Sequential activation of class IB and class IA PI3K is important for the primed respiratory burst of human but not murine neutrophils

Neutrophils are critical components of the immune system and have a vital role in combating bacterial and fungal infections.¹  A key weapon in the neutrophil armory is the so-called “respiratory burst,” the generation of reactive oxygen species (ROS) by a multicomponent oxidase complex.^2,3  Patients with chronic granulomatous disease (CGD) caused by defective expression of active oxidase components experience recurrent, life-threatening infections.⁴  The role of ROS in fighting infections is complex. ROS are involved in the killing process directly through the damaging actions of oxygen radicals and their halogenated derivatives and indirectly through the activation of phagosomal proteases.^5-7  It is also becoming apparent that ROS may regulate the neutrophil lifespan, modify the extracellular matrix through which the neutrophils migrate, and modulate the function of other cells participating in the inflammatory response.^8-12 

Given the potential for self-damage, a key feature of the inflammatory response is to confine ROS generation in time and space to areas where it is required. One way in which this is thought to occur is through a form of signal integration in which prior exposure to local proinflammatory factors is necessary for maximal activation by subsequent oxidase-triggering signals.^13-17  One of best studied examples of this “priming” phenomenon is the ability of tumor necrosis factor-α (TNF-α), a cytokine released primarily by macrophages, to dramatically augment the oxidase response to bacterially derived peptides (eg, N-formyl-methionyl-leucyl-phenylalanine [fMLP]) or components of the complement cascade (C5a). Indeed it is thought that failure to limit priming of the oxidase may play a key role in pathologic conditions in which inflammation is not effectively resolved, such as in acute respiratory distress syndrome (ARDS) or joint involvement in rheumatoid arthritis.

The neutrophil oxidase is regulated by the combined action of several intracellular signaling pathways, including those driven by phosphoinositide 3-kinase (PI3K), phospholipase C (PLC)/Ca²⁺/protein kinase C (PKC), phospholipase D (PLD), phospholipase A₂ (PLA₂), and p38/Erk.^2,3  Presumably, this diversity reflects the need for the oxidase to respond to multiple families of cell-surface receptors (eg, receptors for Fc, integrins, lipopolysaccharide, chemokines, cytokines, and bacterial fragments) that function through different proximal signal-transducing elements such as heterotrimeric G proteins or protein tyrosine kinases. It is clear from the use of specific catalytic site inhibitors and certain mouse knockouts that the PI3K signaling pathway is important for the mechanisms by which G_i-coupled receptors regulate ROS production in neutrophils—for example, downstream of receptors for fMLP, C5a, platelet-activating factor (PAF), histamine, LTB4, and adenosine triphosphate (ATP).^18-21  Moreover, we have previously described the ability of TNF-α to potentiate PI3K activation in response to fMLP, suggesting it may be a mechanism that underlies oxidase priming by this cytokine.^22,23 

Neutrophils contain members of each of the currently defined families of PI3K—classes I, II, and III. The class I family of PI3Ks is responsible for cell-surface receptor-generated PtdIns(3,4,5)P₃ and PtdIns(3,4)P₂ and is thought to play the major role in the regulation of the oxidase.²⁴  Class I PI3Ks are subdivided into class IA and class IB based on the nomenclature of their p110 catalytic subunits (α, β, γ, δ) and mode of regulation; hence, PI3Kγ is subdivided into class IB, and PI3Kα, β, and δ are subdivided into class IA.²⁴  The class IB enzyme has a p101 regulatory subunit and is activated by G-protein-coupled receptors. The class IA enzymes have p55-85 regulatory subunits and are classically activated by tyrosine kinase-coupled receptors.²⁴  Recent studies using mouse PI3Kγ knockouts have defined the class IB isoform as the major player in fMLP-driven PtdIns(3,4,5)P₃ synthesis and ROS production.^19-21  However, there is an apparently contradictory body of work implying the involvement of class IA rather than class IB PI3Ks in the activation of the oxidase by fMLP in human neutrophils.^25-27  Furthermore, our studies describing the effects of TNF-α on the fMLP-stimulated formation of PtdIns(3,4,5)P₃ in human neutrophils suggested a potentiating effect at later, rather than earlier, times of fMLP stimulation,²²  implying a level of PI3K complexity in the response. In light of these observations, we set out to use several recently developed isoform-selective PI3K inhibitors and mouse PI3K “knockout” models to assess the contribution of the various class I PI3Ks to TNF-α priming of the fMLP oxidase response in human and mouse neutrophils.

The PI3K inhibitors TGX-221 (patent WO-2004016607), AS-252424 (5-[5-(4-fluoro-2-hydroxy-phenyl)-furan-2-ylmethylene]-thiazolidine-2,4-dione; synthesis described in the Supplementary Materials, available on the Blood website [see the link at the top of the online article]), IC87114²⁷  (patent WO-0181346), and YM-024 (compound 24 in Table 6; patent EP1277738) were synthesized solely for the purposes of this study. Human TNF-α (hTNF-α) was from R&D (Minneapolis, MN); fMLP, horseradish peroxidase (HRP), luminol, cytochrome c, and Hanks balanced salt solution (HBSS) were from Sigma (Dorset, United Kingdom).

The PI3Kγ^-/-, PI3Kδ^D910A/D910A, and PI3Kδ^-/- mouse strains used in this study have been described previously.^21,28,29  These strains were bred on a C-57BL/6 background for at least 8 generations. The PI3Kγ^-/-, δ^-/- strain was created by interbreeding PI3Kγ^-/- and PI3Kδ^-/- mice on a mixed C-57BL/6, 129/Sv background. In all experiments in which these strains of mice were used, they were compared with appropriate strain-matched wild-type controls. Animals were housed in the small animal barrier unit (SABU) at the Babraham Institute. This work was approved by Home Office Project License PPL 80/1875.

Human neutrophils (at least 95% purity by cytospin) were isolated from the peripheral blood of healthy volunteers by centrifugation over plasma/Percoll gradients, as previously described.³⁰  Murine bone marrow was dispersed in HBSS (without Ca²⁺ and Mg²⁺) with 0.25% fatty acid-free bovine serum albumin (BSA) (HBSS/BSA) and was centrifuged (1256g, 30 minutes, room temperature) over discontinuous gradients composed of 81%, 62%, and 55% Percoll in HBSS. Mature neutrophils were obtained from the 55%/62% interface (purity, 75%-85% by cytospin), and contaminating red blood cells were removed by ammonium chloride lysis; cells were washed twice in HBSS/BSA.

Neutrophils (4 × 10⁷ human; 7 × 10⁷ mouse) were washed twice in inorganic phosphate (Pi)-free HBSS and incubated with [³²P]Pi (6 mCi [222 MBq]; PBS13; Amersham Pharmacia, Bucks, United Kingdom) and TNF-α (200 U/mL and 500 U/mL for human and mouse cells, respectively) in 2 mL Pi-free HBSS for 65 minutes at 37°C. Cells were washed and resuspended in a final volume of 2.4 mL Pi-free HBSS; 150-μL aliquots were warmed to 37°C in the presence of 10 μL PI3K inhibitors (no more than 0.01% dimethyl sulfoxide [DMSO]) for 5 minutes and then challenged with 10 μL fMLP (100 nM and 10 μM final concentrations for human and mouse cells, respectively) for various times. Incubations were terminated, and the lipids were extracted, deacetylated, and quantified by anion-exchange chromatography, as described previously.^31,32  The data represent disintegrations per minute (dpm) in the appropriate fractions corrected between experiments to 1 × 10⁵ dpm GroPIns(4,5)P₂ in the average control samples.

Neutrophils (2 × 10⁶/mL human; 10⁷/mL mouse) were incubated in HBSS with TNF-α at 37°C for 25 minutes before the addition of PI3K inhibitors for another 5 minutes (as described in the previous paragraph). Cells were stimulated with fMLP (as described in the previous paragraph) in the presence of cytochrome c (1.2 mg/mL final), and the optical density of the supernatants was determined by spectrophotometry (550 nm). To obtain data on the kinetics of the oxidative burst, measurements were also obtained using luminol-dependent chemiluminescence. One hundred microliters of neutrophils (4 × 10⁶/mL human; 5 × 10⁶/mL mouse) were primed with TNF-α and preincubated with PI3K inhibitors (as described in the previous paragraph); 100 μL luminol (1 μM final concentration) and HRP (62.5 U/mL final concentration) in HBSS were added, and 150-μL aliquots were transferred to a prewarmed 96-well luminometer plate. Light emission was recorded by a Berthold MicroLumat Plus luminometer (Berthold Technologies, Hartfordshire, United Kingdom) (data output is in relative light units per second); fMLP was added through the injection port.

Human neutrophils (2 × 10⁷/mL) were primed with TNF-α (200 U/mL) or HBSS for 30 minutes at 37°C. Cells were treated with diisopropyl fluorophosphate (DIFP) (7 μM, 5 minutes, RT), pelleted, and resuspended at 4 × 10⁷/mL in HBSS. Aliquots (0.25 mL each) were equilibrated to 37°C and stimulated with fMLP (1 μM) for 30 seconds, unless otherwise indicated. Reactions were terminated by the addition of 1 mL ice-cold phosphate-buffered saline (PBS)/0.5 mM orthovanadate; cells were pelleted and lysed in 1 mL ice-cold lysis buffer (20 mM Tris pH 8.0, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, 1% NP-40, 10% glycerol, 1 mg/mL fatty acid-free BSA, 0.5 mM orthovanadate, 0.1 mM phenylmethylsulfonyl fluoride [PMSF], and 10 μg/mL each of leupeptin, antipain, pepstatin A, and aprotinin). Antiphosphotyrosine antibody-directed immunoprecipitation of PI3K activity was performed with 25 μg FB2 (Cancer Research UK, London, United Kingdom), and PI3K activity was assayed as described previously.³³ 

We measured the levels of [³²P]-phosphoinositides in [³²P]Piprelabeled, TNF-α-primed human neutrophils stimulated with 100 nM fMLP (Figure 1). The low basal levels of PtdIns(3,4,5)P₃ and PtdIns(3,4)P₂ and the dramatic increase in their concentrations on exposure to fMLP are consistent with previous work describing fMLP stimulation of class I PI3K activity, catalyzing the conversion of PtdIns(4,5)P₂ to PtdIns(3,4,5)P₃, followed by 5-phosphatase-linked conversion to PtdIns(3,4)P₂.^32,34  The significantly higher basal level of [³²P]PtdIns3P and the smaller increase on fMLP stimulation are consistent with previous reports of negligible effects of fMLP on class III or class II catalyzed conversion of PtdIns to PtdIns3P.³²  The transient decrease in [³²P]PtdIns(4,5)P₂ on fMLP stimulation is consistent with competing effects of consumption (by PI3K and PLC) and resynthesis of this lipid (by PI4P5K), as previously reported.³² 

The fMLP-stimulated synthesis of PtdIns(3,4,5)P₃ in TNF-α-primed human neutrophils is biphasic, with levels peaking approximately 10 seconds and 60 seconds after the addition of fMLP (Figure 1). We have previously shown that the size of this latter peak of PtdIns(3,4,5)P₃ accumulation is dependent on prior exposure to TNF-α (5- to 6-fold reduced in the absence of TNF-α),²²  suggesting significant complexity in the response. We sought to define the roles of the various class I PI3K isoforms under these conditions by using several recently developed isoform-selective, catalytic site inhibitors. The origins and structures of the compounds chosen for this study are given in “Materials and methods” and Figure S1, and their reported PI3K selectivities in vitro are given in Table 1. We confirmed the relative selectivity of these compounds by assessing their IC₅₀ against the in vitro lipid kinase activities of recombinant PI3Kα, β, δ, and γ in identical assays (Table S1); given that each of these compounds acts as a competitive ATP site inhibitor and that the concentration of ATP used in our assays differed from that of others, the data are in good agreement with those presented for these compounds in the relevant patent applications.

The effect of 0.3 μM TGX-221 on the time course of [³²P]phosphoinositide accumulation in fMLP-stimulated, TNF-α-primed human neutrophils is shown in Figure 1. At this concentration, TGX-221 has a negligible effect on the increasing phase of the first peak of PtdIns(3,4,5)P₃ accumulation (6-second time point; Figure 1) but a substantial effect on the second peak of PtdIns(3,4,5)P₃ accumulation (60-second time point; Figure 1). Concentration-response curves for the effects of TGX-221 on PtdIns(3,4,5)P₃ accumulation after 6- or 60-second stimulation of TNF-α-primed cells are shown in Figure 2; effects of TGX-221 on the other lipids measured are described in Figure S2. The results clearly indicate that TGX-221 inhibits the second phase of PtdIns(3,4,5)P₃ accumulation more potently than the first phase.

Concentration-response curves for the effects of IC87114 and YM-024 on fMLP-stimulated accumulation of PtdIns(3,4,5)P₃ at 6 and 60 seconds are also shown in Figure 2 and on other phosphoinositides in Figure S2. The relative lack of effect of TGX-221, IC87114, and YM-024 on the levels of PtdIns(4,5)P₂ and PtdIns4P suggest that, at the concentrations tested, these compounds have few nonspecific effects on other neutrophil phosphoinositide kinases. The effects of TGX-221 on basal PtdIns3P levels suggest this compound may inhibit a class III PI3K with an IC₅₀ of approximately 1 μM (Figure S2A). The lag in fMLP-stimulated accumulation of PtdIns(3,4)P₂ means that the 6-second data for this lipid is a composite of roughly equal proportions of basal and stimulated synthesis (Figure S2C) and, hence, is difficult to interpret. The results for PtdIns(3,4,5)P₃ accumulation, however, allow several clear conclusions to be drawn. TGX-221, IC87114, and YM-024 each inhibit the second phase of fMLP-stimulated PtdIns(3,4,5)P₃ accumulation more potently than the first, and there is multicomponent inhibition of the second phase (Figure 2). Thus, more than one isoform of PI3K is involved in the synthesis of PtdIns(3,4,5)P₃ downstream of fMLP stimulation in TNF-α-primed cells. Comparison of the effects of TGX-221, IC87114, and YM-024 on fMLP-stimulated accumulation of PtdIns(3,4,5)P₃ at 60 seconds with the respective IC₅₀ against PI3Kδ (Table 1) suggests 50% to 60% of this accumulation is dependent on the δ isoform. The identity of the other PI3Ks contributing to PtdIns(3,4,5)P₃ accumulation at 60 seconds is more ambiguous, but the data suggest limits of no more than 15% for PI3Kβ, 15% for PI3Kα, and 30% for PI3Kγ. By contrast, PtdIns(3,4,5)P₃ accumulation at 6 seconds cannot be significantly dependent on the α, β, or δ isoform (lack of sensitivity to inhibition by YM-024, TGX-221, and IC87114, respectively, at concentrations relevant to inhibition of these isoforms) and, hence, must be largely dependent on PI3Kγ.

To confirm the contribution of PI3Kγ activity to PtdIns(3,4,5)P₃ accumulation downstream of fMLP stimulation in TNF-α-primed cells, we investigated the effects of a recently developed PI3Kγ-selective inhibitor, AS-252424 (Figures 2, S2). In contrast to the other inhibitors used, AS-252424 inhibited both phases of PtdIns(3,4,5)P₃ accumulation in parallel (Figure 2). The IC₅₀ for this inhibition is substantially greater than that predicted by our in vitro PI3K assays (Table 1), but this is likely to reflect the poor cellular accessibility of this compound (M.C., personal communication, January 2005). Given the greater than 15-fold difference between the in vitro IC₅₀ for AS-252424 inhibition of PI3Kγ and that for PI3Kδ, these results indicate that the second, largely PI3Kδ-dependent phase of PtdIns(3,4,5)P₃ accumulation requires the prior PI3Kγ-dependent phase of PtdIns(3,4,5)P₃ accumulation.

Precedent suggests that class I PI3Ks (α, β, δ) are activated downstream of receptors by a tyrosine kinase-based transduction mechanism, with tyrosine phosphorylation of YXXM motifs in an adaptor protein leading to subsequent SH2 domain-mediated recruitment and activation of the p85/p110α, β, and δ PI3K heterodimers.²⁴  Therefore, we measured the presence of PI3K activity associated with antiphosphotyrosine-directed immunoprecipitates isolated from fMLP-stimulated neutrophils. We could indeed measure a TNF-α-dependent, fMLP-stimulated accumulation of PI3K activity in antiphosphotyrosine immunoprecipitates (Figure 3A), albeit with a time course that did not correlate precisely with PtdIns(3,4,5)P₃ accumulation in intact cells (Figure 3B). The sensitivity of the in vitro PI3K activity captured in the antiphosphotyrosine immunoprecipitates to PI3K inhibitors indicates that this activity is largely PI3Kδ (Figure 3C).

We investigated the effects of 2 inhibitors with specificity toward Src and Tec family protein tyrosine kinases PP1²⁶  and LFM-A13,³⁵  respectively, on PtdIns(3,4,5)P₃ accumulation in response to fMLP in TNF-α-primed cells. PP1 inhibited PtdIns(3,4,5)P₃ accumulation at 60 seconds more than at 6 seconds (Figure 3D), with little effect on the levels of PtdIns3P, PtdIns4P, and PtdIns(4,5)P₂ (data not shown), suggesting a Src family kinase may be involved in class IA PI3K recruitment in this response.

fMLP stimulation of human neutrophils leads to the production of ROS by the neutrophil oxidase complex. Extracellular release of ROS was measured by luminometry, which, in the presence of HRP and luminol, provides a rate measure of H₂O₂ production or by cytochrome c reduction, which measures the accumulated production of O₂^- (the direct product of the oxidase). Prior exposure of human neutrophils to TNF-α caused approximately 10-fold priming of the fMLP-stimulated oxidase, as previously described²²  (Figure 4D). We investigated the sensitivity of the fMLP-stimulated ROS response in TNF-α-primed cells to PI3K inhibitors (Figure 4). Similar sensitivities were obtained with both types of ROS assay (AS-252424 could not be used in the luminometry measurements because of its strong coloration). For each of the 4 PI3K inhibitors, the concentration curves describing inhibition of ROS production correlated closely with those describing the inhibition of PtdIns(3,4,5)P₃ accumulation at 60 seconds rather than at 6 seconds of fMLP stimulation. Thus, ROS production correlated with the second phase of fMLP-stimulated PtdIns(3,4,5)P₃ accumulation across TNF-α priming and PI3K inhibition.

We attempted to resolve the relative contribution of PI3Kβ to the oxidase response by investigating parallel concentration curves for inhibition by TGX-221 in the presence or absence of 1 μM IC87114 (a concentration of IC87114 predicted to inhibit at least 70% PI3Kδ). These data revealed a sensitivity of approximately 20% of the oxidase response to concentrations of TGX-221 in the range predicted to inhibit PI3Kβ (Figure 4D). Further experiments with a range of different donors, however, suggested the contribution of PI3Kδ compared with PI3Kβ to ROS production was variable (PI3Kδ, 50%-70%; PI3Kβ, 5%-20%; data not shown).

We next measured fMLP-stimulated changes in the levels of [³²P]-phosphoinositides in [³²P]Pi-labeled, TNF-α-primed mouse bone marrow-derived neutrophils (Figures 5, S3). Mouse neutrophils showed a similar biphasic pattern of [³²P]PtdIns(3,4,5)P₃ accumulation in response to fMLP, but the magnitude of this accumulation was some 5-fold less (relative to basal levels of [³²P]PtdIns(4,5)P₂) than the analogous response in human neutrophils. Other marked differences were observed between mouse and human data. Mouse neutrophils showed a relatively poor fMLP-stimulated accumulation of [³²P]PtdIns(3,4)P₂, had a higher ratio of basal PtdIns(4,5)P₂ to PtdIns4P, and had a more pronounced fMLP-stimulated accumulation of PtdIns4P than human cells. We performed similar experiments with bone marrow-derived neutrophils prepared from mice lacking p110γ (γ^-/-), mice with 110δ replaced with a catalytically inactive allele (δ^D910A/D910A), and mice lacking p110δ and p110γ (γ^-/-, δ^-/-). Neutrophils from each strain of mouse had remarkably similar relative levels of [³²P]-PtdIns3P, -PtdIns4P, and -PtdIns(4,5)P₂ and fMLP-stimulated increases in [³²P]-PtdIns4P (Figure S3).

fMLP-stimulated PtdIns(3,4,5)P₃ synthesis was similar in TNF-α-primed neutrophils derived from the bone marrow of wild-type and PI3Kδ^D910A/D910A mice (Figure 5). The initial PtdIns(3,4,5)P₃ accumulation at 6 seconds was identical within the error of measurement, and PtdIns(3,4,5)P₃ accumulation at 60 seconds was only reduced 10% to 15% in δ^D910A/D910A neutrophils. In contrast, neutrophils from PI3Kγ^-/- and PI3Kγ^-/-, δ^-/- mice had dramatically reduced fMLP-stimulated PtdIns(3,4,5)P₃ responses, though a significant, slowly developing PtdIns(3,4,5)P₃ accumulation amounting to approximately 15% to 20% of the wild-type response at later times could be measured (Figure 5).

The results demonstrating a primary role for PI3Kγ in fMLP-stimulated PtdIns(3,4,5)P₃-accumulation are consistent with previous, less detailed measurements of PtdIns(3,4,5)P₃ accumulation in neutrophils from PI3Kγ^-/- mice.^19-21  The results, however, from the PI3Kδ^D910A/D910A mice are surprising, given previous indications that neutrophils from PI3Kδ^-/- mice have significantly reduced PtdIns(3,4,5)P₃-responses to fMLP.³⁶ 

We sought confirmation of the precise role of PI3Kδ in mouse neutrophil PtdIns(3,4,5)P₃ responses by investigating the effects of TGX-221, IC87114, and YM-024. At concentrations relevant to PI3Kδ inhibition, each of these drugs failed to inhibit fMLP-stimulated PtdIns(3,4,5)P₃ responses in TNF-α-primed cells by more than approximately 15% (Figure 6), in line with the experiments using the genetically modified mice. In contrast, the PI3Kγ-selective inhibitor AS-252424 inhibited fMLP-stimulated PtdIns(3,4,5)P₃ accumulation at 6 seconds and 60 seconds in parallel, with an IC₅₀ similar to that seen in the analogous experiments with human neutrophils (compare Figures 6 and 2). This is consistent with a dominant role for PI3Kγ in both phases of the mouse neutrophil PtdIns(3,4,5)P₃ response. The relative lack of involvement of PI3Kδ in the mouse fMLP response cannot be explained on the basis of p110δ expression levels, which are similar in human and mouse neutrophils (Figure S4). There are also no reported compensatory changes in the expression of other class I PI3K catalytic subunits in PI3Kγ^-/- or PI3Kδ^D910A/D910A mice.^21,28 

We measured the activation of the neutrophil oxidase in mouse bone marrow-derived neutrophils in experiments similar to those performed on human neutrophils. The scale of ROS production to fMLP, the PKC agonist phorbol myristate acetate (PMA), and opsonized zymosan were all substantially lower (approximately 10-fold, based on neutrophil numbers) in mouse neutrophils than in human neutrophils (data not shown). The degree of TNF-α priming of subsequent responses to fMLP or C5a were also lower in mouse neutrophils, in the 3- to 4-fold range compared with 8- to 12-fold in human cells. Furthermore, mouse neutrophils were relatively more responsive to fMLP when first isolated and lost this sensitivity in the absence, but not in the presence, of TNF-α during subsequent incubation at 37°C for 30 minutes, leading to a substantial effect of TNF-α when assayed at this point. In contrast, human neutrophils were relatively unresponsive to fMLP alone when first isolated. They gained sensitivity when subsequently incubated with TNF-α. The time course of fMLP-stimulated ROS production in TNF-α-primed mouse neutrophils was also different from that in human cells (compare Figures 7A and 4A). In mouse cells, ROS production was essentially complete by 1 minute, whereas maximal ROS production in the human cells occurred between 1 and 2 minutes after the addition of fMLP (averaged over at least 20 experiments, mean durations of mouse and human oxidase bursts were 1.2 minutes and 3.3 minutes, respectively).

fMLP stimulated similar oxidase activation in TNF-α-primed neutrophils derived from wild-type or PI3Kδ^-/- mice (Figure 7A; similar results were obtained with PI3Kδ^D910A/D910A mice; data not shown). In contrast, oxidase activation in response to fMLP in TNF-α-primed neutrophils from PI3Kγ^-/- or PI3Kγ^-/-/δ^-/- mice was inhibited by more than 80% (Figure 7A). PMA stimulated similar oxidase responses in all 4 genotypes tested (data not shown). These results are consistent with previously published defects in oxidase activation to G-protein-coupled agonists in PI3Kγ^-/- neutrophils.^19-21 

We investigated the sensitivity of the primed fMLP-stimulated oxidase response in mouse neutrophils to PI3K inhibitors. Concentration-response curves for inhibition of ROS formation by TGX-221 (Figure 7B), IC87114 (Figure S5), and YM-024 (Figure S5) were all shifted to substantially higher concentrations compared with the analogous experiments in human neutrophils. Furthermore, the concentration-response curves for TGX-221 and IC87114 were similar in wild-type and PI3Kδ^D910A/D910A neutrophils (Figures 7B, S5). These results all point to a relatively small role for PI3Kδ in the mouse oxidase response to fMLP, consistent with a minor role for this isoform in PtdIns(3,4,5)P₃ accumulation at 60 seconds. In contrast, the inhibition of fMLP-stimulated ROS production by AS-252424 was similar to that observed in human neutrophils (Figure 7C), supporting a key role for PI3Kγ in the mouse in PtdIns(3,4,5)P₃ accumulation and in activation of the oxidase.

Stimulation of unprimed human neutrophils with fMLP resulted in a rapid pulse of PtdIns(3,4,5)P₃ accumulation, peaking at approximately 6 to 10 seconds, and in minimal activation of ROS production. Pretreatment of human neutrophils with TNF-α resulted in additional, fMLP-dependent accumulation of PtdIns(3,4,5)P₃ at approximately 60 seconds and substantial, parallel activation of ROS production.^22,23  Specific pharmacologic inhibition of this later pulse of PtdIns(3,4,5)P₃ by isoform-selective PI3K inhibitors correlated with ROS production, indicating PtdIns(3,4,5)P₃ accumulation at later times (and PtdIns(3,4)P₂ derived from it) is essential, and indeed of high control strength, for the pathways by which fMLP regulates the oxidase. There are now several suggestions for signaling pathways by which PtdIns(3,4,5)P₃/PtdIns(3,4)P₂ may regulate oxidase activity, including PtdIns(3,4,5)P₃/PtdIns(3,4)P₂-regulated guanine exchange factors (GEFs) for rac (P-Rex1; vav),³⁷  PtdIns(3,4,5)P₃/PtdIns(3,4)P₂-regulated phosphorylation of p47^phox through atypical PKCs or p21-activated kinase (PAK),^2,3  and direct binding of PtdIns(3,4)P₂ to p47^phox itself.³⁸  There is still no clear assessment, however, of the relative importance of these mechanisms to ROS production in vivo.

Our data indicate that the initial phase of PtdIns(3,4,5)P₃ synthesis on the addition of fMLP to human neutrophils is entirely dependent on PI3Kγ. fMLP receptors transduce their signals through direct activation of G_i heterotrimeric G-proteins, leading to substantial release of Gα_i and Gβγ subunits. Gβγ subunits have been shown to elicit direct and substantial activation of PI3Kγ in vitro,³⁹  providing a plausible mechanism for rapid, PI3Kγ-dependent synthesis of PtdIns(3,4,5)P₃. The small GTPase Ras also activates PI3Kγ directly through binding to its p110 catalytic subunit, though the activations measured so far are relatively modest.⁴⁰  fMLP receptors are known to stimulate guanosine triphosphate (GTP) loading of Ras⁴¹ ; therefore, it is also plausible that Ras may play a role in this response.

The second phase of fMLP-stimulated PtdIns(3,4,5)P₃ accumulation is augmented by pretreatment with TNF-α and is driven largely by class IA PI3Ks. Our data suggest PI3Kδ is the predominant (50%-60%) PI3K activity in this phase of the response, with smaller contributions from PI3Kβ (15%) and possibly from PI3Kα (15% or less). It is unclear whether it is simply the timing of this second phase of PtdIns(3,4,5)P₃ accumulation that is critical to oxidase activation or whether its source is also critical. Close comparison of the inhibition curves for PtdIns(3,4,5)P₃ synthesis compared with ROS production suggest the oxidase must recognize PtdIns(3,4,5)P₃ generated by PI3Kδ and β, but the shallower curve for IC87114 inhibition of ROS production compared with that for TGX-221 (and compared with their analogous inhibition curves for PtdIns(3,4,5)P₃ synthesis) suggests the oxidase may recognize PtdIns(3,4,5)P₃ generated by the γ-isoform less effectively. However, given the error in the measurements and the relatively small, direct contribution of the γ isoform to PtdIns(3,4,5)P₃ synthesis at 60 seconds, additional experiments will be needed to clarify this. A striking feature of this second phase of class IA-driven PtdIns(3,4,5)P₃ accumulation is that it is significantly dependent on the activity of PI3Kγ; this idea that PtdIns(3,4,5)P₃ may amplify its own accumulation is also thought to be relevant to understanding the “feed forward” loops in signaling neutrophil migration.⁴² 

The mechanism by which class IA PI3Ks are activated by fMLP receptors and also the point of intervention of TNF-α signaling is unknown. In principle, TNF-α could be activating the relevant PI3Ks or inhibiting the relevant phosphatases (parallel changes in PtdIns(3,4)P₂ and PtdIns(3,4,5)P₃ suggest the phosphatase would have to be a 3-phosphatase of the phosphatase and tensin homolog [PTEN] family). Whatever the point of action of TNF-α, precedent would suggest a tyrosine kinase-based signaling pathway is important at some level for switching on the class IA PI3K.²⁴  There is significant conflict in the literature concerning the contribution that tyrosine kinase-based signaling pathways make to PtdIns(3,4,5)P₃ and O₂^- production elicited by fMLP, though our data suggest some of this confusion is likely to arise from the extent of priming, controlled or otherwise, in the various studies.^{19-21,25-27,35,43-45}  There is also a substantial body of work describing the activation of Src-family tyrosine kinases downstream of fMLP receptors in human neutrophils⁴⁶  and several suggestions that they are important for the activation of PI3K and O₂^- production.^26,47  Our data offer circumstantial support of the idea that there is recruitment of PI3Kδ to phosphotyrosine-containing complexes downstream of fMLP stimulation, that this is dependent on TNF-α priming, and that Src-family kinases may be involved, but more work must be done to establish the relevance of these observations to the accumulations of PtdIns(3,4,5)P₃ measured.

Our observations that class IA PI3Ks do not play a major role in PtdIns(3,4,5)P₃ production and oxidase activation in response to fMLP in TNF-α-primed mouse neutrophils are surprising because of our results in the analogous experiments in human neutrophils and in light of a recent report suggesting there is a substantial role for PI3Kδ in fMLP-stimulated PtdIns(3,4,5)P₃ synthesis in mouse neutrophils.³⁶  Both studies are internally consistent with respect to the effect of a PI3Kδ inhibitor (IC87114) and the PI3Kδ^-/- mouse strain used; at present, we cannot understand this discrepancy. Our results do suggest that mouse bone marrow-derived neutrophils have a superficially similar, biphasic PtdIns(3,4,5)P₃ response to fMLP when compared with human neutrophils; however, the magnitude of this response is much lower in the mouse cells. It is surprising that PI3Kγ is the major isoform responsible for both phases of PtdIns(3,4,5)P₃ production in murine cells; multiple receptors, multiple mechanisms of PI3K regulation (eg, Gβγ vs ras), potential regulation of PtdIns(3,4,5)P₃ hydrolysis, and potential secretion of further active agents all provide plausible mechanisms for obtaining temporal differences in PtdIns(3,4,5)P₃ accumulation driven through this single isoform.

There are also significant differences in fMLP-stimulated ROS production in TNF-α-primed mouse and human neutrophils. In human cells, maximal production of ROS occurs between 1 and 2 minutes after the addition of fMLP, and there is substantial accumulation of class IA PI3K-derived PtdIns(3,4,5)P₃/PtdIns(3,4)P_2. In mouse cells there is a much smaller and shorter burst of ROS production and a much lower PtdIns(3,4,5)P₃/PtdIns(3,4)P₂ accumulation. We do not understand the mechanistic basis for this difference. The basis may be related to the different ways in which oxidase activation is organized in mouse and human cells. However, it is also plausible that our findings resulted from differences in the state of differentiation or in the cellular environment between the pool of “reserve” neutrophils in the bone marrow and those in the circulation. Only small differences in ROS production have been measured between these 2 populations of neutrophils in the mouse.⁴⁸  The basis may also be attributed to a difference in the relative physiologic importance of the particular agonists we have chosen for the mouse and human cells. These are obviously important points to resolve, but, whatever the cause of the differences between our results with mouse and human neutrophils, they raise a cautionary note in using the mouse assays as a model for human neutrophil behavior, particularly with regard to testing the potential efficacy of PI3K inhibitors in PtdIns(3,4,5)P₃ and ROS production.

Given the established importance of the class I PI3K signaling pathway in many aspects of human disease, particularly in cancer and inflammation, the pharmaceutical industry is showing great interest in developing isoform-selective PI3K inhibitors.^49,50  The results presented here and elsewhere^27,51  suggest that it may indeed be possible to generate compounds with the appropriate selectivity and, furthermore, that the complexity of PI3K responses may afford useful operating windows for specificity in the inhibition of a particular cellular response. The results presented here suggest PI3Kγ is a “master” PI3K downstream of fMLP receptors in human neutrophils and that inhibition of this isoform will inhibit all PtdIns(3,4,5)P₃/PtdIns(3,4)P₂-dependent responses to this agonist; however, inhibition of PI3Kδ may inhibit only a subset of these responses (eg, ROS production). Similar levels of complexity have been revealed recently in the roles of the PI3Kγ and PI3Kδ isoforms in the cooperation between immunoglobulin E (IgE) and adenosine receptors in the regulation of mast cell secretion.^52,53  It will be interesting to establish whether this level of PI3K complexity exists in many other contexts of cellular regulation.

We thank Brian Hutchinson for synthesizing Yamanouchi Compound 24; Marion Dorning for measuring the compound IC₅₀s; and Patrick Slocombe, Gillian Hutchinson, Alaina Smith, Carl Doyle, and Kathryn McLaughlin (all employees of Celltech) for producing the purified class I PI3Ks. We thank Cedric Favré for preparing human T cells. TGX-221 was supplied by Kinacia Pty Ltd, a wholly owned subsidiary of Cerylid Biosciences Ltd (Richmond, Victoria, Australia).