Essential role of phosphoinositide 3-kinase gamma in eosinophil chemotaxis within acute pulmonary inflammation

Abstract

We and others have established an important role for phosphoinositide-3 kinase gamma (PI3Kγ) in the chemotactic responses of macrophages and neutrophils. The involvement of this lipid kinase in allergic inflammatory responses is, however, yet to be fully determined. Here we compare wild-type (WT) and PI3Kγ^−/− (KO) mice within a model of ovalbumin (OVA) -specific pulmonary inflammation. Upon OVA aerosol challenge, cell influx into the bronchoalveolar lavage (BAL) fluid consisted of neutrophils, macrophages and, more significantly, eosinophils – which are key effector cells in allergic inflammation. Each population was reduced by up to 80% in KO mice, demonstrating a role for PI3Kγ in cell infiltration into the airways. The mechanism of reduced eosinophilia was analysed within both development and effector stages of the immune response. Comparable levels of OVA-specific T-cell proliferation and immunoglobulin production were established in both strains. Furthermore, no significant differences between WT and KO chemokine production were observed. Having identified the critical point of PI3Kγ involvement, KO eosinophil chemotactic dysfunction was confirmed in vitro. These data are the first to demonstrate the vital role of PI3Kγ in acute allergic inflammation. The profound dependency of eosinophils on PI3Kγ for pulmonary influx identifies this lipid kinase as an attractive target for the pharmacological intervention of asthma.

Introduction

Over the past four decades asthma and allergic disease have steadily increased in prevalence within both industrialized and developing nations.^1,2 Despite clinical benefit from the use of inhaled steroids,³ a significant proportion of patients remain unresponsive to treatment, representing a clear unmet medical need.⁴ Asthma is an inflammatory disease of the airways characterized by eosinophilic infiltration of the bronchial mucosa and mast cell degranulation.⁵ Before site-targeted effector cell activation and the resulting pathology, an inappropriate T helper type 2 (Th2) cell-mediated immune response to the dendritic cell-presented antigen followed by education of B cells to produce antigen-specific immunoglobulin E (IgE) must occur.⁶ Once Th2 cell-driven eosinophils migrate to the lung mucosa, a vast array of degranulation products are released into the microenvironment. These include basic proteins, lipid mediators and reactive oxygen species, which damage nerves and epithelial and mucosal cells, as well as causing bronchoconstriction, vasodilatation and mucus hypersecretion.^7,8 Eosinophils also produce cytokines and chemokines such as interleukin-2 (IL-2), IL-4, IL-8, IL-13, IL-17, granulocyte–macrophage colony-stimulating factor, eotaxin and tumour necrosis factor-α, which serve to perpetuate the ongoing Th2 response and enhance cell infiltration into the inflammatory site.^9–15 There is also evidence implicating eosinophils in airway remodelling processes via the production of transforming growth factor-α, vascular endothelial growth factor, angiogenin, fibroblast growth factor-2, matrix metalloprotease 9 and tissue inhibitor of metalloprotease 1^16–18 as well as intriguing reports suggesting that the antigen-presenting cell function of eosinophils also leads to the augmentation of the inflammatory response.^19,20 There is a large body of evidence supporting the central role of eosinophils within the pathology of asthma, identifying prevention of eosinophil influx into the lung as an attractive strategy for the treatment of this disease.

Members of the phosphoinositide 3-kinase (PI3K) family have been implicated in the regulation of a diverse array of cellular responses including cell survival, mitogenesis, secretion, glucose transport, neurite outgrowth and importantly cell movement.^21,22 The PI3Ks have been classified into three groups, classes I, II and III, according to their amino acid sequence and domain structure, mode of regulation and substrate specificity in vitro.²³ We and others have contributed to an increasing body of evidence identifying the class IB isoform PI3Kγ as being central to the chemotactic responses of neutrophils and macrophages.^24–27 PI3Kγ is a heterodimer of p110γ and a regulatory subunit p101, and is activated by G-protein βγ subunits following the stimulation of G-protein-coupled receptors (GPCRs).^28,29 Both class IA and IB PI3Ks phosphorylate the 3′ position of the inositol ring of phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P₂) to form PtdIns(3,4,5)P₃.²³ Production of the latter lipid leads to the recruitment of pleckstrin homology (PH) domain-containing cytosolic proteins with catalytic and adapter functions to the membrane.^30–32 Of particular relevance to chemotactic responses is the recruitment of protein kinase B (PKB), which colocalizes with PtdIns(3,4,5)P₃ and filamentous actin at the leading edge of a cell.³³ This process, together with a positive feedback loop mediated by Rho family GTPases, produces an internal PtdIns(3,4,5)P₃ gradient, enabling directed movement via this ‘biochemical compass’.^34–36

Mice lacking p110γ are healthy and viable but display impaired neutrophil and macrophage chemotaxis in vitro and within various in vivo models.^{24–26,37,38} Both in vitro and in vivo data suggest that despite the central role of PI3Kγ in neutrophil and macrophage movement, inhibition is not complete in knockout mice,²⁵ and a role for the class IA isoform PI3Kδ was proposed.^39,40 In a recent study, we were able to address this question using PI3Kδ kinase-dead mice,⁴¹ as well as broad-spectrum PI3K inhibitors to demonstrate an exclusive role for the γ isoform in neutrophil chemotaxis within the PI3K family.²⁴ In addition, mast cells derived from p110γ mice fail to degranulate fully in response to GPCR agonists such as adenosine, resulting in reduced oedema formation in a model of passive systemic anaphylaxis.⁴²

Given the profound neutrophil, macrophage and mast cell phenotypes observed in PI3Kγ knockout (KO) mice, assessment of the roles played by this isoform in asthma was undertaken. How lacking PI3Kγ may affect the generation of a Th2-mediated immune response has not been investigated. Moreover, the role played by PI3Kγ in eosinophil movement into a site of inflammation has only been studied using isoform non-specific inhibitor tools such as wortmannin and LY294002.^43,44 In this study we used a murine model of asthma to investigate the role(s) that PI3Kγ may play at every stage, from sensitization to site-specific inflammation. We demonstrate that lacking PI3Kγ causes a profound reduction in eosinophilia, as well as impaired neutrophil and macrophage influx into the bronchoalveolar lavage (BAL) fluid. However, we also demonstrate that PI3Kγ is not involved in the generation of appropriate T-cell and B-cell responses, or in the formation of chemokine gradients within the lung. We show the level of inhibition to be at the effector stage, confirming eosinophil chemotactic dysfunction in vitro– providing a novel rationale for PI3Kγ inhibition as a powerful strategy for the treatment of asthma, without the potential liabilities associated with less focused immunosuppression.

Materials and methods

Mice

BALB/c PI3Kγ KO mice were back-crossed from a 129sv background²⁵ and bred under specific pathogen free conditions at Charles River (Margate, UK). BALB/c wild-type (WT) mice were also produced at Charles River (UK). Animals were housed at 24° in a 12 : 12-hr light : dark cycle. Food and water were accessible ad libitum. The studies reported here conformed to the UK Animals (scientific procedures) Act 1986.

Materials

Ovalbumin (OVA; grade V), endotoxin free fetal bovine serum and bovine serum albumin were obtained from Sigma-Aldrich (Poole, UK). Alum precipitate adjuvant was obtained from Serva (Heidelberg, Germany). Magnetic antibody cells sorting (MACS) reagents were obtained from Miltenyi (Bisley, UK) and fluorescence-acitvated cell sorting (FACS) reagents were obtained from Pharmingen (Oxford, UK). Quantikine enzyme-linked immunosorbent assay (ELISA) kits were obtained from R&D Systems (Abingdon, UK). An IgE OptEIA ELISA kit was obtained from BD Biosciences (Oxford, UK) and OVA-specific IgE and IgG1 ELISAs were optimized in-house using antibodies obtained from Pharmingen. Phosphate-buffered saline (PBS; without Ca²⁺ and Mg²⁺) and RPMI-1640 were purchased from Invitrogen (Paisley, UK). Ficoll–Paque was obtained from Amersham Biosciences (Buckinghamshire, UK). HTS Multiwell Insert System plates were purchased from BD Biosciences. Recombinant murine IL-5 and eotaxin were obtained from R&D Systems.

Models of airway inflammation

Female and male PI3Kγ KO mice and their wild-type equivalents (5–7 weeks old, groups of n = 10) were intraperitoneally sensitized with a 1 mg/kg dose of OVA, diluted and thoroughly mixed in alum adjuvant, on days 0 and 14. The immune sensitization phase was investigated by removal of blood and spleen from mice on day 19. Antigen challenge was achieved by aerosolizing 5% OVA in saline, or saline alone as control, for 20 min. The BAL with 4 × 0·3 ml PBS was performed at various time-points following challenge. Differential cell counts were determined by cytospin of 200 μl BAL fluid, followed by Dif-Quik staining. Total cell numbers were counted on a haemocytometer following centrifugation at 300 g for 10 min, removal of BAL supernatant and resuspension of the cell pellet in 400 μl methyl violet stain (0·01% methyl violet in 1·5% acetic acid). Chemokine and cytokine content was determined in BAL fluid supernatant by ELISA.

Isolation of CD4⁺ T cells, dendritic cell feeders and the assessment of antigen-specific T-cell responses

Wild-type BALB/c and PI3Kγ mice were immunized with an intraperitoneal injection of 100 μg/ml OVA/alum (Sigma). To assess T-cell responses within the sensitization phase, mice were killed 10 days after immunization and their spleens were removed. Splenocytes were prepared by mashing through a nylon sieve, and were treated with red cell depletion buffer (Sigma). CD4⁺ T cells were then positively selected using the Miltenyi MACs system according to the manufacturer’s instructions. Murine bone marrow-derived dendritic cells were prepared as described previously.⁴⁵ Dendritic cells were treated with 200 μg/ml OVA for 2 hr at 37°, after which they were washed twice. Then, 5 × 10⁵ CD4 T cells were mixed with either ovalbumin-treated or untreated bone marrow-derived dendritic cells at a 1 : 32 and 1 : 64 ratio in round-bottom, 96-well cell culture plates and incubated for 3 days at 37°. Each well was then pulsed with 1 μCi thymidine (Amersham) and incubated overnight, after which the plate was frozen. The cells were harvested with a Packard Filtermate 196 harvester and radioactivity was read on a Topcount scintillation counter (Perkin-Elmer, Shelton, CT). Proliferation is expressed as counts per min.

To assess T-cell responses following OVA–antigen challenge, lung-specific draining lymph nodes (posterior mediastinal) were removed at the time of BAL (24 hr postchallenge). Whole lymph node cell populations were prepared by mashing through a nylon sieve, washing and resuspension at 5 × 10⁵ cells/ml in growth medium containing 200 μg/ml OVA or control. Cells were then cultured in round-bottom, 96-well cell culture plates for 6 days at 37°, then proliferation was assessed as described above.

Immunoglobulin, cytokine and chemokine assays

Serum levels of total IgE were measured by OptEIA ELISA kit. Ovalbumin-specific IgE and IgG1 ELISAs were optimized in-house using commercially available antibody pairs, and analysed against a standard curve generated from the pooled serum of high-responding day 19 samples. Levels of murine CXCL-1 keratinocyte chemokine (KC), macrophage inflammatory protein 2 (MIP-2), IL-5 and eotaxin in BAL fluid were quantified using commercially available ELISA kits.

Eosinophil chemotaxis assay

Chemotaxis/migration was determined in an assay system previously described for neutrophils.⁴⁶ Eosinophil-enriched cell populations were obtained from the femur bone marrow of WT and KO mice for chemotaxis studies. Following removal of the distal epiphyses, bone marrow was flushed from the femurs using a syringe containing cold PBS (without Ca²⁺ and Mg²⁺). Cells were then centrifuged at 300 g and resuspended at 1 × 10⁶ cells/ml in supplemented RPMI-1640 containing 5 ng/ml IL-5. Cultures were restimulated every 2 days with fresh RPMI/IL-5 until day 8 when purity was assessed by Dif-Quik analysis. For assay, cells were washed and resuspended at 5 × 10⁵ cells/ml in assay buffer (RPMI + 2% fetal bovine serum). Eotaxin stimulus was diluted in assay buffer to 30, 10 and 3 nm and 1-ml aliquots were placed in the lower chambers of an HTS Multiwell Insert System. Buffer only was included as a negative control, and 500 : 500 μl buffer : cell suspension was also included to represent maximal chemotaxis. The 3-μm pore filter insert plate was then placed in the lower chambers, and 500 μl of cell suspension added (buffer only to maximal wells). For inhibition studies, plates were then incubated at 37°, 5% CO₂ for 90 min. Inserts were then removed and the number of cells that had moved across the filter membrane was quantified by FACS analysis. Autofluorescence was measured, firstly in a chemokine responsive sample (i.e. WT), from which an eosinophil population was gated based on size and granularity, with approximately 10 000 events/cells counted in a positive (WT) control. This gate was then superimposed onto a maximal sample to ascertain the maximum number of eosinophils that could have moved across the filter, so providing a percentage value of eosinophils that did so in the responsive sample.

Statistical analysis

Results are expressed as mean ± SEM. Statistical significance (P < 0·05) was determined using a one-tailed, unpaired Student’s t-test.

Results

Role of PI3Kγ during immune cell influx into the lung in response to antigen challenge

Immune cell influx into the BAL was assessed in OVA/alum sensitized WT and PI3Kγ KO mice following OVA or saline challenge. The vast majority of cells present in a saline-challenged lung are alveolar macrophages. Few infiltrating cells could be detected at the 2 and 6 hr time-points following antigen challenge (data not shown) whereas up to 5 × 10⁵ total cells were observed by 24 hr in wild-type mice. However, in PI3Kγ KO mice, the number of BAL cells was 2555, reduced by 87% (Fig. 1–‘Total’). Morphological analysis revealed that the infiltrating cells in WT mice were mainly eosinophils (40%), macrophages (31%) and neutrophils (27%), and low numbers of lymphocyte (2%). Mice lacking PI3Kγ displayed a profound reduction in all cell types – neutrophils were reduced by 87%, near complete inhibition of macrophages and lymphocytes and, most importantly, eosinophil numbers were inhibited by over 85% (Fig. 1).

Figure 1.

Cell influx into the bronchoalveolar lavage (BAL) of wild-type (WT) and phosphoinositide 3-kinase gamma knockout (PI3Kγ KO) mice. Numbers of neutrophils, macrophages, lymphocytes and eosinophils in sensitized WT (solid bars) and PI3Kγ KO (hatched bars) mice challenged with saline control (grey) or ovalbumin (OVA; black). Data represents four experiments containing group sizes of 10. Significant differences (Student’s t-test) compared to the saline-treated WT response are shown (*P < 0·05, **P < 0·005, ***P < 0·0005).

Role of PI3Kγ during the generation of antigen-specific T-cell and B-cell responses

To investigate the role of PI3Kγ during the sensitization phase of an adaptive immune response, the generation and function of antigen-specific T cells were studied. Spleens were removed and pooled from wild-type and PI3Kγ KO mice 10 days after an OVA/alum sensitizing injection. CD4 T cells were separated using positive selection by MACS, with resulting purities of > 95% based on CD3/CD4 FACS analysis, (data not shown). Bone marrow-derived dendritic cells were differentiated in vitro by incubation for 8 days in granulocyte–macrophage colony-stimulating factor, at which point > 98% were CD11c⁺ (data not shown). These were then loaded with OVA or saline for 2 hr, and subsequently placed in culture with wild-type or PI3Kγ KO T cells at ratios of 1 : 32 or 1 : 64. T-cell responsiveness was determined by the extent of proliferation after 4 days. Dendritic cells cultured alone failed to proliferate, whether loaded with ovalbumin or saline. Compared to saline-loaded controls, wild-type and PI3Kγ KO T cells showed a two- to fivefold increase in proliferation when incubated with OVA-presenting dendritic cells (Fig. 2a). No significant differences were observed between wild-type and PI3Kγ KO proliferation, suggesting that antigen-specific T-cell responses during sensitization were normal in the PI3Kγ KO mice (Fig. 2a).

Figure 2.

Generation of antigen-specific T-cell responses in wild-type (WT) and phosphoinositide 3-kinase gamma knockout (PI3Kγ KO) mice. (a) CD4⁺ T cells were isolated on day 19 from the spleens of sensitized WT (solid bars) and PI3Kγ KO (hatched bars) mice, and placed in culture with ovalbumin (OVA)-pulsed or saline-pulsed WT dendritic cells at ratios of 1 : 32 (grey) and 1 : 64 (black). Proliferation was measured by [³H]thymidine incorporation after 4 days. (b) Whole cell populations were isolated from posterior mediastinal lymph nodes of OVA-challenged WT (solid bars) and PI3Kγ KO (hatched bars) mice, and placed in culture with OVA or saline. Proliferation was measured by [³H]thymidine incorporation after 7 days. Data represents four experiments containing group sizes of 10. Data represents three experiments containing triplicate cultures.

To achieve a site-specific inflammatory response to antigen, T cells must also respond in the appropriate draining lymphoid tissue. Posterior mediastinal lymph nodes, which drain the lung, were removed following BAL. Whole cell populations were then cultured with OVA and proliferation was assessed to determine responsiveness. A threefold increase in proliferation was observed in WT lymph node cells cultured with OVA compared with the saline control. Near identical levels of proliferation were observed in PI3Kγ KO cells, suggesting that antigen-specific T-cell responses following challenge were normal in these mice (Fig. 2b).

Another major function of CD4 T cells within an immune response is ‘helping’ B cells to produce immunoglobulins of the appropriate specificity and class. Following a Th2-biased stimulus such as OVA/alum, high levels of total and OVA-specific IgE, as well as OVA-specific IgG1 should be produced in vivo.⁴⁷ As a measure of the function of both T cells and B cells, serum samples were taken from mice on day 19 and were assayed for immunoglobulin content. Up to 100-fold increases in total IgE levels were seen in both wild-type and PI3Kγ KO mice compared to saline/alum sensitized controls, with no significant difference in absolute values observed between OVA/alum-sensitized groups (Fig. 3a). A near identical pattern was also observed in both OVA-specific IgE (Fig. 3b) and IgG1 (Fig. 3c), though the increases compared with saline/alum-sensitized controls were up to 800-fold and 1 × 10⁶-fold respectively.

Figure 3.

Immunoglobulin responses in wild-type (WT) and phosphoinositide 3-kinase gamma knockout (PI3Kγ KO) mice. Serum samples from saline (grey) or ovalbumin (OVA; black)-sensitized WT (solid bars) and PI3Kγ KO (hatched bars) mice were taken on day 19 and analysed for (a) total immunoglobulin E (IgE), (b) OVA-specific IgE and (c) OVA-specific IgG1. Data represent three experiments containing group sizes of 10.

PI3Kγ and the production of BAL chemokines and cytokines

Following pulmonary antigen challenge in a systemically sensitized animal, appropriate chemokines and cytokines are produced to attract effector cells such as neutrophils and eosinophils to the site.⁴⁸ To determine the effect that lacking PI3Kγ has on the ability of mice to generate the appropriate site-specific chemokine/cytokine milieu, levels of neutrophil and eosinophil chemokines were measured in the BAL over a time–course following OVA challenge. The neutrophil chemokines MIP-2 and KC – murine equivalents of human IL-8 and growth-related oncogene-α (GRO)– peaked in concentration at 2 hr following challenge, were still evident at 6 hr, but had dropped to near saline challenge control levels by 24 hr. No significant differences between wild-type and PI3Kγ KO MIP-2 (Fig. 4a) or KC (Fig. 4b) production could be detected at any time-point. The Th2-derived and eosinophil-active cytokine IL-5 was not evident until 24 hr after challenge, with no difference observed between wild-type and PI3Kγ KO mice (Fig. 4c). Eotaxin, another eosinophil chemokine, was observed at both 6-hr and 24-hr time-points. There is a consistent trend toward lower levels in mice lacking PI3Kγ, although because of the generally low magnitude of response these differences do not achieve significance (Fig. 4d), and therefore do not explain the dramatic 85% reduction in eosinophils observed in BAL of PI3Kγ KO mice (Fig. 1).

Figure 4.

Pulmonary cytokine/chemokine levels in wild-type (WT) and phosphoinositide 3-kinase gamma knockout (PI3Kγ KO) mice. Bronchoalveolar lavage (BAL) chemokine/cytokine levels from sensitized WT (solid lines) and PI3Kγ KO (hatched lines) mice challenged with saline control (grey) or ovalbumin (OVA; black). Neutrophil chemotractants are shown in (a) macrophage inflammatory protein 2 (MIP-2) and (b) murine CXCL-1 keratinocyte chemokine (KC). Eosinophil chemotractants are shown in (c) interleukin-5 (IL-5) and (d) eotaxin. Data represents four experiments containing group sizes of 10.

PI3Kγ-dependent eosinophil chemotaxis

Mice lacking PI3Kγ displayed no apparent dysfunction in their ability to mount a normal antigen-specific allergic immune response with regard to both T-cell and B-cell parameters (Figs 2 and 3). Moreover, upon antigen challenge, no significant differences in the inflammatory cytokine/chemokine milieu were observed between WT and KO mice (Fig. 4). However, a striking reduction in the influx of all immune effector cell populations was observed, the most prominent of them being eosinophils (Fig. 1). It was therefore necessary to determine whether the lack of eosinophil influx into KO mouse BAL fluid was the result of reduced or ablated migratory function within the population itself. Modified analysis by FACS in an HTS Multiwell Insert System was used to investigate the chemotactic function of WT versus KO eosinophils. Eosinophils were matured from bone marrow in vitro with IL-5 until > 70% pure (data not shown). In response to varying concentrations of eotaxin stimulus, a dose response of chemotaxis was observed in WT eosinophils, ranging from 51% (equating to approximately 10 000 recorded events/cells) at 30 nm down to 5% (∼1000 cells) in unstimulated controls, as compared to total eosinophil number (∼20 000 cells). However, < 10% (∼2000) eosinophil movement was observed for PI3Kγ KO cells across the stimulus range – demonstrating a consistent reduction in chemotactic index by up to 87% (Fig. 5).

Figure 5.

Wild-type (WT) and phosphoinositide 3-kinase gamma knockout (PI3Kγ KO) eosinophil chemotaxis to eotaxin. Chemotaxis of bone marrow-derived eosinophils from WT (solid bars) and PI3Kγ KO (hatched bars) mice responding to eotaxin. Data represents three experiments containing duplicate cultures. Significant differences (Student’s t-test) compared to the WT response are shown (*P < 0·05, **P < 0·005, ***P < 0·0005).

Discussion

The novel data presented in this study are the first to demonstrate a profound reduction in eosinophilia, as well as impaired neutrophil and macrophage influx into the BAL fluid of PI3Kγ KO mice within a model of acute allergic pulmonary inflammation. Investigation of sensitization phases of allergic immune responses shows no involvement of PI3Kγ in the generation of appropriate T- and B-cell responses. Furthermore, the generation of site-specific T-cell responses and the formation of chemokine gradients within the lung also appear normal in the absence of PI3Kγ. Finally, we identify the level of dysfunction in vitro as the eosinophil population itself, demonstrating near complete PI3Kγ dependency for migration to eotaxin.

Within an OVA-specific model of asthma, the ability of effector cell populations to migrate into the lung following antigen challenge is profoundly inhibited by a lack of PI3Kγ. By 24 hr, neutrophils and macrophages form approximately half the infiltrating cells – levels which are reduced by up to 87% in PI3Kγ KO mice. This apparent dysfunction in neutrophil and macrophage chemotaxis has not been previously described within a disease model of asthma, but does fit with data generated by ourselves and others within in vitro systems, as well as more mechanistic models of chemokine-stimulated pulmonary inflammation and septic peritonitis.^24–26,37 The largest cell population present in the wild-type BAL fluid are eosinophils – levels which are reduced by 85% in mice lacking PI3Kγ. However, at what level within the sensitization or effector stages of this disease model, and to what extent PI3Kγ plays a role, was key to determining whether the γ isoform represents a viable target for the treatment of asthma.

The specific roles played by the class I PI3K isoforms in T-cell and B-cell responses has been under intense speculation, with often opposing hypotheses proposed, based on differing experimental conditions. Phenotypic analysis of two distinct PI3Kγ KO mouse strains, while not in direct disagreement, put a differing emphasis on the impact of lacking this isoform in T-cell development and function. Hirsch et al. noted a reduction in blood lymphocyte numbers and thymus size, which does correlate with a role for PI3Kγ in thymocyte development proposed by Sasaki et al.^25,26 Further studies using an activating p65^PI3K mutant of the class 1A p85 regulatory subunit could rescue T-cell development in PI3Kγ KO mice,⁴⁹ observations supported by p110^ABD (adaptor binding domain) transgenic mice whose constitutively active class 1A subunit resulted in accumulation of mature thymocytes.⁵⁰ Furthermore, the recently generated PI3Kγ/δ double knockout mouse strain proposes a combined developmental dependency of T cells on both isoforms.⁵¹ However, the role(s) of the different PI3K isoforms in T-cell activation, chemotaxis and function have proved more controversial. It has been suggested that T cells employ migratory mechanisms distinct from PI3K-dominated pathways attributed to monocyte and neutrophil chemotaxis.⁵² Sasaki et al. proposed PI3Kγ to be important in TCR signalling following WT versus KO in vitro experiments investigating proliferation and cytokine production in anti-CD3-stimulated T cells demonstrated up to 80% impairment. The hypothesis was also tested in vivo, with decreases observed in footpad swelling models and IgG1 titres in response to hapten.²⁶ T-cell chemotaxis is a process essential for migration within lympoid organs as well as to the site of inflammation, and is therefore central to function. In vitro transwell studies have suggested a role for PI3Kγ in chemotactic responses to CCL19 (MIP-3β) or CCL12 (monocyte chemotactic protein-5).⁵³ However, other studies have notably found no effect of PI3Kγ inhibition, particularly with regard to Th2 cell migration following stimulation with CCL22 (macrophage-derived chemokine) or CCL17 (thymus and activation-regulated chemokine), which require activation of a phospholipase-C-dependent, calcium-independent mechanism.⁵⁴ Other transgenic studies have proposed DOCK-2 (down-stream of Crk-180 homologue-2) as a likely candidate to explain normal chemotaxis in T cells lacking PI3Kγ.⁵⁵ In this study, rather than attempting to define the potential contribution of PI3Kγ within in vitro mechanistic experiments, which can often omit compensatory mechanisms relevant in vivo, we chose to ask the question of whether mice lacking this isoform could mount an appropriate disease-relevant immunological response. Using ex vivo antigen-specific T-cell proliferation and B-cell production of antigen-specific immunoglobulins, we observed that PI3Kγ KO mice can generate responses of the same magnitude and quality as WT controls, demonstrating that T-cell function with regard to activation and chemotaxis to and within lymphoid organs is independent of PI3Kγ signalling. Our further observations that PI3Kγ KO B cells were able to respond appropriately to T-cell education is in keeping with current literature which has identified class IA isoforms, particularly PI3Kδ, to be principal mediators in B-cell development and function.^41,53

Upon antigen challenge of sensitized individuals, effector mechanisms must be targeted to the relevant site. Vital to this stage is the release of chemokines in response to antigen and stimulation of the innate arm of the immune system, followed by influx of antigen-specific T cells representing the adaptive immune system, both of which function in concert to augment and perpetuate the response, resulting in the influx of effector cells.⁵⁶ Neutrophils are the principal effector cell population of innate responses, and are consequently drawn to the site of inflammation relatively early in response to rapidly produced specific chemokines. Our studies show a rapid production of MIP-2 and KC, in keeping with the literature, the likely source of which being airway epithelial cells.⁵⁷ Importantly, no involvement of PI3Kγ was apparent for the generation of this chemokine gradient.

The key effector cell population in asthma following stimulation of adaptive immunity are eosinophils. We demonstrate that not only does systemic immune sensitization appear normal, but also site-specific T-cell responses following antigen challenge are unaffected by the absence of PI3Kγ. Consequently, chemokines more specific for eosinophils appear later, including eotaxin and IL-5. Eotaxin, probably also epithelial cell-derived, was apparent by 6 hr and peaked by 24 hr, in keeping with the later influx usually seen with this population (personal observations).⁵⁸ By 24 hr, OVA-specific T cells could be present and producing IL-5, in accordance with our findings. Although no significant difference between WT and PI3Kγ KO eotaxin or IL-5 levels was observed, a clear trend toward reduced KO levels was evident. The likely reason for these minor reductions could be the potential multiple sources of these chemokines. Both eotaxin and IL-5 can also be produced by eosinophils themselves, and therefore a trend toward impaired production is likely to be a consequence of the profound inhibition of eosinophil migration into the BAL.^59,60 Taken together, it can be concluded that lacking PI3Kγ does not affect the ability of either the innate or adaptive immune systems to generate the appropriate sensitization phases or mount an inflammatory site-specific response and appropriate chemotactic milieu.

Dysfunction in the responsiveness of PI3Kγ KO eosinophils to eotaxin was unequivocally demonstrated within an in vitro chemotaxis assay. Such experiments have been considered notoriously difficult in the past, because of technological limitations on access to pure and healthy eosinophils for chemotaxis experiments. Together with the use of isoform non-specific compound tools could help to explain conflicting data on the relevance of PI3K in eosinophil migration.⁶¹ However, by using flow cytometry to analyse chemotactic parameters within a transwell system, a specific and consistent degree of near complete PI3Kγ dependency in eosinophil chemotaxis can be demonstrated. The role of this isoform in eosinophil chemotaxis has not been specifically addressed previously, and places this effector cell population within an expanding family of cells, including neutrophils, macrophages and mast cells, whose function within a disease state is reliant on the PI3Kγ.⁴²

These studies support the inhibition PI3Kγ as a viable strategy for the treatment of asthma. Such findings are of particular relevance given the disappointing lack of success from cytokine and anti-cytokine/chemokine therapies. In particular, clinical studies of anti-IL-5 in asthmatic patients have achieved dramatic reductions in circulating eosinophils, whereas the impact on eosinophil migration into the lung was far less striking.^62,63 Possible explanations for these studies come from murine KO experiments. Interleukin-5 KO mice display reduced blood and BAL eosinophil levels;⁶⁴ however, treatment of WT mice with anti-IL-5 antibody, although still able to dramatically reduce blood eosinophilia, could only mildly inhibit migration into the lung.⁶⁵ However, eotaxin KO mice developed normal blood eosinophilia in a similar model, but did not mount a full response in the lung.⁶⁵ Furthermore, a study of eosinophil instillation into the trachea of IL-5 KO mice demonstrated eosinophil survival as well as the reconstitution of the asthma phenotype upon antigen challenge.⁶⁶ In addition, a recent study has investigated the role of PI3Kγ within a model of allergic pleurisy, demonstrating reduced eosinophilia in PI3Kγ KO mice late in the inflammatory response. Though initially encouraging, the proposed mechanism – that of reduced IL-5 production – was based on in vivo inhibition with isoform non-specifc inhibitors, which leaves some conclusions open to interpretation within this unusual model.⁶⁷ It seems that migration to the lung, an eotaxin-dependent process, is more clinically relevant than development of blood eosinophilia, a process more dependent on IL-5.

A growing body of data suggests that the profound influence PI3Kδ has on the development of a functional T/B-cell immune response would make the δ isoform an attractive therapeutic target. Studies have used both genetic⁶⁸ and compound^69,70 strategies in such demonstrations, indicating reduced T-cell and B-cell function, cytokine and chemokine production as well as subsequent airway inflammation. However, an intriguing recent study by Doi et al. used both p85^−/− mice and the PI3Kδ-specific inhibitor IC87114 to propose the δ isoform as a negative regulator of class switching to IgE and that inhibition may lead to raised levels – an eventuality not conducive to asthma treatment.⁷¹ Furthermore, it could also be argued that broad inhibition of such fundamentally important elements of the immune system could lead to adverse effects that would be intolerable for the majority of asthmatic patients, and so limit the use of PI3Kδ inhibition as a therapeutic strategy. More research is required in this area to assess PI3Kδ inhibitors as therapeutics.

Taken together, these data demonstrate the fundamental role of PI3Kγ in the migration of eosinophils, as well as neutrophils and macrophages, into the lung in a model of acute pulmonary inflammation. Combined with the impaired mast cell phenotype observed in PI3Kγ KO mice,⁴² our study presents a strong case for the development of a PI3Kγ blocking strategy to specifically modulate the effector stages of asthma.