Regulation of eosinophil and neutrophil apoptosis - similarities and differences
Apoptosis is the most common form of physiologic cell death and a necessary process to maintain cell numbers in multicellular organisms. In many chronic inflammatory diseases, reduced cell death of different types of granulocytes is one important mechanism for cell accumulation. Granulocytes are constantly produced in large amounts in the bone marrow and the same numbers die, under normal circumstances, within a defined time period. Changing the rate of apoptosis rapidly changes cell numbers in such systems. Overexpression of IL-5 appears to be crucial for delaying eosinophil apoptosis in many allergic disorders, whereas overexpression of GM-CSF and G-CSF is associated with suppression of neutrophil apoptosis in bacterial and non-bacterial inflammations. Cytokine withdrawal leads to the induction of apoptosis in both in vitro and in vivo. In contrast to the role of survival cytokines, little is known about the role of death factors and their receptors in the regulation of granulocyte apoptosis. Recent observations suggest a role for mitochondria in both eosinophil and neutrophil apoptosis, although the mechanisms that trigger mitochondria to release pro-apoptotic factors remain to be determined. Besides similarities, there are differences in the regulation of apoptosis between these granulocyte subtypes that include both expression and function of Bcl-2 and caspase family members. The identification of differences in the apoptosis regulation may help to define new molecular targets that allow specific induction of either eosinophil or neutrophil apoptosis by pharmacological means.
Eosinophils and neutrophils are granulocyte subtypes involved in many chronic inflammatory responses. Granulocytes are able to phagocyte pathogens and secrete toxic mediators, both functions are central to their role as effector cells. They are constantly generated in the bone marrow. About 60% of the leukocytes produced in the marrow are neutrophils, whereas the eosinophils population usually does not exceed 3%. More than 100 billion granulocytes are produced and the same amount enters the blood daily (1,2). Such mature granulocytes are terminally differentiated and short-lived cells that die after 1 to 2 days in tissues in the absence of inflammation. Therefore, the production of high numbers of granulocytes requires an efficient form of physiologic cell death. In eosinophils and neutrophils, the physiologic cell death is apoptosis. Inadequate removal of dead granulocytes would result in the release of their content and consequent tissue damage, a scenario, which does not occur under normal circumstances. We have previously reviewed the regulation of eosinophil apoptosis as well as its morphological features, pathogenic and therapeutic implications (3-7). This review focuses on our current understanding about similarities and differences between the regulation of eosinophil and neutrophil apoptosis.
Overproduction of survival factors at the inflammatory site delays granulocyte apoptosis
The short life of granulocytes can be prolonged by stimulation with cytokines that have anti-apoptotic properties. It has been demonstrated that eosinophil apoptosis is delayed by IL-3, IL-5, and GM-CSF in vitro (8-10) and IL-5 is nasal polyp tissues ex vivo (11). Delayed eosinophil apoptosis appears to contribute to the accumulation of eosinophils in atopic dermatitis (12), bronchial asthma (13,14), and some (15) but not all (16,17) patients with idiopathic eosinophilia. Since IL-5 is crucial for differentiation, activation, and survival of eosinophils, there have been attempts to block this cytokine as a therapeutic approach in allergic diseases. Glucocorticoids (18), cyclosporin A (19), or anti-CD4 monoclonal antibody (20) therapy mostly reduce eosinophil numbers by reducing the numbers of IL-5- producing T cells. Similarly, treatment with IL-10 may shut off IL-5 gene expression in activated T cells (21). Interferon (IFN)-? is a new treatment, which reduces IL-5 - producing T cells, for severe asthma (22) and the Churg-Strauss syndrome (23). Similar mechanisms appear to operate in neutrophilic inflammatory responses. Neutrophil apoptosis is delayed by certain cytokines in vitro (24). Prolonged neutrophil survival ex vivo has also been associated with many neutrophilic inflammatory responses (25). The suppression of neutrophil apoptosis is at least partially due to the overexpression of survival factors, such as GM-CSF and G-CSF (25). Thus it appears that quantitative and qualitative properties of the cytokine microenvironment determine how many and which effector cells accumulate in inflammatory tissues. The resolution of inflammation is at least partially due to downregulation of survival factor gene expression and consequent induction of granulocyte apoptosis.
The regulation of granulocyte apoptosis by death factors
Apoptosis can be induced in response to specific ligands or stress stimuli that engage so-called "death receptors" of the tumor necrosis factor (TNF)/nerve growth factor (NGF) receptor superfamily (26). For instance, both eosinophils (27,28) and neutrophils (29,30) express functional Fas receptors (CD95, APO-1) in in vitro systems. Autocrine and paracrine Fas ligand/Fas receptor interactions have been proposed to be crucial for activation-induced T cell death, an important mechanism for downregulating immune responses. Based on this model, it has been suggested that the same mechanism might drive granulocyte apoptosis and be responsible for the short life span of these cells (30). However, although eosinophils and neutrophils express both Fas receptors and Fas ligands, recent studies using antagonistic anti- Fas receptor antibodies and blocking soluble Fas receptor molecules do not support the idea that spontaneous apoptosis of granulocytes is the consequence of autocrine or paracrine Fas ligand/Fas receptor interactions in purified cell populations in vitro (31,32). Moreover, increasing the chance of such molecular interactions by increasing the cell density of purified granulocytes results in delayed instead of increased cell death (unpublished observations). Finally, granulocyte apoptosis from Fas receptor or Fas ligand deficient mice appears to be normal (33), also arguing against the Fas ligand/Fas receptor hypothesis. However, these data do not exclude the possibility that other members of the TNF/NGF receptor superfamily are involved in promoting autocrine or paracrine granulocyte death. What then could be the function of Fas ligand on the surface of eosinophils or neutrophils? In inflammatory responses, expansion of cytokine-producing antigen- specific T cells occurs. As mentioned above, activated T cells produce, after a few days, molecules required for their own death, such as Fas receptor and Fas ligand (34). However, T cell death via Fas ligand/Fas receptor molecular interactions is prevented by cytokines such as IL-2, IL-4, and IL-15 (35,36). Therefore, as long as inflammation is present, these activated T cells do not undergo apoptosis and other anti-inflammatory mechanisms must exist to limit and/or resolve inflammatory responses. We have recently observed that Fas receptor activation in monocytes/macrophages leads to the production of large amounts of IL-10 in these cells (37). Fas ligand-bearing cells such as granulocytes may initiate the resolution of inflammation by stimulating IL-10 secretion in monocytes/macrophages that reduces the level of inflammatory cytokines (38). When cytokine levels at the inflammatory site are below a certain threshold, T cell apoptosis might occur to further lower the inflammatory response or to restore normal T cell homeostasis in the resolution of inflammation (Fig. 1).
The regulation of granulocyte apoptosis by Bcl-2 family members
There have been several reports suggesting the involvement of members of the Bcl- 2 family in the regulation of apoptosis. The pro-apoptotic Bax molecule was found to be expressed at high levels in both eosinophils (39) and neutrophils (25,40). Apoptotic cell stimulation is associated with translocation of cytosolic Bax into the outer membrane of mitochondria where it forms pores, allowing the release of pro- apoptotic factors such as cytochrome c (41). Clearly, high levels of Bax may contribute but may not be sufficient to the short life-span of granulocytes. The more proximal mechanisms that are responsible for Bax translocation to mitochondria in the absence of sufficient stimulation with survival cytokines remain to be determined. Recently, reduced expression of Bax has been shown to be associated with cytokine-mediated prevention of neutrophil apoptosis (25). However, Bax-/- mice have normal neutrophil numbers in their blood (42), indicating that there might be other important mechanisms responsible for delayed apoptosis in inflammatory responses. ln contrast to neutrophils, cytokine-mediated anti-apoptosis does not appear to involve reduced levels of Bax in eosinophils (39). Besides the pro-apoptotic Bax, granulocytes also express anti-apoptotic members of the Bcl-2 family. Neutrophils have been reported to express Mcl-1 (40), A1 (43), and Bcl-xL (25). Especially A1 appears to be relevant, because neutrophils from A1-/- mice have an accelerated rate of spontaneous apoptosis (44). Although there are some contrasting reports (45), the role of Bcl-2 in delaying apoptosis appears to be little in both eosinophils (39) and neutrophils (25,29). In contrast, IL-5 and GM-CSF have been reported to upregulate Bcl-xL levels in these cells. Since Bcl-xL opposes the activity of Bax, an increased Bcl-xL/Bax ratio may contribute to delayed eosinophil apoptosis in allergic and other eosinophilic diseases. These data suggest that the mechanisms by which survival factors prevent apoptosis are at least partially different in eosinophils and neutrophils (Fig. 2).
The role of caspases in granulocyte apoptosis
Despite a great variety of available apoptotic stimuli, final changes to the cell are similar and many of the signaling events appear to converge into common mechanisms involving activation of cysteine-containing proteases that cleave their target proteins at specific aspartic acids (caspases). Caspases are part of a family comprising 14 members up to now (46). They are present in the cells as inactive zymogens that must be cleaved to generate free catalytic subunits able to associate and form active heterotetramers. The family of caspases can be divided into two functional subgroups, the initiator and the executioner caspases. Caspase zymogens are activated by either autoactivation, transactivation, or other proteases. Caspases form an amplification cascade in which executioner caspases are activated by initiator caspases (46). Many apoptotic responses are initiated by activation of the initiator caspases 8 or 9. Initiator caspases necessitate special mechanisms of activation of zymogens. For instance, caspase 8 can be activated following recruitment and clustering at multicomponent apoptosis signaling complexes, resulting from ligation of cell surface molecules of the TNF/NGF receptor family, presumably by autoprocessing of the zymogens according to the induced-proximity model (47). Caspase 9 is activated by recruitment to Apaf-1 in the presence of ATP following release of cytochrome c from mitochondria (41). Activation of either of these caspases can result in activation of executioner caspases such as caspase 3, leading eventually to apoptosis. Recent studies suggest a critical role of caspase 3 in both spontaneous and Fas receptor-mediated apoptosis in neutrophils (32,48,49). In addition, activation of caspase 8 has been noticed in neutrophils and inactivation of this protease using a specific caspase inhibitor (z-IETD-fmk) resulted in inhibition of apoptosis in these cells (32). Functional inactivation of caspase 9 also resulted in blocking neutrophil apoptosis. Whereas eosinophil apoptosis appears to be also regulated by caspase 9, there is only little evidence that caspases 3 and 8 play a role in both spontaneous and Fas receptor-mediated apoptosis in these cells. For instance, both caspase 3 and 8 inhibitors demonstrate only little apoptosis blocking capacity in eosinophils (32). Moreover, although both caspases are present, it appears that their proteolytic activity is diminished in eosinophils, perhaps due to inadequate processing of the procaspase. For instance, the majority of the fragments generated by processing of caspase 3 does not correspond to the expected size of the active caspase (17 kDa). Instead, the major fragment is 20 kDa in size and might correspond to a fragment constituted of the 17 kDa attached to the 3 kDa prodomain (32) (Fig. 3). As mentioned above, a caspase 9 inhibitor does block eosinophil apoptosis. In addition, broad range caspase inhibitors also block eosinophil apoptosis. These data suggest that caspases are also critical elements of the death machinery in eosinophils. However, at least the effector caspases in these cells appear to be different compared to neutrophils and many other cell types. In contrast, the involvement of caspase 9 and the fact that apoptosis is regulated by members of the Bcl-2 family in both cell types suggest that mitochondria play an important role in the apoptotic pathway in both eosinophils and neutrophils. This has been somewhat surprising since granulocytes have been described as cells with limited numbers of mitochondria (50). The factors that trigger mitochondria to release cytochrome c in these terminally differentiated cells remain to be identified.
Apoptosis is the physiologic cell death of granulocytes. There is increasing evidence from several laboratories that granulocyte numbers are regulated in vivo, not only by their production in the bone marrow, but also by the amount of apoptosis. Granulocyte apoptosis at the inflammatory site appears to be delayed when survival factors are generated by neighboring cells, or, perhaps, due to autocrine production of these cytokines. Another possibility to regulate apoptosis in granulocytes might be achieved via death receptors of the TNF/NGF receptor superfamily. Although both eosinophils and neutrophils demonstrate spontaneous apoptosis, which can be suppressed by survival and accelerated by death factors, it appears that there are differences in the intracellular regulation of apoptosis between these two granulocyte subtypes. Such differences may explain the different effects of glucocorticoids on apoptosis regulation in these two types of granulocytes (51). They may also open up new therapeutic approaches to specifically induce either eosinophil or neutrophil apoptosis in inflammatory disorders. The induction of apoptosis by drugs represents an efficient, non-inflammatory way to remove granulocytes from inflammatory sites. There are many urgent questions to be answered in the near future. For instance, what are the mechanisms that limit granulocyte accumulation by delayed apoptosis? How long do granulocytes act as effector cells in allergic inflammation when they do not undergo apoptosis? Which other death receptors (in addition to Fas and TNF receptors) are expressed in granulocytes? What is their function? Which are the relevant effector caspases in eosinophils? What drives the apoptotic process in mature granulocytes? How does a cell decide its outcome when it receives survival and death signals at the same time, a situation that very likely occurs in vivo? Clearly, there is much more to learn about granulocyte apoptosis.
Fig. 1. A proposed model showing the importance of Fas ligand/Fas receptor molecular interactions in the resolution of allergic inflammatory responses in a sequential manner. (Left) Allergic inflammatory tissues contain increased numbers of T cells (dark blue) and eosinophils (red). Both cells generate large amounts of IL- 4, IL-5, and IL-13. IL-2 is required for allergen-specific T cell expansion and produced by T cells. Activated T cells express both Fas ligand and Fas receptor. However, no apoptosis is induced in these cells as a consequence of Fas ligand/Fas receptor interactions in the presence of high levels of IL-2 and IL-4. (Middle) Fas ligand expressing cells (small arrows on T cells and eosinophils) are able to induce the IL-10 gene in monocytes/macrophages, resulting in the suppression of cytokine expression in both T cells and eosinophils. (Right) In the absence of high levels of cytokines, molecular Fas ligand/Fas receptor interactions on T cells result in apoptosis, reducing furthermore cytokine levels. Eosinophil numbers decrease due to reduced survival factor (IL-5) stimulation and consequent apoptosis.
Fig. 2. Survival factor stimulation results in differential regulation of members of the Bcl-2 family in eosinophils and neutrophils. (A) Normal situation in mature granulocytes: Eosinophils and neutrophils express high levels of Bax, which translocates from the cytosol to the mitochondria. The initial stimulus for Bax translocation is unknown. Bax is involved in the release of cytochrome c (cyto c) from the mitochondria into the cytosol. Cytochrome c is required for the activation of caspase 9 via the adapter molecule Apaf-1. Caspase 9 activates effector caspases, which degrade key cellular substrates and lead to apoptosis. (B) Survival factors (IL- 5, GM-CSF) induce the Bcl-xL gene in eosinophils. Bax levels remain unchanged. Bcl-xL blocks Bax-mediated cytochrome c release and apoptosis. (C) Survival factors (G-CSF, GM-CSF) reduce Bax levels in neutrophils. Less Bax is inserted into mitochondria, resulting in less cytochrome c release and apoptosis.
Fig. 3. Differential cleavage of procaspase 3 in eosinophils and neutrophils. (A) Procaspase is a protein of 32 kDa. Cleavage occurs usually at two aspartic acids. In the case of eosinophils, cleavage occurs only between the large (17 kDa) and small subunit (12 kDa) of the inactive precursor. However, the prodomain (3 kDa) is not cleaved. It is proposed that the observed inefficient cleavage is at least one reason for the inactivity of this enzyme in eosinophils. (B) In neutrophil apoptosis, the prodomain is discarded, and the large and small subunits form the active enzyme.
Hans-Uwe Simon, MD, PhD