Differential dependence of eosinophil chemotactic responses on phosphoinositide 3-kinase (PI3K)


Gordon Dent BSc PhD
School of Life Sciences
Huxley Building
Keele University
Staffordshire ST5 5BG


Background:  Control of eosinophil migration to sites of inflammatory responses is a potentially therapeutic intervention in diseases such as bronchial asthma. Chemoattractants, their receptors and the associated signalling pathways may, therefore, be important targets for novel therapeutics. While several potentially important chemoattractants have been identified, the signalling pathways mediating their actions are incompletely understood.

Aims of the study:  The role of phosphoinositide 3-kinase (PI3K) in responses of human eosinophils to two important eosinophil chemoattractants – platelet-activating factor (PAF) and eotaxin (CCL11) – was studied to determine whether this enzyme activity might be crucial for eosinophil migration.

Methods:  Eosinophils were isolated from atopic donor blood by immunomagnetic selection. Chemotaxis was assayed in a 96-well blind-chamber cell fluorescence assay. Respiratory burst and leukotriene C4 secretion were also assayed.

Results:  Two PI3K inhibitors, wortmannin and LY294002, caused concentration-dependent inhibition of PAF-induced eosinophil chemotaxis (IC50 = 0.54 nM and 0.15 μM, respectively) but exhibited at least 100-fold lower potency against eotaxin-induced responses (IC50 = 48 nM and >100 μM, respectively), indicating that these responses were not dependent upon PI3K. Wortmannin and LY294002 also inhibited PAF induced respiratory burst but not PAF-induced LTC4 secretion.

Conclusions:  We conclude that PI3K-dependence varies with stimulus and response, and that eotaxin-induced eosinophil migration is not controlled by PI3K. This may indicate a limit to the potential of PI3K inhibitors to suppress tissue eosinophilia in diseases such as asthma.

Eosinophils are the characteristic infiltrating cells at sites of IgE-dependent reactions. Through their ability to generate a range of cytokines, basic proteins, lipid mediators and reactive oxygen species eosinophils contribute to the pathology of chronic inflammatory diseases such as bronchial asthma, a condition in which these cells have been defined as an important therapeutic target (1). Eosinophils are capable of participating in multiple facets of asthma pathology, including persistent mast cell activation and remodelling of both the bronchial microvasculature and the submucosal structures of the airway wall (2). The regulation of migration of eosinophils to the inflammatory focus is a critical stage in the processes of chronic inflammation that affect asthmatic airways. Understanding of the mechanisms by which eosinophil migration is stimulated is, therefore, important in defining potential targets for therapeutic interventions, since the chemoattractants, receptors and cell signalling pathways involved in eosinophil attraction and locomotion may each be susceptible to inhibition (3).

Two major chemoattractants for eosinophils are the ether-linked phospholipid, platelet-activating factor (PAF), and the CC chemokine, eotaxin 1 (CCL11) (4, 5). Each of these acts via a single class of receptor – PAF-R and CCR3, respectively – to evoke a variety of responses in the cell. While chemotactic responses to both agents have been shown to be dependent upon activation of mitogen-activated protein (MAP) kinases, it remains unclear whether the signalling events occurring upstream of these enzymes are common to chemotactic agonists at the two classes of receptor (6–8).

In the light of suggestions that phosphoinositide 3-kinase (PI3K) may be a critical enzyme in PAF and chemokine signalling (9, 10), we report here on the role of PI3K in eosinophil chemotactic responses to PAF and eotaxin.

Materials and methods

Eosinophils were isolated from heparinized peripheral blood of healthy, atopic donors and fluorescently labelled, as described (11). Labelled eosinophils (4 × 106/ml) were preincubated with or without the PI3K inhibitors, wortmannin and LY294002 (both from Merck Biosciences, Nottingham, UK), for 30 min, after which chemotaxis assays were performed in a 96-well microplate system (Neuroprobe, Inc., Gaithersburg, MD, USA), essentially as described (11), using PAF (Merck Biosciences) and eotaxin (Albachem, Edinburgh, UK) as stimuli. Chemotactic responses over 1 h are expressed as chemotactic index (CI: number of cells migrating in response to stimulus/number of cells migrating in response to medium). In further experiments, unlabelled eosinophils were incubated with or without wortmannin and LY294002 prior to stimulation with PAF for assay of respiratory burst and leukotriene C4 (LTC4) secretion, as described (12). Data were analysed by unpaired Student t-test or by repeated-measures anova with post hoc Dunnett's test, as appropriate.


Eosinophils exhibited chemotactic responses to both PAF and eotaxin, with the responses to 30-nM PAF and 30-nM eotaxin being sub-maximal and similar in magnitude [geometric mean (95% confidence interval) CI = 2.5 (2.0–3.0) and 2.9 (2.3–3.8), respectively; P = 0.44]. Responses to PAF were inhibited in a concentration-dependent manner by both wortmannin and LY294002, with respective IC50 values of 0.54 (0.30–0.96) nM and 0.15 (0.09–0.23) μM (Figure 1). In contrast, however, wortmannin inhibited chemotactic responses to eotaxin only at the highest concentration tested [IC50 = 48 (23–100) nM) and LY294002 had no effect at all at concentrations up to 100 μM (Figure 1).

Figure 1.

The effects of PI3K inhibitors on eosinophil chemotactic responses to PAF and eotaxin. Eosinophils from atopic non-asthmatic donors were preincubated with or without the indicated concentrations of wortmannin (A) or LY294002 (B) for 30 min before being added to the upper wells of chemotaxis chambers. PAF and eotaxin (each at 30 nM) were used as chemotactic stimuli. Data are shown as mean ± SEM from six experiments conducted in triplicate for each stimulus/inhibitor combination. *P < 0.05; **P < 0.01 compared with control response in the absence of inhibitor.

PAF caused significant stimulation of respiratory burst and LTC4 secretion at higher concentrations than those inducing chemotaxis, with 5 μM being a sub-maximally effective concentration for both responses. Both PI3K inhibitors caused profound suppression of PAF-induced respiratory burst, with superoxide anion (O2·) release reduced to baseline levels at concentrations of 1 nM wortmannin and 10 μM LY294002 (Figure 2A). In contrast, wortmannin had no significant effect upon PAF-induced LTC4 secretion (Figure 2B). Since wortmannin exhibited no effect, it was considered unnecessary to study the less potent drug, LY294002. Stimulation of respiratory burst by eotaxin was highly variable and O2· release in the presence of 100-nM eotaxin (the maximally effective concentration) was not significantly higher than baseline. Eotaxin did not stimulate LTC4 secretion (data not shown).

Figure 2.

The effects of PI-3K inhibitors on respiratory burst and leukotriene C4 secretion in PAF-stimulated human eosinophils. Eosinophils were preincubated with or without inhibitors for 30 min prior to addition of 5-μM PAF. Respiratory burst (A) was measured over 15 min; LTC4 secretion (B) was measured over 5 min. Data are shown as mean ± SEM from four experiments conducted in triplicate (A) or duplicate (B). +P < 0.05, ++P < 0.01 compared with medium; *P < 0.05; *P < 0.01 compared with PAF 5 μM.


PI3K has been identified as a crucial enzyme activity in leukocyte function, with the G protein-coupled PI3Kγ isoforms, in particular, playing an important role in inflammation (13). The induction of actin polymerization, a necessary step in cell migration, has been shown to be dependent upon PI3Kγ in Cos7 cells transfected transiently with the chemoattractant formyl peptide receptor and the p110γ/P101 form of PI3K (14). However, while formyl peptide-induced O2· generation is abolished in the neutrophils of PI3Kγ knockout mice, chemotactic responses of these cells to formyl peptide or interleukin 8 (CXCL8) are only partially inhibited. Although both respiratory burst and chemotactic responses to formyl peptide are dependent upon activation of phospholipase C (PLC) β2/β3, as evinced by the absence of the responses in PLCβ2/inline image neutrophils, it is apparent that alternative PI3Kγ-independent pathways exist downstream of PLCβ that mediate chemotactic responses but not respiratory burst in native neutrophils (15).

Thus, different responses to the same stimulus may involve different signalling pathways. For example, in human eosinophils, respiratory burst is partially dependent upon protein kinase C (PKC), with a parallel PKC-independent pathway also contributing, while generation of thromboxane A2 and LTC4 is entirely PKC-independent (12). In the present study, PI3K was found to be involved in PAF-induced respiratory burst activation, as evinced by the potent inhibition of O2· generation by wortmannin and LY294002, suggesting this as a likely alternative signalling pathway for NADPH oxidase activation. This finding supplements the demonstration that PAF-induced eosinophil adhesion and CD11b/CD18 expression are dependent upon PI3K (16). However, PAF-induced LTC4 secretion was insensitive to wortmannin, indicating that PI3K is not essential for this pathway of arachidonic acid metabolism.

Similarly, common responses to different stimuli may also involve different pathways. This is apparent in the differential sensitivity of PAF- and LTB4-induced eosinophil respiratory burst and formyl/complement peptide and LTB4/IL8-induced neutrophil chemotactic responses to PI3K inhibitors (17, 18). Although chemotactic responses of human eosinophils to both stimuli used in the present study – PAF and eotaxin – have been demonstrated to be dependent upon the activation of Erk1/2 MAP kinases (6, 7), it is possible that events upstream of MAP kinase activation may differ between the two stimuli. Whilst activation of Erk1/2 in response to PAF is blocked by inhibition of PI3K (6), this has not been demonstrated for eotaxin, whose actions appear to involve p38 MAP kinase (8). Significant stimulation of respiratory burst by eotaxin could not be demonstrated in the present study, despite a previous report that this response is evoked (19). While Tenscher et al. used the cytochrome c assay for O2·, the data obtained in this assay were not reported. We found a concentration–response relationship for increased O2· which paralleled that observed by Tenscher et al. in a lucigenin-enhanced chemiluminescence assay (19) but the increases were small and did not achieve statistical significance at any concentration up to 1 μM eotaxin (P > 0.05; n = 6).

The ability of eotaxin to activate PI3K in eosinophils is implied by the sensitivity of CCR3-dependent actions, such as the production of leukotriene-synthesizing lipid bodies, to inhibition by wortmannin and LY294002 (20). However, despite this, the chemotactic response to eotaxin was unaffected by either PI3K inhibitor. This may suggest that different pathways are activated by PAF-R and CCR3, leading to induction of Erk activation and cell locomotion via different routes. Both Erk activation and chemotaxis stimulated by PAF are inhibited by nanomolar concentrations of wortmannin and micromolar concentrations of LY294002 (6). A similar pattern has been observed for the CCR6 agonist, liver and activation-related chemokine/macrophage inflammatory protein 3α (LARC/MIP3α, CCL20) but has not been reported for CCR3 agonists. Since, however, CCR3 agonists do evoke PI3K-dependent responses (20), it may be necessary to consider the possibility that these occur through activation of a different form of the enzyme, and that CCR3-dependent chemotactic responses involve PI3K-independent pathways. Curnock et al. have proposed that multiple PI3Ks are highly involved in chemokine signalling but also raise the possibility that forms other than the widely studied PLCβ-activated class I enzyme, PI3Kγ, contribute to their actions (10). Alternatively, CCR3 activation may recruit additional signalling pathways which interfere with PI3K-dependent responses, so that PI3K activation is not reflected in a functional response. It has been reported that neutrophil chemoattractants can be divided into two classes. Thus, ‘end-stage’ chemoattractants, including complement anaphylotoxin C5a and N-formylmethionyl peptides, act through activation of p38 MAP kinase while ‘intermediary’ chemoattractants such as IL8 and LTB4 act through PI3K. Moreover, p38 activation causes suppression of PI3K-mediated signals (18). Evidence of a role for p38 MAP kinase in the chemotactic action of eotaxin may indicate a similar division of eosinophil chemoattractants (8). Study of the activation of specific isoenzymes through different receptor classes will be necessary to elucidate the precise targets through which eosinophil migration may be inhibited.

In conclusion, while PI3K is clearly involved in a number of eosinophil responses to PAF, the induction of cell migration by a distinct, important chemoattractant, eotaxin, is not dependent upon PI3K. Given that eotaxin appears to play an important role in eosinophil recruitment to the irways in severe asthma (5), PI3K inhibitors may have only a limited effect on the development of tissue eosinophilia in these conditions.


We thank Gilbert Angco for help with volunteer recruitment and Drs Rory O'Donnell, Mark Steel, Paul Beckett and Rami Salib for blood sampling.