A Bcl6 Intronic Element Regulates T Follicular Helper Cell Differentiation

The Bcl6 gene encodes a member of the broad complex, tramtrack, and bric-à-brac–zinc finger family of proteins. As a transcriptional repressor able to recruit corepressors (1), it plays an essential role in the germinal center (GC) reaction for both B and T cells (2–5), and thus constitutes a pivotal control point in adaptive immunity. Consistent with this, Bcl6 is expressed at increased levels in T follicular helper (T_FH) cells as compared with T effector (T_EF) cells, and this is manifest in the enhancement of H3K27 acetylation in a nonpromoter gene element located at the extreme upstream region of the first Bcl6 intron (6).

In a previous study, we identified tandem FOXO1 binding sites in the first intron of the Bcl6 gene, and we further showed that deletion of the Foxo1 gene led to enhanced expression of BCL6. This correlated with enhanced T_FH cell differentiation (CXCR5^hiBCL6^lo), but a profound loss of GC T_FH cells (CXCR5^hiBCL6^hi) (7). Analysis of the Bcl6 intronic region revealed two 9-base consensus forkhead sites located within 28 bases of one another at the point of FOXO1 binding. In our previous study (7), we proposed that FOXO1 acts as a transcriptional corepressor, an activity previously described for FOXO transcription factors despite their strong transcriptional activation domains (8, 9). The notion of FOXO1 mediating repression of Bcl6 is consistent with the observation that a PI3Kδ gain of function in humans and mice, as well as the resulting effects on FOXO1, strongly affects T_FH cell abundance and the GC reaction (10–13). In B cells, FOXO1 was shown to be necessary for GC dark zone development (14).

To determine whether the effects of FOXO1 regulation of T_FH cell differentiation were all or in part due to its direct effects on Bcl6 gene expression, we sought to carry out deletions of the FOXO1 binding sites in the Bcl6 gene in mouse embryos. In this study, we describe the differentiation of T_FH and GC-T_FH cells from two strains of mice harboring selective deletions of the two forkhead binding sites and show that BCL6 expression is increased, resulting in an enhanced GC response. This mechanism of control was found to be conserved because deletion of the very same sequence in cultured human T cells similarly enhanced T_FH cell differentiation.

To delete FOXO1 binding sites in the first intron of the Bcl6 gene in mice, CRISPR/Cas9 technology was carried out in C57BL/6J zygotes. Briefly, two single guide RNAs (sgRNAs), sgRNA6 (5′-ATAATGATCATGAGCAGCGG-3′) and sgRNA11 (5′-GCAGCAACAGCAATAATCACC-3′), were designed to target at the flanking sequences of FOXO1 binding sites in the Bcl6 intronic region. Pronuclear microinjection was carried out using sgRNA6, sgRNA11, and Cas9 mRNA with or without a DNA oligonucleotide to mediate homology-directed repair that would selectively delete the two FOXO1 binding sites. The preparation of sgRNA for pronuclei injection was carried out using in vitro transcription synthesis. The T7 promoter sgRNA scaffold template was produced by PCR (15). The sgRNAs were synthesized and purified by a MEGAshortscript kit (Ambion) and MEGAclear kit purification for large-scale transcription reactions (Ambion), respectively.

All procedures were carried out by the University of California, San Diego (UCSD) transgenic shared resource at the UCSD Moores Cancer Center. Two independent founders, IR18 and IR4, collectively referred to as Bcl6^ΔFKHD, were selected for further study. The differential length and sequence of PCR product confirmed deletion of the Bcl6 locus. IR18 and IR4 mice were crossed to include the P14 TCR transgene, SMARTA TCR transgene, or OT-II TCR transgene. Congenic CD45.1, CD45.1/CD45.2, Foxo1^AAA (7), and CD4Cre-ERT2 (16) mice were bred in our animal facility. CD4Cre-ERT2–mediated activation of Foxo1^AAA alleles was induced by i.p. injection of 2 mg of tamoxifen (Cayman Chemical) emulsified in 200 μl of sunflower seed oil (Sigma-Aldrich) for 5 constitutive days, followed by 3 d of rest. Bcl6^ΔFKHD mice were crossed with TCRα^−/− mice to generate Bcl6^ΔFKHD TCRα^−/− mice. All mouse studies were carried out in a specific pathogen-free facility at the UCSD vivarium and the experiments were performed according to the Institutional Animal Care and Use Committee of the UCSD.

Bcl6 mRNA amounts were determined by quantitative RT-PCR using forward primer (5′-TGAGCAGTTTAGAGCCCATAAG-3′) and reverse primer (5′-GTACATGAAGTCCAGGAGGATG-3′).

The desired amount of citrated sheep blood (Colorado Serum) was washed three times with PBS by centrifugation at 930 × g (2000 rpm) for 10 min at 4°C. The SRBC pellet was resuspended at the concentration of 10¹⁰ cells/ml. Then, 2 × 10⁹ SRBCs per mouse were injected i.p.

Splenocytes from OT-II TCR mice were obtained by mashing total spleens through a 70-μm nylon cell strainer (BD Biosciences), and the OT-II T cells were purified by an EasySep mouse naive CD4⁺ T cell isolation kit (STEMCELL Technologies). The purity of OT-II T cells was verified by FACS, and 10⁵ OT-II cells/mouse were adoptively transferred. The following day, mice were infected i.v. with 10⁵ PFU of vesicular stomatitis virus (VSV)–OVA (gift from Dr. Ananda Goldrath). For lymphocytic choriomeningitis virus (LCMV) infection experiments, the P14 or SMARTA T cells were purified by an EasySep mouse naive CD8⁺ T cell isolation kit or naive CD4⁺ T cell isolation kit (STEMCELL Technologies), respectively. The purity of P14 or SMARTA cells was verified by FACS, and these cells were adoptively transferred into congenic mice. Frozen stocks of LCMV Armstrong and clone-13 were diluted in PBS, and 2 × 10⁵ PFU of LCMV Armstrong was injected i.p. or 2 × 10⁶ PFU of LCMV clone-13 was injected i.v.

Spleens were homogenized and passed through a nylon cell strainer in HBSS (Life Technologies) supplemented with 2% FBS. Surface staining was performed for 30 min at 4°C in PBS supplemented with 2% FCS (FACS buffer) using the following Abs: CD4 (GK1.5), CD8α (53-6.7), CD38 (90), FAS (15A7), GL-7, SLAM (TC15-12F12.2), CD62L (MEL-14), CD8 (53-5.8), CD45.1 (A20), CD45.2 (104), CD44 (IM7), and PD-1 (J43) (BioLegend, San Diego, CA). CXCR5 staining was carried out as described (17) using anti-CXCR5 (2G8; BD Biosciences, San Jose, CA). Each cell-staining reaction was preceded by a 15-min incubation with a purified anti-mouse CD16/32 Ab. Gp33 and Gp66 tetramers were made and provided by the National Institutes of Health tetramer core facilities (Atlanta, GA). Gp66 tetramer staining was performed at 37°C for 90 min. Intracellular transcription factor staining was performed with a FOXP3 staining kit (eBioscience) and stained with Abs specific for TCF7 (C63D9; Cell Signaling) or BCL6 (K112-91; BD Biosciences).

Bone marrow cells were isolated from TCRα^−/−, Bcl6^ΔFKHD TCRα^−/− and wild-type (WT) CD45.1/CD45.2 mice and processed under sterile conditions. A single-cell suspension in PBS was obtained with a 1:1 ratio of TCRα^−/− and WT or Bcl6^ΔFKHD TCRα^−/− and WT mice bone marrow cells. Cells (5 × 10⁶) were injected i.v. into lethally irradiated (10 Gy) CD45.1 hosts in a volume of 200 μl. A second model of bone marrow transplantation was obtained by injecting a 1:1 ratio of WT (CD45.1) and IR18 (CD45.2) bone marrow cells into lethally irradiated (10 Gy) TCRα^−/− hosts in a volume of 200 μl. Radiation chimeras received autoclaved water treated with antibiotics (trimethoprim-sulfamethoxazole) until 4 wk after injection.

Human blood was obtained from the Stanford Blood Center. The CD3⁺ human T cells were isolated using a RosetteSep human T cell enrichment cocktail kit (STEMCELL Technologies) according to the vendor’s manual. Surface staining was performed on enriched samples for 15 min at room temperature in HBSS supplemented with 2% FBS using the following Abs: CD4 (RPA-T4), CD8 (RPA-T8), PD-1 (EH12.2H7), CD45RO (UCHL1), and CD45RA (HI100) (BioLegend). For live/dead discrimination, Zombie NIR dye was used (BioLegend). The purity of CD3 T cells was checked by FACS and reached to 90–98%. T cells were activated with Ab-coated beads specific for CD3, CD28, and CD2 (STEMCELL Technologies) with the addition of IL-2. At 2 d postactivation cells were electroporated using a Neon transfection system under the following conditions (1600 V, 10 ms, and three pulses) to incorporate the empty vector (pSpCas9(BB)-2A-GFP [PX458], Addgene) or two PX458 vectors, with one expressing sgRNA6 and the other expressing shRNA11. The cells were then cultured for 5 d under conditions that promote T_FH cell differentiation (IL-12 at 5 ng/ml, TGF-β at 1 ng/ml, and activin A at 50 ng/ml) (18). CD4⁺ GFP⁺ T cells were FACS sorted and stained for CD4, CXCR5, PD-1, and BCL6.

Flow cytometry measurements of cells were performed on a BD LSRFortessa or BD LSRFortessa X-20 cytometer, and cell sorting was performed on BD FACSAria III. All FACS data were analyzed using FlowJo 10 (Tree Star). Prism 6 software (GraphPad Software) was used to make and to analyze data by a two-tailed unpaired or paired Student t test. Significance was defined as follows: *p < 0.05, ** p < 0.01, ***p < 0.001, and ****p < 0.0001. Data are presented as means ± SEM.

The Bcl6 gene consists of 10 exons with the first 2 noncoding exons separated by a long intron (Fig. 1A). The first exon and the upstream 270 bp of the first intron of Bcl6 constitute a region that is notable for sequence conservation in placental mammals (Supplemental Fig. 1). It is often a site of translocations, deletions, and mutations (kataegis) resulting from somatic hypermutation found in normal GC B cells and GC-derived B cell lymphomas (19–22). This region also constitutes a superenhancer with convergent anti-sense transcription (human LOC106146153) and dense H3K27 modifications that attract activation-induced cytidine deaminase–mediated mutations and translocations (22–27) (Fig. 1B, Supplemental Fig. 1).

To target the forkhead sites bound by FOXO1 (Fig. 1B), candidate flanking Cas9-sgRNAs were validated in fibroblasts in culture followed by DNA sequencing. Two of the sgRNAs flanking the FOXO1 binding sites (sgRNA mit-6 and sgRNA mit-11) were effective in producing indels when used separately and ∼72-bp deletions when used in conjunction (data not shown). These two sgRNA sequences were transcribed in vitro for injection into C57BL/6J single-cell mouse embryos along with mRNA encoding CAS9. Some embryos were also injected with a ssDNA template homologous to the region separating the sgRNA cut sites but with the 9-bp deletions corresponding precisely with the two forkhead consensus sequences.

From a number of mouse strains derived from this procedure (Bcl6^ΔFKHD), two were chosen for analysis. Bcl6^IR4 has a 78-bp deletion corresponding with imprecise nonhomologous end-joining of the two sgRNA cut sites, and Bcl6^IR18 represents a homology-directed repair event similar but not identical to the ssDNA template (Fig. 1C). These alleles were transmitted at a Mendelian frequency and the mice showed no obvious phenotypic effects. For most experiments, heterozygous strains were analyzed.

To test the phenotype of Bcl6^ΔFKHD strains, heterozygous mice were immunized with SRBCs, and the splenic T and B cells were monitored after 7 d. Spleen cells were gated for CD4⁺CD44^hi cells and analyzed for a T_FH phenotype based on the expression of PD-1 or BCL6 versus CXCR5. As shown, both strains displayed a 2-fold increase in the percentage and total number of GC T_FH cells (CXCR5^hiPD-1^hi or CXCR5^hiBCL6^hi) (Fig. 2A, Supplemental Fig. 2A). In addition, BCL6 expression (mean fluorescence intensity [MFI]) in T cells from the Bcl6^ΔFKHD strains was increased by ∼30% over the WT (Fig. 2B). A similar phenotype was seen for the number GC B cells characterized by expression of either FAS⁺GL-7⁺ or CD38^loBCL6⁺ (Fig. 2C, Supplemental Fig. 2B), and the amount of BCL6 expressed by GC B cells (peak MFI) was increased by ∼22% (Fig. 2D). These increases were not seen preimmunization (Supplemental Fig. 2C, 2D). In initial experiments, the Bcl6^IR18 mice homozygous displayed a phenotype that was further exaggerated compared with heterozygous Bcl6^ΔFKHD mice (Supplemental Fig. 2D, left); however, upon testing more mice, this difference was not consistently found (Supplemental Fig. 2D, right). The increase in BCL6 protein expression was reflected in an increase in Bcl6 mRNA expression (Supplemental Fig. 2E).

To characterize the phenotype of a Bcl6^ΔFKHD deletion in the response to a pathogen, mixed bone marrow chimeras were set up in which bone marrow cells from WT and Bcl6^IR18 mice were used to reconstitute lethally irradiated TCRα^−/− mice. Eight weeks following reconstitution, mice were infected with LCMV clone 13 virus and analyzed for tetramer-positive (GP66-specific) T cells 8, 29, and 56 d later (Fig. 3A). As shown, surprisingly, there was no difference in the proportion of WT compared with Bcl6^IR18 GP66-specific GC T_FH cells at day 8, and yet, at later times there was a consistent difference comparing WT and Bcl6^IR18 GC T_FH cells within individual animals (Fig. 3B). Further analysis at day 56 revealed that within the same animal, the numbers of GC T_FH (GP66⁺CXCR5⁺BCL6⁺) cells and total GP66⁺ T cells were significantly increased, and the GC T_FH cells present expressed a higher amount of BCL6 as revealed by the MFI (Fig. 3C).

The process of GC formation involves the comigration of T cells and B cells to the cortical/follicular boundary where B cells present Ag to developing T_FH cells in an exchange that promotes further differentiation of both cell subsets. Thus, an enhancement of either cell type might be predicted to affect the GC reaction and the expansion of both GC T and B cells. In order to determine whether the effects are T cell specific, WT SMARTA or Bcl6^IR18 SMARTA T cells were transferred into C57BL/6 hosts that were then infected with an acutely infectious (LCMV Armstrong) virus (Fig. 4A). Mice were then analyzed at day 9, a time when the virus has been cleared (28).

The analysis of T_FH cells by means of BCL6 or SLAM expression versus CXCR5 revealed an increased proportion of Bcl6^IR18 T_FH cells as compared with WT at day 9 postinfection (Fig. 4B). In this experiment, there was a trend toward increased proportions of GC T_FH cells (CXCR5^hiBCL6⁺) when comparing WT and Bcl6^IR18 T cells, whereas the increase in the T_FH cell compartment as measured by either CXCR5⁺ (upper panel) or CXCR5⁺SLAM⁻ expression (lower panel) achieved significance. Furthermore, the amount of BCL6 expressed per cell was increased in T cells from Bcl6^IR18 mice. These results are consistent with an intrinsic effect of the Bcl6^ΔFKHD allele on the transcription of the Bcl6 locus and the differentiation of T_FH cells.

Similarly, the T cell–intrinsic phenotypic effects of a Bcl6^ΔFKHD allele were analyzed in response to a chronic viral infection (Fig. 4C). In this experiment, genetically distinct WT and Bcl6^IR18 T cells were transferred followed by infection with LCMV clone 13. At days 9 and 29 postinfection, there was a significant increase in the proportion of GC T_FH cells among SMARTA T cells from Bcl6^IR18 mice compared with T cells from WT mice (Fig. 4D). Again, the amount of BCL6 expressed by cells within this subset was also increased. In contrast, the proportions of total transferred SMARTA T cells did not differ between Bcl6^IR18 and WT T cells at either day 9 or 29; why this result at day 29 differs from the expansion found in bone marrow chimeras at day 56 (Fig. 3) is discussed. These results are consistent with an enhanced T cell–intrinsic T_FH cell differentiation in Bcl6^ΔFKHD mice under conditions of equivalent expansion.

To determine whether there was also a B cell–intrinsic component to the Bcl6^ΔFKHD allele, mixed bone marrow radiation chimeras were set up such that the expansion of WT or Bcl6^IR18 B cells could be compared in the presence of WT T cells (Fig. 5A). After 8 wk, the mice were infected with VSV-OVA and GC B cells were analyzed as above. Perhaps surprisingly, there was no difference in the proportion of GC B cells (out of total B cells) when comparing WT to Bcl6^IR18 B cells, and this was found at 8 and 15 d postinfection (Fig. 5B, 5C). We conclude that, although GC B cells from immunized Bcl6^ΔFKHD mice express an increased amount of BCL6, this increase may originate from enhanced T_FH cell maturation and the signals thereby provided.

The deletion of Foxo1 results in increased Bcl6 mRNA and BCL6 protein expression in T_FH cells (7), and deletion of the only FOXO1 binding sites in the Bcl6 gene also enhances BCL6 expression (Figs. 3B, 4B, 4D). We also previously showed that expression of a Foxo1 triple phosphorylation mutant (Foxo1^AAA) that results in the expression of a constitutively nuclear form of FOXO1 prevents T_FH cell differentiation (7). A prediction is that the loss of T_FH cells resulting from Foxo1^AAA expression might be rescued by the Bcl6^ΔFKHD allele; if the effect of constitutively active FOXO1 is through suppression of Bcl6 expression, then the Bcl6^ΔFKHD allele should be resistant to the suppression.

To test this possibility, T cells from OT-II, OT-II Foxo1^AAA, or OT-II Bcl6^IR18 Foxo1^AAA mice were transferred into naive C57BL/6 recipients that were subsequently infected with VSV-OVA. The Ag-specific CD4 T cells were analyzed 8 d later (Fig. 6A). WT OVA-specific T cells were predominantly T_FH cells (CXCR5⁺BCL6^low) (Fig. 6B, left panel, 6C, middle panel, average 68%). With the addition of constitutively nuclear FOXO1 (Foxo1^AAA allele), most T cells instead assumed an effector phenotype (CXCR5⁻) (Fig. 6B, middle panel, 6C, left panel, average 79%). With the further addition of a hemizygous Bcl6^IR18 allele, the differentiation to a T_FH cells phenotype was partially rescued such that the proportions of T_EF and T_FH cells were nearly equivalent (Fig. 6B, right panel, 6C). This result is consistent with the notion that FOXO1 mediates Bcl6 repression; however, in the presence of even a haploid Bcl6^ΔFKHD allele, T_FH cell differentiation was partially restored.

To further study BCL6 regulation, we sought to determine whether human T_FH cells are also regulated by these conserved forkhead consensus sites found in the human BCL6 locus. As shown in (Fig. 1B, the mouse and human regions of interest are identical such that the same sgRNA sequences, that is, sgRNA mit-6 and sgRNA mit-11, that were used for deletion of the mouse locus in fertilized embryos could be used for human T cells. To carry out the deletion, we cloned each of the 21-bp CRISPR RNA (crRNA) targeting sequences individually into the PX458 vector such that the human U6 promoter will transcribe a composite scaffold-guide RNA targeting sequence flanking the forkhead binding sites. The PX458 vector also encodes CAS9 and enhanced GFP separated by a T2A self-splicing sequence and driven by a hybrid CMV-chicken β-actin promoter (29). CD3⁺ T cells from fresh human blood were stimulated for 2 d with Abs specific for CD3, CD28, and CD2 and subjected to electroporation to deliver guide RNA and Cas9-coding plasmids (Fig. 7A). The experimental cells were electroporated with two vectors, one expressing sgRNA mit6-11 and the other sgRNA mit11-10. As a control, T cells were electroporated with the PX458 plasmid expressing the trans-activating crRNA without a crRNA recognition sequence. The cells were further cultured for 5 d in the presence of IL-12, TGF-β, and activin A (18) and sorted for enhanced GFP fluorescence (Fig. 7A).

The analysis of these T cells showed that, for a typical experiment, 32% of the BCL6 alleles were deleted (Fig. 7B). The PCR products were further subjected to sequencing, and this revealed substantial deletions around each of the target sequences (Fig. 7C). The analysis of phenotype further revealed a strong increase in the percentage of CXCR5^hiBCL6^hi or CXCR5^hiPD-1^hi T cells (Fig. 7D).

Large-scale projects to decode the cis-acting sequences that control gene expression have revealed some of the complexity and redundancy of gene elements involved in, for example, tissue specification during development (30–32) or housekeeping functions in mammalian cell lines (33). One of the overriding themes in much of this work is robustness of gene expression and its resistance to perturbation based on short indels or even tiled deletions that encompass up to 1000 bases. More precisely, analyses indicate the existence of widespread enhancer redundancy or the concept of shadow enhancers, that is, two or more enhancers that promote activation of gene transcription in a way that renders deletion of any one enhancer phenotypically silent. Upon this background, deletion of a factor binding site downstream of the transcription start site might not be anticipated to affect expression or cellular differentiation; however, when deletion does result in a change in gene expression, it may identify such an element as being of central importance.

One question is how this type of redundancy extends to enhancers marked by abundant H3K27 acetylation and bound by p300, BRD4, and mediator, defined as superenhancers (34, 35). Such enhancers are essential for the expression of genes that are often lineage defining or key to lineage commitment (25). They attract a diverse array of transcription factors, and recent work links their activity with the formation of phase-separated condensates produced by interaction between the intrinsically disordered domains of BRD4, mediator (MED1), and transcription factor trans-activation domains. A notion is that superenhancers may be excluded from the concept of redundancy, as they appear to be highly sensitive to perturbation (36), and, at least for GC B cells, this applies to the regulatory region located at the first Bcl6 exon and the 5′ region of the first intron (22, 37). In fact, based on the density of H3K27Ac marks listed in Encode, the region surrounding the first noncoding exon of Bcl6 appears to be a superenhancer in several unrelated cell types. A second question is whether there is redundancy of repressive elements, and exactly how repressive elements act mechanistically.

Although we have not attempted to address these complex questions directly, we extended our observations on the role of FOXO1 in T_FH cell differentiation, affirming that the FOXO1 tandem binding sites, mapped by chromatin immunoprecipitation sequencing and unique within a megabase of the Bcl6 gene, mediate nonredundant repressive activity that impacts the differentiation of T cells. Thus, loss of FOXO1 binding to the Bcl6 gene enhanced GC T_FH cell differentiation, whereas constitutively nuclear FOXO1 blocked T_FH and GC T_FH cell differentiation. For example, following infection with VSV-OVA, the average proportion of T_EF cells averaged ∼18%, but with the addition of the Foxo1^AAA allele this number jumped to almost 80%. With the additional Bcl6^IR18 allele the T_EF cells dropped back to 50%. Likewise, the average proportion of T_FH cells was 68% for WT that was reduced to 20% for Foxo1^AAA T cells, and this deficiency was partially rescued to 43% with the further addition of a hemizygous Bcl6^ΔFKHD allele, an addition that should alleviate FOXO1 binding on one chromosome. A possibility is that Foxo1^AAA Bcl6^{ΔFKHD/ΔFKHD} OT-II mice (with a homozygous Bcl6^ΔFKHD allele) might have been completely rescued from the effects of the Foxo1^AAA inhibition, but we were not able to test such mice. We conclude that nuclear FOXO1, bound to these specific sites, is an important determining factor in Bcl6 expression and the differentiation of T_FH cells, although we cannot conclude that this is the only effect exerted by FOXO1 on T_FH or GC T_FH cell differentiation.

A complication to understanding this process is that although T_FH cell differentiation is inhibited by FOXO1, the differentiation from T_FH to the GC T_FH cell phenotype actually requires FOXO1 (7). This, in turn, implies that mechanisms of FOXO1 regulation may vary within the time frame of the initial activation of naive T cells through to the establishment of a T cell–replete GC. Consistent with this notion, studies on the activation of T cells showed that FOXO1 is initially sequestered from the nucleus via triple, AKT-mediated phosphorylation, and we deduce that this is required to relieve cell cycle inhibition via p27^kip1 (38). As this inhibition subsides, it is replaced by posttranscriptional inhibition mediated by miRNA-182, part of the miR-183–miR-96–miR-182 cluster (39). In the differentiation phase, transition to effector Th17 cells was shown to depend on miR-183 as a third means of reducing Foxo1 mRNA (40).

An unanticipated result was that the increases in GC B cells appeared to be cell extrinsic. In the experiments presented, irradiated mice were reconstituted such that the Bcl6^ΔFKHD allele was present in cells other than T cells and then infected with VSV-OVA. Subsequently, we analyzed the total number of GC B cells similar to our initial experiments using SRBCs as an immunogen. We saw neither an increase in the proportion of GC B cells nor an increase in B cell–expressed BCL6. Why there was found to be no difference in BCL6 expression under this protocol is unknown, but the direct implication is that BCL6 expression in GC B cells is affected by the strength of T cell help, and the increased BCL6 expression noted in immunized Bcl6^ΔFKHD mice resulted from the enhanced T cell help provided by Bcl6^ΔFKHD T cells. T cell help includes CD40L–CD40 interactions and the production cytokines such as IL-4 and IL-21, and these signals can have important effects on the dynamics and magnitude of GC B cell differentiation that can include the expression of BCL6 (41, 42).

Another facet of the requirement for FOXO1 in GC T_FH cell differentiation may derive from a requirement for TCF7 and LEF1 acting to repress expression of Prdm1 (Blimp-1) (43–46). In CD8⁺ T cells, TCF7 expression is entirely dependent on FOXO1 postactivation (47, 48). If TCF7 is likewise dependent on FOXO1 in CD4⁺ T cells, GC T_FH cell differentiation would be expected to be FOXO1-dependent. As FOXO1 can have opposing effects at different stages of the process, and it responds to many different extrinsic signals, it is likely to affect T cell differentiation through multiple mechanisms (49).

The Bcl6 gene is a central control point for many aspects of T and B cell differentiation and function (50). It is known to be regulated in T cells by several signaling pathways, including those that result in the activation of STAT3, STAT4, STAT5, TCF7, LEF1, TBET, BCL6, AIOLOS, IKAROS, BOB1, and others (5, 45, 51–53). In addition, the locus is regulated by the organization of its genomic architecture (37). STAT5-inducing cytokines, IL-2 and IL-7, have been found to repress Bcl6 and the differentiation of T_FH cells (54, 55), whereas STAT3/4-inducing cytokines promote Bcl6 expression and the appearance of T_FH cells (51, 56–58). In addition, for an individual naive T cell, the proportion of daughter cells that differentiate to become T_FH cells versus Th₁ or Th₁₇ effector cells in response to infection appears to be determined by the T cell Ag receptor expressed, and experiments show that the strength of signal generated by recognition of dendritic cell–peptide–MHC complexes is the determining factor (59). However, the correlation between dwell time (or affinity) and effector versus T_FH cell differentiation appeared to differ depending on the exact interval of affinities being tested (60). In addition, different TCRs gave rise to different amounts of IL-2Rα-chain expression, implying that IL-2 signals constitute a control point for the differentiation of individual clones of T cells. Because FOXO1 responds to TCR ligation and cytokine signals that activate STAT5- and STAT3/4-inducing cytokines, it clearly constitutes at least one integrating principle in T_FH versus effector cell differentiation.

This publication is dedicated to A.D. in memoriam.

The authors have no financial conflicts of interest.