PerspectiveImmunology

B Cell Receptor Signaling: Picky About PI3Ks

See allHide authors and affiliations

Science Signaling  10 Aug 2010:
Vol. 3, Issue 134, pp. pe25
DOI: 10.1126/scisignal.3134pe25

Abstract

The B cell receptor (BCR) and the pre-BCR control cell fate at many stages of B cell development, survival, and antigen response. Most of these processes require the activation of phosphatidylinositol 3-kinase (PI3K). Previous work has pointed to p110δ as the key catalytic isoform of PI3K for many B cell responses. A study of mice with different combinations of PI3K mutations confirms the central role of p110δ in agonist-mediated signaling, while identifying an unexpected function for the p110α isoform in tonic signaling by the pre-BCR and mature BCR.

Phosphatidylinositol 3-kinases (PI3Ks) are lipid kinases whose products promote the assembly of signaling complexes at cellular membranes (1, 2). There are three classes of PI3K, but only class I enzymes can generate the important second messenger phosphatidylinositol 3,4,5-trisphosphate (PIP3). A key objective has been to determine how the four catalytic isoforms of class I PI3Ks (p110α, p110β, p110γ, and p110δ), which mediate the same enzymatic reaction, are engaged by different signaling systems to generate distinct functional outputs. The p110γ isoform (termed class IB) has unique functions that are determined in part by its association with a distinct subgroup of regulatory subunits (p101 or p84/p87). The other three class I catalytic isoforms (termed class IA) pair with the p85 subgroup of regulatory isoforms (which consists of p85α, p85β, p55α, p50α, and p55γ) without exhibiting any apparent preference. Nonetheless, genetic and pharmacological studies have provided clear evidence that p110α, p110β, and p110δ have unique functions in particular tissues and receptor systems (3). In B cells activated through antigen-mediated clustering of the B cell receptor (BCR), p110δ has a dominant role because its inactivation through gene targeting or with p110δ-specific inhibitors results in a profound blockade of PI3K signaling output and functional responses (4). A study from the Okkenhaug group (5) reports for the first time that p110α has a substantial role in B cells that is revealed when p110δ is inactivated. The redundant function of p110α is restricted to events that are controlled by a specialized form of ligand-independent signaling termed “tonic signaling.” These findings have important implications for future signaling studies and for the pharmaceutical development of PI3K inhibitors.

Previous work has demonstrated that mice with a Pik3cd null mutation (PI3Kδ−/−) or a kinase-deficient knock-in of Pik3cd (p110δD910A/D910A) have severely reduced numbers of marginal zone (MZ) and B-1 B cells relative to those of wild-type mice (4). MZ and B-1 cells are specialized subsets of B cells with a restricted tissue distribution and function; a more abundant B cell subset, referred to as follicular (FO) B cells, recirculates through lymphoid tissues and mediates diverse types of antibody responses as well as B cell memory. FO B cells in p110δD910A/D910A mice proliferate poorly in response to BCR clustering and produce aberrant antibody responses (4); however, inactivation of p110δ does not substantially affect the early development of B cells and only partially diminishes the survival of FO B cells. To explore possible roles for p110α or p110β in B cells, Ramadani et al. crossed mice with floxed alleles of Pik3ca (p110α) or Pik3cb (p110β) with CD2-Cre mice to achieve deletion at an early stage in B cell development (5). Mice lacking only p110α or p110β displayed normal percentages and numbers of B cells in the bone marrow and periphery, indicating that these isoforms of PI3K do not have unique functions. The conditional knockout strains were then crossed with knock-in mice carrying the p110δD910A/D910A mutation to explore the extent of redundancy between the p110 isoforms. Deletion of p110β in B cells had no additional effect when combined with inactivation of PI3Kδ; however, dual loss of p110α and p110δ resulted in a profound arrest of bone marrow development at the pre–B cell stage and a near-total absence of peripheral B cells. These findings pointed to overlapping functions of p110α and p110δ at key stages of B cell development and survival.

To investigate the pre-BCR checkpoint, Ramadani et al. examined the expression of Rag genes and the status of heavy chain gene rearrangements. The products of the Rag genes, Rag-1 and Rag-2, are indispensable components of the recombinase enzyme complex that joins together gene segments (V, D, and J) to produce rearranged antibody genes. In pro–B cells, a productive VDJ rearrangement results in the generation of full-length μ heavy chains that form a signaling-competent pre-BCR with surrogate light chains and immunoglobulin α (Igα) and Igβ. Assembly of the pre-BCR at the large pre–B cell stage causes cells to shut down the expression of Rag, preventing further heavy chain rearrangements so as to enforce allelic exclusion and initiate further B cell development. Through elegant fluorescence-based techniques on single cells, the authors demonstrated that pre–B cells deficient in p110α and p110δ failed to inhibit the expression of Rag and consequently displayed a higher frequency of rearranged heavy chain alleles than did pre-B cells from wild-type mice.

The absence of mature FO B cells in the doubly deficient mice could result simply from a complete block at the pre–B cell stage; however, additional experiments provided evidence that p110α and p110δ also controlled the survival of mature B cells. The authors made use of mice heterozygous for a kinase-deficient knock-in allele of Pik3ca (p110αD933A/WT) (6). Even when crossed with homozygous p110δD910A/D910A mice, the presence of one wild-type allele of Pik3ca was sufficient to support normal early B cell development. However, p110αD933A/WT p110δD910A/D910A mice had markedly fewer peripheral B cells relative to wild-type controls or mice with impaired activity of one or the other isoform (p110δD910A/D910A or p110αD933A/WT). An intermediate reduction in peripheral B cell number was observed in p110αD933A/WT p110δD910A/WT double heterozygous mice. Thus, maintenance of the peripheral pool of B cells requires the function of either p110α or p110δ. A caveat to this conclusion is that the compound knock-in approach does not rule out cell-extrinsic roles for p110 isoforms in B cell survival; however, the increased abundance of CD23 on the surface of surviving B cells from p110αD933A/WT p110δD910A/D910A mice is a sign of reduced PI3K-Akt signaling in the B cells (7).

The maturation and survival of FO B cells are controlled by tonic signaling that seems to be independent of ligand-induced clustering of BCRs (8, 9). Although controversial, some evidence suggests that the pre-BCR checkpoint is also ligand-independent (10, 11). Hence, the two observed defects in p110α and p110δ doubly deficient B cells are both linked to tonic signaling responses. In clear contrast, responses initiated by BCR clustering were entirely dependent on p110δ, with no contribution from p110α. In particular, loss of p110α had no affect on the phosphorylation of Akt, Foxo, or extracellular signal–regulated kinase (ERK) after stimulation with antibody against IgM, whereas the p110δ-specific inhibitor IC87114 completely blocked these events (5), as was seen in p110δD910A/D910A B cells (12). Similarly, proliferation induced by antibody against IgM was unaltered in p110α-deficient B cells (5) yet was abrogated in the absence of p110δ function (1215). Both primary and secondary immune responses to a haptenated protein appeared normal in mice with p110α-deficient B cells (5). In addition, the few peripheral B cells in doubly deficient mice produced antibody responses similar to those of p110δ-deficient animals—namely, disproportional production of IgE despite reduced amounts of IgG and IgA. Finally, the generation and/or maintenance of MZ and B-1 B cells is thought to require BCR clustering by self antigens (9), and these subsets are nearly absent when only p110δ is inactivated. The authors also showed that the increased abundance of CD25, which is triggered by pre-BCR clustering, was blocked in doubly deficient B cells; however, this experiment did not include a comparison to cells lacking only p110δ. Thus, it remains to be determined whether p110α is redundant with p110δ for ligand-initiated responses in pre–B or immature B cells.

This study provides strong evidence that both p110α and p110δ are engaged and contribute to tonic signaling by the BCR and pre-BCR, whereas p110δ is uniquely coupled to pathways initiated by BCR clustering. A key question that emerges from this finding is how selectivity versus redundancy is determined at a molecular level. One possibility is that tonic signals recruit class I PI3Ks primarily through the p85 regulatory subunits, whereas strong BCR clustering nucleates more complex signaling assemblies that engage p110δ directly. In the former case, each class IA catalytic isoform would be recruited to the receptor in direct proportion to its abundance. The absence of a phenotype in p110β and p110δ doubly deficient cells could simply be due to the low abundance of p110β. In the case of BCR clustering, the authors suggest that p110δ may be selectively recruited by certain members of the Ras guanosine triphosphatase (GTPase) family. There are at least two arguments against this model. First, PI3K activation triggered by BCR clustering depends on the binding of p85 molecules to phosphotyrosine motifs within CD19 and B cell adaptor for PI3K (BCAP) (Fig. 1) (16, 17). Second, a report that demonstrated the selective recruitment of p110δ to the BCR through the Ras family member TC21 (encoded by Rras2) linked this interaction to basal (tonic) signaling rather than to BCR clustering responses (18). Hence, further studies are necessary to establish the mechanism for the selective coupling of p110δ to clustered receptor complexes.

Fig. 1

Model for differential coupling to PI3K isoforms in distinct BCR signaling settings. (A) Mature B cells that encounter antigen experience strong clustering of the BCR, often with the CD19-CD21 co-receptor. The activation of PI3K under these conditions requires binding of the p85 regulatory subunits to phosphotyrosine residues in CD19 and the adaptor protein BCAP. However, even though p85 molecules can bind to different class IA catalytic isoforms indiscriminately, the p110δ isoform appears to be exclusively engaged in this setting. This suggests the existence of additional interactions between p110δ and BCR signaling components (denoted by dashed arrows). (B) Tonic signaling through the pre-BCR or BCR occurs in the absence of strong receptor clustering. Basal PI3K signaling clearly occurs under these conditions, possibly also through CD19 and BCAP (16, 24), and mediates crucial developmental steps and peripheral survival. The current study of p110α and p110δ doubly deficient B cells (5) shows that these two isoforms have a shared, redundant function in the setting of tonic signaling.

The PI3K signaling network is important in both cancer and immunological diseases. Consequently, the findings of Ramadani et al. are also relevant to the development of PI3K inhibitors by the drug discovery community. Among the four class I PI3K isoforms, p110α is a particularly important drug target in oncology because of the high frequency of tumors that contain activating mutations in the PI3KCA gene that encodes this isoform (19). The current work suggests that isoform-selective inhibitors of p110α should have little impact on B cell maturation or humoral immunity. However, it remains possible that acute inhibition of the catalytic function of p110α could affect BCR signaling, with the chronic absence of p110α in knockout cells leading to compensatory rewiring of the network. At present, this possibility cannot easily be tested with the kinase-deficient knock-in allele of Pik3ca, because homozygous p110αD933A/D933A embryos die early in development (6). When p110α-selective inhibitors become available, this question should be revisited.

p110δ is considered a promising target for inflammatory diseases because of its key functions in B cells, T cells, and other leukocytes (4, 20). The Ramadani et al. study confirms the central role of p110δ in the responses of B cells to antigen; however, the data also suggest that dual inhibitors of p110α and p110δ might be useful in certain disease settings. Such a target profile might lead to the effective depletion of peripheral B cells in patients with antibody-mediated autoimmune diseases. Moreover, the combined inhibition of p110α and p110δ might block survival signals in B cell malignancies that depend on chronic, active BCR signaling (21). Relative to “pan-PI3K” inhibitors that target the entire class I subgroup, selective inhibitors of p110α and p110δ might also have a favorable tolerability profile by sparing physiological processes that depend on p110β and p110γ.

This work adds an interesting twist to the emerging consensus that different receptors are coupled to distinct PI3K isoforms. In the case of BCR signaling, the same receptor engages distinct sets of catalytic isoforms depending on the nature of the signal. Other work from these authors has identified another interesting phenomenon: that oncogenic transformation abrogates signaling specificity, such that any class IA PI3K isoform can maintain cell proliferation in cancer cells (22, 23).

Acknowledgments

25.Funding sources: Supported by NIH grant AI085462 (D.A.F.) and a Faculty Mentor Program fellowship from the University of California, Irvine (J.J.L.). Competing interests: D.A.F. is a scientific advisor for Intellikine, a company that is developing PI3K inhibitors.

References and Notes

View Abstract

Stay Connected to Science Signaling

Navigate This Article