Transcription factor IRF4 determines germinal center formation through follicular T-helper cell differentiation

Abstract

Follicular T-helper (T_FH) cells cooperate with GL7⁺CD95⁺ germinal center (GC) B cells to induce antibody maturation. Herein, we identify the transcription factor IRF4 as a T-cell intrinsic precondition for T_FH cell differentiation and GC formation. After immunization with protein or infection with the protozoon Leishmania major, draining lymph nodes (LNs) of IFN-regulatory factor-4 (Irf4^−/−) mice lacked GCs and GC B cells despite developing normal initial hyperplasia. GCs were also absent in Peyer’s patches of naive Irf4^−/− mice. Accordingly, CD4⁺ T cells within the LNs and Peyer’s patches failed to express the T_FH key transcription factor B-cell lymphoma-6 and other T_FH-related molecules. During chronic leishmaniasis, the draining Irf4^−/− LNs disappeared because of massive cell death. Adoptive transfer of WT CD4⁺ T cells or few L. major primed WT T_FH cells reconstituted GC formation, GC B-cell differentiation, and LN cell survival. In support of a T-cell intrinsic IRF4 activity, Irf4^−/− T_FH cell differentiation was not rescued by close neighborhood to transferred WT T_FH cells. Together with its known B lineage-specific roles during plasma cell maturation and class switch, our study places IRF4 in the center of antibody production toward T-cell–dependent antigens.

Apart from Th1 and Th2, the family of Th subsets now includes Th17 and Th9 (1). In addition, follicular T-helper (T_FH) cells are defined based on their location within germinal center cells (GCs) of lymphoid organs (2, 3). Here, these cells produce cytokines that normally define other subsets, such as the Th2 product IL-4 (4, 5) or the Th17 product IL-21 (6), which is involved in GC B-cell generation (7–10). T_FH cell propagation is supported by the transcription factor B-cell lymphoma (BCL)-6 and suppressed by Blimp1 (11–13). Further markers used to define T_FH cells include inducible costimulator (ICOS), programmed death-1 (PD-1), and CXC-chemokine receptor 5 (CXCR5), which mediate their migration into GCs (2, 14–16).

The IFN-regulatory factor (IRF) family of transcription factors includes nine members in mammals that bind to related target-gene sequences (17). We and others have described important roles of IRFs during Th cell differentiation. In particular, IRF1 is decisive for Th1 cell generation because it is ubiquitously expressed and redundantly addresses many genes with independent Th1-supporting function (17–19). In contrast, IRF4 controls Th2 and Th17 cell differentiation (20–23), with ensuing total resistance of Irf4^−/− mice in a Th17-dependent mouse model of multiple sclerosis (24). In addition, regulatory T-cell (Treg)-specific IRF4 deficiency or lack of IRF4 binding protein lead to a generalized autoimmune syndrome (25, 26). Finally, we reported on the role of IRF4 during Th9 differentiation (27). Remarkably, IRF4 is also a B-cell intrinsic prerequisite for class switch and plasma cell maturation (28, 29).

Given these pleiotropic activities of IRF4 on B and T cells, we wondered whether IRF4 also contributes to the interaction of T_FH and GC B cells. Herein, we use chronic leishmaniasis, a model infection with prominent T- and B-cell interactions (30) to prove a decisive T-cell intrinsic role of IRF4 for murine T_FH cell development.

Results

Irf4^−/− Mice Fail to Generate GCs.

To study the development of T_FH cells in vivo, we infected Irf4^−/− mice and Irf4-competent control mice with Leishmania major (30). Two weeks later, draining popliteal lymph nodes (LNs) were analyzed (Figs. 1 and 2). By immunohistology, prominent GC formation was observed in WT and Irf4^+/− LNs, including presence of GL7⁺ GC cells (Fig 1A). In contrast, GCs were totally absent in Irf4^−/− LNs and few GL7⁺ cells were dispersed throughout the LN. However, Irf4^−/− LNs did contain normal B and CD4 T-cell areas (Fig. 1A, Bottom Right). We confirmed the lack of GCs in Irf4^−/− mice that were immunized with the myelin oligodendrocyte glycoprotein (MOG) peptide instead of L. major infection (Fig. S1) and in Peyer’s patches (PP) from naive mice (Fig. 3A).

Fig. 1.

Lack of GC formation in Irf4^−/− mice. Mice of the indicated genotypes were infected with L. major. Two weeks later, popliteal LNs were prepared. (A) Tissue sections were stained for CD4 (red), GL7 (green, Left), or B220 (green, Right), and analyzed by fluorescence-microscopy (20× magnification). (Scale bars, 50 μm.) (B and C) LN cell suspensions were stained for the indicated surface molecules and analyzed by flow cytometry. Numbers indicate percentages of positive cells in the respective circles or rectangles. Panels are from one representative mouse per group. Three different experiments, each with two or three mice per group. (B) B220⁺ B-cell gate. Percentages (mean ± SD) of GL7⁺CD95⁺ B cells compiled from all mice tested.

Fig. 2.

Lack of T_FH cells in L. major infected Irf4^−/− mice. Irf4^−/−, WT mice, and Irf4^+/− mice (three per group) were infected with L. major and their popliteal LN cells analyzed 2 wk later for expression of extracellular ICOS, CXCR5, and PD-1 (A) and intracellular BCL-6 (B) and IL-21 (C). (A; B; C, Right; D; and E) Direct ex vivo analysis. (C) Analysis after restimulation for 4 h with PMA and ionomycin (Left). (Right) CD4⁺ cells from pooled LN cells of all mice per group sorted according to ICOS expression (see A), and analyzed for IL-21 compared with HPRT (hypoxanthine-guanine phosphoribosyl transferase) expression by RT qPCR. Bars denote the SD of duplicate qPCR determinations of each sample. The asterisk signifies that this value was set to one. n.t., not tested. (D and E) Single CD4⁺ cells were gated similar to gate B of Fig. S2B. Numbers in A–E indicate percentages of cells in the respective rectangles. Data are from one representative mouse per group. Three experiments with similar outcomes.

Fig. 3.

Lack of T_FH cell differentiation in PP of naive Irf4^−/− mice. (A–C) PP of naive mice were analyzed as described for Figs. 1 and 2.

A strong reduction in GC B cells coexpressing the markers GL7 and CD95 (31, 32) was verified by flow-cytometry of single LN cell suspensions from L. major infected Irf4^−/− mice (Fig. 1B, note the compiled data of three different experiments). Again, the finding was reproduced in PP from naive mice (Fig. 3B). In control FACS-stainings, GL7 was only weakly expressed on WT and Irf4^−/− CD4⁺ cells. Thus, Irf4^−/− mice form the architecture of normal LNs, but lack GC formation. Furthermore, the ICOS ligand (ICOSL) molecule was strongly up-regulated on Irf4^−/− B compared with WT B cells (Fig. 1C), possibly because of a feedback-loop between ICOSL and its partner ICOS (33) expressed on T_FH cells. Together with missing GC formation, these findings suggest a defect in T_FH cells within Irf4^−/− mice.

Irf4^−/− Mice Fail to Generate T_FH Cells.

To directly test this theory, LN cells of L. major infected mice were analyzed for expression of T_FH marker molecules. In Irf4^+/− mice, T_FH cells expressing BCL-6, IL-21, and PD-1 were present and coexpressed ICOS at high (ICOS^hi) or intermediate (ICOS^int) levels. Importantly, Irf4^−/− CD4⁺ cells totally lacked ICOS^hi cells, although ICOS^int cells were present at even enhanced frequency (Fig. 2A). Both findings were confirmed in PP of naive mice (Fig. 3C). As for PD-1, its expression was comparable in CD4⁺ cells of infected Irf4^−/− and Irf4^+/− mice (Fig. 2A), but was also reduced in Irf4^−/− CD4⁺ cells of naive PP (Fig. 3C). Furthermore, amounts of BCL-6 protein, a central molecule for T_FH cell function (11–13), were considerably lower in Irf4^−/− than in Irf4^+/− CD4⁺ cells (Fig. 2B).

An even stronger defect was noted with respect to IL-21 production (Fig. 2C): after short-term restimulation, IL-21 protein was synthesized by many of the ICOS^hi and ICOS^int Irf4^+/− control cells (thus further characterizing them as the source of T_FH cells), but not by Irf4^−/− CD4⁺ cells (Fig. 2C, Left). To measure expression of Il-21 at the mRNA level, we performed quantitative PCR (qPCR) directly ex vivo (Fig. 2C, Right) after sorting ICOS^hi (only control mice), ICOS^int, and ICOS^lo cell populations (Fig. 2A), Again, most Il-21 mRNA was detected in ICOS^hi Irf4^+/− control cells, but the lower levels in ICOS^int cells were further reduced in their Irf4^−/− counterparts. These data demonstrate a striking defect of Irf4^−/− CD4⁺ cells to express T_FH cell markers.

Analysis of CXCR5 Expression.

Expression of the CXCR5 molecule permits T_FH cell migration into GCs (16), but is found at even higher levels in B cells (15). Recently T–B conjugates have been described in FACS analyses of LN cell preparations (5). These conjugates might contain T_FH cells tightly interacting with B cells and complicate testing of CXCR5 expression on T cells. Indeed, we identified CD4⁺ events with considerable CXCR5 costaining (Fig. S2A) as aggregates of CD4⁺ T and B220⁺ B cells (Fig. S2B). In contrast, CXCR5-staining was weaker on single CD4+ T cells, despite nice staining of single B cells within the same sample (Fig. S2B). Mechanical dissociation and reanalysis confirmed lower CXCR5 staining on T cells. Thus, outgating of B–T conjugates did not remove any T_FH cells with particularly strong CXCR5 staining, and was routinely used during data acquisition (with the exception of Figs. S2 and S3). We detected B–T conjugates not only in the Leishmania model, but also in PP of naive mice (Figs. S2 and S3) or after MOG immunization. Although CXCR5 clearly remains a marker of T_FH cells, these T–B conjugates suggest critical care during its staining on T cells.

When we now compared CXCR5 expression on Irf4^−/− and WT cells, we found that CXCR5 staining was totally absent in Irf4^−/− CD4⁺ cells (Fig. 2D), confirming their lack of T_FH cells. Importantly, this conclusion required exclusion of B–T conjugates from the analysis (Fig. S3, gate B), because gating on all viable CD4⁺ events revealed CD4⁺CXCR5⁺ events (Fig. S3A) as a result of aggregating CXCR5-expressing Irf4^−/− B cells (Fig. S3B, gate A). Mechanical disruption of the conjugates and reanalysis confirmed these conclusions (Fig. S3C) and excluded removal of CXCR5⁺ Irf4^−/− T_FH cells. As anticipated, we found a positive correlation for expression of CXCR5 and BCL-6 in Irf4^+/− mice (Fig. 2E). Similar data were gained from PP of Irf4^−/− mice (Fig. 3C, and Fig. S3 D and E). Taken together, these data strongly suggest that Irf4^−/− T cells do not differentiate into T_FH cells in vivo, but CXCR5 deficiency may explain lack of GC formation in Irf4^−/− mice because of altered T-cell migration.

To rule out a defect of Irf4^−/− T cells in their receptor-triggered antigenic response as a trivial reason for T_FH cell deficiency, we restimulated LN cells of infected mice in vitro. In response to Leishmania antigens, Irf4^−/− and Irf4^+/− control CD4⁺ cells secreted the precursor T-cell product IL-2 into their supernatants at the same order of magnitude (Fig. S4A). Similarly, the frequency of intracellularly stained IL-2–producing CD4⁺ cells was similar between the two genotypes (Fig S4B).

Lack of Irf4^−/− T_FH Cells Is Not Caused by a Cell Viability Problem.

Previously, we reported on apoptotic death of Irf4^−/− draining LN cells after about 6 wk of leishmaniasis (34). We therefore aimed to exclude that a cell viability problem hindered GC formation earlier after infection. In naive young Irf4^−/− and Irf4^+/− mice, the size of popliteal LNs was comparable (about 1 × 10⁶ cells). Furthermore, the increase in LN size and histological appearance 2 wk after infection were similar (Fig. 4 A and B), but Irf4^−/− and Irf4^+/− CD4⁺ LN T cells were comparably able to secrete IL-2 (Fig. S4) and to proliferate in response to phorbol 12-myristate 13-acetate (PMA) and ionomycin (Fig. 4C). These data confirm functional integrity of Irf4^−/− LN CD4⁺ T cells at the time point when the lack of GC formation was noted. In contrast, Irf4^−/− LNs had almost totally disappeared 6 wk after infection (Fig. 4A), and severe damage in LN cell morphology was already visible 4 wk after infection (Fig. 4B). Cell death did not occur in other LNs of infected Irf4^−/− mice. The divergence in LN cell viability of WT and Irf4^−/− mice 6 but not 2 wk after infection is underscored by the compiled statistical significance of all mice tested (Fig. 4A). In conclusion, lack of Irf4^−/− GC formation cannot be explained by disturbed cell viability.

Fig. 4.

Disappearance of lesion-draining LNs in L. major infected Irf4^−/− mice. Mice of the indicated genotypes were infected with L. major. At the indicated time points, popliteal LNs were prepared and (A) cell numbers in single-cell suspensions counted (mean ± SD) or (B) tissue sections processed for HE staining and microscopy (40× magnification). (Scale bars, 10 μm.) (C) To control for proliferative capacity 2 wk after infection, cells were labeled with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE), stimulated for 72 h with PMA/ionomycin, and stained for CD4. Data are representative of five (A) or three (B and C) different experiments, each with three mice per group and (B and C) show the results of one representative mouse per group.

Rescue of GC Formation by WT CD4⁺ T Cells.

Although a primary Irf4^−/− T_FH cell defect was likely, a B-cell defect with altered T–B interactions and secondary T_FH cell deficiency remained possible. To directly prove a T-cell intrinsic T_FH promoting IRF4 activity, Irf4^−/− mice were reconstituted intraperitonially with purified CD45 (Ly5.1⁺) congenic CD4⁺ cells from naive mice at the start of infection. Two weeks later—that is, at the maximum of GC formation in WT mice—transferred WT CD4⁺ T cells had perfectly rescued the appearance of GCs and GL7⁺ Irf4^−/− B cells (Fig. 5A). Furthermore, the WT cells tended to accumulate close to the GC areas, but endogenous CD4⁺ T cells mostly remained outside of them. By flow cytometry, WT CD4⁺ T-cell transfer totally rescued the generation of Irf4^−/− GL7⁺CD95⁺ GC B cells (Fig. 5B) and the compensatory up-regulation of ICOSL on Irf4^−/− B cells was almost reverted (Fig. 5C). Because WT CD4⁺ cells can induce GC formation and GC B-cell differentiation in Irf4^−/− mice, their T_FH defect is caused by an intrinsic T- but not B-cell defect. Accordingly, B-cell–specific deficiency in IRF4 leads to disturbed plasma cell differentiation, but no change in GC formation (28). However, this plasma cell defect precludes demonstration of effects of the transferred WT CD4⁺ T cells on antibody formation in our mice.

Fig. 5.

Rescue of GC formation by WT CD4⁺ T cells. Irf4^−/− mice (A–C) or Irf4^+/− mice (B and C) were infected with L. major. Where indicated, Irf4^−/− mice received 8 × 10⁶ Ly5.1 congenic WT CD4⁺ cells by intraperitoneal adoptive transfer on the day of infection. Two weeks later, popliteal LNs were prepared. (A) Tissue sections were stained with antibodies to CD4 (blue), Ly5.1 (red), and GL7 (green) followed by fluorescence-microscopy (20× magnification). (Scale bars, 50 μm.) (B and C) Cells in suspension were stained for B220, GL7, ICOSL, and CD95 and analyzed (B220⁺ gate) by flow cytometry. Numbers refer to percentages in the quadrant below the number (C) or within the indicated gate (B). SSC, side scatter. Data are from one representative mouse per group. Three (A) or two (B and C) different experiments, each with three mice per group, were performed with similar outcome.

Rescue of LN Cell Survival by WT CD4⁺ T Cells.

Like GC formation, LN cell survival 6 wk after infection was rescued to a great extent by transferred CD4⁺ cells and normal histological morphology was regained (Fig. 6 A and B). Most of the surviving LN cells did not express the Ly5.1 marker of the transferred WT cells (Fig. 6C) and Ly5.1-expressing cells were all CD4⁺. Thus, LN cell viability was not secondary to outgrowth of WT CD4⁺ cells. Instead, WT CD4⁺ cells caused improved survival of endogenous Irf4^−/− B and T cells. Of note, WT CD4⁺ T cells were perfectly able to become ICOS^hi cells within Irf4^−/− LNs (Fig. 6C). Thus, the defect in Irf4^−/− T_FH cells is not caused by deficiencies of accessory cells (e.g., in antigen presentation or production of necessary cofactors). Furthermore, even the side-by-side presence of WT T_FH cells did not catalyze ICOS^hi expression, and thus T_FH cell differentiation in endogenous Ly5.1⁻ Irf4^−/− CD4⁺ T cells (Fig. 6C).

Fig. 6.

Rescue of LN cell survival by WT CD4⁺ cells. (A–C) Where indicated, Irf4^−/− mice received 8 × 10⁶ Ly5.1 congenic WT B or CD4⁺ cells or both by intraperitoneal transfer, were infected, and analyzed 6 wk later in comparison with infected WT or Irf4^+/− mice. (A) Popliteal LN cell numbers. (B) H&E staining (40× magnification) (Scale bars, 10 μm.) (C, Upper) total life cell gate. Numbers refer to the percentage of CD4⁺Ly5.1⁻ or CD4⁺Ly5.1⁺ among total cells. (Lower) Relative cell percentages among CD4⁺Ly5.1⁺ cells (Upper numbers) or CD4⁺Ly5.1⁻ cells (Lower) within the respective rectangles. (D and E) T_FH cells (CD4⁺ICOS^hiCXCR5⁺) or control CD4⁺/ICOS⁻/CXCR5⁻ cells sorted (Fig. S6) from 2 wk-infected Ly5.1⁺ WT mice were transferred intraperitoneally (2 × 10⁵ per mouse) into Irf4^−/− mice which, together with control mice, were infected with L. major. After 6 wk, cell numbers of the draining LNs were counted (D) and the proliferative capacity of the cells determined by CFSE dilution (E). Data are representative of three (A–C) or two (D and E) different experiments, each with three (A–C) or two (D and E) mice per group. (D) Accumulated data ± SD of all mice per group in the two experiments or data from one representative mouse (E).

Unlike T cells, transferred WT B cells neither protected LNs from cell death nor modified the effect of WT CD4⁺ cells (Fig. 6A). As a positive control for their functionality, we took advantage of the unrelated Irf4^−/− B-cell defect (28, 35), leading to drastically reduced serum IgM levels. Analysis of mouse sera after B-cell transfer revealed that the transferred WT B cells raised IgM in Irf4^−/− mice almost to WT amounts (Fig. S5A).

Rescue of LN survival mirrored susceptibility toward L. major: transferred WT CD4⁺, but not B cells, led to a healing phenotype, as measured by lesion size and parasite burden and compared with control Irf4^+/− mice (Fig. S5 B and C). However, a high parasite load is probably not the main reason for ensuing cell death in Irf4^−/− LN cells, because susceptible BALB/c or IRF1-deficient mice contain even more Leishmania (18), but keep the original size of the LN and develop a cheesy necrosis inside of it.

To formally link T_FH cells and protection from cell death, we transferred as few as 2 × 10⁵ FACS-sorted (Fig. S6) WT T_FH cells from LNs and spleens of L. major infected Ly5.1⁺ WT mice. For comparison, ICOS⁻CXCR5⁻ CD4⁺ T cells from the same organs were transferred into different Irf4^−/− mice. All mice were infected and the draining LNs analyzed 6 wk later. Transfer of T_FH cells, but not of ICOS⁻CXCR5⁻ CD4⁺ T cells rescued endogenous LN cell viability, as seen from cell numbers and normal proliferative behavior (Fig. 6 D and E). Transferred T_FH cells also rescued a resistant phenotype during leishmaniasis (Fig. S5D), and ICOS⁻CXCR5⁻ CD4⁺ T cells did not. Taken together, these data suggest a link between the T_FH cell defect and cell death in the draining Irf4^−/− LNs during chronic leishmaniasis.

Role of IL-21.

Next, we considered lack of a particular T_FH cell product as primary cause of LN cell death and missing Irf4^−/− GC formation. An important candidate was IL-21, which is induced within T_FH cells via the ICOS–c-Maf axis (9, 36) and is required for GC B-cell differentiation (7–10). We considered IL-21 as well, because expression of ICOS and IL-21 by Irf4^−/− cells is disturbed (Fig. 2 and ref. 21), and because IRF4 binds the ICOS promoter (25) and mediates cell responses to IL-21 together with STAT3 (37). To test for a role of IL-21, we compared the effects of adoptively transferred purified Il21^−/− and WT CD4⁺ T cells on GC formation and LN cell survival in Irf4^−/− mice 2 wk after infection (Fig. 7).

Fig. 7.

Role of IL-21. Irf4^−/− mice were infected with L. major. Where indicated, they also received 8 × 10⁶ purified Il21^−/− or WT CD4⁺ cells by intraperitoneal adoptive transfer. Two weeks (A and B) or 6 wk (C) after infection, draining LNs were prepared (B and C) as cell suspensions. (A) Tissue sections stained with antibodies to CD4 (red) and GL7 (green) and analyzed by fluorescence-microscopy (20× magnification). (Scale bars, 50 μm.) (B) FACS analysis of gated B220⁺ B cells stained for GL7 and CD95. Numbers indicate frequencies of B cells coexpressing GL7 and CD95. (C) Cell numbers within the draining LNs of individual mice with or without transfer of the indicated cells. Data are from one mouse representative for all experiments (A and B) or were compiled from the results of three individual experiments (C).

Although transferred WT cells again perfectly rescued the appearance of GCs and GL7⁺ cells (Fig. 7A), only few GL7⁺ cells were detectable after transfer of Il-21^−/− CD4⁺ cells and were spread throughout the LN as in control Irf4^−/− mice. Furthermore, transferred WT but not Il-21^−/− CD4⁺ cells rescued the frequency of GL7/CD95 coexpressing GC B cells (Fig. 7B), in comparison with Irf4^−/− mice without cell transfer.

To analyze whether IL-21 was also involved in the rescue of LN cell survival by WT cells (Fig. 6), part of the mice was analyzed 6 instead of 2 wk after infection. Transfer of Il-21^−/− CD4⁺ cells created an intermediate phenotype compared with Irf4^−/− mice receiving either no or WT CD4⁺ T cells (Fig. 7C). Accumulation of all data from the different experiments revealed less potent effects of Il-21^−/− than of WT CD4⁺ cells with clear statistical difference. However, even if their effects on Irf4^−/− LN cell survival were weaker, they still were demonstrable with high significance compared with Irf4^−/− mice without cell transfer.

Discussion

In the past, IRF4 has been characterized as an important transcription factor for differentiation of Th2, Th9, and Th17 cells. In addition, aspects of Treg cell function entirely depend on IRF4 (20–23, 25–27). In the B-cell lineage, IRF4 is important for plasma cell differentiation and isotype switching (28, 29). Our results link these previous findings in B and T cells and show an additional important role of IRF4 for development of T_FH cells, which are mainly responsible for the intricate organization of T–B interactions and antibody maturation in vivo.

For our analysis, we used infection of mice with L. major, a model characterized by strong B- and T-cell interactions and LN hyperplasia (30). When analyzing LNs of WT and Irf4^−/− mice at the height of GC formation in WT mice, a striking defect of Irf4^−/− mice became apparent: despite normal structure of B- and T-cell areas, their LNs totally lacked GC formation and differentiation of GC B cells. These findings were confirmed in PPs of naive mice and after immunization with a peptide instead of L. major infection. In parallel, LN CD4⁺ T cells expressed strongly reduced amounts of the T_FH cell-related (2, 11–13, 15, 16) molecules ICOS, IL-21, and BCL-6.

A remarkable result was obtained with respect to the T_FH marker CXCR5 in that LN cell suspensions contained conjugates of adherent B–T cells, which conferred the risk for misinterpreting FACS data on CXCR5-expression in CD4⁺ T cells, because of stronger CXCR5 expression on B cells. Similar B–T conjugates have recently been described and characterized for their B-cell part (5). The lower amounts of CXCR5 on single T cells correlated with high expression of ICOS and IL-21, indicating that CXCR5 can still be used to characterize T_FH cells. These findings were again confirmed in LNs of mice immunized with a peptide or in PPs from naive mice.

Integration of this information in the analysis of Irf4^−/− T_FH cell development revealed that Irf4^−/− LN CD4⁺ cells were completely CXCR5⁻. Importantly, this defect was only notable when the B–T conjugates were either out-gated or mechanically separated, because CXCR5 expression on Irf4^−/− B cells was as high as in WT mice. Lack of CXCR5 on Irf4^−/− CD4⁺ but not B cells suggests a T-cell intrinsic differentiation defect that cannot solely be caused by binding of IRF4 to the cxcr5 gene promoter. With respect to ICOS, its promoter has previously been linked to IRF4 (25). However, ICOS expression is not totally IRF4-dependent, because frequencies of Irf4^−/− cells with intermediate ICOS expression were even increased (Figs. 2 and 3), perhaps in a feedback-loop triggered by deficient terminal T_FH cell differentiation. Lower ICOS expression may be one reason why Irf4^−/− T cells produced almost no IL-21, because Il-21 transcription is induced by an axis via ICOS and c-Maf (36). However, impaired IL-21 production may also mirror an intrinsic deficiency of Irf4^−/− T cells, given that naive Irf4^−/− CD4⁺ T cells stimulated in vitro under Th17-inducing conditions also produced less IL-21 (21). In contrast to ICOS, the T_FH cell marker PD-1 was normally expressed in T cells of L. major infected Irf4^−/− mice. Therefore, missing T-cell stimulation is unlikely to explain absent Irf4^−/− T_FH cell development, a conclusion supported by normal IL-2 production in response to Leishmania antigens.

Importantly, missing Irf4^−/− T_FH cells and GCs correlated with eventual total LN cell death after initial LN hyperplasia. We have previously described enhanced in vitro sensitivity of Irf4^−/− CD4⁺ cells to activation-induced cell death (34). Herein, we observed that death of Irf4^−/− LN cells could be avoided by presence of WT CD4⁺ but not B cells. Only few CD4⁺ WT cells were required for protection, based on their low frequency within the surviving LN cells. LN cell survival was also saved by very few Leishmania primed WT T_FH cells, thus linking T_FH cells and protection from cell death. LN cell survival correlated with resistance to the parasite. Nevertheless, it is difficult to attribute this feature solely to the presence of T_FH cells, because Irf4^−/− mice are also devoid of Th9 or Th17 cells (20, 26, 27). A direct causal link of parasite burden and LN cell survival is not likely, given that draining LNs of mice from other highly susceptible strains behave totally differently.

Of note, adoptive transfer of WT CD4⁺ cells rescued not only LN cell survival, but also appearance of GCs and GC B cells with normalized expression of ICOSL in accordance with the feedback-loop between ICOS and ICOSL (33). These findings underline that the Irf4^−/− T_FH cell defect is T-cell intrinsic and support a report on normal GC formation in mice with a B-cell–specific defect of IRF4 (28). Transfer of Il-21^−/− CD4⁺ T cells proved a central role for IL-21 during the rescue of GC formation. In contrast, IL-21 was necessary but not sufficient for protection from LN cell death, because Il-21^−/− CD4⁺ T cells supported cell survival only partially. LN cell survival during a chronic immune response is of central importance and may be regulated by independent effector molecules, such as IL-4, which is produced locally by T_FH cells (4, 5) and which (like IL-21) cannot be produced by Irf4^−/− Th cells (22, 23).

As for the mechanism how IRF4 affects T_FH cell differentiation, we consider an important role to their key transcription factor BCL-6 (11–13), because its amounts are strongly reduced in Irf4^−/− CD4⁺ T cells. Apparently, this BCL-6–enhancing function of IRF4 is cell-type–specific, because IRF4 suppresses BCL-6 transcription in B cells (38). In addition, IRF4 physically binds to BCL-6 (39) and lack of this interaction may explain why Irf4^−/− Th cells are totally unable to differentiate into T_FH cells, although they still express some BCL-6. Possibly, the T_FH inducing capacity of BCL-6 requires its interaction with IRF4.

In conclusion, we demonstrate a decisive role of IRF4 for development of T_FH cells and, as a consequence, formation of GCs, differentiation of GC B cells, and survival of LN cells during an immune response. Together with the earlier findings that IRF4 also targets class-switching and terminal plasma cell differentiation, our study places IRF4 in the center of B–T cooperation during the formation of an adaptive immune response.

Materials and Methods

Mice and L. major Infection or Immunization.

WT C57BL/6 mice, purchased from Charles River, CD45.1 congenic, Il-21^−/− (from National Institutes of Health Mutant Mouse Regional Resource Centers), and Irf4^+/− and Irf4^−/− mice (bred in our own animal facilities) were used at 6–10 wk of age. Infection with promastigotes of the L. major strain MHOM/IL/81/FEBNI into the footpad, determination of lesion development or parasite burden, and leishmania-specific in vitro cell restimulation were performed as previously described (18, 34). Popliteal LNs or spleens were removed at the indicated timepoints for histology, qPCR, and FACS analysis, as detailed in SI Materials and Methods.

Total IgM levels were measured in a sandwich ELISA by coating with goat anti-mouse IgM followed by application of mouse sera and detection with AP-conjugated goat anti-mouse IgM (both reagents from Chemicon/Millipore). Immunization with MOG peptide and adjuvant was performed as previously described (20). All animal experiments were approved by “Regierungspräsidium Gießen” [permission number: V54-19c20-15(1) MR 20/6 Nr47/2009], the local institutional committee.

Statistics.

For statistical analysis, we used an unpaired t test with Welch’s correction.