LymphoAtlas: a dynamic and integrated phosphoproteomic resource of TCR signaling in primary T cells reveals ITSN2 as a regulator of effector functions

Abstract

T-cell receptor (TCR) ligation-mediated protein phosphorylation regulates the activation, cellular responses, and fates of T cells. Here, we used time-resolved high-resolution phosphoproteomics to identify, quantify, and characterize the phosphorylation dynamics of thousands of phosphorylation sites in primary T cells during the first 10 min after TCR stimulation. Bioinformatic analysis of the data revealed a coherent orchestration of biological processes underlying T-cell activation. In particular, functional modules associated with cytoskeletal remodeling, transcription, translation, and metabolic processes were mobilized within seconds after TCR engagement. Among proteins whose phosphorylation was regulated by TCR stimulation, we demonstrated, using a fast-track gene inactivation approach in primary lymphocytes, that the ITSN2 adaptor protein regulated T-cell effector functions. This resource, called LymphoAtlas, represents an integrated pipeline to further decipher the organization of the signaling network encoding T-cell activation. LymphoAtlas is accessible to the community at: https://bmm-lab.github.io/LymphoAtlas.

Keywords: ITSN2; LymphoAtlas; TCR signaling network; dynamic biological processes; phosphoproteomics.

Figure EV1. Description and quality control of the phosphoproteomic data set

1. Immunoblot of equal protein amounts from total lysates of primary mouse CD4⁺ T cells left unstimulated (−) or stimulated for the indicated times. Global tyrosine phosphorylation was probed with antibody against phosphorylated tyrosine (anti‐P‐Tyr). Phosphorylation of proteins was assessed using phospho‐specific antibodies, as indicated on the right. Anti‐VAV1 was used as loading control.

2. Phosphosites identified in this study (biological replicates A–D). * Determined from the “evidence.txt” tables (MaxQuant search FDR ≤ 0.01). **Number of phosphorylated sites (or combination of sites in the case of multiply phosphorylated peptides) determined from MaxQuant “Phospho (STY)Sites.txt” tables. Classes 1, 2, 3, and 4 correspond to a phospholocalization score of > 75%; 50% < ≤ 75%; 25% < ≤ 50%; and ≤ 25%, respectively.

3. Histogram of the number of phosphosites per protein that were identified (left) and significantly regulated (right) in the phosphoproteome.

4. Protein lengths against the number of phosphosites identified per protein.

5. Log₁₀‐transformed relative abundances (iBAQ) of the proteins in the proteome (Voisinne et al, 2019) against the number of phosphosites identified per protein in the phosphoproteome.

6. Histogram of the log₁₀‐transformed relative abundances (iBAQ) of the proteins in the proteome (gray). The subset of these proteins that are phosphorylated in the phosphoproteome are indicated in red. Dashed lines: median values for each population.

7. UniProt keywords enriched (hypergeometric test P‐value ≤ 0.05, fold change ≥ 1.5, number of annotated proteins ≥ 2) in the set of phosphoproteins compared with the set of all proteins identified in the CD4⁺ T cell proteome. Fold changes and P‐values (*P = 0) are indicated for each term. Keyword terms highlighted in red are not enriched if the same analysis is restricted to phosphosites detected in the unstimulated condition.

Data information: The protein Titin (35,213 amino acids) is omitted in (D) and (E), and the 2 proteins with more than 30 phosphosites (Srrm1: 40 sites; Srrm2: 100 sites) are omitted in (C–E).

Figure 1. Description of the phosphoproteomic data set and its generation

1. A

   Experimental workflow to quantify phosphorylations occurring at the early time points after TCR stimulation. Digested peptides were analyzed by MS before further enrichment to control the impact of TCR stimulation on protein abundances (proteome). Further steps of TiO₂ enrichment and phosphotyrosine immunoprecipitation (pY‐IP) were conducted to quantify changes in the phosphorylated serines and threonines (pST phosphoproteome), as well as phosphorylated tyrosines (pY phosphoproteome), respectively (see the Materials and Methods section for more details). Schematic representations of the measured phosphokinetics are presented on the right.

2. B, C

   Phosphosites identified after TiO₂ enrichment (B) and phosphotyrosine immunoprecipitation (pY‐IP) (C) Left panel: pie chart presenting the proportion of identified phosphorylated serines (pS), threonines (pT), and tyrosines (pY) in the two data sets. Right panel: cumulative number of unique identified phosphosites across the biological replicates (3 and 4 independent experiments for the TiO₂ and pY‐IP, respectively). Phosphorylated serines, threonines, and tyrosines are presented individually next to the total number of identified sites (pSTY). The phosphosites that were not previously reported in any species in the PhosphoSitePlus database (www.phosphosite.org) are reported in isolation in the graph on the right.

Figure 2. Identification and dynamics of TCR‐regulated phosphorylation sites

1. A, B

   Output of the statistical analyses of the TiO₂‐fractionated (A) and the pY‐IP‐fractionated (B) data sets. Left panel: number of phosphosites significantly regulated upon TCR stimulation presented in bar plots for each time point, next to the total number of regulated sites across the entire time course (total). The number of phosphorylated threonines (pT) and serines (pS) regulated in the TiO₂ data set is indicated in yellow and green, respectively. Right panel: representative volcano plot presenting the statistical significance distribution against the log₂‐transformed fold change between 300 s and the unstimulated control. For each condition, phosphosites were considered significantly up‐regulated (red) or down‐regulated (blue) when displaying a corrected P‐value ≤ 0.05 (ANOVA test) and an absolute fold change ≥ 1.75 (see Materials and Methods for more detailed information).

2. C

   Mapping of a subset of TCR‐regulated phosphosites on the canonical TCR signaling network. Phosphosites are color‐coded according to their kinetic of phosphorylation, as presented in (D).

3. D

   t‐SNE view and clustering analysis of phospho‐regulated sites. The 13 clusters were defined using local density within the t‐SNE plot (see Materials and Methods).

Figure EV2. Statistical and functional analysis of the phosphoregulations upon TCR activation

1. Left: number of unambiguously identified proteins across the following data sets: UniProt reviewed entries (www.uniprot.org); a proteome of CD4⁺ T cells (Voisinne et al, 2019) (PXD012826); the phosphoproteome and the set of TCR‐regulated phosphoproteins. Right: repartition of unambiguously identified proteins across the proteome, the phosphoproteome, and the set of TCR‐regulated phosphoproteins. The area of the circles in the Euler diagram illustrates the relative number of proteins per data set without being strictly proportional.

2. Output of the statistical analysis of protein abundances during TCR stimulation. Number of proteins significantly regulated upon TCR stimulation presented in bar plots for each time point next to the total number of regulated proteins across the entire time course (total). The corresponding percentage of the proteome that is regulated is indicated above the bars. Right panel: representative volcano plot presenting the statistical significance distribution against the log₂‐transformed fold change between 30 s (maximum number of regulated proteins) and the unstimulated control. For each condition, proteins were considered significantly up‐regulated (red) or down‐regulated (blue) when displaying a corrected P‐value ≤ 0.05 (ANOVA test) and an absolute log₂‐transformed fold change ≥ 1 (see Materials and Methods for more detailed information).

3. Enriched UniProt–keywords (hypergeometric test P‐value ≤ 0.05, fold change ≥ 1.5, number of annotated proteins ≥ 2) in the set of regulated phosphosites compared with the set of all identified phosphosites. Fold changes and P‐values are indicated for each term.

Figure EV3. Dynamics of phosphorylation and selected functional modules induced by TCR stimulation

1. Dynamics and cluster distribution of regulated phosphotyrosines, phospho–serines, and phosphothreonines.

2. Illustration of TCR‐regulated phosphosites within the polycomb‐repressive complex 1 (PRC1) (Ub: ubiquitination; Ac: acetylation). The corresponding log₂‐transformed MS intensities measured upon TCR activation for each of the biological replicates are shown below. Box plot elements: Center line corresponds to median, box limits correspond to the first and third quartiles, and whiskers indicate variability from Q1–1.5. IQR to Q3 + 1.5 IQR.

3. Short‐time TCR stimulation induces FOXO3 phosphorylation and NFATC2 dephosphorylation and promotes nucleus exit or entry, respectively. CD4⁺ T cells left unstimulated (−) or stimulated for 2 and 5 min with anti‐CD3 plus anti‐CD4 antibodies were subjected to nuclear/cytoplasmic fractionation before immunoblot analysis with antibodies specific for NFATC2 and FOXO3. Arrows indicate phosphorylation (P) and dephosphorylation (deP) forms of the transcription factors. Lamin‐B2 and GAPDH are used to control purity of nuclear and cytoplasmic extracts.

4. t‐SNE plot highlighting phosphosites associated with the UniProt keywords “Protein biosynthesis” and “Translation regulation”. Dot transparency is scaled according to the P‐value corresponding to the local enrichment of the annotation term (hypergeometric test, see method for more detailed information).

5. Left: Dynamics of TCR‐regulated phosphosites associated with the UniProt keywords “Protein biosynthesis” and “Translation regulation” (dots color‐coded by cluster). Right: Schematic representation of the translational initiation and elongation complexes. TCR‐regulated phosphosites are represented as small dots color‐coded by cluster.

Figure 3. Enrichment and dynamics of cellular processes triggered by early TCR signals

1. Heat map showing UniProt keywords (https://www.uniprot.org/) enriched in each cluster. Black dots indicate clusters with significantly enriched keyword terms (hypergeometric test P‐value ≤ 0.01, fold change ≥ 2, number of annotated sites ≥ 2; gray squares: fold change < 1).

2. t‐SNE plot highlighting phosphosites associated with the keyword terms specified above each plot. Dot transparency is scaled according to the P‐value corresponding to the local enrichment of the annotation term (hypergeometric test, see Materials and Methods for more detailed information).

3. Dot plot showing dynamics of scaled intensities over time of selected phosphosites associated with the keyword “Cell projection” (dots color‐coded per cluster). The linear structure of the selected proteins with the corresponding regulated phosphosites is depicted below.

4. Similar representation as in (C) for the terms “Tyrosine‐protein kinase”, “Serine/threonine‐protein kinase”, and “Receptor”.

5. Illustration of the transcription factors for which activation through nucleus‐cytoplasmic shuttling is regulated by phosphorylation. The normalized dynamics/intensities of the corresponding phosphosites are indicated on the right panel (as in C, D).

Figure 4. Dynamics of kinase activities

1. Heat map showing kinase activities based on enrichment of their specific substrates for each cluster. Black dots indicate clusters with significantly enriched kinase activity (hypergeometric test P‐value ≤ 0.05, fold change ≥ 2, number of annotated sites ≥ 2; gray squares: fold change < 1).

2. t‐SNE plot displaying local enrichment of phosphosites corresponding to known substrates associated with the specified kinase. Dot transparency is scaled according to the P‐value corresponding to the local enrichment of the annotation term (hypergeometric test, see Materials and Methods for more detailed information).

3. Illustration of the kinase–substrate interconnections characterizing the AMPK, AKT, and ERK signaling pathways. Substrate phosphosites dynamics are displayed and color‐coded according to their clusters, and the corresponding scaled kinetics are presented on the right.

4. Immunoblot analysis of equal amounts of proteins from total lysates of CD4⁺ T cells treated with DMSO, U0126 (10 μM), or AKTi VIII (10 μM) and left unstimulated (−) or stimulated for 15, 120, or 300 s with anti‐CD3 and anti‐CD4 antibodies, probed with the indicated phospho‐specific antibodies. Anti‐ERK1/2 immunoblot served as a loading control.

5. Equal amounts of proteins from total lysates of CD4⁺ T cells left unstimulated (−) or stimulated for 120, 300, or 600 s with anti‐CD3 and anti‐CD4 antibodies were analyzed by immunoblot with antibodies probing the indicated phosphosites/proteins.

Data information: In the phosphosite labels, “+” and “/” separate the phospholocalizations of doubly phosphorylated peptides or peptides having two proteins identifications with different phosphorylation localizations, respectively. Note that although TBC1D1−pS231 site falls into C1, its t‐SNE coordinates are close to the clusters associated with dephosphorylation, a result consistent with its late dephosphorylation state.

Figure 5. ITSN2 deficiency lowers TCR activation threshold

1. t‐SNE plot highlighting regulated pY sites (for sake of space, only the gene names are reported). Phosphosites of STAM2 and ITSN2 are displayed in red.

2. Cas9‐EGFP OT‐I CD8⁺ T cells were transfected with control sgRNA (sgEGFP) or with two different sgRNA targeting Itsn2 (sgITSN2‐1 or sgITSN2‐2). Transfected cells were stimulated for 48 h with N4 peptide MHC tetramers (0.1 nM) in the presence or absence of soluble anti‐CD28 antibody or with IL‐7 as control. Cell‐surface expression of CD69 and CD5 was analyzed by flow cytometry (Z‐score normalization of the geometric means of fluorescence MFI per experiment) in four independent experiments. Comparison between sgEGFP and sgITSN2 conditions was performed using a paired t‐test. Box plot elements: Center line corresponds to median, box limits correspond to the first and third quartiles, whiskers indicate variability from Q1–1.5. IQR to Q3 + 1.5 IQR. Outliers are shown as black dots.

3. Proliferation of OT‐I Cas9‐EGFP CD8⁺ T cells transfected with EGFP or with ITSN2 sgRNA activated for 48 h with N4 peptide MHC tetramers (0.01–0.1 nM) in the presence or absence of soluble anti‐CD28 antibody or with PMA and ionomycin (PMA/Iono) or with IL‐7. Data are presented as mean ± SD from four proliferative measures of two independent sgEGFP and sgITSN2 transfections. Data are representative of three independent experiments.

4. IFN‐γ production of cells treated as in (C) was assessed by intracellular staining after 24 h of stimulation.

5. Proliferation of similar cells as in (C) activated for 48 h in vitro with a range of increasing doses of N4 peptide MHC tetramers in the presence of soluble anti‐CD28 antibody.

Data information: For panels (C) and (E): Data are representative of at least three independent experiments, and comparison between sgEGFP and sgITSN2 conditions was performed using a paired t‐test (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; NS, non‐significant).

Figure EV4. Comparative analysis of phosphoproteomes performed with OT‐I CD8⁺ and CD4⁺ T cells

1. Euler diagram indicating the number of phosphosites identified (3612) and regulated (640) upon TCR stimulation in CD4⁺ and OT‐I CD8⁺ T cells.

2. Comparison of phosphosite intensities between CD4⁺ and OT‐I CD8⁺ T cells. Only sites regulated upon TCR stimulation in CD4⁺ T cells and identified in OT‐I CD8⁺ T cells were considered. Log₂‐transformed phosphosite intensities with imputed missing values were normalized across biological replicates using the mean intensity and subsequently averaged for each condition of stimulation. Pearson correlation coefficient R = 0.54.

3. Distribution of Pearson's correlation coefficients across phosphosites regulated upon TCR stimulation in CD4⁺ T cells and identified in OT‐I CD8⁺ T cells. Phosphosite intensities were normalized as in (B). The correlation coefficient was computed only for phosphosites with intensity values available in both data sets for a minimum of four stimulatory conditions (n = 584).

4. Overlay dynamics of selected phosphosites in CD4⁺ and OT‐I CD8⁺ T cells.

Figure EV5. ITSN2 inactivation by the CRISPR/Cas9 system in primary mouse T cells

Cas9‐EGFP OT‐I CD8⁺ T cells were transfected with control sgRNA (sgEGFP) or with two different sgRNA targeting Itsn2 (sgITSN2‐1 and sgITSN2‐2).

1. EGFP expression was assessed by flow cytometry in cells transfected with sgEGFP or without guide (none).

2. Equal amounts of total lysates from cells transfected with sgEGFP or sgITSN2 were analyzed by immunoblot using anti‐ITSN2 or anti‐LCK antibodies. The arrows indicate short and long ITSN2 isoforms. Data are presented as mean ± SD from three independent experiments.

3. Transfected cells were stimulated for 48 h with N4 peptide MHC tetramers in the presence of soluble anti‐CD28 antibody or with IL‐7 as control. Surface expression of CD69 and CD5 in cells was analyzed by flow cytometry. A representative FACS profile is shown.

Figure 6. Reduced TCR down‐modulation in ITSN2‐targeted cells

1. Immunoblot analysis of equal amounts of proteins from total lysates of Cas9‐EGFP OT‐I CD8⁺ T cells transfected with sgEGFP, sgITSN2‐1, or sgITSN2‐2 and left unstimulated (−) or stimulated for 2 or 5 min with N4 tetramers (20 nM). Upper panel: membrane probed with antibody to phosphorylated tyrosine (anti‐p‐Tyr) or anti‐LAT (loading control). Lower panel: membrane probed with antibodies to anti‐CBL‐pY774, anti‐ZAP70‐pY319/352, anti‐AKT‐pT308, anti‐ERK1/2‐pY204/T202, or anti‐ERK1/2 (loading control).

2. Heat map showing normalized levels (z‐score) of indicated surface markers and protein phosphorylations (left margin) from Cas9‐EGFP OT‐I CD8⁺ T cells transfected with sgEGFP or sgITSN2‐1 left unstimulated (−) or stimulated for 1, 3, or 10 min with N4 tetramers (1 or 10 nM) and analyzed by mass cytometry. Z‐scores were calculated from hyperbolic arcsine (arcsinh)‐transformed intensities. Results obtained from independent transfections with sgITSN2‐1 and sgITSN2‐2 are compiled in Fig EV5A.

3. Similar cells as in (A) stimulated with 1 nM of N4 peptide for the indicated time points (left panel) or stimulated with increased doses of N4 peptides for 4 h (right panel) were analyzed by FACS for surface TCR expression. Data are presented as mean ± SEM, and comparisons were performed using a paired t‐test (*P ≤ 0.05, **P ≤ 0.01; ***P ≤ 0.001). A representative analysis of three independent experiments is shown.

4. Heat map showing scaled expression levels (z‐score) of indicated surface markers and protein phosphorylations (left margin) from Cas9‐EGFP OT‐I CD8⁺ T cells transfected with sgEGFP or sgITSN2‐1 and stimulated for 6 h with increasing doses of N4 peptides or PMA/ionomycin (PI) analyzed by single‐cell mass cytometry (left). Z‐scores were calculated from hyperbolic arcsine (arcsinh)‐transformed intensities. Signal difference (mean arcsinh difference) between sgEGFP and sgITSN2‐1 transfected cells is depicted on the right panel.