PI3K induces B-cell development and regulates B cell identity

Phosphoinositide-3 kinase (PI3K) signaling is important for the survival of numerous cell types and class IA of PI3K is specifically required for the development of B cells but not for T cell development. Here, we show that class IA PI3K-mediated signals induce the expression of the transcription factor Pax5, which plays a central role in B cell commitment and differentiation by activating the expression of central B cell-specific signaling proteins such as SLP-65 and CD19. Defective class IA PI3K function leads to reduction in Pax5 expression and prevents B cell development beyond the stage expressing the precursor B cell receptor (pre-BCR). Investigating the mechanism of PI3K-induced Pax5 expression revealed that it involves a network of transcription factors including FoxO1 and Irf4 that directly binds to the Pax5 gene. Together, our results suggest that PI3K signaling links survival and differentiation of developing B cells with B cell identity and that decreased PI3K activity in pre-B cells results in reduced Pax5 expression and lineage plasticity.


Results
PI3K regulates Pax5 expression. Analysis of mice with impaired class IA PI3K in B/T lymphocyte progenitors revealed that B cell development is selectively blocked at the pre-B cell stage 24 . To investigate class IA PI3K function in early B cells, we expressed constitutively active versions of p110α (myr-p110α) or AKT (myr-AKT) in a bone marrow (bm)-derived wild type (wt) pre-B cell line (Fig. S1a). GFP-negative and empty vector-transduced cells behaved comparably. For the sake of simplicity only the latter will be shown in further experiments (Fig. 1a). As expected, phosphorylated AKT (pAKT) indicative of activation was increased in pre-B cells expressing myr-p110α or myr-AKT (Fig. 1a). Moreover, we found that Pax5 and CD19 expression was elevated in these cells ( Fig. 1a and Fig. S1b and data not shown). Notably, neither myr-AKT nor myr-p110α did lead to IL-7 independent cell growth (data not shown). To confirm the finding that PI3K signaling induces Pax5 expression, we treated the bm-derived wt pre-B cell line with the PI3K-inhibitor LY294002 at a concentration ensuring a robust decline in AKT activity while not interfering with cell viability at the time of analysis ( Fig. 1b  and Fig. S1c). The results show that inhibition of PI3K signaling interferes with Pax5 expression. Since Pax5 activates SLP-65 expression 4 we tested whether class IA PI3K also regulates SLP-65. Indeed, treatment of cells from the bm-derived wt pre-B cell culture with the PI3K-inhibitor LY294002 led to downregulation of both Pax5 and SLP-65 protein expression (Fig. 1b). Consistent with the negative effect of activated PI3K, FoxO1-levels increased upon LY294002 application (Fig. 1b).
To test whether class IA PI3K-dependent regulation of Pax5 also acts in the presence of continuous pre-BCR signaling as well as for later B cell developmental stages, we utilized a SLP-65-deficient pre-B cell line (Fig. 1c) and primary mature B cells (Fig. 1d, Fig. S1d), respectively. Expression of Pax5 declined in both cell types upon inhibition of PI3K signaling on protein (Fig. 1c,d) and transcript level (Fig. 1e,f). Additionally, PI3K-mediated activation of Pax5 expression was also detected in pre-B and mature B cells of human origin (Fig. S1f,g). Together, class IA PI3K signaling activates Pax5 expression irrespective of the species and of the B cell developmental stage.
PI3K requires the pre-BCR but not the IL-7R. Besides pre-BCR, IL-7R-derived signals play important roles during early B cell development 7 . To test whether IL-7R activates PI3K signaling, we incubated bm-derived wt pre-B cells overnight in the absence of IL-7, treated the cells with IL-7 and after different incubation periods within 60 min we determined AKT phosphorylation. The results show that pAKT was not increased after IL-7 treatment at any time point tested (Fig. 2a). To provide additional evidence for the dispensable role of IL-7 in PI3K activation, we used a bm-derived pre-B cell line carrying loxP-flanked IL-7Rα alleles allowing Cre-mediated deletion of IL-7Rα and thus abrogating IL-7R signaling 28 . Indeed, deletion of IL-7Rα showed no effect on the pAKT-levels ( Fig. 2b and Fig. S2a). Moreover, absence of IL-7 further increased expression of CD19 by myr-AKT and myr-p110α (Fig. S1b). Since pre-BCR expressing cells can be further divided into (IL-7 responsive) large pre-B cells and (IL-7-non-responsive) small pre-B cells, we tested whether IL-7Rα deletion in IL-7-responsive large pre-B cells can affect the phosphorylation level of AKT. Indeed, even with this experimental setup, AKT phosphorylation was not changed after IL-7Rα deletion (Fig. S3a). These data demonstrate that IL-7R is unlikely to be essential for PI3K activation in early B cell development.
To test whether pre-BCR signaling activates PI3K signaling, we reconstituted the pre-BCR in a Rag2-deficient pro-B cell line. Since Rag2-deficient cells express an endogenous surrogate LC, introduction of μHC results in pre-BCR expression in these cells. The data show that μHC leads to increased AKT phosphorylation suggesting that both the pre-BCR and the BCR activate PI3K signaling (Fig. 2c). To confirm these data, we used additional μHCs including 2 randomly cloned μHCs from the spleen of wt mice (Fig. 2d). Interestingly, we found that pre-BCR-induced AKT phosphorylation is detectable within 24-72 h (Fig. 2d) and that AKT phosphorylation is decreased after this time period (Fig. S3b), suggesting that activation of PI3K signaling by the pre-BCR is a regulated process. In addition to the increase in pAKT, μHC expression in Rag2-deficient pro-B cells also resulted in elevated Pax5 levels (Fig. S3c). Together, our data suggest that signaling by the pre-BCR, and not by IL-7R, is required for stimulation of PI3K activity in early B cell development. (a) Cells from a bone marrow (bm)-derived wildtype (wt) pre-B cell culture were transduced with constitutively active forms of AKT (myr-AKT), p110α (myr-p110α), or as control with empty vector (EV) and analyzed for pAKT and Pax5 expression by intracellular FACS. If not indicated otherwise, numbers in the histograms state the mean fluorescence intensity (MFI) of the respective GFP + populations. (b) Cells from a bm-derived wt pre-B cell culture were treated with LY294002 or DMSO for 16 h and analyzed for pAKT, Pax5, SLP-65 and FoxO1 expression by intracellular FACS. (c) Cells from a SLP-65-deficient pre-B cell line were treated with LY294002 or DMSO for 12 h and pAKT and Pax5 expression was analyzed by intracellular FACS. (d) Murine mature B cells (CD43 − ) were isolated and treated with LY294002 or DMSO for 12 h, lysed and subjected to immunoblot for analysis of Pax5 expression. Actin served as a loading control. For original full-length blots see Fig. S1e. (e) Total RNA of SLP-65-deficient cells treated for 12 h with LY294002 or DMSO was isolated. Gapdh and Pax5 mRNA-levels were detected with specific primers by qRT-PCR using the SYBR-Green detection method. Results are shown as mean ± SD of 2 independent experiments, run as duplicates. Statistical significance was calculated using the t-Test. (f) Total RNA of murine mature B cells treated with LY294002 or DMSO for 12 h was isolated. Gapdh and Pax5 mRNA-levels were detected with specific primers by qRT-PCR using the SYBR-Green detection method. Results are shown as mean ± SD of 2 independent experiments, run as duplicates. Statistical significance was calculated using the Mann-Whitney Test. Data shown in Fig. 1a 24 , we used bone marrow cells from a p110α −/− /δ D910A mouse to generate an IL-7-dependent pre-B cell line (Fig. 3a, hereafter referred to as p110dKO cells, Fig. S4a). Signaling through class IA PI3K was affected in p110dKO cells as shown by reduced levels of pAKT (Fig. 3a). In accordance with our previous findings, p110dKO cells showed reduced transcription of Pax5 (Fig. 3b) and of its target genes Cd19 and Blnk (encoding SLP-65) (Fig. 3c).
Since signaling through SLP-65 regulates pre-BCR expression, we tested whether PI3K-deficient p110dKO cells showed defective pre-BCR down-regulation 10,11 . Indeed, we detected increased pre-BCR expression on the surface of p110dKO cells (Fig. 3d). To confirm the block in pre-BCR signaling in p110dKO cells, we tested whether expression of myr-AKT was able to restore expression and function of SLP-65. Indeed, myr-AKT led to increased expression of Pax5, CD19 and SLP-65 as well as down-regulation of surface pre-BCR expression (Fig. 3e). To provide additional evidence for the importance of PI3K signaling in activation of Pax5 expression, we established an IL-7-dependent pre-B cell line from the bone marrow of mice carrying loxP-flanked alleles for p110α and p110δ, the main catalytic p110 subunits acting in B cells 24 (Fig. 3f and Fig. S4b). Deletion of p110α/p110δ by retroviral transduction of Cre expression vector resulted in decreased pAKT and Pax5 expression ( Fig. 3g-i) suggesting that PI3K activity is important for maintaining the transcriptional program of developing B cells.  PI3K deficiency results in unstable B cell commitment. Given the importance of Pax5 for B-cell lineage specific gene expression, the link between class IA PI3K and Pax5 suggested that PI3K-deficient p110dKO pre-B cells show defective B-lineage gene expression because of reduced Pax5 activation. In fact, we detected transcripts of Csf1r and Notch1 (Fig. 3c). Csf1r encodes the colony stimulating factor 1 receptor (CSF1R), also known as macrophage colony-stimulating factor receptor (M-CSFR), which is usually expressed on macrophages and is repressed by Pax5 in B cells. Based on the presence of Csf1r transcripts in p110dKO cells we tested whether p110dKO cells were able to respond to macrophage-colony stimulating factor M-CSF, the ligand of CSF1R and the factor, essential for survival, growth and differentiation of macrophages 29 . In contrast to wt cells, p110dKO pre-B cells survived in medium containing M-CSF ( Fig. 4a and Fig. S5a) and expressed the myeloid specific surface marker CD11b (Mac-1) (Fig. 4b) suggesting that failure to suppress CSF1R expression in p110dKO cells allows survival under myeloid culture conditions. We further confirmed that p110dKO cells surviving in M-CSF-supplemented culture conditions were of B cell lineage origin since both the original IL-7-dependent cell line and the M-CSF-dependent cell line showed the same VDJ recombination pattern (Fig. S5b). Additionally, we tested in vivo the potential of p110dKO to enter the T-cell lineage by injecting p110dKO cells into Rag2/common γ chain double deficient mice. Interestingly, we found that mice injected with p110dKO developed a thymus-like structure in which we could detect CD4 − , CD8 − as well as CD4 + /CD8 + cells (Fig. S5c,d).
Since the analysis of p110dKO cells suggested a role of PI3K in B cell identity, we investigated whether blocking PI3K activity can induce lineage plasticity of PI3K-sufficient cells. Indeed, we detected elevated levels of Csf1r and Notch1 transcripts in cells from a bm-derived SLP-65-deficient pre-B cell line treated with LY294002 (Fig. 4c). To test whether PI3K-inhibitor treatment enabled these cells to survive under non-B cell conditions, LY294002-treated cells were cultured in medium supplemented with M-CSF. In fact, LY294002-treated pre-B cells survived under myeloid conditions (Fig. 4c), indicating that PI3K-sufficient pre-B cells become responsive to a myeloid specific growth factor after inhibition of PI3K.
Together, these data suggest that PI3K-deficiency blocks B cell development at the pre-B cell stage due to the inability of PI3K-deficient pre-B cells to induce Pax5 and thus the differentiation program for developing B cells.

FoxO1 is involved in Pax5 regulation.
To characterize the molecular mechanism underlying the PI3K-mediated activation of Pax5 we investigated the role of FoxO1, which is known to be a highly conserved downstream target of PI3K/AKT signaling and is essential for Ig gene recombination 30,31 . Signaling through class IA PI3K/AKT results in proteasomal degradation of FoxO1, whereas in the absence of PI3K-activity, FoxO1 protein is stabilized 32 . In fact, we detected elevated levels of FoxO1 in p110dKO pre-B cells (Fig. 5a). To test whether FoxO1 was involved in the PI3K-dependent regulation of Pax5, we introduced a constitutively active form of FoxO1 (FoxO1-A3) or an empty control vector (EV) into cells from a bm-derived wt pre-B cell line and found that FoxO1-A3 led to reduced Pax5 expression (Fig. 5b,c). Furthermore, inducible Cre-mediated deletion of loxP-flanked FoxO1 in a bm-derived pre-B cell line resulted in higher expression of Pax5 ( Fig. 5d and Fig. S2c).
Since FoxO1 has been shown to directly bind to the Pax5 gene 33 , we investigated whether FoxO1 represses Pax5 transcription through potential FoxO1-binding sites in the Pax5 gene. Two highly conserved, potential FoxO1-binding sites within the Pax5 gene locus were identified (Fig. 5e) and used for the generation of luciferase expression vectors. These vectors carry, upstream of a basal Vκ-promotor, 1 kb-fragments from the Pax5 gene containing or lacking the potential FoxO1-binding sites (Fig. 5f). However, no significant effects of these conserved FoxO1-binding sites were observed on luciferase expression (Fig. 5g). These results suggest that FoxO1 might regulate Pax5 through an indirect mechanism. FoxO1 represses Pax5 by induction of Irf4. A search for transcription factors that are activated by FoxO1 revealed interferon regulatory factor 4 (Irf4) as a potential candidate for Pax5 regulation. In fact, the Irf4-promoter contains specific sites for FoxO1 binding and subsequent activation of Irf4 expression 34 . Furthermore, Irf4 was shown to bind the Pax5 gene within a recently identified enhancer region 35 . To investigate the role of Irf4 in PI3K-dependent regulation of Pax5 expression, we confirmed that Irf4 is induced upon treatment of SLP-65-deficient pre-B cells with the PI3K inhibitor LY294002 (Fig. 6a) or after introduction of FoxO1-A3 into a bm-derived wt pre-B cell line (Fig. 6b). In line with these findings, p110dKO pre-B cells showed elevated levels of Irf4 transcripts (Fig. 6c). Notably, Irf8, which is structurally highly related to Irf4, was not regulated by PI3K (Fig. 6d) and did not suppress Pax5 expression (Fig. S6). To directly test whether Irf4 links FoxO1 to the regulation of Pax5 gene transcription, we transduced cells from a wt pre-B cell line with Irf4 expression vectors and found that Pax5 protein and transcript amounts were reduced after Irf4 expression (Fig. 6e,f). Additionally, we tested whether the inverse regulation of Irf4 and Pax5 can also be detected in human peripheral from bm of p110α fl/fl /p110δ fl/fl mice. The respective cells were characterized by surface staining for CD19 and B220, or the respective isotype control and analyzed by flow cytometry. (g) Cells described in Fig. 3f were retrovirally transduced with a Cre-encoding expression vector or EV, respectively. Cre-mediated deletion of p110α and δ was confirmed by PCR using specific primers detecting floxed or deleted alleles. PCR for SRP20 served as loading control. For original full-length gel pictures see Figure S4c. (h) Cells described in Fig. 3f were retrovirally transduced with a Cre-encoding expression vector or EV, respectively, and analyzed by intracellular FACS for pAKT and Pax5 expression. Numbers in the histogram plots indicate the MFI, depicted data are representative of at least 4 independent experiments. (i) Average MFIs of pAKT and Pax5 following Cremediated p110α and p110δ deletion. Results are shown as mean ± SD of 4 independent experiments. Statistical significance was calculated using the t-Test.

PI3K-responsive element (PIRE) in Pax5 gene.
To further characterize the molecular link between PI3K, FoxO1, Irf4 and Pax5, we first confirmed the binding of Irf4 to the previously described Pax5 enhancer region 35 containing two Irf binding motifs (Fig. 7a,b). To test whether class IA PI3K and Irf4 influence the activity of this enhancer region, we generated luciferase expression vectors using Pax5-derived DNA sequences containing or lacking the Irf4-binding sites (Fig. 7c, PIRE-basic Luc and mPIRE-basic Luc, respectively). Consistent with the described enhancer function of this Pax5 gene region, we found elevated luciferase expression in presence of the Irf4-binding motifs (Fig. 7d). Notably, inactivating these Irf4-binding motifs (mPIRE-basic Luc) abolished this enhancer function (Fig. 7d). To test whether class IA PI3K activity affects the enhancer function, we treated cells expressing PIRE-or mPIRE-basic Luc with LY294002. Interestingly, luciferase expression declined upon PI3K inhibition in PIRE-but not mPIRE-expressing cells (Fig. 7e), indicating that the activity of class IA PI3K critically influences Pax5 enhancer function. Since our data, presented so far, suggested that class IA PI3K regulates Pax5 via Irf4, we tested next whether increased expression of Irf4 counteracts luciferase expression in PIRE-expressing cells. Indeed, enforced expression of Irf4 repressed PIRE-induced luciferase expression confirming the repressive effect of Irf4 on the fragment containing PIRE (Fig. 7f). Together, these results suggest that class IA PI3K controls B cell development by induction of Pax5 gene expression through the activation of the Pax5 enhancer region.

Discussion
The PI3K signaling pathway is essential for the development of pre-B cells and for the maintenance of mature B cells 36 . In this study, we show that the crucial role of PI3K in B cell differentiation is mediated by activating the expression of Pax5 through a mechanism involving FoxO1 and Irf4. Accordingly, the combined deficiency of p110α and p110δ in p110dKO pre-B cells results in severe alteration in the expression of Pax5 and Pax5-regulated genes, which are required for pre-B cell development. Our data point to FoxO1 and Irf4 as an important molecular link between PI3K activity and Pax5 expression. Although it is conceivable that additional elements might be involved in the PI3K-mediated induction of Pax5, available data support the role of Irf4 as a potential link between PI3K signaling and Pax5 gene expression. For instance, it has been shown that Irf4-levels are elevated in p110δ-deficient mature B cells and that inhibition of p110δ induces Irf4 gene transcription 37 . The finding that FoxO1 directly binds to the Irf4 promoter and induces the expression of Irf4 suggests that PI3K activity represses Irf4 by activating the degradation of FoxO1 34 . This is in agreement with our results showing that class IA PI3K signaling represses Irf4 thereby activating Pax5. Since the ability of Irf4 to function as a transcriptional repressor is well established 38 . Our results suggest that Irf4 represses Pax5 gene transcription by directly binding to the enhancer region of the Pax5 gene. Notably, a previous study showed that a large fragment of the Pax5 enhancer containing the Irf4 binding site in addition to other transcription factors activated the transcription of Pax5 in a transient transfection assay 35 . Together with our data, this suggests that the outcome of Irf4 on Pax5 expression may depend on the activity of additional transcription factors that also bind to the Pax5 enhancer. Consequently, modulating the activity of the involved transcription factors during B cell development determines the effect of Irf4 on Pax5 expression. Moreover, the identification of FoxO1 regulating Irf4 provides a molecular rationale as to how SLP-65-activation results in Irf4-induction and thereby in initiation of κLC-expression 39 . Previous findings showed furthermore that Pax5 induces SLP-65, which downregulates PI3K and activates FoxO1 and Irf4 4,9,39 , the FoxO1/Irf4-mediated repression of Pax5 interferes with SLP-65 expression, thereby regulating FoxO1/Irf4 activation. Thus, SLP-65 amounts control the balance between Pax5 and FoxO1/Irf4 by regulating PI3K signaling activity and therefore, characterizing the exact roles of these transcription factors is essential for understanding pre-B cell commitment and differentiation.
The essential role of PI3K signaling in the initiation of pre-B cell differentiation is supported by studies that directly interfere with PI3K function in developing B cells. For instance, combined deletion of BCAP and CD19 leads to severely impaired AKT activation upon BCR stimulation and the number of early pre-B cells is strongly increased in these mice 16 . In contrast, the number of late pre-B cells is markedly reduced, suggesting that, in the absence of both BCAP and CD19, the impaired activation of PI3K results in defective differentiation of large pre-B cells to the small pre-B cell stage 16 . This observation is in complete agreement with our finding that class IA PI3K-derived signals are required for inducing the expression of Pax5 and the downstream target SLP-65. In fact, p110dKO pre-B cells deficient for PI3K signaling, express reduced amounts of SLP-65 and high amounts of the pre-BCR on their cell surface. This suggests that the residual SLP-65 expression in p110dKO pre-B cells is not sufficient to activate SLP-65-dependent signaling that leads to pre-BCR down-regulation and further differentiation 10 . Therefore, we propose that the amounts of Pax5 and SLP-65 at the onset of pre-BCR expression are insufficient to drive the differentiation processes. Increased PI3K signaling mediated by pre-BCR function results in elevated amounts of Pax5 and SLP-65 thereby activating the signaling cascade for differentiation. In line with this, the B cell developmental block in mice with impaired PI3K signaling, p110dKO mice, is similar to that observed in the absence of BCAP/CD19 or in SLP-65-deficient mice 11,12,16 .
A recently identified patient lacking p85α showed a severe block in B cell development while other hematopoietic lineages were basically not affected 40 . Importantly, the phenotype of this p85α patient resembles that of λ5or SLP-65-deficient patients. Both λ5 and SLP-65 represent well-known components of pre-BCR assembly or signaling 41,42 . In contrast, IL-7Rα-deficient patients lack T cells but have normal B-cell numbers 43,44 . If IL-7R signaling was essential for PI3K activation in developing B cells, PI3K-deficient patients were then expected to reveal a similar phenotype as IL-7R-deficient patients. Together, these findings support the view that PI3K activation is induced by pre-BCR signaling and that IL-7R is not essential for this process.
Notably, while B cell development is completely blocked in p110dKO mice, T cell development is not affected 24 . Moreover, our finding that p110dKO pre-B cells grow in vitro in IL-7 supplemented medium suggests that IL-7R dependent survival and proliferation signals do not require PI3K signaling. Together with our finding that p110dKO pre-B cells have reduced expression of Pax5 and SLP-65, the available data suggest that class IA PI3K activity is specifically important for the development of B cells.
The PI3K-mediated activation of Pax5 together with the well-established roles of PI3K signaling as key signaling axis for B cell survival and of Pax5 as essential factor for B cell development point to an unexpected mechanism, in which Pax5-mediated B cell differentiation is linked to survival. In this scenario, differentiation is a consequence of survival and proliferation. The molecular link between B cell survival and differentiation ensures that pre-BCR-mediated activation of class IA PI3K leads to B cell-specific gene expression and explains the specific importance of PI3K for B cell development. Our finding that PI3K is also required for Pax5 expression in mature B cells suggests that the BCR-generated survival signals, which are mediated by class IA PI3K 14 , maintain Pax5 expression at later developmental stages.
It should be noted, however, that certain amounts of Pax5 are expressed in developing cells before presence of μHC suggesting that other signaling pathways such as IL-7R are involved in the activation of Pax5 expression. IL-7R signals, which are known to activate survival and proliferation of pro-B cells 45 , may contribute to the induction of Pax5 expression by activation of STAT5 and Ebf1 that bind to the Pax5 gene [46][47][48][49] . However, it is unlikely that IL-7R activates Pax5 expression by induction of PI3K signaling. In fact, no change in phosphorylation of AKT was detected upon ablation of IL-7Rα gene expression. Furthermore, the defective PI3K signaling in p110dKO mice led to a developmental block at the pre-B cell stage indicating that pro-B cells, which require IL-7 for survival, proceed normally through early developmental stages in the absence of PI3K. Thus, together with our results, the available data suggest that IL-7R might be involved in the induction of Pax5 expression by activation of STAT5 or Ebf1 but not by PI3K activation, which is specifically activated by the pre-BCR.
Notably, our study is in disagreement with previous findings suggesting that IL-7R signaling activates PI3K signaling and that attenuation of this IL-7R-induced PI3K-signaling is required for induction of Pax5 and thus pre-B cell differentiation 27 . However, the previous study mainly utilized Irf4/Irf8 double deficient pre-B cells and it is unclear how the absence of Irf4 influences the obtained results 27 . Since we identified Irf4 as critical regulator of Pax5 expression and Pax5 as important target for PI3K-mediated activation, it is conceivable that the lack of Irf4 may affect the regulation of PI3K signaling in the Irf4-deficient cells. Importantly, the ability of Irf4 to control Pax5 expression is supported by available reports showing that Irf4 represses Pax5 gene expression 50 . In addition, available data show that the pre-BCR is only transiently expressed during early B cell development suggesting that its signaling activity is tightly regulated. Thus, experiments measuring induction of AKT phosphorylation after 4 days of reconstitution of pre-BCR expression in Rag2-deficent cells cannot detect any pAKT induction most likely because the transient pre-BCR signaling was terminated. Our data suggest that the transient pre-BCR expression and function regulate the switch from proliferation to differentiation. During early phases of of pre-BCR signaling proliferation is induced and at the same time the expression of genes required for differentiation such as SLP-65 is activated. Induction of SLP-65 function downstream of the pre-BCR signaling down-regulates PI3K activity and allows differentiation. This scenario is in full agreement with our previous results elucidating the role of SLP-65 9 and explains why deficiencies in Pax5, SLP-65, Irf4 or PI3K signaling are blocked at the pre-B cell stage of differentiation.
Moreover, we believe that characterization of the role of PI3K in regulating Pax5 expression and B cell commitment improves our understanding of general differentiation processes, as it elucidates how commitment, development and differentiation are regulated by a survival pathway such as PI3K, thereby maintaining the identity of the developing B cells. It is tempting to speculate that similar signaling networks exist in other hematopoietic lineages and might be exploited to regulate the differentiation, survival or proliferation of hematopoietic cells.

Experimental Procedures
Ethics statement. Wildtype (wt), SLP-65-deficient 11 , FoxO1 fl/fl 51 , p110α −/− /p110δ D910A 24 , p110α fl/fl /p110δ fl/fl 52,53 , and IL-7Rα fl/fl 28 mice were used in this study. 8 to 10 weeks old Rag2/common γ chain double deficient mice 54 were used for adoptive transfer of p110dKO cells. All experiments involving animals were reviewed and approved by the institutional animal care and use committee of Ulm University Medical Center and the Max-Planck-Institute of Immunobiology and Epigenetics. All methods were performed in compliance with the humane care and use of laboratory animals and the german animal welfare law. Buffy coats used for the preparation of human peripheral B cells were purchased from the Transfusion Center of Ulm University Medical Center (Institut für Klinische Transfusionsmedizin und Immungenetik Ulm GmbH, Ulm, Germany) and were obtained from anonymized healthy blood donors. All blood donors gave written informed consent to approve and authorize the use of their blood for medical, pharmaceutical, and research purposes.
Cell culture and biochemistry. For generation of pro-/pre-B cell lines, we isolated bone marrow from the respective mice to generate IL-7-dependent pre-B cell lines. Freshly isolated murine cells were therefore cultured in Iscove's medium containing 10% heat inactivated FCS (Vitromex), 2 mM L-glutamine, 100 U/ml penicillin, 100 U/ml streptomycin (Invitrogen), 5 × 10 −5 M 2-mercaptoethanol and IL-7. IL-7 was produced in-house as a sterile-filtered supernatant of J558L cells. Culturing of primary bone marrow cells with medium containing IL-7 results in rapid enrichment of early B cells and finally in stably growing cell lines, which importantly can be propagated independent from feeder cell lines. This system to establish stable IL-7-dependent B cell cultures has been used by us and others extensively. Hallmark of this system is that all generated cell lines are dependent on exogenous IL-7 provided in cell culture medium. Consequently, withdrawal of IL-7 leads to rapid cell death of the cell lines (see Fig. 4a and data not shown). M-CSF was produced with help of the M-CSF producing L929S cell line (kindly provided by M. Freudenberg) or purchased from Immunotools. Mature B cells were isolated from mouse spleens by using negative selection with anti-CD43 MACS-beads (Miltenyi Biotec) and cultured in medium without IL-7. Human peripheral B cells (CD19 + ) were isolated by FACS sorting. The human pre-B cell line Nalm-6 and the mature B cell line SU-DHL-6 were cultured in medium without addition of IL-7. For inhibition of PI3K, SLP-65-deficient pre-B cells and bone marrow derived wt pre-B cells were treated with 30 μM LY294002 (Merck Biosciences), Nalm-6 and SU-DHL-6 cells with 50 μM LY294002 for the indicated time points. Freshly isolated primary mature murine B cells were treated with 15 μM LY294002 for the indicated time points. Stable growing cell lines of FoxO1 fl/fl (either IL-7 dependent or transformed by BCR-ABL) or IL-7Rα fl/fl (IL-7 dependent) bm-derived pre-B cells were retrovirally transduced with tamoxifen-inducible Cre-recombinase (ER T2 -Cre) or empty control vectors and subsequently selected by addition of puromycin. Cre-recombinase was activated by addition of 2 μM 4-hydroxytamoxifen (4-OHT). As a control cells were treated with EtOH (solvent of 4-OHT).

PCR and RT-PCR. Total RNA was isolated from B cells using Trizol reagent (Invitrogen) or RNeasy Plus
Mini KIT MiniPrep (Qiagen). The synthesis of cDNA was performed as previously described 39 . Gene expression was determined with gene specific primers using the SYBR-Green detection method (Applied Biosystems) and a 7500 Fast Real-Time PCR machine (Applied Biosystems). Results were calculated by the ∆C T -Method. Generally, sequences of all primers in this study are available upon request.
Plasmids and retroviral transduction. The plasmids for expression of the constitutively active mutant of AKT (myr-AKT), tamoxifen-inducible form of Cre (Cre-ER T2 ) 9 and FoxO1-A3 55 have been described previously. Open reading frames encoding for the constitutively active mutant of p110α (myr-p110α), Cre-recombinase and the human Irf4 (hIrf4) were subcloned into the pMIG-backbone containing an IRES-GFP cassette, thereby generating pMIG-myr-p110α, pMIG-Cre or pMIG-hIrf4-IRES-GFP, respectively. Target cells were retrovirally transduced as described previously 9 . In summary, the Phoenix retroviral producer cell line was transfected according to the manufacturer's instructions using GeneJuice (Novagen). Retroviral supernatants were harvested after 36 and 60 h and for the subsequent transduction, pre-B cells were mixed with supernatants and centrifuged at 300 g at 37 °C for 3 h.
Electroporation. WEHI cells were transfected with indicated constructs carrying upstream of Luciferase a Vκ21 promoter sequence as well as Renilla vector (rLuc) for normalization using Neon TM transfection System (Lifetechnologies). Luciferase levels were detected by help of Dual Luciferase Assay system (Promega).
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.