Molecular basis for differential activation of p101 and p84 complexes of PI3Kγ by Ras and GPCRs

Abstract

Class IB phosphoinositide 3-kinase (PI3Kγ) is activated in immune cells and can form two distinct complexes (p110γ-p84 and p110γ-p101), which are differentially activated by G protein-coupled receptors (GPCRs) and Ras. Using a combination of X-ray crystallography, hydrogen deuterium exchange mass spectrometry (HDX-MS), electron microscopy, molecular modeling, single-molecule imaging, and activity assays, we identify molecular differences between p110γ-p84 and p110γ-p101 that explain their differential membrane recruitment and activation by Ras and GPCRs. The p110γ-p84 complex is dynamic compared with p110γ-p101. While p110γ-p101 is robustly recruited by Gβγ subunits, p110γ-p84 is weakly recruited to membranes by Gβγ subunits alone and requires recruitment by Ras to allow for Gβγ activation. We mapped two distinct Gβγ interfaces on p101 and the p110γ helical domain, with differences in the C-terminal domain of p84 and p101 conferring sensitivity of p110γ-p101 to Gβγ activation. Overall, our work provides key insight into the molecular basis for how PI3Kγ complexes are activated.

Keywords: CP: Cell biology; GPCR; HDX-MS; PI3K; PIK3CG; PIK3R5; PIK3R6; TIRF; p101; p84; phosphoinositide 3-kinase.

Graphical abstract

Figure 1

The structure of the p110γ-p84 complex and comparison with p110γ-p101 (A) Cartoon schematic of the PI3Kγ catalytic (p110γ) and regulatory subunits (p101 and p84) with domain boundaries indicated. (B) Cartoon of differences in activation between p110γ-p84 and p110γ-p101 complexes downstream of GPCRs and RTKs. (C) Structure of the p110γ-p84 complex based on X-ray crystallography, negative-stain EM, and AlphaFold modeling (PDB: 8AJ8, supporting data in Figures S1 and S2). Domains are indicated from (A), with a cartoon schematic shown in the bottom left. (D) Structure of the p110γ-p101 complex (PDB: 7MEZ). Domains are indicated from (A), with a cartoon schematic shown in the bottom left. (E and F) Differences in the C-terminal domain of p84 (E) and p101 (F) are shown with this domain shown as an electrostatic surface.

Figure 2

Differences in the p110γ interface in p84 versus p101 (A) HDX-MS differences in the p110γ subunit between the p110γ-p101 and p110γ-p84 complexes. Significant differences in deuterium exchange (defined as greater than 5%, 0.4 Da, and a two-tailed t test p < 0.01 at any time point) are mapped on to the structure of p110γ-p84 and cartoon of p110γ according to the legend. (B) Sum of the number of deuteron differences between the p110γ-p101 and p110γ-p84 complexes over the entire deuterium exchange time course. Positive difference is indicative of enhanced exchange in p110γ-p84. Each point is representative of the center residue of an individual peptide. Peptides that met the significance criteria described in (C) are colored red. Error is shown as the sum of the standard deviation across all time points (n = 3 for each time point). All HDX-MS data are provided in the source data. (C) Selected deuterium exchange at 30 s for peptides in p110γ for p110γ-p101 and p110γ-p84 complexes at either a high concentration (1,500 nM) or a low concentration (175 nM). Error is shown as standard deviation (n = 3) with two-tailed p values as indicated: ^∗∗p < 0.01; not significant (ns) > 0.05. Full HDX-MS data for all peptides in this experiment are shown in the source data. (D) Cartoon schematic of the p110γ interface for p101 (top) and p84 (bottom), with a zoom in on the residue’s located at the p110γ interface for both p84 and p101. Dotted lines indicate cation-pi or electrostatic interactions. (E) Sequence alignment of both p101 and p84 residues in the α3 to α6 helices located at the p110γ interface. The residues annotated in panel are indicated on the alignment. A full alignment of p101 and p84 is shown in Figure S3.

Figure 3

Activation of p110γ-p84 and p110γ-p101 by lipidated HRas and Gβγ (A) Cartoon schematic describing PI3Kγ variants tested and the lipidated activators, GTPγS-loaded HRas and Gβγ. (B) Lipid kinase activity assays of different p110γ complexes (concentration, 100–2,000 nM) with and without lipidated Gβγ (1.5 μM) and lipidated HRas (1.5 μM) using 5% phosphatidylinositol 4,5-bisphosphate (PIP₂) vesicles mimicking the plasma membrane (20% phosphatidylserine, 50% phosphatidylethanolamine, 10% cholesterol, 10% phosphatidylcholine, 5% sphingomyelin). Error bars represent standard deviation (n = 3). Two-tailed p values represented by the symbols are as follows: ^∗∗∗∗p < 0.0001, ^∗∗∗p < 0.001; ^∗∗p < 0.01; ^∗p < 0.05; ns > 0.05. (C) Lipid kinase activity assays of p110γ-p84 and p110γ-p101 with varying concentrations of lipidated HRas (n = 3). (D) Lipid kinase activity assays of p110γ-p84 and p110γ-p101 in the presence of lipidated Gβγ (1.5 μM) with varying concentrations of lipidated HRas (n = 3). Experiments in (C) and (D) were performed using the same vesicles as in (B). The dotted red line in the graph for the p110γ-p101 complex shows the peak activity for p110γ-p84 with both activators. The EC₅₀ and 95 confidence intervals (CIs) are indicated for both (C) and (D).

Figure 4

HDX-MS analysis of p110γ-p84 activation by membrane-localized HRas and Gβγ and comparison with p110γ-p101 (A–C) Significant HDX-MS differences in the p110γ and p84 subunits between (A) plasma membrane mimic vesicles and plasma membrane mimic vesicles with 3 μM GTPγS-loaded lipidated HRas, (B) plasma membrane mimic vesicles and plasma membrane mimic vesicles with 3 μM Gβγ, and (C) plasma membrane mimic vesicles with 3 μM GTPγS-loaded lipidated HRas and plasma membrane mimic vesicles with both HRas and Gβγ (3 μM) are mapped on the structure of p110γ-p84 according to the legend in (A). A cartoon model is shown to the right with differences annotated. The sum of the number of deuteron difference is shown for p110γ, with red dots representing peptides showing statistically significant differences. Error is shown as the sum of the standard deviations across all time points (n = 3 for each time point). (D) Significant HDX-MS differences in the p110γ and p101 subunits between plasma membrane mimic vesicles and plasma membrane mimic vesicles with Gβγ mapped on the structure of p110γ-p101 (PDB: 7MEZ) according to the legend. A cartoon model is shown to the right with differences annotated. The HDX-MS data in (D) are from our previous study and are provided as a comparison with (B).

Figure 5

p110γ-p84 and p110γ-p101 exhibit distinct membrane-binding dynamics in the presence of HRas and Gβγ (A) Experimental setup for visualizing DY647-p101-p110γ and DY647-p84-p110γ interactions with a supported lipid bilayer containing membrane-anchored HRas(GDP or GTP) and farnesylated Gβγ. Experiments show a representative experiment, with all quantification generated from 2 to 4 technical replicates (see Table S3). (B and C) Kinetics of PI3K complex membrane recruitment measured in the presence of 10 nM DY647-p101-p110γ or DY647-p84-p110γ using TIRF-M. (D) Representative smTIRF-M images visualizing PI3K complex localization in the presence of 10 pM DY647-p101-p110γ or 100 pM DY647-p84-p110γ. Localization was measured on SLBs containing Ras(GDP), Ras(GTP), Gβγ, or Ras(GTP)/Gβγ. (E and F) Single-molecule dwell time distributions for DY647-p101-p110γ or DY647-p84-p110γ measured in the presence of the indicated binding partners. Plots showing log₁₀(1-cumulative distribution frequency [CDF]) versus dwell time (s). Data are fit to either a single or double exponential decay curve (black dashed lines). Single-molecule imaging of DY647-p84-p110γ yielded the following mean dwell times: 39 (+RasGTP), 74 (+Gβγ), and 188 ms (+RasGTP/Gβγ). Single-molecule imaging of DY647-p101-p110γ yielded the following mean dwell times: 0.146 (+RasGTP), 0.73 (+Gβγ), and 3.09 s (+RasGTP/Gβγ). (G and H) Step size (or displacement) distributions of DY647-p101-p110γ or DY647-p84-p110γ. PI3K complex formation with Ras(GTP), Gβγ, or Ras(GTP)/Gβγ modulates the single-molecule displacement (i.e., step size, μm). Dashed black line represents the curve fit used to calculate the diffusion coefficient (see STAR Methods). The membrane composition for membranes in (B)–(H) was 96% DOPC, 2% PI(4,5)P₂, 2% MCC-PE. See Table S3 for time constants (τ₁ and τ₂), diffusion coefficients (μm²/s), and statistics (n = 2–4).

Figure 6

Model of Gβγ activation of PI3Kγ complexes (A) Model of the activation of p110γ-p101 complex by two different Gβγ subunits. The location of the Gβγ subunits bound to the C-terminal domain of p101 (Figure S6) and the helical domain of p110γ (Figure S7) was based on AlphaFold2-multimer modeling aligned to the structure of the p110γ-p101 complex (PDB: 7MEZ). The domains of p110γ-p101 are annotated, with the Gβ subunit shown as a transparent surface, and the Gγ subunit shown as cartoon, with the C terminus colored in blue. Both Gβγ subunits are positioned in an orientation compatible with membrane binding of p110γ. (B) Model of the C-terminal domain of p101 bound to Gβγ (full details on AlphaFold2-multimer modeling is Figure S6). The unique helical extension between β8-β9 in p101 is annotated. (C) Model of the helical domain of p110γ bound to Gβγ (full details on AlphaFold2-multimer modeling is Figure S7). The N-terminal helix of the helical domain in contact with Gβγ is annotated. (D) Structure of the PH domain of GPCR kinase 2 (GRK2) bound to Gβγ (PDB: 1OMW). (E) Evolutionary conservation of the Gβγ binding site in p101. Alignment of the Gβγ binding extension in p101 between p101 and p84 (top), between different orthologs of p101 (center), and between different orthologs of p84 (bottom). (F) Comparison of the C-terminal domain between p101 and p84. The evolutionarily conserved helical extension that occurs between β8 and β9 in p101 is annotated, with the Gβγ subunit from (B) shown as a transparent surface. The end and start of β8 and β9, respectively, are labeled, highlighting the corresponding loop between p84 and p101, with the loop colored red in p84. In p84, the majority of this loop was disordered in both the X-ray and AlphaFold2-multimer modeling and is indicated as a dotted line. (G) Lipid kinase activity assays of different p110γ complexes (p110γ-p101, p110γ-p101 β8-β9 loop swap mutant, and p110γ-p84) with and without lipidated Gβγ (100 nM) using 5% PIP₂ vesicles (5% PIP₂, 30% phosphatidylserine, 65% phosphatidylethanolamine). Error bars represent standard deviation (n = 3–6). Two-tailed p values represented by the symbols are as follows: ^∗∗∗p < 0.001.

Figure 7

Model of differential activation of PI3Kγ complexes by Gβγ and Ras Schematic of how Ras and Gβγ subunits can activate p110γ-p84 (top) and p110γ-p101 (bottom). Ras in the absence of Gβγ leads to membrane recruitment for both complexes but only weakly activates kinase activity. The Gβγ binding helices in the GBD of p101 and the helical domain interface with Gβγ are shown, with the helical domain hα1 highlighted in green. The C-terminal helix in the kinase domain that reorients upon membrane binding is highlighted in red in the membrane-bound complex.