OPEN
Responsi e popula ion dynamics and wide
seeding in o he duodenal lamina p op ia o
ansglu aminase-2-speci ic plasma cells in
celiac disease
R Di Ni o
1
, O Sni
2
, K Kaukinen
3
, G Yaa i
4
, KEA Lundin
2,5
, NT Gup a
6
, SH Kleins ein
6,7
, M Cols
8
,
A Ce u i
8
,MMa
¨ki
9
, MJ Shlomchik
1
and LM Sollid
2
A hallma k o celiac disease is au oan ibodies o ansglu aminase 2 (TG2). By isualizing TG2-speci ic an ibodies by
an igen s aining o a ec ed gu issue, we iden i ied TG2-speci ic plasma cells in he lamina p op ia as well as an ibodies
in he subepi helial laye , inside he epi helium, and a he b ush bo de . The equency o TG2-speci ic plasma cells we e
ound no o co ela e wi h se um an ibody i e s, sugges ing ha an ibody p oduc ion a o he si es may con ibu e o
se um an ibody le els. Upon commencemen o a glu en- ee die , he equency o TG2-speci ic plasma cells in he
lesion d opped d ama ically wi hin 6 mon hs, ye some cells emained. The equency o TG2-speci ic plasma cells in he
celiac lesion is hus dynamically egula ed in esponse o glu en exposu e. Lase mic odissec ion o plasma cell pa ches,
ollowed by an ibody gene sequencing, demons a ed ha clonal cells we e seeded in dis inc a eas o he mucosa. This
was con i med by immunoglobulin hea y chain epe oi e analysis o plasma cells isola ed om indi idual biopsies o
wo un ea ed pa ien s, bo h o TG2-speci ic and non-TG2-speci ic cells. Ou esul s shed new ligh on he p ocesses
unde lying he B-cell esponse in celiac disease, and he app oach o s aining o an igen-speci ic an ibodies should be
applicable o o he an ibody-media ed diseases.
INTRODUCTION
Celiac disease (CD) is a mul i ac o ial diso de cha ac e ized by
an in es inal in lamma o y esponse o inges ed ce eal glu en
p o eins.
1
The human leukocy e an igen associa ion and he
cen al ole o CD4 þT cells in he pa hogenesis a e ho oughly
in es iga ed.
2
T cells o he lesion ecognize glu en pep ides ha
a e deamida ed in i o by he enzyme ansglu aminase 2
(TG2).
3,4
No ably, he g ea majo i y o CD pa ien s de elop
an au oan ibody esponse, wi h TG2 i sel being he main
au oan igen.
5
I is no known whe he hese an ibodies ha e
a ole in he pa hophysiology o CD, ye an i-TG2 immuno-
globulin A (IgA) an ibodies a e inc easingly used as diag-
nos ic ool
6
and in he ollow-up o he ea men , as upon
commencemen o a glu en- ee die (GFD) au oan ibodies
disappea om se um wi hin mon hs.
7
An i-TG2 an ibodies
a e p oduced by plasma cells (PCs) localized in he lamina
p op ia o he in es inal mucosa,
8–10
bu PCs localized else-
whe e may also con ibu e o he an ibody p oduc ion.
Ko ponay-Szabo e al.
11
de eloped a double-colo immuno-
luo escence me hod ha allowed isualizing an ibody deposi s
in co espondence wi h he subepi helial TG2 laye in he small
in es ine, as well as in o he issues. This me hod builds on
1
Depa men o Immunology, Uni e si y o Pi sbu gh, Pi sbu gh, Pennsyl ania, USA.
2
Cen e o Immune Regula ion and Depa men o Immunology, Uni e si y o Oslo and
Oslo Uni e si y Hospi al-Rikshospi ale , Oslo, No way.
3
Depa men s o In e nal Medicine and Gas oen e ology, Tampe e Uni e si y Hospi al and School o Medicine,
Uni e si y o Tampe e, Tampe e, Finland.
4
Bioenginee ing P og am, Facul y o Enginee ing, Ba -Ilan Uni e si y, Rama Gan, Is ael.
5
Cen e o Immune Regula ion and
Depa men o Gas oen e ology, Oslo Uni e si y Hospi al-Rikshospi ale , Oslo, No way.
6
In e depa men al P og am in Compu a ional Biology and Bioin o ma ics, Yale
Uni e si y, New Ha en, Connec icu , USA.
7
Depa men o Pa hology, Yale Uni e si y School o Medicine, New Ha en, Connec icu , USA.
8
Immunology Ins i u e, Depa men
o Medicine, Icahn School o Medicine a Moun Sinai, New Yo k, New Yo k, USA and
9
Tampe e Cen 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. Co espondence: R Di Ni o ([email p o ec ed])
Recei ed 16 Oc obe 2014; accep ed 12 May 2015; published online 8 July 2015. doi:10.1038/mi.2015.57
ARTICLES na u e publishing g oup
254 VOLUME 9 NUMBER 1 |JANUARY 2016 |www.na u e.com/mi
colocaliza ion be ween IgA and he TG2 p o ein, he la e
usually iden i ied by means o he CUB7402 mu ine mono-
clonal an ibody. Impo an ly, his has been p oposed as a ool
o ea ly diagnosis
12
on a ec ed indi iduals wi hou signs o
illous a ophy
13,14
o wi h nega i e/bo de line se ology.
15
Recen ly, we epo ed a low cy ome y-based me hod o
desc ibe and enume a e TG2-speci ic PCs om eshly
ob ained biopsies o he small in es ine o CD pa ien s.
10
We ound ha in un ea ed CD pa ien s he equency o hese
cells was excep ionally high, and ha hey p oduced an ibodies
wi h limi ed soma ic hype mu a ion bu none heless eason-
able a ini y. In he same wo k we epo ed an immuno luo -
escence-based me hod o isualize TG2-speci ic PCs on
c yosec ions. As his ma e ial can be s o ed o p olonged
pe iods o ime and equi es ela i ely li le amoun o
specimen, s udies on c yosec ions p o ide a simple ye e y
alid al e na i e o he analysis o in es inal humo al
esponses. E en hough signi ican disco e ies o he an i-
TG2 an ibody esponse ha e been done, i s impo ance in he
diagnos ic wo kup o CD and he unknown ole in he
pa hogenesis o he disease sugges ha mo e esea ch on his
opic is necessa y. In his wo k, we expand ou p e ious
indings on he in es inal an i-TG2 esponse by in es iga ing
and desc ibing he loca ion o TG2-speci ic PCs and an ibodies,
he pheno ype o hese cells as well as he niche suppo ing
hem, and he kine ics o hei disappea ance upon commence-
men o a GFD, and we discuss a mechanism o a possible
pa hogenic in ol emen . Ou indings p o ide a new ool o
analyze he an ibody esponse in CD and in gene al he
speci ici y o an ibody esponses di ec ly in a ec ed issues.
RESULTS
TG2-speci ic plasma cells a e abundan and accumula e
close o he gu lumen in he celiac lesion
We ecen ly epo ed ha he equency o TG2-speci ic cells in
he in es inal lesion o un ea ed CD pa ien s is excep ionally
high, a e aging 10%. He e, we u he in es iga e his ma e by
using ou no el immuno luo escence echnique
10
ha akes
ad an age o soluble bio inyla ed TG2 o isualize in acellula
deposi s o an i-TG2 an ibodies ha , in combina ion wi h an
an i-CD138 an ibody, allowed us o di ec ly iden i y TG2-
speci ic PCs (Figu e 1). This me hod di e s om he
CUB7402/IgA cos aining me hod
11
ha e eals he loca ion
o he TG2 p o ein in he issue, and indi ec ly he p esence o
speci ic an ibody deposi s because o he colocaliza ion wi h
IgA. A low magni ica ion ( 10), s ained c yosec ions om
un ea ed CD pa ien s (Figu e 1a,b) showed signi ican
plasmacy osis in he lamina p op ia,
16
and he localiza ion
o he PCs in he lamina p op ia was mo e owa d he in es inal
lumen han in heal hy con ols (Figu e 1d). Figu e 1c is
ep esen a i e o he mos common condi ion obse ed in
GFD- ea ed pa ien s, wi h good eco e y o he in es inal
mo phology, ma kedly dec eased plasmacy osis, and he
p esence o a ew esidual TG2-speci ic PCs. In ac , despi e
he die , we usually do no obse e a comple e disappea ance o
TG2-speci ic PCs.
In e es ingly, in pa ien s wi h ac i e disease we consis en ly
obse ed ha TG2-speci ic PCs we e p e e en ially ound in he
lamina p op ia in p oximi y o he lumen, as e iden om
Figu e 1a,b and u he shown in Figu e 1e a highe
magni ica ion ( 40). The single channel images showed
ha all he TG2 þcells exp ess wi h he an i-CD138 ma ke ,
demons a ing ha hese a e PCs.
Double-colo immuno luo escence s udies o in si u loca-
liza ion o TG2 and IgA sugges ed ha in CD pa ien s, an i-TG2
an ibodies a e deposi ed wi h TG2 in he subepi helial laye .
11
He e by using bio inyla ed TG2 as bai o speci ic an ibodies,
we we e able o isualize an ibodies no only wi hin PCs
(as desc ibed abo e), bu also a hei ex acellula loca ion as
deposi s, p ima ily in he subepi helial laye in associa ion wi h
TG2, as expec ed (Figu e 1 ). Fu he mo e, in many cases TG2-
speci ic an ibodies seemed o be p esen in he epi helial cells
and a he b ush bo de in CD pa ien s, as shown in Figu e 1g.
Al hough i can be a gued ha isualiza ion o subepi helial
deposi s by bio in-labeled ecombinan TG2 may be sensi i e o
a i ac s because o unspeci ic binding o he bai o endogenous
ib onec in,
15
his does no seem o be an issue unde he
condi ions used he e. In ac in heal hy con ols who do no
p oduce an i-TG2 an ibodies, bu also possess he ib onec in-
TG2 subepi helial laye (Figu e 1g), only weak binding o he
bai o ib onec in was obse ed ( o ins ance, compa e
Figu e 1d wi h Figu e 1a,b).
TG2-speci ic plasma cells o he mucosa d op 6 mon hs
a e commencemen o a GFD
The abili y o s ain an igen-speci ic PCs in ozen samples by
immuno luo escence, as well as he a ailabili y o biopsies
aken om he same pa ien s a di e en imes and s o ed in he
eeze , allowed us o s udy he kine ics o he disappea ance o
TG2-speci ic PCs a ea ly s ages o GFD. We collec ed ozen
c yosec ions o in es inal specimens om CD pa ien s a
diagnosis and a e 6, 12, o 24 mon hs o GFD, and enume a ed
TG2-speci ic PCs by immuno luo escence. The esul s a e
epo ed in Figu e 2a as pe cen age o TG2-speci ic cells in he
PC compa men . All pa ien s had ele a ed amoun s o hese
cells a diagnosis (n¼15, a e age 9.6%, ange 3.0–18.6%), and
his equency ma kedly dec eased du ing he pe iod wi h GFD.
A e 6 and 12 mon hs o die , TG2-speci ic cells d opped o an
a e age o 1.7% (n¼6, ange 0–4.1%) and 1.4% (n¼9, ange 0–
5.7%) espec i ely. Fo 3 pa ien s om whom we we e able o
ob ain an addi ional specimen a e 24 mon hs o GFD, we did
no de ec any TG2-speci ic PCs a his ime poin . Howe e , in
hese pa ien s TG2-speci ic PCs we e absen o in low amoun s
al eady a he i s sampling (0 and 0.9% in he 2 pa ien samples
a e 6 mon hs, and 0% o 1 pa ien sampled a 12 mon hs). As
in some cases a ela i ely high equency o TG2-speci ic PCs
was s ill de ec ed a e 6–12 mon hs o GFD, and as some o he
pa ien s included in his s udy had bo de line o posi i e
se ology acco ding o endomysial an ibodies o IgA an i-TG2
despi e GFD (yellow do s in Figu e 2a), we asked whe he
esidual p esence o TG2-speci ic PCs co ela ed wi h se o-
posi i i y. This was no he case a 6 mon hs, as all o he cases a
ARTICLES
MucosalImmunology |VOLUME 9 NUMBER 1 |JANUARY 2016 255
his ime poin we e below cu o o se um IgA an i-TG2
posi i i y (acco ding o he Celikey es ). O he 5 pa ien s who
we e se oposi i e o bo de line se oposi i e a 12 mon hs, 3 had
de ec able TG2-speci ic in es inal plasma cells. These indings
aise he ques ion o whe he in gene al he TG2-speci ic PC
equency in un ea ed pa ien s co ela es wi h hei se um
i e . As shown in Figu e 2b, we ound no co ela ion be ween
PC equency and se um i e . This may be because o he
p esence o pa ches in he in es inal lesion wi h di e en
densi ies o TG2-speci ic PCs, and he e o e one single biopsy
specimen may no be ep esen a i e o he small in es ine in i s
en i e y. In an a emp o add ess his, om one pa ien we we e
Figu e 1 Visualiza ion o ansglu aminase 2 (TG2)-speci ic plasma cells (PCs) and an ibodies by immuno luo escence on c yosec ions om celiac
disease (CD) pa ien s. (a–d) Low-magni ica ion images o sec ions om wo un ea ed pa ien s, one glu en- ee die (GFD)- ea ed pa ien , and one
con ol indi idual. TG2-speci ic plasma cells (iden i ied by cos aining o he CD138 plasma cell ma ke , in ed, and TG2-speci ic an ibody deposi s as
e ealed by he bio inyla ed TG2 an igen, in g een) a e e iden in un ea ed pa ien s, dec eased in he GFD- ea ed pa ien , and absen in he con ol. (e)
Highe -magni ica ion images om an un ea ed pa ien , u he e ealing he apical localiza ion o TG2-speci ic plasma cells (in acellula an ibody
deposi s as e ealed by bio inyla ed TG2, in g een, as shown in cen e and igh panels). ( ) Colocaliza ion o immunoglobulin A (IgA) and an i-TG2
an ibodies (as de ec ed by bio inyla ed TG2), showing PCs and deposi s in he lamina p op ia, as well as addi ional deposi s in he subepi helial laye and
on he b ush bo de . (g) Colocaliza ion o IgA and TG2 (as de ec ed by he an i-TG2 monoclonal an ibody (mAb) CUB7402), showing a di e en pa e n in
CD pa ien s and con ols, wi h ex ensi e o e lap be ween he wo s ainings obse ed in CD pa ien s only.
ARTICLES
256 VOLUME 9 NUMBER 1 |JANUARY 2016 |www.na u e.com/mi
able o ob ain h ee biopsies sampled om di e en in es inal
loca ions. Flow cy ome y quan i ica ion o TG2-speci ic PCs
showed consis en equency among he h ee samples (4.5, 4.0,
and 4.3%, Figu e 2c) and hese da a, al hough hey need o be
subs an ia ed, speak owa d a ai ly homogenous dis ibu ion
o au o eac i e PCs in biopsies o he uppe small bowel o CD
pa ien s.
To u he unde s and he ac o s egula ing TG2-speci ic
plasma cell su i al, we also in es iga ed he PC exp ession o
TACI ( ansmemb ane ac i a o and calcium modula o and
cyclophilin ligand in e ac o ), BCMA (B-cell ma u a ion
an igen), and BAFF-R (B-cell ac i a ing ac o ecep o ) ha
a e ecep o s o BAFF and APRIL (a p oli e a ion-inducing
ligand) in ol ed in B-cell su i al. S aining gu issue sec ions
wi h TACI- and BAFF-R-speci ic an ibodies showed ha TACI
was exp essed in a subse o PCs, and in some cases i s
exp ession was pai ed o ha o BAFF-R (Figu e 3a,b). A
simila he e ogenei y ega ding hese ma ke s was obse ed in
TG2-speci ic cells, wi h no signi ican di e ence compa ed
wi h o he PCs (Figu e 3c). A signi ican ac ion o PCs
appea ed o exp ess BCMA (Figu e 3d). In o de o imp o e
he analysis o BAFF and APRIL ecep o s, we used a low
cy ome y app oach ha yields a mo e quan i a i e and
objec i e se o da a. Figu e 3e and 3 shows he pe cen-
age o IgA PCs exp essing BCMA, BAFF-R, o TACI as judged
by s aining o single-cell suspensions o biopsies o i e
un ea ed CD pa ien s. These da a con i med BCMA
exp ession on mos plasma cells, whe eas BAFF-R and
TACI we e exp essed only by a subg oup o cells and a
a iable le els. The e we e no signi ican di e ences in he
ela i e exp ession o hese ma ke s be ween TG2-speci ic and
non-TG2-speci ic PCs, sugges ing ha hese ecep o s play a
ole in he de elopmen o he PC compa men in he gu , bu i
does no appea ha he au oimmune B-cell esponse in CD
di e s signi ican ly om i s physiological coun e pa .
Simila ly, he exp ession le els o BCMA and BAFF-R we e
simila in CD pa ien s and heal hy con ols, whe eas TACI
appea ed o be exp essed a sligh ly highe le els in CD pa ien s,
al hough he di e ence did no each s a is ical signi icance
(P¼0.055, Mann–Whi ney; Figu e 3g).
Analysis o PCs dissemina ion in he gu mucosa by lase
cap u e and mic odissec ion
To in es iga e whe he in es inal PC popula ions expand
locally o a a dis an si e, we pe o med lase cap u e and
mic odissec ion ollowed by sequencing o immunoglobulin
hea y chain a iable (IGHV) genes on pa ches wi h such cells
om in es inal biopsies o CD pa ien s. To his end, we adap ed
a published se o p ime s o a semines ed PCR o an ibody
genes on genomic DNA. Because o he a ailable mic oscope
con igu a ion, he p ocedu e was op imal o dissec ing
sec ions s ained by immunohis ochemis y unde ansmi ed
ligh . P elimina y expe imen s indica ed ha s aining o he
CD138 ma ke was op imal unde hese condi ions, and we
used his ma ke o s ain sec ions om samples known o ha e
high equency o TG2-speci ic PCs (iden i ied using he
s anda d immuno luo escence me hod shown in Figu e 1). We
dissec ed pa ches comp ising 20 o 30 PCs (an example is
shown in Figu e 4a) om one pa ien wi h ac i e CD. In o de
o inc ease he a iabili y o he analysis, he same pa ch was
picked om h ee consecu i e slides. We ampli ied genomic
DNA and ob ained a PCR p oduc om all o he picks, and
buil and sequenced lib a ies. Impo an ly, se e al IGHV-5
sequences wi h ew o no mu a ions we e ound; IGHV-5
genes had on a e age 2.2 mu a ions (n¼9) as compa ed wi h
o he genes ha a e aged 11.4 mu a ions/sequence (n¼33).
Al hough no a o mal demons a ion, his s ongly sugges s
speci ici y o TG2 o some o he dissec ed cells; in ac , as us
Figu e 2 Quan i ica ion o TG2-speci ic plasma cells. (a) Kine ics o he disappea ance o ansglu aminase 2 (TG2)-speci ic plasma cells (PCs) a ea ly
(6 mon hs) o la e (12 and 24 mon hs) s ages o glu en- ee die (GFD), as de e mined using ou s anda d immuno luo escence s aining echnique
showed in Figu e 1a–e. Ab, an ibody; EMA, endomysial an ibody. (b) Lack o co ela ion be ween he TG2–immunoglobulin A (IgA) se um i e and he
PC coun in he biopsy, as assessed a he ime o small in es inal biopsy sampling in pa ien s wi h ac i e disease. (c) F equency o TG2-speci ic plasma
cells in h ee di e en biopsy samples om he same pa ien , as de e mined by low cy ome y using mul ime ized labeled TG2 as bai o an igen-speci ic
cells.
ARTICLES
MucosalImmunology |VOLUME 9 NUMBER 1 |JANUARY 2016 257
Figu e 3 Analysis o he exp ession o BAFF (B-cell ac i a ing ac o ) and APRIL (a p oli e a ion-inducing ligand) ecep o s on he plasma cell (PC)
popula ions in he lamina p op ia. (a– ) Visualiza ion o ansglu aminase 2 (TG2)-speci ic PCs by immuno luo escence on gu biopsy c yosec ions o
celiac disease (CD) pa ien s, wi h cha ac e iza ion o hei pheno ype and he suppo ing niche. Images a e ep esen a i e o s ainings pe o med on wo
o h ee CD pa ien s and wo heal hy con ols. (e) Quan i a i e analysis o he exp ession o BCMA (B-cell ma u a ion an igen), BAFF-R (B-cell ac i a ing
ac o ecep o ), and TACI ( ansmemb ane ac i a o and calcium modula o and cyclophilin ligand in e ac o ) by low cy ome y using monoclonal
an ibodies and mul ime ized TG2 an igen. Panels indica e pe cen age o cells exp essing BCMA, BAFF-R, and TACI, shown o TG2-speci ic and non-
TG2-speci ic IgA PCs. Box plo s wi h minimum and maximum alues a e shown. The esul s de i e om analysis o i e un ea ed CD pa ien s.
( ) Panels showing low cy ome y plo s o one o he subjec s. Recep o exp ession by TG2-speci ic PCs is shown in blue and non-TG2-speci ic PCs in
ligh ed. The g ay cu es o he le in each low diag am ep esen baseline (Fluo escence Minus One (FMO)). (g) Compa ison o he exp ession le els o
BCMA, BAFF-R, and TACI by low cy ome y o immunoglobulin A (IgA) PCs om gu biopsies o i e un ea ed CD pa ien s and i e heal hy con ols.
Pe cen age posi i e cells wi h box plo s gi ing minimum and maximum alues a e shown.
ARTICLES
258 VOLUME 9 NUMBER 1 |JANUARY 2016 |www.na u e.com/mi

and o he s ha e epo ed,
10,17
IGHV-5 genes wi h low deg ee o
soma ic hype mu a ion a e o he wise a e among he in es inal
PC compa men . In e es ingly, in some cases clonal sequences
we e ound ha showed sha ed and unique mu a ions, hus
aising he possibili y ha such clones mu a e and e ol e
locally. We u he ex ended ou analysis and compa ed
sequences ound in di e en pa ches ha we e sampled om an
addi ional pa ien wi h an ac i e disease. We dissec ed ou PC
pa ches (named A, B, C, and D) ha we e no adjacen o each
o he . The same pa ch was dissec ed om se e al consecu i e
slides. Lib a ies we e ob ained om 14 picks, yielding 53
unique sequences ep esen ing 26 di e en clones; 18 we e
ep esen ed by one single sequence and 8 by 2 o mo e and
allowed us o build lineage ees. Table 1 p o ides a summa y o
he eigh clonal ees ha we e buil wi h sequences om his
pa ien : whe eas a ew simple ees could be buil om clonal
sequences ha we e ound only wi hin one o he pa ches
(example in Figu e 4b, wi h sequences de i ed om one single
pa ch, clone ID 23 acco ding o Table 1), in many o he cases
la ge ees could be buil whe e se e al di e en pa ches
con ibu ed clones (example in Figu e 4c, wi h sequences om
h ee di e en pa ches, clone ID 22 acco ding o Table 1).
Analysis o clonal ela edness by high- h oughpu
sequencing o PCs om mul iple indi idual biopsies
This limi ed sequence analysis sugges s ha clonally ela ed
PCs dissemina e o adjacen e ec o si es in he small in es ine
ollowing di e en ia ing and ma u a ion in a dis al induc i e
si e, as was ecen ly demons a ed in he mouse.
18,19
To u he
subs an ia e his no ion, we de e mined he le el o clonal
sha ing be ween PCs ha we e isola ed om di e en in es inal
biopsies o he same CD pa ien s by high- h oughpu
sequencing analysis o hei IGHV genes. Two pa ien s we e
included and single-cell suspensions om each subjec we e
indi idually p ocessed om i e in es inal biopsies and IgA PCs
we e so ed in o TG2-speci ic and non-TG2-speci ic cells as
p e iously desc ibed
10
using he s aining s a egy shown in
Figu e 2c. Nex , he IGHV genes we e PCR ampli ied and
sequenced. Unique sequences ha o igina ed wi hin each
indi idual biopsy we e g ouped in o clones. The numbe o
so ed cells and numbe o PC clones o each biopsy a e gi en
in Supplemen a y Table S1 online. The deg ee o clonal
ela edness be ween di e en biopsies was u he de e mined
ollowing clonal clus e ing be ween he di e en biopsies.
Among non-TG2-speci ic PCs, he le el o clonal sha ing
be ween di e en biopsies anged om 10.7 o 22.7% and om
8.2 o 13.8% in pa ien s CD1191 and CD1192, espec i ely. In
compa ison, he le els o clonal sha ing among TG2-speci ic
PCs we e highe and anged om 17.9 o 73.1% and om 17.8
o 60% in he 2 pa ien s (Figu e 5a,b). Genealogy analysis
e ealed ha he e exis ed PCs in all i e biopsies ha we e
clonally ela ed and de i ed om one common ances o . A
ep esen a i e clonal ee is depic ed in Figu e 5c.
DISCUSSION
In his wo k we ha e unco e ed se e al no el aspec s o he
an i-TG2 esponse in CD. The eme ging pic u e is ha he
equency o TG2-speci ic PCs cells in he mucosa is
dynamically egula ed and ha he gene a ion o hese PCs
is aking place ou side o he lamina p op ia wi h seeding o
cells, o en clonally ela ed, in o dis inc si es o he gu mucosa.
Recen da a sugges ha PCs in he small in es ine a e longe
li ed han p e iously hough . Hap elmeie e al.
20
demon-
s a ed ha in mice in a ge m- ee en i onmen , and hus in he
Figu e 4 Lase cap u e and mic odissec ion (LCM) o plasma cell
pa ches ich in ansglu aminase 2 (TG2)-speci ic ones. Sequences a e
ob ained om PCR pe o med on genomic DNA om dissec ed pa ches in
he lamina p op ia. (a) Example o a pick comp ising B20 plasma cells. (b,
c) Examples o clonal ees buil om clonally ela ed sequences de i ed
om he same pa ch ((b) clone ID 23 om Table 1) o om h ee di e en
pa ches ((c) clone ID 22 om Table 1) o he same pa ien . Da a a e
summa ized in Table 1. The le e inside he node indica es he pa ch om
which he sequence comes, whe eas di e en colo s indica e sequences
de i ed om di e en , consecu i e c yosec ions. The numbe nex o he
line connec ing he in e ed sequence (g ay node) and ac ual obse ed
sequences (colo ed nodes) ep esen s he numbe o mu a ions by which
he wo connec ed nodes di e .
Table 1 Lase cap u e and mic odissec ion
CLONE
ID
Numbe
o unique
sequences
V-
GENE:
Numbe
o unique
mu a ions
A e age
numbe o
mu a ions
Found in
pa ch(es)
2 3 VH5-51 2 0.7 C, D
4 2 VH5-51 9 5.5 C
7 2 VH5-51 8 5 C
14 2 VH3-66 10 9.5 C
21 3 VH5-51 13 6.7 B, C
22 15 VH5-51 21 5.2 B, C, D
23 3 VH5-51 3 1.3 C
24 2 VH5-a 6 5.5 A, D
Clonal ees o immunoglobulin hea y chain a iable (IGHV) genes o he plasma cells
(PCs) buil om picks.
Clones 22 and 23 we e used o he ep esen a i e ees shown in Figu e 4b,c.
Mu a ions include bo h silen and eplacemen ones.
ARTICLES
MucosalImmunology |VOLUME 9 NUMBER 1 |JANUARY 2016 259
absence o compe i ion, in es inal IgA-sec e ing cells ha e he
capaci y o be long-li ed. By coloniza ion wi h gi en com-
mensal bac e ia, speci ic PC an ibody esponses could be aised,
bu wi hou an igen pe sis ence, he PCs disappea ed wi hin a
ew mon hs, sugges ing ha in es inal an ibody epe oi e is
cha ac e ized by cons an a i ion and hus ep esen s he
dominan an igen cu en ly p esen in he in es ine. Mesin
e al.
21
demons a ed ha PCs om human small in es ine
biopsies su i e and sec e e immunoglobulins o many weeks
in he biopsy cul u es, and in e leukin-6 and APRIL we e
iden i ied o be essen ial o plasma cell su i al and an ibody
sec e ion. TG2-speci ic PCs a e abundan in he lesion o
pa ien s wi h ac i e CD,
10
and o he main enance o his high
abundance, leng hy su i al ime appea s c ucial. We assessed
he exp ession o BAFF and APRIL ecep o s (BCMA, TACI,
and BAFF-R) on PCs. Whe eas signaling h ough TACI
induces class swi ching and an ibody p oduc ion, BAFF-R
and BCMA p omo e B-cell and plasma cell su i al, espec-
i ely.
22–27
As shown by low cy ome y quan i a ion, BCMA
was exp essed on he majo i y o PCs. In humans, bo h APRIL
and BAFF can bind BCMA, he la e wi h low a idi y. BAFF-R
and TACI we e exp essed by a subse o cells and a a iable
le els. In gene al, we did no obse e any pheno ype sugges ing
di e ences be ween TG2-speci ic and o he PCs. Al hough i is
likely ha hese ecep o s play a ole in he de elopmen o he
PC compa men in he gu , i does no appea ha he
au oimmune B-cell esponse in CD di e s signi ican ly om i s
physiological coun e pa . CD pa ien s and heal hy con ols
appea ed o ha e simila exp ession le els o BCMA and BAFF-
R, whe eas we obse ed a mino (and no s a is ically
signi ican ) inc ease in he le els o TACI in CD pa ien s.
We ound ha he numbe o TG2-speci ic PCs in he small
in es inal mucosa d ops d ama ically wi hin 6 mon hs when
glu en is wi hd awn om he die . IgA and IgM PCs o he gu
exp ess cell su ace immunoglobulin.
21
Thus concei ably,
ecogni ion o an igen by su ace immunoglobulin could allow
PCs o sense an igen and his could explain how PCs in he gu
a e main ained by an igen pe sis ence and by a i ion. Remo al
o glu en likely does no lead o emo al o TG2 an igen
a ailable o gu PCs, and hence his is ha dly he explana ion o
he obse ed d op o TG2-speci ic PCs. Possibly, howe e , i
could ela e o cessa ion o p esen a ion o an igen by PCs o T
cells ha would a ec he su i al niche o he PCs. Is has been
sugges ed ha in CD, glu en-speci ic T cells may p o ide help o
TG2-speci ic B cells ia hap en-ca ie -like complexes o glu en
and TG2.
28,29
I such glu en–TG2 complexes can be aken up ia
su ace immunoglobulino TG2-speci icPCsand hePCsp esen
glu en an igen ia sca cely exp essed human leukocy e an igen-
DQ molecules, emo al o glu en om he die would a ec T-cell
ac i a ion and he eby su i al o TG2-speci ic PCs.
Impo an ly, we obse ed ha some TG2-speci ic PCs we e
de ec able a low equency e en a la e ime poin s. One subjec
s ill had 4% o he PCs being TG2 speci ic a e 6 mon hs o
GFD. This subjec was he one wi h he highes equency o
TG2-speci ic PCs a diagnosis (18%). I would be in e es ing o
analyze ea lie and/o close ime poin s and o compa e hem
wi h se um an ibody i e s.
7
Howe e , because o he cu en
guidelines o he CD diagnos ic wo kup, such samples a e
Figu e 5 Clonal sha ing be ween plasma cells (PCs) om di e en duodenal biopsies o pa ien s wi h ac i e celiac disease. Hea maps deno ing he
le el o clonal ela edness be ween PCs ha we e collec ed om i e di e en small in es inal biopsies om wo celiac pa ien s: (a) CD1191 and (b)
CD1192. Single-cell suspensions we e p epa ed om small in es inal biopsies, immunoglobulin A (IgA) PCs we e u he so ed o ansglu aminase2
(TG2)-speci ic and non-TG2-speci ic PCs, and hei immunoglobulin hea y chain a iable (IGHV) genes we e sequenced. Clonal ela edness o PCs
om dis inc biopsies was de e mined ollowing clonal clus e ing be ween he di e en popula ions (see Me hods o de ailed desc ip ion). The le panels
depic esul s om non-TG2-speci ic plasma cells, whe eas he igh panels depic esul s om TG2-speci ic plasma cells. The numbe o PC clones in
each biopsy is shown in he diagonal ac oss each diag am. The numbe s o clones ha a e sha ed be ween biopsies a e displayed and pe cen ages a e
calcula ed as a ac ion om he o al numbe o clones om he biopsy on he ho izon al axis. (c) A ep esen a i e lineage ee ha illus a es clonal
sha ing be ween i e di e en biopsies o subjec CD1191.
ARTICLES
260 VOLUME 9 NUMBER 1 |JANUARY 2016 |www.na u e.com/mi
di icul o ob ain. No wi hs anding, i appea s clea ha many
TG2-speci ic PCs disappea apidly a e emo al o he die
and ew emain in he mucosa. In u u e s udies i will be
in e es ing o analyze in dep h di e ences in pheno ype and
ansc ip ion ac o s be ween PCs ha disappea quickly and
PCs ha emain when CD pa ien s s a he GFD.
I is well known ha TG2-speci ic an ibodies disappea om
he ci cula ion wi hin mon hs a e he commencemen o he
GFD.
7,30
I is hus su p ising ha we did no obse e a clea
co ela ion be ween educ ion o he se um i e and disap-
pea ance o he PCs a he in es inal le el in he GFD- ea ed
pa ien s. TG2-speci ic PCs we e p esen in 5 ou o 6 samples
collec ed a e 6 mon hs o GFD, despi e se onega i i y. Se e al
ac o s a e ele an o his obse a ion. Fi s , in si u de ec ion o
an i-TG2 an ibodies can p ecede appea ance in se um and
hence he wo a e no necessa ily co ela ed. Mo eo e ,
compliance o a s ic GFD, o e en inad e en glu en in ake
(glu en being inc easingly p esen in p ocessed oods), may
su ice o igge local, limi ed sec e ion ha does no ma ch
se ology esul s. Fu he mo e, we did no obse e any
co ela ion be ween he an ibody se um i e and he equency
o TG2-speci ic PCs in he lesion. I is likely ha he absolu e
amoun , a he han he equency, o TG2-speci ic PCs could
co ela e wi h se um i e . The absolu e amoun s o PCs will
depend on he deg ee o plasmacy osis, he size o he a ec ed
small in es ine, he olume o ci cula ing blood, and he deg ee
o spillo e om he in es inal compa men . We did no
analyze o hese ac o s, al hough p elimina y da a om one
pa ien showed compa able amoun s o TG2-speci ic PCs in
indi idual biopsies sampled a he same ime. Finally, i canno
be excluded ha a leas some o he TG2 an ibodies in he
blood a e p oduced a si es ou side o he gu , like lymph nodes,
spleen, and bone ma ow, o om ci cula ing plasmablas s, and
al hough his opic is a duous o add ess because o he
challenges in collec ing p ope samples, i ep esen s a scena io
ha has implica ions ha possibly go beyond CD.
P e iously, we epo ed ha TG2-speci ic PCs did no
exp ess he p oli e a ion ma ke Ki-67, sugges ing ha hei
abundance is no a di ec consequence o sus ained local
expansion.
10
Howe e , he Ki-67 ma ke accumula es upon
sus ained cell di ision, and i may no be exp essed i cells
unde go cell p oli e a ion a a slow, homeos a ic a e. I should
also be no ed ha immuno luo escence may no be sensi i e
enough o his ype o analysis, and ha low cy ome y
app oaches could p o ide be e eadou s. Cell di ision is also
necessa y o accumula ion o mu a ions in an ibody genes.
Al hough no commonly acknowledged, he e is e y clea
e idence o mu a ions occu ing ou side o ge minal cen e s, a
leas in some con ex s in mice.
31
In addi ion, i has been
sugges ed ha p ecu so s o in es inal an ibody- o ming cells
may unde go soma ic hype mu a ion and/o local expansion
di ec ly in he lamina p op ia,
32
al hough e idence emains o
be subs an ia ed. To de e mine whe he in es inal PCs p o-
li e a e locally o a adis an si e we pe o med lase cap u e and
mic odissec ion o TG2-speci ic plasma cell pa ches om wo
pa ien s ollowed by an ibody gene sequencing. We ob ained
and hen compa ed sequences ob ained om se e al di e en
pa ches; mo eo e , he same pa ch was dissec ed, when
possible, om consecu i e slides, hus c ea ing a idimensional
g id o he pa ch and inc easing he a iabili y o he analysis.
The a ionale o he analysis o his expe imen is ha inding
clonal sequences ha di e by one o ew mu a ions wi hin he
same pa ch, and no in pa ches om dis an a eas, would be a
good indica o ha he mu a ions occu ed locally; i his we e
no he case, clonal sequences would no be con ined exclusi ely
o local a eas. Some o he clonal ees we e gene a ed om
sequences ob ained exclusi ely om one pa ch, hus possibly
sugges ing some deg ee o local p oli e a ion and mu a ion.
Howe e , he e we e la ge ees wi h sequences om se e al
di e en pa ches, sugges ing ha hese clones acqui ed
mu a ions and likely expanded a a di e en si e be o e
dissemina ion in he mucosa. To u he subs an ia e his
no ion, we unde ook analysis o he an ibody epe oi e among
TG2-speci ic and non-TG2-speci ic PCs om i e indi idual
biopsies o wo pa ien s wi h un ea ed CD. We indeed
obse ed clonal sha ing among cells o he di e en biopsies,
pa icula ly among he TG2-speci ic PCs whe e he deg ee o
sha ing anged om 20% o 60–70%. These indings sugges
ha he PCs o he duodenum clonally expand be o e seeding
in o he mucosa, al hough he ac ha small ees could be buil
exclusi ely om sequences ound in one pa ch indica es ha
he possibili y o local expansion and mu a ion, a leas o some
ex en , canno be comple ely uled ou . In gene al, ou indings
a e consis en wi h ecen obse a ions in mouse models.
18,19
We in e p e he ex ensi e deg ee o clonal sha ing be ween
TG2-speci ic PCs o indi idual biopsies likely o e lec an
expansion o dominan clones a induc i e si es ha is ollowed
by seeding o PCs in o la ge a eas o he small in es ine.
Al hough an i-TG2 an ibody deposi s a e usually isualized
by he classical CUB7402/IgA s aining,
11
we ound ha deposi s
can be obse ed using he an igen as bai . By s aining wi h
bio in-TG2 in combina ion wi h an an i-IgA (Figu e 1 ), we
con i med he p esence o subepi helial deposi s, bu also
de ec ed TG2-speci ic an ibodies inside epi helial cells and on
he b ush bo de . In ac , mos IgA p oduced by in es inal PCs
a e dime ic, wi h wo molecules bound oge he by he J-chain
polypep ide. The J-chain in e ac s wi h he polyme ic Ig
ecep o a he basal side o epi helial cells, esul ing in anspo
ac oss he epi helium and in o he lumen. I is concei able ha
an i-TG2 an ibodies bind he an igen in he subepi helial laye
and a e co- anspo ed oge he wi h i ac oss he epi helium;
his is u he sugges ed by he p esence o TG2 i sel
(colocalizing wi h IgA) in he epi helium and on he b ush
bo de , as shown in Figu e 1g and as sugges ed by igu es om
a numbe o o he publica ions.
12,33–35
This allows us o
specula e on a model ha builds upon p e ious obse a ions
om Ma ysiak-Budnik e al.,
36
who sugges ed ha an i-gliadin
IgA an ibodies a e expo ed in he gu lumen, whe e hey bind
glu en and anspo i ac oss he epi helium back o he lamina
p op ia, ia a e o anscy osis mechanism media ed by he
ans e in ecep o (CD71). In addi ion o an i-gliadin IgA, i
may also be ha TG2 a he b ush bo de can sample glu en
ARTICLES
MucosalImmunology |VOLUME 9 NUMBER 1 |JANUARY 2016 261
pep ides by enzyma ically ac ing on hem, and hese a e
e o anscy osed wi h an i-TG2 IgA (o IgM
37
). TG2 e ains i s
enzyma ic ac i i y when bound o an ibodies.
10
In addi ion, i
has been shown ha he majo i y o pep ides ep esen ing
p e e ed subs a e o TG2 ha bo known glu en T-cell
epi opes,
38
and ha epi opes equen ly ecognized by pa ien
T cells showed a signi ican highe le el o deamida ion
compa ed wi h hose less equen ly ecognized.
39
O he
g oups ha e p oposed mechanisms ha could media e his
o simila anspo sys ems; Rauha i a e al.
40
also implica ed
a ole o celiac IgA ha necessi a es TG2 enzyma ic ac i i y,
and Leb e on e al.
41
subs an ia ed some o hese inding,
u he ein oducing CD71 in he game o play and also
sugges ing a ole o he enzyma ic ac i i y o TG2. These a e
a ac i e models ha equi e u he s udy, including he
de elopmen o an objec i e quan i a i e measu emen sys em
o an ibody deposi s in ela ion o i s signi icance in CD.
In summa y, he indings desc ibed he e shed new ligh on
he cha ac e is ics o he an i-TG2 B-cell esponse in he
in es inal mucosa o pa ien s a ec ed by celiac disease.
Impo an ly, he gene al app oach can be applied o simila
condi ions whe e he ole and especially he o igin o
(au o)an ibodies is unclea . Fo ins ance, de ma i is he pe i-
o mis and glu en a axia ( he skin and cen al ne ous sys em
mani es a ions o glu en sensi i i y, espec i ely, occu ing in
up o 10% o CD pa ien s) a e cha ac e ized by he local
p esence o TG3-IgA and TG6-IgA deposi s.
42,43
Rele an
diseases whe e local p esence o au oan ibodies is hough o be
impo an bu is sca cely unde s ood include mul iple scle osis
(ne ous sys em), heuma oid a h i is (join s), and o he s. The
app oaches desc ibed he e may help o u he ou knowledge
o hese daun ing diso de s.
METHODS
Pa ien ma e ial. Adul CD pa ien s we e diagnosed acco ding o
s anda d c i e ia, including human leukocy e an igen geno yping,
an i-TG2 se um i e , and his ological analysis o small in es inal
biopsy.
44
The same c i e ia we e used o exclude CD diagnosis in
con ols. E hical app o al o s udy o No wegian subjec s was
ob ained om he Regional E hics Commi ee in Sou h-Eas e n
No way (p ojec S-97201). Each s udy subjec ga e w i en in o med
consen . Duodenal biopsy specimens we e ob ained by gas o-
duodenoscopy. Tissue sec ions o biopsies o se en CD pa ien s wi h
ac i e disease, om h ee CD pa ien s ea ed wi h a GFD and wo
heal hy con ol subjec s we e s udied. In addi ion, biopsies om i e
un ea ed CD pa ien s and i e heal hy con ols we e p ocessed and
used o low cy ome y analysis and PCs o biopsies om wo
un ea ed CD pa ien s we e p ocessed o high- h oughpu
sequencing. E hical app o al o s udy o he Finnish subjec s was
ob ained om he E hics Commi ee o Tampe e Uni e si y Hospi al,
and each subjec p o ided w i en in o med consen . Samples om 15
CD pa ien s and 4 nonceliac con ols su e ing om dyspepsia we e
s udied. F om hese subjec s small-bowel mucosal biopsies we e aken
ei he wi h an adul -size Wa son capsule om he p oximal jejunum o
upon endoscopy wi h o ceps om he dis al duodenum. Fo each
subjec , pa o he samples we e snap- ozen and embedded in op imal
cu ing empe a u e compound (OCT, Tissue-Tec, Miles, Elkha , IN)
o s o ing a 70 1C, whe eas he es o he biopsy specimens we e
pa a in embedded. Se um IgA class an ibodies agains TG2 we e
de ec ed by enzyme-linked immunoso ben assay using human
ecombinan TG2 as an igen, wi h a cu o line o 5.0 U ml
1
(Celikey,
Phadia, F eibu g, Ge many).
Immuno luo escence on in es inal biopsy c yosec ions. Immu-
no luo escence expe imen s we e done as epo ed be o e.
10
The
echnique is based on he abili y o bio inyla ed TG2, ollowed by
labeled s ep a idin, o selec i ely bind TG2-speci ic an ibody deposi s
inside plasma cells. B ie ly, biopsy specimens s o ed a 70 1C we e
hawed and insed wi h phospha e-bu e ed saline (PBS). P ima y
eagen s we e incuba ed in 1.2% bo ine se um albumin in PBS o
45 min a oom empe a u e (RT). Sec ions we e ex ensi ely insed
wi h PBS, and seconda y eagen s we e incuba ed in 12% bo ine se um
albumin in PBS o 30 min a RT. Sec ions we e insed wi h PBS and
moun ed. The ollowing an ibodies and eagen s we e used: mouse
an i-CD138 (1:25, AbD Se o ec, Kidling on, UK), ecombinan human
TG2 p oduced in Esche ichia coli and bio inyla ed as epo ed be o e
10
(5 mgml
1
), mouse an i-TG2 monoclonal an ibody CUB7402
(1:1,000, Neoma ke s, F emon , CA), an i-TACI (1:200, sc-7332
N-19), an i-BCMA (1:200, sc-11743N-16), and an i-BAFF-R (1:150,
eBioscience 8A7, San Diego, CA) p ima y an ibodies. Seconda y
an ibodies we e used as ollows: Alexa Fluo 546/488/649-conjuga ed
an i-goa , an i-mouse, and an i- abbi polyclonal an ibodies (1:1,000,
Li e Technologies, Ca lsbad, CA), s ep a idin-cy2 (1:1,000, GE
Heal hca e Biosciences, Pi sbu gh, PA), s ep a idin-cy3 (1:2,000, GE
Heal hca e Biosciences), an i-mouse-cy3 (1:1,500, Sou he n Bio ech,
Bi mingham, AL), and an i-IgA luo escein iso hiocyana e (1:40,
Dako). DAPI (4’,6-diamidino-2-phenylindole) was used o coun-
e s ain nuclei.
Flowcy ome y. The equency o TG2-speci ic PCs in small in es inal
biopsies was de e mined using low cy ome y as p e iously epo -
ed.
10
The exp ession le els o BCMA, BAFF-R, and TACI on in es inal
PCs ha ei he s ained posi i e o nega i e o TG2 we e also examined
using low cy ome y o single-cell suspensions o biopsies wi h Bi -A
bio inyla ed TG2 mul ime ized on luo escen ly labeled s ep ame s.
In sho , ecombinan human TG2 was exp essed in S 9 insec cells by
baculo i us wi h a Bi -A ecogni ion si e a i s N- e minus. TG2 was
bio inyla ed wi h Bi -A enzyme (A idi y, Au o a, CO), and incuba ed
wi h allophycocyanin-labeled s ep ame s (IBA, Go ingen, Ge many)
in a 4:1 mola a io o 1 h o gene a e TG2-mul ime s. C yop ese ed
single-cell suspensions ha we e p epa ed om gu biopsies we e
hawed and s ained wi h allophycocyanin-labeled TG2-mul ime o
45 min on ice. IgA luo escein iso hiocyana e (Sou he n Bio ech),
B illian Viole 570 CD3, CD14, and CD38 Pe Cp-Cy5.5 (BioLegend,
San Diego, CA), CD19 Paci ic Blue (BD Biosciences), CD27 PE-Cy7
(eBioscience), and ei he BCMA, BAFF-R, o TACI (all phycoe y h in
(PE) conjuga ed, BioLegend) we e added o he las 20 min o
s aining o speci ic cellula lineage ma ke s. FMO con ol (i.e.,
Fluo escence Minus One) was used o p ope ly in e p e he
exp ession o BCMA, BAFF-R, and TACI by s aining single-cell
suspensions o each subjec wi h a combina ion o all an ibodies and
TG2-mul ime s excep he PE-conjuga ed an igens. Samples we e
acqui ed on LSRFo essa (BD Biosciences, San Jose, CA) and he da a
we e analyzed by FlowJo, e sion 10.0.7 (Ashland, OR).
Immunohis ochemis y and lase cap u e and mic odissec ion. Fo
lase cap u e and mic odissec ion, 7 mm hick c yosec ions om
specimens o pa ien s we e cu on 2 mm PEN-memb ane slides (Leica,
Bu alo G o e, IL). Sec ions we e ixed in ace one o 10 min and s o ed
a 80 1C. Fo lase cap u e and mic odissec ion, a Leica LMD6500
ins umen equipped wi h an op ical mic oscope was used, and
sec ions we e he e o e s ained by immunohis ochemis y. B ie ly,
slides we e hawed and insed in PBS. P ima y eagen s (mouse an i-
CD138 o ecombinan human bio inyla ed TG2 p oduced in E. coli,as
epo ed abo e) we e dilu ed in 1% bo ine se um albumin in PBS and
added o he slide o 45 min a RT in a mois chambe . Slides we e
ex ensi ely insed and seconda y eagen s (goa an i-mouse IgG-HRP,
1:500, San a C uz (Dallas, TX), o SA-HRP, 1:500, Sou he n Bio ech)
ARTICLES
262 VOLUME 9 NUMBER 1 |JANUARY 2016 |www.na u e.com/mi