1355
Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
O iginal esea ch
Ac i a ed egula o y T- cells p omo e duodenal
bac e ial ansloca ion in o nec o ic a eas in se e e
acu epanc ea i is
Juliane Glaubi z,1 Anika Wilden,1 Fabian F os ,1 Sabine Ameling,2 Geo g Homu h,2
Hala Mazloum,1 Mal e Ch is oph Rühlemann ,3,4 Co inna Bang ,3
Ali A Aghdassi ,1 Ch is oph Budde,1 Tilmann Picka z,1 And e F anke ,3
Ba ba a M B öke ,5 Uwe Voelke ,2 Julia Maye le ,6 Ma kus M Le ch,1
F ank- Ul ich Weiss,1 Ma hias Sendle 1
To ci e: Glaubi zJ, WildenA,
F os F, e al. Gu
2023;72:1355–1369.
►Addi ional supplemen al
ma e ial is published online
only. To iew, please isi he
jou nal online (h p:// dx. doi. o g/
10. 1136/ gu jnl- 2022- 327448).
Fo numbe ed a ilia ions see
end o a icle.
Co espondence o
D Ma hias Sendle ,
Depa men o Medicine A,
Uni e si y Medicine G ei swald,
G ei swald 17475, Ge many;
ma hias. sendle @ uni-
g ei swald. de
F- UW and MS a e las au ho s.
Recei ed 25 Ma ch 2022
Accep ed 30 Decembe 2022
Published Online Fi s
11Janua y2023
© Au ho (s) (o hei
employe (s)) 2023. Re- use
pe mi ed unde CC BY- NC. No
comme cial e- use. See igh s
and pe missions. Published
by BMJ.
ABSTRACT
Objec i e In acu e panc ea i is (AP), bac e ial
ansloca ion and subsequen in ec ion o panc ea ic
nec osis a e he main isk ac o s o se e e disease and
la e dea h. Unde s anding how immunological hos
de ence mechanisms ail o p o ec he in es inal ba ie
is o g ea impo ance in educing he mo ali y isk o
he disease. He e, we s udied he ole o he T eg/Th17
balance o main aining he in es inal ba ie unc ion in
a mouse model o se e e AP.
Design AP was induced by pa ial duc liga ion in
C57Bl/6 o DEREG mice, in which egula o y T- cells (T eg)
we e deple ed by in ape i oneal injec ion o diph he ia
oxin. By low cy ome y, unc ional supp ession assays
and ansc ip ional p o iling we analysed T eg ac i a ion
and cha ac e ised T- cells o he lamina p op ia as well as
in aepi helial lymphocy es (IELs) ega ding hei ac i a ion
and di e en ia ion. Mic obio a composi ion was examined
in in es inal samples as well as in mu ine and human
panc ea ic nec osis by 16S RNA gene sequencing.
Resul s The p ophylac ic T eg-
deple ion enhanced he
p oin lamma o y esponse in an expe imen al mouse
model o AP bu s abilised he in es inal immunological
ba ie unc ion o Th17 cells and CD8+/γδTCR+ IELs. T eg
deple ed animals de eloped less bac e ial ansloca ion
o he panc eas. Duodenal o e g ow h o he acul a i e
pa hogenic axa Esche ichia/Shigella which associa es
wi h se e e disease and in ec ed nec osis was diminished
in T eg deple ed animals.
Conclusion T egs play a c ucial ole in he
coun e balance agains sys emic in lamma o y esponse
synd ome. In AP, T eg- ac i a ion dis u bs he duodenal
ba ie unc ion and pe mi s ansloca ion o commensal
bac e ia in o panc ea ic nec osis. Ta ge ing T egs in AP may
help o amelio a e he disease cou se.
Regula o y T cells could ep esen a he apeu ic
a ge o he p e en ion o in ec ed nec osis du ing
se e e acu e panc ea i is, which will imp o e he
disease cou se and ou come.
INTRODUCTION
Acu e panc ea i is (AP) is he mos common non-
malignan disease o he gas oin es inal (GI) ac
leading o hospi al admission wi h an inc easing
incidence o e he pas yea s in wes e n coun ies.1
Eigh y pe cen o panc ea i is cases ollow a sel -
limi ing cou se o disease wi hou complica ions o
long- ime hospi alisa ion. Abou 20% o pa ien s
de elop se e e AP (SAP), which is associa ed wi h
sys emic complica ions and inc eased mo bidi y
and mo ali y.2 In ec ed panc ea ic nec osis, which
is belie ed o a ise om ansloca ion o commensal
gu bac e ia in o he panc eas,3 is ega ded o d i e
pe sis en o gan ailu e and subsequen mo ali y.2
Ini ially bac e ia need o o e come he in es inal
ba ie and e ade he immune sys em o cause
in ec ed nec osis. Especially he T eg/Th17- balance
is c ucial o main aining he in es inal ba ie unc-
ion and egula ing issue homoeos asis.4
WHAT IS ALREADY KNOWN ON THIS TOPIC
⇒In ec ed nec osis is a se e e complica ion
du ing acu e panc ea i is (AP) and is associa ed
wi h a signi ican mo ali y.
WHAT THIS STUDY ADDS
⇒In an animal model, he expe imen al induc ion
o AP associa es wi h signi ican changes o he
in es inal mic obio a composi ion and a ma ked
inc ease o acul a i e pa hogenic bac e ia. The
same bac e ial axa a e iden i ied in nec o ic
issue samples o panc ea i is pa ien s.
⇒Du ing AP, ac i a ed egula o y T cells supp ess
he sys emic immune esponse and impai he
immunological in es inal ba ie unc ion.
⇒The deple ion o T egs du ing AP educed he
bac e ial ansloca ion o acul a i e pa hogenic
s ains in o he in lamed panc eas.
HOW THIS STUDY MIGHT AFFECT RESEARCH,
PRACTICE OR POLICY
⇒Regula o y T cells could ep esen a he apeu ic
a ge o he p e en ion o in ec ed nec osis
du ing se e e AP which will imp o e he
disease cou se and ou come.
1356 Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
Panc ea i is is a p ima ily s e ile in lamma ion o igina ing om
p ema u e in acellula p o ease ac i a ion. Ac i a ed zymogens
igge acina cell dea h5 and ac i a e he immune egula ing
ansc ip ion ac o NFκB.6 This in lamma o y p ocess ec ui s
mac ophages and neu ophils o he si e o in lamma ion which
u he inc eases he local damage.7–10 Mac ophages ep e-
sen he majo i y o esiden panc ea ic immune cells and hei
ac i a ion by damage- associa ed molecula pa e ns (DAMPs)
ini ia es a sys emic immune- esponse,8 11–14 including he ac i-
a ion o he adap i e immune sys em h ough he elease o
cy okines.12 Du ing an episode o AP esiden lympha ic T- cells
ini ia e a supp essi e and an i- in lamma o y esponse media ed
by an inc ease in FOXP3+/CD25+ egula o y T- cells (T eg) and
GATA3+ Th2- cells.12 15 This pa allel ini ia ion o a sys emic
in lamma o y esponse synd ome (SIRS) and a compensa o y
an i- in lamma o y esponse synd ome (CARS) is also known
om sepsis o c i ical auma.16 17
Sys emic immunosupp ession (CARS), in pa icula media ed
by T egs, can p omo e seconda y in ec ions18 and hus agg a a e
disease se e i y o cause non- panc ea ic complica ions. To da e,
li le is known abou he ela ionship be ween he gu mic o-
biome and he cou se o AP.19 He e, we in es iga ed in an animal
model o AP how T eg- media ed immunosupp ession a ec s he
gu mic obiome composi ion and he ansloca ion o bac e ia
in o he in lamed panc eas.
RESULTS
AP is associa ed wi h changes o he in es inal mic obio a
composi ion
AP was induced by pa ial duc liga ion in C57Bl/6- J mice.
Duodenal aspi a es and aecal samples om dis al colon and
caecum we e collec ed om AP mice and un ea ed con ol
animals. 16S RNA gene sequencing e ealed mic obio a
di e ences be ween con ol and AP mice ( igu e 1A). P incipal
coo dina e analysis (PCoA) o all GI- ac samples con i med
mic obio a changes, wi h he mos p ominen changes seen
in duodenal aspi a es ( igu e 1B, online supplemen al igu e
S1A,B). A signi ican ly educed Shannon- di e si y and amplicon
sequence a ian (ASV) ichness was obse ed in he duodenal
aspi a es o AP animals. Smalle changes o he mic obio a
composi ion we e obse ed in caecal bu no in dis al colonic
samples ( igu e 1C,D). Analysis o he mos abundan axonomic
uni s iden i ied an inc ease o acul a i e pa hogenic bac e ia like
Esche ichia/Shigella, En e obac e iaceae di e sa, En e ococcus o
S aphylococcus. Especially in duodenal samples all hese axa
we e signi ican ly en iched du ing AP. In colon and caecum
samples, we could also obse e an inc ease o hese acul a-
i e pa hogenic bac e ia, bu only Esche ichia/Shigella showed
a signi ican inc ease in all pa s o he GI- ac ( igu e 1E–G).
Thei high abundance in he duodenal sec ion sugges s bac e ial
o e g ow h (online supplemen al igu e S1C). In con as , he
abundance o bene icial commensal bac e ia such as Lachnospi -
aceae was dec eased, and his e ec was again mos p onounced
in he duodenum ( igu e 1E–G). To e i y he esul s om he
model o AP using pa ial panc ea ic duc liga ion we in es iga ed
he mic obiome changes in duodenum and colon samples in a
second panc ea i is model. To analyse compa able ime poin s
we induced panc ea i is ia 8- hou ly in ape i oneal (i.p.) injec-
ions o Cae ulein (50 µg/kg/bodyweigh ) o e 3 days. Again, we
saw signi ican changes in he in es inal mic obiome composi-
ion wi h he s onges e ec s in he duodenum (online supple-
men al igu e S2A–C). The disease se e i y in his model was
Figu e 1 Acu e panc ea i is (AP) is associa ed wi h changes o he in es inal mic obio a composi ion. (A)Faecal samples om colon, caecum and
duodenum we e collec ed om C57Bl/6 mice wi h AP (n=24) and om un ea ed con ol animals Con (n=10). Isola ed DNA was analysed by 16S
RNA gene sequencing. The mic obio a composi ion and AP- associa ed changes o he axonomic uni s wi h he highes abundance a e illus a ed
by a s acked ba g aph. (B)P incipal coo dina e analysis illus a es he changes o gu mic obiome be ween un ea ed con ols and AP mice. (C,
D)Shannon- Di e si y Index (C)and ichness o obse ed species (D)demons a e signi ican impac o AP on he duodenal mic obiome composi ion.
(E–G)The ba g aph illus a es changes o he mos abundan axa in colonic (E),caecal (F)and duodenal samples (G)o AP ( ed) and con ol mice
(g ey). Facul a i e pa hogenic we e ma ked in ed, bene icial commensal bac e ia we e ma ked in g een. S a is ical signi icance was de e mined by
unpai ed S uden ’s - es and K uskal- Wallis es ollowed by a Dunn’s mul iple compa isons es o analyse di e en ially abundan axa in colon,
caecum and duodenum samples. Signi icance le els o p<0.05 a e ma ked by an as e isk.
1357
Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
signi ican ly lowe compa ed wi h he model induced by pa ial
panc ea ic duc liga ion (online supplemen al igu e S2D,E) and
we obse ed less p onounced changes o he in es inal mic obio a
composi ion (online supplemen al igu e S2A–C) and nea ly no
changes o he sys emic immune esponse (online supplemen al
igu e S2F–H).
In es inal mic obio a changes co ela e wi h he se e i y o
AP and he sys emic immune esponse
In he nex s ep, we in es iga ed how a he duodenal mic o-
biome changes a e in luenced by disease se e i y and h ough
he sys emic immune esponse by analysing splenic lymphocy es.
Unexpec edly LY6G+/LY6Clow/CD11b+- cells in spleen o AP
mice inc eased signi ican ly du ing AP bu he concomi an
dec ease o se um amylase as ma ke o panc ea ic damage
sugges ed an in e se co ela ion o disease se e i y ( igu e 2A).
Splenic LY6G+/LY6Clow/CD11b+- cells showed he same in e se
co ela ion wi h se um lipase ac i i y, his ological damage and
he se um cy okine le el o IL- 6 and TNFα (online supplemen al
igu e S3A–C). Fu he mo e, we de ec ed a posi i e co ela ion
o splenic LY6G+/LY6Clow/CD11b+- cells wi h he spleen weigh ,
and a nega i e co ela ion wi h he numbe o splenic CD25+/
FOXP3+ T egs which a e signi ican ly ele a ed du ing panc ea i is
(online supplemen al igu e S3D,E). Ele a ed apop osis in spleen
Figu e 2 In es inal mic obio a changes co ela e wi h he se e i y o AP and he sys emic immune esponse. (A)Do plo illus a es nega i e
co ela ion o LY6G+/LY6Clow/CD11b+- cells o he ac i i y o se um amylase o mice wi h AP (n=50), signi icance o co ela ion was es ed by
Spea man’s ank co ela ion- coe icien . (B)Disease se e i y in animals was classi ied acco ding o he pe cen age o LY6G+/LY6Clow/CD11b+- cells
in spleen (n=5/g oup). (C)Box plo s illus a e he duodenal mic obiome species ichness and Shannon- di e si y acco ding o mode a e and se e e
panc ea i is. (D)P incipal coo dina e analysis illus a es signi ican di e ences o duodenal mic obiome be ween hese g oups (pe mu a ional
mul i a ia e analysis o a iance; con ol s se e e: p=0.007, R2=59.4%, con ol s mode a e: p=0.015, R2=24.5% and mode a e s se e e: p=0.007,
R2=48.4%). (E)Hea - map illus a es 16S RNA gene sequencing esul s o he duodenal mic obiome analysis, acul a i e pa hogenic we e ma ked
in ed, bene icial commensal bac e ia we e ma ked in g een. (F)H&E- s aining o panc ea ic issue o heal hy and AP mice. (G)Colony o ming uni s
(CFU) we e coun ed om nec o ic issue homogena es and exceeded he cu - o le el o 1 CFU/mg issue weigh in 5 o 21 (23.8%) animals. (I)F om
in ec ed (n=5) and non- in ec ed (n=6) g oups, we analysed in spleen he pe cen age o LY6G+/LY6Clow/CD11b+- cells and (H) he pe cen age o
FOXP3+/CD25+/CD4+- cells. S a is ical e alua ion was done by unpai ed S uden ’s - es o independen samples and signi icance le els o p<0.05 a e
ma ked by an as e isk. AP, acu e panc ea i is.
1358 Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
could be esponsible o he loss o LY6G+/LY6Clow/CD11b+-
cells (online supplemen al igu e S3F). Based on hese indings,
we dicho omised he se o animals in a splenic LY6G+/LY6Clow/
CD11b+ high (mode a e) o low cell (se e e) coho and in es-
iga ed se e i y dependen changes o he duodenal mic obio a
composi ion ( igu e 2B). 16S RNA gene sequencing analysis
con i med a se e i y- dependen loss o ichness and di e si y
( igu e 2C). PCoA o duodenal aspi a es om mode a e and SAP
showed se e i y- dependen signi ican changes o he mic o-
bial composi ion ( igu e 2D). A de ailed analysis o he mos
equen duodenal mic obio a axa is illus a ed in he hea map
( igu e 2E and online supplemen al able 1). Facul a i e pa ho-
genic bac e ial clades like Esche ichia/Shigella, En e ococcus o
S aphylococcus we e again de ec ed wi h inc eased abundance
(online supplemen al igu e S4A).
In ec ed nec osis is a li e- h ea ening complica ion o AP
and ega ded o o igina e om ansloca ion o commensal
gu bac e ia. In ou mouse model o AP we examined, bac e-
ial in ec ion o he liga ed nec o ic pa o he panc eas by
colony g ow h assays o he issue homogena e on aga pla es
( igu e 2F–G). In 5 o 21 animals, we de ec ed a signi ican bac e-
ial load in panc ea ic nec osis wi h mo e han 1 colony- o ming
uni (CFU)/mg panc eas, which is compa able o in ec ion a es
seen in pa ien s ma ched o se e i y.2 A nega i e co ela ion
was de ec ed o he a e o in ec ion o panc ea ic nec osis and
numbe o splenic LY6G+/LY6Clow/CD11b+- cells ( igu e 2I and
online supplemen al igu e S4B). To es a unc ional ela ion-
ship be ween LY6G+/LY6Clow/CD11b+- cells and he in ec ion
o panc ea ic nec osis we deple ed LY6G+- cells by an i- LY6G
an ibody ea men be o e induc ion o panc ea i is. T ea men
wi h an i- LY6G abolished LY6G+/LY6Clow/CD11b+- cells (online
supplemen al igu e S4A–D) and did no a ec bac e ial ans-
loca ion in o he nec osis (S.4E). Ou inding sugges ed ha
educ ion o LY6G+/LY6Clow/CD11b+- cells in spleen can se e
as ma ke o disc imina e be ween mode a e and SAP in mice,
bu appa en ly has no unc ional impac .
Since he deple ion o LY6G+/LY6Clow/CD11b+- cells did no
a ec he adap i e immune cells (online supplemen al igu e S4F)
we in es iga ed in u he expe imen s he cells o he adap i e
immune esponse he spleen. Labelling o he IL- 2 ecep o -α-
chain (CD25) on he su ace o CD4+ T- cells (online supple-
men al igu e S5A) e ealed a signi ican ly s onge inc ease in
he numbe o ac i a ed CD25+ T- cells in hose animals wi h
an in ec ed panc ea ic nec osis, indica ed by a bac e ial load o
mo e han 1 CFU/mg panc ea ic issue. CD69, a second ac i-
a ion ma ke o CD4+ T- cells, showed he same associa ion
(online supplemen al igu e S5B). We nex in es iga ed T- cell
di e en ia ion by labelling o he ansc ip ion ac o s TBET,
GATA3 and FOXP3. Whe eas Th1- cell di e en ia ion ansc ip-
ion ac o TBET was no a ec ed by panc ea i is (online supple-
men al igu e S5C), GATA3+ Th2- cell numbe s we e ound
ele a ed in he spleen, independen o bac e ial in ec ion (online
supplemen al igu e S5D). In e es ingly, he numbe o FOXP3+/
CD25+ T egs was signi ican ly inc eased in animals su e ing om
panc ea i is and was highes in hose wi h in ec ed nec osis
( igu e 2H). Fu he mo e, we de ec ed a di ec co ela ion
be ween T eg a es wi h he bac e ial load (online supplemen al
igu e S5E).
Imbalance o he Th17/T eg a io in duodenal mucosa du ing
AP
To colonise panc ea ic nec osis commensal bac e ia need o o e -
come he in es inal ba ie . Duodenum his ology o AP mice did
no show ob ious di e ences ( igu e 3A). Immuno luo escen
labelling o F4/80 o CD68 demons a ed a signi ican educ ion
o mac ophages wi hin he lamina p op ia, also sIgA p oducing
plasma cells we e signi ican ly educed ( igu e 3B,C). The
numbe o T- cells was also a ec ed by AP, CD3+ T- cells wi hin
he lamina p op ia as well as CD8α+ in aepi helial lymphocy es
(IELs) we e signi ican ly educed ( igu e 3D,E). Addi ionally,
we analysed global mRNA- p o iles o mouse duodenal issues
(AP s con ol) using A yme ix GeneChip- a ays. Analysis
o he gene a ed ansc ip ome da a by QIAGEN’s Ingenui y-
Pa hway- Analysis (IPA) so wa e iden i ied AP- speci ic di e en-
ial gene exp ession in he T- cell- esponse, B- cell- esponse and
o he immune sys em- ela ed pa hways ( igu e 3F). T ansc ip
le els o genes encoding T- cell su ace ma ke s CD3, CD8α,
CD4 and CD28 we e signi ican ly educed, while SOCS3
(supp esso o cy okine signalling 3) speci ic mRNA was signi i-
can ly mo e abundan ( igu e 3G). B- cell- esponse ma ke s
like CD79A, TNFS13/p oli e a ion- inducing ligand APRIL,
TNFSF13B/B- cell ac i a ing ac o BAFF, o JCHAIN (pa o
Immunglobulin A) exhibi ed signi ican ly lowe ansc ip le els
in AP mice ( igu e 3G). In con as o an appa en ly a enu-
a ed immune esponse, he mRNA abundance o an i- mic obial
de ence pep ides such as DEFA1, REG3A, REG3G, DEFA23, o
DEFA21 was signi ican ly highe in he duodenum o AP mice
( igu e 3G). The exp ession o genes encoding s uc u al p o ein
componen s o he in es inal ba ie like igh junc ions, desmo-
somes o adhe ence junc ions we e only in some cases a ec ed
by AP (online supplemen al igu e S6A). We could obse e a
signi ican down egula ion o mRNAs encoding o Claudin15
(Cldn15), he Myosin ligh chain kinase (Mylk), cadhe in ela ed
amily membe 2 (Cdh 2) o Pannexin 1 (Panx1), whe eas o he
mRNAs showed a signi ican up egula ion a e onse o disease
such as Dsp, encoding o Desmoplakin, Cdh5 encoding o
he p o ein Cadhe in 5 o he mRNAs encoding o mucins 2,
3 and 3a (Muc2, Muc3 and Muc3a). C yp - cell p oli e a ion,
ma ked by KI67, was no a ec ed by AP (online supplemen al
igu e S6B). Mos s uc u al p o eins we e una ec ed by AP
like E- Cadhe in (online supplemen al igu e S6C). On he o he
hand, AP exhibi ed signi ican e ec s also on se e al pa hways o
he immune homoeos asis (cy okines, ansmemb ane- ecep o s
and ansc ip ional- egula o s) in he duodenal sys em (online
supplemen al igu e S6D–F).
To e alua e he e ec o disease se e i y on he in es inal
immune esponse, we isola ed and analysed lymphocy es o
he lamina p op ia and duodenal IELs by low cy ome y
wi h espec o disease se e i y ( igu e 3H–L). The pu i y o
bo h cell ypes was con i med de ec ing he CD8α/CD4 a io
(S.7A- B). We obse ed a se e i y- dependen dec ease o CD4+
T cells in he lamina p op ia (online supplemen al igu e
S7C). Wi hin his popula ion o CD4+ T- cells we measu ed an
inc ease o CD25+/FOXP3+ T egs du ing panc ea i is, wi h he
highes numbe s obse ed in SAP ( igu e 3H). The same esul
was ob ained o he popula ion o IELs, bu a a signi ican ly
lowe a io ( igu e 3I). Wi hin he duodenum we obse ed a
se e i y- dependen inc ease in he numbe o CD25+/FOXP3+
T egs ( igu e 3I,J). Fu he mo e, e ec o T- cell (Te ) numbe s
we e signi ican ly ele a ed du ing panc ea i is, he induc ion
o CD25+/RORγ + (Th17) IELs and CD25+/TBET+ (Th1) was
no di e en be ween mode a e o SAP. The a io o CD25+/
RORγ + inc eased o nea ly 10%, whe eas CD25+/TBET+ did
no ise abo e 1% o o al IELs ( igu e 3K,L). Analysis o GFP-
exp essing T egs in DEREG- mice (DEple ion o REGula o y
T- cells)20 e ealed a e induc ion o AP a signi ican edis-
ibu ion o GFP+/CD3+ T- cells wi hin he lamina p op ia,
1359
Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
con i ming he cy ome y analysis ( igu e 3M). No ably he Te /
T eg- a io was signi ican ly lowe in animals su e ing om SAP
( igu e 3N). In summa y, hese esul s sugges ha AP causes
gene al immunosupp ession in he in es inal mucosa and
p onounced changes o he in es inal mic obio a composi ion.
Immune supp essi e unc ion o T egs du ing AP
To illumina e he ole o T egs, we analysed hei panc ea i is-
associa ed supp essi e capaci y. AP was induced by pa ial duc
liga ion in DEREG- mice exp essing a gene encoding a GFP-
diph he ia oxin (DT) ecep o usion p o ein unde he con ol
o a Foxp3- p omo e .20 GFP low cy ome ic analysis con i med
inc eased T eg numbe s a e induc ion o AP ( igu e 4A). The
same inc ease we could obse e in T- cells isola ed om he
duodenum ( igu e 4B). Using a luo escence- ac i a ed cell so e ,
GFP- posi i e T egs we e isola ed om he spleen o DEREG- mice
a e AP- induc ion and o con ol animals. We co- incuba ed
Figu e 3 Imbalance o he Th17/T eg a io in duodenal mucosa du ing AP. (A–E)His ological examina ion o he duodenum illus a es changes
o he in es inal mucosa in C57Bl/6 mice a e induc ion o AP. (A–C)In con as o only sligh changes in H&E- s aining we de ec ed di e ences in
immuno luo escen labelling o F4/80, CD68 as ma ke o mac ophages and sIgA as ma ke o plasma cells (B),all cells we e signi ican ly dec eased
in AP mice (C).(D)CD3 as T- cell ma ke and CD8α as ma ke o IELs we e labelled in small in es ine. (E)Ba g aphs illus a e signi ican changes o
T- cells and IELs in duodenum. (F)The ba g aph shows he -log10(pBH) alues o o e ep esen ed pa hways om an IPA analysis which was based on
di e en ially exp essed ansc ip ome da a o duodenal issue om AP and con ol mice. Hea map illus a es he z- sco e which indica es ac i a ion
(posi i e z- sco e, ed) o inhibi ion (nega i e z- sco e, blue) o his pa hway. (G)Hea - map illus a es old change di e ences o up egula ed ( ed) and
down egula ed (blue) genes in duodenal issue o AP mice compa ed wi h un ea ed con ols. (H)Lymphocy es isola ed om lamina p op ia as well
as om he epi helial laye we e analysed by low cy ome y. In CD4+ T- cells o he lamina p op ia he inc eased le els o CD25 and FOXP3 co ela ed
wi h disease se e i y. (I–J)The numbe s o CD4+/FOXP3+/CD25+- T egs (I)and CD4+/CD25+- T- cells (J) om IELs we e also inc eased in a se e i y-
dependen manne . (K, L)Th17- cells ma ked by RORγ +/CD25+ (K)and Th1- cells ma ked by TBET+/CD25+ (L)did no show a se e i y dependen
inc ease. (M)GFP- p oducing T egs in he duodenum o DEREG- mice we e de ec ed by an i- GFP and an i- CD3 labelling and showed a signi ican
inc ease a e onse o AP. (N)AP induced a shi o he Te /T eg a io. T egs showed a clea inc ease be ween mode a e and se e e AP in con as o Te
cells (Th1, Th2 and Th17). S a is ically signi ican di e ences we e es ed by unpai ed S uden ’s - es o independen samples and signi icance le els
o p<0.05 a e ma ked by an as e isk. AP, acu e panc ea i is.

1360 Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
he T egs a a ios o 1:8, 1:4, 1:2 and 1:1 wi h naï e CM- Dil-
labelled CD3/CD28 ac i a ed T- cells in he p esence o IL- 2.
Dilu ion o CM- Dil luo escence by cell di ision was used o
measu e he supp essi e ac i i ies o T egs on Te - cell p oli e a ion
and showed a signi ican ly s onge supp ession by T egs om AP
animals ( igu e 4C). In addi ion, we analysed he ansc ip ome
o he isola ed T egs using A yme ix GeneChips. Exp ession
o Il10 and Tg bi we e inc eased, whe eas he mRNA le els o
Tn and S a 1 we e dec eased ( igu e 4D). Pa hway analyses o
he ansc ip ion da a using IPA- so wa e indica ed AP- associ-
a ed down egula ion o cell cycle egula ion and o he gene al
immune esponse pa hways in T egs (online supplemen al igu e
S8A). Ups eam egula o s included he cy okines IFNγ, IL- 10,
he ansc ip ion ac o s STAT3, FOXP3 and STAT6, and he
ansmemb ane ecep o s IL10RA, CD28 and CTLA4 (online
supplemen al igu e S8B–D).
We u he analysed AP- induced al e a ions in he global
mRNA p o ile o isola ed T egs om spleen and om duodenal
samples. In bo h subse s he ‘PD- 1, PD- L1 cance immuno-
he apy pa hway’ in ol ed in T eg- media ed immunosupp es-
sion was signi ican ly induced. Analysis o disease- ela ed and
bio unc ion- ela ed pa hways e ealed nega i e ac i a ion sco es
o many immune pa hways. Inna e and adap i e immune cells
seem equally a ec ed and lymphocy e ac i a ion was signi i-
can ly inhibi ed (online supplemen al igu e S9A–D).
Deple ion o T egs s abilises he homoeos asis o CD4+
T cells in he lamina p op ia as well as o CD8α+ IELs in he
duodenum
To e i y he immunosupp essi e e ec o T egs du ing AP we
deple ed T egs in DEREG- mice by i.p. injec ion o DT be o e he
induc ion o panc ea i is. The deple ion o T egs in adul mice
does no esul in au oimmune disease.21 By low cy ome y
( igu e 5A) and immuno luo escence de ec ion o GFP+ cells in
lymph nodes ( igu e 5B) we con i med a signi ican educ ion
o GFP+/CD25+ T egs in DT- ea ed animals, whe eas phospha e-
bu e ed saline (PBS)- ea ed con ol mice showed he ypical
AP- induced inc ease o T egs ( igu e 5C). The Te popula ions in
spleen, GATA3+/CD4+ Th2- cells and TBET+/CD4+ Th1- cells
we e signi ican ly inc eased compa ed wi h con ol animals
(online supplemen al igu e S10A). The popula ion o splenic
LY6G+/LY6Clow/CD11b+- cells, which in e sely co ela e wi h
he disease se e i y we e signi ican ly ele a ed in DT- ea ed
mice, whe eas myelope oxidase ac i i y in lung issue, a ma ke
o lung inju y, was compa able o PBS- ea ed con ols (online
supplemen al igu e S10B). We nex in es iga ed i he T eg deple-
ion a ec s duodenal CD4+ T- cells and CD8α+ IELs popula ions.
Immuno luo escence labelling o CD3+ T- cells and CD8α+ IELs
showed, ha ollowing deple ion o T egs he numbe s o CD3+
cells and CD8α+ IELs we e signi ican ly highe in DT- ea ed
animals compa ed wi h PBS- ea ed mice ( igu e 5D,E). Subse-
quen ly, we isola ed leucocy es om he duodenum o DT- ea ed
DEREG- mice a e induc ion o AP. PBS- ea ed animals we e
used as con ols. In DT- ea ed animals GFP+/CD25+ CD4+
T egs we e absen om he lamina p op ia ( igu e 5F). Exam-
ina ion o CD4+Te cells wi hin he lamina p op ia showed an
inc ease o RORγ +/CD25+ Th17- cells ( igu e 5G), whe eas he
numbe o TBET+/CD25+ Th1- cells was no changed a e T eg
deple ion ( igu e 5H). In e es ingly, we iden i ied a popula ion
o CD4hi cells wi hin he popula ion o IELs which is inc eased
in DT- ea ed animals ( igu e 5I). A de ailed cha ac e isa ion o
his CD4hi popula ion iden i ied mainly GFP+/CD4hi T egs and
RORγ +/CD4hi Th17- cells. Whe eas RORγ +/CD4hi Th17- cells
pe sis ed in DT- ea ed animals, T egs we e comple ely abolished
( igu e 5J,K). We u he analysed he popula ion o CD8α+ IELs,
which accoun o he majo i y o IELs. While immuno luo es-
cence labelling o CD8ɑ+ IELs and o CD3+ T- cells o he lamina
p op ia showed a signi ican educ ion in numbe s ( igu e 5D,E),
hei a io was no a ec ed by DT- ea men ( igu e 5L). Su ace
ma ke s allow o so IELs in o wo di e en popula ions: (A)
Figu e 4 Immune supp essi e unc ion o T egs du ing AP. AP was induced by pa ial duc liga ion in DEREG- and C57Bl/6 mice. (A)The numbe o
splenic GFP p oducing T egs was inc eased in he AP g oup (n=5) bu no in con ol o sham ope a ed mice (n=3). (B)The same inc ease o FOXP3+/
CD25+/CD4+ T egs was obse ed in he small in es ine. (n=5). (C)Supp ession assays showed ha T egs om AP mice ( ed line) ha e an inc eased
supp essi e capaci y on Te cell p oli e a ion. (D)Hea map illus a es old changes o gene ansc ip ion in T egs om AP mice (n=4) compa ed wi h
un ea ed con ols (n=4). S a is ically signi ican di e ences we e es ed by unpai ed S uden ’s - es o independen samples and signi icance le els
o p<0.05 a e ma ked by an as e isk, co ec ed o mul iple es ing (bon e oni- co ec ion). AP, acu e panc ea i is.
1361
Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
induced- IELs posi i e o CD4+/CD8α+ and TCRαβ+ (T- cell
ecep o ), and (B) na u al- IELs which a e speci ic o CD8α+/
TCRγδ+ o CD8αβ-/TCRαβ+.22 Analysis o TCRβ and TCRγδ
on he su ace o CD8α+ IELs in DEREG- mice e ealed ha
he popula ion o induced IELs (CD8α+/TCRβ+) was inc eased
in AP, while he numbe o na u al IELs (CD8α+/TCRγδ+)
was signi ican ly dec eased. This TCRαβ+/TCRγδ+ shi was
less p onounced in T eg- deple ed mice ( igu e 5M). Finally, we
in es iga ed he p esence o he ansc ip ion ac o s TBET and
RORγ wi hin he CD8α+ IELs. No di e ences in he numbe
o TBET syn hesising IELs we e obse ed in he absence o T egs,
whe eas RORγ p oducing IELs we e signi ican ly inc eased
Figu e 5 Deple ion o T egs s abilises he homoeos asis o CD4+ T cells in he lamina p op ia as well as o CD8α+ IELs in he duodenum. AP
panc ea i is was induced in DEREG mice by pa ial duc liga ion, T egs we e deple ed by diph he ia oxin (DT) (n=11), con ols ecei e PBS (n=7).
(A)The e iciency o T eg deple ion wi h DT was e i ied by low cy ome y analysis o splenocy es. (B)Immuno luo escence de ec ion o GFP- p oducing
T egs was pe o med in lymph nodes o DT- ea ed and PBS- ea ed mice. (C)Ba g aphs show he a ios o CD25+ T- cells and GFP+/CD25+ T egs
analysed by low cy ome y o splenocy es. (D, E)Immuno luo escen labelling o CD3 and CD8α showed a signi ican pe sis ence o he T- cell and
IEL popula ion in he absence o T egs in DT- ea ed DEREG- mice compa ed wi h he PBS- ea ed g oup, (E).(F–N)Lymphocy es o he lamina p op ia
and he epi helial laye we e isola ed om duodenum and analysed by low cy ome y. (F)Following T eg- deple ion, no GFP+/CD25+ T egs we e de ec ed
wi hin he lamina p op ia. (G, H)Ba g aph show he numbe s o RORγ +/CD25+ Th17- cells (G)and TBET+/CD25+ Th1- cells (H).(I)The ba g aph
shows a highe a io o CD4hi exp essing IELs in DT- ea ed mice. (J, K)In he CD4hi IELs we obse ed a shi om GFP+ T egs (J) o RORγ + Th17- cells
(K).(L)The a io o CD8α+ IELs was no a ec ed by T eg- deple ion. (M)Do plo and ba g aphs illus a e he a io changes o TCRγδ+ IELs and TCRβ+
a e DT- ea men . (N)Wi hin he popula ion o CD8α+ IELs we measu ed an inc ease in RORγ p oducing cells whe eas TBET was no a ec ed.
S a is ically signi ican di e ences we e es ed by unpai ed s uden ’s - es o independen samples and signi icance le els o p<0.05 a e ma ked by
an as e isk. AP, acu e panc ea i is; IELs, in aepi helial lymphocy es; PBS, phospha e- bu e ed saline.
1362 Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
( igu e 5N). In summa y, he deple ion o T egs p e en ed he
panc ea i is- induced educ ion in T- cells in he lamina p op ia
and o CD8α+ IELs. Flow cy ome y da a showed a signi ican
inc ease o RORγ + Th17- cells, as well as a sus ained popula-
ion o CD8α+/TCRγδ+ na u al- IELs in DT- ea ed mice a e
he onse o AP.
Deple ion o T egs a enua es mic obial dysbiosis du ing
panc ea i is and p e en s bac e ial ansloca ion in o
panc ea ic nec osis
T eg- deple ion esul ed in inc eased Te numbe s hus we anal-
ysed whe he his inc ease also a ec ed mic obial composi ion in
AP. We pe o med 16S RNA gene sequencing based on isola ed
DNA om duodenal aspi a es o DEREG- mice. AP induced a
p onounced educ ion o bac e ial axa wi hin he duodenal
aspi a es, compa able o wha we obse ed in C57Bl/6 mice. In
con as o PBS- ea ed DEREG- con ols, mice ha had ecei ed
DT be o e induc ion o AP e ained a mo e a ied mic obial
composi ion, indica ed by a signi ican ly highe ASV ich-
ness and di e en composi ion shown by PCoA ( igu e 6A–C,
online supplemen al able 2). O e g ow h o acul a i e bac e-
ial pa hogens like Esche ichia/Shigella du ing AP was signi i-
can ly educed in he T eg deple ed g oup (S.10C). Quan i a i e
e e se ansc ip ion- quan i a i e PCR (RT- qPCR) con i med
he esul s om 16S RNA gene sequencing, he abundance o
acul a i e pa hogens like Esche ichia coli and En e ococcus
aecium is signi ican ly inc eased in mice wi h AP, bu o a mino
ex end in he DT- ea ed mice. Bene icial s ains like Lachno-
spi aceae o Lac obacillus emain unchanged (online supple-
men al igu e S10D). We analysed duodenal issue samples o
hese mice by RT- qPCR analysis and con i med he loss o T egs
by a DT- ea men - dependen dec ease in GFP ansc ip le els.
Du ing AP, we obse ed signi ican ly highe ansc ip le els o
he genes encoding IL- 15, IFNγ, TNFα and IL- 17F in DT- ea ed
mice. In e es ingly, ansc ip s o IL- 17A, IL- 22 o IL- 10 did
no di e be ween PBS- ea ed o DT- ea ed DEREG- mice
( igu e 6D). In T eg- deple ed AP mice, he mRNA amoun s o
he bac e ial de ence genes Tn s 13, Tn s 13b, Jchain and De a5
we e signi ican ly ele a ed. On he o he hand, he ansc ip
abundances o De c - s1 and Reg3g encoding an imic obial
pep ides we e no signi ican ly inc eased, while hose o he
genes encoding he bac e ioly ic enzymes Lysozyme 1 and 2,
which a e no mally sec e ed by Pane h cells, we e signi ican ly
highe ( igu e 6E). The exp ession o genes encoding he phys-
ical ba ie p o eins Occludin and Claudin- 1 was signi ican ly
inc eased in DT- ea ed mice whe eas changes in o he ba ie
p o eins like E- Cadhe in, Claudin- 5, ZO- 1 o Mucin- 2 did no
each a signi icance le el in DT- ea ed mice (online supple-
men al igu e S10E). Impo an ly, immuno luo escen labelling
o sIgA demons a ed ha in T eg- deple ed animals he popula-
ion o sIgA p oducing cells wi hin he lamina p op ia emained
cons an in panc ea i is ( igu e 6F).
Finally, we analysed whe he T eg- deple ion imp o es he
s abili y o he in es inal ba ie . We pe o med 16S RNA gene
sequencing in panc ea ic issue DNA o AP mice o de ec any
bac e ial aces. A cu - o o 5000 clean eads was exceeded in
8/25 (32%) o PBS- ea ed mice, indica ing in ec ed panc ea ic
nec osis. In con as , only 1/14 (7.14%) DT- ea ed mice eached
he cu - o le el (online supplemen al igu e S10F). S a is ical
analysis by Mann- Whi ney es con i med a signi ican ly lowe
bac e ial con amina ion o nec o ic issue in T eg-
deple ed mice.
Facul a i e pa hogens like Esche ichia/Shigella, En e obac e-
iaceae di e sa, S aphylococcus o En e ococcus ep esen ed
one o he 15 h mos abundan axa wi hin he nec o ic a eas
( igu e 6G, online supplemen al igu e S11A,B). A compa -
ison be ween panc ea ic samples and gu samples iden i ied
ASVs, which we e obse ed wi h inc eased abundance bo h in
duodenal samples (Esche ichia/Shigella, En e ococcus and Ci o-
bac e ) as well as in nec o ic a eas o he panc eas (online supple-
men al igu e S11C–E). In a compa a i e analysis we obse ed
ha he e y same bac e ial axa ha appea o mul iply in he
duodenal aspi a es om AP- animals we e isola ed om mu ine
panc ea ic nec osis ( igu e 6H). Quan i a i e RT- qPCR e ealed
signi ican ly highe numbe s o 16S RNA gene copies (shown as
lowe dC alue) in he panc eas o PBS- ea ed mice compa ed
wi h DT- ea ed mice wi h AP ( igu e 6I). O al adminis a ion o
an an ibio ic cock ail con aining 10 mg/mL ampicillin, 10 mg/mL
neomycin, 5 mg/mL ancomycin and 10 mg/mL me onidazole,
was pe o med e e y 12 hou s o e he expe imen al pe iod23
and p e en ed bac e ial ansloca ion in o he panc eas compa-
able o he deple ion o T egs by DT ( igu e 6I). 16S RNA gene
sequencing o duodenal samples showed ha an ibio ic ea men
p e en ed bac e ial o e g ow h in he duodenum o acul a i e
pa hogenic bac e ia such as Esche ichia/Shigella, En e ococcus,
S aphylococcus, En e obac e iaceae o Klebsiella, whe eas bene-
icial axa like Lac obacillus emained (online supplemen al
igu e S12A–C). We analysed in DT- ea ed animals whe he he
deple ion o T egs o an an ibio ic he apy amelio a ed he se e i y
o panc ea i is. Enzyma ic ac i i ies o common se e i y ma ke s
such as se um amylase and se um lipase we e signi ican ly lowe
in DT- ea ed and an ibio ic- ea ed animals ( igu e 6J). In
H&E- s ained panc ea ic sec ions less o gan damage was seen in
he T eg deple ed mice as well as in an ibio ic- ea ed mice. His o-
logical sco ing con i med signi ican ly less damage and educed
panc ea ic nec osis in DT- ea ed and an ibio ic- ea ed mice
( igu e 6K, online supplemen al igu e S12D). In conclusion,
he esul s sugges ha he amelio a ing e ec o T eg- deple ion
on disease se e i y is mainly due o he educ ion o bac e ial
ansloca ion. Deple ion o T egs in combina ion wi h an ibio ic
ea men showed no addi ional bene i .
Bac e ial in ec ion o human panc ea ic nec osis
The bac e ial colonisa ion o human panc ea ic nec osis samples
was analysed by 16S RNA gene sequencing. H&E- s aining o
hese samples showed no emaining in ac acina issue a chi-
ec u e (online supplemen al igu e S13A,B). Fo 58 indi idual
pa ien s (online supplemen al igu e S13C), samples ASVs
we e in e ed om PCR- ampli ied 16S RNA gene sequences.
Samples wi h >5.000 clean eads (53/58) o ep esen a i e ASV
sequences we e subsequen ly analysed ( igu e 7A). Fi micu es
was ound o be he mos dominan phylum, ep esen ing mo e
han 50% o he o al ela i e abundance, mainly consis ing o
he gene a En e ococcus, S ep ococcus and S aphylococcus.
The amily o En e obac e iaceae (phylum: P o eobac e ia) also
showed a high abundance wi h Esche ichia/Shigella as he mos
equen axa (online supplemen al igu e S13D), he same axa
we obse ed in mu ine nec osis. O he phyla like Bac e oide es,
Fusobac e ia o Ac inobac e ia we e ound wi h conside ably
lowe abundance. Some pa ien s unde wen mul iple subse-
quen nec osec omy p ocedu es and he analysis o ollow- up
samples showed ha he mic obiome in hese nec o ic samples
was a he s able o e pe iods o se e al weeks (online supple-
men al igu e S14A). In ec ion o panc ea ic nec osis by En e o-
coccus showed a signi ican co ela ion wi h he leng h o
hospi alisa ion ( igu e 7B). None o he o he 15 equen axa
(online supplemen al igu e S13D) o acul a i e pa hogens o
1363
Glaubi zJ, e al. Gu 2023;72:1355–1369. doi:10.1136/gu jnl-2022-327448
Panc eas
he commensal gu bac e ia showed his kind o co ela ion
(online supplemen al igu e S14B). Subg oup analyses ega ding
AP e sus acu e episode o CP showed di e ences, bu he high
a iance in he mic obiome composi ion o panc ea ic nec osis
and he ela i e low numbe o pa ien s did no allow o d aw
i m conclusions (online supplemen al igu e S14C). Addi ional
me agenomic sequencing o eigh nec osis samples ( igu e 7C)
iden i ied he in ec ing bac e ial species En e ococcus aecalis,
En e ococcus aecium, Esche ichia coli, Bac e oides spp and
Ci obac e spp, hese axa a e equen ly ound in he gu .
Figu e 6 Deple ion o T egs a enua es mic obial dysbiosis du ing panc ea i is and p e en s bac e ial ansloca ion in o panc ea ic nec oses. (A)Hea -
map illus a es he majo mic obial axa in he duodenum o DEREG ea ed wi h PBS (n=8) o DT (n=8) a e induc ion o AP. Animals wi hou AP
e e ed as heal hy con ols (0d). Facul a i e pa hogenic we e ma ked in ed, bene icial commensal bac e ia we e ma ked in g een. (B)Box plo s
illus a e Shannon- Di e si y Index and species ichness in DT- o PBS- ea ed DEREG- mice. (C)A p incipal coo dina e analysis illus a es signi ican
di e ences o he duodenal mic obiome be ween hese g oups (pe mu a ional mul i a ia e analysis o a iance; con ol s se e e: p<0.001,
R2=18.2%). (D, E)Box plo s show RT- qPCR analysis o duodenal ansc ip ional changes a e deple ion o T egs. (F)Immuno luo escen labelling o
sIgA p oducing cells in he duodenal mucosa om con ol and AP animals showed signi ican ly dec eased numbe s o sIgA p oducing cells in AP
mice which ecei ed PBS. (G)S aked- ba g aphs illus a es bac e ial axa which could be iden i ied in all samples wi h mo e han 5.000 clean eads
om panc ea ic nec osis in DEREG AP- mice (8 PBS ea ed s 1 DT ea ed DEREG- mice). (H)The same bac e ial axa ha appea o expand in he
duodenum o AP- mice a e ound in mu ine panc ea ic nec osis samples. The s acked ba g aphs show he mean o all mice. (I)We analysed con ol
(con) and duc - liga ed DEREG mice wi h AP (PBS and DT) by 16S RNA gene RT- qPCR analysis o isola ed DNA om panc ea ic issue. Dec eased C
alues indica e bac e ial in ec ion in he panc eas o mice wi h AP wi h a signi ican g ea e ex en in he PBS- ea ed g oup compa ed wi h T eg
deple ed mice. An ibio ic ea men (+AB) signi ican ly educed he copy numbe o bac e ial 16S RNA gene in he panc eas and he e o e p e en ed
bac e ial ansloca ion du ing AP. (J, K)To e alua e how T egs and mic obiome composi ion a ec he disease se e i y DEREG mice we e ea ed a e
AP induc ion wi h an ibio ics (AB), ei he in p esence (+PBS) o absence (+DT) o T egs. (J)Disease se e i y was e alua ed by analysis o se um amylase
and lipase. (K)Panc ea ic his ology was analysed by H&E- s aining o issue sec ions, his ology sco e was e alua ed by quan i ica ion o oedema,
nec osis and leucocy e in il a ion. The hea map illus a es he mean esul o all g oups. S a is ically signi ican di e ences we e es ed by unpai ed
S uden ’s - es o independen samples. Fo mo e han wo g oups, s a is ical signi icance was de e mined by ANOVA one way analysis o a iance
ollowed by Bon e oni co ec ion o mul iple es ing, signi icance le els o p<0.05 a e ma ked by an as e isk. ANOVA, analysis o a iance; DT,
diph he ia oxin; RT- qPCR, e e se ansc ip ion- quan i a i e PCR; PBS, phospha e- bu e ed saline.