B cell priming for extrafollicular antibody responses requires Bcl-6 expression by T cells

We began by investigating whether Bcl-6 expression is confined to GC T cells during T-dependent immune responses by assessing the distribution and kinetics of Bcl-6–expressing cells during a T-dependent response to sheep RBCs (SRBCs). In unimmunized mice, there were virtually no Bcl-6⁺ T cells detected outside rare background GCs (Fig. 1 A). Between days 2 and 3, CD3⁺ Bcl-6⁺ PD-1^lo cells were seen at the T–B border. By day 7, CD3⁺ cells expressing higher amounts of Bcl-6 and PD-1 appeared within newly formed GCs (Fig. 1, B and C). As has been reported before (Haynes et al., 2007; Linterman et al., 2009a), and most were Bcl-6⁺ CD3⁺ PD-1^hi T cells were rarely seen at sites other than the GC. We conclude that high expression of PD-1 identifies GC Tfh cells. Together, these results suggest that the T cells that interact with B cells at the outer T zone have up-regulated Bcl-6, raising the possibility that expression of Bcl-6 influences B cell commitment to both the follicular and the extrafollicular pathways.

To evaluate the influence of T cell–expressed Bcl-6 on the early antibody response to NP-KLH, we constructed mixed chimeras in which sublethally irradiated Rag1^−/− mice were reconstituted with 80% Bcl6^−/− CD45.2:20% Cd28^−/− CD45.1 fetal liver. Cd28^−/− fetal liver cells help maintain the chimeras’ health and provide a source of Bcl-6⁺ B cells and T cells that cannot differentiate into effector cells, including Tfh cells. A control group was reconstituted with 80% Bcl6^+/+ CD45.2:20% Cd28^−/− CD45.1 fetal liver.

Chimeric mice were immunized i.p. with NP-KLH in alum. 7 d after immunization, mice lacking functional Bcl-6–expressing T cells did not form GC (Fig. 2, A and B) or Tfh cells (Fig. 2, C and D). This concurs with previously published data showing that Bcl-6–expressing Tfh cells are required for the formation of GCs (Johnston et al., 2009; Nurieva et al., 2009; Yu et al., 2009). Mice lacking functional Bcl-6–expressing T cells had an ∼50% reduction in switched Syndecan-1⁺ (CD138⁺) plasma cells in the spleen (Fig. S1, A and B). NP-specific IgG1 titers were reduced by >40-fold, and a more modest reduction in NP-specific IgM titers was also observed (Fig. 2 E). By day 7, few extrafollicular foci containing IgG1⁺ plasma cells were visible (Fig. 2 F) in the spleens of chimeras lacking Bcl-6 expression in T cells. Together, these results suggest that Bcl-6–expressing T cells are important for optimal extrafollicular antibody responses to protein antigens.

Although the aforementioned results suggest a role for Bcl-6 in extrafollicular antibody responses, the origin of the antibody-producing cells cannot be precisely identified in this model. For example, an early wave of memory B cells that develops in parallel to the extrafollicular antibody response independently of GC, Bcl-6, ICOS (inducible T cell co-stimulator), and IL-21 has been now described by several groups (Toyama et al., 2002; Inamine et al., 2005; Chan et al., 2009; Linterman et al., 2010; Zotos et al., 2010). Another potential caveat of these experiments is that the absence of Bcl-6 expression in 80% of the B cells may influence the extrafollicular plasma cell response, although Bcl-6–deficient B cells would be expected to give rise to enhanced rather than decreased plasma cell responses as the result of unopposed Blimp-1 expression.

To control for these possibilities and accurately quantify the number of EFPBs produced, we took advantage of the SW_HEL adoptive transfer system. CD45.1 SW_HEL transgenic B cells bearing a rearranged HEL-specific VDJ_H element targeted into the IgH chain locus combined with an HEL-specific k L-chain transgene (Phan et al., 2003) were transferred i.v. into mixed Bcl6^−/− CD45.2:Cd28^−/− CD45.2 or Bcl6^+/+ CD45.2:Cd28^−/− CD45.2 chimeric mice in combination with HEL protein conjugated to SRBC. Thus, transferred SW_HEL B cells have an intact Bcl6 gene, can undergo class switching, and develop into both GC B cell and EFPB populations in response to HEL (Phan et al., 2003; Paus et al., 2006). In this model, GC B cells and EFPBs can be identified by flow cytometry (Phan et al., 2003; Paus et al., 2006).

The development of HEL-binding, CD45.1 EFPBs and GC B cells was quantified 4.5 d after transfer. The gating strategies are shown in Fig. 3; GC B cells derived from transferred SW_HEL cells are CD45.1 B220⁺ GL-7^hi Fas^hi and HEL binding (Fig. 3 A). In the absence of Bcl-6–expressing T cells, the development of antigen-specific GC B cells was reduced to ∼10% of the numbers found in the presence of Bcl-6–expressing T cells (Fig. 3 B), which is consistent with the previously described lack of functional Tfh cells in these chimeric mice (Yu et al., 2009).

The EFPB response derived from donor SW_HEL B cells can be measured by enumerating CD45.1⁺ B220^lo intracellular HEL-binding^hi cells (Fig. 3 C; Phan et al., 2003; Paus et al., 2006). In the absence of Bcl-6–expressing T cells, the percentage of EFPBs was reduced to ∼5% of that produced in the presence of Bcl-6–expressing T cells (Fig. 3 D). On day 5 of the SW_HEL B cell response to HEL-SRBC, anti-HEL antibodies are produced exclusively by EFPBs (Paus et al., 2006). Quantification of serum anti-HEL IgM and IgG1 (the predominant isotype found in SW_HEL B cell responses to HEL-SRBC; Phan et al., 2003) antibodies revealed negligible production of anti-HEL IgM in the mice lacking Bcl-6–expressing T cells (0.18% of the titer found in mice with Bcl-6–expressing T cells; Fig. 3 E) with no detectable anti-HEL IgG1 antibodies. These results demonstrate a requirement of Bcl-6–expressing T cells for the production of switched and unswitched EFPBs in response to a TD antigen.

IL-21 is produced by Tfh cells and acts directly on B cells to maximize Bcl-6 expression and promote GC B cell growth and survival (Linterman et al., 2010; Zotos et al., 2010). IL-21 also promotes Blimp-1 expression and plasma cell formation (Ozaki et al., 2004). We compared side by side the effects of IL-21 in the course of a SW_HEL B cell–derived GC and EFPB response.

SW_HEL B cells sufficient or deficient for the IL-21R (Il21r^−/−) were transferred to CD45.1 congenic C57BL/6 mice, which were immunized with HEL^2x-SRBC (Fig. 4 A). HEL^2x is a mutant HEL protein that binds the SW_HEL BCR with lower affinity. In the absence of IL-21R signaling, the production of GC B cells was reduced to ∼12% of the levels seen with IL-21R–sufficient donor cells at day 4.5 (Fig. 4 B), which is consistent with the described dependency of GC B cells on IL-21. The development of EFPBs was also impaired in the absence of IL-21R signaling, with only around 10% of the cell numbers seen with IL-21R–sufficient donor cells at day 4.5 (Fig. 4 C). In the absence of IL-21–mediated signaling, the extrafollicular response peaked 12 h later, on day 5. At this time point, there was a fourfold reduction in the number of EFPBs in the absence of IL-21R (Fig. 4 C). The GC and EFPB responses to SRBCs elicited by the host CD45.1 Il21r^+/+ B cells were comparable in both groups of adoptively transferred mice (Fig. S2, A and B).

The effects of IL-21 deficiency were apparent after 4 d in the course of both follicular and extrafollicular antibody responses (Fig. 4, A–C), suggesting that IL-21 produced by T cells acts at the stage of T–B interaction to enhance and accelerate B cell activation before their differentiation into either GC cells or extrafollicular plasma cells. IL-21 has recently been shown to promote GC B cell proliferation and survival, at least in part through maximizing Bcl-6 expression in B cells (Linterman et al., 2010; Zotos et al., 2010). Our findings extend this effect of IL-21 to extrafollicular antibody responses and are consistent both with reports that IL-21 also promotes Blimp-1 expression and plasma cells accumulate when overexpressed (Ozaki et al., 2004) and with the paucity of extrafollicular plasma cells 7 d after NP-KLH immunization in IL-21R and IL-21 knockout mice (Zotos et al., 2010).

We next sought to examine whether the requirement for Bcl-6 expression in T cells also occurs in the context of an extrafollicular antibody response to S. enterica infection, which has been shown to occur independently of IL-21 (Linterman et al., 2010). The immune response to infection with S. enterica induces a potent extrafollicular plasma cell response, resulting in the production of T-independent IgM antibodies and T-dependent class switching of B cells to produce IgG2a (IgG2c in C57BL/6 mice) and IgG2b antibodies (Cunningham et al., 2004; Linterman et al., 2009a). T dependency has been previously demonstrated by the absence of S. enterica–specific switched antibody responses in mice lacking CD28 (Cunningham et al., 2004; Linterman et al., 2009a). The GC response is considerably delayed, developing between days 20 and 35 after infection (Cunningham et al., 2007). It has also been established that IFN-γ–producing Th1 cells are responsible for bacterial clearance, starting on week 2 after immunization (Hess et al., 1996). Thus, S. enterica infection is an ideal model in which to investigate whether Bcl-6–expressing T cells rather than conventional Th1 cells are required to drive the extrafollicular plasma cell response without the potential interference of GC B cell–derived antibodies or GC Tfh cells.

Bcl6^+/+ CD45.2:Cd28^−/− CD45.1 or Bcl6^−/− CD45.2:Cd28^−/− CD45.1 mixed fetal liver chimeric mice were infected i.p. with S. enterica. As shown previously (Cunningham et al., 2007), this immunization did not increase GC cells above background 7 d after infection (Fig. 5, A and B). At this same time point, anti–S. enterica IgM and IgG2c titers were measured by ELISA. Bcl-6 deficiency in T cells did not reduce the production of specific IgM (Fig. 5 C) but significantly reduced the titer of anti–S. enterica IgG2c antibodies (Fig. 5 D). This mirrored the requirement of CD28 for specific IgG2c but not IgM production in a cohort of C57BL/6 Cd28^−/− mice (Fig. 5, E and F). Together, these results suggest that switched antibody responses to S. enterica require help from T cells in a Bcl-6–dependent manner.

Our data so far suggest that Bcl-6–expressing T cells rather than conventional Th1 cells are required for extrafollicular antibody responses. To confirm the presence of functional S. enterica–specific Th1 cells in mice lacking Bcl-6 expression in T cells, we investigated bacterial clearance, which has been shown to be critically dependent on the presence of functional Th1 cells (Hess et al., 1996). Although bacterial counts were increased by nearly 2 logs in nonchimeric Cd28^−/− mice that lack Th1 cells (Fig. 5 H), lack of Bcl-6 expression in T cells did not alter bacterial load in the livers of chimeric mice 12 d after infection (Fig. 5 G).

Next, we investigated when and where Bcl-6⁺ T cells appear during the response to S. enterica. A population of CD3⁺ Bcl-6⁺ PD-1^−/lo cells was detected at that outer T zone on day 4 after infection (Fig. 6, A and C), with total numbers increasing ∼10-fold above background. This is consistent with our previous finding of a 20-fold increase in numbers of IgG2c switched plasma cells as early as day 4 after S. enterica infection (Cunningham et al., 2007). At this time point (day 4), CD3⁺ Bcl-6⁺ PD-1^hi cells were absent (Fig. 6, A and C), but they became visible 35 d after immunization, when S. enterica–induced GCs appear, and they were located within the GC (Fig. 6 B). Together, these results highlight a necessary role for T cell–expressed Bcl-6 soon after infection to promote switching. Given that the IgG2c response occurs normally in the absence of IL-21 but requires IFN-γ, it is likely that early Th1 cells producing IFN-γ up-regulate Bcl-6 that facilitates migration to the T–B border and thus cognate interaction with antigen-primed B cells.

The expression of PD-1 and CXCR5 was evaluated on effector (CD4⁺ CD44^hi) cells. On day 7, at which time S. enterica fails to induce GCs regardless of Bcl-6 expression, PD-1^hi CXCR5^hi GC Tfh cells (Fig. 6 D, red gate) that were seen at background levels in PBS-immunized chimeras (Fig. 6 D, top left; and Fig. S4 C) were low or absent after S. enterica immunization in all chimeras. Immunized chimeras lacking Bcl-6 expression in T cells selectively lacked a PD-1^lo CXCR5⁺ population (Fig. 6 D, blue gate; and Fig. S4 B), which is likely to correlate with those cells shown in Fig. 1 A that localize to the T–B border and prime B cells. Non-Tfh effectors identified as CD4⁺ CD44^hi CXCR5⁻ PD-1⁻ (Fig. 6 D, green gate; and Fig. S4 A) were equally represented in both sets of chimeras, irrespective of T cell expression of Bcl-6. Together, these data suggest that CD4⁺ CD44^hi CXCR5⁺ Bcl-6⁺ PD-1^lo cells may be important for the initiation of an extrafollicular antibody response.

A potential confounding factor in the aforementioned experiments is the increased proportion of activated Bcl6^−/− CD44^hi CD4⁺ cells in mixed chimeras reconstituted with Bcl6^−/− fetal liver (Fig. S3). To overcome this issue, we performed adoptive transfers of 10⁵ naive CD45.2 OT-II cells that express a transgenic TCR specific for OVA peptide 323–339 and were either deficient or sufficient in Bcl-6 into CD45.1 Cd28^−/− mice. Recipient mice also received 10⁵ SW_HEL B cells and were immunized with HEL-OVA conjugates. We first compared the phenotype of Bcl-6–sufficient effector (CD44^hi) OT-II cells at the B cell priming stage (pre-GC Tfh, day 2) and the stage in which GC Tfh cells are readily visible (day 4; Fig. 7, A–C). Concordant with the immunohistochemistry data from SRBC-immunized mice (Fig. 1 B), OT-II T cells had up-regulated PD-1 on day 2, but the levels were approximately threefold lower than those of day 4 OT-II cells from immunized recipient mice analyzed on the same day (Fig. 7 B). CXCR5 expression did not increase significantly from day 2 to 4 (Fig. 7 B).

On day 5.5 after immunization, spleens were harvested, and distinct HEL-specific GC B cells and extrafollicular plasma cells were identified by flow cytometry on the basis of size and differential expression of B220, CXCR5, and HEL binding (Fig. 7 C). Recipients of Bcl6^−/− OT-II cells showed a parallel and comparable reduction in both GCs and EFPBs (Fig. 7, D–F). Staining with PD-1 and CXCR5 demonstrated the absence of PD-1^hi CXCR5^hi Tfh cells among OT-II Bcl6^−/− T cells on day 5.5 (Fig. 7, G and H). These results confirm that T cell–expressed Bcl-6 plays a role in B cells differentiating along the extrafollicular pathway. GL-7 has been recently described as a marker of GC Tfh cells (Yusuf et al., 2010). The appearance of GL-7^hi CD4⁺ cells among OT-II cells lacking Bcl-6 expression (these cells expressed low amounts of PD-1 and CXCR5; Fig. S5) suggests that GL-7^hi cells are a heterogeneous population comprised of activated CD4⁺ cells other than GC Tfh cells.

We observed significantly fewer switched EFPBs in the absence of T cell–expressed Bcl-6 across all experimental systems used in this study. A previous report described normal antibody titers in the first 2 wk after immunization of chimeric mice lacking Bcl-6 in T and B cells (Toyama et al., 2002). It is possible that the requirement for Bcl-6⁺ T cells may be overcome by signals from other cells; B cells differentiating along the extrafollicular route are also known to be boosted by IL-6 and/or APRIL (a proliferation-inducing ligand)-producing cells that colocalize with EFPBs (Mohr et al., 2009). It is also conceivable that the plasma cells and antibodies detected on day 14 in mice lacking Bcl-6⁺ T cells have arisen from an alternative route of differentiation. Indeed, several groups have reported a third B cell differentiation route into early memory B cells that is GC⁻, ICOS⁻, Bcl-6⁻, and IL-21⁻ independent (Toyama et al., 2002; Inamine et al., 2005; Chan et al., 2009; Linterman et al., 2010; Zotos et al., 2010).

Our data establish that Bcl-6 expression enables T cells to provide effective help to B cells for both follicular and extrafollicular antibody responses to protein antigens. PD-1 expression distinguishes outer T zone PD-1^lo/inti Bcl-6⁺ pre-GC Tfh cells from the PD-1^hi Bcl-6⁺ Tfh cells that appear within the GC (GC Tfh cells). Bcl-6 has been shown to be critical for up-regulation of CXCR5 and down-regulation of CCR7 (Johnston et al., 2009; Nurieva et al., 2009; Yu et al., 2009), which are required for T and B cell localization to the T–B border (Cyster et al., 2000; Haynes et al., 2007). Thus, expression of Bcl-6 appears important to facilitate the initial T and B cell encounter at this location that results in the provision of T cell help required to initiate T-dependent B cell differentiation and Ig isotype switching. Indeed and contrary to common belief, there is evidence that production of >99% of switch transcripts occurs at this time and therefore precedes GC formation (Toellner et al., 1996; Pape et al., 2003). In agreement with this, we have shown here that in the absence of Bcl-6 expression by CD4⁺ T cells, the switched antibody responses to all antigens are completely abrogated, including the extrafollicular IgG2a response to S. enterica, against which only antigen-specific IgM, of probable T-independent origin, is maintained.

Our finding that T cell–expressed Bcl-6 acts at the time of T–B interaction does not exclude the possibility that Bcl-6 expression in T cells may also be required at the stage of plasmablast growth in extrafollicular foci. CXCR5^lo CXCR4^hi CD4⁺ T cells have been found in splenic extrafollicular foci of autoimmune MRL/Fas^lpr mice (Odegard et al., 2008), and these are also Bcl-6 dependent (Poholek et al., 2010). This work suggests that Tfh are even more heterogeneous than we thought. Bcl-6⁺ T cells in the outer T zones appear competent to help B cells differentiate along either the follicular or extrafollicular pathway. This highlights the need for refining the nomenclature of B follicular helper T cells. As has been previously proposed (McHeyzer-Williams et al., 2009), pre-GC Tfh cells can be used to refer to T cells that prime B cells at the T–B border, which are shown here to be PD-1^lo/int Bcl-6⁺, whereas GC Tfh cells can be used to refer to those CD4⁺ T cells located within GCs, which are generally, albeit not universally, PD-1^hi.

C57BL/6, Cd28^−/− mice and Bcl6^−/− mice backcrossed for 14 generations into the C57BL/6 genetic background were maintained by the Australian National University Bioscience Services in specific pathogen–free conditions and had access to food and water ad libitum. SW_HEL CD45.1 mice, which carry a Vκ10κ light chain transgene and a knocked in VH10 Ig heavy chain in place of the JH segments of the endogenous IgH gene that encode a high-affinity antibody for HEL, were obtained from the laboratory of R. Brink (Garvan Institute, Sydney, New South Wales, Australia) and maintained by the Australian National University Bioscience Services. SW_HEL Il21r^+/+ CD45.2, SW_HEL Il21r^−/− CD45.2, and recipient C57BL/6 CD45.1 mice were maintained in the animal facility at the Garvan Institute. For experiments involving SW_HEL mice, 10⁵ SW_HEL B cells were transferred into recipients, which were simultaneously immunized i.v. with 2 × 10⁸ SRBCs conjugated to HEL or HEL^2x protein (Paus et al., 2006). For experiments involving NP-KLH, immunizations with hapten 4(hydroxy-3-nitrophenyl) acetyl (NP) coupled to keyhole limpet hemocyanin (KLH) at a molar ratio of ∼16:1 were performed. 100 µg NP-KLH was precipitated in alum and injected i.p. HEL-OVA immunizations are described in the OT-II adoptive transfer section. All procedures were performed with appropriate ethical and legal approval by the Australian National University’s Animal Experimentation Ethics Committee.

Recipient Rag1^−/− mice were irradiated with 500 Rad and reconstituted via i.v. injection with 2 × 10⁶ donor fetal liver–derived hematopoietic stem cells. Mice were reconstituted with the following combinations of fetal liver cells: for S. enterica infection and NP-KLH immunization experiments, 80% Bcl6^−/− CD45.2:20% Cd28^−/− CD45.1 and control 80% Bcl6^+/+ CD45.2:20% Cd28^−/− CD45.1 fetal liver cells; and for recipients of CD45.1 SW_HEL B cells, 80% Bcl6^−/− CD45.2:20% Cd28^−/− CD45.2 and control 80% Bcl6^+/+ CD45.2:20% Cd28^−/− CD45.2 fetal liver cells. All groups were maintained on antibiotics for 6 wk, and experiments were performed 8 wk after reconstitution to allow for full reconstitution of the immune system.

S. enterica serovar Dublin strain SL5631 (Segall and Lindberg, 1991) was grown in Luria-Bertani medium overnight. Mice were inoculated with 10⁶ CFUs from a log-phase culture administered i.p. in PBS, with control mice receiving PBS only. Liver bacterial load was measured at day 12 after infection by homogenizing organs, plating serial dilutions in PBS onto Luria-Bertani agar, and incubating at 37°C overnight. For experiments with S. enterica serovar Typhimurium strain SL3261, mice were immunized i.p. with 5 × 10⁵ bacteria in PBS.

ELISAs were used to quantify serum titers of anti-HEL, anti-NP, and anti–S. enterica antibodies. To quantify anti-HEL and anti–S. enterica antibodies, 96-well plates were coated with either 1 µg/ml HEL protein or 250 µg/ml S. enterica lysate prepared as previously described (Valentine et al., 1998) in 0.05 M carbonate buffer (Na₂CO₃ + NaHCO₃, pH 9.6) overnight and subsequently blocked with PBS/1% BSA. Serum was serially diluted onto the 96-well plates in PBS/1% BSA/0.05% Tween 20 buffer and incubated for 2 h at 37°C. After washing, antigen-specific antibodies were detected with anti–mouse IgM, anti–mouse IgG1, or anti–mouse IgG2a^b antibodies conjugated to alkaline phosphatase (Vector Laboratories). Bound antibody was detected with para-nitrophenylphosphate in glycine buffer. Plates were read at 405 nm with a microplate reader (Thermomax; Molecular Devices), and titers for serum samples were calculated as the log serum concentration required to achieve 50% maximum optical density. To detect anti-NP antibodies, 96-well plates were coated with 20 µg/ml NP₁₃-BSA in PBS overnight. Serum samples were serially diluted on plates in block solution (PBS/1% FCS/0.6% skim milk powder/0.05% Tween 20) and incubated for at least 20 h at room temperature. Anti-NP was shown with goat anti–mouse IgG1 conjugated to horseradish peroxidase (HRP; SouthernBiotech) and visualized with ABTS substrate (2’2-Azinobis [3-ethylbenzthiazoline sulfonic acid] diammonium salt; Sigma-Aldrich). Plates were read at 405 nm, with reference wavelength 490 nm (Molecular Devices).

For collaborative responses of SW_HEL B cells and OT-II T cells, 10⁵ sorted CD45.2 SW_HEL B cells and 20 µg HEL-OVA_323–339 peptide (Chan et al., 2009) were injected i.v. into Cd28^−/− CD45.1 mice. 4 h later, 10⁵ sorted naive CD44^lo CD25⁻ CD4⁺ CD45.2 OT-II T cells were injected i.v. into the same recipient mice, and mice were immunized i.p. with 100 µg OVA (Sigma-Aldrich) in alum (Thermo Fisher Scientific) immediately after adoptive cell transfer.

To identify specific cell subsets, single cell suspension of splenocytes in ice-cold FACS buffer (2% FCS and 0.1% NaN₃ in PBS) were stained with the antibodies detailed below, all from BD unless specified, with thorough washing between stain layers. Flow cytometry data were acquired on a flow cytometer (LSRII; BD) running FACSDiva and analyzed with FlowJo version 8 (Tree Star). NP-binding cells were identified by staining with IgM Alexa Fluor 580 (conjugated in house; Invitrogen), NP-APC (conjugated in house; Smith et al., 2000), CD138-PE, PNA-FITC (Vector Laboratories), B220-PE Cy7, and CD45.1–Pacific blue. HEL-binding GC B cells were identified by first staining cells with 200 ng/ml HEL solution, followed by GL-7–FITC, CD95-PE, CD45.2-PerCP Cy5.5, anti-HEL HyHEL9–Alexa Fluor 647 (conjugated in house with an Alexa Fluor 647 antibody labeling kit; Invitrogen), B220-APC Cy7, and CD45.1–Pacific blue (BioLegend). HEL⁺ EFPBs were identified by treating spleens with collagenase and DNase; spleens were cut into ∼1-mm³ pieces and incubated at 37°C for 30 min with 1 ml of 1 mg/ml type II collagenase and 0.1% DNase in RPMI/10% FCS. After washing in FACS buffer as above, cells were stained with CD45.2-PerCP Cy5.5, B220-APC Cy7, and CD45.1–Pacific blue, permeabilized using the Cytofix/Cytoperm kit (BD) according to the manufacturer’s instructions, and stained for intracellular HEL binding and IgG1 with HEL protein conjugated to Alexa Fluor 647 as above and anti–mouse IgG1-FITC, respectively. The presence of GC B cells in S. enterica–infected and uninfected mice was identified by staining cells with GL-7–FITC, CD95-PE, and B220-APC Cy7. T cell phenotypes were investigated using combinations of the following antibodies: CXCR5-biotin + streptavidin-PE Cy7, PD-1 PE, IL-21–APC (R&D Systems), CD4-APC Cy7, B220-FITC, CD44–Pacific blue (BioLegend), and CD25-PerCP Cy5.5.

NP-binding cells were visualized by freezing spleen samples in OCT (Tissue-Tek; Sakura). Spleen sections were prepared and stained as described previously (Toellner et al., 1996). Antibodies and conjugates used were B220-biotin (RA3-6B2), goat anti–mouse IgG1 (SouthernBiotech; HRP), and streptavidin alkaline phosphatase (SouthernBiotech). Antibodies were visualized with substrates AEC (Vector Laboratories) and FastBlue (Vector Laboratories).

To visualize the responses to S. enterica and SRBC immunization, spleen samples were fixed for 20 min in ice-cold acetone (for S. enterica) or 4% paraformaldehyde for 2 h on ice, incubated in six changes of sucrose buffer overnight (for SRBCs), and embedded in Tissue-Tek OCT compound. Sections were blocked with streptavidin and biotin blocking kit (Vector Laboratories) before staining. Sections were stained for CD3 using anti-CD3 (BD) followed by anti–Armenian hamster Cy5 (Jackson ImmunoResearch Laboratories, Inc.). For Bcl-6, anti–mouse Bcl-6 (Santa Cruz Biotechnology, Inc.) was followed by donkey anti–rabbit FITC (Jackson ImmunoResearch Laboratories, Inc.) and then Alexa Fluor 488–conjugated goat anti-FITC (Invitrogen). PD-1 was stained with purified anti–PD-1 (BioLegend) followed by anti–rat Cy3 (Jackson ImmunoResearch Laboratories, Inc.). For AID, anti–mouse AID (eBioscience) was followed by biotinylated anti-rat IgG, and then streptavidin-HRP conjugates (Zymed) were followed by TSATM tetramethylrhodamine (PerkinElmer). Stained sections were mounted in Fluoromount-G (SouthernBiotech) and analyzed with a confocal laser-scanning microscope (DMRXA2; Leica; SRBC experiments) or a laser-scanning confocal microscope (LSM510; Carl Zeiss) with a microscope (AxioVert 100M; Carl Zeiss) using a 40× objective. The latter images were subsequently analyzed on the LSM Image Browser (Carl Zeiss).

Quantification of Bcl-6⁺ CD3⁺ PD-1^hi/lo cells was performed as follows. 10 T zone areas were randomly selected. For each image, the number of CD3⁺ Bcl-6⁺ cells and their PD-1 status (bright/intermediate or negative) was quantified, and the total number of cells in each spleen was calculated as described previously (Cunningham et al., 2007). Cell numbers and densities were estimated using point counting, and counts were adjusted for the different sizes of the spleens seen throughout the study by multiplying the cells/cubic millimeter by the mass of the spleen.

Two-way analysis of variance was used to interrogate whether IL-21R signaling was a significant contributor to variance over a time course (Fig. 4, B and C). For other experiments, the nonparametric Mann-Whitney test (U test) was used.

Fig. S1 demonstrates that splenic switched plasma cells are reduced in mice lacking Bcl-6 expression in T cells in fetal liver chimeric mice immunized 7 d earlier with NP-KLH. Fig. S2 shows host-derived GC and EFPB responses 5 d after SRBC immunization. Fig. S3 shows that there is an increase in the proportion of effector cells among Bcl6^−/− T cells from mixed chimeric mice. Fig. S4 shows that there is a decrease in the numbers of PD-1^lo CD4⁺ cells after S. enterica infection. Fig. S5 shows that PD-1^hi cells but not GL-7^hi cells fail to form in the absence of OT-II cell–expressed Bcl-6 after OVA immunization.

We thank Naresh Verma for help growing S. enterica, Jennifer Hogan for technical assistance, Michelle Townsend for tissue sectioning, and Matthew Cook for critical reading of the manuscript.

This work was funded by a National Health and Medical Research Council (NHMRC) program grant to C.G. Vinuesa and R. Brink and an NHMRC project grant to C.G. Vinuesa. C.G. Vinuesa is supported by a Viertel Senior Medical Research Fellowship. A.F. Cunningham is the recipient of a UK Medical Research Council New Investigator award.

The authors have no conflicting financial interests.