ITK signalling via the Ras/IRF4 pathway regulates the development and function of Tr1 cells

Type 1 regulatory T (Tr1) cells differentiate in response to signals engaging the T cell receptor (TCR), express high levels of the immunosuppressive cytokine IL-10, but not Foxp3, and can suppress inflammation and promote immune tolerance. Here we show that ITK, an important modulator of TCR signalling, is required for the TCR-induced development of Tr1 cells in various organs, and in the mucosal system during parasitic and viral infections. ITK kinase activity is required for mouse and human Tr1 cell differentiation. Tr1 cell development and suppressive function of Itk deficient cells can be restored by the expression of the transcription factor interferon regulatory factor 4 (IRF4). Downstream of ITK, Ras activity is responsible for Tr1 cell induction, as expression of constitutively active HRas rescues IRF4 expression and Tr1 cell differentiation in Itk−/− cells. We conclude that TCR/ITK signalling through the Ras/IRF4 pathway is required for functional development of Tr1 cells. IL-10+Tr1 cells, which do not express stable Foxp3 like conventional Treg cells, are associated with various diseases, but little is known about how their differentiation and function are regulated. Here the authors show that ITK is a central TCR signalling effector of the differentiation and function of Tr1 cells.

T regulatory (Treg) cells promote immune tolerance and suppress inflammation 1,2 . Unlike Treg cells that stably express the transcription factor Foxp3, type 1 regulatory T (Tr1) cells have no or transient expression of Foxp3; however, they produce high levels of IL-10 and can suppress effector cell responses in an IL-10 dependent manner 1,3 , via CTLA-4 and PD-1 interactions, or by directly killing pro-inflammatory cells with granzymes 2,4 . In mice and in humans, induction of antigenic tolerance during haematopoietic stem cell transplantation and specific-antigen immunotherapy are positively correlated with the abundance of Tr1 cells 5,6 , and Tr1 cells can prevent allergic asthma induced by the house dust mite peptidase 1 variant Derp 1 in murine models 7 , and prevent the development of bacterialinduced atopic dermatitis 8 . Thus, like Foxp3 þ Treg cells, Foxp3 À IL-10 þ Tr1 cells have therapeutic potential for inflammatory diseases. Although much is known about the development and function of Treg cells, substantially less is known about Tr1 cells. A better understanding of the development and function of Tr1 cells should provide a wider array of therapeutic options for inflammatory diseases.
IL-2 inducible T cell kinase (ITK) is a Tec family non-receptor tyrosine kinase expressed by T cells, and has a pivotal role downstream of the T cell receptor (TCR); the loss of ITK function leads to attenuated TCR signalling and alters the T cell subset differentiation and function 9 . Naive CD4 þ T cells can differentiate into Tr1 cells upon TCR engagement in the presence of IL-27, and although Tr1 cells can express IFN-g, production of IFN-g or T-bet are not required for Tr1 cell development 10 . Alternatively, Tr1 cells can result from Th17 trans-differentiation during the resolution of inflammation 11 . These findings suggest that Tr1 cell differentiation may share some pathways of regulation with Th1 and Th17 cell development. In mice with ITK deficiency, naive CD4 þ T cells have defects in the differentiation of Th17 cells 12 , and enhanced Th1 differentiation with impaired Th2 and Th9 programming that leads to attenuated allergic asthma [13][14][15] , and have enhanced differentiation of Foxp3 þ Treg cells 16,17 . Whether ITK also has a function in modulating the development and/or function of IL-10-producing Tr1 cells, is unexplored.
Beyond the finding that the cytokine IL-27 and the transcription factors interferon regulatory factor 4 (IRF4), avian musculoaponeurotic fibrosarcoma (cMAF) and aryl hydrocarbon receptor (AHR) are important for Tr1 cell differentiation, we have limited knowledge of the signalling pathways that regulate the development and, importantly, function of Tr1 cells. Here we show that, in the absence of ITK, TCR engagement does not induce optimal differentiation of Tr1 cells in multiple organs and during parasitic or viral infection. The expression and activity of ITK are crucial for Tr1 cell fate programming in both mouse and human, and for Tr1 cell function to suppress effector cell expansion. ITK deficiency impairs IRF4 expression in both mouse and human Tr1 cell development, and restoring IRF4 expression rescues Tr1 cell fate programming and suppressive function in Itk deficient cells. The RAS/MAPK signalling axis is indispensable for Tr1 cell development, and constitutively active RAS signalling completely rescues induction of IL-10 and IRF4 during Tr1 cell differentiation of Itk deficient cells. Our findings identify ITK as a crucial component that bridges extracellular signals, RAS signalling and IRF4 expression during Tr1 cell fate programming, and suggest that ITK signalling components are potential targets for modulating Tr1 cell development and function for clinical benefit.

Results
ITK is required for Tr1 cell development in vivo. We and others have previously shown ITK as a negative regulator of Foxp3 þ Treg cell development 16,17 , but its role in CD4 þ Foxp3 À IL-10 þ Tr1 cell development is unclear. To determine the role of ITK in Tr1 cell differentiation in vivo, we injected WT and Itk À / À IL-10 GFP /Foxp3 RFP dual reporter mice with an anti-CD3e antibody that has been shown to stimulate pronounced Tr1 cell development through TCR activation in vivo 3 . We observed a significant defect in IL-10 þ Foxp3 À LAG3 þ CD49b þ Tr1 cell differentiation systemically in blood, spleen, lung, small intestine and fat, in mice lacking the expression of ITK (Fig. 1), suggesting that ITK is required for Tr1 cell development in vivo. IL-10 production can be significantly elevated in the pulmonary mucosa during the late stages of Nippostrongylus brasiliensis 18 and influenza A 19 infection to prevent tissue damage. To determine whether ITK is required for Tr1 cell development during these infections, we challenged mice with N. brasiliensis larvae or influenza A (WSN) virus, and found that ITK is required for Tr1 cell differentiation during parasitic (Fig. 2a,b) and viral (Fig. 2c,d) infections. Our data support a requisite role for ITK in Tr1 cell development in vivo.
ITK is required for Tr1 cell fate programming in vitro. To determine whether ITK regulates Tr1 cell development in vitro, we isolated naive splenic CD4 þ T cells from WT and Itk À / À mice carrying IL-10 GFP /Foxp3 RFP dual reporters, and cultured them under Tr1-polarizing condition in vitro. We found that in the absence of ITK, differentiation of IL-10 þ Foxp3 À Tr1 cells is severely impaired, which persisted along the time course of our observation (Fig. 3). A similar defect in Tr1 cell differentiation was also observed using naive T cells isolated from the thymus ( Supplementary Fig. 1). ITK is not required for early activation of CD4 þ T cells under Tr1-differentiation condition, as Itk À / À CD4 þ T cells effectively up-regulated early activation markers CD25 and CD69 within 48 h post stimulation (Fig. 3c), and all cells proliferated within 72 h post stimulation (Fig. 3d). However, in the absence of ITK, cells that proliferated failed in the induction of IL-10 production (Fig. 3d, bottom panel), and are impaired in expression of the Tr1-related markers LAG3, CD49b, ICOS and PD-1 (Fig. 3e,f). These data are suggestive of a requisite role for ITK in Tr1 cell fate programming that is not explained by a lack of T cell activation.
ITK kinase activity is required for Tr1 cell differentiation. We have recently shown that the absence of ITK in T cells can have different functional outcomes compared with inhibition of its kinase activity 20 . Furthermore, the ITK inhibitors that have been developed to date all exhibit varying degrees of cross reactivity with other tyrosine kinases (see review ref. 21). To get around these caveats, we developed a unique transgenic mouse system that expresses an altered form of ITK, referred to as allele sensitive ITK (or ITK as ) that allows us to inhibit the kinase activity of ITK using the small molecule 3MBPP1 (refs 16,20). To definitively and unambiguously determine the consequence of inhibiting the kinase activity of ITK during Tr1 cell development, we generated Itk as IL-10 GFP /Foxp3 RFP dual reporter mice. Using naive CD4 þ cells from these mice, we found that specifically inhibiting the kinase activity of ITK resulted in similar defect in Tr1 cell differentiation to that seen in the absence of ITK expression (Fig. 4a). They also exhibited severely impaired expression of classically defined surface markers of Tr1 cells, LAG3 and CD49b 1 (Fig. 4b). Such defective Tr1 cell differentiation is unlikely because of the lack of TCR-mediated cell survival and proliferation but rather impaired Tr1 cell fate programming, because CD4 þ T cells survival and proliferation remained similar in the absence of ITK expression or its kinase activity, compared with the case for WT T cells (Fig. 4b). We also observed that the absence of ITK or its kinase activity resulted in a    Figure 1 | ITK is required for Tr1 cell development in vivo. WT and Itk À / À IL-10 GFP /Foxp3 RFP dual reporter mice were treated with aCD3e antibody or control, and live cells from the indicated organs were analysed. (a) Representative FACS plots showing IL-10 GFP expression by Foxp3 RFP À CD4 þ T cells and summary of IL-10 þ cell percentage over Foxp3 À CD4 þ T cells and number in blood (number per ml), spleen, lung, small intestine and fat (number per gram). (b) Representative FACS plots showing the LAG3/CD49b expression by IL-10 GFP þ Foxp3 RFP À CD4 þ T cells, summary of LAG3 þ CD49b þ cell percentage over IL-10 þ Foxp3 À CD4 þ T cells and number of LAG3 þ CD49b þ IL-10 þ Foxp3 À Tr1 cells from samples shown in (a). y Pr0.05, yy Pr0.01, yyy Pr0.001, compared with the levels in PBS-treated group; *Pr0.05, **Pr0.01, ***Pr0.001, comparing groups connected, by non-parametric significant shift in the production of IFN-g instead of IL-10 under Tr1 polarizing conditions, although this was not induced by TCR re-activation ( Supplementary Fig. 2), suggestive of altered commitment to Tr1 cell fate decision during primary stimulation. In human cells, inhibiting ITK kinase activity using the selective ITK inhibitor BMS-509744 (ref. 22) or the broad Tec kinase inhibitor CNX584 (ref. 23) revealed that the kinase activity of ITK is required for IL-10 production and human Tr1 cell differentiation as well (Fig. 4c,d), suggesting a conserved requisite role for ITK activity in Tr1 cell fate programming in human and mouse.
ITK kinase activity required for Tr1 cell suppressive function. Tr1 cells can suppress effector T cell function in an IL-10dependent manner 1,3 , but we have little knowledge of how this function is regulated. Because of the lack of methods to isolate viable IL-10 producers and specially target them in a co-culture system in the past, the main efforts in the field have been devoted to Tr1 cell development as an end point. In our experimental platform, the use of ITK as and its specific inhibitor 3MBPP1 would allow us to specifically target ITK kinase activity in Itk as Tr1 cells when they are co-cultured with responding effector cells. Using FACS sorted IL-10-producing Itk as Tr1 cells, we were able to examine the role of ITK activity in Tr1 cell suppressive function. We found that Itk as Tr1 cells are able to suppress effector T cell expansion as efficiently as WT Tr1 cells, and importantly, that inhibiting the kinase activity of ITK in the Itk as Tr1 cells by 3MBPP1 significantly diminished their ability to suppress (Fig. 5a,b), accompanied by a reduction in IL-10 production (Fig. 5c). These data indicate that ITK activity is required for functional suppression by Tr1 cells. AHR agonist (or antagonist) was not able to rescue Tr1 differentiation in the absence of ITK (Fig. 6b). cMAF in the other hand, showed an interesting discrepancy in mouse versus human CD4 þ T cells during Tr1 cell differentiation (Fig. 6c,d), which may be due to the difference in mouse and human Tr1 cell polarizing conditions used. Nevertheless, given that ITK kinase activity is required for both human and mouse Tr1 cell development (Fig. 4), it is unlikely that defective cMAF expression down-stream of ITK is responsible for the defect of Tr1 cell development specifically in the mouse. We did however, observe a significant reduction in IRF4 expression when ITK is absent or inhibited in both mouse and human during Tr1 cell development (Fig. 6e,f). In CD4 þ T cells, IRF4 up-regulates IL-4 and IL-10 expression under Th2 polarizing condition 25 ; and IRF4 expression is reduced during activation of Itk À / À CD8 þ T cells 26 and Th9 cell differentiation 15 . However, it is unclear whether IRF4 expression alone is responsible for the Tr1 deficient phenotype we observe in the absence of ITK. Thus to determine whether IRF4 expression is down-stream of ITK signalling during Tr1 cell differentiation, we transduced naive CD4 þ T cells with retroviral particles delivering IRF4-YFP (or YFP alone as control). The transduction restored  IRF4 expression in CD4 þ T cells cultured under Tr1-polarizing condition ( Supplementary Fig. 3a,b) and we found that the reexpression of IRF4 in Itk deficient cells rescued Tr1 cell development (Fig. 7a,b), suggesting that ITK signalling-driven IRF4 expression is essential for Tr1 cell differentiation. To further determine whether re-expression of IRF4 rescued the function of Tr1 cells as well, we isolated cells that were transduced with IRF4-RV YFP þ , along with controls, and performed the in vitro Tr1 suppression assay. Our results showed that Itk À / À Tr1 cells that differentiated upon re-expression of IRF4 were fully functional in suppressing responding T cell proliferation (Fig. 7c). Thus, downstream of ITK, IRF4 is required for Tr1 cell functional developmental.
ITK induced Blimp-1 does not drive Tr1 cell differentiation.
We have also noted that when IRF4 was re-expressed in Itk À / À T cells, we always observed partial rescue of Tr1 cell differentiation ( Fig. 7), and so we sought to find additional regulators that might function downstream of ITK in Tr1 cell development. Blimp-1 has been shown to function jointly with IRF4 in regulating IL-10 production in Foxp3 þ regulatory T cells 27 , and is critical for IL-10 production in CD4 þ T helper effector cells 28 as well as antiviral cytotoxic CD8 þ T lymphocytes 29 . In the absence of ITK, CD4 þ T cells cultured under Tr1-polarizing condition exhibited significantly impaired induction of Blimp-1 expression, compared with WT cells (Fig. 8a,b). To determine whether the lack of Blimp-1 expression contributes to the impaired Tr1 cell differentiation in Itk À / À cells, we re-expressed Blimp-1 using retroviral vectors that co-express the human CD2 marker 30 (to identify Blimp-1-expressing cells), and found that over expression of Blimp-1 (Fig. 8c,d) (Fig. 9). While these inhibitors do not grossly impair T cell activation, they significantly reduced the expression of IRF4 during Tr1 differentiation (Fig. 9). Retroviral transduction-mediated expression of WT HRas can rescue Tr1 cell development and level of IL-10 expression in the absence of ITK, which is further enhanced by the expression of  the constitutively active HRas G12V mutant (Fig. 10a,b). Note that the expression of Ras also rescues the level of IRF4 expression in the Itk deficient cells, suggestive of a functional rescue (Fig. 10c).

Discussion
Our data presented in this report support a requisite role for ITK in Tr1 cell development in both mouse and human. Most importantly, the activity of ITK is required for the suppressive function of Tr1 cells. This function is executed through the kinase activity of ITK, and its down-stream signals through HRas activation and IRF4 expression. Although both Tr1 cells and Foxp3 þ regulatory T cells can produce IL-10 and can suppress T cell proliferation, the role of ITK is different. We and others have reported that ITK, and its kinase activity, suppresses the development of Foxp3 þ regulatory T cells 16,17 . However, our work here indicates that the requirement for ITK in Tr1 cells is more analogous to its role in T effector cells such as Th2, Th9 and Th17 cells 12,15,34 . While the signals downstream of ITK in inducing Th1 differentiation remains unclear, ITK regulation of the transcription factor NFAT has been shown to be critical for its ability to regulate naive cell differentiation to Th2 and Th17 cell fates 12,34 . Furthermore, Schwartzberg and colleagues recently reported that ITK signals via IRF4 to regulate naive cell differentiation to Th9 cell fate and production of IL-9 (ref. 15). Notably, in the absence of ITK, Th9 cell differentiation is fully rescued by the presence of IL-2, production of which is also defective in the absence of ITK. These previous findings suggest that IL-2 may play a role in the function of ITK in Tr1 cell differentiation. Indeed, addition of exogenous IL-2 is able to partially rescue Tr1 cell differentiation in the absence of ITK, although addition of exogenous TGF-b is unable to do so (Supplementary Fig. 4).   Figure 7 | Re-expression of IRF4 rescues Tr1 cell differentiation and function in Itk À / À cells. WT and Itk À / À naive IL-10 GFP /Foxp3 RFP reporter CD4 þ cells were cultured under Tr1 differentiation conditions, and retrovirally transduced with control-YFP or IRF4-YFP. YFP þ (IRF4 þ ) or YFP -(IRF4 -) Foxp3 RFP À CD4 þ T cells were gated for analysis: representative FACS plots of (a) IL-10 GFP expression in Foxp3 À CD4 þ T cells and summary of percentage of IL-10 þ cells over live Foxp3 À CD4 þ T cells; (b) LAG3 and CD49b expression and summary of LAG3 þ CD49b þ Tr1 cells over live Foxp3 À CD4 þ T cells. n ¼ 6. Data represent results of three experiments. (c) WT and Itk À / À Tr1 cells (CD45.2 þ , carrying IL-10 GFP /Foxp3 RFP reporters) with or without IRF4-RV were flow sorted and co-cultured with CFSE stained naive CD4 þ T cell responders (CD45.1 þ ). Representative FACS plots of CFSE dilution of the responders with or without stimulation in the presence or absence of WT or Itk À / À Tr1 cells that are IRF4-RV À or IRF4-RV þ (Tr1: responder cell ratio ¼ 1:2); and summary of percentage of suppression of the responder proliferation by the indicated Tr1 cells. n ¼ 6. *Pr0.05, ***Pr0.001, by one-way ANOVA with Tukey's post-hoc test. Data presented as mean±s.e.m.
We found that the absence of ITK or its kinase activity resulted in defects in expression of the transcription factor AHR, cMAF and IRF4 in T cells activated under Tr1 cell-differentiation conditions. These three transcription factors have all been shown to be important in Tr1 cell differentiation and IL-10 production in T cells 24,25,27,35 . However, an AHR ligand is unable to rescue the Tr1 cell differentiation defects in the absence of ITK, and cMAF expression is not affected in human T cells when ITK kinase activity is inhibited. Our findings that re-expression of IRF4 rescues both Tr1 cell differentiation and function, suggests that IRF4 plays an important role in this process downstream of ITK. AHR can bind to cMAF promoter and regulates cMAF expression during Tr1 cell differentiation 24 . We have also found that re-expression of IRF4 rescued AHR expression in Itk À / À CD4 þ T cell under Tr1 cell-differentiating condition ( Supplementary Fig. 3c,d). Our data suggest that IRF4 is sufficient to compensate for a lack of ITK as re-expression of IRF4 rescues Tr1 cell differentiation and function. Notably, IRF4 has been reported to lie upstream of Blimp-1, and together both are required for IL-10 production in Foxp3 þ regulatory T cells 27 . However, despite the requisite role of ITK signals for the expression of Blimp-1 in Tr1 cells, re-expression of Blimp-1 in ITK deficient cells does not rescue Tr1 cell differentiation. This suggests that IRF4-regulated genes are critical for this process, and the process cannot be rescued by Blimp-1 alone in the absence of ITK.
We found that inhibition of Ras, and the downstream MAP kinase pathways, particularly ERK and JNK, reduced both Tr1 differentiation and the expression of IRF4. Furthermore, overexpression of HRas, or constitutively active HRas, in Itk À / À T cells also rescued Tr1 cell differentiation, along with IRF4 expression. Given that the re-expression of IRF4 in the absence of ITK rescues Tr1 function, we infer that the HRas pathway, by rescuing IRF4, is able to rescue Tr1 cell function as well. The TCR pathways leading to IRF4 are unclear, and our data shows for the first time that downstream of the TCR and ITK, the Ras/MAPK pathways leads to IRF4 expression during Tr1 cell differentiation.
In addition to conventional differentiation of Tr1 cells from naive CD4 þ progenitors, it has recently been shown that Tr1 cells can be derived from cells that had once been Th17 cells but then lost IL-17 production (exTh17 cells) during the resolution of inflammation 11 . Our unique Itk as mice allowed us to be able to generate Th17 cells, where ITK could be specifically targeted by 3MBPP1. We found that similar to the role for ITK in differentiation of Tr1 cells, the kinase activity of ITK is also required for optimal trans-differentiation of exTh17 cell into Tr1  Fig. 5b). However, given that ITK kinase activity is also required for Th17 cell development, survival, proliferation and maintenance of RORgt expression under Tr1polarizing condition ( Supplementary Fig. 5a,c), the effect of inhibiting ITK on down-regulation of IRF4 expression and inhibition of IL-10 production might be the result of a combination of cell survival, proliferation and cell fate reprogramming. Whether ITK functions in trans-differentiation of exTh17 cell into Tr1 cells in vivo remains to be further explored.

cells (Supplementary
Comparing our results in the mouse and human T cells, we found that in human cells, two different ITK inhibitors exhibit differential effects inhibiting the level of IL-10, cell surface markers LAG3/CD49b and transcription factor IRF4 expression. This may be because of the different characteristics and IC 50 s of these compounds, as well as potentially different off target effects. Note that our work with the 3MBPP1 on the murine cells expressing the Itk as allele is much more indicative of what would happen when we only inhibited ITK activity, however, we are not able to use this inhibitor in the human T cells for obvious reasons. Furthermore, it is recently reported that IL-10producing T cells derived from human CD4 þ memory T cells exhibit low levels of LAG3/CD49b expression 36 , in contrast to what was observed in human Tr1 cells derived from naive human CD4 þ T cells 1 . Thus, the effect of inhibiting ITK in human IL-10-producing CD4 þ T cells needs to be further explored to consider those derived from naive precursors versus memory cells. In our experiments, we also examined the responses of Itk À / À mice to infection with N. brasiliensis and Flu. While the response of Itk À / À mice to infection with N. brasiliensis has been previously reported 34 , this work was performed before the discovery of Tr1 cells, and the response of these mice to infection with Flu have not been reported. In both infection models, we found that the Tr1 cell response was quite organ specific, with significant Tr1 cell response found largely in the lungs and to a lesser extent in the mesenteric/draining lymph nodes in both models, with little response in the spleen (Supplementary Fig. 6). Furthermore, while infection with N. brasiliensis led to increased recovery of parasites in the absence of ITK ( Supplementary Fig. 6), there was no difference in the recovery of virus from the lungs of Flu infected mice, and no difference in body weights, although there was reduced survival of Itk À / À mice (Supplementary Fig. 6). The complex immunological phenotype of the Itk deficiency beyond our analysis of Tr1 cell responses in vivo, with multiple cell types exhibiting different phenotypes [12][13][14][15][16][17]23,26,32,34,[37][38][39][40][41][42][43][44][45][46] , makes it more challenging to provide an explanation for these phenotypes without significant additional experiments.
Regulatory T cell function can be a double-edge sword, beneficial in controlling inflammation and promoting immune tolerance, but detrimental in causing immune depression under conditions of infections or in the face or tumours. Tr1 cells have been found to be abundant in chronic hepatitis C virus 47 and Mycobacterium tuberculosis 48 infection, as well as in tumour microenvironments 49,50 , which may lead to a lack of or reduced ARTICLE immune response to clearing viruses, bacteria or tumours. Given the functional requirement for ITK kinase activity in differentiated Tr1 cells, targeting ITK activity and its signalling pathway specifically in Tr1 cells may be a promising strategy in modulating immune suppression due to the functions of Tr1 cells under such circumstances, although further work in humans will have to be done to determine this. These findings place ITK and its signalling axis as promising therapeutic targets to modulate Tr1 cell numbers and activity for treating diseases such as inflammation, chronic infection and tumours.
N. brasiliensis and Influenza A/WSN/1933 (WSN) infection. Mice were given 500 L3 N. brasiliensis larvae per mouse through subcutaneous injection, or 1 LD 50 (10 4 PFU) WSN per mouse through intranasal treatment. Cells from the spleen, lymph nodes and lungs were analyzed at the indicated time points. Parasite burden in mice infected with N. brasiliensis was determined on 7 days post infection (7 d.p.i.) by opening the small intestines longitudinally under an optical microscope. Weight of mice infected with WSN was collected every 24 h, and mice that lost 420% of the original weight were killed and recorded as dead. Plaque-forming units (PFU) of WSN in lungs of mice infected with WSN were determined by plaque assay using MDBK cells 60 . Lungs were sectioned and stained with Periodic Acid-Schiff (PAS) by the Animal Health Diagnostic Center at Cornell University.
In vitro suppression assay. CD45.1 þ CD45.2naive responder cells were labeled with 5 mM CFSE (Thermo Fisher Scientific), and co-cultured with Mitomycin-C treated APCs, in the presence or absence of anti-CD3e (0.5-1 mg ml -1 ). WT and Itkas Tr1 cells (CD45.2 þ ) were differentiated in vitro, sort purified (4 98% IL-10 GFP þ Foxp3 RFP À ), and used as suppressors in the indicated ratios to the naive responders (CD45.1 þ ). 3MBPP1 (1-2 mM) was added to specifically inhibit the kinase activity of ITK in Itk as Tr1 cells, without affecting the kinase activity of ITK in responder T cells. Division of responder cells was analysed 3 days later, and Division Index (DI) was calculated using the FlowJo (Tree Star) 'Proliferation Platform', in which DI is defined as the ratio of 'the total number of divisions/the number of cells at start of culture'. Then the average of DI of responder cells with anti-CD3 stimulation but no Tr1 cell suppression was termed DI 0 , while the DI of responder cells under condition i was termed as DI i . The percentage of suppression (PS) under condition i was defined as 'PS i ¼ 100 (1-DI i /DI 0 )' 61 .
To generate pGC-HRas-IRES-YFP-RV and pGC-HRas G12V -IRES-YFP-RV, the HRas and HRas G12V cDNA (kind gift from Dr R. Cerione, Cornell University) were cloned into the pGC-IRES-YFP-RV via the 5 0 Not I and 3 0 Xho I restriction site ligation respectively, and verified by DNA sequencing. The pMX-IRES-human CD2 (hCD2) RV and pMX-Blimp1-IRES-hCD2 RV are as previously described 30 (kind gifts from Dr T. Malek, University of Miami). HEK 293FT cells were cotransfected with the retroviral vector and the packaging vector. Retroviral supernatants were harvested 48 and 72 h post transfection, concentrated using Retro-X concentrator (Clontech, Takara Bio, Mountain View, CA), and used to infect naive CD4 þ T cells that had been stimulated under Tr1 cell differentiating conditions overnight. 48 h after infection, CD4 þ T cells that expressed the protein encoded by the retroviral vector (YFP þ or hCD2 þ ) were analysed for Tr1 cell differentiation and function.
Flow cytometry. Surface staining of live cells were done in the presence of Fc block and fixable viability dye. For intracellular cytokine staining, cells were stimulated with PMA (100 ng ml -1 ), Ionomycin (0.5 mM), Brefeldin A (5 mg ml -1 ) and Monensin (2 mM) (Sigma-Aldrich, St. Louis, MO) for 4 h, followed by fixation with 2% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA), permeabilization and staining with anti-cytokine antibodies in PBS/0.15% saponin (Sigma). Staining for nuclear factors Ki67, FOXP3, AHR, cMAF, Blimp-1 and IRF4 were performed with Foxp3 staining buffer kit (eBioscience). Flow cytometry data were acquired on LSRII, FACS Aria II or Fusion systems (BD Biosciences), and analyzed in FlowJo (Tree Star, Ashland, OR). All analyses were done gated on fixable viability dye negative singlet population. Representative plots of gating strategies for sorting and cell analyses are shown in Supplementary Fig. 8.
Quantitative real time PCR. Total RNA was extracted using the RNeasy kit (Qiagen) and cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad). To quantify the gene transcript levels, 'Best Coverage' gene probes for mouse Gapdh (Mm99999915_g1), mouse Il10 (Mm01288386_m1), human GAPDH (Hs02786624_g1) and human IL10 (Hs00961622_m1) (TaqMan, Life Technologies, Grand Island, NY) were used on a ViiA 7 real-time PCR system following the manufacturer's recommended program (Hold Stage: 50°C for 2 min, 95°C for 10 min; then PCR Stage with 40 cycles: 95°C for 15 s and 60°C for 1 min). IL-10 relative mRNA level was calculated by normalizing Il10 values to the internal loading control Gapdh values first, and then to the average of Il10/Gapdh values in WT.