A stepwise and digital pattern of RSK phosphorylation determines the outcome of thymic selection

Summary

Developing CD4⁺CD8⁺ double-positive (DP) thymocytes with randomly generated T cell receptors (TCRs) undergo positive (maturation) or negative (apoptosis) selection on the basis of the strength of TCR stimulation. Selection fate is determined by engagement of TCR ligands with a subtle difference in affinity, but the molecular details of TCR signaling leading to the different selection outcomes have remained unclear. We performed phosphoproteome analysis of DP thymocytes and found that p90 ribosomal protein kinase (RSK) phosphorylation at Thr⁵⁶² was induced specifically by high-affinity peptide ligands. Such phosphorylation of RSK triggered its translocation to the nucleus, where it phosphorylated the nuclear receptor Nur77 and thereby promoted its mitochondrial translocation for apoptosis induction. Inhibition of RSK activity protected DP thymocytes from antigen-induced cell death. We propose that RSK phosphorylation constitutes a mechanism by which DP thymocytes generate a stepwise and binary signal in response to exposure to TCR ligands with a graded affinity.

Graphical abstract

Highlights

• •RSK phosphorylation is specifically induced by high-affinity peptides in thymocytes
• •This response operates in a stepwise and digital manner with a sharp threshold
• •RSK triggers phosphorylation and mitochondrial translocation of Nur77 for apoptosis
• •Inhibition of RSK protects cells from death induced by the high-affinity peptides

Immunology; Cell biology

Introduction

A functional immune system in mammals requires the selection of T lymphocytes that are able to distinguish pathogens from self-antigens. This process occurs during T cell development and consists of both positive and negative selection in the thymus, where immature thymocytes at the CD4⁺CD8⁺ double-positive (DP) stage are stimulated by self-peptides presented by major histocompatibility complex (MHC) molecules.¹ Most thymocytes that express a T cell receptor (TCR) that fails to bind to a presented peptide-MHC (pMHC) complex undergo death by neglect, given that the nonfunctional TCR does not initiate downstream signaling. Thymocytes with a TCR that binds with high affinity to pMHC undergo apoptotic cell death (negative selection). Only thymocytes that express a TCR that binds with low affinity to pMHC differentiate into CD4⁺ or CD8⁺ single-positive (SP) cells (positive selection), ensuring functionality and self-tolerance of the differentiating T cells.²

The TCR-ligand engagement in DP thymocytes that distinguishes between positive and negative selection occurs over a narrow affinity threshold^3,4,5 and results in similar but qualitatively and quantitatively different downstream signal transduction. In response to both high- or low-affinity ligand engagement, TCR triggers phosphorylation of the ITAM (immunoreceptor tyrosine-based activation motif)-containing CD3 chains (CD3γ, CD3δ, CD3ε, and CD3ζ) by the SRC family kinase LCK. Subsequent phosphorylation of ZAP70 (zeta chain-associated protein kinase 70) by LCK induces formation of a complex between LAT (linker for activation of T cell) and SLP76 (SH2 domain-containing leukocyte protein of 76 kDa). This signaling complex activates phospholipase Cγ1, which catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate to diacylglycerol and inositol 1,4,5-trisphosphate, the latter of which induces the release of Ca²⁺ from endoplasmic reticulum. High-affinity ligand stimulation during T cell selection results in maximal LAT phosphorylation that is necessary for induction of downstream signaling mediated by the activation of p38 MAPK (mitogen-activated protein kinase) and JNK (c-Jun NH₂-terminal kinase).^6,7,8 High-affinity ligand stimulation also triggers a pronounced Ca²⁺ response and results in the recruitment of the SOS-GRB2 protein complex to the plasma membrane and consequent transient but marked phosphorylation of ERK (extracellular signal-regulated kinase), whereas low-affinity ligand stimulation induces only a weak Ca²⁺ response and compartmentalized RAS signaling that results in a slow and low-level phosphorylation of ERK at the Golgi apparatus.³ A subtle difference in ligand-TCR engagement thus determines the duration and intensity of downstream TCR responses that govern the differential intermediate signal transduction leading to a positive or negative selection outcome. Only a few mediators downstream of TCR that show differential activity and directly link to either cell maturation or cell death have been characterized to date, however.

Recent advances in quantitative phosphoproteomics have allowed investigations of large phosphorylation-mediated signaling networks and the identification of key regulatory molecules and responses in various cell types,^9,10,11 including a T lymphoma cell line,¹² primary T lymphocytes,¹³ and thymocytes.¹⁴ However, TCR signals that are dependent on ligand-TCR affinity in physiologically relevant models have not been widely characterized. DP thymocytes are more sensitive than are mature T cells to ligands with a low affinity for TCR.^15,16,17 Themis is a protein expressed in thymocytes that contributes to selection, with its expression level being highest in DP cells.^{18,19,20,21,22} It selectively dampens TCR signaling triggered by ligands of mid to high affinity and is specifically required for positive selection.^23,24 Thus, whereas studies have indicated that signaling mechanisms in DP thymocytes are essential for positive and negative selection, identification of the global TCR signaling network with the use of relevant mouse models would be expected to provide a valuable resource to support a more comprehensive characterization of these signaling mechanisms.

We have now performed a quantitative phoshoproteomics analysis of DP “preselected” thymocytes from OT-I TCR transgenic mice on the basis of the strength of TCR stimulation. Bioinformatics analysis identified phosphorylation sites that are regulated differentially according to the threshold for thymic positive or negative selection. Functional studies revealed that p90 ribosomal protein kinase (RSK) is specifically phosphorylated and is required for apoptosis induction through the Nur77 axis in response to stimulation with high-affinity peptide ligands.

Results

Validation of the experimental system for study of TCR signaling associated with thymic selection

To globally capture TCR downstream signaling dependent on the ligand-TCR affinity threshold for positive or negative selection, we studied thymocytes of offspring produced by a cross between OT-I transgenic mice and Tap1^−/− mice. OT-I is a mouse TCR that recognizes MHC-I (H2-K^b) loaded with an ovalbumin peptide (OVA: SIINFEKL),²⁵ and Tap1^−/− mice lack antigen peptide transporter 1. Thymocytes of OT-I Tap1^−/− mice are unable to differentiate into CD8⁺ SP cells as a result of low expression of MHC-I on thymic epithelial cells, with most thymocytes in these mice thus being stalled at the preselected CD4⁺CD8⁺ DP stage.²⁶ Stimulation of these thymocytes with H2-K^b tetramers loaded with OVA or OVA mutant peptides allows investigation of TCR signaling dependent on ligand affinity for TCR. With the use of this system, we aimed to identify TCR-mediated signaling that differs between positive and negative selection settings by phosphoproteomics analysis (Figure 1A). OT-I Tap1^−/− thymocytes stimulated in bulk by TCR ligands of graded affinity were thus subjected to quantitative phosphoproteomics analysis, after which bioinformatics analysis was performed to search for phosphorylation sites that were regulated in a manner dependent on ligand-TCR affinity. The biological relevance of the identified phosphorylation sites was then validated by functional analysis.

Figure 1.

Application of the OT-I Tap1^−/− system for study of early TCR signaling associated with thymic positive and negative selection

(A) Experimental design of the study. OT-I Tap1^−/− thymocytes were stimulated with peptide-loaded H2-K^b tetramers that span the TCR affinity threshold for induction of thymic positive or negative selection and were then subjected to phosphoproteomics analysis. Bioinformatics analysis was then applied to identify phosphoproteins that were differentially regulated according to the thymic selection threshold, and biochemical assays were performed to validate the signaling pathways regulated selectively by threshold stimulation.

(B) Fetal thymic organ culture (FTOC) was performed with thymic lobes derived from OT-I Tap1^−/− mouse embryos at embryonic day (E) 14.5 to E15.5 and with OVA or its mutant peptides (Q4, Q4H7, or Q7) for 3 days, after which thymocytes were isolated and subjected to flow cytometric analysis of CD4 and CD8 expression at the cell surface. A peptide derived from vesicular stomatitis virus (VSV) was used as a nonselecting control. Representative plots from two independent experiments are shown.

(C) OT-I Tap1^−/− thymocytes were stained with PE-labeled and peptide-loaded H2-K^b tetramers and then analyzed by flow cytometry. The gray shaded trace represents thymocytes without H2-K^b tetramer staining.

(D) Flow cytometric analysis of the cytosolic Ca²⁺ concentration during stimulation of OT-I Tap1^−/− thymocytes with peptide-K^b tetramers (Tet.). A.U., arbitrary units.

(E) Immunoblot (IB) analysis of the time course of ZAP70, LAT, ERK1/2, and tyrosine (Y) phosphorylation during stimulation of OT-I Tap1^−/− thymocytes with K^b-OVA or K^b-Q7 tetramers. HSP90 was examined as a loading control. Representative plots from two independent experiments are shown. See also Figure S1.

We first attempted to confirm selecting ligands previously defined with OT-I Tap1^−/− thymocytes with the use of the fetal thymus organ culture (FTOC) system.⁴ When we pulsed the FTOC with OVA or the Q4 mutant peptide, almost all DP thymocytes were lost, showing that these ligands induced negative selection (Figures 1B and S1A). Culture with the Q4H7 or Q7 mutant peptides, however, resulted in the appearance of CD8⁺ SP cells, indicating that the corresponding peptide-pulsed DP thymocytes selectively differentiated into SP thymocytes by positive selection. Given the graded affinity of MHC-bound OVA, Q4, Q4H7, and Q7 for TCR (Figure S1B),^3,23 the OT-I Tap1^−/− system thus faithfully reflects the qualitative outcomes of thymic positive or negative selection, which were not directly proportional to ligand affinity for TCR.

Given that phosphoproteomics analysis requires a large number of thymocytes that are uniformly activated by stimulation, we tested K^b tetramers loaded with different selecting peptides for stimulation of such a large number of thymocytes in vitro. Staining of OT-I Tap1^−/− thymocytes with phycoerythrin (PE)-labeled and peptide-loaded K^b tetramers (100 nM) on a large scale revealed that all the thymocytes were PE positive, indicating that the K^b tetramers interacted uniformly with most of the preselected thymocytes regardless of the different TCR affinities of the loaded peptides (Figure 1C). A slightly higher staining level apparent for K^b-OVA reflected the higher ligand potency of OVA for TCR (Figure 1B). Titration of the tetramers also revealed that both high- and low-affinity complexes engaged most (>95%) thymocytes, with geometric mean fluorescence intensity values being similar at the high tetramer concentrations (Figures S1C and S1D). We next examined whether tetramer stimulation in vitro reflects proximal TCR signaling characteristic of thymic positive and negative selection. Calcium flux is a hallmark of TCR proximal signaling and has been used to examine TCR signal strength in a sensitive manner.³ As expected, OT-I Tap1^−/− thymocytes stimulated with the negatively selecting K^b tetramers (OVA and Q4) showed a larger increase in the cytosolic Ca²⁺ concentration than did those stimulated with the positively selecting K^b tetramers (Q4H7 and Q7) (Figure 1D), reflecting quantitative differences in TCR proximal signaling during selection associated with subtle differences in ligand potency. The phosphorylation dynamics for TCR-proximal proteins also differed between K^b-OVA and K^b-Q7, with the increase in the abundance of phosphorylated (p) forms of ERK1/2 and LAT being more rapid and more pronounced with K^b-OVA than with K^b-Q7 (Figure 1E). These results thus confirmed that the OT-I Tap1^−/− system stimulated with K^b tetramers faithfully recapitulated not only quantitative but also a binary response to TCR signaling associated with positive or negative selection and was therefore suitable for quantitative phosphoproteomics analysis.

Quantitative phosphoproteome analysis of positive and negative selection

For application of quantitative phosphoproteomics to the study of early TCR signaling underlying positive or negative selection, OT-I Tap1^−/− thymocytes were stimulated for 2 min with the different peptide-loaded K^b tetramers with graded affinity for TCR, and cell lysates were then mixed with the “spike-in” SILAC (stable isotope labeling by amino acids in cell culture) standard for quantification (Figure 2A). We chose the spike-in SILAC method for quantitation of phosphopeptides because it is a well-established approach to phosphoproteome analysis of primary cell cultures.^27,28 To prepare a spike-in SILAC standard for the present study, we cultured the DP thymoma cell line DPK with the heavy amino acids ([¹⁵N₂,¹³C₆]lysine and [¹⁵N₄,¹³C₆]arginine) for metabolic labeling until >95% of the identified peptides were converted to the heavy forms. A 1:1 mixture of lysates from unstimulated DPK cells and DPK cells stimulated by TCR cross-linking was then prepared for use as the spike-in SILAC standard in order to maximize the coverage of phosphorylation corresponding to before and after TCR stimulation. We monitored robust increases in the phosphorylation of well-established TCR downstream targets including ZAP70 (Tyr²⁹⁰), PAK2 (Ser^192/197), and ERK1/2 (Thr²⁰² and Tyr²⁰⁴) as well as in the phosphorylation of tyrosine residues (pY) of several unidentified proteins in response to stimulation of DP cells with each K^b tetramer-peptide complex (Figure 2B). Lysates from the stimulated thymocytes were mixed with the spike-in standard and digested with trypsin, and the resulting phosphopeptides were enriched before fractionation by basic reversed-phase chromatography and analysis by liquid chromatography and tandem mass spectrometry. We repeated the phosphoproteome analysis with a total of four biological replicates, and we identified and quantified a total of 6456 class I phosphorylation sites (localized score p > 0.75) on 2557 proteins that were detected in at least two of the four replicates (Figure 2C; Table S1), with each replicate including ∼5000 to 7000 phosphorylation sites with similar coverage. The 6456 phosphorylation sites were comprised of 86% phosphoserine, 13% phosphothreonine, and 1% phosphotyrosine residues (Figure 2D), a distribution similar to that previously described for T lymphocytes.²⁹ To examine whether the phosphoproteome faithfully represented the proximal TCR signaling network, we mapped our data to the TCR signaling network in the Kyoto Encyclopedia of Genes and Genomes (KEGG). We found that 40% of the identified proteins were mapped on the TCR signaling network of KEGG (false discovery rate [FDR] q value = 6.6 × 10⁻¹⁰), with 25% of these proteins containing phosphorylation sites whose phosphorylation level was changed from the unstimulated condition by a factor of 1.5 up or down by stimulation with at least one of the peptide-K^b tetramers (Figure 2E; Table S2). These latter proteins included TCR-bound molecules such as CD3 chains (CD3γ, CD3δ, CD3ε, and CD3ζ) as well as proximal TCR signalosome proteins (LAT, SLP76, and ZAP70) and downstream kinases (PAK, p38 MAPK, ERK1/2, PKCθ, AKT, and TAK1), the ubiquitin ligase CBL, and the phosphatase SHP1. Our data thus appeared to cover a broad spectrum of phosphorylated proteins associated with TCR signaling.

Figure 2.

Quantification of the phosphoproteome associated with positive and negative selection of DP thymocytes by the spike-in SILAC method

(A) Experimental protocol. Proteins derived from OT-I Tap1^−/− thymocytes stimulated with peptide-K^b tetramers were mixed with the spike-in standard consisting of proteins derived from DPK cells (heavy isotope labeled), and peptide digests were fractionated by basic reversed-phase (RP) chromatography and analyzed by LC-MS/MS. Changes in phosphorylation induced by the stimuli were quantified on the basis of the relative amounts of experimental phosphopeptides and the corresponding spiked-in phosphopeptides.

(B) Representative immunoblot analysis of pZAP70, pPAK2, pERK1/2, and pY in lysates of OT-I Tap1^−/− thymocytes that had been stimulated with peptide-loaded K^b tetramers for 2 min before phosphoproteomics analysis.

(C) Number of uniquely identified class I phosphorylation sites for each biological replicate of the phosphoproteome.

(D) Distribution of the 6456 phosphorylation sites in (C) by amino acid.

(E) Identified phosphoproteins were mapped to the TCR signaling network in KEGG. Gray shading represents phosphoproteins that were not identified by our analysis, whereas red and orange shading represent identified proteins with phosphorylation sites whose phosphorylation level changed from the unstimulated condition by a factor of 1.5 up or down, or changed to a lesser extent, respectively, in response to stimulation by at least one of the peptide-K^b tetramer stimuli. See also Tables S1 and S2.

Identification of phosphoproteins associated with positive or negative selection

To identify phosphorylation sites consistently regulated by TCR stimulation, we searched for sites that were regulated by stimulation with at least one of the ovalbumin peptide (OVA, Q4, Q4H7, or Q7)-K^b tetramer complexes by application of a permutation-based FDR t test³⁰ (FDR <0.05, S₀ = 0.45, where S₀ represents the artificial within-groups variance) (Figure 3A). The number of regulated phosphorylation sites was found to be similar (∼900) for the different peptide-K^b tetramers (Figures 3A and S2A). About one-fourth of the regulated phosphorylation sites (24%, 410 of 1717 phosphorylation sites regulated by at least one of the peptide-K^b tetramers) were common to stimulation by all the peptide-K^b tetramers (Figure S2B), suggesting that the global signaling pattern did not differ greatly among ligands corresponding to this TCR affinity range. We therefore subjected all of the regulated phosphorylation sites (1717 sites of 1158 proteins) to further analysis (Table S3).

Figure 3.

RSK phosphorylation at Thr⁵⁶² is induced by high-affinity peptide ligands

(A) Volcano plots for phosphorylation sites significantly upregulated or downregulated by peptide-K^b tetramer stimulation in at least two of the phosphoproteome replicates. The black border lines were calculated by the permutation-based FDR method (FDR <0.05, S₀ = 0.45) for significant changes relative to the nonstimulated (None) sample. The numbers at the top right and top left of each plot indicate the numbers of upregulated or downregulated phosphorylation sites, respectively.

(B) Hierarchical clustering of the regulated phosphorylation sites (n = 1717 phosphorylation sites on 1158 proteins) identified as in (A) and Figure S2. The phosphorylation patterns in cluster 2 represent upregulated sites responsive to the selection threshold for peptide-K^b tetramer stimulation. The VSV ratio was used as a denominator to eliminate the basal phosphorylation potentially induced by K^b tetramer binding to CD8.

(C) Logo analysis of phosphorylation site motifs identified in clusters 1 to 4. The n values indicate the number of phosphorylation sites containing each motif sequence.

(D) Phosphoproteins in cluster 2 identified as apoptosis regulators or kinases.

(E) Schematic representation of an RSK phosphorylation network identified among phosphorylation sites of cluster 2 (ERK1-Y204, RSK1-T562, and GSK3β-S9).

(F) Representative immunoblot analysis of RSK phosphorylation at Thr⁵⁶² in lysates of OT-I Tap1^−/− thymocytes stimulated with peptide-K^b tetramers for 2 min.

(G) Quantitative analysis of the pRSK(T562)/HSP90 ratio for immunoblots similar to that in (F). Data are means ± SD (n = 3 independent experiments). ∗p < 0.05; NS, not significant (one-way ANOVA followed by Dunnett’s post hoc test). See also Figures S2 and S3 as well as Table S3.

We next performed hierarchical clustering to identify groups of phosphorylation sites that showed similar response patterns for stimulation with the different peptide-K^b tetramers. We analyzed row-wise clustering of phosphorylation sites and applied a cutoff line with 12 clusters to obtain the most relevant clusters with respect to ligand-TCR affinity. This analysis revealed four prominent clusters with distinctive upward or downward patterns (Figure 3B). Cluster 2 and cluster 3 consisted mostly of phosphorylation sites whose phosphorylation level tended to increase with the TCR affinity of the ligands, with cluster 2 matching most closely the threshold response. Cluster 1 and cluster 4 consisted of phosphorylation sites whose phosphorylation level decreased in relation to ligand-TCR affinity. To examine whether these four clusters were enriched in phosphorylated proteins with distinctive functions, we performed enrichment analysis of each cluster for Gene Ontology (GO) biological terms with the use of the Panther tool (Figures S3A–S3D).³¹ Clusters 2 and 3 (with an upward trend relative to TCR affinity) were enriched in proteins related to cytoskeleton organization, with a significant p value (FDR <0.01) compared with the entire mouse proteome (Figures S3A and S3B). This finding suggested that cytoskeletal reorganization was more pronounced during stronger TCR stimulation, possibly reflecting the formation of tight immunological synapses.³² Clusters 1 and 4 (with a downward trend relative to TCR affinity) were enriched in proteins associated with RNA- or nucleus-related processes, possibly reflecting requirements for the differentiation program. Further exploration of the proteins in these various categories will be necessary to determine their specific roles during thymic selection.

RSK phosphorylation on Thr⁵⁶² induced by high-affinity peptide ligands

Given the narrow range of ligand affinity for TCR corresponding to the qualitative threshold for induction of negative versus positive selection by Q4 and Q4H7 peptides, respectively (Figure 1B), we next focused on cluster 2, the phosphorylation sites of which were upregulated according to the threshold response for negative selection, in order to identify sites that are key to discrimination between thymic positive and negative selection. Cluster 2 consisted of 119 phosphorylation sites of 107 proteins, the genes for which have been annotated according to protein function and subcellular localization (Figure S3E). Analysis of the amino acid sequences surrounding the phosphorylation sites of cluster 2 revealed that many corresponded to the phosphorylation motifs [pS/pT]P (n = 34/119), PX[pS/pT]P (n = 14/119), or RXX[pS/pT] (n = 39/119) (Figure 3C). The proline-directed motif [pS/pT]P, which is targeted by many kinases including MAPKs, cyclin-dependent kinases, and glycogen synthase 3 (GSK3),³³ was also found in clusters 1, 3, and 4. The MAPK consensus motif PX[pS/pT]P, which is a type of proline-directed motif, was also detected in all four clusters. The RXX[pS/pT] motif, which is targeted by AGC family kinases including PKA, PKG, and PKC as well as RSK,³⁴ was detected only in clusters 2 and 3, both of which contained phosphorylation sites with an upward trend according to peptide-K^b tetramer affinity. To examine whether the phosphorylation sites of cluster 2 might contribute to promotion of cell death, we searched the corresponding proteins for kinases as well as molecules with an annotation related to apoptotic process in GO biological terms. This search identified only three proteins—GSK3β, ERK1, and RSK—that fulfilled both criteria in cluster 2 (Figure 3D). The RSK family consists of four isoforms, of which RSK1, RSK2, and RSK3 contain the same phosphorylation site found in the regulated peptide, AENGLLM[pT]PCYTANF (RSK1 at Thr⁵⁶², RSK2 at Thr⁵⁷⁷, RSK3 at Thr⁷⁰⁰). Phosphorylation of RSK1 at Thr⁵⁶² in the C-terminal kinase domain in response to ERK signaling has been shown to initiate activating autophosphorylation of RSK1 at Ser³⁶⁹ and consequent phosphorylation of downstream targets including GSK3β at Ser⁹ (Figure 3E).^35,36,37 Immunoblot analysis with antibodies specific for Thr⁵⁶²-phosphorylated RSK revealed that Thr⁵⁶² phosphorylation was upregulated by stimulation of OT-I Tap1^−/− thymocytes with K^b-OVA or K^b-Q4 tetramers, both of which were associated with negative selection, whereas it was unaffected by stimulation with K^b-Q4H7 or K^b-Q7 tetramers, both of which were associated with positive selection (Figures 3F and 3G). These responses have a sharp threshold between Q4 (negative selecting) and Q4H7 (positive selecting) with regard to TCR affinity, revealing a stepwise and digital response pattern. A similar digital response pattern was apparent with the use of adjusted concentrations of the OVA and Q4 tetramers versus Q4H7 (Figure S3F). Our results thus suggested that RSK phosphorylation at Thr⁵⁶² is specifically upregulated during DP thymocyte stimulation with negatively selecting ligands.

RSK is phosphorylated at Thr⁵⁶² by ERK and translocates to the nucleus in response to high-affinity peptide ligand stimulation

We next investigated how RSK phosphorylation is regulated specifically in response to strong TCR stimulation. Various studies have shown that RSK is phosphorylated by multiple upstream kinases including phosphoinositide-dependent protein kinase 1 (PDK1) and ERK.^34,38 Treatment of OT-I Tap1^−/− thymocytes with the MEK-ERK pathway inhibitor U0126, but not that with the PDK1 inhibitor GSK2334470, blocked the upregulation of RSK phosphorylation at Thr⁵⁶² induced by the K^b-Q4 tetramer (Figures 4A and S4A), indicating that activation of the MEK-ERK axis is required for such phosphorylation. U0126 also inhibited phosphorylation of RSK at Thr⁵⁶² induced by K^b-OVA, whereas it had no effect on the basal level of such phosphorylation apparent in the presence of K^b-Q4H7 or K^b-Q7 (Figure 4B). ERK5, but not ERK1/2, was previously shown to be essential for negative selection.^39,40 Indeed, we found that phosphorylation of both ERK1/2 (Thr²⁰²/Tyr²⁰⁴) and ERK5 (Thr²¹⁸/Tyr²²⁰) was inhibited by U0126 treatment (Figure 4B), and that ERK5 phosphorylation was increased to a greater extent by H2-K^b tetramers loaded with OVA or Q4 than by those loaded with Q4H7 or Q7 (Figure 4C). Of note, quantitative analysis revealed that phosphorylation of only the 115-kDa ERK5 isoform was increased to a greater extent by high-affinity tetramer (OVA, Q4) stimulation than by low-affinity tetramer (Q4H7, Q7) stimulation, with the extent of phosphorylation of the 90-kDa immunoreactive protein being similar in the presence of the different tetramers (Figures 4B and 4C). Phosphorylation of ERK1/2 was similarly affected by the peptide-K^b tetramers without any apparent threshold for affinity, and therefore did not show a digital pattern (Figure S4B). These results suggested that the MEK-ERK5 pathway is responsible for RSK phosphorylation at Thr⁵⁶² in a stepwise and digitally responsive manner across the thymic selection threshold.

Figure 4.

RSK is phosphorylated at Thr⁵⁶² by ERK and translocates to the nucleus in response to high-affinity peptide stimulation

(A) OT-I Tap1^−/− thymocytes were incubated for 15 min on ice in the absence or presence of the MEK inhibitor U0126 (10 μM) or the PDK1 inhibitor GSK2334470 (5 μM) and then stimulated for the indicated times at 37°C with the K^b-Q4 tetramer. Cell lysates were then subjected to immunoblot analysis of pRSK, pERK1/2, and pAKT (examined as an indicator of PDK1 activity). β-actin was examined as a loading control. Data are representative of three independent experiments.

(B) OT-I Tap1^−/− thymocytes were incubated for 15 min on ice in the absence or presence of U0126 (10 μM) and then stimulated for the indicated times at 37°C with peptide-K^b tetramers. Cell lysates were then subjected to immunoblot analysis of pRSK, pERK1/2, and pERK5. A representative blot and quantitative data (means ± SD) for the pRSK/β-actin band intensity ratio from three independent experiments are shown.

(C) Immunoblot analysis of pERK5 and pERK1/2 in OT-I Tap1^−/− thymocytes stimulated with peptide-K^b tetramers for 2 min. The upper and lower bands for pERK5 represent phosphorylation of the long or short ERK5 isoforms. A representative blot and quantitative data (means ± SD) for the pERK5/β-actin band intensity ratio from three independent experiments are shown.

(D) OT-I Tap1^−/− thymocytes were stimulated with peptide-K^b tetramers for 5 min and then subjected to immunofluorescence analysis of RSK1/2/3. Nuclei were stained with Hoechst 33342. Representative images are shown on the left (scale bar, 5 μm), and the proportion of cells showing nuclear localization of RSK as determined from three independent experiments is shown on the right. Quantitative data are means ± SD. ∗∗p < 0.01, ∗∗∗∗p < 0.0001, NS (one-way ANOVA followed by Dunnett’s post hoc test). See also Figure S4.

Phosphorylation of many proteins triggers their translocation to a different cellular compartment for downstream signaling. Indeed, immunofluorescence analysis revealed that RSK underwent translocation to the nucleus in thymocytes stimulated with the K^b-OVA or K^b-Q4 tetramers, but not in those stimulated with K^b-Q4H7 or K^b-Q7 tetramers (Figure 4D), suggesting that the phosphorylation of RSK at Thr⁵⁶² induced specifically by the high-affinity peptide ligands is responsible for the nuclear translocation of RSK. Thus, similar to the induction of RSK phosphorylation (Figures 3F and 3G), induction of the nuclear translocation of RSK showed a sharp threshold between Q4 (negative selecting) and Q4H7 (positive selecting). Collectively, our results suggested that ERK5 phosphorylates RSK at Thr⁵⁶² and thereby induces the nuclear translocation of RSK selectively in response to high-affinity ligand stimulation.

RSK activity is required for antigen-induced thymocyte death

The specific regulation of the phosphorylation and nuclear translocation of RSK by negative selection stimuli prompted us to investigate whether RSK activity is required for the induction of cell death. During negative selection, high-affinity ligand stimulation initiates TCR downstream signaling that finally activates the apoptosis pathway.⁴¹ To test whether RSK activity determines the apoptotic fate of thymocytes, we treated OT-I Tap1^−/− thymocytes with BI-D1870, a potent RSK inhibitor,⁴² and cultured them for 20 h together with B16F1 cells that had been pulsed with the OVA peptide or its mutant versions as antigen-presenting cells (APCs). In the absence of BI-D1870, the cultures containing APCs pulsed with high-affinity peptides (OVA, Q4, or Q4R7) showed both a reduced percentage of viable thymocytes (annexin V⁻/propidium iodide^– cells) (Figure 5A) and an increased proportion of those positive for active caspase-3 (Figure 5B) compared with the cultures containing APCs pulsed with low-affinity peptides (T4, Q4H7, Q7, G4, or VSV) (Figure 5A). Treatment of cultures with BI-D1870 significantly attenuated the induction of apoptotic cell death by the high-affinity peptides (Figures 5A and 5B), indicating that inhibition of RSK activity prevents thymocyte apoptosis in response to strong TCR stimulation. We next examined the effects of BI-D1870 in the FTOC system. The number of surviving OT-I (Vα2⁺) cells was significantly increased in cultures treated with BI-D1870 during OVA or Q4 peptide-induced negative selection (Figures 5C and S5). Similar results were obtained with SL0101, another RSK inhibitor, in the case of Q4-induced negative selection (Figures 5C and S5). In contrast, neither the number of surviving OT-I cells nor the number of differentiated CD8⁺ SP thymocytes was increased by BI-D1870 or SL0101 treatment during stimulation with Q4H7. Collectively, these results indicated that RSK inhibition partially rescued cells from death during negative selection in the FTOC system.

Figure 5.

RSK activity is required for antigen-induced death of thymocytes

(A and B) OT-I Tap1^−/− thymocytes were cultured with peptide-pulsed B16F1 cells and in the absence or presence of the RSK inhibitor BI-D1870 (10 μM) for 20 h, after which the frequency of viable (annexin V⁻/propidium iodide [PI]^–) cells (A) or the proportion of active caspase-3-positive cells (B) among the thymocytes was determined by flow cytometry. Data are means ± SD (n = 4 independent experiments). ∗p < 0.05, ∗∗∗∗p < 0.0001 (two-way ANOVA followed by Bonferroni’s post hoc test).

(C) FTOC was performed as in Figure 1B but in the absence or presence of BI-D1870 (10 μM) or SL0101 (10 μM), after which thymocytes were isolated and subjected to flow cytometric analysis of TCR Vα2, CD4, and CD8 expression at the cell surface. Representative plots and quantitative data (means ± SD) from three independent experiments are shown. ∗p < 0.05, ∗∗p < 0.01, NS (two-way ANOVA followed by Dunnett’s post hoc test). See also Figure S5.

Phosphorylation of Nur77 at Ser³⁵⁴ induced by high-affinity stimulation is associated with its mitochondrial localization for apoptosis

We searched for a molecule that acts downstream of RSK phosphorylation and is responsible for the induction of apoptosis. Given that the nuclear receptor Nur77 was previously identified as a downstream target of RSK in a T lymphocyte cell line⁴³ and that the activity of Nur77 is controlled through phosphorylation by various kinases including RSK,^44,45 we examined whether the RSK-Nur77 axis might contribute to negative selection of thymocytes. Immunoblot analysis revealed that stimulation of OT-I Tap1^−/− thymocytes with K^b-OVA indeed induced Nur77 phosphorylation at Ser³⁵⁴ that was apparent within 2 min and which was correlated with the upregulation of RSK phosphorylation at Thr⁵⁶² (Figure S6A). Furthermore, Nur77 phosphorylation at Ser³⁵⁴ was increased by K^b tetramers loaded with peptides that induce negative selection (OVA or Q4), but not by those loaded with peptides that induce positive selection (Q4H7 or Q7) (Figure 6A). These results thus suggested that Nur77 is a target of RSK that is regulated specifically by high-affinity ligands associated with thymic negative selection.

Figure 6.

Phosphorylation of Nur77 at Ser³⁵⁴ induced by high-affinity ligands triggers its mitochondrial localization for apoptosis induction

(A) Immunoblot analysis of pNur77, pERK1/2, and pERK5 in OT-I Tap1^−/− thymocytes stimulated with peptide-K^b tetramers for the indicated times. A representative blot and quantitative data (means ± SD) for the pNur77/β-actin band intensity ratio from three independent experiments are shown.

(B) Immunoblot analysis of Nur77, α-tubulin (cytosolic marker), lamin B1 (nuclear marker), and cytochrome oxidase IV (COX IV, mitochondrial marker) in subcellular fractions prepared from lysates of OT-I Tap1^−/− thymocytes after stimulation with K^b-Q4 or K^b-Q4H7 tetramers for 10 min. Data are representative of two independent experiments.

(C) OT-I Tap1^−/− thymocytes were incubated in the absence or presence of the RSK inhibitor BI-D1870 (10 μM) for 15 min on ice and then stimulated with the K^b-Q4 tetramer for the indicated times at 37°C, after which cell lysates were subjected to immunoblot analysis of pNur77, pRSK, and pERK1/2. A representative blot and quantitative data (means ± SD) for the pRSK/β-actin or pNur77/β-actin band intensity ratio from three independent experiments are shown.

(D) Immunoblot analysis as in (B) for cells exposed or not to BI-D1870 (10 μM) before stimulation as in (C). Data are representative of two independent experiments.

(E) Model for the role of the ERK-RSK-Nur77 pathway in thymic positive and negative selection. See also Figure S6.

Although Nur77 has been found to be required for negative selection in vivo,^46,47,48 its precise function has been unclear. In T lymphoma cell lines, Nur77 was shown to function as a transcriptional factor that regulates the expression of proapoptotic genes such as those for TRAIL, FAS ligand, NDG1, and NDG2, all of which initiate apoptosis through caspase-8.⁴⁹ Nur77 has also been shown to translocate to the mitochondrial membrane and to interact there with BCL2 during the induction of apoptosis associated with negative selection.^46,50 To examine whether Nur77 phosphorylated at Ser³⁵⁴ indeed undergoes translocation to mitochondria in our experimental setting, we fractionated lysates of stimulated thymocytes into different subcellular compartments and subjected the fractions to immunoblot analysis (Figure 6B). In unstimulated thymocytes, Nur77 was located predominantly in the nuclear fraction. In contrast, stimulation with K^b-Q4, but not that with K^b-Q4H7, induced marked enrichment of Nur77 in the mitochondrial fraction, suggesting that Nur77 translocates to mitochondria in response to stimulation with negatively selecting ligands. To test whether RSK activity is necessary for Nur77 phosphorylation, we examined the effect of BI-D1870 on such phosphorylation at Ser³⁵⁴ (Figure 6C). BI-D1870 treatment indeed inhibited Nur77 phosphorylation at Ser³⁵⁴ induced by K^b-Q4 stimulation. We also examined the effect of BI-D1870 on the mitochondrial translocation of Nur77, and found that RSK inhibition also blocked the localization of Nur77 to mitochondria triggered by K^b-Q4 stimulation (Figure 6D). Kinase inhibitors often have off-target effects, with previous studies having shown that BI-D1870 can influence the mechanistic target of rapamycin (mTOR) and PI 3-kinase-AKT signaling pathways.⁵¹ To exclude the possibility that the observed effects of BI-D1870 on thymocytes might be due to such off-target action, we treated OT-I Tap1^−/− thymocytes with BI-D1870 and monitored the phosphorylation of mTOR downstream targets and AKT. However, no substantial differences in mTOR target phosphorylation or AKT phosphorylation were observed between Q4-stimulated cells treated or not treated with BI-D1870 (Figure S6B), suggesting that the effects of BI-D1870 on thymocytes are unlikely to be mediated through modulation of the mTOR or PI 3-kinase-AKT pathways. Overall, these results suggested that nuclear RSK phosphorylates Nur77 at Ser³⁵⁴ and thereby promotes its translocation to the mitochondrial membrane for apoptosis induction during negative selection.

In summary, our results implicate the ERK5-RSK-Nur77 phosphorylation axis in the induction of apoptosis during thymic negative selection (Figure 6E). Stimulation of thymocytes with high-affinity (negatively selecting) ligands thus elicits TCR signaling that leads to the phosphorylation of ERK5, which in turn phosphorylates RSK at Thr⁵⁶² and induces its translocation to the nucleus. Nuclear RSK then phosphorylates Nur77 at Ser³⁵⁴ and triggers its translocation from the nucleus to the mitochondrial membrane for apoptosis induction. Phosphorylation of RSK at Thr⁵⁶² is not substantially increased by stimulation with low-affinity (positively selecting) ligands, with the result that the RSK-Nur77 phosphorylation axis is differentially controlled by negatively and positively selecting ligands. We propose that phosphorylation of RSK represents a mechanism by which DP thymocytes generate a stepwise and binary signal in response to stimulation by ligands with TCR affinities that differ in a graded manner across the threshold for positive versus negative selection.

Discussion

Despite many studies having investigated the mechanisms underlying thymic selection, how DP thymocytes discriminate ligand stimuli with a wide range of TCR affinity, initiate distinct downstream signaling events, and acquire unique cell fates have remained unclear. We have now generated a quantitative dataset of phosphorylation sites that are regulated in response to stimulation of DP thymocytes by a series of ligands with graded TCR affinities that span the threshold for positive and negative selection. Our data thus provide a valuable resource for future research on known or undefined aspects of signaling downstream of TCR that are key to DP thymocyte fate during thymic positive and negative selection.

We have here shown that 119 phosphorylation sites of 107 proteins in cluster 2 of our hierarchical clustering analysis are regulated differentially by positive or negative selection ligands. Many phosphoproteins present in this cluster have been poorly defined to date with regard to their specific role in thymic selection. Of interest, the phosphorylated proteins of both clusters 2 and 3 show enrichment for proteins related to cytoskeletal organization. Aspects of TCR signaling that differ between positive and negative stimulation have been identified as early as TCR proximal activity, with the maximal TCR signaling required for negative selection being accompanied by longer-lasting and tighter pMHC-TCR complex formation that elicits the entire set of downstream phosphorylation events.^4,32 It is possible that cytoskeletal changes play a distinct role in the signaling events that discriminate between positive and negative selection.

To identify downstream regulators from our data, we focused on kinases with a possible link to apoptosis, and we found that RSK phosphorylation at Thr⁵⁶² is induced by stimulation with negatively selecting ligands but not by that with positively selecting ligands. This phosphorylation event is regulated by MEK-ERK signaling, given that it was blocked by inhibition of this pathway. Phosphorylation of ERK1/2 occurs in a distinctive manner during positive or negative selection as a result of a difference in the compartmentalization of the activation process, which was shown in some studies to be essential for both positive and negative selection.^4,52 However, in vivo studies with ERK1/2 conditional knockout mice revealed that ERK1/2 is required only for positive selection, not for negative selection.^39,53,54,55 A possible explanation for this discrepancy is an overlapping role of ERK1/2 and similar kinases in the phosphorylation of target proteins, with contradictory data having been obtained with regard to the effects of MEK inhibitors. Targets of ERK5 thus overlap with those of ERK1/2,⁵⁶ and ERK5 is phosphorylated in response to TCR stimulation during negative selection.⁴⁰ Forced expression of a dominant-negative form of MEK5, an upstream kinase of ERK5, was also shown to inhibit negative selection.⁴⁰ Given that pharmacological inhibition of MEK1/2 was shown also to result in the inhibition of other MEK-ERK pathways including those mediated by MEK5,⁵⁷ ERK5 may be more important than ERK1/2 for RSK phosphorylation at Thr⁵⁶² in DP thymocytes.

We found that phosphorylation of Nur77 at Ser³⁵⁴ by RSK is a key downstream mechanism that specifically leads to apoptosis during negative selection. Previous studies have shown that Nur77 is required for thymic apoptosis,^48,58 but the mechanism by which it contributes to the apoptotic process during negative selection has remained elusive. Nur77 translocates to the mitochondrial membrane, where it interacts with BCL2, during negative selection,⁴⁶ but it was also shown to contribute to apoptosis through transcriptional regulation of apoptosis-related genes.⁴⁹ Our data support the former mechanism of Nur77 regulation in DP thymocytes, and they suggest that this mechanism is controlled by ERK signaling in response to strong TCR stimulation. Consistent with this notion, ERK5 was shown to be required for negative selection, although its mechanism of action was attributed to transcriptional regulation of apoptosis genes through Nur77.⁴⁰

The phosphoproteome of peripheral T cells was recently characterized with the use of the same system based on K^b tetramer stimulation as in our study.⁵⁹ Although this previous study shares certain aspects with our present study, there are also important differences between the two. First, the previous study used tetramers with a broader range of TCR affinity (OVA, T4, and G4) compared with the narrow range spanning the threshold for positive and negative selection of those used in our study (OVA, Q4, Q4H7, and Q7). Second, although many proximal signaling molecules are shared between thymocytes and peripheral T cells, TCR responsiveness differs between the two cell types. For instance, costimulatory molecules such as CD28 are more important for peripheral T cell responses than for thymocyte responses.^60,61 TCR sensitivity may also differ between preselected thymocytes and mature T cells.¹⁶ Third, the physiological outcomes of TCR stimulation also differ between peripheral T cells and thymocytes. Thymocytes stimulated by high-affinity ligands undergo apoptosis, whereas peripheral T cells typically proliferate or produce cytokines in response to such TCR stimulation. Slight differences in downstream effector responses may contribute to these distinct outcomes, which are related to the developmental and functional processes specific to each cell type. And fourth, the previous study identified Ser³⁶⁹ of RSK as a phosphorylation site that shows a gradual response according to the peptide affinity threshold, whereas we found that RSK phosphorylation at Thr⁵⁶² occurs in a digital manner. Different RSK phosphorylation sites may therefore be regulated in thymocytes and peripheral T lymphocytes, leading to distinct downstream outcomes. Further study to identify and characterize the shared and distinct aspects of TCR proximal signaling between thymocytes and peripheral T cells is warranted.

With the use of physiologically relevant TCR stimulation by ligands spanning the affinity threshold for positive and negative selection, we have now shown that Nur77 is regulated through phosphorylation by RSK, with such phosphorylation triggering Nur77 translocation to mitochondria. Our results thus add a new layer to the mechanism of Nur77 regulation during negative selection.

Limitations of the study

Recent insight into the thymic selection process suggests that its location and timing during thymic development are important.⁶² Given that the effects of OVA peptide ligands in FTOC differ to some extent from those in adult mice, additional factors may also play a role in thymic selection in vivo.⁶³ In addition, the signals that accompany positive and negative selection in vivo or in situ appear to be different than what is observed after in vitro stimulation with tetramers.⁶⁴ However, TCR affinity is a key factor that distinguishes the fates of thymocytes between positive and negative selection, and our finding in this paper using tetramer stimulation in vitro appears to recapitulate at least this aspect of the process.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-pERK1/2(Thr202/Tyr204) (D13.14.4E)	Cell Signaling Technology	Cat#4370; RRID: AB_10694057
anti-pLAT(Tyr171)	Cell Signaling Technology	Cat#3581; RRID: AB_2157730
anti-pRSK(mouseThr562/human Thy573)	Cell Signaling Technology	Cat# 9346; RRID: AB_330795
Anti-pNur77(mouse S354/human S351) (D22G5)	Cell Signaling Technology	Cat#5095; RRID: AB_10695108
Anti-RSK1/2/3(32D7)	Cell Signaling Technology	Cat#9355 RRID: AB_10693963
Anti-pAKT (S308)(C31E5)	Cell Signaling Technology	Cat#2965; RRID: AB_331692
Anti-p4E-BP1(S65) (174A9)	Cell Signaling Technology	Cat#9456; RRID: AB_823413
4E-BP1(53H11)	Cell Signaling Technology	Cat#9644; RRID: AB_10691384
Anti-pp70 S6 Kinase (Thr389)(108D2)	Cell Signaling Technology	Cat#9234; RRID: AB_2269801
Anti-p70 S6 Kinase(49D7)	Cell Signaling Technology	Cat#2708; RRID: AB_390722
Anti-pY(4G10)	Millipore	Cat#16-184; RRID: AB_310790
Anti-HSP90 (68/Hsp90)	BD Biosciences	Cat#610418; RRID: AB_397799
Anti-pZAP70(mouse Tyr318/human Tyr319)	BD Biosciences	Cat#612575; RRID: AB_399864
Anti-pZAP70(mouse Tyr290/human Tyr 292) (A16038)	BioLegened	Cat# 691902 RRID: AB_2632759
Anti-Nur77(E-6)	Santa Cruz Biotechnology	Cat# sc-166166; RRID: AB_2153745
Anti-pERK5 (1.T218/Y220)	Santa Cruz biotechnology	Cat# sc-135760 RRID: AB_2250338
Anti-COXIV(A21348)	Molecular Probes	Cat# A-21348; RRID: AB_221509
Anti-α-Tubulin(TU-1)	Invitrogen	Cat#13-8000; RRID: AB_2533035
Anti-Lamin B1(polyclonal)	Sigma-Aldrich	Cat# SAB2101352; RRID: AB_10605289
Anti-Beta Actin Monoclonal antibody	Proteintech	Cat# 66009-1-Ig RRID: AB_2687938
APC anti-CD4(RM4-5)	BD Biosciences	Cat# 553051; RRID: AB_398528
BV421 anti-CD8a(53-6.7)	BD Biosciences	Cat# 563898; RRID: AB_2738474
V450 anti-Active Caspase3-V450(C92-605)	BD Biosciences	Cat# 560627; RRID: AB_1727415
APC/Cyanine7 anti-mouse CD4 Antibody(RM4-5)	BioLegend	Cat# 100526; RRID: AB_312727
PE anti-Vα2 TCR(B20.1)	BioLegend	Cat# 127808; RRID: AB_1134183
Armenian hamster monoclonal antibodies to CD3ε (145-2C11)	BioLegend	Cat# 100301; RRID: AB_312666
Goat polyclonal antibodies to Armenian hamster IgG (H+L)	BioFX	Cat# 100301; RRID: AB_312666
AlexaFlour-488 goat anti-Rabbit IgG (H+L)	Thermo Fisher scientific	Cat # A124; RRID: AB_2892842
IRDyeR 680RD Donkey anti-Mouse IgG (H + L)	LI-COR Biosciences	Cat# 925-68072 RRID: AB_2814912
IRDye® 800CW Donkey anti-Mouse IgG Secondary Antibody	LI-COR Biosciences	Cat# 926-32212 RRID: AB_621847
IRDye® 680RD Donkey anti-Rabbit IgG Secondary Antibody	LI-COR Biosciences	Cat# 925-68073 RRID: AB_2716687
IRDye® 800CW Donkey anti-Rabbit IgG Secondary Antibody	LI-COR Biosciences	Cat# 926-32213 RRID: AB_621848
Anti–Mouse IgG, HRP-conjugated	Promega	Cat#W402B; RRID: AB_430834
Anti-Rabbit IgG (H+L), HRP Conjugate	Promega	Cat#W4011; RRID: AB_430833
Chemicals, peptides, and recombinant proteins
BI-D1870	Enzo Life Science	Cat#BML-EI407
U0126	Cayman Chemicals	Cat#70970
GSK2334470	Sigma	Cat#SML0217
SL0101	Cayman Chemicals	Cat#11704
MytomycinC	Nakarai	Cat# 20898-21
Hoechst33342	Dojindo	Cat# H342
Trypsin	Thermo	PI90057
Lys-C	Thermo	95001
15N213C6-lysine hydrochloride	Wako	Cat#123-06081
15N413C6-arginine	Wako	Cat#010-24041
TRizol reagent	Thermo Fisher Scientific	Cat#15596018
Calcium indicator Flour-4	Dojindo	Cat#F311
Fluoromount	Diagnoistic BioSystems	Cat#K024
Propidium Iodide	Merck	Cat#P4170
VSV (RGYVYQGL)	Eurofins Genomics	N/A
OVA peptides (SIINFEKL)	Eurofins Genomics	N/A
Q4 (SIIQFEKL)	Eurofins Genomics	N/A
Q4R7 (SIIQFERL)	Eurofins Genomics	N/A
T4 (SIITFEKL)	Eurofins Genomics	N/A
Q4H7 (SIIQFEHL)	Eurofins Genomics	N/A
Q7 (SIINFEQL)	Eurofins Genomics	N/A
G4 (SIIGFEKL)	Eurofins Genomics	N/A
OVA257-264, H-2Kb (SIINFEKL) tetramer	Prepared in house following NIH tetramer core facility protocol	N/A
Q4, H-2Kb (SIIQFEKL) PE-tetramer	NIH tetramer core facility	N/A
Q4H7, H-2Kb (SIIQFEHL) PE-tetramer	Prepared in house following NIH tetramer core facility protocol	N/A
Q7, H-2Kb (SIINFEQL) PE-tetramer	NIH tetramer core facility	N/A
VSV, H-2Kb (RGYVYQGL) PE-tetramer	Prepared in house following NIH tetramer core facility protocol	N/A
Streptavidin-R-Phycoerythrin (monomer tetramerization)	Prozyme	Cat# PJRS25
OVA257-264, H-2Kb (SIINFEKL) PE-monomer	NIH tetramer core facility	N/A
Q4H7, H-2Kb (SIIQFEHL) PE-monomer	NIH tetramer core facility	N/A
VSV, H-2Kb (RGYVYQGL) PE-monomer	NIH tetramer core facility	N/A
Critical commercial assays
APC-Annexin V	BD Biosciences	Cat# 550475 RRID: AB_2868885
C18 Sep-Pak (500 mg) cartridge	Waters	Cat# WAT043395
10 x Annexin V Binding Buffer	BD Biosciences	Cat# 51-66121E
SepPak C18 Sep-Pak (50 mg)	Waters	Cat# WAT054955
ProBond Nickel-Chelating Resin	LifeTechnology	Cat# No.R801-15
Deposited data
Raw Mass Spectrometry Data Files	This paper	JPOSTConsortium via the PRIDE partner repository, JPST001845, (PXD036851 for ProteomeXchange)
Experimental models: Cell lines
Mouse: DPK	A gift from S. Yamasaki (Osaka University)	N/A
Human: B16F1	ATCC	N/A
Experimental Models: Organisms/Strains
Mouse: TAP1-/- OT-I	A gift from S. Yamasaki (Osaka University)	N/A
Software and algorithms
ZEN imaging software	Zeiss	https://www.zeiss.com/microscopy/int/downloads/zen.html
FlowJo v10	Tree Star	https://www.flowjo.com/solutions/flowjo/downloads
GraphPad Prism v8, v9	GraphPad Software	N/A
Image J2	NIH	N/A
MaxQuant v1.3.0.0	Cox and Mann65	https://www.maxquant.org/
Perseus v1.5.1.0	Tyanova et al.66	https://maxquant.net/perseus/
Panther tools	Mi et al.31	www.pantherdb.org/
Seq2Logo-2.0	Thomsen et al.67	https://services.healthtech.dtu.dk/service.php?Seq2Logo-2.0
ImageStudioLite	LI-COR	https://www.licor.com/bio/image-studio-lite/

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Keiichi I. Nakayama (nakayak1@bioreg.kyushu-u.ac.jp).

Materials availability

Mouse lines obtained from other laboratories may require a Material Transfer Agreement with the providing scientists. Commercially available reagents are indicated in the key resources table.

Experimental model and study participant details

Animals

OT-I Tap1^–/– mice were described previously.¹⁶ They were maintained in the specific pathogen–free animal facility at Kyushu University in accordance with institutional guidelines under the following conditions: ambient temperature of 22°C, relative humidity of 50% to 60%, a 12-h-light, 12-h-dark cycle, and free access to water and standard rodent chow. All mice had a C57BL/6 genetic background and were used at 4 to 8 weeks of age for experiments with isolated thymocytes. Fetal thymic lobes were obtained from pregnant mice at E14.5 to E15.5 for FTOC. All experiments were approved by the animal ethics committee of Kyushu University.

Cell lines

DPK cells were cultured under an atmosphere of 5% CO₂ at 37°C in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) before incubation in SILAC medium. B16F1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS.

Method details

Experimental design

Mouse experiments were approved by the animal ethics committee of Kyushu University. Phosphoproteomics analysis was performed with four biological replicates. Other experiments with thymocytes or FTOC were performed with the indicated numbers of replicates. Immunoblot band intensities were measured with the use of ImageJ software (NIH).

Fetal thymus organ culture (FTOC)

FTOC was performed as described.²⁵ In brief, thymic lobes were excised from mice at E14.5 to E15.5 and were cultured for 3 days on 13-mm/0.8-mm Nucleopore membranes (GE Healthcare) in RPMI 1640 medium supplemented with 10% FBS and the OVA(257–264) peptide (SIINFEKL) or its indicated mutant forms at 20 μM. For RSK inhibitor experiments, BI-D1870 (10 μM) or SL0101 (10 μM) was added to the culture medium. Thymocytes were then harvested from the thymic lobes and stained for 30 min on ice with allophycocyanin-conjugated or allophycocyanin/cyanine7-conjugated antibodies to CD4, Brilliant Violet 421 (BV421)–conjugated antibodies to CD8α (53-6.7), and PE-conjugated antibodies to TCR Vα2 (B20.1) in FACS buffer (sterile phosphate-buffered saline supplemented with 4% FBS and 0.1% NaN₃). The cells were then washed three times with FACS buffer, suspended in FACS buffer containing propidium iodide (2 μg/ml), and analyzed with a FACSVerse or FACSAriaII instrument (BD Biosciences). Dead cells were excluded on the basis of propidium iodide staining, and thymocytes were gated according to SSC-A versus FSC-A and the Vα2⁺ population before analysis of CD4 and CD8 expression.

Tetramer staining assay

Thymic lobes were minced with scissors and were subjected to hemolysis as previously described.¹⁴ The thymocytes were then washed with phosphate-buffered saline (PBS), suspended at a density of 1 × 10⁷ cells/ml in FACS buffer, and incubated on ice for 15 min with PE-labeled and peptide-loaded H-2K^b tetramers (100 nM). The cells were then washed twice with FACS buffer before analysis with a FACSVerse instrument (BD Biosciences).

Calcium flux analysis

Isolated thymocytes were suspended at a density of 1 × 10⁷ cells/ml in RPMI 1640 medium supplemented with 10% FBS and were incubated for 30 min at 37°C under 5% CO₂ with the Ca²⁺ indicator Fluo-4 (1.0 μM). The cells were washed twice with RPMI 1640 and once with cHBSS (Ca²⁺- and Mg²⁺-free Hanks’ balanced salt solution supplemented with 1% FBS, 1 mM MgCl₂, 1 mM EGTA, and 10 mM HEPES [pH 7.3]), suspended in cHBSS, warmed to 37°C, and stimulated with peptide-loaded H2-K^b tetramers (100 nM). CaCl₂ (5 mM) was added during event collection with a FACSCalibur instrument (BD Biosciences). The mean fluorescence ratio was calculated with FlowJo software (TreeStar).

Immunoblot analysis

Isolated thymocytes were suspended at a density of 1 × 10⁷ cells/ml in PBS and maintained on ice until stimulation at 37°C with peptide-loaded H2-K^b tetramers (100 nM) for the indicated times. For experiments with kinase inhibitors, the cells were incubated on ice with the inhibitor in PBS for 15 min before stimulation at 37°C in the continued presence of the inhibitor. The stimulated cells were lysed for protein extraction with the TRIzol reagent. The final protein pellet was suspended in a solution containing 2% SDS and 0.1 mM Tris-HCl (pH 7.5), the protein concentration was determined with the bicinchoninic acid (BCA) assay and adjusted to 2 mg/ml, and the suspension was mixed with an equal volume of 2× sample buffer (2% SDS, 100 mM Tris-HCl [pH 6.8], 20% glycerol, and 10% 2-mercaptoethanol), heated at 100°C for 10 min, and subjected to SDS-polyacrylamide gel electrophoresis. Immunoblot analysis was then performed as previously described.¹⁴ Briefly, the separated proteins were transferred to an Immobilon-P or FL membrane (Millipore), which was then probed consecutively with primary antibodies and HRP- or IRDye-conjugated secondary antibodies. Immune complexes were detected using Pierce SuperSignal West Pico reagents for HRP-conjugated antibodies. The blots were scanned with ImageQuant LAS 4000 (GE healthcare) or an Odyssey Infrared Imaging System (LI-COR Biosciences). Quantification of immunoblot band intensity was performed by densitometry with Image Studio Lite software (LI-COR Biosciences), and band intensity ratios were expressed relative to those for the control condition.

Preparation of the spike-in SILAC standard

SILAC labeling (heavy) medium (Wako) was reconstituted with [¹⁵N₂,¹³C₆]lysine hydrochloride (40 mg/l) and [¹⁵N₄,¹³C₆]arginine (200 mg/l) and with 10% FBS that had been dialyzed with the use of a 3.5 kDa–cutoff membrane. DPK cells were cultured in this medium for >2 weeks with at least five passages, washed three times with PBS, suspended in PBS containing Armenian hamster monoclonal antibodies to CD3ε (145-2C11) at 10 μg/ml, and incubated on ice for 30 min. The cells were stimulated by the addition of prewarmed (37°C) PBS containing goat polyclonal antibodies to Armenian hamster immunoglobulin G (10 μg/ml) and incubation for 2 min at 37°C and were then lysed by the addition of 3 volumes of TRIzol reagent. Lysates of nonstimulated cells (those exposed to PBS without antibodies) were also prepared. The protein concentration of the lysates was determined with the BCA assay. The stimulated and nonstimulated cell lysates were mixed in equal amounts, and the mixture was used as a spike-in SILAC heavy standard for phosphoproteomics experiments.

Phosphoproteome sample preparation

OT-I Tap1^–/– thymocytes were suspended on ice in RPMI 1640 medium at a density of 1 × 10⁸ to 2 × 10⁸ cells/ml and were stimulated for 2 min by the addition of prewarmed (37°C) RPMI 1640 containing peptide-loaded H2-K^b tetramers (100 nM) and incubation at 37°C. The cells were lysed for protein extraction with the TRIzol reagent, and the protein concentration of the lysates was determined with the BCA assay and adjusted to 3 mg/ml. The samples (3 mg of protein) were mixed with the same amount (3 mg) of the spike-in standard prepared from stimulated DPK cells, and the mixtures were subjected to sequential digestion at 37°C with LysC for 3 h and with trypsin for 16 h in a final volume of 6 ml. Reduced cysteine/cystine residues of the resulting peptides were blocked by treatment with 5 mM tris-(2-carboxyethyl) phosphine for 30 min at 37°C followed by alkylation with 12.5 mM iodoacetamide for 30 min at room temperature. The digests were desalted with the use of a C18 Sep-Pak (500 mg) cartridge and were then enriched for phosphopeptides by incubation with gentle rotation for 90 min at 37°C with Fe³⁺-immobilized metal ion (IMAC) beads (5:1 w/v, protein:beads). The Fe³⁺-IMAC beads were prepared as previously described.¹⁴ In brief, Nickel-Probond resin was stripped of Ni³⁺ by washing with 10 volumes each of 50 mM EDTA (pH 8.0), water, and 0.1% acetic acid, and the resin was recharged with 100 mM FeCl₃ in 0.1% acetic acid and then washed extensively first with 0.1% acetic acid and then with 0.1% trifluoroacetic acid (TFA) in 60% acetonitrile. The phosphopeptide-enriched Fe³⁺-IMAC beads were washed twice with 3 ml of 0.1% TFA in 60% acetonitrile and once with 0.1% TFA in 2% acetonitrile, and phosphopeptides were then eluted with 4 ml of 1% phosphoric acid, desalted with the use of a C18 Sep-Pak (50 mg) cartridge, and fractionated by high-pH reversed-phase chromatography into eight fractions for LC-MS/MS analysis.

LC-MS/MS analysis

All samples were analyzed with an Orbitrap Velos Pro LC-MS/MS system (Thermo Finnigan) equipped with a Paradigm MS4 HPLC pump and HTC-PAL autosampler (CTC Analytics). L-column C18 material (3 μm, CERI) was packed into an in-house–pulled fused silica capillary (inner diameter of 0.1 mm, length of 15 cm; CERI) at a constant pressure of 230 bar with the use of a high-pressure chamber equipped with an HPLC pump. Samples were dissolved in 2% acetonitrile containing 0.1% TFA and injected into a pre-column (L-column micro, CERI, with an inner diameter of 0.3 mm and length of 5 mm), which was then washed with the same solution before elution at a flow rate of 200 nl/min with a linear gradient of 2% to 35% B for 100 min, 35% to 95% B for 1 min, and 95% B for 1 min (where A is 0.1% formic acid in water and B is 100% acetonitrile). MS and MS/MS spectra were obtained by a data-dependent acquisition method with the following settings: scan range of 300 to 2000 m/z, resolution of 60,000 at 400 m/z, lock mass function enabled, and dynamic exclusion option on. The top 15 most intense MS ions were selected for the MS/MS search.

MS/MS database searching

The raw MS/MS data were processed with MaxQuant software version 1.5.0.0⁶⁶ and the mouse Uniprot database (downloaded on 15 August 2015), and with the following minor changes from the default settings: Oxidized Methionine (M), Acetylation (Protein N-term), and Phospho (STY) as variable modifications, and Carbamidomethyl (C) and SILAC labels (Arg 6/Lys 4 and Arg 10/Lys 8) as fixed modifications. A maximum of two miscleavages was permitted, and ITMS top peaks per 100-Da window was set to 8 and minimum peptide length to 6. The matching between runs option was enabled with a time window of 2 min. Peptide and site FDR thresholds in MaxQuant were each set to a maximum of 1%. Changes in phosphorylation induced by the stimuli were quantified on the basis of the relative amounts of experimental phosphopeptides and the corresponding spiked-in phosphopeptides. The fold change between the unstimulated sample (PBS) and the peptide-K^b tetramer stimulation sample was calculated by dividing the SILAC L (peptide-K^b tetramer)/H (spike-in) ratio by the SILAC L (PBS)/H (spike-in) ratio as previously described.⁶⁹

Bioinformatics analysis

Data analyses were performed with Microsoft Office Excel and Perl and with the bioinformatics platform Perseus (Max Planck Institute of Biochemistry). GO annotation enrichment analysis was performed with the Panther tool (www.pantherdb.org).³¹ The ice-logos motif was analyzed with the use of Seq2Logo 2.0 (https://services.healthtech.dtu.dk/service.php?Seq2Logo-2.0) with default parameters and mouse reference as a background distribution frequency of the amino acids. The identified phosphopeptide sequences of each cluster were aligned from position −6 to +7 relative to the phosphorylated residue.

Immunofluorescence analysis

Thymocytes stimulated as for immunoblot analysis were collected, fixed with 4% formaldehyde for 10 min at 37°C, washed with PBS, and permeabilized for 15 min at 37°C with PBS containing 0.5% saponin and 0.5% bovine serum albumin. The cells were then washed three times with PBS and incubated overnight at 4°C first with antibodies to RSK1/2/3 and then with AlexaFluor 488–conjugated secondary antibodies. Nuclei were stained by exposure of the cells to Hoechst 33342. The cells were mounted onto coverslips with the use of Cytospin (Thermo Fisher) and examined with a Zeiss LSM-700 Meta Confocal Laser-Scanning Microscope. ImageJ software was applied to binary colors to determine the positive areas for Hoechst 33342 and RSK staining in each image. The area of RSK staining overlapping with that of Hoechst 33342 staining was quantified, and if it was >50% the cell was considered to be positive for nuclear RSK. More than 100 cells for each condition were scored in each experiment.

In vitro clonal deletion assay

B16F1 cells were cultured overnight at a confluent density in a 24-well plate and were then exposed to mitomycin C (0.1 mM) for 4 h to block cell proliferation. They were washed with PBS, pulsed for 4 h with the indicated peptides (10 μM) in RPMI 1640 supplemented with 10% FBS, washed, and then cultured together with OT-I Tap1^–/– thymocytes (5 × 10⁵ cells/well) in the absence or presence of BI-D1870 (10 μM) for 20 h. The thymocytes were then collected and stained either with allophycocyanin-labeled annexin V and propidium iodide or, after fixation and permeabilization, with antibodies to active caspase-3. They were finally resuspended in FACS buffer for analysis with a FACSVerse instrument (BD Biosciences).

Subcellular fractionation

Subcellular fractionation was performed as previously described,⁷⁰ with minor modifications. In brief, stimulated thymocytes were washed with PBS, suspended in cytosolic lysis buffer (250 mM sucrose, 70 mM KCl, 137 mM NaCl, 4.3 mM Na₂HPO₄ and 1.4 mM KH₂PO₄ [pH 7.2], digitonin [200 μg/ml], and protease inhibitor cocktail), and incubated for 10 min on ice. The cells were centrifuged at 1000 × g for 5 min at 4°C, and the resulting supernatant was saved as the cytosolic fraction. The pellet was washed twice with cytosolic lysis buffer, suspended in mitochondrial lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 0.2% Triton X-100, 0.3% Nonidet P-40, and protease inhibitor cocktail), and incubated for 10 min on ice. The suspension was centrifuged at 10,000 × g for 10 min at 4°C, and the resulting supernatant was kept as the mitochondrial fraction. The pellet was suspended in SDS lysis buffer (1% SDS and 0.1 mM Tris-HCl [pH 7.5]) and saved as the nuclear fraction. The cytosolic, mitochondrial, and nuclear fractions were subjected to acetone precipitation and dissolved in SDS lysis buffer for immunoblot analysis.

Quantification and statistical analysis

No statistical methods were applied to predetermine sample size. Experiments were not randomized, and investigators were not blinded to allocation during experiments and outcome assessment. The permutation-based FDR of 5% on a t test was applied to correct for multiple testing with an S₀ value³⁰ of 0.45 with the use of Perseus software. Comparisons among more than two groups were performed by one-way analysis of variance (ANOVA) followed by Dunnett’s post hoc test, or by two-way ANOVA followed by Bonferroni’s or Dunnett’s post hoc test. Such statistical analysis was performed with Microsoft Excel or GraphPad Prism 9. A p value of <0.05 was considered statistically significant unless otherwise stated in the figure legends.

Published: August 9, 2023

Data and code availability

• •The phosphoproteome data have been deposited in ProteomeXchange Consortium via the jPOST partner repository⁶⁸ with the dataset identifier JPST001845 (PXD036851 for ProteomeXchange).
• •This paper does not report original code.
• •Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request.