Th17 cells give rise to Th1 cells that are required forthe pathogenesis of colitisStacey N. Harboura, Craig L. Maynarda, Carlene L. Zindla, Trenton R. Schoebb, and Casey T. Weavera,1

aDepartment of Pathology and bDepartment of Genetics, University of Alabama at Birmingham, Birmingham, AL 35294

Edited by Emil R. Unanue, Washington University in St. Louis, St. Louis, MO, and approved April 24, 2015 (received for review August 13, 2014)

Th17 cells reactive to the enteric microbiota are central to thepathogenesis of certain types of inflammatory bowel disease.However, Th17 cells display substantial developmental plasticity,such that some progeny of Th17 cell precursors retain a pre-dominantly IL-17A+ phenotype, whereas others extinguish IL-17expression and acquire expression of IFN-γ, giving rise to “Th1-like” cells. It remains unclear what role these subsets play in in-flammatory bowel disease. Using a Th17 transfer model of colitis,we found that IFN-γ–deficient Th17 cells retained an IL-17A+ phe-notype and were unable to induce colitis in recipients. Develop-ment of disease required the transition of a subset of Th17 pre-cursors to Th1-like cells and was contingent on the expression ofboth Stat4 and T-bet, but not the IL-12 or IFN-γ receptors. More-over, Th17 cells could provide “help” for the development ofpathogenic Th1 cells from naïve precursors. These results indicatethat Th17 cells are potent mediators of colitis pathogenesis bydual mechanisms: by directly transitioning to Th1-like cells andby supporting the development of classic Th1 cells.

Th17 | Th1 | colitis

The discovery that Th17 cells are products of a developmentalpathway distinct from that of Th1 and Th2 cells has had a

major impact on our understanding of immune regulation anddysregulation in the intestines (1). Before this, a commonly heldview was that Th1 cells were the principals in inflammatorybowel disease (IBD) pathogenesis (2–5). Subsequently, however,numerous studies have shown that IL-23, a Th17 pathway cyto-kine, is required for development of IBD (6–8), as is expressionof the IL-23 receptor by CD4+ T cells (9). Moreover, disease-association studies in humans have implicated a number of genesthat are Th17-specific (10), and Th17 cells are now thought to bethe principal effector cell in IBD.Despite this, our own and other studies have found a mix of

IL-17A+ and IFN-γ+ cells during active intestinal inflammationin T-cell transfer models of colitis in mice (9, 11, 12), consistentwith human data showing that CD4 T cells from the diseasedintestines of patients with Crohn’s disease express IL-17A, IFN-γ,or both (13). Accordingly, although the Th17 pathway appears tobe central to IBD pathogenesis, T cells expressing the signatureTh1 lineage cytokine, IFN-γ, are consistently associated with theinflammatory lesions of colitis. Furthermore, there remains a strongassociation of disease with the Th1 pathway in mouse models.Antibody-mediated neutralization of IFN-γ inhibited the devel-opment of colitis in early studies of the CD45RBhi transfer modelof colitis (3), as did transfers of naïve T cells deficient for IFN-γ (4)or the Th1 lineage transcription factors, T-bet and Stat4 (5, 14).A reconciliation of these findings has been suggested by dis-

covery that Th17 cells retain significant late developmentalplasticity (1, 11). Specifically, developing Th17 cells can divergeto acquire Th1-like features contingent on IL-12 or IL-23 sig-naling, both of which activate Stat4 and induce expression ofT-bet that represses expression of the signature Th17 transcrip-tion factor, RAR-related orphan receptor (ROR)γt, and leads toextinction of Il17a and Il17f expression and induction of Ifngexpression (11, 15, 16). Thus, in addition to production of IL-17Aand IL-17F, the Th17 pathway can also give rise to “Th1-like” cells

that produce IFN-γ. Accordingly, we found that Th17 cells isolatedusing an IL-17F reporter transgene induced severe colitis that wasassociated with recovery of Th1-like cells derived from the Th17precursors (11). In view of findings that T cells deficient for IL-17Ainduced colitis that is indistinguishable from (17, 18) or more severethan controls, and this disease was associated with increased fre-quencies of Th1-like cells (19), these results are consistent with apathogenic role for Th1-like cells that emerge from Th17 precursors.Here, we examined mechanisms by which Th1-like cells arise

from Th17 precursors in vivo and their possible contribution tocolitis pathogenesis. We found that IFN-γ production by theprogeny of Th17 cells was required for disease development in atransfer model of Th17-driven colitis: this was T-bet– and, to alesser extent, Stat4-dependent. Moreover, Th17 cells unable toproduce IFN-γ promoted the development of colitis from naïveprecursors deficient in RORγt, establishing Th17-dependentsupport for the differentiation of pathogenic Th1 cells that donot develop via the Th17 pathway.

Results and DiscussionIFN-γ Produced by Th17 Cells Is Indispensible for Colitis Induction.Wehave reported a Th17 transfer model of colitis that enablesstudy of Th17 cell plasticity in the pathogenesis of intestinalinflammation (11). In this model, Th17-polarized cells aregenerated from naive CD4+ T cells of IL-17F reporter mice(Il17fThy1.1) and sorted on the basis of Thy1.1 (IL-17F) expression.IL-17F is the dominant IL-17 family cytokine produced by Th17cells early in development and is thus an ideal marker for Th17commitment. Importantly, Thy1.1+ Th17 cells do not expressIFN-γ after in vitro polarization (11). Transfers of isolatedIL-17F+ Th17 precursor (Th17p) cells into recombination activat-ing gene (RAG)1-deficient recipients results in colitis that is of

Significance

The Th17 subset of CD4+ T cells are important in the patho-genesis of inflammatory bowel disease (IBD), but the mecha-nisms of their actions, particularly the role of the developmentof IFN-γ–producing progeny of Th17 cells (Th1-like cells), areincompletely understood. Here, we show in a mouse model ofTh17-driven IBD that transition of Th17 precursors to Th1-likecells is absolutely required for disease, because Th17 cells de-ficient in IFN-γ fail to induce intestinal inflammation. Thistransition is dependent on the transcription factors T-bet and,to a lesser extent, Stat4. These findings are relevant for clinicalstrategies that target IBD and suggest that focusing on boththe Th17 and Th1-like arms of disease may be beneficial intherapy design.

Author contributions: S.N.H. and C.T.W. designed research; S.N.H., C.L.M., and C.L.Z. per-formed research; S.N.H., C.L.M., C.L.Z., T.R.S., and C.T.W. analyzed data; and S.N.H. andC.T.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed Email: cweaver@uab.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1415675112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1415675112 PNAS | June 2, 2015 | vol. 112 | no. 22 | 7061–7066

IMMUNOLO

GYAND

INFLAMMATION

comparable severity, but accelerated onset, compared with thatcaused by transfers of CD45RBhi CD4+ T cells (11). Impor-tantly, T cells recovered following the onset of colitis in theTh17 transfer model have diverged into distinct subsets, suchthat some progeny of transferred Th17p cells retain expressionof IL-17A, with or without coexpression of IFN-γ, whereasothers silence expression of IL-17A and IL-17F while acquiringexpression of IFN-γ.To probe the role of Th17 cell-derived IFN-γ in the de-

velopment of colitis, we examined the pathogenic potential ofTh17 precursors derived from IFN-γ–deficient IL-17F reportermice. CD4+ T cells from Il17fThy1.1 (WT) or Ifng−/−.Il17fThy1.1

(Ifng−/−) were polarized ex vivo under Th17-inducing conditionsusing Il12b−/− splenic feeder cells to ensure absence of IL-23 andIL-12 (Fig. 1A and Fig. S1A). A large percentage of cells expressedThy1.1 (IL-17F), and the majority of cells that expressed IL-17Awere a subset of those expressing IL-17F (Fig. S1B) (11). Polarizedcells were sorted for Thy1.1 expression (Fig. S1C) and transferredinto Rag1−/− recipients.Transfers of Th17p cells from WT mice induced substantial

disease, indicated by significant weight loss (Fig. 1B) as well assevere colonic inflammation (Fig. 1 C and D) by 4 wk aftertransfer. In contrast, Th17p cells from Ifng−/− mice failed to in-duce wasting disease (Fig. 1B) or significant colonic inflamma-tion (Fig. 1 C and D). Thus, production of IFN-γ by Th17p cellsis essential for development of Th17-mediated colitis.Flow cytometric analysis of T cells recovered from the colonic

lamina propria (CLP) and mesenteric lymph nodes (MLN)showed that four distinct populations of cells emerged from re-cipients of WT Th17p cells (Fig. 1 E and F). A minor fraction ofrecovered cells retained IL-17A expression alone, with a similarpercentage expressing both IL-17A and IFN-γ. Notably, a largepercentage of recovered cells expressed IFN-γ alone, despiteundetectable IFN-γ in the sorted cells before transfer (Fig. S1A).This finding demonstrates that many Th17p cells have transi-tioned in vivo to become IFN-γ–producing Th1-like cells. Incontrast, T cells recovered from the MLN and CLP of recipientsof IFN-γ–deficient Th17 p displayed increased frequencies ofIL-17A–expressing cells, with an absence of cells expressing IFN-γor both cytokines (Fig. 1 E and F). Increased frequencies ofIL-17A+ cells in these recipients were also associated with anincrease in cells expressing Thy1.1 (IL-17F), IL-22, and GM-CSFcompared with WT recipients (Fig. S2B) but not IL-10, whichwas extinguished in both groups. Notably, because there weresignificantly decreased total CD4 T cells recovered from thecolons of these mice, only the number of IL-22 and Thy1.1 cellswere significantly higher than WT recipients, meaning that theabsolute numbers of IL-17A–expressing cells was not signifi-cantly different from WT controls in which there was markedcolonic inflammation (Fig. 1G). Thus, the absence of disease inrecipients of IFN-γ–deficient Th17p cells could not be attributedto a deficit of colonic IL-17A+ Th17 cells.In view of the requirement for IFN-γ expression by transferred

Th17p cells for disease induction, we determined whether IFN-γmight act to enhance the transition of Th17p to Th1-like cells asa mechanism to promote colitis. Transfer studies were performedusing Th17p cells that could, or could not, respond to IFN-γ.Th17p cells from Ifngr1−/−.Il17fThy1.1 (Ifngr1−/−) mice inducedless-severe disease in Rag1−/− recipients, as evidenced by reducedweight loss (Fig. S3A) and histopathology (Fig. S3 B and C)compared with Il17fThy1.1 (WT), but these results did not ach-ieve statistical significance, indicating that actions of IFN-γ ontransferred Th17 cells were of lesser import than its actions onhost cells. T cells recovered from the CLP of Ifngr1−/− recipientshad significantly higher expression of IL-17A compared withWT Th17p recipients (Fig. S3D), suggesting that IFN-γ signalingin Th17 cells inhibits IL-17A production by cells in the colonicmucosa. Interestingly, this pattern did not repeat in the MLN,

suggesting that the actions of IFN-γ might be relatively re-stricted to the intestinal mucosa or that the actions of IFN-γ onTh17 cells promote their retention or localization in the mu-cosa. Despite their deficiency in IFN-γ signaling, recoveredCD4+ cells from CLP of Ifngr1−/− Th17p recipients expressedcomparable levels of T-bet in all cytokine-producing fractions(Fig. S3 F and G), suggesting that other signals, such as type Iinterferons, might partially compensate for the lack of IFN-γsignaling (20).The foregoing findings indicated that transferred Th17p cells

were not the principal targets of the actions of IFN-γ that con-tribute to colitis. Given that IFN-γ acts on a broad range of cells(21, 22), we extended these findings by examining the effect ofneutralizing antibodies to IFN-γ in recipients of Th17p cells thatlacked expression of the IFN-γ receptor. Consistent with theprevious results, administration of neutralizing anti–IFN-γ mAb

WT0

20

40

60WT

Thy1.1 (IL-17F)

IL-1

7A

***

WT

*

0 1 2 3 480

90

100

110

120

WT

IL-17A+ IFN- - IL-17A+ IFN- + IL-17A- IFN- +

CLP MLNCyt

okin

e po

s. c

ells

(%

tota

l CD

4+)

IL-1

7A

WTIfng -/-

Ifng -/-

Ifng -/-

Ifng -/-

0

10

20

30 **

012345 * **

0

10

20

30

40 ****

00.20.40.60.81.0

IL-17A+IFN- +

IL-17A+IFN- -

IL-17A-IFN- +

Cyt

okin

e po

s.ce

lls (n

orm

aliz

ed to

WT)

* *

WT

Ifng-/-

8.7 11.9

32.4

9.2 13.2

30.2

4.3 3.5

34.7

IFN-

Ifng -/-

25.5 1.2

1.4

CLP MLN CLP MLN

% in

itial

bod

y w

eigh

t

Weeks post transfer

Sco

re

A B

C D E

F

G

Fig. 1. Th17-derived IFN-γ is required for Th17-induced colitis. (A) Expres-sion of IL-17A and IL-17F (Thy1.1) in CD4+ T cells from Il17fThy1.1 (WT) andIfng.Il17fThy1.1 (Ifng−/−) cultured with Il12b−/− splenic feeder cells under Th17polarizing conditions (2.5 μg/mL anti-CD3, 2.5 ng/mL TGF-β, 10 ng/mL IL-6,10 μg/mL anti-IFN-γ, 10 μg/mL anti–IL-4) for 7 d. FACS plots are gated on CD4+

cells. (B) Th17 precursor (Th17p) cells from WT or Ifng−/− mice obtained in Awere transferred into Rag1−/− mice (4 × 105 cells per mouse). Weight loss isexpressed as a percentage of starting weight (mean ± SEM). (C) Represen-tative H&E-stained sections of colon from recipients of WT (Upper) or Ifng−/−

Th17p cells (Lower) 4 wk after transfer. H&E stain; original magnification10×. (D) Histological scoring (mean + SEM). (E) Representative FACS plotshowing IL-17A and IFN-γ expression in CLP CD4+ T cells ex vivo after de-velopment of colitis. (F) Relative frequencies of IL-17A+, IL-17A+IFN-γ+, andIFN-γ+ CD4+ T cells in the CLP and MLN of colitic mice (mean ± SEM). (G) Totalnumbers of indicated subsets of CD4+ T cells recovered from CLP of re-cipients of Ifng−/− Th17p cells, expressed as a fraction relative to WT Th17precipients. Data are representative of three independent experiments withfive mice per group. *P < 0.05, **P < 0.01, ***P < 0.001.

7062 | www.pnas.org/cgi/doi/10.1073/pnas.1415675112 Harbour et al.

significantly ameliorated disease in Rag1−/− mice that receivedtransfers of Ifngr1−/− Th17p cells compared with those that re-ceived an isotype control mAb (Fig. S3H).Collectively, these data established that production of IFN-γ

by Th1-like cells that arose from Th17 precursors was requiredfor the development of colitis. Production of IFN-γ by innateimmune cells of Rag1−/− recipients, which were IFN-γ competent,did not complement the IFN-γ deficiency in transferred Th17cells. Although IFN-γ signaling had a detectable impact on thedevelopment of Th17 cells posttransfer, this did not significantlyaffect disease development. These results indicate that the prin-cipal targets of IFN-γ that induce disease are non-T cells.

Stat4, but Not IL-12, Signaling in Th17 Cells Is Required for Colitis.IL-12 signaling via Stat4 is critical for the second “wave” of Th1differentiation, which is believed to sustain and stabilize the Th1phenotype (23). Although not required for Th17 development(24, 25), we have reported that IL-12 can enhance IFN-γ pro-duction from in vitro-derived Th17 cells in a Stat4-dependentmanner (11). However, IL-23, not IL-12, has been linked toimmune-mediated colitis (9, 26), and the IL-23 driven divergenceof Th17p cells to a Th1-like phenotype in vitro is Stat4-dependent(11, 15). Moreover, Th17 cells that develop in vivo lack expressionof Il12rb2 (27), suggesting that the pathway to IFN-γ productionby Th17 cells in vivo is contingent on Stat4 activation downstreamof IL-23, not IL-12, signaling.To directly examine the roles of IL-12 and Stat4 in Th17-derived

IFN-γ production and colitis development, IL-12Rβ2– and Stat4-deficient IL-17F reporter mice were generated. Th17p cells derivedfrom Il17fThy1.1 (WT), Il12rb2−/−.Il17fThy1.1, and Stat4−/−.Il17fThy1.1

have comparable cytokine phenotypes (Fig. 2A). Isolated Thy1.1+

Th17p cells were transferred into Rag1−/− mice and analyzed as be-fore. Although comparable disease developed in WT and Il12rb2−/−

recipients, mice receiving Stat4−/− Th17p cells showed a significantreduction in colonic inflammation (Fig. 2 B and C).Reduced disease in recipients of Stat4−/− Th17p cells corre-

lated with a relative decrease in colonic IFN-γ+ T cells and areciprocal increase IL-17A+ and IL-17A+IFN-γ+ (Fig. 2D). Indraining MLN, there were also decreased frequencies of singleIFN-γ+ T cells, although no significant changes in IL-17A+ orIL-17A+IFN-γ+ cells were detected. Thus, although IFN-γ expres-sion was significantly reduced in T cells recovered from Stat4−/−

Th17p recipients, it was not completely ablated, consistent withreduced, but not absent, colitis development.Concomitant with reduced IFN-γ production by Stat4−/− Th17p

cells was a significant reduction in T-bet expression in both the IL-17A+IFN-γ+ and IFN-γ+ CLP fractions (Fig. 2E). Because T-betexpression can be induced by either Stat1 or Stat4 signaling, Stat1might partially compensate for Stat4 in this setting. Interestingly,although T-bet was also significantly reduced in IFN-γ+ cells re-covered from the CLP of Il12rb2−/− recipients (Fig. 2E), this was notassociated with a significant decrement in recovered IFN-γ+ cells orwith reduction in disease. Accordingly, whereas the pathogenictransition of Th17p cells to Th1-like cells was at least partially de-pendent on Stat4 signaling, IL-12 signaling was dispensable. Thisfinding is consistent with a dominant role for IL-23–driven Stat4signaling in Th1-like transition and disease in the gut.

T-bet Is Indispensable for Th17 Transition to Th1-Like Cells and Induc-tion of Colitis. T-bet is a central transcription factor required forTh1 cell programming and Ifng expression, and was shown inprevious studies to be required for the transition of Th17p cellsinto Th1-like cells (11, 15). To examine the requirement for T-betin Th17-derived IFN-γ production and colitis pathogenicity,Th17p cells derived from Il17fThy1.1 (WT) and Tbx21−/−.Il17fThy1.1

(Tbx21−/−) mice were transferred into Rag1−/− recipients. Fourweeks after transfer there was minimal disease in recipients of

Th17p cells from Tbx21−/− mice (Fig. 3A and Fig. S4A) comparedwith WT controls.Impairment of disease in Tbx21−/− recipients was associated

with a marked reduction in single IFN-γ+ T cells in both the CLPand MLN (Fig. 3 B and C). There were reciprocal increases infrequencies of IL-17A+ cells. This was accompanied by an in-crease in expression of IL-17F by recovered T cells, which wasvirtually undetectable in cells recovered from WT mice (Fig.S4 B and C), demonstrating a role for T-bet in suppressing ofboth IL-17A and IL-17F in vivo. Cells recovered from Tbx21−/−

recipients also expressed increased IL-22 (Fig. S4 D and E).Thus, consistent with results from transfers of Ifng−/− Th17pcells (Fig. 1), T-bet–deficient Th17 cells induced minimal colitis,despite comparable or increased numbers of IL-17A+, IL-17F+

and IL-22+ T cells.Notably, although expression of IFN-γ by recovered T-bet–

deficient Th17p cells was globally repressed, the frequencies ofdual IL-17A+IFN-γ+ cells in the intestine were increased (Fig. 3B and C) and their absolute numbers did not significantly decline(Fig. 3D), despite the marked reduction of inflammatory T-cellinfiltrates in the intestines of recipients of Tbx21−/− Th17 cells(Fig. 3A and Fig. S4). Thus, in contrast to single IFN-γ+ cells,expression of IFN-γ by these double-positive cells was largelyT-bet–independent. Additionally, despite their capacity for IFN-γexpression, this subset was associated with substantially reducedcolitis, suggesting that either the levels of IFN-γ expressionby these cells was insufficient to drive robust disease or thatT-bet–dependent factors in addition to IFN-γ are required.

WT

IFN-

IL-1

7A

0

10

20

30

40IL-17A+ IFN- -

IL-17A+ IFN +

IL-17A- IFN- +

CLP MLN

Stat4 -/-Il12rb2 -/-

WT

Stat4 -/-

Il12rb2 -/-

Cyt

okin

e po

s. c

ells

(% to

tal C

D4+

)

012345

WT

Stat4 -/-Il12rb2 -/-

0

10

20

30

Scor

e

WT Stat4 -/-Il12rb2 -/-

ns***

0

10

20

30

40

**

*

*

*

18.9 0.3

0.6

28.5 0.4

0.7

20.4 0.2

0.6

IL-17A+IFN- -

IL-17A+IFN- +

IL-17A-IFN- +

WT

Stat4 -/-Il12rb2 -/-

0

200

400

600

800

1000 ** *****

T-be

t exp

ress

ion

on C

D4+

(MFI

)

A B

C D

E

Fig. 2. Stat4, but not IL-12, is required for Th17-derived, IFN-γ–dependentcolitis development. Th17p cells from Il17fThy1.1 (WT), Il12rb2−/−.Il17fThy1.1

(Il12rb2−/−), and Stat4−/−.Il17fThy1.1 (Stat4−/−) mice were transferred intoRag1−/− recipients (4 × 105 cells per mouse). Four weeks after transfer, micewere analyzed for development of colitis. (A) Representative FACS plots ofTh17 precursors from WT, Il12rb2−/− or Stat4−/− mice showing expression ofIL-17A and IFN-γ. (B) Histological scoring (mean + SEM). (C) Representativecolon sections. H&E stain; original magnification 10×. (D) Relative frequen-cies of IL-17A+, IL-17A+IFN-γ+, and IFN-γ+ CD4+ T cells in the CLP and MLN ofcolitic mice of indicated groups. (E) Relative expression (median fluorescenceintensity) of T-bet+ cells in the CLP of IL-17A+, IL-17A+IFN-γ+, and IFN-γ+ CD4+

T cells recovered from indicated recipient mice. Data are representative oftwo independent experiments with five mice per group. *P < 0.05, **P <0.01, ***P < 0.001.

Harbour et al. PNAS | June 2, 2015 | vol. 112 | no. 22 | 7063

IMMUNOLO

GYAND

INFLAMMATION

Accordingly, disease induction by Th17p cells is contingent onexpression of T-bet, which is critical to the transition to pathogenicTh1-like cells that repress the Il17a/Il17f locus and up-regulateexpression of IFN-γ (15).Previous work has established that Th17-derived Th1-like cells

differ from classic Th1 cells in their molecular signature,retaining expression of some Th17 lineage-related genes (11). Todetermine whether Th1-like cells and classic Th1 cells arecomparable in their ability to induce colitis, CD4+ T cells fromIfng/Thy1.1 BAC-In or Ifng−/−.Ifng/Thy1.1 BAC-In mice were po-larized under Th1 conditions (Fig. S5A, Left), and CD4+ cells fromIl17fThy1.1 mice were polarized ex vivo under Th17 conditions (Fig.S5A, Right). The Ifng/Thy1.1 BAC-In transgenc mouse reports Ifngexpression via Thy1.1 expression (28), thus the Ifng−/−.Ifng/Thy1.1BAC-In (Tg) is able to report Ifng gene expression but not ex-press IFN-γ. All cells were sorted for Thy1.1 expression (Fig.S5B) to obtain purified Th1 or Th17p cells, respectively, and weretransferred into Rag1−/− recipients.Both Th17p and Th1 cells induced weight loss and inflamma-

tion in the colon of Rag1−/− recipients, compared with recipientsof Ifng−/− Th1 cells, which did not display evidence of wasting ordisease (Fig. S5 C and D). These results suggest that althoughTh1-like and Th1 cells may have different molecular signatures,they induce comparable disease that is IFN-γ–dependent.

Th17 and Th1 Cells Cooperate to Induce Colitis. CD4+ T cells de-ficient in RORγt and IL-23R are unable to induce colitis despitebeing IFN-γ competent, indicating that Th17 development iscritical for colitis induction (9, 29). Nevertheless, Th1 cells de-rived ex vivo were comparable to Th17 cells in their ability todrive disease (Fig. S5). This finding raised the possibility thatTh17 cells might provide an environment wherein the twosubsets cooperate to induce disease. To address this possibility,naïve CD4+CD45RBhi (“RBhi

”) cells from Rorc−/− mice, which

cannot develop into Th17 cells, were cotransferred with Th17pcells from Ifng−/− mice that are nonpathogenic (Fig. 4).In contrast to recipients of WT Th17 or WT RBhi cells, min-

imal disease was seen in recipients of IFN-γ–deficient Th17 cellsor RORγt–deficient RBhi cells alone (Fig. 4A). However, co-transfer of these cells induced disease comparable to recipients ofWT RBhi cells. This finding was associated with significant in-creases in the percentage (Fig. 4C) and number (Fig. 4D) ofIL-17A+IFN-γ– cells recovered from the CLP of recipients of thecotransferred cells compared with Rorc−/− RBhi cells alone, andequivalent numbers of this population compared with recipientsof Ifng−/− Th17 cells alone (Fig. 4D). Notably, the number ofIL-17A–IFN-γ+ cells in the CLP of cotransfer recipients was notsignificantly different from recipients of WT RBhi cells, whereasthe number of these cells in recipients of Rorc−/− RBhi cells alonewas significantly reduced. Thus, deficiency in pathogenicity ofIFN-γ–deficient Th17 cells or RORγt–deficient T cells was re-versed by cotransfers of these populations, implying that Th17cells can support the de novo development and pathogenicity ofclassical Th1 cells in vivo.In this study, we have shown that Th17 cells can play dual roles

in the development of colitis. First, we find that Th1-like cellsthat develop from Th17p cells are required for the pathogenesisof Th17-driven colitis, at least in part because of the IFN-γ theyproduce. In view of previous studies that have identified a centralrole for IL-23 in IBD pathogenesis (6–8), our findings support amodel wherein IL-23 produced by innate cells of the intestineacts on developing Th17 cells to deviate their differentiation toTh1-like cells by up-regulating T-bet and repressing RORγt ex-pression, through a mechanism that is largely Stat4-dependent.

4.7 3.2

29.9

38.8 7.9

4.3

WT

IFN-

IL-1

7A

Tbx21-/-

0

10

20

30

40

0

5

10

15

0

20

40

60

CLP MLN CLP MLN CLP MLN

IL-17A+ IFN- - IL-17A+ IFN- + IL-17A- IFN- +

*** *** ** *** ***WTTbx21-/-

Cyt

okin

e po

sitiv

e (%

tota

l CD

4+)

Sco

re

0

0.5

1.0

1.5

2.0

0

0.5

1.0

1.5

Cyt

okin

e po

sitiv

e (n

orm

aliz

ed to

WT)

Cyt

okin

e po

sitiv

e (n

orm

aliz

ed to

WT)CLP MLN

***

IL-17A+IFN- +

IL-17A+IFN- -

IL-17A-IFN- +

IL-17A+IFN- +

IL-17A+IFN- -

IL-17A-IFN- +

WT0

20

40

60***

Tbx21-/-

A B

C

D

Fig. 3. T-bet is essential for development of Th1-like cells and induction ofcolitis. Th17p cells from Il17fThy1.1 (WT) and Tbx21−/−.Il17fThy1.1 (Tbx21−/−)mice were transferred into Rag1−/− recipients (4 × 105 cells per mouse) andanalyzed for colitis 4 wk after transfer. (A) Histological scores (mean + SEM;one of three similar experiments). (B) Representative FACS plots showingexpression of IL-17A and IFN-γ in CD4+ cells of CLP from Rag1−/− recipients ofWT or Tbx21−/− Th17p. (C) Relative frequencies of IL-17A+, IL-17A+IFN-γ+,and IFN-γ+ CD4+ T cells in the CLP and MLN of colitic mice of indicatedgroups. (D) Total numbers of IL-17A+ Thy1.1+ CD4+ T cells in the CLP andMLN of recipients of indicated groups; fraction relative to WT Th17p re-cipients. Data in C and D are pooled from three independent experimentswith five mice per group. *P < 0.05, **P < 0.01, ***P < 0.001.

Cyt

okin

e po

s. c

ells

(% to

tal C

D4+

) WT Th17pIfng-/- Th17p

Rorc-/- RBhiRorc-/- RBhi + Ifng-/- Th17p

Scor

e

WT Th17pIfng-/- Th17p

Rorc-/- RBhiRorc-/- RBhi + Ifng-/- Th17p

WT RBhi

***

WT RBhi

0

10

20

30

* ***

0

20

40

60

80

100

Cel

l num

ber (

x 10

3 ) WT Th17pIfng-/- Th17p

Rorc-/- RBhiRorc-/- RBhi + Ifng-/- Th17p

WT RBhi

0

20

40

60

80

0

50

100

150

200

***

**** ** **

***

*** **

0

20

40

60

**

*****

0

5

10

15

20

25

0

5

10

15*

8.7 6.5

16.9

21.1 <0.1

0.4

6.2 6.1

34.0

1.8 0.8

26.8

15.3 0.3

10.2

WT Th17p Ifng-/- Th17p Rorc-/- RBhi Rorc-/- RBhi + Ifng-/- Th17pWT RBhi

IL-1

7A

IL-17A-IFN- +

IFN-

IL-17A+IFN- - IL-17A+IFN- +

IL-17A+IFN- - IL-17A+IFN- + IL-17A-IFN- +

A

B

C

D

Fig. 4. Th17 cells cooperate with Th1 cells to induce colitis. Il17fThy1.1 (WT)and Ifng−/−.Il17fThy1.1 (Ifng−/−) Th17p cells derived from naive CD4+ T cells, orCD4+CD45RBhi (CD4+RBhi) cells from WT and Rorc−/− mice, were transferredinto Rag1−/− recipients (4 × 105 cells per mouse). Another group receivedIfng−/− Th17p cells + Rorc−/− CD4+RBhi cells (2 × 105 each). Four weeks aftertransfer, mice were analyzed for development of colitis. (A) Total histologicalscores of indicated recipients (mean + SEM). (B) Representative FACS plots ofCD4+ CLP from recipient Rag1−/− mice showing expression of IL-17A and IFN-γ.(C) Relative frequencies of IL-17A+, IL-17A+IFN-γ+, and IFN-γ+ CD4+ T cells in theCLP of colitic mice of indicated groups. (D) Total numbers of IL-17A+, IL-17A+

IFN-γ+, and IFN-γ+ CD4+ T cells in the CLP of recipient mice of indicated groups.Data are pooled from two independent experiments (mean + SEM). *P < 0.05,**P < 0.01, ***P < 0.001.

7064 | www.pnas.org/cgi/doi/10.1073/pnas.1415675112 Harbour et al.

Second, we find that Th17 cells unable to produce IFN-γ them-selves can support the de novo development of pathogenic Th1cells from precursors blocked for Th17 development. Although acaveat of the current study is the use of effector T cells derived exvivo, which might not completely parallel cells that develop invivo, the results appear to reconcile early studies that identified alink to the Th1 pathway in IBD pathogenesis and subsequentstudies that supported a principal role for the Th17 pathway: theinherent tendency for Th17 cells to deviate to Th1-like progenydriven by IL-23 combined with their capacity to support the de-velopment of pathogenic Th1 cells independently of the Th17pathway provide the basis for a unified hypothesis wherein IL-23,the Th17 pathway, Th17-derived Th1-like cells, and classic Th1cells are all contributory.With regard to mechanisms by which Th1-like cells arise from

Th17p in the intestines, we find that T-cell expression of T-bet iscritical, that Stat4 is contributory, and that IFN-γ and IL-12signaling are dispensable. The finding that deficiency of Stat4significantly reduced disease severity, whereas deficiency ofIL-12Rβ2 did not, indicates that IL-23–driven Stat4 signaling isthe major contributor to the development of pathogenic Th1-likecells from Th17p, consistent with a critical role for IL-23 sig-naling in T cells in colitis (9, 11). Although at least some trans-ferred Th17 cells appeared to have received IL-12 signaling, asevidenced by reduced levels of T-bet expression by Th1-like cellsderived from Il12rb2−/− Th17 cells, this contribution was clearlynot required for colitis. Moreover, the greater impact on diseasecontributed by Stat4, despite comparable reductions in T-betexpression, imparted by deficiencies of IL-12Rβ2 and Stat4, isconsistent with findings that the targets of transcriptional regu-lation by Stat4 and T-bet are not synonymous (30).Our finding that T-bet expression by Th17 cells is required for

their transition to Th1-like cells and colitis pathogenesis is con-cordant with a recent study that identified a similar requirementfor T-bet, as well as Runx factors, in experimental autoimmuneencephalomyelitis (EAE) mediated by Th17 cells transfers (16).Interestingly, other studies suggest that T-bet expression by Tcells is not absolutely required for EAE pathogenesis, althoughdisease was blunted in its absence (31, 32). This finding mightreflect differences in the role of T-bet in the Th17 to Th1 tran-sition of Th17 cells differentiated ex vivo versus those that dif-ferentiate in vivo, which warrants further study. Interestingly,however, in each of these studies, a population of IL-17A+IFN-γ+T cells developed in the absence of T-bet expression by CD4+ Tcells, consistent with findings herein. Whereas these dual positivecells appear to be nonpathogenic in the intestines, their presencewas associated with disease in EAE in at least one study (31).These findings expose likely differences between pathogenesis inthe intestines and CNS. However, a role for IL-23–dependentIFN-γ production in the inflamed CNS was reported (27), sug-gesting the emergence of Th1-like cells from Th17 precursorsmay be a common feature of these two diseases. Nevertheless, auniform deficit of single IFN-γ+ cells in the absence of T-betsupports a central role for T-bet in the generation of this pop-ulation, indicating that T-bet is required to fully inhibit Rorc andthe Il17a/Il17f gene locus and completion of the transition to Th1-like cells (15, 16). The mechanism by which dual IL-17A+IFN-γ+cells arise in the absence of T-bet and what role they play in colitisis a subject for future studies, especially in view of their possibleassociation with disease in T-bet sufficient mice (9). Furthermore,given that expression of IFN-γ by T cells is essential for colitisinduction, and that its pathogenic actions appear to be primarilytargeted to non-T cells, it will be important to determine whichcells targeted by IFN-γ are critical to disease.The relative roles of Th1-like and classic Th1 cells in the path-

ogenesis of colitis have been unclear. Although we found thateffector Th1 cells induce disease equivalent to that of Th17 cells,CD4+ T cells deficient in the Th17-linked genes Rorc and Il23r are

unable to induce colitis, suggesting that the intestinal environmentdoes not support de novo Th1 development. This hypothesisis supported by human data showing intestinal biopsies fromCrohn’s disease patients express a marked increase in IL-23p19transcripts but not IL-12p35 (33). A relative abundance of IL-23relative to IL-12 in the intestine may involve the Notch2-dependent,CD103+CD11b+ subset of intestinal dendritic cells that expresshigh levels of Toll-like receptor 5 (34) and are a major source ofIL-23 in response to bacterial flagellin (35). Flagellin is a dom-inant antigen in Crohn’s disease patients (36), consistent with arole for this pathway of IL-23 production in disease development.Taken together, these findings support a model where initial Th17development in the intestine is promoted, and these cells bothdeviate into pathogenic Th1-like cells as well as support thedevelopment of Th1 cells that do not arise from the Th17 de-velopmental program. Future studies will be necessary to eluci-date the mechanisms by which this support occurs.The role of IFN-γ in colitis has been controversial. Our data,

and that of others, support a central role for IFN-γ in T-celltransfer models (3, 9, 37). However, one report found that diseaseis dependent on T-cell expression of T-bet and Stat4, but notIFN-γ (14). Although the basis for this discrepancy is unclear, oneimportant variable that is inadequately controlled between stud-ies is composition of the microbiota, which can profoundly in-fluence the development and severity of colitis. T-cell transfersinto germ-free (38) and restricted flora recipients (39) demon-strate the requirement for a complex microbiota in the inductionof intestinal inflammation. In addition to commensals that canmodulate the Th17 response, such as segmented filamentousbacteria (40), “pathobionts,” such as Helicobacter spp., which arecommon in animal facilities, can induce innate-driven colitis inimmunodeficient animals (7) and can exacerbate colitis in T-celltransfer models (41). Whether and how microbial factors mightimpact the influence of IFN-γ on disease pathogenesis will needto be explored going forward.The findings herein have potential import for clinical strategies

that target IBD: they suggest that interventions that prevent thetransition of Th17 precursors to Th1-like cells or block effectorfunctions of Th1 or Th1-like cells, such as IFN-γ, deserve atten-tion. Furthermore, as the transition of Th17 precursors to Th1-like progeny involves the loss of IL-23R and RORγt expressionand appears to be irreversible (15), therapies that target RORγt+Th17 precursors or IL-23 alone might not be beneficial to allpatients, at least in the setting of active disease where Th1 andTh1-like cells are abundant. Conversely, because Th17 cells ex-press stem cell markers and appear to have greater capacity forself-renewal than Th1-like cells (42), and might therefore serve asa reservoir of nonpathogenic cells able to transition into patho-genic, IFN-γ producing cells, therapies that deplete Th17 cells arealso warranted. In this regard, antibody-based therapies that blockIL-12p40 (and thus both IL-23 and IL-12) (43, 44) and IFN-γ (45)have shown promise in IBD trials, whereas blockade of IL-17Ahas not (46), despite promising results in clinical trials of otherTh17-driven autoimmune disorders, such as psoriasis, rheumatoidarthritis, and uveitis (47). Thus, although the rationale for tar-geting the Th17 pathway remains promising in IBD, strategiesthat target both the Th17 and Th1 arms should be considered.

Materials and MethodsMice. The generation of Il17fThy1.1 reporter mice and Ifng/Thy1.1 BAC-In (Tg)mice has been described previously (11, 28). The following mice were purchasedfrom the Jackson Laboratories: Tbx21−/−, Ifng−/−, Ifngr1−/−, Il12rb2−/−, Il12b−/−,Rorc−/− and Rag1−/−. B6.Stat4−/− were a gift from M. Kaplan, Indiana University,Indianapolis. Where indicated, crosses between Il17fThy1.1 reporter mice andindividual knockout mice were made in our facility to generate mice homozy-gous for the Il17fThy1.1 and respective gene deletion alleles (e.g., Ifng−/−.Il17fThy1.1;note that for simplicity Il17fThy1.1/Thy1.1 is referred to as Il17fThy1.1). All micewere on a C57BL6 background, were Helicobacter spp.-negative, and weremaintained in a colony in which sentinel mice tested positive for antibodies

Harbour et al. PNAS | June 2, 2015 | vol. 112 | no. 22 | 7065

IMMUNOLO

GYAND

INFLAMMATION

against murine norovirus. Mice were bred and maintained in accordancewith the University of Alabama at Birmingham institutional animal care anduse committee regulations.

CD4 Adoptive Transfer Model of Colitis. CD4+ T cells were cultured under Th17or Th1 polarizing conditions for 6 d. Viable cells were recovered (Ficoll), andThy1.1+ cells were isolated by magnetic sorting according to the manufacturer’sinstructions (Miltenyi Biotec). A total of 4 × 105 Thy1.1+ cells were injected in-traperitoneally into age-matched Rag1−/− recipients. For IFN-γ blocking studies,recipient mice were injected intraperitoneally with 100 μg α-IFN-γ (XMG) 1 dprior and 1 d after T-cell transfers and once weekly thereafter. For the naïveT-cell transfer model, naïve CD4+CD45RBhi cells were FACS sorted and a total of4 × 105 cells were injected into age-matched Rag1−/− recipients. Mice weremonitored regularly for signs of disease andwere weighed weekly. At 4 wk aftertransfer, mice were killed and MLNs and colons were recovered. Colons were cutlongitudinally, and small lengths of tissue were obtained from the proximal,

middle, and distal portions of the colon, fixed in 10% (wt/vol) formalin, andprocessed for histopathological analysis. Samples were scored by a pathologist ina blinded fashion as previously described (48).

Statistical Analysis. Statistical significance was calculated using Prism software(GraphPad). The nonparametric Mann–Whitney test or Kruskal Wallis test wasused to determine significance for pathology data; all other data were ana-lyzed using t test or ANOVA. All P values ≤ 0.05 were considered significant.

Further details are provided in SI Methods and Materials.

ACKNOWLEDGMENTS. We thank members of the C.T.W. laboratory forhelpful comments and suggestions; D. O’Quinn, H. Turner, B. J. Parsons,D. J. Wright, and S. Sinclair for technical assistance; and M. Kaplan (IndianaUniversity) for the provision of B6 Stat4−/− mice. This work was supportedby the National Institutes of Health, the Crohn’s and Colitis Foundation ofAmerica, and Daiichi-Sankyo, Co. Ltd. (C.T.W.).

1. Weaver CT, Elson CO, Fouser LA, Kolls JK (2013) The Th17 pathway and inflammatorydiseases of the intestines, lungs, and skin. Annu Rev Pathol 8:477–512.

2. Powrie F, Leach MW, Mauze S, Caddle LB, Coffman RL (1993) Phenotypically distinctsubsets of CD4+ T cells induce or protect from chronic intestinal inflammation in C.B-17 scid mice. Int Immunol 5(11):1461–1471.

3. Powrie F, Correa-Oliveira R, Mauze S, Coffman RL (1994) Regulatory interactionsbetween CD45RBhigh and CD45RBlow CD4+ T cells are important for the balance be-tween protective and pathogenic cell-mediated immunity. J Exp Med 179(2):589–600.

4. Ito H, Fathman CG (1997) CD45RBhigh CD4+ T cells from IFN-γ knockout mice do notinduce wasting disease. J Autoimmun 10(5):455–459.

5. Neurath MF, et al. (2002) The transcription factor T-bet regulates mucosal T cell ac-tivation in experimental colitis and Crohn’s disease. J Exp Med 195(9):1129–1143.

6. Yen D, et al. (2006) IL-23 is essential for T cell-mediated colitis and promotes in-flammation via IL-17 and IL-6. J Clin Invest 116(5):1310–1316.

7. Hue S, et al. (2006) Interleukin-23 drives innate and T cell-mediated intestinal in-flammation. J Exp Med 203(11):2473–2483.

8. Elson CO, et al. (2007) Monoclonal anti-interleukin 23 reverses active colitis in a T cell-mediated model in mice. Gastroenterology 132(7):2359–2370.

9. Ahern PP, et al. (2010) Interleukin-23 drives intestinal inflammation through directactivity on T cells. Immunity 33(2):279–288.

10. Khor B, Gardet A, Xavier RJ (2011) Genetics and pathogenesis of inflammatory boweldisease. Nature 474(7351):307–317.

11. Lee YK, et al. (2009) Late developmental plasticity in the T helper 17 lineage. Im-munity 30(1):92–107.

12. Huber S, et al. (2011) Th17 cells express interleukin-10 receptor and are controlled byFoxp3⁻ and Foxp3+ regulatory CD4+ T cells in an interleukin-10-dependent manner.Immunity 34(4):554–565.

13. Annunziato F, et al. (2007) Phenotypic and functional features of human Th17 cells.J Exp Med 204(8):1849–1861.

14. Simpson SJ, et al. (1998) T cell-mediated pathology in two models of experimentalcolitis depends predominantly on the interleukin 12/Signal transducer and activatorof transcription (Stat)-4 pathway, but is not conditional on interferon gamma ex-pression by T cells. J Exp Med 187(8):1225–1234.

15. Mukasa R, et al. (2010) Epigenetic instability of cytokine and transcription factor geneloci underlies plasticity of the T helper 17 cell lineage. Immunity 32(5):616–627.

16. Wang Y, et al. (2014) The transcription factors T-bet and Runx are required for theontogeny of pathogenic interferon-γ-producing T helper 17 cells. Immunity 40(3):355–366.

17. Izcue A, et al. (2008) Interleukin-23 restrains regulatory T cell activity to drive T cell-dependent colitis. Immunity 28(4):559–570.

18. Noguchi D, et al. (2007) Blocking of IL-6 signaling pathway prevents CD4+ T cell-mediated colitis in a T(h)17-independent manner. Int Immunol 19(12):1431–1440.

19. O’Connor W, Jr, et al. (2009) A protective function for interleukin 17A in T cell-mediated intestinal inflammation. Nat Immunol 10(6):603–609.

20. Zhu J, et al. (2012) The transcription factor T-bet is induced by multiple pathways andprevents an endogenous Th2 cell program during Th1 cell responses. Immunity 37(4):660–673.

21. Madara JL, Stafford J (1989) Interferon-gamma directly affects barrier function ofcultured intestinal epithelial monolayers. J Clin Invest 83(2):724–727.

22. Dalton DK, et al. (1993) Multiple defects of immune cell function in mice with dis-rupted interferon-gamma genes. Science 259(5102):1739–1742.

23. Schulz EG, Mariani L, Radbruch A, Höfer T (2009) Sequential polarization and im-printing of type 1 T helper lymphocytes by interferon-γ and interleukin-12. Immunity30(5):673–683.

24. Harrington LE, et al. (2005) Interleukin 17-producing CD4+ effector T cells developvia a lineage distinct from the T helper type 1 and 2 lineages. Nat Immunol 6(11):1123–1132.

25. Park H, et al. (2005) A distinct lineage of CD4 T cells regulates tissue inflammation byproducing interleukin 17. Nat Immunol 6(11):1133–1141.

26. Uhlig HH, et al. (2006) Differential activity of IL-12 and IL-23 in mucosal and systemicinnate immune pathology. Immunity 25(2):309–318.

27. Hirota K, et al. (2011) Fate mapping of IL-17-producing T cells in inflammatory re-sponses. Nat Immunol 12(3):255–263.

28. Hatton RD, et al. (2006) A distal conserved sequence element controls Ifng gene ex-pression by T cells and NK cells. Immunity 25(5):717–729.

29. Leppkes M, et al. (2009) RORgamma-expressing Th17 cells induce murine chronicintestinal inflammation via redundant effects of IL-17A and IL-17F. Gastroenterology136(1):257–267.

30. Vahedi G, et al. (2012) STATs shape the active enhancer landscape of T cell pop-ulations. Cell 151(5):981–993.

31. Duhen R, et al. (2013) Cutting edge: The pathogenicity of IFN-γ–producing Th17 cellsis independent of T-bet. J Immunol 190(9):4478–4482.

32. Grifka-Walk HM, Lalor SJ, Segal BM (2013) Highly polarized Th17 cells induce EAE viaa T-bet independent mechanism. Eur J Immunol 43(11):2824–2831.

33. Schmidt C, et al. (2005) Expression of interleukin-12-related cytokine transcripts ininflammatory bowel disease: Elevated interleukin-23p19 and interleukin-27p28 inCrohn’s disease but not in ulcerative colitis. Inflamm Bowel Dis 11(1):16–23.

34. Uematsu S, et al. (2008) Regulation of humoral and cellular gut immunity by laminapropria dendritic cells expressing Toll-like receptor 5. Nat Immunol 9(7):769–776.

35. Kinnebrew MA, et al. (2012) Interleukin 23 production by intestinal CD103(+)CD11b(+)dendritic cells in response to bacterial flagellin enhances mucosal innate immune de-fense. Immunity 36(2):276–287.

36. Lodes MJ, et al. (2004) Bacterial flagellin is a dominant antigen in Crohn disease. J ClinInvest 113(9):1296–1306.

37. Kullberg MC, et al. (2006) IL-23 plays a key role in Helicobacter hepaticus-inducedT cell-dependent colitis. J Exp Med 203(11):2485–2494.

38. Powrie F, Mauze S, Coffman RL (1997) CD4+ T-cells in the regulation of inflammatoryresponses in the intestine. Res Immunol 148(8-9):576–581.

39. Aranda R, et al. (1997) Analysis of intestinal lymphocytes in mouse colitis mediated bytransfer of CD4+, CD45RBhigh T cells to SCID recipients. J Immunol 158(7):3464–3473.

40. Ivanov II, et al. (2009) Induction of intestinal Th17 cells by segmented filamentousbacteria. Cell 139(3):485–498.

41. Kullberg MC, et al. (2002) Bacteria-triggered CD4(+) T regulatory cells suppress Heli-cobacter hepaticus-induced colitis. J Exp Med 196(4):505–515.

42. Muranski P, et al. (2011) Th17 cells are long lived and retain a stem cell-like molecularsignature. Immunity 35(6):972–985.

43. Billich A (2007) Drug evaluation: apilimod, an oral IL-12/IL-23 inhibitor for thetreatment of autoimmune diseases and common variable immunodeficiency. IDrugs10(1):53–59.

44. Burakoff R, et al. (2006) A phase 1/2A trial of STA 5326, an oral interleukin-12/23inhibitor, in patients with active moderate to severe Crohn’s disease. Inflamm BowelDis 12(7):558–565.

45. Reinisch W, et al. (2006) A dose escalating, placebo controlled, double blind, singledose and multidose, safety and tolerability study of fontolizumab, a humanised anti-interferon gamma antibody, in patients with moderate to severe Crohn’s disease. Gut55(8):1138–1144.

46. Hueber W, et al. (2012) Secukinumab, a human anti-IL-17A monoclonal antibody, formoderate to severe Crohn’s disease: Unexpected results of a randomised, double-blind placebo-controlled trial. Gut 61(12):1693–1700.

47. Hueber W, et al.; Psoriasis Study Group; Rheumatoid Arthritis Study Group; UveitisStudy Group (2010) Effects of AIN457, a fully human antibody to interleukin-17A, onpsoriasis, rheumatoid arthritis, and uveitis. Sci Transl Med 2(52):52ra72.

48. Maynard CL, et al. (2007) Regulatory T cells expressing interleukin 10 develop fromFoxp3+ and Foxp3- precursor cells in the absence of interleukin 10. Nat Immunol 8(9):931–941.

7066 | www.pnas.org/cgi/doi/10.1073/pnas.1415675112 Harbour et al.