Proteomic Analysis of Regulatory T Cells Reveals the Importance of Themis1 in the Control of Their Suppressive Function.

of Themis1 the of Abstract Regulatory T cells (Treg) represent a minor sub-population of T lymphocytes which is crucial for the maintenance of immune homeostasis. Here, we present a large-scale quantitative mass spectrometry study that defines a specific proteomic “signature” of Treg. Treg and conventional T lymphocyte (Tconv) sub-populations were sorted by flow cytometry and subjected to global proteomic analysis by single-run nanoLC-MS/MS on a fast-sequencing Q-Exactive mass spectrometer. Besides “historical” proteins that characterize Treg, our study identified numerous new proteins that are up- or down-regulated in Treg versus Tconv. We focused on Themis1, a protein particularly under-represented in Treg, and recently described as being involved in the pathogenesis of immune diseases. Using a transgenic mouse model over-expressing Themis1, we provided in vivo and in vitro evidence of its importance for Treg suppressive functions, in an animal model of inflammatory bowel disease and in co-culture assays. We showed that this enhanced suppressive activity in vitro is associated with an accumulation of Tregs. Thus, our study highlights the usefulness of label free quantitative methods to better characterize the Treg cell lineage and demonstrates the potential role of Themis1 in the suppressive functions of these cells. anti-CD45RB (16A), anti-CD39 (Duha59), anti-CD73 (TY/23), anti-PD-1 (J43), anti-CTLA4 (UC10-4F10-11), anti-GITR (DTA-1) and anti-CD127 (SB/119). Intracellular staining was performed using the Foxp3 staining buffer kit (e-Biosciences) according to the manufacturer’s instructions before data acquisition on a BD flow cytometer and analysis FlowJo software.


Introduction
Regulatory T cells (Treg) are a subset of CD4 + T cells that are characterized by the expression of the transcription factor Foxp3 (Forkhead box protein P3). They play a central role in maintaining peripheral immune tolerance and preventing autoimmune diseases (1). This is best exemplified by the severe systemic autoimmunity and lymphoproliferative disorders observed in Treg deficient Scurfy mice and in human IPEX patients carrying non-functional or hypomorphic alleles of the Foxp3 gene (2)(3)(4)(5). Furthermore, the quantitative or qualitative defect in Treg cells have also been implicated in the development of several common autoimmune and inflammatory diseases. In addition to the maintenance of self-tolerance, Treg population can also be exploited to establish immunologic tolerance to transplanted tissues (6). This has led to an increasing interest in the possibility of using Treg as a target for therapy to preserve and restore tolerance to self-antigens (in autoimmunity), to allergen (in allergy) and to alloantigens (in transplantation). However, an excessive Treg activity could coincidently impair immunity toward pathogens and tumors (7)(8)(9). It is thus critical to understand Treg functions and regulation to avoid potential negative side effects of such therapeutical setups.
In this study, we compared the proteomes of CD4 + Foxp3 + Treg (that include both CD25 + and CD25 -Treg) and CD4 + Foxp3conventional T cells (Tconv) in order to build a data set of proteins differentially regulated in these two cell populations. An important challenge in this context was to achieve sufficient proteomic analytical depth starting from the low protein amounts obtained from highly purified primary murine Treg cells. We herein present an optimized label free LC-MS/MS workflow that allowed us to build an extensive quantitative data set of proteins expressed in Treg and Tconv. Its thorough statistical analysis undercovered a specific proteomic signature of the CD4 + Foxp3 + Treg subset. Most of the differentialy 5 regulated proteins were up-regulated in Treg compared to Tconv, and could be induced by Foxp3 and responsible for Treg development and functions. However, the Treg phenotype also depends on the specific repression of some molecules. For example, the genome organizer SATB1, which is required for the induction of T effector (Teff) cytokines, was actively and continuously suppressed by Foxp3 in Treg in order to prevent the differentiation of Treg into Teff cell (10). Other proteins down-regulated in Treg are involved in their development and/or function, such as TCF7 (11) or ITK (12), which both were shown to modulate TCR signal strength and thereby the commitment of precursors into the Treg lineage. Accordingly, we also identified in the present study many proteins that are down-regulated in Treg compared to Tconv. Among them, Themis1 appeared as a protein particularly down-regulated in Treg cells (4-fold) and was thus selected for further in vitro and in vivo validation studies. We showed that overexpression of Themis1 in Treg led to an increase of their suppressive functions, suggesting its importance as a checkpoint control in the suppressive function of Treg.

11
Proliferation analysis by Cell Trace Violet staining. For proliferation assays, Treg populations (CD4 + CD62L high CD25 bright ) were sorted based on the expression of the endogenous CD25 and CD62L markers to exclude contamination by activated Tconv cells. This population contains more than 99% of Foxp3+ T cells (Fig. S4). Tconv, CD4 + CD62L high CD25naive T cells, were purified from wild type (WT) mice and were cultured in 96-well round-bottomed plates in the presence of CD4 + CD62L high CD25 bright Treg cells purified either from WT mice or Although selection through the GFP marker increases the recovery yield of Treg cells, this population represents only a very minor percentage of the total CD4 + T cells, and we typically isolated around 1.5-2 x 10 5 Treg cells per mouse. In order to optimize protein extraction, we lysed the cells with a strong detergent concentration (4% SDS) combined to sonication. This led to 5 to 10 µg of proteins per sample (from a pool of 5-7 mice). Total protein amounts from Treg and Tconv cell populations were then normalised, samples were trypsin-digested and analyzed in parallel by mass spectrometry using the label-free quantitative analytical workflow described in Fig. 1. In order to increase the Treg and Tconv proteome coverages using low amounts of starting material, several experimental workflows involving or not fractionation of the sample, and using different LC gradient lengths, were tested (Fig. S2A). The highest number of protein identification per run was obtained with a single run analysis of total cell extracts, using a 50 cm reversed-phase column packed in house and a long LC gradient (480min on a LTQ-Velos Orbitrap) to improve chromatographic separation of peptides. This protocol showing good reproducibility of the label-free measurement (Fig. S6). An outline of the data processing steps and statistical analysis is shown in Fig. S3 (24), the Folate receptor 4/Folr4 (also known as Sperm-egg fusion protein Juno/Izumo1r), and the ecto-5'-nucleotidase/Nt5e (CD73) (25). Conversely, proteins known to be under-expressed in Treg were significantly down-regulated in our data set (Fig. 2C). These include for example the DNA-binding protein SATB1, a transcription factor whose down-regulation in Treg was previously shown to be important for Treg phenotype (10).

18
We finally compared the results of our analysis performed on CD4 + Foxp3 + Treg with other available data on Treg, namely a recently published proteomic study performed on FACS sorted CD4 + CD25 + and CD4 + CD25murine T cells (11) and transcriptomic data set obtained from murine T cells, also sorted on CD25 (15).  Table 3). The top 50% of these proteins/genes, based on their mean log2-transformed fold change accross all studies, are shown in Fig. 4. Depending on the data processing method and statistics applied in each study, proteins/genes were reported or not as differentially expressed, as indicated in Fig. 4. A panel of 30 genes were consistently found significantly regulated in all 3 datasets (supplementary Table 3 19 example IRF4 or OX40/Tnfrsf4. Also, proteins highly specific of one cellular subset and undetectable in the other would not have a reported ratio due to dimethyl labelling used in that study (like IL2-RA/CD25 that shows no detectable signal in most of he CD25samples).
Finally, a set of 25 proteins were repeatedely found as variant in both proteomic studies, while showing no significant variation at mRNA level (such as RLTPR), indicating that posttranscriptional mechanisms may be involved in their regulation. In conclusion, the combined analysis of these three high quality data sets (supplementary Table 3

20
TCR signaling but its role as a positive or negative regulator of this pathway is not fully elucidated (35,36). Herein, we showed that Themis1 expression was not abolished in Treg, but maintained at a much-reduced level compared to Tconv, suggesting that the tight control of Themis1 level in the CD4 + Foxp3 + subset might have a functional relevance. We confirmed by immunoblot that the protein Themis1 is less abundant in Treg than in Tconv (Fig. 5A) and we showed that Treg express low amounts of Themis1 mRNA compared to Tconv, which indicates a transcriptional regulation of this gene in Treg (Fig. 5B).
To examine the functional impact of Themis1 expression on Treg development and suppressive functions, we used Themis1 transgenic mice (Themis1-Tg) in which Themis1 expression is driven by the human-CD2 promoter. This model, in which the expression of Themis1 in Treg is increased, was selected to perform gain-of-function experiments and assess wether changing the physiological level of Themis1 impacts Treg phenotype and suppressive activity. Of note, the abundance of Themis1 in Themis1-Tg Treg cells is approximately three fold higher than in wild-type Treg cells, but is similar to the abundance of Themis1 in wild-type Tconv cells (Fig.   5A). At mRNA level, Themis1 expression in Themis1-Tg Treg cells is also higher than in WT Treg and WT Tconv cells (Fig. 5B). We compared by flow cytometry the frequency, the absolute numbers and the phenotypic markers of Treg between WT and Themis1-Tg mice and showed that the overexpression of Themis1 had no major impact on the development of thymic and peripheral Treg ( Fig. 5C and 5D) and on the expression of the main peripheral Treg markers such as Foxp3, CD25, CD39, CD73, CD103, Cytotoxic T-lymphocyte protein 4/CTLA4, GITR and Interleukin-7 receptor subunit alpha/Il7r (CD127) before (Fig. 5E) and after TCR engagement (Fig. S7).
To examine the effect of Themis1 overexpression on the immunosuppressive function of Treg cells in vivo, we used a well-defined adoptive transfer model (37), which involved cotransfer

21
of CD4 + CD25 -CD45RB high colitogenic T cells together with Treg cells purified from either WT or Themis1-Tg mice (Fig. 6A). First, we controlled that Themis1 overexpression had no impact on the percentage of Foxp3 expressing cells among CD4 + CD25 bright sorted Treg, nor on the level of Foxp3 expression (Fig. S5). As expected, adoptive transfer of only CD4 + CD25 -CD45RB high colitogenic T cells into Rag2 -/mice induced severe colitis as assessed by macroscopic score and weight loss, starting from 3 weeks after transfer ( Fig. 6B and C, Fig.   S8A and B). This was associated with pronounced inflammatory cellular infiltration in the mucosa and submucosa, as well as significant epithelial destruction (Fig. S8C). These pathological phenotypes were efficiently blocked when colitogenic CD4 T cells were cotransferred with either WT Treg cells or Themis1-Tg Treg at an optimal ratio of 1:2 between Treg and colitogenic T cells (Fig. S8). To reveal a possible difference in the suppressive capacities of WT Treg and Themis1-Tg Treg cells, we then chose a suboptimal 1:4 ratio between Treg and colitogenic CD4 T cells that did not prevent completly the inflammation of the colon. Although we could not detect any difference in weight loss between animals injected with the two different types of Treg (Fig. 6B), we demonstrated by macroscopic evaluation of colonic tissue damage that Themis1-Tg Treg cells exhibited an enhanced ability to suppress the pathological activity of colitogenic CD4 T cells. This was associated with a significant decrease in colonic wall thickness, suggestive of a decreased inflammation in the colon (Fig. 6C). These observations suggest that Themis1-Tg Treg exhibit an enhanced suppressive activity in vivo.
To directly assess the immune-suppressive activity of Themis1-Tg Treg cells, we used an in vitro suppression assay using Treg cells purified from the spleen and lymph nodes of either WT or Themis1-Tg mice on the basis of their expression of high levels of CD25 and CD62L. This population contains more than 99% of Foxp3+ T cells (Fig. S4). These cells were co-cultured be considered as strong specific markers, and possibly, as promising candidates for functional studies when their role in Treg is not described yet. In addition, the integration of these results allowed to rescue some interesting candidates that would not be identified or not considered as statistically significant in a particular single study.
It is known that Treg cells require stimulation via the TCR in order to be fully functional (27).
Although the importance of Treg for immune tolerance is well recognized, our knowledge is particularly limited with respect to intracellular signaling molecules that regulate/modify Treg 27 acts as a positive or negative regulator of TCR signaling. Indeed, while initial investigations failed to pinpoint a major alteration in TCR signaling in Themis1-deficient thymocytes, the defect of T cell development in Themis1-/-mice suggested that Themis1 might act as an enhancer of TCR signaling (31)(32)(33)(34)50). Contrasting with this interpretation, two recent reports suggest that Themis1 acts as an attenuator of TCR signaling in thymocytes, by reducing the signaling threshold between positive and negative selection (35,52). Themis1 was shown to bind to the inhibitory phosphatase SHP-1/Ptn6 (also known as Tyrosine-protein phosphatase non-receptor type 6) and was proposed to act as an adaptor protein that recruits SHP-1 in the vicinity of proximal TCR signaling complexes, to dephosphorylate them and reduce TCR signal strength (35,52). However, more recently, through the use of TCR signaling reporter mice, a study conversely identified Themis1 as an enhancer of TCR signaling during the positive selection of thymocytes (36). This positive effect of Themis1 was proposed to be mainly mediated through the guanine nucleotide exchange factor Vav1, a protein shown by quantitative mass spectrometry analysis to be a major component of the Themis1 interactome in thymocytes, together with the Growth factor receptor-bound protein 2/Grb2 and SHP-1 (36). Therefore, Themis1's mechanisms appear to be quite refined, with potentially a dual function as a negative regulator of TCR signaling through SHP-1, and a positive regulator of TCR signaling through Vav1.
In this study, through proteomic analysis and complementary functional studies of Themis1 overexpression in Treg, we made several important observations. First, we identified Themis1 as the only TCR signaling molecule, together with ITK, under-represented in Treg. Second, by using Themis1-Tg mice, we showed that the overexpression of Themis1 in Treg (to a similar level as in Tconv) did not modify their expression of Foxp3 but increased their suppressive potential in an in vivo model of IBD. Third, through an in vitro approach, we directly 28 demonstrated that the overexpression of Themis1 in Treg increased their ability to prevent the proliferation of Tconv cells. Finally, we showed that the increased efficacy of Themis1-Tg Treg in vitro is associated with an accumulation of Treg cells. We obtained similar results in vivo in the model of colitis, although the differences did not reach statistical significance (data not shown). As this increased number of Themis1-Tg Treg did not result from an increase of their proliferative capacity, we postulate that it is probably due to enhanced T cell survival, since recent studies showed that Themis1-SHP1 promotes T cell survival (52).
In light of previous studies suggesting that the suppressive abilities of Treg, but also their survival capacity, is correlated with the strength of TCR signaling (26)(27)(28)(29), our data obtained with the Themis1-Tg mice suggest that, in Treg, Themis1 may rather act as an enhancer than as an attenuator of TCR signaling. While SHP-1 was proved before to be an endogenous inhibitor of the suppressive ability of Treg (53) to the spontaneous development of IBD (54,55). This highlights that the cooperation between Themis1 and Vav1 is required for proper Treg function.
On the other hand, our proteomic data showed that the endogenous expression of Themis1 is maintained at low level in Treg compared to Tconv. This would be consistent with a recent report showing that TCR signals are naturally tuned down in Treg cells (30). In the later study,      CD4 + CD25 -CD45RB high colitogenic T cells were sorted from WT mice and injected into Rag2 -/mice to induce colitis. In addition, CD4 + CD25 bright Treg cells were sorted either from Themis1-Tg mice or littermates control and were co-transferred in a ratio of 1:4 (Treg:Tconv).