Full Text

Turn on search term navigation

ARTICLE

Received 22 Sep 2016 | Accepted 19 Apr 2017 | Published 5 Jun 2017

DOI: 10.1038/ncomms15677 OPEN

Ndp1 restricts mTORC1 signalling and glycolysis in regulatory T cells to prevent autoinammatory disease

Awo Akosua Kesewa Layman1,2,*, Guoping Deng3,*, Claire E. OLeary3,*, Samuel Tadros3, Rajan M. Thomas4, Joseph M. Dybas4, Emily K. Moser3, Andrew D. Wells4, Nicolai M. Doliba5 & Paula M. Oliver3,4

Foxp3 T regulatory (Treg) cells suppress immune cell activation and establish normal immune homeostasis. How Treg cells maintain their identity is not completely understood. Here we show that Ndp1, a coactivator of Nedd4-family E3 ubiquitin ligases, is required for

Treg cell stability and function. Ndp1 deletion in Treg cells results in autoinammatory disease. Ndp1-decient Treg cells are highly proliferative and are more likely to lose Foxp3 expression to become IL-4-producing TH2 effector cells. Proteomic analyses indicate altered metabolic signature of Ndp1-decient Treg cells and metabolic proling reveals elevated glycolysis and increased mTORC1 signalling. Ndp1 restricts Treg cell metabolism and IL-4 production via distinct mechanisms, as IL-4 deciency does not prevent hyperproliferation or elevated mTORC1 signalling in Ndp1-decient Treg cells. Thus, Ndp1 preserves Treg lineage stability and immune homeostasis by preventing the expansion of highly proliferative and metabolically active Treg cells and by preventing pathological secretion of IL-4 from Treg cells.

1 Medical Scientist Training Program, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA. 2 Biomedical Graduate Studies, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA. 3 Cell Pathology Division, The Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania 19104, USA. 4 Department of Pathology and Laboratory Medicine, The Childrens Hospital of Philadelphia, Philadelphia, Pennsylvania 19104, USA. 5 Institute for Diabetes, Obesity, and Metabolism, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104, USA. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to P.M.O. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

Foxp3 regulatory T (Treg) cells suppress spontaneous immune cell activation and limit effector cell function, thereby preventing autoimmune and inammatory

disorders1,2. While generally stable, Treg lineage cells can have a high degree of instability in inammatory settings. Treg cell

instability is characterized by the loss of suppressive function, loss of Foxp3 protein or gain of pro-inammatory cytokine production3,4. Dening pathways that help to establish and maintain Foxp3 expression, promote Treg cell suppressive function, prevent Treg cell production of pro-inammatory cytokine and/or maintain Treg cell numbers will aid in the development of new Treg cell-based therapeutic applications.

Appropriate regulation of cellular energetics and metabolism is important for Treg cell function and lineage stability5. Unlike effector T cells, which rely heavily on glycolysis, Treg cells rely on fatty acid oxidation for their energy needs6. Mechanistic target of rapamycin (mTOR) is a serinethreonine kinase that forms part of the mTORC1 and mTORC2 protein complexes and is a critical regulator of cellular metabolic processes. Both complexes can limit glycolysis in Treg cells thereby promoting lineage stability and suppressive functions7,8. Although metabolic state is clearly important for Treg maintenance and function, many factors that impact Treg cell metabolism remain unknown.

Ubiquitylation is a fundamental post-translational modication affecting many aspects of T-cell differentiation and function9,10. Neural precursor cell expressed, developmentally downregulated 4 (Nedd4) family interacting protein 1 (Ndp1) is a transmembrane protein that binds and activates Nedd4 family E3 ubiquitin ligases11. The highly conserved catalytic E3 ligases perform two functions in protein ubiquitylation: binding to the specic ubiquitylation target and catalysing the nal transfer of ubiquitin. Ndp1 activation of the Nedd4 E3 ligase Itch results in ubiquitylation and degradation of the transcription factor JunB, thereby limiting interleukin (IL)-4 cytokine production from T helper type 2 (TH2) cells and TH2-mediated inammatory disease12,13.

Ndp1-decient mice have decreased Treg cell numbers in the small bowel, a site of peripheral Treg generation, likely due to increased IL-4 signalling, which is inhibitory to Treg

differentiation14. However, whether Ndp1 also modulates Treg function after cells have committed to the Treg cell lineage has not been explored. Given that Treg-specic deletion of Itch results in a TH2-biased autoinammatory disease15, it seems plausible that Ndp1 might be required to support Itch function in Treg cells.

Here we show that Ndp1 expression in Treg cells prevents spontaneous inammation at several sites, such as the lungs and skin. Ndp1 limits both the accumulation and proliferation of CD44 effector Treg cells and prevents Treg cell production of

IL-4. Consistent with increased proliferation and exposure to IL-4, Treg cells lacking Ndp1 show increased conserved non-coding DNA sequence 2 (CNS2) methylation and are prone to losing Foxp3 expression in vivo. Increased T-cell proliferation is associated with increased mTORC1 signalling and high glycolytic activity, metabolic programmes that can fuel effector function in Treg cells and contribute to Treg lineage instability.

Thus Ndp1 maintains lineage identity in Treg cells and prevents these cells from aberrant acquisition of effector T-cell function. Ndp1 is therefore a critical molecular sentinel that prevents autoinammatory disease.

ResultsTreg-specic loss of Ndp1 results in inammation. Ndp1 /

mice develop a severe autoinammatory disease by 6 weeks of age, resulting in death12,16. While thymic Treg output in Ndp1-decient mice is not altered14, Treg cell numbers are reduced at the sites of peripheral Treg induction14. Therefore, to test the role

of Ndp1 within committed Treg cells, we generated mice in which Ndp1 is conditionally deleted in Treg cells using the Foxp3-Cre-YFP reporter mice17. We observed that Ndp1 mRNA is induced upon stimulation of control Ndp1 / Foxp3-Cre-sorted YFP Treg cells and that Ndp1 message is effectively ablated in Treg cells from Ndp1/ Foxp3-Cre male mice, which lack Ndp1 in all Treg cells (Fig. 1a).

By 916 weeks of age, Ndp1/ Foxp3-Cre male mice developed pronounced splenomegaly, lymphadenopathy and progressive dermatitis (Fig. 1b). Histologically, the skin, oesophagus and lung showed marked immune inltration inammation (Fig. 1b). Analysis of spleen weight-to-body weight ratios revealed that male Ndp1/ Foxp3-Cre mice showed evidence of lymphoid expansion (Fig. 1c). To determine which immune responses were contributing to the observed inammation, we examined serum immunoglobulin (Ig) levels. We found elevated levels of IgE and IgG1, indicative of type 2 inammation, as well as elevated IgM (Fig. 1d). To determine the contribution of CD4 T cells to this pathology, we examined the spleen and lung of the mice and found increased activated phenotype (CD44 ) CD4 T cells (Fig. 1e). CD4 T cells present in the lung of male Ndp1/ Foxp3-Cre animals were more likely to express the effector cytokines interferon-g (IFNg), IL-4 and

IL-17A upon ex vivo stimulation (Fig. 1f). These data suggested that Ndp1 expression in Treg cells is required for suppression of tissue inammation and pathology.

Ndp1/ Foxp3-Cre mice have more CD44 eTreg cells. The observed pathology could result from a loss of Treg cell numbers

(as occurs with mice lacking Foxp318) or Treg cell function (as is seen in Treg cell-specic CTLA4 deciency19). We therefore examined Treg percentages and numbers in the spleens and lung of Ndp1/ Foxp3-Cre and controls. Surprisingly, Treg cell numbers were increased in Ndp1/ Foxp3-Cre animals (Fig. 2ac). Further, analysis of the Treg cell effector proteins

inducible T-cell costimulator (ICOS), programmed cell death-1 (PD-1) and glucocorticoid-induced TNFR-related protein (GITR)revealed increased expression on Treg cells from

Ndp1/ Foxp3-Cre male mice compared to Cre controls (Fig. 2d). We did not see marked changes in CD25 levels. Taken together, these data suggest that the immunopathology observed in Ndp1/ Foxp3-Cre male mice is not due to decreased Treg numbers or loss of effector proteins known to support Treg

function.

Treg cells can be quiescent or activated; these two subsets can be distinguished by the expression of CD44 and CD62L (ref. 20). Analysis of the lung of male Ndp1/ Foxp3-Cre mice revealed an increase in frequency (Fig. 2e,f) and numbers (Fig. 2g) of Foxp3 cells with an activated or effector (eTreg) phenotype (CD62LloCD44 ). In contrast, Ndp1/ Foxp3-Cre and control mice contained equivalent numbers of cells with a quiescent or central (cTreg) phenotype (CD62LhiCD44 ) (Fig. 2g). ICOS,

PD-1 and GITR are more highly expressed on eTreg cells. In contrast, CD25 is particularly high on cTreg cells20. Thus the observed increase in surface expression of ICOS, PD-1 and GITR observed on total Treg cells from Ndp1/ Foxp3-Cre animals could be due to an increased proportion of eTreg cells. To assess this, we examined the surface expression of ICOS, GITR, CD25 and PD-1 on the eTreg and cTreg populations. cTreg cells, as expected, had low levels of ICOS, GITR and PD-1, and loss of Ndp1 did not alter this (Fig. 2h). However, cTreg cells from

Ndp1/ Foxp3-Cre mice had decreased levels of CD25. Strikingly, in addition to their increased frequency, eTreg cells

from Ndp1/ Foxp3-Cre mice showed higher levels of ICOS, GITR, CD25 and PD-1 compared to wild-type (WT) counterparts (Fig. 2h).

2 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

cTreg cells are thymically derived and undergo peripheral conversion to eTreg cells under the instruction of T-cell receptor (TCR) stimulation and ICOS stimulation20. Therefore, observing increased frequency and number of eTreg cells could indicate either increased conversion of cTreg cells or expansion of eTreg cells. An increased cTreg conversion rate would be expected to decrease the numbers of cTreg cells, unless compensated for by thymic output. As there was no decrease in the peripheral numbers of cTreg cells in Ndp1/ Foxp3-Cre male mice, we next examined thymic output utilizing mixed bone marrow chimera animals in which WT (Ndp1 / Foxp3-Cre)

and Ndp1/ Foxp3-Cre Treg cells develop in the same environment to control for effects of inammation. In control and mixed chimeras, total Treg thymic output was unchanged (Supplementary Fig. 1a). Furthermore, within the mixed chimeras there was no signicant difference in the percentge of thymic Treg cells derived from Ndp1-sufcient or -decient cells relative to the observed reconstitution ratio of all thymic CD4 T cells (Supplementary Fig. 1b). This suggests that the increase in eTreg, but not in cTreg, cell number in the

periphery is not due to an altered rate of cTreg cell conversion to eTreg cells.

0.020

Ndfip1fl/fl Foxp3-Cre

Relative expression

Foxp3-Cre WT

0.015

0.010

Ndfip1fl/fl

0.005

0.000

Skin Oesophagus Lung Lymph nodes

Spleen

CD3/CD28:

Ndfip1 mRNA

+ +

CD44+ CD4 T cells

Inflammation index

Serum antibody

Spleen wt (g) / body wt (kg)

****

Ndfip1fl/fl Foxp3-Cre

600 ****

400

250

g ml1

200

150

100

IgA IgE IgG1 IgG2C IgM

WT Ndfip1fl/fl Foxp3-Cre

Lung

****

***

%cytokine+ CD4+ T cells

% CD44+ CD4+ T cells

***

Spleen Lung

IFN IL-4 IL-17A

Figure 1 | Mice lacking Ndp1 in their Treg cells develop inammatory disease. (a) Ndp1 expression assessed by qPCR before ( ) or after ( )

aCD3/CD28 or PMA/ionomycin (P/I) stimulation of sorted YFP Treg cells from WT and Ndp1/ Foxp3-Cre mice. A representative example of Ndp1 expression relative to Actb after aCD3/CD28 stimulation is shown. (b) Representative Haematoxylin and Eosin-stained histological sections of the skin, oesophagus and lung from genotypes as indicated are shown. Scale bars represent 100 mM. Far right image in panel is a representative image of the spleen and lymph nodes to illustrate size. (c) Inammation index, calculated as a spleen weight/body weight for male Ndp1/ Foxp3-Cre (WT) and Ndp1/

Foxp3-Cre (cKO) mice at 916 weeks of age. (d) Levels of serum antibody isotypes as quantied by ELISA. (e,f) T cells from lung homogenates were analysed by ow cytometry for (e) the percentages of CD44 cells among CD4 cells and (f) the percentages of CD4 cells producing the indicated cytokines after ex vivo (P/I) stimulation. P values determined by Students t-test, with correction for unequal variances as appropriate. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001. For a,cf, bars indicate means.e.m. Data in a are representative of four male animals of each genotype, 916 weeks, and for cf each dot represents an individual male mouse aged between 9 and 16 weeks. All experiments were performed on at least two independent occasions.

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

a b c

WT Ndfip1fl/fl Foxp3-Cre

105

Treg (%) Treg (#)

13.7

18.5

104

WT Ndfip1fl/fl

Foxp3-Cre

***

Spleen

103

102

# Foxp3+ cells

(106)

103

104

105

% Foxp3+ cells

105 104 103 102

8.2

14.5

0.50

Lungs

0.25

Foxp3

0.0

Spleen

Lungs

Spleen

Lungs

0 105

102

103

104

CD4

WT Total Treg

Ndfip1fl/fl Foxp3-Cre Total Treg

ICOS

GITR

CD25 PD-1

100

MFI

0 102

103

104

105

e f g

Treg (%) Treg (#)

Ndfip1fl/fl Foxp3-Cre

100

****

105

104

103

# Foxp3+ cells

(105)

****

58.3

38.6

12.1

86.2

% Foxp3+ cells

1.5

CD62L

1.0

0.5

0.0

102

103

104

105

cTreg eTreg cTreg eTreg

CD44

WT Ndfip1fl/fl Foxp3-Cre

ICOS

GITR

CD25 PD-1

5,000

15,000

5,000 ***

2,500 **

4,000

12,000

2,000

4,000

3,000

9,000

1,500

3,000

MFI

2,000

6,000

1,000

2,000

1,000

3,000

500

1,000

cTreg eTreg cTreg eTreg

Figure 2 | Ndp1-decient eTreg cells are increased in number and Treg cell surface expression. Treg cells from the spleens and lung (ac) or lung (dh) of 916-week-old male WT and Ndp1/ Foxp3-Cre mice were analysed by ow cytometry for numbers and surface markers. (a) Representative ow plots of

Treg cells (previously gated as live CD3 CD4 cells) from the spleen and lung homogenates that were used to determine the percentages and total numbers. (b) Treg percentages and (c) numbers compiled over multiple experiments. (d) Representative histograms of lung Treg cells (gated as in a) analysed for surface expression of ICOS, GITR, CD25 and PD-1. (e) Representative ow plots showing the gating of Treg cells that are effector(eTreg, CD44hiCD62Llo) or central (cTreg, CD62LhiCD44lo). This gating was used to determine the (f) percentages and (g) total numbers of these cells.

(h) Quantication of mean uorescence intensities (MFIs) of ICOS, GITR, CD25 and PD-1 on cTreg cells and eTreg cells. P values were determined by Students t-test, with correction for unequal variances as appropriate. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001. Each dot indicates data acquired from a single male mouse. Graphs with compiled data show means.e.m. All experiments were performed on at least two independent occasions.

Decreased suppressive function of Treg cells is known to lead to inammation. Therefore, we examined Treg cell function. In vitro, Ndp1-decient and WT Treg cells suppressed proliferation of WT T conventional (Tconv) cells to the same degree

(Supplementary Fig. 1c,d). We then examined Treg cell function

in vivo using a model of T-cell transfer-induced colitis. Similar to our results from the in vitro assays, Ndp1-decient and WT Treg

cells were equally able to prevent weight loss due to inammation

4 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

caused by cotransferred Tconv cells (Supplementary Fig. 1e). These data support that the pathology observed in Ndp1/ Foxp3-Cre mice is not due to a loss of Treg number or an overall loss of Treg cell-suppressive function.

Ndp1 limits Treg proliferation and levels of ICOS and GITR. The expansion of CD44 eTreg cells in male Ndp1/ Foxp3-Cre animals in the absence of robust changes in cTreg cells suggested that Ndp1 restricts eTreg cell numbers. Additionally, our data supported that Ndp1 limits the expression of ICOS, GITR, CD25 and PD-1 on eTreg cells. However, both expansion and phenotype of eTreg cells could be altered in inammatory environments, independent of an intrinsic role for Ndp1. To distinguish between these possibilities, we examined Ndp1/ Foxp3-Cre / female animals. These mice have mixtures of

WT (YFP-Cre ) and Ndp1-decient (YFP-Cre ) Foxp3 Treg cells, due to X chromosome inactivation, that can be distinguished by the expression of the YFP Cre reporter.

Surprisingly, despite the presence of WT Treg cells, female Ndp1/ Foxp3-Cre / mice had similar inammatory burden, as dened by the ratio of spleen weight to body weight (Supplementary Fig. 2a) and contained equally high percentages of IFNg-, IL-4- and IL-17A-expressing CD4 T cells, as the males (Supplementary Fig. 2b). Compared to male animals, which are uniformly sick beyond 16 weeks of age, female animals had lower incidence of skin dermatitis. Ndp1/ Foxp3-Cre / mice contained increased numbers of total lung Treg cells compared to Ndp1/ Foxp3-Cre / counterparts (Fig. 3a,b). Similar to their male counterparts, female animals had total lung Treg cells

that were skewed towards the eTreg phenotype (Fig. 3c,d). In control Ndp1 / Foxp3-Cre / female mice, Foxp3-Cre (YFP ) cells were outnumbered by Foxp3-Cre (YFP ) cells by almost 1:3, indicating an effect of Cre expression (Fig. 3f).

Surprisingly, in Ndp1/ Foxp3-Cre / females, this ratio was skewed towards YFP cells, supporting an in vivo competitive advantage of Ndp1-decient Treg cells (Fig. 3f).

To identify intrinsic effects of Ndp1 deciency on Treg cells,

in each female animal, for each parameter examined, we rst determined the ratio of the value for YFP -to-YFP Treg cells to normalize for an effect of Cre. Then we compared this ratio between Ndp1/ Foxp3-Cre / and Ndp1 / Foxp3-Cre/

animals. Our analysis revealed that Ndp1-decient Treg cells

were signicantly more likely to display an eTreg phenotype (CD44 CD62Llo) (Fig. 3g). Furthermore, expression of the proliferative marker Ki67 (Fig. 3h) was increased in Ndp1-decient eTreg cells. eTreg cells lacking Ndp1 also had signicantly higher levels of ICOS (Fig. 3i) and GITR (Fig. 3j). Thus Ndp1 restricts eTreg numbers by limiting their proliferation and their expression of ICOS and GITR. However, levels of CD25 and PD-1 were similar to controls (Fig. 3k,l), suggesting that these markers were not directly affected by the loss of Ndp1 in Treg cells.

The expression of activation markers on Treg cells is known to increase under inammatory settings20. To determine how inammatory conditions could inuence control Treg cells, we looked at the expression of surface markers on CD44 YFP

Treg cells from the Ndp1 / Foxp3-Cre / females or from the inamed environment in Ndp1/ Foxp3-Cre / female mice. CD44 YFP Treg cells from Ndp1/ Foxp3-Cre /

mice showed increased expression of ICOS, GITR, CD25 and PD-1 (Supplementary Fig. 2cf). Taken together with the data in Figs 2 and 3, this supports that the changes in CD25 and PD-1 expression in male Ndp1/ Foxp3-Cre mice are secondary to inammation and that while inammation can increase levels of ICOS and GITR, Ndp1 is also an intrinsic regulator of ICOS and GITR on Treg cells (Supplementary Fig. 2g).

To further address whether the changes observed in the Ndp1-decient Treg cells were driven by inammation, we looked in neonatal mice before the onset of overt inammation. In 13-day-old female neonates, we found no differences between Ndp1/ Foxp3-Cre / and Ndp1/ Foxp3-Cre /

animals in spleen weight (Supplementary Fig. 3a), inammation index (Supplementary Fig. 3b), total lung Foxp3 Treg cell number or cytokine-producing CD4 T cells (Supplementary

Fig. 3c,d). However, lungs from Ndp1/ Foxp3-Cre / female mice contained greater frequencies of YFP Treg cells (Supplementary Fig. 3e) and greater frequencies of eTreg cells

(Supplementary Fig. 3f). Further, these eTreg cells showed higher expression of ICOS and GITR (Supplementary Fig. 3g,h). This further supports that Ndp1 limits eTreg cell frequency

and expression of ICOS and GITR in an intrinsic manner.

Ndp1 limits IL-4 production by Treg cells. Our data indicate that Ndp1 limits eTreg cell expression of ICOS and GITR and restricts eTreg cell proliferation. However, it remained unclear why Ndp1/ Foxp3-Cre / females developed inammation, when approximately half of the Treg cells in these mice are Ndp1 sufcient. This suggested a pathological gain of function in Ndp1-decient Treg cells. We thus investigated whether Ndp1-decient Treg cells could contribute to the pool of cytokine-producing cells in the lung of Ndp1/ Foxp3-Cre male mice shown in Fig. 1. Effector cytokine production by WT Treg cells is relatively rare but has been described under inammatory21 or lymphopenic settings22,23. Strikingly, while WT Treg cells did not produce any IL-4 upon ex vivo stimulation, Ndp1-decient Treg

cells could produce IL-4, and this was detectable at both the protein (Fig. 4a,b) and mRNA levels (Fig. 4c). Ndp1-decient Treg cells were also more likely to produce IL-10, IFNg and

IL-17A relative to WT controls. To determine whether the cytokine production was due to the loss of Ndp1 or tied to the inammatory environment, we generated mixed bone marrow chimeras. Consistent with our data from the female mice, these mixed chimeras developed dermatitis 8 weeks after reconstitution. Upon restimulation of lung homogenate, CD45.1 WT Foxp3 cells did not produce any IL-4 while

CD45.2 Ndp1-decient Foxp3 cells in the same host could produce IL-4 (Fig. 4d,e), supporting an intrinsic role for Ndp1 in limiting IL-4 production from Treg cells. Surprisingly, when we analysed the Foxp3 Tconv cells, we found a signicant population of CD45.2 cells that also expressed IL-4 (Fig. 4f,g), raising the possibility that the IL-4-producing population originated from Ndp1/ Foxp3-Cre Treg cells that had lost Foxp3.

Ndp1-decient Treg cells lose Foxp3 expression. Given the IL-4 production by Ndp1-decient Treg cells and their increased proliferative capacity, we posited that these cells would be likely to become methylated at their Foxp3 locus and become unstable. IL-4 receptor signalling in Treg cells, via signal transducer and activator of transcription factor 6 (STAT6), results in the methylation of the CNS2 region of the Foxp3 locus and repression of Foxp3 gene expression24. Therefore, we compared the methylation at 12 Cytosine-phosphate-guanine (CpG) islands in the Foxp3 CNS2 region in Ndp1-sufcient and -decient Treg cells, as well as WT Tconv cells. As expected, the CNS2

CpG motifs were predominantly methylated in Tconv cells

(Supplementary Fig. 4a)24. Both WT and Ndp1-decient Treg cells had unmethylated CpG motifs in their Foxp3 promoter regions (Supplementary Fig. 4b), consistent with their expression of Foxp3 mRNA (Supplementary Fig. 4c). However, in Ndp1/ Foxp3-Cre Treg cells, there was an increase in methylation at the

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

a b

WT Foxp3-Cre+/

Ndfip1fl/fl Foxp3-Cre+/

Total Treg

WT Foxp3-Cre+/

female Ndfip1fl/fl

Foxp3-Cre+/ female

****

105

10.4

14.7

104

103

%Foxp3+

102

CD4

102

103

104

105

Foxp3

c d

Effector Treg (eTreg) (%)

105

100

%CD44+CD62L

Foxp3+

****

104

20.6

76.4

103

CD62L

103

104

105

CD44

e f

YFP+ : YFP

105

104

103

102 0

64.2

28.4

52.6

43.3

2.0

Ratio of YFP+:YFP

Foxp3+

1.5

1.0

CD4

0.5

0 105

104

103

102

YFP

0.0

g h i

eTreg (%)

Ki67+ eTreg (%) ICOS MFI eTreg

1.25 ****

1.5 ****

1.5 **

Ratio of YFP+:YFP

CD44+ Foxp3+

1.00

1.0

0.75

0.50

0.5

0.25

0.00

0.0

j k l

CD25 MFI eTreg

GITR MFI eTreg

PD-1 MFI eTreg

1.5 *

2.5

2.0 NS

Ratio of YFP+:YFP

CD44+ Foxp3+

2.0

1.0

1.5

1.0

0.5

1.0

0.5

0.0

Figure 3 | Female Ndp1/Foxp3-Cre / mice show similar Treg cell changes as males. (al) Lung homogenates, from 8- to 16-week-old hemizygous female Ndp1/ Foxp3-Cre / mice and Ndp1/ Foxp3-Cre / controls, were analysed ex vivo by ow cytometry. (a) Representative ow plots showing the gating of Treg cells (previously gated as live CD3 CD4 cells). This gating was used to determine the overall frequencies of Treg cells (b) in mice from the indicated genotypes. (c) Representative ow plots showing the gating of Treg cells from a that are eTreg or cTreg cells. This gating was used to determine the (d) frequencies of eTreg cells. (e) Representative ow plot of Treg cells as gated in a and analysed for frequencies of YFP and YFP cells.

(f) Ratio of YFP to YFP Treg cells. (g,h) Percentages of (g) eTreg cells or (h) Ki-67 eTreg cells, shown as a ratio of YFP :YFP cells within each mouse to normalize for a Cre effect. (il) mean uorescence intensity (MFI) of (i) ICOS, (j) GITR, (k) CD25 and (l) PD-1 expression on eTreg cells, again shown as a ratio of YFP:YFP . P values determined by Students t-test, with correction for unequal variances as appropriate. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001. Each dot shows data acquired from a single female mouse. Graphs show means.e.m. All experiments were performed on at least two independent occasions.

6 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

a b

WT Ndfip1fl/fl Foxp3-Cre

IL-4+ Treg (%)

%IL-4+ of Foxp3+

105

104

103

102

0.78

4.82

CD4

102 103 104 105

WT Ndfip1fl/fl

Foxp3-Cre

IL-4

Il4 mRNA

10.0

RelatIve expression

(x104)

8.0

6.0

4.0

WTNdfip1fl/fl Foxp3-Cre

2.0

0.0

CD3/CD28:

+ +

CD45.1 WT

CD45.2 Ndfip1fl/fl Foxp3-Cre

IL-4+ Treg (%)

105

104

103

%IL-4+ of Foxp3+

8.22

IL-4

102

103 104 105

Bone marrow:

CD45.1 WT

CD45.2 Ndfip1fl/fl

Foxp3-Cre

CD4

f g

CD45.1 WT

CD45.2 Ndfip1fl/fl Foxp3-Cre

IL-4+ Foxp3- (%)

8 **

105

104

103

%IL-4+ of Foxp3-

0.42

6.75

IL-4

102

103 104 105

Bone marrow:

0 CD45.1 WT

CD45.2 Ndfip1fl/fl

Foxp3-Cre

CD4

Figure 4 | Ndp1-decient Treg cells make IL-4. (a,b) Lung homogenates from WT and Ndp1/ Foxp3-Cre male mice aged 916 weeks were stimulated ex vivo with P/I and analysed for IL-4 production by ow cytometry. (a) Representative ow plots of Treg cells (Foxp3CD3 CD4 ) producing IL-4.

(b) Compiled percentages of IL-4-producing Treg cells. (c) qPCR analysis of Il4 from sorted YFP Treg cells from WT and Ndp1/ Foxp3-Cre male mice. Cells were unstimulated ( ) or stimulated ( ) with aCD3/CD28. Il4 mRNA is shown relative to Actb. (dg) Lung homogenates from mixed chimeras

were stimulated ex vivo with P/I and analysed for IL-4 production by ow cytometry. (d) Representative ow plots showing IL-4 production from Treg cells

from WT (gated on CD45.1) or Ndp1/ Foxp3-Cre (gated on CD45.2) cells from the same recipient. (e) Graphs showing the percentages of IL-4 Treg cells from CD45.1 and CD45.2 cells in each chimera; cells from the same recipient are connected by a line. (f) Representative ow plots showing IL-4 production from Tconv cells from the same chimeras. (g) Graphs showing the percentages of IL-4 Tconv cells as per e. All experiments were performed on at least two independent occasions. Graphs show means.e.m. Mixed bone marrow chimeras were generated and analysed in two separate experiments using male donors. P values were determined by Students t-test for b,c or paired T-test. *Po0.05, **Po0.01.

12 examined CNS2 CpG sites, compared to WT Treg cells, which remained unmethylated at these sites (Fig. 5a,b).

These data suggested that Ndp1 deciency, via concomitant expression of IL-4, negatively impacts lineage stability of Treg

cells. To test whether Ndp1-sufcient and -decient Treg cells

have differential responses to destabilizing cytokines, we cultured Ndp1-sufcient and -decient Foxp3 cells in stabilizing (IL-2)

or destabilizing (IL-4 plus anti-IL-2) conditions and analysed Foxp3 protein expression as a surrogate for lineage stability. We found that Ndp1-decient Treg cells were equally stabilized

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

by IL-2 and destabilized by IL-4 compared to their Ndp1-sufcient counterparts (Supplementary Fig. 4d,e). Thus Ndp1-decient Treg cells are not uniquely sensitive to the

destabilizing effects of IL-4; in vivo, however, Ndp1-decient IL-4 producing Treg cells are more likely to encounter destabilizing cytokine milieus.

a b

Unmethylated CpG

Methylated CpG

Foxp3 CNS2

100

4,2754,3114,3274,3614,3884,3934,4424,4614,4654,4734,5354,605 0

% of alleles methylated at

the indicated cytosine position

+4275

+4,311

+4,327

+4,361

+4,388

+4,393

+4,442

+4,461

+4,465

+4,473

+4,535

+4,605

Cytosine position

Ndfip1fl/fl

Foxp3-Cre

WT Treg

Ndfip1fl/fl

Foxp3-Cre Treg

CD45.2 YFP+ Treg + CD45.1 naive Tconv

13 weeks

Analyse

d e

CD45.2 WT Treg

CD45.2 Ndfip1fl/fl Foxp3-Cre Treg

100 **

105

104

103

102

% of total CD45.2

CD4+ T cells

35.9 36.2

61.2

61.3

CD4

102 103 104 105

+ +

Foxp3:

CD45.2+ WT

CD45.2+ Ndfip1fl/fl Foxp3-Cre

Foxp3

f g h

WT Ndfip1fl/fl Foxp3-Cre WT Ndfip1fl/fl Foxp3-Cre

IL-4+ Current Treg IL-4+ Former Treg IL-4+ CD45.1 Tconv cell

WT Treg Ndfip1fl/fl Foxp3-Cre Treg

80 ****

% CD45.2+ IL-4+

Foxp3+ CD4+

%CD45.2+ IL-4+

Foxp3- CD4+

%CD45.1+ IL-4+

Foxp3

Foxp3:

CD45.2

+ +

Foxp3:

With:

CD45.2

Figure 5 | Ndp1-decient Treg cells have Foxp3 locus instability and lose Foxp3 in vivo. (a,b) YFP eTreg cells were sorted from 9- to 12-week-old WTor Ndp1/ Foxp3-Cre male mice and assessed for CNS2 methylation using bisulte sequencing. (a) Representative data of methylation at the Foxp3 CNS2 locus in Treg cells from the two genotypes. (b) Quantication of methylation at 12 CpG islands in the Foxp3 CNS2 locus. Data shown are compiled from three mice of each genotype. Data are the percentage of alleles methylated at each position. (ch) YFP Treg cells from CD45.2 WT or Ndp1/

Foxp3-Cre mice were mixed 1:5 with CD45.1 WT naive Tconv cells and transferred to Rag1 / recipients. Cells from lung homogenates were analysed for Foxp3 and IL-4 expression using ow cytometry 13 weeks after transfer. (d) A representative ow plot showing CD45.2 WTand Ndp1/ Foxp3-Cre cells that were analysed for the expression of Foxp3. (e) Compiled data from multiple mice analysed as in d. (f) Percentages of IL-4-producing cells that remained Foxp3 (current Treg cells ) and (g) that had become Foxp3 following transfer (former Treg cells ). (h) Percentages of IL-4-producing CD45.1

Tconv cells. All experiments were performed on at least two separate occasions. Each dot represents cells from an individual mouse. P values were calculated by one-way ANOVA. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001. Graphs show means.e.m.

8 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

To test Treg lineage stability in vivo, we sorted WT or Ndp1-decient CD45.2 YFP Treg cells and cotransferred them with sorted naive CD45.1 WT Tconv cells into Rag1 / hosts (Fig. 5c). As we had previously observed (Supplementary Fig. 1e), mice receiving either Ndp1-decient or -sufcient Treg cells

did not develop colitis (Supplementary Fig. 5a). At harvest, we determined the frequencies of CD45.2 cells that were still

Foxp3 (current Treg cells) or that had lost Foxp3 (former Treg cells). In mice that received WT Treg cells, on average, 30% had lost Foxp3, similar to previous reports by other groups8,22.

Strikingly, in mice that had received Ndp1-decient Treg cells,

60% on average had lost expression of Foxp3 (Fig. 5d,e). These Foxp3 CD45.2 cells could represent Ndp1-decient Treg cells that had lost Foxp3 expression in vivo or a very proliferative

Tconv cell contamination from cell sorting. If the Foxp3 cells represented former Treg cells, they would lack Ndp1 mRNA, due to excision following the expression of Foxp3-Cre. Contaminating Tconv cells, on the other hand, would continue to express Ndp1.

We sorted current Treg cells, former Treg cells and Tconv cells and stimulated to induce the Ndp1 mRNA expression. We found that the WT Tconv cells expressed Ndp1 mRNA, while neither the Ndp1/Foxp3-Cre current Treg cells nor former Treg cells expressed detectable levels of Ndp1 mRNA (Supplementary

Fig. 5b). Thus Treg cells lacking Ndp1 were more likely to lose Foxp3 expression than their WT counterparts.

Current (Fig. 5f) and former (Fig. 5g) Treg cells lacking Ndp1 produced IL-4, while current and former WT Treg cells did not. Importantly, none of the WT Tconv cells produced IL-4 (Fig. 5h). This was not true for all cytokines since WT and Ndp1-decient cells were equally likely to make IFNg regardless of whether they were Treg cells (Supplementary Fig. 5c,d) or cotransferred Tconv cells (Supplementary Fig. 5e).

Additionally, compared to WT Treg cells, Ndp1-decient Treg cells expanded to greater numbers in vivo (Supplementary Fig. 5f,g) suggesting an advantage in proliferation and/or survival of both current and former Ndp1-decient Treg cells. However, at steady state, it is likely that only a small percentage of Ndp1/Foxp3-Cre Treg cells are actively losing their Foxp3 at any point of time. In support of this, Ndp1-decient total Treg

cells, eTreg cells and cTreg cells from female animals do not show a decrease in Foxp3 mean uorescence intensity at steady state (Supplementary Fig. 6).

In summary, in a setting where Treg cells are pushed to undergo lymphopenia-induced proliferation, Ndp1-decient Treg cells

have an advantage in growth and expansion, which leads to a large increase in total numbers of Ndp1-decient current and and former Treg cells. Since the majority of Ndp1-decient former Treg cells (B50%) produce IL-4, this may explain the dramatic loss of Foxp3 observed in vivo.

IL-4 is dispensable for Ndp1-decient eTreg expansion. To assess whether IL-4 is required for the phenotypic changes in Treg

cells lacking Ndp1, we examined Treg cells in mice that lack both Ndp1 and IL-4 (Ndp1 IL-4 double knockout or DKO mice)25.

These animals do not show the overt signs of inammation such as dermatitis25,26 observed in age-matched Ndp1/ Foxp3-Cre animals. We observed increased frequencies of Treg cells

(Supplementary Fig. 7a,b) compared to IL-4-decient controls. The Ndp1-decient Treg cells that lacked IL-4 were more proliferative than controls, as determined by Ki67 (Supplementary Fig. 7c). Similar to the Ndp1/ Foxp3-Cre mice, Treg cells in the Ndp1 IL-4 DKO mice were predominantly eTreg cells (Supplementary Fig. 7d,e). Additionally, the Ndp1-and IL-4-decient eTreg cells expressed higher levels of Ki67 (Supplementary Fig. 7f) and ICOS (Supplementary Fig. 7g)

relative to control eTreg cells. Thus, while IL-4 from current and former Treg cells likely contributes to the inammation observed in Ndp1/ Foxp3-Cre mice, increased Treg cell activation, proliferation and ICOS expression are not due solely to IL-4 signalling. Thus IL-4 is insufcient to explain the increased tness of Ndp1-decient Treg cells.

Ndp1-decient eTreg cells have altered metabolic activity. To identify molecular pathways underlying the increased proliferation and altered eTreg cell phenotype of Ndp1 decient

Treg cells, we used label-free quantitative proteomics to compare WT and Ndp1-decient cTreg and eTreg cells. We sorted YFP eTreg and cTreg cells from young WT or Ndp1/ Foxp3-Cre male mice, subjected these cells to liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis and quantied protein abundance using intensity-based absolute quantication2729 (Fig. 6a). We achieved good reproducibility of identied proteins across three experiments, pooling two to three mice of each genotype per experiment (Fig. 6b). Comparing proteins identied in control eTreg and cTreg proteomes, our data t well with what has been published previously on proteins unique to each subtype. Namely, among proteins that were signicantly more abundant in cTreg than eTreg proteomes or that were found exclusively in cTreg proteomes, dened here as having a cTreg bias, we identied CD62L and Bcl2 (ref. 20) (Fig. 6c, Supplementary Data 1). Proteins that were increased in, or biased toward, the eTreg proteome included CD44 and several proteins associated with eTreg cells, such as Integrin alpha E/itgae (CD103)30 (Fig. 6c, Supplementary Data 1).

To characterize differential protein expression in Ndp1-sufcient and -decient Treg cells, we compared the ratio of protein abundance in eTreg:cTreg for each of the three

experiments. This revealed widespread changes in the proteomes of Ndp1-decient Treg cells. Given that Ndp1 deciency drove changes predominantly in eTreg cells, we focussed our comparisons on the levels of proteins identied in WT and Ndp1-decient eTreg cells (Supplementary Data 2 and Fig. 6d).

Our ow cytometric nding of increased GITR expression was supported by these Treg proteomes (Supplementary Data 2 and

Fig. 6d). Surprisingly, proteins associated with increased mTORC1 activity were increased in Ndp1-decient eTreg cells.

This included Lamtor1 (Fig. 6d), as well as the V-ATPase subunit d1 and Lamtor3, which are components of the v-ATPase-regulator complex that drives mTORC1 activation in cells3133. We then performed network visualization of gene-ontology (GO) enrichment analysis for differentially regulated proteins in Ndp1-sufcient and -decient eTreg cells and identied clusters of nodes relating to metabolic processes that were signicantly enriched (Fig. 6e and Supplementary Data 2). Pathway analysis further supported changes in the metabolic state of Ndp1-decient eTreg cells and, more specically, an increase in mTORC1 activity. Thus loss of Ndp1 led to an altered proteomic prole indicative of altered metabolic activity.

Ndp1-decient Treg cells have increased glycolysis. Based on this proteomics proling, we sought to analyse the metabolic capacity of Ndp1-decient Treg cells. To obtain sufcient cell numbers for metabolic testing, we used in vitro IL-2-expanded Treg cells. We evaluated the bioenergetics of these expanded

WT and Ndp1/ Foxp3-Cre Treg cells at rest and also upon restimulation. We measured the extracellular acidication rate under glycolytic stress conditions (Fig. 7ad) and oxygen consumption rate under mitochondrial stress conditions (Fig. 7eh). Prior to stimulation, WT Treg cells had low basal rates

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

a b

WT cTreg overlap n =3

WT eTreg overlap n =3

105

104

103

105

104

103

102

YFP

CD62L

17.9

102 103 104 105

0 102 103 104 105

CD4

CD44

SDSPAGE Gel fractionation In-gel digest LC/MS-MS analysis

Log2 fold change (Ndfip1fl/fl Foxp3-Cre eTreg/ WT eTreg)

c d

Log10 (Pvalue)

CD62L

Log10 (Pvalue)

Bcl-2

Lamtor1 GITR

CD103

CD44

Neuropilin-1

CTLA4

V-ATPase subunit d1

4 2 0 2 4 6 8

Log2 fold change

(WT eTreg/ WT cTreg)

Regulation of metabolic process

Metabolic process

organization

Intracellular transport

Immune response

Figure 6 | Ndp1/Foxp3-Cre eTreg cells have altered metabolic whole-cell proteome. (a) Representative ow cytometry dot plots and gel pixelation to illustrate procedure used to analyse cTreg and eTreg cell proteomes using mass spectrometry. (b) Area-proportional Venn diagrams illustrating the reproducibility of proteins identied in cTreg (left) or eTreg (right) cells from three independent experiments. WT cells were compared for this analysis and data were compiled using the intensity-based absolute quantication method. (c) A volcano plot illustrating differentially expressed proteins between eTreg

versus cTreg cells in WT mice. (d) A volcano plot representing differentially expressed proteins between eTreg cells from Ndp1/ Foxp3-Cre or WTanimals. (e) Network diagram of enriched GO terms with nodes representing GO annotations and edges connecting similar terms based on the GO hierarchy. The size of the nodes corresponds to the number of genes associated with the respective GO term and the colour of the nodes corresponds to the level of signicance of the enrichment of the respective term in the data set (darker colour corresponds to a higher level of signicance). Clusters of GO terms were manually analysed and annotated to identify broad functional similarity. Data are shown for YFP eTreg cells that were sorted from 9- to 12-week-old WTor

Ndp1/ Foxp3-Cre male mice.

of glycolysis, but rates increased when cells were stimulated (Fig. 7a,b). In Ndp1-decient Treg cells, the resting glycolytic rate was only modestly increased over WT cells (Fig. 7a,c). However, restimulated Ndp1-decient Treg cells had a considerably increased glycolytic rate and glycolytic capacity compared to restimulated WT Treg cells (Fig. 7bd). To assess mitochondrial function, we measured cell respiratory control, which is a general test of mitochondrial function in cells. Loss of Ndp1 did not change the maximum respiratory capacity of Treg

cells and caused a slight decrease in the spare respiratory capacity of Treg cells (Fig. 7eh). Taken together, these data indicate that the loss of Ndp1 drives a metabolic switch in activated Treg cells,

promoting a more effector cell-like reliance on glycolysis to supply cellular energetic demands.

Ndp1-decient Treg cells have elevated mTORC1 activity. Effector T cells utilize glycolysis while Treg cells are more dependent on oxidative phosphorylation as their main source of energy3438. High glycolytic activity in Treg cells, as can occur when mTORC1 activity is increased, has been associated with Treg

cell dysfunction6,7,39. To investigate whether mTORC1 activity was increased in Ndp1-decient Treg cells, we cultured WT and

Ndp1/ Foxp3-Cre Treg cells in vitro. We found that Ndp1/ Foxp3-Cre Treg cells quickly outnumbered their WT counterparts (Fig. 8a), were increased in cell size (Fig. 8b) and proliferated more (Fig. 8c). This, together with observed increases in ICOS and GITR expression (Fig. 8d,e), suggested that this in vitro culture system recapitulates the phenotype of Ndp1-decient Treg

cells in vivo. Given that ICOS can be driven by mTORC1 (ref. 39)

10 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

a b

Unstimulated cells CD3/CD28-stimulated cells

150

100

Ndfip1fl/fl

Foxp3-Cre

fl/fl

-Cre

ECAR (mpH min1)

100

ECAR (mpH min1)

Time (mins):

Treatment:

0 1 8 16 23 29 36 42 49 55 62 68 75 81

Basal Glucose Oligomycin 2DG

Time (mins):

Treatment:

0 1 8 16 23 29 36 42 49 55 62 68 75 81

Basal Glucose Oligomycin 2DG

c d

Glycolysis

Glycolytic capacity

100

ECAR (mpH min1)

Ndfip1fl/fl

Foxp3-Cre

CD3/CD28:

+ +

Ndfip1fl/fl

Foxp3-Cre

CD3/CD28:

Basal Oligomycin FCCP Rot./Ant A Treatment: Basal Oligomycin FCCP Rot./Ant A

0 + +

e f

Unstimulated cells

CD3/CD28 -stimulated cells

200

0 0

200

fl/fl

-Cre

fl/fl

-Cre

OCR (pmoles min1)

150

100

OCR (pmoles min1)

150

100

1 8 16 23 29 36 42 49 55 62 68 75 81 1 8 16 23 29 36 42 49 55 62 68 75 81

Time (mins):

Treatment:

g h

Maximum respiratory capacity

Spare respiratory capacity

150

Ndfip1fl/fl

Foxp3-Cre

OCR (pmoles min1)

Ndfip1fl/fl

Foxp3-Cre

100

CD3/CD28:

+ +

CD3/CD28:

+ +

Figure 7 | Ndp1-decient Treg cells have a signicantly increased rate of glycolysis. (ah) YFP Treg cells were sorted from Ndp1/ Foxp3-Cre WT or Ndp1/ Foxp3-Cre mice, expanded in culture and then were left unstimulated or were restimulated before metabolic function was assessed.

(a,b) Extracellular acidication rate (ECAR) was measured during a glycolysis stress test in a unstimulated or (b) stimulated Treg cells treated with drugs as indicated. (c) Rate of glycolysis is the difference in ECAR between post-glucose addition and baseline. (d) Glycolytic capacity is the difference between post-Oligomycin ECAR and baseline ECAR. (e,f) Oxygen consumption rate (OCR) changes during a mitochondrial function assessment test in e unstimulated or (f) stimulated WT or Ndp1/ Foxp3-Cre Treg cells treated with drugs as indicated. (g) Maximum respiratory capacity is the difference in

OCR after addition of rotenone/Antimycin A versus after addition of uoro-carbonyl cyanide phenylhydrazone (FCCP). (h) Spare respiratory capacity is the difference in increased OCR following addition of FCCP compared to baseline. Graphs show means.e.m. (ad) Represents n 4 mice (male or female,

712 weeks) per genotype in two independent experiments. Final Foxp3% after in vitro expansion was 74.45%8.03 for Ndp1/ Foxp3-Cre versus74.89.80 for Ndp1 / Foxp3-Cre. (eh) Represents n 68 mice (male and female, 712 weeks) in three experiments. Final Foxp3% was 88.12.08

for Ndp1/ Foxp3-Cre versus 86.431.94 for Ndp1/ Foxp3-Cre. P values were calculated by one-way ANOVA. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001.

activity, we examined other proteins in the mTORC1 pathway, including the amino acid transporter, CD98, and phosphorylated S6 (pS6). CD98 and pS6 (Fig. 8f,g) were both

increased in Ndp1/ Foxp3-Cre Treg cells. Coculture conrmed that these changes are not directly linked to IL-4 signalling (Fig. 8h).

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

a b c

FSCA Ki67

1.5 ****

Normalized cell count

FSCA normalized MFI

Ki67 normalized MFI

****

1.0

0.5

0 2 3 4 5 6 7

0.0

WT Ndfip1fl/fl Foxp3-Cre

Days in culture

ICOS

e f

GITR

CD98

4 *

2.0

ICOS normalized MFI

GITR normalized MFI

CD98 normalized MFI

1.5

1.0

0.5

0.0

WT Ndfip1fl/fl Foxp3-Cre

g h

Coculture

pS6

100

WT Ndfip1fl/fl Foxp3-Cre

% of max

Fold change relative to

cocultured ctrl cells

***

0 0 102 103 104 105

Ki67

MFI

ICOS MFI

GITR MFI

CD98 MFI

pS6 MFI

p-S6

WT Ndfip1fl/fl Foxp3-Cre

Figure 8 | Loss of Ndp1 leads to elevated mTORC1 signalling and metabolic tness in Treg cells. (ah) Treg cells were sorted from congenically distinct CD45.1 WT or CD45.2 Ndp1/ Foxp3-Cre mice. Treg cells were expanded in vitro (ag) individually or (h) in mixed cocultures. (a) Cell numbers in the cultures were analysed daily for 7 days. (bf) After 7 days, cells were analysed by ow cytometry for cell size (forward scatter (FSC)), (b) Ki67 expression, (c) or their surface levels of: (d) ICOS, (e) CD98, and (f) GITR. (g) Cells were also restimulated at the end of the culture period and analysed for expression of pS6. (h) Cocultured WT and Ndp1/ Foxp3-Cre Treg cells were similarly examined for the expression of Ki67, ICOS, GITR, CD98 and pS6. Eachdot represents an individual mouse. All experiments were performed on at least two independent occasions using age-matched male or female, aged 712 weeks, mice of each genotype. P values were calculated by multiple t-tests. *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001. Error bars indicate means.e.m.

Increased responsiveness to IL-2 could explain the tness of Ndp1-decient Treg cells in this IL-2 expansion system. To test this, we sorted Ndp1-sufcient and -decient Treg cells and rested them overnight without cytokine or TCR stimulation. Upon brief IL-2 stimulation, we detected STAT5 phosphorylation (pSTAT5) by ow cytometry. pSTAT5 levels (mean uorescence intensity) were similar between WT and Ndp1-decient Treg

cells before and after IL-2 stimulation (Supplementary Fig. 8). This is consistent with our recent report showing that both Ndp1 and its homologue, Ndp2, are required in Tconv cells to

drive the degradation of Jak1 and decrease STAT5 phosphorylation in response to IL-2 (ref. 40). Therefore, the increased metabolic activity of Ndp1/Foxp3-Cre Treg cells is cell intrinsic and not just because of enhanced responsiveness to IL-2.

DiscussionTherapies using adoptively transferred Treg cells or that target Treg cell function are now being developed, making it essential to understand mechanisms that maintain Treg cell identity and how these pathways integrate with TCR activation and cellular

metabolic processes. Our data identify Ndp1, an activator of E3 ubiquitin ligases, as a new molecular target impacting Treg

stability and function. We demonstrate that Ndp1-decient Treg

cells are highly proliferative and concomitantly gain the capacity to produce the proinammatory cytokine IL-4. Importantly, Ndp1 deciency can lead to the expansion of Treg cells and loss of Foxp3 in vivo.

Both mTORC1 and mTORC2 play important functions in preventing Treg cells from adopting a glycolytic metabolic prole. mTORC1-mediated glycolytic activity in Treg cells is modulated by the serinethreonine phosphatase, PP2A, and by the autophagy proteins, Atg5 and Atg7 (refs 7,8). The lipid phosphatase, PTEN, inhibits mTORC2-mediated metabolic pathways in Treg cells to prevent Treg cell instability and maintain suppression of Th1 and Tfh effector T-cell functions41,42. Ndp1-decient Treg cells have increased protein levels of

ICOS, CD98 and pS6, all indicators of high mTORC1 signalling. mTORC1 deletion in Treg cells leads to defects in efciently synthesizing lipids from glucose and to defects in homeostatic proliferation in vivo39. Therefore, it is not surprising that Ndp1-decient Treg cells, with high mTORC1 signature,

12 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

show greater homeostatic expansion in vivo, show greater metabolic tness in vitro and in vivo and are more efcient at glycolysis. This altered glycolytic metabolism fuels dysfunction in vivo by driving the expansion of eTreg cells that are prone to produce IL-4 and lose Foxp3.

We propose that Treg cells lacking Ndp1 progress through a series of changes that is precipitated by increased proliferation of activated (CD44 ) Treg cells and accompanied by the acquisition of IL-4 production. These changes are likely initiated in eTreg

cells since Ndp1 is inducibly expressed upon T-cell activation. The elevated glycolytic metabolism in these cells drives further expansion of IL-4-producing current and former Treg cells.

Autocrine IL-4 signalling can then lead to loss of Foxp3. Supporting this, CD44 Ndp1-decient Treg cells have increased methylation at the CNS2 region of the Foxp3 locus, indicating susceptibility towards instability. Ndp1-decient TH2

cells are capable of driving tissue damage and inammation as already described by our laboratory and others12,14,40,43.

Loss of IL-4 did not affect the activation, proliferation and in vivo expansion of Ndp1-decient Treg cells leading us to conclude that the metabolic tness in Ndp1-decient Treg cells

precedes or is independent of the gain in Treg cell-intrinsic IL-4 production. Furthermore, IL-4-overexpressing transgenic mice have been described and have increased recruitment of different immune cell subsets into the skin but do not develop spontaneous dermatitis44 as observed in Ndp1/ Foxp3-Cre mice. Last, loss of the mTORC1 subunit, Raptor, in Treg cells does not lead to a cell-intrinsic increase in Treg cell IL-4 or other effector cytokines39, suggesting that cytokine regulation and metabolic regulation by Raptor in Treg cells may be two independent events39. An attractive hypothesis that warrants further future investigation is whether two parallel (but perhaps cross-talking) cellular pathways are regulated by Ndp1: one which dampens mTORC1 activity and glycolysis in order to limit Treg cell

activation and proliferation, and a second that limits Treg cell IL-4

production. These pathways, though independent, may cooperate to account for the severity of the inammation observed in Ndp1/ Foxp3-Cre mice.

Ndp1 is a known activator of Itch40,45,46. Mice bearing a Treg-specic deletion of Itch show some overlapping features with the Ndp1/ Foxp3-Cre mice described here15: Itch-decient Treg cells have an effector Treg cell phenotype, an increase in proliferation and intact suppression in traditional in vivo and in vitro suppression assays. Itch-decient Treg cells are unstable as dened by their increased production of IL-4 but these Treg cells

do not have defects in their ability to maintain Foxp3 expression in a lymphoreplete animal15. Future work will be needed to determine whether mTORC1 signalling is elevated in Itch-decient Treg cells and whether this contributes to their dysfunction. Future studies on Itch and Ndp1 will be of particular interest as therapies that activate or inhibit these pathways are being developed. If such therapeutic strategies can be used to regulate Treg cell functions, their use in the clinic could have broad potential.

Methods

Mouse strains. Ndp1/ mice47 were crossed to Foxp3-YFP-Cre reporter mice (Jackson Laboratory, stock# 01695917), also referred to as Foxp3-Cre, and bred for only one copy of the Cre gene to generate Ndp1/ Foxp3-Cre male mice and Ndp1/ Foxp3-Cre / female mice. WT control mice were: Ndp1 /

Foxp3-Cre male, and Ndp1 / Foxp3-Cre / female mice. We generated congenic CD45.1/CD45.2 control mice by crossing Ndp1 / Foxp3-Cre mice to CD45.1 mice (Jackson Laboratory, Stock# 002014). CD45.1 and Rag1 / are maintained in house. Ndp1 IL-4 DKO mice25,26 are maintained in house. The gender and ages of mice used are indicated in each gure legend. Mice were maintained in a barrier facility at the Childrens Hospital of Philadelphia. All procedures were approved by the Institutional Care and Use Committee of the Childrens Hospital of Philadelphia. Genotyping primer sequences used were:

Ndp1 ox alleles (Ndp1 oxed forward 50-TGAGGAAACAGACACACAATG-30, Ndp1 oxed reverse 50-TGGAATGAACCTGAGGTCTCC-30)48 and Foxp3-Cre (Jackson Laboratory, stock# 016959, WT Forward: 50-CCTAGCCCCTAGTTCCA ACC-30, WT Reverse: 50-AAGGTTCCAGTGCTGTTGCT-30, Mutant Forward: 50-AGGATGTGAGGGACTACCTCCTGTA-30, Mutant Reverse: 50-TCCTTCACT CTGATTCTGGCAATTT-30)17. Sequences for Il4 genotyping (Jackson Laboratory, stock# 002253) are: Il4 common 50-GTGAGCAGATGACATGGGGC-30; Il4 WT 50-CTTCAAGCATGGAGTTTTCCC-30 and Il4 mutant 50-GCCGATTGTCTGTT GTGCCCAG-30.

Flow cytometry. The following ow cytometric antibodies were purchased from Biolegend: CD4 (GK1.5), CD8a (53-6.7), CD44 (IM7), IL-17A (TC11-18H10.1), CD3e (17A2), CD45.2 (104), CD45.1 (A20) GITR (YGITR765), ICOS (C398.4A), and PD-1 (RMP1-30), all used at 1:200 dilution. aYFP/GFP (A-21311) was purchased from Life Technologies and used at 1:400 dilution. Antibodies against IFNg(XMG1.2) used at 1:300 and Ki67 (B56) used at 1:200 dilution were purchased from BD Biosciences. Biotinylated aphospho-S6 was purchased from Cell Signalling Technologies (D57.2.2E) and used at 1:150 dilution. The remaining antibodies for ow cytometry were purchased from eBioscience: CD25 (PC61.5), CD62L (MEL-14), and IL-10 (JES5-16E3) all used at 1:200 dilution; IL-4 (11B11) used at 1:100 dilution; and Foxp3-biotin (FJK-16s). Biotinylated aFoxp3 and apS6 were used at 1:150 dilution and detected with uorophore-conjugated streptavidin at 1:750 dilution (S32354 Invitrogen).

Tissue processing and ow cytometry. Serum was obtained by cardiac puncture immediately following CO2 killing. Spleens and lymph nodes were harvested and mashed through 70 mM nylon lters using cold Hanks Balanced Salt Solution (HBSS). Spleens were red blood cell (RBC) lysed using ammonium-chloride-potassium lysis buffer, washed, ltered again via a 70 mM lter and resuspended in cold HBSS for sorting or resuspended in complete DMEM (cDMEM) for either phorbol myristate acetate (PMA)/Ionomycin stimulation or direct ow cytometry. Lungs were ushed with cold phosphate-buffered saline (PBS) immediately after killing, minced in cDMEM medium containing 0.9 mg ml 1 of collagenase A (Sigma), 0.8 mg ml 1 of collagenase 1A (Sigma) and 20 mg ml 1 of DNase I (Sigma) and rotated for 1 h at room temperature. FCS was added at 15% v/v and the mixture was ltered through a 100 mM and then 40 mM lter before centrifugation. Lung pellets were RBC lysed using in-house-made ammonium-chloride-potassium lysis buffer, washed, resuspended in complete DMEM media and ltered through a 70 mm lter. The cells were then stimulated for 4 h in 5 ml round bottom polystyrene test tubes with PMA (30 ng ml 1, Calbiochem), ionomycin (1 mM, Abcam) and Brefeldin A (1 mg ml 1, Sigma) for intracellular cytokine staining or directly stained for ow cytometry. For ow cytometry, cells were washed in serum-free HBSS or PBS, stained with live/dead xable blue dead cell stain (L-23105; Invitrogen, 1:60 dilution), Fc blocked (aCD16/32, 2.4G2; BD

Biosciences, 1:25 dilution) and stained with the appropriate antibody mixtures. After staining 25 min at 4 C, samples were washed in FACs buffer (1xPBS, 2.5% v/ v FCS and 0.1%w/v Sodium azide) and then xed overnight using the Foxp3 Fix/ Perm Kit (00-5523-00, eBioscience). After washing with the provided Foxp3 kit perm buffer, intracellular staining was carried out for 1 h at 4 C. Cells were washed again with perm buffer and resuspended for ow cytometry in FACs buffer. In Fig. 8, expanded Treg cells, cultured as described below, were restimulated with mouse T-cell activator beads (Gibco, 11456D) at a 1:1 cell:bead ratio in complete DMEM medium for 4 h. Samples were immediately xed at a nal concentration of 2% PFA, permeabilized in 90% methanol and then stained for pS6 and surface markers together in Fc block. The same PFA xation protocol was used for the pSTAT5 experiment. Samples were acquired on an LSRFortessa (BD Biosciences) and analysed using FlowJo version 9.8 (Flowjo LLC). Events analysed were singlets, (FSC-A FSC-H, SSC-W SSC-H), live (viability dye negative) and within the

lymphocyte gate (FSC-A SSC-A) (Supplementary Fig. 9).

Enzyme-linked immunosorbent assay (ELISA). Serum was isolated from 8- to 16-week-old male Ndp1-sufcient and -decient Foxp3-Cre mice.

The concentrations of Ig subclasses in mouse sera were determined using isotypespecic antibodies with a sandwich ELISA protocol. Monoclonal anti-mouse IgA-HRP, IgM-HRP, IgG1-HRP, IgG2C-HRP and IgE-HRP were purchased from SouthernBiotech. These were used in twofold serial dilutions starting at10 ng ml 1. Flat-bottom, 96-well plate (Nunc) was coated with capture antibody overnight at 1:250 dilution. Plates were blocked with 10% FCS in PBS buffer for 2 h and incubated with sample serum (1:20,000 for IgA, IgM and IgE and 1:40,000 for IgG1 and IgG2C) for 2 h at room temperature and detected with HRP-conjugated Ig subclass antibody (SouthernBiotech, 1:10,000 dilution) for 1 h at room temperature. Plates were developed with TMB substrate solution (eBioscience) and read at 450 nM using a Synergy HT Microplate Reader (BioTek).

T-cell isolation. YFP CD4 Treg cells were isolated from the spleen and lymph node, enriched for CD4 T cells by negative selection using 5 ml per mouse of rat amouse CD8a (2.43; Biolegend,) and 5 ml per mouse of rat amouse I-A/I-E (M5/ 114.15.2; Biolegend) in 57.5 ml of complete DMEM medium and incubated for 30 min with end-over-end rotation at 4 C. Cells were then washed and

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

incubated with 1.5 ml of Biomag goat arat IgG magnetic beads (Qiagen) for 10 min with end-over-end rotation at room temperature; unbound cells were collected using a Dynal magnet (Invitrogen). These enriched and untouched CD4 T cells were then stained with antibodies against CD4, CD44, CD25 and CD62L all at 1:200 dilution. Cells were ltered through 35 mM lter cap polystyrene FACS tubes (BD Biosciences) and sorted under high speed on a MoFlow Astrios (Beckman Coulter) or a FACS Aria (BD Biosciences). Total Treg cells were sorted as CD4

YFP , cTreg cells as CD4 YFP CD62LhighCD44 and eTreg cells as CD4 YFP CD62LlowCD44 . Naive cells were identied as CD4 CD62LhighCD44 CD25 .

Bisulte sequencing. Sodium bisulte conversion, followed by PCR amplication and sequencing of individual clones, was used to determine the extent of CpG DNA methylation at the CNS2 and promoter regions of the Foxp3 gene. First, CD4 YFP CD62LlowCD44 eTreg cells were sorted from YFP WT and

Ndp1/ Foxp3-Cre mice. Naive CD4 T cells were sorted from WT mice. Next, DNA was extracted using the Qiagen DNeasy Blood and Tissue Kit.

Approximately 1 mg of DNA puried from CD4 Treg cells was bisulte converted via the following procedure49: the DNA was precipitated in ethanol, washed with 70% ethanol, and resuspended in 20 ml of Tris-EDTA buffer. This DNA was then denatured at 37 C for 15 min using 0.3 M NaOH and then incubated at 55 C for 16 h in a mixture containing 5.36 M Urea, 1.72 M sodium metabisulte and 0.5 mM hydroquinone. The DNA was then desalted using the Wizard DNA clean-up system (Promega, A7170), desulfonated with 0.3 M NaOH, neutralized with 3 M ammonium acetate and precipitated with 100% ethanol. The CNS2 region of the Foxp3 gene was PCR amplied using the following primers: Foxp3 Intron Forward1 50-GGGTTTTGGGATATTAATATATATAGTAAG-30, Foxp3 Intron

Reverse1 50-CCACTATATTAACTTAACCCATATAACTAA-30, Foxp3 Intron Forward2 50-TTGAGTTTTTGTTATTATAGTATTTGAAGAT-30, and Foxp3 Intron Reverse2 50-ACTAAAAACCTAAAAAACTAAACTAACCAA-30. The promoter region was amplied with the following primers: Foxp3 Promoter Forward1 50-GTTTGGAGTAGAAGGAAGTTTTTGGAGAT-30, Foxp3 Promoter

Reverse1 50-TATCTAAAAACCAACTACTCCACCTATCTA-30, Foxp3 Promoter Forward2 50-GGTTGTTTTTTATTTATATGGTAGGT-30, and Foxp3 Promoter Reverse2 50-CCAAAATCCTTACCTAAAATAACTA-30. Following nested PCR, bands of appropriate sizes were gel puried and cloned into pGEM-T Easy vector (Promega, A1360). Single colonies were miniprepped with the QIAprep Spin Miniprep DNA Purication Kit (Qiagen, 27104) and sequenced with SP6 primer. Only sequences that were derived from fully converted alleles were used for methylation analysis.

In vitro cell culture. For the in vitro Treg cell suppression assay, sorted naive Tconv CD45.1 CD4 CD62LhighCD44 CD25 cells were stained with 2.5 mM carboxyuorescein succinimidyl ester (CFSE) by incubating 400 ml of cells (10-20million cells per ml) in PBS with 400 ml of 5 mM CFSE in PBS solution for 3 min with continuous shaking. Next, 800 ml of FCS was added to the reaction mixture for 30 s at room temperature, followed by adding 10 ml of complete DMEM medium and spinning down in a centrifuge at 1300 r.p.m. for 5 min at 46 C. The CFSE-labelled naive Tconv cells were then resuspended to achieve1.5 104 cells per 50 ml, and 50 ml of these cells were added to the appropriate wells

of a 96-well plate. Sorted YFP CD4 CD45.2 Treg cells were resuspended at1.5 104 cells per 50 ml. A total of 50 ml of twofold serial dilutions of these Treg cells

were added to the Tconv cells in each well of the 96-well plate to achieve a nal Treg:Tconv ratio per well ranging from 0:1 to 1:32 (ref. 50). Next, irradiated bulk CD45.2 splenocytes (2,500 rads, X-ray irradiator) were resuspended at 1.5 105

cells per 50 and 50 ml was added. Last, 50 ml of a 4 solution containing aCD3

(145-2C11; Biolegend) and rhIL-2 were added to achieve a nal concentration of 1 mg ml 1 soluble aCD3 and 50 U ml 1 IL-2 per well. The cells were cultured for 4 days for the suppression assay. For expanded Treg cultures, sorted YFP CD4 total Treg cells or CD4 YFP CD62LhighCD44 cTreg cells were resuspended at 1 106 cells per ml in media containing 2,000 U ml 1 IL-2 and stimulated at a 1:3

(cell:bead) ratio with mouse aCD3/aCD28 T-cell activator Dynabeads (Gibco, 11456D). After 48 h, cultures were counted and split daily with IL-2 media. Cells were assayed by ow cytometry after 6 or 7 days in cultures. Due to the 48-h activation, results from experiments where we started with sorted YFP cTreg cells were virtually identical to those in which we started with sorted YFP total Treg cells due to in vitro conversion of c Treg to e Treg cells. We therefore present all

of that data pooled together. For analysis of RNA, expanded Treg cells were restimulated on day 7 of culture at a 1:1 ratio with aCD3/CD28 beads for the indicated periods. Supernatant was saved for ELISA and cells were harvested for quantitative PCR (qPCR). For cocultures, sorted WT and Ndp-1 decient Treg

cells were mixed in a 1:1 ratio before being plated. For the in vitro stability assay using IL-2, IL-4 and aIL-2, WT or Ndp1-decient YFP Treg cells were sorted from matched male or female donors. The Treg cells were cultured for 96 h on plate-bound 5 mg ml 1 aCD3/aCD28 and either 500 U ml 1 IL-2 or 20 ng ml 1

IL-4 with 20 ng ml 1 a-IL-2. All cultured cells, unless otherwise noted, were cultured at 10% CO2 in complete T-cell medium, that is, DMEM (Mediatech) supplemented with 10% FCS (Atlanta Biologicals, premium FCS), 1% pen/strep (Invitrogen), 1% Glutamax (Invitrogen), 1% Minimum Essential Mediumnon-essential amino acids (Invitrogen), 20 mM HEPES (Invitrogen), 1 mM Sodium Pyruvate and 0.12 mM betamercaptoethanol (Sigma) supplemented with

recombinant human 50 U ml 1 IL-2 (originally obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of

Allergy and Infectious Diseases, National Institutes of Health).

Metabolic function assays. YFP CD44 CD62L Treg cells (cTreg cells ) or YFP CD4 Treg cells (total Treg cells ) were sorted from 7- to 12-week-old, age-matched male or female, congenically marked CD45.1 WT or CD45.2

Ndp1/ Foxp3-Cre (cKO) mice. After sorting, the WT or cKO Treg cells were expanded at a 1:3 (cell:bead) ratio with mouse aCD3/aCD28 beads for 7 days. On day 7, cells were harvested and resuspended at 5 106 cells per ml in appropriate

Seahorse media. In all, 40 ml of cells were added to 140 ml of appropriate seahorse media or 140 ml of Seahorse media containing fresh mouse aCD3/aCD28 beads at a 1:1 (cell:bead) ratio. Cells were plated in XF96 cell culture microplates that were precoated with 25 ml of Cell tak reagent overnight according to the manufacturers instructions (Corning, 354240).

For examining glycolysis, the glyco-stress test kit (Seahorse Bioscience) was used with additions of the following reagents to cells in glucose-free Seahorse media: 10 mM glucose, 2 mM oligomycin, and 50 mM 2-DG. To determine the rate of glycolysis, Treg cells were incubated in glucose-free medium and the increased extracellular acidication rate response upon stimulation of glycolysis by glucose and oligomycin was determined. Oligomycin inhibits mitochondrial ATP synthase (complex V) forcing glycolysis to compensate for the lack of ATP production in oxidative phosphorylation. Finally, 2-dexoglucose was added to inhibit glycolysis. For examining mitochondrial function, a mito-stress kit was used with additions of the following reagents to cells in appropriate Seahorse media: 2 mM Oligomycin,1.5 mM uoro-carbonyl cyanide phenylhydrazone (an uncoupler of respiration and oxidative ATP synthesis), and premixed 100 nM rotenone 1 mM antimycin A

solution (Electron Transport Chain complex I and III inhibitors). Plates were run on an XF96 Seahorse assay instrument. Settings used were: calibration, equilibration, and baseline readings (loop three times). Before the baseline reading, and after each injection from a port, the following procedure was followed: mix, wait 3 min, and measure for 2 min with 3 min end loop51. Each reagent was added according to the volumes recommended by the Seahorse assay manufacturer.

Quantitative PCR. Samples for qPCR were lysed in Trizol (Ambion). RNA was extracted using chloroform and qPCR was performed as follows: 10 ng of cDNA was added to TaqMan Gene Expression Master Mix and TaqMan Gene Expression primer/probe mix specic for Ndp1 (Applied Biosystems), according to the manufacturers protocol for a nal reaction volume of 20 ml. qPCR was performed using an Applied Biosystems 7500 Real-Time PCR system. Each sample was assayed in triplicate along with the endogenous control (actin). Actb primer/probe (4352933E) was obtained from Applied Biosystems. Ndp1 custom primers: Ndp1_F 50-GCTCCTCCACCATACAGCAGC-30; Ndp1_R 50-CGATGGGGGCT

TTGGAAATCCAG-30, and Ndp1 Taqman MGB probe: 50-TTTGGAAATCCAG ATTCATCTTTG-30 were obtained from Applied Biosystems40. Relative mRNA expression of each gene of interest was calculated as 2dCt, where dCt represents threshold cycle (Ct) of Actin beta minus Ct of gene of interest.

Bone marrow chimeras. Bone marrow from male Ndp1/ Foxp3-Cre (CD45.2 ) or WT mice: Ndp1 / Foxp3-Cre (CD45.1/CD45.2) or CD45.1 mice was obtained by ushing femur and tibia with cold HBSS, RBC lysing and passing through a 70 mM lter. Cells were processed into a single-cell suspension,

T cell depleted, resuspended in freezing media (90% FCS, 10% DMSO) and kept at

80 C until used. Thawed cells were washed, counted, resuspended in sterile PBS, mixed 1:1 CD45.1:CD45.2 and injected intravenously into sublethally irradiated (400 rads, X-Ray irradiator) 6-week-old male Rag1 / recipients at 1 106 cells

per mouse. Chimeras were kept on 5.3 ml of Trimethoprim/Sulfamethoxazole (200 mg sulfamethoxazole 40 mg trimethoprim per 5 ml) per 400 ml drinking

water for the rst 2 weeks after bone marrow transfer. Chimeras were analysed 8 weeks after transfer.

Histology. Skin (ear) and oesophagus were dissected and xed in 10% neutral buffered formalin for at least 24 h. Lung were obtained after manual transcardial perfusion with 10 ml syringes containing PBS and xed in 10% neutral buffered formalin. All organs were then embedded in parafn, sectioned to 5 mm thickness and stained with Haematoxylin and Eosin. Images were obtained using a Leica DM4000B upright scope paired with a Spot RT/SE Slider camera (Childrens Hospital of Philadelphia Pathology Core).

T-cell transfer colitis. Naive CD4 T cells from WT mice (CD45.1 or CD45.1/CD45.2) and CD4 YFP Treg cells from CD45.2 Ndp1/ Foxp3-Cre and CD45.2 Ndp1 / Foxp3-Cre male mice were isolated from the spleens and lymph nodes. Cells were resuspended in sterile PBS and congenic naive T cells were mixed at a 5:1 ratio with YFP Treg cells. A total of 0.36 106 cells were

injected intraperitoneally into 68-week-old Rag1 / mice. Mice were weighed weekly. Recipients of both genotypes were cohoused. At killing, mice were weighed, spleens were weighed and the spleens and lung were processed for ow cytometry as described above. A fraction of the total lung/spleen single-cell suspensions were

14 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677 ARTICLE

stained with antibodies against CD4, CD45.1 and CD45.2 and sorted using a Moow Astrios as described above for analysis of Ndp1 mRNA in cells from Ndp1/ Foxp3-Cre and Ndp1 / Foxp3-Cre donor Treg cells. Sorted cells were CD4 CD45.2 CD45.1 YFP (current Treg cells ) or CD4 CD45.2

CD45.1 YFP (former Treg cells) or CD4 CD45.2 CD45.1 YFP- (Tconv cells). Cells were stimulated for 5 h with plate-bound 5 mg ml 1 aCD3/CD28 at 3 105

cells per ml and harvested in Trizol for qPCR. The remaining fractions of total lung/spleen single-cell suspensions were either stained directly for ow cytometry or stimulated for 4 h with PMA/ionomycin in the presence of Brefeldin A to analyse cytokine production by ow cytometry.

Preparation of lysates for whole-cell proteome analysis. Ndp1-sufcient and -decient cTreg cells and eTreg cells were isolated and lysed using a lysis buffer containing 100 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 5 mM EDTA, 1% NP-40, 0.5% Triton-X 100, protease inhibitor cocktail (Roche, 11697498001), 5 mM of 1,10-phenyanthroline (o-PA), 5 mM N-ethymaleimide and 0.1 mM PR-619. In all, 10 ml of lysis buffer was used for every 10 106 cells. Protein was quantied by

BCA (Pierce, 23227). Approximately 15 mg of lysate was mixed 1:1 with 4

Laemmli sample buffer for whole proteome analysis. Samples were boiled and run B2 cm past the stacking gel in 10% Criterion precast Tris-HCL gels (Biorad). Gels were xed overnight and stained briey with Coomassie blue. Each lane of the Coomassie-stained gel was divided into 810 2 9 mm2 pixels, each cut into

1 mm3 cubes52. They were destained with 50% Methanol/1.25% Acetic Acid, reduced with 5 mM dithiothreitol (Thermo) and alkylated with 40 mM iodoacetamide (Sigma). Gel pieces were then washed with 20 mM ammonium bicarbonate (Sigma) and dehydrated with acetonitrile (Fisher). Trypsin (5 ng ml 1 in 20 mM ammonium bicarbonate, Promega) was added to the gel pieces and proteolysis was allowed to proceed overnight at 37 C. Peptides were extracted with0.3% triouroacetic acid (J.T. Baker), followed by 50% acetonitrile. Extracts were combined and the volume was reduced by vacuum centrifugation.

Mass spectrometric analysis. Tryptic digests were analysed by LC-MS/MS on a hybrid LTQ Orbitrap Elite mass spectrometer coupled with a Dionex Ultimate 3000 (Thermosher Scientic San Jose, CA). Peptides were separated by reverse phase (RP) HPLC on a nanocapillary column, 75 mm id 30 cm Reprosil-pur

1.9 mM (Dr Maisch, Germany). Mobile phase A consisted of 0.1% formic acid (Thermo) and mobile phase B of 0.1% formic in acetonitrile. Peptides were eluted into the mass spectrometer at 300 nl min 1 with each RP-LC run comprising a 90 min gradient from 3% to 45% B in 90 min. The mass spectrometer was set to repetitively scan m/z from 300 to 1,800 (R 240,000 for LTQ-Orbitrap Elite)

followed by data-dependent MS/MS scans on the 20 most abundant ions, with a minimum signal of 1,500, dynamic exclusion with a repeat count of 1, repeat duration of 30 s, exclusion size of 5,000 and duration of 60 s, isolation width of 2.0, normalized collision energy of 33 and waveform injection and dynamic exclusion enabled. Fourier transform MS full-scan AGC target value was 1e6, while MSn AGC was 1e4. Fourier transform MS full-scan maximum ll time was 10 ms, while ion trap MSn ll time was 100 ms; microscans were set at one. FT preview mode; charge state screening and monoisotopic precursor selection were all enabled with rejection of unassigned and 1 charge states.

Proteomic data analysis. Proteomic data were analysed using Maxquant version1.5.0.30 searching against the Uniprot complete mouse reference proteome, including isoforms, (updated 19 September 2013) and common laboratory contaminants. A minimum peptide length of six amino acids and a peptide and protein false discovery of 1% was required. The three biological replicates for Ndp1-sufcient and -decient eTreg cells and cTreg cells were analysed together, with match between runs and requantify turned on. Label-free quantication via intensity-based absolute quantication was calculated in MaxQuant. Quantication data were normalized by the mean quantication value of identied proteins in each replicate and log2 transformed. Comparisons of protein hits from WT or Ndp1-decient eTreg or cTreg cells were calculated by log2 fold changes and evaluated for signicance using a one-sample t-test. Area-proportional Venn diagrams were generated with eulerAPE version 3.0 (http://www.eulerdiagrams.org/eulerAPE/

Web End =http://www.eulerdiagrams.org/ http://www.eulerdiagrams.org/eulerAPE/

Web End =eulerAPE/ )53. The GO analysis and corresponding network diagram in Fig. 6e was generated with the BiNGO application54 in Cytoscape55. The differentially regulated proteins identied in WT vs cKO eTreg comparisons were compared against the Mus musculus GO Biological Process annotation using a hypergeometric signicance test with BenjaminiHochberg false-discoveryrate correction. A threshold of adjusted P value of o0.01 was required for enrichment.

Statistical analysis. Data were graphed and analysed for statistical signicance in Prism version 6 or version 7 (Graphpad Software, Inc) or Excel (Microsoft). The heatmap in Fig. 5 was generated in prism 7. The following statistical tests were used as appropriate and as noted in the gure legends: t-test, one-way analysis of variance (ANOVA), two-way ANOVA, and repeated-measures ANOVA. All data are shown as averages.e.m., with a cutoff of Po0.05 for statistical signicance: *Po0.05, **Po0.01, ***Po0.001, ****Po0.0001.

Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information les). The mass spectrometric proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE56 partner repository with the data set identier PXD006251.

References

1. Sakaguchi, S., Sakaguchi, N., Asano, M., Itoh, M. & Toda, M. Immunologic self-tolerance maintained by activated T cells expressing IL-2 receptor a-chains (CD25). Breakdown of a single mechanism of self-tolerance causes various autoimmune diseases. J. Immunol. 155, 11511164 (1995).

2. Brunkow, M. E. et al. Disruption of a new forkhead/winged-helix protein, scurn, results in the fatal lymphoproliferative disorder of the scurfy mouse. Nat. Genet. 27, 6873 (2001).

3. Overacre, A. E. & Vignali, D. A. A. T reg stability: to be or not to be. Curr. Opin. Immunol. 39, 3943 (2016).

4. Sakaguchi, S., Vignali, D. a. a., Rudensky, A. Y., Niec, R. E. & Waldmann, H. The plasticity and stability of regulatory T cells. Nat. Rev. Immunol. 13, 461467 (2013).

5. Newton, R., Priyadharshini, B. & Turka, L. A. Immunometabolism of regulatory T cells. Nat. Immunol. 17, 618625 (2016).

6. Michalek, R. D. et al. Cutting edge: distinct glycolytic and lipid oxidative metabolic programs are essential for effector and regulatory CD4 T cell

subsets. J. Immunol. 186, 32993303 (2011).7. Apostolidis, S. A. et al. Phosphatase PP2A is requisite for the function of regulatory T cells. Nat. Immunol. 17, 556564 (2016).

8. Wei, J. et al. Autophagy enforces functional integrity of regulatory T cells by coupling environmental cues and metabolic homeostasis. Nat. Immunol. 17, 277286 (2016).

9. Layman, A. A. K. & Oliver, P. M. Ubiquitin ligases and deubiquitinating enzymes in CD4 T cell effector fate choice and function. J. Immunol. 196,

39753982 (2016).10. OLeary, C. E., Lewis, E. L. & Oliver, P. M. Ubiquitylation as a rheostat for TCR signaling: from targeted approaches toward global proling. Front. Immunol. 6, 618 (2015).

11. Harvey, K. F., Shearwin-Whyatt, L. M., Fotia, A., Parton, R. G. & Kumar, S. N4WBP5, a potential target for ubiquitination by the Nedd4 family of proteins, is a novel Golgi-associated protein. J. Biol. Chem. 277, 93079317 (2002).

12. Oliver, P. M. et al. Ndp1 protein promotes the function of itch ubiquitin ligase to prevent T cell activation and T helper 2 cell-mediated inammation. Immunity 25, 929940 (2006).

13. Fang, D. et al. Dysregulation of T lymphocyte function in itchy mice: a role for Itch in TH2 differentiation. Nat. Immunol. 3, 281287 (2002).

14. Beal, A. M., Ramos-Hernndez, N., Riling, C. R., Nowelsky, E. A. & Oliver, P. M. TGF-b induces the expression of the adaptor Ndp1 to silence IL-4 production during iTreg cell differentiation. Nat. Immunol. 13, 7785 (2011).

15. Jin, H. S., Park, Y., Elly, C. & Liu, Y. C. Itch expression by Treg cells controls Th2 inammatory responses. J. Clin. Invest. 123, 49234934 (2013).

16. Ramon, H. E. et al. The ubiquitin ligase adaptor Ndp1 regulates T cell-mediated gastrointestinal inammation and inammatory bowel disease susceptibility. Mucosal Immunol. 4, 314324 (2011).

17. Rubtsov, Y. P. et al. Regulatory T cell-derived interleukin-10 limits inammation at environmental interfaces. Immunity 28, 546558 (2008).

18. Godfrey, V. L., Wilkinson, J. E. & Russell, L. B. X-linked lymphoreticular disease in the scurfy (sf) mutant mouse. Am. J. Pathol. 138, 13791387 (1991).

19. Wing, K. et al. CTLA-4 control over Foxp3 regulatory T cell function.

Science 322, 271275 (2008).20. Smigiel, K. S. et al. CCR7 provides localized access to IL-2 and denes homeostatically distinct regulatory T cell subsets. J. Exp. Med. 211, 121136 (2014).

21. Yang, X. O. et al. Molecular antagonism and plasticity of regulatory and inammatory T cell programs. Immunity 29, 4456 (2008).

22. Duarte, J. H., Zelenay, S., Bergman, M. L., Martins, A. C. & Demengeot, J. Natural Treg cells spontaneously differentiate into pathogenic helper cells in lymphopenic conditions. Eur. J. Immunol. 39, 948955 (2009).

23. Pandiyan, P. & Zhu, J. Origin and functions of pro-inammatory cytokine producing Foxp3 regulatory T cells. Cytokine 76, 1324 (2015).

24. Feng, Y. et al. Control of the inheritance of regulatory T cell identity by a cis element in the foxp3 locus. Cell 158, 749763 (2014).

25. Ramon, H. E., Beal, A. M., Liu, Y., Worthen, G. S. & Oliver, P. M. The E3 ubiquitin ligase adaptor Ndp1 regulates Th17 differentiation by limiting the production of proinammatory cytokines. J. Immunol. 188, 40234031 (2012).

26. Layman, A. A. K., Sprout, S. L., Phillips, D. & Oliver, P. M. Ndp1 restricts Th17 cell potency by limiting lineage stability and proinammatory cytokine production. Sci. Rep. 7, 39649 (2017).

27. Nagaraj, N. et al. Deep proteome and transcriptome mapping of a human cancer cell line. Mol. Syst. Biol. 7, 548 (2011).

28. Schwanhausser, B. et al. Global quantication of mammalian gene expression control. Nature 473, 337342 (2011).

NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15677

29. Arike, L. et al. Comparison and applications of label-free absolute proteome quantication methods on Escherichia coli. J. Proteomics 75, 54375448 (2012).

30. Huehn, J. et al. Developmental stage, phenotype, and migration distinguish naive- and effector/memory-like CD4 regulatory T cells. J. Exp. Med. 199,

303313 (2004).31. Zhang, C. S. et al. The lysosomal v-ATPase-ragulator complex is a common activator for AMPK and mTORC1, acting as a switch between catabolism and anabolism. Cell Metab. 20, 526540 (2014).

32. Jewell, J. L., Russell, R. C. & Guan, K.-L. Amino acid signalling upstream of mTOR. Nat. Rev. Mol. Cell Biol. 14, 133139 (2013).

33. Bar-Peled, L. & Sabatini, D. M. Regulation of mTORC1 by amino acids. Trends Cell Biol. 24, 400406 (2014).

34. Gerriets, V. & Kishton, R. Metabolic programming and PDHK1 control CD4

T cell subsets and inammation. J. Clin. Invest. 125, 114 (2014).35. Chang, C. H. et al. Posttranscriptional control of T cell effector function by aerobic glycolysis. Cell 153, 12391251 (2013).

36. Yang, K. et al. T cell exit from quiescence and differentiation into Th2 cells depend on Raptor-mTORC1-mediated metabolic reprogramming. Immunity 39, 10431056 (2013).

37. Coquillard, C., Vilchez, V., Marti, F. & Gedaly, R. mTOR signaling in regulatory T cell differentiation and expansion. SOJ Immunol. 3, 110 (2015).

38. Pearce, E. & Pearce, E. Metabolic pathways in immune cell activation and quiescence. Immunity 38, 633643 (2013).

39. Zeng, H. et al. mTORC1 couples immune signals and metabolic programming to establish T(reg)-cell function. Nature 499, 485490 (2013).

40. OLeary, C. E. et al. Ndp-mediated degradation of Jak1 tunes cytokine signalling to limit expansion of CD4 effector T cells. Nat. Commun. 7, 11226

(2016).41. Shrestha, S. et al. Treg cells require the phosphatase PTEN to restrain TH1 and TFH cell responses. Nat. Immunol. 16, 178187 (2015).

42. Huynh, A. et al. Control of PI(3) kinase in Treg cells maintains homeostasis and lineage stability. Nat. Immunol. 16, 188196 (2015).

43. Altin, J. A. et al. Ndp1 mediates peripheral tolerance to self and exogenous antigen by inducing cell cycle exit in responding CD4 T cells. Proc. Natl

Acad. Sci. 111, 20672074 (2014).44. Elbe-Brger, A. et al. Overexpression of IL-4 alters the homeostasis in the skin.J. Invest. Dermatol. 118, 767778 (2002).45. Mund, T. & Pelham, H. R. B. Control of the activity of WW-HECT domain E3 ubiquitin ligases by NDFIP proteins. EMBO Rep. 10, 501507 (2009).

46. Riling, C. et al. Itch WW domains inhibit its E3 ubiquitin ligase activity by blocking E2-E3 ligase trans-thiolation. J. Biol. Chem. 290, 2387523887 (2015).

47. Ramos-Hernndez, N. et al. Ndp1 enforces a requirement for CD28 costimulation by limiting IL-2 production. J. Immunol. 191, 15361546 (2013).

48. Kurzweil, V., Tarangelo, A. & Oliver, P. M. Gastrointestinal microbiota do not signicantly contribute to T cell activation or GI inammation in Ndp1-cKO mice. PLoS ONE 7, e34478 (2012).

49. Thomas, R. M., Gao, L. & Wells, A. D. Signals from CD28 induce stable epigenetic modication of the IL-2 promoter. J. Immunol. 174, 46394646 (2005).

50. Collison, L. W. & Vignali, D. A. A. In vitro Treg suppression assays. Methods Mol Biol. 707, 2137 (2011).

51. van der Windt, G. J. W., Chang, C.-H. & Pearce, E. L. Measuring bioenergetics in T cells using a seahorse extracellular ux analyzer. Curr. Protoc. Immunol. 113, 3 (2016).

52. Shevchenko, A., Wilm, M., Vorm, O. & Mann, M. Mass spectrometric sequencing of proteins from silver-stained polyacrylamide Gels. Anal. Chem. 68, 850858 (1996).

53. Micallef, L. & Rodgers, P. euler APE: drawing area-proportional 3-Venn diagrams using ellipses. PLoS ONE 9, e101717 (2014).

54. Maere, S., Heymans, K. & Kuiper, M. BiNGO: a Cytoscape plugin to assess overrepresentation of Gene Ontology categories in Biological Networks. Bioinformatics 21, 34483449 (2005).

55. Christmas, R. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Am. Assoc. Cancer Res. Educ. Book 1216 (2005).

56. Vizcano, J. A. et al. 2016 update of the PRIDE database and its related tools. Nucleic Acids Res. 44, D447D456 (2016).

Acknowledgements

We thank Steve Seeholzer, Hossein Fazelinia, Hua Ding and Lynn Spruce of the Childrens Hospital of Philadelphia Research protein core. We thank Stephanie Sprout for mouse colony management. This work was supported by the National Institutes of Health Grants R01AI093566 and R01AI114515 to P.M.O.; T32CA009140-39 to A.A.K.L.; the American Asthma Foundation Grant 13-0020 to P.M.O.; and by the University of Pennsylvania Diabetes Center Islet Cell Biology Core P30-DK19525 to N.M.D.

Author contributions

A.A.K.L., G.D., C.E.O., S.T., R.M.T. and N.M.D. performed experiments, analysed data and/or assembled gures. N.M.D. carried out the metabolic function assays. E.K.M. assisted with in vivo injections and discussions. J.M.D. assisted with proteomic bioinformatics analysis. A.A.K.L. wrote the manuscript. P.M.O. conceived the project, provided guidance and edited the manuscript. All authors read/edited the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Layman, A. A. K. et al. Ndp1 restricts mTORC1 signalling and glycolysis in regulatory T cells to prevent autoinammatory disease. Nat. Commun.8, 15677 doi: 10.1038/ncomms15677 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the articles Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/ http://creativecommons.org/licenses/by/4.0/

Web End =licenses/by/4.0/

r The Author(s) 2017

16 NATURE COMMUNICATIONS | 8:15677 | DOI: 10.1038/ncomms15677 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

Word count: 13943

Show less

Abstract

Translate

Foxp3⁺ T regulatory (T_reg ) cells suppress immune cell activation and establish normal immune homeostasis. How T_reg cells maintain their identity is not completely understood. Here we show that Ndfip1, a coactivator of Nedd4-family E3 ubiquitin ligases, is required for T_reg cell stability and function. Ndfip1 deletion in T_reg cells results in autoinflammatory disease. Ndfip1-deficient T_reg cells are highly proliferative and are more likely to lose Foxp3 expression to become IL-4-producing T_H 2 effector cells. Proteomic analyses indicate altered metabolic signature of Ndfip1-deficient T_reg cells and metabolic profiling reveals elevated glycolysis and increased mTORC1 signalling. Ndfip1 restricts T_reg cell metabolism and IL-4 production via distinct mechanisms, as IL-4 deficiency does not prevent hyperproliferation or elevated mTORC1 signalling in Ndfip1-deficient T_reg cells. Thus, Ndfip1 preserves T_reg lineage stability and immune homeostasis by preventing the expansion of highly proliferative and metabolically active T_reg cells and by preventing pathological secretion of IL-4 from T_reg cells.

Details

Title

Ndfip1 restricts mTORC1 signalling and glycolysis in regulatory T cells to prevent autoinflammatory disease

Author

Layman, Awo Akosua Kesewa; Deng, Guoping; O'leary, Claire E; Tadros, Samuel; Thomas, Rajan M; Dybas, Joseph M; Moser, Emily K; Wells, Andrew D; Doliba, Nicolai M; Oliver, Paula M

Pages

15677

Publication year

2017

Publication date

Jun 2017

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms15677

ProQuest document ID

1905706279

Ndfip1 restricts mTORC1 signalling and glycolysis in regulatory T cells to prevent autoinflammatory disease

Jump to:

Full Text

Abstract

Details

Suggested sources