Central Role for G Protein-Coupled Phosphoinositide 3-Kinase γ in Inflammation

See allHide authors and affiliations

Science  11 Feb 2000:
Vol. 287, Issue 5455, pp. 1049-1053
DOI: 10.1126/science.287.5455.1049


Phosphoinositide 3-kinase (PI3K) activity is crucial for leukocyte function, but the roles of the four receptor-activated isoforms are unclear. Mice lacking heterotrimeric guanine nucleotide-binding protein (G protein)–coupled PI3Kγ were viable and had fully differentiated neutrophils and macrophages. Chemoattractant-stimulated PI3Kγ−/− neutrophils did not produce phosphatidylinositol 3,4,5-trisphosphate, did not activate protein kinase B, and displayed impaired respiratory burst and motility. Peritoneal PI3Kγ-null macrophages showed a reduced migration toward a wide range of chemotactic stimuli and a severely defective accumulation in a septic peritonitis model. These results demonstrate that PI3Kγ is a crucial signaling molecule required for macrophage accumulation in inflammation.

Chemoattractant-mediated recruitment of leukocytes is a key step in the progress of acute and chronic inflammation. Chemokines and chemotactic peptides, such as N-formyl-Met-Leu-Phe (fMLP), C5a, and interleukin-8 (IL-8), bind to G protein–coupled receptors (1). Receptor activation induces the release of Gβγ subunits from trimeric G proteins. In phagocytic cells, this triggers a series of signaling events that culminate in directional cell movement, phagocytosis, degranulation, and superoxide generation (1). The production of phosphatidylinositol 3,4,5-trisphosphate [PtdIns (3,4,5)P3] appears to have an essential role, because inhibitors of PI3K prevent these responses (2). Leukocytes express all the four known class I PI3K isoforms (PI3Kα, β, γ, and δ), but it is presently not clear which enzyme(s) relay inflammatory signals (3).

To assess the physiologic role of the PI3Kγ isoform, we generated PI3Kγ-deficient mice by homologous recombination. The targeting vector disrupted the PI3Kγ gene by the insertion of an IRES (internal ribosomal entry site)-LacZ and a neomycin resistance cassette in the first coding exon (exon 2) (4). Embryonic stem (ES) cell clones showing heterozygous gene disruption were used to generate germ line chimeras (5).

Mice homozygous for the PI3Kγ-targeted allele were viable, fertile, and displayed a normal life-span in a conventional mouse facility. Whereas wild-type (WT) mice expressed PI3Kγ in neutrophils, macrophages, and splenocytes, homozygous mutant cells showed no expression of the protein. In the same cells, the lack of PI3Kγ did not alter the expression of other class I PI3Ks (namely α, β, and δ) (Fig. 1A) (6).

Figure 1

PI3K downstream signaling by seven–transmembrane helix receptors. (A) Failure of PI3Kγ−/− hematopoietic cells to express PI3Kγ. Murine PI3Kγ was detected in WT bone marrow–derived neutrophils (NØ), thioglycollate-elicited peritoneal macrophages (MØ), macrophage-depleted splenocytes (Spl.), and cells from the upper Percoll layer [light bone marrow fraction (6), LBM; consisting mainly of lymphocytes and monocytes] by immunoblotting with monoclonal antibodies produced against the NH2-terminal part of human PI3Kγ. PI3Kα, β, and δ were detected with antibodies described in (6). Murine hematopoietic 32D cells (32D) and human embryonic kidney (HEK) 293 cells were applied as positive and negative controls. (B) Neutrophil PI3K activity after C5a (10 nM), fMLP (1 μM), and IL-8 (100 nM) stimulation. Lipids were extracted from PMNs labeled with inorganic phosphate (32Pi) at the indicated times, deacylated, and separated by HPLC (2). To obtain the kinetic profile in the left panel, peaks as shown in the right panel (15-s stimulation) were integrated, and the ratio of PtdIns(3,4,5)P3/PtdIns(4,5)P2 was calculated as a percentage (n ≥ 3, mean ± SE). The elution times of the deacylation products of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 are indicated by (4,5) and (3,4,5), respectively. (C) PKB phosphorylation in response to heptahelical receptor agonists. Bone marrow–derived PMNs (106/ml) of the indicated genotypes were incubated for 10 min at 37°C and subsequently stimulated with 1 μMfMLP, 10 nM C5a, or 10 nM IL-8 for 30 s. Samples were probed for the presence of PI3Kγ, PKB phosphorylated on Ser-473, and total PKB. (D) PKB phosphorylation by G protein–independent signaling pathways. Human serum–opsonized zymosan (C3bi and immunoglobulin G–coated) and GM-CSF (100 ng/ml) were used to stimulate neutrophils for 15 and 5 min, respectively. C5a stimulation was done as in (C). (E) G protein–dependent intracellular calcium release. Fura-2–loaded PMNs obtained from wild-type (wt) and PI3Kγ-null mice were stimulated with IL-8 (50 nM), C5a (1 nM), platelet activating factor (PAF, 100 nM), and fMLP (100 nM).

Peripheral blood cell counts of PI3Kγ-deficient mice showed no statistically significant differences in the hematocrit or in the distribution of lymphocytes, monocytes, basophils, and eosinophils compared with those in WT animals. By contrast, PI3Kγ-null mice had about twice as many of circulating polymorphonuclear neutrophils (PMNs) as WT mice (Table 1). Microscopic examination of blood smears did not show any morphological abnormality in any leukocyte population. Fluorescence-activated cell sorter (FACS) analysis of bone marrow PMNs and resident peritoneal macrophages with antibodies to distinct cell surface markers (Gr-1 and CD11b for PMNs; CD11b, F4/80, CD80, and CD86 for macrophages) revealed matching cell distribution and expression patterns in WT and PI3Kγ-null mice, indicating that differentiation of myeloid cells is independent of PI3Kγ (7).

Table 1

Blood parameters. Data are the mean ± SE of 15 mice from each group.

View this table:

The coupling of PI3Kγ to seven–transmembrane receptor signaling was assessed in morphologically mature bone marrow PMNs. Phosphoinositides were analyzed by metabolic labeling with32P–inorganic phosphate and subsequent analysis of deacylated lipids on high-performance liquid chromatography (HPLC). After stimulation of cells with C5a, fMLP, or IL-8, PtdIns(3,4,5)P3 was produced in WT but not in PI3Kγ−/− cells (Fig. 1B). The serine-threonine protein kinase B (PKB/Akt) is a major target of PI3K (8). Whereas C5a, fMLP, and IL-8 triggered PKB phosphorylation in response to chemoattractants in WT cells, in PI3Kγ-null PMNs PKB phosphorylation did not rise above background levels (Fig. 1C). Serum–opsonized zymosan and granulocyte-macrophage colony-stimulating factor (GM-CSF), by contrast, were still capable of signaling to PKB (Fig. 1D). Therefore, protein tyrosine kinase–dependent processes successfully activate PI3Kα, β, or δ isoforms in PI3Kγ-null cells. The intact calcium release initiated byfMLP-, C5a-, platelet-activating factor (PAF), and IL-8 receptors (Fig. 1E) illustrates that PI3Kγ-independent G protein–coupled signaling pathways are not affected. PI3Kγ is thus the sole PI3K isoform coupled to fMLP, C5a, and IL-8 receptors.

PI3Kγ functions in cytoskeletal remodeling and leukocyte mobility (9). In addition, the chemotactic response to agonists of G protein–linked heptahelical receptors is correlated with PI3K-dependent activation of PKB (10). We thus examined the ability of mutant PMNs to adhere and migrate. Mutant cells showed no increase in cell adhesion on fibronectin in response to IL-8 (11). Because adhesion and cytoskeletal remodeling are prerequisites for cell motility, the chemotactic response to IL-8,fMLP, and C5a was assayed. PMNs from PI3Kγ-null mice displayed a reduction in chemotaxis in response to IL-8,fMLP, and C5a in vitro (Fig. 2A). The in vivo impact of this chemotactic defect was assessed by measurement of agonist-induced PMN infiltration into subcutaneous air pouches. IL-8 at doses of 0.3 and 1 μg caused recruitment of 60% fewer PMNs in PI3Kγ-null mice than in WT animals after 2 and 4 hours (Fig. 2B). The response to carrageenan, a pleiotropic inflammatory stimulus (12), was not altered in PI3Kγ-null mice (Fig. 2B). These data indicate that lack of PI3Kγ leads to impaired recruitment of PMNs in response to chemokines.

Figure 2

Chemotaxis and respiratory burst of PMNs. (A) Chemotaxis to IL-8 (50 nM), fMLP (100 nM), and C5a (3 nM) was measured with fluorescently labeled cells (18). The data (mean ± SE, n = 6 to 13) are expressed as the percentage of total cells that migrated. Indicated P values were calculated by ANOVA and represent comparisons of WT and PI3Kγ-null populations. (B) Migration to air pouches. IL-8 (1 and 0.3 μg/ml in sterile apyrogenic saline) and carrageenan (1% in sterile apyrogenic saline) were injected into 6-day-old air pouches (12). Mice were killed 2 or 4 hours later and the exudate collected in 1 ml of saline. Results are the mean ± SE of PMN counts from five to six different mice per group. (C to E) Respiratory burst in PMNs. (C) PMNs were preincubated in the absence or presence of lipopolysaccharide (LPS, 100 ng/ml) before stimulation with 1 μMfMLP (mean ± SE; n = 8 to 13). (D) Resting PMNs were stimulated with human serum–opsonized zymosan (mean ± SE; n = 3). (E) Resting PMNs were stimulated with 100 nM PMA (mean ± SE; n = 6). Chemiluminescence was measured according to published methods (19), and data represent integrated responses (fMLP, 3 min; opsonized zymosan, 30 min; PMA, 30 min).

Of the various neutrophil responses, the agonist-induced respiratory burst is the most sensitive to PI3K inhibitors (2). Resting neutrophils only weakly respond with a respiratory burst to seven–transmembrane receptor agonists, but can be “primed” with tumor necrosis factor–α, GM-CSF, or lipopolysaccharide (LPS) to markedly increase their response (13). Consistent with these data, WT murine bone marrow–derived PMNs responded to fMLP after a prolonged incubation with LPS, whereas PI3Kγ-null cells remained less responsive (Fig. 2C). Prolonged adhesion restored the sensitivity of PI3Kγ-null neutrophils to fMLP in a wortmannin-sensitive manner (7). Activation of the NADPH (nicotinamide adenine dinucleotide phosphate, reduced) oxidase by serum–opsonized zymosan or phorbol 12-myristate 13-acetate (PMA) was intact in PI3Kγ−/− cells (Fig. 2, D and E). These results, and the reported sensitivity of fMLP and zymosan-induced respiratory burst to wortmannin (2), suggest that fMLP triggers PtdIns(3,4,5)P3 production required for the respiratory burst exclusively by way of PI3Kγ. The zymosan signal (complement 3bi-mediated stimulation) or priming events, on the other hand, act by way of protein tyrosine kinases on p85-associated PI3Ks, thus relieving an essential requirement for PI3Kγ (e.g., for the activation of PKB; see Fig. 1D).

PI3Kγ-null macrophages, obtained from peritoneal exudate of thioglycollate-treated mice, were tested in an in vitro chemotaxis assay with various chemoattractants. First, we evaluated the chemotactic response toward endotoxin-activated mouse serum (EAMS) as a source of chemotactic complement fractions. In this assay, the chemotactic response of PI3Kγ-null cells was reduced by 60% (Fig. 3A). In contrast, peritoneal macrophages showed a similar migration toward PMA, indicating that the defect resided in receptor signaling rather than in leukocyte locomotion ability. To further characterize the migratory deficiency observed with EAMS, we stimulated PI3Kγ-deficient macrophages with G protein–coupled serpentine receptor agonists such as RANTES (regulated on activation, normal T cell expressed and secreted), macrophage inflammatory protein–5 (MIP-5), macrophage-derived chemokine (MDC), stromal cell–derived factor–1 (SDF-1), and C5a. Migration toward all of these chemotactic agents was reduced in mutant macrophages. Chemotaxis of PI3Kγ−/− macrophages was decreased in response to C5a (85% reduction), SDF-1 (85%), RANTES (80%), MDC (70%), and MIP-5 (52%) compared with WT cells (Fig. 3A). PI3Kγ−/− peritoneal macrophages also exhibited an 85% reduction in chemotaxis toward vascular endothelial growth factor (VEGF), an agonist known to bind a tyrosine kinase receptor (Fig. 3A) (14). This observation is consistent with the report that VEGF-stimulated migration can be inhibited by pertussis toxin (15) and establishes a crucial role of PI3Kγ in this G protein–mediated response. Upon stimulation with G protein–coupled receptor agonists RANTES, MIP-5, SDF-1, C5a, andfMLP, mutant and WT macrophages showed similar increases in intracellular Ca2+ release (Fig. 3B). Consistent with a specific PI3K signaling defect, PI3Kγ−/−-purified resting peritoneal macrophages showed no increase in PKB phosphorylation after C5a stimulation (Fig. 3C).

Figure 3

Role of PI3Kγ in thioglycollate-elicited peritoneal macrophages. (A) Chemotaxis was elicited by using optimal concentrations of agonists: 5% EAMS, 16 nM PMA, 100 ng/ml SDF-1, 100 ng/ml MDC, 1 μg/ml MIP-5, 100 ng/ml RANTES, 100 ng/ml C5a, and 10 ng/ml VEGF. Data represent the percentage of total cells that migrated through the filter pores (12). Results are shown as the mean ± SD of two triplicate experiments with eight mice of each genotype. (B) Increase of intracellular calcium concentration in Fura-2–loaded macrophages from WT and PI3Kγ−/− mice. Concentrations of agonists were 300 ng/ml (MIP-1α and SDF-1), 50 ng/ml (C5a), and 10 μM (fMLP). Results are representative of two independent experiments. (C) Phosphorylation of PKB after C5a stimulation. Experiments were performed as in Fig. 1C.

A model of aseptic peritonitis induced by intraperitoneal injection of thioglycollate was used to evaluate the impact of the lack of PI3Kγ on the onset of an inflammatory response in vivo. The number of thioglycollate-elicited peritoneal leukocytes was measured at various time points in mutant and control mice. There was a 36% decrease in total PI3Kγ−/− peritoneal leukocytes (n= 7; P = 0.09) at 120 hours, but no differences were present at earlier time points (4 and 48 hours). In contrast, induction of septic peritonitis by injection of Gram-positive and Gram-negative bacteria resulted in an impaired inflammatory response in PI3Kγ-deficient mice. FACS analysis and microscopic inspection of the elicited cell populations indicated that the lack of PI3Kγ affected the recruitment of both neutrophils and macrophages. The number of peritoneal PI3Kγ-null macrophages as early as 12 hours after bacteria administration was reduced by 90% compared with that in WT animals (Fig. 4A). Similar results were obtained with Gram-positive bacteria such as Staphylococcus aureus[5 × 108 colony-forming units (CFU)]. Microscopic analysis of peritoneal leukocytes revealed that PI3Kγ−/− macrophages did normally phagocytose bacteria. Because macrophage recruitment is essential to purge peritoneal infections, we tested whether PI3Kγ-deficient mice were able to clear peritoneal bacteria after administration of sublethal doses of S. aureus (16). Forty-eight hours after intraperitoneal injection of 5 × 107 CFU per mouse, bacteria persisted in the abdominal cavity of PI3Kγ−/−mice, with a concentration 10 times that of WT mice (Fig. 4B).

Figure 4

Impairment of leukocyte recruitment during septic peritonitis in PI3Kγ−/− mice. (A) Kinetics of PMN and macrophage recruitment after intraperitoneal administration of 107 CFU of E. coli [American Type Culture Collection (ATCC) 25922]. Bacteria were grown to exponential phase, and the optical density at 600 nm was used to extrapolate cell number. Cells were washed and resuspended in 200 μl of phosphate-buffered saline before injection. Three to six mice were used for each time point (mean ± SD). (B) Clearance of viable S. aureus (ATCC 25923) from the peritoneal cavity. The peritoneal cavity was washed with 5 ml of sterile saline and the number of CFU/ml was evaluated. Results represent the mean ± SD of three mice of each genotype.

Our data are consistent with a central role of PI3Kγ in linking G protein–coupled receptor signaling to PtdIns(3,4,5)P3 production, which in turn rigorously governs cell motility in macrophages and to some extent in neutrophils. The control of macrophage infiltration in chronic inflammatory diseases such as rheumatoid arthritis, pulmonary fibrosis, atherosclerosis, and autoimmune disorders is a major task of present pharmacological research. Our results indicate that PI3Kγ might be a suitable target for development of drugs that could specifically modulate phagocyte functions without generating severe side effects.

  • * To whom correspondence should be addressed. E-mail: hirsch{at} and matthiaspaul.wymann{at}

  • These authors contributed equally to this work.


Stay Connected to Science

Navigate This Article