Differentiation of B lymphocytes into Ab-secreting plasmablasts and plasma cells is Ag driven. The interaction of Ag with the membrane-bound Ab of the BCR is critical in determining which clones enter the plasma cell response. However, not much is known about the coupling between BCR activation and the shift in transcription factor network from that of a B cell to that of ASC differentiation. Our genome-wide analysis shows that Ab-secreting cell differentiation of mouse B cells is induced by BCR activation through very fast regulatory events from the BCR. We identify activation of IFN regulatory factor-4 and down-regulation of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B gene expression as immediate early events. Furthermore, the transcription factor E2A is required for the rapid key down-regulations after BCR activation, and the Ca2+ sensor protein calmodulin has the corresponding regulatory effect as BCR activation. Moreover, mutants in the calmodulin binding site of E2A show that Ca2+ signaling through calmodulin inhibition of E2A is essential for the rapid down-regulation of immediate early genes after BCR activation in initiation of plasma cell differentiation.
When encountering Ags, B lymphocytes can adapt to produce a highly specific and potent Ab response. The affinity maturation of Abs and differentiation of B cells into Ab-secreting cells (ASC)3 are crucial components of the immune response. B cells that successfully respond to Ag can differentiate into plasmablasts (short-lived proliferating ASC), into long-lived noncycling Ab-secreting plasma cells, and into memory B cells that can rapidly differentiate into ASC after re-exposure to Ag (1). It is well established that plasma cell development is Ag driven (1, 2, 3): signaling from Ag receptors (BCRs) with their membrane-bound Abs plays a key role in the start of ASC differentiation. The BCR is a fundamental determinant of early plasma cell differentiation (4), and the strength of the BCR interaction with Ag plays a critical role in determining which clones enter both early extrafollicular plasma cell differentiation and plasma cell differentiation after successful affinity maturation in the germinal center (4, 5, 6). Only germinal center B cells that have acquired high affinity for the immunizing Ag form plasma cells. Thereby, only cells that can produce optimal Ab specificities become plasma cells, which ensures effective immunity (4, 6).
The developmental program leading to ASC differentiation from B cells is controlled by a gene-regulatory network in which transcriptional repression plays a key role (1). A small group of transcription factors expressed in the B cells, including Pax5, Bcl-6, MITF, Ets-1, and BACH2, are repressors of the ASC differentiation program (1, 3, 7). Pax5 is essential to establish and maintain B cell identity. It activates many B lineage-specific genes and simultaneously represses many genes associated with stem cell and non-B lineage programs as well as a number of genes involved in ASC differentiation, including Prdm1, the gene encoding B lymphocyte-induced maturation protein 1 (Blimp1), and Xbp-1. Therefore, Pax5 must be repressed to allow ASC differentiation (1, 2, 3). Bcl-6, Ets-1, Spi-B, and BACH2 are also suppressors of ASC development, in part by repressing Blimp1 and Xbp-1 (1, 3, 7, 8). In addition, MITF represses IFN regulatory factor (IRF)-4, another key regulator of the ASC development (1, 3, 9). In ASC, however, IRF-4 and Blimp1 inhibit Pax5 and Bcl-6 (1, 3, 9, 10). The Fli-1 transcription factor has recently been implicated in development and maintenance of B cells (11). The existence of these two different gene-regulatory transcription factor networks based on mutual repression has been established mainly by studies that examine the effects of overexpression or loss of expression of the different transcription factors. However, not much is known about the coupling between BCR activation and the shift in transcription factor network from that of a B cell to that of ASC differentiation.
In this study, we report that ASC differentiation of mouse B cells is induced by BCR activation through very fast regulatory events from the BCR. Primary regulatory events were identified as inhibition of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B gene expression as well as the activation of IRF-4 gene expression. We further show that Ca2+ signaling through inhibition of E2A by the Ca2+ sensor calmodulin (CaM) is essential for the rapid down-regulation of immediate early genes after BCR activation in initiation of plasma cell differentiation.
Materials and Methods
Activation of B lymphocytes from mouse spleens
Primary B lymphocytes were purified from mouse spleens and maintained as previously described (12), and stimulated with LPS (Calbiochem), IL-4 (PeproTech), and/or CD40L (R&D Systems), as indicated, at 10 μg/ml, 5 ng/ml, and 200 ng/ml, respectively, unless otherwise indicated. The BCR of stimulated cells at 106 cells/ml was (where indicated) activated by incubation with goat F(ab′)2 anti-mouse IgM (Southern Biotechnology Associates) at 2.5 μg/ml (5 μg/ml for Figs. 1 and 6) for the indicated times.
DNA microarray analysis
The analysis of gene expression changes in B cells upon BCR activation was performed using the Illumina BeadChip system. For in vitro transcription amplification, 200 ng of RNA was used with the Illumina RNA Amplification Kit (Ambion). Amplified RNA (1.5 μg) was hybridized to the Sentrix MouseRef-6 Expression Beadchip (Illumina) containing 47,667 probes. The primary data were collected from the BeadChips using the manufacturer’s BeadArray Reader and analyzed using the supplied scanner software. Data normalization was performed by cubic spline normalization using Illumina’s Beadstudio v3 software. Clustering of certain selected genes was done using cluster 3 software, and data were visualized using tree view v1.2 software. All raw CEL dataset files are available at www.ncbi.nlm.nih.gov/geo (accession no. GSE15606).
Plasmids and viruses
The pcDNAI-based expression vector for mammalian CaM, the short hairpin RNA (shRNA) that interferes with E12/E47-specific human RNA, the EBV-based shuttle vector pMEP4 derivatives encoding wild-type and CaM-resistant m847 and m8N47 mutants of mouse E12, and the murine stem cell virus (MSCV)-internal ribosome entry site-GFP-based retroviruses have all been described previously (12, 13, 14).
Infection of mouse B lymphocytes
Retrovirus concentrated by centrifugation was added with 5 μg/ml polybrene to 0.5 × 106 purified B cells after activation with LPS plus IL-4 for 14 h (24 h in Fig. 6). After a 12-h incubation, the infection was repeated for 12 h, followed by incubation for a further 22 h postinfection in fresh complete medium with the stimulants to allow for expression of GFP and E12 or CaM. In experiments after activation with CD40L plus IL-4, the medium was supplemented with LPS (2.5 μg/ml) during retroviral infection incubations to improve infection efficiency. Where indicated, anti-mouse IgM was added for the indicated time. Intracellular immunostaining of harvested cells was done with 2% paraformaldehyde and ethanol, as previously described (15), using Abs as listed in Materials and Methods. Stainings were done for at least 30 min at room temperature in the dark. Flow cytometry was with a FACSCalibur instrument and analysis with CellQuest software (BD Biosciences).
Cell culture and transfections
The human B cell lymphoma line DG75 (16) was maintained in RPMI 1640 medium supplemented with 5% FCS and antibiotics. DG75 cells were transfected by electroporation with 2 μg of CaM expression plasmid or the empty pcDNAI/amp vector. Eight hours later, live cells were separated using Lymphoprep (Axis-Shield), and 12 h later anti-IgM was added to one-half of the cells, followed by continued culture for 3 h. DG75 cells transfected to express empty pMEP4 derivative, or pMEP4 derivative encoding the shRNA, E12, or m847 or m8N47 mutant of E12, were selected with hygromycin B (Roche) for 5 days, as described previously (12). The BCR of DG75 was activated by stimulation of 1 × 107 cells for 3 h, unless otherwise specified, with 5 μg of goat F(ab′)2 anti-human IgM (Southern Biotechnology Associates) in 1 ml of complete RPMI 1640 medium supplemented with 5% FCS.
Total RNA was extracted using TRIzol reagent (Invitrogen), and real-time PCR analysis was performed, as previously described (12), using GAPDH as an internal control. The specificity of the real-time PCR was analyzed by melt-curve analysis, as described by the manufacturer, and the sizes of the PCR products with the different primer pairs were verified by 2.5% agarose gel electrophoresis. Some of the primer pairs used have been described previously (12), and the additional RT-PCR primer pairs used are listed in supplemental Materials and Methods.4
Very fast gene-regulatory events upon activation of the BCR
To analyze whether ASC differentiation is Ag driven in an in vitro cultivation system, B cells purified from mouse spleen were first activated using LPS and IL-4, and the effect of activation of the Ag receptor with anti-IgM after 2 days of cultivation was analyzed. To examine ASC differentiation, FACS analyses were performed after immunostaining for syndecan-1 (CD138), a marker for plasmablasts and plasma cells. Compared with the cells not given BCR activation, BCR activation for 1 day produced many more plasmablasts/plasma cells, identified by increased expression of syndecan-1 (Fig. 1,A). Furthermore, compared with cells not treated with anti-IgM, the average level of syndecan-1 expression increased between 2.5 and 3 times in this increased population of highly syndecan-1-expressing cells. After a second day of BCR activation, the highly syndecan-1-expressing cells increased in size, a change expected for differentiation into ASC cells. In the control without BCR activation, the frequencies of large and strongly syndecan-1-expressing cells changed little (Fig. 1,A). Thus, activation of the Ag receptor increases plasmablast/plasma cell differentiation in this system. The increase in frequency of syndecan-1 high cells and their syndecan-1 expression level at day 1 and the shift to larger highly syndecan-1-expressing cells at day 2 were also obtained when stimulating the BCR of cells activated with 4-fold lower levels of LPS and IL-4 (Fig. 1 B).
In the germinal center, the Ag receptor works with Th cells to determine selection of B cells for plasma cell differentiation, and CD40L (CD154) is the most important cytokine from the Th cells (2, 17). Therefore, we analyzed the effect of activation of the BCR with anti-IgM also when cultivating the splenic mouse B cells with CD40L instead of LPS together with the IL-4. This condition also resulted in a profound increase in large and highly syndecan-1-expressing cells when the BCR was stimulated (Fig. 1,C). The increase in large and highly syndecan-1-expressing cells by stimulating the BCR was even stronger when the levels of CD40L and IL-4 were reduced 4- or 10-fold (Fig. 1 D, and data not shown). Thus, Ag receptor stimulation increases plasmablast/plasma cell differentiation both in Th cell-independent and Th cell-dependent differentiation and over a range of concentrations of LPS, CD40L, and IL-4.
BCR activation results in a large number of very rapid transcription changes
To search for rapid changes in gene expression following BCR activation, we performed DNA microarray analysis of activated splenic B cells with and without anti-IgM treatment for 3 h. Cells activated with CD40L and IL-4 were used because this activation resulted in the largest increase in ASC differentiation (cf Fig. 1, A and B with Fig. 1, C and D) and because most Ab responses are Th cell dependent, mimicked by the main cytokine CD40L. The change in expression for each gene was computed as a ratio of expression in anti-IgM-treated B cells vs untreated control B cells. Using the Illumina Mouse Beadchip system, the expression of 31,492 genes was examined. Dendrogram analysis of these genes demonstrated a high degree of reproducibility between the three mice analyzed (data not shown). The expression of a remarkably large set of genes differed significantly (≥1.5-fold change, p < 0.05) between the BCR-stimulated and nonstimulated B cells: 2,259 genes were up-regulated, and 2,345 genes were down-regulated. The cluster plots of certain genes with a coupling to B cell and/or ASC differentiation are shown in Fig. 2, and the complete result of BCR activation for the 4,604 genes with significantly changed expression is shown in supplemental Table I.4 The analysis reveals that BCR activation, which promotes terminal differentiation of B cells, reduces expression within 3 h of major classes of genes that are highly expressed in primary and germinal center B cells, including genes important for B cell identity, such as secretory factors (e.g., IL-16 and Ltb), surface proteins (e.g., CD22 and Btla), and transcription factors (e.g., Bcl-6, Pax5, Oct-2, BACH2, Ets-1, MTA3, and Spi-B) (Fig. 2). The levels of the decreases in expression of the transcription factors within 3 h varied from ∼2-fold to a 10-fold down-regulation for Ets-1.
Plasmacytic differentiation requires the induction of several transcription factors that are important for this development, and one of these is IRF-4. Interestingly, IRF-4 showed more than a 3-fold up-regulation within 3 h of BCR activation (Fig. 2 and supplemental Table S1).4 We found also that the BCR activation increased expression of many surface proteins, including CD27, which has been correlated with the commitment to the plasma cell lineage (18). The finding that ∼7% of the genes were up-regulated and 7% were down-regulated within 3 h of BCR activation (supplemental Table S1)4 is comparable to the reported difference in expression of 15% of the genes (≥2-fold) between purified mouse germinal center B cells and plasma cells (19). Thus, a significant part of the change in mRNA level happens already within 3 h for most of the genes that change expression in differentiation from a B cell to an ASC.
To study the earliest Ag receptor-driven transcriptional changes in splenic B cells, the mRNA levels were followed by quantitative real-time PCR after the BCR was stimulated for selected genes identified in our genome-wide microarray analysis (Fig. 2) and for certain genes previously reported to change expression during ASC differentiation or to have a role in this process (1, 2, 3, 20). First, cells preactivated with LPS and IL-4 were analyzed (Fig. 3,A). The mRNA levels did not change significantly, or changed only relatively slowly, during 12 h of activation of the BCR for many of the proteins, including IgH, XBP-1, MTA3, BACH2, IRF-8, STAT6, EBF, BHLH B3, and Aiolos (Fig. 3,A, and data not shown). However, this was not the case for all genes. The mRNA level for Pax5, a key transcription factor for B cell identity and repressor of the ASC differentiation program, was significantly reduced already within 30 min of BCR stimulation. The reduction was ∼3-fold within 1 h and 4- to 5-fold within 3 h (Fig. 3,A). The mRNA level of Bcl-6, another repressor of ASC differentiation, also decreased significantly within 30 min of BCR stimulation, and this decrease was also ∼3-fold within 1 h and 4- to 5-fold within 3 h (Fig. 3,A). The level of mRNA for MITF, one more repressor of ASC differentiation, was reduced by ∼50% within 30 min of the BCR stimulation (Fig. 3,A). The mRNA levels of the Ets family members Ets-1, Fli-1, and Spi-B, all implicated in maintenance of B cells (1, 3, 7, 8, 11), were also significantly reduced within 30 min of the BCR stimulation (Fig. 3,A). Their mRNA levels were reduced between 2 and 3 times within 1 h, and they decreased further within 3 h (Fig. 3,A). The reduction was especially pronounced for Ets-1, which had only 18 ± 6% of the mRNA remaining after 3 h. Thus, down-regulation of the mRNA for several repressors of ASC differentiation are very early regulatory events that are initiated within less than 30 min and carried far within 1 h when activating the Ag receptor of splenic B cells. The mRNA levels for IRF-4 and Blimp1, important regulators of ASC differentiation and repressors of Pax5 and Bcl-6, increased relatively fast, although perhaps slightly slower than the six down-regulations. These mRNA increases reached 2-fold within less than 2 h (Fig. 3 A).
The very fast reductions of the mRNA levels for several repressors of ASC differentiation could be through inhibition of the transcription, provided that the mRNA is constitutively short-lived, or, alternatively, through anti-IgM-induced degradation of the mRNA. Therefore, the effect of BCR activation was compared with that of an inhibitor of transcription, actinomycin D. The mRNA level for each of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B was found to fall with approximately the same rate as after anti-IgM treatment also after actinomycin D treatment (data not shown). Thus, the mRNAs of these genes are all constitutively short-lived, and the fast reduction in mRNA level after Ag receptor stimulation appears to be through inhibition of transcription for each of these genes. To determine whether the rapid reductions in the mRNA of these genes and the increase in Blimp1 mRNA after BCR activation resulted in corresponding changes at the protein level, we performed Western blot analyses. A typical Western for each protein is shown in Fig. 3,B. All reductions in mRNA levels led also to reductions in the protein levels, although somewhat delayed compared with the mRNA levels. At 30 min, when the mRNA levels were already reduced (Fig. 3 A), the Pax5, Bcl-6, MITF, Ets-1, and Fli-1 protein levels remained at 105 ± 10, 90 ± 21, 100 ± 15, 95 ± 3, and 92 ± 9%, respectively (n ≥ 3). However, the levels of these proteins did decrease to 85 ± 9, 76 ± 27, 51 ± 8, 93 ± 17, and 80 ± 8%, respectively, after 1 h, and to 56 ± 12, 62 ± 18, 38 ± 4, 73 ± 7, and 48 ± 5%, respectively, after 2 h (n ≥ 3). In contrast, the increase in Blimp1 protein was several hours delayed compared with the increase in Blimp1 mRNA, indicating the presence of posttranscriptional regulation of the protein level after the BCR activation for this protein.
We analyzed the time courses of the effects of activation of the BCR also after combining the IL-4 treatment of the splenic mouse B cells with CD40L instead of LPS. The BCR stimulation resulted in the corresponding down-regulation of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B mRNA, as well as a reduction in MTA3 and BACH2 and up-regulation of IRF-4 and Blimp1 mRNA when using CD40L instead of LPS in the activation (Fig. 3 C). Notably, the decrease in Bcl-6 mRNA was more pronounced when using CD40L. This agrees with our observation that CD40L increases Bcl-6 mRNA expression to a higher level than LPS does (data not shown), and with Bcl-6 being a strongly expressed master regulator of germinal centers whose formation depends on Th cells and CD40L (2). The effects of BCR stimulation on the expression of the 10 genes were similar also after combined stimulation with CD40L and LPS (at 4-fold reduced levels) together with IL-4 (data not shown). Thus, Ag receptor stimulation results in the rapid key regulatory changes both in Th cell-independent and Th cell-dependent ASC differentiation.
The early regulatory events upon activation of the BCR depend on Ca2+ signaling and CaM
Activation of Ag receptor leads to formation of a large complex of proteins below the receptor that rapidly increases the intracellular Ca2+ concentration and results in a combination of Ca2+ signaling and a cascade of phosphorylations (21). To investigate whether Ca2+ signaling and/or serine protein kinases were essential for the effects of BCR stimulation on the expression of Pax5, Bcl-6, MITF, Ets-1, Fli-1, IRF-4, Spi-B, and Blimp1 mRNA, the activation with anti-IgM was performed in the presence of various inhibitors of the signaling pathways. The Ca2+ chelator BAPTA-AM completely blocked the effect of BCR stimulation on the mRNA level of all the eight genes studied, and inhibition of L-type Ca2+ channels with Nifedipine or inositol 1,4,5-triphosphate receptor Ca2+ channels with TMB-8 partially or completely blocked the effect, especially when combined (supplemental Fig. S1).4 Thus, Ca2+ signaling was essential for the effect of BCR stimulation on the expression of mRNA for each of these transcription factors. However, the effects of inhibitors of Ca2+/CaM-dependent protein kinase (CaMK) or Ca2+/CaM-dependent phosphatase calcineurin varied for the different proteins and were in most cases smaller or absent (supplemental Fig. S1).4 The MAPK inhibitor PD98059 blocked the reductions in Pax5, Bcl-6, MITF, and Spi-B mRNA and the increase in IRF-4 mRNA, and it partially blocked the increase in Blimp1 mRNA and the reduction in Ets-1 and perhaps also Fli-1 mRNA (supplemental Fig. S1).4 This suggests that, in addition to the Ca2+ signaling, MAPK signaling is important to different degrees for the BCR-induced changes in expression of all or most of the analyzed genes. The protein kinase C (PKC) inhibitor bisindolylmaleimide had little effect compared with Ca2+ chelator or Ca2+ channel blockers on the reductions in Pax5, Bcl-6, MITF, and Ets-1 mRNA and the increase in Blimp1 mRNA after BCR stimulation, whereas Fli-1 and Spi-B reductions and IRF-4 induction were affected by both types of treatment (supplemental Fig. S1).4
Both stimulated splenic B cells and the easily manipulated human B cell lymphoma line DG75 (16) were used to investigate the Ca2+ signaling-dependent effects of BCR activation. As in stimulated splenic B cells, activation of the BCR with anti-IgM rapidly reduced the levels of Pax5, Bcl-6, Ets-1, and Fli-1 mRNA and increased the level of Blimp1 mRNA in the B cell line (Fig. 4 A). MITF mRNA could not be detected in this cell line, and Spi-B expression was not inhibited (data not shown). The cause of these defects in MITF and Spi-B is unknown, but they might be coupled to the lymphoma phenotype of this cell line. The mechanism of IRF-4 induction was not analyzed because IRF-4 is induced by NF-κB (2) and the main Ca2+-sensor protein CaM and CaMK II are essential for NF-κB activation after Ag receptor activation (14, 22).
To examine whether CaM could affect the expression of the Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Blimp1 genes, the effects of its overexpression were investigated. The efficiency of the CaM overexpression system in DG75 cells has been verified previously by Western blot analysis (12). The overexpression of CaM reduced the Pax5, Bcl-6, and Fli-1 mRNA levels ∼2-fold and the Ets-1 mRNA level ∼3-fold compared with vector control, and CaM increased the level of Blimp1 mRNA ∼2-fold (Fig. 4,B). The effects of overexpression of CaM were also analyzed in LPS- and IL-4-activated splenic B cells at the protein level by FACS. The cells were infected with retrovirus-expressing CaM, followed by an internal ribosome entry site and GFP. The FACS plots of a representative experiment with Pax5 are shown in supplemental Fig. S2A.4 The signals obtained with Pax5 Ab were up to 100-fold higher than without primary Ab, and 30–60% of the cells were infected. Analysis of the FACS plots showed that CaM overexpression not only reduced Pax5 of GFP-positive cells, but also had some effect on noninfected cells of the same plot when compared with cells not exposed to CaM-overexpressing cells. This reproducible finding is probably a bystander effect, i.e., that infected cells can signal to their neighboring cells. It is notable that Ca2+ and MAPK signaling pathways, both of which regulate Pax5 and the other transcriptional regulators studied in this work, are involved in production of bystander effects on neighboring cells (23). To avoid any influence of bystander effects on the results, the effects of overexpressions on Pax-5 and the other regulators were compared with vector control infection and not with noninfected cells of the same FACS plot. The results of all experiments are summarized in Fig. 4,C. As seen in the figure, up to 2-fold decreases in Pax5, Bcl-6, and MITF and increase in Blimp1 were obtained upon BCR activation of uninfected control samples, a finding that agrees with our Western blot results (Fig. 3,B). The CaM overexpression system, previously verified by Western blot analysis (12), reduced Pax5, Bcl-6, and MITF, and increased Blimp1 protein levels also in the primary splenic B cells (Fig. 4,C). Effects of overexpression of CaM were also analyzed after activation with CD40L and IL-4 (Fig. 4,D and supplemental Fig. S2B),4 and this resulted in ∼2-fold decreases of Bcl-6, Ets-1, and Fli-1 expression. The level of Bcl-6 protein was higher at this condition compared with in presence of LPS plus IL-4, and the reductions in Bcl-6 protein on CaM overexpression and after BCR activation were larger (Fig. 4, cf C and D), findings that agree with our observations that CD40L plus IL-4 activate Bcl-6 mRNA much more than LPS plus IL-4 (data not shown). In summary, Fig. 4, B–D, shows that overexpression of CaM has the corresponding effect as BCR activation on Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Blimp1 expression.
Regulation of expression of repressors of plasma cell differentiation and Blimp1 by the BCR depends on CaM sensitivity of E2A
Because Ca2+-loaded CaM can inhibit DNA binding of E2A and other E proteins by directly binding to the DNA-binding basic sequences in their basic-helix-loop-helix domains (13, 24, 25), we examined whether E2A participates in the initiation of ASC differentiation when BCR is activated. Previously, we have shown that BCR activation does not affect the level of E2A mRNA or protein in DG75 B cells for several hours (12), and activation of the BCR did not significantly affect the level of E2A mRNA for at least 12 h in mouse splenic B cells (supplemental Fig. S3).4 Id proteins, which are inhibitors of DNA binding of E proteins, participate at several regulatory steps in B cell lineage development (26, 27). However, none of the Id proteins showed a large increase in mRNA level during 5 h of BCR activation that could explain the pronounced rapid decreases in Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B or the increase in Blimp1 mRNA in mouse splenic B lymphocytes (supplemental Fig. S3)4 or DG75 B cells (12). In addition, no significant change is seen in the expression level of any of the Id proteins in DG75 B cells (12). Furthermore, BCR activation does not decrease the amount of E2A that can bind to DNA in the absence of Ca2+ (12); therefore, no decrease in the level of E2A or increased inhibition of its DNA binding by Id proteins mediates the rapid transcriptional effects of BCR stimulation.
We have reported a series of mutants in the basic DNA and CaM-binding sequence of the E2A isoform E12 that through combinations of mutations are resistant to CaM to different extents (13). To investigate the possible role of Ca2+/CaM inhibition of E2A in the regulation of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Blimp1, we used the most CaM resistant of these, the m847 mutant, which has three amino acid substitutions. This mutant, however, is also resistant to Id proteins, whereas a second mutant used, m8N47, is sensitive to Id proteins and almost as resistant as the m847 mutant to CaM (12). We used a hygromycin-selectable EBV-based shuttle vector that stably directs synthesis of shRNA that interferes with human E2A mRNA. It reduces expression of both E2A mRNA and E2A protein over 10-fold in DG75 B cells (12). This shRNA expression plasmid reduced expression of Pax5 and Ets-1 mRNA in DG75 B cells by ∼2-fold, Bcl-6 by ∼1.3-fold, and Fli-1 by 3-fold, whereas Blimp1 expression was increased by ∼2-fold (Fig. 5,A), showing that the level of expression of these genes is E2A dependent. The reduction in E2A and the resulting effects on the expression of these genes were specific, because the shRNA expression plasmid did not affect the expression of a number of other BCR-regulated genes that have not been reported to be E2A regulated (Fig. 5 A) (12) (data not shown).
BCR activation with anti-IgM reduced Pax5, Ets-1, Bcl-6, and Fli-1 mRNA 1.3- to 3-fold and increased Blimp1 mRNA 2- to 3-fold in nontransfected DG75 cells or in the presence of vector control plasmid (Fig. 5,A). In contrast, expression of these genes was approximately at the level of the anti-IgM-treated cells transfected with vector control both with and without anti-IgM treatment when the shRNA expression plasmid was present (Fig. 5,A). Thus, BCR activation had no further effect on the expression of any of these genes when E2A expression was inhibited (Fig. 5,A). This loss of effect of anti-IgM was specific, because expression of the shRNA against E2A did not influence the effect of anti-IgM for any control gene analyzed that was induced or inhibited by BCR activation and not E2A regulated (Fig. 5 A) (12) (data not shown).
The decrease in E2A mRNA was reversed by cotransfection with expression plasmid for the mouse E12 isoform of E2A, because the expressed shRNA does not interfere with mouse E2A. This complementation did not significantly change the expression of any control gene, again confirming that they are not E protein regulated (Fig. 5,A) (12) (data not shown). This complementation fully restored both the levels of expression of Pax5, Bcl-6, Ets-1, Fli-1, and Blimp1 mRNA and the sensitivities to anti-IgM treatment (Fig. 5,A). Importantly, the sensitivity to BCR stimulation was completely lost when complementing with a CaM-resistant mutant of E12. Anti-IgM had no effect on expression of any of the genes upon complementation with either the m847 or the m8N47 mutant of E12 (Fig. 5,A). The loss of the effects of BCR stimulation on these genes was attributed to the loss of CaM sensitivity and not to loss of sensitivity to an Id protein, because the losses of effect of anti-IgM were as complete with the m8N47 mutant that is only resistant to CaM as with the m847 mutant that is also resistant to Id proteins. This loss of effect of anti-IgM on the five genes was specific, because CaM-resistant m847 or m8N47 mutant of E12 did not affect the result of anti-IgM treatment for any control gene stimulated or inhibited by BCR activation, including IRF-4 (Fig. 5 A) (12) (data not shown).
To examine further whether the Ca2+ signaling from the BCR affected expression of Pax5, Bcl-6, MITF, Blimp1, Ets-1, and Fli-1 through CaM-mediated inhibition of E2A, we performed FACS analyses of primary splenic B cell cultures infected with retrovirus expressing wild-type or CaM-resistant E12 (Fig. 5, B and C, and representative FACS plots in supplemental Fig. S4).4 We found a clear difference between primary splenic B cells infected with wild-type E12 that showed Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Blimp1 expressions sensitive to anti-IgM treatment and cells infected with CaM-resistant m847 or m8N47 mutant of E12 that showed clearly anti-IgM-resistant expression of these genes (Fig. 5, B and C). The decrease in Bcl-6 expression upon BCR activation after infection with retrovirus expressing wild-type E12 was larger in cells activated with CD40L and IL-4 than with LPS and IL-4 (Fig. 5, cf B and C), a finding that agrees with the results from noninfected or vector-infected cells (Fig. 4, C and D). This larger decrease is probably due to the much higher initial activation of Bcl-6. In summary, the inhibitions of Pax5, Bcl-6, MITF, Ets-1, and Fli-1 expression and the activation of Blimp1 expression after BCR activation all depend on CaM sensitivity of E2A.
Plasma cell differentiation is regulated by CaM inhibition of E2A
To investigate whether plasma cell differentiation is regulated by CaM and dependent on CaM sensitivity of E2A, we used the in vitro system for plasma cell differentiation (Fig. 1) and infected the purified splenic B cells with retrovirus expressing either CaM, wild-type E12, or CaM-resistant E12, followed by staining for the plasma cell marker syndecan-1 and gating for GFP-positive cells (infected cells) in the FACS analysis. Two days of anti-IgM stimulation of the BCR of B cells activated with LPS and IL-4 increased the frequency of large highly syndecan-1-expressing cells in the vector control infection (Fig. 6,A), similarly to the results illustrated in Fig. 1. The generally higher frequencies of large syndecan-1-positive cells in Fig. 6 occur because gating for infected cells enriches for large highly syndecan-1-expressing cells, presumably as a combined effect of more efficient infection of cells prone to become larger and more syndecan-1 expressing together with cells becoming more active and thereby larger due to the infection. Nevertheless, the effect of BCR stimulation on the frequency of large highly syndecan-1-expressing cells was even larger in Fig. 6,A than in Fig. 1 (14.2% (33.7-19.5%) vs 6.4% (7.5-1.1%)). Importantly, overexpression of CaM resulted in an increase in large highly syndecan-1-expressing cells even without anti-IgM treatment. This increase, approximately as large as upon stimulation of the BCR (Fig. 6,A), shows that overexpression of CaM has a corresponding positive effect on plasma cell differentiation as BCR activation. This result agrees with our findings on the effects of overexpression of CaM on the regulation of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Blimp1 (Fig. 4, B–D). A corresponding effect of overexpression of CaM on the frequency of large highly syndecan-1-expressing cells was obtained also in cells grown in presence of CD40L and IL-4 (Fig. 6,B). Cells grown with either LPS and IL-4 or CD40L and IL-4 and overexpressing wild-type E12 were clearly affected by BCR activation, and a ∼1.5- to 2-fold higher frequency of large highly syndecan-1-expressing cells was observed, a frequency similar to the vector control-infected cells (Fig. 6). Importantly, in contrast to expression of the wild type, expression of a CaM-resistant E12, either the Id-resistant m847 mutant or m8N47 that is sensitive to the Id proteins, resulted in a reduced plasma cell differentiation both with and without BCR stimulation, and this reduction appeared to be slightly larger upon BCR stimulation (Fig. 6). In summary, overexpression of CaM stimulates plasma cell differentiation, and the efficiency of initiation of Ag receptor-driven plasma cell differentiation depends on CaM sensitivity of E2A.
Plasmacytic differentiation is induced in response to appropriate signals to generate specific Abs upon Ag exposure. The BCR plays a key role, and the strength of the interaction with Ag is critical in determining which clones enter the plasma cell response (4, 5, 6). Additional signals participating in plasma cell differentiation include pathogen-associated molecular patterns, which signal through TLRs, and cytokines and other T cell signals in which CD40L is of special importance (2, 17, 28). A large number of studies have identified several transcription factors that have to be up-regulated or down-regulated in ASC differentiation. The differentiation has been seen as a chain of changes in expression of transcription factors (1, 2, 3, 17, 20), but it has been unclear which regulatory change is primary and leads to subsequent changes in expression of a series of other transcription factors. Reviews of this differentiation have argued for one or the other as the primary regulatory change (1, 2, 3, 17, 20). The up-regulation of Blimp1 has recently been excluded as the primary event, because initiation of the differentiation is Blimp1 independent, and up-regulation of Blimp1 is very slow and secondary to down-regulations of repressors during this differentiation (28, 29, 30). For a long time, Bcl-6 and Pax5 have been key candidates for being the primary transcription factor whose expression is changed first (1, 2, 3, 17, 20). In this study, we show that the key signal in initiation of the plasma cell response, activation of the Ag receptor, does not lead to a key initial change in expression of one transcription factor, followed by a chain of secondary events, but instead leads to parallel changes in expression of several factors. The primary regulatory events identified were inhibition of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B gene expression. These inhibitions were very rapid; much of the reduction in their mRNA levels occurred within 30 min. The result was essentially the same after TLR activation with LPS and after stimulation with the main Th cell cytokine CD40L. Thus, initiation of Ag-driven ASC development appears to occur in the same way both in T cell-independent and T cell-dependent differentiation.
Why are there many parallel changes rather than one key regulatory change? The former alternative has several advantages. It becomes more robust, because variation in expression or activity of one single repressor of ASC differentiation will not alone have a decisive importance when several repressors together determine the critical developmental decision if the Ab of a particular B cell is good enough for ASC differentiation and thereby mass production of the Ab. Furthermore, the decision can be more fine-tuned by combining the strength of the BCR signal with the strengths of many other signals when several transcription factors, which are influenced to different degrees by different signals, participate in the initial decision to differentiate or not to differentiate. In addition, BCR stimulation together with additional signals can also lead to differentiation into memory B cells. The decision to differentiate into a memory B cell also depends on transcription factor changes (2, 19), and parallel inhibition of expression of several transcription factors would enable control of this decision between ASC or memory B cell differentiation by the level of the changes in different important transcription factors. A key difference between these differentiations is that Pax5 is not down-regulated in memory B cell differentiation (2, 19). This difference could be the reason for the observed limitation of the Pax5 reduction (4- to 5-fold) upon the BCR stimulation, because a small fraction of the BCR-stimulated cells might differentiate into memory cells instead of plasma cells.
We found that the inhibitions of Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B gene expression all depend on Ca2+ signaling, because they were blocked by Ca2+ chelator or a combination of Ca2+ channel blockers, and expression of all analyzed genes was reduced by the Ca2+ sensor CaM. Furthermore, the inhibitions of expression by BCR activation depended on CaM sensitivity of E2A when analyzed at the mRNA level and/or at the protein level. The resistance of the expression of these genes to BCR activation when E2A is mutated does not exclude that the other E proteins, E2-2 and HEB, may also contribute to the expression and become CaM inhibited like E2A when the BCR is activated. However, the reduction in mRNA for the analyzed proteins by shRNA against E2A in the DG75 cells and the dominant effect of CaM-resistant E12 over all endogenous CaM-sensitive E proteins suggest that E2A is the dominant E protein regulating these genes. Thus, Ca2+ signaling leading to inhibition of E2A by Ca2+-loaded CaM is needed for many gene expression changes in initiation of plasma cell differentiation.
The very rapid and parallel reductions in the mRNAs for Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B by BCR stimulation, before reduction of the protein levels, strongly suggest that these mRNA reductions are not secondary to another change in expression. E2A functions in hematopoietic progenitor cells to activate expression of EBF and Pax5 and establishment of the program of B cell-specific gene expression (26, 31). The genes shown to be down-regulated through inhibition of E2A by Ca2+-loaded CaM could all be direct targets of E2A. However, it is presently not known whether they have an important E2A binding site in a regulatory DNA sequence of the gene. Therefore, we cannot exclude that one or more of them are indirect E2A targets. The direct alternative is, however, the most likely, because the levels of mRNA were reduced by one-half within approximately the same time after BCR activation as after blocking RNA synthesis with actinomycin D. Thus, the limiting factor is the decay rates of the mRNA, and any delay before reduction in transcription is at most a few minutes. This fact strongly argues against the possibility that CaM inhibition of E2A down-regulates a hypothetical unknown primary gene, which subsequently would down-regulate these genes. Instead, this suggests that Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B are direct targets of E2A.
BCR stimulation increased IRF-4 and Blimp1 mRNA levels relatively fast, but perhaps slightly slower than the rapid transcription factor down-regulations. IRF-4 and Blimp1 are important regulators of ASC differentiation and repressors of Pax5 and Bcl-6 (1, 3, 9, 10), but IRF4 participates in plasma cell differentiation also independently of down-regulation of Bcl-6 (9). Induction of IRF-4 expression after BCR activation depends on NF-κB activation and not on inhibition of E proteins by Ca2+-loaded CaM. The increase in IRF-4 was sensitive to Ca2+ chelator, Ca2+ blockers, PKC inhibitor, and CaMK II inhibitor, findings that agree with reports that IRF-4 is induced by NF-κB (2) and that PKCβ and the main Ca2+-sensor protein CaM and CaMK II are essential for NF-κB activation after BCR activation (14, 22, 32). Furthermore, neither shRNA against E2A nor expression of CaM-resistant E12 changes the level of IRF-4 mRNA or the effect of BCR activation on this expression in DG75 B cells (12).
The increase in Blimp1 mRNA, and even more in Blimp1 protein, after BCR activation was slower than the down-regulations of several repressors of the ASC development. Nevertheless, the induction of Blimp1 mRNA and protein expression did, like the down-regulations of the repressors, depend on CaM sensitivity of E12. This can be explained by the rapid inhibition of Pax5 and Bcl-6, which are direct repressors of Blimp1 expression (28), through this mechanism. Furthermore, MITF is also rapidly inhibited through this mechanism, and MITF is an inhibitor of IRF4, which is an activator of Blimp1 expression (28). In addition to induction of Blimp1 expression secondary to relief of repression dependent on CaM sensitivity of E2A, Blimp1 induction could also be the result of a shift from CaM-sensitive E protein homodimers to a CaM-resistant E protein heterodimer with another basic helix-loop-helix transcription factor at an important gene-regulatory DNA element. Such a scenario would be analogous to the important Ca2+/CaM-dependent shift from E protein homodimers to E protein heterodimers with MyoD in muscle development (33).
In summary, we report that initiation of plasma cell differentiation is Ag driven through many very fast primary regulatory events from the BCR. Down-regulation of expression of many key transcription factor genes, including Pax5, Bcl-6, MITF, Ets-1, Fli-1, and Spi-B, occurs within 30 min. Ca2+ signaling leading to inhibition of E proteins by Ca2+-loaded CaM is needed for many of the gene expression changes in initiation of plasma cell differentiation.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Swedish Research Council and the Swedish Cancer Society.
Abbreviations used in this paper: ASC, Ab-secreting cell; Blimp1, B lymphocyte-induced maturation protein 1; CaM, calmodulin; CaMK, CaM-dependent protein kinase; IRF, IFN regulatory factor; MSCV, murine stem cell virus; PKC, protein kinase C; shRNA, short hairpin RNA.
The online version of this article contains supplemental material.