TNFα-Signaling Modulates the Kinase Activity of Human Effector Treg and Regulates IL-17A Expression

Maintenance of regulatory T cells CD4+CD25highFOXP3+ (Treg) stability is vital for proper Treg function and controlling the immune equilibrium. Treg cells are heterogeneous and can reveal plasticity, exemplified by their potential to express IL-17A. TNFα-TNFR2 signaling controls IL-17A expression in conventional T cells via the anti-inflammatory ubiquitin-editing and kinase activity regulating enzyme TNFAIP3/A20 (tumor necrosis factor-alpha-induced protein 3). To obtain a molecular understanding of TNFα signaling on IL-17 expression in the human effector (effTreg, CD25highCD45RA−) Treg subset, we here studied the kinome activity regulation by TNFα signaling. Using FACS-sorted naïve (naïveTreg, CD25highCD45RA+) and effTreg subsets, we demonstrated a reciprocal relationship between TNFα and IL-17A expression; effTreg (TNFαlow/IL-17Ahigh) and naïveTreg (TNFαhigh/IL-17Alow). In effTreg, TNFα-TNFR2 signaling prevented IL-17A expression, whereas inhibition of TNFα signaling by clinically applied anti-TNF antibodies led to increased IL-17A expression. Inhibition of TNFα signaling led to reduced TNFAIP3 expression, which, by using siRNA inhibition of TNFAIP3, appeared causally linked to increased IL-17A expression in effTreg. Kinome activity screening of CD3/CD28-activated effTreg revealed that anti-TNF-mediated neutralization led to increased kinase activity. STRING association analysis revealed that the TNF suppression effTreg kinase activity network was strongly associated with kinases involved in TCR, JAK, MAPK, and PKC pathway signaling. Small-molecule-based inhibition of TCR and JAK pathways prevented the IL-17 expression in effTreg. Together, these findings stress the importance of TNF-TNFR2 in regulating the kinase architecture of antigen-activated effTreg and controlling IL-17 expression of the human Treg. These findings might be relevant for optimizing anti-TNF-based therapy and may aid in preventing Treg plasticity in case of Treg-based cell therapy.

Effector Treg cells were shown to express pro-inflammatory cytokines such as the autoimmune associated pro-inflammatory cytokine IL-17A, but also naïve Treg was found to produce IL-17A albeit at lower frequencies (5,8). IL-17A-producing Treg have been observed in human inflammatory diseases such as psoriasis and IBD, suggesting that they contribute to the inflammatory process as has been demonstrated in mouse models (9)(10)(11)(12)(13)(14). Although some cues that regulate IL-17A expression by Treg have been identified, including mTOR inhibition (15), CD28 superagonist stimulation (16), and platelet microparticle interaction (17), our mechanistic understanding of IL-17A expression by Treg is limited, let alone that this information is available for naïve and effector Treg. Recently, it has been elucidated that TNFR2 signaling is vital to establish Treg stability by promoting FOXP3 expression and inhibiting secretion of pro-inflammatory cytokines like IL-17A and IFNγ (18,19). In conventional CD4+ memory T cells, inhibition of TNFR2 signaling by anti-TNF led to reduced expression of the anti-inflammatory regulator tumor necrosis factor-alphainduced protein 3 (TNFAIP3, also known as A20), and as a consequence, this resulted in increased IL-17A expression (20). TNFAIP3/A20 acts as a ubiquitin-editing enzyme that regulates multiple other signaling pathways such as IL-17R (21) signaling and kinase activity [e.g., PKC (22), TCR (23), and MAPK (24)].
TNF-TNFR2 signaling appears essential for human Treg expansion and proper function and additionally an autologous TNFα signaling feedback loop has been proposed that regulates IL-17A expression in human Treg (18,19,(25)(26)(27)(28)(29). Anti-TNF therapy is successfully used for the treatment of severe chronic inflammatory diseases such as inflammatory bowel diseases, psoriasis, psoriatic arthritis, and rheumatoid arthritis (30)(31)(32)(33). Paradoxically, it has been observed that in 0.6-5% of the patients treated with anti-TNF medication, this might unintentionally trigger specific forms of immune pathology, suggesting that inhibition of anti-TNF therapy affects Treg function (34)(35)(36)(37). If and how naïve and effector Treg are affected by inhibition of TNFα is not known.
We hypothesize that TNFα signaling controls IL-17A expression in Treg by interfering at the level of kinase activity, which we here explored in eff Treg. We demonstrate that inhibition of TNFα signaling by anti-TNF in vitro led to increased IL-17A expression. Down-regulation of the anti-inflammatory mediator TNFAIP3 played a role in this process. Comprehensive kinome analysis revealed that inhibition of TNFα signaling in eff Treg unexpectedly led to an increase of a kinase activity network containing TCR-linked kinases and immune signaling pathway such as the JAK. Small-molecule-based inhibition of these pathways prevented the anti-TNF-induced IL-17A expression in eff Treg.

RESULTS naïve Treg and eff Treg Cells Reveal a Reciprocal IL-17A-TNFα Relationship
To investigate the link between TNFα and IL-17A expression in naïve and effector Treg, FACSsorted naïve Treg (CD4 + CD45RA + CD25 + ) and eff Treg (CD4 + CD45RA − CD25 high ) ( Figure 1A) derived from healthy volunteers were stimulated with PMA plus ionomycin, and subsequently TNFA, IL17A, IL17F, and RORC (RORÈt) expression was accessed by RT-qPCR ( Figure 1B). As compared to eff Treg, naïve Treg expressed significantly lower levels of IL17A, IL17F, and RORC (p = 0.0005, p = 0.0093, and p = 0.0016, respectively), while TNFA expression was higher (p = 0.0002) ( Figure 1B). Next, we compared the fold change in gene expression between the Treg subsets and observed a reciprocal gene expression signature for TNFA, IL17A, IL17F, and RORC ( Figure 1C). Correlation analysis revealed a reciprocal relationship between TNFA and IL17A (r = −0.50), IL17F (r = −0.42), and RORC (r = −0.68) ( Figure 1D). As expected, a strong positive correlation between IL17A/IL17F (r = 0.81), IL17A/RORC (r = 0.74), and IL17A/RORC (r = 0.54) was observed. The inverse relationship was also confirmed at the protein level upon PMA plus ionomycin stimulation ( Figure 1E) or αCD3/CD28 stimulation of FACSsorted Treg ( Figure 1F). As compared to eff Treg, naïve Treg hardly produced IL-17A, but showed an increased production of TNFα. Analysis of conventional T cells further supported the uniquely high production of IL-17A in these eff Treg, as the numbers of IL-17A/FOXP3-positive cells in FACSsorted naïve or memory CD4 + CD25 − T cells were very low ( Figure S1).   (Figure 2A). At the transcriptional level, we demonstrated that supplementation of rhTNFα suppressed IL-17A, IL-17F, and RORC gene expression in eff Treg ( Figure 2B). These data support the idea that TNFα signaling controls IL-17A expression in eff Treg. TNFα binding to its receptors (TNFR1 and TNFR2) leads to a cascade of intracellular events that culminate in NFκB translocation to the nucleus and subsequent transcription of NFκB target genes NFKBIA (encode Iκβα), NFKB1 (encode p50), and NFKB2 (encode p52) (38,39). Therefore, we analyzed the effect on the expression of NFκB target genes in eff Treg after αCD3/CD28 stimulation with and without supplementation of rhTNFα or anti-TNF. Supplementation with rhTNFα led to a significant increase of NFKBIA and NFKB2 expression, indicating that TNFα signaling promotes the expression of NFκB target genes, an indication of NFκB activation during Treg activation, while anti-TNF suppressed the NFκB pathway ( Figure 2C). We previously found that TNFα signaling enhanced TNFAIP3 (tumor necrosis factor-induced protein 3) expression in conventional T cells (20). TNFAIP3 encodes the ubiquitinediting enzyme A20, which in turn regulates NFκB activity. Here, we also observed that TNFα signaling regulated TNFAIP3 expression in eff Treg ( Figure 2D). To demonstrate causality between suppression of TNFAIP3 and enhanced expression of IL-17A, we carried out a small interfering RNA assay (siRNA) to inhibit TNFAIP3 transcription. siRNA-mediated TNFAIP3 inhibition of eff Treg, similar to TNFα signaling inhibition by anti-TNF treatment, led to enhanced IL-17A gene expression (Figures 2E,F).
As TNFα can bind to both TNFR1 and TNFR2, we measured the expression of these receptors on freshly isolated eff Treg and demonstrated that they expressed TNFR2, but TNFR1 was hardly detected ( Figure 2G). The latter agrees with previous studies (20,40) and suggests that TNFα-mediated regulation of IL-17A expression in eff Treg might be primarily mediated via the TNFR2. To examine this, αCD3/CD28stimulated eff Treg were cultured in the absence and presence of a specific TNFR2 agonist for 5 days. TNFR2 agonist stimulation led to a reduction in the percentages of IL-17A expressing FOXP3+ cells ( Figure 2H). This indicates that IL-17A expression in eff Treg subsets is regulated via TNFα-TNFR2 signaling. Together, these data suggest that TNFα signaling via TNFR2 promotes the expression of the anti-inflammatory mediator TNFAIP3/A20, which seems to prevent IL-17A expression in regulatory T cells, as ablation of TNFα signaling suppresses TNFAIP3/A20 and results in increased IL-17A expression in human Treg.

TNFα Suppresses T-Cell Receptor and Janus Kinase Protein Activity and Regulates IL-17A Expression in Effector Regulatory T Cells
TNFAIP3/A20 has been demonstrated to regulate critical proteins involved in TCR (23), TNFα (41), IL-17R (21), and Wnt signaling (20,42). Recently, we demonstrated that the prevention of TNFα signaling in conventional CD4 + memory T cells leads to inhibition of TNFAIP3/A20 expression, which subsequently leads to enhanced IL-17A expression (20). TNFAIP3/A20 has been shown to regulate kinase activity (21,23). To better understand kinase regulation by TNFα signaling in eff Treg, we here profiled the activity of ∼300 kinases in FACS-sorted eff Treg following stimulation with αCD3/CD28 beads in the absence or presence of anti-TNF or rhTNFα. Subsequently, we analyzed the threonine/serine and tyrosine kinase activity using a multiplex human kinase activity array. This kinome array employs ∼300 peptide substrates with known phosphorylation sites and provides a reliable and high-throughput kinase profiling tool for further pathway elucidation (see Materials and Methods) (43). We found 30 unique and differentially activated kinases following anti-TNF vs. rhTNFα supplementation comparison ( Figure S2). For the kinase activity profiling, we focused on Treg that were stimulated with αCD3/CD28 beads plus rhIL-2 with or without TNFR2 agonist for 5 days (n = 9). All data are shown as mean ± SEM. For statistical analysis, a Friedman test followed by Dunn's multiple comparison test (A), a two-way ANOVA followed by a Bonferroni posttest (B,C), and a Wilcoxon matched-pairs signed-rank test (D,G,H) were used. *p < 0.05, **p < 0.01,***p < 0.001, ****p < 0.0001, ns, not significant. the two most extreme states of TNF pathway signaling and addressed the differential kinase activity profile following eff Treg activation following TNF vs. anti-TNF supplementation. The obtained kinome data were visualized using a volcano plot that shows the fold change of kinase activity and the associated level of significance (p-values) ( Figure 3A, left panel; raw data Table S1). We found that inhibition of TNFα signaling, as compared to the supplementation of rhTNFα, in activated eff Treg significantly promoted the activity of multiple kinases (red symbols indicate p < 0.05). The ranked log2-fold changes of kinase activity are shown in the right panel of Figure 3A. Notably, several of the kinases were related to TCR signaling [CD3ζ (CD247), CD3ε, ZAP70, and Lck] (44). Also, cell cycle regulating (CALM, CD28, GSK3B, MAPK3, PGR, and JAK3) (45,46) and apoptosis (ANXA2, Annexin V) (47)-related kinases were induced.
To obtain a more comprehensive understanding of the kinase network and cellular pathways regulated by neutralization of TNFα, the kinases that were significantly activated following anti-TNF mAb treatment were analyzed using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins). STRING is a web-based biological resource (https://string-db.org) of known and predicted protein-protein interactions enabling prediction of the functional protein association network of a group of given proteins by estimating the likelihood of meaningful biological interactions (48). In our analysis, we used the highest confidence interaction score (0.900) to associate all kinases that were significantly activated following anti-TNF treatment as listed in the right panel of Figure 3A. STRING association analysis demonstrated that inhibition of TNFα signaling in activated eff Treg involved prominent immune signaling pathways such as the PKC, p38-MAPK, and JAK pathways, which were all linked to TCR signaling [CD3ζ (CD247) and CD3ε] ( Figure 3B). Previously, these pathways were shown to be associated with the induction of IL-17A expression (49)(50)(51)(52).
To validate if the predicted pathways were indeed involved in rhTNFα-induced suppression of IL-17A expression in eff Treg, FACS-sorted eff Treg were activated in the presence or absence of anti-TNF and specific kinase inhibitors of JAK/STAT (Tofacitinib), PKC (AEB071, Sotrastaurin), or p38 MAPK (UR13870). For the inhibition of TCR signaling, an Lck inhibitor (A420983) was applied. We demonstrated that suppression of JAK, and Lck kinases, but not PKC and p38, prevented the expression of IL-17A expression in eff Treg that were activated under TNFα signaling inhibiting or not (Figure 3C). In fact, suppression of JAK and Lck inhibited the expression of IL-17 similar to the TNF supplementation condition. The inhibitors tested did not affect FOXP3 expression ( Figure S3).
Next, we performed a functional ontology enrichment analysis of the most significant biological process networks, processes, and diseases by submitting the kinase data that we identified in activated eff Treg following supplementation vs. inhibition of TNFα to MetaCore TM database analysis. Significant enriched MetaCore TM GO process networks involved immune response-TCR signaling, cell cycle regulation, and lymphocyte proliferation ( Figure 4A). The most significantly enriched MetaCore TM Go processes based on the submitted kinases were kinase signaling pathways via transmembrane receptor protein tyrosine, signal transduction processes, and tyrosine phosphorylation and modification ( Figure 4B). Furthermore, there was an enrichment of cell communication and cell development processes. MetaCore TM Go diseases indicated a strong enrichment of autoimmune disease, next to other pathological conditions ranging from the nervous system, nutritional, and metabolic disorders ( Figure 4C). Together, these data demonstrated that CD3 and CD28 activation of eff Treg in the absence of TNF-signaling by anti-TNF treatment promotes tyrosine kinase activity of relevant TCR-associated signaling pathways.

DISCUSSION
Human Treg can express the pro-inflammatory cytokine IL-17A under specific conditions; a phenomenon referred to as Treg plasticity (5,8). The molecular mechanisms regulating this phenomenon are not well-understood. In our current work, we demonstrate that TNFα signaling regulates IL-17A expression in eff Treg by controlling a kinase activity network that includes TCR linked kinases and other prominent immune signaling kinase pathways such as the JAK pathway. Also, TNFα-mediated regulation of the anti-inflammatory mediator TNFAIP3/A20 appeared crucial to control IL-17A expression by eff Treg. TNFR2 is the main receptor for TNFα signaling in Treg. TNFR2 stimulation has been demonstrated to support Treg stability (18,19,25,53), whereas the effect of TNF signaling on the stability of Treg is ambiguous (54,55). Here, we show that TNFR2 is highly expressed on human eff Treg, and TNF-TNFR2 signaling in eff Treg acts as a negative regulator of IL-17A expression by controlling TCR and JAK signaling.
STRING association analysis revealed that inhibition of TNFα signaling is associated with increased TCR associated signaling of CD3ζ, CD3ε, ZAP70, and Lck, indicating that TNFα signaling in eff Treg functions as a rheostat of TCR signal transmission. Although information of TNFα stimulation on the TCR signaling in Treg is lacking, it has been shown in CD4+ T cells of both mice and man that TNFα stimulation results in specific down-regulation of TCRζ expression and impaired TCR/CD3 signaling, including phosphorylation of the TCRζ, CD3ε, ZAP-70 tyrosine kinase, and linker for activation of T cells (LAT) (56). TCR signaling is essential for both effector and regulatory T cells (57). Treg have a more extensive TCR repertoire than effector T cells, and TCR signaling is crucial for proper Treg function (58)(59)(60)(61). Signaling via the T cell antigen receptor of Treg is critical for FOXP3 expression and their suppressive activity. Mutations resulting in signaling-deficient TCRζ chains led to increased Treg numbers with higher suppressive activity (62)(63)(64). Reduced TCR signaling will alleviate downstream signaling and favor Treg cell lineage commitment. TNFα signaling, as we demonstrate here, seems to safeguard TCR-related kinase activity in eff Treg and stabilize Treg function as illustrated by preventing IL-17A expression. Note that anti-TNF had a mild effect on the induction of IL-17A expression in eff Treg, which is in contrast to its clear induction of IL-17A in conventional memory T cells (20). This phenomenon may be caused by the poor intrinsic  Pathway inhibition validation assays applying small chemical molecules in the stimulation assay as described above (mean ± SEM, n = 7). JAKi, JAK inhibitor (tofacitinib); Lcki, Lck inhibitor (A420983); PKCi, PKC inhibitor (AEB071); and p38i, p38MAPK inhibitor (UR13870). ANOVA Dunnett's testing (A) and Friedman test followed by Dunn's multiple comparisons test (C) were used. *p < 0.05, **p < 0.001, ***p 0.0001, ns, not significant.
capacity of eff Treg to produce TNFα in vitro. In fact, highly pure FACS-sorted eff Treg barely produced TNFα (41.35 pg/ml ± 6.75), whereas memory conventional T cells produced significantly higher levels (335.7 pg/ml ± 65.33, n = 4) (data not shown). Next to TCR-derived signals, Treg integrates inputs from cytokine, chemotactic, and metabolic cues to fulfill their function optimally. Proximal cytokine signaling often takes place via JAK-STAT signaling (65). IL-17A gene transcription is associated with JAK-STAT3 signaling (66). Inhibition of TNFα signaling using anti-TNF inhibitor ETN was associated with increased JAK1 and JAK3 kinase activity in αCD3/CD28 stimulated eff Treg. Inhibition of JAK1 and JAK3 kinase activity by the clinically applied JAK inhibitor tofacitinib prevented IL-17A expression in anti-TNF-treated eff Treg, suggesting that TNFα signaling is involved in driving JAK/STAT signaling. Although TNFα is not a prototypic JAK/STAT activating cytokine, the anti-inflammatory molecule A20 (encoded by TNFAIP3) that is a downstream target of TNFα signaling acts as a regulator of STAT (67,68). The absence of A20 in myeloid cells resulted in enhanced STAT1dependent inflammation (68). This relationship needs to be confirmed in eff Treg.
Although anti-TNF therapy is improving the life quality of many patients with chronic inflammatory diseases, 10-20% of patients do not respond to the treatment while 0.6-5% of patients treated with TNF inhibitors reveal paradoxical immunemediated inflammatory side effects (36,37). Although the mechanism of the latter phenomenon is not fully understood, it might be of interest to consider an additional JAK inhibitor treatment such as tofacitinib or other JAK inhibitors to prevent the putative IL-17A expression by Treg. Also, regarding Tregbased immune therapy in transplantation or autoimmunity, the clinical design has started to consider strategies to minimize the risks of Treg plasticity (69) at the time of ex vivo production and following in vivo administration (70,71). Our results suggests that TNFα-TNFR2 signaling or inhibition of JAK signaling might favor Treg stability. Along with this line of reasoning, it has been demonstrated that JAK inhibition (72) as well as TNFR2 stimulation (18,19) support human Treg function and prevent Treg plasticity.
In conclusion, we demonstrated an inverse production of TNFα and IL-17A between human naïve and effector Treg cells. Supplementation of rhTNFα led to a down-regulation in the frequency of IL-17A-producing eff Treg, mainly via the activation of NFkB pathway as well as the up-regulation of TNFAIP3/A20 expression. TNFR2 receptor seems to play a crucial role since we hardly detected any expression of TNFR1 on eff Treg and treatment of eff Treg with TNFR2 specific agonist resulted in a similar inhibition of IL-17A production. Accordingly, inhibition of TNFα signaling using the clinically applied anti-TNF inhibitor ETN led to decreased TNFAIP3 and increased IL-17A expression, a phenomenon similar to what is observed in human conventional memory CD4+ T cells. Kinome activity screening of αCD3/CD28 stimulated eff Treg revealed that anti-TNF led to an increase in kinase activity of multiple kinases including CD3ζ (CD247) and LcK. A functional ontology enrichment analysis indicated that these kinases were highly associated with different immune response signaling pathways including TCR-, JAK-mediated pathways. We propose that these findings might be relevant for optimizing anti-TNFbased therapy and may aid in preventing Treg plasticity in case of Treg-based cell therapy.

Study Approval
The protocols of this study were performed in agreement with the Declaration of Helsinki and in accordance with the Radboud university medical center (Radboudumc) in Nijmegen, the Netherlands.

Subjects
Blood buffy coats from voluntary donors were purchased from the Sanquin Blood Bank, Nijmegen, the Netherlands. The volunteers gave written informed consent.  Figure 1A. The purity of the sorted cell populations was 95.3 ± 4.1% (mean ± SD).

RPMI-1640
Dutch modified (Gibco, Massachusetts, United States) culture medium, containing sodium bicarbonate FIGURE 4 | Enrichment analysis of the kinome array data. Functional ontology enrichment analysis using the MetaCore TM database reveals (A) the distinct biological networks, (B) the different biological processes, and (C) various diseases related to kinases identified in Figure 3. The probability of a random intersection between the set of kinases with ontology entities was estimated with the "p" value of the hypergeometric intersection. A lower "p" value means higher relevance of the entity to the dataset, which appears in higher rating for the entity.

Small Interfering RNA Transfection
For small interfering RNA (siRNA) knockdown of TNFAIP3, Accell SMARTpool siRNA (Dharmacon, Colorado, United States) was used according to the manufacturer's instructions. Briefly, 1 × 10 5 eff Treg cells per well were stimulated with αCD3/CD28 beads (1:5 of bead:cell ratio) in Accell Delivery Medium (Dharmacon) supplemented with rhIL-2 (100 U/ml) and incubated with 1 mmol cyclophilin B siRNA (positive control), or 1 mmol non-targeting control siRNA, or 1 mmol TNFAIP3 siRNA for 120 h (for siRNA sequences, see Table S2). Quantitative real-time PCR (RT-qPCR) was performed to confirm the knockdown of the target gene expression.

RT-qPCR
Total RNA was extracted by using the RNeasy Plus Micro Kit (Qiagen) followed by cDNA synthesis using the SuperScript III First-Strand Synthesis System and Oligo(dT)20 primer (Thermo Fisher Scientific, Massachusetts, United States). TaqMan gene expression assays were purchased from Thermo Fisher Scientific (Table S3). RT-PCR was acquired in a 7500 Real-Time PCR System (Applied Biosystems). RT-qPCR cycle values (C T ) obtained for specific mRNA expression in each sample were normalized to the C T values of human HPRT1 (endogenous control), resulting in C T values (log ratio of the gene concentrations) that were used to calculate the relative gene expression. C T= Mean CT − Housekeeping gene Mean CT Then, we performed an exponential conversion of C T , namely, 2 − CT using the following formula: 2 ∧ (exponential) − ∆CT 2 − CT representing the relative gene expression was used in Figures 1B,E,F. eff Treg stimulated in the absence of anti-TNF or rhTNFα were used as a baseline to calculate the relative gene expression in fold change ( C T ) for eff Treg stimulated in the presence of rhTNFα vs. ETN treatment.
Subsequently, we performed an exponential conversion of ∆∆C T , namely, 2 − CT using the following formula: representing the relative gene expression in fold change was employed for Figures 2B-D. In Figures 1C,D, log10 C T was employed. The Relative Quantification app (Thermo Fisher Scientific cloud) was used for data analysis.

Measurement of Cytokines Secretion
The cell culture supernatants were analyzed for the presence of IL-17A, IFNγ, and TNFα using Bio-Plex Pro Human Th17 Cytokine Assays (Bio-Rad, California, United States) according to the manufacturer's instruction. The cytokine concentrations were measured using a Luminex 100 machine (Luminex Corp., Texas, United States). The lowest limit of detection was <1.870 pg/ml for IL-17A, <2.411 pg/ml for IFNγ, and <2.231 pg/ml for TNFα.

Protein Kinase Chip Assay
After sorting and stimulations of cells, samples were frozen for further analysis. The protein isolation was performed according to the manufacturer's instruction (P1160, PamGene International B.V., 's-Hertogenbosch, the Netherlands). Kinase activity was measured with PamGene's Protein Tyrosine Kinase (PTK) PamChip (Cat. number 86402) and Serine Threonine kinase (STK) PamChip (Cat. number 87102). Each PTK PamChip array contains 196 peptides immobilized on a porous membrane, whereas each STK PamChip array contains 144 peptides (see the full list of peptides at www.pamgene.com). The peptide sequences (13 amino acids long) harbor phosphorylation sites, defined based on literature or derived from computational predictions and are correlated with one or multiple upstream kinases. A fluorescently labeled anti-phospho-Tyr antibody (PY20) is used to detect the phosphorylation activity of tyrosine kinases present in the sample. For the STK assay, an antibody mix is used to detect the phosphorylated Ser/Thr, and the 2nd FITC-conjugated antibody is used in a detection mix to quantify the phosphorylation signal. BioNavigator software 6.3 (PamGene) was used to determine signal intensities, peptide quality control (QC) and preselection (phosphorylation kinetics, or increase in signal over time, in 25% of the arrays analyzed), Log 2 transformation, ANOVA-Dunnett's testing, and data visualization. Mapping and pathway elucidation analysis were performed using METACORE TM (Clarivate Analytics, PA, USA) and STRING (73). As described by the GeneGo manufacturer's report, the analysis consists in matching the protein IDs of possible targets for the "common, " "similar, " and "unique" sets with protein IDs in functional ontologies in MetaCore (73). The lower p-value means a higher relevance of the entity to the dataset, which shows a higher rating for the entity.

Statistics
Statistical analysis was performed using GraphPad Prism 5.0 for Windows (GraphPad Software, San Diego, California, USA) and R. For experiments with more than two groups of matched samples, we used non-parametric Friedman test followed by Dunn's Multiple Comparison Test, whereas for experiments with only two groups of matched samples, we employed nonparametric Wilcoxon matched-pairs signed-rank test.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Radboud university medical center (Radboudumc) in Nijmegen, the Netherlands. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
PU, IJ, and HK designed the research. PU, OF, XH, HT, and BH performed the experiments. PU, OF, IJ, BH, XH, RS, and HK analyzed the data. PU, XH, IJ, and HK prepared and wrote the final manuscript. All the authors reviewed the paper.

ACKNOWLEDGMENTS
PU and OF were supported by a scholarship provided by the Brazilian mobility program Science Without Borders. We thank PAMgene especially Rob Ruijtenbeek and Dirk Pijnenburg for supporting the kinome analysis. We also thank Roslyn Kemp from the University of Otago, Department of Microbiology and Immunology, who helped review the manuscript.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.03047/full#supplementary-material Figure S1 | Expression of IL-17A in Conventional and regulatory T cells. Sorted naïve T cells (CD4 + CD45RA + CD25 − ), memory T cells (CD4+CD45RA − CD25 − ), naïve Treg (CD4 + CD45RA + CD25 low ) and effector Treg (CD4+CD45RA − CD25 high ) were stimulated with anti-CD3/CD28 bead plus rhIL-2. Cells were harvested on day 5 and intracellular FOXP3 and IL-17A expression were detected by FACS staining. Figure S2 | Kinome data analysis. Kinase activity of eff Treg following αCD3/CD28 bead plus rhIL-2 activation in the absence or presence of anti-TNF or rhTNFα. Significant changes of kinase activities are presented in a Venn diagram (A, left panel) and a bar plot that ranked based on the log2 fold-change of kinase activities. (B,C) Bar graphs showing significant changes of kinase activity between anti-TNF and αCD3/CD28 control (B) or rhTNF and αCD3/CD28 control (C). Figure S3 | The Janus kinase, Lck, PKC and p38 MAPK inhibitors do not affect FOXP3 expression in eff Treg. eff Treg were stimulated with αCD3/CD28 beads in the presence or absence of rhTNFα or anti-TNF or small chemical molecules such as JAK inhibitor (tofacitinib), Lck inhibitor (A420983), PKC inhibitor (AEB071) and p38MAPK inhibitor (UR13870) for 5 days. Flow cytometry analysis of intracellular FOXP3 expression (n = 5). Data are shown as mean ± SEM.