Bach2 negatively regulates Tfh cell differentiation and is critical for CD4⁺ T cell memory

Abstract

T follicular helper (Tfh) cells are essential for germinal center B cell responses. The molecular mechanism underlying the initial Tfh cell differentiation, however, is still incompletely understood. Here, we show that in vivo, despite enhanced non-Tfh effector functions, the deletion of transcription factor Bach2 results in preferential Tfh cell differentiation. Mechanistically, the deletion of Bach2 leads to the induction of CXCR5 expression even before the up-regulation of Ascl2. Subsequently we have identified a novel regulatory element in the murine CXCR5 locus that negatively regulates CXCR5 promoter activities in a Bach2-dependent manner. Bach2-deficiency eventually results in a collapsed CD4⁺ T cell response with severely impaired CD4⁺ T cell memory including Tfh cell memory. Our results demonstrate that Bach2 critically regulates Tfh cell differentiation and CD4⁺ T cell memory.

Introduction

T follicular helper (Tfh) cells are a unique CD4⁺ T cell subset that plays an essential role in the formation of germinal centers and the generation of high-affinity antibodies (1, 2). The molecular mechanism underlying the Tfh cell differentiation, in particular, the initial stage involving the regulation of CXCR5 expression, is still not well understood.

Induction of CXCR5 expression is a hallmark of Tfh cell differentiation (3, 4). For activated conventional CD4⁺ T cells to move out of the T-cell zone and migrate into the B-cell follicle, studies have shown that the induction of CXCR5 as well as the down-regulation of CCR7 is critical for this process (5). Although cytokine IL-21 and co-stimulatory molecule ICOS have been shown to be important for CXCR5 induction (6, 7), the molecular mechanism underlying the regulation of CXCR5 expression is not clear. An early report on transcription factor Ascl2 has shown that remarkably Ascl2 alone is sufficient to induce CXCR5 expression, and the study suggests that the Ascl2 induction initiates the Tfh cell programming (8).

In lymphocytes, the transcription factor Bach2 was initially discovered as a key player in antibody class switching (9). Subsequently, in T cells, Bach2 has been found to play a critical role in maintaining Treg function and homeostasis (10, 11); and Bach2 suppresses CD4⁺ T effector functions by constraining Th1, Th2, and Th17 cell differentiation (10, 12). In addition, studies have also shown that Bach2 is required to suppress effector memory-related genes to maintain the naïve T cell state (13). Without Bach2, CD8⁺ T cell memory is severely impaired (14). In Tfh cell differentiation, Bcl6 has been demonstrated to play a central role (15–17), and Blimp-1 has been shown to antagonize the function of Bcl6 in Tfh cell development (15). Interestingly, Blimp1 has been shown to be a direct target of Bach2, and Bach2 negatively regulates Blimp1 expression (13, 18, 19). The function of Bach2 in Tfh cell differentiation (20, 21), however, has not been studied.

Here we report that, surprisingly, Bach2 is a negative regulator of Tfh cell differentiation. Despite the increased effector functions and Blimp1 expression, the deletion of Bach2 in CD4⁺ T cells leads to preferential Tfh cell differentiation. We find that Bach2 negatively regulates CXCR5 expression, and the enhanced CXCR5 upregulation in the absence of Bach2 occurs before the induction of Ascl2. Bach2-deficiency eventually results in a collapsed CD4⁺ T cell response and Bach2 is important for CD4⁺ T cell memory including Tfh cell memory.

Materials and Methods

Mice.

All mice were maintained in specific pathogen-free barrier facilities and were used in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham. ACTB:FLPe B6J, Cre-ERT2⁺, Rosa^YFP, OT-II and B1–8i transgenic mice were from The Jackson Laboratory. Cd4-Cre transgenic mice and Ly5.1⁺ (CD45.1) C57BL/6 congenic mice were from Taconic. Bach2^{tm1a(EUCOMM)WTsi} (termed Bach2^f/f in this study) mice were recovered by The Jackson Laboratory from cryopreserved sperms. Bach2^f/f mice were bred with ACTB:FLPe B6J to delete the neo-cassette, then were bred with Cre-ERT2⁺, Rosa^YFP, OT-II and Cd4-Cre mice to generate various control and experimental mice.

Flow cytometry, cell sorting and intracellular staining.

These procedures were carried out as described previously (22, 23). The sorted population were >98% pure. Antibodies were as follows: Alexa Fluor 647-anti-mouse Bcl6 (K112–91), Alexa Fluor 647-hamster anti-mouse CD95 (Fas) (Jo2) from BD Biosciences; Brilliant Violet 421-anti-mouse CD185 (CXCR5) (L138D7), PE-Cy7-anti-mouse PD-1 (29F.1A12), PE-Cy7-anti-mouse CD38 (90), APC-Cy7-anti-mouse CD4 (GK1.5), Alexa Fluor 700-anti-mouse CD45.2 (104), Brilliant Violet 510-anti-mouse/human CD45R/B220 (RA3–6B2), Percp-cy5.5-anti-mouse/rat CD90.1 (Thy1.1)(OX-7), APC-anti-human CD271 (NGFR) (ME20.4) from Biolegend. For multi-color flow cytometry analysis, cells were first gated on size and singularity, followed by excluding dead cells using Live/Dead Fixable Blue Dead Cell staining Kit (Invitrogen). Donor OT-II cells were gated on the congenic marker CD45.2 and/or YFP, and donor B1–8i B cells were gated on the congenic marker CD45.2 for further analysis. Flow cytometry results were analyzed using Flowjo software (Tree Star).

T cell stimulation and retroviral transduction.

T cell stimulation and retrovirus transduction were done as described previously (22, 23). The open reading frames of Bach2 were sub-cloned into the retroviral vector MSCV-IRES-Thy1.1 (MIT). The plasmid MigR1-CXCR5 was a gift from Dr. Hai Qi. The plasmid MIT-Bach2-ΔZIP was a gift from Dr. Rahul Roychoudhuri. The self-inactivating retroviral CXCR5 promoter reporter was a gift from Dr. Di Yu. The CXCR5 5’ regulatory element Site 1 (chr9:44370828–44371436) and Site 2 (chr9:44345810–44346556) were sub-cloned into the CXCR5 promoter reporter vector. Retroviruses containing sequences encoding Bach2, CXCR5 and CXCR5 reporter were produced in Plat-E cells (a gift from Dr. Matthew Pipkin) by co-transfection with retroviral vectors and helper plasmids.

Adoptive transfer.

Six- to ten- week old mice were treated with tamoxifen (Sigma-Aldrich) as described previously (22, 23). YFP⁺ or YFP⁻ CD44^loVα2^hiCD4⁺ naive OT-II T cells were sorted with a BD FACSAria III sorter (BD Biosciences). 0.5×10⁶ naive T cells alone or together with 10×10⁶ total B cells (≈ 5% Igλ⁺) from B1–8i mice were transferred into age- and sex-matched SMARTA recipient mice through tail vein (i.v.) injection, followed by immunization by intraperitoneal injection (i.p.) of 100 μg 4-hydroxy-3-nitrophenyl acetyl OVA (NP-OVA, Biosearch Technologies) emulsified in alum adjuvant (Thermo Fisher Scientific). For mixed transfer experiments, 0.5×10⁶ naive OT-II Bach2-WT T cells mixed with 0.5×10⁶ naive OT-II Bach2-cKO T cells were co-transferred into the SMARTA recipient mice through i.v. injection, followed by immunization by i.p. injection of 100 μg NP-OVA in alum. For the transfer of activated T cells, 1×10⁶ T cells were co-transferred with 10×10⁶ total B cells (≈ 5% Igλ⁺) from B1–8i mice into age- and sex-matched SMARTA recipient mice followed by immunization with 100 μg NP-OVA in alum. For the re-challenge experiments, the recipient mice were first immunized with 100 μg NP-OVA in alum after cell transfer; 30 days later, mice were re-challenged by i.p. injection of 100 μg NP-OVA in PBS.

Infection with influenza virus.

Mice were immobilized with isoflurane and were infected intranasally with mouse-adapted influenza virus strain A/Puerto Rico/8/34-OVA_323–339 (PR8-OVA; H1N1) at a dose of 200 VFU or mouse-adapted influenza virus strain A/Puerto Rico/8/34 (PR8) at a dose of 15000 VFU.

Real-time RT-PCR.

Total RNA purification and some PCR primers were as described previously (22, 23). Expression of mRNA was normalized to Rpl32 expression and presented as relative to wild-type naive OT-II or CD4⁺ T cells. The new primers were as follows: for Bach2 (forward, 5’-ACTGGTTGGACAGACGAAAG-3’, and reverse, 5’-AGTAACAGCTTGGCAGTGTAG-3’); for Ascl2 (forward, 5’-CGCTGCCCAGACTCATGCCC-3’, and reverse, 5’-GCTTTACGCGGTTGCGCTCG-3’); for CXCR5 (forward, 5’-GACCTTCAACCGTGCCTTTCTC-3’, and reverse, 5’-GAACTTGCCCTCAGTCTGTAATCC-3’).

RNA-seq analysis

Total RNA was isolated by using miRNeasy Mini Kit (QIAGEN). Illumina library preparation and sequencing were performed at La Jolla Institute sequencing center. Sequencing adapters were trimmed from reads using TrimGalore! (Babraham Bioinformatics). Trimmed reads were aligned to mouse mm10 using STAR (24), and reads mapping to individual genes were counted using HTSeq-Count. Raw read counts were normalized, and differential expression analyzed using DESeq2 (25). The scatterplot comparing expression between wild-type and Bach2-deficient T cells was constructed by plotting the log2 of the mean of normalized counts plus a pseudo-count of 1 for each gene. The Venn diagram was built by generating a list of genes that differed by at least 1.5-fold and had an adjusted p-value less than 0.01, which was compared to a list of 145 Tfh-associated genes (22). Pathway analysis was performed using the Gene Set Enrichment Analysis (GSEA) program available from the Broad Institute.

Accession codes

GEO: RNA-seq data, GSE123350 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE123350)

Histology.

These procedures were carried out as described previously (22, 23). Antibodies used for staining: purified rat anti-mouse CD35 (8c12; BD Biosciences), biotin-anti-CD45.2 (104; BD Biosciences), Alexa Fluor 555-conjugated goat polyclonal anti-rat (Invitrogen), Alexa Fluor 488-Streptavidin (Invitrogen), Alexa Fluor 647-conjugated rat Ab to mouse IgD (11–26C.2a; Biolegend). Mounted sections were imaged with 10 × and 20 × objective on a Nikon Eclipse Ti-S microscope.

Statistics.

Unpaired two-tailed Student’s t-test, one-way analysis of variance (ANOVA) and two-way analysis of variance (ANOVA) was used for calculation of P values. Statistical analysis was performed with Graphpad Prism 6.

Results

Bach2 negatively regulates Tfh cell differentiation

To study how Bach2 may regulate Tfh cell differentiation, we first examined the Bach2 expression in CD4⁺ T cells after activation in vivo in the OT-II/NP-OVA model system (22, 23). We found that compared with naive OT-II T cells, the activated non-Tfh OT-II T cells had reduced Bach2 mRNA expression, with PD-1⁺CXCR5⁺ OT-II Tfh cells having the lowest levels (Fig. 1A). To verify such results, we employed a PR8 influenza virus infection model in C57BL/6 mice. We found that on day 14 after PR8 intranasal infection, the Bach2 mRNA levels were also the lowest in PD-1⁺CXCR5⁺ Tfh cells (Fig. 1B), suggesting that the Bach2 expression needs to be suppressed for Tfh cell differentiation.

Figure 1. Bach2 negatively regulates Tfh cell differentiation.

(A) Real-time-PCR analysis of Bach2 mRNA in wild-type (WT) naive OT-II T cells, day 7 splenic donor PD-1-CXCR5- non-Tfh OT-II and PD-1⁺CXCR5⁺ Tfh OT-II T cells from the Ly5.1⁺SMARTA recipient mice given transfer of naive OT-II T cells followed by immunization with NP-OVA in alum. (B) Real-time-PCR analysis of Bach2 mRNA in naive CD4⁺ T cells, day 14 CD44⁺PD-1-CXCR5- non-Tfh cells and CD44+PD-1+CXCR5+ Tfh cells from the mediastina lymph nodes of Ly5.1+ mice infected with influenza virus PR8. (C-F) Naive OT-II Bach2-WT or OT-II Bach2-cKO T cells were co-transferred with B1–8i B cells into the Ly5.1⁺SMARTA recipient mice followed by immunization with NP-OVA in alum. The number of total splenic donor OT-II T cells was analyzed at d 3, 5 and 7 (C). The Tfh cell differentiation of donor OT-II T cells and the number of total splenic donor Tfh cells (D-E), and the GC B cell differentiation of B1–8i cells and the number of total splenic GC B1–8i cells (F) were analyzed 5 d after immunization. Bars in (C) represent average ± standard deviation (SD), (day 3, n=4; days 5 and 7, n=6). Bars in (D-F) represent average ± SD, n=6. (G) Naive OT-II Bach2-WT and OT-II Bach2-cKO T cells were co-transferred into the Ly5.1⁺SMARTA mice followed by immunization with NP-OVA in alum. The Tfh cell differentiation of donor T cells was analyzed 4 d later. Bars in (G) represent average ± SD, n=3. Data in (A-G) are representative (or pooled results) of two independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001.

We obtained from the European Conditional Mouse Mutagenesis Program (EUCOMM) the cryopreserved sperms of the mouse line with Bach2 allele conditionally targeted, and generated the Bach2^f/f mice accordingly (Supplemental Fig. 1A–C). Studies have shown that Bach2 is expressed at high levels in single-positive thymocytes (13). To circumvent potential thymocyte development issue, we generated OT-II^TgBach2^f/fCre-ERT2⁺Rosa^YFP mice, in which the Cre recombinase expressed from the gene encoding a tamoxifen-sensitive estrogen receptor variant (Cre-ERT2) will induce conditional deletion of both the Bach2 alleles and the stop codon in front of the gene encoding yellow fluorescent protein (YFP) inserted in the Rosa locus (22, 23); the Bach2-deleted OT-II cells will express YFP from the ubiquitously expressed Rosa26 locus.

We sorted CD44^low YFP⁺ Bach2-deficient naive OT-II (OT-II Bach2-cKO) or wild-type naive OT-II (OT-II Bach2-WT) T cells from tamoxifen-treated OT-II^TgBach2^f/fCre-ERT2⁺Rosa^YFP mice or from control OT-II^TgBach2^f/fRosa^YFP (or OT-II^TgBach2^+/+Cre-ERT2⁺Rosa^YFP) mice, respectively. Sorted cells were then co-transferred with B1–8i B cells into lymphocytic choriomeningitis virus (LCMV) epitope gp66–77-specific (SMARTA) TCR transgenic recipient mice followed by immunization with NP-OVA in alum. SMARTA mice were used as recipients to reduce the competition between transferred donor OT-II T cells and host T cells in response to NP-OVA challenge (22, 23, 26). B1–8i B cell receptor (BCR) is derived from the (4-hydroxy-3-nitrophenyl) acetyl (NP)-binding antibody B1-8i and the B1–8i B cells will receive help exclusively from the donor OT-II T cells after NP-OVA immunization in the SMARTA recipient mice (23, 27).

Whereas OT-II Bach2-cKO T cells and OT-II Bach2-WT T cells expanded with similar kinetics for 5 days after NP-OVA challenge, the response of OT-II Bach2-cKO T cells seemed to collapse by day 7 (Fig. 1C). At day 5 when the total donor OT-II T cell numbers were at the same levels, Bach2-deficient OT-II T cells contained significantly higher frequency and number of PD1⁺CXCR5⁺ Tfh cells than those of the OT-II Bach2-WT cells (Fig. 1D). Consistently, Bach2-deficient OT-II T cells also contained higher frequency of Bcl-6⁺CXCR5⁺ Tfh cells (Fig. 1E), suggesting that Bach2 is a negative regulator of Tfh cell differentiation. In agreement with the enhanced Tfh cell responses for the first 5 days, we found that both the frequency and number of germinal center (GC, CD38^loFAS⁺) Bi-8i B cells in the recipient mice that received OT-II Bach2-cKO T cells were also higher than those in mice that received OT-II Bach2-WT T cells (Fig. 1F), suggesting that Bach2-deficient Tfh cells do function like Tfh cells. Published studies have shown that Bach2-deletion causes more T cell death (14). By day 7, although the Tfh cell frequency of donor OT-II Bach2-cKO T cells appeared to be at the same level of OT-II Bach2-WT T cells (Supplemental Fig. 2A), which is likely due to the cell death of Bach2-deficient OT-II T cells (Fig. 1C), the Tfh cell number of OT-II Bach2-cKO T cells was much lower than that of OT-II Bach2-WT T cells (Supplemental Fig. 2A). And correspondingly, the GC B cell percentage and number were also lower (Supplemental Fig. 2B).

The experiments of mixed co-transfer of OT-II Bach2-cKO T cells and OT-II Bach2-WT T cells in the same recipient mice showed that Bach2-deficient T cells still contained higher frequency of Tfh cells (Fig. 1G), demonstrating that Bach2-mediated regulation of Tfh cell differentiation is cell-intrinsic.

To further verify the role of Bach2 in Tfh cell differentiation, in addition to the loss-of-function approach, we used the gain-of-function approach as well. The retroviral over-expression of Bach2 in wild-type OT-II T cells led to increased total donor T cell recovery but almost completely abolished the Tfh cell differentiation and the subsequent GC B cell responses after antigen challenge in the OT-II/NP-OVA model (Fig. 2A, B and Supplemental Fig. 2C). In a PR8-OVA influenza virus infection model in which the PR8 viruses express OVA (28), we also found that the retroviral over-expression of Bach2 drastically suppressed an otherwise very strong Tfh cell response of OT-II T cells to influenza virus infection (Fig. 2C). For transcriptional repression, Bach2 is dependent upon a DNA-binding basic leucine zipper region (14). We found that the over-expression of Bach2-ΔZIP which lacks the DNA-binding basic leucine zipper region, failed to suppress the Tfh cell differentiation (Fig. 2D), suggesting that the DNA binding of Bach2 is important for its regulation of Tfh cell development, though it is not clear how Bach2-ΔZIP may seem to induce enhanced Tfh cell differentiation. Taken together, our results demonstrate that Bach2 functions as a critical negative transcriptional regulator of Tfh cell differentiation.

Figure 2. Bach2 over-expression suppresses Tfh cell differentiation and Bach2 ZIP domain is important for the inhibition of Tfh cell differentiation.

(A-C) Wild-type naive OT-II T cells were activated in vitro and infected with control retrovirus (RV-Thy1.1) or retrovirus expressing Bach2 (RV-Bach2). (A-B) Retrovirally infected OT-II T cells were sorted and co-transferred with B1–8i B cells into the Ly5.1+SMARTA recipient mice followed by immunization with NP-OVA in alum 1 d later. The Tfh cell differentiation and the total number of splenic donor OT-II T cells (A), and the GC B cell differentiation and the total number of splenic GC B1–8i cells (B), were analyzed 7 d after immunization. (C) Retrovirally infected OT-II T cells were sorted and transferred into the Ly5.1+SMARTA recipient mice followed by infection with influenza virus PR8-OVA 1 d later. The Tfh cell differentiation of donor OT-II T cells in the mediastinal lymph nodes was analyzed 7 d after infection. (D) Wild-type naïve OT-II T cells were activated in vitro and infected with control retrovirus (RV-Thy1.1) and retrovirus expressing Bach2 (RV-Bach2) or Bach2 (ΔZIP) (RV-Bach2 (ΔZIP)). Retrovirally infected OT-II T cells were sorted and transferred into the Ly5.1⁺ SMARTA recipient mice followed by immunization with NP-OVA in alum 1 d later. PD-1⁺CXCR5+ Tfh and Bcl6⁺CXCR5+ Tfh cell percentage of donor OT-II T cells was analyzed 7 d later. Data in (A-D) are representative (or pooled results) of two independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001.

Bach2 deletion leads to CXCR5 upregulation before the induction of Ascl2

To understand the molecular mechanism underlying Bach2-mediated suppression of Tfh cell differentiation, we performed the transcriptome profiling by RNA sequencing (RNA-seq) in naive and day 3 OT-II Bach2-cKO and OT-II Bach2-WT T cells after NP-OVA challenge. We found 590 genes differentially expressed between day 3 Bach2-deficient and wild-type donor OT-II T cells (Supplemental Fig. 3A and Supplemental Table 1). When compared with the 145 Tfh-associated genes previously described (22), 33 of the 590 differentially expressed genes overlapped (Supplemental Fig. 3A). The signaling pathway analysis showed that the overall T cell activation, metabolism and oxidative responses were enhanced in the absence of Bach2 (Supplemental Fig. 3B).

Studies have shown that Bach2 suppresses the effector functions of CD4⁺ T cells (10, 12). We found that in the OT-II/NP-OVA model, the mRNA levels of Tbx21, Ifng and Gzmb were increased in the Bach2-deficient OT-II T cells (Fig. 3A and Supplemental Fig. 3D), suggesting that without Bach2, the CD4⁺ T cell effector function is enhanced. Interestingly, at day 3, an early time point of the OT-II T cell response to NP-OVA challenge, we found that the mRNA levels of CXCR5 already started to be higher in OT-II Bach2-cKO T cells than in wild-type control OT-II T cells (Fig. 3B). A higher proportion of activated Bach2-deficient OT-II T cells were found in the B cell follicles by day 4 (Fig. 3C). By gating on PD-1⁺CXCR5⁺ OT-II Tfh cells on day 5, the cell surface expression levels of CXCR5 of OT-II Bach2-cKO Tfh cells were higher than those of OT-II Bach2-WT Tfh cells (Fig. 3D). These results suggest that despite the enhanced effector functions, the deletion of Bach2 preferentially leads the activated CD4⁺ T cells towards Tfh cell differentiation and Bach2 may directly regulate CXCR5.

Figure 3. Bach2 deletion leads to enhanced CXCR5 expression before the induction of Ascl2.

Naive OT-II Bach2-WT or OT-II Bach2-cKO T cells were transferred into the Ly5.1⁺SMARTA mice followed by immunization with NP-OVA in alum. (A) Transcriptome analysis was performed in d 3 donor OT-II T cells. Scatterplot of the average signal of d 3 donor OT-II Bach2-cKO versus OT-II Bach2-WT T cells is shown. Expression of Th1 associated genes including Tbx21, Ifng, Prdm1, Gzmb and Tfh-associated genes including Ascl2, CXCR5 and Bcl6 is listed. The dash lines indicate gene expression change by 1.5 fold. Data shown in (A) are normalized from two replicates. (B) Real-time-PCR analysis of Ascl2 and CXCR5 mRNA in naïve, d 3 splenic donor OT-II T cells, and d 7 donor wild-type OT-II non-Tfh cells and Tfh cells. (C) Confocal microscopy of B cell follicles (IgD⁺ B cells and CD35⁺ follicular dendritic cells) and localization of donor (CD45.2⁺) OT-II T cells in the recipient mice 4 d after immunization. Scale bar, 100 μm. (D) Histogram of CXCR5 cell surface expression on d 5 donor OT-II Bach2-WT Tfh and OT-II Bach2-cKO Tfh cells. CXCR5 expression on endogenous T cells was used as control. (E) Bach2 binding to the promoter and two potential regulatory regions upstream of the murine CXCR5 locus by bioinformatics analysis of published dataset (GSE45975) (left panel). Diagrams of retroviral Thy1.1 reporter vectors containing CXCR5 promoter (CXCR5_p), CXCR5 promoter plus 5’ regulatory Site 1 region (CXCR5_1+P), and CXCR5 promoter plus 5’ regulatory Site 2 region (CXCR5_2+P) (right panel). (F) Naive Bach2-WT or Bach2-cKO (from Bach2^f/fCd4-Cre mice) T cells were activated in vitro and infected with retrovirus expressing CXCR5_P, CXCR5_1+P, or CXCR5_2+P, and the reporter activities were analyzed on d 4 by flow cytometry. Data in (B-F) are representative (or pooled results) of at least two independent experiments. ** p < 0.01, *** p < 0.001.

Ascl2 has been reported to initiate the Tfh cell differentiation and directly regulate CXCR5 expression (8). Surprisingly, we found that at the early time point of the response, the increased CXCR5 expression in both wild-type and Bach2-deficient activated CD4⁺ T cells seemed to occur in the absence of Ascl2 expression (Figs. 3B and Supplemental Fig. 3C). Studies of Bach2 global binding in the murine genome have been carried out (10). We performed bioinformatics analysis of this published dataset (GSE45975) and found that Bach2 binds to the promoter and two potential regulatory regions upstream of the murine CXCR5 locus (Fig. 3E, left panel), in which the Site 2 element has been reported to have a Blimp1-mediated suppression function on the CXCR5 promoter (29). After obtaining the murine CXCR5 promoter reporter construct (29), we cloned the Site 1 and Site 2 elements containing the Bach2 binding sites and performed the reporter assays to examine their functions in regulating the CXCR5 promoter activity (Fig. 3E, right panel). The CXCR5 reporter vector uses eGFP for infection indication, which is difficult to be separated from the YFP marker of OT-II Bach2-cKO T cells in staining. Thus, in the reporter assay experiments, we used the naive CD4⁺ T cells from the wild-type and Bach2^f/fCd4-Cre mice.

We found that in vitro, although there was no CXCR5 induction in the activated wild-type or Bach2-deficient CD4⁺ T cells (data not shown), a significant proportion of the activated CD4⁺ T cells were positive for the CXCR5 promoter reporter activities and no difference was observed between the wild-type and Bach2-deficient groups (Fig. 3F). The Site 2 element, 11 kb upstream of the transcription start site (TSS) of murine CXCR5, reduced the CXCR5 promoter reporter activities but in a Bach2-independent manner. The Site 1 element, 36 kb upstream of murine CXCR5 TSS, also helped suppress the promoter reporter activities (Fig. 3F). In the absence of Bach2, however, the Site 1 element lost its suppression function (Fig. 3F), suggesting that Bach2 likely suppresses CXCR5 up-regulation via its control of the Site 1 element.

CXCR5 over-expression does not rescue Bach2-mediated suppression of Tfh cell differentiation

To examine whether Bach2 may regulate Tfh cell differentiation by controlling genes other than CXCR5, we retrovirally over-expressed both CXCR5 and Bach2 in wild-type OT-II T cells (Fig. 4A). Bach2 over-expression alone in OT-II T cells almost completely suppressed the generation of PD-1⁺CXCR5⁺ Tfh cells (Fig. 2A and C). In the OT-II T cells over-expressing both CXCR5 and Bach2, the CXCR5 levels were restored to the levels of the OT-II T cells over-expressing CXCR5 alone (Fig. 4A). Thus, we wanted to determine whether the retroviral CXCR5 expression is sufficient to rescue Bach2-mediated suppression of Tfh cell differentiation.

Figure 4. CXCR5 over-expression does not rescue Bach2-mediated suppression of Tfh cell differentiation.

(A) Wild-type naive OT-II T cells were activated in vitro and infected with control retrovirus (RV-Thy1.1), or retrovirus expressing CXCR5 (RV-CXCR5) or CXCR5 plus Bach2 (RV-CXCR5+RV-Bach2). The CXCR5 expression was analyzed on d 4 by flow cytometry. (B-C) Wild-type naive OT-II T cells were activated and infected with different retroviruses. Subsequently, retrovirally infected T cells were sorted and co-transferred with B1–8i B cells into the Ly5.1⁺SMARTA recipient mice followed by immunization with NP-OVA in alum 1 d later. At d 7, Tfh cell differentiation of donor OT-II T cells and the GC B cell differentiation of donor B1–8i B cells were analyzed by flow cytometry (B), and the localization of donor OT-II T cells in the spleens of the recipient mice was examined by fluorescence confocal microscopy (C). Scale bar, 50 μm. (D) Naive OT-II Bach2-WT or OT-II Bach2-cKO T cells were transferred into the Ly5.1⁺SMARTA mice followed by immunization with NP-OVA in alum. Bcl6 and Prdm1 mRNAs in d 3 and d 4 splenic donor OT-II T cells were examined by real-time PCR analysis. Data in (A-D) are representative (or pooled results) of at least two independent experiments. ** p < 0.01, *** p < 0.001.

We found that the PD-1⁺CXCR5⁺ and Bcl6⁺CXCR5⁺ Tfh cell percentages of the OT-II T cells over-expressing both CXCR5 and Bach2 were still significantly lower than those of the OT-II T cells over-expressing CXCR5 only (Fig. 4B), suggesting that CXCR5 over-expression is not sufficient to rescue Bach2-mediated suppression of Tfh cell differentiation. Interestingly, whereas the Tfh cell percentage of the OT-II T cells over-expressing both CXCR5 and Bach2 was significantly higher than that of the control RV-Thy1.1 OT-II T cells, there was almost no GC B cell formation (Fig. 4B), which was also reflected by the almost exclusive T cell zone localization of activated OT-II T cells over-expressing CXCR5 and Bach2 (Fig. 4C). These results suggest that Bach2 likely regulates Tfh cell differentiation by controlling genes other than CXCR5 as well, and potentially cell migration is involved.

Studies have shown that Blimp1 is a direct target of Bach2 and Bach2 negatively regulates Blimp1 expression (13, 18, 19). We found that in day 3 Bach2-deficient OT-II T cells, the mRNA levels of Blimp1 were increased (Fig. 4D), and at the same time the mRNA levels of Bcl6 were also increased (Fig. 4D). The simultaneous increase of both Blimp1 and Bcl6 in the absence of Bach2 became more obvious in the day 4 Bach2-deficient OT-II T cells (Fig. 4D), suggesting that the Tfh cell differentiation of Bach2-deficient CD4⁺ T cells may be resistant to Blimp1-mediated inhibition.

Bach2 is important for Tfh cell memory

Studies have shown that Bach2 is important for CD8⁺ T cell memory (29). In our study, we found that the number of Bach2-deficient OT-II T cells decreased dramatically on day 7 (Fig. 1C), suggesting that Bach2 may regulate CD4⁺ T cell memory formation as well, including Tfh cell memory. To test this idea, we transferred OT-II Bach2-WT and OT-II Bach2-cKO T cells into the SMARTA recipient mice followed by NP-OVA immunization in alum; 30 days after the first antigen challenge, we re-challenged the recipient mice and examined the CD4⁺ T cell memory responses and Tfh cell differentiation.

We found that after the antigen re-challenge, the OT-II Bach2-cKO T cells still had higher percentages of PD1⁺CXCR5⁺ and Bcl6⁺CXCR5⁺ Tfh cells than did the OT-II Bach2-WT T cells (Fig. 5A). However, the numbers of total donor OT-II Bach2-cKO T cells, PD1⁺CXCR5⁺ Tfh cells and PD1⁻CXCR5⁻ non-Tfh cells, were all significantly lower than those of the OT-II Bach2-WT T cells (Fig. 5B). To confirm the role of Bach2 in CD4⁺ T cell memory, we also used the over-expression approach and found that the retroviral over-expression of Bach2 helped generate more robust OT-II T cell memory responses with continued suppressed Tfh cell differentiation (Fig. 5C and D). Taken together, these results suggest that whereas Bach2 still functions as a negative regulator of Tfh cell differentiation in memory CD4⁺ T cell responses, the deletion of Bach2 is overall detrimental to CD4⁺ T cell memory including Tfh cell memory.

Figure 5. Bach2 is important for CD4⁺ T cell memory.

(A, B) Naive OT-II Bach2-WT or OT-II Bach2-cKO T cells were transferred into the Ly5.1⁺SMARTA mice followed by immunization with NP-OVA in alum. 30 days later, the recipient mice were re-challenged by immunization with NP-OVA in PBS. At d 5 after re-challenge, the Tfh cell differentiation of donor OT-II T cells was analyzed by flow cytometry (A), and the numbers of total splenic donor OT-II T cells, OT-II Tfh cells and non-Tfh cells were calculated (B). (C, D) Wild-type naïve OT-II T cells were activated in vitro and infected with control retrovirus (RV-Thy1.1) or retrovirus expressing Bach2 (RV-Bach2). Retrovirally infected OT-II T cells were sorted and transferred into the Ly5.1⁺SMARTA recipient mice followed by NP-OVA immunization and re-challenge as in (A). At d 5 after re-challenge, the Tfh cell differentiation of donor OT-II T cells was analyzed by flow cytometry (C), and the numbers of total splenic donor OT-II T cells, OT-II Tfh cells and non-Tfh cells were calculated (D). Data in (A-D) are representative (or pooled results) of at least two independent experiments. * p < 0.05, *** p < 0.001.

Discussion

Previous studies have shown that Bach2 inhibits CD4⁺ T effector functions of multiple lineages and is critical for Treg homeostasis (10). Surprisingly, in our study, we have found that despite the enhanced effector functions of Bach2-deficient conventional CD4⁺ T cells, the deletion of Bach2 leads the conventional CD4⁺ T cells to a preferential Tfh cell differentiation. Bach2-deficient Tfh cells have increased CXCR5 expression levels, and the induction of CXCR5 expression occurs before the Ascl2 expression.

The regulation of CXCR5 expression after naïve CD4⁺ T cell activation is complicated. Although we have identified a regulatory element 36 kb upstream of the murine CXCR5 locus which suppresses the CXCR5 promoter activity in a Bach2-dependent manner in the reporter assays in vitro, neither wild-type nor Bach2-deficient OT-II T cells upregulated CXCR5 expression in vitro despite the reporter activities, suggesting that there are other mechanisms involved in regulating the CXCR5 expression in vivo. At the same time, the results also suggest that during the activation of naive CD4⁺ T cells, the CXCR5 promoter may have acquired enhanced basal activity which is suppressed by a Bach2-mediated regulation via the Site 1 negative regulatory element. The down-regulation of Bach2 seems to be necessary yet not sufficient for the CXCR5 expression. The mechanisms underlying the Bach2 down-regulation and the induction of CXCR5 expression before Ascl2 expression in activated CD4⁺ T cells are still not clear.

That CXCR5 over-expression alone is not sufficient to rescue Bach2-mediated suppression of Tfh cell differentiation suggests that there are other important genes regulated by Bach2 in Tfh cell development. Studies have shown that Bcl6 and Blimp1 antagonize each other in regulating Tfh cell differentiation and Blimp1 negatively regulates Tfh cell differentiation (10). It is very intriguing that in day 3 and day 4 Bach2-deficient CD4⁺ T cells after antigen challenge, both Bcl6 and Blimp1 expression levels are increased, indicating that the Tfh cell differentiation of Bach2-deficient CD4⁺ T cells may somehow be resistant to Blimp1-mediated suppression. Whether it is Bcl6 expression that is resistant to Blimp1-mediated suppression or Blimp1 may suppress Tfh cell differentiation via additional mechanisms (which presumably are abolished in the Bach2-deficient situation) is not clear. Further investigation is warranted for this interesting phenomenon.

In our study, we found that the impact of Bach2-deletion on CD4⁺ T cell memory is quite dominant, likely due to a cell survival mechanism that is particularly manifested at the stages after the peak of the primary response. Our loss-of-function and gain-of-function experiments show that Bach2 greatly regulates the magnitude of the CD4⁺ T cell memory responses including both non-Tfh and Tfh responses. Taken together, our study establishes Bach2 as a critical regulator of Tfh cell differentiation and CD4⁺ T cell memory including Tfh cell memory.

Key points.

• Bach2 deletion in CD4⁺ T cells results in preferential Tfh cell differentiation.

• The CXCR5 induction in the absence of Bach2 occurs before the upregulation of Ascl2.

• Bach2 is critical for CD4⁺ T cell memory including Tfh cell memory.