Plasma cell differentiation is initiated by Ag stimulation of BCR. Until BCR stimulation, B lymphocyte-induced maturation protein 1 (BLIMP1), a master regulator of plasma cell differentiation, is suppressed by PAX5, which is a key transcriptional repressor for maintaining B cell identity. After BCR stimulation, upregulation of BLIMP1 and subsequent suppression of PAX5 by BLIMP1 are observed and thought to be the trigger of plasma cell differentiation; however, the trigger that derepresses BLIMP1 expression is yet to be revealed. In this study, we demonstrated PAX5 phosphorylation by ERK1/2, the main component of the BCR signal. Transcriptional repression on BLIMP1 promoter by PAX5 was canceled by PAX5 phosphorylation. BCR stimulation induced ERK1/2 activation, phosphorylation of endogenous PAX5, and upregulation of BLIMP1 mRNA expression in B cells. These phenomena were inhibited by MEK1 inhibitor or the phosphorylation-defective mutation of PAX5. These data imply that PAX5 phosphorylation by the BCR signal is the initial event in plasma cell differentiation.
PAX5 is a member of the highly conserved paired-box (PAX) domain family of transcription factors. PAX5 is exclusively expressed from the pro-B to mature B cell stages and is downregulated during terminal differentiation into plasma cells (1). PAX5 is not only indispensable for B-lineage commitment; its continuous expression is essential for maintaining the identity of B cells (2–4). PAX5 functions as both a transcriptional activator of B lineage-specific genes and a repressor of B lineage-inappropriate genes (1) [i.e., it activates CD19 (5), CD79A (6), and B cell linker protein (7) and represses CSF1 receptor (8), Notch1 (9), and FMS-like tyrosine kinase 3 (10)]. In addition, it checks the initiation of plasma cell differentiation and the terminal differentiation of B cells by repressing B lymphocyte-induced maturation protein 1 (BLIMP1) and X box-binding protein 1 (11, 12).
BCR signal plays important roles in the activation, survival, and differentiation of B lymphocytes. The initial event after BCR engagement is the activation of Lyn and Syk. These kinases trigger a complex network of signaling pathways downstream of the receptor, including the Ras-Raf-MEK-ERK1/2 pathway, the Vav-cell division cycle 42-JNK pathway, and the NF-κB pathway (13). The resulting signals quickly reach the nucleus and alter gene expression. The ultimate effects on B cells are profound and vary depending on the maturation state of the cell and on additional signals that the cell receives. For germinal center B (GCB) cells, BCR signal after encountering Ag is known to initiate PAX5 downregulation, BLIMP1 upregulation, and eventually, plasma cell differentiation (14, 15). In these cells, PAX5 suppresses BLIMP1 expression and checks plasma cell differentiation. After BCR stimulation by Ag, BLIMP1 repression by PAX5 is abolished, and once BLIMP1 is expressed, it suppresses PAX5. Eventually, PAX5 is replaced by BLIMP1, which initiates plasma cell differentiation (14). The abolition of PAX5-mediated repression was thought to be the first event to initiate plasma cell differentiation (16); however, the mechanism is still unknown. In this study, we demonstrated that PAX5 was phosphorylated by ERK1/2 in vitro and in vivo at serines 189 and 283. This phosphorylation attenuated the transcriptional repression of BLIMP1 by PAX5. Finally, BCR stimulation induced the phosphorylation of ERK1/2 and PAX5, as well as BLIMP1 mRNA expression in B cells, which were inhibited by MEK1 inhibitor or the phosphorylation-defective mutation of PAX5. These data imply that PAX5 phosphorylation by the BCR signal is an initial event in plasma cell differentiation.
Materials and Methods
Cells, Abs, and reagents
Burkitt lymphoma cell line Ramos cells were cultured in RPMI 1640 medium supplemented with 10% FBS. Anti-HA Ab, anti-ERK2 Ab, and anti-PAX5 Ab (C-20) for immunoblotting, and anti-PAX5 Ab (N-19) for the supershift assay in EMSA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-human IgM Ab and anti-mouse IgM Ab for BCR stimulation and anti–phospho-ERK1/2 Ab were from Abcam (Cambridge, U.K.), Jackson ImmunoResearch Laboratories (West Grove, PA), and Cell Signaling (Beverly, MA), respectively. U0126 was obtained from Calbiochem (San Diego, CA).
PAX5/pCDNA, the expression vector for PAX5, was described previously (17). PAX5/pGEX and PAX5(1-279)/pGEX, expression vectors for GST-fused full-length and partial PAX5, as well as PAX5/pBGJR, were made by subcloning PAX5 cDNA digested from PAX5/pCDNA with appropriate restriction enzymes into pGEX 5X-1 vector (Pharmacia, Uppsala, Sweden) and pBGJR, a lentivirus expression vector kindly provided by Dr. Stefano Rivella (Memorial Sloan-Kettering Cancer Center). Mutations for serine/threonine-to-alanine substitutions were introduced into PAX5/pCDNA using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). All clones were subjected to sequence analysis to confirm the introduction of the correct mutations and to exclude PCR artifacts. The expression vector for HA-tagged CA-MEK1, HA–CA-MEK1/pCMV, was described previously (18). The −1921 to +138 region (relative to the translation start site) of the BLIMP1 promoter containing putative NF-κB- and PAX5-binding sequences was amplified by PCR, according to a previous report (19), and subcloned into pGL4.20 (Promega, Madison, WI). Two specific primer genes, 5′-TAACAGTGAGTTGATTCACTGGC-3′ (sense) and 5′-CTCGGCGGTCCCTCCTCG-3′ (antisense), were selected on the basis of the sequence of the human BLIMP1 genomic DNA (accession number AL358952, http://www.ncbi.nlm.nih.gov/nuccore/Al358952). This reporter gene was designated as BLIMP1-luc/pGL4. Expression vectors for NF-κB p50 and p65 were purchased from Addgene (Cambridge, MA).
Transient transfection, lentivirus infection, immunoblotting, immunofluorescence, EMSA, luciferase assay, and in vitro kination assay
The small interfering RNA (siRNA) targeting the 5′-GACTATCCATCCATCATAA-3′ sequence in the 3′ untranslated region (UTR) of PAX5 was purchased from Sigma-Aldrich (St. Louis, MO) and introduced into Ramos cells with nucleofector (Lonza, Wuppertal, Germany), according to the manufacturer’s instructions.
Phosphate-affinity SDS-PAGE was performed similarly to SDS-PAGE, except that Phos-tag acrylamide-containing acrylamide gel was used. Phos-tag acrylamide was obtained from Wako Laboratory Chemicals (Osaka, Japan). In this system, Phos-tag acrylamide binds to phosphorylated amino acids during electrophoresis and slows the migration of phosphorylated proteins, according to the number of phosphorylation sites. Autoradiography or immunoblotting following electrophoresis can detect the phosphorylation of the target protein as band shifts.
Mouse spleen cell isolation
Mouse spleen cells were collected from 10-wk-old BALB/c mice and were used for immunoblotting analysis to detect PAX5 phosphorylation. Mouse spleen B cells were purified from the spleen cells using CD45R (B220) MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and subjected to RT-PCR to detect BLIMP1 expression. These cells were cultured in RPMI 1640 medium supplemented with 10% FCS, 100 U/ml penicillin, and 100 U/ml streptomycin.
PAX5 DNA sequencing of clinical samples
Lymph nodes or other tissues containing tumor cells were collected with informed consent from 85 patients diagnosed with diffuse large B cell lymphoma. The sequencing study was approved by the institutional review board of Nagoya University Graduate School of Medicine. Genomic DNA was extracted from those samples with the QIAamp DNA Micro Kit (QIAGEN), according to the manufacturer’s instructions. Pyrosequencing was performed for analysis of DNA mutation surrounding codons 189 and 283, according to the manufacturer’s instructions. Briefly, a 125-bp sequence of exon 5, including 189 codon, or a 128-bp sequence of exon 7, including 283 codon, was amplified by PCR using biotin-tagged primers. The biotinylated PCR strands were immobilized and purified by streptavidin-Sepharose beads, denatured, and added to annealing buffer containing 250 nM sequencing primer. Sequencing was carried out with the PyroMark Q96 ID system (QIAGEN). DNA sequences corresponding to codons 181–199 and codons 273–292 were analyzed using two sequence primers. Primers used in this analysis are described in Supplemental Table I.
PAX5 is phosphorylated by ERK2 in vitro
It was recently reported that ERK1/2 signal was a key initiation signal for BLIMP1 expression and plasma cell differentiation of GCB cells (21); however, the detailed mechanism of ERK1/2 signal that induces BLIMP1 expression is still unknown. Combined with the open question about the initial event in plasma cell differentiation, these findings gave rise to the speculation that PAX5 phosphorylation by ERK1/2 negatively affected BLIMP1 repression by PAX5 and is the trigger of plasma cell differentiation. To test whether PAX5 can be phosphorylated in response to ERK1/2 signal, we performed an in vitro kinase assay using GST-PAX5 as a substrate. rERK2 efficiently phosphorylated full-length PAX5 and the N-terminal region of PAX5, whereas JNK did not (Supplemental Fig. 1A). Inspection of the PAX5 sequence revealed the presence of eight ERK1/2 consensus sites, S/T-P. Therefore, we introduced a series of alanine substitutions at these sites to map the actual phosphorylation sites by ERK2 and to check the phosphorylation status with the phosphate-affinity SDS-PAGE system. In this system, PAX5 phosphorylation by ERK2 was detected as two shifted bands (Fig. 1A), suggesting that PAX5 had two phosphorylation sites. Substitutions of alanine at codons 189, 283, and 285 (combined mutation), as well as at 283, caused the disappearance of one of the shifted bands, whereas other substitutions, including at codon 285, did not affect the shifted bands (Fig. 1A, Supplemental Fig. 1B, 1C), indicating that the phosphorylation sites were serines 189 and 283. Consistently, all shifted bands disappeared by combined substitution at codons 189 and 283 (Fig. 1A).
PAX5 phosphorylation occurred in vivo through ERK1/2 signaling
We next examined whether PAX5 phosphorylation occurred by the activation of ERK1/2 in vivo. Coexpression of the constitutively active mutant of MEK1 (CA-MEK1, an upstream activator of endogenous ERK1/2) in 293T cells caused similar PAX5 phosphorylation as in vitro (i.e., two shifted bands were observed in phosphate-affinity SDS-PAGE and disappeared by the mutation at ERK2 phosphorylation sites determined in vitro) (Fig. 1B). These results indicated that ERK1/2 could also phosphorylate PAX5 in vivo. We demonstrated the schema of the PAX5 structure, pointing out the phosphorylation sites. The amino acid sequences surrounding the phosphorylation sites are conserved evolutionarily, except in zebrafish (Fig. 1C).
Next, we attempted to determine whether endogenous PAX5 was phosphorylated by BCR stimulation. We stimulated BCR by anti-IgM Ab in Ramos cells, a cell line of Burkitt lymphoma, which is thought to be a tumor of GCB cells. Strikingly, BCR stimulation induced strong ERK1/2 phosphorylation and PAX5 phosphorylation in Ramos cells. This phosphorylation was inhibited by the MEK1 inhibitor, U0126, indicating the mediation of PAX5 phosphorylation by ERK1/2 signal (Fig. 2A). Furthermore, to examine whether BCR signal-induced PAX5 phosphorylation occurred at ERK2 phosphorylation sites determined in vitro, we established stable Ramos transfectants of control vector and the expression vectors of wild-type and phosphorylation-defective mutant of PAX5, designated as Control-Ramos, PAX5 Wild-Ramos, and PAX5 189/283A-Ramos, respectively. Because phosphorylation of endogenous PAX5 masked the difference in the phosphorylation status between exogenously expressed wild-type and mutant PAX5 (data not shown), we further introduced the siRNA targeting 3′ UTR of endogenous PAX5 to specifically knockdown endogenous PAX5 of these transfectants. Successful specific knockdown of endogenous PAX5 is demonstrated in Fig. 2B. In this system, BCR signal-induced PAX5 phosphorylation was completely diminished by the mutation at ERK2 phosphorylation sites (Fig. 2C). These results indicated that BCR signal-induced PAX5 phosphorylation was mediated by ERK1/2.
We also performed the same experiment in Fig. 2A using mouse spleen cells that were rich in primary B cells. The results were similar to that in Ramos cells. BCR stimulation of mouse spleen cells induced ERK1/2 phosphorylation and PAX5 phosphorylation, which was inhibited by U0126 (Fig. 2D). These results suggested that PAX5 phosphorylation by ERK1/2 in response to BCR stimulation also occurred in primary normal B cells.
ERK1/2 signal canceled PAX5-dependent transcriptional repression of BLIMP1
Next, we set out to identify the effect of phosphorylation on BLIMP1 repression by PAX5. To replicate BLIMP1 repression by PAX5 in the luciferase assay, we constructed a reporter gene containing an ∼2-kbp region of BLIMP1 promoter, including putative binding sites for PAX5, and NF-κB, one of the BLIMP1 expression activators, according to a previous report (19) (Fig. 3A). Before the assay, we focused on enhancement of PAX5 expression by CA-MEK1 coexpression observed in Fig. 1B, because it may affect the luciferase assay comparing PAX5 function with and without CA-MEK1 coexpression. We judged that this enhancement of PAX5 expression is due to a nonspecific effect of ERK1/2 signal on the transcriptional machinery on the T7 or CMV promoter of PAX5/pCDNA, because CA-MEK1 coexpression also enhanced the expression of the phosphorylation-defective mutant of PAX5 (Fig. 1B), indicating that this phenomenon was independent of PAX5 phosphorylation, and BCR stimulation or CA-MEK1 expression in Ramos cells did not affect the expression level of endogenous PAX5 (Fig. 2A, data not shown). Therefore, we investigated the effect of CA-MEK1 coexpression on PAX5 expression in detail (Supplemental Fig. 2A) and adjusted the amount of PAX5 expression vector used for the luciferase assay when it was cotransfected with the CA-MEK1 expression vector to keep the PAX5 expression level constant. A constant expression of PAX5 among luciferase samples was confirmed by immunoblotting (Fig. 3B, lower panel).
Luciferase expression was strongly induced by NF-κB expression, which was decreased to 15% of the control level by wild-type PAX5 coexpression (Fig. 3B, lane 2 versus lane 4). Importantly, CA-MEK1 coexpression increased the luciferase expression suppressed by wild-type PAX5 to 80% of the control level (Fig. 3B, lane 3 versus lane 5), indicating that ERK1/2 signal canceled transcriptional repression by PAX5. Furthermore, transcriptional repression by mutant PAX5 was attenuated by CA-MEK1 coexpression to a significantly lesser extent than that by wild-type PAX5 (Fig. 3B, lane 5 versus lane 7, p < 0.05), indicating its resistance to ERK1/2 signal-dependent cancellation of the transcriptional repression. These data suggested that PAX5 phosphorylation by ERK1/2 signal played an important role in the abolition of BLIMP1 repression by PAX5.
Of note, mutant PAX5 suppressed the luciferase expression more strongly than did wild-type PAX5 when CA-MEK1 was not coexpressed (Fig. 3B, lane 4 versus lane 6). This is probably due to mild attenuation of wild-type PAX5 ability by weak phosphorylation of PAX5 by constitutively activated ERK1/2 signal in 293T cells, because ERK1/2 was weakly, but constitutively, phosphorylated in 293T cells without CA-MEK1 coexpression (Supplemental Fig. 2B), and administration of U0126 enhanced the transcriptional repression by wild-type PAX5 but did not affect that by mutant PAX5 and diminished the difference between them (Supplemental Fig. 2C).
BCR signal-induced PAX5 phosphorylation increased BLIMP1 expression in B cells
To further confirm the BLIMP1 derepression by BCR stimulation, we first examined BLIMP1 mRNA expression after BCR stimulation in Ramos cells. BLIMP1 mRNA expression was induced by BCR stimulation following ERK1/2 and PAX5 phosphorylation, and U0126 inhibited the induction of BLIMP1 expression (Fig. 4A). Next, we used Ramos transfectants with specific knockdown of endogenous PAX5. Exogenous expression of wild-type and mutant PAX5 reduced BLIMP1 expression. BCR stimulation relieved BLIMP1 repression by wild-type PAX5, whereas repression by mutant PAX5 was resistant to BCR stimulation (Fig. 4B). These data indicated that BCR signal-induced BLIMP1 expression was mediated by PAX5 phosphorylation by ERK1/2. Notably, in contrast to the luciferase assay in Fig. 3B, no difference in the repression of BLIMP1 was observed between wild-type and mutant PAX5 transfectants of Ramos cells when there was no BCR stimulation (Fig. 4B, lane 3 versus lane 5). This occurs because Ramos cells have no constitutive ERK1/2 activation, which is different from 293T cells (Supplemental Fig. 2B).
We also examined BCR signal-induced BLIMP1 expression in mouse spleen B cells. BCR stimulation of mouse spleen B cells induced BLIMP1 mRNA expression, which was inhibited by U0126 (Fig. 4C). Taken together with BCR signal-induced PAX5 phosphorylation in these cells (Fig. 2D), these data implied that BCR signal-induced BLIMP1 derepression through PAX5 phosphorylation by ERK1/2 might also occur in primary B cells.
PAX5 phosphorylation did not affect DNA-binding ability or cellular localization
To clarify how PAX5 phosphorylation attenuated its function, we used EMSA to investigate the effect of PAX5 phosphorylation on DNA-binding ability. PAX5, synthesized in vitro, was incubated with radiolabeled oligomers containing the PAX5-binding sequence in the CD19 promoter. The obtained single band was competed by the presence of a 200-fold molar excess of nonradiolabeled oligomers and was supershifted by anti-PAX5 Ab but not by control rabbit IgG (Fig. 5A). Wild-type and mutant PAX5 were subjected to in vitro kination, with or without ERK2, and then to EMSA. No difference in DNA-binding activity was observed, regardless of phosphorylation by ERK2 and phosphorylation-defective mutation (Fig. 5A). These results indicated that PAX5 phosphorylation did not affect its DNA-binding ability.
Next, we examined the alteration of PAX5 cellular localization by coexpression of CA-MEK1 or phosphorylation-defective mutation. Overexpressed wild-type and mutant PAX5 localized diffusely in the nucleus, and this was not affected by coexpression of CA-MEK1, suggesting that cellular localization of PAX5 was not altered by its phosphorylation (Fig. 5B). The molecular mechanisms through which PAX5 phosphorylation abolishes transcriptional repression of BLIMP1 are unknown.
The data presented in this article support our speculation that PAX5 phosphorylation by ERK1/2 negatively affected BLIMP1 repression by PAX5. It is reported that BLIMP1 repression by PAX5 is abolished after BCR stimulation by Ag. Once BLIMP1 is expressed, it suppresses PAX5, which allows greater expression of BLIMP1. This positive feedback loop enables quick replacement of PAX5 with BLIMP1, which initiates plasma cell differentiation (14). Kallies et al. (16) investigated the first event of plasma cell differentiation in detail; they reported that the abolition of PAX5-mediated repression of BLIMP1 was the first event to initiate plasma cell differentiation and that the mechanism was neither a decrease in the DNA-binding ability of PAX5 nor downregulation of the PAX5 expression level and was yet to be revealed. PAX5 phosphorylation by ERK1/2 could be a clue to this uncertainty. The schema of this putative model is shown in Fig. 6.
It should be noted that Ramos cells or primary B cells from mouse spleen did not undergo plasma cell differentiation as the result of BCR stimulation with anti-IgM Ab (data not shown), despite the stimulation-induced phosphorylation of ERK1/2 and PAX5 and BLIMP1 expression. We could not keep the primary B cells alive for longer than a few days and could not estimate the differentiation. With regard to Ramos cells, one possible reason is that ERK1/2 phosphorylation induced by anti-IgM Ab stimulation was transient, peaking 10 min after stimulation and returning to basal levels ∼2 h later. The kinetics of PAX5 phosphorylation were similar to those of ERK1/2 phosphorylation and, consistent with these kinetics, BLIMP1 mRNA expression was also transient and returned to the basal level within 24 h (data not shown). After anti-IgM Ab stimulation, BLIMP1 expression in Ramos cells is not sufficient to suppress PAX5 and initiate the above-described positive-feedback loop to replace PAX5 with BLIMP1. BCR stimulation with Ag is not the only stimulation required for plasma cell differentiation. Stimulation with cytokines, such as IL-2, IL-4, IL-10, and IL-21, and contact-dependent engagement of CD40 on B cells by CD40L (CD154), expressed by activated CD4+ T cells, are required for plasma differentiation (22–26). Costimulation with these cytokines and T cells might prolong and enhance ERK1/2 phosphorylation and PAX5 phosphorylation and enable enough BLIMP1 expression to initiate plasma cell differentiation. The other possible reason is that Ramos cells, a lymphoma cell line, have impaired differentiation, as do tumor cells. BLIMP1 expression is induced by the cooperation of transcription factors, such as STAT3, IRF-4, and NF-κB (27, 28). Normal GCB cells express these factors properly and are ready to respond to BCR stimulation, which might enable a rapid and substantial increase in BLIMP1 expression in response to even transient PAX5 phosphorylation.
Other researchers reported the phosphorylation of PAX family proteins by the MAPK superfamily. PAX2 is phosphorylated by JNK at the transactivation domain (29), and both ERK1/2 and p38 phosphorylate PAX6 at the same sites: serines 376 and 413 and threonine 323 (30). This phosphorylation enhances the transcriptional activities of PAX family proteins. In addition, PAX6 phosphorylation by homeodomain-interacting protein kinase 2 (31), sumoylation of PAX6 (32), and acetylation of PAX5 by p300 (33) are reported to be posttranslational modifications of PAX family proteins, and all enhance the transactivation of PAX family proteins; therefore, negative regulation of PAX5 function by phosphorylation seems to be unique. Serines of PAX5 phosphorylation sites are not conserved in any other PAX family proteins. PAX5 is the only PAX family protein that regulates the differentiation of hematopoietic cells and might obtain a unique method to respond to extracellular signals. Furthermore, the cancellation of PAX family-dependent transcriptional repression by phosphorylation may also be unique to PAX5, although the effect of phosphorylation on transcriptional repression by other members of the PAX family has not been investigated.
The aberrant expression of normal PAX5 protein by the fusion gene between the potent enhancer of the IGH gene and the PAX5 promoter was found in non-Hodgkin’s lymphoma patients with the chromosomal translocation, t(9;14)(p13;q32) (34, 35). The oncogenicity of this fusion gene might be explained by impaired initiation of plasma cell differentiation due to sustained repression of BLIMP1 by overexpressed PAX5. Similarly, phosphorylation-defective mutation of PAX5 might impair plasma cell differentiation and cause lymphoma; therefore, we examined the genome DNA sequence surrounding PAX5 phosphorylation sites in 85 cases of diffuse large B cell lymphoma, but no mutation was found (data not shown; information on samples and sequencing methods is described in Materials and Methods). PAX5 mutations in lymphoma cells, if they exist, might be at the ERK1/2 binding site of PAX5, which is currently unknown.
In summary, our study provides evidence for an ERK1/2 pathway that phosphorylates PAX5 in response to BCR stimulation; this increase in PAX5 phosphorylation may associate with attenuated transcriptional repression by PAX5, derepression of BLIMP1, and initiation of plasma cell differentiation. This work provides new insight into the regulation of PAX5 function and establishes a novel relationship among the BCR signal, ERK1/2, and PAX5.
T.N. received research funding from Otsuka Pharmaceutical Co., Ltd., Kyowa Hakko Kirin Co., Ltd., Wyeth, and Chugai Pharmaceutical Co., Ltd. K.S. is a graduate student at Nagoya University and an employee of Otsuka Pharmaceutical Co., Ltd.
We thank Tomoko Kawake, Yoko Matsuyama, Asako Watanabe, and Chika Wakamatsu for technical assistance.
This work was supported by Grants-in-Aid from the National Institute of Biomedical Innovation and the Ministry of Education, Culture, Sports, Science, and Technology of Japan.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- B lymphocyte-induced maturation protein 1
- germinal center B
- small interfering RNA
- untranslated region.
- Received October 21, 2011.
- Accepted April 17, 2012.
- Copyright © 2012 by The American Association of Immunologists, Inc.