G-protein-regulated PI3Kγ (phosphoinositide 3-kinase γ) plays a crucial role in inflammatory and allergic processes. PI3Kγ, a dimeric protein formed by the non-catalytic p101 and catalytic p110γ subunits, is stimulated by receptor-released Gβγ complexes. We have demonstrated previously that Gβγ stimulates both monomeric p110γ and dimeric p110γ/p101 lipid kinase activity in vitro. In order to identify the Gβ residues responsible for the Gβγ–PI3Kγ interaction, we examined Gβ1 mutants for their ability to stimulate lipid and protein kinase activities and to recruit PI3Kγ to lipid vesicles. Our findings revealed different interaction profiles of Gβ residues interacting with p110γ or p110γ/p101. Moreover, p101 was able to rescue the stimulatory activity of Gβ1 mutants incapable of modulating monomeric p110γ. In addition to the known adaptor function of p101, in the present paper we show a novel regulatory role of p101 in the activation of PI3Kγ.
Class I PI3Ks (phosphoinositide 3-kinases) are second-messenger-generating enzymes, which transform extracellular signals into the principle product PtdIns(3,4,5)P3 in order to control a plethora of fundamental cellular responses, including proliferation, differentiation, growth and chemotaxis [1–8]. On the basis of their structural features and modes of regulation, class I PI3Ks have been grouped into the class IA and class IB subfamilies. Class IA PI3Ks are heterodimeric lipid kinases composed of one out of five non-catalytic p85-type adaptor subunits and a catalytic subunit classified as p110α, p110β or p110δ [2,7,9–12]. Members of the class IA PI3K subfamily are recognized by the nature of their catalytic subunit, which currently appears to be more important for assigning signalling specificity than the adapter subunit [7,11,13]. All class IA enzymes are tightly and directly regulated by RTKs (receptor tyrosine kinases), other tyrosine kinases and Ras GTPases, whereas class IB PI3Ks are under the control of GPCRs (G-protein-coupled receptors) via direct interaction with Gβγ [14–19]. Only one class IB catalytic subunit, p110γ, is known . It forms dimers with one of two non-catalytic subunits, p101 or p87 (also known as p84) [14,18,21,22]. Although the p110γ subunit defines PI3Kγ as a GPCR-controlled Gβγ-dependent effector, recent evidence suggests that Ras proteins, together with the non-catalytic subunits p101 and p87, also contribute to signalling specificity [23–25].
Class IA PI3K regulation depends on tyrosine phosphorylation of activated RTK recognised by the SH2 (Src homology 2) domains of p85 subunits of the PI3K dimers which mediate translocation of the enzyme from the cytosol to the plasma membrane [10,11,19,26–30]. The recognition process is associated with conformational changes within the PI3K dimer, resulting in abrogation of p85-mediated inhibition of the catalytic p110 subunits [28,29,31–34]. Molecular biological and crystal structure studies argue that the mode and strength of the interaction between the p85 adaptor and catalytic subunits provides the basis for the autoinhibitory function of the p85 subunit, which is responsible for the differences in activity and regulation within class IA PI3K enzymes [7,13,35–37].
In contrast with the class IA PI3Ks, the mechanism of PI3Kγ regulation is poorly understood. Initially, PI3Kγ was discovered as a Gβγ-sensitive monomer [15,20], but shortly afterwards p101 was described as an indispensable complex partner of p110γ responsible for sensitizing PI3Kγ to Gβγ . In fact, the data suggest that Gβγ may stimulate PI3Kγ solely through interaction with p101, which somewhat resembles the scenario known for the activation of class IA PI3Ks. To elucidate how p101 sensitizes Gβγ for PI3Kγ, we and others have studied PI3Kγ in its monomeric or heterodimeric form in vitro and in vivo [16–18,22,24,38,39].
We have found evidence that Gβγ interacts with both PI3Kγ subunits in a selective manner in order to stimulate PtdIns(3,4,5)P3 formation. This prompted us to propose a model of Gβγ-induced activation of PI3Kγ, in which Gβγ has to bind to the non-catalytic p101 subunit for translocation of the enzyme to the membrane, enabling PtdIns(3,4,5)P3 formation by direct interaction of Gβγ with p110γ . The underlying data suggest that the latter step is independent of p101; however, we could not exclude the possibility that p101 may be involved in Gβγ-induced stimulation of membrane-attached p110γ. Therefore Gβγ may interact with p101 and p110γ through individual or common binding sites.
In order to validate our hypothesis and to identify the structural determinants of Gβγ involved in the membrane recruitment and regulation of PI3Kγ enzymatic activity, we addressed this question by using Gβ1 mutants where the amino acids involved in interactions with GDP-bound Gα or downstream effectors were substituted by alanine [40–43].
Expression and purification of recombinant proteins
Sf9 (Fall Armyworm Ovary; Gibco) cells were cultured in suspension with TNM-FH medium (Sigma) supplemented with 10% (v/v) FBS (fetal bovine serum; Gibco), lipid medium supplement (1:100 dilution; Sigma), penicillin (100 units/ml) and streptomycin (0.1 mg/ml). For protein expression, Sf9 cells (1.5×106 cells/ml) were infected with viruses encoding the subunits of PI3Kγ and/or wild-type or mutant Gβ together with Gγ [42,44]. After 48 (PI3Kγ) or 60 (Gβγ variants) h of infection, the cells were collected by centrifugation at 1000 g for 5 min and washed twice with PBS. Subsequent purification of lipidated recombinant Gβ1γ2 variants was performed as detailed previously . Expression and purification of recombinant His6-tagged PI3Kγ were carried out according to protocols published previously [16,45] with some modifications. After elution from a Resource 15Q 5/5 column, fractions containing PI3Kγ were pooled, concentrated and loaded on to a gel filtration Superdex HR 10/30 column. Proteins were eluted using a buffer containing 20 mM Tris/HCl, pH 8, 150 mM NaCl, 2 mM dithiothreitol and 0.033% C12E10 (polyoxyethelene-10-lauryl ether). Purified proteins were quantified by Coomassie Brilliant Blue staining following SDS/PAGE (10% or 15% gel) with BSA as the standard. The proteins were stored at −80°C.
Copurification of Gβ1γ2 with PI3Kγ subunits
For the copurification experiments , viruses encoding His6-fused subunits of PI3Kγ, p101 or p110γ were co-infected with viruses encoding Gβ1γ2. After 55 h, the cells were harvested and lysed by forcing the Sf9 cell suspension through a 22-gauge needle five times and subsequently through a 26-gauge needle 10 times. The suspension was incubated for 30 min with a buffer containing 20 mM Hepes/NaOH, pH 7.5, 150 mM NaCl, 10 mM 2-mercaptoethanol and 0.5% C12E10, and incubated with Ni2+-NTA (nitrilotriacetic acid) Superflow for 2 h. After several washing steps, the proteins of interest were eluted using a buffer containing 20 mM Tris/HCl, pH 8.0, 150 mM NaCl, 10 mM 2-mercaptoethanol, 0.1% C12E10 and 200 mM imidazole.
Gel electrophoresis, immunoblotting and antibodies
Generation and characterization of the antiserum against Gβ1 subunit (AS 398) are detailed elsewhere . Monoclonal anti-PI3Kγ antibodies directed against intact p110γ were described previously . Preparations containing Gβ1 and p110γ proteins were fractionated by SDS/PAGE (10% or 15% gel) and transferred on to nitrocellulose membranes (Hybond™-C Extra, Amersham Biosciences). Visualization of antibodies was performed using an ECL (enhanced chemiluminescence) system (Amersham Biosciences) or the SuperSignal® West Pico Chemiluminescent Substrate (Pierce) according to the manufacturers' instructions.
Proteolysis of Gβ1γ2 variants with trypsin
The digestion assay, with some modifications, was performed as detailed previously . Proteins were cleaved with Tos-Phe-CH2Cl (tosylphenylalanylchloromethane, also known as TPCK)-treated trypsin. The assays were conducted in a final reaction volume of 30 μl containing 20 mM Tris/HCl, pH 8.0, 150 mM NaCl, 2 mM dithiothreitol and 0.033% C12E10.
The G-protein concentration in the reaction mixture was 167 μg/ml. Tos-Phe-CH2Cl-treated trypsin was diluted in the same buffer and added to the sample at a 1:25 trypsin/substrate ratio. The samples were incubated for 40 min at 30°C. Proteolysis was terminated by the addition of 4× Laemmli sample buffer and the samples were boiled for 1 min. The reactions were analysed by SDS/PAGE (15% gel).
In vitro assay for lipid kinase activity
The assays were conducted in a final volume of 50 μl containing 40 mM Hepes/NaOH, pH 7.4, 0.1% BSA, 1 mM EGTA, 7 mM MgCl2, 120 mM NaCl, 1 mM dithiothreitol and 1 mM β-glycerophosphate (vesicle buffer) as described previously [24,45] with some modifications. A 30 μl lipid vesicle mixture, containing 320 μM phosphatidylethanolamine, 300 μM phosphatidylserine, 140 μM phosphatidylcholine and 30 μM sphingomyelin supplemented with 40 μM PtdIns(4,5)P2 was dried using N2 gas and sonicated in vesicle buffer. Subsequently, the phospholipid vesicles were mixed with Gβ1γ2 and incubated on ice for 10 min. The samples containing different amounts of Gβ1γ2 were adjusted to identical detergent concentrations, such as 0.002% of C12E10. Thereafter 10 ng of PI3Kγ was added, and the mixture was incubated for another 10 min at 4°C in a final volume of 40 μl. Then, the assay was started by adding 40 μM ATP (1 μCi of [γ-32P]ATP, Hartmann Analytic) in 10 μl of the above-mentioned assay buffer at 30°C. After 15 min, the reaction was stopped with 150 μl of ice-cold 1 N HCl and the tubes were placed on ice. The lipids were extracted by vortexing the samples with 500 μl of a 1:1 chloroform/methanol solution. After centrifugation (4000 g for 1 min and 4°C), the organic phase was washed with 200 μl of 1 N HCl. Subsequently, 25–70 μl of the organic phase was resolved on a potassium oxalate-pretreated TLC plate (Whatman) with 35 ml of 2 N acetic acid and 65 ml of n-propyl alcohol as the mobile phase. Dried TLC plates were exposed to Fuji imaging plates, and autoradiographic signals were quantified with a Fujifilm FLA-5000 imaging system (Raytest).
Differences in the lots of the phospholipids used and variability in various experimental parameters made it difficult to assure the precision of reproducibility necessary for analysing the results of the different Gβ1γ2 variants. For that reason, the ability of Gβ1γ2 variants to simulate PI3Kγ in all experiments was studied in side-by-side experiments using wild-type Gβ1γ2. The simultaneous determination of PI3Kγ activity induced by wild-type Gβ1γ2 enabled the calculation of the correlation coefficient of PI3Kγ stimulation between each tested concentration of the Gβ1γ2 variants and wild-type Gβ1γ2. On the other hand, the different data sets of PI3Kγ activity induced by wild-type Gβ1γ2 allowed us to determine the maximal stimulation (Vmax) and EC50 values of our experimental set-up. The results are means±S.D. from at least three independent experiments. The maximal stimulation (Vmax) of p110γ and p110γ/p101 in the presence of wild-type Gβ1γ2 was 3.2±1.3 and 28.1±7.6 nmol PtdIns(3,4,5)P3 per mg of protein/min respectively. The EC50 values were 202.6±28.3 and 8.7±2.8 nM for p110γ and p110γ/p101 respectively. On the basis of the mean values, the data for the Gβ1γ2 variants were normalized by replotting the data corresponding to the template curve using the mean of the correlation coefficients estimated in individual experiments.
In vitro assay for protein kinase activity
The protein kinase activity of PI3Kγ was measured as described previously for the lipid kinase activity with some modifications . The assay volume was 25 μl (2 μCi of [γ-32P]ATP per tube). The phospholipid vesicles were prepared without PtdIns(4,5)P2. The reaction was stopped after an incubation period of 30 min at 30°C by adding 10 μl of 4× Laemmli sample buffer. Following separation by SDS/PAGE (10% gel), the proteins were transferred on to nitrocellulose membranes. Dried membranes were exposed to Fuji imaging plates, and autoradiographic signals were measured using a FLA-5000 Fuji-Imager (Raytest). Generation and presentation of the dose-response curves was done as described for the lipid kinase activity.
Lipid vesicle pull-down assay
The experimental conditions for the determination of Gβ1γ2 and PI3Kγ association in phospholipid vesicles were similar to the measurements of the enzymatic activity of PI3Kγ . The assay did not contain radioactively labelled ATP and had a higher amount of PI3Kγ (200–400 ng). After an incubation period of 15 min at 30°C, the mixture was put on ice and centri-fuged at 12000 g for 2 min at 4°C. The supernatant and pellet were separated. The supernatant was supplemented with 4× Laemmli sample buffer. The pellet was resuspended and washed twice with vesicle buffer. Subsequently the pellet was resolved in 1× Laemmli sample buffer. The samples were subjected to SDS/PAGE (10% gel) and transferred on to nitrocellulose membranes. Semiquantitative analysis of immunoblots was performed using specific antisera against p110γ and Gβ1 subunits.
RESULTS AND DISCUSSION
Sensitivity of Gβ1γ2 variants to trypsin digestion
Gβ1 mutants co-expressed with the His6-tagged and isoprenylated Gγ2 subunit formed heterodimers and were purified from Sf9 cells as described previously . Proper protein folding of the purified mutants was confirmed by a partial trypsin digestion assay. This approach has been described previously as a useful tool to examine protein integrity as a surrogate for correct folding of mutant proteins [42,47]. The tryptic digestion of the intact Gβ1γ2 dimer yielded only two proteolytic Gβ1 fragments of 26 kDa and 14 kDa, despite the presence of 32 potential tryptic sites in the primary sequence (Figure 1). In contrast, thermal denaturation of the protein prior to tryptic digestion resulted in a protein smear with a multiplicity of protein fragments indicative of proteolysis of the unfolded Gβ1 protein (Figure 1, WT panel, middle lane). Following this approach, we checked all purified isoprenylated Gβ1γ2 variants for correct folding and dimerization with Gγ prior to further functional analysis (Figure 1). Only the Gβ1K78Aγ2 mutant showed more than two proteolytic fragments (Figure 1, K78A panel). Nevertheless, purified Gβ1K78Aγ2 was able to fully stimulate PI3Kγ activities (see below), including protein kinase activity. This may indicate that the K78A mutation is not directly involved in PI3Kγ interactions, regardless of whether or not the K78A mutation destabilizes the overall Gβ1 structure. Despite its enhanced sensitivity towards trypsin treatment, the conformation of Gβ1K78Aγ2 may be stabilized upon interaction with PI3Kγ.
Characterization of purified recombinant Gβ1γ2 variants by partial trypsin digestion
The purified Gβ1γ2 variants underwent a first round of evaluation to determine PI3Kγ lipid kinase activity under maximal stimulatory conditions. The extent of stimulation of dimeric p110γ/p101 activity by the different variants was plotted against the Gβγ-stimulated activity of monomeric p110γ (Figure 2). Each symbol reflects an individual Gβ1 mutant. Their distribution allowed the assignment of the Gβ1γ2 variants to different groups, highlighted by grey or black colour coding. The grey symbols represent the group of Gβ1γ2 variants showing more or less wild-type features, which can be clearly distinguished from the mutants represented by black symbols.
Ability of Gβ1γ2 variants to stimulate p110γ/p101 or p110γ lipid kinase activity
Gβ1γ2 variants with wild-type phenotype
Gβ1γ2 variants with an alanine residue within the Gα-binding region of Gβ1 at the positions Leu55, Lys57, Lys78, Ile80, Lys89, Ser98, Asn119 or Thr143 stimulated the lipid kinase activity of both p110γ and p110γ/p101 in a similar manner to the experiments performed in parallel using wild-type Gβ1γ2 (Figures 3A and 3B). Alanine mutations of these amino acids did not affect the Gβ1γ2-dependent recruitment of p110γ/p101 to phospholipid vesicles (Figure 3C and results not shown). Since the Gβ1γ2-dependent translocation of monomeric p110γ to phospholipid vesicles (in the pellet) was weak (Figure 3D), we excluded this aspect from the present study. It is remarkable that this group of Gβ1 residues are apparently not essential for the interaction with PI3Kγ, as they are integral elements for binding to both the N-terminal (Lys57, Ser98, Asn119 and Thr143) and switch II interfaces (Leu55, Lys78, Ile80 and Lys89) of the GDP-bound Gα subunit [40,41]. Alanine mutations of these residues have been shown to inhibit the formation of the heterotrimeric complex with the Gα subunit and/or modulate the activity of effector proteins, including ACII (adenylyl cyclase II), PLCβ2 (phospholipase Cβ2), GIRK (G-protein-activated inward rectifier potassium channel) and the Ca2+ channel Cavα1B [42,48,49]. Taken together, these residues cluster into an interacting surface on Gβ1 that is involved in the modulation of effectors other than PI3Kγ.
Gβ1γ2 variants leave PI3Kγ enzymatically unaltered
In addition to the Gβ1γ2 variants showing more or less indistinguishable effects from wild-type Gβ1γ2, we identified variants which differed in their capacity to stimulate PI3Kγ activity, suggesting that these Gβ1 residues are important for the Gβ1γ2-dependent regulation of PI3Kγ (Figure 2). Within this group of regulatory relevant mutants, two mutants represented in Figure 2 by black triangles (L117A and Y145A) can be distinguished from the mutant groups represented in Figure 2 by black circles (W99A and M101A) and black squares (Y59A, D228A and W332A) based on their mode of PI3Kγ activation.
Key residues of Gβ1 necessary for PI3Kγ regulation
Within the panel of mutants tested, two variants attracted the most attention (Figures 4A and 4B, black triangle symbols). These mutants, Gβ1L117Aγ2 and Gβ1Y145Aγ2, failed to activate p110γ at any of concentrations tested (Figure 4A). This finding points to a direct interacting role of residues Leu117 and Tyr145 in the activation process of the catalytic p110γ subunit of PI3Kγ. Similar observations for Gβ1L117Aγ2 were reported previously for the regulation of ACII, PLCβ2 and PLCβ3 [42,48]. Additionally, Gβ1 residues Leu117 and Tyr145 are involved in interactions with the GRK2 PH (GPCR kinase 2 pleckstrin homology) domain . Surprisingly Gβ1Y145Aγ2 regained its ability to stimulate p110γ when complexed to p101, albeit with significantly lower potency and efficiency (Figure 4B). Correspondingly Gβ1Y145Aγ2 recruited p110γ/p101 to phospholipid vesicles, although with less efficiency than wild-type Gβγ (Figure 4C). In contrast, the Gβ1L117Aγ2 mutant, which also failed to stimulate p110γ, only revealed a very weak ability to stimulate it (up to ~20%) in the presence of p101 (Figure 4B). The strong reduction in the PI3Kγ-stimulatory activity of Gβ1L117Aγ2 correlated with an impaired ability to recruit p110γ/p101 to phospholipid vesicles (Figure 4C). Furthermore in the copurification experiment (Figure 4D), Gβ1L117Aγ2 showed remarkably decreased interactions with p101 as compared with wild-type Gβ1γ2, and did not apparently copurify with the p110γ subunit of PI3Kγ.
Gβ1γ2 variants with altered characteristics of PI3Kγ lipid kinase activation
These results readily suggest that the severe reduction in Gβγ mutant stimulatory capacity was caused by an impairment in the interaction with PI3Kγ subunits rather than specific interference in the activation process. This conclusion may be premature since the significance of both the recruitment and copurification analysis is limited, since these methods detect static rather than dynamic protein–protein interactions and their semi-quantitative character. Therefore we decided to check the interaction of Gβγ mutants with PI3Kγ by testing the ability of Gβγ to stimulate the protein kinase activity of PI3Kγ. Although the physiological role of PI3Kγ autophosphorylation is still unclear, this feature is attractive to exploit as a read-out because the complexity of the assay is dramatically reduced due to the identity of the enzyme and substrate. Previously, we found that Gβγ enhances the autophosphorylation of Ser1101 of p110γ in a dose-dependent manner [45,51]. Although basal autophosphorylation is visible in the absence of lipid vesicles, Gβγ-dependent stimulation requires their presence . Using this approach, we tested the monomeric p110γ protein first (Figure 5A). Interestingly, it exhibited a clearly visible basal phosphate incorporation, which was in the same stoichiometric range as reported previously . Gβγ stimulated autophosphorylation of the monomer less than 2-fold (Figure 5A). In contrast, p101 suppressed the basal autophosphorylation of PI3Kγ, but also enabled wild-type Gβγ to stimulate protein kinase activity in a concentration-dependent manner by more than 12-fold (Figures 5A and 5B). All mutants stimulated the autophosphorylation of heterodimeric PI3Kγ (Figure 5B and results not shown). In particular, the Gβ1L117Aγ2 variant increased the protein kinase activity of PI3Kγ. This clearly demonstrates that Gβ1L117Aγ2, similar to all other Gβγ mutants, still physically interacts with PI3Kγ.
Gβ1γ2 variants stimulating PI3Kγ protein kinase activity
Gβ1γ2 variants depend on p101 to stimulate PI3Kγ activity
The mutants Y59A, W332A, D228A, M101A and W99A formed a set of Gβ1γ2 variants exhibiting decreasing efficiency to stimulate the catalytic p110γ subunit in its monomeric form (Figure 4A, square and circle symbols). Compared with wild-type Gβ1γ2, the maximum effect ranged between 48% (Y59A) and 3% (W99A). The mutants Y59A, W332A and D228A showed intermediate efficacy, whereas the potency of stimulation (EC50) was similar to wild-type Gβ1γ2 (Figure 4A). The results with the Gβ1M101Aγ2 and Gβ1W99Aγ2 mutants were most striking. In fact, Gβ1W99Aγ2 failed to significantly stimulate lipid kinase activity at all, although it still bound to p110γ (Figures 4A and 4D). However, we noted that binding of the mutant to p101 was blunted (Figure 4D). In order to examine whether this finding has an impact on the membrane recruitment of dimeric PI3Kγ, we assessed the recruitment of p110γ/p101 to phospholipid vesicles by Gβ1γ2 variants (Figure 4C). Gβ1W99Aγ2 showed the most diminished recruitment capability, whereas all other Gβ1γ2 mutants of this group (Y59A, M101A, D228A and W332A) were more efficient (Figure 4C and results not shown).
Much to our surprise, the Gβ1W99Aγ2 mutant was still able to efficiently stimulate PI3Kγ in the presence of p101 (Figure 4B). Actually, under the experimental conditions employed, the non-catalytic p101 subunit fully rescued maximal enzymatic activity in the presence of Gβ1W99Aγ2 as well as all other Gβ1γ2 variants of this set. However, the potencies of the mutants to stimulate PI3Kγ activity was reduced (from 22.3 nM for D228A to 98.6 nM for W99A). In order to strengthen the true rescue effect of the p101 subunit, rather than an indirect effect due to the presence of an extra protein in the reaction mixture, we analysed the stimulation of lipid kinase activity of p110γ/p87 by the Gβ1W99Aγ2 mutant. It is known that the p87 subunit binds the catalytic p110γ subunit, but is not directly involved in its activation by the Gβγ dimer [21,22,24]. In contrast with p110γ/p101 (Figure 4B), the Gβ1W99Aγ2 mutant almost completely lost its ability to stimulate p110γ/p87 (Figure 4B, inset).
These results not only validate our current understanding of the Gβγ-dependent activation of PI3Kγ, but also highlight novel unexpected aspects. The largely hydrophobic interaction of Gβ1 with Gα is provided by Trp99 of Gβ1 [40,41,52]. The fact that Gα and effectors share a common binding surface on Gβ suggests that mutation of Trp99 will also disturb the interaction between Gβγ and its effectors. Indeed, it has been shown that the mutation of Trp99 to an alanine residue affects the regulation of PLCβ2, ACII and GIRK1/GIRK4 . Consistently, the catalytic p110γ subunit of PI3Kγ represents a prototypical Gβγ effector, which is insensitive to stimulation by the Gβ1W99Aγ2 mutant. Remarkably p101 resolved the incapability of the disabled Gβ1γ2 variant to restore maximal stimulatory PI3Kγ activity. This phenomenon may be explained by a co-regulatory function of the p101 subunit. Moreover, the assumed regulatory function of p101 appears to depend on Gβ1γ2, even if the G-protein is unable to directly stimulate the catalytic subunit. These conclusions refine our current understanding of how Gβ1γ2 regulates PI3Kγ activity via p101, i.e. in addition to its adapter function which enables the membrane-bound Gβ1γ2 to recruit cytosolic PI3Kγ, p101 should also be considered as a Gβ1γ2-sensitive regulatory subunit.
Specific PI3Kγ fingerprint on the surface of Gβ1γ2
All of the tested Gβ1γ2 variants harbouring mutations within the Gα–GDP binding region could be divided into two main groups (Figure 6): (i) variants with the phenotype of wild-type Gβ1γ2, which are not involved in the regulation of PI3Kγ (white coloured residues, L55A, K57A, K78A, I80A, K89A, S98A, N119A, T143A and D186A); and (ii) variants with an impact on PI3Kγ regulation (Y59A, W99A, M101A, L117A, Y145A, D228A and W332A). Amino acid residues phenotypically resembling wild-type PI3Kγ stimulation (Figure 6, amino acids shown in white) are integral elements of both the N-terminal (Leu55, Lys78, Ile80 and Lys89) and switch interfaces (Lys57, Ser98, Asn119 and Thr143) on the surface of Gβ1 interacting with the GDP-bound Gα subunit [40,41,53]. Despite the impact of these Gβ1γ2 variants in G-protein-dependent signalling [42,48,49], they were not critical in the modulation of PI3Kγ.
Gβ1 fingerprint in the regulation of PI3Kγ enzymatic activities
Amino acids (Figure 6, shown in colour: Tyr59, Trp99, Met101, Leu117, Tyr145, Asp228 and Trp332) belonging to the Gα/Gβγ switch interface of Gβ1 clustered at the Gβ1γ2–PI3Kγ interaction site. On the basis of the functional results shown above, the interaction site is composed of at least three sectors constituted by adjacent amino acids, which represent functionally defined rather than structurally defined groups.
Sectors 1 and 2 (Figure 6)
Exchange of Tyr59, Trp99, Met101, Asp228 or Trp332 for an alanine residue results in stimulation of p110γ/p101 with less potency (Figure 4B), whereas the efficiency of stimulating the p110γ monomer is reduced (Figures 2 and 4A). The amino acids of this group can be divided into two functionally defined clusters. Sector 1 clusters amino acids (Tyr59, Asp228 and Trp332) with auxiliary roles determined by their intermediate effects in the stimulation of monomeric p110γ (Figure 4A, square symbols and Figure 6, amino acids shown in orange). In contrast, Sector 2 residues exchanged for an alanine (Trp99 and Met101) lose their ability to stimulate p110γ (Figure 4A, circle symbols and Figure 6, amino acids shown in red). All of the residues share an interesting feature: although alanine mutations partially or completely abrogate their capability to stimulate p110γ lipid kinase activity, p101 rescued their stimulating activity on heterodimeric PI3Kγ, arguing for a coregulatory function of p101 (Figure 4B).
Sector 3 (Figure 6)
The exchange of Leu117 or Tyr145 for an alanine residue allows for clear phenotypic discrimination from the effects mediated by the other amino acids (Figures 4A and 4B, triangle symbols, and Figure 6, amino acids shown in blue). Leu117 or Tyr145 mediate the stimulation of PI3Kγ, largely independently of p101, thus underlining the crucial role of these amino acids in the activation process of p110γ.
We studied the molecular mechanism of Gβ1γ2-mediated stimulation of PI3Kγ. In our approach, several parameters important for Gβγ-induced stimulation of PI3Kγ were analysed, including stimulation of monomeric p110γ, dimeric p110γ/p101 and phospholipid vesicle recruitment of dimeric p110γ/p101. Using mutation of amino acids localized in the Gα/Gβγ switch interface, we determined the fingerprint of those amino acids relevant to Gβ1–PI3Kγ interaction. Accordingly, three amino acid sectors were defined with distinct impacts on PI3Kγ regulation. The amino acids of Sectors 1 and 2 all exert p101-dependent PI3Kγ stimulation. Exchange of these amino acids with alanine resulted in reduced (Sector 1) or complete loss (Sector 2) of efficacy to stimulate monomeric p110γ, whereas the efficacy of stimulating dimeric p110γ/p101 was maintained and only the potency was reduced, which argues for a Gβγ-dependent regulatory function of p101. Leu117 and Tyr145 in Sector 3 are involved in the stimulation of p110γ, which when mutated to an alanine residue was insufficiently rescued by p101. The results of the present study validate and extend our earlier observations demonstrating a direct interaction of Gβ with p110γ and rebut the view of a linear activation chain, in which p101 solely mediates signals from Gβγ to p110γ. More interestingly, we have provided the first evidence that within this dual interaction of Gβγ with PI3Kγ subunits, p101 functions not only as a sole Gβγ adapter for membrane anchoring of PI3Kγ, but also exhibits regulatory functions by participating in the Gβγ-induced activation process of p110γ.
Aliaksei Shymanets and Bernd Nürnberg designed the research. Aliaksei Shymanets performed the research. Aliaksei Shymanets, Mohammad R. Ahmadian, Katja T. Kössmeier and Reinhard Wetzker contributed new reagents/analytical tools. Aliaksei Shymanets, Mohammad R. Ahmadian, Katja T. Kössmeier, Christian Harteneck and Bernd Nürnberg analysed the data. Aliaksei Shymanets, Mohammad R. Ahmadian, Christian Harteneck, Reinhard Wetzker and Bernd Nürnberg wrote the paper.
We thank Professor Heidi Hamm (Vanderbilt University, Nashville, U.S.A.) for providing the baculoviruses encoding the Gβ1 mutants. The expert technical assistance of Sonja Weinmann and Renate Riehle is greatly appreciated. We also thank Dr Roger Williams, Dr Len Stephens and Dr Phill Hawkins (Babraham Institute, University of Cambridge, Cambridge, U.K.) for fruitful discussions. All members of the Nürnberg laboratory previously located in Düsseldorf and in Tübingen are thanked for discussion and support.
The study was financed in part by the Deutsche Forschungsgemeinschaft (DFG). Mohammad R. Ahmadian and Katja T. Kössmeier were supported by the NGFNplus program of the German Ministry of Science and Education (BMBF).