Environmental sensing by mature B cells is controlled by the transcription factors PU.1 and SpiB

Humoral immunity requires B cells to respond to multiple stimuli, including antigen, membrane and soluble ligands, and microbial products. Ets family transcription factors regulate many aspects of haematopoiesis, although their functions in humoral immunity are difficult to decipher as a result of redundancy between the family members. Here we show that mice lacking both PU.1 and SpiB in mature B cells do not generate germinal centers and high-affinity antibody after protein immunization. PU.1 and SpiB double-deficient B cells have a survival defect after engagement of CD40 or Toll-like receptors (TLR), despite paradoxically enhanced plasma cell differentiation. PU.1 and SpiB regulate the expression of many components of the B cell receptor signaling pathway and the receptors for CD40L, BAFF and TLR ligands. Thus, PU.1 and SpiB enable B cells to appropriately respond to environmental cues. Although important for early development, PU.1 is dispensable for mature B cell function, possibly owing to compensation by the related transcription factor SpiB. Here the authors show PU.1 and SpiB are collectively required for humoral immunity, through regulation of germinal centre formation and plasma cell differentiation.

A ntibody-mediated immunity relies on the ability of B cells to respond to multiple environmental stimuli including antigen, Toll-like receptor (TLR) ligands, and T-cellderived help, including CD40L and the cytokines interleukin-4 (IL-4) and IL-21. The survival of mature B cells and plasma cells also depends on members of the tumor necrosis factor receptor superfamily (TNFRSF), including the B-cell-activating factor receptor (BAFF-R) 1 . Mature B cells, including follicular and marginal zone (MZ) B cells, are quiescent and relatively long lived. After exposure to cognate antigen, B cells re-enter the cell cycle and undergo multiple rounds of division, as well as initiating immunoglobulin class switch recombination (CSR) 2 . Proliferating B cells have the potential to differentiate into short-lived plasmablasts that provide the immediate, but low affinity, antibody that is important early in the immune response. Alternatively, in response to antigen and T cell help, activated B cells can enter a structure termed the germinal center (GC), where they undergo clonal amplification and somatic hypermutation and differentiation into plasma cells that secrete high-affinity antibodies 2 . GCs also produce memory B cells that can rapidly differentiate into plasma cells upon re-exposure to antigen.
A complex network of transcription factors controls each aspect of the B cell response to antigen. This network includes factors that are essential for B cell proliferation and the GC response, including PAX5, BACH2, IRF4/BATF, IRF8, NFκB, Eproteins (E2A, E2-2) and Oct2/OBF1, whereas a smaller group, including high concentrations of IRF4, BLIMP-1/PRDM1, ZBTB20 and XBP1, are required for plasma cell differentiation and antibody production (reviewed in refs. [3][4][5]. We have reported a role for a complex of the transcription factors PU.1 and IRF8 in negatively regulating plasma cell differentiation in cell culture, although the role of these factors in vivo is unclear 6,7 . The Ets family transcription factor PU.1, encoded by the Spi1 gene, is a major regulator of haematopoiesis, controlling the expression of hundreds of genes including growth factor receptors, adhesion molecules, transcription factors and signaling components 8 . PU.1-deficient mice lack all lymphocytes, including B cells, suggesting that PU.1 is an essential regulator of the B cell developmental pathway [9][10][11][12] ; however, this requirement is limited to early lymphopoiesis as conditional deletion of PU.1 in CD19expressing B cells is compatible with normal development and function 10,[13][14][15][16] . This minimal consequence of PU.1 loss in B cells is surprising, as PU.1 is well-known to bind tens of thousands of sites in the B cell genome. One possible explanation for this discrepancy is the strong expression of SpiB, the most closely related Ets family member in B cells, that binds to the identical nucleotide sequence GGAA 17,18 . Indeed, Spi1 +/− Spib −/− mice have markedly reduced B-cell numbers 19 , and inactivation of PU.1 and SpiB in bone marrow B-cell progenitors results in a block in development at the pre-B-cell stage and the formation of pre-B acute lymphoblastic leukemia (ALL) 15 . A similar developmental block and leukemogenesis is also observed in B cells lacking PU.1/IRF4 or IRF4/IRF8, demonstrating that an Ets/IRF complex controls bone marrow B-lymphopoiesis 13,20 .
To better define the function of PU.1 and SpiB in mature B cells, we here utilize the Cd23-Cre strain 21 to inactivate PU.1 in Spib −/− mice at the transitional B-cell stage in the spleen, thus allowing us to bypass the developmental block that occurs when both PU.1 and SpiB are deleted in immature B cells. We find that PU.1/SpiB redundantly control mature B cell survival and differentiation in response to a diverse range of stimuli. PU.1/SpiBdeficient B cells cannot signal efficiently through the B-cell receptor (BCR) and thus mutant mice lack antigen-specific B-cell responses entirely. PU.1 and SpiB directly control many components of the BCR pathway as well as important receptors for Tcell-derived signals and microbial products. In keeping with this observation, PU.1/SpiB-deficient B cells are not very responsive to these stimuli in vitro, but plasma cell lineage differentiation is increased, a cellular stage in which Spi1 is lowly expressed and the Spib gene is silenced. These findings highlight PU.1 and SpiB as cell intrinsic regulators of B cell responsiveness to environmental cues, a critical process for humoral immunity.

Results
PU.1 and SpiB control follicular B cell homeostasis. To investigate the function of PU.1 and SpiB in mature B cells we have generated mice that carry floxed alleles of Spi1 14 , null alleles of Spib 22 and a BAC transgene where Cre recombinase is brought under the control of the Cd23 (Fcer2) regulatory sequences (Spi1 fl/fl Spib −/− Cd23-Cre T/+ ), hereafter termed PU.1 SpiB double knockout (DKO). Cd23-Cre is active from the transitional stage of B-lymphopoiesis, which results in highly efficient gene inactivation in both follicular and MZ B cells 21 . The frequency and phenotype of splenic B cells from the PU.1 SpiB DKO mice was compared to control (Cd23-Cre), PU.1 conditional knockout (cKO; Spi1 fl/fl Cd23-Cre T/+ ) and SpiB knockout (KO; Spi1 fl/fl Spib −/− Cd23-Cre +/+ ) genotypes. Although the singular loss of either PU.1 or SpiB did not substantially alter the proportion, total number or subset make up of the splenic B cell compartment, the proportion and number of follicular B cells was decreased 2-fold to 3-fold in PU.1 SpiB DKO spleens (Fig. 1a, b), whereas the number of MZ B cells remained unaltered. In keeping with a previous report 23 , PU.1 and SpiB were redundantly required for cell surface expression of CD23, which was significantly downregulated in the absence of SpiB ( Supplementary Fig. 1) and more so in the absence of both factors (Fig. 1a). As a consequence, follicular B cells were defined as B220 + IgD + IgM int (Fig. 1a). Taken together, these data suggest that PU.1 and SpiB are redundantly required for follicular B cell survival.
Increased plasma cells without PU.1 and SpiB. In contrast to the relative loss of follicular B cells, the combined deficiency of PU.1 and SpiB resulted in a significantly increased frequency and proportion of plasma cells (B220 − CD138 + CD98 + ) in both the spleen and bone marrow (Fig. 2a, b). PU.1 cKO mice displayed a normal plasma cell frequency while SpiB KO mice harbored significantly more plasma cells in the spleen, with a similar but non-significant increase in the bone marrow. Analysis of the serum immunoglobulin titers and isotype make up of the plasma cell compartment revealed a significant increase in IgM and IgG3 in SpiB KO and PU.1 SpiB DKO sera and decreased IgA and IgG2b ( Supplementary Fig. 2). Curiously, PU.1 cKO mice displayed increased titers of IgG2b and decreased titers of IgG2c that were not seen in the PU.1 SpiB DKO mice. In keeping with the increase in IgM and IgG3, isotypes that arise from extrafollicular sources, such as MZ B cells 24 , immunofluorescence on the spleens of naive PU.1 SpiB DKO mice revealed a marked increase in IgM + plasma cells in the red pulp (Fig. 2c). Taken together these data demonstrate that PU.1 and SpiB act both individually and redundantly to negatively regulate plasma cell homeostasis.
The germinal center reaction requires PU.1 and SpiB. Previous studies found that SpiB was dispensable for the establishment of a GC, but that the GC could not be sustained beyond 4 weeks 22 , while mice lacking a single allele of Spi1 and both copies of Spib throughout B-cell development generated few mature B cells that could not initiate a GC reaction 19 . However, in neither study was the fate of the antigen-specific B cells tracked.
Analysis of control mice 14 days after immunization with the T cell dependent antigen NP-KLH in alum revealed robust production of NP-binding B cells that had undergone CSR to   Supplementary Fig. 8A. Data are representative of 7-10 mice per genotype. b Absolute cell numbers are graphed as the mean ± s.d. P-values compare the indicated samples (two tailed t-test). Each circle represents the results from an individual mouse. *P < 0.05, **P < 0.01, ****P < 0.0001 IgG1 and near uniformly upregulated the GC regulator Bcl6. (Fig. 3a, (Fig. 3e). Similarly, ELISpot analysis of the NP-specific antibody response revealed a normal frequency of NP-binding IgG1 antibody-secreting cells at day 14 post immunization in mice lacking SpiB, a non-significant reduction in mice lacking PU.1 and very few NP reactive cells in the DKO spleen (Fig. 3c, d). Together these data indicate that PU.1 and SpiB are redundantly required for the GC reaction and the production of high-affinity antibody.

PU.1 and SpiB negatively regulate plasma cell differentiation.
Our previous studies using Cd19-Cre to inactivate PU.1 revealed that PU.1 and IRF8 acted together to repress plasma cell differentiation in vitro 6 . It that model, the singular deficiency of PU.1 resulted in a relatively mild enhancement of plasma cell formation after stimulation with CD40 L and IL-4, a finding that agreed with a second study in which inactivation of a miRNA binding site in the 3′ untranslated region of Spi1 resulted in an increased concentration of PU.1 in activated B cells and impaired plasma cell formation 25 .
To address the combined importance of PU.1 and SpiB for B cell differentiation in vitro we cultured follicular B cells of the four genotypes, generated exclusively from lymph nodes to exclude any possible MZ B cell contamination, in CD40 L + IL-4, conditions that promote B cell proliferation, CSR to IgG1 and plasma cell differentiation 26 . Although B cells of each genotype proliferated similarly in CD40 L + IL-4, as assessed by CTV dilution (Fig. 4a), the yield of cells was consistently reduced in the PU.1 cKO and PU.1 SpiB DKO cultures (Fig. 4b). Analysis of the   To investigate how PU.1 and SpiB interact with the gene regulatory networks controlling late B cell and plasma cell differentiation, we interrogated our RNA-seq data from activated B cells for the expression of the transcriptional regulators known to be important in these processes 5 . Although all components of the B cell network were expressed in the absence of PU.1 and SpiB, indicating that the network is intact at a global level, we observed partial reduction in several key regulators, including Pax5 and Bcl6 in PU.1 cKO and PU.1 SpiB DKO B cells (Fig. 5d). PU.1 and SpiB also appeared to redundantly regulate the expression of Bach2, a critical negative regulator of Blimp1 and plasma cell differentiation 28 . In contrast the expression of other important regulators, including Irf4, Irf8, and Pou2af1 (Obf1) was similar in all genotypes. Curiously, we observed a small but significant increase in Tcf3 (E2A) and a similar decrease in Tcf4 (E2-2) in PU.1 SpiB DKO B cells (Fig. 5d). Similar results were also observed in activated B cell cultures following stimulation with LPS + IL-4 ( Supplementary Fig. 6D).
Analysis of the transcriptional regulators of the plasma cell fate identified premature de-repression of Prdm1, Ell2, Irf4 and Cebpb without PU.1 or PU.1 and SpiB and Xbp1 in the absence of both factors (Fig. 5d). As no bona fide plasma cells are observed at this early time point ( Supplementary Fig. 3), we interpret the derepression of these genes to be early markers of the increased propensity of the PU.1 cKO and PU.1 SpiB DKO B cells to undergo terminal differentiation. Notably, we also observed a robust increase in Aicda, encoding Activation induced cytidine deamidase (AID) in PU.1 cKO and PU.1 SpiB DKO B cells, making it unlikely that a deficiency in this enzyme is the underlying cause of the defective CSR observed in vitro and in vivo without PU.1 and SpiB (Fig. 5d, Supplementary Fig. 6D).
Gene ontology analysis of the DE genes revealed enrichment for "Metabolic pathways" in each gene set. More informatively the pathways downregulated in PU.1 cKO and PU.1 SpiB DKO largely overlapped and included several gene sets associated with lymphocyte signaling, including the BCR signaling pathway (Fig. 5e). Genes upregulated showed a strong increase in gene frequency in the DKO comparison and included the "protein processing" gene set that contains many genes involved in the unfolded protein response, which is critical for high level antibody secretion by plasma cells (Fig. 5e).
Given the strong deficiency we observed in PU.1 SpiB DKO mice after T-dependent immunization and the existing literature in this area, we decided to focus on the B cell receptor (BCR) signaling pathway. Strikingly, although the core constituents of the BCR (Igh, Igk, Igl, Cd79a, and Cd79b) were normally expressed, many downstream components of the BCR signal transduction cascade were significantly downregulated (most between 2-fold and 3-fold) in the absence of PU.1 and SpiB (Fig. 6a). A similar trend, albeit with a lower fold change was also observed in PU.1 cKO B cells. Western analysis confirmed these findings, with modest reductions in protein expression for several BCR signaling components including Btk, Rasgrp3, and Card11 (Fig. 6c) in the absence of PU.1 or PU.1 and SpiB. Analysis of the PU.1 ChIPseq data identified the presence of PU.1 binding in the proximal promoter of most of the PU.1 and PU.1/SpiB dependent components of the BCR pathway (Fig. 6b). A similar trend, in  Supplementary Fig. 8C, D. c Representative ELISpots show the number of NP-specific IgG1 + antibody-secreting cells (ASC) per million splenocytes plated. Data are representative of 5 or 6 mice per genotype from 2 independent experiments d Frequencies of NP + IgG1 + B cells are shown relative to total B220 + cells (left), proportion of B220 + NP + cells expressing Bcl6 (middle), frequency of total NP-specific IgG1 secreting ASC in the spleen per million cells (right) are shown. Data are graphed as mean ± s. d. of between five and six mice per genotype from two independent experiments. P-values compare the indicated samples (two tailed t-test). **P < 0.01, ****P < 0.0001. e Spleen sections from mice of each of the genotypes examined mice were stained with antibodies to CD3 to detect T cells (blue), B220 to identify B cell follicles (green) and GL7 for GCs (red). Scale bar = 200 μm several BCR signaling components was also observed following activation with LPS + IL-4 ( Supplementary Fig. 7).
To determine if these quantitative changes in the expression of multiple components of the BCR resulted in impaired signaling and cellular responses, we cultured follicular B cells of the four genotypes in the presence of anti-IgM + IL-4 for 3 days. Proliferation was similar between the control, PU.1 cKO and SpiB KO B cells, however in keeping with the gene expression data, PU.1 SpiB DKO B cells displayed extremely low viability after BCR engagement (Fig. 6d). To assess more proximal signaling events after BCR crosslinking we could not use the traditional cell source of naive B cells as the inactivation of Spi1 is incomplete in PU.1 SpiB DKO follicular B cells ( Supplementary  Fig. 3A). To circumvent this problem, we first cultured naive follicular B cells for 48 h in CD40 L + IL-4, a time point where exon 5 of Spi1 was absent (Fig. 5a), rested them for one hour and then exposed the cells to anti-IgM. This approach showed clear inhibition in the BCR response, as assessed by calcium mobilization, in the absence of PU.1, and to a lesser extent SpiB (Fig. 6e). These data demonstrate that PU.1 and SpiB are redundant regulators of multiple components of the BCR signaling pathway.
Regulation of immune receptors by PU.1 and SpiB. In addition to the regulation of the genes encoding members of the BCR signaling pathway, it was striking that PU.1 or PU.1 and SpiB were required for the normal expression of several other immune receptors critical for the ability of a B cell to sense its immunological environment (Fig. 7a, Supplementary Fig. 7B). This included Tnfrsf13c, encoding the BAFF-R, essential for peripheral B cell survival and homeostasis 1 and Cd40, a critical component of T cell help 29 . Flow cytometric analysis confirmed the downregulation of both the BAFF-R and CD40 in the absence or PU.1, with a further reduction in expression in the absence of both PU.1 and SpiB (Fig. 7b, c). Analysis of the PU.1 ChIP-seq data again showed the presence of PU.1 binding in the proximal promoter of both of these targets (Fig. 7d).
This quantitative reduction in Tnfrsf13c is likely to be responsible for the 2-fold to 3-fold reduction in follicular B cells in PU.1 SpiB DKO mice (Fig. 2b), while the deceased CD40 expression may be responsible for the lower cell yields in cultures containing CD40 L and IL-4, despite the increased propensity to differentiate into ASCs (Fig. 4b). In contrast, cytokine receptors, such as the IL4ra and Il5r were marginally upregulated in the absence of PU.1 and SpiB (1.6-fold each).
Several important microbial sensors were also directly regulated by PU.1 and SpiB including Tlr4 and Tlr9, with each having clear proximal promoter-binding sites for PU.1 (Fig. 7a, d,  Supplementary Fig. 7B). To examine the relevance of the reduced TLR4, we cultured naive lymph node follicular B cells of each genotype in LPS. Although all genotypes robustly proliferated, as assessed by CTV dilution, there was a decrease in cell yields without PU.1 and SpiB (Fig. 7e). LPS induced a strong differentiation response to CD138 + plasma cells in control B cells that was augmented as a proportion of live cells at day 4, in the PU1 SpiB DKO cells (Fig. 7e), in a manner similar to that observed for CD40 L + IL-4 cultures (Fig. 5a). However, after adjustment for the different cell numbers obtained from the various genotypes, it was apparent that the total number of CD138 + cells was reduced without PU.1 and SpiB (Fig. 7e). LPSinduced CSR to IgG3 was dependent on PU.1 (Fig. 7e) manner similar to what we observed for IgG1. Taken together these data demonstrate that PU.1 and SpiB redundantly control the receptors for several major pathways by which B cells sense and respond to environmental cues.

Discussion
Humoral immunity is underpinned by the ability of B cells to sense and integrate signals from their environment and respond with the appropriate intrinsic differentiation program. These environmental cues include the cognate antigen, T cell help and survival factors such as BAFF. In this study, we show that the Ets family transcription factors, PU.1 and SpiB, well-known regulators of early lymphopoiesis, are redundantly required for the normal expression and function of virtually all pathways by which B cells sense and respond to the surrounding milieu. In contrast plasma cell development was either unaffected or accelerated by the loss of both factors. Thus PU.1 and SpiB act redundantly to balance the proliferative and differentiation response of B cells that is essential for protective humoral immunity.
Within the Ets family, PU.1 and SpiB are the genes most highly related to each other. They bind to the same consensus DNA recognition site, either on their own or in complex with IRF4 or IRF8 17,18 . Both factors are constitutively expressed throughout B cell development, with SpiB expression being rapidly extinguished upon plasma cell differentiation, while PU.1 remains expressed at a low concentration and can, at least in culture, still occupy many thousands of binding sites in the genome 27,30 . The similarity of this pair of transcription factors has long complicated the interpretation of their function in B cells. PU.1 null embryos die in late gestation and lack all lymphocytes 12 , while removal of PU.1 in adult hematopoietic stem cells ablates B-cell development 9,10 . Surprisingly, conditional inactivation of PU.1 in committed B cells using Cd19-Cre, which is first active during B cell development in the bone marrow, results in no pronounced defects in B cell production or function beyond a role in B1 cell formation 10,[13][14][15][16] . Loss of SpiB is also compatible with normal B cell development, however SpiB KO mice have defects in the GC reaction at later times points after immunization 22 . In this study, we observed normal numbers of peripheral B cell subsets in mice lacking PU.1 or SpiB, as well as functional GC, at the peak time post immunization (day 14). In contrast to our previous studies using Cd19-Cre 6, 14 , we observed a partial impairment in GC responses in vivo and enhanced plasma cell differentiation in vitro in the absence of PU.1. The major difference between the studies is that here we have utilized Cd23-Cre that is active from the transitional stage in the spleen 21 , sparing the bone marrow development process entirely. This, as discussed below may impact on the ability of SpiB to compensate for the loss of PU.1.
Previous attempts to investigate the degree of functional compensation between PU.1 and SpiB have been hampered by the complete absence of B cells in PU.1-deficient mice 12 . As a compromise, B cell development was examined in Spi1 +/− Spib −/− mice, which show a 9-fold decrease in splenic follicular B cells 19,23 . The residual B cells in those mice, many of which were MZ B cells, appeared to be functionally defective, unable to contribute to GCs and impaired in their signaling responses through the BCR 19 and TLRs 31 . More recently, conditional inactivation of PU.1 on a SpiB KO background using Cd19-Cre, revealed a dramatic loss of follicular B cells and an expansion of the pre-Bcell compartment that ultimately generated B-ALL at high frequency 15,32 . By utilizing Cd23-Cre our study has bypassed the important function of these proteins at the pre-B-cell stage and allowed the first robust analysis of the functionality of PU.1 SpiB DKO B cells. Although we found a 2-fold to 3-fold reduction in the frequency of follicular B cells in the DKO mice, these data should be interpreted cautiously, as we observed strong retention of the floxed exon 5 of Spi1 in B cells only in the absence of Spib, suggesting that these two proteins are redundantly essential for follicular B cell survival. Interestingly, the gene encoding the BAFF-R, Tnfrsf13c, is a direct PU.1/SpiB target gene providing a likely mechanism for the reduced follicular B cell numbers 33 .
The redundancy between PU.1 and SpiB was also apparent on the molecular scale, with both the number of DE genes and the magnitude of change being increased in the PU.1 SpiB DKO activated B cells. Importantly, exon 5 of Spi1 was efficiently excised in the DKO B cells after in vitro activation for 48 h, allowing a full examination of the transcriptional consequences of gene inactivation at this stage. As with the cellular data, redundancy between PU.1 and SpiB was observed. This agrees well with ChIPseq studies that show that the vast majority of SpiB occupied sites in the B cell genome also bind PU.1 18 . Pathway analysis of these data revealed a striking role for PU.1 and SpiB in directly regulating the expression of many components of the BCR signaling pathway, including the previously characterized targets, Blnk 32 and Btk 34,35 . Although the reduced expression of the individual target genes was in each case modest, combined reductions resulted in grossly impaired in vitro responsiveness to anti-IgM and a total ablation of the antigen-specific immune response. The function of PU.1 and SpiB was not restricted to the BCR pathway, as the ability of the double mutant follicular B cells to respond to antigen-independent simulation, either via T cells derived signals such as CD40 L or microbial stimuli such as LPS was also diminished, again likely due to the direct regulation of these key signaling pathways by PU.1 and SpiB.
In contrast to the essential function of PU.1 and SpiB to allow B cell responsiveness to environmental cues, both factors act as negative regulators of plasma cell development, in steady-state conditions in vivo and after activation in vitro. The increased Gene names with PU.1 peaks (P < 10 −6 ) are indicated in red. e KEGG pathway analysis of genes regulated by PU.1 and PU.1/SpiB. Genes were first divided into those that were down or upregulated in the indicated genotypes. The number of genes that are associated with each statistically enriched (P < 0.05) functional category is shown plasma cell frequency in vivo was predominantly IgM secreting cells, with a milder increase in IgG3 production, isotypes that often derive from MZ B cells 24 . In keeping with this conclusion, MZ B cell frequency was unaltered in PU.1 SpiB DKO and mildly increased in Spi1 +/− Spib −/− spleens 23 . The preferential retention of MZ B cells is potentially also due to the role of PU.1 and SpiB in activating the expression of key components of the BCR pathway, as impaired BCR signaling is known to favor MZ over follicular B cell fate 36 . The splenic MZ B cells from Spi1 +/− Spib −/− mice have been shown to be impaired in their responses to several microbial products, including LPS, because of reduced Nfkb1 expression 31 . In contrast our RNA-seq and ChIPseq data derived from in vitro activated lymph node follicular B cells pointed to defective Tlr4 expression as the underlying cause of the reduced responsiveness to LPS in PU.1 SpiB DKO B cells.
PU.1 cKO and PU.1 SpiB DKO B cells also produced markedly more plasma cells after in vitro activation with CD40 L + IL-4 or LPS + IL-4. This is in keeping with our recent report of increased plasma cell differentiation in the absence of PU.1 and IRF8 6 and occurred despite the lower overall cell yield without PU.1 and SpiB. PU.1/SpiB loss resulted in reduced expression of key B cell program regulators such as Pax5, Bcl6, and Bach2, while plasma cell promoting factors such as Prdm1, Xbp1, Ell2, and Cebpb were increased in expression. As with PU.1/IRF8 DKO B cells 6 , Aicda expression was increased in the absence of PU.1 and SpiB suggesting that reduced expression of this enzyme was not the    Fig. 5) for the indicated genotype is plotted as the mean fold change (log 2 , from 2 experiments) relative to the expression in the control genotype. *Indicates genes that are DE relative to controls (defined as log 2 fold change >0.6-fold, log 2 RPKM (reads per kilobase of exon model per million mapped reads) >0, P < 0.15). b PU.1 ChIPseq identifies promoter-binding sites for PU.1 in many components of the BCR signaling pathway. Gene structures are shown below. Gene names in a, b with PU.1 peaks (P < 10 −6 ) are indicated in red. c Lysates were generated from activated B cells (cultured for 48 h in CD40 L + IL-4) of the indicated genotype. Western blotting was performed for the indicated proteins. Actin serves as a control for protein loading. Data in b, c are representative of two experiments each. d Loss of PU.1 and SpiB leads to decreased viability in response to BCR signaling. Resting lymph node B cells from mice of the indicated genotype were labeled with CTV and cultured in the presence of anti-IgM + IL-4 for 3d before flow cytometric analysis. The number of live cells is plotted for each genotype examined. Data are the mean ± s.d. for three independent experiments. P-values compare the indicated samples (two tailed t-test). ****P < 0.0001. e Resting B cells of the indicated genotype were cultured with CD40 L + IL-4 for 48 h before being washed and rested for one hour. Cells were then loaded with Indo-1 AM before basal fluorescence was measured for 30 s. Anti-IgM was then added and fluorescence changes recorded for an additional 2.5 min. Results are presented as arbitrary units of changes in fluorescence ratio over time. Data are representative of two experiments underlying cause of the decreased CSR in PU.1 SpiB DKO B cells. Collectively these data suggest that PU.1 and SpiB act to balance B cell proliferation with the need to produce sufficient numbers of plasma cells, with PU.1 and SpiB deficiency tipping the balance away from clonal expansion and towards terminal differentiation.

ARTICLE
There is to date only limited understanding of the unique function(s) of SpiB, although it is of note that Spib expression is restricted to B cells and pDCs 37 , lineages that have a distinctly low rate of PU.1 expression. Our finding of increased expression of genes associated with pDCs, macrophages and T cells in naive B cells lacking SpiB is reminiscent of similar observations of PU.1 repression of natural killer cell associated genes in pro-B 38 and pro-T 39 cells and suggests a previously unappreciated function of SpiB to repress lineage inappropriate gene expression in B cells that warrants closer examination in a future study. PU.1, in contrast, has a broad expression pattern and has been intensively studied in many blood cell lineages. One of the recurrent themes of PU.1 function in haematopoiesis is its pivotal role in activating the expression of a range of receptors, including the genes encoding MCSFR, GMCSFR, IL3R, IL7R, and Flt3 and many homing molecules such as integrins 8,9,[40][41][42][43] . The data presented in this study provide a novel example of this paradigm, with PU.1 controlling the interplay between the extrinsic signals that control B cell homeostasis and responsiveness and intrinsic programs that govern differentiation events essential for humoral immunity. . Cells were analyzed on FACSCanto flow cytometers and cell sorting was carried out using FACSDiVa or Aria flow cytometers (BD Biosciences). Intracellular staining for BCL6 (7D1) was performed using the FoxP3 staining buffer set (eBiosciences). All Abs were produced in-house unless otherwise indicated. Antigen-specific B cells were identified by binding NP coupled to phycoerythrin.
Calcium flux. Naive resting B cells were isolated as described above and cultured for 48 h in CD40 L + IL-4, to ensure complete deletion of the Spi1 allele in PU.1 SpiB DKO B cells. Cells were then washed extensively and rested for one hour before being resuspended in loading buffer (KDS-BSS buffer containing 0.1% glucose and 0.1% FCS) containing Indo-1 acetoxymethyl (Molecular Probes). Following an incubation for 40 min at 37°C, loaded cells were washed extensively and then resuspended in the loading buffer indicated above. Calcium flux was measured by flow cytometry. Baseline fluorescence was established for 30 s before anti-IgM (10 μg/ml; Jackson Immunoresearch) was added and the subsequent calcium flux followed for the indicated times.
Whole transcriptome analysis. RNA was isolated from either naive or activated (48 h in CD40 L + IL-4 or LPS + IL-4) lymph node derived B cells using the RNeasy Plus Mini kit (74,134; Qiagen). Two biological replicates were generated and sequenced for each sample. mRNA was subjected to transcriptome re-sequencing on an Illumina NextSeq sequencer. Between 15 and 90 million reads were analyzed per sample. Each sample's read set was aligned to the GRCm38/mm10 build of the Mus musculus genome using the Subread aligner 44 . Only uniquely mapped reads were retained. Genewise counts were obtained using featureCounts 45 . Reads overlapping exons in annotation build 38.1 of NCBI RefSeq database were included. Genes were filtered from downstream analysis if they failed to achieve a CPM (counts per million mapped reads) value of at least 0.5 in at least two libraries. Counts were converted to log 2 counts per million, quantile normalized and precision weighted with the "voom" function of the limma package 46,47 . A linear model was fitted to each gene, and empirical Bayes moderated t-statistics were used to assess differences in expression 48 . Genes were called differentially expressed if they achieved a false discovery rate of 0.15 or less with a fold change >1.5. Pathway analysis was performed using the KEGG (Kyoto Encyclopedia of Genes and Genomes) database (http://www.genome.jp/kegg/) 49 while analysis for association with particular cell lineages was performed with Enrichr 50, 51 .
PU.1 binding analysis. The PU.1 ChIPseq dataset from LPS activated plasmablasts was downloaded from the Gene Expression Omnibus (GEO) database (GSM1843342). Sequence reads were mapped to mouse genome mm10 using Subread aligner. Only uniquely mapped reads were kept. PU.1 binding peaks were called by using MACS2 program with a false discovery rate cutoff of <10 −6 . Called peaks must also have at least 200 support reads in the 200 bp region centered on the binding summit. Peaks were assigned to genes if they occurred <20 kb upstream of the transcriptional start sites, within the gene body an <5 kb downstream of the end of transcription. Within these parameters multiple gene assignment was allowed.
Statistical analysis. Statistical significance of non RNA-Seq data was assessed with Prism6 (GraphPad). t tests for all single comparisons were unpaired, assumed Gaussian distribution and that both populations have the same standard deviation (SD). Enrichment of PU.1 binding in DE gene lists was determined using χ 2 -tests. Bar graphs display the arithmetic mean ± s.d. unless otherwise stated.
Data availability. Sequence data that support the findings of this study have been deposited in GEO with the primary accession code GSE90094.