Mutual expression of the transcription factors Runx3 ...

Articles

Mutual expression of the transcription factors Runx3 and ThPOK regulates intestinal CD4+ T cell immunity © 2013 Nature America, Inc. All rights reserved.

Bernardo Sgarbi Reis1, Aneta Rogoz1, Frederico Azevedo Costa-Pinto1,2, Ichiro Taniuchi3 & Daniel Mucida1 The gut mucosa hosts large numbers of activated lymphocytes that are exposed to stimuli from the diet, microbiota and pathogens. Although CD4+ T cells are crucial for defense, intestinal homeostasis precludes exaggerated responses to luminal contents, whether they are harmful or not. We investigated mechanisms used by CD4 + T cells to avoid excessive activation in the intestine. Using genetic tools to label and interfere with T cell–development transcription factors, we found that CD4 + T cells acquired the CD8-lineage transcription factor Runx3 and lost the CD4-lineage transcription factor ThPOK and their differentiation into the TH17 subset of helper T cells and colitogenic potential, in a manner dependent on transforming growth factor-b (TGF-b) and retinoic acid. Our results demonstrate considerable plasticity in the CD4 + T cell lineage that allows chronic exposure to luminal antigens without pathological inflammation. Environmental cues are part of the differentiation processes of all cell types, including T cells. Such cues, including the cytokine milieu, influence mature CD4+ T cells to differentiate into various subsets with many functions in the periphery, including proper control of infection (helper T cells) and the prevention of progressive activation of the immune system (regulatory T cells (Treg cells)). In contrast, mature CD8+ T cells are mainly cytotoxic T lymphocytes (CTLs) and are essential in the protection against intracellular pathogens. The transcription factor ThPOK (Zbtb7b or cKrox) drives the development of CD4+ T cells from CD4+CD8+ double-positive precursors, whereas the development of CD8+ T cells requires mainly the transcription factor Runx3 and the zinc-finger transcription factor MAZR (also called PATZ1 or Zfp278)1–3. These transcription factors bind to each other, and their dynamic interaction ultimately determines thymic T cell fate. Ablation of the Runx complex in developing thymocytes results in derepression of Zbtb7b (which encodes ThPOK; called ‘Thpok’ here) at the double-positive stage and considerably fewer mature CD8+ single-positive T cells in the periphery4. Conversely, inactivation of Thpok by conditional deletion, hypomorphic expression or loss-of-function helper-deficient mutation results in a nearly complete absence of peripheral CD4 + T cells5–7. In the intestine, where a large amount of diverse antigens can be constantly perceived as stimuli, the immune system has developed particular pathways to deal with such rich luminal content without generating progressive inflammation8. Although Treg cells and other regulatory cells can be found in the intestinal tissue, not much is known about cell-intrinsic mechanisms that regulate the helper function of CD4+ T cells at this environmental intersection.

Peripheral mature CD4+ T cells and CD8+ T cells express ThPOK and Runx3, respectively, in a mutually exclusive way3,5. However, ThPOK expression by CD4+ T cells may not be as stable as previously believed, as intestinal CD4+ T cells show consistent postthymic downregulation of ThPOK expression9. To address whether such a pattern is associated with changes in Runx3 expression by intestinal CD4+ T cells, we analyzed the expression of ThPOK and Runx3 through the use of reporter strains with sequence encoding green fluorescent protein (GFP) or yellow fluorescent protein (YFP) knocked in to the Thpok locus (Thpok-GFP mice) or Runx3 locus (Runx3-YFP mice), respectively3,5. We observed that both lower expression of ThPOK and high expression of Runx3 were associated with acquisition of the expression of genes related to the CD8 lineage and less differentiation into the TH17 subset of helper T cells. Loss of ThPOK function resulted in dampening of the inflammatory potential of CD4+ T cells, although it did not directly regulate TH17 differentiation. In contrast, loss of Runx3 function resulted in higher expression of ThPOK by intestinal CD4+ T cells and enhanced TH17 differentiation. Our experiments provide mechanistic evidence of how transcription factors involved in T cell lineage choice continue to have a decisive role in cell function in the periphery. RESULTS Reciprocal expression of ThPOK and Runx3 by CD4+ T cells Through the use of the knock-in reporters for ThPOK and Runx3 described above (Thpok-GFP and Runx3-YFP), we found that whereas these transcription factors were expressed by CD4+ T cells and CD8+ T cells, respectively, in peripheral tissues (Fig. 1a), intestinal CD4+

1Laboratory

of Mucosal Immunology, The Rockefeller University, New York, New York, USA. 2Department of Pathology, School of Veterinary Medicine, University of Sao Paulo, Sao Paulo, Brazil. 3Laboratory for Transcriptional Regulation, RIKEN Research Center for Allergy and Immunology, Yokohama, Kanagawa, Japan. Correspondence should be addressed to D.M. ([email protected]). Received 18 September 2012; accepted 11 December 2012; published online 20 January 2013; doi:10.1038/ni.2518

nature immunology  aDVANCE ONLINE PUBLICATION



Articles a

Figure 1  Reciprocal regulation of ThPOK and Runx3 in intestinal tissue. (a,b) Expression of ThPOK and Runx3 by CD45+TCRβ+ cells from spleen (a) or CD45+TCRβ+CD4+ IELs from the small intestine (b) of naive Thpok-GFP (top) or Runx3-YFP (bottom) reporter mice (n = 5 per group). Right (b),expression of 2B4 (CD244) by ThPOK+CD8α−CD4+, ThPOK−CD8α−CD4+ and ThPOK−CD8α+CD4+ cells (top) or Runx3– CD8α−, Runx3+CD8α− and Runx3+CD8α+ cells (bottom); circled areas at left correspond to populations analyzed at right. (c) Expression of mRNA from various genes (vertical axes) in IELs sorted from the small intestine of naive Runx3-YFP reporter mice as in b; results are presented relative to those for Rpl32 (which encodes the ribosomal protein L32). (d) Expression of CD8αα and of ThPOK (top) or Runx3 (bottom) by CD45+TCRβ+CD4+ cells from the spleen, IELs from the small intestine (sIEL) and lymphocytes from the lamina propria of the small intestine (sLPL) of Rag1−/− recipients (n = 5 per group) 40 d after adoptive transfer of sorted naive CD4+ T cells isolated from spleens of Thpok-GFP mice (top) or Runx3-YFP mice (bottom). (e) Expression of intracellular IL-17A and of ThPOK (top) or Runx3 (bottom) by CD45 +TCRβ+CD4+ IELs from the small intestine of recipient mice as in d. Numbers in top right corners (d,e) indicate percent cells in each corresponding quadrant. Data are representative of at least three independent experiments (a,b), are pooled from three representative of two independent experiments with similar results (c; error bars, s.e.m. of triplicates) or are representative of three independent experiments (d,e).

b

CD4+T cells CD8+T cells

CD8α–ThPOK+ CD8α–ThPOK– CD8α+ThPOK– 3

4

0 10

5

10

0

10

ThPOK

4

4

3

4

0 10

CD8α+Runx3+ 0

10

10

3

4

10

10

5

Runx3

Tbx21

2.6 0.6 0.2 0 Spleen

3

4

10

2

3

4

80 40 0

800 600 400 200 0

CD8α–Runx3– CD8α–Runx3+ CD8α+Runx3+

29.5 0.34 42.8 27.3

sLPL

e

sIEL

3.07 1.48 20.9 74.5

2.63 10.6 43.0 43.8

5

10

0

ThPOK

3

10

10

4

5

10

ThPOK 0.48 0.03 95.6 3.83

5

10

4

10

1.81 24.2 55.8 18.1

5

1.32 4.49 82.6 11.6

16.6 0.94 51.1 27.4

10

4

10

3

10

3

10

0

IL-17

CD8αα

5

10 10 10

120

sIEL

1.02 0.41 5.87 92.7

10

4,000 3,000 2,000 1,000 0 400 300 200 100 0

0 10

2B4

Foxp3

Cd8a

40 30 20 10 0

0

CD8α–Runx3+

0

5

10

Runx3

d

10

Il17a

CD8αα

CD8α

3

0

Thpok

CD8α–Runx3–

10

3

10

Runx3

2B4

5

10

10

© 2013 Nature America, Inc. All rights reserved.

5

ThPOK

5

10

c

103 104 10

0

Runx3

3

10

4

10

5

10

0 0

3

10

10

4

5

10

Runx3

T cells did not follow the same pattern (Fig. 1b). Most CD4+ intraepithelial lymphocytes (IELs) had modest expression of ThPOK but high expression of the long isoform of Runx3 derived from the distal Runx3 promoter5 (Fig. 1b,c). Upregulation of Runx3 expression by CD4+ T cells was directly associated with expression of the coreceptor CD8αα (CD8α+CD8β− cells; Fig. 1b,c). Furthermore, acquisition of Runx3 expression paralleled upregulation of expression of the natural killer (NK) cell– and CTL-related molecule 2B4 (CD244; Fig. 1b) and also of Tbx21, which encodes the transcription factor T-bet (Fig. 1c). In contrast, Runx3hi CD4+ IELs had low expression of Thpok, Foxp3 (which encodes the transcription factor Foxp3) and Il17a (which encodes interleukin 17A (I-17A; Fig. 1c), a pattern that resembled the gene signature of CD4+ IEL populations that lose ThPOK expression9; this suggested that this pathway diverted CD4+ T cells toward developing into IELs or innate-like cytotoxic T cells10–12. To extend the analysis reported above to a proinflammatory setting, we used the T cell–transfer model of colitis13. We sorted pure naive (CD25−CD62LhiCD44lo) CD4+ T cells from Thpok-GFP reporter mice (as ThPOK+ naive CD4+ T cells) or from Runx3-YFP reporter mice (as Runx3− naive CD4+ T cells) and adoptively transferred those cells into mice deficient in recombination-activating gene 1 (Rag1−/− mice). The expansion of CD4+ T cell populations in such lymphopenic hosts induces severe colitis and inflammatory cytokine 

e­ xpression by CD4+ T cells, particularly interferon-γ (IFN-γ) and IL-17 (refs. 13,14). Consistent with this being a post-thymic event, CD4 + T cells upregulated Runx3 expression after migrating to the intestinal environment, particularly in the intraepithelial compartment (Fig. 1d). As observed in the steady state, CD8αα expression by CD4+ T cells was restricted to ThPOKlo or Runx3hi populations (Fig. 1d). Additionally, only cells that sustained high expression of ThPOK9 or had little or no expression of Runx3 were able to produce IL-17 (Fig. 1e). These data indicated that environmental cues in the gut led to a distinct program of CD4+ T cells in which loss of ThPOK expression and acquisition of Runx3 expression were both associated with suppression of features of CD4+ helper T cells and differentiation toward a CTL- or IEL-like phenotype. Intestinal cues influence the differentiation of CD4 + T cells IELs have high expression of the integrin CD103 (αE), and the heterodimer αEβ7 binds to E-cadherin expressed by intestinal epithelial cells. CD103 expression can be modulated by transforming growth factor-β (TGF-β) signaling via Runx3 (refs. 15,16). Retinoic acid (RA) is another environmental factor that modulates the differentiation of CD4+ T cells and their migration to the intestine, and it is known to act in synergy with TGF-β—for example, by inducing Foxp3-­expressing induced Treg cells and by enhancing T-bet expression in conditions of TGF-β plus IL-6, which suppresses TH17 differentiation17–19. We used both in vitro and in vivo models to evaluate the environmental cues involved in modulation of the expression of ThPOK and Runx3 by CD4+ T cells. Initially, we cultured OT-II CD4+ T cells (which have transgenic expression of an ovalbumbin (OVA)-specific T cell antigen receptor) with splenic dendritic cells (DCs) and OVA peptide in the presence of soluble cytokines. As described before20, exogenous TGF-β induced some expression of CD8α in CD4+ T cells (Fig. 2a). However, whereas TGF-β ‘preferentially’ induced CD8αβ expression, the combination of TGF-β and RA induced mostly CD8αα+ (CD8β−) OT-II cells (Fig. 2a,b). To determine whether these factors were involved in the peripheral modulation of T cell–lineage transcription factors, we interbred OT-II mice with Thpok-GFP or Runx3-YFP mice. Indeed, CD4+CD8αα+ cells induced in vitro had lower ThPOK expression than did CD4+CD8αα− cells stimulated in the same way (Fig. 2c). The addition of either aDVANCE ONLINE PUBLICATION  nature immunology

Articles

6.24

6.12

4

CD8β

103

0.62

0

5.44

0.77

10.1

9 6 3 0

103 104 105

0

200 150 100 50 0

CD8αα–Foxp3+

0

CD8αα+Foxp3– 0

103 104 105

CD8αα

0

103 104 105

ThPOK

Foxp3

CD8αα Foxp3

103

Foxp3

–

200 150 100 50 0

150 100 50 0

Tbx21

–

Cd8a

10

4

Gzmb

f

TGF-β + RA

105

d

8 6 4 2 0

CD4+ CD4+ + TGF-β CD4+CDαα+ + TGF-β

100 80 60 40 20 0 0

– T TG GF β Fβ + R A

CD8α

e

c

3

4

10 10 10

5

OT-II Thpok-GFP

TGF-β RA TGF-β + RA 3

4

10 10 105

0

ThPOK

25 20 15 10 5 0 2.0 1.5 1.0 0.5 0

ThPOK

g CD8αα+Foxp3– –

OT-II Runx3-YFP

–

CD8αα Foxp3

0

3

10

10

4

5

10

Runx3

TGF-β + RA 105 5.98

+

6.39

TGF-β + RA + IL-12 13.4

4.51

104

CD8αα

14.6

12

CD8αα+ CD8αβ+

Thpok

10

4.37

b

Runx3

105

TGF-β + RA

RA

Events (% of max)

TGF-β

–

Cells (%)

a

103 0 19.8 0

30 103 104 105

ThPOK

Figure 2  TGF-β and RA concomitantly induce Treg cells and ThPOKloRunx3hiCD4+ TGF-β + RA + IL-12 TGF-β + RA 6 T cells. (a,b) Expression of CD8α and Runx3–CD8αα– 6.83 105 105 Runx3+CD8αα– CD8β by sorted naive Vα2+CD4+ T cells 104 4 104 CD8αα–CD103– Runx3+CD8αα+ isolated from OT-II Thpok-GFP reporter 103 103 2 CD8αα–CD103+ mice and cultured with DCs and OVA 0 0 peptide alone (−) or in the presence of 23.4 68 CD8αα+CD103+ 0 TGF-β or RA or both (above plots (a) or 3 4 5 3 4 5 103 104 105 0 0 10 10 10 0 10 10 10 Runx3 below graph (b)). Numbers in or adjacent CD8αα ThPOK to outlined areas (a) indicate percent CD8β+CD8α+ cells (top) or CD8β−CD8α+ cells (bottom). (c) ThPOK expression by T cells obtained as in a,b and stimulated without additional cytokines (−) or gated as CD4+ cells or CD4+CD8αα+ cells and stimulated in the presence of TGF-β (+TGF-β). (d) Expression of ThPOK (left) or Runx3 (right) by sorted naive Vα2+CD4+ T cells isolated from OT-II Thpok-GFP mice (left) or OT-II Thpok-GFP mice (right) and cultured as in a,b (left margin). (e) Expression of Foxp3 and CD8αα by sorted naive Vα2+CD4+ T cells isolated from OT-II Thpok-GFP–Foxp3-RFP reporter mice and cultured with DCs, OVA peptide and TGF-β plus RA (left), and ThPOK expression (right) by CD8α+Foxp3−, CD8α−Foxp3+and CD8α−Foxp3− populations gated at left (circled areas). (f) Expression of mRNA from various genes (vertical axes) in the cells gated in e, presented relative to Rpl32 expression. (g,h) Expression of CD8αα and ThPOK (g) or CD103 (h) by sorted naive Vα2+CD4+ T cells isolated from OT-II Thpok-GFP reporter mice and cultured with DCs and OVA peptide in the presence of TGF-β and RA and then restimulated (above plots) in the presence of either TGF-β and RA or TGF-β, RA and IL-12. Right (h), ThPOK expression by CD8α+CD103+, CD8α−CD103+ and CD8α−CD103− populations gated at left (circled areas). (i) Expression of Runx3 and CD8αα by sorted naive Vα2+CD4+ T cells isolated from OT-II Runx3-YFP reporter mice and cultured with DCs and OVA peptide in the presence of TGF-β and RA (left), and mean fluorescence intensity (MFI) of CD103 in those gated cells (right). Numbers in quadrants (g,i) indicate percent cells in each. Data are representative of at least three independent experiments (error bars (b,f,i), s.e.m. of duplicates).

TGF-β alone or the combination of TGF-β and RA efficiently suppressed ThPOK expression while enhancing that of Runx3 (Fig. 2d). The induction of CD4+CD8αα+ cells paralleled the induction of Foxp3 expression, visualized through the use of cells from a reporter strain with sequence encoding monomeric red fluorescent protein (RFP) knocked in to the Foxp3 locus (Foxp3-RFP mice), mostly in a reciprocal manner (Fig. 2e). However, only CD4+CD8αα+ cells had lower ThPOK expression, whereas induced Treg cells maintained high expression of ThPOK (Fig. 2e,f). Consistent with their naturally occurring IEL counterparts, CD4 +CD8αα+ cells induced in vitro with TGF-β plus RA had higher expression of Tbx21, Gzmb (encoding granzyme B) and Runx3 than did CD4+CD8αα−Foxp3+ cells induced in the same way (Fig. 2f). As intestinal T cells are chronically stimulated, we restimulated OT-II cells initially primed with TGF-β plus RA. To address whether the higher Tbx21 expression of ThPOKlo or Runx3hi CD4+ T cells correlated with deviation toward the CD4+CD8αα+ phenotype, we did secondary stimulation in the presence of IL-12, a potent inducer of T-bet21. We found that this condition mirrored that of intestinal CD4+ T cells in naive mice, in which a substantial part of CD4+ T cells downmodulated their expression of ThPOK and upregulated their expression of CD8αα, CD103 and Runx3 (Fig. 2g–i). Additionally, overexpression of the long Runx3 isoform from Runx3 in TH17 cells resulted in higher Tbx21 expression, whereas enforced Tbx21 expression reciprocally resulted in upregulation of expression of the long Runx3 isoform nature immunology  aDVANCE ONLINE PUBLICATION

CD103 (103 MFI)

i

CD8αα

CD103

© 2013 Nature America, Inc. All rights reserved.

h

from Runx3 by TH17 cells (Supplementary Fig. 1a,b). In either case, IL-17 production was suppressed (Supplementary Fig. 1c,d). Next we addressed the in vivo role of signaling by TGF-β and RA in modulation of the peripheral CD4+ T cell lineage. For this we used mice with conditional knockout of Tgfbr2 in T cells (mice with loxPflanked Tgfbr2 alleles deleted by Cre recombinase expressed from the Cd4 promoter)22 on the Thpok-GFP reporter background. To avoid the systemic inflammation and premature death induced by the lack of TGF-β signaling in CD4+ T cells22, we backcrossed those mice onto OT-II Rag1−/− mice (to generate progeny called ‘OT-II(∆Tgfbr2) mice’ here). As expected, OT-II(∆Tgfbr2) mice had a normal lifespan and no signs of inflammation (data not shown). To induce the differentiation of naive T cells and their migration to the intestinal tissue, we maintained OT-II and OT-II(∆Tgfbr2) mice for 7 d on an OVAcontaining diet23. The OT-II(∆Tgfbr2) showed signs of inflammation of both the small and large intestine with polymorphonuclear infiltrates and epithelial cell damage and developed severe diarrhea, which indicated that their oral tolerance was impaired, but the OT-II mice did not (Fig. 3a–c). We found that whereas in OT-II mice, about 50% of CD4+ T cells in the IEL compartment downregulated their expression of ThPOK, lack of TGF-β signaling resulted in only about 10% of cells with loss of ThPOK expression and much lower CD8αα expression in the CD4+ T cell compartment (Fig. 3d). Conversely, feeding mice OVA induced a high frequency of CD103 + and 2B4+ CD4+ T cells in OT-II mice but not in OT-II(∆Tgfbr2) mice (Fig. 3e). 

Articles

3

4

2.33

30

*

0

sI

CD103

5

*

PL

sI EL

sL

60 45 30 15 0

* * PL

4.29 2

0 10 10 10 10

60

1.95 0.61

2.57

97.6 4.71 3

4

92 2.98

93.8

5

10 10 10

10

5

10

4

10

3

0

1.37 0.49

42.5

22.2

1.46

1.94 3

4

0 10 10 10

5

ThPOK

i

h

dnRaralsl/lsl

30 15

*

0

*

sI EL sL PL

0.94 1.62

dnRaralsl/lsl Cd4(dnRaralsl/lsl)

*

sL

0

90

0.56

0

g

*

ND

0

L

3

0.65

0.66

0.6

LN

10

2.31

0

1.2

SP

4

*

1.39

53.2

ThPOK

M

5

10

*

1.8

IFN-γ+ cells (%)

2B4

OT-II(∆Tgfbr2)

10

16.8

25

sI EL

29.4

50

PL

OT-II

75

PL

43.5

OT-II OT-II(∆Tgfbr2)

sL

9.04

EL

62.8

f Foxp3+ cells (%)

OT-II OT-II(∆Tgfbr2)

LPL 2B4+ cells (%)

IEL 1.95

4 10 OT-II (∆Tgfbr2) 103 0

0

59.7 35.6

45

60 40 20

*

0

*

sI EL sL PL

3

sL

1 2 Time (d)

10 0.025 5

97.3 21.9

ThPOKlo cells (%) CD8αα+ cells (%)

1.17

OT-II OT-II(∆Tgfbr2)

dnRaralsl/lsl Cd4(dnRaralsl/lsl) 40 30 20 10 0

*

*

CD8α+ CD8α–

Cd4(dnRaralsl/lsl) CCR9+ cells (%)

4 2

0

Spleen sIEL sLPL 0.11 1.44 17.7 1.72 0.62 9.56

ThPOKlo cells (%)

Score

25

CD103+ cells (%)

Diarrhea-free (%)

50

d OT-II

6

75

0

*

8

100

e

15 Figure 3  Signaling by TGF-β and RA in intestinal CD4+ T cells is required for downmodulation of ThPOK 75 45 expression and for CD8α expression in vivo. (a) Frequency of diarrhea-free mice among OT-II or 10 50 30 OT-II(∆Tgfbr2) mice fed OVA-containing chow for 7 d. (b) Hematoxylin-and-eosin staining of the proximal * 5 25 15 * colon of mice as in a. Original magnification, ×20. (c) Histological scores of the colon of mice as in a. * 0 0 0 Each symbol represents an individual mouse; small horizontal lines indicate the mean. (d–f) Expression of CD8α and ThPOK (d), 2B4 and CD103 (e), and intracellular IFN-γ and Foxp3 (f) by CD45+Vα2+CD4+CD8β− cells from the spleen (SPL) and mesenteric lymph nodes (MLN), IELs from the small intestine and lymphocytes from the lamina propria of the small intestine of mice as in a (n = 4–6 per group), assessed by flow cytometry. Right, frequency of gated cells. ND, not detected. (g) Expression of CD8α and ThPOK by CD45+TCRβ+CD4+CD8β− IELs isolated from the small intestine of naive dnRaralsl/lsl (control) mice and Cd4(dnRaralsl/lsl) mice (n = 3–4 per group). (h,i) Frequency of cells expressing ThPOK (h) or 2B4, CD103 or CCR9 (i) among cells gated in g. *P < 0.05 (Student’s two-tailed t-test). Data are representative of two experiments (a–f) or three independent experiments (g–i; error bars (d–f,h,i), s.e.m.). 2B4+ cells (%)

© 2013 Nature America, Inc. All rights reserved.

OT-II OT-II(∆Tgfbr2)

CD103+ cells (%)

c OT-II(∆Tgfbr2)

OT-II

CD8α

b OT-II OT-II(∆Tgfbr2)

CD8α

a

The lack of upregulation of CD103 expression was consistent with less induction of Runx3 in OVA-fed OT-II(∆Tgfbr2) mice. This regimen induced modest numbers of Foxp3-expressing cells in the gut, but absence of the TGF-β receptor TGF-βRII abolished the development of Treg cells (Fig. 3f). Although we did not observe induction of IL-17 expression in either group after OVA exposure (data not shown), OT-II(∆Tgfbr2) mice produced substantial IFN-γ (Fig. 3f). These results established a role for TGF-β signaling in the tolerance to foodderived antigens that went beyond the induction of induced Treg cells, which suggested that modulation of ThPOK and Runx3 may also be needed to avoid excessive immune response. To assess the role of RA signaling in the plasticity of the CD4+ lineage in vivo, we used a strain carrying sequence encoding a dominantnegative form of the RA receptor RARα (RAR403) downstream of a loxP-flanked ‘stop’ cassette (dnRaralsl/lsl mice)24. We crossed those mice with mice expressing Cre recombinase from the Cd4 promoter to generate Cd4(dnRaralsl/lsl) progeny, in which only T cells express the dominant-negative form of RARα, and with Thpok-GFP reporter mice to allow visualization of ThPOK expression after disruption RARα signaling in T cells. We compared the CD4+ IEL populations in dnRaralsl/lsl and Cd4(dnRaralsl/lsl) littermates and found that whereas about 60–70% of CD4 IELs were ThPOKlo in dnRaralsl/lsl mice, less than 5% of CD4+ T cells downregulated ThPOK expression in Cd4(dnRaralsl/lsl) mice (Fig. 3g,h). Consistent with that, Cd4(dnRaralsl/lsl) mice lacked CD4+CD8αα+ cells in the IEL compartment (Fig. 3g,h). The expression of other molecules associated with intestinal modulation of ThPOK and Runx3, such as 2B4 and CD103, was also significantly lower in the presence of the dominantnegative form of RARα (Fig. 3i). In accordance with the role of RA signaling in the homing of T cells to the gut25, expression of the 

chemokine receptor CCR9 was also suppressed (Fig. 3i). In addition to their published role in regulating the differentiation of Treg cells and effector CD4+ T cells17,19,20,26–29, a role for TGF-β and RA in the modulating the expression of Runx3 and ThPOK by CD4+ T cells in vivo was established by these data. CD4+ T cell–induced inflammation requires ThPOK The physiological loss of ThPOK in the intestine observed in intact mice, as well as after the transfer of T cells, is inversely correlated with TH17 differentiation and is directly correlated with acquisition of a CTL-like gene signature9. To address whether ThPOK loss itself was required for these effects, we forced downmodulation of ThPOK expression in CD4+ T cells in vivo. Given that ThPOK is required for the thymic differentiation of CD4+ T cells, the progeny of mice expressing Cre from its usual drivers (Cd4 or Lck) crossed with mice with loxP-flanked Thpok alleles (Thpokfl/fl mice) have impaired CD4+ T cell development6. To circumvent that issue, we developed various strategies. First, we crossed Thpokfl/fl mice with mice expressing Cre from the promoter of the gene encoding the Treg cell marker OX40 (Tnfrsf4; called ‘Ox40’ here) to generate ‘Ox40(∆Thpok)’ offspring. OX40 expression does not occur during the developmental stages of CD4+ T cells but is restricted to Treg cells and activated CD4+ T cells30. Therefore, the expression of Cre in Ox40(∆Thpok) mice allows the development of CD4+ T cells, as confirmed by the similar ratio of CD4+ T cells to CD8+ T cells in the periphery of Ox40(∆Thpok) mice, Ox40Cre–Thpokfl/+ mice and wild-type mice (Supplementary Fig. 2a). In contrast, Ox40(∆Thpok) mice had a high frequency of CD4+ IELs that expressed CD103, CD8αβ and CD8αα (Fig. 4a), which reinforced the proposal that loss of ThPOK was an important driving event in this program. Because in naive wild-type mice CD8α is generally expressed aDVANCE ONLINE PUBLICATION  nature immunology

Articles b

10

5

36

48.5

0.71

83.8

4

10

3

CD8α

10

0

1.59 103 104 105

0

CD8β

12 0

0

sIEL

10 20 30 40 50 Time (d)

MLN

f

27.6

1.05

WT

d

Ox40(∆Thpok)

WT

* * *

CD103 MLN

e

140 130 120 110 100 90

103 104 105

c

WT Ox40(∆Thpok)

Score

Ox40(∆Thpok)

Body weight (%)

a

1.15

sIEL

0.42

25.9

WT Ox40(∆Thpok) 15 12 9 6 3 0

*

g

3.02

WT

WT Ox40(∆Thpok)

0

2.03

+

*

*

*

*

*

*

h 0.73

j

sIEL 1.49

0.21

MLN 30.1

WT

13.3

k

0

+

CD4 CD8β

sIEL 1.51

14.5

27.1

7.63

39.7

3.15

9.22 105

62.4

4

10

0

40.1

32.3

103 104 105

14.2

+

+

CD4 CD8β 80.2

6.17

103 2

10 0

16.8 103 104 105

IFN-γ

40.1

4.32

26.7

40

0 102 103 104 105

30

IL-17

*

20

10

10 0

0

CD103

14.7

10

IL-17

3

10.2

4

Ox40 (∆Thpok)

10

19.4

41.2

+

IL-17

+

IL-17 IFN-γ+

IFN-γ

5

11.3

23.7

25.3

64.2

104

+

10

CD8α

10.1

63.5

lLPLs (%)

50 45.7

–

6.72

WT Ox40(∆Thpok)

WT

0

*

*

104 105

i MLN

Ox40 (∆Thpok)

*

20

*

*

CD8β

Foxp3

105

40

sI EL lIE L sL PL lL PL

3.39 0 103

5

*

60

SP L LN

0

103 104 105

10

80

M

3

10

CD8αα cells (%)

24.5

LN sI EL lIE L sL PL lL PL

0.2

1.45

0

39.6

15

SP L

3

25.6

10

Ox40 (∆Thpok)

10

22.8

4

20

M

53.2

5

1.04

+

34.8

0

CD8αα

© 2013 Nature America, Inc. All rights reserved.

10

5

4

10

CD8αα

Ox40 (∆Thpok)

0.6

CD8α

10

0.39

Foxp3 cells (%)

13.1

3

0

29.4 0

5.56

103 104 105

IFN-γ

CD4+

Figure 4  Loss of ThPOK by activated T cells hinders the development of colitis. (a) Expression of CD8α, CD8β and CD103 by CD45+TCRβ+CD4+ IELs isolated from the small intestine of a naive Ox40(∆Thpok) mouse. (b) Body weight of Rag1−/− recipients of adoptively transferred naive CD4+ T cells isolated from the spleens of wild-type (WT) or Ox40(∆Thpok) mice. (c) Hematoxylin-and-eosin staining of the proximal colon of recipients as in b. Original magnification, ×4. (d) Histological scores of colon from recipient mice as in b (presented as in Fig. 3c). (e,f) Expression of CD8αα and intracellular Foxp3 (e) or CD8αα and CD8β (f) by CD45+TCRβ+CD4+ cells from mesenteric lymph nodes and IELs from the small intestine of recipient mice as in b. (g) Frequency of Foxp3+ cells or CD8αα+ cells among CD45+TCRβ+CD4+ cells from the spleen and mesenteric lymph nodes, IELs from the small intestine (sIEL) or large intestine (lIEL), and lymphocytes from the lamina propria of the small intestine (sLPL) or large intestine (lLPL) of recipient mice as in b, gated as in e,f. (h,i) Surface expression of CD8α and CD103 (h) or intracellular expression of IL-17 and IFN-γ (i) by CD45+TCRβ+CD4+ cells as in e,f. (j) Frequency of IL-17+ cells, IL-17+IFN-γ+ cells or IFN-γ+ cells among the CD45+TCRβ+CD4+ lymphoytes from the lamina propria of the large intestine of recipient mice as in b, gated as in h,i. (k) Surface expression of CD8α and intracellular expression of IL-17 and IFN-γ by CD45+TCRβ+CD4+ CD8β− or CD8β+ cells isolated from the lamina propria of the large intestine of recipients of naive Ox40(∆Thpok) CD4+ T cells. Numbers in quadrants (a,f,h,i,k) or in outlined areas (e) indicate percent cells in each. *P < 0.05 (Student’s two-tailed t-test (b,g,j) or Mann-Whitney test (d)). Data are representative of two experiments with five mice per group (a) or two independent experiments with three to six mice per group (b–k; error bars (b,g,j), s.e.m.)

as CD8αα homodimers, it is likely that CD4+CD8αβ+ cells developed as consequence of incomplete differentiation toward the CD4+CD8αα+ phenotype, which requires the intestinal environment. To study the consequences of the loss of ThPOK in activated CD4+ T cells in vivo, we transferred naive CD4+ T cells from Ox40(∆Thpok) mice into Rag1−/− hosts. We found that wild-type CD4+ T cells induced severe colitis and wasting disease, but Ox40(∆Thpok) CD4+ T cells did not (Fig. 4b–d). The lack of disease in recipients of Ox40(∆Thpok) cells was not caused by less proliferation or population expansion of donor cells, as we recovered similar cell numbers from recipients of naive wild-type CD4+ T cells (Supplementary Fig. 2b). We analyzed Foxp3 expression by donor cells to address the possibility that Ox40(∆Thpok) cells ‘preferentially’ converted into Treg cells, which would also result in protection from nature immunology  aDVANCE ONLINE PUBLICATION

disease13. Instead, we found that Ox40(∆Thpok) donor T cells did not efficiently upregulate Foxp3 expression, particularly in the lamina propria and in peripheral tissues (Fig. 4e). In contrast, the transfer of Ox40(∆Thpok) cells resulted in more differentiation of CD4+CD8α+ T cells (CD8αβ+ and CD8αα+ cells) than did the transfer of wildtype cells (Fig. 4f,g and Supplementary Fig. 2c). In confirmation of the proposal that Ox40(∆Thpok) cells underwent CD4+CD8αα+ ‘programming’, possibly mediated by upregulation of Runx3, we found higher CD103 expression in these cells than in wild-type cells (Fig. 4h). However, we did not find a significantly lower frequency of IL-17+ or IFN-γ+ intestinal CD4+ T cells isolated from recipients of Ox40(∆Thpok) cells than in recipients of wild-type cells (Fig. 4i). In fact, mice that received Ox40(∆Thpok) CD4+ T cells had more cells 

Articles

Veh

TAM (day 0)

**

105

12

*

d Veh

*

105 2.19

6 3

0

10

20 30 Time (d)

40

0

103 0

5.45

9.7 4 0 103 10 10 IL-17

e

8.42

5

Veh



+

+

that produced both IL-17 and IFN-γ than did recipients of wild-type CD4+ T cells (Fig. 4j). Nonetheless, IL-17 production was mostly restricted to CD8α− cells, and IFN-γ-producing cells were mostly CD8α+ (Fig. 4k). To independently confirm that activated CD4+ T cells lacking ThPOK expression were less prone to induce intestinal inflammation, we transduced Thpokfl/fl CD4+ T cells with a retroviral construct encoding a fusion of Cre and GFP, activated the cells in vitro and adoptively transferred them into Rag1−/− recipients (Supplementary Fig. 2d,e). In contrast to results obtained with mock-transduced cells, CD4+ T cells with deletion of ThPOK were not able to induce wasting disease, had abundant expression of CD8αα and did not have substantially lower IL-17 expression (Supplementary Fig. 2f–h). We next sought to determine if deletion of Thpok during the course of inflammatory responses would also suppress colitis. We interbred Thpokfl/fl mice with Rosa26-CreERT2 mice (which express a fusion of Cre and a tamoxifen-sensitive variant of the estrogen receptor from the ubiquitous Rosa26 locus) to generate a strain (ThpokERT2) in which Cre could be induced through the administration of tamoxifen, which would allow temporal regulation of gene expression but avoid deletion of the gene in embryos. The administration of increasing doses of tamoxifen to CD4+ T cells from ThpokERT2 mice resulted in a gradual loss of ThPOK in vitro (Supplementary Fig. 2i). We transferred naive CD4+ T cells from ThpokERT2 mice into Rag1−/− hosts. The administration of tamoxifen immediately after that transfer resulted in significant prevention of wasting disease in Rag1−/− recipients, but the administration of tamoxifen 20 d after the transfer did not (Fig. 5a). However, both groups treated with tamoxifen had much less intestinal inflammation than did vehicle-treated control mice (Fig. 5b,c). PCR analysis of isolated intestinal cells confirmed the deletion of Thpok in vivo (data not shown). In addition, consistent with the proposal of a role for ThPOK in preventing the differentiation of CD4+CD8αα+ cells, tamoxifen administration starting at day 0 or day 20 led to enhanced differentiation of these cells (Fig. 5d,e). However, again we did not detect changes in the production of IL-17 or IFN-γ by CD4+ T cells after enforced deletion of Thpok (Supplementary Fig. 2j). We concluded that ThPOK was involved in maintaining the inflammatory ability of CD4+ T cells while preventing the development of the cytotoxic machinery, although additional factors must have acted concomitantly to suppress TH17 differentiation.

CD103 cells (%)

TAM (day 0) Figure 5  Continuous ThPOK expression is required for the inflammatory activity of CD4 + T cells. TAM (day 20) (a) Body weight of Rag1−/− recipient mice (n = 3 per group) given adoptive transfer of naive CD4 + * 20 * 50 * T cells isolated from the spleens of ThpokERT2 mice, then given vehicle (Veh) or tamoxifen (TAM) * * 40 15 intraperitoneally three times every 3 d, starting on day 0 or day 20 after transfer. (b) Hematoxylin30 and-eosin staining of the proximal colon of recipient mice as in a, 40 d after transfer. Original 10 20 * magnification, ×10. (c) Histological scores of recipient mice as in a, 40 d after transfer (presented 5 10 as in Fig. 3c, with mean and s.e.m.). (d) Expression of CD8α and intracellular IL-17 () by 0 0 MLN lIEL lLPL MLN lIEL lLPL CD45+TCRβ+CD4+ cells from recipient mice as in a, 40 d after transfer. Numbers in quadrants + + indicate percent cells in each. (e) Frequency of CD8α cells and CD103 cells from the mesenteric lymph nodes, IELs from the large intestine and lymphocytes from the lamina propria of the large intestine of recipient mice as in a, gated as in d. NS, not significant; *P < 0.05 (Student’s two-tailed t-test (a,e) or Mann-Whitney test (c)). Data are representative of two independent experiments (error bars (a,e), s.e.m.) CD8α cells (%)

© 2013 Nature America, Inc. All rights reserved.

TAM (day 0) TAM (day 20) 0.68 11.5 0.86 0.34 18.5

104

NS

9

95 85

Veh TAM (day 0) TAM (day 20)

TAM (day 20)

Score

Body weight (%)

115

c

b

Veh TAM (day 0) TAM (day 20)

CD8α

a

Runx3 modulates ThPOK in peripheral CD4+ T cells The functional implications of the downregulation of ThPOK expression noted above prompted us to examine its underlying mechanisms. The binding of Runx3 to transcriptional silencer elements of Thpok leads to the suppression of ThPOK expression by developing CD4+CD8+ double-positive thymocytes, which directs them to the CD8 lineage4. Moreover, in agreement with the proposal of a role for Runx3 in peripheral regulation of CD4+ T cells as well, we found that ThPOKlo CD4+ T cells were CD103+, whereas some ThPOKhi cell groups included both CD103+ and CD103− cells (Fig. 6a), which suggested that Runx3-induced upregulation of CD103 expression preceded the loss of ThPOK by intestinal CD4+ T cells. To address the issue of whether upregulation of Runx3 expression indeed occurred in ThPOKhi cells, we interbred Runx3-YFP reporter mice with ThpokGFP reporter mice and analyzed intestinal CD4+ T cells in their ‘Runx3-YFP–Thpok-GFP’ offspring. Although some of the CD4+ cells were Runx3hiThPOKhi, we did not observe Runx3loThPOKlo CD4+ T cells (Fig. 6b), which suggested that the upregulation of Runx3 expression occurred before CD4+ T cells downregulated ThPOK expression. We obtained similar results after transferring naive CD4+ T cells from Runx3-YFP–Thpok-GFP mice into Rag1−/− recipients (Fig. 6b). The development of Runx3hiThPOKhi cells might have represented an initial event in the development of CD4+CD8αα+ cells. To directly investigate the requirement for Runx3 in the loss of ThPOK by intestinal CD4+ T cells, we crossed mice with conditional deletion of Runx3 in T cells (through deletion of loxP-flanked Runx3 alleles by Cre expressed via Cd4 (‘Cd4(∆Runx3)’ mice)) with Thpok-GFP reporter mice to generate ‘Cd4(∆Runx3)–Thpok-GFP’ offpsring. Cd4(∆Runx3) mice have normal development of CD4+ T cells, although their development of CD8+ T cells is considerably impaired31. Consistent with the proposal of a role for Runx3 in directly suppressing ThPOK expression in intestinal CD4 + T cells, we found that Cd4(∆Runx3)–Thpok-GFP mice had a much smaller ThPOKlo CD4+ IEL population than did Thpok-GFP mice, as well as a lower frequency of CD4+ T cells expressing CD8αα, CD103 and 2B4 (Fig. 6c–e). These data provided evidence that Runx3 was required for the downmodulation of ThPOK expression in intestinal CD4+ T cells. To independently confirm the requirement for Runx3 in the peripheral downmodulation of ThPOK expression during inflammation, aDVANCE ONLINE PUBLICATION  nature immunology

Articles

Transfer model

c

10 4 10 3 10

5

10

4

10

0 0

3

10

4

10

2

5

22.5

1.72

2.16

10

5

10

4

3

0

10

3

97.4

0.86

3

0

ThPOK

3

10

4

10

10

10

0

0.44

5

41.5

0 10

ThPOK

3

10

4

10

0.43

95.3

30.4

2.82

0.43

0.86

57.6

1.72

96.6

0

9.12

5

0

ThPOK

10

3

10

4

10

5

ThPOK

f

g

Cd4(∆Runx3)–

h

Thpok-GFP

Thpok-GFP

120

d *

(∆ R

C trl un Cd x3 4 )

10 20 30 40 50 Time (d)

e Thpok-GFP Cd4(∆Runx3)–Thpok-GFP

SI LI 12 9 6 3 0

* * 0

c

Score

110 100 90 80

Cd4(∆Runx3)– Thpok-GFP

8 6 4 2 0

*

sIEL

SPL

15 10 CD45.2

Thpok-GFP

IFN-γ+ cells (%)

b Thpok-GFP Cd4(∆Runx3)–Thpok-GFP

IL-17+ cells (%)

a

Runx3 modulates intestinal inflammatory responses Next we investigated whether Runx3 was involved in the helper function of CD4+ T cells in vivo, particularly in the development of colitis. As in the experiments described above, we transferred naive CD4+ T cells fromThpok-GFP or Cd4(∆Runx3)–Thpok-GFP mice into Rag1−/− recipients. Mice that received Cd4(∆Runx3)–ThpokGFP cells showed accelerated wasting disease relative to that of mice that received Thpok-GFP cells (Fig. 7a). Mice that received both Thpok-GFP and Cd4(∆Runx3)–Thpok-GFP cells eventually developed intestinal inflammation, although histopathological analysis showed that of recipients of Cd4(∆Runx3)–Thpok-GFP cells had enhanced crypt abscesses and neutrophil infiltrates, and that the latter was directly associated with TH17 responses (Fig. 7b,c). Consistent

5 0

sIEL

f

10

5

10

4

10

3

sIEL 35.4

43.5

63.6

55.1

0 2

3

4

0 10 10 10 10

5

CD45.1

WT Cd4(∆Runx3)–Thpok-GFP

g

Frequency (%)

we adoptively transferred naive Thpok-GFP or Cd4(∆Runx3)– Thpok-GFP T cells into Rag1−/− hosts and analyzed recipient mice 4–6 weeks after transfer. We observed that Cd4(∆Runx3)– Thpok-GFP donor cells did not develop along the CD4+CD8αα+ pathway, and we found a lower frequency of ThPOKlo CD4+ T cells in the intestine of recipients of Cd4(∆Runx3)–Thpok-GFP donor cells than in the intestine of recipients of Thpok-GFP donor cells (Fig. 6f–h). However, the finding that about 15% of Cd4(∆Runx3)– Thpok-GFP T cells lost ThPOK expression during colitis suggested that during chronic inflammation, additional factors such as MAZR9 also contributed to the modulation of ThPOK. Together these data established a role for Runx3 in the modulation of ThPOK expression by peripheral CD4+ T cells.

CD103+ cells (%)

CD8αα

ThPOKlo cells (%)

Figure 6  Upregulation of Runx3 expression precedes the Cd4(∆Runx3)– Cd4(∆Runx3)– Thpok-GFP Thpok-GFP Thpok-GFP Thpok-GFP downmodulation of ThPOK expression and the expression of 1.73 2.35 1.14 75 45 5 15 10 + CD8αα by intestinal CD4 T cells. (a) Expression of CD103 and 4 10 50 30 ThPOK by CD45+TCRβ+CD4+ IELs isolated from the small intestine * 3 * of a naive Thpok-GFP mouse. (b) Expression of Runx3 and ThPOK 10 25 15 by CD45+TCRβ+CD4+ IELs isolated from the small intestine of a 0 24.3 59 16.9 79.6 0 0 naive Runx3-YFP–Thpok-GFP mouse (Ex vivo) or from a Rag1−/− 5 4 3 sIEL sIEL 0 10 10 10 + recipient 40 d after adoptive transfer of sorted naive CD4 T cells ThPOK isolated from Runx3-YFP–Thpok-GFP spleens (Transfer model). (c) Expression of CD8αα and ThPOK by CD45+TCRβ+CD4+CD8β− IELs isolated from small intestine of a naive Cd4(∆Runx3)–Thpok-GFP mouse. (d,e) Expression of ThPOK and CD103 (d) or 2B4 (e) by CD45+TCRβ+CD4+ IELs isolated from the small intestine of a naive Thpok-GFP mouse (control (Ctrl)) or Cd4(∆Runx3)–Thpok-GFP mouse (Cd4(∆Runx3)). (f) Expression of CD8αα and ThPOK by CD45+TCRβ+CD4+CD8β− IELs isolated from small intestine of Rag1−/− recipients (n = 4 per group) 40 d after adoptive transfer of sorted naive CD4 + T cells isolated from spleens of Thpok-GFP or Cd4(∆Runx3)–Thpok-GFP mice. (g,h) Frequency of ThPOKlo cells (g) and CD103+ cells (h) gated in f. Numbers in quadrants indicate percent cells in each. *P < 0.05 (Student’s two-tailed t-test) . Data are representative of three independent experiments with five mice in each (a,b), two independent experiments with three mice per group in each (c–e) or two independent experiments (f–h; error bars (g,h), s.e.m.).

Body weight (%)

WT Cd4(∆Runx3) –Thpok-GFP 75 50 25 0

* SPL sIEL

WT

nature immunology  aDVANCE ONLINE PUBLICATION

IL-17+ cells (%)

Cd4(∆Runx3)–Thpok-GFP Figure 7  Runx3 expression by T cells inversely correlates with their inflammatory 28 25 potential. (a) Body weight of Rag1−/− recipients (n = 5 per group) of sorted naive CD4+ 20 21 T cells isolated from spleen of Thpok-GFP or Cd4(∆Runx3)–Thpok-GFP mice. 15 14 10 (b) Hematoxylin-and-eosin staining of the proximal colon of recipient mice as in a, 7 5 analyzed 40–50 d later. Original magnification, ×20. (c) Histological scores of small 0 0 sIEL lIEL sLPL lLPL sIEL lIEL sLPL lLPL intestine and colons of recipient mice as in a, analyzed 40–50 d later. (d) Frequency + + + + + of IL-17 and IFN-γ cells among CD45 TCRβ CD4 IELs isolated from the small intestine of recipient mice as in a, analyzed 40–50 d later. (e) Expression of CD45.1 and CD45.2 by TCRβ+CD4+ cells isolated from spleen or from the small intestine of Rag1−/− recipients (n = 3 per group) of sorted naive CD4+ T cells isolated from the spleens of wild-type (CD45.1 +) mice or Cd4(∆Runx3) (CD45.2+) mice and adoptively transferred together at a ratio 1:1, analyzed 40 d later. Numbers adjacent to outlined areas indicate percent CD45.2+CD45.1− cells (Cd4(∆Runx3); top left) or CD45.2−CD45.1+ cells (wild-type; bottom right). Right, frequency of cells gated at left. (f,g) Frequency of CD8αα+ (CD8β−) cells (f) and IL-17+ cells (g) among TCRβ+CD4+ CD45.1+ (WT) or CD45.2+ (Cd4(∆Runx3)) IELs isolated from the small and large intestine and lymphocytes from the lamina propria of the small and large intestine of recipient mice as in e. Lines connect data from each individual recipient mouse. *P < 0.05 (Student’s two-tailed t-test (a,d,e) or one-way analysis of variance (c)). Data are representative of three independent experiments (a–d) or two experiments (e–g; error bars (a,c,d,e), s.e.m.).

CD4+

CD8αα+ cells (%)

© 2013 Nature America, Inc. All rights reserved.

Cd4(∆Runx3)– Thpok-GFP

Thpok-GFP

4

10

3

10

4 5 0 10 10 10 10

10

35.4

5

e

Cd4(∆Runx3)– Thpok-GFP

Thpok-GFP 10

4

10 0

ThPOK

0

1.29

5

10

CD8αα

Runx3

10

d

Cd4(∆Runx3)– Thpok-GFP 10

2

3

CD103

Ex vivo 5

2B4

b

CD103

a



Articles

© 2013 Nature America, Inc. All rights reserved.

95

0

4

7 11 14 18 Time (d)

*

103

10

5



9

10

6

10

Cd4(∆Runx3) 11.5 17.9

WT

4

107

*

104

0

10.2

16.9

0 102 103 104 105 IFN-γ

103

Figure 8  Enhanced resistance of Cd4(∆Runx3) mice to infection with C. rodentium. (a) Body weight of wild-type and Cd4(∆Runx3) mice 18 d after oral infection with C. rodentium. (b) Hematoxylinand-eosin staining of the proximal colon of mice infected as in a. Original magnification, ×20. (c) Histological scores of the proximal colon of mice infected as in a (presented as in Fig. 3c). (d) Colony-forming units (CFU) of C. rodentium in fecal pellets of mice infected as in a. (e) Intracellular expression of IL-17 and IFN-γ by CD45+TCRβ+CD4+ lymphocytes isolated from the lamina propria of the large intestine of mice infected as in a (left) and frequency of IL-17+ cells among those at left (right). Numbers in quadrants (left) indicate percent cells in each. (f) Intracellular IL-17 expression by sorted naive CD4+ T cells isolated from wild-type or Cd4(∆Runx3) mice (left margin) and cultured for 4.5 d with splenic DCs and soluble antibody to CD3ε in the presence of various cytokines (above plots). Numbers adjacent to outlined areas indicate percent IL-17+CD4+ cells (± s.e.m.). *P < 0.05 (MannWhitney test (c) or Student’s two-tailed t-test (d,e)). Data are representative of three independent experiments with three to five mice per group (a,b,d,e; error bars (a,d,e), s.e.m.) or three independent experiments (f; mean and s.e.m. of duplicate wells) or are from two combined experiments (c).

with our experiments reported above, the transferred Cd4(∆Runx3)– Thpok-GFP cells had less loss of ThPOK and lower CD103 expression (data not shown). Additionally, in the absence of Runx3, IL-17 expression was significantly higher, whereas IFN-γ production was not altered (Fig. 7d). To specifically investigate the cell-autonomous effects of Runx3 deficiency in CD4+ T cells, we transferred naive CD4+ T cells from wild-type (CD45.1+) mice and Cd4(∆Runx3)–ThpokGFP (CD45.2+) mice together into Rag1−/− hosts at a ratio of 1:1 and analyzed the recipient mice 6 weeks after transfer. We found a similar survival and population expansion of Cd4(∆Runx3)–Thpok-GFP and wild-type donor CD4+ T cells in peripheral tissue but a slightly lower ratio of these cells in the intestine (Fig. 7e), which indicated that the accelerated disease in Cd4(∆Runx3)–Thpok-GFP mice was not due to more homing of donor cells to the gut or proliferation of donor cells. Analysis of the expression of CD8αα and IL-17 confirmed the experiments reported above (Fig. 7a–d) and indicated a cellautonomous role for Runx3 in modulating the differentiation of CD4+ T cells (Fig. 7f,g). Although TH17 responses can be detrimental in some models of colitis, they are required for the control of infection with several extracellular bacteria, particularly at the mucosal surfaces32. Therefore, we investigated whether Cd4(∆Runx3) mice had altered responses to the extracellular bacteria Citrobacter rodentium, a mouse counterpart to enteropathogenic Escherichia coli32. We found that although wild-type and Cd4(∆Runx3) strains developed similar weight loss after infection (Fig. 8a), histopathological analysis of the colon showed that infected Cd4(∆Runx3) mice had more inflammation, with transmural infiltrates and epithelial cell damage, than did wild-type mice (Fig. 8b,c), which was probably the result of exaggerated inflammatory T H17 responses. Indeed, Cd4(∆Runx3) mice cleared the C. rodentium more efficiently than wild-type mice did, with a lower pathogen burden than that of wild-type mice, and had a greater frequency of IL-17-producing CD4+ T cells than did wild-type mice (Fig. 8d,e). Consistent with that, naive Cd4(∆Runx3) T cells were more efficient in differentiating into pathogenic TH17 cells (induced by low concentrations of TGF-β, IL-1β, IL-6 and IL-23) but not into nonpathogenic TH17 cells (induced by TGF-β and IL-6)33,34 than were wild-type CD4+ T cells (Fig. 8f). These data indicated that in addition to its effects on ThPOK expression, Runx3 expressed by CD4+ T cells may have been

105 5.43

f

TGF-β + IL-6 + IL1β

WT Cd4(∆Runx3) IL-17+ cells (%)

100

10 8 6 4 2 0

e WT Cd4(∆Runx3)

40 30 20 10 0

*

lLPL TGF-β + IL-6 + IL1β + IL23

34.05 ±1.48

32.25 ±2.75

38.00 104 ±2.82

59.35 ±12.5

WT

105 Cd4(∆Runx3)

103

IL-17

WT Cd4(∆Runx3)

110 105

d

Cd4(∆Runx3)

IL-17

115

c WT

C.rodentium (CFU)

b WT Cd4(∆Runx3)

Score

Body weight (%)

a

0 0 102 103 104 105 CD4

involved in repression of the TH17 pathway. To confirm the inverse association between Runx3 expression and severity of colitis and also TH17 differentiation, we overexpressed Runx3 in primary CD4+ T cells through the use of a retrovirus that encodes the long isoform of Runx3 (fused to an internal ribosomal entry site and GFP; ‘CD4 (Runx3) cells’) and transferred those cells into Rag1−/− hosts. CD4(Runx3) cells induced wasting disease in recipient mice with slower kinetics than that induced by mock-transduced cells (Supplementary Fig. 3a). Additionally, CD4(Runx3) donor cells had higher expression of CD103 and CD8αα but less production of IL-17 than did mocktransduced cells (Supplementary Fig. 3b,c). We concluded that upregulation of Runx3 expression, rather than loss of ThPOK itself, drove the suppression of proinflammatory TH17 responses. Together, the combined effects of Runx3 and loss of ThPOK resulted in substantial changes in the function of CD4+ T cells, which deviated from their classic helper phenotype toward a CTL-like phenotype with less ability to induce tissue damage and inflammation. DISCUSSION The intestinal tissue offers a unique environment for the immune system. It represents the main site for nutrient absorption while forming the largest surface exposed to environmental allergens. It interacts with beneficial microbiota in the midst of invading pathogens. Here we have described an alternative fate for CD4+ T cells that migrate to the intestinal tissue, particularly to the intraepithelial compartment, where cues such as TGF-β and RA are required for a post-thymic suppression of ThPOK and upregulation of Runx3 expression. Cells that underwent this pathway had a gene signature that resembled that of CTLs or innate-like lymphocytes10–12 and, conversely, these intestinal CD4+ T cells had lower expression of molecules associated with CD4+ helper T cells. We addressed whether that considerable change in the ‘CD4 program’ was driven by downregulation of ThPOK expression, upregulation of Runx3 expression or both. Published studies have demonstrated that forced suppression of ThPOK expression leads to derepression of the ‘CTL program’ in CD4+ T cells5–7. The extensive in vivo data presented here have shown that modulation of ThPOK in CD4 + T cells was a physiological process with consequences for the differentiation and function of T cells. Using several novel approaches for the aDVANCE ONLINE PUBLICATION  nature immunology

© 2013 Nature America, Inc. All rights reserved.

Articles deletion of Thpok post-thymically, we demonstrated a causal relationship between expression of ThPOK and suppression of the CTL program in CD4+ T cells in vivo. More notably, we showed that loss of ThPOK in activated CD4+ T cells dampened their inflammatory potential. Nevertheless, unlike ThPOKlo CD4+ IELs, activated CD4+ T cells with deleted Thpok alleles did not lose their ability to produce IL-17, even though they became inefficient in causing colitis. Indeed, the role of TH17 cells in the transfer model of colitis is controversial. Although they are generally associated with inflammation35, some populations of IL-17-producing T cells are also associated with regulatory immune responses, including those in colitis models33,34,36,37. Mutual expression of ThPOK and Runx3 may regulate ‘nonpathogenic’ and ‘pathogenic’ TH17 cells differently33,34, a possibility supported by our in vitro data. It should be noted that Ox40(∆Thpok) CD4+ T cells did not cause colitis, even though their conversion into Treg cells was impaired, which indicated the relevance of this pathway to CD4+ T cell function. It is possible that induced CD4+CD8αα+ cells could suppress inflammation in a cell-extrinsic manner (that is, they are regulatory), although we have not found evidence for that. Additionally, the upregulation of CTL- and natural killer cell–related genes in CD4+ T cells in which Thpok was deleted and their greater tendency to divert to the CD4+CD8αα+ pathway could itself explain their diminished ability to cause intestinal inflammation. In contrast to the diminished ability of Thpok-deficient CD4+ T cells to induce colitis, we found that Runx3-deficient CD4+ T cells induced an aggravated wasting disease with more TH17 differentiation. However, Cd4(∆Runx3) mice were more efficient than wild-type mice at clearing C. rodentium while sustaining a greater inflammatory infiltrate, known to be a TH17- or TH22-dependent process32. This is a notable example of how intestinal immune responses must constantly deal with the two facets of inflammation: protection and tissue damage. As Thpok-deletion experiments did not result in much lower IL-17 production by CD4+ T cells, we concluded that upregulation of Runx3 expression, rather than downregulation of ThPOK expression, drove the suppression of TH17 differentiation. Therefore, Runx3 could have a role both in the initiation of this program (suppressing ThPOK expression) and downstream (suppressing TH17 differentiation), in contrast to Runx1, which is positively associated with TH17 development38,39. Cooperation between Runx3 and T-bet is proposed to mediate activation of Ifng in TH1 cells40, and our data indicated that such cooperation could also have a role in the suppression of TH17 differentiation. What are the mechanisms for physiological downregulation of ThPOK expression in CD4+ T cells? The identification of regulatory elements in Thpok and demonstration of Runx3-binding sites at its silencers, which suppress Thpok transcription, served as a basis for the present understanding of T cell lineage fate4,41. Here we have described a physiological process that resulted in a post-thymic suppression of ThPOK via upregulation of Runx3 expression in intestinal CD4+ T cells. We provided mechanistic insights into this process, which were mediated at least in part by signaling via TGF-β and RA. We also provided ample evidence of a physiological role for this TGF-β-dependent pathway, from modulating susceptibility to intestinal infection, to the development of wasting disease and colitis, to oral tolerance. The last, however, is also influenced by TGF-β-induced peripheral Treg cells23,42. Although we do not understand the mechanisms that direct naive T cells to develop into CD4+CD8αα+ T cells rather than induced Treg cells, our data indicated that differences in the expression of ThPOK, Runx3 and T-bet are involved. Our study has provided information about the functional relevance of and mechanistic evidence of a continuous role for T cell–development nature immunology  aDVANCE ONLINE PUBLICATION

transcription factors in the periphery, which compose a tightly regulated process that substantially changes CD4+ T cell activity in a sitespecific manner. The functional studies of induced changes in the expression of ThPOK and Runx3 raise the possibility that disruption of this pathway may result in uncontrolled CD4+ T cell responses. Reinforcement of this pathway, in contrast, led to attenuated intestinal inflammation and greater susceptibility to infection with extracellular bacteria. It is also possible, however, that excessive triggering of the CTL function of CD4+ T cells results in breakdown of the epithelial layer, as observed in celiac disease, but paradoxically it could also increase resistance to viral infections that shut down the pathway of presentation via major histocompatibility complex class I. Factors that influence this pathway in one way or another are a potential target for therapeutic interventions in inflammatory disorders. Methods Methods and any associated references are available in the online version of the paper. Note: Supplementary information is available in the online version of the paper. Acknowledgments We thank K. Velinzon and Y. Shatalina for sorting cells; members of the Nussenzweig, Steinman, Marraffini and Tavazoie laboratories and employees of the The Rockefeller University for assistance; R. Noelle (Dartmouth University) for dnRaralsl/lsl mice; D. Littman (New York University) for Runx3-YFP mice; L. Glimcher (Cornell University) for Tbx21 vectors; and L. Marraffini, S. Tavazoie, G. Kim, M. Kronenberg, H. Cheroutre and members of the Mucida laboratory, particularly L. Feighery, for discussions and critical reading and editing of the manuscript. Supported by the Ellison Medical Foundation (D.M.), the Irma T. Hirschl Trust (D.M.), the Crohn’s & Colitis Foundation of America (D.M.), the US National Institutes of Health (R01 DK093674-01 to D.M.), the Leona M. and Harry B. Helmsley Charitable Trust (D.M.), Fundação de Amparo à Pesquisa do Estado de São Paulo (F.A.C.-P.) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (F.A.C.-P.). AUTHOR CONTRIBUTIONS D.M. conceived of and supervised this study and wrote the paper; B.S.R. and D.M. designed experiments; B.S.R., A.R. and D.M. did experiments; B.S.R. prepared figures and helped with manuscript preparation; F.A.C.-P. analyzed and assigned scores to intestinal tissue for inflammation and helped with manuscript preparation; and I.T. provided mouse strains and constructs for overexpression of genes and helped with manuscript preparation. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/doifinder/10.1038/ni.2518. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html.

1. He, X., Park, K. & Kappes, D.J. The role of ThPOK in control of CD4/CD8 lineage commitment. Annu. Rev. Immunol. 28, 295–320 (2010). 2. Xiong, Y. & Bosselut, R. CD4–CD8 differentiation in the thymus: connecting circuits and building memories. Curr. Opin. Immunol. 24, 139–145 (2012). 3. Sakaguchi, S. et al. The zinc-finger protein MAZR is part of the transcription factor network that controls the CD4 versus CD8 lineage fate of double-positive thymocytes. Nat. Immunol. 11, 442–448 (2010). 4. Setoguchi, R. et al. Repression of the transcription factor Th-POK by Runx complexes in cytotoxic T cell development. Science 319, 822–825 (2008). 5. Egawa, T. & Littman, D.R. ThPOK acts late in specification of the helper T cell lineage and suppresses Runx-mediated commitment to the cytotoxic T cell lineage. Nat. Immunol. 9, 1131–1139 (2008). 6. He, X. et al. The zinc finger transcription factor Th-POK regulates CD4 versus CD8 T-cell lineage commitment. Nature 433, 826–833 (2005). 7. Wang, L. et al. The zinc finger transcription factor Zbtb7b represses CD8-lineage gene expression in peripheral CD4+ T cells. Immunity 29, 876–887 (2008). 8. Cheroutre, H., Lambolez, F. & Mucida, D. The light and dark sides of intestinal intraepithelial lymphocytes. Nat. Rev. Immunol. 11, 445–456 (2011). 9. Mucida, D. et al. Transcriptional reprogramming of mature CD4+ helper T cells generates distinct MHC class II–restricted cytotoxic T lymphocytes. Nat. Immunol. advance online publication, doi:10.1038/ni2523 (20 January 2013).



© 2013 Nature America, Inc. All rights reserved.

Articles 10. Denning, T.L. et al. Mouse TCRαβ+CD8αα intraepithelial lymphocytes express genes that down-regulate their antigen reactivity and suppress immune responses. J. Immunol. 178, 4230–4239 (2007). 11. Yeh, J.H., Sidhu, S.S. & Chan, A.C. Regulation of a late phase of T cell polarity and effector functions by Crtam. Cell 132, 846–859 (2008). 12. Yamagata, T., Mathis, D. & Benoist, C. Self-reactivity in thymic double-positive cells commits cells to a CD8αα lineage with characteristics of innate immune cells. Nat. Immunol. 5, 597–605 (2004). 13. Ahern, P.P. et al. Interleukin-23 drives intestinal inflammation through direct activity on T cells. Immunity 33, 279–288 (2010). 14. Sujino, T. et al. Regulatory T cells suppress development of colitis, blocking differentiation of T-helper 17 into alternative T-helper 1 cells. Gastroenterology 141, 1014–1023 (2011). 15. Shi, M.J. & Stavnezer, J. CBFα3 (AML2) is induced by TGF-β1 to bind and activate the mouse germline Ig alpha promoter. J. Immunol. 161, 6751–6760 (1998). 16. Grueter, B. et al. Runx3 regulates integrin αE/CD103 and CD4 expression during development of CD4−/CD8+ T cells. J. Immunol. 175, 1694–1705 (2005). 17. Mucida, D. et al. Reciprocal TH17 and regulatory T cell differentiation mediated by retinoic acid. Science 317, 256–260 (2007). 18. Mucida, D. et al. Retinoic acid can directly promote TGF-β-mediated Foxp3+ Treg cell conversion of naive T cells. Immunity 30, 471–472 (2009). 19. Takahashi, H. et al. TGF-β and retinoic acid induce the microRNA miR-10a, which targets Bcl-6 and constrains the plasticity of helper T cells. Nat. Immunol. 13, 587–595 (2012). 20. Konkel, J.E. et al. Control of the development of CD8((+ intestinal intraepithelial lymphocytes by TGF-β. Nat. Immunol. 12, 312–319 (2011). 21. Lazarevic, V. & Glimcher, L.H. T-bet in disease. Nat. Immunol. 12, 597–606 (2011). 22. Li, M.O., Sanjabi, S. & Flavell, R.A. Transforming growth factor-β controls development, homeostasis, and tolerance of T cells by regulatory T cell-dependent and -independent mechanisms. Immunity 25, 455–471 (2006). 23. Mucida, D. et al. Oral tolerance in the absence of naturally occurring Tregs. J. Clin. Invest. 115, 1923–1933 (2005). 24. Rajaii, F., Bitzer, Z.T., Xu, Q. & Sockanathan, S. Expression of the dominant negative retinoid receptor, RAR403, alters telencephalic progenitor proliferation, survival, and cell fate specification. Dev. Biol. 316, 371–382 (2008). 25. Iwata, M. et al. Retinoic acid imprints gut-homing specificity on T cells. Immunity 21, 527–538 (2004).

10

26. Mucida, D., Park, Y. & Cheroutre, H. From the diet to the nucleus: vitamin A and TGF-β join efforts at the mucosal interface of the intestine. Semin. Immunol. 21, 14–21 (2009). 27. DePaolo, R.W. et al. Co-adjuvant effects of retinoic acid and IL-15 induce inflammatory immunity to dietary antigens. Nature 471, 220–224 (2011). 28. Hall, J.A. et al. Essential role for retinoic acid in the promotion of CD4+ T cell effector responses via retinoic acid receptor α. Immunity 34, 435–447 (2011). 29. Pino-Lagos, K. et al. A retinoic acid-dependent checkpoint in the development of CD4+ T cell-mediated immunity. J. Exp. Med. 208, 1767–1775 (2011). 30. Klinger, M. et al. Thymic OX40 expression discriminates cells undergoing strong responses to selection ligands. J. Immunol. 182, 4581–4589 (2009). 31. Naoe, Y. et al. Repression of interleukin-4 in T helper type 1 cells by Runx/Cbfβ binding to the Il4 silencer. J. Exp. Med. 204, 1749–1755 (2007). 32. Mangan, P.R. et al. Transforming growth factor-β induces development of the TH17 lineage. Nature 441, 231–234 (2006). 33. Ghoreschi, K. et al. Generation of pathogenic TH17 cells in the absence of TGF-beta signalling. Nature 467, 967–971 (2010). 34. Lee, Y. et al. Induction and molecular signature of pathogenic TH17 cells. Nat. Immunol. 13, 991–999 (2012). 35. Leppkes, M. et al. RORγ-expressing Th17 cells induce murine chronic intestinal inflammation via redundant effects of IL-17A and IL-17F. Gastroenterology 136, 257–267 (2009). 36. O’Connor, W. Jr. et al. A protective function for interleukin 17A in T cell-mediated intestinal inflammation. Nat. Immunol. 10, 603–609 (2009). 37. Ono, Y. et al. T-helper 17 and interleukin-17-producing lymphoid tissue inducer-like cells make different contributions to colitis in mice. Gastroenterology 143, 1288–1297 (2012). 38. Zhang, F., Meng, G. & Strober, W. Interactions among the transcription factors Runx1, RORγt and Foxp3 regulate the differentiation of interleukin 17-producing T cells. Nat. Immunol. 9, 1297–1306 (2008). 39. Lazarevic, V. et al. T-bet represses TH17 differentiation by preventing Runx1mediated activation of the gene encoding RORγt. Nat. Immunol. 12, 96–104 (2011). 40. Djuretic, I.M. et al. Transcription factors T-bet and Runx3 cooperate to activate Ifng and silence Il4 in T helper type 1 cells. Nat. Immunol. 8, 145–153 (2007). 41. Muroi, S. et al. Cascading suppression of transcriptional silencers by ThPOK seals helper T cell fate. Nat. Immunol. 9, 1113–1121 (2008). 42. Curotto de Lafaille, M.A. et al. Adaptive Foxp3+ regulatory T cell-dependent and -independent control of allergic inflammation. Immunity 29, 114–126 (2008).

aDVANCE ONLINE PUBLICATION  nature immunology

ONLINE METHODS

© 2013 Nature America, Inc. All rights reserved.

Mice. C57BL/6 (CD45.1+ and CD45.2+) mice, OT-II mice, Rag1−/− mice, Thpokfl/fl mice, Cd4-Cre mice, Rosa26-CreERT2 mice, Ox40-Cre mice and Foxp3-RFP mice (Jackson Laboratories) were maintained at the Rockefeller University animal facilities. Mice with loxP-flanked Tgfbr2 alleles were from the National Cancer Institute. The dnRaralsl/lsl and Runx3-YFP reporter mice were provided by R. Noelle24 and D. Littman, respectively. Several of these lines were interbred at the Rockefeller University animal facilities to obtain the final strains described. Mice were maintained at the Rockefeller University animal facilities under specific pathogen–free conditions, and sentinel mice from the Rag1−/− mouse colony were determined to be negative for Helicobacter species and C. rodentium. Mice were used at 7–12 weeks of age for most experiments. Animal care and experimentation were consistent with guidelines of the US National Institutes of Health and were approved by the Institutional Animal Care and Use Committee of Rockefeller University. Antibodies and flow cytometry. Fluorescent dye–conjugated antibodies were as follows: antibody to CD4 (anti-CD4; 550954), anti-CD25 (553866), antiThy1.1 (561401), anti-CD103 (557495), anti-CD244.2 (553306), anti-IL-17a (559502), anti-T-bet (561312; all from BD Pharmingen); and anti-CD8α (560081) anti-CD44 (56-0441), anti-CD45.1 (25-0453), anti-CD45.2 (47-0454), anti-CD62L (48-0621), anti-TCR-β (47-5961), anti-IFN-γ (25-7311), antiFoxp3 (17-5773) and anti-Vα2 (48-5812; all from eBioscience). Flow cytometry data were acquired on a LSR II flow cytometer (Becton Dickinson) and were analyzed with FlowJo software (TreeStar). A Foxp3 Mouse Regulatory T cell Staining Kit (eBioscience) was used for intracellular staining of Foxp3 and granzyme B. For flow cytometry of cytokine-secreting cells, cells were incubated for 4.5 h in the presence of 100 ng/ml PMA (phorbol 12-myristate 13-acetate; Sigma) and 500 ng/ml ionomycin (Sigma) and with 10 µg/ml brefeldin A (Sigma) for the final 2.5 h before staining. Cell populations were first stained with antibodies to cell surface markers, followed by permeabilization in Fix/Perm buffer and intracellular staining in Perm/Wash buffer (BD Pharmingen). In vitro T cell culture. Naive (CD4+CD25−CD62hiCD44lo) T cells were sorted with a FACSAria (Becton Dickinson) and were cultured for 4.5 d in 96-well plates precoated with 2 µg/ml of anti-CD3ε (17A2; eBioscience) and 1 µg/ml of soluble anti-CD28 (37.51; eBioscience). Cells were then stimulated with cytokines (10 ng IL-1β, 20 ng IL-6, 10 ng IL-12, 10 ng IL-23, 10 nM RA, 2 ng TGF-β (Treg cells) and/or 0.2 ng TGF-β (TH17 cells; all from R&D Systems). OT-II cells were cultured with CD11c+ splenic DCs in the presence of 500 nM OVA peptide and cytokines. For restimulation experiments, cells were cultured for 4.5 d as described above and were resuspended for another 72 h in fresh medium containing cytokines. Retroviral transduction of CD4+ T cells. Retroviral vectors for Runx3 and Cre have been described31,41. Retroviral vectors for T-bet were provided by L. Glimcher (Cornell) and have been described39. Transduction of CD4+ T cells with retroviral vectors was done as described31,41. Sorted naive T cells were activated in vitro for 36–48 h with 2 µg/ml of plate-bound anti-CD3ε and 1 µg/ml of soluble anti-CD28. Cells were then ‘spin-transduced’ with retrovirus for 2 h at 1,250g. Quantitative PCR. Quantitative PCR was done as described17. The housekeeping gene Rpl32 (encoding the ribosomal protein L32) was used for normalization of samples. Primers used were as follows: Thpok forward, 5′-ATGGGATTCCAATCAGGTCA-3′, and reverse, 5′-TTCTTCCTACA CCCTGTGCC-3′; Tbx21 forward, 5′-ATCCTGTAATGGCTTGTGGG-3′,

doi:10.1038/ni.2518 

and reverse, 5′-TCAACCAGCACCAGACAGAG-3′; Rpl32 forward, 5′-GAAA CTGGCGGAAACCCA-3′, and reverse, 5′-GGATCTGGCCCTTGAACC TT-3′; Cd8a forward, 5′-ACTGCAAGGAAGCAAGTGGT-3′, and reverse, 5′-CACCGCTAAAGGCAGTTCTC-3′; Rnx3 forward, 5′-ACAGCATC TTTGACTCCTTCC-3′, and reverse, 5′-TGTTCTCGCCCATCTTGC-3′; Foxp3 forward 5′-CCCATCCCCAGGAGTCTTG-3′, and reverse, 5′-ACCA TGACTAGGGGCACTGTA-3′; and Il17a forward, 5′-TGAGAGCTGCCC CTTCACTT-3′ and reverse, 5′-ACGCAGGTGCAGCCCA-3′. Experimental colitis model. Colitis was induced after transfer of 5 × 105 sorted naive T cells into Rag1−/− mice as described17. For cotransfer experiments, 2.5 × 105 sorted naive T cells from Cd4(∆Runx3) (CD45.2+) mice were injected together with 2.5 × 105 sorted naive T cells from C57BL/6 (CD45.1+) mice. Recipient mice were monitored regularly for signs of disease, including weight loss, hunched posture, and piloerection of the coat and diarrhea, and were analyzed at various times after the initial transfer or when they reached 80% of their initial weight. Colitis was assigned scores by a researcher ‘blinded’ to sample identity according to the following combination of parameters (with a maximum score of 17): extent of inflammation (0, none; 1, lamina propria; 2, submucosal; 3, transmural), degree of inflammation (0, none; 1, mild; 2, moderate; 3, severe), abnormal crypt morphology (0–3), neutrophil infiltration (0–4), goblet cell loss (0, none, 1, moderate, 2, severe), mucosal erosions or ulceration (0 or 1) and crypt abscesses (0 or 1). Tamoxifen treatment. For in vitro treatment, sorted naive T cells were cultured for 4 d with 10 nM 4-hydroxytamoxifen (Sigma). DNA and RNA were extracted for confirmation of treatment efficiency. For in vivo treatment, mice were given intraperitoneal injection every 3 d of 2 mg tamoxifen-citrate (Acros) dissolved in vegetable oil (Sigma) at a concentration of 10 mg/ml, for a total dose of 6 mg of tamoxifen per mouse. C. rodentium infection. Mice were infected with 2 × 108 C. rodentium per mouse, as described32. Bacteria were inoculated by gavage in recipient mice in a total volume of 200 µl. After infection, mice were monitored daily for weight loss and abundance of colony-forming units in feces and liver. Mice were killed and analyzed 18 d after infection. Inflammation induced by C. rodentium infection was assigned scores by a researcher ‘blinded’ to sample identity according to the following combination of parameters (with a maximum score of 14): extent of inflammation (0, none; 1, lamina propria; 2, submucosal; 3, transmural), severity of epithelial injury (0–3), goblet-cell loss (0 or 1), mucosal hypertrophy (0 or 1), neutrophil infiltration (0–3) and mononuclear cell infiltration (0–3). Oral OVA treatment. Mice were fed regular chow containing 1% chicken OVA for 7 d and analyzed 3 d later. For measurement of diarrhea, mice were challenged daily by gavage with 50 mg OVA solution (250 mg/ml) until diarrhea was detected. Mice were classified as positive when signs of diarrhea were detected two consecutive times after challenge. Tissue inflammation was assigned scores as described above for colitis. Preparation of intraepithelial and lamina propria lymphocytes. Intraepithelial and lamina propria lymphocytes were isolated as described17. Statistics. GraphPad Prism software was used for statistical analysis. Data were analyzed by one-way analysis of variance or an unpaired Student’s t-test whenever necessary. For analysis of histological scores, a nonparametric Mann-Whitney test was used. A P value of less than 0.05 was considered significant.

nature immunology