Grb2-Mediated Recruitment of USP9X to LAT Enhances Themis Stability following Thymic Selection

The selection of thymocytes that express a functional self-tolerant TCR and the commitment of immature CD4⁺CD8⁺ (double-positive [DP]) thymocytes to the CD4⁺ or the CD8⁺ T cell lineage constitute two major decision checkpoints during T cell development (1). The affinities of TCRs for self-peptides bound to the MHC (self-pMHCs) determine the fate of T cells during selection in the thymus. Thymocytes that express TCRs that bind with high affinity to self-pMHCs are negatively selected and are triggered to undergo apoptotic cell death, whereas thymocytes that express TCRs that bind with low affinity to self-pMHCs are positively selected and become CD4 or CD8 single-positive (SP) thymocytes, according to the MHC class restriction of their TCRs (2).

The fate of thymocytes during these processes is dependent upon signals transmitted by the TCR and by coreceptors, such as CD4 and CD8, that bind together with the TCR to self-pMHCs. The strength of the TCR–self-pMHC interaction and the differential kinetics of CD4 and CD8 surface expression during the transition from the DP stage to the SP stage govern the intensity and the duration of TCR signals, which, in turn, dictate distinct developmental outcomes (3). Although positive selection and CD4 lineage commitment are promoted by sustained or repeated TCR signaling waves (4–8), negative selection and CD8 lineage commitment are preferentially induced by short-lived or interrupted TCR signaling events, respectively (5, 8–10).

Signals triggered by TCRs and coreceptors propagate through a complex network of intracellular signaling proteins that exhibit enhancing or inhibitory functions. TCR engagement by pMHC leads to activation of the protein tyrosine kinases Lck and ZAP70, which phosphorylate the transmembrane adaptor protein LAT on multiple tyrosine residues (11). Phosphorylated LAT proteins serve as docking sites for cytosolic adaptors (e.g., Grb2, GADS, and SLP76) that enable the recruitment of effectors proteins, such as PLCγ1, Vav1, and Itk, which, in turn, trigger distinct signaling pathways important for the regulation of T cell development (12). Inhibitory proteins, such as the protein tyrosine phosphatase SHP-1, function to attenuate or interrupt TCR-mediated signals, according to the strength of the TCR–pMHC interactions (13–15).

Themis is a recently identified TCR signaling protein that has an important role during T cell development (16–18). It contains a proline-rich C-terminal sequence (PRS), which binds to the adaptor protein Grb2 (19, 20), and two globular cysteine-containing, all-β in Themis (CABIT) domains (17). Themis is recruited to LAT through Grb2 following TCR engagement (21). Although the molecular function of Themis has long remained elusive, it was shown recently that Themis enhances TCR signaling in thymocytes by blocking the inhibitory activity of SHP-1 (22). This action is mediated by the CABIT domains of Themis, which bind to the phosphatase domain of SHP-1 and promote or stabilize oxidation of the SHP-1 catalytic cysteine (22).

The phenotype of Themis^−/− mice shows that Themis is essential for the positive selection of DP thymocytes (16–18) and for their commitment to the CD4 lineage (18), suggesting that Themis might be important to perpetuate and sustain TCR signals during these steps of T cell development. However, previous studies reported that the transcription of the gene encoding for Themis is strongly downregulated in DP cells at the early stages of positive selection (16, 17). We show in this article that, despite the strong downregulation of Themis mRNA, the amount of Themis protein increases following positive selection. We found that Themis stability is controlled by the ubiquitin-specific protease USP9X, which removes K48-linked ubiquitin chains on Themis following TCR engagement. Biochemical analyses reveal that USP9X binds directly to the N-terminal CABIT domain of Themis. Using Grb2-deficient thymocytes, we show that Grb2 promotes the recruitment of Themis/USP9X to LAT, which stabilizes Themis expression during T cell development.

C57BL/6 mice are from Janvier (Le Genest-Saint-Isle, France). Themis^−/− (18), Grb2^+/− (23), and Usp9X^flox/flox (24) mice were described previously. AND TCR-transgenic mice were from Taconic Farms. OT-1 and OT-2 TCR-transgenic mice were provided by Prof. R. Liblau (INSERM UMR 1043, Toulouse, France), and CD4-Cre mice were provided by Dr. O. Joffre (INSERM UMR 1043). All of the experiments were conducted with sex- and age-matched mice between 6 and 10 wk old that were housed under specific pathogen–free conditions at the INSERM animal facility (Zootechnie US-006; accreditation number A-31 55508), which is accredited by the French Ministry of Agriculture to perform experiments on live mice. All experimental protocols were approved by the local ethics committee and are in compliance with the French and European regulations on the care and protection of laboratory animals (European Commission Directive 2010/63).

The following Abs and reagents were used in this study: anti-Vav1 (C-14), anti–c-Cbl (A9), anti-GAPDH (FL-335), anti-ERDj3 (C-7), and anti-GFP (B-2) (all from Santa Cruz Biotechnology); anti-V5 and anti-USP9X (Bethyl Laboratories); anti–p-SHP1(Y564), anti–p-ZAP (Y319), anti-ubiquitin (P4D1), anti-K48 (D9D5), and anti-K63 (D7A11) polyubiquitin (all from Cell Signaling Technology); anti-hemagglutinin (HA) and anti-FLAG (M2) (Sigma-Aldrich); and anti-Themis (Q13-1103), anti-Grb2, anti–protein tyrosine phosphatase 1C/SHP-1, and anti-LAT Y226 (all from BD Biosciences). Anti-Themis (3A3) Ab was provided by Dr. P.E. Love (National Institutes of Health, Bethesda, MD). Anti-FLAG M2 and anti-V5 Agarose Affinity Gel were from Sigma-Aldrich. Thymocytes were incubated for the indicated period of times with MG132 (20 μM) or with PR-619 (10 μM) (Sigma-Aldrich). Fluorochrome-conjugated Abs against CD4 and CD69 were from BioLegend, and those against CD8α, CD5, TCRβ, and Themis were from eBioscience. Biotin-conjugated Abs against CD3ε (145-2C11) and CD4 (GK1.5) were purchased from BD Biosciences. For flow cytometry analysis, single-cell suspensions from thymus, spleen, or lymph nodes were incubated in PBS, 0.5% BSA, and 2 mM EDTA containing the appropriate Abs. Intracellular staining was performed by fixing cells with 4% formaldehyde and permeabilizing them with 0.5% Triton X-100. Cell detection was performed on a BD LSR II flow cytometer (BD Biosciences), after analysis was performed with FlowJo software (TreeStar).

For the purification of thymocyte subpopulations, total thymocytes were labeled with the appropriate Abs. Each population was isolated with a BD FACSAria cell sorter. For gene-expression studies, total cell RNA was isolated with an RNeasy Kit (QIAGEN). RNA samples (500 ng of each) were reverse transcribed with SuperScript III Reverse Transcriptase (Invitrogen) and were assayed by real-time PCR. Transcripts were quantified with a Roche LightCycler 480 System. Duplicates were run for each sample in a 96-well plate; β2-microglobulin (β2m) served as the endogenous reference gene. The relative quantification method was used, with the mRNA abundance of the gene of interest normalized to the abundance of β2m mRNA; the average of control thymocyte samples served as the calibrator value. Themis primers were exon-3 forward 5′-TGAAATCCAAGGTGTGCTGA-3′, exon-4 reverse 5′-CGTCCGTAGACAGCAACTGA-3′, β2m forward 5′-ACATACGCCTGCAGAGTTAAGCAT-3′, and β2m reverse 5′-CGATCCCAGTAGACGGTCTTG-3′.

Plasmids encoding for FLAG-ThemisWt, FLAG-ThemisΔPRS, GFP-Themis, and V5-USP9X were described previously (20, 25). HA-Ubiquitin, HA-UbiquitinK48R, and FLAG-USP19 were purchased from Addgene. Plasmids encoding for FLAG-ThemisΔCABIT1(1–243), FLAG-ThemisΔCABIT2(267–494), HA-UbiquitinK63R, HA-UbiquitinK33R, and HA-UbiquitinK29R were generated by site-directed mutagenesis using a QuikChange kit (Stratagene).

For the analysis of Themis ubiquitylation, 6 × 10⁷ thymocytes or 5 × 10⁶ Jurkat cells were stimulated or not with anti-CD3ε, anti-CD4, and streptavidin (30 μg/ml) at 37°C. Stimulation was stopped on ice, and cells were immediately centrifuged and resuspended in 140 μl of ice-cold lysis buffer (10 mM Tris, 150 mM NaCl, 1% Triton X-100, 2 mM Na₃VO₄, 5 mM NaF, 1 mM EDTA, 1% SDS, 10 mM iodoacetamide, 10 mM N-ethylmaleimide, and Protease Inhibitor Cocktail Tablet [Roche]). Lysates were boiled at 95°C for 10 min and sonicated two times for 5 s (30% duty cycle). Lysates were diluted in 1.4 ml of ice-cold lysis buffer without SDS and cleared by centrifugation at 12,000 rpm for 15 min at 4°C. Themis was immunoprecipitated from cleared lysate for 2 h at 4°C with 15 μl of protein A–Sepharose resin coated with 10 μl of serum containing polyclonal rabbit anti-Themis Abs. Resins were washed and incubated on a shaker for 15 min with elution buffer (100 μg/ml Themis antigenic peptides, 50 mM Tris, and 150 mM NaCl). Samples were further processed for Western blot analysis. For any other immunoprecipitation, thymocytes or Jurkat cells were resuspended in ice-cold lysis buffer (10 mM Tris, 150 mM NaCl, 1% Triton X-100, 2 mM Na₃VO₄, 5 mM NaF, 10% glycerol, and Protease Inhibitor Cocktail Tablet [Roche]) for 20 min on ice. Lysates were cleared by centrifugation at 12,000 rpm for 15 min at 4°C, and proteins were immunoprecipitated from cleared lysate with 15 μl of Sepharose resin coated with the indicated Abs for 2 h at 4°C.

Jurkat-TAg cells were maintained in RPMI 1640 medium supplemented with 10% FBS (Sigma-Aldrich). Human embryonic kidney epithelial cells (HEK293) were maintained in DMEM and 10% FBS. For transfection, 10⁷ Jurkat-Tag cells were washed three times with RPMI 1640 medium without FBS and resuspended in 100 μl of Ingenio electroporation solution (Euromedex) containing 20 μg of the indicated plasmids. Cells were electroporated with the Nucleofector II Device (H10 program). For lipofection, HEK293 cells were incubated for 6 h with 500 μl of DMEM containing 5 μg of the indicated plasmids and 15 μl of LipoD293 DNA (Tebu-bio).

Ten micrograms of purified human USP9X (E-552; Boston Biochem) and 10 μg of purified mouse Themis were incubated for 45 min in buffer containing 150 mM NaCl and 10 mM Tris (pH 7.4). The solution was adjusted with 0.2% Triton X-100, and USP9X was immunoprecipitated for 3 h at 4°C with 15 μl of protein A–Sepharose resin coated with anti-USP9X Abs. Samples were further processed for Western blot analysis.

Proteins were resolved by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane was blocked with 20% FCS for the analysis of Themis ubiquitylation and 5% milk for coimmunoprecipitation analysis in TBST for 1 h at room temperature prior to incubation with primary Abs at 4°C overnight. After washing, membranes were incubated with secondary Abs for 1 h at room temperature. Subsequently, membranes were incubated with ECL solution for 5 min in the dark, and luminescence was captured using a Bio-Rad XRS+ imager. Images were analyzed using Bio-Rad Image Lab software.

Statistical comparisons were carried out using the Fisher test to verify equal variance of the populations, followed by an unpaired or paired two-tailed t test or Mann–Whitney U test if variance was not equal. The p values are indicated in the figure legends.

We first compared the protein and mRNA levels of Themis in DP thymocytes, before selection (“preselection,” TCR^loCD69^loCD5^lo) or soon after they were stimulated by self-pMHC ligand (TCR^loCD69^hiCD5^hi). In accordance with previous studies (16, 17), we found that Themis mRNA levels are strongly decreased in TCR^loCD69^hiCD5^hi DP thymocytes compared with those in preselection DP thymocytes (Fig. 1A). Unexpectedly, we found that the amount of Themis protein is increased in DP thymocytes undergoing positive selection, indicating a posttranslational regulation of Themis expression (Fig. 1B). The amount of Themis protein was similarly increased in TCR^loCD69^hiCD5^hi DP thymocytes expressing class I (OT-1) or class II (OT-2) restricted TCRs, suggesting that the increased amount of Themis observed in these cells does not result from a shift in the TCR repertoire and is not specific to thymocytes positively selected by MHC class I or MHC class II self-pMHC interactions (Fig. 1C, 1D). Further analyses at later stages of T cell development show that Themis mRNA levels are decreased by ∼80% in CD69^hi CD4 SP thymocytes compared with those in preselection DP thymocytes, whereas the amounts of Themis protein are similar in these two populations (Fig. 1B). Finally, Themis protein levels are decreased in mature CD69^lo CD4 and CD8 SP thymocytes compared with preselection DP thymocytes, confirming the previously reported decrease in Themis protein expression in these mature thymocyte subsets (Fig. 1B).

The discrepancy between Themis mRNA and protein levels during the positive selection of DP thymocytes suggests that posttranslational events might be important to stabilize Themis expression at this stage of T cell development. Ubiquitin-specific proteases (USPs) remove mono- or polyubiquitin chains on proteins and, thus, enhance their stability by preventing them from degradation by the proteasome (26). Previously, mass spectrometry analysis of proteins that coimmunoprecipitate with Themis in thymocytes led us to identify several USPs (USP9X, USP24, USP19, and USP15) (Fig. 2A), among other proteins, as potential binding partners of Themis in these cells (27). To investigate the potential role of these interactions on Themis expression, we incubated thymocytes with PR619, a well-characterized inhibitor that blocks the activity of most deubiquitylating enzymes (DUBs) (28). We found that incubation of thymocytes with PR619 results in a rapid and dramatic decrease in the amount of Themis protein, whereas the amount of other signaling proteins, such as c-Cbl, was not affected, indicating that Themis expression is highly sensitive to DUB-mediated stabilization (Fig. 2B). We identified USP9X, but not the other USPs, as a binding partner of Themis in a yeast two-hybrid screen using a cDNA bank of murine splenocytes (Fig. 2C). Transfection of HEK293T cells with cDNA encoding tagged versions of Themis and USP9X confirmed the binding of USP9X to Themis (Fig. 2D).

To investigate whether USP9X controls the stability of Themis, we analyzed Themis protein levels in thymocytes from Usp9X^flox/flox mice expressing the Cd4-Cre transgene, which begins to be expressed in DP thymocytes (29). We found that the amount of Themis was similar in Usp9X^flox/flox (Usp9X^+/+) and Usp9X^flox/flox;Cd4-Cre (Usp9X^−/−) CD4⁻CD8⁻ double-negative (DN) thymocytes but was decreased in Usp9X^−/− DP thymocytes (Fig. 3A, 3B). Themis protein levels were further decreased in positively selected Usp9X^−/− CD4 and CD8 SP thymocytes and in peripheral CD4⁺ and CD8⁺ T cells (Fig. 3A–C). However, the amount of Themis remained higher in Usp9X^−/− SP thymocytes than in thymocytes treated with PR619 (Fig. 2B), indicating that USPs other than USP9X contribute to regulate Themis expression. Quantification of Themis mRNA amounts by real-time PCR showed similar expression levels in Usp9X^+/+ and Usp9X^−/− DP and CD4 SP thymocytes, showing that the decreased protein abundance of Themis in Usp9X^−/− thymocytes does not result from a transcriptional defect (Fig. 3D). The adaptor protein Grb2 and the tyrosine phosphatase SHP-1, which are known binding partners of Themis, were expressed normally in Usp9X^−/− CD4 SP thymocytes, indicating a selective effect of USP9X on Themis (Fig. 3E).

To examine whether the reduced expression of Themis in Usp9X^−/− thymocytes has a functional effect on Themis-mediated signaling, we compared the phosphorylation of SHP-1 and ZAP70, which are known to be impaired in Themis^−/− thymocytes, in Usp9X^+/+ and Usp9X^−/− CD4 SP thymocytes. We found that SHP-1 phosphorylation was decreased in Usp9X^−/− CD4 SP thymocytes compared with that in Usp9X^+/+ control cells (Fig. 3F). The phosphorylation of SHP-1 was more strongly impaired in Themis^−/− thymocytes than in Usp9X^−/− thymocytes, in accordance with the partial decrease in Themis expression levels observed in Usp9X^−/− thymocytes (Fig. 3F). Although the phosphorylation of ZAP70 was reduced in Themis^−/− thymocytes compared with that in control cells, it was not affected in Usp9X^−/− thymocytes, suggesting that the inhibition of SHP-1 by Themis remains sufficient in Usp9X^−/− thymocytes to promote the activity of potential targets of SHP-1, such as ZAP70 (Fig. 3F). To determine the impact of USP9X deficiency on positive selection, we next crossed Usp9X^−/− mice with transgenic mice expressing a fixed MHC class II–restricted αβ-TCR transgene (AND). We found similar proportions of CD4 SP thymocytes in AND Usp9X^+/+ and Usp9X^−/− mice, indicating that the reduced expression level of Themis in Usp9X^−/− mice is not sufficient to significantly impair positive selection (Fig. 3G). CD4 SP/CD8 SP ratios in thymocytes in Usp9X^−/− mice were also comparable to those in Usp9X^+/+ mice, suggesting that the choice for CD4 or CD8 lineage was also normal in Usp9X^−/− mice (Fig. 3H).

We next investigated whether Themis is directly regulated by ubiquitin-mediated modifications. Analysis of Themis in unstimulated thymocytes and in Jurkat cells transfected with cDNA encoding for tagged versions of Themis and ubiquitin showed that Themis is polyubiquitylated (Fig. 4A, 4B). Incubation of thymocytes with the proteasome inhibitor MG132 resulted in an increase in ubiquitylated Themis, supporting that Themis is targeted to the proteasome through ubiquitin chain modifications (Fig. 4C). We next investigated the type of linkage by which Themis–ubiquitin chains are assembled. We found that ubiquitylated Themis was recognized by ubiquitin K48 chain–specific Abs, but not by Abs specific for ubiquitin K63 chains, in freshly isolated thymocytes (Fig. 4D). Confirming these results, we found that Themis ubiquitylation was strongly reduced in Jurkat cells expressing tagged versions of ubiquitin mutated on its K48 residue (K48R), whereas it was only moderately affected in Jurkat cells expressing K63R, K29R, and K33R ubiquitin mutants (Fig. 4E).

We noticed that Themis ubiquitylation was increased when thymocytes were rested for 1 h in culture medium, suggesting that stimuli occurring in the thymic environment might dampen or prevent Themis ubiquitylation (Fig. 4F). Thus, we examined the effect of TCR cross-linking on Themis ubiquitylation and found that the amount of ubiquitylated Themis decreases following the stimulation of thymocytes or peripheral CD4⁺ T cells with anti-CD3 + anti-CD4 Abs (Fig. 4G, 4H). In contrast, we found that Themis ubiquitylation was enhanced in Usp9X^−/− thymocytes after stimulation, suggesting that ubiquitin ligases may further ubiquitylate Themis following TCR engagement but that USP9X overcomes this process and reduces Themis ubiquitylation to less than its baseline level by removing polyubiquitin chains on Themis (Fig. 4I).

We next investigated the molecular mechanism by which USP9X controls Themis expression and ubiquitylation. We found that USP9X coimmunoprecipitates with Themis in unstimulated thymocytes (Fig. 5A). TCR cross-linking with anti-CD3 and anti-CD4 Abs did not result in a significant increase in this interaction (Fig. 5A). Additionally, we found that Grb2 and Themis coimmunoprecipitate with USP9X in thymocytes (Fig. 5B). Transfection of HEK293T cells with cDNA-encoding tagged versions of Grb2 and USP9X confirmed the binding of USP9X to Grb2 (Fig. 5C), indicating that Grb2 and USP9X may interact in the absence of Themis. However, the coimmunoprecipitation of Grb2 with USP9X was reduced in Themis^−/− thymocytes, suggesting that part of the interaction between Grb2 and USP9X occurs through Themis in thymocytes (Fig. 5B). The mutation of Themis PRS, required for the binding of Themis to Grb2, did not prevent Themis interaction with USP9X in 293T cells, supporting a direct interaction between Themis and USP9X (Fig. 5D, 5E). Accordingly, we found that recombinant proteins of Themis and USP9X could coprecipitate in cell-free assays (Fig. 5F). We next investigated whether the CABIT modules of Themis were important for its interaction with USP9X and found that deletion of Themis’ CABIT1(1–260) domain, but not of its CABIT2(260–493) domain, impaired the coimmunoprecipitation of USP9X with Themis (Fig. 5D, 5E).

The recruitment of USP9X to LAT promotes its phosphorylation on one of its serine residues (S1600) and enhances its catalytic activity (30). Because Grb2 is important for the recruitment of Themis to LAT, we suspected that Grb2 might also be involved in the translocation of USP9X to LAT signaling complexes through its interaction with Themis. To address this possibility, we analyzed the interaction of USP9X with LAT in thymocytes partially deficient for Grb2 (Grb2^+/−) to avoid the substantial perturbations of early TCR signaling observed in thymocytes fully deficient for Grb2 (31). We found that the amounts of LAT that coimmunoprecipitate with USP9X following TCR engagement were decreased in Grb2^+/− heterozygote thymocytes (Fig. 6A). We then analyzed Themis ubiquitylation in Grb2^+/− thymocytes, suspecting that Grb2 might be important to regulate USP9X activity and, thereby, to positively regulate Themis deubiquitylation. We found that the ubiquitylation of Themis was increased in Grb2^+/− thymocytes compared with that in Grb2^+/+ thymocytes when thymocytes were stimulated with anti-CD3 + anti-CD4 Abs (Fig. 6B). Accordingly, we found that the expression levels of Themis were comparable in Grb2^+/+ and Grb2^+/− DN thymocytes but were decreased in Grb2^+/− DP thymocytes (Fig. 6C, 6D). The amounts of Themis protein were further decreased in SP thymocytes, whereas the relative amount of Themis mRNA remained unchanged in these cells (Fig. 6C–E). Expression of other binding partners of Grb2, such as Cbl or Vav1, remained unchanged in Grb2^+/− thymocytes, indicating that Grb2 selectively stabilizes Themis in thymocytes (Fig. 6F). Collectively, these results identify Grb2 as an important regulator of Themis stability that promotes the deubiquitylation of Themis by USP9X following TCR engagement.

The strong defect in T cell development initially reported in Themis^−/− mice and the concomitant lack of a clear molecular model to explain this phenotype have led to intensive investigations to resolve the molecular function of this protein. Recent findings suggesting that Themis enhances TCR signaling through the selective inhibition of the tyrosine phosphatase SHP-1 constitute an important breakthrough in the comprehension of this enigmatic molecule (22). Although the role of Themis in TCR signaling begins to be better understood, the mechanisms that regulate Themis expression and/or function have not yet been investigated. In this article, we show that Themis binds to the ubiquitin-specific protease USP9X, which sustains Themis expression transiently during positive selection, whereas the gene encoding for Themis is transcriptionally shutdown. USP9X associates with Themis/Grb2 signaling complexes and enhances Themis stability by removing K48-ubiquitin chains on Themis following TCR engagement. Our study suggests that Grb2 is required for the recruitment of Themis into TCR signaling complexes, as well as contributes to stabilize Themis expression to sustain TCR signals by recruiting USP9X to LAT, where its activity is enhanced following initial signaling events (30).

We show that Themis is ubiquitylated predominantly by K48-linked chains that have a well-established function in the targeting of proteins to proteasomes for degradation (32). Supporting this mechanism, ubiquitylated Themis accumulates when thymocytes are incubated with the proteasome inhibitor MG132. Themis ubiquitylation is detected in unstimulated thymocytes and Jurkat cells, suggesting that Themis is constitutively ubiquitylated in these cells and may have a high turnover rate. These results are supported by proteomic resources obtained in Jurkat cells that identify up to 21 ubiquitylation sites on Themis (33). The mechanism that promotes Themis ubiquitylation is unknown. We previously used mass spectrometry to characterize the interactome of Themis in thymocytes and identified several ubiquitin ligases, such as c-Cbl, Arih2, and Nedd4, as potential binding partners of Themis (27). The cellular levels of Themis remain unchanged in c-Cbl^−/− thymocytes (data not shown), suggesting that Arih2, Nedd4, or another ubiquitin ligase might be responsible for Themis ubiquitylation.

In this study, we show that the amount of ubiquitylated Themis is decreased following TCR cross-linking, suggesting that TCR-mediated signals promote deubiquitylation and stabilization of Themis. We identified Themis as a new target for the DUB USP9X, a protein that was recently shown to positively regulate TCR signaling (34). USP9X enhances TCR signaling by removing inhibitory monoubiquitylation from ZAP70 and by preventing the sequestration of ZAP70 in early endosomes (30). Interestingly, it was recently shown that the recruitment of USP9X to LAT in peripheral CD4 T cells induces its phosphorylation on S1600, which enhances its catalytic activity (30); however, the mechanism of USP9X recruitment to LAT was not identified in that study. Our biochemical analysis demonstrates that USP9X binds directly to Themis through its CABIT domains, and it may also bind directly to Grb2. Themis binds to the C-terminal SH3 domain of Grb2 via its PRS and is recruited by Grb2 to tyrosine phosphorylated LAT following TCR engagement (21). By facilitating the recruitment of Themis/USP9X complexes to LAT, Grb2 might promote activation of USP9X by serine phosphorylation, which then deubiquitylates Themis. Supporting this mechanism, we found that Themis ubiquitylation is increased in Grb2^+/− thymocytes. Thus, in addition to recruiting Themis to LAT signaling complexes, Grb2 is important for the stabilization of Themis once it becomes recruited to this transmembrane adaptor.

We show that the amounts of Themis protein are increased in DP thymocytes stimulated by self-pMHC ligand even though Themis mRNA is decreased, suggesting that positive-selection signals stabilize Themis protein expression and shut down Themis gene expression. The stabilization of Themis protein at this stage of T cell development might be important to compensate for the profound drop in Themis mRNA levels that could lead to a sudden decrease in TCR signaling and, consequently, impair positive selection or CD4 lineage commitment, which are known to be dependent upon long-lasting TCR signals (35). In contrast, the downregulation of Themis mRNA levels might be important in the longer term, in peripheral T cells, to persistently reduce the amount of Themis protein and, thereby, to dampen the responsiveness of these cells to self-ligands, which is required for the prevention of autoimmunity. It is interesting to note in this context that Themis and SHP-1 expression profiles are inversely correlated, with SHP-1 being more highly expressed in mature T cells than in immature DP thymocytes (36). Reducing the Themis/SHP-1 ratio in mature T cells might be important to relieve the brake that Themis exerts on SHP-1 and enable this phosphatase to inhibit weak TCR signals that may potentially lead to self-recognition (15).

We previously showed that positive selection is impaired when Themis expression level is decreased by 2-fold in AND TCR-transgenic mice hemideficient for Themis (Themis^+/−), indicating that a quantitative variation in Themis expression may affect positive selection (27). However, we found that positive selection and CD4 lineage choice occur normally in Usp9X^−/− mice, suggesting that the reduced expression of Themis observed in Usp9X^−/− DP thymocytes might not be sufficient to significantly impair these developmental processes. Notably, the partial reduction in Themis expression in Usp9X^−/− thymocytes only moderately attenuated SHP-1 phosphorylation in comparison with the strong reduction observed in Themis^−/− thymocytes, and it had no impact on Grb2 expression level, which is decreased in Themis-deficient thymocytes (27). It is possible that USP9X acts redundantly with other USPs to stabilize Themis expression during these late stages of T cell development. Supporting this hypothesis, the expression levels of Themis are only moderately decreased in Usp9X^−/− thymocytes in comparison with the dramatic decrease in Themis expression observed in thymocytes treated with the pan-DUB inhibitor PR619. Mass spectrometry analysis of Themis binding partners showed that Themis binds other USPs (USP24, USP19, and USP15), in addition to USP9X, that have no reported function in T cell development (27). Further studies will be required to evaluate the role of these proteins in Themis-mediated signaling events and more generally on thymic selection processes.

We previously showed that Themis enhances Grb2 stability by reducing its ubiquitylation in thymocytes (27). This suggests that Themis and Grb2 are more stable as a complex than as free unbound proteins. This mutual stabilization mechanism might be important to enrich thymocytes with effective Themis/Grb2 signaling complexes, which may facilitate the transmission of TCR signals and their persistence during the development of T cells. We showed, in the same article, that the transgenic expression of Themis in Grb2^+/− thymocytes expressing the class II–restricted AND TCR rescues the defect in positive selection resulting from the partial loss of Grb2 (27). We speculated at that time that Themis acts by readjusting the amount of intracellular Grb2 close to its level in wild-type thymocytes. This new study provides a different interpretation for these data, suggesting that the defect in positive selection in Grb2-deficient mice might result from the destabilization of Themis and from the subsequent decrease in its expression. Thus, an unexpected function of Grb2 could be to maintain the expression of Themis above a certain level and under which level TCR signals are not sufficient to effectively promote positive selection. Whether the transgenic expression of Themis rescues the defect in positive selection in thymocytes that are fully deficient for Grb2 remains to be addressed.

We thank P.E. Love for critical reading of the manuscript, the Centre de Physiopathologie de Toulouse Purpan flow cytometry facility, and the INSERM UMS006-CREFRE animal care facility.

The authors have no financial conflicts of interest.