Distinct regulation of Nr4a receptors by NFAT and ERK signalling during T cell activation

Nr4a gene family members are activated by T cell receptor (TCR) signalling and play key roles in Regulatory T cell differentiation and have also been implicated in promoting T cell exhaustion in cancer. The precise signalling pathways that govern their transcription, however, remain unclear. Here we utilise Nr4a3-Tocky mice to elucidate the signalling pathways that govern Nr4a1, 2 and 3 expression in response to physiological stimulation of CD4+ and CD8+ T cells. Our findings reveal that Nr4a1-3 are Src family kinase-dependent. Moreover, Nr4a2 and Nr4a3 are abolished by calcineurin inhibitors and bind NFAT1, highlighting a necessary role for NFAT in the control of Nr4a2 and Nr4a3. Interestingly, the NFAT pathway is redundant for Nr4a1 activation, but all three Nr4a members require ERK signalling for optimal expression. Analysis of T cells expressing constitutively active NFAT1 reveals that NFAT1 alone is sufficient to induce expression of Nr4a2 and Nr4a3, but not Nr4a1. These findings further our understanding of Nr4a regulation, highlighting key differences in the sensitivity of Nr4a1 and Nr4a3 to distal TCR signalling pathways.


Introduction
The Nr4a family of orphan nuclear receptors consists of three members: Nr4a1 (Nur77), Nr4a2 (Nurr1) and Nr4a3 (Nor1). Nr4a receptors have been shown to be ligand-independent and their structure is set to a constitutively active form (Wang et al., 2003). Nr4a1 and Nr4a3 have been shown to be rapidly upregulated in peripheral T cells (Ashouri and Weiss, 2017;Bending et al., 2018b;Moran et al., 2011;Zikherman et al., 2012) and thymocytes (Cheng et al., 1997), following T cell receptor (TCR) signalling, and are considered a more specific marker of T cell activation than the cell surface marker CD69, which can be upregulated by non TCR stimuli (Moran et al., 2011). Expression of Nr4a1 (Moran et al., 2011) and Nr4a3 (Bending et al., 2018b) in CD4 + T cells is lost in major histocompatibility complex (MHC) Class II knockout mice in vivo, highlighting the key role of TCR signalling in regulating expression of the Nr4a receptors.
Early studies of Nr4a1 identified a key role for it in the induction of apoptosis by TCR-induced Nur77 protein expression (Liu et al., 1994), which associates with Bcl-2 at the mitochondria to mediate negative selection (Thompson and Winoto, 2008).
Nr4a2 can bind Foxp3 regulatory elements and regulate the differentiation of CD4 + T cells (Sekiya et al., 2011), and persistent Nr4a3 expression is a hallmark of Treg undergoing differentiation in the thymus (Bending et al., 2018b). Genetic ablation of Nr4a receptors abolishes Treg development and can lead to multiorgan autoimmunity (Sekiya et al., 2013). Nuclear Nr4a1 can modulate Treg differentiation and clonal deletion (Fassett et al., 2012), highlighting a critical role for Nr4a family members in thymic T cell development and immune regulation.
More recent studies of the Nr4a family have found that nuclear factor of activated T cells (NFAT) (Martinez et al., 2015) and Nr4a receptors (Mognol et al., 2017;Scott-Browne et al., 2016) are intimately linked to the development of CD8 + T cell exhaustion, suggesting that NFAT and Nr4a family members may cooperate during chronic antigen stimulation to adapt T cell programmes. Indeed, Nr4a family members have been shown to limit chimeric antigen receptor (CAR) T cell function in solid tumours  and that Nr4a1 may be a key mediator of T cell dysfunction through binding and repressing expression of the activator protein one (AP-1) transcription factor (Liu et al., 2019). Nr4a family member function is complex since they can also promote CD8 + T cell exhaustion through cooperation with other transcription factors such as thymocyte selection-associated high mobility group box protein (TOX) and TOX2 . In addition, Nr4a1 can alter the metabolic state of T cells, acting as a break during T cell activation to dampen inflammation (Liebmann et al., 2018).
Given the central roles of Nr4a members in autoimmunity and cancer, they are an emerging therapeutic target. For example, pharmacological inhibition of Nr4a nuclear receptors can enhance antitumor immunity (Hibino et al., 2018). Therefore, understanding the regulation of the Nr4a transcription factors is of both fundamental and therapeutic interest. Whilst Nr4a members are paralogues, the signalling pathways that regulate their expression have not been clearly defined, although a role for NFAT1 has been proposed for CD8 + T cells (Martinez et al., 2015;Mognol et al., 2017;Scott-Browne et al., 2016). Here we utilised the state-of-the-art Nr4a3-Timer of cell kinetics and activity (Tocky) system (Bending et al., 2018a;Bending et al., 2018b) to determine the pathways regulating Nr4a transcription in CD4 + and CD8 + T cells. Our findings highlight that the calcineurin (CaN)/NFAT pathway is necessary and sufficient for the induction of Nr4a2 and Nr4a3, but redundant for Nr4a1. Additionally, the mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) pathway is required for optimal Nr4a1, Nr4a2 and Nr4a3 transcriptional activity.

Peptide stimulation induces Src family kinase-dependent persistent dynamics of Nr4a3 transcription in CD4 + and CD8 + T cells
In order to interrogate the signalling pathways that control Nr4a3 expression, Nr4a3-Tocky mice were crossed with the Tg4 Tiger TCR transgenic line (Burton et al., 2014), which express an autoreactive CD4 + TCR specific for myelin basic protein (MBP) (Liu et al., 1995). This system allows a sophisticated dissection of the temporal patterns of TCR signalling in response to self-agonist peptides ( Supplementary Fig. 1A). In addition, in order to study CD8 + TCR signalling dynamics Nr4a3-Tocky mice were bred to the ova-peptide specific OTI TCR transgenic system (Hogquist et al., 1994) (Supplementary Fig. 1B). Analysis of spleen and thymus of these two transgenic lines revealed selective expression of the transgenic beta chains Vb8 TCR (Tg4) and Vb5 in OTI ( Supplementary Fig. 1).
Stimulation with myelin basic protein (MBP) Ac1-9[4Y] peptide (Tg4) or ova peptide (OTI) triggered activation of Nr4a3-Timer expression, resulting in early Blue fluorescence, before time dependent maturation to a Blue + Red + fluorescent state indicative of persistent dynamics of Nr4a3 expression (Fig. 1A&B). Expression of Nr4a3-Blue (i.e. active TCR signalling) was shown to be dependent on the Src family kinase, since incubation of Tg4 Nr4a3-Tocky T cells with the inhibitor PP2 (Hanke et al., 1996) abolished peptide-induced expression of Nr4a3 and CD69 in both Tg4

Nr4a3 is calcineurin-pathway dependent in CD4 + and CD8 + T cells
TCR signalling results in the activation of many signalling intermediaries but converge on the activation of at least three key transcription factors: NFAT, AP1 and NFkB (Brownlie and Zamoyska, 2013). Given that membrane depolarisation and calcium flux are central to T cell activation and that Nr4a receptors have previously been linked to NFAT activation (Scott-Browne et al., 2016), we investigated the sensitivity of Nr4a3 to NFAT pathway inhibitors. To do this we utilised two distinct inhibitors of calcineurin, a key enzyme in TCR signalling which dephosphorylates NFAT in response to Ca 2+ signalling (Hogan et al., 2003) (Fig. 2). Interestingly, incubation with either cyclosporin A (a fungal derived product that forms a complex with cyclophilin to block the phosphatase activity of calcineurin (Matsuda and Koyasu, 2000)) or FK506 (Tacrolimus, a macrolide calcineurin inhibitor (Thomson et al., 1995)) completely abolished TCR stimulation-induced Nr4a3-Timer Blue expression in CD4 + Tg4 T cells (Fig. 2A&B), whilst not significantly affecting the frequency of CD69 + T cells (Fig. 2C). Cyclosporin A did show a significant effect on CD69 MFI levels ( Fig. 2D), suggesting that CD69 is in part regulated by the CaN/NFAT pathway. Identical findings were mirrored for CD8 + T cells ( Fig. 2E-H).

Nr4a family members bind NFAT1, but only Nr4a2 and Nr4a3 are sensitive to calcineurin inhibitors
The calcineurin inhibitor data suggested that Nr4a3 is an NFAT responsive distal TCR signalling reporter. Three isoforms of NFAT are expressed in T cells (NFAT1, NFAT2 and NFAT4, (Macian, 2005)). To ascertain NFAT binding, we identified a previously published ChIP-Seq data (GSE64409, (Martinez et al., 2015)) for the binding of NFAT1 to Nr4a family members (Fig. 3A). To control for non-specific binding NFAT1 -/cells were compared to NFAT1 intact mice in unstimulated and stimulated CD8 + T cells. Analysis of NFAT1 binding peaks revealed clear NFAT1 binding across the Nr4a2 and Nr4a3 genes, which were only evident in stimulated CD8 + T cells from NFAT1 wild type mice. Nr4a1 showed specific peaks upstream of the gene but also displayed considerable non-specific binding throughout the Nr4a1 gene region. These findings together confirm that Nr4a family members bind NFAT1 within activated CD8 + T cells.
To test the sensitivity of the other Nr4a family members to calcineurin inhibitors, splenocytes from Tg4 mice were stimulated for four hours with soluble anti-CD3 (to activate CD4 + and CD8 + T cells) and RNA extracted. Endogenous levels of Nr4a1, Nr4a2 and Nr4a3 mRNA were quantified in response to TCR stimulation in order to rule out a potential artefact of the BAC transgenic system. Intriguingly, whilst Nr4a2 and Nr4a3 showed a complete inhibition of transcriptional upregulation in response to both cyclosporin and FK506 (comparable to the effect of PP2, Src family kinase inhibitor). Nr4a1 was insensitive to both cyclosporin A and FK506 calcineurin inhibitors (Fig. 3B). Using an antibody to Nr4a1, intracellular staining was performed to verify these findings at the protein level (Fig. 3C). These data suggest that NFAT is necessary for Nr4a2 and Nr4a3 expression but is redundant in the regulation of Nr4a1.

ERK signalling is required for optimal Nr4a1, 2 and 3 expression in CD4 + and CD8 + T cells
NFAT is known to extensively co-operate with AP-1 (Jain et al., 1993;Peterson et al., 1996), and together the two activate many important genes in response to TCR stimulation, such as Ifng and Il2. AP-1 is dependent on MAP kinase activity (Karin, 1995), in particular ERK pathway activation can lead to the upregulation of c-Fos and enhanced AP-1 transcription factor activity. Past reports have suggested that Nr4a1 is ERK sensitive in ovarian cells (Stocco et al., 2002). To test the sensitivity of Nr4a3 and CD69 to ERK/AP1 pathway inhibition, splenocytes from Tg4 mice were peptide stimulated in the presence of a MEK/ERK inhibitor (PD0325901, (Barrett et al., 2008)) and compared to the effects of DMSO or cyclosporin A (Fig. 4A). TCR induction of Nr4a3 was, as previously shown, abolished by CsA, but also showed a significant attenuation in the presence of ERK pathway inhibition (Fig. 4A&B). CD69  (Fig. 4A). OTI CD8 + T cells largely mirrored Tg4 CD4 + T cells in the sensitivity of Nr4a members to NFAT and ERK pathway inhibitors, with all showing partial ERK dependence, but only Nr4a2 and Nr4a3 being sensitive to calcineurin inhibitors (Fig. 4E). These findings were verified by flow cytometric analysis in CD8 + OTI Nr4a3-Tocky mice, revealing that OTI CD8 + T cells also require ERK pathway signalling for optimal Nr4a3 expression, but are absolutely dependent on the CaN/NFAT pathway ( Supplementary Fig. 2).

Constitutively active NFAT1 is sufficient to induce Nr4a2 and Nr4a3 expression
The previous findings suggested that Nr4a2 and Nr4a3 are NFAT-dependent transcription factors, which require ERK signalling for optimal expression. Nr4a1 on the other hand was not dependent on the NFAT pathway, but had a partial dependency on ERK activity. Given that NFAT:AP-1 complexes are critical to many facets of T cell biology, we were interested to know whether NFAT activity alone would be sufficient to drive Nr4a2 and Nr4a3 expression. We therefore performed in silico analysis of a previously published RNA-seq dataset of T cells expressing a constitutively active form of NFAT1 that was modified to abrogate its binding to AP1  Fig. 5B). These data therefore reveal that NFAT1 activity alone is sufficient for the induction of Nr4a2 and Nr4a3 expression, but not Nr4a1 and Cd69, in response to TCR signalling ( Supplementary   Fig. 3).

Discussion
Nr4a receptors are emerging as an exciting target for immunomodulation (Flemming, 2019), in particular as a potential strategy to fine-tune CAR T cell attack of solid tumours (Li and Zhang, 2019). Here we have shown that Nr4a1, Nr4a2 and Nr4a3 are distinctly regulated by the CaN/NFAT pathway, but that all three transcription factors require MEK/ERK pathway activation for optimal expression in response to TCR stimulation in both CD4 + and CD8 + T cells. Interestingly, analysis of NR4A1 protein regulation in anti-CD3 stimulated human T cells has shown that they are also insensitive to calcineurin inhibitors (Ashouri and Weiss, 2017), but show a partial ERK dependence, suggesting that regulation of Nr4a1 may be conserved between mice and humans.
Remarkably, despite being insensitive to calcineurin inhibitors, the Nr4a1 gene region exhibited binding of NFAT1. Surprisingly, both cyclosporin and FK506 had no effect on the transcriptional dynamics of Nr4a1 expression in response to peptide stimulation of either CD4 + and CD8 + T cells. This could represent redundancy in the regulation of Nr4a1 by NFAT1. In particular, it has been previously observed that cyclosporin may interfere with Nr4a1 biology at the level of its DNA binding activity through its N-terminal protein region (Yazdanbakhsh et al., 1995), without interfering with its transcription. In this way, cyclosporin may abrogate the biological effects of all Nr4a family members through disparate mechanisms of action. On the other hand, constitutively active NFAT1 did not significantly alter Nr4a1 expression in CD4 + or CD8 + T cells transduced with the CA-RIT-NFAT1 vector, suggesting that Nr4a1 cannot be activated by NFAT1 activity alone. However, as the CA-RIT-NFAT1 construct is incapable of binding AP-1 (Martinez et al., 2015), it is still possible that NFAT:AP1 complexes could redundantly activate Nr4a1, or that NFAT2, and NFAT4 may play roles. For instance, Cd69 was not significantly upregulated in CA-RIT-NFAT1 T cells but is abolished by co-administration of CaN inhibitors and MEK/ERK pathway inhibitors. These suggest that CD69 has multiple redundant mechanisms controlling its transcription in T cells downstream of TCR signalling, including a partial regulation by JAK inhibitors (Ashouri and Weiss, 2017).
The discovery that ERK has a role in optimal Nr4a expression is in keeping with studies in other tissues. NOR1 (Nr4a3) has been shown to be regulated by the MAPK/ERK pathway in response to platelet derived growth factor in vascular smooth muscle cells (Nomiyama et al., 2006). Similarly, in ovarian cells, calcium-dependent activation of ERK mediates AP-1 induction of Nur77 (Nr4a1) (Stocco et al., 2002), again establishing a common link between ERK signalling and the regulation of Nr4a expression in a wide range of cell types.
The finding that all Nr4a members are regulated by the Src family kinase suggests that the initiation of the pathway to Nr4a induction is by the activity of lymphocytespecific protein tyrosine kinase (Lck). Lck associates with the cytoplasmic tails of the CD4 and CD8 co-receptors (Veillette et al., 1988). Peptide-MHC engagement of the TCR leads to Lck mediated phosphorylation of the ITAM containing TCR and CD3 chains (Straus and Weiss, 1992). Interestingly, past work has suggested that more CD4 coreceptors are loaded compared to CD8 molecules within T cells, which can affect the dwell time of T cells in response to negatively selecting antigens (Stepanek et al., 2014). Comparison of Nr4a transcriptional dynamics in this study suggested that CD8 + T cells receive an initial shorter, sharper and stronger activation of Nr4a transcription, which peaks at 1hr. Interestingly Nr4a receptor transcription in CD4 + T cell peaked around 2hrs and appeared to plateau. These findings warrant further investigation to understand the temporal regulation of Nr4a expression following antigen experience, but could also reflect differences in early negative feedback pathways, differential affinity of the TCR and antigen, and/or differences in MHC Class I and II expression levels.
Despite the different sensitivities of Nr4a1 and Nr4a2/3 to NFAT pathway inhibitors, the overall kinetics of expression in peptide stimulated CD4 + or CD8 + T cells was fairly uniform. However, it is important to note that these readouts are from primary stimulation of a largely naïve population of TCR transgenic T cells. For instance, antigen-experienced T cells in humans show altered ERK and p38 phosphorylation patterns compared to naïve T cells (Adachi and Davis, 2011). Nr4a family receptors may therefore be useful in studying the re-tuning of T cell receptor signalling pathways -in particular comparing Nr4a1 to Nr4a2/Nr4a3 transcriptional responses.
Currently within the field there exists several Nr4a1 (Nur77) GFP transgenic reporters (Moran et al., 2011;Zikherman et al., 2012). These reporters have been used to study Treg and iNKT development as well as address B cell tuning to endogenous antigen. However, as previously alluded to by Weiss and colleagues, temporal analysis of Nr4a1-GFP reporters may be hampered by the persistence of GFP expression following antigen encounter (Au-Yeung et al., 2014). The Nr4a3-Tocky system employed in this study does not suffer from this same issue, since the half-life of Timer Blue protein is 4hrs (Bending et al., 2018b), allowing a sensitive readout of distal TCR signalling changes over much shorter periods of time. Given that NFAT in the absence of AP-1 induces a chronic T cell exhaustion phenotype, such as PD1 high Lag3 high CD8 + T cells (Martinez et al., 2015), it is likely that Nr4a3-Tocky will prove a useful model for understanding NFAT pathway activity in animal models of T cell dysfunction and cancer in vivo. Furthermore, assessing Nr4a3 and Nr4a1 co-regulation may be a method to interrogate alterations in TCR signalling in T cells during chronic antigen exposure or memory T cell responses.

In silico ChIP-Seq
Processed bigwig data files deposited in GSE64409 (Martinez et al., 2015) were downloaded and hosted in CyVerse Discovery Environment (de.cyverse.org/de) and then mapped against the mm9 genome using the UCSC genome browser. These files contain analysis of NFAT1 ChIP-Seq in CD8 T cells from WT and NFAT1 -/mice either unstimulated or stimulated for 1hr with PMA and Ionomycin as described in (Martinez et al., 2015).

RNAseq analysis
Log2 fold change estimates of Nr4a1, Nr4a2, Nr4a3 and Cd69 expression was extracted from DESeq data deposited in (Martinez et al., 2015) for CD4 + or CD8 + T cells either transfected with mock vector or CA-RIT-NFAT1.

Statistical analysis and data visualisation
Statistical analysis was performed on Prism 7 or 8 software (GraphPad).
For comparison of more than two means a one-way ANOVA with Tukey's multiple comparisons test was used. For comparison of more than two means over time, a two-way ANOVA with Tukey's multiple comparison's test was used. Adjusted p values for Figure 5 were extracted from DESeq analysis in (Martinez et al., 2015).