Strong basal/tonic TCR signals are associated with negative regulation of naive CD4⁺ T cells

Abstract

T cells experience varying intensities of tonic or basal TCR signaling in response to self-peptides presented by MHC (self-pMHC) in vivo. We analyzed four subpopulations of mouse naive CD4⁺ cells that express different levels of Nur77-GFP and Ly6C, surrogate markers that positively and inversely correlate with the strength of tonic TCR signaling, respectively. Adoptive transfer studies suggest that relatively weak or strong Nur77-GFP intensity in thymocytes tends to be maintained in mature T cells. Two-dimensional affinity measurements were lowest for Nur77-GFP^LO Ly6C⁺ cells and highest for Nur77-GFP^HI Ly6C⁻ cells, highlighting a positive correlation between apparent TCR affinity and tonic TCR signal strength. Despite experiencing the strongest tonic TCR signaling, Nur77-GFP^HI Ly6C⁻ cells were least responsive to multiple concentrations of a cognate or suboptimal pMHC. Gene expression analyses suggest that Nur77-GFP^HI Ly6C⁻ cells induce a gene expression program that has similarities with that of acutely stimulated T cells. However, strong tonic TCR signaling also correlates with increased expression of genes with inhibitory functions, including coinhibitory receptors. Similarly, ATAC-seq analyses suggested that increased tonic TCR signal strength correlated with increased chromatin accessibility associated with genes that have positive and inhibitory roles in T cell activation. Strikingly, Nur77-GFP^HI Ly6C⁻ cells exhibited differential accessibility within regions of Cd200r1 and Tox that were similar in location to differentially accessible regions previously identified in exhausted CD8⁺ T cells. We propose that constitutive strong tonic TCR signaling triggers adaptations detectable at both the transcriptional and epigenetic levels, ultimately contributing to the tuning of T cell responsiveness.

INTRODUCTION

T cells are positively selected for expressing TCRs that weakly recognize self-peptide antigens and MHC (self-pMHC) (1). TCR:self-pMHC interactions stimulate “tonic” or “basal” TCR signaling in mature T cells in vivo (2), resulting in constitutive tyrosine phosphorylation of the TCR complex and association of the tyrosine kinase ZAP-70 with the TCR ζ chain (3, 4). It has been appreciated that the strength of tonic TCR signaling influences T cell activation, though the mechanisms underlying this phenomenon are incompletely understood (5, 6).

Studies of variations in tonic TCR signaling are enabled by surrogate markers of TCR signaling. Such indicators include reporter transgenes of Nr4a1 (encoding Nur77), which are rapidly expressed upon TCR stimulation (7, 8). Upregulation of Nur77-GFP reporter transgene expression is detected in response to self-pMHC during positive selection in the thymus (7–9). Phenotypically naive CD44^LO CD4⁺ T cells located in the spleen and lymph nodes constitutively express Nur77-GFP in MHCII sufficient, but not MHCII-deficient hosts, suggesting that basal Nur77-GFP expression depends on continuous exposure to self-pMHC II (7, 10). We showed previously that the intensity of basal TCR ζ chain phosphorylation is lowest in Nur77-GFP^LO T cells and highest in Nur77-GFP^HI T cells, suggesting that Nur77-GFP fluorescence intensity can also reflect the intensity of tyrosine phosphorylation of the TCR complex (10). Our previous work also showed that the fluorescence intensity of Nur77-GFP is sensitive to variations in TCR affinity for pMHC and to inhibition of ZAP-70, a tyrosine kinase required for activation of the major signal transduction pathways downstream of the TCR (9, 11, 12). Together, these findings suggest that Nur77-GFP intensity can reflect relative differences in the strength of TCR signal transduction experienced by T cells in response to self- or foreign pMHC.

Previous work suggests that Ly6C can also be used as a marker of reactivity to self-pMHC, although its expression inversely correlates with tonic TCR signal strength (13). In combination, Nur77-GFP and Ly6C enable the visualization of a broad range of tonic TCR signal strengths that are experienced by individual naive T cells. Our previous studies investigated the activation of subpopulations of cells across the entire spectrum of Nur77-GFP and Ly6C expression. In those studies, naive Nur77-GFP^HI Ly6C⁻ cells generated fewer IL-2 secreting cells and diminished proliferative responses relative to Nur77-GFP^LO Ly6C⁺ cells. In the present study, we find that strong tonic TCR signaling is relatively stable and correlates with the relative affinity to pMHC. Cells that experience the strongest tonic TCR signals exhibit gene expression and chromatin accessibility patterns with features similar to both activated T cells and hypo-functional (anergic, exhausted) T cells. These results support a model whereby naive CD4⁺ T cell responsiveness to TCR stimuli is tuned in response to the strength of tonic TCR signaling.

MATERIALS AND METHODS

Mice

A mouse strain with the Tg(Nr4a1-EGFP)GY139Gsat transgene, referred to as Nur77-GFP here, has been previously described (8). A compound mouse strain with the Nur77-GFP transgene and the targeted Foxp3-IRES-RFP reporter allele, Foxp3tm1Flv/J, has been previously described (10). Mouse strains with a combination of the OT-II TCR transgene (Tg(TcraTcrb)425Cbn), Nur77-GFP transgene, Foxp3-IRES-RFP reporter allele, with or without the TCRα chain knockout allele Tcra^tm1Mom, have been described (10). A strain with the combination of the Tg(TcrAND)53Hed transgene (AND TCR transgene), Nur77-GFP transgene, and Foxp3-IRES-RFP targeted mutation has been described (10). A strain with the combination of the Zap70*M413A)2Weis transgene (referred to here as ZAP-70^AS) and the endogenous Zap70 targeted null mutation Zap70^tm1Weis and Nur77-GFP transgene has been described previously (9). WT CD45.2⁺ C57BL/6 mice were purchased from the Jackson Laboratory. The mice used in these studies were housed in the Division of Animal Resources at Emory University. Experiments were performed according to protocols approved by the Emory University Institutional Animal Care and Use Committee.

Antibodies and cellular analysis

Thymocyte and T cell samples were analyzed using BD LSR Fortessa or FACS Symphony analyzers. Antibodies used for flow cytometry analysis were from either BD: CD4 [clone RM4-5]; CD44 [IM7]; CD127 [SB/199]; CD200 [OX-90]; Bcl6 [K112-91]; Helios [22F6]; Ly6C [AL-21]; TCR Va2 [B20.1]. Biolegend: CD5 [53-7.3]; CD25 [PC61]; CD45.1 [A20]; CD45.2 [104]; CD45RB [c363-16A]; CCR7 [4B12]; PD-1 [29F.1A12]; TCRβ [H57-597]; MHCI (H-2K^b) [AF6-88.5]; eBioscience: CD8α [53-6.7]; CD69 [H1.2F3]; FR4 [eBio12A5]. Cell viability was assessed by exclusion of LIVE/DEAD fixable yellow or near-IR stains according to the manufacturer’s protocol. Intracellular staining was performed after fixation and permeabilization using the Foxp3 staining kit according to the manufacturer’s protocol (Thermo Fisher). For staining of splenic red pulp cells, 3 μg anti-CD45.2 APC antibody was injected intravascularly 3 minutes prior to euthanasia, as described (14). Total thymocytes were pre-enriched for CD4 single positive (CD4SP) thymocytes by negative selection of CD8⁺ cells by anti-CD8-biotin antibodies and Streptavidin magnetic beads according to the manufacturer’s protocol (Stem Cell Technologies).

Cell sorting and adoptive transfers

CD4SP cells were enriched from the thymi of Nur77-GFP mice, followed by FACS sorting for CD4⁺ CD8⁻ Nur77-GFP^LO or Nur77-GFP^HI populations using a FACSAria sorter (BD Biosciences). The sort gates were set on the highest and lowest 15% of cells based on GFP fluorescence intensity. Between 5 x10⁵ and 7.5 x 10⁵ sorted CD4SP thymocytes were adoptively transferred into congenic WT CD45.1⁺ hosts by intravenous injection. Naive CD4⁺ T cells were sorted from the spleens and lymph nodes of OT-II TCR transgenic mice with the phenotype CD4⁺ CD44^LO CD62L^HI Foxp3-RFP-negative, and further sorted into Populations A (Nur77-GFP^LO Ly6C⁺); Population B (Nur77-GFP^MED Ly6C⁺); Population C (Nur77-GFP^MED Ly6C⁻); and Population D (Nur77-GFP^HI Ly6C⁻). A total of 2 x 10⁵ sorted Population A-D cells were transferred into WT congenic CD45.2⁺ mice by intravenous injection.

T cell stimulation

Sorted Population A-D OT-II cells were stimulated with T cell-depleted splenocytes and chicken ovalbumin 323-339 (OVA^323-339) peptide (ISQAVHAAHAEINEAGR) or an OVA^H331R altered peptide (ISQAVHAARAEINEAGR) (Genscript). ZAP-70 analog-sensitive (ZAP-70^AS) T cells were stimulated with T cell depleted splenocytes and 0.1 μg/ml soluble anti-CD3 (clone 145-2C11, Biolegend). Cells were cultured in the presence of varying concentrations of the inhibitor HXJ42, kindly provided by Dr. Kevan Shokat, University of California, San Francisco (9). For in vivo stimulation, sorted CD45.1⁺ OT-II Population A-D cells were labeled with CellTrace Violet (Thermo Fisher) and 6 x 10⁵ cells were adoptively transferred into WT CD45.2⁺ hosts by intravenous injection. The following day, WT splenocytes were pulsed ex vivo with 100 μM OVA^H331R peptide at 37C for 1 hour and washed with PBS. A total of 5 x 10⁶ cells were transferred by intravenous injection into the hosts containing OT-II cells. Cell Trace dye dilution was detected by flow cytometry 3 days after injection of the peptide-pulsed splenocytes.

2-dimensional micropipette adhesion frequency assay (2D-MP)

CD4⁺ cells were pre-enriched from the spleen and lymph nodes of OT-II TCRα-deficient Nur77-GFP Foxp3-IRES-RFP mice with CD4⁺ T cell negative selection kits according to the manufacturer’s protocol (Miltenyi). Naive CD44^LO CD62L^HI RFP⁻ Population A-D cells were sorted from enriched CD4⁺ cells as described above. In these experiments, the T cells were not labeled with anti-CD4 antibodies, which may affect the affinity measurement assay. The relative 2-dimensional affinities (A_c K_a μm⁴) were measured using the previously described 2D-MP assay (15–17). Briefly, human RBCs were coated with I-A^b monomers loaded with WT OVA^328-337 peptide, which were obtained from the NIH Tetramer Core Facility. Quantification of binding events, pMHC density TCR surface density and TCR:pMHC affinity calculations were calculated as previously described (15–17).

RNA-seq and ATAC-seq libraries

RNA-seq libraries were prepared from 1 x 10⁵ CD4⁺ CD44^LO RFP⁻ Population A-D cells from the spleen and lymph nodes of Nur77-GFP Foxp3-IRES-RFP mice. Cells were sorted directly into RLT lysis buffer (Qiagen) with 1% β-mercaptoethanol, sequenced, quality checked, and mapped to the mm10 genome as previously described (18). Genes expressed at 3 reads per million in all samples of one group were considered expressed. ATAC-seq libraries were prepared from 2 x 10⁴ CD4⁺ CD44^LO RFP⁻ Population A-D cells from the spleen and lymph nodes of Nur77-GFP Foxp3-IRES-RFP mice, sequenced, and mapped to the mm10 genome as previously described (18).

RNA-seq and ATAC-seq analysis

Raw fastq files were mapped to a custom mm10 genome containing the GFP sequence using STAR (19) with the Gencode vM17 reference transcriptome. PCR duplicate reads were marked with PICARD MarkDuplicates and removed from downstream analyses. Reads mapping to exons for all unique ENTREZ genes and GFP was summarized using GenomicRanges (20) in R v3.5.2 and normalized to reads per kilobase per million. Genes expressed at 3 reads per million in all samples of one group were considered expressed. Differentially expressed genes were determined using edgeR (21), and genes that displayed an absolute fold change > 1.5 and Benjamini-Hochberg false-discovery rate (FDR) corrected p-value < 0.05 were considered significant.

Raw ATAC-seq reads were mapped to the mm10 genome using Bowtie2 (22), enriched accessible regions for each sample determined using MACS2 (23). A composite set of unique peaks called across all samples was annotated to the nearest gene using HOMER (24), the read depth from each sample annotated using the GenomicRanges package (20), and data normalized to reads per peak per million (rppm) in R v3.5.2. Differentially accessible regions were determined using edgeR (21). Regions that displayed an absolute fold change > 2.0 and Benjamini-Hochberg false-discovery rate (FDR) corrected p-value < 0.05 were considered significant. Nur77 binding site accessibility was computed by first identifying the location of all Nur77 motifs using the annotatePeaks.pl [DAR file] mm10 -size given -noann -m nur77.motif -mbed nur77.motifs.bed HOMER command. Next, coverage at the 50bp surrounding each motif was calculated for all samples and the group mean for each motif summarized as a box plot using custom R scripts. Box plots were generated using the boxplot function in R. Taiji analysis was performed using the standard workflow (25).

RESULTS

A broad range of tonic TCR signal strength

In these studies, we analyzed four subpopulations of naive CD4⁺ cells: Population A (Nur77-GFP^LO Ly6C⁺), Population B (Nur77-GFP^MED Ly6C⁺), Population C (Nur77-GFP^MED Ly6C⁻), and Population D (Nur77-GFP^HI Ly6C⁻) (Fig. 1A). These four subpopulations were detected within the naive CD4⁺ cell population, as defined by a CD44^LO CD25⁻ phenotype that excludes antigen-experienced cells and Tregs. The relative expression pattern was skewed differently in antigen-experienced CD44^HI cells, with the majority being Ly6C⁻. CD25⁺ cells, a population that s enriched for Tregs, included Ly6C⁻ and Ly6C⁺ populations, both of which were skewed toward high Nur77-GFP expression (Fig. 1B). Populations A-D were also detected in naive CD4⁺ cells as defined by a CD45RB^HI phenotype (Fig. 1C). Population A-D cells were also detected using the combination of CD44 and CD62L to define the naive T cell population (10). The percentages of Population A-D cells were similar in inguinal/axial lymph nodes, splenic red and white pulp, and peripheral blood (Fig. 1D). Population D was the only subpopulation that exhibited a small but consistent decrease in peripheral blood relative to lymph nodes and spleen. This finding is consistent with previous studies indicating weaker tonic TCR signaling in T cells harvested from peripheral blood relative to T cells from lymph nodes (3).

Figure 1.

Strong tonic TCR signal strength is relatively stable. A) Schematic shows the gating scheme for Populations A-D. B) Contour plot on the left shows CD44 and CD25 expression by viable CD4⁺ cells from the spleen of Nur77-GFP mice. The contour plots on the right show the expression of Nur77-GFP and Ly6C for antigen-experienced (CD44^HI CD25⁻), naive (CD44^LO CD25⁻), and Treg (CD25⁺) populations. Numbers indicate the percentage of cells within each color-coded gate. C) Histogram on the left shows the expression of CD45RB by viable CD4⁺ cells. CD45RB^LO and CD45RB^HI gates are shown. and the contour plots on the right show the expression of Nur77-GFP and Ly6C by CD45RB^LO and CD45RB^HI cells. Numbers indicate the percentage of cells within each gate. For experiments in panels B-C, n = 3 independent experiments. D) Bar graph shows the mean percentage ± SEM of Population A-D cells in the Red (RP) and White (WP) pulp of the spleen; Inguinal plus axial lymph nodes (LN); and peripheral blood (PB). Red pulp cells were identified by positive staining and white pulp cells were identified by the lack of staining following intravenous injection of Nur77-GFP Foxp3-IRES-RFP mice with anti-CD45.2 antibodies 3 minutes prior to euthanasia. N= 3 independent experiments. E) Contour plot shows CD4 and CD8 expression by total thymocytes from a Nur77-GFP Foxp3-IRES-RFP mouse. Histograms show Nur77-GFP expression by CD4 CD8 double positive (DP) cells, CD4 single positive (CD4SP) cells, and CD8 single positive (CD8SP) cells. F) Contour plot on the left shows TCRβ and CCR7 expression by total thymocytes from a Nur77-GFP mouse. The arrow indicates that the contour plot to the right is gated on the TCRβ⁺ CCR7⁺ population. The contour plot in the middle shows CD69 and MHCI expression on the TCRβ⁺ CCR7⁺ populations. CD69⁺ MHC-I⁻ semi-mature (SM), CD69⁺ MHC-I⁺ Mature 1 (M1), and CD69⁻ MHC-I⁺ populations are indicated. Right, Histograms show Nur77-GFP expression by the indicated populations. For experiments in panels E-F, n= 3 independent experiments. G) Histograms show Nur77-GFP expression by total CD4SP thymocytes (top), and FACS-purified Nur77-GFP^LO and Nur77-GFP^HI CD4SP cells (bottom) from Nur77-GFP mice. H) Bar graph shows the mean percentage ± SEM of donor cells within the Population A-D gates 2 weeks post-adoptive transfer. For experiments in panels G-H, n = 3 independent experiments. Statistical significance, * P < 0.05; ** P < 0.005; **** P < 0.0001; n.s. = not significant, Student’s t-Test. I) Contour plot shows Nur77-GFP and Ly6C expression by naive CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ cells. Numbers indicate the percentage of cells within the Population A-D gates. J) Graph shows the 2D TCR affinities of naive OT- II Population A-D cells for WT OVA^228-337 peptide/I-A^b monomers as measured by 2-dimensional micropipette adhesion frequency assays. Each symbol represents an individual measurement and cell. Data from 3 independent experiments, n = 5 mice. Statistical significance, * P < 0.05, **** P < 0.0001, n.s. = not significant. Ordinary one-way ANOVA Tukey multiple comparison test.

We next analyzed the range of basal Nur77-GFP expression in thymocytes. The mean fluorescence intensity of Nur77-GFP is greater in CD4 single-positive (CD4SP) and CD8 single-positive (CD8SP) thymocytes relative to CD4 CD8 double positive (DP) cells (Fig. 1E) (7, 8). Within the CD4SP and CD8SP populations, Nur77-GFP fluorescence intensity spanned more than 2 orders of magnitude, suggesting that a broad range of Nur77-GFP expression is possible at a single cell level after positive selection. Semi-mature (SM), mature 1 (M1), and mature 2 (M2) thymocytes exhibited a broad range of Nur77-GFP, suggesting that heterogeneity of TCR signal strength is detectable throughout maturation (Fig. 1F) (26). To determine whether relatively low or high Nur77-GFP intensity is stable in the transition from CD4SP to mature T cell stages, we sorted CD4SP Nur77-GFP^LO or CD4SP Nur77-GFP^HI thymocytes and adoptively transferred them into congenic CD45.1 WT hosts (Fig. 1G). Two weeks post-transfer, most Nur77-GFP^LO donor cells exhibited a Population A phenotype (Fig. 1H). In contrast, Nur77-GFP^HI donor cells were skewed toward the Population C phenotype (Fig. 1H). These results suggest that relatively low or high Nur77-GFP expression in positively selected CD4⁺ T cells tends to persist in the peripheral lymphoid organs.

To determine how T cells may be skewed toward relatively low or high Nur77-GFP expression, we analyzed Population A-D cells that express the same transgenic TCR. We previously generated a mouse strain that combines the OT-II TCR transgene, Nur77-GFP transgene, Foxp3-IRES-RFP reporter allele, and the null allele of the endogenous TCR α chain to preclude expression of endogenously rearranged TCRs (10). Naive OT-II CD44^LO CD62L^HI Foxp3-IRES-RFP⁻ cells with Population A-D phenotypes were detected (Fig. 1I). These findings suggested that variations in tonic TCR signal strength are possible within populations that express identical transgenic TCRs. Among the variations that could account for the differences in tonic TCR signal strength and intensity of Nur77-GFP expression is the relative TCR:pMHC affinity within the context of a cell-cell interaction. We used 2-dimensional micropipette adhesion assays to determine the relative affinities of individual Population A-D OT-II cells for OVA^228-337 peptide/I-A^b monomers bound to human red blood cells (16). The mean apparent affinity to OVA pMHC was lowest for Population A, intermediate for Populations B and C, and highest for Population D, highlighting a marked positive correlation with Nur77-GFP fluorescence intensity (Fig. 1J). These results imply that Nur77-GFP fluorescence intensity in naive T cells, in addition to reflecting the intensity of TCR signal transduction, also reflects the relative affinity of individual T cells for self-antigens.

Strong tonic TCR signaling is associated with hypo-responsiveness to pMHC

We next investigated how Populations A-D respond to variations in pMHC concentration or affinity of an altered peptide ligand (APL). Naive CD4⁺ CD44^LO Foxp3⁻ Population A-D cells were sorted from OT-II TCRα-deficient Nur77-GFP Foxp3-IRES-RFP TCR transgenic mice. Sorted T cells were stimulated with antigen presenting cells plus WT OVA^323-339 peptide or OVA^H331R APL, a less potent altered peptide (27). In response to high concentrations of OVA^323-339 (≥ 10 μM), the percentages of Population A-D cells that expressed CD25 and CD69 were similar (Fig. 2A,B). However, low concentrations of OVA^323-339 ≤ 1 μM revealed graded responses where the strongest response was by Population A cells and decreased stepwise to Populations B, C, and D. Graded responses to OVA^H331R peptide were detected at all concentrations of peptide that were tested (Fig. 2C,D). We next aimed to determine the responsiveness of Population A-D OT-II cells to the altered peptide ligand in vivo. Population A-D OT-II cells were labeled with CellTrace Violet and adoptively transferred to congenic WT hosts that were subsequently challenged with splenocytes pulsed with OVA^H331R peptide. After 72 hours, the majority of Population A and B cells had undergone multiple rounds of cell division, whereas the majority of Population C and D cells remained undivided (Fig. 2E). These results are consistent with in vitro studies and highlight the hypo-responsiveness of naive T cells that experience strong tonic TCR signaling.

Figure 2.

Strong tonic TCR signaling correlates with hyporesponsiveness to TCR stimulation. A) Bar graph shows the mean percentage ± SEM of OT-II Population A-D cells expressing CD25 and CD69 after 24 hours of stimulation with T cell depleted splenocytes pulsed with the indicated concentrations of WT OVA peptide. CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ Population A-D cells were sorted from the spleen and lymph nodes of OT-II Nur77-GFP Foxp3-IRES-RFP Tcra^−/− mice. B) Histograms show the expression of Nur77-GFP, CD69, and CD25 by OT-II Population A-D cells after stimulation with T cell-depleted splenocytes plus the indicated concentrations of WT OVA peptide. C,D) Similar to panels A and B, except OVA H331R peptide was added to the cells. For experiments in panels A-D, n = 3 independent experiments. E) Overlaid contour plot shows Nur77-GFP and Ly6C expression by CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ Population A-D cells sorted from OT-II Nur77-GFP Foxp3-IRES-RFP mice prior to adoptive transfer into congenic WT CD45.2⁺ hosts. The dot plots show CellTrace Violet and Nur77-GFP fluorescence of Population A-D cells 72 hours after adoptive transfer of splenocytes that were pulsed with 100 μM OVA H331R peptide. n = 2 independent experiments. F) Histograms show the expression of Nur77-GFP, CD69, and CD25 by Population A-D cells sorted from ZAP-70^AS mice, following stimulation with T cell depleted splenocytes and anti-CD3ε antibodies and in the presence of the indicated concentrations of HXJ42 inhibitor. G) Bar graph shows the mean percentage ± SEM of Population A-D cells from ZAP-70^AS mice expressing both CD25 and CD69 after 24 hours of stimulation. The concentration of HXJ42 in each condition is indicated. G) Bar graph shows the mean percentage ± SEM of total CD4⁺ cells from control Zap70^+/− mice expressing both CD25 and CD69 following stimulation with T cell depleted splenocytes and anti-CD3ε antibodies for 24 hours in the presence of the indicated concentrations of HXJ42. N = 3 independent experiments. Statistical significance, * P < 0.05; ** P < 0.005; *** P < 0.0005; n.s. = not significant. Student’s t-Test.

To determine the sensitivity of Population A-D cells to incremental decreases in TCR signal strength, we used a specific inhibitor of ZAP-70, a tyrosine kinase required for activation of signaling downstream of the TCR. Titration of ZAP-70 catalytic activity can be achieved with a compound mouse strain that has a transgene encoding a Methionine 413 to Alanine mutant of ZAP-70 and is on a Zap70 knockout background (9). T cells from this strain express only the M413A mutant ZAP-70, which is catalytically active but gains sensitivity to the compound HXJ42, an analog of the kinase inhibitor PP1. Purified Population A-D cells expressing the analog-sensitive ZAP-70 (ZAP-70^AS) mutant were stimulated with antigen presenting cells and the same concentration of anti-CD3, in the presence of varying concentrations of HXJ42. Following stimulation of Population A-D cells in the absence of inhibitor, the percentage of CD25 and CD69 expressing cells progressively decreased from Population A to Population D (Fig. 2F,G). Increases in inhibitor concentration correlated with reductions in the responses of each Population. However, the pattern of graded decreases in CD25, CD69 expression from Population A to Population D was consistently detected. The concentrations of HXJ42 used in these experiments (≤ 0.5 μM) did not result in significant reductions in the upregulation of CD25 and CD69 by control Zap70^+/− T cells, which express the wild-type, HXJ42-insensitive, ZAP-70 protein (Fig. 2H). This result suggested that the effects of HXJ42 inhibition were specific to inhibition of the ZAP-70^AS mutant kinase. Together, these results suggest that as tonic TCR signaling increases, responsiveness to TCR stimuli decreases, the end result of which renders Population D cells most susceptible to inhibition of TCR signal transduction.

Transcriptome analysis of Population A-D cells

To determine whether Population A-D cells exhibit distinct gene expression patterns, we performed bulk RNA sequencing analysis of naive CD4⁺ CD44^LO Foxp3-IRES-RFP^neg cells sorted from three non-TCR transgenic mice. A total of 232 genes were differentially expressed (>1.5-fold) between any two populations. Principal Component Analysis (PCA) of all differentially expressed genes (DEGs) revealed that Populations A and C were more similar to each other in Principal Component 1 (PC1), whereas Populations B and D were most similar to each other, also in PC1 (Fig. 3A). This pattern may reflect the relatively high levels of Nur77-GFP expression by Populations B and D amongst the Ly6C⁺ cells and Ly6C⁻ cells, respectively. The comparison between Populations C and D yielded the most DEGs (174), 144 of which were upregulated in Population D (Fig. 3B,C). We next performed pathway analysis of the DEGs identified for each pairwise comparison. For the (Population D vs. Population A) and (Population C vs. Population B) comparisons, there were no statistically significant gene ontology (GO) terms identified (not shown). Due to the absence of DEGs between Population B and Population D, there were also zero GO terms identified. The top GO terms identified for the (Population C vs. Population A) and (Population B vs. Population A) comparisons were overlapping and largely associated with transcription and biosynthesis (Fig. 3D). In contrast, the (Population D vs. Population C) comparison highlighted several GO terms associated with immune cell activation, proliferation, adhesion, development, and signaling. In light of these findings, we decided to focus on the (Population D vs. Population C) comparison and the transcripts that were upregulated in Population D. The GO terms associated with the genes upregulated in Population D are consistent with an activated T cell, suggesting that the gene expression profile induced by strong tonic TCR signaling shares features with that of activated T cells.

Figure 3.

Increasing tonic TCR signal strength correlates with gene expression changes. Bulk RNA-seq analysis was performed on CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ Population A-D cells purified from the spleen and lymph nodes of 3 Nur77-GFP Foxp3-IRES-RFP mice. A) Principal Component Analysis of 232 differentially expressed genes identified for the six pairwise comparisons between cell populations. Individual replicates corresponding to Populations A-D are indicated by separate color-coded dots. Circles denote 99% confidence intervals. B) Heat map shows the relative expression levels of the 232 differentially expressed genes with greater than 1.5-fold change in expression. C) Table shows the number of differentially expressed genes between each pairwise comparison. D) Table shows the top gene ontology (GO) terms identified from pathway analysis of DEGs identified in the indicated pairwise comparisons. The top terms are shown with associated false discovery rate (FDR); n.s. = not significant. E) Volcano plots highlight subsets of differentially expressed genes within the indicated categories. Selected genes are labeled. F,G) Contour plots show Nur77-GFP and Ly6C expression by total naive CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ cells from F) Nur77-GFP Foxp3-IRES-RFP mice, or G) AND TCR transgenic Nur77-GFP Foxp3-IRES-RFP mice. Histograms show expression of the indicated cell surface molecules for Populations A-D. For panels F-G, n=3 independent experiments. H) Graphs of Gene Set Enrichment Analysis (GSEA) comparing the genes upregulated in Population D compared to Population C, versus the indicated gene sets. NES; Normalized Enrichment Score.

Transcripts upregulated in Population D include the Nr4a1, which was predicted by the high levels of Nur77-GFP expressed by this cell population, and Cd5, which has also been characterized as a surrogate marker of tonic TCR signal strength or TCR reactivity to self-pMHC. Also upregulated were co-stimulatory receptors (Icos, Tnfrsf4, Tnfrsf9) and inhibitory receptors (Pdcd1, Lag3, Cd200). Several transcription factors were upregulated in Population D, including Bcl6, Eomes, Foxp3, and Ikzf2. Despite excluding Foxp3-IRES-RFP expressing cells from our analysis, Foxp3 was identified as a DEG in Population D. This observation correlates with previous work, showing a higher propensity of Population D cells to induce Foxp3-IRES-RFP expression when stimulated in culture with TGFβ and IL-2 under Treg polarizing conditions (28). We verified elevated expression of selected DEGs including CD200, Helios, Bcl6, CD5, PD1, FR4, and CD127 at the protein level in naive CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ T cells from non-TCR transgenic mice (Fig. 3F) or AND TCR transgenic mice (Fig. 3G). Previous work showed that AND TCR transgenic cells in H-2^b background mice exhibit high basal expression of Nur77-GFP and are highly skewed toward Population C and Population D phenotypes (10). Elevated expression of Bcl6 at the transcript and protein level in Population D suggests that these cells express a lineage-defining transcription factor associated with Tfh cells. Consistent with this concept, recent studies showed a correlation between strong tonic TCR signaling and Tfh lineage differentiation (29).

Gene Set Enrichment Analysis suggested there was a high degree of overlap between the genes upregulated in Population D and genes upregulated in CD4⁺ cells following acute TCR stimulation (28) or CD4⁺ cells overexpressing Nr4a1 (Fig. 3H) (30). Recent evidence indicates that Nr4a1 expression is required for the maintenance of tolerance (28, 30–32). Together, these observations are consistent with a model whereby strong tonic TCR signaling induces increased expression of Nr4a1 by Population D cells. These findings are also consistent with gene expression profile associated with CD4⁺ T cell dysfunction induced by chronic stimulation of CD4⁺ cells with cognate antigen (33), and the gene expression program in naturally occurring CD73⁺ FR4⁺ anergic T cells (Fig. 3H).

Chromatin accessibility in Population D shares features with exhausted T cells

Considering that there were differences at the transcript level, we decided to investigate whether there were differences in chromatin accessibility associated with changes in tonic TCR signal strength. We performed ATAC-seq analysis of purified CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ Population A-D cells from 3 individual Nur77-GFP Foxp3-IRES-RFP mice. A total of 3,234 differentially accessible chromatin regions (DAR) with ≥ 2-fold difference in accessibility were detected. Most DARs can generally be clustered into two groups: one in which accessibility is highest in Population A with a graded downward trend from Population A to Population D, and a larger group of DARs with progressive increases in accessibility from Population A to Population D (Fig. 4A). This trend is also apparent in the PCA analysis of all DARs, which shows a progression along PC1 that parallels the increases tonic TCR signal strength (Fig. 4B). Most DARs (3,101), were differentially accessible between Populations A and D and the least (33) between Populations A and B (Fig. 4C). There were 490 known Nur77 binding sites within DARs in this dataset. The mean accessibility score of these DARs increased progressively from Population A to D, implying a positive correlation between Nur77-GFP expression and Nur77-dependent activity (Fig. 4D). Pathway analysis of the DARs in the (Population D vs. Population C) comparison identified GO terms including Cell activation, Biological adhesion, and Cytokine production, which are suggestive of a program of T cell activation (Fig. 4E). These processes were consistent with the top GO terms identified in the RNA-seq analysis of transcripts upregulated in Population D, further supporting the concept that Population D cells have experienced the strongest tonic TCR signals. The top GO terms identified for the (Population D vs. Population C) DARs were also highlighted by analyses of DARs in the comparisons between (Population C vs. Population A), (Population C vs. Population B), and (Population D vs. Population C) (not shown), which suggests that these comparisons have many overlapping DARs.

Figure 4.

Increasing tonic TCR signal strength correlates with differences in chromatin accessibility. ATAC-seq analysis was performed on CD4⁺ CD44^LO Foxp3-IRES-RFP⁻ Population A-D cells purified from the spleen and lymph nodes of 3 Nur77-GFP Foxp3-IRES-RFP mice. A) Heat map shows normalized chromatin accessibility at 3234 chromatin regions that are differentially accessible identified for the six pairwise comparisons between cell populations. B) PCA analysis of individual replicates from Population A-D. Circles denote 99% confidence intervals. Individual replicates corresponding to Populations A-D are indicated by separate color-coded dots. C) Table shows the number of differentially accessible regions (DARs) for the six pairwise comparisons between cell populations. D) Box and whisker graph shows the accessibility of 490 DARs containing Nur77 binding motifs. Rppm=reads per peak per million. Box indicates inner quartile range and error bars represent 1.5 times the inner quartile range. E) Table shows the top Gene Ontology terms identified in pathway analysis of the genes containing chromatin regions that are differentially accessible in the Population D vs. Population C comparison. F) Box and whisker graphs show the z-scores for accessibility of DARs associated with the indicated categories of genes. Box indicates inner quartile range and error bars represent 1.5 times the inner quartile range. To the right side of each graph is a partial list of genes with DARs in that group.

Examples of DARs that exhibit decreasing accessibility from Population A to Population D include regions in Il7r and Ly6c1 (Fig. 4F, left). These were also validated to be differentially expressed at the protein level (Fig. 3F,G). In contrast, the majority of DARs increase in accessibility from Population A to Population D. We next analyzed DARs that are within genes encoding cell surface receptors, intracellular signaling molecules, transcription factors, and function in cell adhesion or migration. The average accessibility score of DARs within these groups of genes increased from Population A and peaked in Population D (Fig. 4F). Notably, several DARs were within or near genes encoding costimulatory (Tnfrsf4 and Tnfrsf9) and coinhibitory receptors (Ctla4, Havcr2, Lag3, and Pdcd1) (Fig. 4F). These data suggest that strong tonic signaling promotes greater accessibility of genes associated with positive and inhibitory roles in T cell activation.

To determine whether there was a correlation between gene expression and chromatin accessibility for the Population D versus Population C comparison, we analyzed the fold change in gene expression and the fold change in chromatin accessibility. Out of 232 DEGs, 161 had associated DARs. This subset of 161 DEGs had a total of 929 associated DARs. For this comparison, there was a positive correlation between the change in gene expression versus the change in chromatin accessibility (Fig. 5A). Examples of genes with a positive correlation between chromatin accessibility and gene expression include Arhgap20, Bcl6, Lag3, and Pdcd1 (Fig. 5B). DARs in Arhgap20 and Lag3 were near the first exon and accessibility of both DARs was highest in Population D cells (Fig. 5C,D). There was an additional DAR in an intron in the Lag3 locus that also was most accessible in Population D (Fig. 5D). DARs associated with Bcl6 and Pdcd1 were upstream of the first exon, and were most accessible in Population D (Fig. 5E,F).

Figure 5.

Positive correlation between gene expression and chromatin accessibility. A) Graph plots the log₂ fold change in gene expression versus the log₂ fold change in chromatin accessibility for 161 out of 232 total DEGs that had associated DARs. Dots represent DARs associated with DEGs. Red dots represent DARs within genes that are expressed >1.5-fold higher in Population D compared to Population C; Blue dots represent DARs within genes expressed >1.5-fold lower in Population D compared to Population C; black dots represent DARs associated with genes that are not differentially expressed in the Population D vs. Population C. The line represents the correlation with 95% confidence interval and p- value is indicated. B) Graph is similar to the one in A), except the highlighted dots represent DARs associated with four DEGs, Arhgap20, Lag3, Pdcd1, and Bcl6, as indicated. C-F) Genome plots represent chromatin accessibility score (y-axis) as a function of chromosome location (x-axis) for the indicated genes, as determined by ATAC-seq. Each trace includes a schematic of the coding region of each gene at the top. Label at bottom shows the chromosome coordinates for each region. DARs with >2-fold differential accessibility between any of the six pairwise comparisons between populations are enclosed by boxes. G) Taiji analysis of transcriptional targets. Bar graph shows the transcription factors with the highest fold change in PageRank scores between Population D versus Population C.

To analyze which transcription factor networks may be most active in Population D cells, we performed Taiji analysis, which integrates RNA-seq and ATAC-seq data (25). Analysis of the fold-change in PageRank scores between Population D and Population C datasets suggested that Nr4a3 (Nor1), NFATc1 and NFATc3, transcription factors activated by TCR signaling, were among the differentially activated factors (Fig. 5G). Also differentially active in Population D cells was Eomes, which is expressed in Th1 effector cells but also expressed in dysfunctional CD4⁺ cells that have been chronically stimulated (33).

We next investigated selected individual DARs detected near exon 4 of Havcr2 (Tim3) and in introns 1 and 3 of Ctla4, both of which were more accessible in Population D (Fig. 6A,B). These DARs are similar in location to DARs that have been reported to occur selectively in antigen-experienced, including effector, memory, and effector CD8⁺ cells, but not naive T cells (34, 35). It has also been reported that DARs in intron 1 of Cd200r1 and near exon 1 of Cxcr5 are selectively detected in exhausted, but not naive, effector, or memory CD8⁺ cells (36). Strikingly, we detected DARs in Population D cells that were in similar regions of Cd200r1 and Cxcr5 (Fig. 6C,D). Moreover, multiple DARs throughout the Tox locus have been selectively detected in exhausted CD8⁺ T cells (37). In our analysis, two DARs were detected in intron 1 and intron 3 of the Tox locus in Population D cells, the locations of which were similar to DARs detected in exhausted CD8⁺ cells (Fig. 6E) (37). These data suggest that naive CD4⁺ T cells that experience the strongest tonic TCR signals share some epigenetic features with antigen-experienced T cells. Furthermore, constitutive strong tonic TCR signaling may induce some chromatin accessibility patterns that are shared with exhausted T cells. These data also support a model where naive T cells adapt to intense tonic TCR signals to induce hypo-responsiveness.

Figure 6.

Some differentially accessible regions of chromatin in Population D are also associated with antigen-experienced cells. Genome plots for Havcr2 (encoding Tim3), B) Ctla4, C) Cd200r1, D) Cxcr5, and E) Tox represent chromatin accessibility score as a function of chromosome location, as determined by ATAC-seq analysis. A schematic of the coding region of each gene is located above each trace. Labels below each trace display the chromosome coordinates for each region shown. DARs are enclosed by boxes. Asterisks indicate DARs that have been previously reported to occur in effector, memory, or exhausted CD8⁺ cells. E) In the Tox locus, the DARs highlighted in yellow boxes are enlarged below to more clearly show the DARs within the selected regions.

DISCUSSION

The strength of tonic TCR signaling can vary broadly between individual naive CD4⁺ cells. Our data support a model whereby a subpopulation of CD4⁺ T cells adapt to sustained, strong tonic TCR signaling, as a result of relatively high affinity to self-pMHC. Furthermore, such T cells adapt to the strength of tonic TCR signaling in part through changes in gene expression and chromatin accessibility, and ultimately in reduced responsiveness to subsequent TCR stimulation.

In previous 2D micropipette studies, the relative affinities of OT-II and Smarta TCR transgenic cells to their respective cognate pMHC antigens were measured across multiple individual cells (38). The relative affinities of differed between OT-II and Smarta T cells, however, there was also heterogeneity within each TCR transgenic population. The relative affinities for individual OT-II T cells spanned an approximate 10-fold range. These studies revealed heterogeneity of individual T cell affinities, even in populations that expressed identical transgenic TCRs. Here, we show that the hierarchy of relative affinities within this 10-fold range correlates with the intensity of basal Nur77-GFP expression. Importantly, the OT-II T cells used in these studies were from mice that were also homozygous for the null allele of the endogenous TCR α chain, which prevents expression of endogenously rearranged TCRs. Previous work showing that in polyclonal T cell populations, clones with relatively low affinities to a given antigen can comprise a substantial proportion of a T cell response (39). This is consistent with our finding that Population A cells are most responsive to stimulation with varying concentrations of peptide, with a less potent peptide, or under conditions where ZAP-70 catalytic activity is partially inhibited.

In light of their increased apparent affinities, we propose that mature naive Nur77-GFP^HI T cells may arise from immature T cells expressing TCRs that are near the affinity threshold of negative selection. This concept is consistent with previous work implying that subpopulations of peripheral T cells are derived from precursors that survived incomplete negative selection (40–43). As a result of relatively high affinity for self-pMHC, Nur77-GFP^HI cells and Population D cells in particular likely experience strong tonic TCR signaling chronically. Recent work described the functional and gene expression changes associated with chronic stimulation of naive CD4⁺ cells in the absence of inflammation (33). In these studies, chronic stimulation resulted in a hypofunctional state characterized by attenuated effector cytokine potential, and a gene expression pattern associated with anergy or T cell exhaustion (Cblb, Tox, Nr4a1, Eomes, Lag3, Nt5e, Izumo1r, and Pdcd1). We found these genes were also upregulated in Population D, suggesting that chronic TCR stimulation in response to self-antigens may promote a gene expression program with common features. In addition to adaptations at the transcriptional level, we also detected differential chromatin accessibility patterns in Population D cells that shares some features with exhausted CD8⁺ T cells. Among the genes with differentially accessible chromatin regions in exhausted CD8⁺ cells and Population D cells were Cd200r1, Cxcr5, and Tox. Though it is currently unknown precisely which signals induce differential accessibility at these loci, we propose that the chronicity of TCR signaling experienced by both Population D cells and exhausted CD8⁺ T cells may have a role. It is notable that the transcriptional signatures of Population B and D cells were similar, yet there were multiple differences in chromatin accessibility between these populations. We propose that RNA-seq provides a transcriptional profile of cells at the time they were lysed, whereas, ATAC-seq provides information on chromatin accessibility, which can reflect the cumulative experience of Population A-D cells from previous time points up to the time the cells were lysed. Thus, the differences between Population B and D detected by ATAC-seq could reflect previously experienced signals that continue to influence chromatin accessibility.

Among the differentially expressed genes that were upregulated in Population D were transcription factors Foxp3 and Bcl6, which are associated with Treg cells and Tfh cells, respectively. The cells that were sorted for RNA-seq analysis were sorted to exclude expression of a Foxp3-RFP reporter gene. Therefore, we conclude that Population D cells are not Treg cells but may exhibit a bias toward peripheral Treg differentiation. Previous studies have indicated that strong tonic TCR signaling, as marked by high expression of basal Nur77-GFP and CD5, or absence of Ly6C expression, correlates with a higher propensity for inducing Foxp3 expression under Treg polarizing conditions in vitro (10, 44, 45). Similarly, the presence of elevated Bcl6 transcript levels in Population D is suggestive of a correlation between strong tonic TCR signaling and Tfh lineage differentiation, which is consistent with recent studies (29). These findings add support to the concept that the strength of tonic TCR signaling influences the effector differentiation of naive CD4⁺ T cells.

Data Availability

RNA-seq and ATAC-seq data is available under accession number GSE206074 in the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/).