Journal(club( - Mucosal Immunology · LETTER doi:10.1038/nature11462 Pregnancy imprints regulatory memory that sustains anergy to fetal antigen Jared H. Rowe1, James M. Ertelt1{,

Journal club

5th October 2012 Yasmin

LETTERdoi:10.1038/nature11462

Pregnancy imprints regulatory memory thatsustains anergy to fetal antigenJared H. Rowe1, James M. Ertelt1{, Lijun Xin1{ & Sing Sing Way1{

Pregnancy is an intricately orchestrated process where immuneeffector cells with fetal specificity are selectively silenced. Thisrequires the sustained expansion of immune-suppressive maternalFOXP31 regulatoryT cells (Treg cells), because even transient partialablation triggers fetal-specific effector T-cell activation and preg-nancy loss1,2. In turn, many idiopathic pregnancy complicationsproposed to originate from disrupted fetal tolerance are associatedwith bluntedmaternal Treg expansion3–5. Importantly, however, theantigen specificity and cellular origin of maternal Treg cells thataccumulate during gestation remain incompletely defined. Herewe show that pregnancy selectively stimulates the accumulation ofmaternal FOXP31 CD4 cells with fetal specificity using tetramer-based enrichment that allows the identification of rare endogenousT cells6. Interestingly, after delivery, fetal-specificTreg cells persist atelevated levels, maintain tolerance to pre-existing fetal antigen,and rapidly re-accumulate during subsequent pregnancy. Theaccelerated expansion of Treg cells during secondary pregnancywas driven almost exclusively by proliferation of fetal-specificFOXP31 cells retained from prior pregnancy, whereas inducedFOXP3 expression and proliferation of pre-existing FOXP31 cellseach contribute to Treg expansion during primary pregnancy.Furthermore, fetal resorption in secondary compared with primarypregnancy becomes more resilient to partial maternal FOXP31 cellablation. Thus, pregnancy imprints FOXP31 CD4 cells that sustainprotective regulatory memory to fetal antigen. We anticipate thatthese findings will spark further investigation on maternal regula-tory T-cell specificity that unlocks new strategies for improvingpregnancy outcomes and novel approaches for therapeuticallyexploiting Treg cell memory.The accumulation of maternal Treg cells during pregnancy

parallels the need for expanded tolerance to encompass ‘non-self’ fetalantigens3–5,7,8. However, one consequence of sustained FOXP31 cellexpansion is susceptibility to prenatal infection2. Given the increas-ingly recognized importance of Treg specificity in regulating the fluidbalance between immune activation that maintains host defence andimmune suppression that prevents autoimmunity9–14, we reasonedthat establishing the specificity of maternal Treg cells that expand duringpregnancy could unravel ways to dissociate their beneficial and detri-mental impacts. Furthermore, extending this analysis post-partummayallow the regulatorymemory recently described for Treg cells responsiveto an induced self antigen to be investigated in a more physiologicalcontext15. To address these questions, we developed a mating strategywhere the I-Ab 2W1S55–68 peptide (a variant of peptide residues 55–68for the alpha chain of the mouse major histocompatibility complex(MHC) class II, I-Ed) becomes a surrogate fetal antigen using malemice (H-2d; Balb/c or H-2b C57BL/6 [B6]) engineered to co-expressthis peptide with b-actin to impregnate non-2W1S-expressing B6females16. In turn, the high precursor frequency of CD4 cells with2W1S55–68 specificity allows endogenous maternal Treg cells to thissurrogate fetal antigen to be identified using MHC class II tetramerenrichment6.

Using this approach, maternal CD4 cells with fetal-2W1S specificitywere found to sharply upregulate CD44 expression, progressivelyaccumulate throughout pregnancy, and persist at approximately ten-fold increased levels through day 100 post-partum compared withnon-pregnant controls (Fig. 1a). Maternal 2W1S1 cell expansionwas specific to mating with 2W1S-expressing mice because they didnot accumulate in females impregnated by non-transgenic Balb/cmales (Supplementary Fig. 1). Because seminal fluid also contains cellsof paternal origin17, 2W1S1 cells in female mice rendered infertile withlow-dose irradiation were also enumerated. We found that althoughmating without pregnancy stimulated modest 2W1S1 cell expansionandCD44upregulation, themagnitudewasmarkedly reduced comparedwith pregnantmice (Supplementary Fig. 1). Thus,maternal 2W1S1CD4cell expansion during pregnancy reflects an antigen-specific responseto cells of fetal origin.Given the essential requirement for Treg cells in maintaining fetal

tolerance2,7,18,19,we investigatedFOXP3expressionamongmaternal cellswith fetal-2W1S specificity. Beginning mid-gestation, 2W1S1 com-pared with 2W1S2 CD4 cells became enriched for FOXP3 expressionin allogeneic (Fig. 1a, b), as well as syngeneic pregnancy (Supplemen-tary Fig. 2). As pregnancy progressed, FOXP3 expression among2W1S1 cells becameprogressivelymore pronounced, peaking at around50% late gestation through to the first 48 h post-partum (embryonic day18.5 (E18.5) to post-partum day 2 (PP2)) (Fig. 1a, b and SupplementaryFig. 3). Furthermore, 2W1S1FOXP31 cells, and to a lesser extent2W1S1FOXP32 cells, upregulated the proliferation marker Ki67 thatparalleled expanding fetal tissue (Fig. 1c). Reciprocally after expulsionof the fetus (PP14 to PP100), Ki67 expression among 2W1S1FOXP31

and 2W1S1FOXP32 cells became reduced (Fig. 1c). However, despitediminished Ki67 levels, FOXP3 expression among 2W1S1 cellswas sustained at ,20% through day 100 post-partum (Fig. 1a, b).Accordingly, maternal Treg cells with fetal specificity selectively accu-mulate during pregnancy and persist following parturition.Interestingly, maternal Treg cells with fetal-2W1S specificity also

progressively downregulatedHelios (also known as IKZF2) expressionthat dropped to its lowest level of ,40% Helioshi by late gestation,whereas the few 2W1S1FOXP31 cells in non-pregnant mice were uni-formlyHelioshi (Fig. 1d). Comparatively,Helios expression amongbulkmaternal Treg cells did not shift significantly. Although this discordancein Helios expression may suggest conversion of fetal-specific FOXP32

cells into FOXP31 cells20, the recent finding that some peripherallyinduced Treg cells also expressHelios led us tomore definitively investi-gate the origin ofmaternal Treg cells with fetal specificity21. In particular,we asked whether mating with 2W1S-expressing males can convert2W1S1FOXP32 CD4 cells from Foxp3DTR/DTR donors ablated of Tregcells with diphtheria toxin22 (Foxp3DTR/DTR Treg cells carry the humandiphtheria toxin receptor (DTR) fused to an internal ribosome entrysite into the 39untranslated regionofFoxp3, rendering themsusceptibleto ablation with low-dose diphtheria toxin) into FOXP31 cells afteradoptive transfer into virgin Foxp3WT/WT recipient mice. By mid-gestation, 2W1S1FOXP31 among Treg-ablated donor Foxp3DTR/DTR

1University of Minnesota School of Medicine, Departments of Pediatrics and Microbiology, Center for Infectious Disease and Microbiology Translational Research, Center for Immunology, Minneapolis,Minnesota 55455, USA. {Present addresses: Cincinnati Children’s Hospital Medical Center, Division of Infectious Diseases, Cincinnati, Ohio 45229, USA (J.M.E., L.X. and S.S.W.).

0 0 M O N T H 2 0 1 2 | V O L 0 0 0 | N A T U R E | 1

Macmillan Publishers Limited. All rights reserved©2012

First pregnancy Second pregnancy

Treg Treg Treg

Virgin

Introduc-on:

•  Tregs are important during pregnancy in avoiding maternal immune reac@on towards the fetus

•  Controversy: General vs an-gen-‐driven expansion?

•  If tolerance to fetus fails: Pre-‐eklampsia in women (similar to fetal resorp-on in mice)

•  6-‐8% of all pregnancies worldwide, 2nd-‐3rd trimester

•  Hypertension and proteinuria, remodelling of spiral arteries

•  maternal immune reac@on to the placenta

•  less frequent in secondary pregnancies! But recurrent with long period between pregnancies

•  Treg memory is not well studied

Supplementary Fig. 8)24. By extension, the restored responsiveness ofpost-partum 2W1S1 cells after adoptive transfer into Treg-sufficientnaive mice most likely represents dilution of co-transferred maternalTreg cells with fetal-2W1S specificity (Fig. 3c).Lastly, to establish how maternal Treg cells with fetal specificity

retained post-partum have an effect on subsequent pregnancy out-comes, the frequency of fetal resorption triggered by partial maternalFOXP31 cell ablation usingFoxp3DTR/WTmicewas compared betweensecondary and primary pregnancy2. We found secondary pregnancybecame significantlymore resilient to partial Treg ablation because fetalresorption was reduced by,60% compared with primary pregnancy(Fig. 4c). In turn, fetal resorption in Treg-sufficient Foxp3

WT/WT miceduring secondary pregnancywas also significantly reduced from back-ground levels compared with primary allogeneic pregnancy. MaternalTreg cells were essential for these protective effects because wholesaleFOXP31 cell ablation using Foxp3DTR/DTR mice triggered pervasivefetal resorption equally in secondary andprimaryallogeneic pregnancy(Fig. 4c). Importantly, fetal wastage with maternal Treg ablation in thiscontext was driven by antigen heterogeneity, and not poor maternalhealth, because the frequency of resorption was sharply reduced withTreg ablation in mice bearing syngeneic pregnancy (SupplementaryFig. 9).Together, these findings establish amodelwhereby pregnancy primes

the selective accumulation and activation ofmaternal Treg cellswith fetal

specificity (Supplementary Fig. 10), and extend the role of antigen-experienced Treg cells from primary into subsequent pregnancies2,7,18.In this regard,whereasmaternal Treg cells havebeendescribed to expandup to twofold when examined in a non-antigen-specific fashion2–5, ourresults demonstrate that FOXP31 cells with fetal specificity expand.100-fold through parturition (Fig. 1a and Supplementary Fig. 3).After delivery, maternal Treg cells with fetal specificity are sustainedat enriched levels, and are functionally distinct as they re-accumulatewith accelerated kinetics and out-compete ‘naive’ Treg cells duringsecondary pregnancy. Similar to discordant functional properties ofnaive and activated effector T cells27, these results uncover the excitingpossibility of exploiting antigen-specific ‘memory’ Treg cells to disso-ciate detrimental and beneficial immune responses. Applied to humanpregnancy, these datamay explainwhy rates of pre-eclampsia, and othercomplications associated with disrupted fetal tolerance, are reduced insecondary compared with primary pregnancy28. However, given theincreased risk of pre-eclampsia in recurrent human pregnancy whenthe inter-pregnancy interval is extended, waning Treg memory similar

Figure 3 | The post-partum environment maintains anergy for maternalCD4 cells with pre-existing fetal specificity. a, IFN-c-producing 2W1S1CD4cells 5 days after Lm-2W1S inoculation in virgin or pregnant mice mid-gestation by Balb/c-2W1S or Balb/c males. b, IFN-c-producing 2W1S1 CD4cells 5 days after Lm-2W1S inoculation in post-partum mice previouslyimpregnated by Balb/c-2W1S or Balb/c males. c, IFN-c-producing post-partum donor (CD45.21CD90.21), naive donor (CD45.21CD90.11) or naiverecipient (CD45.11) CD4 cells 5 days after Lm-2W1S inoculation andstimulation with phorbol myristate acetate/ionomycin. Bars, means6 onestandard error.

Figure 4 | Maternal post-partum Treg cells mitigate IFN-c responsivenessand mediate resiliency to fetal resorption in secondary pregnancy.a, Representative plots illustrating the majority (.96%) of FOXP31 cells arederived from adoptively transferred CD4 in diphtheria-toxin-treatedFoxp3DTR/DTR mice. b, IFN-c-producing 2W1S1 cells amongCD45.11CD45.22 cells, accumulation of 2W1S1FOXP31 cells, and Heliosexpression among 2W1S1FOXP31 Treg cells 5 days after Lm-2W1Sinoculation. c, Percentage fetal resorption during primary (open) or secondary(shaded) allogeneic pregnancy forFoxp3WT/WT, Foxp3DTR/WT orFoxp3DTR/DTR

females 5 days after diphtheria toxin initiation beginning mid-gestation. Bars,means6 one standard error.

RESEARCH LETTER

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 2


a b

IFN

-γ

T-bet

CD

44

2W1S

Virgin Balb/c-2W1S Balb/c

13.6 12.0 1.6 1.4 16.2

T-bet

CD

44

2W1S

Balb/c-2W1S Balb/c

Virgin Balb/c-2W Balb/c0

10203040

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Balb/c-2W1S Balb/c0

10

20

30

CD

44

2W1S

IFN

-γ

T-bet

CD

45.1

CD45.2

Lm-2W1S

CD45.2 CD90.2 (post-partum donor) CD45.2 CD90.1 (naive donor)

CD

90.2

CD90.1

6.7

91.5

18.9

13.7

86.2

16.3

15.8

Naive (recipient)

Naive (donor)

Post-partum (donor)

c

P = 0.003 P = 0.004 P = 0.001

CD45.1

CD45.2 CD90.2

CD45.2 CD90.10

10

20

30

2W1S

+ ce

lls

P = 0.08 P = 0.98

IFN

-γP

erce

ntag

e IF

N-γ

+

2W1S

+ ce

lls

CD45.1 recipient

Per

cent

age

IFN

-γ+

CD45.1 post-partum Foxp3DTR/DTR recipient

a

b

CD45.2 naive donor

Sustained DT

CD

45.1

CD45.2

FOXP

3

Helios

14.8

84.2

5.7

93.9

96.1

3.8

Foxp3+

FOXP3–

FOXP

3

Helios

CD

44

2W1S

IFN

-γ

T-bet

23.5

0.8 17.5

2.4 Intact post-partum

Reconstituted post-partum

Intact

Reconstituted

0

25

50

75

100

Intact

Reconstituted

0

10

20

30

40

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Intact

Reconstituted

101

102

103

104

2W1S

+ FO

XP3+

cells

per

mou

se

P = 0.003 P = 0.02 P = 0.003

Foxp3

WT/WT

Foxp3

DTR/WT

Foxp3

DTR/DTR0

20

40

60

80

100P

erce

ntag

e re

sorb

ed fe

tus

Primary pregnancySecondary pregnancy

P = 0.02

P = 0.002

P = 0.95 c

Per

cent

age

Hel

ioshi

♂ ♀

MHC-‐II I-‐ED alpha chain (Balb/c) Not present in B6!

2W1S55-‐68

Balb/c H-‐2d

Transgenic for 2W1S

C57BL/6 H-‐2b

Transgenic for 2W1S

C57BL/6 H-‐2b

C57BL/6 H-‐2b

Wild type (Normally 2W1S is not present)

H2b/d

Transgen +/-‐

H2b/b

Transgen +/-‐

Recognised by B6

X

X

2W1S55-‐68 presented on MHC-‐II I-‐Ab

MHC-‐II I-‐Ab

Fetus

Tools and experimental setup:

by chance a large propor@on of the B6 intrinsic CD4+ cell repertoire recognise 2W1S presented on I-‐Ab

Any maternal 2W1S+ T cell expansion during pregnancy reflects an an@gen-‐specific response to the fetus.

Allogenic pregnancy

Syngenic pregnancy

Fetal-‐specific maternal CD4+ T cells increase the level of FoxP3 expression during gesta@on.

Maternal CD4+ Cell adhesion,

Binds hyaloronic acid

Pregnancy Aaer birth – post-‐partum (PP)

2W1S+ cells -‐ spleen, all pLN

Mated with Balb/c 2W1S

Where do FoxP3 cells originate from?

Prolifera-on of FoxP3+ cells contributes to fetus-‐specific T cell pool

Gated on 2W1S+ CD4+

Can an@gen-‐specific maternal T cells (endogenous) convert into FoxP3+ cells during pregnancy?

Prolifera@on of FoxP3+ cells

Induc@on of Helios nega@ve cells

Gated on 2W1S+ CD4+FoxP3+

Can an@gen-‐specific maternal T cells convert into FoxP3+ cells during pregnancy?

cellswere readily recovered, illustrating inductionofmaternal Treg cellswith fetal specificity (Fig. 1e). This conversion was pregnancy-specificand not due to incomplete donor Treg ablation because FOXP31 cellswere undetectable among Treg-ablated donor cells in unmated controlmice (Supplementary Fig. 4). Importantly, however, FOXP31 amongTreg-ablated donor CD4 cells was also consistently reduced (by,50%)compared with either 2W1S1FOXP31 donor cells in mice withoutdiphtheria toxin treatmentoramong recipientCD4cells not susceptibleto diphtheria toxin (Fig. 1e). Thus, FOXP3 induction among FOXP32

precursors and proliferation of pre-existing FOXP31 cells each con-tribute to the accumulation of maternal Treg cells with fetal specificityduring primary pregnancy.To further characterize maternal Treg cells with specificity to pre-

existing fetal antigen that persist post-partum, these cells were trackedduring subsequent pregnancy. After secondary mating, maternalFOXP31 cells with fetal-2W1S specificity accumulatedwith acceleratedkinetics in an antigen-specific fashion (Fig. 2a and SupplementaryFig. 5). The more rapid expansion of maternal Treg cells in separategroups ofmice was recapitulatedwithin the samemouse bymeasuring

2W1S1 Treg accumulation among donor CD4 cells from post-partummice (secondary expansion) adoptively transferred beforemating with2W1S-expressing males, compared with cells in virgin recipient mice(primary expansion) (Fig. 2b). By substituting CD4 cells from post-partum Foxp3DTR/DTR mice for adoptive transfer and using diphtheriatoxin to eliminate donor Treg cells before mating, we also addressedwhether the accelerated secondary expansion of maternal Treg cellswith fetal-2W1S specificity reflects more vigorous induction amongFOXP32 cells or proliferation of pre-existing FOXP31 cells.We foundthat, in sharp contrast to primary pregnancy, the ablation of donorTreg cells from post-partum Foxp3DTR/DTR mice almost uniformlyeliminated their expansion in subsequent pregnancy (Fig. 2c). Thus,recurrent pregnancy primes the accelerated accumulation of maternalFOXP31 cells that expand from pre-existing Treg cells retained fromprior pregnancy.Expanding this model, the responsiveness of maternal CD4 cells

with fetal specificity was also investigated. We found 2W1S1 cellsrecovered from mice mid-gestation or post-partum each comparedwith non-pregnant controls did not produce appreciable IFN-c ex vivo

Figure 1 | Accumulation of maternal CD4 and FOXP31 Treg cells with fetalspecificity during gestation. a, Total 2W1S1 or 2W1S1FOXP31 CD4 cells inB6 females impregnated by Balb/c-2W1S males. b, Percentage FOXP31

among 2W1S1 or 2W1S2 CD4 cells. c, Percentage Ki671 among2W1S1FOXP31 or 2W1S1FOXP32 CD4 cells. d, Percentage Helioshi among

2W1S1FOXP31or 2W1S2FOXP31 CD4 cells. e, Percentage FOXP31 amongFoxp3DTR/DTR donor (CD45.11) or Foxp3WT/WT recipient (CD45.21) 2W1S1

CD4 cells mid-gestation (E11.5) by Balb/c-2W1S males, with diphtheria toxin(DT) treatment (top) or no diphtheria toxin controls (bottom). Bars,means6 one standard error.

RESEARCH LETTER

2 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 2


b

29.4 58.6 51.1 17.6 18.5 58.7 19.1

Virgin E11.5 E18.5 PP2 PP14 PP30 PP1000

20406080

Per

cent

age

Ki6

7+

FOXP3–FOXP3+

Ki67

59.1 40.5 63.6 65.6 51.9 97.3 49.2


20406080

100

Per

cent

age

Hel

ioshi


Helios



Tota

l num

ber

per m

ouse

2W1S+

2W1S+ FOXP3+


CD

44

2W1S


20406080

Per

cent

age

FOXP

3+ 2W1S+

2W1S–

45.1 21.1 7.0 60.1 19.6 20.1 18.4 Virgin E11.5 E18.5 PP2 PP14 PP30 PP100

Foxp

3

Helios

a

c d

CD45.1 CD4 cells from Foxp3DTR/DTR donor

CD45.2 recipient

DT, then mate Balb/c-2W1S

No DT, mate Balb/c-2W1S

CD

44

2W1S

9.8

89.5 C

D45

.1

CD45.2

13.9

24.5

FOXP

3

Helios

CD45.1

CD45.2

22.1

25.1

FOXP

3

Helios

CD45.1

CD45.2

CD45.2CD45.1

0

20

40

60

Per

cent

age

FO

XP3+

CD45.2CD45.1

0

20

40

60

e

9.5

89.6

CD

45.1

CD45.2

CD

44

2W1S

P = 0.001

P = 0.57

2W1S+

2W1S–

Per

cent

age

FO

XP3+

FoxP3DTR/DTR

CD45.1 congenic

Number of FoxP3+ cells newly induced

CD4 CD45.1

CD4 CD45.1

CD4 CD45.1

CD4 CD45.1 FoxP3

CD4 CD45.1 CD4 CD45.1 FoxP3

CD4 CD45.1

CD4 CD45.1

CD4 CD45.1

CD4 CD45.1 FoxP3

CD4 CD45.1 CD4 CD45.1 FoxP3

Ctrl: Number of FoxP3+ cells recovered aaer transfer

Conversion of FoxP3+ cells contributes to fetus-‐specific T cell pool

DT abla@on in upper panel

Mate with transgenic male

mid-‐gesta@on E11.5

following stimulation, consistent with previously described anergy ofmaternal cells with fetal specificity (Supplementary Fig. 6)2,23. Therefore,tomore fully evaluate the responsiveness ofmaternal T cells with fetal-2W1S specificity, wemeasured their in vivo response to Listeriamono-cytogenes engineered to express the 2W1S55–68 peptide (Lm-2W1S)that potently stimulates TH1-differentiation in other contexts24,25. Wefound that 2W1S1 cells expand and upregulate T-bet (also known asTBX21) expression each in an antigen-specific fashion in both naivemice andmice impregnated by 2W1S-expressingmales after Lm-2W1Sinoculation, similar to other intracellular pathogens (SupplementaryFig. 7)26. Interestingly, however, 2W1S1 cells in pregnant mice where2W1S represents a surrogate fetal antigen produced only backgroundlevels of IFN-c and other effector cytokines, with reciprocal accumula-tion of FOXP31 cells (Fig. 3a and Supplementary Fig. 8). Comparatively,.15%of 2W1S1 cells inLm-2W1S-inoculated virginmicewere IFN-c1

(Fig. 3a). This hypo-responsiveness was specific to fetal-2W1S stimu-lation, because 2W1S1CD4 cells inmice impregnated with non-2W1S-expressing males produced IFN-c levels comparable to non-pregnantcontrols (Fig. 3a andSupplementary Fig. 8).Given the sustained enrich-ment of fetal-specific Treg cells after delivery (Fig. 1a, b), these studieswere extended to investigate whether diminished IFN-c productionamong maternal CD4 cells with specificity to pre-existing fetal antigenis similarly maintained. Remarkably, IFN-c production remainedanaemic in post-partum mice previously exposed to 2W1S as a fetalantigen, whereas post-partummice without prior fetal-2W1S exposureproduce IFN-c comparable to non-pregnant controls (Fig. 3b).Accordingly, pregnancy imprints functional anergy for maternal CD4cells with fetal specificity that is sustained post-partum.

To dissociate whether pregnancy-induced T-cell anergy was cell-intrinsic or imposed by features associated with the post-partum envir-onment, wemeasured IFN-cproduction byCD4 cells frompost-partumor virginmice after adoptive transfer intonaive recipientmice.We foundIFN-cproductionbydonorpost-partumandeach groupof naive (donorand recipient) 2W1S1 CD4 cells were similar, and notably increasedcompared with 2W1S1 cells in un-manipulated post-partum mice fol-lowing Lm-2W1S inoculation (Fig. 3b, c). Thus, anergy amongmaternalCD4 cells with specificity to pre-existing fetal antigen is not cell-intrinsic,but maintained by the post-partum environment.In complementary studies we addressed the importance ofmaternal

Treg cells in sustaining anergy to cells with specificity to pre-existingfetal antigen by investigating the effect of replacing the entire Tregcompartment in post-partum mice previously exposed to fetal-2W1Swith naive FOXP31 cells from virgin mice. Consistent with recentstudies using adoptively transferred Foxp3WT/WT CD4 cells to refillthe cellular compartment in Foxp3DTR/DTRmice sustained ondiphtheriatoxin treatment2, Treg cells fromnaivemice efficiently reconstitutedTreg-ablated Foxp3DTR/DTR post-partummice (Fig. 4a). Using this approach,we found that replacing maternal FOXP31 cells in post-partum micewith Treg cells from naive mice restored IFN-c production for 2W1S1

CD4 cells (Fig. 4b). Furthermore, whereas only rare 2W1S1FOXP31

cells that were Helioshi were found among post-partum mice recons-tituted with naive Treg cells, a significant proportion of 2W1S1 cellsexpanded in response to Lm-2W1S in intact post-partum miceremained FOXP31 (,20%) and Helioslo (,40%) (Fig. 4b). Thus, themuted expansion of naive FOXP31 CD4 cells with L. monocytogenesinfection is overcome by pregnancy-induced Treg activation (Fig. 4b and

Figure 2 | Accelerated expansion of maternal Treg cells with fetal specificityduring secondary pregnancy. a, Percentage FOXP31 among virgin (primarypregnancy) or post-partum (secondary pregnancy) females before mating ormid-gestation (E11.5) by Balb/c-2W1S males. b, Percentage FOXP31 among

post-partum donor (CD45.11) or naive recipient (CD45.21) 2W1S1CD4 cellsmid-gestation by Balb/c-2W1S males. c, Percentage FOXP31 among Treg-ablated Foxp3DTR/DTR post-partum donor (CD45.11) or naive recipient(CD45.21) 2W1S1CD4 cellsmid-gestation. Bars,means6 one standard error.

LETTER RESEARCH

0 0 M O N T H 2 0 1 2 | V O L 0 0 0 | N A T U R E | 3


CD45.1 CD4 cells from post-partum donor

CD45.2 recipient

Mate Balb/c-2W1S

CD

44

2W1S

CD

45.1

CD45.2

CD45.1

CD45.2

CD45.1

CD45.2

a

b

c

CD

45.1

CD45.2

CD

44

2W1S

60.1 19.3

16.3 9.1

CD

44

2W1S

CD

44

2W1S

FOXP

3

Helios

FOXP

3

Helios

VirginE11.5

020406080

Per

cent

age

FOXP

3+

Primary pregnancy

Secondary pregnancy

CD45.1 CD4 cells from Foxp3DTR/DTR post-partum donor

CD45.2 recipient


9.9

89.8

FOXP

3

Helios

26.9

FOXP

3

Helios

0

20.3

6.9

92.8

CD45.2CD45.1

0

20

40

60

80

CD45.2CD45.1

0

20

40

60

80

Virgin E11.5 Virgin E11.5

Post-partum E11.5 Post-partum E11.5

P = 0.006

P = 0.01

Post-partum

E11.50

20

40

60

80 P = 0.001

P = 0.009

Per

cent

age

FOXP

3+P

erce

ntag

e FO

XP3+

Per

cent

age

FOXP

3+

62.2

Tregs persist post-‐partum – Secondary pregnancy?

1 2

3 4

1 2

3 4

Fetal-‐specific FoxP3+ Tregs expand from las@ng FoxP3+ cells from prior pregnancy

B6 CD45.1 Aaer primary pregnancy

CD4+ cell isola@on (rich in fetal-‐specific FoxP3+ cells)

Transfer into congenic CD45.2 recipient

Secondary pregnancy

ablated in C

following stimulation, consistent with previously described anergy ofmaternal cells with fetal specificity (Supplementary Fig. 6)2,23. Therefore,tomore fully evaluate the responsiveness ofmaternal T cells with fetal-2W1S specificity, wemeasured their in vivo response to Listeriamono-cytogenes engineered to express the 2W1S55–68 peptide (Lm-2W1S)that potently stimulates TH1-differentiation in other contexts24,25. Wefound that 2W1S1 cells expand and upregulate T-bet (also known asTBX21) expression each in an antigen-specific fashion in both naivemice andmice impregnated by 2W1S-expressingmales after Lm-2W1Sinoculation, similar to other intracellular pathogens (SupplementaryFig. 7)26. Interestingly, however, 2W1S1 cells in pregnant mice where2W1S represents a surrogate fetal antigen produced only backgroundlevels of IFN-c and other effector cytokines, with reciprocal accumula-tion of FOXP31 cells (Fig. 3a and Supplementary Fig. 8). Comparatively,.15%of 2W1S1 cells inLm-2W1S-inoculated virginmicewere IFN-c1

(Fig. 3a). This hypo-responsiveness was specific to fetal-2W1S stimu-lation, because 2W1S1CD4 cells inmice impregnated with non-2W1S-expressing males produced IFN-c levels comparable to non-pregnantcontrols (Fig. 3a andSupplementary Fig. 8).Given the sustained enrich-ment of fetal-specific Treg cells after delivery (Fig. 1a, b), these studieswere extended to investigate whether diminished IFN-c productionamong maternal CD4 cells with specificity to pre-existing fetal antigenis similarly maintained. Remarkably, IFN-c production remainedanaemic in post-partum mice previously exposed to 2W1S as a fetalantigen, whereas post-partummice without prior fetal-2W1S exposureproduce IFN-c comparable to non-pregnant controls (Fig. 3b).Accordingly, pregnancy imprints functional anergy for maternal CD4cells with fetal specificity that is sustained post-partum.

To dissociate whether pregnancy-induced T-cell anergy was cell-intrinsic or imposed by features associated with the post-partum envir-onment, wemeasured IFN-cproduction byCD4 cells frompost-partumor virginmice after adoptive transfer intonaive recipientmice.We foundIFN-cproductionbydonorpost-partumandeach groupof naive (donorand recipient) 2W1S1 CD4 cells were similar, and notably increasedcompared with 2W1S1 cells in un-manipulated post-partum mice fol-lowing Lm-2W1S inoculation (Fig. 3b, c). Thus, anergy amongmaternalCD4 cells with specificity to pre-existing fetal antigen is not cell-intrinsic,but maintained by the post-partum environment.In complementary studies we addressed the importance ofmaternal

Treg cells in sustaining anergy to cells with specificity to pre-existingfetal antigen by investigating the effect of replacing the entire Tregcompartment in post-partum mice previously exposed to fetal-2W1Swith naive FOXP31 cells from virgin mice. Consistent with recentstudies using adoptively transferred Foxp3WT/WT CD4 cells to refillthe cellular compartment in Foxp3DTR/DTRmice sustained ondiphtheriatoxin treatment2, Treg cells fromnaivemice efficiently reconstitutedTreg-ablated Foxp3DTR/DTR post-partummice (Fig. 4a). Using this approach,we found that replacing maternal FOXP31 cells in post-partum micewith Treg cells from naive mice restored IFN-c production for 2W1S1

CD4 cells (Fig. 4b). Furthermore, whereas only rare 2W1S1FOXP31

cells that were Helioshi were found among post-partum mice recons-tituted with naive Treg cells, a significant proportion of 2W1S1 cellsexpanded in response to Lm-2W1S in intact post-partum miceremained FOXP31 (,20%) and Helioslo (,40%) (Fig. 4b). Thus, themuted expansion of naive FOXP31 CD4 cells with L. monocytogenesinfection is overcome by pregnancy-induced Treg activation (Fig. 4b and

Figure 2 | Accelerated expansion of maternal Treg cells with fetal specificityduring secondary pregnancy. a, Percentage FOXP31 among virgin (primarypregnancy) or post-partum (secondary pregnancy) females before mating ormid-gestation (E11.5) by Balb/c-2W1S males. b, Percentage FOXP31 among

post-partum donor (CD45.11) or naive recipient (CD45.21) 2W1S1CD4 cellsmid-gestation by Balb/c-2W1S males. c, Percentage FOXP31 among Treg-ablated Foxp3DTR/DTR post-partum donor (CD45.11) or naive recipient(CD45.21) 2W1S1CD4 cellsmid-gestation. Bars,means6 one standard error.

LETTER RESEARCH

0 0 M O N T H 2 0 1 2 | V O L 0 0 0 | N A T U R E | 3


CD45.1 CD4 cells from post-partum donor

CD45.2 recipient

Mate Balb/c-2W1S

CD

44

2W1S

CD

45.1

CD45.2

CD45.1

CD45.2

CD45.1

CD45.2

a

b

c

CD

45.1

CD45.2

CD

44

2W1S

60.1 19.3

16.3 9.1

CD

44

2W1S

CD

44

2W1S

FOXP

3

Helios

FOXP

3

Helios

VirginE11.5

020406080

Per

cent

age

FOXP

3+

Primary pregnancy

Secondary pregnancy

CD45.1 CD4 cells from Foxp3DTR/DTR post-partum donor

CD45.2 recipient


9.9

89.8

FOXP

3

Helios

26.9

FOXP

3 Helios

0

20.3

6.9

92.8

CD45.2CD45.1

0

20

40

60

80

CD45.2CD45.1

0

20

40

60

80

Virgin E11.5 Virgin E11.5

Post-partum E11.5 Post-partum E11.5

P = 0.006

P = 0.01

Post-partum

E11.50

20

40

60

80 P = 0.001

P = 0.009

Per

cent

age

FOXP

3+P

erce

ntag

e FO

XP3+

Per

cent

age

FOXP

3+

62.2

Where does great popula@on of an@gen-‐specific Tregs in secondary pregnancy originate from? (Memory pool or new expansion?)

DONOR – 2nd pregnancy

RECIPIENT – 1st pregnancy

DONOR

RECIPIENT









RESEARCH LETTER

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 2


a b

IFN

-γ

T-bet

CD

44

2W1S


13.6 12.0 1.6 1.4 16.2

T-bet

CD

44

2W1S

Balb/c-2W1S Balb/c


10203040

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Balb/c-2W1S Balb/c0

10

20

30

CD

44

2W1S

IFN

-γ

T-bet

CD

45.1

CD45.2

Lm-2W1S


CD

90.2

CD90.1

6.7

91.5

18.9

13.7

86.2

16.3

15.8

Naive (recipient)

Naive (donor)

Post-partum (donor)

c

P = 0.003 P = 0.004 P = 0.001

CD45.1

CD45.2 CD90.2

CD45.2 CD90.10

10

20

30

2W1S

+ ce

lls

P = 0.08 P = 0.98

IFN

-γP

erce

ntag

e IF

N-γ

+

2W1S

+ ce

lls

CD45.1 recipient

Per

cent

age

IFN

-γ+


a

b

CD45.2 naive donor

Sustained DT

CD

45.1

CD45.2

FOXP

3

Helios

14.8

84.2

5.7

93.9

96.1

3.8

Foxp3+

FOXP3–

FOXP

3

Helios

CD

44

2W1S

IFN

-γ

T-bet

23.5

0.8 17.5



Intact

Reconstituted

0

25

50

75

100

Intact

Reconstituted

0

10

20

30

40

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Intact

Reconstituted

101

102

103

104

2W1S

+ FO

XP3+

cells

per

mou

se

P = 0.003 P = 0.02 P = 0.003

Foxp3

WT/WT

Foxp3

DTR/WT

Foxp3

DTR/DTR0

20

40

60

80

100

Per

cent

age

reso

rbed

fetu

s


P = 0.02

P = 0.002

P = 0.95 c

Per

cent

age

Hel

ioshi

Anergy of fetal-‐specific pregnancy-‐induced Tregs

Th1

LM Listeria monocytogenes

Model: Virgin x Balb/c-‐2W1S Balb/c

Treg

LM Listeria monocytogenes

Th1

LM

Th1

Hyporesponsiveness of fetal-‐specific CD4+ cells persists aaer pregnancy! (b)

2W1S+ cell

Infected mid-‐gesta-on Or post-‐partum

Infected mid-‐gesta-on Or post-‐partum

Responsiveness of fetal-‐specific cells:

Analysis pep-de-‐specific response

X

?









RESEARCH LETTER

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 2


a b

IFN

-γ

T-bet

CD

44

2W1S


13.6 12.0 1.6 1.4 16.2

T-bet

CD

44

2W1S

Balb/c-2W1S Balb/c


10203040

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Balb/c-2W1S Balb/c0

10

20

30

CD

44

2W1S

IFN

-γ

T-bet

CD

45.1

CD45.2

Lm-2W1S


CD

90.2

CD90.1

6.7

91.5

18.9

13.7

86.2

16.3

15.8

Naive (recipient)

Naive (donor)

Post-partum (donor)

c

P = 0.003 P = 0.004 P = 0.001

CD45.1

CD45.2 CD90.2

CD45.2 CD90.10

10

20

30

2W1S

+ ce

lls

P = 0.08 P = 0.98

IFN

-γP

erce

ntag

e IF

N-γ

+

2W1S

+ ce

lls

CD45.1 recipient

Per

cent

age

IFN

-γ+


a

b

CD45.2 naive donor

Sustained DT

CD

45.1

CD45.2

FOXP

3

Helios

14.8

84.2

5.7

93.9

96.1

3.8

Foxp3+

FOXP3–

FOXP

3

Helios

CD

44

2W1S

IFN

-γ

T-bet

23.5

0.8 17.5



Intact

Reconstituted

0

25

50

75

100

Intact

Reconstituted

0

10

20

30

40

Per

cent

age

IFN

-γ+

2W1S

+ ce

lls

Intact

Reconstituted

101

102

103

104

2W1S

+ FO

XP3+

cells

per

mou

se

P = 0.003 P = 0.02 P = 0.003

Foxp3

WT/WT

Foxp3

DTR/WT

Foxp3

DTR/DTR0

20

40

60

80

100

Per

cent

age

reso

rbed

fetu

s


P = 0.02

P = 0.002

P = 0.95 c

Per

cent

age

Hel

ioshi

B6 CD45.2 naïve no exposure to 2W1S CD4+ transfer T reg pool replenished by donor CD4+?

Ablated (via DT in FoxP3DTR/DTR) CD45.1 recipient FoxP3DTR/DTR

Fetal-‐specific FoxP3+CD4+ cells (from 1st preg)

Are these naïve (as opposed to fetal-‐specific) Tregs responsive to infec@on?

Replacement of post-‐partum fetal-‐specific FoxP3+ CD4+ by naïve CD4+ cells

Naïve CD4+ FoxP3+ Tregs are responsive to infec@on.

Discussion:

SUPPLEMENTARY INFORMATION

1 0 | W W W. N A T U R E . C O M / N A T U R E

RESEARCH

Supplementary Figure 10. Models comparing maternal regulatory CD4 cell accumulation. In model 1 (top), pregnancy stimulates non-specific expansion of maternal FOXP3+ regulatory T cells before investigation using antigen-specific tools. In model 2 (bottom), maternal regulatory T cells with fetal-antigen specificity selectively expand and accumulate during pregnancy. This model is supported by data presented in this paper where maternal regulatory CD4 cells with specificity for a single peptide antigen expressed by the developing fetus are found to expand greater than 100-fold, while FOXP3 expression among bulk maternal CD4 cells accumulate less than twofold. Furthermore, we show pregnancy-induced maternal regulatory T cells with fetal specificity are pheno-typically distinct, persist after delivery, and rapidly re-expand and provide protection from fetal resorption during secondary pregnancy.

•  during pregnancy a pool a fetal-‐specific Tregs accumulate in the dam

•  accumula@on via prolifera@on and conversion of CD4+ cells

•  upregulate CD44 and a propor@on is Heliosneg

•  are anergic

•  cons@tute a memory pool that expands greatly in a secondary pregnancy

Th17 cells in inflamma-on AND homeostasis

•  Th17 cells produce IL-‐17A, IL-‐17F, IL-‐21 and IL-‐22 and TNFα

•  Th17 cells in the intes@ne are almost undetectable in germ-‐free mice

microbiota induce Th17 differen@a@on

–  SFB can specifically induce Th17 (Ivanov, Cell 2009)

–  SFB also specifically induce IgA (Talham, Infect.Immun. 1999 and Klaasen, Infect.Immun. 1993)

•  LINK between Th17 and IgA?

•  Inflamma@on vs. Homeostasis

–  Homeostasis via IL-‐17-‐driven induc-on of pIgR and IgA secre-on

IL-‐17R-‐/-‐ mice display lower levels of IgA and pIgR

Fecal content ELISA

Small intes-ne

qPCR of Pigr Whole -ssue?

Large intes-ne

IL-‐17 regulates pIgR independent of microbiota

B6 and IL-‐17R-‐/-‐ mice were cohoused from 3 weeks of age experiment at 8weeks

TCRβ/δ-‐/-‐ mice display lower levels of IgA and pIgR

Fecal content ELISA

qPCR of Pigr in „intes-nal -ssues“ 30d post-‐transfer

T-‐cell dependent induc@on of IgA is the prominent pathway of IgA iduc@on

TCRβ/δ KO

Th17

DONOR:

Spleen CD4 MACS Culture for 10d with irradiated splenic APCs S-m with PMA/Iono Enrich for 45.2+ and IL-‐17A+ FACS sorted

CBir1 transgenic mice flagellin (present in recipient) OTII transgenic mice OVA (not present in recipient)

Can in vitro differen@ated T cells restore IgA and pIgR?

Cell number???

Fold change rela-ve to TCRb/d/-‐/-‐ without transfer

Fecal content ELISA

Can microbiota-‐specific Th17 cells rescue IgA and pIgR?

Spleen, CD4 MACS BirC transgenic (microbiota-‐specific Th17 cells) OTII transgenic

Culture for 5 days with irradiated APCs

10ng/ml TGF-‐b 20ng/ml IL-‐6 10mg/ml an@-‐IFNg 10mg/ml an@-‐IL-‐4

Res@mula@on with PMA/Iono Enrichment: CD45.2, IL-‐17A Sorted

i.V transfer (cell number???) of in-‐vitro differen@ated Th17 cells

Recipient: TCRβ/δ/-‐/-‐

qPCR of fecal contents

Polarized Th17 cells induce IgA produc@on in vitro

ELISA of culture supernatant d5

T cells B cells IgD+ splenic

CBir1

Culture for 5 days

Th0, Th1, Th17-‐condi@ons CBir1 CD4 OTII CD4

The effect of IgA induc@on is dependent on Th17 effector cells

HT29

Rela-ve to untreated cells

+ 10ng/ml TNF-‐a + 20ng/ml IL-‐17A

HT29

+ 10ng/ml TNF-‐a + 20ng/ml IL-‐17A

+ NFkB inhibitor + PI3K inhibitor for 1 h

HT29

Epithelial cells’ response to IL-‐17

Signalling response to IL-‐17

pigr

In vivo: In the absence of IL-‐17 signalling mice develop a more severe coli@s

Chronic coli-s 1.75% DSS in drinking water

1 ip injec-on of an-body 7d 2% DSS IL-‐17R-‐/-‐ Neutralizing IL-‐17 an@body

IgA and pIgR expression under healthy condi@ons correlates with suscep@bility to DSS-‐induced weight loss

n=4/group

DiscussionDespite enormous bacterial challenge, the host intestine establishesa mutualistic relationship with the microbiota. Multiple mecha-nisms have evolved to regulate this relationship. The intestinal tracthas been shown as a natural site for Th17 cell development, whichis stimulated by specific species of microbiota (14), with SFB beingrecently identified as one such stimulator (13). Although bothproinflammatory and anti-inflammatory functions of Th17 havebeen demonstrated in different experimental systems (4–8), theenrichment of Th17 cells in the intestine suggests a role for thesecells in mucosal homeostasis and more specifically in the con-tainment of the vast local microbiota. In consistency with thisargument, our data demonstrated that Th17 cells are able to pro-mote intestinal IgA secretion via induction of epithelial cell pIgRexpression, thereby contributing to the maintenance of host im-mune homeostasis to microbiota.One of the most important strategies to generate immune pro-

tection and maintain intestinal homeostasis is the production ofIgA (9), which is the primary Ab in the gut. IgA regulates themicrobiota, and bacteria in turn adapt to IgA by altering their gene

expression patterns (38). Although IgA also plays a role in hostresistance to infection, it has been argued that the major role ofIgA in the intestine is in maintaining the balance between the hostand its microbiota (39). In the absence of pathogen exposure,specific pathogen–free mice have abundant levels of IgA, whereasgerm-free mice have very low levels of IgA (9). B cell IgA pro-duction can be stimulated by dendritic cell–B cell or epithelialcell–B cell interactions via BAFF, APRIL, inducible NO synthase,and TLR ligands, or utilizing T cell help and a number of cyto-kines including TGF-b, IL-4, IL-6, and IL-10 (10). Although therelative contribution of T cell-dependent and -independent regu-lation to intestinal IgA production is still not completely under-stood, decreased levels of intestinal IgA in T cell-deficient TCR-bxd2/2 mice compared with wild type mice indicates a predomi-nant role of the T cell-dependent pathway (20, 39). However, it isstill unclear which T cells provide help and which sources ofcytokines are needed for intestinal IgA production in the mucosa.

FIGURE 5. IL-17R2/2 mice suffer worsened colitis as a result of de-creased pIgR and IgA secretion. (A) Age-matched wild type and IL-17R2/2

mice, which had been cohoused from 3 wk old, were administered 1.75%DSS in drinking water. Weight was measured every 2 d. After 7 d of DSS,drinking water was replaced with fresh water for 3 d, and the cycle wasrepeated over 60 d. Weights are shown as a percentage of individual weighton day 0. Significant differences are compared between strains on DSS.*p , 0.05 compared with wild type mice; n = 4 mice per group. (B)Colonic histopathology of DSS-treated mice after 60 d of DSS adminis-tration. H&E, original magnification 310. (C) IgA and pIgR in fecalpellets were quantified from mice by ELISA before DSS administrationand plotted against their individual body weight after 54 d.

FIGURE 6. Blockade of IL-17 induces more severe colitis from DSSadministration, and bacterial translocation is increased in IL-17R2/2 mice.(A) C57BL/6 mice were injected i.p. with a neutralizing Ab to IL-17A, orisotype control, and administered DSS for 7 d. Weights are shown asa percentage of body weight on day 0. (B) Pathologic score of colitis wasexamined by blind histologic observation 10 d after DSS administration.**p , 0.01 compared with the mice treated with control mAb; n = 4 miceper group. (C) Colonic histopathology of the DSS-treated mice after 10 dof DSS administration. H&E, original magnification 310. (D) Mesentericlymph nodes were harvested from wild type or IL-17R2/2 mice underaseptic conditions. MLN homogenates were cultured onto blood agarplates and incubated in aerobic and anaerobic conditions at 37˚C. Aerobiccultures were incubated overnight; anaerobic cultures were incubated for3 d. *p , 0.05 compared with wild type mice; n = 3 mice. (E) Serum IgGagainst commensal bacterial lysate were quantified from wild type orIL-17R2/2 mice by ELISA. Wild type mice were injected i.v. with 200 mgA4 bacteria to indicate relative amount of serum IgG. *p , 0.05 comparedwith wild type mice; n = 4 mice per group.

6 Th17 REGULATION OF INTESTINAL IgA

at Universitaetsspital B

iblio Bern on Septem

ber 28, 2012http://jim

munol.org/

Dow

nloaded from

Increased bacterial transloca@on with systemic priming and commensal-‐specific IgG response

MLN

Serum

200µg A4 bacteria i.v

Summary

•  TCRβ/δ-‐/-‐ and IL-‐17R-‐/-‐ mice display lower levels of fecal IgA and (secreted) pIgR •  IgA and pIgR can be induced upon transfer of microbiota-‐specific Th17 cells •  In vitro:

•  Th17 cells can induce IgA switch •  IL-‐17 signals via NF-‐κB and PI3K on epithelial cells to induce pIgR •  IL-‐17 can act in synenrgy with TNF-‐α

•  DSS: •  fecal IgA prior to DSS can determine suscep@bility to DSS-‐induced weight loss •  IL-‐17R-‐/-‐ are more suscep@ble to DSS, display bacterial transloca@on to the MLNs and systemic IgG priming against commensals

Although TGF-b has been shown as a crucial cytokine in pro-moting IgA class switching (10), and Treg production of TGF-bgreatly contributes to intestinal IgA production (20), it cannotcompletely explain why high levels of IgA are present only in theintestine, but not other lymphoid tissues even though TGF-b arealso present in those sites. Our data indicated that repletion ofTh17 cells promoted intestinal IgA secretion in the TCR-bxd2/2

mice. Blockade of Th17 cytokine IL-17 decreased intestinal IgA(Fig. 2). In addition, IL-17R deficiency resulted in lower intestinalIgA secretion compared with wild type mice (Fig. 1), indicatingthat Th17 cells and their signature cytokine IL-17 greatly con-tribute to intestinal IgA secretion. Promotion of IgA secretion isnot due to Tregs that were converted from Th17 cells, because theintestinal tissues produced TGF-b at a similar level. Several typesof innate cells have been identified recently in the intestines thatcould also provide sources of IL-17 to promote intestinal IgAproduction (40–42). Indeed, a previous report showed thatRORgt+ LTi cells but not RORgt+ CD4+ T cells induced T cell-independent LP IgA production in the absence of Peyer patches(43). In RORgt-deficient mice, transfer of RORgt+ LTi cells in-duced isolated lymphoid follicle (ILF) formation as well as LPIgA. However, transfer of RORgt+ CD4+ T cells did not induceILF or PP formation, nor intestinal IgA, indicating that in theabsence of PP and ILF, Th17 cells would not be activated and thuswould not produce cytokines required for induction of intestinalIgA. Several recent studies demonstrated that communal micro-biota greatly affect intestinal Treg, Th17 cell, and IgA responses.SFB preferably induces intestinal Th17 cells (13) and IgA (12, 13),whereas colonization with Clostridium species and Schaedler flora,which contain eight known commensal bacteria including Clos-tridium, induces Tregs (44, 45). Interestingly, failure to activateTregs results in the induction of Th17 cells; therefore, commensalbacteria regulate the balance between Tregs and Th17 cells. AsTregs have been shown to promote intestinal IgA response (20), andwe now show that Th17 cells are also able to upregulate intestinalIgA, the microbiota greatly influence intestinal IgA responses atleast partially through regulation of Tregs and Th17 cells.IgA translocation across the intestinal epithelium is mediated

by the pIgR (9). IgA function in the intestinal lumen is dependenton pIgR expression, and reduction in pIgR expression has beenshown to lead to decreased IgA-mediated protection against lu-minal Ags (15). Intestinal pIgR expression was lower in TCR-bxd2/2 mice compared with wild type mice, indicating a role forT cells in the induction of pIgR (Fig. 2). Consistent with a previ-ous report describing IL-17–mediated pIgR expression in airwayepithelial cells (19), repletion of Th17 cells restored intestinalpIgR expression in TCR-bxd2/2 mice, and IL-17R deficiency re-sulted in lower expression of intestinal pIgR, demonstrating thatTh17 and IL-17 signaling regulate intestinal epithelial pIgR ex-pression. Indeed, treatment with IL-17 greatly increased HT-29epithelial cell expression of pIgR, alone or synergistically withTNF-a. IL-17 was able to activate NF-kB p65 signaling in in-testinal epithelial cells (Fig. 4). Blockade of NF-kB signaling andPI3 kinase activity with selective chemical inhibitors inhibited IL-17 induction of pIgR. Interestingly, both pathways work inde-pendently in IL-17 signaling as the inhibition of either pathwaydid not result in strong abrogation of PIGR transcription; onlyblockade of both pathways resulted in significant downregulationof PIGR mRNA. Intestinal Th17 cells require cognate luminalAg stimulation to produce effector cytokines. Once cytokines areproduced by the activated T cells, they regulate intestinal IgAproduction in an Ag-nonspecific manner.Both intestinal pIgR and IgA have been implicated in mainte-

nance of intestinal immune homeostasis, because deficiency of

either pIgR or IgA results in greater commensal bacterial trans-location across the intestinal epithelium and more severe intestinalinflammation in response to DSS (17, 18, 36). Thus, Th17 cellregulation of intestinal pIgR and IgA could have a crucial rolein protection against intestinal inflammation induced by mucosalbreach by commensal flora. Indeed, there was higher level ofsystemic anti-commensal bacterial IgG in IL-17R2/2 mice but notin wild type mice (Fig. 6E), which is indicative of the presence ofcommensal bacteria in the systemic immune system. This findingrevealed that deficiency of IL-17 signaling resulted in morecommensal bacterial translocation from lumen, and sequentially,to more severe intestinal inflammation in response to DSS (Fig. 5).Consistent with these observations, we also found higher numbersof bacteria in the mesenteric lymph nodes of IL-17R2/2 mice(Fig. 6D). This finding is likely due to impaired intestinal pIgRexpression and IgA secretion, although the induction of a numberof cytokines and antimicrobial peptides from epithelial cells byIL-17 could also contribute to IL-17–mediated protection againstintestinal inflammation. However, we cannot exclude the possi-bility that wild type and IL-17R2/2 mice may have differences inthe composition of their respective gut microbiota, which couldhave contributed to our results.In summary, our data demonstrate that enriched microbiota Ag-

specific Th17 cells protect the host from chronic inflammation andcontribute to intestinal immune homeostasis by regulating epi-thelial pIgR expression, thereby promoting intestinal IgA. How-ever, it certainly does not mean that this is the only function of Th17cells that contributes to intestinal immune homeostasis, becauseTh17 cells and IL-17 have been shown to stimulate a number ofcytokines and antimicrobial peptides that also contribute to theregulation of host immune responses to microbiota (33). Tregs havebeen shown to greatly promote intestinal IgA production via di-rectly promoting B cell IgA class switching through production ofTGF-b. We now show that Th17 cells promote IgA translocationacross the intestinal epithelium via induction of pIgR by IL-17.Thus, Tregs and Th17 cells coordinately regulate intestinal IgAproduction and secretion (Fig. 7). A deficiency in either pathway

FIGURE 7. Coordinate regulation of intestinal IgA production and se-cretion by Treg and Th17 cells. TGFb produced by Treg cells drives naiveB cells to differentiate into IgA-producing cells. IL-21 from Th17 cellsaccentuates the effect of TGFb and increases IgA+ B cell differentiation.Polymeric IgA then binds to pIgR expressed on intestinal epithelial cells,causing transcytosis of pIgR-bound pIgA, and the IgA complex is secretedinto the lumen as sIgA. IL-17 from Th17 cells increases pIgR expressionfrom IECs and increases the rate of sIgA secretion into the lumen.

The Journal of Immunology 7

at Universitaetsspital B

iblio Bern on Septem

ber 28, 2012http://jim

munol.org/

Dow

nloaded from

Model:

Th17 cells can induce pIgR and IgA, thereby contribu@ng to homeostasis

Journal(club( - Mucosal Immunology · LETTER doi:10.1038/nature11462 Pregnancy imprints regulatory memory that sustains anergy to fetal antigen Jared H. Rowe1, James M. Ertelt1{,

Documents