Main

During infection, antibody responses are one of the important host-defense mechanisms used to clear invading pathogens. Antibodies produced by B cells mediate the destruction of extracellular microorganisms and prevent the spread of intracellular infectious agents. The activation of naive B cells is triggered by antigen and usually requires follicular helper T cells (TFH cells) for sustained proliferation and differentiation1,2. Germinal centers (GCs) are transient structures in the B cell follicles of secondary lymphoid tissues in which B cells undergo somatic hypermutation, affinity maturation and differentiation into memory B cells and long-lived plasma cells3. The formation of GCs is controlled by TFH cells, which are required for the initial help provided to B cells as well as for the maintenance of GCs by positive selection of B cells expressing B cell antigen receptors (BCRs) of the highest affinity. TFH cells also regulate the generation of plasma cells and memory B cells. TFH cells localize in the follicles and GCs because their expression of a set of genes encoding migration-related molecules, most notably the chemokine receptor CXCR5, is different from that of other helper T cells.

TFH cells express interleukin 21 (IL-21), IL-4 and the chemokine CXCL13 and are characterized by high expression of surface markers required for cognate T cell–B cell interactions, including CXCR5, the inducible costimulator ICOS, the T cell inhibitory receptor PD-1, the ligand for the costimulatory receptor CD40 and members of the SLAM family of receptors1. Three independent groups have identified the transcription repressor Bcl-6, a member of the BTB-POZ ('bric-a-bric, tramtrack, broad complex–poxvirus zinc finger') family of zinc-finger–containing transcription factors, as a master transcription factor for TFH cells4,5,6. These results clearly established TFH cells as a subset distinct from other TH cell subsets. However, the regulation of the differentiation of TFH cells is only partially characterized. Accumulating evidences suggests that several cytokines, including IL-6 and IL-21, contribute to the differentiation of TFH cells through mechanisms dependent on the transcription factors STAT3 and STAT1 (refs. 7,8,9,10,11,12). Signaling via ICOS and its ligand is required for Bcl-6 expression and the differentiation of TFH cells13,14 and is important for the migration of TFH cells15. In contrast, signaling via IL-2 potently inhibits the differentiation of TFH cells by mechanisms dependent on the transcription factors STAT5 and Blimp-1 (refs. 16,17,18). Blimp-1 is a powerful repressor of Bcl-6 expression4. However, details of the mechanisms by which Bcl-6 expression is induced and Bcl-6 is regulated are not well understood19, nor are the interactions between Bcl-6 and other key transcription factors involved in the differentiation of TFH cells, including c-Maf, BATF, IRF4 and others13,20,21,22.

The ubiquitination of proteins is a post-translational modification in which a protein substrate is 'tagged' with the 76–amino acid small polypeptide ubiquitin as mono- or poly-ubiquitin chains, an event catalyzed by a cascade of enzymes, including E1, E2 and E3 (ref. 23). The modified proteins can be subjected to proteasomal degradation or endocytosis, or the ubiquitin modification can instead alter protein function, analogous to phosphorylation or acetylation. Ubiquitin ligases are critical regulators of many biological processes23. In T cells, ubiquitin ligases control signaling via the T cell antigen receptor24, anergy25, differentiation into T helper type 2 (TH2) cells26, differentiation into regulatory T cells27 and other processes. On the basis of the structural features of their E2-binding domain, most E3 ligases can be classified into two families: RING ('really interesting new gene')-type E3 ligases, and HECT ('homologous to the E6-associated protein carboxyl terminus')-type E3 ligases28. Itch belongs to the HECT family of E3 ligases. The locus encoding Itch is disrupted by an inversion in mice of the itchy strain, which develop severe immunological and inflammatory disorders and constant itching of the skin29. Itch targets the transcription factors JunB and c-Jun for degradation and inhibits the production of TH2 cell cytokines26. In this study, we present evidence of a critical role for Itch in the differentiation of TFH cells. Itch-deficient mice underwent a decrease in the abundance of GC B cells and antigen-specific antibody production after viral infection, due to a cell-intrinsic defect in TFH cell differentiation. This defect in TFH cell differentiation was not associated with TH2 cells and was independent of signaling via IL-2. Enforced expression of Bcl-6 restored the TFH differentiation of Itch−/− T cells, which suggested that Itch functioned upstream of Bcl-6. Unexpectedly, the defective TFH cell differentiation was rectified by ablation of the gene encoding the transcription factor Foxo1. We further demonstrated that Itch associated with Foxo1 and promoted its ubiquitination and subsequent degradation. Our data suggest that Itch is essential for inducing the differentiation of TFH cells and humoral immunity by targeting Foxo1 for degradation.

Results

TFH cell differentiation requires Itch

To investigate the role of E3 ubiquitin ligases in TFH cell differentiation and humoral immune responses, we did a small-scale screen of mice deficient in genes encoding E3 ligases known to be expressed in CD4+ T cells, including Itch, Cblb, Cbl and Wwp2, in a model of infection with vaccinia virus (VACV)14. At day 8 after infection, we analyzed T cell and B cell responses by flow cytometry and measured virus-specific antibody responses by enzyme-linked immunosorbent assay (ELISA). Mice of the itchy strain (called 'Itch−/− mice' here) exhibited a significantly lower frequency and absolute number of both TFH cells (CXCR5+SLAMlo) and GC TFH cells (CXCR5+PD-1hi or CXCR5+Bcl-6hi) than that of wild-type mice (Fig. 1a,b). We observed the considerably defective TFH cell phenotype in Itch−/− mice but not in mice with other ubiquitin-ligase deficiencies (Supplementary Fig. 1a–c), which indicated a selective role for Itch in the differentiation of TFH cells. To further explore such differentiation in Itch−/− mice, we isolated activated CD4+ T cells from VACV-infected wild-type and Itch−/− mice and analyzed their TFH cell–related gene-expression profiles by real-time quantitative PCR. The expression of TFH cell–related genes, including Cxcr5, Icos, Bcl6 and IL21, was lower in activated Itch−/− CD4+ T cells than in their wild-type counterparts (Supplementary Fig. 1d). Il4 expression was higher in activated Itch−/− CD4+ T cells than in their wild-type counterparts (Supplementary Fig. 1d), consistent with a published report26.

Figure 1: Itch deficiency results in T cell–intrinsic defective differentiation of TFH cells.
figure 1

(a) Flow cytometry of activated (CD44hi) CD4+ T cells from wild-type (WT) and Itch−/− mice 8 d after infection with VACV. Numbers adjacent to outlined areas indicate percent CXCR5+SLAMlo polyclonal TFH cells (top row) or CXCR5+PD-1hi GC TFH cells (middle row) or CXCR5+Bcl-6hi GC TFH cells (bottom row). (b) Frequency (among activated (CD44hi) CD4+ T cells) and total number of TFH cells and GC TFH cells in spleens of mice as in a (n = 7 per group). (c) Flow cytometry of total B220+ B cells from mice as in a. Numbers adjacent to outlined areas indicate percent GL7+Fas+ GC B cells (top row) or CD138+IgDlo plasma cells (bottom row). (d) Frequency (among total B220+ B cells) and total number of GC B cells and plasma cells in the spleen of mice as in a (n = 7 per group). (e) ELISA of VACV-specific IgG in serum from infected mice as in a or uninfected B6 mice (Naive) (n = 4 per group), presented as absorbance at 450 nm (A450). (f) Flow cytometry assessing polyclonal TFH cells and GC TFH cells from wild-type and Itch-cKO mice 8 d after infection with VACV (numbers in plots as in a). (g) Frequency (among activated (CD44hi) CD4+ T cells) and total number of TFH cells and GC TFH cells in the spleen of mice as in f (n = 6 per group). (h) Flow cytometry assessing GC B cells and plasma cells from mice as in f (numbers in plots as in c). (i) Frequency (among total B220+ B cells) and total number of GC B cells and plasma cells in the spleen of mice as in f (n = 6 per group). (j) ELISA of VACV-specific IgG in serum from mice as in f (n = 3 per group). Each symbol (b,d,g,i) represents an individual mouse; small horizontal lines indicate the mean (±s.d.). *P < 0.01 and **P < 0.001 (Student's t-test). Data are representative of at least three independent experiments (error bars, s.d.).

Because TFH cells are the main cognate helpers of antiviral B cell responses4, we next examined GC formation in Itch−/− mice after infection with VACV. As expected, we observed a robust abundance of GC B cells in wild-type C57BL/6J (B6) mice after infection with VACV, but the frequency and absolute number of GC B cells were much lower in their Itch−/− counterparts after such infection (Fig. 1c,d and Supplementary Fig. 1e). We then assessed differentiation into plasma cells in wild-type and Itch−/− mice after infection with VACV. Consistent with the lower abundance of GC B cells, the number of plasma cells was also much lower in Itch−/− mice (Fig. 1c,d). To assess the consequences of the defective TFH cell and GC B cell responses of Itch−/− mice, we measured VACV-specific serum concentrations of immunoglobulin G (IgG). VACV-specific IgG concentrations were 57-fold lower the serum of Itch−/− mice than in that of wild-type mice (Fig. 1e and Supplementary Fig. 1f). Collectively, these data suggested that Itch was required for the generation of TFH cells, GC B cells and high-affinity antibodies.

We next investigated whether Itch deficiency affected the production of inflammatory cytokines and type I interferons during acute viral infection, which could possibly result in enhanced clearance of the infecting virus and indirectly diminish the abundance of TFH cells. We collected serum from wild-type and Itch−/− mice at days 1, 2 and 8 after infection with VACV and measured the concentration of inflammatory cytokines (for example, IL-6) and type I interferons in the sera. We observed no substantial differences between wild-type mice and Itch−/− mice (Supplementary Fig. 1g and data not shown). Furthermore, there was no significant difference in viral load in spleens from wild-type and Itch−/− mice 4 d after viral infection (P = 0.87; Supplementary Fig. 1h). These data suggested that any differences between Itch−/− mice and wild-type mice in their innate immune responses did not have much influence on the defective differentiation of TFH cells.

Intrinsic regulation of TFH cell differentiation by Itch

Next we investigated whether Itch regulates TFH cell differentiation in T cell–dependent manner. To address this issue, we crossed mice with a loxP-flanked Itch exon (Itchfl/fl)30 with mice carrying a transgene encoding Cre recombinase under the control of the Cd4 enhancer-promoter (CD4-Cre) to generate mice with T cell–specific (conditional) Itch deficiency (Itch-cKO). We then analyzed the T cell and B cell responses of Itch-cKO mice after infection with VACV. Similar to Itch−/− mice, Itch-cKO mice had a much lower frequency and absolute number of both TFH cells (CXCR5+SLAMlo) and GC TFH cells (CXCR5+PD-1hi or CXCR5+Bcl-6hi) than did their wild-type counterparts (Fig. 1f,g). Moreover, the frequency and number of GC B cells were much lower in Itch-cKO mice than in wild-type mice (Fig. 1h,i and Supplementary Fig. 1i). Furthermore, the generation of plasma cells was also significantly lower in Itch-cKO mice than in wild-type mice (Fig. 1h,i). Finally, VACV-specific IgG concentrations were 41-fold lower in the serum of Itch-cKO mice than in that of wild-type mice (Fig. 1j and Supplementary Fig. 1j). The similar phenotypes of Itch−/− mice and Itch-cKO mice suggested that Itch regulated TFH cell differentiation and humoral immunity in a T cell–intrinsic manner.

Itch in various stages of TFH cell differentiation

TFH cell differentiation is a multistage, multifactorial process1. To investigate at what phases Itch regulates such differentiation, we crossed Itch−/− mice with SMARTA mice (which have transgenic expression of a T cell antigen receptor specific for the epitope of lymphocytic choriomeningitis virus (LCMV) glycoprotein amino acids 66–77, presented by the major histocompatibility complex (MHC) class II molecule I-Ab) to generate Itch−/− SMARTA mice31. We isolated naive CD4+ T cells from Itch+/+ SMARTA mice (called 'wild-type SMARTA mice' here) and Itch−/− SMARTA mice and then adoptively transferred them separately into B6 recipient mice, followed by subsequent acute infection of the host mice with LCMV. Although Itch−/− SMARTA CD4+ T cells showed normal proliferation (Supplementary Fig. 2a), they almost completely failed to differentiate into TFH cells by day 3 after infection, as measured by gating of CXCR5+SLAMlo, CXCR5+PD-1hi or CXCR5+Bcl-6hi cells (Fig. 2a,b). Moreover, expression of both CXCR5 protein (P = 0.0012) and Bcl-6 protein (P = 0.0046) was significantly lower in Itch−/− SMARTA CD4+ T cells than in wild-type SMARTA CD4+ T cells (Supplementary Fig. 2b). Consistent with that, Itch−/− SMARTA CD4+ T cells had lower levels of Cxcr5, Bcl6 and Il21 mRNA and higher levels of Prdm1 mRNA (which encodes Blimp-1) at day 3 after infection (Fig. 2c). These results suggested that the defective TFH differentiation of Itch−/− cells was associated with a failure to express Bcl-6 at early time points.

Figure 2: Itch is intrinsically required for TFH cell differentiation.
figure 2

(a) Flow cytometry of donor (CD45.1+) CD4+ T cells obtained from B6 host mice given transfer of naive CD4+ T cells from wild-type and Itch−/− SMARTA donor mice, followed by infection of host mice with LCMV and analysis 3 d after infection. Numbers adjacent to outlined areas indicate percent early-stage CXCR5+SLAMlo TFH cells (top row) or CXCR5+PD-1hi TFH cells (middle row) or CXCR5+Bcl-6hi TFH cells (bottom row). (b) Frequency (among SMARTA (CD45.1+) CD4+ T cells) and total number of TFH cells in the spleen of host mice as in a (n = 6–7 per group). (c) Real-time PCR analysis of mRNA from TFH cell–related genes in mice as in a (pool of 15 mice per group); results were normalized to those of Actb mRNA (encoding β-actin) and are presented relative to those of naive wild-type SMARTA CD4+ T cells. (d) Flow cytometry of donor (CD45.1+) CD4+ T cells from B6 mice given transfer of naive wild-type and Itch−/− SMARTA transgenic CD4+ T cells, followed by infection of host mice with LCMV and analysis 8 d after infection (numbers in plots as in Fig. 1a). (e) Frequency (among SMARTA (CD45.1+) CD4+ T cells) and total number of TFH cells and GC TFH in the spleen of mice as in e (n = 7 per group). Each symbol (b,e) represents an individual mouse; small horizontal lines indicate the mean (±s.d.). *P < 0.05, **P < 0.01 and ***P < 0.001 (Student's t-test). Data are representative of three independent experiments (error bars, s.d.).

As ICOS is required for Bcl-6 expression at day 3 after infection in vivo14, we quantified Icos expression in wild-type and Itch−/− SMARTA CD4+ T cells. Although Itch−/− SMARTA CD4+ T cells failed to differentiate into TFH cells, their expression of Icos mRNA and ICOS protein was similar to that of wild-type SMARTA CD4+ T cells (Fig. 2c and Supplementary Fig. 2c). Moreover, the expression of genes encoding some transcription factors upstream of Bcl-6, such as Batf and Irf4, was also intact in Itch−/− SMARTA CD4+ T cells (Fig. 2c). These results indicated that the defective differentiation of Itch−/− TFH cells was not due to changes in the expression of ICOS or known transcription factors upstream of Bcl-6.

We next investigated whether the early defect in Itch−/− TFH cells continued through the peak of the LCMV-specific response (day 8) and whether it affected GC TFH cell development. Itch−/− SMARTA CD4+ T cells showed normal proliferation at day 8 after infection with LCMV (Supplementary Fig. 2d). Notably, we observed an almost complete loss of TFH cells among Itch−/− SMARTA CD4+ T cells (Fig. 2d,e). There were almost no GC TFH cells (CXCR5+PD-1hi, CXCR5+Bcl-6hi or CXCR5+GL7hi) among Itch−/− SMARTA CD4+ T cells (Fig. 2d,e and Supplementary Fig. 2e). In addition, Itch−/− SMARTA CD4+ T cells were unable to upregulate expression of CXCR5 or Bcl-6 at day 8 after infection with LCMV (Supplementary Fig. 2f). Collectively, these results demonstrated that Itch was required for the differentiation of TFH cells both at early stages and late stages and that Itch regulated TFH cell differentiation in a cell-intrinsic manner.

Defective differentiation of TFH cells unrelated to TH2 bias

Itch associates with JunB and c-Jun and promotes their ubiquitination and subsequent degradation. The increased amount of JunB protein in Itch−/− T cells drives TH2-biased differentiation and leads to elevated production of canonical TH2 cell cytokines, particularly IL-4 (ref. 26). IL-4 is also one of the cytokines produced by TFH cells, especially by GC TFH cells32,33, and has long been recognized as a factor involved in the survival and differentiation of B cells. However, IL-4 and its signaling pathways are not required for TFH cell differentiation11,34,35. Il4−/− mice exhibited a frequency and number of total TFH cells and GC TFH cells similar to that of wild-type mice in response to acute infection with VACV (Fig. 3a,b). Nevertheless, we next investigated further to confirm that the defect in TFH cell differentiation in Itch−/− mice was not due to TH2-biased differentiation or chronic inflammation caused by overproduction of IL-4. To address this, we crossed Itch−/− mice or Itch-cKO mice with Il4−/− mice to generate mice deficient in both Itch and IL-4 (Itch–IL-4–DKO mice). We infected the progeny with VACV and analyzed their T cell and B cell responses. We first assessed the differentiation of TFH cells in Itch–IL-4–DKO mice. The differentiation of TFH cells and GC TFH cells was considerably impaired in Itch–IL-4–DKO mice compared with that in wild-type mice (Fig. 3c,d). The differentiation of TFH and GC TFH cells in Itch–IL-4–DKO was not greater than that in Itch−/− mice (Fig. 3c,d). Consistent with the reduced number of GC B cells and plasma cells in Itch−/− mice, the number of GC B cells and plasma cells in Itch–IL-4–DKO mice was also much lower than that of wild-type mice (Fig. 3e,f and Supplementary Fig. 3a). Furthermore, the GC B cell and plasma cell responses of Itch–IL-4–DKO mice were not enhanced relative to those of Itch−/− mice (Fig. 3e,f and Supplementary Fig. 3a), which suggested that the defect in the development of GC B cells and plasma cells in Itch−/− mice could not be 'rescued' by deletion of IL-4. Collectively, these data suggested that IL-4 did not affect TFH cell differentiation and that the defect in TFH cell differentiation in Itch−/− mice was independent of TH2 cells.

Figure 3: IL-4 is dispensable for the differentiation of TFH cells in Itch−/− mice.
figure 3

(a) Flow cytometry of activated (CD44hi) CD4+ T cells from wild-type and Il4−/− mice 8 d after infection with VACV (numbers in plots as in Fig. 1a). (b) Frequency (among activated (CD44hi) CD4+ T cells) and total number of TFH cells and GC TFH cells in the spleen of mice as in a (n = 7 per group). (c) Flow cytometry of activated (CD44hi) CD4+ T cells from wild-type, Itch−/− and Itch–IL-4–DKO mice 8 d after infection with VACV (numbers in plots as in Fig. 1a). (d) Frequency (among activated (CD44hi) CD4+ T cells) and total number of TFH cells and GC TFH cells in the spleen of mice as in c (n = 6–7 per group). (e) Flow cytometry of total B220+ B cells from mice as in c (numbers in plots as in Fig. 1c). (f) Frequency (among total B220+ B cells) and total number of GC B cells and plasma cells from mice as in c (n = 6–7 per group). Each symbol (b,d,f) represents an individual mouse; small horizontal lines indicate the mean (±s.d.). NS, not significant; *P < 0.01 and **P < 0.001 (Student's t-test). Data are representative of three independent experiments.

As an additional investigation of TH2 signaling, we next assessed whether the defective differentiation of TFH cells of Itch−/− mice could be rectified by JunB deficiency. For this, we knocked down JunB in Itch−/− and Itch+/+ (wild-type) SMARTA bone marrow through the use of short hairpin RNA (shRNA) and generated chimeras reconstituted with that bone marrow. We isolated naive SMARTA CD4+ T cells from the reconstituted mice and then adoptively transferred these cells into B6 recipient mice, followed by infection of the hosts with LCMV. Deficiency of JunB did not restore the TFH-differentiation defect of Itch−/− T cells (Supplementary Fig. 3b,c). These data further confirmed that the defect in TFH cell differentiation in Itch−/− mice was independent of TH2 cells.

IL-2 signaling is not responsible for the TFH cell defect

It has been reported that IL-2 inhibits differentiation into TFH cells16,17 and is in fact dose-limiting for differentiation into TH1 cells versus TFH cells in response to an acute infection with LCMV, as twofold lower expression of the receptor for IL-2 (IL-2R) is sufficient to double the frequency of TFH cells19. At day 3 after infection of wild-type and Itch−/− SMARTA mice with LCMV, expression of the α-chain of IL-2R (IL-2Rα) in SMARTA Itch−/− non-TFH (TH1) CD4+ T cells was similar to that in their wild-type counterparts (Supplementary Fig. 3e). However, additional results suggested that Itch might target IL-2Rγ for ubiquitination (data not shown). We explored the possibility that the TFH-differentiation defect of Itch−/− CD4 T cells was due to enhanced IL-2 signaling. We transferred wild-type SMARTA CD4+ T cells or Itch−/− SMARTA CD4+ T cells into B6 mice that we subsequently infected with LCMV. We treated the recipient mice with neutralizing antibody to IL-2 (anti-IL-2) or isotype-matched control antibody 1 d before and 1 d after infection. Consistent with a published report16, neutralization of IL-2 significantly enhanced the commitment of wild-type SMARTA cells to TFH differentiation at day 3 after infection, from 17% to 36% (Fig. 4). However, neither the frequency of TFH cells (CXCR5+SLAMlo cells or CXCR5+Bcl-6hi) nor the expression of CXCR5 and Bcl-6 protein was restored in Itch−/− CD4 T cells by neutralization of IL-2 (Fig. 4). We also knocked down IL-2Rγ in primary SMARTA CD4+ T cells through the use of shRNA and examined TFH cell differentiation in vivo. Knockdown of IL-2Rγ increased the frequency of wild-type SMARTA TFH cell but did not rectify the TFH-differentiation defect of Itch−/− SMARTA CD4+ T cells (Supplementary Fig. 3d,e). Collectively, our data indicated that IL-2 signaling was probably not the key factor that led to the impaired TFH differentiation of Itch−/− cells.

Figure 4: Blockade of IL-2 signaling does not restore the TFH differentiation of Itch−/− cells.
figure 4

(a) Flow cytometry of donor (CD45.1+) CD4+ T cells obtained from B6 host mice given adoptive transfer of naive wild-type or Itch−/− SMARTA CD4+ T cells, followed by infection of the recipient mice with LCMV 1 d later, along with treatment of the recipient mice with neutralizing antibody to IL-2 (α-IL-2) or isotype-matched control antibody (Isotype) 1 d before and 1 d after infection, and analysis of splenocytes 3 d after infection. Numbers adjacent to outlined areas indicate percent SLAMloCXCR5+ cells (top) or Bcl-6hiCXCR5+ cells (bottom), identified as early-stage TFH cells. (b) Quantification of SLAMloCXCR5+ TFH cells among total SMARTA CD4+ T cells (left), and expression of CXCR5 (middle) and Bcl-6 (right), presented as mean fluorescence intensity (MFI), for cells from host mice in a. Each symbol represents an individual mouse; small horizontal lines indicate the mean (±s.d.). *P < 0.05 and **P < 0.001 (Student's t-test). Data are representative of two independent experiments two mice per group.

'Rescue' by enforced expression of Bcl-6

As Bcl-6 has been identified as the critical transcription factor in the differentiation of TFH cells and its expression was substantially reduced in Itch−/− CD4+ T cells, we explored whether Bcl-6 is a potential target of Itch. We found that Itch associated with Bcl-6 both in vivo by coimmunoprecipitation and in vitro by precipitation assay, and we further identified a Pro-Pro-X-Tyr motif (where 'X' is any amino acid) at positions 182–185 in Bcl-6 that was responsible for the interaction (Supplementary Fig. 4a,b). In addition, Itch promoted both monoubiquitination and polyubiquitination of Bcl-6 (Supplementary Fig. 4c). To investigate the physiological function of the modification of Bcl-6 by Itch, we transduced wild-type SMARTA CD4+ T cells with a retroviral vector expressing green fluorescent protein (GFP) alone (empty vector) or GFP and either wild-type Bcl-6 or mutant Bl-6 with replacement of phenylalanine with tyrosine, then sorted the transduced cells and transferred them into B6 recipient mice, followed by infection of the host mice with LCMV. Expression of the mutant Bcl-6 induced differentiation into TFH cells and GC TFH cells similar to that induced by wild-type Bcl-6 (Supplementary Fig. 4d,e). These results suggested that modification of Bcl-6 by Itch might not have an apparent physiological function in TFH cell differentiation.

We then investigated whether enforced expression of Bcl-6 was able to rectify the defective TFH differentiation of Itch−/− CD4+ T cells. We transduced wild-type or Itch−/− SMARTA CD4+ T cells with retroviral vector expressing Bcl-6 or empty vector (as described above). Bcl-6 expression drove more robust TFH differentiation of wild-type SMARTA CD4+ T cells in vivo (80%) than did expression of GFP only by the empty vector (37%)4 (Fig. 5). Notably, Bcl-6 expression also substantially enhanced the TFH differentiation of Itch−/− SMARTA CD4+ T cells (70% versus 6%), nearly to the extent of that of wild-type SMARTA CD4+ T cells (70% versus 80%; Fig. 5). Furthermore, Bcl-6 expression induced similar CXCR5 expression in wild-type and Itch−/− SMARTA CD4+ T cells (Fig. 5b). These results indicated that Itch might function mainly upstream of Bcl-6 expression and might be required for the induction of Bcl-6 expression and TFH cell differentiation.

Figure 5: Restoration of the defective TFH differentiation of Itch−/− cells by overexpression of Bcl-6.
figure 5

(a) Flow cytometry of donor (CD45.1+) CD4+ T cells obtained from B6 host mice given adoptive transfer of wild-type or Itch−/− SMARTA CD4+ T cells transduced with retroviral vector expressing GFP only (empty vector (EV)) or GFP and Bcl-6 (Bcl-6), followed by infection of recipient mice with LCMV and analysis 8 d after infection. Numbers adjacent to outlined areas indicate percent SLAMloCXCR5+ cells (TFH cells). (b) Proliferation of SMARTA CD4+ T cells, quantified as total SMARTA cells in the spleen (left), frequency of TFH cells among SMARTA (CD45.1+) CD4+ T cells (middle), and CXCR5 expression, presented as mean fluorescence intensity (right), for cells from host mice in a. Each symbol represents an individual mouse; small horizontal lines indicate the mean (±s.d.). *P < 0.05, **P < 0.01 and ***P < 0.001 (Student's t-test). Data are representative of three independent experiments with three to four mice per group.

Foxo1 as a target for Itch

We next looked for other potential targets of Itch that may be involved in regulating the differentiation of TFH cells, with a particular interest in upstream regulators of Bcl-6. Foxo proteins belong to the forkhead-box family of transcription factors, which are characterized by a conserved winged-helix DNA-binding domain called the 'forkhead' domain. Foxo transcription factors are subject to extensive and varied post-translational modifications that affect their abundance, localization and transcriptional activity, with ubiquitination being one major pathway by which Foxo factors are regulated36. It has been well documented that the Foxo family can be negatively regulated by the signaling pathways of phosphatidylinositol-3-OH kinase (PI(3)K) and the kinase Akt in response to insulin, growth factors or the engagement of costimulatory receptors (CD28 and ICOS)36,37. Phosphorylation of Foxo factors at three conserved sites (Tyr24, Ser256 and Ser319) by the kinases Akt and SGK ('serum and glucocorticoid-induced kinase') causes their export from the nucleus and degradation and thereby prevents Foxo factors from transactivating or repressing their target genes36. Ubiquitination and degradation of Foxo1 can be mediated by SKP1–CUL1–F-box protein–SKP2 complex38. Relevant to our study here, Foxo1 has been linked to the regulation of TFH cells either in a positive manner or a negative manner18,39. However, the importance of the involvement of Foxo1 or its homologs in the differentiation of TFH cells remains unclear, and the underlying molecular mechanisms by which Foxo1 is regulated in this process have not been elucidated so far.

We then investigated whether Foxo1 is a substrate of Itch. Given the presence of proline-rich sequences in Foxo proteins, we assessed the ability of Itch to recognize Foxo proteins. We first coimmunoprecipitated proteins from Jurkat human T lymphocyte cells that overexpressed Foxo proteins and Itch. Only Foxo1 immunoprecipitated together with Itch, whereas Foxo3a protein did not (Fig. 6a), which suggested that Itch selectively interacted with Foxo1. Notably, we also detected endogenous interaction between Foxo1 and Itch in purified CD4+ T cells from VACV-infected B6 mice (Fig. 6b), which suggested that this interaction was physiologically functional. More notably, the interaction between Foxo1 and Itch in CD4+ T cell blasts was rapidly induced by restimulation with monoclonal anti-ICOS and monoclonal anti-CD3 (Fig. 6c).

Figure 6: Itch interacts with Foxo1 and targets it for ubiquitination and degradation.
figure 6

(a) Immunoassay of lysates of Jurkat T cells transfected with plasmids encoding Itch and either Myc-tagged (Myc-Foxo1) or Myc-tagged Foxo3a (Myc-Foxo3a), assessed by immunoprecipitation (IP) with anti-Itch (α-Itch) and immunoblot analysis (IB) with anti-Myc or anti-Itch. Lysate (bottom), immunoblot analysis of an aliquot of the cell lysate without immunoprecipitation; analysis with antibody to β-actin (α-actin) serves as a loading control (throughout). Right margin, molecular size in kilodaltons (kDa). (b) Immunoassay of combined cytosolic and nuclear fractions of CD4+ T cells purified from B6 mice 8 d after infection with VACV, assessed by immunoprecipitation with IgG (isotype-matched control antibody) or anti-Foxo1 and immunoblot analysis with anti-Itch or anti-Foxo1. (c) Immunoassay of combined cytosolic and nuclear fractions of CD4+ T cell blasts stimulated for 0 or 15 min with anti-CD3 (3 μg/ml) and anti-ICOS (2 μg/ml), assessed by immunoprecipitation with anti-Foxo1 and immunoblot analysis with anti-Itch or anti-Foxo1. (d) Immunoassay of Jurkat T cells transfected with plasmids encoding Foxo1 and wild-type Itch and/or hemagglutinin-tagged ubiquitin (HA-Ub), then left untreated (−) or treated (+) for 1 h with MG132 (25 μM), assessed by denaturation of lysates in 1% SDS, immunoprecipitation with anti-Foxo1 and immunoblot analysis with anti-hemagglutinin (anti-HA) and anti-Foxo1 (lysates without immunoprecipitation probed with anti-Itch or anti-Foxo1). Ub(HA)-Foxo1 (top right), hemagglutinin-tagged ubiquitinated Foxo1. (e) Immunoassay of wild-type or Itch−/− CD4+ T cell blasts pretreated with MG132 and stimulated for 20 min with anti-CD3 (3 μg/ml) and anti-ICOS (2 μg/ml), assessed by denaturation of lysates in 1% SDS, immunoprecipitation with polyclonal anti-ubiquitin (α-Ub) and immunoblot analysis with monoclonal anti-ubiquitin (P4D1; left blot) or anti-Foxo1 (top right blot); lysates without immunoprecipitation (bottom right blot) probed with anti-Foxo1, anti-Itch or anti-β-actin. Poly-Ub, polyubiquitination; Ub-Foxo1, ubiquitinated Foxo1. (f) Immunoblot analysis of lysates of wild-type or Itch−/− CD4+ T cell blasts stimulated for 0–60 min (above lanes) with anti-CD3 (3 μg/ml) and anti-ICOS (2 μg/ml), probed with anti-Foxo1 (α-Foxo1), antibody to Foxo1 phosphorylated at Ser256 (α-p-Foxo1(S256)), antibody to Akt phosphorylated at Ser473 (α-p-Akt(S473)) or Thr308 (α-p-Akt(T308)), anti-Itch (α-Itch) or anti-β-actin (α-actin). Data are representative of two independent experiments.

To investigate whether Itch promoted the ubiquitination of Foxo1, we transfected Jurkat T cells with plasmids expressing hemagglutinin-tagged ubiquitin and Myc-tagged Foxo1 (and Xpress-tagged wild-type Itch). We then either left the cells untreated or treated them with the proteasome inhibitor MG132 and then immunoprecipitated proteins from cell lysates with anti-Foxo1. Overexpression of wild-type Itch enhanced the conjugation of ubiquitin to Foxo1 in the presence of MG132 (Fig. 6d). To assess the in vivo ubiquitination of Foxo1, we generated a new rabbit polyclonal antibody to ubiquitin and used this antibody in an assay in which we immunoprecipitated ubiquitinated protein. In these experiments, we pretreated CD4+ T cell blasts with MG132 and then restimulated them with monoclonal anti-CD3 and monoclonal anti-ICOS. After restimulation, we immunoprecipitated proteins from lysates of Itch−/− or wild-type CD4+ T cells with the polyclonal antibody to ubiquitin. We detected polyubiquitinated Foxo1 as slowly migrating, high-molecular-weight species in wild-type cells that was almost completely absent from Itch−/− cells (Fig. 6e). Together these results suggested that Itch acted as an E3 ligase for the ubiquitination of Foxo1.

It has been reported that the PI(3)K-Akt pathway induces the phosphorylation of Foxo1 and promotes its degradation and that ICOS regulates TFH cell differentiation through the PI(3)K-Akt pathway37,40. These reports prompted us to investigate whether an ICOS-PI(3)K-Akt pathway induces the degradation of Foxo1 by Itch. We examined the signal-transduction events and their consequences in ICOS-stimulated wild-type and Itch-deficient T cells. As described above, we restimulated CD4+ T cell blasts with monoclonal anti-CD3 and monoclonal anti-ICOS. Consistent with published findings40, engagement of ICOS substantially enhanced the activation of Akt mediated via the T cell antigen receptor, as indicated by increased phosphorylation of Akt at Ser473 in wild-type CD4+ T cells (Fig. 6f). Phosphorylation of Foxo1 was also enhanced by ligation of ICOS in wild-type CD4+ T cells. However, phosphorylation of Akt Ser473 and Foxo1 was similarly increased by ligation of ICOS in Itch-deficient CD4+ T cells. Notably, engagement of ICOS resulted in decreased expression of Foxo1 protein after 10 min of stimulation in wild-type CD4+ T cells. In contrast, the expression of Foxo1 protein remained largely unchanged in Itch-deficient T cells. Together these results suggested that Itch was not involved in the ICOS-triggered signaling events that resulted in the phosphorylation of Foxo1 but that Itch was critically required for the degradation of Foxo1 protein.

Foxo1 ablation rectifies defective TFH cell differentiation

The expression of Foxo1 protein was much lower in TFH cells than in naive cells or non-TFH cells and was even lower in GC TFH cells (Fig. 7a). Notably, the expression of Foxo1 mRNA was substantial in each population (Fig. 7a), consistent with a central role for post-translational degradation in the control of Foxo1 expression. However, the expression of Itch protein and Itch mRNA was retained in all populations (Fig. 7a). This indicated that TFH cell differentiation might require downregulation of Foxo1 expression through post-translational modification by Itch. We therefore tested mice conditionally deficient in Foxo1 or Foxo3a. The differentiation of TFH cells in response to acute infection with VACV was enhanced in Foxo1fl/flCD4-Cre mice compared with that of their wild-type (Foxo1+/+ CD4-Cre or Foxo1fl/fl) counterparts (Supplementary Fig. 5a). Consequently, the number of GC B cells and plasma cells was also greater in Foxo1fl/flCD4-Cre mice than in their wild-type counterparts after infection with VACV (Supplementary Fig. 5b). However, Foxo3afl/flCD4-Cre mice showed normal differentiation of TFH cells, as well as normal development of GC B cells and plasma cells after infection (Supplementary Fig. 5c,d). These results suggested that Foxo1 was a negative regulator of the differentiation of TFH cells.

Figure 7: Foxo1 deficiency rectifies the defective differentiation of TFH cells in Itch−/− mice.
figure 7

(a) Immunoblot analysis of Foxo1 and Itch (left) and real-time PCR analysis of Foxo1 and Itch mRNA (right) in naive cells, non-TFH (CXCR5PD-1) cells, TFH (CXCR5+PD-1lo) cells and GC TFH (CXCR5hiPD-1hi) cells sorted from wild-type mice 8 d after infection with VACV; mRNA results were normalized to those of Actb mRNA and are presented relative to those of naive cells, set as 1. (b) Real-time PCR analysis of the mRNA products of Foxo1 targets in donor CD4+ T cells from B6 recipient mice given transfer of CD4+ T cells from wild-type or Itch−/− SMARTA donor mice, followed by infection of recipients with LCMV and analysis 3 d later; mRNA results were normalized to those of Actb mRNA and are presented relative to those of naive wild-type SMARTA CD4+ T cells, set as 1. (c) Flow cytometry of splenocytes from B6 mice not treated with poly(I:C) (far left) or from wild-type (Itchfl/fl) mice (WT), Itch-iKO mice and Itch-Foxo1-iDKO mice given intraperitoneal injection of poly(I:C), followed by infection of all mice with VACV 2 weeks later and analysis 8 d after infection (numbers in plots as in Fig. 1c). (d) Frequency of GC B cells and plasma cells among B220+ B cells from mice as in c (n = 4–5 per group). Each symbol represents an individual mouse; small horizontal lines indicate the mean (±s.d.). (e) Flow cytometry of activated (CD44hi) CD4+ T cells from mice as in c (numbers in plots as in Fig. 1a). (f) Frequency of TFH cells and GC (CXCR5+Bcl-6hi) TFH cells among activated (CD44hi) CD4+ T cells from mice as in c (n = 4–5 per group). *P < 0.05, **P < 0.01 and ***P < 0.001 (Student's t-test). Data are representative of two experiments (a,b; error bars, s.d.) or at least three experiments (cf).

We next sought to determine whether Itch affects Foxo1-mediated gene expression. We sorted wild-type and Itch−/− SMARTA CD4+ T cells from B6 recipient mice at day 3 after infection with LCMV and analyzed expression of Foxo1 targets by real-time PCR. Itch−/− cells had much higher expression of several known Foxo1 targets41,42, such as Bcl2l11, Il7r, Klf2, Sell, Selplg and S1pr1, than did wild-type SMARTA CD4+ T cells (Fig. 7b). These data suggested that Itch suppressed the expression of at least some of Foxo1 targets.

We then investigated whether the defective differentiation of TFH cells in Itch−/− mice could be restored by genetic ablation of Foxo1. Because of the technical difficulty in generating Itchfl/flFoxo1fl/fl CD4-Cre+ mice, we crossed Itchfl/flFoxo1fl/fl mice with transgenic mice in which Cre recombinase is under control of the interferon-responsive Mx1 promoter to generate Itchfl/flFoxo1fl/flMx1-Cre mice with inducible deficiency in both Itch and Foxo1 (Itch-Foxo1-iDKO mice). Immunoblot analysis revealed that the expression of Itch and/or Foxo1 was almost completely eliminated in splenocytes 2 weeks after injection of the synthetic RNA duplex poly (I:C) (polyinosine-polycytidylic acid) (Supplementary Fig. 6a). Following infection with VACV, the development of GC B cells in Itchfl/flMx1-Cre mice with inducible deficiency in Itch (Itch-iKO) was much lower than that in their wild-type counterparts (Fig. 7c,d and Supplementary Fig. 6b,c), consistent with the phenotype of Itch−/− mice and Itch-cKO mice. Likewise, the development of plasma cells was also significantly lower in Itch-iKO mice than in their wild-type counterparts (Fig. 7c,d and Supplementary Fig. 6b,c). Unexpectedly, the development of both GC B cells and plasma cells in Itch-Foxo1-iDKO mice was increased to their development in wild-type mice (Fig. 7c,d and Supplementary Fig. 6b,c). We next examined the differentiation of TFH cells and GC TFH cells. The frequency and absolute number of TFH cells was much lower in Itch-iKO mice than in wild-type mice (Fig. 7e,f and Supplementary Fig. 6c). The frequency and number of GC TFH cells was also lower in Itch-iKO mice than in wild-type mice. Notably, the differentiation of TFH cells and GC TFH cells was restored in Itch-Foxo1-iDKO mice (Fig. 7e,f and Supplementary Fig. 6c). In contrast, IL-4 production in CD4+ T cells from Itch-iKO mice was similar to that in CD4+ T cells from Itch-Foxo1-iDKO mice (Supplementary Fig. 6d). These results indicated that deletion of Foxo1 did not affect Itch-mediated production of IL-4. Together these data suggested that the defect in the differentiation of GC B cells and TFH cells in Itch−/− mice could be 'rescued' by genetic ablation of Foxo1.

To investigate whether the restoration of TFH cell differentiation in Itch−/− mice via deficiency in Foxo1 was cell intrinsic, we knocked down Foxo1 expression in wild-type or Itch−/− SMARTA CD4+ T cells with a retrovirus encoding shRNA and transferred transduced and untransduced cells together into B6 recipient mice, followed by infection of the host mice with LCMV. Transduction of control shRNA (targeting CD8α) did not alter the TFH differentiation of either wild-type cells or Itch−/− cells (Supplementary Fig. 7a–c). In contrast, transduction of shRNA targeting Foxo1 substantially enhanced the TFH differentiation of wild-type cells (Supplementary Fig. 7a–c). Notably, knockdown of Foxo1at least partially restored the TFH differentiation of Itch−/− cells (Supplementary Fig. 7a–c). The data further confirmed a critical role for Foxo1 in the regulation of TFH cell differentiation by Itch.

Discussion

Accumulating evidence has established TFH cells as a distinct and important type of CD4+ helper T cell that has a crucial role in humoral responses to pathogen infection and damaging roles in autoimmune diseases. Bcl-6 has been identified as a master regulator of TFH cell differentiation4,5,6. However, how Bcl-6 is induced and how TFH cell differentiation is regulated still has many unknown aspects19. In this study, through the use of a combination of genetic, cellular and molecular approaches, we have identified a previously unknown and critical function for Itch in TFH cell differentiation and humoral immunity. We have shown that Itch was intrinsically required for both early stages and late stages of TFH cell differentiation and was associated with a substantial reduction in Bcl-6 expression. Finally, we have also shown that Itch regulated TFH cell differentiation by targeting Foxo1, a negative regulator of such differentiation, for degradation.

On the B6 background, Itch−/− mice develop chronic inflammatory diseases and constant itching of the skin43. Itch inhibits TH2 differentiation by targeting JunB for ubiquitination and degradation26. This might all contribute to the chronic inflammatory diseases of Itch−/− mice. However, Itch−/− mice unexpectedly showed a substantial defect in TFH cell differentiation in response to viral infection. Although IL-4 is one of the cytokines produced by GC TFH cells that is required for the optimal provision of help to B cells, it has been shown that IL-4 is dispensable for TFH cell differentiation. Our data support that notion and clarify that the defective differentiation of TFH cells in Itch−/− mice was not due to TH2 bias, because TFH cell differentiation was also unaffected in Il4−/− mice; the defect in the differentiation of TFH cells in Itch−/− mice was not rectified by IL-4 deficiency; and genetic ablation of Foxo1 did rectify the defect in TFH cells in Itch-deficient mice without affecting IL-4 production.

Foxo1 can be targeted for ubiquitination by several ubiquitin E3 ligases, including the SKP1–CUL1–F-box protein–SKP2 complex, MDM2, COP1 and CHIP ('C terminus of Hsc70-interacting protein'), in various cell types44. In this study, we identified Itch as an additional E3 ligase that targeted Foxo1 for ubiquitination and degradation. We also established that the ICOS-PI(3)K-Akt pathway led to the ubiquitination and degradation of Foxo1 by Itch. However, future studies are needed to explore details of the mechanism, including the sites modified and the polyubiquitination chain of Foxo1. In addition, the partial 'rescue' of TFH cell differentiation in the experiment in which Foxo1 was knocked down by shRNA might suggest that Itch targets other substrates. Although we found that Bcl-6 itself might be a target, the Itch–Bcl-6 association was not required for the development of TFH cells. The data reinforce the notion that Foxo1 is a critical substrate of multiple Itch targets for TFH cell differentiation. Future efforts should provide a more comprehensive understanding of the molecular interactions among the potential participants.

Emerging evidence has shown that Foxo1 and Foxo3a are involved in immunological regulation. Foxo1 and Foxo3a function redundantly as transcription factors important in the promotion of Foxp3 expression27,39,45. Although Foxo3a can bind and activate the Bcl6 promoter in B cell lymphoma lines46, mice with T cell–specific deficiency in Foxo3a exhibited normal TFH cell differentiation. Published studies and also our study here have shown that large numbers of TFH cells accumulate in mice with T cell–specific deficiency in Foxo1 maintained under standard housing conditions39 or infected with a specific pathogen. However, whether this excessive formation of TFH cells is cell intrinsic or is due to loss of regulatory T cells has remained unclear39. A scan of the Bcl6 promoter identified Foxo-binding motifs in the DNA18. Although chromatin-immunoprecipitation experiments have suggested that Foxo1 binds directly to putative Foxo-binding motifs in the Bcl6 promoter18,42, the consequence of such binding remains controversial. Luciferase experiments have suggested that Foxo proteins, including Foxo1, activate the Bcl6 promoter18. However, the in vivo data we have presented here indicated a negative role for Foxo1 in Bcl-6 expression and TFH cell differentiation. Future studies are needed to clarify this issue. In addition to Bcl6, other TFH cell–related genes, such as Cxcr4, Batf, Icos and Prdm1, have also been reported as containing Foxo1-binding sites identified by chromatin immunoprecipitation followed by deep sequencing42. Therefore, Foxo1 may regulate TFH cell differentiation by directly controlling the expression of TFH cell–related genes. Foxo1 can also regulate lymphocyte trafficking by inducing L-selectin and the chemokine receptor CCR7 (ref. 41). The proposal that Foxo1 directly binds to the promoter, untranslated region, introns or intergenic regions of some genes encoding homing molecules, such as Ccr7, S1pr1, Sell and Selplg, has been further supported by chromatin immunoprecipitation followed by deep sequencing42. Two other studies have reported that microRNAs of the miR1792 family regulate TFH cell differentiation by targeting the PI(3)K antagonist PTEN47,48 and the phosphatase PHLPP2 (ref. 48) in the ICOS-PI(3)K-Akt pathway. One of those studies also showed that the miR1792 microRNAs are required for the ability of TFH cells to migrate to B cell follicles and GCs, although no direct mechanism was shown in that study48. It is reasonable to speculate that Foxo1 and Itch also regulate TFH cell differentiation at least partially through T cell migration.

In summary, our findings have established a function for Itch as a crucial regulatory factor in TFH cell differentiation. In addition, we have shown that Itch positively regulated such differentiation by promoting the conjugation of ubiquitin to Foxo1 and subsequent degradation of Foxo1. Given that published studies have shown that Itch has a negative role in regulating the differentiation of TH2 cells and inflammatory signaling, we propose that Itch is a key participant in the control of both TH2 cell–mediated allergic inflammation and TFH cell–promoted B cell immunity. Understanding how Itch regulates such different processes may be useful in both rational vaccine design for human infectious diseases and therapeutic intervention in human inflammatory diseases.

Methods

Mice.

Itch−/− mice on a C57BL/6J (B6) background have been described26. Heterozygous Itch+/− mice were intercrossed to generate Itch+/+, Itch+/− and Itch−/− littermates. Itchfl/fl mice on the B6 background have been described30. SMARTA mice (with transgenic expression of a T cell antigen receptor specific for the epitope of LCMV glycoprotein amino acids 66–77 presented by I-Ab) were fully backcrossed to the B6 background31. Itch−/− SMARTA mice were generated by the crossing of wild-type SMARTA to Itch−/− mice. Il4−/−, CD4-Cre and Mx1-Cre mice were from the Jackson Laboratory. Foxo1fl/fl and Foxo3afl/fl mice on the 129 background were provided by R.A. DePinho and were bred onto the B6 background for six generations. All animal protocols were approved by members of the Institutional Animal Care and Use Committee of the La Jolla Institute for Allergy and Immunology.

Antibodies and reagents.

Anti-CD4 (GK1.5 and RM4-5), anti-CD25 (PC61.5), anti-CD44 (IM7), anti-CD45.1 (A20), anti-CD45.2 (104), anti-ICOS (15F9), anti-IgD (11-26) and streptavidin were from eBioscience. Anti-B220 (RA3-6B2), anti-CD62L (MEL-14), anti-CD69 (H1.2F3), anti-CD150 (anti-SLAM; TC15-12F12.2), anti-interferon-γ (XMG1.2) anti-ICOS purified to be low endotoxin and azide free (used for in vitro stimulation; C398.4A) and anti-CD3 (2C11) were from Biolegend. Anti-CD8α (53-6.7), anti-CD95 (FAS, Jo2), anti-CD138 (281-2), anti-PD-1 (J43), anti-CXCR5 (2G8), biotin-conjugated anti-CXCR5 (2G8), anti-Bcl-6 (K112-91), anti-CD95 (Jo2), antibody to the T cell– and B cell–activation antigen (GL7) and anti-IL-4 (BVD4-1D11) were from BD Pharmingen. Anti-Itch (32/Itch) was from BD Transduction Laboratories. Antibody to Akt phosphorylated at Ser473 (D9E) or Thr308 (244F9), anti-Foxo1 (L27 and C29H4) and antibody to Foxo1 phosphorylated at Ser256 (9461) were from Cell Signaling Technology. Anti-Foxo1 (ab39670) was from Abcam. Anti-Myc (9E10), anti-actin (C4) and anti-ubiquitin (P4D1) were from Santa Cruz. Purified anti-CD28 (37.51) was from Bio-X-Cell. Purified antibody to Armenian hamster IgG (127-005-099) and biotin–conjugated goat antibody to rat IgG (112-065-167) were from Jackson ImmunoResearch. Polyclonal antibody to ubiquitin was raised in a rabbit immunized with ubiquitin peptide, in collaboration with Millipore. Recombinant human IL-2 and mouse IL-7 were from Peprotech.

Plasmids and cell transfection.

cDNA encoding Itch or hemagglutinin-tagged ubiquitin has been described26. cDNA encoding Foxo1 or Foxo3a (Addgene) was subcloned into plasmid pEF4mychisB. For the construction of shRNA expression vectors, oligonucleotides were cloned into the vector pLMP-Ametrine.

For protein expression in Jurkat T cells, cells were transfected with the appropriate amount of plasmid (5–20 μg) by electroporation (260 V and 950 μF; Bio-Rad). Cells were lysed for 30 min on ice in NP-40 lysis buffer (1% NP-40, 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 5 mM NaPiP, 5 mM NaF, 2 mM Na3VO4 and 10 μg/ml each of aprotinin and leupeptin) and cell debris were removed by centrifugation at 13,200 r.p.m. for 10 min at 4 °C.

Retroviral transduction and cell transfer.

shRNA-containing oligonucleotides (Supplementary Table 1) were cloned into the plasmid pLMPd-Ametrine. Plat-E packaging cells were transfected with 3 μg of retroviral vector (pMIG-GFP or pMIG–Bcl-6 (ref. 4), or pLMPd-Ametrine) along with 9 μl of TransIT-LT1 transfection reagent (Mirus). 48 h after transfection, the culture supernatant containing retrovirus was collected. Naive CD4+CD44CD62L+CD25 T cells were stimulated with plate-bound anti-CD3 (2C11; Biolegend) and anti-CD28 (37.51; Bio-X-Cell). 24 h after stimulation, CD4+ T cells were infected with retrovirus together with 10 μg/ml polybrene and 100 U/ml IL-2 by centrifugation of cells at 2,000 r.p.m. for 60 min at room temperature. A second transduction was repeated 24 h after the first transduction. 24 h after the second transduction, cells were transferred to fresh medium with 100 U/ml IL-2, followed by incubation for 2 d. Cells were then transferred to new medium with 1 ng/ml IL-7, followed by incubation for 24 h before being sorted. Naive or retrovirus-transduced SMARTA CD4+ T cells were transferred into recipient mice by intravenous injection into the retro-orbital sinus. Naive SMARTA CD4+ T cells were transferred for experiments on day 3 (5 × 105 cells) and day 8 (2.5 × 103 cells). 5 × 105 or 2.5 × 104 retrovirus-transduced SMARTA CD4+ T cells were transferred for experiments at day 3 or day 8, respectively.

Bone marrow chimeras with knockdown by shRNA.

For the generation of chimeric mice expressing shRNA, Plat-E cells were transfected with 3 μg of pLMPd-Ametrine vector through the use of 9 μl of TransIT-LT1 (Mirus). 48 h after transfection, the culture supernatant containing retrovirus was collected. SMARTA (CD45.1+) bone marrow depleted of mature T cells was cultured for 24 h in complete DMEM containing 10 ng/ml IL-3, 10 ng/ml IL-6, and 100 ng/ml stem cell factor (all from Peprotech) before initial retroviral infection. Bone marrow depleted of mature T cells as infected with retrovirus along with 5 μg/ml polybrene (Sigma-Aldrich) by centrifugation (2,000 r.p.m. for 1 h). 24 h after infection, retrovirus-transduced bone marrow cells were injected into lethally irradiated (900 rads) B6 recipient mice.

Infection.

Stocks of VACV (Western Reserve strain) and LCMV (Armstrong strain) were prepared and 'titrated' as described49. Virus was prepared in plain DMEM. Each mouse was infected with 5 × 105 or 1 × 105 plaque-forming units of LCMV (Armstrong strain) for experiments at day 3 or day 8, respectively, by bilateral intraperitoneal injection. Each mouse was infected with 2.5 × 105 VACV (Western Reserve strain) by bilateral intraperitoneal injection.

Neutralization of IL-2 in vivo

Anti-IL-2 (S4B6) and isotype-matched control antibody (rat IgG2a; 2A3) were from Bio-X-Cell. Each B6 mouse was treated with 0.5 mg anti-IL-2 or isotype-matched control antibody by both intraperitoneal and retro-orbital injection 24 h before infection with LCMV and then again 24 h after infection with LCMV.

Inducible deletion of Itch and/or Foxo1.

For inducible deletion of Itch and/or Foxo1, Itchfl/fl and/or Foxo1fl/fl mice were crossed with mice expressing Cre recombinase under control of the interferon-responsive Mx1 promoter (Mx1-Cre). For activation of the interferon-inducible Mx1-Cre, 6-week-old mice were injected intraperitoneally with 250 μl of 1 mg/ml poly(I:C) every the other day for a total of three injections. All mice were infected with VACV 2 weeks after the final poly(I:C) treatment.

ELISA.

Concentrations of cytokines in serum were measured by sandwich ELISA according to the instructions from Biolegend. Anti-VACV IgG was quantified by ELISA in lysates of cells infected with VACV that had been inactivated by ultraviolet irradiation as the capture antigen. 96-well PolySorp microtiter plates (Nunc) were coated overnight lysates of cells infected with VACV inactivated by ultraviolet irradiation, in PBS. After incubation of sample serum, plates were incubated with biotin-conjugated goat antibody to mouse IgG (1030-08; Southern Biotech), followed by horseradish peroxidase–conjugated streptavidin (7100-05; Southern Biotech) and then tetramethylbenzidine substrate solution (172-1068; Bio-Rad).

Real-time quantitative PCR.

Total RNA was extracted with an RNeasy mini or RNeasy plus micro kit (Qiagen). cDNA was synthesized with SuperScript III Reverse Transcriptase (Invitrogen) and oligo(dT). Quantitative PCR was done in duplicate with iTaq Universal SYBR Green Supermix (Bio-Rad) on a Roche LightCycler 480 (Roche). β-actin was used as the reference for normalization. Primers for Itch (QT01048684) were from Qiagen; all other primers are in Supplementary Table 2.

Immunoprecipitation, glutathione S-transferase precipitation assay and immunoblot analysis.

Proteins were immunoprecipitated by incubation of the cell lysates overnight at 4 °C with the appropriate antibodies (1 μg; all identified above), followed by the addition of protein A/G–Magnetic beads (88802; Thermo Scientific) and incubation for another 2 h at 4 °C. Immunoprecipitates were washed five times with NP-40 lysis buffer and were boiled in 50 μl SDS loading buffer. For endogenous coimmunoprecipitation, a combination of cytosolic and nuclear fractions from primary CD4+ T cells was incubated with anti-Foxo1 (ab39670; Abcam). For glutathione S-transferase (GST) precipitation assays, a GST fusion protein or GST alone (5 μg) was added to lysates of Jurkat cells, followed by incubation for 2 h at 4 °C, then 50 μl of glutathione-Sepharose beads (17-0756-01; GE Healthcare) was added, followed by incubation for another 1 h. Precipitates were washed five times with NP-40 lysis buffer and were boiled in 50 μl SDS loading buffer. For visualization of ubiquitinated protein, 1.0% SDS was added to lysis buffer for disruption of nonspecific protein interactions. Cell lysates were denatured by being boiled for 15 min and then were diluted to a concentration of 0.1% SDS before immunoprecipitation. Samples were separated to 10–12% SDS-PAGE, followed by electrotransfer to PVDF membranes (Millipore). Membranes were analyzed by immunoblot with the appropriate antibodies (all identified above), followed by horseradish peroxidase-conjugated second antibody (NA931V or NA934V; GE Healthcare) and development with an enhanced chemiluminescence detection system (RPN2106, RPN2232 or RPN2235; GE Healthcare).

T cell–restimulation assay.

Primary T cells were cultured in RPMI-1640 medium supplemented with 10% FBS, glutamine, HEPES (10 mM), sodium pyruvate (1 mM), β-mercaptoethanol, penicillin and streptomycin. Primary CD4+ T cells were isolated from spleens and lymph nodes of 2- to 3-month-old wild-type and Itch−/− mice with a negative selection kit from BD. For the preparation of CD4+ T cell blasts, T cells were activated for 2 d by plate-bound anti-CD3 (3 μg/ml) and anti-CD28 (3 μg/ml) (both identified above) and were allowed to 'rest' for 1 d in medium alone. The CD4+ T cell blasts were harvested and then were restimulated by combinations of anti-CD3 (3 μg/ml) and anti-ICOS (2 μg/ml) (both identified above). The antibodies were crosslinked at 37 °C by goat antibody to hamster IgG (20 μg/ml; identified above) and the cells were restimulated for various times. After being restimulated, cells were lysed for 20 min in 2× SDS loading buffer and were boiled for 10 min. Cell lysates were then subjected to immunoblot analysis with the appropriate antibodies (all identified above).

Flow cytometry.

Single-cell suspensions of splenocytes were prepared by mashing of spleens through a cell strainer. After treatment with red-blood-cell–lysis buffer, surfaces of cells in suspension were stained with fluorochrome-conjugated antibodies (all identified above) in flow cytometry buffer (0.5% BSA and 0.05% NaN3 in PBS). For experiments at day 8 of infection with VACV, a three-step CXCR5 staining was performed with purified anti-CXCR5 (2G8; BD PharMingen), followed by biotinylated goat antibody to rat IgG (112-065-167; Jackson ImmunoResearch) and then phycoerythrin-indotricarbocyanine–labeled streptavidin (25-4317-82; eBioscience), with each staining step done at 4 °C in CXCR5 staining buffer (0.5% BSA, 2% FCS and 2% normal mouse serum in PBS). For adoptive transfer experiments with LCMV infection, a two-step CXCR5 staining was performed with biotinylated anti-CXCR5 (2G8, BD PharMingen), followed by phycoerythrin–indotricarbocyanine–labeled streptavidin.

Intracellular cytokines were stained after stimulation of cells for 4 h with 50 ng/ml PMA (phorbol 12-myristate 13-acetate; Sigma-Aldrich) and 1 μg/ml ionomycin (Sigma-Aldrich) in the presence of GolgiStop. Cells were incubated with antibodies to cell surface markers (all identified above), and then were fixed and permeabilized with Cytofix/Cytoperm Buffer (51-2090KZ; BD Biosciences). Cells were then stained with antibodies to cytokines (all identified above). Intracellular Bcl-6 (K112-91; BD Bioscience) was stained after cell-surface staining. Samples were fixed and permeabilized with Foxp3 staining buffer according to the manufacturer's manual (00-5523 ; eBioscience). Samples were incubated for 45–60 min at 4 °C with Fixation/Permeabilization buffer and washed with 1× permeabilization buffer. Samples were incubated for another 45–60 min with Alexa Fluor 647–conjugated monoclonal antibody to Bcl-6 (identified above) in permeabilization buffer.

Statistical analysis.

All data were analyzed by a paired or unpaired t-test with GraphPad Prism 5.0.