A threshold level of NFATc1 activity facilitates thymocyte differentiation and opposes notch-driven leukaemia development

NFATc1 plays a critical role in double-negative thymocyte survival and differentiation. However, the signals that regulate Nfatc1 expression are incompletely characterized. Here we show a developmental stage-specific differential expression pattern of Nfatc1 driven by the distal (P1) or proximal (P2) promoters in thymocytes. Whereas, preTCR-negative thymocytes exhibit only P2 promoter-derived Nfatc1β expression, preTCR-positive thymocytes express both Nfatc1β and P1 promoter-derived Nfatc1α transcripts. Inducing NFATc1α activity from P1 promoter in preTCR-negative thymocytes, in addition to the NFATc1β from P2 promoter impairs thymocyte development resulting in severe T-cell lymphopenia. In addition, we show that NFATc1 activity suppresses the B-lineage potential of immature thymocytes, and consolidates their differentiation to T cells. Further, in the pTCR-positive DN3 cells, a threshold level of NFATc1 activity is vital in facilitating T-cell differentiation and to prevent Notch3-induced T-acute lymphoblastic leukaemia. Altogether, our results show NFATc1 activity is crucial in determining the T-cell fate of thymocytes.

D ifferentiation of CD4 À CD8 À double-negative (DN) cells to the CD4 þ or CD8 þ single-positive (SP) T cells in the thymus is regulated by a complex network of signalling pathways involving multiple transcription factors at various stages of development. On the basis of their differentiation status, DN thymocytes consist of four distinct populations, CD44 þ CD25 À DN1, CD44 þ CD25 þ DN2, CD44 À CD25 þ DN3 and CD44 À CD25 À DN4 (ref. 1). DN3 thymocytes upon rearrangement of their T-cell receptor b (Tcrb) locus express a functional TCRb chain, which, in combination with the preTa (pTa) chain forms the preTCR (pTCR). PreTCR signalling is essential for the differentiation of DN3 cells to DN4 and later stages. We have shown recently that the transcription factor NFATc1 plays a critical role during early stages of thymocyte development 2 . A haematopoietic lineage cell-specific ablation of NFATc1 activity blocked thymocyte development at DN1 stage, the earliest stage of thymic T-cell development 2 . Mice deficient for NFATc2, NFATc3 or both, do not show any apparent defect in thymocyte development 3,4 , suggesting a critical role played by NFATc1 during thymic T-cell development.
Ligation of TCR with cognate peptide-MHC complex in vivo, or stimulation of T cells with anti-CD3 plus anti-CD28 Abs in vitro increases intracellular Ca 2 þ levels, which in turn activate the serine/threonine phosphatase calcineurin. Active calcineurin dephosphorylates multiple serine/threonine residues in NFAT proteins and facilitates their nuclear translocation. We have previously elucidated a novel NFAT activation pathway in pTCR-negative thymocytes, which plays an indispensable role in their survival and differentiation 2 . In contrast to the calcineurin-mediated dephosphorylation pathway, in pTCRnegative thymocytes IL-7-JAK3 signals activated NFATc1 via phosphorylation of tyrosine 371 in the regulatory domain 2 . Both, the calcineurin-mediated 'conventional' and IL-7-JAK3-mediated 'alternative' pathways though explained the post-translational mechanisms of NFAT activation, the transcriptional regulation of Nfatc1 itself is poorly understood. A previous study showed Nfatc1 expression in T cells is autoregulated by NFATc1 (ref. 5). In this study, we have delineated the signalling pathways that regulate Nfatc1 expression with distinct promoter usage at pTCRnegative and -positive thymocytes. Further, we provide evidence in support of a critical role of NFATc1 in suppressing lineage plasticity of immature thymocytes towards non-T lineages, and the essentiality of a threshold level of NFATc1 activity at the pTCR-positive DN3 stage in facilitating the T-cell fate of thymocytes by preventing the development of T-Acute Lymphoblastic Leukaemia (T-ALL).

Results
Differential usage of Nfatc1 promoters in thymocytes. In T cells, two distinct promoters, a distal (P1) and a proximal (P2), initiate Nfatc1 expression. Due to alternative splicing and usage of two different poly-adenylation (pA) sites, six Nfatc1 isoforms; three from P1 promoter (Nfatc1aA, aB and aC) and three from P2 promoter (Nfatc1bA, bB and bC), respectively, are synthesized ( Supplementary Fig. 1a) 6,7 . To distinguish whether any particular NFATc1 isoform is prevalent in pTCR-negative and -positive thymocytes we analysed wild-type (WT) DN1-DN4 cells for Nfatc1 expression. Interestingly, we observed an exclusive P2 promoter activity in the pTCR-negative DN3 cells, whereas pTCR-positive DN4 cells exhibited both P2 as well as P1 promoter activity ( Fig. 1a; Supplementary Fig. 1b). Exclusive P2 activity in pTCR-negative thymocytes was supported by an active chromatin configuration, as indicated by histone modifications and a concomitant recruitment of RNA polymerase II (Pol II) at the Nfatc1 P2 promoter in Rag2 À / À DN3 cells, whereas only little H3K4me3 was detectable at the P1 promoter (Fig. 1b). Analysis of small RNA-seq data 8 , to score for bidirectional paused transcripts enriched at promoters further confirmed significant enrichment of transcriptional start site RNAs at P2 promoter over P1 (Fig. 1b). Using an additional set of primers to detect total aor b-specific transcripts derived from the P1 or P2 promoters respectively, again we observed an exclusive P2 activity in DN3, and both P1 and P2 activity in DN4 cells ( Supplementary Fig. 1c). Further, in agreement with our observation regarding induction of P1 promoter activity at the pTCR-positive stages, we detected NFATc1a proteins in DN4 cells but not in DN3 cells ( Fig. 1c; Supplementary Fig. 1d). Keeping in view the findings of a previous report 5 , that in naive T cells, constitutive basal level TCR signalling maintains the P2 promoter activity, and TCR-antigen ligation induced signals are responsible for the P1 promoter activity, it was quite intriguing how thymocytes lacking even the pTCR exhibit such a robust Nfatc1 expression.
To further substantiate our observation regarding selective promoter usage, and the role of pTCR signalling in inducing P1 activity, we investigated three different mouse models, where either there was no pTCR (Rag1 À / À ), pTCR was absent but a downstream signalling molecule calcineurin was constitutively active (Calcineurin transgenic; DCam) 9 or there was enhanced pTCR signalling (Notch3 transgenic; N3 tg) 10 . In both Rag1 À / À and DCam mice, T-cell development is blocked at the DN3 stage due to lack of pTCR signals 9 . In contrast, N3 tg mice showed an enhanced DN3 to DN4 transition due to strong pTCR signals (Fig. 1d,e). Confirming our ChIP-Seq and RNA-Seq observations, analysis of Nfatc1 expression in Rag1 À / À DN3 cells showed the presence of only Nfatc1b isoforms (P2 activity), similar to that in WT DN3 cells (Fig. 1f). Stimulating Rag1 À / À DN3 cells with anti-CD3 Abs to mimic pTCR signalling, we readily detected Nfatc1a transcripts in addition to the Nfatc1b transcripts implying that pTCR signals can induce P1 promoter activity (Fig. 1f). Further, P1 promoter activity was autoregulated by NFATc1, as P1 activity disappeared without affecting P2 activity when anti-CD3 antibodies stimulated Rag1 À / À DN3 cells were treated with cyclosporine A (CsA; Fig. 1f). In contrast to the WT and Rag1 À / À mice, DCam DN3 cells showed a robust P1 activity in addition to the P2 promoter activity (Fig. 1g). The strong P1 activity in DCam DN3 cells was due to the autoregulatory loop maintained by constitutive calcineurin activity, as CsA treatment specifically extinguished Nfatc1 P1 activity without affecting the P2 activity (Fig. 1g). Further, we observed a similar pattern of strong P1 and P2 promoter activity as that in DCam mice, in N3 tg DN3 cells (Fig. 1h). Thus, it was evident that Nfatc1 is expressed from distinct promoters in pTCR-negative and -positive thymocytes, and pTCR signalling is necessary for the induction of Nfatc1 P1 promoter activity.
NFATc1 activity is vital for DN thymocyte differentiation. Due to exclusive P2 activity in pTCR-negative thymocytes, we investigated whether NFATc1b is solely critical for the differentiation of early DN thymocytes. To clarify this, we have generated a mutant mouse with floxed Nfatc1 P2 promoter element (P2 fl/fl ) to abolish NFATc1b activity in a tissue-specific manner ( Supplementary Fig. 2a-d). We bred P2 fl/fl mice with mice expressing cre-recombinase under Vav promoter (Vav-Cre) 11 to abolish Nfatc1 P2 activity during thymocyte development. Surprisingly, analysis of Vav-CreP2 fl/fl mice showed completely normal T-cell development, which was indistinguishable from that of littermate controls (Fig. 2a-c; Supplementary Fig. 2e). The normal T-cell development in Vav-CreP2 fl/fl mice was quite surprising, as loss of NFATc1 activity was expected to block thymocyte development at the DN1 stage 2 . PCR with reverse transcription (RT-PCR) analysis on pTCR-negative thymocytes showed a total lack of Nfatc1 P2-derived transcripts indicating abolition of Nfatc1 P2 promoter activity (Fig. 2d). However, NFATc1 activity was not lost rather in absence of P2 activity we observed a robust P1 promoter activity in the Vav-CreP2 fl/fl pTCR-negative thymocytes (Fig. 2d).
We have previously shown that Bcl-2 is a target of NFATc1 in pTCR-negative thymocytes 2 . However, analysis in Vav-CreP2 fl/fl DN3 cells showed similar levels Bcl-2 proteins as that in littermate controls ( Supplementary Fig. 2f), suggesting that most likely NFATc1a and -b isoforms regulate a similar set of target genes that are critical for early thymocyte survival and differentiation. Thus, in the Vav-CreP2 fl/fl mice, Nfatc1 P1 promoter activity functionally compensated for the loss of P2 activity, underlining the indispensability of NFATc1 activity in thymocyte differentiation.
NFATc1 activity is essential for T-cell development. To investigate the physiological significance of the distinct pattern of NFATc1 promoter activity in pTCR-negative and -positive thymocytes, we explored what impact NFATc1a will have on thymocyte development if it is co-expressed with NFATc1b in pTCR-negative thymocytes. To address this issue, we used mice in which a constitutively active version of Nfatc1aA was knockedin into Rosa-26 locus flanked by a floxed stop cassette (R26-caNfatc1aA-Stop fl/fl : designated hereafter as Nfatc1aA fl/fl ) 12 . To activate Nfatc1aA expression in early thymocytes we bred Nfatc1aA fl/fl mice with Vav-Cre mice. Surprisingly, analysis of Vav-CreNfatc1aA fl/fl mice showed severely impaired thymocyte development as evident from a dose-dependent reduction in the size of the thymus, spleen and lymph nodes (Fig. 3a). Accordingly, the cellularity in these organs was drastically reduced in Vav-CreNfatc1aA fl/fl mice leading to T-cell lymphopenia (Fig. 3b,c). Analysis of DN cells from Vav-creNfatc1aA fl/fl mice to understand the reason behind the low thymic cellularity revealed a dose-dependent block in the transition of DN3 cells to the DN4 stage ( Fig. 3d  cells (Fig. 3e). This observation further suggests that the reduced thymic cellularity in the Vav-creNfatc1aA fl/fl mice was not because of an increase in cell death of the DN thymocytes. Similar to the DCam mice 9 , enforced NFATc1aA expression in DN3 cells resulted in the lack of rearranged TCRb expression (Fig. 3f), leading to the loss of pTCR signalling, although the expression of CD3e a component of the pTCR complex remained unaffected ( Supplementary Fig. 3b). In addition, NFATc1aA activity in pTCR-negative cells strongly retarded their rate of differentiation to the double-positive (DP) and later stages. In in vitro co-culture assays on OP9-DL1 (bone marrow stromal cells expressing Notch ligand delta-like 1) monolayer, Vav-CreNfatc1aA fl/fl DN1-DN4 cells showed inefficient differentiation to DP stage compared with the cells from WT mice (Fig. 3g). These observations suggest that the combined activity of NFATc1a and NFATc1b, or rather an increase in total NFATc1 activity, before pTCR-positive stages is detrimental for T-cell development.
To rule out the possibility that the negative impact on thymocyte development in the Vav-CreNfatc1aA fl/fl mice was due to an excess in NFATc1 activity derived from both the endogenous as well as the transgene, we generated a mouse model where only NFATc1aA was expressed without any contribution from the endogenous Nfatc1 gene. We bred Vav-CreNfatc1aA fl/fl mice with Nfatc1 fl/fl mice to eliminate all NFATc1 activity derived from the endogenous Nfatc1 gene. Analysis of Vav-CreNfatc1aA fl/fl Nfatc1 fl/fl mice revealed that even in the absence of any endogenous NFATc1, NFATc1a activity derived from the knocked-in gene only resulted in a similar block in thymocyte differentiation at the DN3 stage compared with littermate controls (Supplementary Fig. 3c,d), as observed in the Vav-CreNfatc1aA fl/fl mice. This could still be due to an above threshold level of NFATc1 activity derived from the transgene. Thus, from our analysis it is apparent that a certain threshold level of NFATc1 activity is essential for T-cell development.
NFATc1 suppresses lineage plasticity of immature thymocytes. T-cell development in the thymus follows a sequential process of T-lineage specification, and commitment. While DN1 cells retain the potential to differentiate into B cells, natural killer (NK) cells, dendritic cells (DCs) and macrophages, the lineage plasticity gets restricted to NK and DC lineages in DN2 cells, and finally is completely lost at the DN3 stage 13 . Accordingly, only in WT DN1 cells we observed Ebf1 and Pax5 expression necessary for B-cell development, whereas, expression of Id2 necessary for NK cell, or Spf1 (PU.1), Csf1r and Cebpa, necessary for myeloid-lineage development were maintained in DN2 cells (Fig. 4a). We assumed that NFATc1b activity most likely play a role in suppressing the lineage plasticity during DN1 to DN3 differentiation and thereby consolidate T-lineage commitment. To investigate this possibility, we analysed lineage-specific gene expression in DN1-DN4 cells from Vav-CreP2 fl/fl mice. We observed a similar suppression of lineage plasticity towards non-T lineages in the absence of NFATc1b activity (Fig. 4b), as in the WT mice (Fig. 4a). As in the Vav-CreP2 fl/fl mice P1 promoter-driven NFATc1a activity compensated for the loss of NFATc1b during T-cell development, we concluded that a threshold level of NFATc1 activity irrespective of any particular isoform might be crucial in suppressing lineage plasticity and thereby facilitate T-lineage commitment.
DN1 cells are a heterogeneous population, and based on CD24 (heat-stable antigen) and CD117 (c-Kit) expression they are characterized to have five distinct populations termed as DN1a, DN1b, DN1c, DN1d and DN1e (ref. 13). DN1a and DN1b cells maintain highest lineage plasticity, whereas, DN1d and DN1e cells are more specified to take the T-cell lineage. We presumed that the combined activity of NFATc1a, and NFATc1b is essential for irreversible T-lineage commitment at the pTCR-positive stage, and NFATc1a if expressed along with NFATc1b early at the pTCR-negative stages will suppress the lineage plasticity more effectively. Interestingly, analysis of Vav-CreNfatc1aA fl/fl mice showed a strong reduction in DN1a and DN1b populations compared with littermate WT mice (Fig. 4c,d). Accordingly, we observed a dose-dependent specific loss of B cells in the thymus from Vav-CreNfatc1aA fl/fl mice compared to WT mice whereas, NK, DC or macrophage lineages  were not affected (Fig. 4e). Corroborating this, Ebf1 and Pax5 expression were suppressed in Vav-CreNfatc1aA fl/fl DN1 cells, whereas gene expression necessary for NK and myeloid lineages was unaffected (Fig. 4f). However, the suppression of B-lineage potential of the Vav-CreNfatc1aA fl/fl DN1 cells was neither due to enhanced cell death nor due to any toxic effect of NFATc1 activity on the precursor cells. In a B-lineage permissible environment, FACS sorted DN1 thymocytes from Vav-CreNfat-c1aA fl/fl mice developed comparable proportion of B220 þ B cells as that in case of WT DN1 cells, when co-cultured on OP9 bone marrow stromal cell layer ( Supplementary Fig. 3e). This observation ruled out that there was any inherent developmental restriction towards B-lineage differentiation in the Vav-CreNfat-c1aA fl/fl DN1 cells.
Integrin signalling induces Nfatc1 P2 promoter activity. The exclusive P2 activity in the WT pTCR-negative thymocytes led us to ask how the P2 promoter is regulated in these cells. Surprisingly, when we cultured DN4 thymocytes in vitro, not only all P1-directed transcripts disappeared but we also observed the absence of all P2-derived transcripts as well (Fig. 5a). However, annexin V analysis revealed that majority of cells in this culture condition were alive, ruling out cell death being the reason for the Gated on CD4 -CD8 -CD44 + CD25 -DN1 cells loss of Nfatc1 expression ( Supplementary Fig. 4a). We have previously reported the responsiveness of Nfatc1 promoter activity to Forskolin (FSK) treatment in vitro 5 . To investigate whether cyclic-AMP (cAMP) signalling is responsible for Nfatc1 P2 activity, we treated WT DN3 and DN4 cells with a cAMP analogue 8-CPT-cAMP. Interestingly, 8-CPT-cAMP treatment specifically induced the expression of Nfatc1b, both in DN3 and DN4 cells (Fig. 5b). To prove that P2 activity is cAMP signallingdependent, we explored for a physiological context where the intracellular cAMP level is high. Several recent studies have reported increased cAMP levels in CD4 þ CD25 þ Foxp3 þ regulatory T (T reg ) cells [14][15][16] . The effect of cAMP in inducing Nfatc1 P2 promoter activity was confirmed as T reg cells exhibited stronger P2 promoter activity compared with CD4 þ CD25 - αA βA αB βB αC βC αA βA αB βB αC βC Foxp3effector T (T eff ) cells (Fig. 5c). This observation was corroborated with higher level of green fluorescent protein (GFP) expression reflecting higher NFATc1 levels in T reg cells against T eff cells from Nfatc1-eGfp-Bac tg reporter mice 17 (Fig. 5d).
Next, we investigated which signalling pathway is involved in generating intracellular cAMP that drives Nfatc1 expression in thymocytes and in T cells. Cell-cell interaction is mediated by various integrins, and thymocytes, as well as T cells express a number of integrins. CD2 (LFA-2) expressed on thymocytes and T cells has been shown to activate T cells 18,19 . Murine CD2 binds to its ligand CD48 expressed on T cells as well as on the interacting cells and thereby can transduce signals both in cis and in trans 20-23 . In addition, CD2 signalling in T cells have been reported to induce cAMP 24,25 , and thereby the activation of cAMP response element binding (CREB) proteins 26 . Mice lacking cAMP-CREB signalling show impaired thymocyte proliferation and altered fetal T-cell development 27,28 . To check whether CD2-CD48 interactions regulate Nfatc1 expression in DN thymocytes, we analysed CD2 and CD48 expression on various thymocyte subsets. We observed only a fraction of DN thymocytes expressed CD2 on their surface, which was increased in DP and SP thymocytes as well as in peripheral T cells (Fig. 5e). However, CD48 expression was highest on DN cells and lowest in DP and CD4 þ SP cells (Fig. 5f). Among DN thymocytes, DN1 cells expressed the highest level of CD2, whereas CD48 expression was high on all pTCR-negative DN populations (Fig. 5g,h). To prove whether CD2-CD48 signals could induce Nfatc1 expression, we stimulated WT DN3 cells with CD48 antibodies and analysed for the generation of P1-and P2-directed transcripts. CD48 stimulation at a higher concentration induced the synthesis of both P1-and P2-directed Nfatc1 transcripts in DN3 cells, while stimulation with PMA, reported to substitute CD2 co-stimulation 29 , did not generate any Nfatc1 RNA (Fig. 5i)  activity in the pTCR-negative thymocytes unraveling the effect of signal strength on the inducibility of P1 and P2 promoters ( Supplementary Fig. 4b).
Although CD2-CD48 signalling induced Nfatc1 expression in vitro, these are unlikely to be the only integrins responsible for Nfatc1 expression in vivo. This was supported by the reported normal T-cell development in Cd2 À / À mice [30][31][32] , as well as in mice injected with anti-CD2 antibodies 33 , and also in Cd48 À / À mice 34 , indicating the involvement of additional integrins in regulating Nfatc1 expression. Accordingly, analysis of thymocytes and T cells showed a differential expression pattern for various integrins (Supplementary Fig. 4c; Fig. 5j). CD4 þ CD25 þ T reg cells both in the thymus and in LNs expressed much higher levels of various integrins compared to CD4 þ CD25 À T eff cells ( Fig. 5k; Supplementary Fig. 4d), concurring with our observation regarding higher Nfatc1 expression in T reg cells over T eff cells (Fig. 5c). Among the CD4 þ CD25 þ population, CD25 hi cells expressed highest level of various integrins, the most prominent being CD18, CD48, CD49d and CD62L ( Supplementary  Fig. 5a,b). This integrin expression pattern directly correlated to that of Nfatc1 expression in CD25 hi cells as revealed from higher GFP levels in the Nfatc1-eGfp-Bac transgenic reporter mice ( Supplementary Fig. 5c). To further consolidate our observation, we stimulated WT CD4 þ T cells with fibronectin, the ligand for CD18. Interestingly, fibronectin stimulation induced robust Nfatc1 expression, which was mainly derived from the P2 promoter (Fig. 5l). In addition, similar to that in DN3 thymocytes (Fig. 5i), CD2-CD48 signalling also induced Nfatc1 P2 activity in peripheral CD4 þ T cells (Fig. 5l). Thus, the above observations established integrin-cAMP signalling as a critical component in regulating Nfatc1 gene expression in thymocytes and in T cells.
An intronic cis-regulatory element controls P1 activity. To address the question why P1 promoter is inactive in the pTCR-negative thymocytes, we analysed DNA methylation status at the P1 promoter in these cells. However, methylation analysis did not reveal any significant difference between WT DN3 and DN4 cells (Supplementary Fig. 6a). To investigate whether epigenetic modifications on pTCR signalling at the DN3 to DN4 transition induce P1 activity we performed ChIP-Seq analysis of Rag2 À / À DN3 cells stimulated with anti-CD3 Abs to mimic pTCR signals. However, no additional epigenetic changes were detected at the P1 promoter in the anti-CD3 stimulated Rag2 À / À DN3 cells compared to the unstimulated cells ( Supplementary Fig. 6b; Fig. 1b). These observations suggested that P1 promoter induction still requires involvement of additional cis-regulatory elements and/or trans-acting factors. To identify cis-regulatory elements essential for P1 promoter activity, we studied the DNase I hypersensitivity analysis of the human NFATc1 locus (Roadmap Epigenomics Project, www.roadmapepigenomics.org). Besides the P1 and P2 promoter elements, two very distinct DNase I hypersensitive sites designated as E1 and E2 were evident in the intron between exons 10 and 11 in the CD34 þ human haematopoietic stem cells (Fig. 6a). Significantly, these elements were tissue-specific as compared with the hematopoietic stem cells, CD3 þ human T cells showed only the presence of E2 site, which was conserved in murine T cells as well (Fig. 6a).
To investigate the influence of E1 and E2 elements on Nfatc1 expression, we generated luciferase reporter constructs with the P1 or P2 promoter in combination with the E1 or E2 element (Fig. 6b). On stimulation with PMA þ Ionomycin (I), in EL-4 thymoma cells the E2, but not the E1 element induced a strong Nfatc1 P1 promoter activity compared with a mild increase in P2 promoter activity (Fig. 6b). We also tested two additional DNAse I hypersensitive sites, E3 positioned in the intron between exons 3 and 4, and E4 positioned at the extreme 3 0 region of Nfatc1 locus for their potential regulatory effect on Nfatc1 promoter activity. However, both these elements were ineffective in inducing Nfatc1 P1 promoter activity ( Supplementary Fig. 6c), suggesting the essentiality of the E2 element in the context of P1 promoter activity.
To extend our characterization of the E2 enhancer, we analysed for the trans-acting factors, which might play a role in enhancing the E2-mediated P1 promoter activity. We detected binding motifs for three prominent lymphoid-specific transcription factors; NFATc1, GATA3 and PU.1 in the E2 element ( Supplementary Fig. 6d). To investigate whether any of these factors positively modulates the E2 enhancer activity, we mutated the binding sites for individual factors and checked their enhancer potential for Nfatc1 P1 promoter activity. In reporter assays, only the NFATc1 mutant showed loss of enhancer activity, whereas GATA3 or PU.1 mutants exerted no influence on Nfatc1 P1 promoter activity (Supplementary Fig. 6d). This was in line with the previous reports from our laboratory that NFATc1 autoregulates the P1 promoter activity 5 , however, the involvement of the E2 element was hitherto unknown.
Further, to study the effect of E2 element on Nfatc1 promoter in vivo, we generated Nfatc1-eGfp-Bac-DE2 tg reporter mice by deleting 1Kb DNA from intron 10 harbouring the E2 element ( Supplementary Fig. 6e) 17 . Loss of E2 enhancer activity in Nfatc1-eGfp-Bac-DE2 tg reporter mice, showed a clear reduction in Nfatc1 expression in CD4 þ T cells as evident from reduced GFP expression both in unstimulated as well as in stimulated cells compared with that in the Nfatc1-eGfp-Bac tg reporter mice (Fig. 6c). This reduction in NFATc1 levels was due to a specific loss of P1-derived NFATc1a, as we observed a strong decrease in NFATc1a proteins in immunoblot analysis with antibodies against GFP as well as NFATc1a, whereas loss of E2 enhancer activity did not influence NFATc1b protein levels ( Fig. 6d;  Supplementary Fig. 7). Thus, we have identified a novel cis-regulatory element E2, acting as an enhancer, specifically for the P1 promoter activity in T cells.
NFATc1 activity prevents experimentally induced T-ALL. At the DN3 stage, though pTCR signalling is vital for proliferation and differentiation of thymocytes, it is equally crucial that the pTCR signals are switched off in the differentiated cells. Failure to do so will lead to the development of T-ALL, an aggressive form of leukaemia observed in both mouse and humans with constitutive Notch signalling [35][36][37][38][39][40] .
We have previously shown that T-cell development in DCam mice is severely blocked at DN3 stage due to a defect in pTCR formation 9 . Further analysis revealed a strong downregulation in Ptcra (pTa) expression (Fig. 7a), in DCam DN3 cells. Ptcra expression is regulated by Notch signalling 41 , and accordingly, in N3 tg T cells pTa expression is not extinguished 38 . The negative effect of NFATc1a on Ptcra expression in DCam DN3 cells suggests that Nfatc1 P1 activity might be involved in downregulating pTa expression once the cells have received pTCR signals. Inhibition of NFATc1 activity by CsA treatment restored Ptcra expression in DCam DN3 cells (Fig. 7b), further strengthening the possibility that NFATc1a suppresses Ptcra expression. This was confirmed by a dose-dependent suppression of Ptcra expression in DN4 cells in Vav-CreNfatc1aA fl/ þ and Vav-CreNfatc1aA fl/fl mice compared with WT cells (Fig. 7c). Further, ChIP assays confirmed NFATc1a binding at the Ptcra promoter in vivo in WT DN and DCam DN3 cells (Fig. 7d). Also, in reporter assays NFATc1a failed to induce Ptcra promoter activity whereas, it efficiently activated the murine Il2 promoter shown as mean ± s.d., *** Po0.0001, one-way analysis of variance. ** P ¼ 0.0039 or 0.0049 and *P ¼ 0.0317 or 0.0207, unpaired t-test. (Fig. 7e). However, in human T-ALL cell line Jurkat, in reporter assays we observed a strong downregulation in Notch-induced Ptcra promoter activity by NFATc1a (Fig. 7f). Thus, our observations suggest that Nfatc1 P1 activity critically influences Notch-pTCR signalling, necessary for normal T-cell development.
If this is the case, modulation of Nfatc1 P1 activity might help to prevent T-ALL pathogenesis. We used N3-induced T-ALL as a model to check the role of NFATc1 in T-ALL development. N3 tg mice developed severe T-ALL with CD4 À CD8 À CD44 À CD25 þ DN3 cells accumulating in the thymus and peripheral lymphoid organs (Fig. 7g). As a result of tumorigenic transformation the differentiation of DN3 cells to T cells was drastically blocked in N3 tg tumour mice (Fig. 7g). Interestingly, Nfatc1 expression in these N3-induced T-ALL cells was strongly downregulated compared with WT DN3 cells (Fig. 7h). We hypothesized that if the suppressed Nfatc1 expression is the 'cause' and not the 'effect' of T-ALL, then restoring NFATc1 activity in N3 tg mice will help prevent leukemogenesis. To check this possibility in vivo, we crossed DCam mice with N3 tg mice to generate DCam Â N3 double-tg mice. Surprisingly, in contrast to N3 tg mice, none of the DCam Â N3 double-tg mice analysed, showed any sign of T-ALL (Fig. 7i). Rather, the double-tg mice showed a thymic phenotype similar to that observed in DCam mice ( Fig. 7i; Supplementary Fig. 8a). Importantly, in DCam Â N3 double-tg mice even though thymocytes were accumulated at the DN3 stage ( Supplementary Fig. 8b), still they were not tumorigenic suggesting a strong tumour-preventing activity of NFATc1 in this experimental model.

Discussion
We have shown previously the indispensability of NFATc1 activity in early thymocyte development 2 . Here we have delineated the signals that control Nfatc1 expression initiated from two distinct promoters in a thymocyte developmental stagespecific manner and, how NFATc1 activity simultaneously facilitates the T-cell fate of the thymocytes, and prevents the pathogenesis of T-ALL. The finding that NFATc1 expression directed from only P2 promoter at the pTCR-negative stages and from both P1 and P2 promoters at the pTCR-positive stages establish the differential threshold of NFATc1 activity required for thymocyte differentiation. Once the DN3 cells receive pTCR signals, they are irreversibly committed towards T-lineage only. Hence, the NFATc1a activity at the pTCR-positive stages is the key to enhance the threshold of total NFATc1 activity through which T-lineage commitment is established on pTCR signals. Thus, the switch from 'NFATc1b only' to both 'NFATc1a and b' at the pTCR-positive DN3 cells is absolutely necessary for T-lineage commitment in the thymus. In this regard our observation that NFATc1 activity specifically suppresses the B-lineage plasticity of the immature thymocytes provides an important clue as to how activities of various transcription factors stabilize T-cell fate during development. Importantly, the characterization of a novel enhancer element specifically regulating the P1 promoter activities holds immense significance in the context of T-lineage commitment. From our observations, it appears that epigenetically E2 element is in an open conformation at the DN3 stage, but it is the specific transcription factor occupancy, which is critical in inducing P1 promoter activity at the pTCR-positive stage. As we have shown earlier that NFATc1 nuclear levels increases from DN1 to DN3 stage thus, most likely making it possible to bind to the E2 element and autoregulates its own expression by inducing P1 promoter activity. The elucidation of the signalling pathways regulating Nfatc1 gene regulation will help to better understand T-cell development and will provide insight for therapeutic mani-pulations of NFATc1 activity to achieve immune reconstitution in situations like T-cell lymphopenia (Supplementary Fig. 8c).
The loss of Nfatc1 expression in in vitro cultures though quite intriguing was not due to an enhanced cell death of the immature thymocytes ( Supplementary Fig. 4a). Rather, because of very small number of sorted cells their density was too low in the in vitro cultures, which led not only to a lack of interaction between thymocytes and the thymic stroma, but it also resulted in a loss of interaction among thymocytes themselves. In the normal thymus, millions of thymocytes are very densely packed into the available space, which facilitates both thymocyte-stroma and also thymocyte-thymocyte interactions. In our opinion both types of interactions provide the signals essential for Nfatc1 expression. In line with this Nfatc1 expression was not lost when thymocytes were cultured in vitro at a higher density.
We have previously reported a severe T-cell lymphopenia in mice lacking NFATc1 activity in the early thymocytes 2 . Here we show overexpression of NFATc1 also results in T-cell lymphopenia (Fig. 3a-c; Supplementary Fig. 8c), indicating a threshold level of NFATc1 activity is critical for T-cell development. The increase in total NFATc1 activity in the Vav-CreNfatc1aA fl/fl mice not only prevented the differentiation of DN3 cells to DN4 stage ( Fig. 3d) but it also severely affected several key T-cell-specific genes expression. The dose effect of NFATc1, not only on the DN thymocyte development, but also on the DP to SP differentiation is clearly reflected in the cellularity of the thymus, LNs and spleen in the Vav-CreNfatc1aA fl/fl mice compared with WT mice (Fig. 3b,c) underlining the influence of NFATc1 activity on thymic T-cell development.
Our results from DCam Â N3 double-tg mice show the tumour suppressor activity of NFATc1 in the experimental setting of hyperactive Notch, as it prevented T-ALL development despite these mice having constitutive Notch3 signalling. In vivo, once the DN3 cells receive pTCR signals they undergo a limited cell cycle and differentiate to DN4 stage ( Supplementary Fig. 8c). Unless the pTCR signalling is switched off, it will lead to uncontrolled proliferation resulting in T-ALL development ( Supplementary  Fig. 8c). To prevent pTCR formation, Ptcra expression needs to be silenced. Here we show that NFATc1a activity suppresses the Ptcra expression (Fig. 7a-c,e,f). As Nfatc1 P1 activity is initiated only at the pTCR-positive thymocytes and pTCR signals induce NFATc1 activation 42 , this is most likely the natural-feedback mechanism to suppress pTCR expression and thereby prevent T-ALL development once the cells have received optimal pTCR signals ( Supplementary Fig. 8c). However, loss of function mutations involving NFAT has not been reported in spontaneous cases of either human or murine T-ALL; thus, the tumour suppressive activity of NFATc1 may be specific to the background of the additional oncogenic events. Altogether, our study shows that a threshold level of NFATc1 activity is critical for T-cell development. Any alteration, such as above or below threshold levels of NFATc1 activity will push the delicate balance towards an unwanted extreme, either resulting in T-cell lymphopenia or, likely in combination with other oncogenic events, leading towards the development of leukaemia ( Supplementary Fig. 8c). Generation of Nfatc1 P2 fl/fl mice. To generate the Nfatc1 P2 promoter floxed allele, the 5 0 end of the murine NFATc1 cDNA was used to isolate genomic cosmid clones of the Nfatc1 gene. The 8.5 kb XbaI fragment containing the Nfatc1 P2 promoter was identified in DNase I hypersensitive site analysis. The blunt ended left (XbaI to XhoI) and right (XhoI to XbaI) arms were subcloned into the SmaI site of pGL3 basic vector. To incorporate the loxP sites into the left or right arm, loxP sites cut out from the pBS112 SX vector were inserted into the NdeI (448 bp), Eco72I (5,020 bp) and NdeI (7536 bp) sites, respectively. The pTKNeoloxP plasmid containing the neomycin (neo) gene for positive selection and thymidine kinase gene for negative selection was used as basic vector in constructing the targeting vector. Embryonic stem cell clones positive for the targeted P2 floxed allele were selected and were used to generate the chimeric Nfatc1 P2 fl/fl mice by blastocyst injection method. Genomic tail DNA from the chimeric mice was analysed by long-distance PCR to confirm the genomic integration of the floxed Nfatc1 P2 allele, and mice were bred with C57BL/6 mice for several generations. WT, Nfatc1 P2 fl/ þ and Nfatc1 P2 fl/fl mice were genotyped using the following primers: 5 0 -TCTCCACCTGACTTTCTGTTCC-3 0 (forward) and rev: 5 0 -CTCTTCCCAA TGGTTGTCTCTC-3 0 (reverse).
ChIP-Seq and RNA-Seq analysis. ChIP-seq 43 and Small RNA-seq 8 were performed as described previously using same standards and quality controls.
ChIP-Seq. Briefly, DN cells were isolated from Bl.6 Rag2 À / À mice and crosslinked with 1% formaldehyde. Histone ChIPs were performed by using crosslinked nuclear extracts from 5 million cells with 2 mg specific antibodies and 20 ml Dynabeads Protein G suspension (Life Technologies, USA), whereas 50 million cells were used for Pol II ChIPs. Antibodies used are described in Koch et al., with the exception of H3K27ac (ab4729; Abcam). DNA was extracted from immunoprecipitated chromatin and was used to prepare sequencing library using TrueSeq ChIP-seq Library Preparation Kit (Illumina Inc., USA) according to manufacturer's instructions, and sequenced on a Genome Analyzer II sequencing platform (Illumina Inc., USA). Data processing, including input subtraction and wig files generation was performed as previously described 43 .
Small RNA-Seq. Total RNA was extracted from DN cells isolated from Rag2 À / À cells using TriZol Reagent (Life Technologies, USA). Total RNA of 10 mg was fractionated by TBE-Urea PAGE to isolate RNA fragments ranging from 15 nucleotides to 70 nucleotides length. These size-selected small RNAs were used to prepare sequencing library using Small RNA-Seq Library Preparation Kit (Illumina Inc., USA) according to manufacturer's instructions. Libraries were then sequenced on a Genome Analyzer II sequencing platform (Illumina Inc., USA). The RNA-Seq data used is accessible in gene omnibus under the accession number GSE44578.
Semiquantitative RT-PCR. For RT-PCR, cDNA was synthesized from cells either freshly isolated or cultured as indicated in each case using Miltenyi Biotech cDNA synthesis kit. To check CD2-CD48 signalling-induced Nfatc1 expression, 1 Â 10 6 sorted WT DN3 cells were cultured in cRPMI-1640 medium (10% fetal calf serum) supplemented with 10 ng ml À 1 rhIL-7. Cells were either left unstimulated or stimulated with plate bound CD48 antibodies (5 mg), PMA (100 ng) or with CD48 þ PMA for 12 h at 37°C. Similarly, 8 Â 10 6 CD4 þ T cells from lymph nodes of WT mice were left unstimulated or stimulated with plate bound Fibronectin (1 mg), CD48 (0.5 mg) or Fibronectin þ CD48 for 24 h at 37°C. Semiquantitative RT-PCR was performed to check the levels of expression of indicated genes. b-actin expression levels were used as control. RT-PCRs to check the expression of six Nfatc1 isoform were performed using the Fermentas Long distance PCR Kit (Cat. # K0181). Primer sequences are available (Supplementary Table 1) in the Supplemental Information online.
Methylation analysis. Genomic DNA was isolated from FACS sorted WT DN3 and DN4 cells following standard protocol. Sodium bisulphite treatment of genomic DNA was performed using ZYMO Research EZ DNA Methylation Gold kit (Cat. # D5005) according to manufacturer's instructions. All PCR were performed on MWG AG Biotech Primus 96 plus thermocyclers (Ebersberg, Germany) in a final volume of 25 ml including 1 Â PCR Master Mix (Thermo Scientific, Cat. # K0171), 10 pmol each of forward and reverse primers (Nfatc1 P1d forward: (-1297 -818) 5 0 -GGAGATTTGAAAGAGAAAAA-3 0 and Nfatc1 P1d reverse: (-1297 -818) 5 0 -AAARCTACTCTCCCTTTTAAT-3 0 ), and 10 ng of bisulphite-treated genomic DNA. The following amplification protocol was used: 95°C for 10 min and 40 cycles of 95°C for 1 min, 52°C for 45 s, and 72°C for 1 min, and a final amplification step of 72°C for 10 min. PCR products were eluted and purified using innuPREP DOUBLEpure kit (Analytik Jena AG, Cat. # 845-KS-5050050) and cloned into a pGEM T-Easy Vector (Promega, Cat. # A1360) according to manufacturer's protocol. E. coli (Top-F) were transformed with plasmid ligates and were selected by amp R on Lurient-Broth Agar plates. Colonies expressing the full-size amplificate were detected by blue-white selection and DNA was prepared according to standard protocols. Positive clones were further determined by restriction digestion (EcoRI) and subsequently sequenced. PCRs were performed with primers for T7 and SP6 (Sigma-Aldrich) in both 5 0 and 3 0 direction using Big Dye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Product P/N 4336935). Sequence data was generated with an ABI Prism 3130xl, and analysed with BiQ Analyzer v2.0 (Christoph Bock).
Immunoblotting. Positively selected LN CD4 þ T cells (2 Â 10 7 ) from Nfatc1-eGfp-Bac tg or Nfatc1-eGfp-Bac-DE2 tg reporter mice were left unstimulated or stimulated with aCD3 (3 mg ml À 1 ) plus aCD28 (5 mg ml À 1 ) for 48 h. Subsequently, whole-cell protein extracts were prepared and 20 mg protein from each sample was subjected to immunoblotting in 12% polyacrylamide gels. Proteins were transferred to a nitrocellulose membrane followed by immunodetection of GFP with Abs directed against GFP (Santacruz; sc-9996), or NFATc1a with Abs raised against the a-peptide of NFATc1 (ImmunoGlobe; IG-457). Images in Fig. 6d are cropped for presentation excluding the lanes 3 and 6, which represents B-cell extracts. Full-size images from original scan with molecular weight marker are presented in Supplementary Fig. 7.
Luciferase reporter assays. One microgram of a murine Ptcra luciferase reporter construct containing 400 bp enhancer region ( À 4,400 to 4,000 bp) and 1.1 kb DNA fragment spanning murine Ptcra promoter region ( À 1034 to þ 66 bp), or 1 mg of a luciferase reporter construct containing 293 bp DNA fragment spanning murine proximal Il2 promoter region was co-transfected along with 4 mg control vectors into EL-4 thymoma cells by DEAE/Dextran method. Thirty six hours post-transfection cells were left unstimulated or stimulated with PMA plus Ionomycin (100 ng ml À 1 each, Calbiochem) for 12 h.
Afterwards, luciferase activity representing the Ptcra promoter transactivation was measured using a MicroLumat LB 96P (EG&G Berthold) luminometer. For Nfatc1 P1 and P2 promoter constructs, a 1,430 bp (-1155 to þ 275) DNA fragment in the region around exon 1, and a 1260 bp (-1205 to þ 55) DNA fragment in the region around exon 2, respectively, were cloned as NheI/XhoI-digested PCR products into the pGl3basic vector. The potential enhancer elements E1-E4 were cloned 3 0 of the luciferase reporter gene as SalI-digested PCR products. Mutations of potential NFAT, GATA and PU.1 binding sites in the E2 element were introduced by site-directed mutagenesis. The details about the NFATc1 (Nc1A) and Notch1 intracellular domain (N1-IC) expression constructs have been described previously 5,44 . To assess the effects of NFATc1 activity on Notchinduced Ptcra promoter activation, 1 Â 10 6 Jurkat cells were transfected by the ViaFect Transfection Reagent (Promega, Wisconsin, USA) with 5 mg DNA containing all necessary constructs according to the manufacturer's protocol. Twenty four hours later fresh medium was added and the cells were divided into untreated or ionomycin (I) treated samples. After further 16 h cells were collected to estimate luciferase activity.