Allosteric activation or inhibition of PI3Kγ mediated through conformational changes in the p110γ helical domain

PI3Kγ is a critical immune signaling enzyme activated downstream of diverse cell surface molecules, including Ras, PKCβ activated by the IgE receptor, and Gβγ subunits released from activated GPCRs. PI3Kγ can form two distinct complexes, with the p110γ catalytic subunit binding to either a p101 or p84 regulatory subunit, with these complexes being differentially activated by upstream stimuli. Here, using a combination of cryo electron microscopy, HDX-MS, and biochemical assays, we have identified novel roles of the helical domain of p110γ in regulating lipid kinase activity of distinct PI3Kγ complexes. We defined the molecular basis for how an allosteric inhibitory nanobody potently inhibits kinase activity through rigidifying the helical domain and regulatory motif of the kinase domain. The nanobody did not block either p110γ membrane recruitment or Ras/Gβγ binding, but instead decreased ATP turnover. We also identified that p110γ can be activated by dual PKCβ helical domain phosphorylation leading to partial unfolding of an N-terminal region of the helical domain. PKCβ phosphorylation is selective for p110γ-p84 compared to p110γ-p101, driven by differential dynamics of the helical domain of these different complexes. Nanobody binding prevented PKCβ-mediated phosphorylation. Overall, this work shows an unexpected allosteric regulatory role of the helical domain of p110γ that is distinct between p110γ-p84 and p110γ-p101 and reveals how this can be modulated by either phosphorylation or allosteric inhibitory binding partners. This opens possibilities of future allosteric inhibitor development for therapeutic intervention.


Introduction
The class I phosphoinositide 3 kinases (PI3Ks) are master regulators of myriad functions through their generation of the lipid signaling molecule phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) downstream of cell surface receptors (Burke and Williams, 2015;Rathinaswamy and Burke, 2020; Initial promise has been reported for the development of class IA p110α oncogene-specific allosteric inhibitors. However, further investigation of the molecular mechanisms underlying p110γ regulation will be required for the discovery of regions that can be targeted for allosteric inhibitor development. Here, we report critical roles of the helical domain of p110γ in both activation and inhibition of lipid kinase activity. We characterized an allosteric inhibitory nanobody (NB7) that potently inhibits p110γ activity. Cryo-EM was used to define the inhibitory interface, which is composed of the helical domain, the ABD-RBD linker, and the regulatory motif of the kinase domain of p110γ. The region that the nanobody binds to is in close spatial proximity to a previously identified PKCβ phosphorylation site (S582) in the helical domain, and oncogenic activating mutants in the regulatory motif. We fully characterized the activity and dynamics of stoichiometrically PKCβ phosphorylated p110γ, leading to the discovery of a novel additional phosphorylation site (either S594 or S595). PKCβ phosphorylation was highly selective for p110γ and p110γ-p84, with limited phosphorylation of p110γ-p101. HDX-MS analysis showed that phosphorylation of p110γ leads to unfolding of the N-terminal region of the helical domain, and increased kinase activity. The presence of the inhibitory nanobody significantly blocks PKCβ phosphorylation, while phosphorylation of p110γ prevented binding to NB7. Overall, this work provides unique insight into the critical role of the helical domain in controlling p110γ activity, and how phosphorylation and binding partners can modify this regulation. It also reveals a unique binding site located at the interface of the helical and kinase domain that can be targeted for future allosteric inhibitor design.

Molecular mechanism of nanobody inhibition of p110γ
We previously identified multiple nanobodies that inhibited the activity of p110γ. One from this group (denoted NB7 throughout the manuscript) potently inhibited the membrane-mediated activation of p110γ-p84 by both Ras and Gβγ, with HDX-MS experiments mapping the NB7 binding interface to the RBD, helical and kinase domains (Rathinaswamy et al., 2021b). We originally hypothesized that NB7 worked by sterically inhibiting Ras binding to the RBD of p110γ. To further explore the molecular mechanism of inhibition we purified all complexes of p110γ (p110γ apo, p110γ-p84, p110γ-p101) along with the NB7 nanobody. The SDS-PAGE of all proteins utilized in this study are shown in Source data 1.
To define the mechanism concerning how NB7 inhibits PI3Kγ activity we analyzed how this nanobody inhibited all class IB PI3Kγ complexes (p110γ, p110γ-p84, p110γ-p101) upon activation by lipidated Gβγ subunits. Intriguingly, we found that all three forms of p110γ were potently inhibited by NB7 ( Figure 1A). While the IC 50 measured for the three complexes was different, this is likely mainly due to the dramatic difference in protein required to measure lipid kinase activity in vitro (~300 nM for p110γ apo/p110γ-p84, and ~10 nM for p110γ-p101). This suggested that the mechanism of inhibition was not driven by a steric block of Ras association through the RBD, as previously proposed (Rathinaswamy et al., 2021b). We examined the binding of this nanobody to all complexes using biolayer interferometry (BLI). The nanobody bound equivalently tightly to all complexes, with ~2 nM potency for p110γ, p110γ-p84, and p110γ-p101 ( Figure 1B). We also tested binding of the nanobody to all class IA PI3Ks, and there was no detectable binding to p110α, p110β, and p110δ ( Figure 1C).
To further understand the mechanism by which this nanobody blocked lipid kinase activity we measured the bulk membrane recruitment dynamics of fluorescently labeled Dy647-p84-p110γ on supported lipid bilayers (SLBs) using total internal reflection fluorescence (TIRF) microscopy. We found that the nanobody had no effect on membrane recruitment of p110γ-p84 on bilayers containing membrane-tethered Ras (GTP) and Gβγ ( Figure 1D-F). Membrane binding was not affected when the nanobody was spiked into samples containing membrane-associated Dy647-p84-p110γ ( Figure 1D). Similarly, pre-incubation of Dy647-p84-p110γ with 500 nM NB7 did not perturb membrane association of the kinase when flowed over a supported membrane ( Figure 1E).
We wanted to define the molecular mechanism of how nanobody NB7 was a potent allosteric inhibitor of lipid kinase activity. We purified the complex of nanobody NB7 bound to p110γ-p84 to homogeneity by gel filtration. Using this sample, we obtained a cryo-EM reconstruction of the complex of nanobody (NB7)-bound p110γ at 3.0 Å overall resolution from 149,603 particles (  quality to perform automated and manual construction of the p110γ-NB7 complex, with unambiguous building of the interfacial contacts between NB7 and p110γ ( Figure 2-figure supplement 2). Nanobody binding did not induce any large-scale conformational changes of p110γ, as the structure of p110γ bound to NB7 was similar to the apo p110γ crystal structure or p110γ-p101 cryo-EM structure ( Figure 2-figure supplement 3). The lowest local resolution was in the ABD, with increased B-factors of the ABD in the p110γ-NB7 structure compared to p110γ-p101 (Figure 2-figure supplement 3). This is consistent with the concept that ABD flexibility plays an important role in class I PI3K regulation (Liu et al., 2022).   The interface between NB7 and p110γ was extensive, with ~1200 Å of buried surface area, with interactions of the ABD-RBD linker, N-terminus of the helical domain, and the regulatory motif at the turn between k8 and k9 (1022-1026aa). This location in the regulatory motif is where both activating oncogenic (R1021C) and inhibitory loss of function mutants have been identified (R1021P) (Takeda et al., 2019), as well as a putative inhibitory phosphorylation site (T1024) (Perino et al., 2011). The resolution was sufficient to unambiguously build the three complementarity determining region (CDR) loops of NB7 that mediate target selectivity ( Figure 2E-F). The interface is primarily hydrophobic, with only 8 hydrogen bonds, and 1 electrostatic interaction among the 33 interfacial residues of NB7. A pocket formed between the helical domain and the ABD-RBD linker forms the majority of the interface, with extensive interactions with the long CDR3. The CDR1 loop packed up against the N-terminal section of the helical domain, with the CDR2 loop forming the interface with the regulatory motif. Previous study of oncogenic mutants in the regulatory motif of p110γ showed that increased dynamics mediated by these mutants increased kinase activity, putatively by breaking the autoinhibitory tryptophan lock in k12 of the regulatory motif (Rathinaswamy et al., 2021c). Therefore, rigidifying the regulatory motif likely explains the molecular basis for how it prevents kinase activity. The nanobody interface is distinct from the predicted Gβγ interface  and the experimentally resolved Ras interface (Pacold et al., 2000), explaining why it can still be membrane recruited by these stimuli.
The interface of NB7 with p110γ is distant from both the putative membrane binding surface and the catalytic machinery of the kinase domain. To further understand how this nanobody could potently inhibit PI3Kγ activity we examined any other potential modulators of PI3Kγ activity localized in this region. There are two regulatory phosphorylation sites in the helical (Walser et al., 2013) and kinase domain (Perino et al., 2011) localized at the NB7 interface. This is intriguing as helical domain phosphorylation is activating, and kinase domain phosphorylation is inhibitory. This suggested that a critical role in the regulation of p110γ is the dynamics of this kinase-helical interface. To fully define the role of NB7 in altering the dynamics of the helical domain we needed to study other modulators of helical domain dynamics.

p110γ activation by helical domain phosphorylation
To further understand the potential role of helical domain dynamics in controlling p110γ activity we examined the role of S582 helical domain phosphorylation by the protein kinase PKCβII (encoded by the gene PRKCB2, referred to as PKCβ for simplicity throughout the manuscript) (Walser et al., 2013). S582 is located on the interior of the helical domain and would not be expected to be exposed when the N-terminal region of the helical domain is folded ( Figure 3A). To understand this better at a molecular level, we purified a catalytic fragment of PKCβ and performed protein phosphorylation reactions on p110γ apo, p110γ-p84, and p110γ-p101. We identified a phosphorylated peptide containing S582, and surprisingly, we found an additional p110γ phosphorylation site at either S594/ S595 ( Figure 3B, Figure 3-figure supplement 1). The modification at this site results in a single phosphorylation event, but due to CID MS/MS fragmentation we cannot determine which site is modified, and we will refer it as S594/S595 throughout the manuscript. The S594/S595 site is also located in the N-terminal region of the helical domain, and is even more buried than S582, and would not be expected to be exposed when this region is folded ( Figure 3A). Dose-response curves of PKCβ treatment were carried out for p110γ ( Figure 3C), p110γ-p84 ( Figure 3D), and p110γ-p101 ( Figure 3E). Both p110γ and p110γ-p84 showed similar dose-response curves for PKCβ treatment, with similar curves for S582 and S594/S595. The p110γ-p101 complex was only very weakly phosphorylated, with <100-fold lower levels compared to p110γ and p110γ-p84 ( Figure 3E). This is consistent with the helical domain in p110γ being more rigid when bound to p101, compared to either bound to p84 or p110γ alone.
To provide additional insight into the molecular mechanisms underlying p110γ phosphorylation we carried out HDX-MS experiments on p110γ and phosphorylated p110γ (90.8% phosphorylated S594/595, 92% phosphorylated S582) ( Figure 4A). The full data underlying the experiment is available in Source data 1, and data processing information is in Supplementary file 2. We have previously observed that the N-terminal region of the helical domain of apo p110γ (residues spanning 557-630aa) shows isotope profiles that are consistent with EX1 hydrogen-deutriuim (H/D) exchange kinetics (Rathinaswamy et al., 2021b;Rathinaswamy et al., 2021c;Vadas et al., 2013;Walser et al., 2013). This is indicative of cooperative unfolding of extended protein regions, with H/D exchange occurring faster than the refolding event. This region is where the PKCβ phosphorylation sites are located and may explain how the buried residues S582 and S594/S595 can be exposed to PKCβ. This is compatible with the observation that p110γ-p101 is protected from phosphorylation, as it does not show EX1 kinetics in this region, whereas both p110γ and p110γ-p84 do (Rathinaswamy et al., 2021a     When we compared phosphorylated p110γ (>90.8% as measured by mass spectrometry at both sites) to unphosphorylated p110γ we observed extensive increases in dynamics in the C2, helical domain, and kinase domain ( Figure 4A-D). The largest increases in exchange upon phosphorylation were located in the N-terminal region of the helical domain, with the peptides directly adjacent to the phosphorylation site showing almost complete deuterium incorporation at the earliest time points of exchange. This is indicative of significant disruption of the alpha helical secondary structure in this region. When we examined the exchange profiles in this region, they still underwent EX1 kinetics ( Figure 4C), however, phosphorylated p110γ was enriched in the more fully deuterated species. In addition to the regions in the helical domain, a portion of the regulatory motif of the kinase domain also showed increased deuterium exposure. This included the k9-k12 helices that surround the activation loop of p110γ. These increases in exchange were similar to those we had observed in a R1021C oncogenic activating mutant of PIK3CG (Rathinaswamy et al., 2021c).
To further explore the potential role of phosphorylation in mediating p110γ activity, we examined the kinase activity of p110γ under two conditions: basal ATP turnover, and with PIP 2 containing lipid membranes. The experiments in the absence of PIP 2 measure turnover of ATP into ADP and phosphate and is a readout of basal catalytic competency. Experiments with PIP 2 measured ATP consumed in the generation of PIP 3 , as well as in non-productive ATP turnover. The p110γ enzyme in the absence of stimulators is very weakly active toward PIP 2 substrate with only ~2-fold increased ATP turnover compared to in the absence of membranes. This is consistent with very weak membrane recruitment of p110γ complexes in the absence of lipid activators . PKCβ-mediated phosphorylation enhanced the ATPase activity of p110γ ~2-fold in both the absence and presence of membranes ( Figure 4E). This suggests that the effect of phosphorylation is to change the intrinsic catalytic efficiency of phosphorylated p110γ, with limited effect on membrane binding.

Nanobody decreases p110γ phosphorylation
As NB7 bound at the interface of the helical and kinase domains that is exposed upon PKCβ phosphorylation of p110γ, we hypothesized that the nanobody would likely alter phosphorylation. We carried out PKCβ phosphorylation of p110γ, p110γ-p101, and p110γ bound to NB7. The presence of NB7 showed a significant decrease in p110γ phosphorylation at both sites ( Figure 5A-B). We also wanted to determine whether p110γ phosphorylation reciprocally perturbed NB7 binding. BLI experiments . Nanobody NB7 blocks PKCβ phosphorylation, and phosphorylation prevents nanobody binding. (A) Extracted ion chromatograms for p110γ, p110γ-p101, and p110γ bound to NB7 are shown for the S594 or S595 phosphorylation site in p110γ. A bar graph showing the intensities of phosphorylated and non-phosphorylated p110γ peptide (593-607) for p110γ (black), p110γ with NB7 (red), and p110γ-p101 (purple) are shown to the right of the extracted ion chromatograms (n=3, right). In all experiments in panels A+B, PKCβ was present at 500 nM. Significance is indicated by ***(<0.0001%). (B) Extracted ion chromatograms for p110γ, p110γ-p101, and p110γ bound to NB7 are shown for the S582 phosphorylation site in p110γ. A bar graph showing the intensities of phosphorylated and non-phosphorylated p110γ peptide (579-592) p110γ (black), p110γ with NB7 (red), and p110γ-p101 (purple) are shown to the right of the extracted ion chromatograms (n=3, right). Significance is indicated by *(<0.01%) and ***(<0.0001%). The putative phosphorylation site is shown in red in the sequence above the bar graphs for both panels A+B. (C) Cartoon schematic of biolayer interferometry (BLI) analysis of the binding of immobilized His-NB7 to phosphorylated and non-phosphorylated p110γ. (D) Association curves for phosphorylated and non-phosphorylated p110γ (25 nM) binding to His-NB7 are shown (n=3). (E) ATPase kinase activity assays comparing the activation/ inhibition of phosphorylated and non-phosphorylated p110γ (1000 nM) with or without nanobody (3000 nM final) in the absence of PIP 2 membranes. Significance is indicated by *(<0.05%) and NS (>0.05%). showed that there was no detectable binding of NB7 to phosphorylated p110γ ( Figure 5C-D), consistent with phosphorylation disrupting the N-terminal region of the p110γ helical domain. In addition, lipid kinase assays using phosphorylated p110γ showed no detectable difference in activity when measured in the absence and presence of nanobody ( Figure 5E), consistent with NB7 being unable to bind to phosphorylated p110γ.

Discussion
Here, we find that the helical domain is a central regulator of the p110γ catalytic subunit of class IB PI3Kγ, with modulation of helical dynamics through binding partners or PTMs able to either increase or decrease lipid kinase activity. These results expand on previous work defining the helical domain as a central regulator of class IA PI3Ks, where the nSH2 domain of the p85 regulatory subunits makes inhibitory interactions that significantly inhibit lipid kinase activity of all class IA catalytic subunits (p110α, p110β, and p110δ) (Mandelker et al., 2009;Miled et al., 2007;Burke and Williams, 2013;Burke, 2018). This inhibitory interaction in class IA PI3Ks is disrupted in human cancers (helical hotspot Figure 6. Comparison of nanobody binding site compared to p85 inhibition of class IA phosphoinositide 3 kinases (PI3Ks) and class IB activation sites.
(B) Comparison of the nanobody NB7 binding site in p110γ compared to the X-ray structure of the Ras binding site (PDB: 1HE8) (Pacold et al., 2000) and the Alphafold model of Gβγ bound to p110γ . (C) Oncogenic mutations and post-translational modifications in spatial proximity to the nanobody binding site. mutations in PIK3CA) (Samuels et al., 2004) and immune disorders (helical mutations in PIK3CD in APDS1) (Angulo et al., 2013;Lucas et al., 2014). Class IB PI3Ks are uniquely compared to class IA PI3Ks, as they are not inhibited by p101 and p84 regulatory subunits, but instead potentiate GPCR activation. This lack of inhibition is due to the distinct binding interface of class IB PI3K regulatory subunits compared to class IA regulatory subunits, with only class IA regulatory subunits making direct inhibitory interactions with the kinase and helical domains of p110 catalytic subunits (Rathinaswamy et al., 2021a). Here, we show that a unique surface at the interface of the helical and kinase domains of p110γ is a potential site for the development of novel allosteric inhibitors that modulate p110γ activity.
The previously identified inhibitory nanobody (NB7) (Rathinaswamy et al., 2021b) bound with high affinity and inhibited all complexes of p110γ. The nanobody interface is distinct from how the nSH2 inhibits class IA PI3K activity, as its binding site is on the opposite face of the helical domain ( Figure 6A). The mechanism of inhibition is also distinct, as the nSH2-helical interaction plays a critical role in preventing membrane recruitment of inhibited class I PI3Ks, with removal of this interface either through pYXXM motif binding or oncogenic mutations leading to increased membrane recruitment (Burke et al., 2012;Burke et al., 2011;Zhang et al., 2011). Analysis of the nanobody binding site compared to the structure of HRas-p110γ or the HDX-MS supported Alphafold-multimer prediction of Gβγ-p110γ  shows that nanobody binding does not sterically block complex formation ( Figure 6B). This is consistent with it not blocking membrane recruitment by Ras/Gβγ. The nanobody inhibited ATP turnover both in solution and on membranes, suggesting that it prevents formation of a catalytically competent conformation of p110γ, but still allows for membrane recruitment. Further development of small molecule allosteric binding partners in this allosteric pocket between the kinase and helical domain may reveal the specific molecular interactions in this pocket that mediate inhibition.
Oncogenic mutations are frequent in the class IA PI3K encoded by PIK3CA, with this being the second most frequently mutated gene in human cancer (Lawrence et al., 2014). Mutations in p110γ encoded by PIK3CG in cancer are less frequent, however, they can still provide insight into regulatory mechanisms that control activity. Oncogenic mutations in the kinase domain (R1021C) and helical domain (E581K) are in close proximity to the nanobody binding site, and both would be expected to disrupt the stability of the helical domain or regulatory motif of the kinase domain ( Figure 6C). In addition to these mutations there are also multiple post-translational modifications that occur in this region, including inhibitory phosphorylation at T1024 (Perino et al., 2011) and activating phosphorylation at S582 (Walser et al., 2013). PKCβ is activated downstream of the IgE receptor in mast cells (Walser et al., 2013), but the full details of how this activates PI3Kγ have been unclear. We identified an additional PKCβ phosphorylation site located in the helical domain (S594/S595) ( Figure 6C). Both the S582 and S594/S595 sites are not surface accessible and would require a transient opening of the helical domain for kinase accessibility. HDX-MS analysis of the helical domain of p110γ has shown that it is more dynamic than other class I PI3K isoforms Walser et al., 2013), with the presence of the p101 regulatory subunit dramatically decreasing helical domain dynamics . This putative mechanism of helical domain dynamics driving PKCβ phosphorylation is consistent with our observation that p101 subunits decreased p110γ phosphorylation >100-fold. PKCβ phosphorylation of p110γ leads to increased dynamics in both the helical and kinase domains with increased kinase activity, although only weakly compared to full activation by either membrane localized Ras or Gβγ. This increase was observed with both membrane and soluble substrate, so likely is not driven by altered membrane recruitment.
Overall, our biophysical and biochemical analysis of modulators of helical domain dynamics reveal the critical role of this domain in regulating class IB PI3Kγ activity. This raises possibilities for development of small molecule modulators that may either increase or decrease helical domain dynamics, leading to either activation or inhibition. The high-resolution structure of an allosteric inhibitor nanobody provides initial insight into which pockets can specifically be targeted. Multiple ATP competitive p110γ selective inhibitors are in clinical trials for human cancers (Li et al., 2021), with many having significant side effects. The identification of novel inhibitory strategies provides new opportunities for targeting p110γ dysregulation in human disease.

Plasmid generation
Plasmids encoding Homo sapiens p110γ (human), Mus musculus p84 (mouse), Sus scrofa p101 (porcine), and Gβγ were used as previously described . The plasmids encoding the class IA PI3Ks were also used as previously described (Dornan et al., 2017;Siempelkamp et al., 2017). The pDONR223-PRKCB2 (PKCβII, uniprot identifier: P05771-2) was a gift from William Hahn & David Root (Addgene plasmid #23746; http://n2t.net/addgene: 23746; RRID:Ad-dgene_23746) (Johannessen et al., 2010). The PKCβII construct contains an internal TEV site that cleaves the catalytic domains from the C1/C2 regulatory domains (TEV site inserted between residues 320 and 321 of PKCβ). This construct was subcloned into a pACEBAC Sf9 expression vector for Sf9 protein production. All constructs were cloned to include a 10× histidine tag, a 2× strep tag, and a tobacco etch virus protease cleavage site on the N-terminus. For p110γ and PKCβII this tag was included at the N-terminus, with this tag included at the N-terminus of either p84 or p101 for purification of p110γ-p101 and p110γ-p84. Full details of the plasmids are included in Appendix 1-key resources table.

Virus generation and amplification
The plasmids encoding genes for insect cell expression were transformed into DH10MultiBac cells (MultiBac, Geneva Biotech) to generate bacmid containing the genes of interest. Successful generation was identified by blue-white colony screening and the bacmid was purified using a standard isopropanol-ethanol extraction method. Bacteria were grown overnight (16 hr) in 3-5 mL 2xYT (BioBasic #SD7019). Cells were spun down and the pellet was resuspended in 300 μL of 50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 100 mg/mL RNase A. The pellet was lysed by the addition of 300 μL of 1% sodium dodecyl sulfate (SDS) (wt/vol), 200 mM NaOH, and the reaction was neutralized by addition of 400 μL of 3.0 M potassium acetate, pH 5.5. Following centrifugation at 21130 RCF and 4°C (Rotor #5424R), the supernatant was mixed with 800 μL isopropanol to precipitate bacmid DNA. Following centrifugation, the pelleted bacmid DNA was washed with 500 μL 70% ethanol three times. The pellet was then air dried for 1 min and re-suspended in 50 μL Buffer EB (10 mM Tris-Cl, pH 8.5; all buffers from QIAprep Spin Miniprep Kit, QIAGEN #27104). Purified bacmid was then transfected into Sf9 cells. Two mL of Sf9 cells at 0.6×10 6 cells/mL were aliquoted into a six-well plate and allowed to attach to form a confluent layer. Transfection reactions were prepared mixing 8-12 μg of bacmid DNA in 100 μL 1×PBS and 12 μg polyethyleneimine (Polyethyleneimine 'Max' MW 40.000, Polysciences #24765, USA) in 100 μL 1×PBS and the reaction was allowed to proceed for 20-30 min before addition to an Sf9 monolayer containing well. Transfections were allowed to proceed for 5-6 days before harvesting virus containing supernatant as a P1 viral stock. Viral stocks were further amplified by adding P1 to Sf9 cells at ~2 × 10 6 cells/mL (2/100 volume ratio). This amplification was allowed to proceed for 4-5 days and resulted in a P2 stage viral stock that was used in final protein expression. Harvesting of P2 viral stocks was carried out by centrifuging cell suspensions in 50 mL Falcon tubes at 2281 RCF (Beckman GS-15). To the supernatant containing virus, 5-10% inactivated fetal bovine serum (FBS; VWR Canada #97068-085) was added and the stock was stored at 4°C.

Expression and purification of PI3Kγ, PI3Kα/β/δ, and PKCβ constructs
PI3Kγ and PKCβ constructs were expressed in Sf9 insect cells using the baculovirus expression system. Following 55 hr after infection with P2 viral stocks, cells were harvested by centrifuging at 1680 RCF (Eppendorf Centrifuge 5810R) and the pellets were snap-frozen in liquid nitrogen. The complex was purified through a combination of nickel affinity, streptavidin affinity, and size-exclusion chromatography.
Frozen insect cell pellets were resuspended in lysis buffer (20 mM Tris pH 8.0, 100 mM NaCl, 10 mM imidazole pH 8.0, 5% glycerol [vol/vol], 2 mM ME), protease inhibitor (Protease Inhibitor Cocktail Set III, Sigma) and sonicated for 2 min (15 s on, 15 s off, level 4.0, Misonix sonicator 3000). Triton X was added to the lysate to a final concentration of 0.1% and clarified by spinning at 15,000 RCF at 4°C for 45 min (Beckman Coulter JA-20 rotor). The supernatant was loaded onto a 5 mL HisTrap FF crude column (GE Healthcare) equilibrated in NiNTA A buffer (20 mM Tris pH 8.0, 100 mM NaCl, 20 mM imidazole pH 8.0, 5% [vol/vol] glycerol, 2 mM ME). The column was washed with high salt NiNTA A buffer (20 mM Tris pH 8.0, 1 M NaCl, 20 mM imidazole pH 8.0, 5% [vol/vol] glycerol, 2 mM ME), NiNTA A buffer, 6% NiNTA B buffer (20 mM Tris pH 8.0, 100 mM NaCl, 250 mM imidazole pH 8.0, 5% [vol/ vol] glycerol, 2 mM ME) and the protein was eluted with 100% NiNTA B. The eluent was loaded onto a 5 mL StrepTrap HP column (GE Healthcare) equilibrated in gel filtration buffer (20 mM Tris pH 8.5, 100 mM NaCl, 50 mM ammonium sulfate and 0.5 mM tris(2-carboxyethyl) phosphine [TCEP]). To purify PI3Kα/β/δ, the purification protocol was performed as described above but instead the protein was eluted in PI3Kα gel filtration buffer (20 mM HEPES 7.5, 150 mM NaCl, 0.5 mM TCEP). The column was washed with the corresponding gel filtration buffer and loaded with tobacco etch virus protease. After cleavage on the column overnight, the PI3K protein constructs were eluted in gel filtration buffer. The protein was concentrated in a 50,000 MWCO Amicon Concentrator (Millipore) to <1 mL and injected onto a Superdex 200 10/300 GL Increase size-exclusion column (GE Healthcare) equilibrated in gel filtration buffer. After size exclusion, the protein was concentrated, aliquoted, frozen, and stored at -80°C. For PKCβ, the protein was eluted from the strep column in gel filtration buffer, and the eluate was then loaded on a 1 mL HisTrap FF column to remove his tagged LipTev. The flowthrough was collected, and the column was washed with 2 mL of gel filtration buffer. These fractions were pooled and concentrated and stored at -80°C.
To purify phosphorylated p110γ, the purification protocol as described above was performed but PKCβ was added to the strep column at a molar ratio of 1:3 (PKCβ:p110γ) along with LipTEV, 20 mM MgCl 2 , and 1 mM ATP and allowed to incubate on ice for 4 hr. The protein was eluted by adding 7 mL of gel filtration buffer and treated with a second dose of PKCβ (same ratio as above) and allowed to incubate on ice for another 3 hr. For non-phosphorylated p110γ, same protocol was followed with the exception in the addition of PKCβ. Both the proteins were concentrated in a 50,000 MWCO Amicon Concentrator (Millipore) to <1 mL and injected onto a Superdex 200 10/300 GL Increase sizeexclusion column (GE Healthcare) equilibrated in gel filtration buffer. The final phosphorylation level of the two sites was characterized by mass spectrometry, with these values being 92% and 90.8%, for S582 and S594/S595, respectively. After size exclusion, the protein was concentrated, aliquoted, frozen, and stored at -80°C.

Expression and purification of lipidated G for kinase activity assays
Full-length, lipidated human Gβγ (Gβ1γ2) was expressed in Sf9 insect cells and purified as described previously. After 65 hr of expression, cells were harvested, and the pellets were frozen as described above. Pellets were resuspended in lysis buffer (20 mM HEPES pH 7.7, 100 mM NaCl, 10 mM βME, protease inhibitor [Protease Inhibitor Cocktail Set III, Sigma]) and sonicated for 2 min (15 s on, 15 s off, level 4.0, Misonix sonicator 3000). The lysate was spun at 500 RCF (Eppendorf Centrifuge 5810R) to remove intact cells and the supernatant was centrifuged again at 25,000 RCF for 1 hr (Beckman Coulter JA-20 rotor). The pellet was resuspended in lysis buffer and sodium cholate was added to a final concentration of 1% and stirred at 4°C for 1 hr. The membrane extract was clarified by spinning at 10,000 RCF for 30 min (Beckman Coulter JA-20 rotor). The supernatant was diluted three times with NiNTA A buffer (20 mM HEPES pH 7.7, 100 mM NaCl, 10 mM imidazole, 0.1% C12E10, 10 mM ME) and loaded onto a 5 mL HisTrap FF crude column (GE Healthcare) equilibrated in the same buffer. The column was washed with NiNTA A, 6% NiNTA B buffer (20 mM HEPES pH 7.7, 25 mM NaCl, 250 mM imidazole pH 8.0, 0.1% C12E10, 10 mM ME) and the protein was eluted with 100% NiNTA B. The eluent was loaded onto HiTrap Q HP anion exchange column equilibrated in Hep A buffer (20 mM Tris pH 8.0, 8 mM CHAPS, 2 mM dithiothreitol [DTT]). A gradient was started with Hep B buffer (20 mM Tris pH 8.0, 500 mM NaCl, 8 mM CHAPS, 2 mM DTT) and the protein was eluted in ~50% Hep B buffer. The eluent was concentrated in a 30,000 MWCO Amicon Concentrator (Millipore) to <1 mL and injected onto a Superdex 75 10/300 GL size-exclusion column (GE Healthcare) equilibrated in Gel Filtration buffer (20 mM HEPES pH 7.7, 100 mM NaCl, 10 mM CHAPS, 2 mM TCEP). Fractions containing protein were pooled, concentrated, aliquoted, frozen, and stored at -80°C.

Expression and purification of nanobody
Nanobody NB7-PIK3CG with a C-terminal 6× His tag was expressed from a pMESy4 vector in the periplasm of WK6 E. coli. A 1 L culture was grown to OD600 of 0.7 in Terrific Broth containing 0.1% glucose and 2 mM MgCl 2 in the presence of 100 μg/mL ampicillin and was induced with 0.5 mM isopropyl-β-D-thiogalactoside (IPTG). Cells were harvested the following day by centrifuging at 2500 RCF (Eppendorf Centrifuge 5810R) and the pellet was snap-frozen in liquid nitrogen. The frozen pellet was resuspended in 15 mL of TES buffer containing 200 mM Tris pH 8.0, 0.5 mM ethylenediaminetetraacetic acid (EDTA), and 500 mM sucrose and was mixed for 45 min at 4°C. To this mixture, 30 mL of TES buffer diluted four times in water was added and mixed for 45 min at 4°C to induce osmotic shock. The lysate was clarified by centrifuging at 14,000 rpm for 15 min (Beckman Coulter JA-20 rotor). Imidazole was added to the supernatant to final concentration of 10 mM loaded onto a 5 mL HisTrap FF crude column (GE Healthcare) equilibrated in NiNTA A buffer (20 mM Tris pH 8.0, 100 mM NaCl, 20 mM imidazole pH 8.0, 5% [vol/vol] glycerol, 2 mM β-mercaptoethanol [ME]). The column was washed with high salt NiNTA A buffer (20 mM Tris pH 8.0, 1 M NaCl, 20 mM imidazole pH 8.0, 5% [vol/vol] glycerol, 2 mM ME), followed by 100% NiNTA A buffer, then a 6% NiNTA B wash buffer (20 mM Tris pH 8.0, 100 mM NaCl, 250 mM imidazole pH 8.0, 5% [vol/vol] glycerol, 2 mM ME) and the protein was eluted with 100% NiNTA B. The eluent was concentrated in a 10,000 MWCO Amicon Concentrator (Millipore) to <1 mL and injected onto a Superdex 75 10/300 GL Increase sizeexclusion column (GE Healthcare) equilibrated in gel filtration buffer (20 mM Tris pH 8.5, 100 mM NaCl, 50 mM ammonium sulfate, and 0.5 mM TCEP). Following size exclusion, the protein was concentrated, frozen, and stored at -80°C.

Lipid vesicle preparation for kinase activity assays
Lipid vesicles containing 5% brain phosphatidylinositol 4,5-bisphosphate (PIP2) and 95% brain phosphatidylserine (PS) were prepared by mixing the lipids solutions in organic solvent. The solvent was evaporated in a stream of argon following which the lipid film was desiccated in a vacuum for 45 min. The lipids were resuspended in lipid buffer (20 mM HEPES pH 7.0, 100 mM NaCl, and 10% glycerol) and the solution was vortexed for 5 min followed by sonication for 15 min. The vesicles were then subjected to 10 freeze thaw cycles and extruded 11 times through a 100 nm filter (T&T Scientific: TT-002-0010). The extruded vesicles were sub-aliquoted and stored at -80°C. Final vesicle concentration was 2 mg/mL.

Kinase assays
All kinase assays were done using Transcreener ADP2 Fluorescence Intensity (FI) assays (Bellbrook labs) which measures ADP production. All assays contained ATP at a final concentration of 100 μM, and those with membranes used vesicles containing 5% PIP 2 and 95% PS at a final concentration of 0.5 mg/mL.
After the 60 min incubation, all reactions were stopped with 4 μL of 2× stop and detect solution containing Stop and Detect buffer (20 mM HEPES,0.02% Brij-35, 40 mM EDTA pH 7.5), 8 nM ADP Alexa Fluor 594 Tracer and 93.7 μg/mL ADP2 Antibody IRDye QC-1, covered and incubated at 20°C for 1 hr before reading the fluorescence. The fluorescence intensity was measured using a SpectraMax M5 plate reader at excitation 590 nm and emission 620 nm. All data was normalized against the appropriate measurements obtained for 100 μM ATP and 100 μM ADP with no kinase. The percent ATP turnover was interpolated using a standard curve (0.1-100 μM ADP). Interpolated values were then used to calculate the specific activity of the enzyme.

Biolayer interferometry
All BLI experiments were performed using the Octet K2 instrument (Fortebio Inc). For all experiments His-tagged nanobody (500 nM) was immobilized on an Anti-Penta-His biosensor for 600 s, and the sensor was dipped into varying concentrations of the protein complex being measured. A dose response was carried out for p110γ, p110γ-p84, and p110γ-p101 (50 nM -1.9 nM), with association occurring for 600 s, followed by a 1200s dissociation in Octet Buffer (20 mM Tris pH 8.5, 100 mM NaCl, 50 mM ammonium sulfate, 0.1% bovine serum albumin, and 0.02% Tween 20). Experiments comparing class IA PI3K versus class IB PI3Kγ used 50 nM of each class IA PI3K.
When comparing nanobody binding to phosphorylated and unphosphorylated p110γ, we used a final concentration of 25 nM for both phosphorylated and non-phosphorylated p110γ with association occurring for 600 s, followed by a 600 s dissociation. The K D (dissociation constant) for the different p110γ complexes was calculated from the binding curves based on their global fit to a 1:1 binding model using ForteBio data analysis 12.0 (Fortebio Inc).
TIRF-M experiments were performed using an inverted Nikon Ti2 microscope with a 100× Nikon (1.49 NA) oil immersion objective. The x-axis and y-axis positions were controlled using a Nikon motorized stage. Dy647-p84-p110γ was excited with a 637 nm diode laser (OBIS laser diode, Coherent Inc Santa Clara, CA, USA) controlled with an acousto-optic tunable filter and laser launch built by Vortran (Sacramento, CA, USA). The power output measured through the objective for single particle imaging was 1-3 mW. Excitation light passing through quad multi-pass dichroic filter cube (Semrock). Fluorescence emission passed through Nikon emission filter wheel containing the following 25 mm ET700/75 M emission filters (Semrock) before being detected on iXion Life 897 EMCCD camera (Andor Technology Ltd., UK). All TIRF-M experiments were performed at room temperature (23°C). Microscope hardware was controlled using Nikon NIS elements. Data analysis was performed using ImageJ/Fiji and Prism graphing program.

Cryo-EM sample preparation and data collection
Three µL of purified nanobody-bound p110γ at 0.45 mg/mL was adsorbed onto C-Flat 2/2-T grids that were glow discharged for 25 s at 15 mA. Grids were then plunged into liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific) with the following settings: -5 blot force, 1.5 s blot time, 100% humidity, and 4°C. Vitrified specimens were screened for ice and particle quality at the UBC high-resolution macromolecular electron microscopy (HRMEM) facility using a 200 kV Glacios transmission electron microscope equipped with a Falcon 3EC direct electron detector (DED). Clipped grids were sent to the Pacific Northwest Cryo-EM Center (PNCC) where 7322 movies were collected using a Titan Krios equipped with a Gatan K3 DED and a BioQuantum K3 energy filter with a slit width of 20 eV. The movies were collected at a physical pixel size of 0.830 Å/pix and a total dose of 50 e -/ Å 2 over 50 frames.

Cryo-EM image analysis
The data were processed using cryoSPARC v.3.3.2 (Punjani et al., 2017). The movies were preprocessed by patch motion correction using default settings except Fourier cropping by a factor of 2, followed by patch CTF estimation using default settings. A 3D map of PI3K p110γ-p101 complex (EMD-23808) was used to create 2D projections for use as templates to auto-pick 1,463,553 particles. Particles were extracted with a box size of 380 pixels, Fourier cropped to a box size of 96 pixels and subjected to 2D classification. After discarding classes with obvious noise and no features, 795,162 particles were used for multiple rounds of ab initio reconstruction and heterogeneous refinement using four or five classes; 365,178 particles, which generated the two best 3D reconstruction, were used to carry out per-particle local-motion correction with 760 pixels box size later downsized to 380 pixels followed by several rounds of ab initio reconstruction and heterogeneous refinement using three or five classes; and 149,603 from best class were further refined by homogeneous refinement and a final non-uniform-refinement which generated a reconstruction with an overall resolution of 3.02 Å based on the Fourier shell correlation 0.143 criterion.

Building the structural model of p110γ-NB7
The previous structural model of full-length p110γ from the complex of p110γ-p101 (PDB: 7MEZ) (Rathinaswamy et al., 2021a) was fit into the map using Chimera (Pettersen et al., 2004). A model of the nanobody was generated using Alphafold2 using the Colabfold v1.5.2 server (Mirdita et al., 2022). The CDR loops were removed from this initial model, and the remaining nanobody was fit into the map using Chimera. The final structure was built by iterative rounds of automated model building in Phenix, manual model building in COOT (Emsley et al., 2010), and refinement in Phenix. real_space_refine using realspace, rigid body, and adp refinement with tight secondary structure restraints (Afonine et al., 2012). This allowed for unambiguous building of the CDRs of the nanobody, and their interface with p110γ. The full refinement and validation statistics are shown in Supplementary file 1.
Phosphorylation of all proteins was confirmed using mass spectrometry and PEAKS7 analysis. The LC-MS analysis of these samples was carried out using the same pipeline as used in the HDX-MS section. The phosphorylated and non-phosphorylated peptide ratios were determined by generating extracted ion chromatograms for each phosphorylated or non-phosphorylated peptide using their molecular formula and charge state in the Bruker Compass Data Analysis software. The area under each extracted curve was then extracted. The full MS quantification of each of the phosphorylated and non-phosphorylated peptide is provided in Source data 1.

Hydrogen-deuterium exchange mass spectrometry
Exchange reactions to assess differences in p110γ upon phosphorylation were carried out at 20°C in 10 µL volumes with final concentrations of 1.6 µM for both apo and phosphorylated p110γ. A total of two conditions were assessed: p110γ apo and PKCβ phosphorylated p110γ. The H/D exchange reaction was initiated by the addition of 8 µL D 2 O buffer (94.3% D 2 O, 100 mM NaCl, 20 mM HEPES pH 7.5) to the 2 µL protein for a final D 2 O concentration of 75.4%. Exchange was carried out over five time points (3 s on ice, and 3 s, 30 s, 300 s, and 3000 s at 20°C) and the reaction was quenched with addition of 60 µL of ice-cold acidic quench buffer (0.7 M guanidine-HCl, 1% formic acid). After quenching, samples were immediately frozen in liquid nitrogen and stored at -80°C. All reactions were carried out in triplicate.

Protein digestion and MS/MS data collection
Protein samples for both HDX-MS and phosphorylation analysis were analyzed using the same LC-MS setup. Samples were rapidly thawed and injected onto an integrated fluidics system containing an HDx-3 PAL liquid handling robot and climate-controlled chromatography system (LEAP Technologies), a Dionex Ultimate 3000 UHPLC system, as well as an Impact HD QTOF Mass spectrometer (Bruker). The protein was run over two immobilized pepsin columns (Applied Biosystems; Poroszyme Immobilized Pepsin Cartridge, 2.1 mm × 30 mm; Thermo Fisher 2-3131-00; at 10°C and 2°C respectively) at 200 μL/min for 3 min. The resulting peptides were collected and desalted on a C18 trap column (Acquity UPLC BEH C18 1.7 mm column [2.1×5 mm]; Waters 186003975). The trap was subsequently eluted in line with an ACQUITY 1.7 µm particle, 100×1 mm 2 C18 UPLC column (Waters 186002352), using a gradient of 3-35% B (buffer A, 0.1% formic acid; buffer B, 100% acetonitrile) over 11 min immediately followed by a gradient of 35-80% B over 5 min. MS experiments acquired over a mass range from 150 to 2200 mass/charge ratio (m/z) using an electrospray ionization source operated at a temperature of 200°C and a spray voltage of 4.5 kV.

Peptide identification
Peptides were identified using data-dependent acquisition following tandem MS/MS experiments (0.5 s precursor scan from 150 to 2000 m/z; twelve 0.25 s fragment scans from 150 to 2000 m/z). MS/ MS datasets were analyzed using PEAKS7 (PEAKS), and a false discovery rate was set at 0.1% using a database of purified proteins and known contaminants. The same approach was used to identify phosphorylated and non-phosphorylated peptides for our in vitro phosphorylation experiments, with variable phosphorylation of STY residues was added to the search. The search parameters were set with a precursor tolerance of 20 parts per million, fragment mass error 0.02 Da, and charge states from 1 to 8, with a selection criterion of peptides that had a −10logp score of >24.03 for phosphorylated and >23.05 for non-phosphorylated. The MS/MS spectra of the PKCβ phosphorylated peptides are included in Figure 3-figure supplement 1.

Mass analysis of peptide centroids and measurement of deuterium incorporation
HD-Examiner Software (Sierra Analytics) was used to automatically calculate the level of deuterium incorporation into each peptide. All peptides were manually inspected for correct charge state, correct retention time, and appropriate selection of isotopic distribution. Deuteration levels were calculated using the centroid of the experimental isotope clusters. HDX-MS results are presented with no correction for back exchange shown in Source data 1, with the only correction being applied correcting for the deuterium oxide percentage of the buffer used in the exchange (75.4%). Changes in any peptide at any time point greater than specified cut-offs (5% and 0.45 Da) and with an unpaired, two-tailed t-test value of p<0.01 was considered significant. A number of peptides in the helical domain showed isotope distributions consistent with EX1 H/D exchange. Attempts to define the relative percentages of each population using HDExaminer were extremely noisy, so representative EX1 profiles are shown in Figure 4C. The raw peptide deuterium incorporation graphs for a selection of peptides with significant differences are shown in Figure 4D, with the raw data for all analyzed peptides in Source data 1. To allow for visualization of differences across all peptides, we utilized number of deuteron difference (#D) plots ( Figure 4B). These plots show the total difference in deuterium incorporation over the entire H/D exchange time course, with each point indicating a single peptide. The data analysis statistics for all HDX-MS experiments are in Appendix 1-key resources table according to the guidelines of Masson et al., 2019. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Perez-Riverol et al., 2022) with the dataset identifier PXD040765.