Gβγ is a direct regulator of endogenous p101/p110γ and p84/p110γ PI3Kγ complexes in mouse neutrophils

Abstract

The PI3Kγ isoform is activated by Gi-coupled GPCRs in myeloid cells, but the extent to which the two endogenous complexes of PI3Kγ, p101/p110γ and p84/p110γ, receive direct regulation through Gβγ or indirect regulation through RAS and the sufficiency of those inputs is controversial or unclear. We generated mice with point mutations that prevent Gβγ binding to p110γ (RK552DD) or to p101 (VVKR777AAAA) and investigated the effects of these mutations in primary neutrophils and in mouse models of neutrophilic inflammation. Loss of Gβγ binding to p110γ substantially reduced the activation of both p101/p110γ and p84/p110γ in neutrophils by various GPCR agonists. Loss of Gβγ binding to p101 caused more variable effects, depending on both the agonist and cellular response, with the biggest reductions seen in PIP₃ production by primary neutrophils in response to LTB4 and MIP-2 and in the migration of neutrophils during thioglycolate-induced peritonitis or MIP2-induced ear pouch inflammation. We also observed that p101^VVKR777AAAA neutrophils showed enhanced p84-dependent ROS responses to fMLP and C5a, suggesting that competition may exist between p101/p110γ and p84/p110γ for Gβγ subunits downstream of GPCR activation. GPCRs did not activate p110γ in neutrophils from mice lacking both the p101 and p84 regulatory subunits, indicating that RAS binding to p110γ is insufficient to support GPCR activation in this cell type. These findings define a direct role for Gβγ subunits in activating both of the endogenous PI3Kγ complexes and indicate that the regulatory PI3Kγ subunit biases activation toward different GPCRs.

Fig. 1. Schematic diagram of mutations created in PI3Kγ subunits and their effect on expression levels in isolated bone marrow neutrophils (BMN).

(A) Cartoon illustrating Gβγ binding to p84-p110γ and p101-p110γ complexes and the mutations created in the Gβγ binding domains of p101 regulatory and p110γ catalytic subunits. A single Gβγ binding site is depicted in p101-p110γ but it is currently unknown whether there is a single or multiple binding sites for Gβγ in this complex. (B to E) Relative abundance of PI3Kγ subunits in BMN derived from the indicated strains of mice; (B) Western blot representative of five independent experiments that were combined and quantified for the abundance of p84 (C), p110γ (D) and p101 (E). Western blotting analysis was performed on 1x10⁶ BMNs from mice of the indicated genotypes. The abundances of p84, p110γ and p101 protein in samples from mice of each genotype were quantitated by densitometry and normalized to p67^phox to correct for neutrophil input (β-COP is shown as an additional loading control). Data are means ± SEM of n= 5 independent experiments and are expressed as a ratio to the relative protein band density of WT samples on each blot. Normalized data was analyzed by ratio paired t tests with Holms-Sidak corrections applied to account for multiple comparisons, with comparisons made to WT samples.

Fig. 2. GPCR-driven signaling responses in BMNs and effects of defects in Gβγ binding or knockout of PI3Kγ subunits.

(A and B) BMNs (1 × 10⁶ cells) from p101/p84-dKO (dashed-purple), p110γ-KO (dashed-orange), p110γ/101-dGβγ (solid-brown), p110γ-Gβγ (solid-blue), p101-Gβγ (solid-red), and their relevant WT controls (solid-black) were stimulated at 37°C with 10μM fMLP for the indicated times. (C-E) BMNs from WT (black) and p101-Gβγ (red) mice were stimulated with indicated GPCR agonists (or vehicle controls) for 10 s (C), with 10 μM surfen (or vehicle control) for 10 min (D), or 3 nM MIP-2 for the indicated times (E). Reactions were quenched, lipids were extracted, and the amounts of C38:4-PIP₃ and C38:4-PIP₂ were quantitated by mass spectrometry. Data are means ± SEM of at least three experiments and are expressed as the ratio of the abundance of C38:4-PIP₃ to that of C38:4-PIP₂ in the same sample (C38:4 PIP₃/C38:4 PIP₂) to account for any variations in cell input, expressed as a percentage of WT response at 10 s (A and E) or WT 10 μM fMLP response (C), or WT surfen response (D). (B) Areas under the curve (AUC) for fMLP time courses shown in (A) were calculated for each genotype response and compared to its relevant WT control in the same experiment using ratio paired t tests. A one-way ANOVA on log-transformed data was used for the comparisons between genotypes. Holm-Sidak correction was applied to account for multiple comparisons. Similar analysis was performed for the MIP-2 time course (E). Non-normalized data for each genotype and their relevant WT controls from (A) are shown in fig. S4. A 2-way ANOVA on log transformed data followed by Sidak’s multiple comparison test was used in comparing PIP₃ responses to multiple agonists (C) whereas a t test was used to compare surfen PIP₃ data (D). (F-H) BMNs (0.5 × 10⁶ cells) from WT (black), p101-Gβγ (red), p110γ-Gβγ (blue), and p110γ/101-dGβγ (brown) mice were stimulated for 1 min at 37°C with 10 μM fMLP. Cells were analysed for DAG (G), or lysates run on SDS-PAGE and Western blot analysis performed for phospho-PKB or ERK (F and H). Abundances of phospho-PKB and phospho-ERK protein in samples from mice of each genotype were quantitated by densitometry and normalized to total p67^phox to correct for neutrophil input protein. A representative Western blot is shown in fig. S5. Data are means ± SEM of n≥4 independent experiments and are expressed as the ratio of abundance of C38:4-DAG to that of C38:4-PI in the same sample to account for variations in cell input (G), or as a percentage of protein band density of WT fMLP samples on each blot (F, H). Data was analyzed by 2-way ANOVA with Holm-Sidak correction applied to account for multiple comparisons (G) or ratio paired t tests with Holms-Sidak corrections applied to account for multiple comparisons, with comparisons made to WT samples (F and H).

Fig. 3. Characterization of GPCR-driven superoxide responses in BMN and effects of defects in Gβγ binding or knockout of PI3Kγ subunits.

(A-F). BMNs (0.5 × 10⁶) from p110γ/101-dGβγ, p101-Gβγ, p110γ-Gβγ, or p101/p84-dKO mice, and their relevant WT controls, were preincubated with horseradish peroxidase (HRP) and luminol before being added to a 96-well plate. Agonists were either injected (A and D to F) or manually added to wells (B and C). ROS generation was then measured in duplicate for each genotype by chemiluminescence and recorded with a Berthold MicroLumat Plus luminometer. Data are means ± SEM for accumulated light emission (RLU) per second (A and C to F, left hand panels) for at least three experiments (except for p101/p84-dKO where n=3 experiments for PMA stimulation and n=2 experiments for fMLP stimulation (B and C)); data were normalized to the WT maximal response for each experiment (C to F, left hand panels) or accumulated ROS over the recording times, expressed as a percentage of accumulated ROS in WT cells (B to F right hand panels). Non-normalized superoxide responses from each genotype were compared to their relevant WT control in the same experiment using ratio paired t tests. A one-way ANOVA on log-transformed data was used for the comparisons between genotypes. Holm-Sidak correction was applied to account for multiple comparisons. Non-normalized data for each genotype and relevant WT control (C) are found in fig. S5.

Fig. 4. Effect of Gβγ binding to PI3Kγ subunits on BMN migration in vitro and in in vivo models of inflammation.

(A). Centre-zeroed tracks of individual WT, p101-Gβγ, p110γ-Gβγ, p110γ/101-dGβγ, p101/p84-dKO or p110γ-KO BMNs in an EZ-Taxiscan Chamber migrating towards reservoirs located at the top of the diagrams containing 3 μM fMLP. Isolated cells were added to chambers containing fibrinogen-coated glass coverslips. Shown are combined data from at least 4 experiments with at least 9 tracks per experiment. (B). Analysis of EZ-Taxiscan migratory tracks showing relative percentage of migratory cells and migration speed normalized to WT values (WT migration 31.9 ± 2.8% of cells, WT velocity 11.4 ± 0.44 μm/min), distance travelled, straight line distance and migratory index. Thirty five to 250 cells were analyzed for each genotype in each experiment. Data are mean ± SEM of average track data from each experiment. Data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparison test, comparing each genotype against WT data. (C) Mice of the indicated genotypes were injected with 3% thioglycolate (I.P.) or left uninjected. 3.5hours later the peritoneum was flushed and cells collected for FACS analysis gating for CD45+ cells. Quantitation of other cell types in the peritoneum flush in the absence or presence of thioglycolate-induced peritonitis are shown in fig. S6. (D) For ear-pouch inflammation models, anesthetised mice of the indicated genotypes were injected with either 1.5 μM MIP-2 or vehicle intradermally in the pinna (until a blister about 4 to 5 mm in size formed) and mice were sacrificed 4 hours later. The pinna was stained with Ly6G, and the extent of Ly6G staining was evaluated by microscopy with at least 10 images taken from each pinna. In (C) and (D), BMN accumulation from p101-Gβγ, p110γ-Gβγ, or p110γ/101-dGβγ mice, and their relevant WT controls was quantitated using FACS analysis and data shown are mean ± SE from at least three mice per genotype group. Data were analyzed by one-way ANOVA followed by Dunnett’s multiple comparisons test, comparing each genotype to WT data.