Research ArticleImmunology

PI3Kβ Plays a Critical Role in Neutrophil Activation by Immune Complexes

See allHide authors and affiliations

Science Signaling  12 Apr 2011:
Vol. 4, Issue 168, pp. ra23
DOI: 10.1126/scisignal.2001617

Abstract

Neutrophils are activated by immunoglobulin G (IgG)–containing immune complexes through receptors that recognize the Fc portion of IgG (FcγRs). Here, we used genetic and pharmacological approaches to define a selective role for the β isoform of phosphoinositide 3-kinase (PI3Kβ) in FcγR-dependent activation of mouse neutrophils by immune complexes of IgG and antigen immobilized on a plate surface. At low concentrations of immune complexes, loss of PI3Kβ alone substantially inhibited the production of reactive oxygen species (ROS) by neutrophils, whereas at higher doses, similar suppression of ROS production was achieved only by targeting both PI3Kβ and PI3Kδ, suggesting that this pathway displays stimulus strength–dependent redundancy. Activation of PI3Kβ by immune complexes involved cooperation between FcγRs and BLT1, the receptor for the endogenous proinflammatory lipid leukotriene B4. Coincident activation by a tyrosine kinase–coupled receptor (FcγR) and a heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptor (BLT1) may provide a rationale for the preferential activation of the β isoform of PI3K. PI3Kβ-deficient mice were highly protected in an FcγR-dependent model of autoantibody-induced skin blistering and were partially protected in an FcγR-dependent model of inflammatory arthritis, whereas combined deficiency of PI3Kβ and PI3Kδ resulted in near-complete protection in the latter case. These results define PI3Kβ as a potential therapeutic target in inflammatory disease.

Introduction

Class I phosphoinositide 3-kinases (PI3Ks) play major roles in the eukaryotic signaling pathways that link cell surface receptors to the control of cell function (1, 2). The receptors that signal through PI3Ks are highly diverse and include receptors for many different growth factors, hormones, antigens, and inflammatory stimuli. Upon activation by receptors, these enzymes generate the lipid messengers phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3, also known as PIP3] and phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2] in target cell membranes. These two lipids are recognized by numerous effector proteins by binding to conserved domains, the best established of which are a subgroup of pleckstrin homology (PH) domains. These PH domain–containing effector proteins include protein kinases [such as Akt, also known as protein kinase B (PKB), and Bruton’s tyrosine kinase (BTK)], guanine nucleotide exchange factors (for example, Grp-1 and PRex-1), and guanosine triphosphatase (GTPase)–activating proteins (GAPs), such as ARAP3, which play central roles in the regulation of important cellular responses, such as growth, survival, and movement (3). Moreover, the adenosine triphosphate (ATP)–binding site of PI3Ks is amenable to selective inhibition; thus, members of this gene family are attracting increasing attention as potential therapeutic targets, particularly in oncology and inflammation (4, 5).

There are currently four isoforms of class I PI3Ks defined (2, 3). Each is a heterodimer of a regulatory subunit and a homologous p110 (110 kD) catalytic subunit. PI3Kα, PI3Kβ, and PI3Kδ consist of p110α, p110β, and p110δ subunits, respectively, in addition to one member of a family of five related, Src homology 2 (SH2) domain–containing regulatory subunits, p50-p85. PI3Kγ consists of the p110γ catalytic subunit and either a p101 or a p84 regulatory subunit. Heterodimers of PI3K that contain p110γ and either p101 or p84 are classically activated by heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors (GPCRs) that couple to Gi proteins, through direct binding to Gβγ subunits and Ras proteins (6). Heterodimers of PI3K that contain p50 or p85 and either p110α or p110δ are activated by protein tyrosine kinase–coupled receptors, through p50- or p85-dependent binding to specific phosphorylated tyrosine residues, and also by direct binding to Ras proteins (7). Confusingly, heterodimers of PI3K that contain p110β appear to be engaged by either GPCRs or tyrosine kinase–coupled receptors, depending on the receptor or cellular context (811).

Studies involving a combination of mouse genetics and isoform-selective inhibitors have started to reveal specific, nonredundant roles for individual PI3K isoforms. In most cases, however, the molecular explanation for the specific role of a particular isoform is unclear, particularly for the selective engagement of the class IA enzymes PI3Kα, PI3Kβ, and PI3Kδ, given that these isoforms apparently share the same regulatory subunits. Although widely expressed, PI3Kα has emerged as a critical regulator of angiogenesis and insulin-dependent metabolism (12, 13). It is also a potent oncogene, carrying activating mutations in up to 40% of certain human tumors (14). PI3Kδ is highly abundant in cells of the hematopoietic lineage and plays important, nonredundant roles in the development and function of T cells and in mast cell activation (15, 16). PI3Kγ is also highly abundant in hematopoietic cells and is an important regulator of the functions of macrophages, neutrophils, and mast cells (17). More recently, PI3Kγ has also emerged as a critical modulator of cardiac function (18).

Compared to the other isoforms of PI3K, less is known about PI3Kβ. Specific catalytic site inhibitors of PI3Kβ are reported to block the responses of platelets to shear stress and to regulate purinergic receptor–mediated stabilization of arterial thrombi (11, 1921). PI3Kβ is thought to be widely expressed, and its deletion in the mouse was initially reported to cause an early block in embryonic development (22). Subsequently, tissue-specific knockouts and a kinase-defective knock-in of p110β have been generated, and it now appears that a permissive genetic background enables some p110β-null and p110β kinase–deficient embryos to escape death (8, 10). Studies of mouse embryonic fibroblasts (MEFs) derived from these embryos indicate that PI3Kβ is surprisingly insensitive to activation by growth factors that act through classical receptor tyrosine kinase signaling, such as platelet-derived growth factor, insulin, and epidermal growth factor, but is required for the activation of PI3K signaling downstream of certain Gi-coupled receptors, such as receptors for lysophosphatidic acid (LPA), sphingosine 1-phosphate (S1P), and stromal cell–derived factor 1α (SDF-1α, also known as CXCL12) (810). Furthermore, p110β appears to have a kinase-independent function in the proliferation of MEFs (9, 10). There is also evidence that p110β may play an important role in PI3K signaling in mouse models of prostate and mammary gland tumorigenesis (9, 10).

Neutrophils are key components of our innate immune system, which are primarily responsible for combating bacterial and fungal infections, but they are increasingly implicated in the pathogenesis of several inflammatory conditions (for example, rheumatoid arthritis, glomerulonephritis, and acute respiratory distress syndrome), in which their infiltration into inflammatory sites exacerbates local inflammation and contributes to tissue damage (23, 24). Neutrophils have various cell surface receptors that are coupled to PI3K-dependent processes, including chemotaxis, phagocytosis, cell spreading, and the generation of reactive oxygen species (ROS) by the NOX2-containing NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase (25). Although neutrophils have all four PI3K isoforms, most research to date has focused on the γ isoform. PI3Kγ plays an important role in the signaling pathways that link Gi-coupled GPCRs for soluble, proinflammatory stimuli, including interleukin-8 (IL-8), N-formyl-Met-Leu-Phe (fMLP), fragment of complementary protein C5 (C5a), and leukotriene B4 (LTB4), to the regulation of neutrophil chemotaxis, extravasation, and the production of ROS (17). However, the extent to which PI3Kγ may be more critical in pathology rather than in the normal control of infection is still unclear.

The specific role of neutrophils in pathology is probably clearest in mouse models of the “effector phase” of autoimmune disease, in which inflammation is caused by the passive transfer of antibodies to self-antigens, causing the deposition of immune complexes (ICs) that cannot be cleared effectively, which leads to tissue damage and nonresolving inflammation, as for example, in mouse models of experimentally induced autoimmune arthritis, glomerulonephritis, and skin-blistering disease (2629). In these models, the involvement of neutrophils depends on their ability to exit the vasculature and be recruited to sites of IC-mediated complement fixation and production of cytokines, chemokines, or both. Once at sites of IC deposition, neutrophils bind directly to the ICs through low-affinity antibody receptors (FcγRs), which results in the secretion of extracellular ROS, proteases, and further proinflammatory stimuli, which lead, in turn, to further cycles of tissue damage and inflammation (30). Recent work, largely focused on other cell types, has revealed that FcγRs signal through the immunoreceptor tyrosine-based activation motif (ITAM)–containing, γ chain adaptor and Src- or Syk-driven signaling pathways, including those initiated by phospholipase C (PLC), PI3K, and Ras (31). The extent to which the various PI3K isoforms are involved in the activation of neutrophils downstream of these receptors, however, is unknown.

Here, we investigated the effects of PI3Kβ gene disruption and inhibition of PI3Kβ activity on neutrophil ROS production as an initial, highly sensitive screen for receptor-specific PI3Kβ function in these cells. We provide evidence that the PI3Kβ isoform plays an essential, nonredundant role in the efficient activation of mouse neutrophils by immobilized ICs in vitro. Moreover, activation of neutrophils by ICs involved the engagement of FcγRs and a paracrine-autocrine loop of LTB4 synthesis and activation of its Gi-coupled GPCR, BLT1. In this context, the coincident activation of neutrophils by both a tyrosine kinase–linked receptor and a GPCR may provide a potential rationale for the selective engagement of the PI3Kβ isoform. Furthermore, loss of PI3Kβ conferred protection in mouse models of the effector phase of arthritis and a skin-blistering disease, suggesting that PI3Kβ might be a useful therapeutic target in combating autoimmune disease.

Results

PI3Kβ plays a selective role in neutrophil activation by ICs

We investigated a potential role for PI3Kβ in neutrophil function in experiments with bone marrow–derived neutrophils (BMNs) isolated from mice carrying a homozygous germline deletion of exons 21 and 22 of the gene encoding p110β (Pik3cbΔ21,22/Δ21,22; hereafter referred to as βKO mice) or from mice homozygous for a kinase-deficient knock-in allele of p110β (Pik3cbD931A/D931A; hereafter referred to as βKD mice). βKO and βKD mice were born at lower than the expected Mendelian ratios. From a heterozygote cross, βKO mice were 7.43% of littermates compared to an expected 25% from 794 live births, whereas βKD mice represented 3.45% of littermates compared to an expected 25% from 290 live births. Adult mice were, on average, slightly smaller than their wild-type littermates (wild-type mice weighed 27.6 ± 5.9 g compared to βKO mice at 22.6 ± 3.9 g, for 29 mice per genotype; P < 0.001). These mice had no other obvious abnormalities and had normal blood cell counts (table S1). We found no evidence for the expression of any theoretically possible N-terminal fragment of p110β in βKO cells (fig. S1 and table S2), and the abundance of p110β was unchanged in βKD BMNs compared to that in wild-type mice (fig. S1). Furthermore, the abundances of p110α, p110γ, and p110δ were unaffected in both βKO and βKD BMNs (fig. S1).

Relative to wild-type BMNs, βKO and βKD BMNs exhibited substantially reduced ROS production upon interaction with submaximal densities of immobilized ICs prepared with antibodies against bovine serum albumin (BSA-ICs) (Fig. 1 and fig. S2). Further, acute pharmacological inhibition of PI3Kβ with a cell-permeable, ATP site inhibitor of p110β (TGX221; inhibition constant Ki ~5 to 10 nM) (11) reduced the analogous wild-type responses to a similar extent as that observed in βKO and βKD BMNs (Fig. 1 and fig. S2). The scale of the effects of inhibition of PI3Kβ declined progressively as the density of BSA-IC was increased, from ≥80% reduction at lower densities to an ~40% reduction at the highest density of BSA-IC tested (Fig. 1).

Fig. 1

PI3Kβ has a nonredundant role in IC-elicited ROS production. (A) Wild-type (WT) or p110 mutant BMNs pretreated with DMSO or TGX221 (40 nM) were applied to BSA-ICs formed with a 1:10,000 dilution of anti-BSA antibody, and ROS production was measured in duplicate. Data are means ± SEM (n ≥ 3) and are expressed as the percentage of the peak ROS production observed in DMSO-treated WT BMNs. (B) ROS production in response to BSA-ICs that contained increasing concentrations of anti-BSA antibody. Data are means ± SEM (n ≥ 3) and are expressed as the percentage of the total ROS production observed in DMSO-treated WT BMNs adhering to ICs generated with a 1:2500 dilution of anti-BSA antibody. BSA-coated and milk-blocked wells provided controls devoid of ICs (No Ab).

In contrast, genetic ablation of PI3Kγ (γKO) or PI3Kδ activity (δKD) had a much smaller effect on ROS production in response to BSA-IC (Fig. 1), indicating a selective role for PI3Kβ in these responses. We did notice, however, that over the entire course of this study, responses in the γKO BMNs were characteristically more variable than those of the other genotypes investigated, sometimes reaching as low as 35 to 40% of those of wild-type BMNs at early times and low doses of IC. Nevertheless, even the highest extent of inhibition observed in γKO BMNs failed to approach the extent of inhibition that was consistently seen with the loss of PI3Kβ. Unfortunately, it was difficult to interpret the effects of acute inhibition of PI3Kγ or PI3Kδ, because the best PI3Kγ- and PI3Kδ-selective inhibitors available (AS605240 and IC87114, respectively) (32, 33) had substantial off-target activities in these assays (as revealed by inhibition in their respective genetic knockout backgrounds, fig. S3). It was clear, however, that the addition of TGX221 to δKD BMNs, but not γKO BMNs, consistently inhibited ROS production at higher densities of BSA-IC to a greater extent than did addition of this compound to wild-type BMNs (Fig. 1B). Further, combined genetic deletion of PI3Kβ and PI3Kδ activity also produced more severe inhibition at the higher densities of IC than did deletion of PI3Kβ alone (Fig. 1B). These results indicate that some “redundancy” developed between PI3Kβ and PI3Kδ as the density of BSA-IC was increased, such that the loss of PI3Kβ was substantially compensated for by PI3Kδ at higher densities of BSA-IC.

Inhibition of PI3Kβ did not affect ROS responses to phorbol 12-myristate 13-acetate (PMA) (Fig. 2A) or Staphylococcus aureus (Fig. 2B) or the killing of serum-opsonized S. aureus in vitro (Fig. 2E). This is consistent with the prevailing view that activation of the neutrophil oxidase by PMA bypasses receptors and that activation of the oxidase by phagocytosis of S. aureus is dependent on class III but not class I PI3Ks (34). ROS production in response to fMLP (Fig. 2C) or LTB4 (Fig. 2D) and the chemotaxis of BMNs toward a source of fMLP on a glass surface (35) were also unaffected by specific inhibition of PI3Kβ; however, ROS production in response to fMLP and LTB4 was substantially inhibited by the loss of PI3Kγ (Fig. 2, C and D), consistent with a previous study that indicated that the GPCRs for these chemoattractants signal predominantly through the γ isoform of PI3K (5).

Fig. 2

PI3Kβ is not required for ROS production in response to GPCR agonists alone, to serum-opsonized S. aureus, or for S. aureus killing in vitro. (A to D) ROS production in response to (A) PMA (300 nM), (B) serum-opsonized S. aureus [BMN/S. aureus (S.A.) = 1:20], (C) fMLP (10 μM), or (D) LTB4 (100 nM). Data are means ± SEM (n ≥ 3) or ± range (n = 2) for PMA (n ≥ 3); S.A. (n = 2 except for βKO + S.A. + TGX221, which was performed once in duplicate); fMLP (n ≥ 3); LTB4 (n = 3). (E) Primed BMNs, pretreated with DMSO or the oxidase inhibitor, DPI, were mixed with S. aureus for 15 min at 37°C and the number of surviving S. aureus was determined. Data are the means ± range (n = 2) number of live S. aureus.

Previous studies have indicated that ROS production by BMNs in response to ICs is driven by the low-affinity antibody receptors FcγRIII and FcγRIV, phosphorylation of the common Fc γ chain by Src family kinases, and recruitment of the protein tyrosine kinase, Syk (30). Further, a study has suggested that β2 integrins can also use the Fc γ chain as an adaptor for Syk in pathways that regulate the NOX2-containing oxidase (36). We found that inhibition of PI3Kβ did not substantially affect the abundances of FcγRIII or FcγRIV (fig. S4, A and B), the extent of FcγR-induced global tyrosine phosphorylation (Fig. 3A), or the phosphorylation of the ITAM motif in the Fc γ chain (Fig. 3B). In addition, inhibition of PI3Kβ only partially compromised the ability of BMNs to adhere to and spread on BSA-ICs (fig. S5). However, specific inhibition of PI3Kβ substantially reduced FcγR-mediated phosphorylation of the PI3K effector, Akt, extracellular signal–regulated kinase 1 (ERK1) and ERK2, and the secretion of the tertiary granule marker, gelatinase (Fig. 3, C to G). Inhibition of PI3Kβ also substantially inhibited ROS production in response to integrin-mediated adhesion of cells to tumor necrosis factor (TNF)–fibrinogen and poly-Arg-Gly-Asp (poly-RGD) (fig. S6), without affecting the amount of surface integrin expression (fig. S4, C and D). Combined inhibition of PI3Kβ and PI3Kδ reduced these responses still further (Fig. 3 and fig. S6). Collectively, these results suggest that PI3Kβ operates at a point between phosphorylation of the Fc γ chain and responses further downstream, and plays a particularly prominent role in driving ROS production at low doses of IC.

Fig. 3

PI3Kβ regulates FcγR-activated phosphorylation of Akt and ERK1/2 and gelatinase secretion downstream of phosphorylation of the Fc γ chain. (A and B) WT or βKO BMNs pretreated with DMSO or TGX221 (40 nM) were activated by FcγR cross-linking (XL) for 30 s. (A) Total cellular tyrosine phosphorylation (pY). Arrows represent bands that observably increased in intensity upon FcγR cross-linking. Data are from one experiment that is representative of two independent experiments performed in duplicate. β-Actin is a loading control. (B) Western blot depicts Fc γ chain with phosphorylated ITAMs captured from lysates with a Syk(SH2)2 construct in a pull-down assay. β-Actin from whole-cell lysates served as a control for protein loading. Control samples received no anti-Fc F(ab′)2. (C and D) BMNs were applied to immobilized IC (1:10,000 dilution of anti-BSA antibody) and lysed, and the phosphorylation of Akt and ERK1/2 was determined. Data in (D) are means ± SEM (n ≥ 4) and expressed as a percentage of the Akt phosphorylation detected in DMSO-treated WT BMNs adhering in the presence of antibody (Ab). Western blots in (C) are representative of four independent experiments. (E) Gelatinase release from WT BMNs in response to increasing concentrations of BSA-IC in the absence or presence of TGX221 (40 nM); the zymograph shown is a representative example taken from three independent experiments. (F) Gelatinase release from WT and βKO BMNs in response to BSA-IC (1:2000 dilution of anti-BSA antibody); the zymograph shown is a representative example taken from three independent experiments. (G) Densitometric analysis of gelatinase zymographs in WT or βKO BMNs in response to ICs (1:2000 anti-BSA antibody). Data are means ± SD (n = 4 for WT; n = 3 for βKO).

IC-elicited ROS production is dependent on a paracrine-autocrine loop involving LTB4 and BLT1

Previous work has suggested that 5-lipoxygenase (5-LO)–derived LTB4 may play a role in FcγR-stimulated signaling in alveolar macrophages through binding to its Gi-coupled receptor BLT1 (37). Low doses of the 5-LO inhibitor, AA861, or the selective BLT1 antagonist, CP105,696, potently inhibited ROS production in response to BSA-IC by wild-type BMNs (Fig. 4). IC-elicited ROS production in γKO BMNs was similarly suppressed by these inhibitors (Fig. 4), despite the critical role for this isoform in promoting ROS production in response to LTB4 alone (Fig. 2). Further, MK886, a potent inhibitor of the arachidonic acid transfer protein FLAP (which is required for 5-LO activity), also strongly suppressed BSA-IC–elicited ROS production (Fig. 4C). These results suggested that an autocrine-paracrine loop involving LTB4 operated in BMN responses to immobilized BSA-IC in a manner largely independent of PI3Kγ.

Fig. 4

Binding of LTB4 to its receptor BLT1 promotes IC-induced ROS production independently of PI3Kγ. (A to E) WT or γKO BMNs were pretreated with an inhibitor of 5-LO (AA861) or the AA-transfer protein, FLAP (MK886), or, alternatively, a BLT-1 antagonist (CP105,696) at the indicated concentrations, and IC-induced ROS production was measured. Data are expressed as means ± SEM (n ≥ 4) for WT and γKO mouse data as a percentage of the peak ROS response measured in the corresponding WT BMNs treated with vehicle alone (Control).

The addition of LTB4 alone to 5-LO–inhibited BMNs stimulated rapid, transient ROS production (with a peak of ~10 s) and Akt activity that were selectively dependent on PI3Kγ (Fig. 5), consistent with our earlier results for cells in which 5-LO was not inhibited (Fig. 2). In contrast, the addition of LTB4 to 5-LO–inhibited BMNs in the presence of BSA-IC stimulated more slowly developing, but persistent, ROS (peak of ~20 min) and Akt responses that were selectively dependent on PI3Kβ (Fig. 5). Thus, BLT1 drove rapid signaling through PI3Kγ, but cooperated with FcγRs to drive sustained responses through PI3Kβ.

Fig. 5

PI3Kβ is cooperatively engaged by FcγR and BLT1 to mediate sustained BMN responses. (A to F) AA861-treated (0.3 μM) WT or βKO BMNs were further treated with vehicle or TGX221 and ROS production (A and B, D and E) or Akt phosphorylation (C and F) assessed in the absence (Suspension) or presence (Immune Complex) of IC (1:10,000 dilution of anti-BSA antibody). BMNs were concurrently stimulated by exogenous LTB4 (10 nM) where indicated. Data in (A) and (B) are presented over a 30-s time period because of the rapid and transient nature of the ROS response (mean ± range from a single representative experiment performed in duplicate), whereas data in (D and E) are presented over a prolonged period because ROS production was sustained (mean ± SEM, n = 4). (C and F) Western blots are from single representative experiments (n = 3), whereas data in (F) are means ± SEM (n = 4). Note that LTB4 added to cells in suspension induced ROS production to a much lower extent than when in the presence of IC. Furthermore, exogenous LTB4 failed to elicit sustained ROS production on BSA alone in the absence of an anti-BSA antibody.

Loss of PI3Kβ confers substantial protection in a mouse model of epidermolysis bullosa acquisita

FcγRs on neutrophils can play a cell-autonomous and critical role in the development of the effector phase of certain autoimmune diseases, that is, the phase during which most tissue damage occurs because of ROS or proteolytic activity in response to IC recognition (38). We therefore sought to establish whether PI3Kβ was important in IC-elicited damage in vivo. We initially chose a mouse model of the human autoimmune blistering disease epidermolysis bullosa acquisita (EBA), which can be given to naïve mice by the transfer of rabbit polyclonal antibodies against mouse collagen VII (designated anti-mcVII) (29). Immobilized mcVII-IC stimulated a large ROS response from BMNs in vitro, and this response showed a similar sensitivity to inhibition of PI3Kβ as did the BSA-IC–elicited responses described earlier (Fig. 6A).

Fig. 6

βKO mice are resistant to experimentally induced EBA (A) BMNs from WT or βKO mice were applied to wells containing rabbit anti-mcVII antibody alone or antibody complexed with mcVII (10 μg/ml), and ROS production was measured in duplicate (mean ± SEM, n ≥ 3). (B) Comparison of skin blistering induced by injection of rabbit anti-mcVII antibody into WT or βKO mice (left-hand panel; n = 16 to 18 mice) and into WT/WT or βKO/WT bone marrow chimeras (right-hand panel; n = 6 mice). Data were arbitrarily binned into three indices of disease severity for all genotypes and the mean percentage of mice falling within each index was calculated. The data were subjected to t test and the resulting P values are given above the clinical score plots. (C) Histological hematoxylin and eosin (H&E)–stained sections of skin from trunks of mice injected with anti-mcVII Ig were scored for inflammatory cell infiltration (0, none, to 3, severe). White circles represent infiltration scores from mice in (B) (left-hand panel), whereas black circles represent data from mice in (B) (right-hand panel). Black lines represent the median infiltration score for combined data. (D) Representative photographs of WT and βKO mouse trunk appearance (top panels) and micrographs of H&E-stained skin sections 12 days after injection with nonimmune antibody (NR-Ig) or rabbit anti-mcVII Ig (bottom panels). Bottom left panels show sections in lower magnification, whereas bottom right panels show magnifications of the highlighted areas on the left. Arrows indicate dermal-epidermal junctions. Note: In some βKO mice administered with anti-mcVII Ig, dermal-epidermal junctions were preserved despite exhibiting moderate inflammatory cell infiltrates.

βKO mice were resistant to the development of blisters in response to injections of rabbit anti-mcVII antibodies (Fig. 6, B to D). In wild-type mice, the hallmarks of such blisters are scaly skin patches, hair loss, or both. Microscopically, these are seen as the separation of the dermal-epidermal junction as well as a dense inflammatory infiltrate and a thickened epidermal layer. Immunohistochemical analyses of mouse skin for the presence of deposits of rabbit immunoglobulin G (IgG) and complement (C3) and measurements of anti-mcVII antibody titers from injected mice revealed no differences between wild-type and βKO mice (fig. S7). Furthermore, βKO radiation chimeric mice, which were generated by the injection of βKO bone marrow into lethally irradiated C57BL/6J mice, were similarly protected from the development of blisters in this model (Fig. 6B), indicating that this protection was autonomous to the hematopoietic lineage.

We also carefully evaluated the extent of inflammatory infiltrate in the skin of these mice (Fig. 6C). Whereas the median extent of infiltration in both βKO mice [Fig. 6, C (white circles) and D] and βKO radiation chimeras [Fig. 6, C (black circles) and D] was substantially reduced, there were numerous examples in the βKO mice where dermal-epidermal separation failed to occur despite the presence of normal amounts of infiltrating leukocytes. Moreover, isolated wild-type neutrophils elicited damage upon local reconstitution in the skin of βKO mice that had been injected with anti-mcVII antibody, but not nonimmune antibody (fig. S8). These data suggest that there is a primary, cell-autonomous role for PI3Kβ in granulocyte-mediated tissue damage at sites of deposition of anti-mcVII antibodies, which probably influences further rounds of recruitment of inflammatory cells.

We also investigated the PI3Kβ dependency of human neutrophil responses to ICs. Human neutrophil ROS production in response to BSA-IC was sensitive to inhibition by low doses of TGX221, although the extent of inhibition was slightly lower than with mouse neutrophils (fig. S9, A and B). This may reflect a different balance of PI3K isoform usage between human and mouse neutrophils, as has been previously noted downstream of fMLP stimulation (39). We also prepared immobilized EBA-IC by coating plates with recombinant human collagen VII (hcVII) and adding sera from six normal donors or from seven donors diagnosed with EBA and containing increased amounts of anti-hcVII immunoreactivity (fig. S10). EBA, but not normal, sera elicited a large ROS response in primary human neutrophils (fig. S9C). Moreover, these ROS responses were inhibited by TGX221 (fig. S9D), which suggested that EBA sera–induced ROS production by human neutrophils was PI3Kβ-dependent.

Loss of both PI3Kβ and PI3Kδ activity confers substantial protection in the K/BxN mouse model of rheumatoid arthritis

The transfer of arthritogenic K/BxN serum to naïve mice, with the resulting neutrophil-dependent joint inflammation that occurs, is an established model of the effector phase of rheumatoid arthritis (40). βKO chimeric mice were partially, but significantly, protected from the development of clinical signs of arthritis, including joint swelling, in response to a dose of K/BxN serum (150 μl), which was at the lower end of the range normally used in this model; this was particularly true of cohorts of mice that responded with low to medium indices of severity (Fig. 7 and figs. S11 and S12). No protection was seen in βKO chimeric mice that were subjected to a higher dose of arthritogenic serum, and δKD chimeric mice exhibited normal disease progression in response to either dose (Fig. 7 and figs. S11 and S12). Chimeric mice prepared with bone marrow lacking both PI3Kβ and PI3Kδ activity (βKO/δKD) were highly protected in this model at both low and high doses of K/BxN serum (Fig. 7 and figs. S11 and S12). In contrast to the clear role of PI3Kβ in the effector phase of both the K/BxN and the EBA mouse models of autoantibody-driven inflammation, we found no evidence that PI3Kβ contributed to antigen receptor signaling in B cells or T cells, and humoral immune responses were unaffected by lymphocyte-specific deletion of PI3Kβ (fig. S13).

Fig. 7

p110β and p110δ are critical for the clinical symptoms of arthritis. βKO, δKD, βKO/δKD mice, and their corresponding WT counterparts were injected with arthritogenic K/BxN serum (150 or 400 μl), and the hindlimbs of animals were scored for clinical signs of arthritis. Data were arbitrarily binned into three indices of disease severity for all genotypes and the mean percentage of mice falling within each index was calculated. Data are from the most severely affected limb in every animal over the entire experimental period. The results of statistical tests are indicated adjacent to a given WT-p110 mutant cohort. Data are derived from n = 22 (150 μl) and n = 9 mice (400 μl) for βKO mice; n = 8 (150 μl) and n = 4 mice (400 μl) for δKD mice; and n = 11 (150 μl) and n ≥ 11 mice (400 μl) for βKO/δKD mice.

Discussion

Genetic or pharmacological inhibition of PI3Kβ substantially reduced several IC-elicited responses in BMNs, indicating that this isoform played a role in coupling the activation of neutrophil FcγRs to the activation of Akt, ERK1/2, the secretion of gelatinase, and the production of ROS. Genetic ablation of PI3Kγ or PI3Kδ alone had much smaller effects on these responses, although the combined inhibition of PI3Kβ and PI3Kδ characteristically produced more profound effects than did inhibition of PI3Kβ alone. IC-elicited ROS production, which had the advantage that it could be quantified accurately, was particularly sensitive to the inhibition of PI3Kβ at lower densities of IC, but combined loss of both PI3Kβ and PI3Kδ activity was required to achieve ≥80% inhibition at higher densities of IC, suggesting that increased redundancy developed between these isoforms as the strength of the stimulus increased. In contrast, PI3Kγ was unable to compensate for the loss of PI3Kβ at any dose of IC. These results imply that PI3Kβ and PI3Kδ isoforms provide a common “currency” in support of oxidase activation, but that PI3Kβ activity is specifically required to provide sufficient signal strength when the extent of FcγR occupancy is low. This idea is consistent with the expectation that all of the class IA PI3Ks (β, δ, and α) share a common family of regulatory subunits, which govern recruitment to phosphotyrosine-containing complexes, and a common catalytic product (PIP3), yet their catalytic subunits have distinct molecular properties (for example, differential sensitivities to Ras family proteins or Gβγ subunits) that may enable them to be better suited to a particular signaling niche.

To better describe the circumstances that define a selective requirement for PI3Kβ in FcγR-driven ROS responses, we considered the possibility that additional receptors might be involved. The activation of relatively pure populations of secretory cells in vitro is often complicated by paracrine or autocrine loops that involve the release of active substances that in turn contribute to the response being measured. This has been a particularly confusing aspect of understanding the signaling pathways downstream of individual receptors in platelets and mast cells, and it is becoming increasingly appreciated in studies of primary isolates of macrophages and neutrophils (4144). Of relevance here, stimulation of FcγR on macrophages involves an autocrine-paracrine loop of 5-LO–generated LTB4 and its action through the receptor BLT1 (37). Our data suggest that a similar autocrine-paracrine loop involving LTB4 and BLT1 operates in IC-elicited ROS production in BMNs. The precise contribution of this loop to IC-induced ROS responses varied to some extent between BMN preparations, but it was always in the range of 60 to 80% of the total ROS generated and thus represented a substantial contribution to this response. The ability to block endogenous LTB4 synthesis with 5-LO inhibitors enabled us to add back LTB4 to 5-LO–inhibited BMNs and thus compare the effects of LTB4 in the presence or absence of IC. The simple addition of LTB4 to quiescent BMNs stimulated extremely rapid, transient ROS and Akt responses that were blocked by a specific BLT1 antagonist and by the absence of PI3Kγ, but not PI3Kβ or PI3Kδ (Fig. 2 and fig. S14). Thus, on its own, LTB4 acted through BLT1 to elicit rapidly desensitizing responses through PI3Kγ, in agreement with previous work indicating that Gi-coupled GPCRs in neutrophils signal primarily through PI3Kγ (5). In the presence of ICs, however, LTB4 supported more prolonged ROS and Akt responses that persisted in the absence of PI3Kγ, but were highly dependent on PI3Kβ activity (Fig. 5 and fig. S15). This suggests that in the presence of IC, BLT1 did not become completely desensitized and could couple to PI3Kβ.

The molecular mechanisms that enable cell surface receptors to couple to PI3Kβ are unclear. There is compelling evidence that both protein tyrosine kinase–coupled receptors (for example, the GPIIb/IIIa integrin and GPVI in platelets) and some Gi-coupled GPCRs (for example, P2Y12 in platelets and the S1P and LPA receptors in MEFs) can engage PI3Kβ (811, 20, 21, 45). It has also been shown that the p110β-p85 heterodimer is synergistically activated by Gβγ subunits and p85-binding phosphotyrosine peptides in vitro (46); however, clear evidence for cooperative signaling to PI3Kβ through tyrosine kinases and GPCR-based activation in cells is lacking. Our results, which show that BLT1 and FcγR acted in concert through PI3Kβ, support this model and suggest that a specific adaptation of this isoform might be to amplify coincident signals through protein tyrosine kinases and GPCRs. A natural consequence of adopting this idea would be to search for phosphotyrosines that bind to p85 in cases in which GPCRs have previously been shown to engage PI3Kβ. PI3Kγ, by contrast, is still seen as the main PI3K that transduces purely GPCR-derived signals. Thus, where Gi, PI3Kβ, and PI3Kγ are found together, many situations are likely to involve parallel activation of both PI3Kβ and PI3Kγ, albeit with each isoform driving PIP3 responses with different magnitudes and kinetics. For example, the release of GPCR ligands that act directly through PI3Kγ may explain the variable PI3Kγ dependence of some of our IC-elicited ROS assays.

We investigated the PI3Kβ dependence of neutrophil activation by ICs in vivo in a mouse model of the human blistering disease, EBA. The effector phase of the EBA model exhibits robust inflammation in various genetic backgrounds and is caused by neutrophil accumulation at sites of IC and complement deposition in the dermis (29). Tissue damage and blistering is dependent on β2 integrins, FcγRs, protease secretion, and NADPH oxidase (NOX2)–generated ROS (47). βKO bone marrow chimeric mice were highly protected from the development of clinical symptoms of this disease, despite having normal amounts of IC and complement deposition and, in some mice, moderate inflammatory infiltrate in skin tissue, which suggests that PI3Kβ participates in neutrophil-mediated damage in this model. Human neutrophils have a different repertoire of activating FcγRs from that of mouse neutrophils (FcγRIIA and FcγRIIIB versus FcγRIII and FcγRIV, respectively), although receptors from both species are thought to signal through an ITAM-Syk axis (30). IC-elicited ROS responses in human neutrophils were also dependent on PI3Kβ activity, including those generated with recombinant hcVII antigen and EBA patient sera.

We also investigated the role of PI3Kβ in the K/BxN mouse model of experimentally induced arthritis. The passive transfer of arthritogenic K/BxN serum induces IC- and neutrophil-mediated damage to the joints, which depends on the local generation of LTB4 (48), but is independent of ROS generated by NOX2 (26). PI3Kγ and PI3Kδ play partial roles in the development of clinical symptoms of arthritis in this model (49). Our δKD chimeric mice were not substantially protected from disease, possibly because of subtle differences in experimental conditions, mouse strains, or both. βKO chimeric mice were partially, but significantly, protected from the development of arthritis to low doses of arthritogenic serum; however, βKO/δKD chimeric mice exhibited almost complete protection, even at high doses of serum. The K/BxN model involves several receptors, cell types, and cell responses; thus, it cannot be assumed that a quantitative model of “signaling redundancy” built from in vitro assays of BMNs will be directly applicable to conditions in vivo, but it is interesting that the combined deletion of PI3Kβ and PI3Kδ activities was required to elicit maximum protection in this model.

There is now a considerable body of evidence from genetic linkage studies, animal models, and in vitro studies that points to a critical balance between activating FcγRs and the inhibitory FcγRIIB receptor in determining human susceptibility to infection and autoimmune disease (50). These receptors are found on several cells that regulate the activity of both the innate and the adaptive arms of our immune system, and the precise contribution that any individual receptor-cell combination makes in a particular pathological context is likely to be both subtle and complex. Nevertheless, there is mounting evidence that when resident ICs are not cleared effectively, the FcγRs on neutrophils play a disproportionately “negative” role in the cycles of damage and inflammation that ensue. Our data suggest that the PI3Kβ isoform plays a major role in determining the sensitivity of neutrophil activation by ICs in enabling integration between FcγRs and GPCRs for proinflammatory mediators found in the local milieu. Loss of PI3Kβ did not markedly affect humoral antibody responses or the ability of neutrophils to ingest and kill complement-opsonized bacteria. Class I PI3Ks are highly “druggable” targets, and there exists the potential to generate both selective and combined inhibitors. Therefore, we suggest that inhibition of PI3Kβ, possibly in combination with that of PI3Kδ, might offer an opportunity to block a major step in IC-elicited tissue damage without adversely affecting adaptive or innate immune functions that are normally required for managing infection.

Materials and Methods

Reagents

TGX221 was purchased from Cayman Chemicals; IC87114 was a gift from P. Depledge (UK); AS605240, AA861, and MK886 were from Enzo Life Sciences; and CP105,696 was a gift from Pfizer. Antibodies against p110β (N-terminal), pAkt (Ser473), pERK1/2, and total Akt were from Cell Signaling Technology; antibodies against p110α, p110β (C-terminal), and p110δ were obtained from Santa Cruz Biotechnology; antibody against p110γ was from Onyx; antibody against pan-p85 was from Upstate Biotechnology; antibody against FcγRII/FcγRIII (detects both receptors) was from eBioscience; monoclonal antibody against FcγRIV (9E9) was a gift from J. Ravetch (Rockefeller University, USA); antibody against β-COP was a gift from N. Ktistakis (Babraham Institute, UK); antiserum against JNK was a gift from S. Cook (Babraham Institute, UK); antibody against rat IgG F(ab′)2 was from Jackson Immunoresearch; and antibodies against horseradish peroxidase (HRP)–conjugated phosphotyrosine and the FcεRI γ chain were from Millipore. Recombinant murine type VII collagen NC1 fragment (mcVII) was generated as described previously (51, 52). Serum samples were obtained from patients with EBA before the initiation of treatment under the approval of the Ethics Committee of the Medical Faculty of the University of Freiburg, Freiburg, Germany (Institutional Board Projects 318/07 and 407/08). We generated hcVII as previously described (53); however, amino acid residues 1528 to 2602 were removed from the protein to provide increased stability. All other reagents, including endotoxin-free or low-endotoxin buffer components, were from Sigma or as detailed elsewhere (34, 39, 54).

Mice

The Pik3cdD910A/D910A (δKD) (55) and Pik3cg−/− (γKO) (56) mouse strains have been previously described. The mouse strain carrying the p110β gene in which exons 21 and 22 are flanked by loxP sites (p110βflox/flox) has also been previously described (8). Cre-mediated deletion of exons 21 and 22 are theoretically predicted to generate an in-frame 112-kD splice variant of p110β (although we found no evidence for the expression of this variant; see fig. S1). To obtain mice with a heritable deficiency in p110β, we crossed p110βflox/flox mice with a Cre deleter strain to generate Cre+/+;p110βΔ21-22/Δ21-22 mice. Further crossing was performed to obtain Cre-negative mice that were homozygous for the Cre-recombined allele of p110β (βKO). The Pik3cbD931A/D931A (βKD) strain, wherein the DFG site in p110β is point mutated to AFG to obtain a kinase-deficient mutant, was generated by B. Vanhaesebroeck et al. (Centre for Cell Signalling, Institute of Cancer, Queen Mary University of London, UK) (8). βKO/δKD double-mutant mice, which were generated by crossing βKO and δKD mice, showed birth ratios, fertility, and growth characteristics similar to those of the βKO strain. βKO, βKD, and βKO/δKD mice yielded BMNs in normal number and morphology (table S1). All strains, except the CD2-Cre-p110βΔ/Δ (see Supplementary Materials and Methods) and the corresponding wild-type animals that were on a C57BL/6J background, were on a mixed (C57BL/6J and 129Sv) background and were housed in the small-animal barrier unit, with the δKD, γKO, and βKO/δKD strains housed under specific pathogen-free conditions at the Babraham Institute, under Home Office Project Licence PPL 80/2335. All animals were routinely genotyped by polymerase chain reaction (PCR) assay with primers that confirmed targeted mutations. In all experiments, p110 mutant mice were compared with appropriate strain-matched wild-type controls. Radiation chimeras for βKO, δKD, and βKO/δKD and their associated wild-type counterparts were created in C57BL/6J mouse recipients with established protocols under the approval of the appropriate ethics committees at the facilities in which they were generated. All radiation chimeric mice were confirmed to exhibit 92 to 97% reconstitution with donor-derived cells.

Isolation of neutrophils

Human neutrophils and murine BMNs were isolated as described previously (34). Use of human blood was approved by the Cambridge Research Ethics Committee (06/Q0108/165). Where indicated, BMNs were pretreated with vehicle [dimethyl sulfoxide (DMSO) or ethanol; 0.1% maximum final concentration] or the indicated inhibitors for 10 min at 37°C.

Immobilization of adhesive proteins and ICs

Fibrinogen (150 μg/ml) and poly-RGD (20 μg/ml) were adsorbed onto 96-well plates overnight at 4°C (for fibrinogen) or at room temperature for ≥2 hours (for poly-RGD). ICs were immobilized by coating the plates overnight with phosphate-buffered saline (PBS) containing BSA (100 μg/ml), mcVII (10 μg/ml), or hcVII (10 μg per well) at 4°C. Wells were blocked with either 1% fat-free milk (for BSA) or 100% heat-inactivated fetal calf serum (FCS) (for mcVII and hcVII) for 45 min at room temperature. ICs were formed on BSA, mcVII, and hcVII through the addition of the indicated dilutions of antibody against BSA (mouse antibody for experiments with mouse BMNs and rabbit antibody for experiments with human neutrophils), antibody against mcVII (20 μg/ml), or the indicated dilutions of sera from normal donors or EBA patients, for 1 hour at room temperature.

Measurement of ROS

Chemiluminescent detection of ROS production was performed as previously described with unprimed BMNs, except for assays with S. aureus, wherein cells were primed (34, 54). BMNs were added to 96-well plates precoated with adhesion proteins or ICs or that contained fMLP (10 μM), LTB4 (10 or 100 nM), PMA (300 nM), TNF-α (20 ng/ml), or S. aureus (1:20 ratio of BMN/S. aureus) and assayed for ROS production. Data are expressed as the accumulated light emission [relative light units (RLUs)] over time, or in some cases as the rate kinetics of ROS production in relative light units per second (RLU/s). In other experiments, data are expressed as the percentage of peak RLU/s or total RLUs observed in vehicle-treated wild-type cells.

In vitro neutrophil killing of S. aureus

Primed wild-type or βKO BMNs (6.2 × 106) were pretreated with DMSO or diphenyleneiodonium (DPI, 3 μM) for 15 min before being mixed with 3.5 × 106 serum-opsonized S. aureus. In vitro killing of the bacteria was then assessed as described previously (54).

Total cellular tyrosine and Fc γ chain phosphorylation

Total phosphotyrosine and Fc γ chain phosphorylation were assessed as described previously, with minor modifications (36, 57). FcγRs (first bound by antibodies against FcγRIII/II) were cross-linked with goat antibody (100 μg/ml) against rat IgG F(ab′)2 for 30 s.

Measurement of Akt and ERK phosphorylation in BMNs

BMNs (5 × 106/ml) were applied to ICs for 20 min at 37°C, and lysates were generated as described previously, with minor modifications (58). Lysates were analyzed by Western blotting for the presence of pAkt (Ser473), pERK1/2, or β-COP (as a loading control). Quantitation of pAkt was performed with AIDA software, and the data are expressed as the percentage of pAkt in the indicated control condition. In experiments examining Akt phosphorylation in the presence of AA861 and that were reconstituted with LTB4, cells in the supernatant at the end of the adhesion period were discarded and only those cells that were firmly adhered to the IC matrix were lysed and processed.

In-gel gelatin zymography

Assessment of released gelatinase activity was performed as described by Jakus et al. (58) for 30 min at 37°C on ICs generated with a 1:2000 dilution of anti-BSA antibody.

In vivo models of autoimmune disease

The in vivo model of EBA was performed as described (52). Blisters and erosions in skin over various parts of the body were counted, and the extent of skin disease was scored by two independent investigators as follows: 0, no lesions; 1, 10 lesions or 1% of skin surface; 2, 10 lesions or 1 to 5% of the skin surface; 3, 5 to 10%; 4, 10 to 20%; and 5, 20% involvement of the skin surface. Biopsies of lesional and perilesional skin were obtained 2 days after the last injection of IgG and were prepared for examination by histopathology and immunofluorescence microscopy as described previously (51). Inflammatory cell infiltration in skin sections was visually scored as follows: 0, no infiltration; 1, low-level infiltration; 2, moderate infiltration; and 3, severe infiltration. In some experiments, βKO mice were injected with 15 mg of rabbit anti-mcVII IgG or normal rabbit IgG (both 15 mg) every second day (three injections). On day 6 after the first antibody injection, 4 × 106 isolated wild-type BMNs were injected intradermally into the ears of βKO mice, and we assessed blistering 36 hours after injection. The K/BxN model was performed as described (59) with wild-type, βKO, δKD, or βKO/δKD mice and 150 or 400 μl of control or arthritogenic K/BxN sera. Given the categorical nature of the clinical scores in the EBA and K/BxN experiments, data were arbitrarily binned into three indices of disease severity, and the mean percentage of mice falling within each severity index was calculated. Data are thus displayed by “geyser” plots. Data for ankle thickness in the K/BxN model, being continuous in nature, are presented as mean ± 95% confidence interval (CI) with classic line graphs.

Statistical analysis

Two main statistical approaches were used in this study. When the data met the assumptions for parametric tests, t tests and analysis of variance (ANOVA) followed by post hoc tests were applied to the data. When there was every reason to expect heterogeneous random error, a logarithmic transformation of the data was carried out before analysis to stabilize the errors. When the assumptions were not met, Kruskal-Wallis and Mann-Whitney tests were applied, with Bonferroni correction in case of multiple comparisons in the same data set.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/4/168/ra23/DC1

Materials and Methods

Fig. S1. Analysis of p110 and p85 protein abundance in p110 mutant mice.

Fig. S2. βKD BMNs are defective in their ability to produce ROS in response to ICs.

Fig. S3. The PI3Kδ inhibitor, IC87114, and the PI3Kγ inhibitor, AS605240, exhibit off-target activities in IC-mediated mouse BMN responses.

Fig. S4. Inhibition of p110β does not affect the surface expression of FcγR or β2 integrin on resting or stimulated neutrophils.

Fig. S5. Lack of p110β activity does not markedly affect BMN adhesion or spreading on ICs.

Fig. S6. PI3Kβ, but not PI3Kδ or PI3Kγ, activity is required for ROS production by BMNs in response to adhesion to integrin ligands.

Fig. S7. Normal rabbit anti-mcVII-IgG and C3 deposition in mouse skin, and serum IgG amounts in wild-type and βKO mice.

Fig. S8. Skin blistering in βKO mouse skin upon local reconstitution with wild-type mouse neutrophils.

Fig. S9. PI3Kβ is important in promoting ROS formation in response to autoimmune sera from EBA patients.

Fig. S10. Immunoreactivity of normal human or EBA patient sera toward recombinant human collagen VII (hcVII) as assessed by enzyme-linked immunosorbent assay (ELISA).

Fig. S11. Combined deficiency of p110β and p110δ activities confers substantial protection from arthritic symptoms in the K/BxN model of autoimmune arthritis.

Fig. S12. Combined deficiency of p110β and p110δ activities confers substantial protection from joint swelling in the K/BxN model of autoimmune arthritis.

Fig. S13. Normal Akt phosphorylation and antibody production in mice with lymphocytic deletion of p110β.

Fig. S14. LTB4-dependent ROS production and Akt phosphorylation in BMNs.

Fig. S15. PI3Kβ is cooperatively engaged by FcγRs and the GPCR, BLT1, to drive sustained neutrophil responses.

Table S1. Hematological analysis of βKO mouse peripheral blood.

Table S2. Isoforms of p110 identified in p85 immunoprecipitates by mass spectrometry.

References

References and Notes

  1. Acknowledgments: We thank members of the Small Animal Breeding Unit at the Babraham Institute for animal husbandry and J. Webster for help with mass spectrometry experiments. We also thank K. Boyle, H. Welch, S. Vermeren, F. Colucci, and M. Turner for critical review of the manuscript and J. Swanson, M. Peters-Golden, and L. Gambardella for helpful discussions. Funding: This study was supported by grants from the Biotechnology and Biological Sciences Research Council (BB/DO13593/1 and BBS/B/01979), British Lung Foundation, Wellcome Trust (WT085889MA), and the CJ Martin Fellowship awarded to S.K. by the National Heart Foundation (Australia) and the National Health and Medical Research Council (Australia), and funding awarded to C.S. and D.Z. by the Dr. Robert Pfleger-Stiftung (15.11.06) and Excellence Initiative of the German Federal and State Governments (EXC 294). Z.J. and A.M. were funded by the European Research Council (Starting Independent Investigator grant 206283). Z.J. was also funded by a Bolyai Research Fellowship from the Hungarian Academy of Sciences. R.L. and M.H. were funded by DFG EXC 306/1 and DFG LU 877/5-1. The gene-targeted mice used in this study were funded by the Ludwig Institute for Cancer Research, USA. Author contributions: S.K. designed, performed, and analyzed the experiments and helped to prepare the manuscript; C.S. designed and performed the EBA experiments; Z.J. designed, performed, and analyzed all of the K/BxN experiments; K.E.A., G.D., K.D., M.H., D.O., J.J., T.A.M.C., F.R., and H.G. performed the experiments; A.F. generated hcVII; G.E.J. provided FcR γ chain knockout mice; A.S.-P. performed statistical analysis; K.O. supervised and designed the lymphocyte experiments; R.L. and D.Z. supervised and designed the EBA experiments; A.M. supervised and designed the K/BxN experiments; B.V. supervised and designed the generation of PI3Kβ transgenic mice; and L.R.S. and P.T.H. were responsible for overall supervision of the project, designed the experiments, and prepared the manuscript. Competing interests: Use of the p110β and p110δ mice requires a material transfer agreement (MTA) from the Ludwig Institute for Cancer Research. Use of the KRN transgenic mice requires an MTA from Semmelweis University and IGBMC (Institut de Génétique et de Biologie Moléculaire et Cellulaire), France. B.V. is a consultant and Science Advisory Board member of Intellikine, San Diego, CA. K.O. is a paid consultant for GlaxoSmithKline.
View Abstract

Navigate This Article