ably, all iden ified pep ides we e es ic ed o jus one die a y challenge (Fig. 3b), sugges ing ha ou me hod eadily di e - en ia es pep ides o med om di e en die a y p o eins. Indeed, when he sequences we e mapped on o he g ain p o eomes, he majo i y we e p edomina ely de i ed om he co esponding die a y challenge (Fig. 3c and Supplemen a y Fig. 7). Ou obse a ion ha ewe sequences could be mapped o he ye p o eome om ye challenge u ine is a likely a i ac o he size o he a ailable ye p o eome: he UniP o esou ce con ains ~100× ewe sequences han he whea o ba ley p o eomes, despi e hei genomes ha ing simila sizes. Ou da a sugges ha many o he de ec ed sequences in he ye challenge u ine may be de i ed om ye unanno a ed ye p o eins whose sequences a e also p esen in he whea p o eome. Taken oge he , hese esul s demons a e ha he pep ides de ec ed by ou me hod o igina e om he die . They also confi m he specifici y o ou me hod o dis inguish closely ela ed die a y componen s. P elimina y analysis o banked u ine om pa ien s wi h CeD. To es ou abili y o iden i y whea -de i ed pep ides in he u ine o indi iduals wi h CeD, we analyzed a se o nine u ine samples banked om a p io s udy27. These u ine samples we e collec ed om single oids o pa ien s wi h confi med CeD diagnoses and we e p e iously assayed o glu en p esence by la e al flow immunoassay. This immunoassay used G12 and A1 monoclonal a c Die Fas Glu en Human Volun ee s Time (h) 0 spo u ine die a y glu en challenge pooled u ine ~10 ~18 0 10 20 30 40 Glu en F ee Die Glu en Challenge # Whea Pep ide Sequences Die Glu en-F ee Glu en Human Volun ee s Time (h) sample ex ac ion LC-MS/MS analysis 024 4832 56 pooled u ine die a y glu en challenge pooled u ine b d 0 10 20 30 40 50 # Whea Pep ide Sequences O e nigh Fas Glu en Challenge sample ex ac ion LC-MS/MS analysis Fig. 2 LC–MS/MS enables he iden ifica ion o chemically defined die a y whea pep ides in human u ine. a Pilo s udy design. bNumbe o whea - de i ed pep ide sequences de ec ed in he u ine o 4 heal hy pa icipan s (4 males, 24-28 yea s o age) a e a 24-h glu en- ee die and a e a die a y challenge wi h whea glu en. Ho izon al line ep esen s median. cSimplified s udy design. dNumbe o whea -de i ed pep ide sequences de ec ed in 8 heal hy pa icipan s (5 males, 3 emales, 24-28 yea s o age) a e an o e nigh as and a e a die a y challenge wi h whea glu en. Ho izon al line ep esen s median. In all LC–MS/MS expe imen s, pep ide iden ifica ions we e con olled a a alse disco e y a e o 1% using he PEAKS so wa e decoy- usion app oach59. Sou ce da a a e p o ided as a Sou ce Da a file. ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 4NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions Table 1 Examples o whea -de i ed pep ide sequences iden ified in human u ine. Whea pep ide sequenceaUnip o p o ein accession(s)bP o ein name(s) Mass (Da) # o pa icipan s wi h pep ide GQQQPFPPQQPYPQPQPFPS GDA0_WHEAT; GDA1_WHEAT; GDA2_WHEAT; GDA9_WHEAT α/β-gliadin; α-gliadin 2292.0962 8 GQQQPFPPQQPYPQPQPFP GDA0_WHEAT; GDA1_WHEAT; GDA2_WHEAT; GDA9_WHEAT α/β-gliadin; α-gliadin 2205.0642 7 SQQPEQTISQQPQQPFPQQPHQPQQPYPQQQPYGSSL A0A290XYU3_WHEAT; U5UA46_WHEAT; R9XT67_WHEAT; R9XUY1_WHEAT ω-gliadin 4284.0259 6 SCHVMQQQCC GDB1_WHEAT; GLTC_WHEAT; GDB3_WHEAT; GLTB_WHEAT γ-gliadin; LMW glu enin 1336.4781 6 LGQQQPFPPQQPYPQPQPFPSQQP GDA1_WHEAT; A0A0K2QJX_WHEAT; A0A0E3Z7F7_WHEAT, A0A1K0JNE4_WHEAT α/β-gliadin; α-gliadin 2758.3500 5 SQQPQQPFPQQPHQPQQPYPQ B6ETR9_WHEAT; A0A290XYU3_WHEAT; U5UA46_WHEAT; R9XT67_WHEAT ω-gliadin; LMW glu enin 2512.1882 5 SCHVMQQQCCQ GDB1_WHEAT; GLTC_WHEAT; GDB3_WHEAT; GLTB_WHEAT γ-gliadin; LMW glu enin 1464.5367 5 Py QTFPHQPQQQVPQPQQPQQP GDB2_WHEAT; B6DQB5_WHEAT; R9XWD0_WHEAT; R9XUB9_WHEAT γ-gliadin 2348.1296 5 PQQPPFSQQQQQQQQQQQPPFSQQQQPVL Q5MFQ2_WHEAT; D0EVP4_WHEAT; Q5MFQ1_WHEAT; Q5MFQ6_WHEAT LMW glu enin 3457.6763 3 TQQPQQPFPQQPQQPFPQQPQQPFPQ R9XUS6_WHEAT; A0A290XZ20_WHEAT; A0A290XZ51_WHEAT; A0A290XZ34_WHEAT γ-gliadin 3098.4998 3 TQQPQQPFPQQPQQPFPQTQQPQQPFPQ GDB2_WHEAT; B6DQB5_WHEAT, R9XV87_WHEAT; R9XWD0_WHEAT γ-gliadin 3327.6060 3 QPFPPQQPYPQPQPFP GDA0_WHEAT; GDA1_WHEAT; GDA2_WHEAT; GDA9_WHEAT α/β-gliadin; α-gliadin 1891.9257 2 P[I/L]QPQQPFPQQPQQPFPQPQcB6ETR9_WHEAT; A0A290XYW6_WHEAT; R9XT67_WHEAT; R9XUY1_WHEAT ω-gliadin 2354.1807 2 FLQPQQPFPQQPQQPYPQQPQQPFPQ GDB0_WHEAT, R9XV78_WHEAT; Q9FS77_WHEAT, B6UKM8_WHEAT γ-gliadin 3145.5410 1 CHVMQQQCCQdGDB1_WHEAT; GLTC_WHEAT; GDB3_WHEAT; GLTB_WHEAT γ-gliadin; LMW glu enin 1393.5000 1 Py QQQQPPFSQQPPISQQQQPPFSQQQQPQF GLTA_WHEAT, D2DII3_WHEAT, Q6SPZ1_WHEAT, Q6SPZ3_WHEAT LMW glu enin 3416.6174 1 aResidues in bold indica e CeD- ele an T-cell epi opes, as defined in e . 37. All Cys esidues we e de ec ed in hei ca bamidome hyla ed o ms due o de i a iza ion wi h iodoace amide du ing sample wo kup. “Py Q”indica es glu amine esidues ha had unde gone cycliza ion in o py oglu amic acid. bFo pep ides mapping o mo e han ou p o eins in he UniP o da abase, only he fi s ou accession codes a e displayed. All iden ified pep ide sequences and ull de ails ela ing o pep ide iden ifica ion in each pa icipan s’u ine a e p o ided in Supplemen a y Da ase s 3 and 4. cThe UniP o da abase con ained o he wise iden ical pep ides con aining Leu and Ile a he same posi ion. These wo amino acids a e isoba ic and indis inguishable by ou MS/MS me hod and hus he sequence is epo ed as [I/L]. dThe Me in his pep ide was de ec ed in i s oxidized o m. NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 ARTICLE NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions 5 an ibodies, which specifically ecognize he amino acid mo i s QPQLP(Y/F) and QLP(Y/F)PQP, espec i ely36. Samples (n=4) om pa ien s epo ing adhe ence o a GFD and in which he measu ed glu en pep ide concen a ions we e below he la e al flow assay limi o quan ifica ion we e analyzed as nega i e con- ols. In acco dance wi h he la e al flow esul s, we did no measu e pep ides mapping o he whea p o eome by LC–MS/MS. In con as , all samples (n=5) o which la e al flow assays indica ed high appa en glu en pep ide concen a ions, yielded a leas 17 whea pep ide sequences (Fig. 4)byLC–MS/MS. The mean numbe o iden ified whea pep ides (74 pe sample) was much highe han hose desc ibed in ou p oo -o -concep s udies wi h heal hy pa icipan s (24 pe sample; Fig. 2b, d) while he numbe o human pep ides was simila (Supplemen a y Fig. 8). S ikingly, one sample om a pa ien wi h CeD yielded 206 dis- inc whea pep ides, a quan i y ou imes g ea e han in any heal hy pa icipan analyzed p e iously. These in iguing obse - a ions led us o hypo hesize ha he u ina y whea pep ide epe oi es o pa ien s wi h CeD a e much mo e di e se han non- CeD indi iduals. We, he e o e, ini ia ed a p ospec i e clinical s udy o o mally es his hypo hesis, desc ibed below. Clinical s udy. Adul pa ien s wi h gas oin es inal symp oms including dyspepsia, bloa ing and dia hea who we e ea ing glu en-con aining die s and we e unde going e alua ion o CeD we e ec ui ed wi hin he Celiac Disease P og am a he S an o d Diges i e Heal h Cen e , along wi h gende -ma ched heal hy con ols (Fig. 5a). CeD s. non-CeD diagnoses we e no known a he ime o en ollmen and we e subsequen ly es ablished by se ology and endoscopy wi h biopsy. All pa icipan s unde ook a die a y glu en challenge and u ine collec ion acco ding o he op imized s udy design depic ed in Fig. 2c. A he end o he ec ui men pe iod, u ine samples we e p ocessed and analyzed by LC–MS/MS. Compa ed o heal hy con ols (n=8) and pa ien s ul ima ely diagnosed wi h non-celiac gas oin es inal diso de s (n=5), pa ien s wi h CeD (n=6) had app oxima ely 5 imes mo e unique glu en pep ides in hei u ine (Fig. 5b), while he numbe o endogenous human pep ides was simila in all h ee g oups (Supplemen a y Fig. 9). Consis en wi h ou pilo s udies, a ian s o he inna e immune esponse s imula ing-pep ide GQQQPFPPQQPYPQPQPFPS we e de ec ed in all indi iduals, ega dless o clinical s a us (Supple- men a y Da ase 7). We analyzed whe he pa ien s wi h CeD had inc eased di e si y o pep ides wi h T-cell epi opes known o s imula e he adap i e immune esponse. Indeed, he numbe o pep ides wi h a leas one CeD- ele an T-cell epi ope was Fig. 3 LC–MS/MS analysis o u ine om wo heal hy pa icipan s challenged wi h die a y whea , ye, and ba ley. a Illus a ion o sequence di e ences be ween closely ela ed gliadin, secalin, and ho dein p o eins om whea , ye, and ba ley, espec i ely. The boxes highligh sou ce- specific sequence egions. bAll possible in e sec ions o he h ee se s o pep ide sequences iden ified a e he g ain challenges a e depic ed as an UpSe plo 60. The indi idual se size is plo ed o he le o each ow, wi h he black po ion o he ba ep esen ing sequences ha map uniquely o he espec i e die a y challenge p o eome, and he o ange po ion ep esen ing hose ha do no . The e a e no in e sec ions be ween hese se s o pep ides, sugges ing he pep ides depic ed in o ange a e none heless likely o ha e o igina ed om he specific g ain used o he die a y challenge. cNumbe o dis inc pep ide sequences uniquely mapping o he whea , ba ley, and ye p o eomes ollowing die a y challenges. Sequences ha mapped o mo e han one o he p o eomes we e excluded om his figu e. Sequences ha mapped o mo e han one o he p o eomes we e excluded om his figu e. In he ye die , sequences mapping uniquely o he whea p o eome we e only ound in he ye die , sugges ing hey a e ye unanno a ed ye pep ides. A ull lis o obse ed pep ide sequences and hei espec i e p o eome mappings is p o ided in Supplemen a y Da ase 5. Pa icipan s in his expe imen we e males, ages 27 and 28. Sou ce da a a e p o ided as a Sou ce Da a file. 0 50 100 150 200 # Whea Pep ide Sequences by LC-MS/MS Appa en Glu en Pep ide Concen a ion by La e al Flow Assay <6.3 ng/µL >100 ng/µL Fig. 4 LC–MS/MS analysis o banked u ine samples om pa ien s wi h CeD27.U ine collec ed om a single oid was es ed o he p esence o glu en by an A1 and G12 monoclonal an ibody-based la e al flow assay. Each da a poin ep esen s a single u ine sample, and he ho izon al ba indica es he median. In o ma ion associa ed wi h he iden ifica ion o hese pep ide sequences by LC–MS/MS is p o ided in Supplemen a y Da ase 6. Sou ce da a a e p o ided as a Sou ce Da a file. ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 6NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions app oxima ely 4- old highe in pa ien s wi h CeD (Fig. 5c). CeD- ele an T-cell epi opes occu ed mo e equen ly in pa icipan s wi h CeD, and se en di e en T-cell epi opes comple ely absen in con ols we e ound in pa ien s (Table 2and Supplemen a y Table 2). O e all, we measu ed 289 unique whea pep ides om pa ien s wi h CeD ha we e no de ec ed in heal hy indi iduals o pa ien s wi h non-celiac gas oin es inal diso de s, whe eas only 37 pep ides we e sha ed be ween all h ee g oups (Fig. 5d and Supplemen a y Da ase 7). To gain insigh in o he di e si y o pep ide sequences ound in pa ien s wi h CeD, we anked he equency by which CeD- specific pep ide sequences we e de ec ed (Supplemen a y Da a- se 8). Jus ou pep ides we e ound in he majo i y o (≥4/6) pa ien s (Table 3), whe eas mos (201) we e de ec ed only in single indi iduals. To unde s and he chemical space occupied by he pep ides ha we e ound in pa ien s wi h CeD, we aligned he de ec ed pep ide sequences o he whea p o eome (Fig. 5e). S ikingly, pa ien s wi h CeD no only had mo e a ian s (i.e., sligh ly longe o sho e e sions) o pep ides ound in con ols, bu hey also possessed pep ides ha mapped o dis inc egions o he whea p o eome (Fig. 5 –h and Supplemen a y Fig. 10). Fo example, h ee pa ien s had pep ides spanning esidues 194–199 o γ-gliadin (GDBX) ha we e absen in con ols. Taken oge he , hese esul s demons a e ha pa ien s wi h CeD ha e pep i- domes ha a e mo e di e se in hei chemical and biological p ope ies compa ed o con ol pa icipan s. Discussion Despi e ex ensi e esea ch in o he mechanisms by which glu en p o eins play a ole in CeD pa hogenesis (Fig. 6), he iden i ies o he ac ual die -de i ed molecules ha in e ac wi h he human 1262 GQQQP F P P Q KM Residue Numbe alpha-gliadin e e ence sequence de ec ed unique pep ide sequences sequence alginmen and mapping Max F equency 3 no malized o max equency ac oss all subjec s Heal hy Con ols (No GI Complain s) (n=8) Pa ien s Unde E alua ion o CeD (n=11) se ology endoscopy CeD Pa ien s (n=6) O he GI Pa ien s (n=5) 0 50 100 150 200 CeD Pa ien s Heal hy Con ols # Whea Pep ide Sequences O he GI Pa ien s * *ns 0 20 40 60 CeD Pa ien s Heal hy Con ols O he GI Pa ien s * *ns # Whea Pep ide Sequences wi h T-cell Epi opes ab c e d gh CeD Pa ien s 289 O he GI Pa ien s 5 Heal hy Con ols 16 20 68 1 37 NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 ARTICLE NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions 7 immune sys em in he gu and o he o gans ha e been a long- s anding mys e y. As a s ep owa d illumina ing hese molecula in e ac ions, we de eloped a LC–MS/MS-based assay ha di ec ly de ec s he me abolic p oduc s o ood g ains. In con as o in i o glu en p epa a ions ha ha e been s anda d esea ch ools in he CeD field o se e al decades, ou me hod is com- pa ible wi h in i o die a y challenge. I he e o e can e eal he p ecise chemical s uc u es o die a y pep ides ha may d i e CeD in humans. Fu he mo e, unlike adi ional an ibody eagen s, he un a ge ed LC–MS/MS eadou eadily dis inguishes he p ecise amino acid sequences o diges ion- esis an pep ides om whea , ba ley, and ye. Thus, his me hod p o ides a unique window in o glu en’s ADME cha ac e is ics and immunogenici y. Indeed, we obse ed mo e a ied whea pep ide epe oi es om pa ien s wi h CeD han was p e iously app ecia ed –and ones ha we e much mo e di e se han hose ound in non-CeD indi iduals. Ini ially, ou e o s we e hinde ed by LC–MS/MS me hods ha we e unsui ed o he unique challenges o u ina y pep idome analysis. U ine con ains high concen a ions o me aboli es and sal s ha a e di ficul o sepa a e om pep ides and a e incom- pa ible wi h LC–MS/MS33,34. The e o e, we de eloped an ex ac ion echnique o emo e hese in e e ing compounds. This allowed us o achie e ou main goal o whea pep ide iden ifica ion, while also imp o ing ou abili y o measu e u ina y pep ides o igina ing om o he endogenous and die a y sou ces (Supplemen a y Fig. 3). Ou wo kflow (Fig. 1) is compa ible wi h s anda d e e sed-phase LC–MS/MS ins umen a ion a ailable in mos p o eomics labo a o ies32. Compa ed o published app oa- ches employing solid-phase ex ac ion echniques, ou me hod iden ifies app oxima ely 2 o 10 imes mo e endogenous human pep ides om ypical 1–10 mL u ine samples33,34,43,44. Mo eo e , ou sample p epa a ion echnique equi es only ~6 h, acili a ing su ficien h oughpu o us o unde ake a compa a i e analysis o he whea -de i ed u ina y pep idomes o pa ien s wi h CeD and heal hy con ols. Gi en i s high specifici y (Fig. 3), ou wo kflow has po en ial u ili y o o he s udies equi ing analysis o he u ina y pep idome. Al hough he e we ocused on he iden ifica ion o whea -de i ed pep ides, we also iden ified o e 30,000 human pep ides (Supplemen a y Da ase s 1–7), which is o ou knowledge, is he la ges collec ion o u ina y pep ides sequenced by LC–MS/MS o da e. Pa en he ically, we no e ha p e ious su eys o he u ine pep idome ound ha hyd oxyp oline-modified collagen pep ides we e among he mos abundan 33,34,43,44. Howe e , he e we did no allow hyd o- xyp oline as a a iable modifica ion in ou da abase sea ches, as his modifica ion was no ele an o ou o e iding goal o de ec ing biologically ele an , whea -de i ed pep ides. The e o e, we ha e deposi ed ou aw da a in he PRIDE da abase o acil- i a e he iden ifica ion o addi ional pep ides (e.g., by including o he a iable modifica ions o using al e na i e sea ch engines). This s udy has iden ified, o he fi s ime, he specific amino acid sequences and pos - ansla ional modifica ions o pep ides esul ing om in i o diges ion o glu en. In ou pilo iden ifi- ca ion o die a y whea pep ides in heal hy pa icipan s (Fig. 2), we iden ified sequences wi h known CeD ele ance, as well as o he s ha had ne e been implica ed in CeD o iden ified as being esis an o gas oin es inal diges ion. These findings unde sco ed he alue o ou un a ge ed LC–MS/MS me hod and mo i a ed us o analyze he u ine o pa ien s wi h CeD. Using banked u ine specimens, we ound ha whea - consuming pa ien s wi h CeD had subs an ially g ea e glu en pep ide di e si y han non-CeD indi iduals (Figs. 2and 4). We confi med his finding wi h a subsequen con olled clinical s udy (Fig. 5) in which pa icipan s ollowed ou well-defined glu en challenge p o ocol (Fig. 2d). Mechanis ically, his pep ide Fig. 5 The u ina y whea pep idomes o pa ien s wi h CeD a e significan ly mo e di e se han in heal hy con ols o pa ien s wi h non-celiac gas oin es inal diso de s. a Clinical s udy design. Pa icipan s we e ec ui ed o e app oxima ely 2 yea s wi hin he Celiac Disease P og am a he S an o d Diges i e Heal h Cen e . Pa icipan cha ac e is ics a e epo ed in Supplemen a y Tables 3 and 4. bNumbe o whea pep ides de ec ed in pooled u ine samples collec ed o 8 h subsequen o a die a y challenge wi h wo bagels (~18 g glu en). Pa ien s wi h CeD had subs an ially mo e whea pep ide sequences compa ed o heal hy con ols (p=0.017) o pa ien s wi h non-celiac gas oin es inal diso de s (p=0.017). cNumbe o unique whea -de i ed pep ide sequences wi h a leas one T-cell epi ope in pa ien s wi h CeD compa ed o heal hy con ols (p=0.019) and pa ien s wi h non-celiac gas oin es inal diso de s (p=0.046). b,cS a is ics we e de i ed om n=6 pa ien s wi h celiac disease, n=5 pa ien s wi h non-celiac gas oin es inal diso de s, and n=8 heal hy con ols using a one-way K uskal–Wallis ANOVA/Dunn’s mul iple compa ison es ; ho izon al ba ep esen s he median. dVenn diag am compa ing he unique pep ide sequences de ec ed in heal hy con ols, pa ien s wi h CeD, and o he GI pa ien s. A ull lis ing o pep ide sequences is p o ided in Supplemen a y Da ase 7. eSchema ic desc ibing how he equencies a which u ina y whea -de i ed pep ides map o p o eins in he whea p o eome we e plo ed as hea maps. –hHea map ep esen a ion o de ec ed pep ide sequences in alpha/be a gliadin 2 (GDA2), γ-gliadin 2 (GDB2), and gamma-gliadin X (GDBX) demons a es ha while many pep ides map o he same egion o he whea p o eome, he u ina y whea pep idomes o pa ien s wi h CeD also occupy dis inc chemical space. Below he maps, he loca ions o CeD T-cell epi opes a e highligh ed in ed, and he ela i e densi y o Gln and P o is indica ed in yellow. Hea maps o o he p o eins in he whea p o eome a e p o ided in Supplemen a y Fig. 10. In b–h, all samples we e analyzed by LC–MS/MS in duplica e and he agg ega ed esul s a e shown. Analyses o indi idual eplica es a e p o ided in Supplemen a y Fig. 11. Full de ails on LC–MS/MS iden ifica ion o pep ide sequences a e p o ided in Supplemen a y Da ase 7. Sou ce da a a e p o ided as a Sou ce Da a file. Table 2 Numbe o de ec ed T-cell epi opes in pa ien s wi h celiac disease (CeD), non-celiac gas oin es inal diso de s (GI), and heal hy con ols (HC). Epi ope name(s)aEpi ope sequence CeD GI HC DQ2.5-glia-α1a PFPQPQLPY 2 0 0 DQ2.5-glia-α1b PYPQPQLPY 1 0 0 DQ2.5-glia-α2 PQPQLPYPQ 2 0 0 DQ2.5-glia-α3 FRPQQPYPQ 1 0 0 DQ2.5-glia-γ1/DQ8.5-glia-γ1/ DQ8-glia-γ2 PQQSFPQQQ 1 0 1 DQ2.5-glia-γ3/DQ8-glia-γ1b QQPQQPYPQ 6 1 0 DQ2.5-glia-γ4a SQPQQQFPQ 1 0 0 DQ2.5-glia-γ4b PQPQQQFPQ 8 4 3 DQ2.5-glia-γ4c/DQ8-glia-γ1a QQPQQPFPQ 105 29 36 DQ2.5-glia-γ4d PQPQQPFCQ 2 0 1 DQ2.5-glia-γ4e LQPQQPFPQ 28 5 4 DQ2.5-glia-γ5 QQPFPQQPQ 37 10 15 DQ2.5-glia-ω1/DQ2.5-ho -1/ DQ2.5-sec-1 PFPQPQQPF 3 0 1 DQ2.5-glia-ω2 PQPQQPFPW 1 0 0 DQ2.5-glu -L1/DQ2.2-glu -L1 PFSQQQQPV 17 2 11 DQ2.5-glu -L2 FSQQQQSPF 1 0 0 DQ8.5-glu -H1/DQ8-glu -H1 QGYYPTSPQ 2 0 1 aEpi ope sequences and nomencla u e a e defined in e . 37. ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 8NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions di e si y may s em om he ac ha indi iduals wi h ac i e CeD ha e ele a ed in es inal pe meabili y, which likely allows whea - de i ed pep ides o mo e eadily c oss he epi helial ba ie 5. This phenomenon could lead o inc eased pep ide concen a ions in he sys emic ci cula ion and in u ine, he eby enhancing hei abili y o be de ec ed. A di e en (o complemen a y) explana ion is ha pa ien s wi h CeD diges and/o me abolize whea di e - en ly han o he indi iduals. Such po en ial di e ences in whea me abolism could con ibu e o he onse o CeD. Fu u e analyses o pa ien s wi h o he non-CeD en e opa hies known o cause “leaky gu ”such as opical sp ue as well as diso de s in which he mucosa emains g ossly in ac such as i i able bowel synd ome wi h dia hea should be insigh ul in pinpoin ing he o igins o he inc eased pep ide di e si y obse ed in pa ien s wi h CeD45,46. Mo e undamen ally, analysis o u ine om indi iduals con- suming die s wi h a ious ood addi i es, such as mic obial ansglu aminase, should help o add ess he eme ging hypo hesis ha hese addi i es con ain inc eased le els o immunogenic pep ides and/o di ec ly con ibu e o inc eases in in es inal pe meabili y47,48. We iden ified 17 dis inc immuno oxic epi opes known o be p esen ed on MHC class II molecules and ecognized by T cells ha eside in he lamina p op ia o pa ien s wi h CeD. Many o hese epi opes occu ed mo e equen ly in pa ien s wi h CeD han in con ols (Table 2). No ably, howe e , we we e unable o de ec in i s in ac o m an ex ensi ely s udied and highly inflamma o y 33- esidue pep ide om α2-gliadin ha accumu- la es in he mammalian gu lumen unde physiological condi ions by i ue o i s esis ance o gas oin es inal p o eolysis4. I s appa en absence om human u ine could be due o physico- chemical p ope ies ha hinde abso p ion, p omo e sys emic me abolism, o make i unamenable o chemical ex ac ion and/ o de ec ion by mass spec ome y. None heless, i is jus as no ewo hy ha an analogous 26- esidue pep ide om γ-gliadin, FLQPQQPFPQQPQQPYPQQPQQPFPQ, was obse ed in 2/7 pa ien s wi h CeD. This pep ide is also p o eoly ically esis an and highly inflamma o y o CeD-specific T cells on accoun o i s poly alency38. Despi e he abundance o immunogenic whea pep ides in u ine, hus a we ha e been unable o defini i ely iden i y any pep ide ha unde wen egioselec i e Gln deamida ion by ansglu aminase 2 (TG2), an impo an s ep in CeD pa hogenesis (Fig. 6). Because TG2-ca alyzed deamida ion inc eases he immunogenici y o glu en-de i ed pep ides by se e al o de s o magni ude4, i is possible ha a ela i ely small mola ac ion o abso bed glu en is deamida ed in pa ien s wi h CeD, making he concen a ions o hese pep ide species oo low o be de ec able. Al e na i ely (o addi ionally), deamida ion may enhance he me abolic labili y o ci cula ing glu en pep ides. The possibili y ha hei high a fini y o HLA-DQ2 o -DQ8 al e s hei ADME cha ac e is ics should no be o e looked. Rega dless, u u e de elopmen o a ge ed LC–MS/MS me hods could acili a e he de ec ion o low-abundance deamida ed glu en pep ides. The e is inc easing in e es in non-in asi e moni o ing o GFD compliance ia u inalysis26–28,49, as a la ge p opo ion o pa ien s wi h CeD expe ience symp oms e en when a emp ing o adhe e o a GFD50. A ailable me hods p edominan ly exploi he G12 monoclonal an ibody, which p e e en ially ecognizes he immuno oxic QPQLP(Y/F) mo i ound in se e al CeD- ele an T-cell epi opes36. Ou da a sugges ha his mo i is a e in u ine om indi iduals who consume die a y whea (Table 2). In con- as , he sequence GQQQPFPPQQPYPQPQPFPS appea ed in whea p oduc pa ially diges ed whea p oduc whea pep ides p o eases ansloca ion ansloca ion deamida ion TG2 GI T ac Lamina P op ia Blood S eam il a ion ( o be exc e ed) Kidneys Bowman’s Space Glome ula Capilla ies Glome ulus Fig. 6 Schema ic ep esen a ion o ele an me abolic p oduc s o die a y glu en in humans. Whea -de i ed p o eins such as glu en a e esis an o gas oin es inal p o eolysis. Incomple e diges ion p ese es disease-causing T-cell epi opes in pep ides ha accumula e in he gu lumen. A po ion o hese glu en pep ides ansloca e o he lamina p op ia whe e ansglu aminase 2 (TG2) deamida es selec ed Gln esidues (b own do s) in some sequences, he eby con e ing hem in o Glu esidues (yellow do s). Abso bed pep ides can also en e he bloods eam, and a e u he me abolized and/ o exc e ed in u ine. Table 3 Whea -de i ed pep ide sequences de ec ed in he majo i y o pa ien s wi h CeD bu no in con ol g oups. Whea pep ide sequence P o ein name # o CeD pa ien s wi h pep ide PQQQIPQQHQIPQQPQQFPQQ ω-gliadin 5 GQQQQQFPGQQQPFPPQQPYPQPQP α/β-gliadin 4 PELQQPIPQQPQQPFPLQPQQPFPQQ ω-gliadin 4 PTSPQQSGQGQQPGQWQQPGQGQPG HMW-glu enin 4 NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-022-28353-1 ARTICLE NATURE COMMUNICATIONS | (2022) 13:888 | h ps://doi.o g/10.1038/s41467-022-28353-1 | www.na u e.com/na u ecommunica ions 9