The mTORC1-4E-BP-eIF4E axis controls de novo Bcl6 protein synthesis in T cells and systemic autoimmunity

Post-transcriptional modifications can control protein abundance, but the extent to which these alterations contribute to the expression of T helper (TH) lineage-defining factors is unknown. Tight regulation of Bcl6 expression, an essential transcription factor for T follicular helper (TFH) cells, is critical as aberrant TFH cell expansion is associated with autoimmune diseases, such as systemic lupus erythematosus (SLE). Here we show that lack of the SLE risk variant Def6 results in deregulation of Bcl6 protein synthesis in T cells as a result of enhanced activation of the mTORC1–4E-BP–eIF4E axis, secondary to aberrant assembly of a raptor–p62–TRAF6 complex. Proteomic analysis reveals that this pathway selectively controls the abundance of a subset of proteins. Rapamycin or raptor deletion ameliorates the aberrant TFH cell expansion in mice lacking Def6. Thus deregulation of mTORC1-dependent pathways controlling protein synthesis can result in T-cell dysfunction, indicating a mechanism by which mTORC1 can promote autoimmunity.

P recise regulation of T follicular helper (T FH ) cell numbers is critical for optimal humoral responses, and aberrant expansion of T FH cells is associated with autoimmune diseases, including systemic lupus erythematosus (SLE) 1,2 . The transcriptional repressor Bcl6 is a lineage-defining factor for T FH cells [3][4][5] . Bcl6 is necessary to specify the T FH cell program and overexpression of Bcl6 is sufficient to drive T FH cell differentiation, indicating that tight control of Bcl6 expression is essential to ensure proper regulation of T FH cell numbers. Bcl6 expression in T FH cells has, until now, been shown to be primarily regulated by transcriptional mechanisms 6 . The expression of Bcl6, however, can be controlled by complex regulatory networks that fine-tune Bcl6 expression by targeting both Bcl6 mRNA and protein 7    transcriptional mechanisms, which control Bcl6 protein stability and its activity 7 . Among post-transcriptional mechanisms, translational control has a major function in regulating protein abundance and can influence protein levels to an extent similar to transcription 8 . A critical controller of protein synthesis is mammalian target of rapamycin (mTOR), a serine/threonine kinase that exists in two distinct complexes, mTORC1 and mTORC2, distinguished by the presence of unique components such as raptor and rictor, respectively 9,10 . mTORC1 activation occurs in response to diverse environmental cues, including growth factors, energy status, and amino-acid availability. Growth factors activate mTORC1 mainly through the phosphoinositide-3 kinase (PI3K)-AKT pathway, whereas the energy status of a cell regulates mTORC1 activation via AMP-activated protein kinase (AMPK) [9][10][11] . mTORC1 activation by PI3K-AKT and AMPK occurs via the TSC complex and the small GTPAse Rheb [9][10][11] . By contrast, amino acids regulate a different set of GTPases, the Rag proteins, which recruit mTORC1 to the lysosomes enabling subsequent activation by Rheb. Although activation of the Rags normally depends on their interaction with the Ragulator complex, an alternative docking system that depends on the central signaling hub p62 can also control activation [11][12][13] . p62 interacts with and activates the Rags, helps recruit mTORC1 to the lysosomes by binding Raptor and also mediates the assembly of a trimolecular complex with TRAF6, which can then activate mTOR kinase activity via K63-linked polyubiquitination 12,13 .
mTOR is a major coordinator of T H cell fate decisions and regulates the differentiation of several T H subsets 9,10 . mTOR plays a complex role in T FH differentiation. Whereas the interleukin (IL)-2-mTORC1 axis shifted differentiation away from T FH cells toward the T H 1 lineage in an acute viral infection model 14 , mTORC1 activation is required for the spontaneous formation of T FH cells in Peyer's patches and for the induction of T FH cells upon immunization with a foreign antigen 15,16 . mTORC2 activity is also important for T FH differentiation, particularly in Peyer's patches 16 . The varying requirements of T FH cells on mTOR activity are probably due to differences in the precise environmental cues to which T FH cells are exposed 16 . mTOR has been shown to regulate T H cell differentiation by controlling the transcription of master regulators and metabolic reprogramming. Although regulation of protein synthesis is also a major downstream function of mTORC1, its role in T H cells is less well understood. mTOR has been implicated in the pathogenesis of autoimmune disorders, like SLE 17 . The pathways resulting in mTOR deregulation and T H cell dysfunction in autoimmunity are, however, not fully understood. Def6 is a an SLE risk variant 18 , which together with its only homolog SWAP-70, comprises the SWEF family of molecules 19 . Unlike SWAP-70, which is expressed by B cells but not naive T H cells 20 , Def6 is highly expressed by naive T H cells. Notably, double knockout (DKO) of Def6 and SWAP70 in C57BL/6 mice results in development of lupus, predominantly in female mice as in human SLE 21 . Autoimmunity in DKO mice results from dual abnormalities in T and B cells, whereby the lack of Def6 alone is responsible for the T-cell abnormalities, and the absence of both Def6 and SWAP70 contributes to the deregulated B-cell responses 21 .
In this study, we demonstrate that the robust humoral autoimmune responses observed in Def6 tr/tr Swap70 −/− DKO mice are accompanied by cell-intrinsic expansion of the T FH cell compartment. Importantly, we show that DKO T cells have aberrant control of Bcl6 protein synthesis, which occurs in an mTORC1 and eukaryotic initiation factor 4E (eIF4E)-dependent manner. Enhanced mTORC1 activation in DKO T cells is a result of dysregulated interaction of raptor with both p62 and TRAF6, critical regulators of an amino-acid-sensing pathway of mTORC1 activation 12,13 . We also demonstrate that Def6 controls the assembly of the raptor-p62-TRAF6 complex and that this pathway selectively regulates the abundance of a specific subset of proteins. Consistent with these findings, rapamycin administration or T-cell deletion of raptor significantly decreased the accumulation of T FH cells in DKO mice. Thus abnormalities in the mechanisms by which mTORC1 regulates protein abundance can result in T-cell dysfunction and contribute to autoimmunity.

Spontaneous expansion of T FH cells in lupus-prone DKO mice.
Precise control of T FH cell differentiation is essential to prevent SLE 1, 2 . Mice lacking Def6 and Swap70 (here termed DKO mice) spontaneously develop a systemic autoimmune disorder characterized by increased numbers of effector T cells, germinal center (GC) B cells, plasma cells, and autoantibody production 21 . This robust humoral autoimmune response led us to speculate that many of the effector T cells that accumulate in these mice might be T FH cells. A detailed analysis indeed revealed that young DKO mice spontaneously display increased frequencies and numbers of T FH cells based on staining for either CXCR5 hi PD1 hi CD4 + T cells or CXCR5 hi Bcl6 hi CD4 + T cells (Figs. 1a, b). Similar results were also obtained when T FR cells, a subset of CXCR5 hi PD1 hi cells that expresses Foxp3 and specialize in the inhibition of the GC reaction 22,23  To establish whether DKO T FH cells were located within the spontaneous GCs that develop in these mice, we stained splenic sections from unimmunized or immunized wild-type (wt) mice . e, f Flow cytometric analysis of T FH cells in wt, Def6 tr/tr , Swap70 −/− , and DKO mice. e Representative FACS plots for CXCR5 and PD1 (upper panel) or CXCR5 and Bcl6 (lower panel) expression gated on CD4 + T cells in the spleens from mice of the indicated genotype (>18 weeks). f Quantification of T FH cells (CD4 + CXCR5 hi PD1 hi ) (n = 5-7). Combined data from three independent experiments. g Representative immunofluorescence images of splenic sections from wt unimmunized mice, wt mice 8 days post-immunization with NP-CGG and DKO unimmunized mice (two independent experiments, n = 2-3 each). PNA, green; PD1, red; CD3, magenta; Scale bar, 100 μm. h, i Quantification of CD4 + T cells producing IL-17, IL-21, and IFN-γ in the spleens of aging wt and DKO mice (n = 4-8) analyzed by intracellular flow cytometry. Each dot represents an individual mouse. Error bars indicate mean ± s.d., P-value by two-tailed ttest. Data were combined from at least two independent experiments or unimmunized DKO mice with PNA and PD1, markers previously utilized to identify GC B cells and T FH cells, respectively 24 . Few PD1 + cells were detected in unimmunized wt mice while PD1 + cells could easily be visualized within the PNA + GCs of immunized wt mice (Fig. 1g, top panel). Costaining of PD1 + cells with CD3 confirmed their T-cell identity (Fig. 1g, bottom panel). Similarly to immunized wt mice, young unimmunized DKO mice contained well-formed PNA + GCs with PD1 + T cells, which appeared to be diffusely distributed throughout the GCs (Fig. 1g and Supplementary Fig. 3a) 25 . A similar localization pattern was also observed in aging DKO female mice ( Supplementary Fig. 3b). T FH cells in DKO mice thus accumulate within GCs although some PD1 + T cells could also be observed at extrafollicular sites.
T FH cells produce various cytokines that have been implicated in autoimmune pathogenesis including interferon (IFN)-γ and IL-21 1, 2 . Excessive IFN-γ production has been associated with the accumulation of pathogenic T FH cells in sanroque mice 26 .
Evaluation of the cytokine profile of DKO T FH cells revealed an expansion of T FH cells that produce both IFN-γ and IL-21 ( Fig. 1h and Supplementary Figs. 3c, d). While DKO mice also exhibited an increased number of CD4 + T cells producing both IL-17 and IL-21, this effect was primarily restricted to non-T FH effector T cells ( Fig. 1i and Supplementary Fig. 3e). The T FH cells that accumulate in DKO mice are thus capable of producing cytokines that have been linked to the development of systemic autoimmunity.
To evaluate whether the spontaneous expansion of T FH cells in DKO mice is cell-intrinsic, we generated mixed bone marrow chimeras. Rag1 −/− mice reconstituted with only CD45.2 + DKO bone marrow cells were also generated as control. The recipient mice were then analyzed for the presence of CXCR5 hi PD1 hi T FH cells in the spleen and peripheral lymph nodes. CD45.2 + DKO T cells contained an increased proportion of T FH cells, which reached levels comparable to those observed in mice reconstituted with CD45.2 + DKO bone marrow cells alone (Figs. 2a,   Similar results were also obtained by staining for CXCR5 hi Bcl6 hi or PD1 hi ICOS hi T cells (Figs. 2b, c). A slightly greater propensity of DKO B cells to acquire a GC B-cell phenotype as compared to wt B cells could also be observed (Fig. 2f). Collectively, these data support the notion that the spontaneous expansion of DKO T FH cells is cell-autonomous.
Aberrant control of Bcl6 protein synthesis in DKO T cells. The spontaneous cell-intrinsic expansion of pathogenic T FH cells in the lupus-prone DKO mice suggested that DKO T cells inappropriately acquire features of T FH cells. In view of the critical role of Bcl6 in the differentiation of T FH cells 3-5 , we focused our attention on this transcription factor. To investigate whether Bcl6 expression might be dysregulated in DKO T cells, we sorted naive CD4 + T cells from wt and DKO mice, which expressed similar low levels of Bcl6 protein ( Supplementary Fig. 4a), and activated them in vitro. Remarkably, as compared to wt T cells, DKO T cells exhibited much higher Bcl6 protein levels ( Fig. 3a and Supplementary Fig. 4b).
In line with the finding that Def6 is the primary member of this small family of proteins expressed in T cells 27,28 , similar results were obtained in T cells lacking Def6 alone while T cells from SWAP-70-deficient mice were indistinguishable from wt T cells ( Supplementary Fig. 4c). No phosphorylation of Stat1, Stat3 or Stat4 was detected in DKO T cells, suggesting that the high levels of Bcl6 expression were not due to inappropriate STAT activation or aberrant production of IL-6 or IL-21 ( Supplementary Fig. 4d). Furthermore, the expression of interferon regulatory factor 4 (IRF4), which has been implicated in the differentiation of several T H cell lineages including T FH cells, was similar in wt and DKO T cells (Fig. 3a). Interestingly, we observed a similar increase in Bcl6 protein expression in DKO T cells cultured under T H 1 conditions although T-bet levels were upregulated to a similar extent in wt and DKO T cells upon T H 1 skewing ( Supplementary Fig. 4e). Surprisingly, Bcl6 mRNA was not upregulated in DKO T cells as compared to wt T cells as assessed by quantitative reverse transcription-PCR (RT-PCR) with two different sets of primers and at different time points ( Fig. 3b and Supplementary Figs. 4f, g). Thus the expression of a key controller of T FH cell differentiation, Bcl6, is selectively dysregulated in DKO T cells and this effect is not due to a transcriptional mechanism or to global alterations in the ability of DKO T cells to undergo T H differentiation. In B cells, post-transcriptional mechanisms play an important role in regulating Bcl6 protein levels 7 . In particular, Bcl6 contains three PEST domains and can be rapidly degraded. To evaluate whether the elevated levels of Bcl6 protein in DKO T cells could be due to alterations in the stability of the Bcl6 protein, wt or DKO T cells were treated with cycloheximide, a protein synthesis inhibitor, and the levels of Bcl6 protein monitored over time (Figs. 3c,d). In contrast to IRF4, which was very stable, Bcl6 protein was rapidly degraded following cycloheximide treatment. The rate of degradation of Bcl6, however, was similar between wt and DKO T cells (Figs. 3c, d), indicating that the accumulation of Bcl6 protein in DKO T cells is not secondary to changes in protein stability.   (Fig. 3e). Interestingly, the [ 35 S]methionine incorporation assays did not reveal a global dysregulation in protein synthesis in DKO T cells ( Supplementary   Fig. 4h). Thus the increased levels of Bcl6 protein expression in DKO T cells are due to aberrancies in the control mechanisms that regulate the de novo synthesis of Bcl6 protein.
mTORC1 controls Bcl6 protein levels in DKO T cells. mTORC1 is a crucial controller of protein synthesis due to its ability to modulate the activity of translational regulators, such as 4E-BP proteins (4E-BP1, 4E-BP2, and 4E-BP3) and S6K1 9, 10 . To assess whether mTORC1 controls the expression of Bcl6 protein in DKO T cells, we utilized rapamycin, a well-known mTORC1 inhibitor. While rapamycin treatment did not affect the low levels of Bcl6 protein in wt T cells, it markedly decreased the high levels of Bcl6 protein in DKO T cells ( Fig. 4a and Supplementary  Fig. 5a). The inhibitory effects of rapamycin were not due to effects on Bcl6 transcription ( Supplementary Fig. 5b) and did not represent global effects on protein translation since the expression of IRF4 was unaffected by rapamycin treatment (Fig. 4a). Thus the increased expression of Bcl6 protein in DKO T cells is dependent on mTORC1 activity. The finding that mTORC1 regulated the abundance of Bcl6 protein suggested that mTORC1 activation might be dysregulated in DKO T cells. An examination of the phosphorylation status of 4E-BP proteins, key downstream effectors of mTORC1, demonstrated that, as compared to wt T cells, DKO T cells exhibited higher levels of 4E-BP phosphorylation, which was sensitive to rapamycin inhibition (Fig. 4b, Supplementary Fig. 5c). Higher phosphorylation levels were particularly noticeable in the case of the faster migrating form of p-4E-BP proteins, which may represent phosphorylation of the 4E-BP2 isoform 29 . The phosphorylation levels of S6K1 were also increased in DKO T cells but to a lesser extent ( Supplementary Fig. 5d). To further evaluate the activation of mTORC1 in wt and DKO mice, we  Fig. 6 Proteomics-based identification of mTORC1 targets in DKO T cells. a Nuclear extracts from naive wt T and DKO T cells were cultured as described in Fig. 3 and treated with vehicle control or rapamycin (20 nM) for the final 24 h of the 3-day culture were subjected to LC-MS/MS. Heatmap of the 20 mTORC1-regulated proteins whose levels were upregulated in DKO T cells as compared to wt and were downregulated upon rapamycin treatment, and the 18 mTORC1-regulated proteins that were downregulated in DKO T cells and whose downregulation was rapamycin sensitive. Yellow bars indicate upregulated proteins and blue bars indicate downregulated proteins. Color codes for the Z-scored LFQ values are shown. Proteins were selected based on the criteria that one comparison needed to be statistically significant by >4-fold difference and that both comparisons needed to be at least >1.7-fold different. b, c Representative western blottings from at least two independent experiments of mTOR-regulated proteins, ASF1a and XPC. d, e Quantitative RT-PCR analysis of the expression of Asf1a and Xpc mRNA in sorted naive T cells cultured as in Fig. 3 and treated with control or rapamycin (20 nM) for the final 24 h of the 3-day culture. The data were normalized relative to β-Actin mRNA expression and are combined from three independent experiments. Error bars indicate mean±s.d., statistical significance was determined using Wilcoxon matched-pairs test. ns not significant SLE patients exhibit elevated mTORC1 activity including increased 4E-BP phosphorylation 30,31 , these data indicate that mTORC1 activity is dysregulated in DKO T cells and that high levels of phosphorylation of the mTORC1 target, 4E-BP, characterize the spontaneous accumulation of T FH cells in two different murine lupus models. 4E-BP proteins regulate cap-dependent translation by binding to eIF4E, a key rate-limiting factor in translation initiation, disrupting its interaction with eIF4G, and thereby preventing the formation of the eIF4F complex 32,33 . Although eIF4E is involved in the translation of all mRNAs, changes in its levels and/or activity can selectively alter the translation of a subset of "eIF4Esensitive" mRNAs without impacting global protein synthesis 32 . The function of 4E-BP can be mimicked by a small-molecule inhibitor, 4EGI-1, which binds to eIF4E and interferes with its interaction with eIF4G, preferentially inhibiting the translation of "eIF4E-sensitive" mRNAs 33 . To evaluate whether the increased Bcl6 protein expression in DKO T cells depends on eIF4E, T cells from wt and DKO mice were cultured in the presence or absence of 4EGI-1. Expression of the Bcl6 protein in DKO T cells decreased significantly upon exposure to 4EGI-1 in a dosedependent manner (Fig. 4f, Supplementary Fig. 5h). This effect was not due to a decrease in Bcl6 mRNA, which, in line with the known ability of Bcl6 to negatively regulate its own expression 7 , increased upon 4EGI-1 treatments (Supplementary Fig. 5i). The expression of IRF4 and HDAC1 was not affected by 4EGI-1 treatments, suggesting that global protein synthesis was not altered by these doses of the inhibitor (Fig. 4f). These data thus support the idea that increased Bcl6 protein levels in DKO T cells are secondary to a deregulated mTOR-4E-BP axis, leading to aberrant Bcl6 protein expression due to eIF4E hyperactivity.
Def6 inhibits assembly of a raptor-p62-TRAF6 complex. mTORC1 can be activated by growth factors, energy status, and availability of amino acids. To dissect the mechanisms leading to mTORC1 activation in DKO T cells, we first focused on the PI3K-AKT axis, the key pathway responsible for growth factormediated activation of mTORC1 in T cells 9,10 . DKO T cells exhibited lower levels of AKT phosphorylation on Ser473 than wt T cells (Fig. 5a). Phosphorylation of AKT on Thr308 was absent in both wt and DKO T cells although it was upregulated after rapamycin treatment, consistent with the known interference of rapamycin with mTORC1-dependent feedback inhibitory mechanisms (Fig. 5a). Consistent with AKT activation levels, the phosphorylation of the AKT target site in PRAS40 (Thr246) was lower in DKO T cells than in wt T cells (Fig. 5a). Thus the PI3K-AKT axis is not responsible for the deregulated mTORC1 activation in DKO T cells.
To determine the contribution of energy-dependent pathways, we assessed the activation state of AMPK, an energy sensor that monitors the intracellular AMP/ATP ratio. AMPK is activated by phosphorylation at Thr172 by liver kinase B1 (LKB1) and, in turn, can phosphorylate raptor at Ser792, inducing its dissociation from mTORC1 and leading to mTORC1 inactivation 10 . The phosphorylation levels of both AMPK Thr172 and raptor Ser792 were similar in T cells from wt and DKO mice, suggesting that the regulation of mTORC1 by energy-dependent pathways is not altered in DKO T cells (Fig. 5b).
Interaction of raptor with p62 and TRAF6 has been implicated in the regulation of nutrient-sensitive pathways that lead to the recruitment of mTORC1 to the lysosomes 12,13 . To examine this pathway, we first assessed whether assembly of the raptor-p62-TRAF6 complex was appropriately regulated in DKO T cells. Only a minimal interaction of raptor with either p62 or TRAF6 was observed in wt T cells (Fig. 5c). In contrast, raptor strongly interacted with both p62 and TRAF6 in DKO T cells (Fig. 5c). Thus formation of the raptor-p62-TRAF6 complex, a critical step in mTORC1 activation, is deregulated in T cells from DKO mice.
Since T cells lacking Def6 alone but not those lacking SWAP-70 alone exhibited deregulated Bcl6 protein expression (Supplementary Fig. 4c), we next explored the possibility that Def6 controls the assembly of this trimolecular complex. Def6 coimmunoprecipitated with p62 and TRAF6 in wt T cells, indicating that endogenous Def6 interacts with components of this complex under physiological settings (Figs. 5d, e). To further dissect the role of Def6 in the formation of the raptor-p62-TRAF6 complex, we cotransfected HA-tagged p62 with Myc-tagged raptor in 293T cells in the presence/absence of Def6 (Fig. 5f). As expected, in the absence of Def6, raptor coimmunoprecipitated with p62. Consistent with the results observed with the endogenous proteins, Def6 and p62 strongly interacted (Fig. 5f). Strikingly, the presence of Def6 completely abrogated the interaction between raptor and p62 (Fig. 5f). When HA-tagged p62 was cotransfected with Flag-tagged TRAF6, interaction between p62 and TRAF6 was also significantly diminished by the presence of Def6 (Fig. 5g). Cotransfection of HA-tagged Def6 with either Flag-tagged TRAF6 or Myc-tagged raptor moreover demonstrated that Def6 directly interacts with TRAF6 but not with raptor (Figs. 5h, i). A mutational analysis mapped the association of p62 with Def6 to amino acids 225-266, a region of p62 that contains the TRAF6-binding sequence ( Supplementary Figs. 6a, b). By directly interacting with p62 and TRAF6, Def6 can thus prevent the assembly of a raptor-p62-TRAF6 complex and inhibit the activation of mTORC1.
mTOR regulates a subset of proteins in DKO T cells. To identify additional proteins that might be aberrantly regulated in DKO T cells in an mTOR-dependent manner, we employed a global proteomic approach. Given that our metabolic labeling experiments had revealed that enhanced [ 35 S]methionine incorporation by DKO T cells could be primarily observed in nuclear extracts, we focused on this cellular compartment. Nuclear extracts from biological replicates were obtained from wt T cells and DKO T cells as well as DKO T cells treated with rapamycin. The different samples were subjected to in-solution trypsin digestion followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) based proteomics analysis. Using label free quantitation (LFQ) we quantitated 2989 proteins. Statistical analysis led to the identification of 20 proteins whose expression was upregulated in DKO T cells in an mTOR-dependent manner ( Fig. 6a and Supplementary Data 1). Eighteen proteins were expressed at lower levels in DKO T cells than in wt T cells and their downregulation was rapamycin-sensitive ( Fig. 6a and Supplementary Data 1). Performing the classification without imputation revealed an additional seven proteins that were upregulated in DKO T cells in an mTOR-dependent manner and an additional six proteins, whose abundance was lower in DKO T cells than in wt T cells and whose downregulation was rapamycin-sensitive (Supplementary Data 1).
Several of the identified proteins were previously shown to be translationally regulated mTOR targets, such as MYH14 and LMF2 (Fig. 6a and Supplementary Data 1) 34,35 . The expression of the Bcl6 interaction partner SMRT (also known as NCOR2) 36 was also increased in DKO T cells in an mTOR-dependent manner ( Fig. 6a and Supplementary Data 1). Furthermore, a number of proteins involved in the DNA damage response like ASF1a and XPC 37,38 were identified in this analysis (Fig. 6a and Supplementary Data 1). The expression pattern of some of the targets identified in the MS experiment, ASF1a and XPC, was further corroborated by western blotting (Figs. 6b, c). Quantitative real-time PCR for Asf1a and Xpc confirmed that the differences in expression levels observed in the different samples were not due to transcriptional effects (Figs. 6d, e). Aberrant activation of mTORC1 in DKO T cells thus targets the expression of a subset of proteins in addition to Bcl6. mTORC1 regulates T FH cell expansion in lupus-prone mice. To start assessing whether abnormalities in mTORC1 activation in  DKO T cells contributed to the aberrant expansion of T FH cells in DKO mice, we next evaluated whether inhibition of mTORC1 by rapamycin treatment could decrease the accumulation of T FH cells in these mice. Daily injection of rapamycin intraperitoneally for 10 days into aging DKO female mice resulted in a significant reduction in the frequency and numbers of T FH cells in DKO mice (Figs. 7a, b and Supplementary Fig. 7a). Consistent with this finding, phosphorylation of 4E-BP in DKO T FH cells was significantly reduced by rapamycin treatment (Fig. 7c). The changes in the T FH cell compartment were accompanied by a profound decrease in GC B cells although total B-cell numbers were unaffected ( Supplementary Figs. 7b-d). These studies thus support the idea that mTORC1 activation plays a critical role in the aberrant humoral responses of DKO mice. Given that rapamycin could affect the activation of mTORC1 in multiple cellular compartments, we then proceeded to directly evaluate the impact of T cell-mTORC1 in DKO mice by selectively deleting raptor in the DKO T-cell compartment. As an initial step, we generated CD4 cre raptor f/f mice, lacking raptor in CD4 + T cells 39 and immunized them with a T-dependent antigen (NP-KLH). T FH cells were decreased in CD4 cre raptor f/f mice ( Supplementary Figs. 7e, f). We then crossed CD4 cre raptor f/f mice to DKO mice to generate CD4 cre raptor f/f DKO mice. In line with the notion that Bcl6 is necessary to drive T FH cell differentiation and that mTORC1 regulates Bcl6 in DKO T cells, CD4 cre raptor f/f DKO mice exhibited fewer CXCR5 hi PD1 hi CD4 + T cells and CXCR5 hi Bcl6 hi CD4 + T cells, indicating that expansion of this compartment in DKO mice is mTORC1dependent (Figs. 7d, e, Supplementary Figs. 7g, h). T cells producing both IL-21 and IFN-γ were also significantly decreased in CD4 cre raptor f/f DKO mice (Supplementary Fig. 7i). Interestingly, an analysis of the B-cell compartment in CD4 cre raptor f/ f DKO mice revealed that plasma cells in these mice were reduced to a greater extent than GC B cells (Figs. 7f-i and Supplementary  Figs 7j, k). Notably, autoantibody production was significantly diminished in CD4 cre raptor f/f DKO mice (Figs. 7j, k), confirming a crucial role for T-cell mTORC1 in the accumulation of T FH cells and the systemic autoimmune responses that characterize the DKO mice.

Discussion
The acquisition of a T FH cell phenotype is critically dependent on Bcl6 3-5 and aberrancies in T FH cell homeostasis have been linked to autoimmune disorders, including SLE 1, 2 . While previous studies have elegantly outlined the transcriptional control of Bcl6 expression 6 , here we demonstrate that the total levels of Bcl6 in T cells can also be regulated by controlling the de novo synthesis of the Bcl6 protein in an mTORC1-dependent manner. We furthermore show that mTORC1 activation in this autoimmune setting proceeds via an alternative pathway, which involves the assembly of a raptor-p62-TRAF6 complex, and controls the abundance of a selected group of proteins. Thus, in addition to effects on the transcriptional landscape and the metabolic programs of T cells, the ability of mTORC1 to regulate the expression of lineage-defining transcription factors by controlling protein synthesis can also be an important contributor to T-cell dysfunction and the development of autoimmunity.
The ability of 4EGI-1, an inhibitor of the eIF4E-eIF4G interaction, to downregulate the expression of the Bcl6 protein in DKO T cells supports the idea that this Ragulator-independent docking system may be particularly important for the regulation of the mTORC1-4E-BP-eIF4E axis. Notably, eIF4E has been shown to enhance the translation of a specific group of transcripts during cellular stress and knockdown of eIF4E selectively affects these transcripts without global decreases in protein levels 40 . Transcripts that are particularly dependent on eIF4E levels are believed to contain highly structured 5′ untranslated regions that require unwinding by eIF4A and/or encode proteins, which, like Bcl6, have a short half-life. In line with our data, eIF4E was recently shown to control the abundance of the Bcl6 protein in diffuse large B-cell lymphomas 41 . Thus employment of this alternative pathway to regulate mTORC1 activity could permit T FH cells to maintain optimal translation of selected key targets, like Bcl6, Ncor2, and DNA damage response proteins, in the GC light zones where hypoxic conditions have recently been shown to occur 42 . In line with this notion, increased levels of 4E-BP phosphorylation could be detected in T FH cells generated upon immunization of wt mice with a T-dependent antigen. Inappropriate control of this mTORC1-4E-BP-eIF4E axis, in turn, may enable autoimmune T FH cells to bypass an important checkpoint and endow them with enhanced competitive fitness.
Strong T-cell receptor (TCR) engagement is a key controller of the acquisition of a T FH cell phenotype 43 and the requirement for distinct APC-T-cell interactions can be altered by the presence of an abundant source of antigen 44 . Def6 contains multiple phosphorylation sites, which can be phosphorylated by Lck and ITK, and undergoes profound conformational changes upon TCR engagement 27,28 . In particular, ITK-mediated phosphorylation of Def6 promotes its aggregation and the formation of Def6 cytoplasmic granules that co-localize to P-bodies 45 . The involvement of Def6 in controlling assembly of the raptor-p62-TRAF6 complex could thus provide a direct mechanistic link by which TCR engagement could regulate the mTORC1-4E-BP-eIF4E axis. By post-translationally modifying Def6, T cells could selectively change the inhibitory effects of Def6 on the p62-TRAF6 interaction. Indeed, Def6 and p62 exhibited a similar localization in T FH cells from immunized wt mice (Supplementary Fig. 8a). Intriguingly the expression of Slc7a5 and other amino-acid transporters is also controlled by TCR engagement and the lack of Scl7a5 results in defective high-affinity T-dependent humoral responses 46 . Thus TCR engagement may control this axis via its dual ability to regulate Def6 activity and amino-acid availability.
The capacity of Def6 to control protein synthesis in addition to cytoskeletal reorganization and gene expression 27,28 supports the idea that employment of multifunctional signaling hubs like Def6 may enable T cells to precisely coordinate these critical processes in a rapid manner when faced with quickly changing environmental conditions such as those encountered in the GCs. Importantly, the involvement of Def6 in the integration of these diverse processes may be a crucial feature underlying the ability of Def6 variants to be associated with the pathogenesis of human SLE. In this regard, the availability of mouse models where the lack of Def6 leads to the development of a disease that recapitulates key features of human SLE, like the aberrant expansion of T FH cells, can provide an invaluable platform for an in-depth understanding of the mechanisms by which human SLE variants could impact disease pathogenesis.
Recent studies have uncovered a surprising level of complexity in the reliance of T FH cells on the mTOR pathway with differential requirements for the mTORC1 complex depending on the stimulus triggering their formation and their localization [14][15][16] . We now show that, despite the well-known ability of autoimmune T cells to deregulate multiple signaling pathways 47 , a feature that can potentially lessen their dependency on any one pathway, the inappropriate accumulation of T FH cells in the chronic inflammatory environment characteristic of autoimmune settings requires T cell-mTORC1. It is likely that this dependency results from both a direct role of mTORC1 in maintaining high levels of Bcl6 protein expression and from the known effects of mTORC1 on T-cell metabolism [48][49][50] . Importantly, the T FH cells that expand in DKO mice remain amenable to therapeutic interventions aimed at inhibiting mTORC1 since administration of rapamycin to aging DKO mice results in a decreased T FH compartment and a marked and rapid amelioration of their exuberant humoral responses. Given that rapamycin is being investigated as a treatment for SLE 17 , these findings could have important therapeutic implications and help define the patient subsets that might be most responsive to this treatment.
Our data thus indicate that aberrancies in mechanisms controlling de novo protein synthesis represent a novel mechanism leading to T-cell dysfunction in autoimmunity. The ability of post-transcriptional mechanisms to selectively modify the expression of key regulators of T H cell fate decisions furthermore suggest that, in order to fully understand the functional phenotype of autoimmune T cells, transcriptomic and epigenomic evaluations will need to be complemented by proteomic analyses. Together with previous work demonstrating that abnormalities in mRNA stability can also lead to deregulation of the T FH cell compartment 2, 51 , post-transcriptional mechanisms may thus emerge as major contributors to ensure T FH cell homeostasis.
Western blotting and immunoprecipitation. Nuclear and cytoplasmic extracts from naive T cells cultured for 3 days were prepared with NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce), as previously described 21 . The purity of the nuclear and cytoplasmic fractions was verified by probing with antibodies against Lamin B1 (D4Q4Z; 1,000×; Cell Signaling Technology). Whole-cell extracts were immunoprecipitated with an anti-Raptor (24C12; 100×; Cell Signaling