IL-4 prevents adenosine-mediated immunoregulation by inhibiting CD39 expression

The ectonucleotidase CD39 functions as a checkpoint in purinergic signaling on effector T cells. By depleting eATP and initiating the generation of adenosine, it impairs memory cell development and contributes to T cell exhaustion, thereby causing defective tumor immunity and deficient T cell responses in older adults who have increased CD39 expression. Tuning enzymatic activity of CD39 and targeting the transcriptional regulation of ENTPD1 can be used to modulate purinergic signaling. Here, we describe that STAT6 phosphorylation downstream of IL-4 signaling represses CD39 expression on activated T cells by inducing a transcription factor network including GATA3, GFI1, and YY1. GATA3 suppresses ENTPD1 transcription through prevention of RUNX3 recruitment to the ENTPD1 promoter. Conversely, pharmacological STAT6 inhibition decreases T cell effector functions via increased CD39 expression, resulting in the defective signaling of P2X receptors by ATP and stimulation of A2A receptors by adenosine. Our studies suggest that inhibiting the STAT6 pathway to increase CD39 expression has the potential to treat autoimmune disease while stimulation of the pathway could improve T cell immunity.


Introduction
Purinergic signaling is an important mechanism in fine tuning immune cell functions (1)(2)(3). Extracellular ATP, released by damaged and dying cells, is a danger signal that plays a proinflammatory function. In T cell differentiation, ATP is secreted by activated T cells and stimulates P2X receptors (P2RX) (4). Calcium influx through P2RX7 has been shown to be pivotal for generating long-lived memory and tissue-resident effector CD8 T cells (5). P2RX4 is recruited to the T cell receptor (TCR) synapse, provides costimulatory signals, and regulates migration (6). CD39 and CD73 are the 2 most critical molecules with enzymatic activities to calibrate the purinergic signals. CD39 depletes secreted ATP, thereby inhibiting P2RX stimulation. The cooperation between CD39 and CD73 produces extracellular adenosine that acts as a mostly immunosuppressive signal to modulate the functions of immune cells. In addition, extracellular AMP produced by CD39 in the absence of CD73 is also immunosuppressive through binding to P1 receptors (7). Taken together, CD39 expression on the cell surface largely determines whether the microenvironment is proinflammatory or immunoregulatory (1)(2)(3).
CD73 is constitutively expressed on most naive CD8 T cells and 20%-40% of naive CD4 T cells (8). Upon activation, it is downregulated but re-expressed on a smaller fraction of CD4 and CD8 central and effector memory T cells (8,9). In contrast, CD39 expression is inducible in naive and memory T cells by TCR stimulation. CD39 + CD4 + T cells have functional characteristics of effector T cells with increased apoptosis susceptibility, at least in part due to stimulation of the A2A receptor (A2AR) by adenosine (10). They have a reduced potential to differentiate into long-term memory T cells. T cells from older adults are more prone to CD39 expression due to more sustained mTORC1 signaling, which may explain their The ectonucleotidase CD39 functions as a checkpoint in purinergic signaling on effector T cells. By depleting eATP and initiating the generation of adenosine, it impairs memory cell development and contributes to T cell exhaustion, thereby causing defective tumor immunity and deficient T cell responses in older adults who have increased CD39 expression. Tuning enzymatic activity of CD39 and targeting the transcriptional regulation of ENTPD1 can be used to modulate purinergic signaling. Here, we describe that STAT6 phosphorylation downstream of IL-4 signaling represses CD39 expression on activated T cells by inducing a transcription factor network including GATA3, GFI1, and YY1. GATA3 suppresses ENTPD1 transcription through prevention of RUNX3 recruitment to the ENTPD1 promoter. Conversely, pharmacological STAT6 inhibition decreases T cell effector functions via increased CD39 expression, resulting in the defective signaling of P2X receptors by ATP and stimulation of A2A receptors by adenosine. Our studies suggest that inhibiting the STAT6 pathway to increase CD39 expression has the potential to treat autoimmune disease while stimulation of the pathway could improve T cell immunity.
JCI Insight 2022;7(12):e157509 https://doi.org/10.1172/jci.insight.157509 defect in establishing long-lived memory T cells (10,11). CD39 expression also accounts for defective T follicular helper (TFH) cell differentiation in older age; and stimulation of the A2AR activates the cAMP/ PKA/pCREB pathway, which in turn suppresses BCL6 function (12). Moreover, Gupta et al. found CD39 to be coexpressed with PD-1 on effector T cells in chronic lymphocytic choriomeningitis virus (LCMV) infection but not in acute infection. They proposed CD39 to be a T cell exhaustion marker (13). Canale et al. described tumor-infiltrating CD8 + T cells with high expression of CD39 that were absent in lymphoid organs but increased in frequencies with tumor growth (14). Taken together, CD39 expression on T cells and the associated depletion of ATP and increased production of adenosine contribute to T cell dysfunction in immune aging and antitumor responses. Therefore, targeting this pathway is an opportunity to enhance immunity, and several forms of intervening with purinergic signaling, in part in combination with checkpoint inhibition, are being explored (15,16).
While interfering with adenosine signaling appears currently more feasible, targeting CD39 directly has the advantage of maintaining extracellular ATP (eATP) concentrations that are important for the differentiation of memory T cells (5). An alternative to pharmacological inhibition of the enzyme activity is to prevent CD39 expression in T cells. Several studies have explored the transcriptional control of ENTPD1. On endothelial cells, hypoxia-induced Sp1 upregulates CD39 expression (17). Liao et al. found that the cAMP/PKA/pCREB pathway mediated the transcriptional regulation of CD39 expression on murine macrophages (18). This later pathway is also pertinent for T cells, where adenosine generated by CD39/CD73 augments ENTPD1 transcription in a positive feedback loop (12). IL-6-activated pSTAT3 and TGF-β-driven downregulation of GFI1 promoted CD39 and CD73 expression on murine TH17 cells (19). BCL6 can essentially abrogate CD39 expression through directly binding to the ENTPD1 promoter, consistent with the observation that TFH cells do not express CD39. In contrast, RUNX3, essential for effector cell differentiation, transcriptionally upregulated CD39 expression on T cells (12). Therefore, transcription factor networks quintessential for T cell differentiation and largely driven by cytokines drive CD39 expression.
To identify actionable cytokine-transcription factor networks, we screened cytokines for their ability to influence CD39 expression induced by TCR stimulation and found that IL-4 inhibited ENTPD1 transcription through the phosphorylation of STAT6 and the upregulation of GATA3, GFI1, and YY1. GATA3 functioned by interacting with RUNX3 and blocking RUNX3 binding to the ENTPD1 promoter. The ability of STAT6 inhibition to upregulate CD39 expression may be useful for treating autoimmune disease, while activation of the pathway in an immune response has the potential to prevent T cell dysfunction.

IL-4 signaling inhibits CD39 expression.
CD39 is a checkpoint in purinergic signaling that is induced in selected subsets of effector T cells. The preferential induction on TH1 and TH17 cells and the virtual absence on TFH cells (12,19) suggest the importance of the initial cytokine milieu for the transcriptional regulation of ENTPD1. To identify cytokines that prevent expression, we activated total T cells in the presence of IL-2, IL-4, IL-7, IL-15, TNFα, Type I IFN, and Type II IFN. IL-4 was the only cytokine that downregulated CD39 expression (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.157509DS1). It did so in both naive and memory T cells, suggesting that the effect was not limited to TH2 differentiation (Supplemental Figure 1B). Subsequent experiments in a larger cohort confirmed inhibition of CD39 expression on both CD4 and CD8 T cells by IL-4 ( Figure  1A). Inhibition occurred at the level of ENTPD1 transcription ( Figure 1B) and was not limited to T cells but was also found upon treating B cells with IL-4 ( Figure 1C). The finding on B cells demonstrated that IL-4 not only prevented the induction but also decreased the constitutive expression of CD39. However, exhausted T cells had downregulated the IL-4R, suggesting that they are no longer susceptible to the IL-4 effect. We analyzed splenic and tumor-infiltrating T cells (TILs) from B16 melanoma/Ova-bearing mice ( Figure 1D). Most TIL-expressing PD-1 and TIM3 coexpressed CD39. These cells had reduced IL-4R expression compared with CD39 -CD4 + and CD8 + splenic T cells. In contrast to TILs, splenic CD39 + CD4 + T cells lacking TIM3, likely resembling Tregs, had normal IL-4R.
STAT6 inhibition in a T cell response upregulates CD39 expression. IL-4R stimulation results in STAT6 phosphorylation. In cultures of purified total T cells activated by anti-CD3/CD28 Dynabeads, we observed STAT6 (Y641) phosphorylation starting on day 1 and subsiding on day 4 after stimulation ( Figure 2A). In parallel, GATA3 was upregulated upon activation and downregulated on day 4 after stimulation JCI Insight 2022;7(12):e157509 https://doi.org/10.1172/jci.insight.157509 ( Figure 2A). This initial peak in STAT6 phosphorylation was inhibited by blocking the IL-4R (Figure 2A), suggesting autocrine or paracrine activity of IL-4. CD39 expression started to rise on day 3 and then progressively increased, suggesting that the delay in the expression of CD39 was due to the transient production of IL-4 in these cultures (Figure 2A). Western blotting confirmed the expression of GATA3 for 3 days  Figure 2B). GATA3 expression was inhibited by adding IL-4R blocking Abs to the culture; the addition of IL-4 served as a positive control ( Figure 2B). AS1517499 is a potent and selective STAT6 inhibitor through inhibition of STAT6 phosphorylation and blocking pSTAT6 binding to DNA sequence (20)(21)(22). Inhibition of GATA3 expression was seen even at the low concentration of 100 nM ( Figure 2B).
Culture of T cells purified from human peripheral blood with AS1517499 for 4 days induced CD39 expression on a markedly higher fraction of CD4 and CD8 T cells compared with control-treated cells ( Figure 2C). Furthermore, transfecting T cells with STAT6-specific siRNA SMARTPOOL increased ENT-PD1 transcription ( Figure 2D). To examine whether STAT6 inhibition upregulated CD39 expression on T cells in vivo, we used the LCMV infection mouse model. Naive CD45.1 + SMARTA CD4 T cells, specific for LCMV GP61-80, were adoptively transferred into CD45.2 + B6 WT mice, followed by LCMV infection and daily AS1517499 treatment for 6 days. Flow cytometric analysis showed increased CD39 expression on SMARTA cells in STAT6 inhibitor-treated mice compared with the control group ( Figure 2E). Taken together, pSTAT6 suppresses activation-induced CD39 expression in T cells.
GATA3-mediated inhibition of ENTPD1 transcription is independent of DNA binding. Lentiviral transduction of human T cells with GATA3 significantly reduced CD39 expression after activation on both CD4 and CD8 T cells ( Figure 3A). We predicted 2 GATA3 binding sites in the CD39 promoter sequence at -448 bp and -1424 bp relative to transcription start site (TSS) using public online software Matinspector. However, we did not detect GATA3 binding to these sequences in ChIP assays of T cells activated for 3 days, in contrast to the IL4 gene promoter, which served as a positive control ( Figure 3B). GATA3 ChIP-seq data on in vitro differentiated human TH1 and TH2 cells from the public resource of Cistrome (23) did not yield any peaks in the entire ENTPD1 gene while peaks in the IL4 gene region proved the quality of the data set (Supplemental Figure 2). Luciferase reporter assays provided further evidence that GATA3 does not directly regulate the ENTPD1 promoter. Compared with controls, GATA3 overexpression in HEK293T cells did not reduce the luciferase intensity regulated by the ENTPD1 promoter ( Figure 3C). Taken together, transcription factor GATA3 downstream of IL-4 signaling indirectly suppresses CD39 expression.
GATA3 reduces ENTPD1 expression through inhibiting RUNX3. Several groups have described the physical interaction between GATA3 and RUNX3 to block each other's binding to DNA sequences (24,25). RUNX3 is a positive regulator of CD39 expression on T cells through binding to the ENTPD1 promoter (12). Lentiviral overexpression of RUNX3 during T cell activation significantly upregulated CD39 expression on CD4 and CD8 T cells ( Figure 4A). Conversely, RUNX3 knockdown using a commercial shRNA lentivector plasmid reduced CD39 induction compared with the control group ( Figure  4B). To determine whether GATA3 inhibits CD39 expression through preventing RUNX3 binding to the ENTPD1 promoter, we first performed CoIP assays of tagged RUNX3 and GATA3 overexpressed in HEK293T cells. When using anti-FLAG magnetic beads to pull down RUNX3, a weak but clear GATA3 band was detected, confirming the physical interaction between RUNX3 and GATA3 ( Figure  4C). Previous work identified RUNX3 binding to 2 sites in the ENTPD1 promoter (12). When ChIP assays were performed with T cells activated for 4 days in the presence and absence of IL-4, a significant suppression of RUNX3 binding to these 2 sites was observed with IL-4 ( Figure 4D). The reduced binding was not due to reduced RUNX3 expression. We collected T cells activated for 3 days in the presence of IL-4, STAT6 inhibitor, and IL-4R blocking Abs, respectively. High doses of IL-4R blocking Abs may slightly reduce, but certainly did not increase, RUNX3 expression. The other treatments had no effect ( Figure 4E). To further prove that the inhibition of IL-4R signaling on CD39 expression depends on the presence of RUNX3, we activated splenocytes from WT and Cbfb -/-B6 mice for 4 days with plate-coated anti-CD3/CD28 Ab in the presence or absence of IL-4. IL-4 reduced CD39 expression on CD4 and CD8 T cells from WT but not in Cbfb-deficient mice ( Figure 4F).
IL-4-induced transcription factor network repressing ENTPD1 includes GFI1 and YY1. In addition to GATA3, transcription factors GFI1 and YY1 were also induced by IL-4 ( Figure 5, A and B). Binding sites of GFI1 and YY1 are both predicted in the ENTPD1 promoter sequence ( Figure 5C). GFI1 is a transcriptional repressor that has been shown to suppress CD39 expression in mouse Th17 cells (19). YY1 can be a transcriptional repressor or activator; its activity is reduced in T cells from older adults, which are more prone to CD39 expression (10,26,27). We cloned the full length of human GFI1 and YY1 cDNA into pCDH lentivectors containing a GFP reporter gene. Lentiviral overexpression of GFI1 or YY1 in activated human T cells significantly reduced CD39 induction ( Figure 5D). STAT6 inhibition suppresses T cell function through CD39 upregulation. CD39 on activated effector T cells functions as a checkpoint of purinergic signaling by depleting eATP and initiating the generation of adenosine that binds to P1 receptors. eATP can be detected in day 4 culture of T cells after activation, but not in medium in the absence of cells, suggesting active secretion of ATP (Supplemental Figure 3). To determine whether the increased expression of CD39 upon STAT6 inhibition is sufficient to change eATP, we added 20 μM ATP to day 4-T cell cultures and monitored eATP over the subsequent hour. T cells with the AG or GG genotype of the ENTPD1 SNP at rs_10748643 that allows for CD39 expression on effector T cells were used. Cells activated in the presence of 100 nM STAT6 inhibitor AS1517499 for 4 days were more efficient at degrading ATP than control cultures ( Figure 6A). We had 1 sample from an adult with the A/A genotype which fails induction of CD39 expression. Here, no effect of the STAT6 inhibitor on eATP was detected (Supplemental Figure 3). The relative changes in CD39 expression on day 4 after IL-4 treatment were small ( Figure 1). The addition of 20 ng/mL IL-4 to the cultures had no effect on eATP depletion after 30 minutes and had a small but significant effect at 1 hour ( Figure 6B).
To determine whether the increased expression of CD39 has functional consequences, we assessed cytokine production upon restimulation of T cells that were activated in the presence of the STAT6 inhibitor and found reduced generation of IL-2, IFN-γ, and TNFα ( Figure 6C). STAT6 is known to regulate other transcription factors, such as IRF4 (28), BLIMP1 (29), and TCF1 (30) that could affect the ability to produce cytokines independent of CD39. We therefore performed experiment to determine whether the decline is due to purinergic signaling. Restimulation of activated T cells by anti-CD3/CD28 Dynabeads in the presence of 500 μM adenosine for 6 hours reduced production of cytokines in CD4 and CD8 T cells including IFN-γ, TNFα, and IL-2 ( Figure 6D) suggesting that generation of adenosine may account for the defective effector function of these T cells; however, required adenosine concentrations were high. Results using the A2AR selective agonist CGS21680 at 100 nM in the same system showed consistent inhibition for TNFα on CD4 and CD8 T cells but less so for IL-2 and not for IFN-γ ( Figure 6E). Conversely, restimulation of cells in the presence of the P2RX inhibitor oxidized ATP (oATP) also suppressed cytokine production upon restimulation, suggesting that the defective effector function could be directly caused by eATP depletion ( Figure 6F). To directly address the question whether the inhibition of cytokine secretion by the STAT6 inhibitor was through upregulated CD39 expression and generation of adenosine from endogenous ATP, purified T cells were treated with STAT6 inhibitor (AS1517499), A2AR antagonist (SCH442416), or a combination of both during T cell activation. Addition of the A2AR inhibitor improved cytokine production in both settings, but even more so in effector T cells that were generated in the presence of the STAT6 inhibitor. The inhibitory effect of STAT6 inhibition was nearly completely abrogated by A2AR blocking ( Figure 6G). These data support the notion that the inhibition of T cell function by AS1517499 was mainly mediated by elevated CD39 expression and adenosine production, although a contribution of additional eATP depletion by CD39 is possible.

Discussion
The ectonucleotidase CD39 is the rate-limiting enzyme in the degradation of extracellular ATP. As a major checkpoint in purinergic signaling, it is a target for immune intervention; inhibiting its enzymatic activity or preventing its expression on activated T effector cells is immunostimulatory and would therefore be beneficial in particular settings of T cell exhaustion. Here, we describe transcription factors downstream of IL-4 function as repressors of ENTPD1, thereby improving T cell responses upon restimulation through preventing adenosine production or ATP depletion. IL-4-induced GATA3 inhibited ENTPD1 transcription indirectly by physically interacting with RUNX3, blocking its recruitment to the ENTPD1 promoter. GFI1 and YY1, both also TFs induced by IL-4, reduced CD39 expression when overexpressed in T cells. Conversely, STAT6 inhibition greatly increased the frequency of T cells that expressed CD39 in an immune response, indicating its application in the treatment of autoimmune diseases.
In recent years, several strategies have been explored to manipulate the ATP-adenosine pathway to improve immune responses to tumors. CD73 antagonists and inhibitors of A2A receptors have shown efficacy naive SMARTA cells were isolated and injected into CD45.2 B6 WT mice via tail vein, followed by LCMV Armstrong infection and daily i.p. injection of the STAT6 inhibitor AS1517499 or DMSO solvent control. After 6 days, SMARTA cells were isolated from spleen and examined for CD39 expression using flow cytometry. Representative contour plots (upper panel) and summary data (lower panel). Data in D are shown as mean ± SEM and compared by 2-tailed paired or unpaired Student's t test in C and E.
JCI Insight 2022;7(12):e157509 https://doi.org/10.1172/jci.insight.157509 in improving tumor immunity. More recent studies have employed inhibitors of CD39 that have the advantage of not only inhibiting the generation of adenosine but also the degradation of ATP, thereby preserving ATP's proinflammatory and immune stimulatory activity (16,31). An alternative approach to pharmacological inhibition of CD39 is to prevent its expression. CD39-deficient mice generated by genetic KO of the Entpd1 gene exhibited dysregulation of many physiological processes (32,33), consistent with the wide tissue distribution of CD39. Moreover, human siblings with a stop-codon mutation in ENTPD1 developed neurological and gastrointestinal complications (34). These data suggest that significant side effects must be expected from global CD39 repression or inhibition. However, human adults with a frequent SNP at the position rs10748643 of the ENTPD1 gene are largely healthy with the exception of an increased susceptibility to inflammatory bowel disease (35). The SNP is responsible for different levels of CD39 expression in T cells. Individuals with the GG variant express high CD39 on T cells upon activation, while those with the AA allele induce CD39 only on a very small number of T cells and only with a low cell surface density. Consequently, individuals with an AA allele are able to generate more TFH cells (12) and have better vaccine responses, while in those with the GG allele, effector T cells are more short-lived (10). Interestingly, the effect of the allelic polymorphisms is mainly limited to T cells and not seen with B cells or monocytes, suggesting cell-type specific transcriptional regulation, which, therefore, opens the opportunity for selective targeting that appears to be safe.
Several transcription factors have been implicated in regulating ENTPD1 transcription. Initial studies have been spearheaded by Pinsky's group, mostly in cell types that have constitutive expression of CD39. Hypoxia was shown to induce CD39 expression through the transcription factor SP1 (17). Regional expression on endothelial cells was in part determined by KLF2 under laminar shear stress (36). pCREB regulated  the ENTPD1 promoter in mouse macrophages (18). Also, several studies have examined the transcriptional control of ENTPD1 in T cells. pCREB, also important in T cells, is a possible target for intervention. Inhibition of cyclooxygenase and prostaglandin synthesis impaired pCREB-dependent transcription of ENTPD1. pCREB is also part of an important feedback loop downstream of adenosine-mediated A2AR stimulation, thereby susceptible to A2AR blockers (12). In CD8 T cells, CD39 expression is regulated by NAD(P)H oxidases, possibly through an ROS-dependent activation of NF-κB and JNK pathways (37). In vitro generation of murine TH17 cells is associated with the expression of CD39 and CD73 through IL6-mediated STAT3 signaling and TGF-β-mediated downregulation of Gfi-1 (19); however CD39 and CD73 are rarely coexpressed in human T cells and the vast majority of human CD4 + CD39 + effector T cells are not TH17 cells. A key transcription factor in CD4 T cells is BCL6. Loss of BCL6 or gain in the antagonistic transcription factor BLIMP1 is required for CD39 to be expressed on effector T cells (12). Increased expression of BLIMP1, unleashing ENTPD1 transcription in T cells from older adults appears to be 1 mechanism that curtails effector T cell survival and impairs their function in older age (10,11).
An earlier study has found reduced CD39 expression in TH2 cell-driven diseases, indicating a role of the IL-4 signaling pathway in ENTPD1 transcription. ENTPD1 transcripts were significantly lower in PBMC of patients with asthmatic diseases compared with healthy controls and were inversely correlated with IL-4 and GATA3, but positively correlated with TGF-β and FOXP3 transcripts (38). Here, we show that IL-4 is indeed actively repressing ENTPD1 by inducing GATA3, GFI1, and YY1. This effect was seen irrespective of TH2 differentiation; in fact, studies were done on total T cells and, therefore, mostly lineage-committed T cells that continued to produce IFN-γ and TNF upon restimulation. GATA3 functions by competing with the activity of RUNX3 that is absolutely required for the induction of CD39 expression after T cell activation. The IL-4 signaling pathway, therefore, emerges as a promising druggable target to manipulate purinergic signaling in effector T cells. Moreover, the amount of IL-4 produced at the initiation of an immune response may have downstream effects through attenuating CD39 expression, preventing this arm of T cell exhaustion. Lymphocytes mainly express the type I IL-4 receptor, which is formed by IL-4Rα with the γc chain. The binding affinity of IL-4 to IL-4Rα is very high; therefore IL-4 at a very low concentration can maximally occupy the receptor chains (39). In our polyclonal activation assay, stimulation of total T cells under TH0 conditions in the absence of exogenous IL-4 induced phosphorylation of STAT6, likely due to the activation of infrequent IL-4-producing cells. Elevated pSTAT6 during T cell activation under nonpolarizing conditions has been shown to play an important role in T cell differentiation, such as the downregulation of TCF1 (30) and maintaining IRF4 (28) and FOXP3 (40,41) expression. Our experiments in the LCMV infection model showed that STAT6 inhibition upregulated the expression of CD39 on LCMV-specific SMARTA cells, providing additional evidence that the IL-4-STAT6-GATA3 pathway is activated and relevant in antiviral responses in vivo. Stimulation with IL-4 during the first days of T cell activation could therefore prevent CD39 expression that contributes to exhaustion. A previous study has shown that CD39 inhibition restored cytokine production, consistent with our findings, but that PD-1 inhibition induced proliferation (42). Thus, different interventions may be needed to improve functionality of exhausted T cells. It should be noted that expression of IL-4R was low on exhausted TIM3 + CD8 T cells, rendering them resistant. IL-4, therefore, can only be used to prevent, not inhibit, CD39 expression on exhausted cells.
While the major focus of targeting CD39 is on improving or preventing T cell exhaustion in antitumor responses or optimizing vaccine response in older adults, its immunosuppressive function is likely beneficial in the setting of autoimmune disease. The rs10748643 G/G SNP causing high expression on T cells is associated with reduced production of effector cytokines IL-17 and IFN-γ (43), reduced graft versus host disease in a human mouse chimera model (44), and reduced generation of TFH cells (12). The AA SNP associated with low CD39 expression was linked to increased susceptibility to Crohn's disease (35). ENTPD1 KO mice have a high susceptibility to murine colitis (35). It is undetermined whether this effect is related to the absence of CD39 on professional Tregs, but upregulation of CD39 on effector T cells may curtail pathological inflammatory T cell responses. We found that the potent and selective STAT6 inhibitor AS1517499 is a collected for CoIP assay using anti-FLAG magnetic beads to pulldown RUNX3. Western blots of total lysates (upper panel) and anti-FLAG pull downs (lower panel) are shown. (D) Purified total T cells were activated for 4 days by anti-CD3/CD28 Dynabeads in the presence or absence of IL-4. Cells were collected for RUNX3 ChIP, followed by qPCR of 2 RUNX3 binding sites in the ENTPD1 promoter. Data are shown as a percentage of input. (E) Purified total T cells were activated by anti-CD3/CD28 Dynabeads under indicated conditions and collected on day 3 for Western blotting of RUNX3. (F) Splenocytes were isolated from WT and Cbfb -/-B6 mice and cultured on plates precoated with anti-CD3 and CD28 Abs (both 8 μg/mL) in the presence or absence of IL-4 (20 ng/mL). After 4 days, CD39 expression was compared by flow cytometry. Results are representative of 2 experiments. Data in D are shown as mean ± SEM and compared by 2-tailed paired or unpaired Student's t test in A, B, and D. powerful inducer of CD39 expression on in vitro cultured CD4 and CD8 T cells (Figure 2A). In vivo experiments also showed upregulation in antigen-specific CD4 T cells during acute LCMV infection ( Figure 2E). T cells treated with AS1517499 had decreased cytokine production in CD4 and CD8 T cells, including IL-2, IFN-γ, and TNF, that was likely a combined effect of ATP depletion and reduced P2RX stimulation plus adenosine generation and A2AR stimulation. While a direct effect of STAT6 inhibition on T cell differentiation, for example by a failure of inducing IRF4 (45), cannot be excluded, the reversibility by the A2AR antagonist SCH442416 implicates the increased CD39 expression. In addition to potential applications of STAT6 inhibitors in TH2-driven diseases, they may also be useful to prevent excessive inflammatory responses in viral infection or treat autoimmune disease by increasing adenosine signaling.
Cells were analyzed on an LSRII or LSR Fortessa (BD Biosciences), and flow cytometry data were analyzed using FlowJo (TreeStar).
Quantification of extracellular ATP. Purified total T cells were activated with anti-CD3/CD28 Dynabeads for 4 days. eATP in the supernatant was quantified using the ATP determination kit (Thermo Fisher Scientific). Briefly, supernatants were incubated with the premade mix including luciferase and its substrate luciferin for 10 minutes at room temperature. Luminescence was read using BioTek Synergy H1. ATP concentrations were calculated from a standard curve.
adenosine (500 μM for D), or the A2AR agonist CGS21680 (100 nM for E), or the P2RX antagonist oATP (100 μM for F) for 6 hours and then stained for intracellular cytokines. (G) Purified T cells were treated with STAT6 inhibitor AS1517499, A2AR antagonist SCH58621, or both during T cell activation by Dynabeads. On day 4, cells were restimulated by PMA/IONO for 4 hours before being collected for intracellular cytokine staining. Results are pooled from 3 experiments. Data in C are shown as mean ± SEM and compared by paired 2-tailed Student's t test in A-G.
Mice, LCMV infection, and AS1517499 treatment. C57BL/6J (B6) mice were purchased from the Jackson Laboratory. LCMV-Armstrong was grown in BHK cells and titered in Vero cells. Naive CD4 + cells were isolated from CD45.1 SMARTA mice and injected into CD45.2 B6 WT mice through the tail vein. After 2 days, mice were infected with LCMV Armstrong at a dose of 2 × 10 5 plaque-forming units, followed by daily i.p. injection of 10mg/kg STAT6 inhibitor AS1517499 (AXON Medchem, catalog Axon 1992) or DMSO solvent control. Mice were sacrificed after 1 week and CD39 expression on splenic SMAR-TA cells was determined. All mice were housed in the Stanford Research Animal Facility according to Stanford University guidelines.
Characterization of TILs for CD39 and IL-4R expression. C57BL/6J mice were reconstituted with OT-I cells via tail vein injection. After 1 day, mice were implanted with B16-OVA cells in the right flank. Tumor growth was monitored 2 times a week using digital calibers. Animal studies were performed with approval by the Mayo Clinic Institutional Animal Care and Use Committee.
Statistics. Statistical analysis was performed using Prism (GraphPad). Paired or unpaired 2-tailed Student's t tests were used for comparing 2 groups, and 1-way ANOVA with Tukey's post hoc test was used for multigroup comparisons. P < 0.05 was considered statistically significant.
Study approval. Human studies were approved by the Stanford and the Mayo Clinic IRBs. Animal experiments were approved by the Stanford University Institutional Animal Care and Use Committee.

Author contributions
FF, WC, YM, CMW, and JJG designed the study. FF, WC, YM, HO, LL, and JQ performed the experiments. WC, FF, WC, YM, CMW, and JJG analyzed and interpreted the data. FF and JJG wrote the manuscript with all authors providing feedback. FF, WC, and YM contributed equally to this study. FF initiated this study and therefore is listed first.