EOSINOPHIL APOPTOSIS: MECHANISMS and CLINICAL RELEVANCE IN ASTHMATIC and ALLERGIC INFLAMMATION

The eosinophil possesses a considerable array of histotoxic substances which contribute to the initiation and maintenance of the allergic inflammatory response; these include cytotoxic granule proteins, cytokines and lipid mediators, each of which plays a substantial though differing role. Although eosinophils are thought to be important in adaptive immunity to helminthic parasitic worms, there is now a wide consensus that eosinophil-derived products make a major contribution to allergic and asthmatic disease (Walsh, 1999). Infiltration of the bronchial mucosa by large numbers of proinflammatory cells is characteristic of the airway inflammation which is central to the pathogenesis of all forms of asthma. Despite being a minority constituent of circulating leucocytes, the eosinophil is a prominent feature of this infiltrate. Indeed, it has recently been suggested that there is as much as a 50- to 100-fold increase in the accumulation in eosinophils over neutrophils in the airways of patients with asthma (Wardlaw, 1999). Importantly, there are many studies which have demonstrated a correlation between asthma severity and levels of eosinophils and their products in blood, induced sputum and in bronchial biopsy or lavage samples. Moreover, eosinophil granule-derived mediators are heavily implicated in bronchial epithelial cell damage; this leads to cilial dysfunction and cell loss, both of which are thought to be central to the development of bronchial hyper-responsiveness, one of the cardinal features of asthma (Bousquet et al, 1990). Recently, attention has focused on increasing our understanding of the factors involved in eosinophil clearance, not least because dissection of the factors which facilitate the apoptosis-induced clearance of eosinophils from the lung and their subsequent engulfment by phagocytes is a rational therapeutic aim for asthma. In the time since my previous reviews of the subject of eosinophil apoptosis (Walsh, 1997a, b), there has been a great increase in our knowledge, with the publication of some interesting and insightful studies. The current review is aimed at updating the field of eosinophil apoptosis and clearance – including the relationship with prolonged eosinophil survival – and will also discuss the relevance of these processes to allergic and asthmatic inflammation. Our knowledge of the complex progression involved in eosinophil accumulation is now considerable. Critical stages include augmented production of eosinophils in the bone marrow, their increased release into the circulation and their selective accumulation in the conducting airways (Leung, 1998; Walsh, 1999; Wardlaw, 1999). Once eosinophils have entered the tissues, there is no mechanism available by which they can emigrate when their contribution to the allergic inflammatory response is complete. Thus, their removal is dependent upon their death by the tightly controlled mechanism of apoptosis, or programmed cell death, followed by their recognition and phagocytosis by macrophages or resident cells. Apoptosis is fundamental to the maintenance of a viable immune system (Cohen, 1999) and there is an evolving hypothesis that the tissue load of eosinophils in allergic and asthmatic disease is related to inhibition of or defects in the apoptotic process. Eosinophils are terminally differentiated end cells which die by apoptosis when cultured in vitro, being rapidly recognized and ingested as intact cells by autologous macrophages. Engulfment is vital to the resolution of inflammation for two reasons. First, it prevents spillage of the eosinophil's histotoxic contents, as seen in cellular necrosis or cytolysis (Perssen & Erjefalt, 1997). Even in the situation where eosinophils become apoptotic, failure or defects in the phagocytic removal process would result in the release of their proinflammatory contents via secondary necrosis. Indirect evidence for this is provided by the observation that dermal eosinophils in patients with atopic dermatitis do not become apoptotic but undergo cytolytic degeneration, which was shown to be associated with the deposition of eosinophil granule products in the dermis (Cheng et al, 1997). Second, engulfment of apoptotic eosinophils induces an anti-inflammatory cytokine and mediator secretory profile in the macrophage, i.e. interleukin 10 (IL-10), transforming growth factor β (TGF-β) and prostaglandin E2; this is in direct contrast to ingestion of necrotic eosinophils which results in a proinflammatory cytokine and mediator profile, i.e. release of thromboxane B2 and granulocyte–macrophage colony-stimulating factor (GM-CSF) (Stern et al, 1996). Recognition and engulfment is dependent on the phagocyte receptors recognizing several changes in the membrane of the apoptotic cell which are often termed ‘eat me’ signals (Savill, 1997). To date, the best characterized of these is loss of membrane phospholipid asymmetry in which phosphatidylserine, which is normally confined to the inner leaflet of the cell membrane, is relocated to the outer layer of the plasma membrane (Fadok et al, 1992). Annexin V is a Ca2+-dependent phospholipid-binding protein with a high affinity for phosphatidylserine and when labelled with a fluorescent tag provides a sensitive and early apoptosis signal in eosinophils (Walsh et al, 1998). It is apparent that a number of non-professional phagocytes, including dendritic cells, smooth muscle cells and lung fibroblasts, also have the capacity to recognize and ingest apoptotic cells (Platt et al, 1998). Although the bronchial epithelium is generally considered to be the target for cell damage and loss by eosinophil-derived mediators, we have recently made the novel observation that human small airway epithelial cells also have the capacity to ingest apoptotic eosinophils. This is a specific lectin- and integrin-mediated process which was enhanced by the proinflammatory cytokines IL-1α and tumour necrosis factor α (TNF-α) (Walsh et al, 1999). This observation may have important consequences for the clearance of lung eosinophils in asthma. It has been appreciated for some time that IL-3, GM-CSF and IL-5 enhance eosinophil survival when cultured in vitro for up to 2 weeks, and it might thus be that in vivo eosinophil persistence in the tissues may be prolonged in their presence. Additionally, IL-13 has been shown to enhance eosinophil survival (Luttmann et al, 1996; Horie et al, 1997), an effect enhanced by co-incubation with TNF-α (Luttmann et al, 1999). The significance of these findings is emphasized by ample evidence that these cytokines are present in asthmatic airway tissue (Leung, 1998) and that eosinophils isolated from patients with atopic dermatitis and, to a lesser extent, inhalant allergy displayed enhanced survival compared with normal controls (Wedi et al, 1997). Eosinophil interactions with the proteins of the extracellular matrix are also thought to make a significant contribution to their persistence as a consequence of signalling through integrin receptors mediating autocrine production of GM-CSF, IL-3 and IL-5 (Walsh & Wardlaw, 1997). The finding that bronchial epithelial cells recognize and ingest apoptotic eosinophils is also interesting in the light of evidence demonstrating that under certain circumstances eosinophil persistence in the airways might also be enhanced by their interaction with epithelial cells (Cox et al, 1991) or epithelial cell-derived mediators (Peacock et al, 1999). Moreover, other resident lung cells including mast cells (Levi-Schaffer et al, 1998) or IL-1β-stimulated airway smooth muscle cells (Hallsworth et al, 1998) elaborate GM-CSF, which in turn enhances eosinophil survival, whereas upper airway tissue eosinophils isolated from nasal polyps show enhanced survival when cultured in vitro compared with peripheral blood eosinophils (Ramis et al, 1995). Indeed, an ex vivo system which utilized eosinophilic nasal polyp tissue as a model of allergic inflammation demonstrated delayed eosinophil apoptosis which was associated with the presence of IL-5 (Simon et al, 1998), and nasal polyp tissue from allergic, but not non-allergic subjects, spontaneously released significant quantities of GM-CSF which enhanced eosinophil survival in vitro (Park et al, 1997). Furthermore, eosinophils have been shown to express the membrane receptor CD40, ligation of which resulted in enhanced eosinophil survival as a consequence of autocrine GM-CSF release. This study also demonstrated that tissue eosinophils resident in nasal polyp tissue had a high constitutive expression of CD40 (Ohkawara et al, 1996). The ligand for CD40, CD40L, is expressed by CD4+ T cells which are also present in nasal polyp tissue, which suggests an intriguing potential for a further relationship between eosinophils and T cells. Eosinophil numbers in the airways can be related to a balance between incoming recruited cells, those cells whose viability is enhanced by the apoptosis-inhibiting effects of IL-3, IL-5, GM-CSF or IL-13, and apoptotic death followed by phagocytic clearance. It can be appreciated therefore that the interactions between these different but inter-related processes represent a fine balance which determines the tissue load of eosinophils. These concepts are summarized in Fig 1. Eosinophil pathways of apoptosis-induction (anti-inflammatory) or activation and degranulation (proinflammatory). Glucocorticoids such as dexamethasone can cause a striking reduction in eosinophil numbers in vivo (Schleimer, 1990) and in their inhaled form remain the mainstay of anti-inflammatory therapy in asthma (Barnes, 1998). Although the precise mechanism of action of steroids remains to be determined, glucocorticoids are likely to exert their effects on eosinophils by accelerating their apoptosis and engulfment by lung macrophages (Meagher et al, 1996), by inhibiting the production of survival-enhancing cytokines or both. One recent study has demonstrated that in vitro corticosteroid treatment of eosinophils in nasal polyp tissue sections enhanced their apoptosis induction (Saunders et al, 1999), suggesting that eosinophil apoptosis induction by glucocorticoids might be relevant to their anti-inflammatory effects in asthma. Interestingly, there is evidence that the bronchodilating β-adrenoreceptor agonists block the antiapoptotic effects of corticosteroids on eosinophils (Nielson & Hadjokas, 1998), which might have important implications for the overusage of β2-agonists in asthma therapy. Moreover, the bronchodilator theophylline, which is widely used in asthma therapy, has demonstrable anti-inflammatory effects (Rabe & Dent, 1998) and will also trigger apoptosis in human eosinophils (Ohta et al, 1996). Finally, the rationale for pursuing eosinophil apoptosis as a valid option for targeting asthmatic inflammation is strengthened by the observation that a GM-CSF receptor analogue E21R induced apoptosis in human eosinophils (Iversen et al, 1997). These findings suggest that the GM-CSF receptor actively controls the death as well as the survival of eosinophils, thereby regulating their numbers at sites of allergic inflammation in the tissues. Other drugs which have also been shown to inhibit eosinophil survival by inducing apoptosis include the sulphonylureas (Bankers-Fulbright et al, 1998) and the cyclosporins A and H (Kitagaki et al, 1997), whereas the local anaesthetic lidocaine was also found to induce apoptosis in eosinophils maintained in IL-5 (Okada et al, 1998). More recently, incubation of eosinophils with IL-4 was shown to enhance their constitutive rate of apoptosis and also overcame the antiapoptotic effects of IL-3, IL-5 and GM-CSF (Wedi et al, 1998). The last is an interesting finding given that IL-4 has profound effects in promoting allergic inflammation, most notably as a requirement for the class switch in B cells from IgG to IgE production and, as mentioned, that the closely related cytokine IL-13 enhances eosinophil survival. These findings suggest the existence of a negative feedback mechanism by which IL-4 can exert anti-inflammatory effects via induction of eosinophil apoptosis. Evidence for this notion was provided by a study which demonstrated that an inhaled allergen challenge of mildly asthmatic subjects resulted in a reduction in IL-4-positive cells in inflamed airway tissue (Nonaka et al, 1995). A number of studies have investigated the effects of membrane receptor ligation on the induction of apoptosis in eosinophils. These include monoclonal antibody-dependent ligation of the type II integral membrane signalling receptor CD69 (Walsh et al, 1996) or the tyrosine phosphatase-linked pan-leucocyte receptor CD45 (Blaylock et al, 1999). Other receptors appear to be able to differentially regulate eosinophil survival and/or apoptosis depending on the manner in which the signal is delivered to the receptor. For example, monoclonal antibody-dependent or aggregated IgG-dependent ligation of the FcγRII (CD32) low-affinity IgG receptor resulted in eosinophil survival as a result of autocrine production of GM-CSF. In contrast, if either IgG or the monoclonal antibodies were immobilized on tissue culture plates, ligation of the FcγRII induced eosinophil apoptosis, an effect which was blocked by anti-β2 integrin (CD18) monoclonal antibodies (Kim et al, 1999). Several studies have demonstrated that ligation of the death receptor Fas (APO-1/CD95) with monoclonal antibodies or with Fas ligand also induces eosinophil apoptosis (for a review, see Walsh, 1999). Interestingly, Fas expression appears to be differentially regulated in human eosinophils. Incubation of eosinophils with IFN-γ and TNF-α, both alone and synergistically, caused an increase in Fas expression, an effect reversed by addition of IL-3, IL-5 or GM-CSF (Luttman et al, 1998). The increase in Fas expression was functional as enhanced Fas ligand-mediated apoptosis was observed in eosinophils stimulated with IFN-γ or TNF-α (Luttman et al 2000). Thus, eosinophils have increased susceptibility to Fas-induced apoptosis in the presence of the TH1 cytokines IFN-γ and TNF-α, whereas the TH2 cytokines IL-3, IL-5 or GM-CSF inhibit this effect. Moreover, nitric oxide (NO) was found to disrupt Fas receptor-mediated eosinophil apoptosis (Hebestreit et al, 1998). Patients with bronchial asthma have increased levels of NO in their expired air (Alving et al, 1993) and eosinophils themselves have the potential for NO production (del Pozo, 1997). Thus, increased levels of NO might disrupt eosinophil clearance by Fas receptor ligation-dependent apoptosis. Furthermore, eosinophils might contribute to their own persistence via NO production in addition to autocrine production of IL-3, IL-5 or GM-CSF. As mentioned, these Th2 cytokines are present in the airways of asthmatics and a T-cell response skewed towards the Th2 pathway is thought to be an important factor in asthma pathogenesis (Utmetsu & DeKruyff, 1997). Thus, it is interesting to note that a recent report demonstrated that mitogen-stimulated peripheral blood T cells from asthmatic subjects expressed cell-surface Fas but were resistant to Fas-dependent apoptosis (Jayaraman et al, 1999). The intracellular pathways important in the inhibition of eosinophil apoptosis and their subsequent enhanced survival by IL-3, GM-CSF and IL-5 include the Lyn, Jak2, Raf1 and mitogen-activated protein kinases (for reviews, see Simon & Alam, 1999; Walsh, 1999). Indeed, intracellular levels of protein tyrosine phosphorylation appear to be vital in determining whether an eosinophil will exhibit prolonged survival or undergo apoptosis, and a role for both tyrosine phosphorylation and the tyrosine kinase Lyn was recently demonstrated in Fas receptor-mediated apoptosis in eosinophils (Simon et al, 1998). The picture is complicated further by the observation that levels of intracellular oxygen species in human eosinophils also appear to be involved in regulating their apoptosis and that antioxidants blocked Fas-mediated eosinophil death (Wedi et al, 1999). In other cell types, important regulators of apoptosis have been well characterized. These include the proto-oncogene bcl-2 and related molecules, the family of proteases which share homology with IL-1β-converting enzyme (ICE) and another protein family which interacts with products of the proto-oncogene c-myc. These regulators act at various points on the apoptosis pathway and it appears that their interactions are vital in determining whether a cell will survive or die. In particular, the ICE-like proteases have been implicated as universal apoptosis effectors associated with the downstream events of programmed cell death (Patel et al, 1996). Other factors important in the control of apoptosis in many cellular systems include the Bcl-2 family of proteins. Bcl-2 was first identified as a key contributor to neoplastic B-cell expansion as its overexpression prevents B-cell death (Yang & Korsmeyer, 1996). A number of members of the Bcl-2 family also inhibit cell death and these include Bcl-xL, whereas other members such as Bax and Bcl-xs promote apoptosis. It is thought that the relative levels of these proteins and their molecular interactions are crucial in determining whether a cell will survive or become apoptotic. The presence of Bcl-2 in eosinophils has been reported in two studies (Ochiai et al, 1997; Saita et al, 1997). In contrast, another study (Druilhe et al, 1998a) found little or no Bcl-2 or Bcl-xL in peripheral blood eosinophils, although easily detectable levels of both proteins were found in eosinophils cultured from umbilical cord blood mononuclear cells. Moreover, a study which utilized reverse transcription polymerase chain reaction, immunoblotting, immunochemistry and flow cytometry found no evidence of Bcl-2 expression either in unstimulated eosinophils or in cells cultured in GM-CSF or IL-5. These same workers found that eosinophil apoptosis was associated with a decrease in Bcl-xL mRNA and protein levels, whereas culture of eosinophils in IL-5 or GM-CSF maintained or enhanced Bcl-xL expression (Dibbert et al, 1998). We have recently confirmed that freshly isolated eosinophils do not appear to express Bcl-2 and have also demonstrated constitutive expression of Bax and Bcl-x. However, we also demonstrated that culturing eosinophils with IL-5 reduced their apoptosis and induced modest expression of Bcl-2 mRNA and protein with no detectable change in Bcl-x or Bax (Dewson et al, 1999). Thus, eosinophil expression of the Bcl-2 family of proteins appears to be variable and is likely to be dependent upon the state of maturation and/or activation of the eosinophils under examination. A natural question that arises from the above is what degree of significance does eosinophil apoptosis have in the real world, i.e. is there any evidence that the resolution of asthmatic and allergic inflammation, either spontaneously or after treatment, is associated with eosinophil apoptosis and their removal by phagocytosis. Although approaches to this problem have utilized both human and animal-based models, there are surprisingly few studies available. In humans, corticosteroid treatment of asthmatic patients with exacerbations of their symptoms not only resulted in clinical improvement but was also accompanied by the appearance in their induced sputum of apoptotic eosinophils, together with alveolar macrophages which exhibited evidence of eosinophil engulfment (Woolley et al, 1996). Another study examined the expression of Bcl-2, Fas and Fas ligand in bronchial biopsies from asthmatic subjects and normal subjects. Compared with normal controls or untreated patients, steroid-treated asthmatics exhibited reduced airway eosinophilia which was associated with augmented apoptotic eosinophils and increased expression of Bcl-2, Fas and epithelial cell Fas ligand (Druilhe et al, 1998b). Thus, steroid treatment of asthmatic patients induces eosinophil apoptosis not only by a direct effect but also via enhancement of the expression of Fas ligand by epithelial cells, whereas enhanced expression of Bcl-2 might contribute to survival of the epithelium. More recently, Vignola et al (1999) have demonstrated that the absolute numbers of apoptotic eosinophils and macrophages in bronchial biopsies from asthmatic subjects were inversely correlated with their asthma symptoms, as measured by the Aas score, which grades the different forms of chronic asthma from very mild (score 1) to very severe (score 5). In other words, the milder the symptoms of these asthmatic subjects, the greater the percentage of apoptotic eosinophils in their biopsy specimens. These workers also demonstrated that the severity of asthma correlated with increased GM-CSF production, which in turn was associated with the presence of non-atopic eosinophils, thereby providing additional evidence that GM-CSF inhibition of eosinophil apoptosis is an important component of eosinophilic inflammation in asthma. Eosinophil infiltration of the dermis is a feature of atopic dermatitis and the lesions of patients with this condition contain eosinophil major basic protein deposits (Leifferman, 1991). Therefore it is of interest that resolution of the late-phase reaction in antigen-induced cutaneous skin reactions in atopic subjects was associated with the ingestion of apoptotic eosinophils and neutrophils by macrophages. This study demonstrated measurable levels of apoptotic eosinophils as early as 6 h after allergen challenge, which peaked at 48 h, whereas the number of macrophages containing apoptotic cells or bodies increased from 24 h and peaked at 72 h (Ying et al, 1997). These findings are in contrast to a more recent study by Erjefält et al (1999), which examined biopsies from nasal allergen-challenged allergic subjects. These workers demonstrated that at 24 h after challenge there was evidence of an intense eosinophilia associated with actively degranulating cells together with evidence of large numbers of free eosinophil granules. In contrast, no evidence of the presence of apoptotic eosinophils was observed. The differences between these two studies might be related to variation in the kinetics of eosinophil attraction, activation and apoptosis in tissue sites as diverse as the skin and the nose. However, the more likely explanation lies in the very different protocols for allergen provocation used in the two studies. In the skin study, there was a single cutaneous allergen challenge, whereas in the nasal study two daily provocations were administered over a period of 1 week. Given these circumstances, it would be expected that a more radical and prolonged allergen exposure would result in higher levels of mucosal inflammation which could be expected to take longer to resolve. In this regard, it would have been interesting had Erjefält et al (1999) extended their time-course beyond the 24-h time point on which their observations are based, to 48 or even 72 h. Under these circumstances, there might have been evidence of both apoptotic eosinophils and their engulfment by macrophages associated with the resolution of nasal mucosal inflammation in their allergen-challenged subjects. Inhalation of aerosolized allergen by sensitized animals results in an inflammatory response and a sustained airway eosinophilia similar to that seen in the pathogenesis of human asthma. Eosinophils obtained by bronchoalveolar lavage of sensitized mice after an aerosol allergen challenge expressed Fas antigen and were sensitive to Fas-induced apoptosis. More importantly, inhalation of anti-Fas monoclonal antibody after induction of lung eosinophilia resulted in an increase in the number of peroxidase-positive macrophages in the bronchoalveolar lavage fluid and a marked reduction in the number of eosinophils in the airways which was associated with the resolution of eosinophilic inflammation (Tsuyuki et al, 1995). A recent study also demonstrated that the spontaneous resolution of lung inflammation in sensitized mice challenged with ovalbumin was associated with the presence of apoptotic eosinophils in the subepithelium of the bronchi (Kodama et al, 1998). The opposing effects of glucocorticoids on eosinophil and neutrophil survival have been demonstrated in vivo using a rat model. Dexamethasone, prednisilone and hydrocortisone induced apoptosis in peritoneal eosinophils, whereas neutrophil apoptosis was inhibited (Nittoh et al, 1998). The body of knowledge regarding the mechanisms controlling the programmed cell death and clearance of eosinophils has increased considerably in recent years and, importantly, is complemented by several clinical studies. Together, the results of these investigations provide evidence that induction of eosinophil apoptosis and their subsequent engulfment by phagocytes represents a rational avenue for the development of novel and more specific anti-inflammatory therapies for asthma. I thank Mrs Catherine Convery-Walsh for extensive editorial assistance and am grateful for the support of the Wellcome Trust (grant no. 044988/2/95/2) and Tenovus Scotland.