MiR-125a targets effector programs to stabilize Treg-mediated immune homeostasis

Abstract

Although different autoimmune diseases show discrete clinical features, there are common molecular pathways intimately involved. Here we show that miR-125a is downregulated in peripheral CD4⁺ T cells of human autoimmune diseases including systemic lupus erythematosus and Crohn's disease, and relevant autoimmune mouse models. miR-125a stabilizes both the commitment and immunoregulatory capacity of Treg cells. In miR-125a-deficient mice, the balance appears to shift from immune suppression to inflammation, and results in more severe pathogenesis of colitis and experimental autoimmune encephalomyelitis (EAE). The genome-wide target analysis reveals that miR-125a suppresses several effector T-cell factors including Stat3, Ifng and Il13. Using a chemically synthesized miR-125a analogue, we show potential to re-programme the immune homeostasis in EAE models. These findings point to miR-125a as a critical factor that controls autoimmune diseases by stabilizing Treg-mediated immune homeostasis.

ARTICLE

Received 8 May 2014 | Accepted 2 Apr 2015 | Published 12 May 2015

Wen Pan1,2,3,*, Shu Zhu4,*, Dai Dai1,2,*, Zheng Liu5, Dan Li6, Bin Li6, Nicola Gagliani4, Yunjiang Zheng4,

Yuanjia Tang1,2, Matthew T. Weirauch7, Xiaoting Chen8, Wei Zhu5, Yue Wang5, Bo Chen5, Youcun Qian1, Yingxuan Chen9, Jingyuan Fang9, Ronald Herbst5, Laura Richman5, Bahija Jallal5, John B. Harley7,10, Richard A. Flavell4, Yihong Yao5 & Nan Shen1,2,7,11

Although different autoimmune diseases show discrete clinical features, there are common molecular pathways intimately involved. Here we show that miR-125a is downregulated in peripheral CD4 T cells of human autoimmune diseases including systemic lupus erythematosus and Crohns disease, and relevant autoimmune mouse models. miR-125a stabilizes both the commitment and immunoregulatory capacity of Treg cells. In miR-125adecient mice, the balance appears to shift from immune suppression to inammation, and results in more severe pathogenesis of colitis and experimental autoimmune encephalomyelitis (EAE). The genome-wide target analysis reveals that miR-125a suppresses several effector T-cell factors including Stat3, Ifng and Il13. Using a chemically synthesized miR-125a analogue, we show potential to re-programme the immune homeostasis in EAE models. These ndings point to miR-125a as a critical factor that controls autoimmune diseases by stabilizing Treg-mediated immune homeostasis.

1 State Key Laboratory of Oncogenes and Related Genes, Shanghai Cancer Institute, Renji Hospital, Shanghai Jiaotong University School of Medicine, Shanghai 200032, China. 2 Joint Molecular Rheumatology Laboratory of the Institute of Health Sciences and Shanghai Renji Hospital, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, and Shanghai Jiaotong University School of Medicine, Shanghai 200025, China. 3 Department of Genetics, Yale University School of Medicine, New Haven, Connecticut 06520, USA. 4 Department of Immunobiology, Yale University School of Medicine, New Haven, Connecticut 06520, USA. 5 MedImmune LLC, Gaithersburg, Maryland 20878, USA. 6 Institute Pasteur of Shanghai, Chinese Academy of Sciences, Shanghai 200025, China. 7 Center for Autoimmune Genomics and Etiology (CAGE) and Divisions of Rheumatology, Cincinnati Childrens Hospital Medical Center, Cincinnati Ohio 45229, USA. 8 School of Electronic and Computing Systems, University of Cincinnati, Cincinnati, Ohio 45221, USA. 9 Division of Gastroenterology and Hepathology, Renji Hospital, Shanghai Institute of Digestive Disease, Shanghai Jiaotong University School of Medicine, Shanghai 200001, China. 10 US Department of Veterans Affairs Medical Center, Cincinnati, Ohio 45220, USA. 11 Collaborative Innovation Center of Systems Biomedicine, Shanghai Jiaotong University, Shanghai 200240, China. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to N.S. (email: mailto:[email protected]

Web End [email protected] ) or to Y.Y. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms8096

MiR-125a targets effector programs to stabilize Treg-mediated immune homeostasis

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

The clinical presentation of any specic autoimmune disease is the culmination of complex interactions between genetics, primary and secondary immune effector

mechanisms, and environmental triggers1,2. Although different autoimmune diseases show discrete clinical features, there are common molecular pathways intimately involved, which ultimately dene specic diseases3,4. Distinct lineages of CD4

T cells can function to both drive and constrain immune-mediated pathology. While T effector cells are often implicated in multiple autoimmune diseases, Treg cells suppress inammation and control the development of multiple autoimmune diseases, such as type 1 diabetes, multiple sclerosis and inammatory bowel disease57. Foxp3 acts as a lineage specication factor of Treg cells. It binds to multiple genes and plays a dual role as both a transcriptional activator and repressor8. Sustained expression of Foxp3 amplies and stabilizes molecular features of Treg precursor cells. Simultaneously, it represses the molecular features deleterious to Treg cell differentiation or function, thereby preventing deviation of Treg cells into effector T-cell lineages9,10.

The extensive involvement of microRNAs (miRNAs) across human diseases helps to understand the pathogenesis of various ailments and might lead to eventual development of effective therapeutics against these diseases11. A subset of miRNAs have shown therapeutic promise and are currently actively pursued either in preclinical or clinical trials12,13. However, most studies and patents in the miRNA eld are related to the treatment of cancer, viral infections, ocular disorders, cardiovascular disorders and neurological disorders. Promising miRNA candidates for therapeutic development in autoimmune diseases are still very limited12.

In this study, we show that miR-125a is downregulated in peripheral CD4 T cells of multiple human autoimmune diseases and in relevant mouse models. We further utilize miR-125a-decient mice and show that miR-125a stabilizes both the commitment and immunosuppressive capacity of Treg cells. Deciency of miR-125a results in more severe colitis and encephalomyelitis (EAE) pathogenesis. The genome-wide target analysis reveals that miR-125a suppresses several effector T-cell factors such as Stat3, Ifng and Il13. Moreover, a chemically synthesized miR-125a analogue not only contributes to EAE prevention but also alleviates established EAE. Together, our ndings demonstrate miR-125a as a critical regulator that stabilizes Treg-mediated immune homeostasis and controls autoimmunity.

ResultsMiR-125a is downregulated in autoimmune CD4 T cells.

CD4 T-helper cells play a variety of important roles in the development and maintenance of various autoimmune diseases.

Our miRNA microarray data showed that miR-125a-5p (named miR-125a here) was among top 10 downregulated miRNAs in the splenic CD4 T cells isolated from MRL/lpr mouse strain, the most commonly studied mouse model for human lupus at 5 and 16 weeks of age (Supplementary Fig. 1a,c), as well as in the splenic CD4 T cells isolated from different progressive stages of experimental autoimmune EAE, which is a mouse model for human multiple sclerosis (Supplementary Fig. 1b,d). MiR-125a was also the most signicantly downregulated miRNA among all commonly altered miRNAs in the two sets of microarray data. Moreover, previous studies have shown that downregulation of miR-125a in peripheral T cells is associated with human multiple sclerosis14. Therefore, we hypothesized that miR-125a might be a commonly dysregulated miRNA in some human autoimmune diseases. To test the hypothesis, we collected samples from

another two different human autoimmune diseases and some mouse models, and conrmed that miR-125a was indeed signicantly decreased in disease CD4 T cells compared with healthy CD4 T cells. These samples included 22 patients with systemic lupus erythematosus (SLE) versus 20 healthy controls

0.0 CTRL CD

2.5 ***

8 ***

2.0

miR-125a

(relative expression)

6

miR-21

(relative expression)

1.5

4

1.0

2

0.5

0.0 CTRL SLE

0 CTRL SLE

2.0

*

3

**

(relative expression)

1.5

2

miR-125a

1.0

miR-155

(relative expression)

1

0.5

0 CTRL CD

1.5 **

NS

6 ** **

miR-125a

(relative expression)

(relative expression)

1.0

4

miR-21

(relative expression)miR-326

0.5

2

0.0 0 d 16 d 32 d

0 0 d 16 d 32 d

2.0 NS **

5 week 16 week

0 CTRL LPR CTRL LPR

20

***

miR-125a

(relative expression)

1.5

15

1.0

10

***

0.5

5

0.0

CTRL LPR CTRL LPR

5 week 16 week

Figure 1 | MiR-125a is downregulated in autoimmune CD4 T cells.(a) Quantitative realtime (RT)PCR validation of miR-125a expression, with miR-21 as a positive control, in peripheral CD4 T cells from 22 SLE patients and 20 healthy controls (CTRLs). (b) Quantitative RTPCR validation of miR-125a expression, with miR-155 as a control, in peripheral CD4 T cells from 12 patients with CD and 14 healthy controls. (c)

Quantitative RTPCR validation of miR-125a expression, with miR-326 as a positive control, in splenic CD4 T cells isolated from six naive control

C57BL/6 mice, six EAE mice 16 days after immunization or six EAE mice 32 days after immunization. (d) Quantitative RTPCR validation of miR-125a expression, with miR-21 as a positive control, in splenic CD4 T cells isolated from MRL/lpr mice at 5 or 16 weeks of age and normal controls.

n 8 Mice per group. The relative miRNA expression was normalized to the

expression of RNU48 (a,b) or snoRNA202 (c,d). All data are means.e.m. *Po0.05, **Po0.01, ***Po0.001 (nonparametric MannWhitney U-test).

NS, not signicant. Data are representative of two (a,b) or three (c,d) independent experiments. For additional clinical information, refer to Supplementary Tables 1 and 2.

2 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096 ARTICLE

(Fig. 1a), 12 patients with Crohns disease (CD) versus 14 healthy controls (Fig. 1b), myelin olygodendrocyte glycoprotein (MOG)-induced EAE versus controls (Fig. 1c) and MRL/lpr mouse strain versus control mice (Fig. 1d). Together, these data suggest that miR-125a may be a commonly downregulated miRNA in multiple human autoimmune diseases and related mouse models.

Treg differentiation is defective in miR-125a / T cells. To further explore the function of miR-125a in autoimmunity, we generated miR-125a transgenic (TG) mice (Supplementary Fig. 2) and knockout (KO) mice (Supplementary Fig. 3). Percentages of T (CD4 or CD8 ), B (B220 ) or myeloid (Mac-1 ) lineages in peripheral lymph nodes or spleens from 6- to 8-week-old miR-125a KO and control wild-type (WT) mice were initially analysed by ow cytometry. The results showed that there were no statistical differences in these lineages between WT and KO mice (Supplementary Table. 3). We next checked whether miR-125a affected CD4 T-cell activation or proliferation. Flow cytometric analysis showed that the percentages of different activation markers in CD4 T (Supplementary Fig. 4a) or proliferation of carboxyuorescein succinimidyl ester (CFSE)-labelled CD4 T cells (Supplementary Fig. 4b) were not signicantly altered between KO and WT mice. As a previously reported miRNA prole data have shown that miR-125a was highly expressed in mouse CD4 CD25 GITR

Treg cells15, we next examined and found that miR-125a was indeed highly expressed in CD4 CD25hi population compared with naive T population (Fig. 2a) and was also markedly induced in in vitro-derived Treg cell subset (Fig. 2b). These data implied that miR-125a might have a function in Treg cells. We further examined the percentages of different T-cell subsets in CD4

T population from the spleens of KO or WT mice, and found that the proportion and absolute number of Foxp3 Treg cells showed a signicant reduction in miR-125a KO mice compared with that in WT mice (Fig. 2ce), while the other T-cell subsets such as TH1, TH2 and TH17 showed no signicant differences (Supplementary Table. 3). To conrm the defective differentiation of Treg in miR-125a-decient mice, we also sorted naive CD4 T cells (CD4 CD25 CD44loCD62Lhi)

from 6- to 8-week-old miR-125-decient mice and their WT littermates, cultured the cells under different T-cell subset-skewing conditions for 4 days and then assessed their differentiation efciency. Flow cytometric analysis revealed that miR-125a-decient T cells were indeed impaired in their ability to differentiate into Foxp3 population compared with WT T cells, even at a high concentration of transforming growth factor (TGF)-b (Fig. 2f). Furthermore, the Foxp3 mRNA level was also signicantly decreased in miR-125a-decient Treg cells relative to that in their WT controls (Fig. 2g). Consistent with the above in vivo observation, ow cytometric analysis showed miR-125a had no obvious effects on in vitro TH1, TH2 and TH17 differentiation (Supplementary Fig. 5). Together, these results show that miR-125a-decient T cells may be defective in their ability to differentiate into Treg cells, both in vitro and in vivo.

MiR-125a / Treg cells fail to prevent colitis effectively. We next tried to examine whether there was any defect in the immunoregulatory capacity of Treg cells. To achieve that, we examined Treg cells from miR-125a-decient and WT mice in overcoming the immune pathogenesis caused by adoptive transfer of naive CD4 T cells into immunodecient Rag1/ mice.

Briey, naive T cells (CD4 CD25 CD44loCD62Lhi) were sorted from C57BL/6 WT mice and Treg cells (CD4 CD25hi) were sorted from WT or miR-125a-decient mice. Then, naive T cells were transferred into Rag1/ mice alone or in combination

with WT or miR-125a-decient Treg cells (Fig. 3a). Disease progression was assessed by weight loss within 40 days. We observed that Rag1/ mice transferred with WT naive T cells alone gradually developed severe body weight loss. Co-transfer of sufcient numbers of WT Treg cells prevented the disease much more effectively than co-transfer of miR-125a-decient Treg cells, indicating a defect of Treg cell-mediated suppression in miR-125a-decient mice (Fig. 3b). Haematoxylineosin staining of the colon sections also showed denser inltration of leukocytes and more severe epithelial hyperplasia in Rag1/ mice co-transferred with miR-125a-decient Treg cells than those co-transferred with WT Treg cells (Fig. 3c). Taken together, these data suggest that miR-125a-decient Treg cells fail to effectively prevent the immunopathology of T-cell transfer-induced colitis. MiR-125a may contribute to sustain Treg cell-mediated immunosuppressive function.

MiR-125a / mice are susceptible to autoimmune EAE. To test whether failure to maintain immune homeostasis in the absence of miR-125a was associated with some autoimmune disorders, we induced EAE model in miR-125a-decient and WT mice. In EAE, Treg cells are required for the suppression of both TH1- and TH17-cell subsets16. We rst evaluated a set of parameters in MOG-induced EAE pathology, and observed that the clinical score and amounts of inammatory inltration and demyelination were signicantly increased in the miR-125adecient mice compared with those in their WT controls (Fig. 4a,b). Consistently, the mRNA levels of genes related to TH1, such as Ifng, and those related to TH17 lineage, such as Il17a,

Il17f, Il21 and Il23r, were signicantly upregulated in splenic CD4 T cell from miR-125a-decient mice compared with those from WT controls (Fig. 4c). The concentration of interferon (IFN)-g and interleukin-17a (IL-17a) in the supernatants from splenocyte culture challenged with MOG peptide was also increased in miR-125a-decient mice compared with WT controls (Fig. 4d). Furthermore, intracellular cytokine staining showed the percentage of Treg cells in the CD4 T-cell population isolated from the spleens (Fig. 4e,f) and the central nervous system (CNS) inltrates (Fig. 4g,h) were signicantly decreased in miR-125a-decient mice compared with those in the WT controls. In contrast, the percentages of inammatory IFN-gand IL-17a- producing subsets in the CD4 population were higher in miR-125-decient mice than that in WT mice in both the spleens (Fig. 4e,f) and the CNS inltrates (Fig. 4g,h). Consistently, the clinical score of EAE was lower in the miR-125a TG mice than in WT mice (Supplementary Fig. 6a). There were higher levels of in vitro-derived Foxp3 Treg cells and Foxp3 mRNA in miR-125a TG mice than in WT mice (Supplementary Fig. 6b,c). The percentage of IFN-g was signicantly decreased in the CD4 T cells isolated from the

CNS of TG mice compared with WT mice (Supplementary Fig. 6d). Collectively, these experiments show that miR-125a-decient mice are susceptible to autoimmune EAE, manifesting fewer Treg population and greater production of proinammatory cytokines in the periphery and CNS.

MiR-125a / Treg cells display increased effector T genes. To gain insights into the general functional features of miR-125adecient Treg cells, we sorted CD4 CD25hi Treg population from the spleens of 6- to 8-week-old miR-125a-decient and their littermate WT mice, and then performed a genome-wide gene-array analysis. We then predicted the biological functions and mapped the biomolecular networks of the differentially expressed gene transcripts using Ingenuity Pathway Analysis tool. The most signicant enrichment of the differentially expressed genes was

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

*

Relative miR-125a levels

8

10 **

9

Relative mir-125a levels

**

8

6

7

6

4

5

4

2

3

2

0

0

Naive T

Treg

Naive T

1 T T H1

H0

T iTreg

H2

T Non-T

H17

WT

13.1 7.2

KO

15 *

20 *

Foxp3

10 10 10 10

10 10 10 10

Foxp3+ cells (%CD4+ )

Foxp3+ cells (105 )

15

10

10

5

5

10 10

10

10

10

10

10

10

10

0 WT KO

0 WT KO

CD4

TGF- (ng ml1)

1 5 10

10 10 10 10

10 10 10 10

10 10 10 10

27.8

11.2

54.4

38.3

71.2

44.1

WT

KO

10

10

10

10

10 10

10

10

10

10

10

Foxp3

(relative mRNA levels)

10 WT 125a KO

* *

*

*

Foxp3

8

10 10 10 10

10 10 10 10

10 10 10 10

6

4

2

0 1 5 10

10 10

10

10

10 10

10

10

10

10

10

10

TGF- (ng ml1)

CD4

Figure 2 | Treg differentiation is defective in miR-125a / T cells. (a) Naive T cells (CD4 CD25 CD44loCD62hi) and Treg cells (CD4 CD25hi) were isolated from the spleens of C57BL/6 mice. Quantitative RTPCR analysis of miR-125a expression was performed on these two populations.

(b) Quantitative RTPCR analysis of miR-125a expression in different CD4 T-helper cell subsets differentiated in vitro for 4 days under different skewing conditions. (c) Flow cytometric analysis of Foxp3 Treg cells from the spleens of 6- to 8-week-old miR-125a-decient mice and their WT littermates.

(d) Frequency of Foxp3 Treg cells in the CD4 population in c (n 6 per group). (e) Absolute numbers of Foxp3 Treg cells in c (n 6 per group).

(f) Flow cytometric analysis of naive CD4 Tcells (CD4 CD25 CD44loCD62Lhi) isolated from the spleens of 6- to 8-week-old miR-125a-decient mice and their WT littermates, cultured for 4 days with IL-2 (5 U ml 1), anti-CD3 antibody (2 mg ml 1) and anti-CD28 antibody (2 mg ml 1), plus various concentrations of TGF-b. (g) Quantitative RTPCR analysis of Foxp3 expression in Treg cells differentiated as in f from miR-125a-decient mice or their WT littermate controls. The relative miR-125a expression was normalized to the expression of snoRNA202 (a,b), while the relative expression levels of Foxp3 were normalized to the expression of mouse Rpl13a. Data are displayed as means.e.m.*Po0.05, **Po0.01 versus control (two-tailed Students t-test).

Data are representative of three (af) or four (g) independent experiments.

related to T-helper cell pathways. A salient characteristic of Treg cells is their inability to produce proinammatory cytokines. However, in the absence of miR-125a, Treg cells showed increased expression of genes related to TH1-, TH2- and TH17 subsets, including some lineage-specic transcriptional factors, cytokines and cytokine receptors. Consistently, Treg lineagespecic transcription factor Foxp3 was decreased, together with some cytokine genes such as Tgfb1 (Fig. 5a). We further examined the expression of most of the genes cited above with real-time PCR. The results conrmed the array data (Fig. 5b), showing that in the absence of miR-125a, Treg cells showed increased expression of effector lineage genes and reduced expression of Treg lineage genes such as Foxp3. These data also implied that sustained expression of miR-125a in Treg cells may restrain the differentiation programs of effector lineages to stabilize Treg lineage commitment. Although deciency of miR-125a in Treg cells upregulated TH1-, TH2- and TH17-related genes, miR-125a had no obvious effects on in vitro TH1, TH2 and TH17

differentiation, as well as in vivo proportions of these T effector subsets. Therefore, we postulated that the exaggerated proinammatory cytokine production in miR-125a-decient mice at pathogenesis conditions such as EAE might be mainly attributable to the break of immune tolerance caused by reduced differentiation or suppressor function of Treg cells, but might not be a cell-intrinsic property of miR-125a-decient effector T cells.

Mir-125a inhibits several effector T lineage factors. The above results raised several questions such as how exactly miR-125a restrained effector lineage programs in Treg cells, and whether the elevated genes in miR-125-decient Treg cells were the direct targets of miR-125a or the consequence of a secondary effect caused by the reduced expression of Foxp3. To answer these questions, we screened the direct targets of miR-125a by combining target prediction by TargetScan with microarray gene expression data to identify the overlapping and upregulated gene transcripts. Using a combination of these two approaches, we

4 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096 ARTICLE

WT

WT

KO

CD4+CD25CD44loCD62Lhi Naive T cells

CD4+CD25hi treg cells

4105

2105

Control Naive

Naive + WT treg Naive + KO treg

Rag1/

Naive Tn (n=6)

Naive Tn+WT treg (n=6)

Naive Tn+KO treg (n=6)

Days post transfer

110

% Of initial weight

100

* * * *

90

80

0 5 10 15 20 25 30 35 40

Figure 3 | MiR-125a / Treg cells fail to prevent colitis effectively. (a) Schematic illustration of how T-cell transfer model of colitis was investigated. Each Rag1/ recipient mouse was transferred with 5 105 CD4 CD25 CD44loCD62Lhi naive T cells isolated from WT mice, and co-transferred with

2.5 105 CD4CD25hi Treg cells isolated from WT or miR-125a-decient mice. (b) The weight loss of the mice in a was monitored every 4 days for

40 days from the day after injection. n 6 Mice for each group. *Po0.05 (multiple Students t-test). (c) Representative haematoxylineosin staining

of colon sections from untreated WT mice (control), Rag1/ mice transferred with WT naive T cells (naive), Rag1/ mice co-transferred with WT naive T cells and WT Treg cells (naive WT Treg), and Rag1/ mice co-transferred with WT naive T cells and miR-125a-decient Treg cells (naive KO Treg;

original magnication, 20). Data are representative of two independent experiments (means.e.m.).

identied 74 genes. Among them, only ve predicted genes (Stat3, Tnfsf4, Tnfaip3, Mapkapk2 and Ppp5c) have been reported to have the potential functions in immune responses or pathways (Supplementary Table 4). We conrmed the inhibitory effect using a luciferase reporter assay and found that miR-125a signicantly suppressed the luciferase activity of the 30 untranslated region (UTR) of Stat3 and Tnfsf4 (Supplementary Fig. 7). To avoid potentially signicant miR-125a targets from being excluded, we next examined the genes that belonged to T-helper cell pathway and whose expression were inversely correlated to the expression of miR-125a in the array data, including those encoding major transcription factors, cytokines, cytokine receptors and signalling molecules, and focused on 21 potential genes (Supplementary Table 5). We next cloned the 30UTRs of all these genes into luciferase reporter constructs, and found that miR-125a markedly repressed the reporter activities of several genes, including Stat3, Il13 and Ifng, whereas the activity of others was either slightly affected or unaffected by miR-125a (Fig. 6a). We further identied the potential binding sites of miR-125a in the 30UTRs of Stat3, Il13 and Ifng by TargetScan (Fig. 6b). Mutation of these binding sites conrmed the suppressive roles of miR-125a on these genes (Fig. 6c). To further demonstrate that Stat3, Il13 and Ifng were the direct targets of miR-125a, we next detected the relative mRNA and protein levels of the targets after overexpression of miR-125a. Since Stat3 is constitutively expressed in some human and mouse cell lines, and the target sites are highly conserved between different species, we rst

transfected HEK293T cells with two different doses of miR-125a agomirs that are chemically synthesized double-stranded RNA (dsRNA) oligonucleotides used to mimic miRNA. We also infected mouse CD4 T cells with retrovirus-mediated over-expression of miR-125a. The results that showed miR-125a suppressed the protein levels of endogenous Stat3 expression in a dose-dependent manner (Fig. 6d). Considering Il13 and Ifng are secreted cytokines, we next generated retrovirus-based constructs to overexpress miR-125a, and infected CD4 T cells under

Th1- or Th2-skewing conditions for 4 days, respectively. The infection efciency, calculated as the percentage of GFP cells, was B40% when the cells were infected once, and 485% when they were infected repeatedly (Supplementary Fig. 8). The amounts of IFN-g and IL-13 in the supernatants of the cells infected with miR-125a were signicantly reduced compared with that in cells infected with the empty vector (Fig. 6e). The mRNA levels of Stat3, Ifng and Il13 were also signicantly reduced after overexpression of miR-125a (Fig. 6f). Considering the fact that Stat3 is a critical transcriptional factor required for TH17 differentiation, which inhibits Smad3 binding to enhancer I to limit Treg differentiation17,18, Ifng and Il13 are TH1/TH2 polarizers that can also potently inhibit Foxp3 expressing Treg cells differentiation from naive CD4 precursors19,20, these data therefore show that miR-125a directly represses several targets of effector T lineages that are detrimental to Treg differentiation, which may help to stabilize Treg lineage commitment and function.

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

H&E Luxol fast blue

WT (n=6) MiR-125 KO (n=5)

*

4

WT

KO

****

Mean clinical score

3

2

1

0 1 5 10 15 20 25 30

Day after immunization

5 WT 125 KO ** *

*

* *

20 WT

125 KO

*

Concentration (ng ml1 )

15

mRNA (relative)

4 Il17a Il17f Il21 Il22 Il23r Ifng Il4 Tgfb

3

10

*

*

IL17a

2

5

1

0

0 IFN-

WT

KO

WT KO

9.0

0.3

3.4

18.7

0.8

20.8 15.1

103

102

101

100

103

102

101

100

103

102

101

100

103

102

101

100

Foxp3

IFN-

87.3

74.5

6.0

100 103

102

101

100 103

102

101

102

103

100 103

102

101

100

103

101

CD4

IL-17

WT KO26.5 14.4

IFN-

WT KO

16.8

1.8

7.1

31.6

4.1

12.3

103

102

101

100

103

102

101

100

103

102

101

100

103

102

101

100

Foxp3

73.3

52.0

100 103

101

102

103

100

101

102

100

101

102

103

100

101

102

CD4

IL-17

* *

* *

25

25

**

10

*

40

40

Foxp3+ cells (%CD4+ )

15

IL-17+ cells (%CD4+ )

IFN-+ cells (%CD4+ )

IL-17+ cells (%CD4+ )

Foxp3+ cells (%CD4+ )

IFN-+ cells (%CD4+ )

20

20

8

30

30

15

15

6

10

20

20

10

10

4

5

5

5

2

10

10

0 WT KO

0 WT KO

0 WT KO

0 WT KO

0 WT KO

0 WT KO

Figure 4 | MiR-125a / mice are susceptible to autoimmune EAE. (a)Mean clinical scores (s.e.m.) of EAE induced in 6- to 8-week-old WT and miR-125a-decient mice. (b) Representative haematoxylineosin (H&E) staining or Luxol Fast Blue staining of spinal cord sections on day 21 after immunization of the mice in a. (c) Quantitative RTPCR analysis of gene expression in splenic CD4 T-cell samples on day 21 after immunization of the mice in a. The relative expression levels of different genes were normalized to the expression of mouse Rpl13a. (d) Enzyme-linked immunosorbent assay examination of the concentration of IFN-g and IL-17a in the supernatants from the splenocyte culture challenged with MOG peptide. (e,g) Representative ow cytometric analysis of the percentage of Treg, TH1 and TH17 cells in the CD4 Tcells isolated from the spleens (e) and CNS (g) of the mice (a) on day 21 after immunization. (f,h) The percentages of Foxp3 Treg cells, IFN-g T cells and IL-17 T cells in the CD4 populations from the spleens (e) and

CNS (g), respectively. *Po0.05, **Po0.01 versus control (two-tailed Students t-test). Data are representative of three (a,b, eh) or four (c,d) independent experiments (means.e.m.).

6 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096 ARTICLE

Transcription factors Secreted moleculars Cell surface moleculars

Nfat5

Klf9

Bcl6

4.41

2.60

1.41

Il9

Il4

Il17a

Tgfb1

11.07

5.36

1.98

0.74

Il10ra

Il17rb

Cxcr3

4.55

1.40

0.52

Batf3

2.88

Il5

9.71

Il17re

2.71

Gata3

Rorc

Stat3

2.67

1.541.41

Il13

Il21 Il24

5.60

4.192.46

Tnfsf4

Icos Cd28

1.67

0.630.60

Batf

2.57

Il22

5.24

Il17ra

0.64

Stat1

0.59

Il11

1.62

Tgfbr2

0.37

Klf4

0.45

Il6

1.57

Foxp3 Klf2

0.430.38

Ifng Il23a

1.351.34

TH1

TH2

TH17

Treg

Tbet

Ifng

Gata3

Il4

Il5

Il9

Il13

Il24

Rorc

Stat3

Batf

Batf3

Irf4

Il6

Il17a

Il21

Il22

Foxp3

Cd28

Icos

Tgfb1

Tgfbr2

5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Relative mRNA levels (fold KO/WT)

Figure 5 | MiR-125a / Treg cells display increased effector T genes. (a) Genome microarray analysis of CD4CD25hi Treg cells from miR-125adecient and WT mice. Splenic CD4 T cells from miR-125a-decient and

WT mice were enriched by Miltenyi Biotec beads. Then, Treg cells (CD4 CD25hi) were sorted by ow cytometry and harvested in Trizol.

Samples from three independent experiments were pooled for the microarray analysis. Ingenuity Pathway Analysis analysis was used to investigate which molecular pathways were affected by miR-125a deciency. Genes that are associated with the T-helper cell pathways show the most signicant enrichment, among which gene transcripts up- or downregulated by 430% are listed in a. Left, gene symbols. Right, fold change of gene expression in miR-125a-decient Treg compared with WT

Treg. (b) Quantitative RTPCR analysis of the expression levels of genes associated with T-helper cell differentiation pathway in a. Different T-helper cell subsets are marked in different colours. Data are representative of three (b) independent experiments (means.e.m.). Array data are available in GEO under accession code GSE66608.

MiR-125a agomir treatment ameliorates EAE pathogenesis. A series of studies have supported the systemic administration of miRNA as a therapeutic strategy21,22. Our current studies showed the suppressive roles of miR-125a on the pathogenesis of

autoimmune animal models. Therefore, we sought to evaluate the efcacy and potential application of a miR-125a-based therapy for autoimmune pathogenesis, such as EAE. Agomirs or antagomirs are synthetic RNA molecules with improved stability and delivery, and have been shown to achieve therapeutic overexpression or silencing of an endogenous miRNA gene in vivo23,24. To conrm the efcacy of a systemic administration of miR-125a agomir or antagomir in vivo, we rst injected mice intravenously with miR-125a agomir, antagomir or their corresponding controls. Different tissues were collected 24 h after the injection, and miR-125a expression was determined by quantitative PCR. The results showed that miR-125a can be efciently overexpressed or silenced in most tissues, including the liver, kidney and spleen, but not in the brain, suggesting that agomirs and antagomirs minimally cross the bloodbrain barrier (Supplementary Fig. 9). We also used orescence-labelled miR-125a agomirs to show the systemic transfection efciency of miRNA into CD4 T cells. Cy5-labelled miR-125a agomirs was rst transfected into CD4 T cells in vitro, and confocal microscopy image showed that the Cy5 uorescence label was successful (Supplementary Fig. 10a). Cy5-labelled agomirs were further intravenously injected into WT mice for continuous 3 days, and the lymphocytes were harvested 1 day later to test the transfection efciency. Flow data showed that the transfection efciency of Cy5-labelled agomir into CD4 T cells in vivo was

B58% in lymph nodes and 53% in spleen after three injections (Supplementary Fig. 10b). Since miR-125a sustains Treg cell differentiation and suppressive function during the pathogenesis of EAE, we postulated that treatment with miR-125a agomir could ameliorate the pathogenesis of EAE. As expected, after intravenously injecting a high dose of miR-125a agomir three times at the onset of EAE (within 1 day of initial symptoms), the disease severity was markedly reduced, accompanied by decreased CNS inammation and demyelination (Fig. 7a,b). Consistently, the percentage of Treg cells was signicantly increased(11.61.08% versus 18.71.59%; Po0.01; Fig. 7c), whereas the percentage of IFN-g (28.82.82% versus 20.72.07%, respectively; Po0.05) and IL-17 proinammatory subsets(12.81.06% versus 8.91.05%, respectively; Po0.05) were signicantly decreased in the CD4 T cells isolated from the

CNS after treatment with miR-125a agomirs (Fig. 7df). In contrast, the disease severity of EAE increased signicantly with the antagomir-mediated silencing of miR-125a at the onset of EAE (Supplementary Fig. 11), which is consistent with the results of miR-125a-decient mice. Moreover, to test the role of miR-125 in a therapeutic approach in EAE25,26, we also administered miR-125a agomir for continuous 3 days after acute EAE was established, at a clinical score 12. The results showed that miR-125a agomir treatment also signicantly ameliorated established EAE symptoms (Fig. 7g and Supplementary Fig. 12). Taken together, these data suggest that manipulation of miR-125a expression in vivo not only contributed to EAE prevention but also alleviated established EAE.

DiscussionThe extensive involvement of miRNAs across human diseases helps to understand the pathogenesis of various ailments and lead to therapeutics. However, the promising miRNA candidates for controlling autoimmune pathogenesis are still limited. Our previous work has that shown miR-125a is downregulated in the peripheral CD3 T cells of SLE patients27. Interestingly, miR-125a was also downregulated in oral lichen planus, which is a T-cell-mediated autoimmune disease of oral mucosa28. In this study, we further identied miR-125a as a commonly downregulated miRNA in peripheral CD4 T cells of multiple human autoimmune diseases such as SLE and CDs, as well as

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

150 NC Mir-125a

*

Luciferase activity (% control)

125

100

75

*

**

50

0 Tbx21

Il17re

Il17rb

Nfat5a Il6 Il17a Il21 Il22 Il23a

Bcl6

Stat5a

Batf3

Batf

Stat3

25

Psicheck

Ifng

Gata3

Il4

Il5

Il9

Il10ra

Il13

Rorc

NC+ wt UTR

Mir-125a+ wt UTR

NC+ mut UTR

Mir-125a+ mut UTR

** * *

0.0 Stat3 Il13 Ifng

H.saplensM.musculus

H.saplensM.musculus

H.saplensM.musculus

STAT3 Stat3

IL13 Il13

IFNG Ifng

150

8mer 7mer-m8 7mer-1A

CDS 3UTR

Luciferase activity (% control)

125

100

75

50

25

0 Psicheck

Stat3

Il13

Ifng

HEK293T CD4+T

IL-13 concentration (ng ml)

0.0

20

2.5 *

1.5 pMSCV-NC pMSCV-125a

*

* *

Agomir-125a 100nmol

Agomir-148a 100nmol

pMSCV-ctrl

pMSCV-125a

IFN- concentration (ng ml)

*

mRNA (relative)

15

2.0

1.0

1.5

10

1.0

0.5

5

0.5

Stat3

Actin

Agomir-NC 100nmol

Agomir-125a 10nmol

pMSCV-NC

0 pMSCV-125a

pMSCV-125a

pMSCV-NC

Figure 6 | Mir-125a inhibits several effector T lineage factors. (a) MiR-125a targets Stat3, Il13 and Ifng. Dual luciferase assay of HEK293T cells co-transfected with reporter vectors containing the 30UTRs of different genes and miR-125a agomir or control agomir. The renilla luciferase activity was normalized to the Firey luciferase activity. Data are shown as relative luciferase activities of cells transfected with miR-125a related to those of the control cells. (b) Comparison of miR-125a-responsive elements in human and mice transcripts encoding Stat3 (STAT3 and Stat3, respectively), Il13 (IL13 and Il13, respectively) and Ifng (IFNG and Ifng, respectively) by TargetScan. Colours in the key indicate branch lengths (and match to miR-125a). CDS, coding sequence. (c) Dual luciferase assay of HEK293T cells co-transfected the indicated WT or point-mutated 30UTR-driven reporter constructs (wt UTR or mut

UTR, respectively) with miR-125a or control agomirs. (d) Western blotting of endogenous STAT3 protein levels after transfection of HEK293T cells or infection of CD4 T cells with different doses of miR-125a agomirs or miR-125a retrovirus for 2 days. MiR-148a was used as a control. (e) Enzyme-linked immunosorbent assay examination of IFN-g and IL-13 in the supernatants of CD4 Tcells transfected with retrovirus-based miR-125a or empty vector, and differentiated under TH1- or TH2-skewing conditions for 4 days. (f) Quantitative RTPCR analysis of the mRNA levels of Stat3, Il13 and Ifng after the retrovirus-based overexpression of miR-125a in CD4 Tcells for 4 days. The relative expression levels were normalized to the expression of mouse Rpl13a.

Data are representative of four (a) or three (cf) independent experiments (means.e.m.). *Po0.05, **Po0.01 versus control (two-tailed Students t-test; a,e,f) or (one-way analysis of variance and least signicant difference t-test; c).

their related animal disease models. These ndings thus provided a very valuable candidate with potential as a biomarker of some autoimmune diseases. Furthermore, we identied miR-125a as a regulator that stabilizes the commitment and immunoregulatory capacity of Treg cells by restraining the differentiation programs of other effector lineages, consequently for the suppression of inammation. This nding identied miR-125a as a key regulator of CD4 T-cell differentiation that prevents autoimmune pathogenesis by controlling the balance between tolerance and autoimmunity.

During the early stage of T-cell differentiation, the cells potentially integrate different environment clues and ultimately allow cells to move towards committing to a particular T-cell lineage29,30. The fate decision involves the competition of various transcription factors that dictate different T-cell differentiation programs3133. Recent studies regarding Treg cells development suggests that the decision of antigen-stimulated cells to

differentiate into different T cells subtypes may depend on balancing polarizing cytokines regulated by T cells subtypespecic transcription factors32,34. It has been shown that Foxp3 and Gata3, a specic TH2 transcription factor, are able to physically associate with each other to determine T-cell differentiation to either Treg or TH2 (ref. 35). Similarly, interaction between Foxp3 and Rorc appears to reciprocally regulate the induction of Treg and TH17 cells3638. Foxp3 is able to stabilize the molecular features of the Treg lineage and inhibit effector T-cell developmental factors10,39,40. In this study, in the absence of miR-125a, Treg cells showed increased expression of TH1-, TH2- and TH17-related effector lineage genes, as well as reduced expression of Treg lineage genes such as Foxp3. As such, miR-125a appears to operate as a crucial regulator in modulating the fate decision of Treg lineage and in the maintenance of immune homeostasis. Further target screen demonstrated that miR-125a directly suppressed several targets, including Stat3, Il13

8 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096 ARTICLE

H&E Luxol fast blue

5 Ago-NC (n=6)

Ago-125a (n=6)

0 5 i.v.

Ago-NC

Ago-125a

Mean clinical score

4

**

********

30

3

**

* **** ****

*

** **

2

1

1

10

15

20

25

Day after immunization

Ago-NC

Ago-125a

Ago-NC Ago-125a

12.4

5.4

8.2

3.2

103

102

101

100

103

102

101

100

Foxp3

103

102

101

100 103

102

101

100 103

102

101

100 103

102

103

102

101

100

12.4

21.2

IL-17

54.4

27.8 70.8

17.8

101

100 103

102

101

100

CD4

IFN-

Ago-NC (n=7)

Ago-125a (n=6)

10 i.v.

4

*

Mean clinical score

***** **

*** ****

*

Foxp3+ cells (%CD4+ )

IFN-+ cells (% CD4+ )

10

IL-17+ cells (% CD4+ )

30 **

40 *

20 *

3

30

20

15

2

20

10

10

1

5

0 Ago-125a

0 Ago-125a

0

0

1

5

Ago-NC

Ago-NC

Ago-125a

Ago-NC

15

20

25

30

Day after immunization

Figure 7 | MiR-125a agomir treatment ameliorates EAE pathogenesis. (a) Mean clinical scores (s.e.m.) for EAE induced in 6- to 8-week-old WT mice after three intravenous (i.v.) injections of miR-125a agomir or control agomir (5 m mol kg 1 per day per mouse) at the onset of EAE. (b) Representative haematoxylineosin (H&E) staining or Luxol Fast Blue staining of spinal cord sections on day 21 after immunization of the mice in a. (c) Flow cytometric analysis of the percentage of Treg cells in the CD4 T cells isolated from the CNS of the mice in a on day 21 after immunization. (d) Flow cytometric analysis of the percentages of TH1 and TH17 in the CD4 Tcells isolated from the CNS of the mice (a) on day 21 after immunization. (e) Mean percentages (s.e.m.) of Foxp3 Treg cells in the CD4 population of c (n 6 per group). (f) Mean percentages (s.e.m.) of IL-17 or IFN-g T cells in the CD4

population of d (n 6 per group). (g) Mean clinical scores (s.e.m) for EAE induced in 6- to 8-week-old WTmice with three injections of miR-125a agomir

(5 m mol kg 1 per day per mouse) after acute EAE was established, at a clinical score 12. n 7 or 6 per group. The EAE symptoms were evaluated every

day. Data are representative of three (a) or two (cg) independent experiments (means.e.m.). *Po0.05, **Po0.01 (two-tailed Students t-test).

and Ifng, which are critical factors of effector lineages and also detrimental to Treg differentiation. Thus, miR-125a exerts its roles in a manner partially similar to Foxp3. This nding not only explains how sustained expression of miR-125a in Treg cells is required for restraining effector lineage programs to stabilize Treg lineage commitment and function at a molecular level but also provides insights into the mechanism of miR-125a in controlling immune homeostasis in the context of autoimmunity.

However, although deciency of miR-125a upregulates some lineage-specic transcriptional factors, cytokines and cytokine receptors in Treg cells, we did not observe obvious effects of miR-125a on TH1, TH2 and TH17 differentiation. The potential mechanism may be attributable to the following factors:(1) miR-125a is induced and highly expressed in Treg cells but not in effector T cells, so it is probably that only loss function of miR-125a in Treg cells has obvious effects; (2) miR-125a may inhibit other potential targets that have not been identied yet. miR-125a might have a combinational effect on effector T-cell differentiation, and probably happen to target some negative

regulators of TH1 and TH17 differentiation, in addition to targeting the positive regulators such as Stat3 and IFN-g. Nevertheless, although the absence of miR-125a expression was not associated with a modulation of TH2 differentiation in vitro and in vivo on normal conditions, it might be also interesting to test the inuence of miR-125a in Th2-prone models such as asthma, in addition to EAE model. Moreover, as we only observed the upregulation of the transcripts of some lineagespecic transcriptional factors, cytokines and cytokine receptors in Treg cells, it also raised a question to measure protein expression of these genes. As Treg cells usually express very low levels of inammatory cytokines at basal level, a better way is to cross miR-125a KO mice with reporter mice such as Foxp3-Red uorescent protein (RFP), IL-17-green uorescent protein (GFP) or IFN-g-yellow uorescent protein (YFP) reporter mice. In that case, the relative expression levels of Foxp3 and effector cytokines such as IL-17 and IFN-g can be easily monitored by the uorescence intensity of RFP, GFP or YFP under different conditions, such as inammatory environments.

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

Although previous studies have shown that several miRNAs, such as miR-146a (ref. 41), miR-155 (ref. 42) and miR-10a (ref. 43), regulate Treg cells development, suppressor function or homeostasis, the expression patterns of these miRNAs in different autoimmune diseases are still unclear. Our study demonstrated that miR-125a is a commonly dysregulated miRNA in peripheral CD4 T cells of multiple human autoimmune diseases and animal models. Although the exact mechanisms for the dysregulation of miR-125a in autoimmune pathogenesis are still unknown, we postulate that there might be a potential possibility involved. Previous works have shown that a single-nucleotide polymorphism is located within the precursor ( 8) of the mature

miR-125a, which signicantly blocks the processing of pri-miRNA to pre-miRNA44. Whether the genetic polymorphisms within this single locus are associated with numerous autoimmune diseases might be worth further exploring.

Despite the recent discovery of miRNAs, several candidates have already progressed into clinical/preclinical development. The most clinically advanced miRNA-based therapy at the present time is a miR-122 antagonist, SPC3649, that targets an infectious diseasehepatitis C virusand is currently being evaluated in phase 2 clinical trials4547. Other miRNAs currently in development are in the area of cardiovascular disease and include miR-208 (ref. 48), miR-195 and so on. Our work showed that miR-125a agomir not only contributes to EAE prevention but also alleviates established EAE, thus providing very valuable promise for pharmacological intervention of miR-125a in autoimmune diseases. Although the systemic injection of miR-125a agomir for consecutive three times obtained transfection efciency B58% in lymph nodes and 53% in spleen, the therapeutic effect of mir-125a agomir on EAE model was quite efcient. One possible reason is that once miR-125a agomir is overexpressed in T cells, it promotes the differentiation of Treg cells and stabilizes Treg commitment and function, therefore altering the ratio of Treg to T effector cells in vivo. Treg cells further produce large amount of cytokines such as TGF-b, which subsequently upregulates miR-125a expression in more CD4 T cells (Supplementary Fig. 13). Under this condition, mir-125a functions as an initiator and promotes the balance to shift from inammation to immune suppression through a positive feedback loop.

In summary, our study demonstrates miR-125a as a critical factor that controls autoimmune diseases through stabilizing Treg-mediated immune homeostasis by targeting effector programs. And this nding also provides insights into developing potential miRNA-based therapies in autoimmunity.

Methods

Mice. MiR-125a KO mice were generated by replacing a 630-bp-long genome DNA fragment that only encodes mmu-miR-125a with a neomycin selection cassette. The mice were created on the C57BL/6 background by Shanghai Biomodel Organism Science and Technology Development Co., Ltd. The primer sequences used for genotyping were as follows: sense primer 50-CTGAGGTCTGGATCTCT

GGC-30 and antisense primer 50-GCTCAGAAACCCAGTGTCACA-30 for WT mice; sense primer 50-CTGAGGTCTGGATCTCTGGC-30 and antisense primer 50-GCTCAGAAACCCAGTGTCACA-30 for KO mice. The mice were crossed with Flp mice to get rid of neomycin gene. Six- to 12-week-old miR-125a KO mice and their littermate controls were used for experiments. To generate miR-125a TG (miR-125 TG) mice, a 352-bp-long genome DNA fragment encoding mmu-miR-125a was cloned into the plasmid pSuper, using the forward primer: 50-

TCCCTCTTATTCTGGCTTTCTA-30 and reverse primer: 50-GAGGTCAGGT TTCAGTTGGTG-30. The mice were created on the C57BL/6 background by Shanghai Biomodel Organism Science and Technology Development Co., Ltd. The primer sequences used for genotyping were as follows: sense primer 50-GTAAAACGACGGCCAGT-30 and antisense primer 50-AACAGCTATGAC

CATG-30 on the mouse miR-125a-coding sequence. Six- to 12-week-old miR-125a KO or TG mice and their littermate controls were used for experiments. C57BL/6 WT mice were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Sciences. Rag1 / mice were purchased from Model Animal

Research Center of Nanjing University. All mice were maintained in specic

pathogen-free conditions. All animal experiments were performed in compliance with the guide for the care and use of laboratory animals, and were approvedby the institutional biomedical research ethics committee of the Shanghai Institutes for Biological Sciences (Chinese Academy of Sciences).

Subjects. A total of 34 patients (22 SLE patients and 12 CD patients) were included in the study. Thirty-four normal subjects were matched with the patients for age, sex and race. All participants are from Chinese Han population. The participants with concurrent infections were excluded from the study. For additional clinical information, refer to Supplementary Tables 1 and 2. The study was approved by the Research Ethics Board of Renji Hospital. Peripheral blood samples were collected from each subject in tubes containing acid citrate dextrose formula A.

miRNA and DNA microarray. Total RNA was isolated using TRIzol reagent (Invitrogen). RNA quality was assessed with an Agilent 2100 Bioanalyzer (Agilent), and only samples with an RNA integrity number 48 were used. For the miRNA expression analysis, a TaqMan Low Density Array v3.0 (Applied Biosystems) was used to detect and quantied up to 754 human miRNAs or 641 mouse miRNAs using an Applied Biosystems real-time instrument, according to the manufacturers protocol. Global mRNA expression in iTreg samples from miR-125a-decient and WT mice were assayed with the Affymetrix GeneChip Mouse Genome 430 2.0 Array. Data were further analysed with Agilent GeneSpring GX 10 Microarray Data Analysis Software. Array data are available in GEO under accession code GSE66608.

Reagents. The following antibodies were purchased from BD Pharmingen: anti-mouse CD4FITC (GK1.5; used at 1:100 dilution), anti-mouse CD25APC (PC61; used at 1:200 dilution), anti-mouse CD62LPE (MEL-14; used at 1:200 dilution), anti-mouse CD44PE/Cy5 (IM7; used at 1:200 dilution), anti-mouse IFN-gAPC (XMG1.2; used at 1:100 dilution) and anti-mouse IL-17PE (TC11-18H10; used at 1:100 dilution). The following antibodies were purchased from eBioscience: anti-mouse CD3e Functional Grade Puried (145-2C; nal concentration used at 2 mg ml 1), anti-mouse CD28 Functional Grade Puried (37.51; nal concentration used at 2 mg ml 1), anti-mouse/rat Foxp3PE (FJK-16S; used at 1:100 dilution) and antibody against Stat3 (sc-482; Santa Cruz; used at 1:1,000 dilution).

T-cell purication and activation. Naive CD4 CD25 CD44loCD62Lhi T cells were FACS (uorescence-activated cell sorting) sorted from the spleens of 6- to 8-week-old C57BL/6 mice. FACS-sorted cells were activated with plate-coated anti-CD3 antibody (2 mg ml 1) and soluble anti-CD28 antibody (2 mg ml 1), and then induced to differentiate into TH1 cells by supplementation with IL-12 (10 ng ml 1; Peprotech) plus anti-IL-4 antibody (11B11; BD Pharmingen), TH2 cells with IL-4 (40 ng ml 1; Peprotech) plus anti-IFN-g antibody (XMG1.2; BD

Pharmingen), TH17 differentiation, cells usually received anti-IL-4 and anti-IFN-g plus a TH17 cocktail, containing TGF-b1 (2 ng ml 1; R&D), IL-6 (30 ng ml 1;

Peprotech), tumour necrosis factor (10 ng ml 1; Peprotech), IL-1b (10 ng ml 1; Peprotech) and IL-23 (20 ng ml 1; Peprotech), inducible Treg cells with TGF-b1 (5 ng ml 1; R&D) and recombinant mouse IL-2 (50 U ml 1; Peprotech).

Neutralizing anti-IFN-g and anti-IL-4 were both used at a concentration of10 mg ml 1. Cells stimulated under neutral conditions (anti-IL-4 plus anti- IFN-g, but no additional cytokines) were considered TH0 cells.

Intracellular staining and ow cytometry. For intracellular cytokine staining, cells obtained from in vitro culture or from spleens or CNS inltrates of mice with EAE were incubated in a tissue culture incubator for 5 h at 37 C with phorbol 12-myristate 13-acetate (50 ng ml 1; Sigma), ionomycin (750 ng ml 1; Sigma) and brefeldin A (10 mg /ml 1; Sigma). Surface staining was performed as described above with the corresponding uorescence-labelled surface antibodies. After surface staining, the cells were resuspended in xation/permeabilization solution (Cytox/Cytoperm Kit; BD Pharmingen), and intracellular cytokine staining was performed according to the manufacturers protocol. For Foxp3 staining, the cells were not stimulated with phorbol 12-myristate 13-acetate and ionomycin, instead the cells were stained according to the manufacturers protocol (Foxp3 Staining Buffer Set, eBioscience). FACS Calibur (BD Biosciences) and FlowJo software were used for analysis. Sample data were acquired by FACS Calibur cytometer (BD Biosciences) and analysed using FlowJo software.

Transfer model of colitis. Naive T cells (CD4 CD25 CD44loCD62Lhi) from C57BL/6 mice were sorted by FACS and then transferred into Rag1/ mice alone or in combination with CD4 CD25hi Treg cells. After T-cell reconstitution, the mice were weighed weekly and monitored for signs of disease. Mice were killed at 6 weeks after T-cell transfer and their colons were subjected to histopathological analysis. Epithelial cell hyperplasia and leukocyte inltration were mainly evaluated.

10 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096 ARTICLE

EAE. The induction of EAE was described previously49,50. Briey, C57BL/6 mice at 68 weeks of age were induced by a subcutaneous immunization in the back region with 300 mg of the MOG peptide (3555) in Complete Freunds Adjuvant (CFA;

Difco). Pertussis toxin (List Biological Laboratories, Inc.) at a dose of 200 mg per mouse in PBS was administered intraveously on the day of immunization and once more 48 h later. The disease score was evaluated as previously described (ref. 49): 0, no clinical signs; 1, limp tail; 2, paraparesis (weakness, incomplete paralysis of one or two hind limbs); 3, paraplegia (complete paralysis of two hind limbs);4, paraplegia with fore limb weakness or paralysis and 5, moribund state or death. For histology, spinal cords were removed from MOG-immunized mice and were immediately xed in 4% paraformaldehyde. Parafn-embedded 5-mm sections of spinal cord were stained with haematoxylineosin or Luxol Fast Blue and then examined by light microscopy.

Quantitative PCR. Total RNA was extracted from the cells and mouse tissues with TRIzol reagent (Invitrogen) according to the manufacturers instructions. cDNA was synthesized with PrimeScript RT reagent kit (Takara). The expression of the genes encoding mouse Foxp3, Tgfb1, Stat3, Rorc, Il17a, Il17f, Gata3, Il4, Il5, Il13, Ifng, Il21, Il22, Il23r and so on were quantied by real-time PCR using the SYBR Premix ExTaq kit (Takara) (sequence data are available upon request). All gene expression results were normalized to the expression of the housekeeping gene Rpl13a. Amplication of cDNA was performed on an AbiPrism 7900 HT cycler (Applied Biosystems). Mature miR-125a and other miRNAs were assayed with a specic TaqMan kit from Applied Biosystems, and their expression was normalized to the expression of snoRNA202 or RNU48.

MiRNA agomir and antagomir. MiR-125a agomir and antagomir were purchased from Ruibo Biotech. MiRNA agomirs are chemically synthesized dsRNA oligo-nucleotides used to mimic miRNA. The terminal nucleotides at both ends of the dsRNAs are modied by an O-methyl moiety at the 20-ribose position. Another modication is cholesteryl functionality at the 50 end of the RNA at sense strand. MiRNA antagomirs are single-stranded oligonucleotides used to silence endogenous miRNAs. Similarly, different chemical modications are used to enhance the stability and cellular uptake, including partial phosphorothioate backbone modications, fully 20-OMe sugar modications and 30-cholesterol conjugation.

Universal negative controls for both agomirs and antagomirs are based on the sequences of cel-miR-67 in Caenorhabditis elegans, which has been conrmed to have minimal sequence identity with miRNAs in human, mouse or rat.

Luciferase assays. For the 30UTR reporter assay, psiCHECK-2 vector (Promega) was used to clone the 30UTRs of all the predicted targets of miR-125a. Site-specic mutants were generated by PCR in psiCHECK-2 vector. HEK293T cells were seeded in the wells of a 96-well plate 1 day before transfection, and then each well was transfected with a mixture of 10 ng 30UTR luciferase reporter vector and 0.5 ml of 20 mM miRNA agomirs. Twenty-four hours after transfection, the cells were lysed and the luciferase activity was measured. The ratio of renilla luciferase to rey luciferase was calculated for each well.

Immunoblots. Lysates were resolved by electrophoresis, transferred to polyvinylidene uoride (PVDF) membrane and probed with antibodies directed against Stat3 or Actin. Images have been cropped for presentation in Fig.6d. Full-size images are presented in Supplementary Fig.14

Statistics. The nonparametric MannWhitney U-test was applied for the statistical comparison of health and patient samples. A two-tailed Students t-test was applied for statistical comparison of two groups or where appropriate. The one-way analysis of variance and least signicant difference t-test was used for comparison of multiple groups. P values o0.05 were considered statistically signicant.

References

1. Gregersen, P. K. & Olsson, L. M. Recent advances in the genetics of autoimmune disease. Annu. Rev. Immunol. 27, 363391 (2009).

2. Baumgarth, N. & Bevins, C. L. Autoimmune disease: skin deep but complex. Nature 449, 551553 (2007).

3. Nanda, S. Genetic variants suggest shared pathwaysto autoimmune disease. Nat. Rev. Rheumatol. 6, 675 (2010).

4. Littman, D. R. & Rudensky, A. Y. Th17 and regulatory T cells in mediating and restraining inammation. Cell 140, 845858 (2010).

5. Haas, J. et al. Prevalence of newly generated naive regulatory T cells (Treg) is critical for Treg suppressive function and determines Treg dysfunction in multiple sclerosis. J. Immunol. 179, 13221330 (2007).

6. Xavier, R. J. & Podolsky, D. K. Unravelling the pathogenesis of inammatory bowel disease. Nature 448, 427434 (2007).

7. Brusko, T. & Atkinson, M. Treg in type 1 diabetes. Cell Biochem. Biophys. 48, 165175 (2007).

8. Zheng, Y. et al. Genome-wide analysis of Foxp3 target genes in developing and mature regulatory T cells. Nature 445, 936940 (2007).

9. Josefowicz, S. Z. & Rudensky, A. Control of regulatory T cell lineage commitment and maintenance. Immunity 30, 616625 (2009).

10. Samstein, R. M. et al. Foxp3 exploits a pre-existent enhancer landscape for regulatory T cell lineage specication. Cell 151, 153166 (2012).

11. Mendell, J. T. & Olson, E. N. MicroRNAs in stress signaling and human disease. Cell 148, 11721187.

12. van Rooij, E., Purcell, A. L. & Levin, A. A. Developing microRNA therapeutics. Circ. Res. 110, 496507 (2012).

13. Jackson, A. L. & Levin, A. A. Developing microRNA therapeutics: approaching the unique complexities. Nucleic Acid Ther. 22, 213225 (2012).

14. Jernas, M. et al. MicroRNA regulate immune pathways in T-cells in multiple sclerosis (MS). BMC Immunol. 14, 32 (2013).

15. Cobb, B. S. et al. A role for Dicer in immune regulation. J. Exp. Med. 203,

25192527 (2006).

16. Lock, C. et al. Gene-microarray analysis of multiple sclerosis lesions yields new targets validated in autoimmune encephalomyelitis. Nat. Med. 8, 500508 (2002).

17. Laurence, A. et al. STAT3 transcription factor promotes instability of nTreg cells and limits generation of iTreg cells during acute murine graft-versus-host disease. Immunity 37, 209222 (2012).

18. Xu, L. L. et al. Positive and negative transcriptional regulation of the Foxp3 gene is mediated by access and binding of the Smad3 protein to enhancer I. Immunity 33, 313325 (2010).

19. Chang, J. H., Kim, Y. J., Han, S. H. & Kang, C. Y. IFN-gamma-STAT1 signal regulates the differentiation of inducible Treg: potential role for ROS-mediated apoptosis. Eur. J. Immunol. 39, 12411251 (2009).

20. Wei, J. et al. Antagonistic nature of T helper 1/2 developmental programs in opposing peripheral induction of Foxp3 regulatory T cells. Proc. Natl Acad.

Sci. USA 104, 1816918174 (2007).21. Kota, J. et al. Therapeutic microRNA delivery suppresses tumorigenesis in a murine liver cancer model. Cell 137, 10051017 (2009).

22. Lanford, R. E. et al. Therapeutic silencing of microRNA-122 in primates with chronic hepatitis C virus infection. Science 327, 198201 (2010).

23. Krutzfeldt, J. et al. Silencing of microRNAs in vivo with antagomirs. Nature 438, 685689 (2005).

24. Hou, J. et al. Identication of miRNomes in human liver and hepatocellular carcinoma reveals miR-199a/b-3p as therapeutic target for hepatocellular carcinoma. Cancer Cell 19, 232243 (2011).

25. Weber, M. S. et al. Type II monocytes modulate T cell-mediated central nervous system autoimmune disease. Nat. Med. 13, 935943 (2007).

26. Youssef, S. et al. The HMG-CoA reductase inhibitor, atorvastatin, promotes a Th2 bias and reverses paralysis in central nervous system autoimmune disease. Nature 420, 7884 (2002).

27. Zhao, X. et al. MicroRNA-125a contributes to elevated inammatory chemokine RANTES levels via targeting KLF13 in systemic lupus erythematosus. Arthritis Rheum. 62, 34253435 (2010).

28. Hu, J. Y. et al. Increasing CCL5/CCR5 on CD4 T cells in peripheral blood of

oral lichen planus. Cytokine 62, 141145 (2013).29. Maldonado, R. A., Irvine, D. J., Schreiber, R. & Glimcher, L. H. A role for the immunological synapse in lineage commitment of CD4 lymphocytes. Nature 431, 527532 (2004).

30. OShea, J. J. & Paul, W. E. Mechanisms underlying lineage commitment and plasticity of helper CD4 T cells. Science 327, 10981102 (2010).

31. OShea, J. J. & Paul, W. E. Mechanisms underlying lineage commitment and plasticity of helper CD4( ) T cells. Science 327, 10981102 (2010).

32. Vahedi, G. et al. STATs shape the active enhancer landscape of T cell populations. Cell 151, 981993 (2012).

33. Nakayamada, S., Takahashi, H., Kanno, Y. & OShea, J. J. Helper T cell diversity and plasticity. Curr. Opin. Immunol. 24, 297302 (2012).

34. Wan, Y. Y. & Flavell, R. A. Regulatory T-cell functions are subverted and converted owing to attenuated Foxp3 expression. Nature 445, 766770 (2007).

35. Dardalhon, V. et al. IL-4 inhibits TGF-beta-induced Foxp3 T cells and,

together with TGF-beta, generates IL-9 IL 10 Foxp3( ) effector T cells.

Nat. Immunol. 9, 13471355 (2008).36. Bettelli, E. et al. Reciprocal developmental pathways for the generationof pathogenic effector TH17 and regulatory T cells. Nature 441, 235238 (2006).

37. Zhou, L. et al. TGF-beta-induced Foxp3 inhibits T(H)17 cell differentiation by antagonizing RORgammat function. Nature 453, 236240 (2008).

38. Ichiyama, K. et al. Foxp3 inhibits RORgammat-mediated IL-17A mRNA transcription through direct interaction with RORgammat. J. Biol. Chem. 283, 1700317008 (2008).

39. Rudra, D. et al. Transcription factor Foxp3 and its protein partners form a complex regulatory network. Nat. Immunol. 13, 10101019 (2012).

40. Rudensky, A. Y. Regulatory T cells and Foxp3. Immunol. Rev. 241, 260268 (2011).

NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8096

41. Lu, L. F. et al. Function of miR-146a in controlling Treg cell-mediated regulation of Th1 responses. Cell 142, 914929 (2010).

42. Lu, L. F. et al. Foxp3-dependent microRNA155 confers competitive tness to regulatory T cells by targeting SOCS1 protein. Immunity 30, 8091 (2009).

43. Takahashi, H. et al. TGF-beta and retinoic acid induce the microRNA miR-10a, which targets Bcl-6 and constrains the plasticity of helper T cells. Nat. Immunol. 13, 587595 (2012).

44. Duan, R., Pak, C. & Jin, P. Single nucleotide polymorphism associated with mature miR-125a alters the processing of pri-miRNA. Hum. Mol. Genet. 16, 11241131 (2007).

45. Jopling, C. L., Yi, M., Lancaster, A. M., Lemon, S. M. & Sarnow, P. Modulation of hepatitis C virus RNA abundance by a liver-specic MicroRNA. Science 309, 15771581 (2005).

46. Gebert, L. F. R. et al. Miravirsen (SPC3649) can inhibit the biogenesis of miR-122. Nucleic Acids Res. 42, 609621 (2014).

47. Janssen, H. L. et al. Treatment of HCV infection by targeting microRNA.N. Engl. J. Med. 368, 16851694 (2013).48. van Rooij, E. et al. Control of stress-dependent cardiac growth and gene expression by a microRNA. Science 316, 575579 (2007).

49. Zhu, S. et al. Modulation of experimental autoimmune encephalomyelitis through TRAF3-mediated suppression of interleukin 17 receptor signaling.J. Exp. Med. 207, 26472662.50. Zhu, S. et al. The microRNA miR-23b suppresses IL-17-associated autoimmune inammation by targeting TAB2, TAB3 and IKK-alpha. Nat. Med. 18, 10771086 (2012).

Acknowledgements

We thank Robert Georgantas and Philip Brohawn for their helpful discussions, technical

expertise and/or review of this manuscript. This work is supported by grants from 973

programme (2014CB541902 and 2014CB541901), National Natural Science Foundation

of China (81230072, 81421001, 81025016 and 31370880), the Key Research Program of

the Chinese Academy of Sciences (KSZD-EW-Z-003-3), the grants from the State Key

Laboratory of Oncogenes and Related Genes (No.91-14-05) as well as Chinese Ministry

of Health (201202008) and the Program of the Shanghai Commission of Science and

Technology (12431900703, 12JC1406000 and 12ZR1435900). W.P. is supported by a

fellowship from Leukemia and Lymphoma Society. S.Z. is supported by a fellowship from

Howard Hughes Medical InstituteThe Helen Hay Whitney Foundation and Shenzhen

Foundation of Science and Technology (grant number GJHZ2014041417082192).

Author contributions

W.P. and N.S designed the research; N.S. and Y.Y supervised the research; W.P., S.Z. and

D.D. conducted the experiments; W.P. wrote the manuscript; N.S, Y.Y and Z.L. revised

the manuscript; Y.C. and J.F provided clinical samples; D.L., B.L., Y.Z., Y.T., M.T.W.,

W.Z., Y.W., B.C. and Y.Q. helped the experiments; J.B.H., R.A.F., N.G., R.H., L.R. and

B.J. commented on the research.

Additional information

Accession codes: Array data are available in GEO under accession code GSE66608. 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 nancial 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: Pan, W. et al. MiR-125a targets effector programs to stabilize

Treg-mediated immune homeostasis. Nat. Commun. 6:7096 doi: 10.1038/ncomms8096

(2015).

12 NATURE COMMUNICATIONS | 6:7096 | DOI: 10.1038/ncomms8096 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.