Arginine methylation catalyzed by PRMT1 is required for B cell activation and differentiation

Arginine methylation catalyzed by protein arginine methyltransferases (PRMT) is a common post-translational modification in mammalian cells, regulating many important functions including cell signalling, proliferation and differentiation. Here we show the role of PRMT1 in B-cell activation and differentiation. PRMT1 expression and activity in human and mouse peripheral B cells increases in response to in vitro or in vivo activation. Deletion of the Prmt1 gene in mature B cells establishes that although the frequency and phenotype of peripheral B cell subsets seem unaffected, immune responses to T-cell-dependent and -independent antigens are substantially reduced. In vitro activation of Prmt1-deficient B cells with a variety of mitogens results in diminished proliferation, differentiation and survival, effects that are correlated with altered signal transduction from the B cell receptor. Thus PRMT1 activity in B cells is required for correct execution of multiple processes that in turn are necessary for humoral immunity.

PRMT1 is the major arginine methyltransferase active in mammalian cells and is required for normal embryogenesis, cell cycle progression, cell viability, and signal transduction 1, 2, 6, 10 . PRMT1 methylates histones, RNA-binding proteins (RBPs), cell cycle proteins and proteins involved in regulating gene transcription, including high-mobility group proteins (HMGA) and runt-related transcription factor 1 (RUNX1) 1,2 . PRMT1 is also an active component of signal transduction pathways including those from the B-cell receptor (BCR), T-cell receptor (TCR), nerve growth factor receptor, type 1 interferon receptor, and the pathway involving NF-AT [11][12][13][14][15][16] . PRMT1 is also known to directly regulate the activity of forkhead box protein O1 (FOXO1) 18 and Bcl-2-associated death promoter (BAD) 17 , proteins associated with cell viability. In these examples, the motif RxRxxS/T recognized by protein kinase B (PKB) and present in both FOXO1 and BAD, overlaps with sites recognized by PRMT1, leading to competition in which methylation by PRMT1 inhibits phosphorylation by PKB, thereby prolonging the localization of FOXO1 in the nucleus in one case and inhibiting the proapoptotic activity of BAD in the other 17,18 . Clearly, arginine methylation catalyzed by PRMT1 is crucial to a multitude of pathways although the regulation of its activity is unclear in the majority of cases.
We described previously a PRMT1 target motif within the Ig-α signalling component of the BCR 14 . This evolutionarily conserved arginine is proximal to the immunoreceptor tyrosine activation motif (ITAM) of Ig-α, is methylated directly by PRMT1 in immature B cells, and regulates the activation and differentiation of these cells following BCR ligation 14 . Interestingly, arginine methylation of Ig-α in immature B cells is transiently diminished following BCR ligation, indicating this methylation could be a controlled event and therefore, arginine methylation could contribute to the regulation of B-cell activation in physiological settings 14 . To investigate the possible involvement of PRMT1 in Bcell differentiation further, we undertook an analysis of immune function in mice with Prmt1 deleted specifically in mature B cells. This analysis revealed the post-translational modification of proteins catalysed by PRMT1 to be a regulated event, and to be essential for normal proliferation, differentiation and survival of activated B cells in vitro and for normal antibody responses to antigen in vivo.

Results
PRMT1 is dispensable in mature B cells. We analyzed the function of PRMT1 in mature B cells by generating mice in which Prmt1 was deleted only in the periphery, which also circumvented the congenital lethality of its deficiency 4,19 . This was done by crossing Prmt1 f/f mice with mice in which Cre recombinase was expressed under the control of CD23 regulatory elements 20 , thereby initiating deletion at the T2 stage of B-cell development to produce a peripheral B-cell compartment that was Prmt1 deficient. Analysis of B-cell development in the spleens of these Prmt1 f/f CD23Cre mice revealed no abnormalities in B-cell number, phenotype or distribution (Fig. 1). This was true for both immature and mature B cells, distinguished by CD93 expression, and for marginal zone and follicular B cells, resolved by CD21 and CD23 expression (Fig. 1a-c). The localization of B cells in the splenic white pulp also was unaffected by loss of PRMT1 (Fig. 1d). Thus, despite the absolute requirement for PRMT1 in embryogenesis 4 , it was not required for the appearance or maintenance of B-cell subsets in the periphery.
B-cell activation increases PRMT activity. We next examined the amount of PRMT1 in B cells, both resting and after activation. B cells were purified from the spleens of control and Prmt1 f/f CD23Cre mice and the amount of PRMT1 assessed by western blot before and after stimulation with CD40L in the presence of interleukins (IL) 4 and 5. PRMT1 was detected in unstimulated control B cells and in increased amounts following activation ( Fig. 2a). Amounts of PRMT1 increased after stimulating control B cells with either lipopolysaccharide (LPS) or F(ab′) 2 anti-IgM, albeit to a greater extent with LPS (Fig. 2b). As expected, PRMT1 was not detected in Prmt1 f/f CD23Cre B cells (Fig. 2a, b). The presence of PRMT1 in control B cells and its increase following activation suggested that PRMT1 activity, and thus the distribution of proteins containing asymmetrically dimethylated arginine, would also change following B-cell stimulation. We assessed PRMT1 activity in resting and activated B cells by two methods. First, total cell lysates from resting and activated, control and Prmt1 f/f CD23Cre B cells were separated by gel electrophoresis and probed for the presence of proteins containing asymmetric dimethylated arginines using a specific antibody. In resting, control B cells, several bands were revealed, indicating constitutive arginine methylation of a subset of proteins (Fig. 2c).
Despite the absence of PRMT1, asymmetrically dimethylated proteins were detected in lysate from unstimulated Prmt1 f/f CD23Cre B cells, but at a frequency and intensity that was less than in control B-cell samples (Fig. 2c), and presumably reflected the activity of other type I PRMTs in these cells. The intensity of asymmetric dimethylated arginine-containing protein bands increased in control B cells following activation with CD40L, coincident with the increased amounts of PRMT1 (Fig. 2a, c). Some bands corresponded in molecular weight to those present in the unstimulated control B-cell sample, but the intensity was increased and new bands were visible (Fig. 2c). The number and intensity of arginine methylated protein bands also increased in CD40L-stimulated Prmt1-deficient B cells relative to their unstimulated sample, but again their frequency and intensity were less than in equivalently treated control samples (Fig. 2c). Importantly, unique bands were detected in the stimulated control samples that were not present in the equivalent Prmt1-deficient sample, indicating unique PRMT1 substrates in B cells. The second approach assessed ongoing or active methylation by culturing B cells in the presence of both radioactive L-[methyl-3H] methionine and protein synthesis inhibitors. This assay did not detect ongoing methylation in unstimulated B cells from either Prmt1 f/f CD23Cre or control animals, but the activity increased significantly in both genotypes following stimulation (Fig. 2d). Again, the extent and intensity of labelling differed between control and Prmt1-deficient B cells, reflecting increased methyltransferase activity in the control B-cell sample. Collectively these results indicate that PRMT1 is a major asymmetric dimethytransferase in B cells, that its activity increases following stimulation and that it has unique substrates in B cells.
In light of the role PRMT1 activity has in regulating signalling events downstream of the BCR in immature B cells 14 , we examined the consequences of PRMT1 deficiency on signal transduction following BCR ligation on mature, naive B cells. Control and Prmt1-deficient B cells were stimulated with F(ab′) 2 anti-IgM, then lysed after 0, 5 or 30 min and the distribution of tyrosine phosphorylated proteins determined (Fig. 2e). At time zero, the few tyrosine phosphorylated protein bands were distributed similarly in both B-cell samples. After 5 min stimulation, the intensity and number of such bands had increased substantially in both Prmt1-deficient and control B-cell samples, again with similar distributions. At 30 min post stimulation, the intensity of bands in both samples was reduced from their 5 min peaks, however the reduction was less in Prmt1-deficient B cells than in controls, resulting in a relative hyper-phosphorylation in the absence of PRMT1 (Fig. 2e). Interestingly we also found increased phosphorylation of ERK1/2, FOXO1 and PKB in equivalently anti-IgM stimulated Prmt1deficient B cells (Fig. 2e, f). These results revealed activationinduced increases in PRMT1 activity under a variety of conditions and an apparent prolongation of signalling from the asy-R m BCR in the absence of PRMT1. We also examined CD19 + B cells isolated from human peripheral blood for PRMT1 amounts before and after CD40L + IL-4 activation and found the amount also increased following stimulation ( Supplementary Fig. 1a).
PRMT1 deficiency affects B-cell proliferation and differentiation. The increased amount of PRMT1 in B cells after stimulation suggested a possible role for protein arginine methylation in B-cell activation. To assess this, we labelled control and Prmt1deficient mouse B cells with the division tracking dye, CTV, and then cultured the cells with CD40L, IL-4 and IL-5. The extent of B-cell proliferation and differentiation were compared after 4 days using, respectively, the distribution of cell divisions and the frequency of antibody secreting cell (ASC) as indicated by CD138 expression 21 . This analysis revealed reduced proliferation and differentiation in the Prmt1 f/f CD23Cre B-cell cultures (Fig. 3a, b). The impact of Prmt1 deficiency on proliferation was apparent throughout the culture, as assessed by counting the number of B cells on successive days (Fig. 3c). To separate effects on proliferation from differentiation, which are intimately linked 22 , we assessed the division profiles of control and Prmt1-deficient B cells after 4 days stimulation with CD40L, IL-4 and IL-5 for the relative frequency of CD138 + cells, thereby normalizing for proliferation (Fig. 3d). While this revealed division-dependent increases in differentiation in both control and Prmt1-deficient B-cell cultures, there was significantly less differentiation per division in the latter. Equivalent experiments using LPS stimulation ( Fig. 3e-g) also revealed significantly less proliferation and differentiation in Prmt1-deficient B cells compared to controls, which was apparent again over the course of the culture ( Fig. 3e-g). Thus Prmt1 deficiency affected both B-cell proliferation and differentiation, although the effect on the former appeared to be more marked.
Lymphocyte activation induces metabolic changes that have profound effects on cell proliferation and differentiation 23 . In B cells, for example, metabolic reprogramming is required for antibody production 24 . Given the requirement for PRMT1 in normal B-cell activation, next we compared the metabolic pathways utilized in control and Prmt1-deficient B cells. This   revealed that the basal oxygen-consumption rate (OCR) in unstimulated B cells was not influenced by Prmt1 deficiency (Supplementary Fig. 2a) but conversely, PRMT1 was required for basal and maximal respiratory capacity as well as glycolytic capacity in activated B cells ( Supplementary Fig. 2a-c). Thus, PRMT1 activity was required for normal B-cell responses to stimuli that mimic aspects of T-cell dependent (TD) or T-cell independent (TI) responses and this included the metabolic reprogramming that follows activation.
PRMT1 is required in B cells for humoral immunity. The increased amounts of PRMT1 and asymmetrically arginine methylated proteins following in vitro activation of B cells suggested that similar changes might occur in in vivo activated B cells. We purified therefore naive B cells, germinal centre (GC) B cells and ASC from mice 7 days after immunization with a TD antigen, prepared lysates and probed each sample for both PRMT1 and proteins containing asymmetrically dimethylated arginines (Fig. 4a). The amount of PRMT1 was low in non-GC splenic B cells, but increased in both GC B cells and, to a lesser extent, in ASC. Probing lysates for asymmetrically argininemethylated proteins revealed few bands in naive B cells, but these increased in number and amount in both the GC B cell and ASC samples (Fig. 4a). We purified also naive (CD20 + CD27 − CD38 lo ), GC (CD20 hi CD27 + CD38 hi ) and memory (CD20 + CD27 + CD38 lo ) B cells from human tonsils and found PRMT1 to be readily detected in all samples, although the amount was slightly increased in GC B cells (Supplementary Fig. 1b). Examining these samples for proteins containing asymmetric dimethlyated arginines revealed a similar pattern in all cell types, but with some bands of increased intensity in the memory and GC B-cell fractions ( Supplementary Fig. 1c). Thus, human B cells ex vivo also showed activation-dependent increases in PRMT1 amounts and protein arginine methylation.
To assess possible functions of PRMT1 in B-cell responses in vivo, Prmt1 f/f CD23Cre and control mice were immunized intraperitoneally with the TD antigen (4-hydroxy-3-nitrophenyl) acetyl coupled to keyhole limpet hemocyanin (NP-KLH) delivered with an alum adjuvant. In the ensuing immune  Supplementary Fig. 9 response, we monitored the frequency of isotype-switched, antigen-specific B cells and total GC B cells (Fig. 4b). Remarkably, NP + IgG1 + B cells, PNA + FAS + GC B cells and physical GC structures were not detected in the spleens of immunized Prmt1 f/ f CD23Cre mice, in contrast to their abundance in control mice (Fig. 4b, c). Whether Prmt1-deficient B cells generated NPspecific IgG1 + ASC in spleen or NP-specific serum antibody in vivo was determined by ELISpot assays and ELISA, which revealed a more than 100-fold reduction in both ASC frequency and antibody titres in Prmt1 f/f CD23Cre mice (Fig. 4d-f). Importantly, these differences were reflected as absolute numbers (Fig. 4g-i) with similar differences recorded 5 and 14 days after immunization. To determine whether a defective immune response was restricted to hapten antigens, we infected Prmt1 f/f CD23Cre mice with influenza virus and found similar defects in GC and ASC production ( Supplementary Fig. 3). In this case, both Prmt1 f/f CD23Cre and control mice increased the number of CD8 + D b NP366-binding cells, confirming successful infection and an absence of any global immune suppression in Prmt1 f/ f CD23Cre mice ( Supplementary Fig. 3d, e).
Prmt1 f/f CD23Cre mice were also challenged with NP-Ficoll, a TI antigen, and again failed to generate a normal response (Fig. 5). While the frequency of NP + IgM + B cells increased in the spleens of Prmt1 f/f CD23Cre mice following this immunization (Fig. 5b), indicating a degree of clonal expansion, this was substantially less than occurred in controls, as was the amount of NP-specific IgG3 antibody in serum (Fig. 5d-h). Thus, responses to both TD and TI antigens were defective in mice with Prmt1deficient B cells.
Immune defects are strictly B-cell intrinsic. The immune defects identified in mice with Prmt1-deficient B cells could in principle arise from an inability of such B cells to induce appropriate CD4 + T-cell help, potentially compounding an otherwise modest B-cell defect. We examined this possibility by creating BM radiation chimeras with 50% of hematopoietic cells derived from Ly5.2 Prmt1 f/f CD23Cre BM and 50% from Ly5.1 congenic BM that was Prmt1 wild type. Control chimeras were made with equal amounts of Ly5.2 CD23Cre and Ly5.1 BM, all cells being Prmt1 sufficient. Nine weeks after reconstitution, both groups of mice were immunized with NP-KLH in alum and analyzed 7 days later for the presence of antigenspecific B cells of either allotype in the spleen (Supplementary Fig 4). Prmt1-deficient B cells, although equally represented in the naive B-cell population, failed to generate detectable NP + IgG1 + B cells, despite the presence of control NP + IgG1 + B cells and T-cell help ( Supplementary Fig. 4a, b). The same Prmt1-deficient B-cell deficit was apparent in the NP-specific ASC compartment, identified as NP + CD138 + B220 lo and then partitioned by expression of Ly5.1 ( Supplementary Fig. 4c, d). In control chimeras, both allotypes were represented equally in the NP-reactive B-cell and ASC populations ( Supplementary Fig. 4). The immune response defect was still apparent when we examined all GC B cells, irrespective of specificity. Thus, Prmt1-deficient B cells were unable to either initiate or enter into a GC reaction.

PRMT1 is required in B cells for a memory response.
The previous experiments revealed a requirement for PRMT1 in B cells for normal humoral immune responses. These data, however, did not test whether PRMT1 was required in memory B cells to mount a recall response. To assess this requirement, we first created mice in which Prmt1 deficiency was both inducible and restricted to the B-cell lineage. This was done by reconstituting lethally irradiated recipients with a BM mixture comprising 80% from mice homozygous for the µMT mutation, and thus unable to generate B cells 25 , and 20% from mice that were Prmt1 f/f and carried the Rosa26CreERT2 allele encoding a constitutively expressed, Tamoxifen inducible form of Cre 26 . In control chimeric mice, the 20% BM donor was Prmt1 +/+ Rosa26CreERT2. After reconstitution, these two groups of mice were immunized with NP-KLH in alum and 9 weeks later, dosed on 3 successive days with Tamoxifen to induce activation of Cre and deletion of the Prmt1 f/f alleles. The mice were then boosted with soluble NP-KLH and examined 5 days later for the presence of both NP-specific B cells and GC B cells (Fig. 6). Boosted mice in which  Prmt1 had been deleted from memory B cells showed a significant deficit in NP + IgG1 + and GC B cells compared to controls in which Prmt1 was retained (Fig. 6a, b). Importantly, this treatment did not significantly reduce the overall number of B cells in spleen or BM (Fig. 6c). Thus, as in a primary immune response, PRMT1 activity was required for a normal memory B-cell response.
PRMT1 regulates survival following BCR ligation. B-cell responses to TD antigens depend on stimulation through CD40L and the BCR. While the reduced proliferation and differentiation of Prmt1-deficient B cells in response to stimulation with CD40L and cytokines could contribute to defective immunity, it was possible that other pathways were affected. To determine whether Prmt1 deficiency altered proliferation following BCR ligation, B cells purified from spleens of Prmt1 f/f CD23Cre and control mice were labelled with CTV and cultured for 4 days in the presence of F(ab′) 2 anti-IgM to crosslink the BCR (Fig. 7a, b). On days 2, 3 and 4, the number of cells present in the cultures and the extent of cell division were assessed, revealing a reduced number of divisions, a reduced number of cells in each division and significantly reduced overall viability at   Supplementary Fig. 11 all times for Prmt1-deficient B cells compared to controls (Fig. 7a,  b). The nature of the cell death induced by BCR ligation was examined by repeating the cultures in the presence of the pancaspase inhibitor, QVD, which reduced cell death at 24 h such that viability in the Prmt1 f/f CD23Cre and control cultures was equal (Fig. 7c). That QVD blocked the increased apoptosis in Prmt1-deficient B cells following BCR stimulation suggested a connection between PRMT1, BCR ligation and the proteins that protect against activation induced cell death 27 . We examined therefore the amounts of BCL2 family members MCL1, BCL2, A1 and BCLX in Prmt1 f/f CD23Cre and control B cells before and after stimulation with anti-IgM (Fig. 7d-f). MCL1, which was equally abundant in unstimulated Prmt1 f/f CD23Cre and control B cells, was reduced in Prmt1-deficient B cells following BCR ligation. BCL2 decreased in both control and Prmt1 f/f CD23Cre B cells following BCR ligation, but to a greater extent in Prmt1-deficient B cells. A1 was induced in both groups, but less so in Prmt1 f/f CD23Cre B cells (Fig. 7e). BCLX was detectable in equal amounts prior to and after both stimulations in control B cells, but was barely detectable in unstimulated Prmt1-deficient B cells and increased marginally following BCR ligation (Fig. 7f). Thus, PRMT1 was required for appropriate regulation of expression of pro-survival members of the BCL2 family in B cells before and after BCR stimulation such that in its absence, apoptosis was significantly increased.
Discussion PRMT1, considered the predominant type I arginine methyltransferase activity in mammalian cells, regulates a multitude of molecular and cellular functions 2 . This includes gene transcription, responding to DNA damage, mRNA translation, cell division, apoptosis and signal transduction 14,18,19,28,29 . Despite such individual examples, potential roles for PRMT1 in B-cell immune responses have not previously been reported. In the present study we have examined the role of PRMT1 in mature B cells. We have shown that the increased activity of PRMT1 upon B-cell activation serves to co-ordinate a number of crucial processes that are required for a successful immune response and that are dysregulated in its absence. These dysregulated processes include cell cycle progression, metabolism and survival, the last two being novel findings of this work. Indeed, the altered expression of BCL2 family pro-survival proteins in the absence of PRMT1 could be a major contributor to the immunodeficiency we have observed in these mice, as MCL1, one of the altered proteins, is essential for GC B-cell and plasma cell survival 26,30 .
Asymmetrically dimethylated arginine proteins are present in naive B cells before increasing in amount significantly following B-cell activation. Given that we found protein methylation activity to be essentially undetectable in naive B cells over a 3-h interval in vitro, this result implies that arginine methylation is a particularly stable post-translational change in these cells. Interestingly, we observed several novel asymmetrically methylated protein bands in both control and Prmt1-deficient B-cell samples following stimulation, indicating unique and redundant substrates of PRMT1 that presumably participate in the processes that are dysregulated in the enzyme's absence. Identifying these substrates and the activities that are dependent on PRMT1 will be the basis of significant future work, determining how methylation modulates their roles in proliferation, differentiation and survival following B-cell activation by a variety of stimuli.
Activation of Prmt1-deficient B cells is associated also with significant apoptosis. This cell death could reflect the defect in mitochondrial respiratory capacity, meaning that PRMT1 is required for either the metabolic changes that sustain B-cell growth and viability following activation, or to integrate those changes into a viability signal. The marked dysregulation in expression of pro-survival members of the BCL2 family of proteins observed in response to BCR stimulation of Prmt1-deficient B cells could also contribute to increased apoptosis. Specifically, the inability of Prmt1-deficient B cells to sustain MCL1 and BCL2 proteins following BCR ligation, or to upregulate and/or maintain A1 protein and BCLX could all contribute to increased cell death after BCR ligation. Indeed, the ability of the broad caspase inhibitor, QVD, to block apoptosis of Prmt1-deficient B cells is consistent with such a role. These results reveal a previously undescribed role for PRMT1 in cell death. Collectively, impairments in proliferation, differentiation, metabolism and survival could combine to create the near-complete block in B-cell immune responses observed in vivo.
The consequences on in vivo B-cell behaviour of Prmt1 deletion are not restricted by the nature of the antigen. We observed a deficit of plasma cells and GC B cells in mice containing Prmt1deficient B cells following either immunization with a protein antigen in adjuvant or infection with influenza virus. In response to immunization, antigen-specific, isotype-switched antibody is detected in the serum of Prmt1-deficient mice along with rare IgG1 ASC in spleen, suggesting at a minimum the initiation of an immune response. That we measured equal expansion of influenza specific CD8 + T cells in control and Prmt1 f/f CD23Cre mice following infection rules out technical issues or systemic immune suppression in the Prmt1-deficient group. Similarly, observing a degree of clonal expansion in response to NP-Ficoll, indicates that antigen specific Prmt1-deficient B cells were exposed to and at least transiently activated by antigen. Reflecting these in vivo results, in vitro proliferation of Prmt1-deficient B-cells in response to a range of mitogens, although significantly less than in controls, does occur to some extent. We also investigated the requirement for PRMT1 in a memory B-cell response to a TD antigen and found it again to be indispensable. Our finding that PRMT1 protein expression and activity in human B cells mirrored that in mice B cells-being higher in GC and plasma cells compared to naive B cells and being induced by in vitro activation -suggests the roles of PRMT1 in B cells will be conserved between these species and that inhibiting PRMT1 activity may be beneficial in situations of pathogenic B-cell hyperactivity.
Arginine methylation within the cytoplasmic tail of Ig-α alters the consequences of ligation of the pre-BCR, increasing the activity of PI3K and thus promoting differentiation over proliferation 14 . We showed here in mature B cells that BCR ligation in the absence of PRMT1 also leads to elevated and prolonged PI3K activation, as indicated by PKB phosphorylation. Similarly, and perhaps counterintuitively given the overall hyporesponsiveness at the cell level, total protein tyrosine phosphorylation was increased and prolonged in PRMT1-deficient B cells, as was phosphorylation of FOXO1 and ERK1/2. These results suggest that in the period shortly after BCR stimulation, PRMT1 acts as a negative regulator of signalling in a role that is apparent in both pre-B and follicular B cells.
PRMT1 regulates the activity and/or persistence of FOXO1 by controlling the access of PKB 17,18 . Interestingly, FOXO1 is active in GC B cells where its predominant role is to establish the dark zone (DZ) 28,31 . FOXO1 is required also for class switch recombination and affinity maturation in vivo 28,31 . Equally, PI3K activity is an important regulator of both GC formation and class switch recombination, promoting cell survival and proliferation on one hand and repressing AID activity on another 32 . The compromised appearance of GC and class switched, antigenspecific B-cells in mice carrying Prmt1-deficient B cells could be explained in part by reduced FOXO1 arginine methylation leading to increased phosphorylation by PKB and thus inactivation. These consequences of not methylating FOXO1 could be amplified by the extended PI3K activity we observed in Prmt1-deficient B cells following BCR ligation. Thus, the combined effects of dysregulated FOXO1 and PI3K activity could perturb B-cell proliferation, class switch recombination, plasma cell differentiation and ultimately the GC reaction. Determining the precise mechanisms by which PRMT1 activity regulates different components of B-cell activation will be a challenging but essential part of understanding how this enzyme coordinates multiple aspects of B-cell biology.
PRMT1 has many substrates 2 , which may explain why its absence has such profound, heterogeneous and often cell typeand stage-specific effects. For example, PRMT1 could be involved in regulating proliferation through its action on MRE11, which is thought to contribute to the chromosomal instability and cell cycle checkpoint abnormalities seen in PRMT1 deficient embryonic fibroblasts 19,33 . More recently, PRMT1 was shown to methylate cyclin-dependent kinase 4 (CDK4) in FOXO1expressing pre-B cells, preventing or disrupting complexes between CDK4 and cyclin-D3, and thereby blocking cell cycle progression and promoting differentiation 34 . It is possible that either or both these activities contribute to the deficit of B-cell expansion observed here in vitro and in vivo and further work will identify the relevance of these mechanisms in mature B cells.
It is possible also that the consequences of Prmt1 deletion in B cells are mediated not by the loss of PRMT1 activity per se, but rather through a concomitant alteration in the activity of another PRMT, which is what actually affects the GC reaction and plasma cell differentiation 35 . The activity of the transcription factor E2F-1, for example, varies depending on the relative amounts of PRMT1 and PRMT5, which compete for access to the same methylation site, promoting apoptosis and proliferation, respectively 36 . In addition to the role of PRMT1 in GC formation reported here, PRMT7 has been shown to repress transcription of Bcl-6 via H4R3 methylation in its promoter 37 . Deletion of Prmt7 in B cells increases Bcl6 expression, promoting GC development and repressing Irf4 and Prdm1, genes that are required for plasma cells differentiation 37 . Conceivably, any interplay between PRMT1 and PRMT7 in which one affected the activity of the other, could contribute to the in vivo B-cell phenotype observed here. The intersection of the activities of the different PRMT, which may be substrate specific, suggests an assessment of all remaining arginine methylation activity may be required to fully understand the mechanisms underlying the phenotypes observed in the absence of any individual PRMT.
A previous report has described normal TD responses and abnormal TI responses in B-cell Prmt1-deficient mice 38 . B cells from these mice were also found to be normal or even augmented in their proliferation in response to mitogen stimulation in vitro. However, these earlier studies differed from those reported here in two respects. First, Prmt1 was deleted in the earliest stages of B-cell development using a CD19-Cre, potentially enabling compensation of its absence during development. Second, Prmt1 deletion was incomplete with PRMT1 protein detected in peripheral B cells, which means some of the reported results may be dosage related 38 .
In conclusion, our results identify PRMT1 as a central regulator of humoral immunity, establishing a previously unidentified role of this enzyme in the activation, proliferation and differentiation of B cells. Our findings have demonstrated the dynamic induction of arginine methylation in activated B cells and have revealed striking similarities in PRMT1 expression and activity between mouse and human B cells, suggesting conserved roles in supporting proper immune responses in both species. We have also established that, in activated B cells, PRMT1 is required to sustain MCL1 and BCL2 protein expression, revealing a previously unknown role of PRMT1 in apoptosis. Overall, our studies indicate that PRMT1 activity may be essential to integrate multiple pathways into a coherent outcome that is manifest as both B cell activation and an immune response.

Methods
Mice and immunizations. Cd23Cre mice 20 were provided by Meinrad Busslinger and the creation of Prmt1 fl/fl mice has been described 19 . All mice were maintained at the Walter and Eliza Hall Institute of Medical Research (WEHI) on a C57BL/6 background. Animal procedures were approved by the WEHI Animal Ethics Committee and mice used in the study were male or female, and 8 weeks of age on commencement. No animals were specifically excluded from the study and the investigators were not blinded to the groups during the experiments or analyses. For primary immune responses, mice were injected intraperitoneally (i.p.) with 100 μg of 4-hydroxy-3-nitrophenyl)acetyl (NP) conjugated to keyhole limpet hemocyanin (KLH) at a molar ratio of (13-20):1, precipitated on 10% alum. For secondary responses, mice were boosted i.p. with 50 μg of NP-KLH in PBS. Sample sizes were calculated to detect with 80% power a difference of at least 30% in the means with a variance of 20%, with equal numbers in all groups. For influenza infections mice were inoculated with 10 4 p.f.u. of HKx31 (H3N2) influenza virus 39 .
Histology, ELISPOT and ELISA. Portions of spleen were frozen in OCT (Tissue-Tek Sakura), and 7 μm sections were cut with a microtome (Leica) at −20°C. Sections were fixed in ice-cold acetone for 10 min, air-dried, then stained as indicated in the figures with goat anti-mouse IgM conjugated to HRP (Southern Biotech) at 1:200 dilution, biotinylated anti-IgD (clone 1126c; Southern Biotech) at 1:100 dilution, or anti-GL7 (clone GL7; made in-house) at 1:200 dilution. Biotinylated antibodies were revealed using streptavidin alkaline phosphatase (Southern Biotech) at 1:200 dilution. Unlabelled antibody was detected with anti-rat kappa HRP (Southern Biotech) at 1:200 dilution. Staining was visualised with an AEC substrate kit (Vector Laboratories) and Vector Blue substrate kit (Vector Laboratories) 41 and slides were mounted using Aqua Polymount (Polysciences, Inc.). ASC and serum antibody titres were analyzed by ELISPOT and ELISA, respectively 41 . For ELISA, at the times indicated blood was collected and serum separated. 96-well plates were coated with 20 µg ml −1 NP 13 -BSA (conjugated in-house), and diluted serum was incubated for at least 20 h at room temperature. NP-specific IgG1 was