Diverging regulation of Bach2 protein and RNA expression determine cell fate in early B-cell response

Summary:

During the early phase of primary humoral responses, activated B cells can differentiate into different types of effector cells, dependent on B cell receptor affinity for antigen. However, the pivotal transcription factors governing these processes remain to be elucidated. Here, we show that transcription factor Bach2 protein in activated B cells is transiently induced by affinity-related signals and mTORC1-dependent translation to restrain their expansion and differentiation into plasma cells while promoting memory and germinal center (GC) B cell fates. Affinity-related signals also down-regulates Bach2 mRNA expression in activated B cells and their descendant memory B cells. Sustained and higher concentrations of Bach2 antagonize the GC fate. Repression of Bach2 in memory B cells predisposes their cell-fate choices upon memory recall. Our study reveals that differential dynamics of Bach2 protein and transcripts in activated B cells controls their cell-fate outcomes and imprints the fates of their descendant effector cells.

Introduction

After immunization with T cell-dependent (TD) antigens, antigen-specific B-cells are activated by the binding of antigens to B cell antigen receptor (BCR) and migrate to the periphery of B cell follicles and interfollicular zones of secondary lymphoid organs, where they form cognate interactions with T follicular helper (Tfh) cells within 1-3 days (Cyster and Allen, 2019; De Silva and Klein, 2015). Antigen binding alone initiates a cascade of signaling events that are necessary but not sufficient to induce B cell expansion and further differentiation (Turner et al., 2020). With the help of T cells, antigen-engaged activated B cells rapidly proliferate and differentiate into three types of effector cells: extrafollicular short-lived plasma cells (PCs) secreting low-affinity antibodies, follicular germinal center (GC) B-cells and early GC-independent memory B cells (Cyster and Allen, 2019; Dhenni and Phan, 2020; Kurosaki et al., 2015; Taylor et al., 2012; Taylor et al., 2015). GC-independent memory B cells are dominantly IgM⁺ and CD80⁻ population, which preferentially participate in secondary GCs upon secondary immunization (Gitlin et al., 2016; Pape et al., 2011; Taylor et al., 2012; Zuccarino-Catania et al., 2014). Understanding molecular pathways controlling these processes are critical for vaccine development, interpretation of vaccine efficacy and the treatment for autoimmune diseases (Elsner and Shlomchik, 2020).

At the pre-GC stage, the expansion of activated B cells and subsequent differentiation into diversified effector cells are largely controlled by the affinity of BCRs for their cognate antigens and require the help of T cells (Chan et al., 2009; Paus et al., 2006; Schwickert et al., 2011; Shih et al., 2002). High-affinity crosslinking drives rapid B cell expansion and predisposes them to differentiate towards PCs (Chan et al., 2009; Paus et al., 2006; Phan et al., 2006). B cells with intermediate BCR affinity preferentially enter the GC pathway, whereas clones with weaker antigen reactivity are supposed to be directed to the memory pools (Kaji et al., 2012; Taylor et al., 2015). Higher BCR affinity is likely to induce stronger BCR signals and subsequently stronger T cell help in the form of CD40L and cytokines, such as interleukin-4 (IL-4) and IL-21(Ise and Kurosaki, 2019; Victora and Nussenzweig, 2012). T cell help and CD40 engagement are limited to B cells that can present cognate antigens via MHCII, thus allowing for a dynamic affinity threshold to be imposed on antigen-binding B cells (Schwickert et al., 2011). Affinity-based signals have been demonstrated to regulate the expression and activity of key “fate-determining” transcription factors to orchestrate cell-fate choices. Either BCR or CD40 signals can induce the expression of the transcription factor IRF4, and the intensity of their strength scales with IRF4 concentrations (Luo et al., 2018; Sciammas et al., 2006). IRF4 is critically involved in the fate decision of activated B cells by regulating the expression of two counter-antagonistic cell fate determinants, Bcl6 and Blimp1 (encoded by Prdm1) (Ochiai et al., 2013; Sciammas et al., 2006). Similar to GC B cells, memory B cells originate from activated B cells with lower BCR affinity. However, the essential transcription factors that direct memory B cell differentiation versus other cell fates at the pre-GC stage remain poorly understood.

The transcription factor Bach2 is highly expressed within GC B cells and regulates the GC program via direct suppression of Prdm1 (Huang et al., 2014; Muto et al., 2010; Muto et al., 2004). Repression of Bach2 expression in GC-derived IgG1⁺ memory B cells has been shown to predispose them to differentiate into plasma cells upon memory recall (Kometani et al., 2013). Bach2 is also required for the formation of GC-dependent memory B cells (Shinnakasu et al., 2016). Lower-affinity light zone GC B cells, which are a memory-prone population, express higher levels of Bach2 transcripts (Shinnakasu et al., 2016). However, it remains unclear whether increased Bach2 transcripts in this population are an indicator of BCR affinity or actually contribute to selection into the memory pool. It is also unknown whether Bach2 protein is expressed by these cells in the similar model as its transcript. Paradoxically, GC B cells express more Bach2 transcripts than memory B cells (Wang et al., 2017). More recently, we showed that human lymphoma-derived TBL1XR1 mutations skew the humoral immune response towards producing memory B cells by increasing the transcriptional activity of Bach2 (Venturutti et al., 2020). These results suggest that precise Bach2 expression kinetics might influence the fate decision of GC B cells in the light zone. Finally, the precise functions of Bach2 in alternative cell-fate decisions during the early phase of primary humoral responses have never been explored. Herein, we explore affinity-based precise Bach2 protein and mRNA expression kinetics in various B subsets during this early phase and its impact on alternative cell-fate decisions.

Results

Bach2 restrains activated B cell expansion and PC differentiation

To trace the dynamics of very rare antigen-engaged B cells in vivo at the pre-GC stage, we performed adoptive transfer experiments using B1-8^hi B cells in which the VH186.2 heavy chain is knocked into the heavy chain locus and recognizes hapten 4-hydroxy-3-nitrophenylacetyl (NP) when paired with endogenous λ light chains (Shih et al., 2002). Transferred NP-specific B cells were activated and underwent rapid expansion in the first 3 days after immunization with NP-OVA (Figures S1A and S1B). CD38⁺GL7⁻ memory B cells and CD19⁺CD138⁺ plasmablasts could be detected as early as day 2, and the majority of transferred NP⁺ cells became CD38⁺GL7⁺ active precursors on day 3 (Figure S1C). Starting on day 3, the donor NP-specific B cells gave rise to bursts of various effector subsets comprising CD19^lo/−CD138⁺ antibody-secreting PCs, CD19⁺CD38⁻GL7⁺ GC B cells and CD19⁺CD38⁺GL7⁻ memory B cells (Figure S1C), as described previously (Taylor et al., 2012; Taylor et al., 2015). The identity of PCs and GC B cells was further confirmed by assessment of the lineage-specific transcription factors IRF4 and Bcl6, respectively (Figure S1D).

To analysis the roles of Bach2 on alternative cell-fate choices at the pre-GC stage, we analyzed the differentiation of transferred B1-8^hi wild-type and Bach2-deficient cells under competitive conditions (Figure 1A). Although donor non-NP-binding Bach2-deficient B cells showed a slight reduction in homing to the spleens, they had no competitive disadvantage when compared to their wild-type counterparts after immunization (Figure 1A). However, the ratio between Bach2-deficient and wild-type compartments among donor NP-specific B cells was significantly increased on days 2-3 after immunization (Figure 1A). Interestingly, this increase rapidly declined after day 3, and Bach2-deficient NP-specific B cells showed an obvious competitive disadvantage on day 7 (Figure 1A). After contact with cognate T cells, activated B cells undergo rapid expansion in the first 3 days after immunization, and then produce a burst of effector cell differentiation (Figure S1C). The BrdU pulse labeling experiments demonstrated that more Bach2-deficient NP-specific activated B cells were in the S phase on day 2 after immunization (Figure 1B). Although Bach2-deficient B cells were prone to die upon BCR activation in vitro, they exhibited activated caspase-3 expression similar to that of their wild-type counterparts in vivo on day 2 after immunization (Figure S2A). Bach2-deficient memory B cells and active precursors underwent more rounds of division than their wild-type counterparts on day 3, as indicated by increased CFSE dilution (Figure 1C), indicating that Bach2 constrains B cell proliferation prior to full differentiation into effector cells. We next asked whether Bach2 deficiency disrupts the differentiation of effector cells. Bach2 deficiency markedly accelerated the expansion and differentiation of CD19⁻CD138⁺ PCs on day 3, and this tendency obviously declined on day 7 (Figure 1D). This result is probably because early extrafollicular PCs are short-lived and die several days after birth. The generation of Bach2-deficient CD38⁺GL7⁻ memory and CD38⁺GL7⁺ active precursors tended to decrease on day 3 (Figure 1D). Strikingly, the Bach2-deficient compartment almost lost CD38⁺GL7^™ memory and CD38⁻GL7⁺ GC B cells and their precursors under competitive conditions on day 7 (Figure 1D). Moreover, Bach2 deficiency markedly promoted the generation of IRF4⁺ cells on day 3 and suppressed the formation of Bcl6⁺ cells on day 7 (Figure S2B). In a noncompetitive environment, donor Bach2-deficient NP⁺ cells generated a greater percentage of PCs with lower percentages of memory and GC B cells (Figure S2C). Therefore, Bach2 prevents expedited PC differentiation while facilitating memory and GC cell fate at the pre-GC stage.

Figure 1. Bach2 restrains activated B cell expansion and PC differentiation.

(A) Diagrammatic representation of the experimental protocol. Flow cytometric analysis of the ratio of Bach2^fl/fl relative to Bach2^+/+ compartments in donor NP⁻ and NP⁺ cells at indicated days and quantifications. (B) Flow cytometric analysis and quantifications of percent BrdU⁺ cells in donor NP-specific cells and activated B cells on day 2. (C) Flow cytometric analysis of CFSE dilution in active precursors and memory B cells on day 3 and quantifications. (D) Flow cytometric analysis of the ratio of Bach2^fl/fl relative to Bach2^+/+ compartments in indicated donor NP-specific effector subsets and quantifications. Data are means ± SEM and are representative of at least two independent experiments. ND, undetected. *P <0.05 and **P <0.01 (two-tailed t-test). See also Figures S1 and S2.

Bach2-deficient activated B cells enforce the transcriptional program of PCs

To gain mechanistic insight into the suppressive effects of Bach2 on PC differentiation, wild-type and Bach2-deficient GL7⁺CD38⁺ activated NP-specific B cells were sorted on day 2 after immunization and subjected to transcriptional profile analysis. In total, we detected 706 upregulated genes and 436 downregulated genes (FDR<0.01, fold change>2) in Bach2-deficient activated B cells compared to wild-type control cells (Figure 2A). Notably, PC-specific transcription factors, including Prdm1 and Irf4, were markedly upregulated in Bach2-deficient cells. Consistent with a previous report (Muto et al., 2004), Aicda was also downregulated in these cells. We selected representative genes that were upregulated in PCs (FDR<0.01, fold change>4) relative to memory or GC B cells for gene set enrichment analysis (GSEA). This analysis revealed that two PC-upregulated gene signatures were significantly overexpressed in Bach2-deficient activated B cells (Figure 2B). Gene ontology analysis further showed that Bach2-deficient activated B cells upregulated gene signatures associated with the endoplasmic reticulum (ER) remodeling and unfolded protein response (UPR) and downregulated gene sets related to B cell activation and proliferation (Figure S3A). In particular, the expression of IRE1α/XBP1, a key axis of the UPR pathway, was enhanced in Bach2-deficient activated B cells. The IRE1α-XBP1 axis is essential for ER expansion and plays a central role in the development and antibody production of PCs (Benhamron et al., 2014; Iwakoshi et al., 2003). A recent study showed that activated B cells up-regulate ER remodeling- and UPR-genes to prepare for antibody production and secretion when differentiating towards PCs (Gaudette et al., 2020). These results suggest that the transcriptional profile of Bach2-deficient activated B cells has switched towards that of PCs. By analyzing published BACH2 ChIP-seq data in human activated B cells (Hipp et al., 2017), we identified approximately 12,968 genomic regions bound by BACH2. We performed LiftOver analysis with human BACH2 binding peaks and the mouse genome and obtained 6,844 conserved Bach2 binding peaks. A total of 131 out of 706 upregulated genes, including Irf4 and Prdm1, in Bach2-deficient cells contained at least one Bach2 binding site within 5 kb of their transcription start site (TSS) (Figures 2C and S3B), indicating that they are direct targets of Bach2. Analysis of the enriched motif of these binding sites at 131 genes identified the Bach2 consensus binding motif, which embeds the canonical AP-1 motif (Figure 2C).

Figure 2. Bach2 directly represses the Irf4-Prdm1 axis in activated B cells.

(A) Volcano plots showing differential expression genes in Bach2^fl/fl versus Bach2^+/+ activated B cells. (B) GSEA of up-regulated genes in PCs relative to their expression in GC B cells or memory B cells. (C) The diagram showing Bach2 direct target genes which were up-regulated in Bach2-deficient activated B cells and contained at least one Bach2 binding site within 5 kb of TSS. Bottom panel displays overrepresented top sequence motif enriched at these Bach2-bound regions. (D) Flow cytometric analysis and quantifications of percent IRF4^hiBlimp1⁺ subset in B cells stimulated with 2.5 μg/ml α-IgM antibody for 72 h in the presence of the c-Fos inhibitor T-5224 (50 μM). (E) CTV dilution in IRF4^hiBlimp1⁺ and IRF4^lo Blimp1⁻ activated B cells as described in D. The graph depicts the division index of each genotype. Data are means ± SEM (D, E) and are representative of at least two independent experiments, ns, not significance and **P <0.01 (two-tailed t-test). See also Figure S3.

We and other reported that Bach2 directly binds to the promoter region of Prdm1 to repress its transcription in human and murine B cells (Huang et al., 2014; Ochiai et al., 2008). ChIP-QPCR confirmed the enrichment of Bach2 at the predicted Bach2-binding site at the promoter of Irf4 gene in activated murine B cells (Figure S3C). Moreover, up-regulation of Irf4 mRNA abundance occurred in activated Bach2-deficient B cells as early as 16 h after BCR ligation, a time-point before the onset of PCs differentiation and Blimp1 up-regulation (Figures S3D and S3E). These findings collectively suggest that Bach2 is capable to directly repress the transcription of Irf4 in B cells.

The c-Fos/AP-1 transcription factors are activated upon B cell activation, and required for B cell proliferation and PC differentiation in part by promoting IRF4 and Blimp1 expression (Ohkubo et al., 2005). Treatment with T-5224, a specific inhibitor of c-Fos/AP-1 transcription factors that inhibits DNA binding activity, significantly impaired PC differentiation and cell division in the absence of Bach2 (Figures 2D and 2E). We reason that Bach2 constrains B cell expansion and PC differentiation in part by antagonizing the transcription activity of c-Fos/AP-1 transcription factors.

Bach2 represses the Irf4-Prdm1 loop to enforce memory and GC B cell fates

Starting from day 3 after immunization, active precursors gave rise to burst of different types of effector subsets (Figure S1C). Bach2-deficient active precursors contained a higher proportion of Blimp1⁺IRF4^hi PC-prone cells than their wild-type counterparts on day 3 (Figure 3A). Furthermore, in adoptive transfer experiments Bach2-deficient active precursors displayed a higher propensity to differentiate into PCs at the expense of memory and GC B cells (Figure 3B). These results collectively suggest that Bach2 deficiency induces the generation of Blimp1⁺IRF4^hi PC-prone active precursors to prevent memory and GC B cells differentiation. Blimp1 and IRF4 forms a reciprocal transcription regulatory loop to initiate PC differentiation and both are essential for the generation of PCs (Sciammas et al., 2011; Sciammas et al., 2006; Shapiro-Shelef and Calame, 2005; Tellier et al., 2016). Reduction of either Blimp1 or IRF4 is able to damage the loop to inhibit PC differentiation. To examine whether Bach2 deficiency prevents memory and GC B cell fates by activating the Irf4-Prdm1 loop, we performed adoptive transfer using Bach2-deficient B1-8^hi cells expressing scramble (scr), shRNA against Prdm1 (shPrmd1) or Irf4 (shIf4). Either Prdm1 or Irf4 knockdown markedly reduced the fractions of PCs, accompanied with increased proportions of memory and GC B cells among donor NP-specific B cells (Figure 3C). Thus, Bach2 restrains the Irf4-Prdm1 loop to allow memory and GC B cell differentiation.

Figure 3. Bach2 represses the Irf4-Prdm1 axis to enforce memory and GC B cell fates.

The adoptive transfer experiment and immunization was described above (Figure 1A). (A) Flow cytometric analysis and quantifications of percent IRF4^hiBlimp1⁺ cells in indicated donor NP-specific populations on day 3 after immunization. (B) Donor Bach2^+/+ (CD45.2/CD45.1) and Bach2^fl/fl (CD45.2) NP-specific active precursors were isolated 3 days post-immunization and mixed at a ratio of 1:1, and adoptively transferred into wild-type mice (CD45.1⁺) immunized at the same time as the donor animals. Flow cytometric analysis and quantifications of the percentages of each compartment among indicated donor effector subsets. (C) Bach2-deficient B1-8^hi cells were infected with retrovirus virus expressing scr (scramble), shPrdm1 or shIrf4, and adoptively transferred into the recipients and immunized as described in Figure 1A. Flow cytometric analysis of and quantifications of indicated effector subsets in donor transfected NP⁺ cells. Data are means ± SEM and are representative of at least two independent experiments. *P <0.05 and **P <0.01 (two-tailed t-test).

BCR stimulation represses Bach2 transcription, but increases its protein abundance

Having demonstrated that the importance of Bach2 expression in the B cell-fate decision at the pre-GC stage, we sought to understand how Bach2 expression is regulated. Consistent with previous reports (Kometani et al., 2013; Shinnakasu et al., 2016), BCR stimulation reduced Bach2 mRNA abundance in B cells in a dose-dependent manner, and this effect occurred as early as 6 h after treatment (Figure 4A). To our surprise, BCR stimulation dramatically upregulated Bach2 protein expression 24 h after treatment (Figure 4B). IRF4 upregulation occurred as early as 3 h after treatment (Figure 4B). Intriguingly, upregulation of either Bach2 or IRF4 protein was proportional to the concentrations of anti-IgM antibody at 24 h after treatment (Figure 4C). Similar results were observed in B cells stimulated with anti-CD40 antibody, the Toll-like receptor 4 ligand lipopolysaccharide (LPS) and the Toll-like receptor 9 ligand CpG oligonucleotides (Figure 4D). Interestingly, coactivation of BCR and CD40 signaling synergistically induced Bach2 protein expression and downregulated its mRNA abundance (Figure 4D). These results indicate that Bach2 protein and mRNA levels are controlled by a range of B cell stimulators, which can result in opposite directions of regulation.

Figure 4. BCR stimulation represses Bach2 transcription, but increases its protein abundances.

(A) qPCR analysis of Bach2 mRNA abundance in splenic B cells treated with various concentrations of α-IgM antibody for the indicated times. No stim, no stimulation. (B-C) Immunoblot analysis of indicated protein in splenic B cells as described in A. (D) Immunoblotting and qPCR analysis of Bach2 protein and mRNA abundance in splenic B cells stimulated with various stimulators. All qPCR data were normalized within each sample to β-Actin and further normalized to the un-stimulated ice sample, which is set to 1 (A, D). (E) The illustration depicting ChIP-seq tracks of BACH2 and IRF4 at the Bach2 gene in human and mouse activated B cells, respectively, by analyzing published dataset (GSE102460 and GSE46607). (F) qPCR analysis of Bach2 mRNA abundance in indicated splenic B cells treated with α-IgM antibody (10 μg/ml) for 24 h. Data are means ± SEM (A, D, F) and are representative of at least two independent experiments. *P <0.05 and **P <0.01 (two-tailed t-test).

By analyzing published chromatin immunoprecipitation sequencing (ChIP-seq) data from our previous study and others (Hipp et al., 2017; Huang et al., 2014; Ochiai et al., 2013), we found that IRF4 and BACH2 bind to the Bach2 gene (Figure 4E). Upon BCR activation either Irf4- or Bach2-deficient B cells reduced Bach2 transcripts but to a lesser extent than did their wild-type counterparts (Figure 4F), suggesting that IRF4 and Bach2 directly suppress Bach2 transcription. Of note, to assess Bach2 transcription in Bach2-deficient cells, primers were designed to amplify the exon 1 of Bach2, which is not deleted in these cells. In conclusion, BCR stimulation increases Bach2 and IRF4 protein expression which in turn inhibits Bach2 transcription.

Differential expression of Bach2 mRNA in activated B cells and various effector subsets

To monitor the kinetics of Bach2 mRNA expression in vivo, we constructed a Bach2-GFP (Bach2^EGFP) mRNA reporter mouse (Figures S4A and S4B). On day 2 after immunization, the expression of the Bach2 mRNA reporter in memory and activated B cells remained largely unchanged, and its levels were obviously attenuated in CD19⁺CD138⁺ plasmablasts, compared to non-NP binding donor B cells (Figure 5A). On day 5, CD38⁺GL7⁻ memory B cells and CD38⁺GL7⁺ active precursors maintained lower but modest levels of Bach2 mRNA reporter than naïve NP⁻ B cells (Figure 5B). CD19⁻CD138⁺ PCs had almost lost expression of the Bach2 mRNA reporter, whereas CD38⁻GL7⁺ GC B cells had largely restored their expression to levels similar to those of naïve NP⁻ B cells (Figure 5B). The differential amounts of Bach2 transcripts in various effector subsets were further validated by analyzing the transcriptional profiles of the cells isolated 5 days post-immunization (Figure 5C).

Figure 5. Kinetics of Bach2 mRNA expression in activated B cells and various effector subsets.

B cells from the B1-8^hi Bach2^EGFP mRNA reporter mice were adoptively transferred into allotypically marked recipients, and then these recipients were immunized with NP-OVA (A-F). (A-B) Flow cytometric analysis of the Bach2^EGFP mRNA reporter in donor NP⁺ cells at indicated days after immunization and quantifications. (C) FPKM values of Bach2 in indicated cell populations based on RNA-seq data. (D) Flow cytometric analysis of CCR6 expression in various effector cells on day 5. (E) Identification of CD138⁺CD19⁺ plasmablasts, CD138⁻CCR6⁻CD19⁺ GC precursors, CD138⁻CCR6⁺CD19⁺ memory precursors in CD38⁺GL7⁺active precursors on day 3 and day 5 after immunization. (F) Flow cytometric plots showing Bach2 mRNA reporter expression in various precursors and quantifications. The relative MFI of Bach2 mRNA reporter was calculated after normalization to donor NP⁻ B cells (A-B, F). Data are means ± SEM and are representative of at least two independent experiments. *P <0.05 and **P <0.01 (two-tailed t-test). See also Figure S4.

CD19⁺CD38⁺GL7⁺ active precursors are heterogenous populations comprising various progenitors for PCs, memory and GC B cells (Taylor et al., 2012; Taylor et al., 2015). We next analyzed the kinetics of Bach2 mRNA expression in various precursor subsets. The chemokine receptor CCR6 marks memory B precursors in both mouse and human GCs (Suan et al., 2017a). Fully differentiated PCs and GC B cells shut down CCR6 expression on day 5 (Figure 5D), suggesting that they originate from CCR6⁻ active precursors. Based on CCR6 and CD138 expression, CD19⁺GL7⁺ CD38⁺ active precursors on day 3 and day 5 were classified into three subsets: CCR6⁻CD138⁺ plasmablasts, CCR6⁻CD138⁻ and CCR6⁺CD138⁻ (Figure 5E). CCR6⁻CD138⁻ active precursors did not express Bcl6 on day 3 after immunization, but on day 5, ~40% of them gained Bcl6 expression, a hallmark of GC B cells (Figure 5E). Bcl6 protein was not expressed by CCR6⁺CD138⁻ active precursors on either day 3 or day 5 (Figure 5E). These results collectively indicate that CCR6⁻CD138⁻ and CCR6⁺CD138⁻ populations can be considered GC and memory precursors, respectively. As expected, CCR6⁻CD138⁺ plasmablasts expressed the lowest amounts of Bach2 mRNA reporter (Figure 5F). Intriguingly, CCR6⁻CD138⁻ GC precursors expressed lower levels of Bach2 mRNA reporter than CCR6⁺CD138⁻ memory precursors on day 3 post-immunization; however, this expression feature was reversed on day 5 (Figure 5F). These results indicate that activated B cells with intermediate Bach2 transcript levels preferentially differentiate into early GC precursors, which gradually restore Bach2 transcription when differentiating into GC B cells.

Dynamics of Bach2 protein in activated B cells and various effector subsets

The above results indicate that BCR activation promotes Bach2 protein expression at the post-transcriptional levels (Figures 4A–D). The 5’ untranslated region (UTR) of Bach2 contains GC-rich complex secondary structures and has a predicted ΔG value of −235 (Figure S5A), suggesting a cap-dependent translational control of Bach2 protein. Antigen binding, as well as other B cell stimulators such as CD40 engagement, activates the extracellular-signal-regulated kinase (ERK) and phosphoinositide 3-kinase (PI3-kinase)/AKT pathways to stimulate cap-dependent translation via mechanistic target of rapamycin complex 1 (mTORC1) (Luo et al., 2018; Reth and Wienands, 1997). mTORC1 stimulates the p70S6K/ eIF4A axis to induce cap-complex formation and translation initiation, especially for genes harboring complex secondary structures in 5’ UTRs. BCR stimulation-induced Bach2 protein up-regulation was potently blocked by treatment of inhibitors for ERK, AKT or mTORC1 (Figure 6A). Silvestrol, a pharmacological inhibitor of eIF4A (Wolfe et al., 2014), exhibited the similar activity (Figure 6A). Interestingly, other B cell stimulators including CpG, CD40 and LPS up-regulated Bach2 protein levels via mTORC1 (Figure 6B). We further observed enhanced loading of Bach2 mRNA in the higher molecular weight polysomal fractions upon BCR activation, which was suppressed by silvestrol treatment (Figure 6C). To evaluate the direct contribution of the 5’ UTR of Bach2 to protein translation, we cloned it into a dual fluorescence retroviral reporter construct (Figure S5B). In this reporter construct, the mCherry gene was placed under the control of the MSCV promoter and fused to the 5’ UTR sequence of Bach2. The transcription of the internal control GFP was also controlled by the MSCV promoter, but its translation was controlled by the eIF4A-independent IRES element. B cells were infected with retrovirus expressing these reporter constructs, and the expression of GFP and mCherry was measured. After normalization to GFP, the Bach2 5’ UTR repressed the expression of mCherry protein by 4-fold (Figures S5C and S5D). As retroviral transduction requires stimulated B cells, infected B cells harbored active BCR signaling, partially unleashing the Bach2 5’ UTR. We observed that silvestrol treatment significantly diminished mCherry protein expression, but not GFP, in the presence of the Bach2 5’ UTR (Figures S5C and S5E). In contrast, silvestrol treatment had no effect on the protein abundance of either mCherry or GFP in the absence of the Bach2 5’ UTR (Figures S5C and S5E). These results demonstrate that the 5’ UTR of Bach2 impedes cap-dependent protein translation and is sensitive to eIF4A activity. Therefore, BCR stimulation up-regulates Bach2 protein abundance by activating the mTORC1/eIF4A axis-dependent translation.

Figure 6. Kinetics of Bach2 protein expression in activated B cells and various effector subsets.

(A) Immunoblot analysis of Bach2 protein in B cells treated with or without indicated inhibitor in the presence of 10 μg/ml α-IgM antibody for 24 h. (B) Immunoblot analysis of Bach2 protein in B cells treated as indicated for 24 h. (C) Polysome profiles of B cells treated as indicated for 12 h. The graph showing the ration of Bach2 transcript on pooled polysomal fractions relative to that on polysome-free fraction after normalization with β-Actin. (D-E) Flow cytometric analysis of phospho-S6 and Bach2^EGEP protein reporter in indicated subsets and quantifications. Data are means ± SEM and are representative of at least two independent experiments. **P <0.01 (two-tailed t-test). See also Figures S5 and S6.

BCR engagement and T cell help are transient, and mTORC1 activation and active Bach2 protein translation are supposed to limit to specific B cell subsets at the pre-GC stage. To test this notion, we firstly analyzed the mTORC1 activity in various B cell subsets. mTORC1 stimulates the phosphorylation of ribosomal protein S6 (phospho-S6) at several serine sites via p70S6K. S6 phosphorylation, which can be used to monitor the kinetics of mTORC1 activity in GC responses (Ersching et al., 2017). S6 phosphorylation was obviously up-regulated in both NP-binding activated and memory B cells on day 2 after immunization, although the latter increased less (Figure 6D). On day 5 memory B cells and active precursors reduced phospho-S6 expression to levels similar to those of naïve B cells (Figure 6D). S6 phosphorylation was continually and highly expressed in plasmablasts and PCs, but barely detected in GC B cells on day 5 (Figure 6D).

To examine the kinetics of Bach2 protein expression in vivo, we constructed a Bach2 protein reporter mouse strain (Figures S6A and S6B). Although naïve B cells contained high amounts of Bach2 mRNA, they expressed very low levels of Bach2 protein due to the lack of mTORC1 activation (Figure 6E). The Bach2 protein reporter was obviously up-regulated in both NP-binding activated and memory B cells on day 2 after immunization, although the latter increased less (Figure 6E). Of note, some plasmablasts displayed an increase of Bach2 protein possibly because they still expressed substantial levels of Bach2 transcripts (Figure 5A). On day 5 expression of the Bach2 protein reporter in memory B cells and memory precursors returned to the similar levels as those in naïve B cells (Figure 6E). Therefore, these results indicate the similar dynamics of the mTORC1 activity and Bach2 protein abundance in these cells, suggesting a critical role of mTORC1 activity in regulating Bach2 protein abundance. Unexpectedly, GC B cells and their precursors expressed high levels of Bach2 protein reporter (Figure 6E), although the mTORC1 activity is largely silenced in these cells, suggesting that the translation of Bach2 protein is independent of the mTORC1 activity in GC B cells.

Sustained and higher concentrations of Bach2 protein facilitates the generation of memory B cells while antagonizing the GC fate

Transient up-regulation of Bach2 protein in activated B cell favors memory and GC B cell fates (Figure 1). We next asked whether sustained Bach2 protein expression in B cells can disrupt their differentiation programs. To do so, congenically marked CD45.1/2 and CD45.2 wild-type B1-8^hi cells were infected with retrovirus expressing GFP and Bach2-IRES-GFP, respectively. Transfected cells were mixed and adoptively transferred into wild-type hosts (CD45.1), followed by NP-OVA immunization 24 h post-transfer (Figure 7A). Enforced Bach2 protein expression did not influence the expansion of either non-NP-binding or NP-binding B cells 3 days after immunization (Figure 7B). However, Bach2-overexpressing NP-binding B cells had an increased propensity to differentiate toward memory B cells at the expense of PC and GC B cells (Figure 7C). Transient IRF4 expression is sufficient to induce the GC program (Ochiai et al., 2013). We observed that Bach2-overexpressing NP-binding B cells significantly down-regulated IRF4 protein expression on day 3 (Figure 7D). Thus, sustained and higher concentrations of Bach2 protein terminate the PC and GC programs while promoting the generation of memory B cells, probably by suppressing IRF4 expression.

Figure 7. Bach2 concentration regulates memory B cell fate and memory recall.

(A) Diagrammatic representation of the experimental protocol. (B) Flow cytometric analysis and quantifications of the ratio of Bach2-GFP expressing cells relative to GFP expressing cells in donor NP⁻ and NP⁺ B cells at day 3 after immunization. (C) Flow cytometric analysis of percentages of indicated effector subsets in GFP or Bach2-GFP expressing NP⁺ donor cells. (D) Flow cytometric analysis and quantifications of IRF4 expression in GFP or Bach2-GFP expressing donor NP⁺ cells on day 3. (E) Diagrammatic representation of the experimental protocol. Flow cytometric analysis and quantifications of the ratio of Bach2^fl/fl relative to Bach2^+/+ compartment in each memory B cell subset. (F-G) Flow cytometric analysis and quantifications of the percentages of each compartment among indicated donor effector subsets on day 5 after antigen recall. Each circle represents one mouse. Data are means ± SEM and are representative of at least two independent experiments. *P <0.05 and **P <0.01 (two-tailed t-test). See also Figure S7.

Bach2 concentrations control the cell-fate of memory B cell upon recall

Our above results demonstrate that memory B cells down-regulate Bach2 transcripts and barely express Bach2 protein due to lack of mTORC1 activation. We next sought to explore the roles of Bach2 in the maintenance and recall of memory B cells. Induced Bach2 ablation did not significantly disrupt the total memory pool or in two functionally distinct memory subpopulations based on the expression of PD-L2 and CD80 (Zuccarino-Catania et al., 2014) (Figure 7E), indicating that Bach2 is dispensable for the maintenance of memory B cells. We next conducted secondary adoptive transfer experiments to investigate whether Bach2 expression in GC-independent memory B cells influences their cell-fate outcomes upon memory recall (Figure S7A). Donor Bach2-suffcient memory B cells formed equal numbers of PCs and GC B cells upon memory recall, and had greatly reduced capacity for memory re-entry (Figure 7F). Reporter assay showed that activation-induced cytidine deaminase (Aicda)-driven Cre can ablate gene in over 75% of memory B cells (Figure S7B), indicating Aicda-Cre can used to delete gene in memory B cells. Donor Bach2-defcient memory B cells generated higher frequencies of PCs and lower frequencies of GC B cells (Figure 7F). Moreover, they almost abolished the ability to return to the memory pool (Figure 7F). Interestingly, Bach2 overexpression facilitated memory B cells to differentiate towards GC B cells at the expense of near abolishment of PC differentiation (Figure 7G). Moreover, Bach2 overexpression forced memory B cells to enter the memory pool (Figure 7G), a characteristic of activated B cells. Therefore, Bach2 expression level influences the cell-fate of memory B cells upon antigen recall.

Discussion

BCR affinity-related signals establish a sophisticated transcriptional network governing the expansion of activated B cells and subsequent differentiation into diversified effector cells at the pre-GC stage. The key components of this transcriptional network controlling the PC and GC B cell fates have been partially identified (Cook et al., 2020; Dhenni and Phan, 2020; Sciammas et al., 2011; Suan et al., 2017b; Willis and Nutt, 2019). In this study we demonstrate that BCR affinity-related signals control Bach2 protein expression kinetics in activated B cells to precisely shape their cell-fate outcomes (Figure S7C). Bach2 serves as a key “fate-determining” transcription factor for memory and GC B cell fates at the pre-GC stage. Concomitantly, BCR affinity-related signals suppress Bach2 transcription in activated B cells and GC-independent memory B cells, predisposing their cell-fate outcomes during memory recall (Figure S7C). Our study demonstrates that how affinity-related signals control Bach2 expression kinetics in a manner which links to B cell-fate diversification in primary and second immune responses.

Bach2 mRNA contains a complex 5’ UTR, and its effective translation requires mTORC1 activation and eIF4A activity. BCR and CD40 signals transiently activate the mTORC1 and consequently promote the translation of Bach2 protein in activated B cells (Figure S7C). Higher-affinity activated B cells are predisposed to receive stronger BCR and CD40 signals when the availability of antigen and subsequent T cell help are limited (Schwickert et al., 2011), and are expected to generate more mTORC1 activity and Bach2 protein. This is evidenced that plasmablasts expressed the highest levels of active mTORC1 and Bach2 protein abundance, and memory B cells expressed the lowest amounts of them among all response B cells. Memory B cells and active precursors gradually reduce Bach2 protein expression when further differentiation due to diminished mTORC1 activation.

BCR and CD40 signaling coordinately drive the expansion of activated B cells and generate lineage-committed transcriptional program by inducing the expression and activity of a series of transcription factors, including Myc, IRF4, and Blimp1 (Calado et al., 2012; Dominguez-Sola et al., 2012; Luo et al., 2018; Ochiai et al., 2013; Sciammas et al., 2006). The Irf4-Prdm1 loop promotes the expansion of activated B cells and PC differentiation, and our study suggests that this loop can be directly repressed by Bach2. Moreover, in activated B cells IRF4 binds to the AP-1-IRF composite element (AICE) that also embeds canonical AP-1 motif (Ochiai et al., 2013). Bach2 also binds to AP-1 or AICE motif in activated B cells (Hipp et al., 2017) . Therefore, although Bach2 and IRF4 protein are concomitantly up-regulated in activated B cells, they are antagonistic transcription factors and play central but opposing roles for cell-fate decisions. Thus, transient up-regulation of Bach2 protein represents an important mechanism which partially counteracts proliferation of activated B cells and delay their terminal differentiation into PCs, providing a time window for initiating memory and GC B cell differentiation. RNA-seq reveals that Bach2-deficient activated B cells reduced the transcripts of IL-4R and IL-21R (date not shown). IL4 and IL21 signaling is critical for activated B cells to express Bcl6 protein and subsequently differentiate into GC B cells (Chevrier et al., 2017). Bach2 might strengthen IL4 and IL21 signaling in activated B cells to facilitate GC B cell fate.

Several studies have reported that either BCR or CD40 signals down-regulate Bach2 at the transcript levels (Kometani et al., 2013; Muto et al., 2004; Shinnakasu et al., 2016). Our data further demonstrate that activation of BCR and CD40 induces expression of Bach2 and IRF4 protein, which in turn suppress Bach2 transcription. Plasmablasts rapidly reduced Bach2 mRNA expression prior to activated B cells and memory B cells. Consistent with a previous report (Muto et al., 2004), mature PCs almost silence Bach2 mRNA expression, resulting in loss of Bach2 protein although they exhibit high mTORC1 activity. Early GC precursors receive stronger BCR and CD40 signals than memory counterparts, and expressed lower amounts of Bach2 transcript. However, early GC precursors gradually restored Bach2 transcript when further maturation. The underlying mechanism may be linked to up-regulation of Bcl6 when differentiation towards GC B cells (Shinnakasu et al., 2016). Our results explain the confusing observation that mature GC B cells express higher levels of Bach2 mRNA than GC-derived memory B cells (Wang et al., 2017). Restoration of Bach2 transcription in GC B cells appears to be critical for the GC programs. Memory B cells also gradually reduced, but not lost Bach2 transcript. Graded concentrations of Bach2 in GC-independent memory B cells controlled their cell-fate choice between PCs and GC B cells upon memory recall. Therefore, stimulatory history down-regulates expression of preexisting Bach2 mRNA in GC-independent memory B cells, predisposing them to generate optimal numbers of PCs and GC B cells upon memory recall. Therefore, at the pre-GC stage, signaling-induced repression of Bach2 transcript in memory B cells, but not loss, optimizes Bach2 protein expression and cell-fate choices in second immune responses.

In conclusion, our study identifies translational control of Bach2 as a key mechanism underlying affinity-based B cell fate decisions, and highlights a crucial role of Bach2 protein expression kinetics in controlling memory and GC B cell fates at the pre-GC stage. Affinity-based signals up-regulate Bach2 and IRF4 protein levels to suppress Bach2 transcription in activated B cells and their descendant memory B cells, predisposing the optimal fate choice upon memory recall. Therefore, affinity-related signals induce differential dynamics of Bach2 protein and mRNA expression at the pre-GC stage via distinct mechanisms, and controls B cell fate specification at separate phases of humoral responses.

Limitations of the Study

The main limitation of this study is the precise regulation of IRF4 by Bach2. Bach2 appears to be dispensable for repression of IRF4 in IRF4^low activated B cells. However, Bach2 seems to be critical to suppress the transition of IRF4^low activated B cells to IRF4^high PC-committed B cells via direct repression of Irf4 transcription. IRF4 and Bimp1 form the reciprocal transcription regulatory loop to drive PC differentiation. It is possible that Bach2 indirectly regulates IRF4 via suppression of Blimp1. It would be of great interest to investigate the effect of Bach2 on IRF4 in the absence of Blimp1.

STAR*METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chuanxin Huang (huangcx@shsmu.edu.cn).

Materials availability

All unique reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability

• RNA-seq data in this study have been deposited in the GEO repository (GSE180277 and GSE180460). They are publicly available as of the date of publication.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the Lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice

All mice were crossed on C57BL/6 background. Bach2^fl/fl mice have been previously described (Zhang et al., 2019). B6 CD45.1 (B6.SJL-Ptprc^a Pepc^b/BoyJ), B6 CD45.2 (C57BL/6J), Irf4^fl/fl (B6.129S1-Irf4^tm1Rdf/J) Rosa26^tdTomato, Rosa26-CreER (B6.129-Gt(ROSA)26Sor^{tm1(cre/ERT2)Tyj}/J). Aicda^cre (B6.129P2-Aicda^tm1(cre)Mnz/J), OT- II (B6.CgTg(TcraTcrb)425Cbn/J), MD4 (C57BL/6-Tg(IghelMD4)4Ccg/J) and B1-8^hi (B6.129P2-Ptrpc^aIgh^tm1Mnz/J) mice were from Jackson Laboratory. The generation of Bach2^EGFP mRNA and protein reporter mice was described in Figures S4 and S6. All mice were housed in specific pathogen-free (SPF) facilities. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of shanghai Jiao Tong University, School of Medicine. Both female and male mice at 8–12 weeks of age were used in the experiments.

Cell lines

Phoenix-Eco cells (ATCC, CRL-3214) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS), 100 unit/mL penicillin and 100 mg/mL streptomycin. Cells are mycoplasma free.

METHOD DETAILS

Adoptive transfers and immunization

Spleens were homogenized by filtering through a 70 μ m cell strainer and red blood cells were lysed with ACK buffer. Resting B and CD4⁺ T cells were purified by negative magnetic-activated cell sorting (MACS) using the mouse B cell isolation Kit and mouse CD4⁺ T cell isolation kit, respectively. To study the early TD response, −5 × 10⁶ B1-8^hi B cells (~5 × 10⁵ Igλ NP-specific B cells) were transferred intravenously into recipient mice, along with 2 × 10⁵ OT- II CD4⁺ T cells. The recipients were immunized intraperitoneally with 100μg NP(16)-OVA plus 15μg LPS precipitated with Imject alum 24h post-transfer.

To trace the cell-fates of active precursors, NP-specific IgD⁻CD19⁺GF7⁺CD38⁺ active precursors were FACS-sorted from the animals transferred as described above and immunized with NP-OVA for 3 days. Equal numbers of allotypically marked Bach2-sufficient and deficient active precursor B cells (6×10³ in total) were transferred into the recipients which received 2 × 10⁵ OT- II CD4⁺ T cells, and were immunized with NP-OVA at the same day as the donor animals. The recipients were sacrificed for analysis on day 4 after transfer of active precursor cells.

For memory B cell maintenance, the adoptive transfer experiments were performed as described above. Six weeks later, deletion of loxP-flanked Bach2 alleles was induced by oral administration (p.o.) of 2 mg tamoxifen daily for 3 consecutive days. 7 days post-administration, mice were scarified for analysis of memory B cells.

For memory recall, IgD⁻CD19⁺GF7⁻CD38⁺ NP-specific memory B cells were FACS-sorted from the animals transferred as described above and immunized with NP-OVA for 5 days. Sorted memory B cells were transferred into the recipients which were primed with 50μg OVA precipitated with Imject alum 2-4 weeks ago. 24h later, the recipients were boosted with NP(16)- OVA alone, and sacrificed on day 5 post-immunization.

In vitro B cell culture and stimulation

Purified B cells were plated at a concentration of 1×10⁶ cells per 400 μl in round-bottom 48-well plates in complete RPMI media with stimuli for 1-3 d. B cells were stimulated with 10μg/ml goat anti-mouse IgM F(ab’)2, 10μg/ml anti-CD40, 5μg/ml LPS, 0.2 μM CpG, 10μM U0126-EtOH, 2μM MK-2206, 100ng/ml Rapmaycin, 50μM T-5224, 10nM Silvestrol as indicated in Figure legends. In some experiments, anti-IgM and anti-CD40 were used at different concentrations. To monitor B cell division, B cells were labeled using CFSE or CTV (Invitrogen) according to the manufacture’s protocol, and then transferred to complete medium for culture.

Flow cytometry analysis

Splenocytes were harvested into single-cell suspension as described above. Cells were first blocked with anti-CD 16/32(BD Bioscience), and stained with LIVE/DEAD™ Fixable Violet Dead Cell Stain Kit according to the manufacturer’s instructions. Surface staining was performed in the PBS containing 2% fetal bovine serum with fluorescence-tagged antibodies (KEY RESOURCE TABLE). After surface staining, cells were fixed and permeabilized with Foxp3/transcription factor staining buffer set to detect IRF4, Blimp1, Bc16. All the flow cytometry analysis was performed on a LSRFortessa X-20 (BD), and data were analyzed with FlowJo software (Tree Star).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti- CD45R/B220	BD Biosciences	Cat# 553092, RRID:AB_398531
anti- CD45R/B220	BD Biosciences	Cat# 553087, RRID:AB_394618
anti- CD45R/B220	BD Biosciences	Cat# 561878, RRID:AB_394619
anti- CD45R/B220	BD Biosciences	Cat# 561881, RRID:AB_394458
anti- CD45R/B220	BioLegend	Cat# 103241, RRID:AB_11204069
anti- CD19	BioLegend	Cat# 115543, RRID:AB_11218994
anti- IgD- Biotin	Thermo Fisher Scientific	Cat# 13-5993-82, RRID:AB_466860
anti- IgD	Thermo Fisher Scientific	Cat# 48-5993-82, RRID:AB_1272202
anti- IgD	BioLegend	Cat# 405715, RRID:AB_10660304
anti- CD138 (Syndecan-1)	BioLegend	Cat# 142514, RRID:AB 2562198
anti- CD138	BD Biosciences	Cat# 561070, RRID: AB_395000
anti- CD79b (Igβ)	BioLegend	Cat# 132808, RRID:AB_2632916
anti- CD38	Thermo Fisher Scientific	Cat# 17-0381-81, RRID:AB_469381
anti- CD38	BioLegend	Cat# 102722, RRID:AB_2563333
anti- CD38	BioLegend	Cat# 102718, RRID:AB_2275531
anti- GL7 (T and B cell Activation Marker)	BioLegend	Cat# 144614, RRID: AB_2563292
anti- GL7 (T and B cell Activation Marker)	BioLegend	Cat# 144606, RRID:AB_2562185
anti- GL7 (T and B cell Activation Marker)	BioLegend	Cat# 144603, RRID:AB_2561696
anti- CD80 (B7-1)	Thermo Fisher Scientific	Cat# 17-0801-81, RRID:AB_469416
anti- CD273 (B7-DC, PD-L2)	BioLegend	Cat# 107217, RRID:AB_2728125
anti- CD196 (CCR6)	BioLegend	Cat# 129813, RRID:AB 1877148
anti- CD45.1	Thermo Fisher Scientific	Cat# 11-0453-81, RRID:AB 465057
anti- CD45.1	Thermo Fisher Scientific	Cat# 45-0453-82, RRID: AB_1107003
anti- CD45.1	BioLegend	Cat# 110737, RRID:AB_11204076
anti- CD45.2	Thermo Fisher Scientific	Cat# 56-0454-81, RRID:AB 657753
anti- CD45.2	Thermo Fisher Scientific	Cat# 45-0454-82, RRID:AB_953590
anti- CD45.2	BioLegend	Cat# 109841, RRID:AB_2563485
anti- CD45.2	Thermo Fisher Scientific	Cat# 11-0454-81, RRID:AB_465060
anti- CD4	BioLegend	Cat# 100449, RRID:AB_2564587
anti- CD4	Thermo Fisher Scientific	Cat# 17-0041-82, RRID:AB_469320
anti- IRF4	BioLegend	Cat# 646406, RRID:AB_2563267
anti- IRF4	BioLegend	Cat# 646404, RRID:AB_2563005
anti- IRF4	BioLegend	Cat# 646414, RRID:AB_2728480
anti- BrdU	BioLegend	Cat# 339808, RRID:AB_10895898
anti- Bcl6	BD Biosciences	Cat# 563363, RRID:AB_2738159
anti- Blimp1/Prdm1	BD Biosciences	Cat# 563643, RRID:AB 2738342
Phospho-Akt (Ser473) (D9E) Rabbit mAb	Cell Signaling Technologies	Cat# 4060S, RRID:AB_2315049
Phospho-S6 Ribosomal Protein (Ser235/236) (D57.2.2E) Rabbit mAb	Cell Signaling Technologies	Cat# 4858T, RRID:AB_916156
Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody	Thermo Fisher Scientific	Cat# A32733, RRID:AB_2633282
Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody	Thermo Fisher Scientific	Cat# A11034, RRID:AB_2576217
anti-β-Actin	Sigma-Aldrich	Cat# A5441, RRID:AB_476744
IRF-4 (D9P5H) Rabbit mAb	Cell Signaling Technologies	Cat# 15106S, RRID:AB_2798709
HRP-labeled Goat Anti-Mouse IgG (H+L) antibody	Beyotime	Cat# A0216, RRID:AB_2860575
HRP-labeled Goat Anti-Rabbit IgG (H+L) antibody	Beyotime	Cat# A0208, RRID:AB_2892644
anti-Bach2 polyclonal antibody	Huang et al. (2014)	N/A
Chemicals, Peptides, and Recombinant Proteins
AF647 sreptavidin	BioLegend	Cat# 405237
PE-CY7 sreptavidin	BioLegend	Cat# 405206
Fixable Viability Stain 700	BD Biosciences	Cat# 564997
Fixable Viability Stain 780	BD Biosciences	Cat# 565388
Imject™ Alum Adjuvant	Thermo Fisher Scientific	Cat# 77161
NIP(16)-BSA-Biotin	Biosearch Technologies	Cat# N-1027
NP(29)-PE	Biosearch Technologies	Cat# N-5070
NP(16)-OVAL(Ovalbumin)	Biosearch Technologies	Cat# N-5051
Ovalbumin(Oval)	Biosearch Technologies	Cat# O-1000
LPS(055:B5)	Sigma	Cat# L2880
Tamoxifen	Sigma	Cat# T5648
Polybrene Transfection Reagent	Millipore	Cat# TR-1003-G
BRDU	Sigma	Cat# B5002
Anti-CD180	BD Biosciences	Cat# 562191
Anti-CD40	eBioscience	Cat# 16-0402-38
Affinipure F(ab’)2 Fragment Goat anti-mouse IgM	Jackson ImmunoResearch	Cat# 115-006-020
CPG	InvivoGen	Cat# tlrl-1826
U0126-EtOH	selleck	Cat# S1102
MK-2206 2HCL	selleck	Cat# S1078
Silvestrol	Biovision	Cat# B2417
DTT	VWR	Cat# 0281
DNase I	Qiagen	Cat# 79256
Murine RNase Inhibitor	Vazyme	Cat# DD4102
cycloheximide	selleck	Cat# S7418
Rapamycin	selleck	Cat# S1039
T-5224	selleck	Cat# S8966
Critical Commercial Assays
CaspGLow™ Fluorescein Active Caspase-3 Staining Kit	Biovision	Cat# K183
eBioscience™ Foxp3 / Transcription Factor Staining Buffer Set	ThermoFisher Scientific	Cat# 00-5523-00
Phosflow™ Fix Buffer I	BD Bioscience	Cat# 557870
Immobilon Western Chemiluminescent HRP Substrate	Millipore	Cat# WBKLS0500
EasySep™ Mouse CD4+ T Cell Isolation Kit	STEMCELL Technologies	Cat#19852
B Cell Isolation Kit	Miltenyi Biotec	Cat# 130-090-862
CellTrace CFSE Cell Proliferation Kit	Thermo Fisher Scientific	Cat# C34554
APC-Brdu kit	BD Biosciences	Cat# 552598
RNA Clean & Concentrato-5	Zymo Research	Cat# R1015
Deposited Data
Raw data files for RNA-seq of B subsets	This paper	GSE180460
Raw data files for RNA-seq of activated B	This paper	GSE180277
Datasets for BACH2 Chip-seq	Hipp et al. (2017)	GSE102460
Datasets for IRF4 Chip-seq	Ochiai et al. (2013)	GSE46607
Experimental Models: Cell Lines
Phoenix-Eco cell line	ATCC	Cat# CRL-3214, RRID:CVCL_H717
Experimental Models: Organisms/Strains
Mouse: B6 CD45.2: C57BL/6J	The Jackson Laboratory	RRID:IMSR_JAX:000664
Mouse: B6 CD45.1: B6.SJL-Ptprca Pepcb/BoyJ	The Jackson Laboratory	RRID:IMSR_JAX:002014
Mouse: AicdaCre: B6.129P2-Aicdatm1(cre)Mnz/J	The Jackson Laboratory	RRID:IMSR_JAX:007770
Mouse: Rosa26-CreER: B6.129-Gt(ROSA)26Sortm1(cre/ERT2)Tyj/J	The Jackson Laboratory	RRID:IMSR_JAX:008463
Mouse: B1-8hi: B6.129P2-Ptrpcalghtm1Mnz/J	The Jackson Laboratory	RRID:IMSR_JAX:007594
Mouse: MD4: C57BL/6-Tg(IghelMD4)4Ccg/J	The Jackson Laboratory	RRID:IMSR_JAX:002595
Mouse: Rosa26Tdtomato: B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J	The Jackson Laboratory	RRID:IMSR_JAX:007914
Mouse: OT-II : B6.CgTg(TcraTcrb)425Cbn/J	The Jackson Laboratory	RRID:IMSR_JAX:4194
Mouse: Irf4fl/fl: B6.129S1-Irf4tm1Rdf/J	The Jackson Laboratory	RRID:IMSR_JAX:009380
Mouse: Bach2fl/fl	Zhang et al (2019)	N/A
Mouse: Bach2EGFP RNA reporter mice	This paper	N/A
Mouse: Bach2EGFP protein reporter mice	This paper	N/A
Oligonucleotides
Primers for RT-qPCR, see Table S1	This paper	N/A
Primers for CHIP-qPCR, see Table S1	This paper	N/A
Primers for cloning, see Table S1	This paper	N/A
Recombinant DNA
pSIREN-RetroQ-dsRED	Clontech	Cat# 632455
MIGR1-MSCV retro-expression vector	Addgene	Plasmid #27490
Mouse Bach2 sequence-verified cDNA	This Paper	N/A
Mouse Bach2 5’ UTR sequence-verified cDNA	This Paper	N/A
Software and Algorithms
FlowJo	FlowJo LLC	https://www.flowjo.com/solutions/flowjo/downloads
Graphpad	Prism Graphpad	https://www.graphpad.com/

Construction of plasmids

Expressing construct for mouse Bach2 was generated by inserting cDNA encoding mouse Bach2 into retroviral Migr1-GFP or Migr1-Thy1.1 vector. cDNA encoding mCherry was cloned into the Migr1-GFP vector to generate Migr1-mCherry-GFP (control reporter). To generate the Bach2 5’UTR reporter, cDNA encompassing the Bach2 5’UTR was synthesized, and cloned into upstream of mCherry of control reporter. The sequence of cloning primers was found in KEY RESOURCE TABLE. Short hairpin RNA targeting Prdm1 sh(Prdm1), Irf4 (shIrf4) or scramble (scr) was cloned into pSIREN-RetroQ-DsRed (Clontech). The targeting sequences were found in KEY RESOURCE TABLE.

Polysome profiling

B cell lysate was prepared using polysome lysis buffer (20mM Tris-HCl pH 7.4, 150mM NaCl, 5 mM MgCl2, 1mM DTT, 100 μg/ml cycloheximide, 1% Triton X-100, 25 U/ml DNase I , 20 U/ml RNase inhibition). Cell lysate was cleared by centrifugation at 20,000×g at 4 °C for 10 min. Clarified lysate was loaded onto a 5–50% linear sucrose gradient (sucrose in 20mM Tris-HCl pH 7.4, 150mM NaCl, 5 mM MgCl2, 1mM DTT, 100 μg/ml cycloheximide, 20 U/ml RNase inhibition) and centrifuged for 2 h in a SW60Ti rotor (Beckman) at 36000 rpm at 4°C. During fractionation using a Gradient fractionator (Biocomp) the UV profile (260 nm) was measured. Obtained fractions were further purified using RNA Clean & Concentrator™-5 (Zymo Research) and then reverse-transcribed for qPCR analysis.

Retrovirus production and transduction

Virus were prepared by transfecting Phoenix cells (ATCC) with retroviral vectors. Naïve B cells were stimulated with anti-CD180 (0.25 μg/ml) for 24h, then centrifuged at 2000rpm for 1.5 h at 32°C with retroviruses in the presence of polybrene (8μg/ml). and cultured overnight for the second infection as one day before. Transduced cells were then cultured for analysis or transferred immediately into the recipients via tail vein injection.

Computational prediction of RNA structure

The sequence of the 5’UTR of mouse Bach2 were obtained from the Ensembl Genome Browser. The most abundant transcript was selected for predicting. Prediction of the Bach2 mRNA 5’ UTR was performed using the Mfold program (http://www.unafold.org/).

RNA sequencing and GSEA

A total of ~1 × 10⁴ different B cell subset was sorted using a FACSAria III cell sorter (BD Biosciences) at the core facility of Shanghai Institute of Immunology. cDNA was synthesized and amplified using the SMART-Seq v4 Ultra Low input RNA Kit for Illumina Sequencing. The library for mRNA sequencing was generated using TurePrep DNA Library Prep Kit V2 for Illumnia and sequenced using Illumnia NovaSeq 6000 sequencer. The mapping rate was 96% overall across all the samples in the dataset. HTSeq was used to quantify the gene expression counts from HISAT2 alignment files. Differential expression analysis was performed on the count data using R package DESeq2. P values obtained from multiple tests were adjusted using Benjamini-Hochberg correction. Significantly differentially expressed genes are defined by a Benjamini-Hochberg corrected P value < 0.01 and fold change > 2. Gene set enrichment analysis (GSEA) software was provided by Broad Institute and was used for analysis accordingly to instructions.

Immunoblot analysis

Cells were lysed in 1×SDS lysis buffer and denatured at 95 °C for 10 min. Protein lysates then were resolved in 8–10% Tris-Glycine gels. Samples were incubated with primary antibodies overnight at 4° C. Horseradish peroxidase–conjugated secondary antibodies and ECL reagent (Millipore) were used for detection. Each western blotting was repeated for three times and the representative results were shown.

Real-time PCR

MACS-purified B cells were cultured at 37 °C with varying stimuli and harvested into TRIzol (Invitrogen), then stored at −80° C. RNA was extracted via phenol phase separation and then reverse-transcribed with PrimeScript RT reagent Kit (Takara). Quantitative real-time PCR was carried out with TB Green® Premix Ex Taq™ (Takara). All quantitative data were normalized to the amount of β-Actin. Gene specific primers are listed in the supplementary information (KEY RESOURCE TABLE).

QUANTIFICATION AND STATISTICAL ANALYSIS

The number of mice per group, number of replicates and the nature of error bars are indicated in the legend of each figure. Center bars always indicate mean. Statistical significance was determined with Graphpad Prism Version 7.0 using the tests indicated in each figure legend. P value < 0.05 is considered statistically significant, and the level of significance was indicated as *P < 0.05 and **P < 0.01. P-values >0.05 are considered to be not significant (ns).