I. Introduction and Historical Perspective

Top Next References

Using a compound microscope and the blood of several animals (including the elephant!), the English anatomist, T. W. Jones, discovered, in 1846, that some white blood cells contained granules that became visible when immersed in hypotonic solutions (Jones, 1846). Although it has been claimed that Jones had discovered the eosinophil (Archer, 1963), it is more likely that he visualized the more abundant neutrophil (Spry, 1988). It was Brown (1898) who probably first detected eosinophils in the blood and bone marrow of patients with eosinophilic leukemia in the latter decades of the 19th century, although the lack of appropriate dyes and staining techniques at that time prevented formal identification. Full credit for the discovery of the "eosinophile" is thus given to Paul Ehrlich (1879) who first noticed that a certain population of white blood cells was stained with a negatively charged, brominated fluorescein compound, eosin, and was so named for that property.

Despite the discovery of eosinophils almost 120 years ago, still relatively little is known of their biochemistry and pharmacology when compared to their highly studied sister cell, the neutrophil. This is perhaps surprising given the critical role of these cells both in host defense (Butterworth and Townley, 1993; Allen and Davis, 1994) and, under certain circumstances, in a variety of diseases, including many, if not all, of those indicated in Table 1. However, a persuasive argument (and one that, through experience, is vigorously championed by the authors of this review!) for the lack of investigation almost certainly reflects the difficulty in obtaining eosinophils in sufficient numbers and of a purity required for detailed studies to be performed and from which unambiguous conclusions can be drawn. Moreover, the process of purification and the effect of previous drug therapy on the ex vivo behavior of human eosinophils invariably leads to alterations in cell function and can make interpretation of results difficult. With the refinement of separation and purification techniques, in particular the use of "negative selection" to remove unwanted leukocytes (Hansel et al., 1989, 1990, 1991b), has come a marked increase in the number of articles published relating to eosinophil biology. Indeed, according to PubMed records, more than 3500 articles have been published since 1990 with a noticeable increase in pharmacological and biochemical content. It thus seems timely to attempt a comprehensive treatise of the pharmacology of the eosinophil, and the authors make no apology for omitting much of the immunology and parasitology which has been elegantly reviewed elsewhere (Capron, 1991, 1992; Weller, 1991; McEwen, 1992; Butterfield and Leiferman, 1993; Butterworth and Thorne, 1993; Wardlaw et al., 1995).

	II. Gross Morphology and Ultrastructure

Top Previous Next References

Eosinophils are actively motile, terminally differentiated leukocytes derived from the bone marrow, and have been identified in many mammalian and nonmammalian species (Table 2). Human eosinophils are approximately 8 µm in diameter, have a volume of 275 fl and, in addition to their avidity for eosin, exhibit several distinct characteristics that distinguish them from other granulocytes (Sokol et al., 1988; Dvorak, 1991). Generally, normal healthy eosinophils have a bi-lobed nucleus that is filled with partially condensed chromatin (Figs. 1 and 2). In some diseases, however, the number of lobes is increased to more than four (Sokol et al., 1987). A prominent feature of the eosinophil is the presence of many spherical or ovoid granules (Figs. 1 and 2) that occupy approximately one-fifth of the cytoplasm. Four distinct populations of granule (secondary granules, small granules, primary granules, lipid bodies) have been recognized that house a plethora of proteins, many with enzymatic activity (Fig. 2; Table 3). The first morphological marker of the eosinophil is the appearance of granules that are visible at the promyelocyte stage (Zucker Franklin, 1980). Several proteins are found within these structures, including eosinophil peroxidase (EPO),² acid phosphatase, and arylsulphatase B. Despite earlier descriptions to the contrary (Bainton and Farquhar, 1970), these granules are probably precursors of the specific, or secondary, granules that are first seen at the myelocyte stage of maturation (Hardin and Spicer, 1970; Gleich and Loegering, 1984). In eosinophils harvested from humans (Zucker Franklin, 1980; Tavassoli, 1981; Cohen and Ottesen, 1983) and from many other species, including the dog, mouse, rat, goat, guinea pig and rhesus monkey (Jain, 1986), the specific granules feature a prominent crystalloid core containing major basic protein (MBP). Specific granules, containing multiple cores, also have been visualized (Newman et al., 1996) but their occurance is relatively rare (Fig. 3). In addition, other highly charged cationic proteins typified by eosinophil cationic protein (ECP), eosinophil-derived neurotoxin (EDN), and EPO (Egesten et al., 1986) are located within the noncrystalloid matrix along with a number of cytokines (Fig. 2; Table 3). Differences in the gross morphology of the secondary granules are apparent between species. Thus, in cats, the core is lamellar rather than crystalloid, whereas in cattle, horses, mink, and gorillas the granules lack a central core and appear to be homogeneous when visualized under the electron microscope (Henderson et al., 1983; Jain, 1986; McEwen, 1992).

View larger version (116K): [in this window] [in a new window]   Fig. 1.   Electron micrograph of untreated eosinophils purified from the peritoneal cavity of guinea pigs. A bi-lobed nucleus containing condensed chromatin is shown (large arrow) along with cytoplasm packed with many large, membrane-enclosed, dense crystalloid-containing ovoid granules (smaller arrows). Cells were conventionally fixed (glutaraldehyde/osmium tetroxide). Bar, 1 µm. Original magnification, 18,000×. See II for further details

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Cardinal structures of a human eosinophil. Shown are the typical bi-lobed nucleus (BLN) and the four main granules. The primary (1°) granule is the principle site of Charcot-Leyden crystals, whereas MBP, ECP, EDN, and EPO reside within the classically crystalloid secondary (2°) granule along with a number of cytokines and a host of other proteins many with enzymatic activity. Lipid bodies (LB), which represent a site of lipid mediator biosynthesis, also are found in resting and activated eosinophil where their number is increased along with small granules (SG) that store proteins such as arylsulfase B and acid phosphatase. See II for further details. COX, cyclooxygenase; 5-LO, 5-lipoxygenase; LPLase, lysophospholipase.

View larger version (180K): [in this window] [in a new window]   Fig. 3.   Identification of single and multiple crystalloid cores in specific granules from streptolysin O-permeabilized guinea pig peritoneal eosinophils stimulated with GTPS and Ca2+. Bar, 500 nm. Original magnification, 55,000×. See II for further details.

A population of small granules also has been identified in human tissue eosinophils that are not seen within circulating cells or those in the bone marrow (Parmley and Spicer, 1974). These structures stain intensely for arylsulphatase B and acid phosphatase (Parmley and Spicer, 1974; Dvorak, 1991) and may also contain catalase (Iozzo et al., 1982).

The third type of storage organelle that has secretory properties is the primary granule, which accounts for approximately 5% of all eosinophil granules (Fig. 2). Morphologically, they are roughly spherical, of variable size, and contain no discernible core. In resting eosinophils, primary granules provide the sole location for Charcot-Leyden crystals (Dvorak et al., 1988), which are colorless, and have a characteristic hexagonal, bipyramidal structure with intrinsic lysophospholipase activity (Ackerman et al., 1980; Weller et al., 1980). In activated cells, trace amounts of Charcot-Leyden crystals have been identified within the nucleus and cytoplasm, implying that this protein can be released intracellularly.

The final population of granules is known as lipid bodies and these structures are not membrane-bound (Fig. 2). Approximately five lipid bodies are found per resting eosinophil, although the number can increase when the cell is activated. Lipid bodies are spherical (0.5-2 µm in diameter), electron-dense organelles, and, as the name implies, provide a principle store for arachidonic acid (AA) that is esterified into glycerophospholipids (Weller and Dvorak, 1985; Weller et al., 1991a).

Further description of the morphology of eosinophils in health and disease is beyond the scope of this review, but interested readers should consult articles by Dvorak (1991) and Sokol et al. (1987) which provide a comprehensive treatise of the subject.

	III. Life Cycle, Maturation, and Tissue Distribution

Top Previous Next References

Eosinophil turnover, or eosinopoeisis, occurs almost exclusively in the bone marrow although ancillary sites of production can include the spleen, thymus, and lymph nodes (Till and McCulloch, 1961; Jain, 1986; McEwen, 1992). The bone marrow from normal individuals contains about 3% eosinophils, of which 37% are mature, nondividing granulocytes, and the remainder are promyelocytes/myelocytes (37%) and metamyelocytes (26%) that exist in "storage" compartments (Spry, 1988, 1993). At any one time, it has been estimated that about 16% of myelocytes are undergoing DNA synthesis (i.e., are in the S phase of the cell cycle), which lasts about 13 h, and that the time taken from the last mitosis until they appear in the blood as mature cells (the emergence time) is approximately 2.5 days (Spry, 1988). The migration of eosinophils from the bone marrow to the blood takes about 3.5 days (Parwaresch et al., 1976). Using [³H]thymidine flash-labeled peripheral blood eosinophils, Walle and Parwaresch (1979) performed studies in three hematologically normal men to estimate the eosinophil reserve capacity in the postmitotic granulocyte compartment in the bone marrow and the effective eosinopoeisis. The results of those experiments demonstrated that mean turnover of eosinophils is approximately 2.2 × 10⁸ cells/kg/day and that the bone marrow provides the largest postmitotic eosinophil reserve capacity (9-14 × 10⁸ cells/kg).

There is compelling evidence that eosinophils are derived from small populations of self-regenerating, hematopoietic stem cells that also are capable of differentiation into the individual lymphomyeloid lineages. The ultimate commitment of stem cells to unipotential progenitors, and their subsequent survival and expansion into mature eosinophils, has been studied extensively, although a complete understanding of the factors and processes by which this occurs still is lacking. It has been suggested that the fate of a hematopoietic stem cell to regenerate or to commit to a multipotential progenitor is purely stochastic (Till et al., 1964; Nakahata et al., 1982; Nakahata and Ogawa, 1982). In the latter scenario, a host of cytokines and other factors are required including interleukin (IL)-6, IL-11, IL-12, granulocyte colony-stimulating factor (G-CSF), stem cell factor (SCF; CD117; formerly known as c-kit ligand and Steel factor), and leukemia inhibitory factor (LIF) (Ploemacher et al., 1993; Ogawa, 1994). Further development of multipotential cells into eosinophil progenitors is under the influence of SCF, IL-3, IL-4, granulocyte/macrophage (GM)-CSF, and eotaxin (Kobayashi, 1993, Peled et al., 1998). Interleukin-5 and possibly eotaxin then provide the major driving force for the terminal stages of maturation and release into the blood stream (Clutterbuck et al., 1989; Sanderson, 1993; Palframan et al., 1998a).

In the guinea pig, IL-5 releases eosinophils from the bone marrow by a mechanism that is blocked by the phosphatidylinositol 3-kinase (PtdIns 3-kinase) inhibitors wortmannin and LY 294002, although the downstream substrates involved in this process are currently undefined (Palframan et al., 1998b). Moreover, the emigration of eosinophils from the marrow precipitated by IL-5 is associated with adhesive interactions involving ₄ and ₂ integrins that act in an opposing manner. In vivo the expression of ₂ integrins is reduced in response to IL-5, whereas the ₄ integrin level remains unchanged. The observation that a ₂ integrin-blocking antibody suppresses IL-5-driven eosinophil mobilization suggests that these adhesion molecules are necessary for effective migration. In contrast, an ₄ integrin-blocking antibody enhances the release of eosinophils from the marrow in response to IL-5, and it has been speculated that this prevents their normal attachment to the bone marrow sinus endothelium (Palframan et al., 1998b). Thus, the egress of mature eosinophils from the marrow involves a number of discrete steps.

Once in the circulation, eosinophils have a half-life of approximately 18 h and a mean blood transit time (26 h) similar to neutrophils (Steinbach et al., 1979). However, the half-life of eosinophils is prolonged when an eosinophilia is precipitated which might be due to an increase in the concentration of certain circulating cytokines that enhance survival (see XII. H) and/or to the saturation of sites through which eosinophils migrate into tissue.

In humans and many domestic animals, eosinophils comprise 2 to 10% of the peripheral leukocytes, but in cows the average titer is approximately 20% (Duncan and Prasse, 1986; McEwen, 1992). The circulating eosinophil count exhibits diurnal variation in some species; thus, in normal human subjects the highest and lowest levels are seen in the evening and the morning, respectively (Horn et al., 1975), whereas the opposite occurs in horses (McEwen, 1992). Eosinophils are predominantly tissue cells and do not reenter the circulation. The gastrointestinal tract, lung, and skin and, in rats, the uterus during dioestrus or oestrogen treatment (see XIV.F) are the principle sites of accumulation (Dembele Duchesne et al., 1991), and histological studies with human tissues have identified columnar epithelial surfaces as particularly rich in eosinophil infiltrates. Large numbers of eosinophils can be found in tissues even when the peripheral blood count is low, which suggests that their longevity is enhanced once they leave the circulation. It has been estimated that the number of eosinophils in the bone marrow and tissues of rats is 300 times higher than the circulating count (Rytomaa, 1960). The tissue distribution of eosinophils in subjects with disease has not been systematically quantified, although it is curious that pathogen-free animals have no eosinophils in their blood and tissue eosinophils are difficult to find. This strongly suggests that an increase in the circulating eosinophil count and retention of eosinophils in tissues is disease-related (Spry, 1993), although this might not apply to the gut (see V.E.1).

	IV. Transcription Factors and Eosinophilia

Top Previous Next References

Gene transcription is regulated in a highly coordinated and complex fashion by a diverse family of DNA-binding proteins known collectively as transcription factors. In diseases such as those associated with peripheral blood eosinophilia, transcription factors may play a key role in inducing or repressing critical genes that control eosinopoiesis. Perhaps the most universal and ubiquitously distributed transcription factors are activator protein 1 (AP-1) and nuclear factor B (NF-B), which are involved in the regulation and coregulation of many genes. In contrast, other transcription factors have a more cell-specific distribution and regulate the expression of a restricted number of genes. For example, the transcription factors nuclear factor of activated T cells (NFAT) (Rao et al., 1997), guanine-adenine-thymine-adenine (GATA-3) (D. H. Zhang et al., 1997) along with NF-B (Yang et al., 1998) are critically important in controlling the IL-5 and eotaxin genes that are probably essential for the differentiation, maturation, and trafficking of eosinophils (see III.). Similarly, in the lungs of mice that are deficient in the p50 subunit of NF-B, lymphocyte recruitment after allergen provocation is attenuated compared to wild type animals due to a reduction in the secretion of macrophage inflammatory protein (MIP)-1 and MIP-1 (Yang et al., 1998).

Additional genes are also likely to regulate circulating eosinophil number and eosinophilia associated with disease. In individuals with familial eosinophilia, a rare disease encompassed by the generic term hypereosinophilic syndrome that has no allergic or parasitic basis, a locus (or loci) has been identified on region q31-q33 of chromosome 5 which contains the cytokine gene cluster for IL-3, IL-5, and GM-CSF (Rioux et al., 1998). Since no functional polymorphisms were found within the enhancer, promoter, exons, or introns of any of these genes, it has been speculated that a main cause of familial eosinophilia is due to a novel gene that is situated within region q31-q33 (Rioux et al., 1998). Indeed, this idea is supported by the knowledge that greater than 100 anonymous transcripts have been found in that region of human chromosome 5 (Schuler et al., 1996). Martinez et al. (1998) also have identified markers on the same region of chromosome 5 that controls for circulating eosinophil number as a percentage of total white leukocytes.

	V. G Protein-Coupled Receptors and Their Ligands

Top Previous Next References

G protein-coupled receptors are characterized by an extracellular amino-terminal sequence followed by seven transmembrane-spanning domains, with three extracellular and three intracellular loops, and an intracellular carboxyl terminus. Conserved cysteine residues within the amino-terminal sequence and in the third extracellular loop are thought to form a disulfide bond which is required for ligand binding, while a second disulfide bond is probably formed between conserved cysteine residues within the first and second extracellular loops. The functional responses that result from ligand binding are transduced by G proteins. These are heterotrimeric proteins consisting of , , and  subunits that each exist in multiple isoforms (20 , five , 10 ) in mammalian cells. Several G proteins and/or subunits thereof have been identified in human and guinea pig eosinophils including G_s, G_i3, G₀, G_q/11, and G (Agrawal et al., 1992; Lacy et al., 1995).

In excess of 17 G protein-coupled receptors have thus far been identified on eosinophils (Table 4). These receptors can couple to a vast array of effector proteins that ultimately produce a host of functional responses resulting both in stimulation and suppression of eosinophil activity. These are identified and discussed in detail below.

A. Platelet-Activating Factor

1. Receptors and Signaling. The ether lipid, platelet-activating factor (PAF), evokes its biological effects by interacting with a classical seven transmembrane-spanning receptor that is composed of 342 amino acids and has a molecular mass of approximately 39 kDa (Honda et al., 1991; Nakamura et al., 1991). Radioligand-binding experiments have identified PAF receptors on many cells, including eosinophils. However, until the early 1990s ³H-labeled PAF was the only radioligand available for this purpose and proved to be unsatisfactory for several reasons. Notably, it causes activation of cells and, with prolonged exposure, receptor down-regulation. Furthermore, the lipophilicity of PAF gives rise to high levels of nonspecific binding, "specific" nonreceptor binding, and the labeling of intracellular or internalized receptors, factors that hamper its utility for accurate determination of cell surface receptor density and ligand affinity (Dent et al., 1989). Nevertheless, estimates of K_d (2.3 nM) and B_max (104 fmol/10⁶ cells) have been made for ³H-labeled PAF in human eosinophils (Korth, 1996), and it seems likely that the ligand-labeled sites represent specific receptors because binding was reversed by unlabeled PAF and the PAF antagonist apafant (WEB 2086) (Korth, 1996).

2. In Vitro Effects. PAF is a potent chemoattractant and selectively promotes the migration of eosinophils over neutrophils. The ability of PAF to promote directional migration is significantly increased in eosinophils taken from asthmatic subjects both in remission and during an attack when compared with healthy volunteers (Shindo et al., 1997), suggesting that they have been primed in vivo. Possible candidate-priming agents include GM-CSF, which enhances PAF-induced pulmonary and cutaneous eosinophilia in guinea pigs (Sanjar et al., 1990a) and mice (Yukawa et al., 1992), and IL-3 and IL-5, which prime murine eosinophils for enhanced chemotactic activity induced by PAF (Yukawa et al., 1992). Other proinflammatory effects of PAF include the generation of a plethora of other bioactive lipids (Table 5) and the release of preformed mediators from both the specific and small granules.

3. In Vivo Effects. In guinea pigs, rabbits, and primates, aerosol and systemic administration of PAF results in the extravascular infiltration of eosinophils into the lungs which resembles, both in amplitude and duration, that seen in response to allergen in sensitized animals (Denjean et al., 1983; Arnoux et al., 1988; Lellouch Tubiana et al., 1988; Sanjar et al., 1990b; Gundel et al., 1991; Herd et al., 1992; Wegner et al., 1992). Comparable observations have been made with rats given PAF directly into the pleural cavity (Silva et al., 1989) and in atopic individuals where intradermal administration of PAF produces a cellular infiltrate rich in eosinophils that is reminiscent of the eosinophilia seen in the same subjects after antigen provocation (Henocq and Vargaftig, 1988). Similarly, in individuals with seasonal allergic rhinitis PAF, given intranasally and outside the pollen season, evokes a marked increase in the number of eosinophils (Klementsson and Andersson, 1992) and in the concentration of ECP (Tedeschi et al., 1994) in the nasal lavage fluid.

B. Leukotriene B₄

1. Receptors and Signaling. The BLT, or LTB₄, receptor, which is expressed on guinea pig, mouse, and probably human eosinophils, was cloned in 1997 from retinoic acid-differentiated HL-60 cells. This human receptor is composed of 352 amino acids and is a member of the seven transmembrane-spanning family of G protein-coupled receptors (Yokomizo et al., 1997). A cDNA that encodes a 351-amino acid murine glycoprotein that is 78% identical with the human BLT receptor has also been identified and expressed in Chinese hamster ovary (CHO) cells (Huang et al., 1998). An analysis of [³H]LTB₄ binding to membrane fractions prepared from CHO cells, and retinoic acid-differentiated HL-60 and COS-7 cells transfected with the cDNA for the human and murine LTB₄ receptor show similar binding characteristics, with K_d values of 0.1, 0.14, and 0.15 nM, respectively (Yokomizo et al., 1997; Huang et al., 1998). Binding studies also have identified and partially characterized the BLT receptor on murine and guinea pig eosinophils (Maghni et al., 1991; Ng et al., 1991; Sehmi et al., 1992a; Huang et al., 1998) also using [³H]LTB₄ as the radioligand. However, notable differences are apparent between these studies. Using intact peritoneal eosinophils from guinea pigs, Ng et al. (1991) reported that [³H]LTB₄ interacts with an apparently homogeneous population of binding sites with a B_max of 40,000 sites per cell and a K_d of 2.8 nM, which is approximately 10-fold lower than that reported in the transfection experiments described by Yokomizo et al. (1997). Similar results were reported for the murine receptor (Huang et al., 1998). The sites labeled on eosinophils probably represent functional receptors since various compounds related structurally to LTB₄ compete with the radioligand with affinities that correlate closely with their ability to induce chemotaxis and to evoke the formation of superoxide anions (Ng et al., 1991). Intriguingly, the rank order of potency for the displacement of [³H]LTB₄ from intact peritoneal eosinophils [LTB₄ > 20-hydroxy-LTB₄ > 12R-hydroxyeicosatetranoic acid (HETE) > 12S-HETE > 20-carboxy-LTB₄ > 5S,12S-dihydroxyeicosapentanoic acid (diHEPE) (Ng et al., 1991)] differs from the rank order obtained using membranes from COS-7 cells transfected with the LTB₄ receptor [LTB₄ > 20-hydroxy-LTB₄ > 20-carboxy-LTB₄ > 5S,12S-diHEPE > 12R-HETE > 12S-HETE (Yokomizo et al., 1997)] which might indicate species difference, LTB₄ receptor heterogeneity (see below), and/or the existence of different conformations of a single LTB₄ receptor. With respect to the two latter possibilities, Maghni et al. (1991) reported that [³H]LTB₄ interacts with a heterogeneous population of binding sites on guinea pig alveolar eosinophils; approximately 1000 sites/cell are labeled with high affinity (K_d = 1 nM), whereas 5500 sites/cell are labeled with low affinity (K_d = 63 nM). Identical results have been obtained with guinea pig peritoneal eosinophils (Sehmi et al., 1992a). Thus, a small population (B_max = 900 sites/cell) of receptors for which LTB₄ has high affinity (K_d = 0.3 nM) were identified by radioligand binding along with a large number of sites (60,000/cell) at which LTB₄ has relatively low affinity (K_d = 140 nM). Again, the finding that various metabolites of LTB₄ competed with [³H]LTB₄ for binding to alveolar eosinophils with a rank order of potency in good agreement with their ability to induce chemotaxis (Maghni et al., 1991) supports the belief that the high-affinity sites represent bona fide receptors. Of considerable interest is the role of the two populations of receptor expressed by these cells. Maghni et al. (1991) have considered the hypothesis, posed originally by Goldman and Goetzl (1984), that they mediate different functional responses: the receptor for which LTB₄ has high affinity subserving chemokinesis and chemotaxis, the receptor for which LTB₄ has low-affinity mediating respiratory burst and prostanoid generation. Support for this idea derives from affinity estimates of the LTB₄ antagonist U-75302, which differs significantly (~17-fold) between the two populations of receptor (Maghni et al., 1991). Collectively, the available data suggest that peritoneal eosinophils express the same LTB₄ receptor that is labeled with high affinity by [³H]LTB₄ on guinea pig alveolar eosinophils [albeit at a much higher (~ 40-fold) density]. A reason for the inability of Ng et al. (1991) to identify receptors on guinea pig peritoneal eosinophils for which LTB₄ has low-affinity may relate to the fact that in those studies [³H]LTB₄ was not used at concentrations that would detect the low-affinity sites.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   LTB4-induced signaling in guinea pig peritoneal eosinophils. Scheme A, low concentrations (1 pM to 10 nM) of LTB4 induce a PTX-sensitive increase in the [Ca2+]I and activate the src-related tyrosine kinase lyn and the raf-1/MEK-1/2/ERK-1/2 protein kinase cascade. The increase in [Ca2+]I is due exclusively to influx of extracellular Ca2+, whereas the activation of lyn, is thought to mediate the activation of a Ca2+-dependent PLA2 (possibly cPLA2) and the subsequent release of AA. Scheme B, higher concentrations (100 nM to 1 µM) of LTB4 activate PLC, with attendant generation of Ins(1,4,5)P3, and a Ca2+-independent PLA2 (possibly iPLA2) with a further liberation of AA. LTB4 also is thought to stimulate a tyrosine kinase-dependent pathway that is implicated in the activation of the NADPH oxidase. See V.B for additional details.

2. In Vivo Effects. A number of in vivo animal models have been developed to establish the potential pathogenic role of LTB₄ in allergic eosinophil inflammatory disorders and autoimmune diseases such as multiple sclerosis and asthma. In 1996, Gladue et al. reported that the LTB₄ antagonist CP 105,696 abolished the ability of encephalogenic T lymphocytes, injected into naïve mice, to evoke two cardinal features of experimental allergic encephalomyelitis (multiple sclerosis): paralysis and weight loss. Moreover, the protection was associated with a 97% reduction in the accumulation of eosinophils to the lower spinal cord as determined by light and electron microscopy, and by the level of EPO (Gladue et al., 1996). Those findings have important implications since they show that agonism of LTB₄ receptors results in eosinophil recruitment and that they play a hitherto unrecognized role in the pathogenesis of experimental allergic encephalomyelitis. Clearly, the possible utility of LTB₄ antagonists in the treatment of human multiple sclerosis, and the part eosinophils play in that disease, merits evaluation.

Two receptors (Cys-LT₁ and Cys-LT₂) for the cysteinyl LTs, which include LTC₄, LTD₄, and LTE₄, have been classified pharmacologically but supporting molecular evidence is still awaited. Both receptors couple predominantly through the G_q/11 class of heterotrimeric GTP-binding proteins, and it is highly likely that they are members of the seven transmembrane-spanning family of receptors (see Coleman et al., 1995 for additional details). Currently, selective antagonists are available only for the Cys-LT₁ receptor and these have been used to identify those receptors on eosinophils. However, antagonist affinities have not been calculated and the assignment of eosinophil leukotriene receptors as Cys-LT₁ is equivocal.

1. In Vitro Effects. Relatively little is known of the pharmacological actions of cysteinyl-leukotrienes on eosinophil function compared to those of LTB₄. Although early studies failed to demonstrate that LTD₄ possessed chemoattractant activity (Nagy et al., 1982; Camp et al., 1983), convincing evidence is now available to the contrary. Using a novel in vitro method, which allows the quantification of migration distance and vectorial orientation, it has been shown that LTD₄ is a potent chemoattractant for human eosinophils, with activity in the subnanomolar range. Moreover, LTD₄-induced chemotaxis is antagonized by SK&F 104353, suggesting that Cys-LT₁ receptors are involved (Spada et al., 1994, 1997). In contrast, LTD₄ does not increase the chemokinetic response of eosinophils above spontaneous migratory activity (Spada et al., 1994).

2. In Vivo Effects. In laboratory animals, LTD₄ and LTE₄ given locally and systemmically can stimulate the accumulation of eosinophils into various sites including the skin, eye, and lungs (Spada et al., 1986, 1988; Chan et al., 1990; Foster and Chan, 1991; Woodward et al., 1991; Wegner et al., 1993; Underwood et al., 1996). For example, in one study guinea pig eosinophils were labeled with [¹¹¹In]oxime and injected (i.v.) into recipient animals (naïve and sensitized), and the effect of LTD₄ and allergen on their emigration into the conjunctiva was monitored. Using that model, it was consistently found that LTD₄ and allergen significantly enhanced conjunctival radioactivity by a mechanism that was abolished (LTD₄) and reduced by 50% (allergen) by the Cys-LT₁ receptor antagonist MK-571 (Chan et al., 1990). Significantly, LTD₄ neither promotes the infiltration of eosinophils into the skin of guinea pigs following intradermal administration nor is it active in other ocular anterior segment structures such as the iris, cornea, and ciliary body after topical or intracameral administration (Woodward et al., 1991). Thus, it appears that LTD₄ regulates eosinophilia in a tissue-dependent manner.

Two variants of the human fMLP receptor have been isolated from a CDM8 expression library prepared from mRNA extracted from dibutyryl cyclic AMP-differentiated HL-60 cells (Boulay et al., 1990a). Both recombinant forms of the receptor are composed of 350 amino acids, have a predicted molecular mass of 38 kDa, but differ from each other by two residue changes at positions 101 and 346; significant differences also are apparent at the 3'- and 5'-untranslated regions (Boulay et al., 1990a). Expression of these proteins in COS-7 cells results in the appearance of two populations of a highly glycosylated receptor for which the hydrophilic ligand N-fMLP-lysine has high affinity with K_d values of 0.5 to 1 nM and 5 to 10 nM (Boulay et al., 1990b). Moreover, several transcripts have been identified by Northern blot analysis, suggesting that the fMLP receptor is a family of closely related proteins.

At present, there are no molecular data concerning the nature of the fMLP receptor(s) expressed by eosinophils of any species. However, functionally, fMLP elicits a variety of effects in isolated cells, some of which are listed in Table 8. Less is published on the in vivo effects of formylated peptides, although in guinea pigs fMLP promotes lumenal chemotaxis of eosinophils as assessed by histology and differential cell counts (Munoz et al., 1997a). Intriguingly, that effect is attenuated by the LTB₄ antagonist LTB₄ dimethyl amide, by zileuton, a 5-lipoxygenase inhibitor, and by zafirlukast, a Cys-LT₁ antagonist, indicating that the ability of fMLP to facilitate the migration of eosinophils from the lamina propria to the airway lumen of guinea pigs is indirect and requires the liberation of LTB₄ and LTD₄ (Munoz et al., 1997a).

Compared to other G protein-coupled receptors, relatively little is known of the signaling pathways recruited following ligation of the fMLP receptor on eosinophils. It is established that fMLP promotes a rapid and transient increase in [Ca²⁺]_i in both human (Yazdanbakhsh et al., 1987b; Wymann et al., 1995) and guinea pig eosinophils (Kroegel et al., 1990b) that is believed to be important for the generation of oxygen-derived free radicals (Kernen et al., 1991). Furthermore, many of the functional effects elicited by fMLP including degranulation (Kita et al., 1991a), activation of the NADPH oxidase (Kernen et al., 1991), the release of IL-8 (Miyamasu et al., 1995), as well as intracellular markers of activation (stimulation of PLC, Ca²⁺ mobilization) are mediated by a PTX-sensitive mechanism(s), indicating the involvement of one of more members of the G_i or G_o family of heterotrimeric GTP-binding proteins.

Chemokines are an expanding superfamily of proteins with molecular masses of between 8 and 10 kDa (for reviews, see Horuk, 1994; Power and Wells, 1996; Raport et al., 1996a). Characteristically, human chemokines contain four distinct and conserved cysteine residues that have provided the basis of their classification either as CXC or  chemokines, where the first two cysteine residues are separated by an amino acid, or CC or  chemokines, where the first two cysteines are adjacent. Two other chemokine families have been described: C (or ) chemokines that contain a single cysteine residue and include lymphotactin, and the CX₃C chemokine family (also known as  chemokines) where three amino acids separate the two cysteine residues, of which fractalkine and neurotactin are examples. The CXC chemokines generally are involved in the recruitment of neutrophils and have been implicated in acute inflammatory responses. In contrast, CC chemokines exert their actions upon other leukocyte populations, including eosinophils, monocytes, and T lymphocytes, and are believed to be involved in the pathogenesis of chronic inflammation. Four CXC and eight CC chemokine receptors have been cloned thus far that are recognized by a selective range of chemokines with characteristic rank orders of potency (see Gerard and Gerard, 1994; Murphy, 1994; Ben Baruch et al., 1995; Combadiere et al., 1995, Gao and Murphy, 1995; Power et al., 1995; Hoogewerf et al., 1996; Ponath et al., 1996a,b; Power and Wells, 1996; Raport et al., 1996b; Samson et al., 1996; Heath et al., 1997). It is this diversity of chemokine receptor expression and the selective release of chemokines that provide a mechanism for the recruitment of different leukocyte populations to inflammatory sites. Moreover, in the context of asthma, chemokines activate distinct cellular and biochemical pathways that act in a coordinated fashion to elicit complex pathophysiological changes such as eosinophilia and airways hyperreactivity (Gonzalo et al., 1998).

Of the multitude of chemokine receptors thus far identified, human eosinophils express CCR1, CCR3, and possibly a receptor for IL-8 that is either CXCR1 or CXCR2 (Table 9). The pharmacological properties of these receptors and the functional responses they subserve are discussed below.

1. CC Chemokines. The eotaxin receptor, or CCR3, is selectively expressed upon eosinophils, basophils, and CD4⁺ T lymphocytes (Ponath et al., 1996a), and is a major binding site for CC chemokines (Daugherty et al., 1996; Gao et al., 1996; Kitaura et al., 1996; Ponath et al., 1996b; Forssmann et al., 1997; Heath et al., 1997). CCR3 has been cloned from guinea pig (Sabroe et al., 1998), murine (Gao et al., 1996), and human eosinophils (Ponath et al., 1996a), and the latter has been transfected into AML14.3D10 (Daugherty et al., 1996) and murine pre-B lymphoma cell lines (Ponath et al., 1996a) where it binds eotaxin, regulated on activation, normal T-expressed and secreted (RANTES), and monocyte chemotactic protein (MCP) 3 at levels that are indistinguishable from those achieved in binding studies with primary eosinophils. Furthermore, a study using an antagonistic monoclonal antibody demonstrated that >95% of the eosinophil's response evoked by eotaxin, RANTES, MCP-2, MCP-3, and MCP-4 was mediated through CCR3 (Heath et al., 1997). Eosinophils also express low levels of the chemokine receptor CCR1, which appears to mediate the effects of MIP-1 (Daugherty et al., 1996). The expression of CCR1 and CCR3 is up-regulated during the maturation of eosinophilic HL-60 cells, although the kinetics of these effects is different with CCR1 levels rising first (Tiffany et al., 1998). Significantly, increased CC chemokine receptor expression correlates with the development of specific binding sites for MIP-1 and eotaxin, and the accompanying ability of the cells to generate Ca²⁺ and chemotactic responses (Tiffany et al., 1998). CCR3 expression also is increased on eosinophilic HL-60 cells by IL-5, suggesting that the chemokine receptors represent a marker of late eosinophilic differentiation (Tiffany et al., 1998).

2. CXC Chemokines. Only one CXC chemokine, IL-8, is known to activate eosinophils. However, whether it mediates its effects through CXCR1 (IL-8_A-R) or CXCR2 (IL-8_B-R) is unexplored. Several inconsistencies exist in the literature with respect to the sensitivity of eosinophils to IL-8. Erger and Casale (1995) found that IL-8 promoted eosinophil migration across micropore filters and through monolayers of HUVECs and A549 cells. However, those findings contradict the results obtained in a previous study (Ebisawa et al., 1994). This discrepancy may have resulted from cell priming during the isolation of eosinophils (Rozell et al., 1996). Indeed, this explanation would be in agreement with a number of in vitro studies where IL-8-induced chemotaxis is observed only after preincubation of eosinophils with IL-5 or GM-CSF (Warringa et al., 1991, 1992b; Schweizer et al., 1994; Heath et al., 1997). Interleukin-8-induced chemotaxis is associated with an increase in the [Ca²⁺]_i (Collins et al., 1993) and actin polymerization (Schweizer et al., 1994). In vivo studies have established that IL-8 elicits eosinophil migration into the BAL fluid (Lagente et al., 1995) and skin of guinea pigs (Collins et al., 1993), although it is uncertain whether this is a direct or indirect response.

Complement is a collective term that refers to a group of at least 30 proteins including proteolytic proenzymes, nonenzymatic components from which active enzymes are derived, and receptors that together form part of an intricate enzyme system found in plasma (Ember and Hugli, 1997). Triggering of these systems sets in motion an amplification cascade that ultimately results in the formation of the, so-called, terminal attack sequence that promotes cell lysis and is central to protecting the host from invading parasites and microbes. In excess of nine receptors for complement fragments have been identified and characterized to some extent (see Ross, 1989; Krych et al., 1992; Wetsel, 1995; Ember and Hugli, 1997). The 74- to 77-amino acid anaphylatoxins (C3a, C4a, C5a) are known to signal through G protein-coupled receptors and these are described below. The remaining complement receptors relevant to eosinophil biology are discussed in XI. M.

2. Complement 4a Anaphylatoxin. The second component of the classical complement pathway, C4, is split by C1 into the anaphylatoxin C4a and the larger C4b. Controversy surrounds the mechanism by which C4a anaphylatoxin elicits its functional effects. Indeed, evidence is available that C4a interacts with a structurally distinct C4a receptor (Murakami et al., 1993; Ames et al., 1997) and that it shares the same receptor as C3a (Hugli, 1984; Gerard and Gerard, 1994). To the authors' knowledge, nothing is known of the effect of C4a on eosinophil function. However, a C3a receptor-mediated effect of C4a seems unlikely given that it fails to promote Ca²⁺ mobilization in RBL-2H3 cells stably transfected with an expression plasmid encoding the murine C3a receptor (Tornetta et al., 1997).

3. Complement 5a Anaphylatoxin. Activation of the complement cascade can result in enzymatic cleavage of complement C5 and the release of a small polypeptide, C5a, from the remainder of the molecule, C5b, which remains loosely attached to the catalyst C5 convertase. One possible source of C5 degradation is provided by the eosinophil itself when exposed to immune complexes or SOZ (Ogawa et al., 1981a). Under those conditions, eosinophils can secrete a neutral protease that cleaves C5 to yield an eosinophil chemotactic activity that may well be C5a (Ogawa et al., 1981a). In 1977, Klebanoff et al. reported that eosinophils taken from the peritoneum of a child with eosinophilic gastroenteritis were activated by C5a, suggesting that receptors for this anaphylatoxin were expressed. Subsequent studies confirmed the expression of C5a-binding sites on human eosinophils using [¹²⁵I]C5a as a ligand and identified two apparently distinct populations of saturable sites (Gerard et al., 1989). One of these is present in relatively low abundance (B_max = 15,000-20,000 sites/cell) for which C5a has high affinity (K_d = 31 pM). The other constitutes the majority (>90%) of the total binding capacity (B_max = 375,000 sites/cell) although the affinity (K_d = 100 nM) of C5a is considerably (>300-fold) lower.

The lipids 5-oxo-ETE (Powell et al., 1995; Schwenk and Schroder, 1995; O'Flaherty et al., 1996a; Czech et al., 1997), 5-oxo-15-HETE (Schwenk et al., 1992; Powell et al., 1995; O'Flaherty et al., 1996a; Czech et al., 1997), 5-HETE (O'Flaherty et al., 1996b), and 8,15-diHETE (Morita et al., 1990a; Sehmi et al., 1991) are powerful eosinophil chemoattractants. In addition 5-oxo-ETE, the most potent of these novel lipid mediators (Powell et al., 1995; O'Flaherty et al., 1996a), induces degranulation of GM-CSF-treated eosinophils and enhances, by up to 10,000-fold, the ability of C5a, LTB₄, PAF, and fMLP to effect secretion of stored proteins (O'Flaherty et al., 1996a). Similarly, 5-oxo-ETE, at substimulatory concentrations, potentiates the chemotactic activity of PAF (Powell et al., 1995). Interestingly, 5-oxo-ETE, but not 5-HETE or 15-HETE, is approximately 100 times more potent as an eosinophil stimulant than its activity on neutrophils, suggesting that this compound may act selectively to induce eosinophil margination and activation (O'Flaherty et al., 1996a).

The cell surface receptor(s) on human eosinophils at which 5-oxo-ETE and 5-HETE act are not defined but their ligation results in rapid actin polymerization, intracellular Ca²⁺ mobilization, and the generation of oxygen radicals via a PTX-sensitive mechanism (Czech et al., 1997). Thus, the receptor for 5-oxo-ETE is likely to be G_i protein-coupled. 5-Oxo-ETE also enhances the expression of CD11b and the shedding of L-selectin by a mechanism that is insensitive to PD 098059, wortmannin, and staurosporine (Powell et al., 1999). Some 5-oxo-ETE-elicited responses might be attributable to its ability to promote the phosphorylation of ERK-1 and ERK-2 (O'Flaherty et al., 1996b).

In vivo, 5-oxo ETE, given by the intratracheal route to Brown Norway rats, produces a drammatic (5- to 8-fold) increase in the number of eosinophils around the airway wall that is not blocked by LTB₄ or PAF antagonists but is attenuated (~ 75%) by monoclonal antibodies directed against the adhesion molecules very late antigen (VLA) 4 and CD11a (Stamatiou et al., 1998). The magnitide of this effect is significantly greater than that effected by LTB₄.

Sensory neuropeptides represent a host of biologically active mediators, many of which have a variety of effects on eosinophil function. The most studied of these peptides include SP, NKA, NKB, calcitonin gene-related peptide (CGRP), gastrin-releasing peptide, peptide histidine isoleucine, secretin, helodermin, secretoneurin, cholecystokinin octapeptide, and vasoactive intestinal peptide (Goetzl and Sreedharan, 1992), and some of these are discussed below.

1. Substance P. Substance P (SP), NKA, and NKB comprise the tachykinins and exert many (if not all) of their effects by acting through at least three structurally distinct, seven transmembrane-spanning receptors denoted neurokinin (NK) ₁, NK₂, and NK₃. In humans, the NK₁, NK₂, and NK₃ receptors are composed of 407, 398, and 468 amino acids, respectively, represent distinct gene products and couple primarily through the G_q/11 family of GTP-binding proteins. Heterogeneity of tachykinin receptors also is seen in cells and tissues from mice and rats. See Regoli et al. (1994) for additional details.

3. Secretoneurin. Dunzendorfer et al. (1998a,b) have reported that secretoneurin, a novel 33-amino acid peptide derived from secretogranin II (Kirchmair et al., 1993) that is released from sensory afferent C-fibers by capsaicin (Kirchmair et al., 1994), is an effective chemoattractant for human eosinophils with a potency 10 to 50 times less than SP, RANTES, and IL-8. Preliminary studies designed to evaluate the signaling pathway(s) utilized by secretoneurin established that chemotaxis was abolished by the PtdIns 3-kinase inhibitor wortmannin, but not by tryphostin-23, whereas the same response evoked by SP was inhibited by both pharmacological agents (Dunzendorfer et al., 1998a,b). Thus, it would appear that secretoneurin-induced human eosinophil chemotaxis is mediated, in part, by mechanisms distinct from those recruited by SP. Studies with the phosphodiesterase (PDE) inhibitor 3-isobutyl-1-methyl-xanthine (IBMX), which significantly attenuated secretoneurin-, but not SP-, induced chemotaxis support this idea (Dunzendorfer et al., 1998a,b).

4. Vasoactive Intestinal Peptide. In humans and rats, three receptors (PAC, VPAC₁, and VPAC₂) unequivocally have been defined at which vasoactive intestinal peptide (VIP) is an agonist. Molecular genetics has established that each receptor is encoded by a distinct gene that activates effector elements by coupling exclusively through G_s. See Harmar et al. (1998) for additional details.

Two subtypes (B₁ and B₂) of the bradykinin receptor have been defined by pharmacological and molecular techniques and additional evidence for a B₃ receptor has been provided from antagonist studies (Farmer, 1995). In humans, the B₁ and B₂ receptors are composed of 353 and 364 amino acids, respectively, represent distinct gene products and couple primarily through the G_q/11 family of GTP-binding proteins. See Hall (1997) for additional details.

Bradykinin has no known direct effect on eosinophil function although, in vivo, it promotes localized eosinophilia in several species, including the guinea pig (Fechter et al., 1986; Farmer et al., 1992) and rat (Pasquale et al., 1991; Bowden et al., 1994; Pires et al., 1994; Ferreira et al., 1996). Bradykinin B₂ receptors are implicated in the cavine model since eosinophil accumulation is suppressed by the B₂-selective antagonists NPC 567 and NPC 16731. In rats, bradykinin acts, in large part, by effecting the generation of lipoxygenase products (Pasquale et al., 1991).

Two receptors (ET_A and ET_B) for the endothelins have been classified in a number of species that couple to intracellular effectors through the G_q/G₁₁ family of GTP-binding proteins. In humans, the ET_A and ET_B receptors are composed of 427 and 442 amino acids, respectively, and are distinct gene products. Pharmacological evidence is available for ET_B receptor heterogeneity but this is yet to be confirmed at the molecular level. See Masaki et al. (1994) for additional details.

Fujitani et al. (1997) have reported that the appearance of eosinophils in the BAL fluid of allergen-challenged, sensitized BALB/c mice was significantly suppressed by BQ 123 (ET_A antagonist), SB 209670 (ET_A/ET_B antagonists), and a neutralizing anti-endothelin antibody. The additional finding that the ET_B antagonist BQ-788 was inactive suggests that endogenously released endothelin promotes pulmonary eosinophilia through an action at ET_A receptors. It is not known whether eosinophils express functional endothelin receptors but the mechanism of action of BQ 123 and SB 209670 in the murine model probably resides in their ability to release interferon  (IFN) from Th₁ T lymphocytes (Fujitani et al., 1997). Indeed, the i.v. administration of endothelin to guinea pigs does not cause pulmonary leukocyte accumulation (Macquin-Mavier et al., 1989).

Currently, four receptors for adenosine have been unequivocally defined in human tissues and are denoted A₁, A_2A, A_2B, and A₃. Each, so-called, purinoceptor is a member of the seven transmembrane-spanning family of receptors and couples to multiple effectors through G_i, G_o, and G_s. Thus, adenosine can act as an excitatory and inhibitory ligand depending upon the receptor subtype expressed by the cell or tissue of interest. In humans, the A₁, A_2A, A_2B, and A₃ purinoceptor are composed of 326, 412, 332, and 318 amino acids, respectively, and represent distinct gene products. Adenosine receptor multiplicity also is found in cells and tissues from mice and rats. See Fredholm et al. (1994) for detailed description of classification.

Human and guinea pig eosinophils generate and release adenosine spontaneously in biologically active quantities. This phenomenon is seen in cells pretreated with adenosine deaminase and the adenosine receptor antagonist 8-phenyltheophylline (which does not inhibit PDE). Both of these pharmacological interventions augment the generation of superoxide anions in response to SOZ, indicating that adenosine acts in an autocrine manner to suppress, tonically, the activity of the NADPH oxidase (Yukawa et al., 1989a). Pharmacological experiments designed to determine the adenosine receptor coupled to the inhibition of the NADPH oxidase implicate the A₂ subtype since 5'-N-ethyl-carboximide adenosine has a greater inhibitory effect than R-N-phenyl-isopropyl adenosine (Yukawa et al., 1989a). Furthermore, adenosine has been shown to increase the [Ca²⁺]_i in fura-2/AM-loaded guinea pig eosinophils and to significantly enhance PAF-induced superoxide anion generation and Ca²⁺ mobilization (Walker, 1996). The receptor that mediates these latter effects is not defined but is unlikely to be either of the A₂ purinoceptor subtypes since they couple predominantly through G_s.

In situ hybridization and polymerase chain reaction (PCR) studies have localized transcripts of the adenosine A₃ receptor to human eosinophils from normal and atopic donors (Kohno et al., 1996; Walker et al., 1997). More detailed experiments have established that eosinophil membranes express a homogeneous population of noninteracting, high-affinity (K_d = 3.2 nM) binding sites for ²⁵I-labeled N⁶-(4-aminobenzyl)-adenosine-5'-N-methyluronamide, an adenosine A₃ receptor agonist, with a B_max of 1.3 pmol/mg protein (Kohno et al., 1996). Intriguingly, the density of adenosine A₃ receptor transcripts is higher in lung tissue taken from subjects with airway inflammation than from normal donors (Walker et al., 1997), although whether this is associated with an increase (or decrease) in functional receptors is currently unknown. In addition to suppressing the activation of the NADPH oxidase via A₂ receptors (see above), adenosine exerts effects in eosinophils through agonism of the adenosine A₃ receptor that are considered to be both proinflammatory and anti-inflammatory. Thus, PAF-, RANTES-, and LTB₄-induced chemotaxis of human eosinophils is prevented by 3-(3-iodo-4-aminobenzyl)-8-(4-oxyacetate)phenyl-1-propyxanthine, a selective antagonist at A₃ receptors (Knight et al., 1997; Walker et al., 1997). In contrast, the highly potent and selective A₃ agonist CI-IB-MECA mobilizes Ca²⁺ from both intracellular stores and from the extracellular compartment, suggesting that the A₃ purinoceptor can couple to a PLC (Kohno et al., 1996).

Histamine (2-(4-imidazole)ethylamine) can act at three distinct receptors denoted H₁, H₂, and H₃. Classical pharmacology, allied with molecular techniques, has identified the H₁ and H₂ receptor in humans, mice, and rats and has established that they belong to the seven transmembrane-spanning family of receptors. The human H₁ and H₂ receptors are composed of 487 and 359 amino acids, respectively, and are the products of different genes. Although the H₁ receptor couples primarily through a PTX-insensitive G protein, probably of the G_q/11 class, the H₂ receptor is linked to effector enzymes via G_s. Thus, like adenosine, histamine can activate or inhibit depending upon the receptor subtype expressed by the cell or tissue of interest. The H₃ receptor has not yet been cloned but has pharmacology distinct from the other histamine receptors. The effector molecules involved in H₃ receptor signaling are unknown, although radioligand-binding experiments imply a possible link to a G protein. See Hill et al. (1997) for detailed review.

Pharmacological evidence points to the expression of H₁, H₂, and H₃ histamine receptors on human eosinophils and much of this is derived from experiments assessing locomotion in vitro. Although many studies have examined the chemotactic potential of histamine in several species (Clark et al., 1975, 1977; Bryant et al., 1977; Jones and Kay, 1977; Wadee et al., 1980; McEwen et al., 1990; Foster and Cunningham, 1998), much of those data are contradictory with respect to the receptor subtype(s) involved. Thus, in the late 1970s, the chemotactic activity of histamine on guinea pig eosinophils was attributable to an interaction at receptors of the H₂ subtype (Jones and Kay, 1977). In contrast, similar experiments performed at the same time with human cells failed to corroborate that finding and, instead, proposed the existence of a novel receptor based on the finding that histamine-induced chemotaxis was not blocked by H₁ or H₂ antagonists (Clark et al., 1975, 1977). Further discrepancy is provided by the results of additional studies where histamine was shown to augment human eosinophil chemokinesis (random migration), effected by endotoxin-activated serum, through a pyrilamine (H₁)-sensitive receptor (Clark et al., 1977; Wadee et al., 1980), whereas, in the same experimental setup, directional motility (chemokinesis) was mediated through H₂ receptors (Clark et al., 1977; Wadee et al., 1980). In equine eosinophils, histamine promotes migration and adherence to serum- and fibronectin-coated plastic solely through the histamine H₁ subtype (Foster and Cunningham, 1998). Thus, although species differences should not be discounted, the histamine receptor subtype(s) that promotes eosinophil locomotion still is equivoval.

Histamine H₃ receptors were identified on human eosinophils by Raible et al. in the early 1990s. Using Ca²⁺ mobilization as an index of activation, the affinity of the selective H₃ antagonists burimamide, thioperamide, and impromidine were similar to those calculated for the H₃ receptors in the central nervous system (Raible et al., 1992, 1994). However, R--methylhistamine and N- methylhistamine (H₃-selective agonists) were less active than histamine itself which led Raible et al. (1994) to suggest that the eosinophil H₃ receptor is different from those expressed in other tissues. However, low receptor expression or poor receptor-effector-coupling efficiency equally could explain this apparently anomalous result.

With the possible exception of motility, the functional effects of histamine in eosinophils are surprisingly little studied. Reports that histamine evokes superoxide anion generation from guinea pig and human eosinophils (Pincus et al., 1982) and enhances C3b rosette formation (Anwar and Kay, 1980) by a H₁ receptor-mediated mechanism have been suggested but not corroborated.

One of the first investigations to address the in vivo effects of histamine was published by Vegad and Lancaster (1972) who reported that local application produced cutaneous eosinophilia in sheep. That finding has since been confirmed in guinea pigs (Woodward et al., 1985) and in humans, where the chemoattraction was greater in atopic subjects when compared to normal individuals (Bryant and Kay, 1977). A role for histamine in eosinophil recruitment is not restricted to the skin. Histamine promotes the emigration of eosinophils to the conjunctiva of guinea pigs (Woodward et al., 1986; Spada et al., 1986) and also is implicated in allergen-induced pulmonary eosinophilia in sensitized dogs (Johnson et al., 1992). In a guinea pig model of cutaneous and conjunctival eosinophilia, pyrilamine and cimetidine administered concurrently is necessary to significantly blunt eosinophil infiltration, indicating that histamine H₁ and H₂ receptors are involved (Woodward et al., 1985, 1986). However, eosinophil trafficking was not abolished by that treatment, tempting speculation that H₃ receptors also play a role (Woodward et al., 1986). Paradoxically, local application of histamine to unroofed heat-suction blisters of ragweed-sensitive subjects inhibited allergen-induced cutaneous eosinophilia (Ting et al., 1981). An important role for inhibitory H₂ receptors is, therefore, proposed.

Elegant studies performed since the mid-1970s have provided pharmacological evidence for five main classes of receptor for the naturally occurring prostanoid agonists (reviewed in Coleman et al., 1994). These receptors have been given the prefix DP-, EP-, FP-, IP-, and TP- and belong to the G protein-coupled receptor superfamily. Because of the lack of selective antagonists, this taxonomy was formulated predominantly from rank orders of agonist potencies obtained in various pharmacological preparations where each prostanoid is at least one order of magnitude more potent than the others at a specific prostanoid receptor. Molecular biological techniques have recently confirmed this pharmacological classification with the cloning and expression of cDNAs for representatives of the five prostanoid receptors in a number of species including humans (Hirata et al., 1991; Abramovitz et al., 1994; Boie et al., 1994,1995; Kunapuli et al., 1994; Regan et al., 1994a,b; Yang et al., 1994).

In vitro studies suggest that eosinophils might express excitatory DP receptors based on the finding that prostaglandin (PG) D₂ (but not PGF₂ or TX mimetics) enhances zymosan-activated serum-induced eosinophil migration (Butchers and Vardey, 1990). This possibility is supported by an earlier description of the chemokinetic activity of PGD₂ (Goetzl et al., 1979) and its ability to promote Ca²⁺ mobilization in fura-2-loaded human eosinophils (Raible et al., 1992). In vivo, PGD₂ promotes eosinopenia and the accumulation of eosinophils in the airways of dogs (Marsden et al., 1984; Emery et al., 1989) in a manner that is attenuated by the nonselective prostanoid receptor antagonist SK&F 88046. Thus, it seems likely that the chemokinetic action of PGD₂ results from a direct action on the eosinophil (Emery et al., 1989). Furthermore, PGD₂ (acting through TP receptors on the airways smooth muscle) evokes potent bronchoconstriction in humans (Beasley et al., 1989; Johnston et al., 1992). This effect raises important clinical considerations given that PGD₂ is present in the BAL fluid of mild asthmatic subjects and is released into the lungs following acute allergen provocation (Murray et al., 1986; Liu et al., 1990).

Evidence derived from pharmacological studies suggests that eosinophils express a population of prostanoid receptors that, when activated, suppress several indices of activation. Butchers and Vardey (1990) reported that fMLP-induced ECP release from a mixed population of human granulocytes was suppressed by PGD₂, PGE₂, and PGF₂ with a rank order of potency in good agreement with that found with other cells and tissues that express DP receptors such as human platelets (Keery and Lumley, 1988). Similarly, the synthetic PGD₂ agonist BW 245C was more potent than the natural ligand at blocking degranulation (Butchers and Vardey, 1990). In complete agreement with those data, Sturton and Norman (1991) noted that PGD₂ was the most effective natural prostaglandin at preventing fMLP-induced respiratory burst (assessed as luminol-enhanced chemiluminescence) in human eosinophils. Thus, it appears that DP receptors can mediate both excitatory and inhibitory effects in human eosinophils that might reflect DP receptor heterogeneity (see Fernandes and Crankshaw, 1995).

PGE₂ inhibits, albeit modestly (20-30%), PAF-induced CD11b expression and the shedding of L-selectin on human eosinophils (Berends et al., 1997), implying that their interaction with the appropriate counter ligands on vascular endothelial cells would be reduced. A similar result was documented for PGE₁ which attenuated the up-regulation by PAF and C5a of the ₂ integrin CD18 in guinea pig eosinophils (Teixeira et al., 1996a). This action would temper the directional migration of eosinophils in response to chemoattractants and might attenuate eosinophil-driven inflammatory responses. The identity of the prostanoid receptor subtype at which E-series prostaglandins suppress adhesion molecule expression has not been determined, although it is curious that the PDE4 inhibitor rolipram is inactive, which tempts speculation that EP receptors coupled positively to adenylyl cyclase are not involved. Nevertheless, eosinophils may express inhibitory prostanoid receptors of the EP₂ subtype (Butchers and Vardey, 1990; Teixeira et al., 1997a). In human cells, this is suggested by the finding that PGE₂ increases the cAMP content (indicative of agonism at EP₂ or EP₄ receptors (Coleman et al., 1994)), and that misoprostol (EP₂-/EP₃-selective agonist), but not sulprostone (EP₁-/EP₃-selective agonist), inhibits fMLP-induced ECP release. In guinea pig eosinophils, pharmacological evidence based on the rank order of agonist potency (PGE₂ > PGE₁ > 11-deoxy-PGE₁ > misoprostol > butaprost > AH 13205) also implicates EP₂ receptors in the inhibition of PAF-induced homotypic aggregation (Teixeira et al., 1997a). In those studies, the selective EP₂ agonists butaprost and AH 13205 were uniformly weak, which might question the classification of the inhibitory eosinophil EP receptor as an EP₂ subtype. However, comparable results have been described in rat neutrophils (Wise and Jones, 1994) and human monocytes (Meja et al., 1997) that express EP₂-like receptors. Thus, given the high selectivity of butaprost for EP₂ receptors, an alternative possibility is that guinea pig eosinophils express a modest number of EP₂-binding sites at which butaprost and AH 13205 have low efficacy. Regardless of their precise identity, the inhibitory EP receptors are apparently coupled positively to adenylyl cyclase since inhibition of PKA reduced the ability of PGE₁, 11-deoxy-PGE₁, and AH 13205 to suppress PAF-induced aggregation (Teixeira et al., 1996a, 1997a).

In vivo, E-series prostaglandins inhibit cutaneous eosinophilia in guinea pigs in response to PAF and compound 48/80 and after passive cutaneous anaphylaxis under conditions where local edema formation is enhanced (Teixeira et al., 1993). Prostaglandins exert several direct effects on eosinophils that could contribute to their ability to reduce eosinophil number to sites of an inflammatory insult (see above). However, the accumulation of neutrophils in the skin of guinea pigs is enhanced by PGE₁ and PGE₂, whereas in vitro neutrophil activation is generally attenuated (Teixeira et al., 1996a, 1997a; Berends et al., 1997). Thus, E-series prostaglandins might affect eosinophil emigration indirectly. However, studies with the long-acting ₂ adrenoceptor agonist salmeterol (Teixeira and Hellewell, 1997a) has provided persuasive evidence that agents that elevate cAMP can prevent eosinophil locomotion; thus, the mechanism of action of E-series prostaglandins in vivo remains to be elucidated.

Neither functional nor radioligand-binding experiments have provided any evidence for IP, FP, or TP receptors on human or guinea pig eosinophils (Butchers and Vardey, 1990; Giembycz et al., 1990; Sturton and Norman, 1991). As described in XII. C.2, the major cyclooxygenase products generated by PAF- and LTB₄-stimulated eosinophils are TX and PGE₂ (Giembycz et al., 1990; Perkins et al., 1995). However, exposure of guinea pig eosinophils to the cyclooxygenase inhibitor flurbiprofen, at a concentration that abolished PGE₂ generation, did not affect LTB₄- or PAF-induced functional responses (Giembycz et al., 1990; Rabe et al., 1992), indicating that this prostanoid is not generated in an amount sufficient to act in an autocrine manner.

Although formal identification (by radioligand binding or pharmacological means) of cell surface  adrenoceptors is lacking, Masuyama and Ishikawa (1985) suggested that they might be expressed on guinea pig eosinophils based on the finding that noradrenaline (-selective) inhibited eosinophil phagocytosis and free radical production under conditions where isoprenaline (-selective) was less active. However, in the absence of data obtained with selective agonists and antagonists, the expression of ₁ or ₂ adrenoceptors (or subtypes thereof) on eosinophils is equivocal.

1. Receptors. Three distinct  adrenoceptor subtypes (₁, ₂, and ₃) have been unequivocally classified. Each subtype is a member of the seven transmembrane-spanning family of receptors and is the product of a different gene. In humans, the ₁, ₂, and ₃ adrenoceptor are composed of 477, 413, and 408 amino acids, respectively, and interact predominantly, but not exclusively, with G_s-linked effectors (see Bylund et al., 1994 for details). Pharmacological evidence also is available for ₄ adrenoceptors (Molenaar et al., 1997).

2. Activation of the NADPH Oxidase. In human and guinea pig eosinophils, ₂ adrenoceptor agonists effectively suppress the activation of the NADPH oxidase (Rabe et al., 1993; Dent et al., 1994; Hadjokas et al., 1995; Ezeamuzie and Al-Hage, 1998). In the latter species this effect may not be mediated by receptors of the ₁ or ₂ subtype since the affinities of propranolol (pA₂ = 7.2), atenolol (pA₂ > 5), and ICI 118,551 (pA₂ ~ 7.1) in antagonizing LTB₄-induced H₂O₂ generation (a reliable measure of the respiratory burst) are considerably less than would be predicted from an interaction with classical ₁ or ₂ adrenoceptors (Rabe et al., 1993). Moreover, the long-acting ₂ adrenoceptor agonist salmeterol is inactive at suppressing oxidant production in response to LTB₄ and actually behaves as an antagonist at this "atypical" receptor subtype with reasonable affinity (pA₂ = 5.9) (Rabe et al., 1993). This finding also provides additional evidence for atypical  adrenoceptors on eosinophils. Indeed, logic dictates that if ₂ adrenoceptors were involved, salmeterol should inhibit H₂O₂ generation since it has essentially the same efficacy as salbutamol (Dougall et al., 1991).

3. Degranulation. Another in vitro functional response where  adrenoceptor agonists demonstrate an inhibitory effect is on degranulation. In human normodense eosinophils, isoprenaline, salbutamol, and eformoterol inhibit (albeit weakly) the secretion of products (ECP, EDN, or EPO) stored within the specific granules in response to fMLP (Munoz et al., 1995; Ezeamuzie and Al-Hage, 1998), PAF (Eda et al., 1993a), and Ig (IgG and secretory IgA)-coated Sepharose beads (Kita et al., 1991b). Curiously, salmeterol is inactive at blocking fMLP-induced EPO release (Munoz et al., 1995; Ezeamuzie and Al-Hage, 1998) and actually blocks the inhibitory effect of salbutamol under the same experimental conditions, although the nature of the antagonism was not elucidated (Munoz et al., 1995). Those findings confirm previous observations with guinea pig eosinophils (Rabe et al., 1993) that salmeterol can act as a competitive  adrenoceptor antagonist.

4. Chemotaxis and Chemokinesis. Salmeterol and formoterol partially inhibit PAF- and fMLP-induced chemotaxis of human eosinophils under experimental conditions where salbutamol is inactive (Koenderman et al., 1992; Eda et al., 1993a; Tool et al., 1996). However, the concentrations required to achieve this effect are very high (1-100 µM) and in excess of those required to increase maximally the cAMP content of eosinophils, inhibit homotypic aggregation (see below), and effect airways smooth muscle relaxation. Thus, the relevance of these findings in relation to the concentration of ₂ adrenoceptor agonist achieved in clinical practice is questionable. Isoprenaline similarly inhibits eosinophil chemotaxis stimulated by endotoxin-activated serum using two indices of migration, the Zigmond-Hirsch assay and a nucleopore filter assay (Clark et al., 1977). It would appear that species or the nature or concentration of the activating stimulus has a profound effect on whether or not  adrenoceptor agonists are active given that isoprenaline does not inhibit chemotaxis of guinea pig eosinophils (Sugasawa and Morooka, 1992).

5. Adhesion and Adhesion Molecule Expression. In anesthetized, pathogen-free F344 rats, the i.v. administration of SP and bradykinin produces an inflammatory response in the airways characterized by the adherence of proinflammatory leukocytes to venular endothelial cells along with plasma extravasation and edema (see V.H.1 and V.I). Bowden et al. (1994) demonstrated that acute administration of rats with eformoterol reduced the number of eosinophils adherent to venules in the airway mucosa in response to both inflammatory stimuli. This effect was mediated by ₂ adrenoceptors since it was abolished by ICI 118,551 (Bowden et al., 1994). A clue to the mechanism of action of eformoterol in that model can be inferred from a study by Berends et al. (1997) in which isoprenaline, at a maximally effective concentration, suppressed the up-regulation of the adhesion of CD11b (by 43%) and the shedding of L-selectin (by 34%) on human eosinophils evoked by PAF.

6. Membrane Lipid Metabolism. Few reports have appeared in the literature describing the effect of ₂ adrenoceptor agonists on the liberation of lipid mediators from eosinophils and the little information available is contradictory. For example, the short-acting ₂ adrenoceptor agonist salbutamol has been reported to inhibit fMLP-, C5a-, and PAF-induced LTC₄ generation from human eosinophils (Munoz et al., 1994; Tenor et al., 1996), whereas salmeterol was inactive under roughly comparable experimental conditions at concentrations that suppressed chemotaxis (Tool et al., 1996). Salmeterol similarly failed to prevent fMLP-induced PAF generation (Tool et al., 1996).

7. Homotypic Aggregation. The ability of guinea pig eosinophils to undergo homotypic aggregation in response to PAF and C5a is effectively antagonized by  adrenoceptor agonists (Teixeira et al., 1996a; Teixeira and Hellewell, 1997a). In fact, salbutamol is significantly more potent at suppressing aggregation than H₂O₂ formation with an EC₅₀ similar to that required for cAMP accumulation. Moreover, in contrast to studies on the NADPH oxidase, the PDE4 inhibitor rolipram markedly potentiates the inhibitory effect of salbutamol at a concentration that has no effect on aggregation per se (Teixeira et al., 1996a), suggesting that cAMP-dependent mechanisms regulate this response. It is intriguing that whereas salmeterol fails to inhibit H₂O₂ generation from LTB₄-stimulated eosinophils (Rabe et al., 1993) and actually behaves as a  adrenoceptor antagonist, PAF- and C5a-induced homotypic aggregation are, paradoxically, suppressed in a propranolol-sensitive manner (Teixeira et al., 1996a; Teixeira and Hellewell, 1997a). Several explanations can be advanced for this discrepancy, although no firm conclusion can be drawn at the present time. The first is that guinea pig eosinophils express two populations of inhibitory  adrenoceptor that regulate, independently, the cell-signaling pathways responsible for the activation of the NADPH oxidase and homotypic aggregation. This hypothesis would be consistent with the anomalous pA₂ values that have been calculated for a range of  adrenoceptor antagonists in chemotaxis and respiratory burst assays (Sugasawa and Morooka, 1992; Rabe et al., 1993). Alternatively, the sensitivity of the signal transduction pathway that ultimately promotes homotypic aggregation to the inhibitory action of cAMP might be considerably greater that those mechanisms that govern the activation of the NADPH oxidase. However, it is noteworthy that the failure of PDE4 inhibitors to potentiate the inhibitory effect of salbutamol on H₂O₂ generation is not consistent with a cAMP-dependent mechanism of action. Thus, as in other tissues,  adrenoceptor agonists might recruit multiple and distinct signal transduction cascades that negatively regulate eosinophil activation (Maguire and Erdos, 1980; Barber et al., 1989; Rooney et al., 1991; Vaziri and Downes, 1992; Wu et al., 1995; Xiao et al., 1995) which can theoretically involve signaling via G and G heterodimers (Daaka et al., 1997).

8. In Vivo Effects. The effect of  adrenoceptor agonists on stimulus-induced eosinophil recruitment in vivo is the subject of some debate. When acute studies are performed in laboratory animals, short- and long-acting ₂ adrenoceptor agonists are generally active (but see Banner et al., 1995; Namovic et al., 1996) at preventing pulmonary and cutaneous eosinophilia in response to a variety of stimuli including allergen (Fugner, 1989; Whelan and Johnson, 1990, 1992; Sanjar et al., 1991; Whelan and Johnson, 1990, 1992; Sugiyama et al., 1992; Teixeira et al., 1993, 1995a; Whelan et al., 1993; Howell et al., 1995; Teixeira and Hellewell, 1997a). Similarly, in humans, the systemic administration of isoprenaline can decrease circulating eosinophil number (Ohman et al., 1972) which may be responsible, at least in part, for the ability of ₂ adrenoceptor agonists to abolish cutaneous eosinophilia in sensitized human volunteers (Ting et al., 1983). It is likely that part of the inhibitory effect of ₂ adrenoceptor agonists on eosinophil recruitment is due to a direct effect on the eosinophil (Teixeira and Hellewell, 1997a). This is suggested from a study performed with salmeterol-treated, ¹¹¹In-labeled guinea pig eosinophils (where the inhibitory effect persists for many hours even after extensive washing) which, when injected into recipient guinea pigs, do not migrate to skin sites exposed to proinflammatory stimuli (Teixeira and Hellewell, 1997a).

Five distinct somatostatin receptors (denoted sst₁ to sst₅) have been identified in humans and mice and belong to the seven transmembrane-spanning family of receptors. Each sst receptor is the product of a different gene and couples primarily to G_o/i. See Bruns et al. (1995) for additional details.

Eosinophils have the capacity to synthesize, store, and release (Aliakbari et al., 1987) somatostatin, although it is not known whether they express cognate sst receptors. However, the somatostatin antagonist lanveotide effectively inhibits the peripheral blood and peritoneal eosinophilia precipitated in rats by i.p. administration of Sephadex beads, cyclophosphamide, PAF, or allergen (in sensitized animals) (Etienne et al., 1989a,b). Since somatostatin is known to affect T lymphocyte proliferation, and since T cells are involved in the differentiation of hematopoetic cells to eosinophils, it is possible that somatostatin decreases, indirectly, eosinophil availability and recruitment.

Structurally, lipoxins are acyclic eicosanoids that contain a conjugated tetraene structure and three alcohol groups (Serhan et al., 1984a,b; Serhan and Samuelsson, 1988; Steinhilber and Roth, 1989; Serhan, 1991). The two major lipoxins in this series of eicosanoids are positional isomers and have been named LXA₄ (5S,6R,15S-trihydroxy-7,9,13-trans-11-cis-eicosatetraenoic acid) and LXB₄ (5S,14R,15S-trihydroxy-6,10,12-trans-8-cis-eicosatetraenoic acid). Other lipoxins also have been identified and are known as LXC₄, LXD₄, and LXE₄ (Steinhilber and Roth, 1989). The human and murine LXA₄ receptors have been cloned, expressed, and their distribution at the mRNA level mapped (Fiore et al. 1994; Serhan et al., 1994; Takano et al., 1997). In the mouse, LXA₄ receptor mRNA transcripts are most abundantly expressed in blood leukocytes followed by the spleen and lung (Takano et al., 1997). Both receptors have a sequence indicative of a seven transmembrane-spanning G protein-coupled receptor and share 73% identity at the amino acid level (Fiore et al. 1994; Serhan et al., 1994; Takano et al., 1997). Binding studies have established that ³H-labeled LXA₄ interacts with LXA₄ receptors with high affinity (K_d = 1-2 nM); LTD₄ effectively competes for this site whereas LXB₄ does not, indicative of lipoxin receptor heterogeneity. In CHO cells transfected with the murine or human LXA₄ receptor, LXA₄ promotes GTP hydrolysis and the release of esterified arachidonic acid by a pertussis toxin-sensitive mechanism (Fiore et al., 1994, Takano et al., 1997). These results are consistent with findings in human neutrophils where LXA₄ evokes functional responses through G_i/G_o-coupled receptors (Fiore et al., 1994). Phylogenetically, the murine and human LXA₄ receptor belongs to the CC chemokine family of G protein-coupled receptors rather than to the eicosanoids such as the prostanoids (Toh et al., 1995).

Receptors for LXA₄ have not been unequivocally identified on eosinophils but they are probably expressed based on functional studies. Thus, although little is known of the biological activities of the lipoxins, LXA₄ is weakly chemotactic for human eosinophils, evoking responses about 20% of that produced by PAF and fMLP. In addition, LXA₄ inhibits PAF- and fMLP-induced eosinophil migration (Soyombo et al., 1994) but has no effect on ECP release per se or on degranulation effected by fMLP (Soyombo et al., 1994). LXA₄ has been shown to activate PKC with potency greater than DAG (Hansson et al., 1986). However, it displays selectivity for PKC (Shearman et al., 1989) which predominates in the central and peripheral nervous systems but is not expressed by human eosinophils (Evans et al., 1999). The biological activities of LXB₄, C₄, D₄, and E₄ equally are obscure.

	VI. Interleukin-3, Interleukin-5, and Granulocyte/Macrophage Colony-Stimulating Factor

Top Previous Next References

The hematopoietins, which include IL-3, IL-5, and GM-CSF, are important regulators of eosinophil function and exert both distinct and overlapping effects (Tables 11, 12, and 13 for details and Miyajima et al., 1992a,b). The IL-5 receptor in humans is selectively expressed by eosinophils and basophils but not neutrophils or monocytes (Chihara et al., 1990; Ingley and Young, 1991). This contrasts with cell surface receptors for IL-3 and GM-CSF that have a more ubiquitous distribution (Clutterbuck et al., 1989; Ogawa, 1993).

Radioligand-binding experiments using ¹²⁵I-labeled IL-5 have demonstrated cross-competition among IL-3, IL-5, and GM-CSF (Lopez et al., 1989, 1991) due to a structural similarity in hematopoeitic cytokine receptors. Thus, all three receptors are composed of two subunits: a 60- to 80-kDa  subunit, that is unique to each receptor, and a common  subunit (_c), which has a mass between 120 and 140 kDa (Tavernier et al., 1991). Interleukin-3, IL-5, and GM-CSF interact with the  subunit of their respective receptors with low affinity, whereas the additional interaction with the _c subunit results in the formation of a high-affinity ligand-receptor complex, thereby permiting cell signaling to occur (Miyajima et al., 1992b; Murata et al., 1992; Koike and Takatsu, 1994). It is possible that the cross-competition between cytokines results from a limited number of _c subunits that would limit the extent of eosinophil activation. In human eosinophils, a single population of receptors for IL-5 has been identified although the binding constants are variable. Thus, IL-5 has been reported to interact with eosinophils with an affinity of 19 pM (Tagari et al., 1993), 120 pM (Lopez et al., 1991), 170 to 330 pM (Migita et al., 1991), and 400 pM (Ingley and Young, 1991); a broad spectrum of B_max values (260-1500 sites/cell) also has been reported (Migita et al., 1991; Lopez et al., 1991; Okada et al., 1998). Less research has been done with IL-3 and GM-CSF but they appear to interact with a single class of noninteracting sites with K_d values of 470 pM and 44 pM, respectively (Lopez et al., 1989).

Regulation of the IL-5 receptor, as well as of the synthesis, storage, and release of IL-5, clearly is important in determining eosinophil responses. However, relatively little is known of the factors that control the transcription and expression of these proteins. It has been reported that transforming growth factor (TGF) ₁ and phorbol 12-myristate 13-acetate (PMA) down-regulate IL-5 receptor  chain mRNA transcripts in vitro in a remarkably stimulus-specific manner (Zanders, 1994). Indeed, a host of other stimuli including ILs 1 to 11, G-CSF, GM-CSF, LIF, stem cell factor (SCF), erythropoetin, IFN-, RANTES, MIP-1, EGF, platelet-derived growth factor (PDGF), dexamethasone, forskolin, retinoic acid, and cyclosporin A were inactive. Conversely, up-regulation of IL-5 mRNA was observed in bronchial biopsies taken from asthmatic individuals (Yasruel et al., 1997). In that study, the majority of the IL-5 receptor mRNA was associated with eosinophils, suggesting that they represent the major target for IL-5-induced responses. The gene encoding the IL-5 receptor  subunit is located in region 3p26 of chromosome 3 (Tavernier et al., 1991) and encodes a membrane-anchored form that is produced by alternative mRNA splicing (Tavernier et al., 1992). In addition, two novel soluble isoforms, which are secreted into body fluids, also are produced that arise from either normal mRNA splicing or from the absence of a splicing event (Tavernier et al., 1992). Although the soluble isoforms bind IL-5 in in vitro assays, their role in vivo is presently unclear; however, it is likely that they neutralize the effect of IL-5 on target tissues (Tavernier et al., 1992; Devos et al., 1993). Recent studies have identified two functional promoter regions, P1 and P2, in the gene encoding the IL-5 receptor  subunit that are located in the 5' upstream regions of exon 1 (L. Sun et al., 1995) and exon 2 (J. Zhang et al., 1997), respectively. Using the eosinophilic cell line AML14, P1 promoter activity has been localized within a 561-base pair (bp) sequence proximal to the transcriptional start site (Z. Sun et al., 1995). 5'-Deletion mutants within that region have identied a 34-bp domain (432 to 398) that confers approximately 80% promoter activity and is highly active in a myeloid cell- and eosinophil-specific manner (Z. Sun et al., 1995). However, consensus sequences for known transcription factors are absent indicative of unique myeloid cell- and, possibly, eosinophil-specific, regulatory elements. Subsequent studies identified an enhancer element (EOS1) within the P1 promoter (Sun et al., 1996). A comparison with other models of transcription factor binding shows that EOS1 is similar to the bacterial helix-turn-helix phage  and 434 repressor-operator complexes, and the Cys4 zinc finger glucocorticoid response element (GRE) motifs. The possibility that the enhancer element may function as a GRE is supported by the identification of an AP-1-binding site adjacent to the EOS1 domain. This is significant as AP-1:GRE is a composite response element in the regulation of a number of genes (Sun et al., 1996). The P2 promotor is located within a 66-bp region (31 to +35) of exon 2 and features a 5'-CCAAT-3'-binding domain for the transcription factor CCAAT-enhancer binding protein (C/EBP), and two consensus motifs (5 to +1 and +13 to +18) for the oncogene c-ets (J. Zhang et al., 1997). However, of particular interest is the presence of a novel 6-bp element (5'-CTAATT-3'), spanning 19 to 14, that is essential for P2 promotor activity and which is activated by a transcription factor specific to the eosinophil lineage (J. Zhang et al., 1997).

The binding of IL-3, IL-5, and GM-CSF to their cognate receptors leads to the activation of multiple signaling pathways (Fig. 5; Koenderman et al., 1996; van der Bruggen and Koenderman, 1996; Yousefi et al., 1997). Although the  and _c subunits of hematopoietic receptors do not exhibit intrinsic kinase activity, activating cytokines cause rapid changes in the tyrosine phosphorylation of a number of cellular proteins (van der Bruggen et al., 1993a) through the recruitment of cytoplasmic tyrosine kinases and phosphatases. Ligation of the IL-5 receptor on human eosinophils induces a rapid recruitment of the tyrosine kinases lyn, syk, and Jak-2 to the _c subunit of the receptor (Alam et al., 1995; Pazdrak et al., 1995a,b; van der Bruggen et al., 1995; Bates et al., 1996) along with the tyrosine phosphatase SHPTP-2 (Pazdrak et al., 1997). Similarly, GM-CSF activates lyn and Jak-2 (Simon et al., 1997b). In addition, IL-5 promotes the phosphorylation of p52^shc, an adapter protein that physically links cell surface receptors to downstream signaling elements, and enhances its association to another adapter protein, Grb (Bates et al., 1998). Other early signaling events that occur in eosinophils exposed to IL-5 include the activation of PtdIns 3-kinase and the subsequent phosphorylation of PKB (Coffer et al., 1998). Despite these data, the down-stream biochemical events or the functional responses they ultimately promote are not clearly defined. However, it has been established that IL-5 stimulates the Ras-Raf1-MEK-ERK protein kinase cascade in human eosinophils (Alam et al., 1995; Pazdrak et al., 1995a; Bates et al., 1996; Coffer et al., 1998), although, at present, there are contradictory reports concerning the ERK isoform that is activated. Independent studies by Bates et al. (1996) and Hiraguri et al. (1997) found that anti-ERK antibodies immunoprecipitated three proteins of molecular weights 42, 44, 45 kDa and 40, 42, 44 kDa, respectively and, consistent with Pazdrak et al. (1995a), found that IL-5 activated the higher molecular weight species, that is probably ERK-1. However, those data contrast to the recent report of Coffer et al. (1998) who found that IL-5 only activated ERK-2. The upstream events linking the Ras-Raf1-MEK-ERK pathway to the IL-5 receptor have not been fully characterized but antisense studies have implicated a role for SHPTP-2 in ERK-2 activation (Pazdrak et al., 1997). Similarly, PtdIns 3-kinase and, possibly, PKB also are involved since the activation of ERK-1 by IL-5 and GM-CSF is inhibited by wortmannin (Hiraguri et al., 1997).

View larger version (23K): [in this window] [in a new window]   Fig. 5.   IL-5-induced signaling in human eosinophils. The binding of IL-5 with its cognate receptor and dimerization of the  and c subunits is believed to trigger the phosphorylation of tyrosine residues upon the latter by an, as yet, undefined mechanism. The phosphotyrosine residues then permit the binding and activation of a number of kinases including lyn, syk, and Jak-2, the phosphatase SHPTP-2, and, possibly, the scaffold or adapter proteins shc and Grb. A number of downstream kinase cascades and transcription factors then are activated including ras/raf1/MEK/ERK, PtdIns 3-kinase/PKB, JNK-54, and STAT-1 with resultant gene transcription. See VI.B for further details.

Other proteins necessary for signaling through the IL-5 receptor include the transcription factor signal transducers and activators of transcription (STAT) 1, which probably is activated by Jak-2 (Alam and Grant, 1995; Pazdrak et al., 1995b; van der Bruggen et al., 1995). de Groot et al. (1997) also have provided evidence that TPA-responsive element (TRE)- and diad symmetry element (DSE)-dependent transcription is regulated by Jak-2 and JNK-54.

Specific domains within the common  subunit of the IL-5 receptor initiate signaling to the cells' interior. Using truncated mutants of the cytoplasmic domain of _c subunit Sato et al. (1993) identified two functional regions: a membrane proximal domain (amino acid residues 456-517) essential for proliferation, activation of Jak-2 and induction of c-myc, and a second domain (amino acid residues 627-763) that is required for the binding of shc, activation of the p21^ras-Raf-l-MEK-ERK kinase cascade, and the induction of c-fos and c-jun. The association of SHPTP-2 with the IL-5 receptor _c has been demonstrated in a cell-free reconstituted system using a synthetic peptide (residues 605-624) of the latter incorporating Y⁶¹² (Pazdrak et al., 1997). Binding to this phosphotyrosine-containing peptide, but not a peptide in which the phosphorylated Y⁶¹² had been mutated to F, increases SHPTP-2 activity implying that direct binding can induce enzyme activation (Pazdrak et al., 1997). Three additional tyrosine residues (Y⁷⁵⁰, Y⁸⁰⁶, Y⁸⁶⁹) located carboxyl-terminal to amino acid 589 on the IL-5 receptor _c also have been found that are surrounded by a consensus sequence that favors the binding of SHPTP-2. Thus, the exact site at which SHPTP-2 binds remains unresolved, although Pazdrak et al. (1997) have speculated Y⁶¹² and/or Y⁷⁵⁰ are likely candidates. A pentapeptide sequence at amino acids 577 to 581 also has been identified that is central to the activation of JNK-54 and DSE-dependent transcription (de Groot et al., 1997).

In addition to the _c subunit, the cytoplasmic domain of the IL-5 receptor  subunit is apparently essential for IL-5-induced proliferation and the activation of c-jun, c-fos, and Jak-2 (Takaki et al., 1994; Cornelis et al., 1995; Muto et al., 1996).

A current model of the IL-5/IL-3/GM-CSF signaling pathway predicts that activating ligands induce a conformational change in their cognate receptors, which thereby activate receptor-bound tyrosine kinases (Fig. 5). These then tyrosine phosphorylate the common  subunit of the receptor to provide the binding sites for the recruitment and subsequent activation of lyn, syk, and SHPTP-2. The tyrosine kinases responsible for this event are likely to be bound to a proline-rich domain, also called a box-1 motif, at residues 458 to 465 as deletion of these amino acids prevents the tyrosine phosphorylation of the IL-5 receptor _c subunit (Itoh et al., 1996). Currently, the identity of this tyrosine kinase(s) is unknown but a case can be made for Jak-2 based on the finding that mutant cells lacking this kinase are unable to phosphorylate the IL-5 receptor _c subunit after stimulation with GM-CSF (Watanabe et al., 1997).

Interleukin-3, IL-5, and GM-CSF exert a range of effects on eosinophils (see Tables 11, 12, and 13). In particular, they are central in determining the number of eosinophils in the circulation and in tissues through their ability to promote production, proliferation, and differentiation (see III.) and to enhance their survival by suppressing apoptosis (see XII.H). Hematopoeitic cytokines are also implicated in the priming of mature eosinophils to a range of stimuli that evoke chemotaxis (see XII.A.3), degranulation (see XII.B), adhesion (see XII.A.2), and activation of the NADPH oxidase (see XII.G).

It is well established that administration of IL-5 to laboratory animals induces blood eosinophilia (e.g., Iwama et al., 1992) and IL-5 transgenic mice show life-long eosinophilia in organs without overt pathology, indicating that eosinophils require other factors for activation (Dent et al., 1990). The importance of IL-5 in allergen-induced tissue eosinophilia in laboratory animals also has been examined extensively and similar investigations now are emerging in humans. Generally, exposure of sensitized mice, rats, and guinea pigs to allergen results in the appearance of IL-5 and eosinophils in the BAL fluid. The pulmonary eosinophilia is dependent upon circulating, not locally produced, IL-5 (Wang et al., 1998) and is associated with an increase in airways reactivity to a variety of stimuli including acetylcholine (ACh), arecholine, histamine, and 5-hydroxytryptamine (Chand et al., 1992a; Gulbenkian et al., 1992; Nagai et al., 1993, 1996; Brunjzeel et al., 1993). Similar effects are seen in the pleural cavity of antigen-challenged sensitized mice (Bozza et al., 1994a). Almost without exception, the effect of neutralizing IL-5 with antibodies inhibits eosinophil infiltration but has a variable effect on airways responsiveness (Gulbenkian et al., 1992; Chand et al., 1992a; van Oosterhout et al., 1993; Nagai et al., 1993, 1996). Using the technique of adoptive transfer, it has been found that IL-5-secreting CD4⁺ Th₂-type cells in mice play a pivotal role in generating blood and airways eosinophilia and in the subsequent development of bronchial hyperreactivity and lung damage that occurs in response to aeroallergens (Hogan et al., 1998).

The effect of anti-IL-5 antibodies has not been reported in humans. However, a similar activity to that described in animals might be prediced given the reports of Shi et al. (1997, 1998) who found that IL-5 given to asthmatic subjects by the inhaled route, or instilled directly into the airways, produced pulmonary eosinophilia, and increased the number of eosinophils and the level of ECP in the induced sputum.

IL-5 is also involved in parasitosis and in helminth-induced airway hyperresponsiveness (Hall et al., 1998). Indeed, administration of the anti-IL-5 antibody TRFK-5 to mice inoculated with microfilariae of the filarial nematode Onchocerca lienalis reduces the ability of the animals to resist re-infection (Folkard et al., 1996). A similar approach has been adopted to show that IL-5 is important in driving eosinophilia and reducing parasite burden in mice exposed to Aspergillus fumigatus (Murali et al., 1993; Kurup et al., 1997), Toxocara canis (Buijs et al., 1995), and Angiostrongylus cantonesis (Sasaki et al., 1993).

	VII. Interferon Receptor Superfamily

Top Previous Next References

The IFN receptor superfamily, which includes receptors for IFN/, IFN, and IL-10, characteristically are single transmembrane-spanning glycoproteins with either one (IFN and IL-10) or two (IFN/) homologous extracellular regions that feature two fibronectin domains. Although, IFN/ (type I interferons) and IL-10 (a type II interferon) exert biological actions on human eosinophils (Table 14), only a receptor for IFN (type II interferon) has been convincingly identified (Aldebert et al., 1996; Ishihara et al., 1997). ¹²⁵I-labeled IFN labeled a single population of noninteracting sites on intact eosinophils with a K_d and B_max of 3.9 pM and 183 to 233 sites per cell, respectively (Aldebert et al., 1996). Although IFN binds with high affinity, the ability of the agonist-occupied receptor to signal requires a species-specific accessory protein that associates with an epitope on the intracellular domain of the receptor protein.

	VIII. Tumor Necrosis Factor Superfamily

Top Previous Next References

The tumor necrosis factor (TNF) or nerve growth factor (NGF) superfamily is composed of cytokine receptors and leukocyte surface glycoproteins. Members of this family are characterized by three to four cysteine-rich repeats of 40 amino acids in the extracellular portion of the molecule (Mallett and Barclay, 1991).

The type I (CD120a) and type II (CD120b) TNF receptors have respective molecular masses of 55 and 75 kDa and have been identified on human eosinophils by fluorescence-activated cell sorting analysis and immune electron microscopy (Zeck Kapp et al., 1994). Generally, the actions of TNF on eosinophils in culture or isolation usually are proinflammatory (Table 15). In vivo, antibodies against TNF significantly attenuated the development of fibrosis elicited by bleomycin in mice and the associated pulmonary eosinophilia, suggesting that TNF plays an important pathogenic role in that model (K. Zhang et al., 1997).

View this table: [in this window] [in a new window]   TABLE 15 Functional effects evoked by TNF- in human eosinophils

CD30 is a transmembrane receptor that was originally identified as a surface antigen on Reed-Sternberg cells in Hodgkin's disease and found subsequently to be preferentially expressed by human activated CD4⁺ T lymphocytes (Del Prete et al., 1995; Manetti et al., 1994). Eosinophils are CD30 cells but express an activating ligand CD30L (CD153) (Pinto et al. 1996) that has homology only with members of the TNF superfamily (Falini et al., 1995). The demonstration that native CD30L can transduce proliferative signals in CD30⁺ targets such as Hodgkin and Reed-Sternberg cells has suggested a possible role for eosinophils in the pathology of Hodgkin's disease (Pinto et al., 1996, 1997). This contention is supported by the higher than normal levels of CD30L expression on circulating and tissue eosinophils in patients with Hodgkin's disease and hypereosinophilic syndrome compared to normal subjects. In this respect, it is interesting that the expression of CD30L on eosinophils is increased by IL-3, IL-5, and GM-CSF (Pinto et al., 1996).

Originally identified on B lymphocytes and some carcinoma cell lines, CD40 is expressed on a variety of cells including eosinophils (Ohkawara et al., 1996). Structurally, CD40 is a 45- to 50-kDa transmembrane-spanning glycoprotein and, together with its activating ligand, CD40L (CD154 also called gp39), is thought to be important for the full expression of allergic inflammatory responses in the airways of animals and possibly humans (Lei et al., 1998). mRNA and surface protein for CD40 are expressed constitutively on circulating eosinophils of allergic patients and are up-regulated in response to IgA immune complexes and down-regulated by IL-10 (Ohkawara et al., 1996). Similarly, constitutive expression of CD40L on cells obtained from a hypereosinophilic patient has been reported along with the finding that normal eosinophils and the eosinophilic cell line Eol-3 will produce CD40L in response to fMLP, PMA, and ionomycin (Gauchat et al., 1995).

Functionally, cross-linking of CD40 increases eosinophil survival in a concentration-dependent manner by stimulating the release of GM-CSF (Ohkawara et al., 1996). In the presence of IL-4, eosinophils are able to induce CD40L-dependent B lymphocyte proliferation in vitro (Gauchat et al., 1995).

The CD69 antigen is a phosphorylated 28- to 32-kDa disulfide-linked homodimer that was first identified on activated T lymphocytes and natural killer cells in the late 1980s (for review, see Testi et al., 1994). Complementary DNA clones encoding human and mouse CD69 have been isolated and identified the antigen as a C-type lectin (Ziegler et al., 1994). Gene-mapping studies have placed CD69 on mouse chromosome 6 and the p13 region of human chromosome 12 (Ziegler et al., 1994). The role of CD69 as a possible marker of activated eosinophils was proposed shortly after it was originally described following the detection of significant levels of CD69⁺ cells in the BAL fluid, but not peripheral blood, of patients with eosinophilic pneumonia (Nishikawa et al., 1992). It is now known that CD69 is expressed on eosinophils taken from the BAL fluid of patients with mild asthma (Hartnell et al., 1993, Matsumoto et al., 1998) and on peripheral blood eosinophils during human parasitosis (Mawhorter et al., 1996) consistent with an activated phenotype. Indeed, CD69⁺ eosinophils are rapidly induced in vitro in response to IL-3, IL-5, GM-CSF, IFN, and IL-13 (Nishikawa et al., 1992; Hartnell et al., 1993; Luttmann et al., 1996; Mawhorter et al., 1996; Matsumoto et al., 1998). The induction of CD69 by GM-CSF is inhibited by cycloheximide, suggesting that new protein synthesis is required (Hartnell et al., 1993). However, it has been reported that protein and mRNA for CD69 are found within unstimulated eosinophils (Luttmann et al., 1996), although those data were not corroborated in a subsequent investigation (Matsumoto et al., 1998). The function of CD69 is largely unexplored but it might be involved in regulating longevity based on the finding that anti-CD69 antibodies promote apoptosis of GM-CSF-stimulated eosinophils (Walsh et al., 1996b).

Human CD95 (Fas/APO-1) is a membrane-associated polypeptide, has an approximate molecular mass of 48 kDa, and is comprised of 335 amino acids with a glycosylated amino-terminal extracellular domain, a hydrophobic middle, and an intracellular carboxyl terminus (Oehm et al., 1992; Smith et al., 1994). The amino terminus contains three cysteine-rich regions that are characteristic of the TNF/NGF receptor family whereas a 70-amino acid sequence at the carboxyl terminus features a, so-called, "death domain" that is necessary and sufficient for the transduction of signals that effect apoptosis (Itoh and Nagata, 1993).

Freshly purified eosinophils express CD95 at a low but consistent level (Matsumoto et al., 1995; Druilhe et al., 1996). However, following culture of eosinophils in the absence of cytokines the level of CD95 increases in a time-dependent manner that is associated temporally with reduced viability and an increase in the number of apoptotic nuclei (Druilhe et al., 1996). Similarly, cross-linking of CD95 with specific monoclonal antibodies produces a time- and concentration-dependent increase in apoptosis (Matsumoto et al., 1995; Tsuyuki et al., 1995; Druilhe et al., 1996). mRNA and protein for CD95 are up-regulated in human eosinophils cultured for 24 h with IFN and TNF, and synergy occurs when both cytokines are used concurrently. These effects are functionally relevant as eosinophils now display an enhanced rate of apoptosis in response to CD95L (Luttmann et al., 1998b). Significantly, IL-3, IL-5, and GM-CSF prevent CD95 expression by an unknown mechanism and this presumably contributes to their survival-prolonging activity (Luttmann et al., 1998b; see XII.H for additional details). Unlike human neutrophils, the activating ligand CD95L is not constitutively expressed on eosinophils (Liles et al., 1996). However, ligation of CD95 by CD95L present on activated T lymphocytes, for example, recruits a number of intracellular pathways in human eosinophils including JNK-54, lyn, and IL-1-converting enzyme-like proteases that are believed to couple the activation of an upstream sphingomyelinase to the degradation of lamin B₁ (Hebestreit et al., 1998; Simon et al., 1998). Indeed, the broad-spectrum tyrosine kinase inhibitors genistein and lavendustin A prevent CD95-mediated death in human and murine eosinophils in vitro and partially resolve CD95L-induced eosinophilia in an in vivo model of inflammation in the mouse (Simon et al., 1998). Lavendustin A also inhibits CD95-mediated lamin B₁ degradation which might account in part for its antiapoptotic activity (Simon et al., 1998).

Relatively little is known of the functional actions of NGF on eosinophils although chemotaxis, lavicidal activity and degranulation (Hamada et al., 1996; Solomon et al., 1998) all are accredited activities. NGF also suppresses fMLP-stimulated LTC₄ release (Takafuji et al., 1992).

	IX. Adhesion Molecules

Top Previous Next References

Adhesion molecules or receptors are thought to be central to the process of eosinophil migration from the systemic circulation into tissue (see XII.A). A number of adhesion molecules are expressed by eosinophils (Fig. 6) and can be divided into three families: the selectins, integrins, and immunoglobulins.

View larger version (31K): [in this window] [in a new window]   Fig. 6.   Eosinophil adhesion molecules/receptors and their counterligands. See IX for additional details.

Three selectin families (denoted E, P, and L) have been described. E-selectin (CD62E) and P-selectin (CD62P) are expressed on endothelial cells whereas L-selectin (CD62P) is found on the cell surface of leukocytes including eosinophils (see Bevilacqua and Nelson, 1993; Lasky, 1995). Structurally, the selectins are characterized by an amino-terminal C-type (Ca²⁺-dependent), lectin-like, binding domain, an EGF-like region, two to nine concensus repeats of sequence similar to those appearing in complement-regulatory proteins, such as decay-accelerating factor, a membrane-spanning domain, and a short cytoplasmic tail.

The expression of E-selectin by endothelial cells is induced by certain cytokines and requires gene transcription and protein synthesis (Bevilacqua et al., 1987). P-selectin is stored within cytoplasmic Weibel-Palade bodies from where it translocates to the plasma membrane within minutes of stimulation (Johnston et al., 1989; Geng et al., 1990). In contrast, L-selectin is constitutively expressed by eosinophils but is shed upon activation with stimuli such as A23187, PAF, fMLP, and IL-5 (Smith et al., 1992; Neeley et al., 1993). Those in vitro observations are entirely consistent with the lower than normal expression of L-selectin on eosinophils harvested from the sputum of asthmatic subjects when compared with blood eosinophils (in't Veen et al., 1998). The counterligands for selectins are a family of sialylated, fucosylated, and, in many cases, sulfated, glycosaminoglycans typified by the moiety sialyl Lewis X (Springer and Lasky, 1991). The precise carbohydrate moieties recognized by the selectins are presently unknown, although the peptide backbone appears to be important in conferring selectin specificity. The majority of selectin counterligands contain mucin regions that are characteristically serine/threonine/proline-rich peptide sequences with rigid backbones and are decorated with O-linked carbohydrates (Shimizu and Shaw, 1993). To date, three L-selectin counterligands have been identified upon endothelial cells: 1) GlyCAM-1, 2) MadCAM-1, and 3) CD34, which contain mucins or mucin-like domains (Lasky et al., 1992; Briskin et al., 1993; Baumhueter et al., 1994). Studies examining P-selectin-mediated binding of eosinophils to nasal polyp endothelial cells and to soluble P-selectin identified PSGL-1, a sialylated, homodimeric glycoprotein, as the eosinophil counterligand (Wein et al., 1995; Symon et al., 1996). PSGL-1 has been isolated by expression cloning from an HL-60 library and shown to be a 220-kDa homodimer with a heavily O-glycosylated mucin-like structure (Sako et al., 1993). Further structural analyses revealed that, in contrast to the 15-decapeptide repeat found in neutrophil PSGL-1, the corresponding eosinophil variant is 10-kDa heavier due to an extra repeat (Symon et al., 1996). The counterligand for E-selectin was identified from examining the interaction of eosinophils with soluble E-selectin immobilized upon plastic plates and identified as a sialylated, protease-resistant structure (Bochner et al., 1994).

The integrins constitute a superfamily of gene products that are composed of two noncovalently linked  and  transmembrane heterodimeric glycoproteins. Eosinophils express the ₁ (CD29) integrins: VLA-2 (CD49b), VLA-4 (CD49d), VLA-5 (CD49e), and VLA-6 (CD49f); the ₂ (CD18) integrins (Kuijpers et al., 1993): leukocyte function-associated antigen (LFA) 1 (CD11a, _L2), CR3 (CD11b, _M2, Mac-1), and complement receptor (CR) 4 (CD11c, _X2, p150,95); the novel integrin, _d2, which interacts with intercellular adhesion molecule (ICAM) 3, and the ₇ integrin, ₄₇ (CD49d/CD103) (Kuijpers et al., 1993; Walsh et al., 1996a). The integrins bind to members of the Ig superfamily expressed upon endothelial cells as well as components of the extracellular matrix (see Fig. 6).

1. ₁ Integrins. The most extensively studied ₁ integrins are VLA-4 and VLA-6 which bind to the extracellular matrix proteins fibronectin and laminin, respectively. Relatively little is known about VLA-2 and VLA-5. It has been demonstrated that VLA-4 binds to the, so-called, IIICS region of fibronectin that features a 25-amino acid alternatively spliced connecting segment, CS-1, which is recognized by the integrin through a characteristic LDV motif (Anwar et al., 1993, 1994). VLA-4 also can bind to VCAM-1, an Ig superfamily member expressed upon cytokine-exposed endothelial cells (Bochner et al., 1991a; Dobrina et al., 1991; Weller et al., 1991b; Atsuta et al., 1998). The interaction occurs at sites within the first and fourth Ig-like domains of the protein (Osborn et al., 1992; Vonderheide et al., 1994). The VLA-4:VCAM-1 and VLA-4:fibronectin interaction is encouraged when eosinophils are preincubated with GM-CSF (Sung et al., 1997), SCF (Yuan et al., 1997), and PAF (Anwar et al., 1994) and is due to an increase in ligand affinity rather than an up-regulation of receptor expression or changes in receptor distribution (Neeley et al., 1993; Sung et al., 1997; Yuan et al., 1997). The chemoattractants, RANTES, MCP-3, and C5a transiently increase VLA-4-mediated adhesion to purified VCAM-1 and fibronectin (Weber et al., 1996). However, those data contrast with the result of other experiments. In particular, Burke-Gaffney and Hellewell (1996) have found that the activation of human lung microvascular endothelial cells by TNF was associated with increased VLA-4-mediated adhesion of eosinophils after their exposure to eotaxin but not to RANTES or MIP-1.

2. ₂ Integrins. Human eosinophils express the common ₂ chain CD18 and the  chains CD11a (LFA-1), CD11b (CR3), and CD11c (p150,95) that bind to ICAM-1 (CD54) (Fischer et al., 1986; Hartnell et al., 1990; Walsh et al., 1990a,b; Kyan Aung et al., 1991a; Grayson et al., 1997). A fourth integrin, _d2, also is expressed on human eosinophils, is up-regulated by IL-5 and Ca²⁺ ionophore A23187, and can function as an alternative ligand for VCAM-1 (Grayson et al., 1998). The expression of CD11b is greater on eosinophils found in the sputum of asthmatic subjects when compared with their peripheral blood counterparts (in't Veen et al., 1998), and this is consistent with the ability of proinflammatory mediators to up-regulate ₂ integrins in general. CR3 numbers are increased on the surface of human eosinophils by PAF, IL-3, IL-5, GM-CSF (Thorne et al., 1990; Walsh et al., 1991a; Hartnell et al., 1992a; Lundahl et al., 1993; Neeley et al., 1993; Tsai et al., 1993; Fattah et al., 1996), and, to a lesser extent, TNF (Thorne et al., 1990), fMLP (Lundahl et al., 1993; Neeley et al., 1993), LPS (Lundahl et al., 1993), C5a (Gerard and Gerard, 1991; Lundahl et al., 1993), and RANTES (Alam et al., 1993). In contrast, the expression of LFA-1 seems to be more tightly controlled in that it is up-regulated only by PAF (Hayashi et al., 1994). A number of investigators have reported that in addition to receptor number, IL-3, IL-5, GM-CSF, and RANTES also increase the affinity of ₂ integrins for their counterligands (Blom et al., 1994; Kakazu et al., 1995).

3. ₄₇ Integrin. In addition to the formation of VLA-4, the ₄ subunit has been shown by immunostaining and flow cytometry to associate with the  subunit, ₇ to form ₄₇ in eosinophils (Erle et al., 1994; Walsh et al., 1996a). This integrin is expressed at the same level as ₄₁ and is believed to bind to the Ig, MadCAM-1 (Berlin et al., 1993; Briskin et al., 1993) as well as VCAM-1 and the CS-1 region of fibronectin (Chan et al., 1992; Ruegg et al., 1992; Postigo et al., 1993). Although ₄₇ is constitutively expressed upon eosinophils, it is poorly active since the presence of Mn²⁺ is required to demonstrate adherence (Seminario and Bochner, 1997). This is despite the fact that exposure of eosinophils to PAF has been shown to induce ₄₇-mediated binding to MadCAM-1 but not VCAM-1 in L-12 cells transfected with the appropriate cDNAs (Walsh et al., 1996a). Thus, in the absence of stimuli, the binding of eosinophils to VCAM-1 and fibronectin is mediated predominantly by VLA-4.

Intercellular adhesion molecule 1 (CD54) is a member of the C2-type Ig family of proteins that have been implicated in cell adhesion and complement binding. Human ICAM-1 is composed of a 55-kDa core protein, an extracellular-facing fragment containing five Ig-like domains and up to eight possible sites for N-linked glycosylation, and a 28-amino acid cytoplasmic tail rich in lysine and arginine that is thought to be responsible for binding to the cytoskeleton (Staunton et al., 1990; Carpen et al., 1992; Kirchhausen et al., 1993).

ICAM-1 is not constitutively expressed by circulating blood eosinophils (Hansel et al., 1992; Czech et al., 1993) but has been detected on sputum eosinophils (Hansel et al., 1991a) and in eosinophils recovered from the BAL fluid of patients with eosinophilic pneumonia (Azuma et al., 1996). In contrast to data published by Hansel et al. (1992), Czech et al. (1993) reported that the inflammatory cytokines IL-3, IFN, and TNF induce ICAM-1 expression on normal circulating eosinophils. Despite that discrepancy, there is a consensus that TNF [and TNF (Hansel et al., 1992)] acts synergistically with IL-3, IL-5, GM-CSF, and IFN to up-regulate ICAM-1 expression by a mechanism that involves de novo protein synthesis (Hansel et al., 1992; Czech et al., 1993). Burke-Gaffney and Hellewell (1998) have shown that ICAM-1 mediates adhesion of eosinophils to human bronchial epithelial cells which would aid their accumulation and retention in the airways in diseases such as asthma.

Little is known of the affect of ICAM-1 binding upon eosinophil function, although it has been implicated in GM-CSF- and TNF-induced degranulation (Horie et al., 1997a).

	X. Immunoglobulins

Top Previous Next References

Eosinophils can express Fc receptors for IgA, IgD, IgG, and IgM. Receptors for the Fc portion of IgE also have been detected on eosinophils but controversy surrounds the precise nature of the IgE-FcR interaction (Kita and Gleich, 1997).

The Fc receptor for IgA (CD89) is a transmembrane glycoprotein expressed by several granulocytes including neutrophils and eosinophils (Capron et al., 1988a). Molecular genetics has mapped the human CD89 gene to chromosome 19 (Kremer et al., 1992), which contrasts to the genes that encode other Fc receptors that are localized to chromosome 1. Structurally, FcR is composed of five exons spanning approximately 12 kb (de Wit et al., 1995). The first two exons (denoted S1 and S2) encode a leader sequence, the third and fourth (termed EC1 and EC2) each encode a homologous Ig-like domain and the final exon (TM/C) codes for a short intracellular region, a transmembrane segment and a short cytoplasmic tail (de Wit et al., 1995). Evidence exists for at least seven transcripts of FcR designated FcRa.1 to 6 and FcRb that arise from alternative mRNA splicing (Patry et al., 1996; Pleass et al., 1996; van Dijk et al., 1996). That discovery is entirely consistent with what has been established for FcRI (Porges et al., 1992) and FcRII (Brooks et al., 1989) and suggests that the expression of several closely related proteins by alternative splicing provides a means of diversifying function (see X.D).

The cDNAs derived from the two major Fc transcripts of both human eosinophils and neutrophils have been cloned and sequenced, and the neutrophil variants stably transfected in CHO-K1 cells (Pleass et al., 1996). The largest clone, FcRa.1, represents the previously described full-length receptor (Maliszewski et al., 1990), whereas the splice variant, FcRa.3, is a truncated form lacking the entire second, membrane-proximal Ig domain. The long and short forms do not bind anti-FcR monoclonal antibodies equally or serum IgA, supporting the idea that alternative splicing of FcR gene could provide a means of altering FcR receptor function.

van Dijk et al. (1996) also have reported a novel isoform of the FcR that is expressed in human eosinophils and neutrophils. The cloned receptor, FcRb, differs from previously described splice variants in that it lacks the exon (TM/C) encoding the transmembrane/intracellular region of the wild-type receptor and exon EC2 is extended to encode 23 additional amino acids. Transfection of IIA1.6 murine pro-B lymphocytes with the cDNA for FcRb results in high levels of expression at the plasma membrane, along with the secretion of a significant amount of protein. The expression of FcRb at the cell surface is not affected by phosphatidylinositol-specific PLC, indicating that glycosyl phosphatidylinositol (GPI) linkage of FcRb is unlikely. In IIA1.6 murine pro-B lymphocytes expressing FcRb and FcR, which is necessary for signal transduction by wild-type FcR, neither Ca²⁺ mobilization nor tyrosine phosphorylation is observed upon receptor cross-linking (van Dijk et al., 1996), suggesting that FcRb has a different functional role to FcR.

The molecular mass of FcR on eosinophils (70-100 kDa) is significantly higher when compared to its counterpart on human neutrophils (55-75 kDa). However, removal of N-linked carbohydrates from both cell types yields a protein of 32 kDa, indicating differential degrees of glycosylation between neutrophils and eosinophils (Monteiro et al., 1993). The expression of FcR is up-regulated approximately 3-fold on human eosinophils exposed to the Ca²⁺ ionophore A23187 (Monteiro et al., 1993). Similarly, eosinophils harvested from allergic individuals express higher levels of FcR when compared with control subjects (Monteiro et al., 1993).

Functionally, anti-IgA induces eosinophil migration in atopic and healthy volunteers (Rihoux et al., 1990). A number of studies also have demonstrated the expression of functional IgA receptors on human eosinophils (Capron et al., 1988a; Abu Ghazaleh et al., 1989; Kita et al., 1991b) with particular reference to degranulation. Secretory IgA and IgA-coated Sepharose beads are particularly effective at promoting EDN release by a mechanism that is enhanced by IL-3 and GM-CSF (Abu Ghazaleh et al., 1989; Kita et al., 1991b).

It has been reported (Wardlaw et al., 1995) that human eosinophils express Fc receptors for IgD but this assertion has not received further documentation. In contrast, normal blood eosinophils lack FcµR (CD7) (Ottesen et al., 1977; Walsh and Kay, 1986) although binding of IgM can apparently occur when cells are cultured in vitro (De Simone et al., 1982a).

It has been known for some time that certain allergic diseases and parasitic infections are associated with peripheral blood and tissue eosinophilia along with an increase in total and antigen-specific IgE. Indeed, there is a close correlation between serum IgE and the prevalence and severity of allergic diseases such as asthma (Sears et al., 1991). Similarly, the acquisition of immunity against Schistosoma hematobium is positively correlated with the appearance of anti-schistosome IgE antibodies (Hagan et al., 1991). IgE is known to promote mast cell and eosinophil degranulation (Khalife et al., 1986; Galli et al., 1991; Tomassini et al., 1991), enhance antigen presentation to, and internalization by, T lymphocytes when bound to antigen-presenting cells (APCs) (Mudde et al., 1995; Maurer et al., 1996) and mediate killing of invading parasites by acting as a ligand for antibody-dependent, cell-mediated cytotoxicity (Capron et al., 1982; Truong et al., 1993; Capron and Capron, 1994; Gounni et al., 1994a). Thus, these and other data have led to the general view that IgE is implicated in the direct and indirect activation of eosinophils in allergic diseases and following parasite infestation.

Arbesman and coinvestigators were the first to demonstrate that complexed IgE binds to human eosinophils (Ishikawa et al., 1974; Fujita et al., 1975), an observation that was extended several years later by the identification of cell surface receptors for IgE (Capron et al., 1981). Subsequently, it was established that IgE bound to a receptor on eosinophils with low affinity (K_d ~ 100 nM) that was similar, but not identical, to CD23 (FcRII) expressed by B lymphocytes (Capron et al., 1986, 1991; Jouault et al., 1988; Yokota et al., 1988; Capron and Joseph, 1991). However, in 1998, definitive evidence was provided that human eosinophils express FcRII that is identical with CD23 expressed by B lymphocytes (Abdelilah et al., 1998). Structurally, FcRII is a 45-kDa type II glycoprotein that can exist in at least two isoforms, FcRIIa and FcRIIb, that differ only in their amino-terminal cytoplasmic tail and arise through differential mRNA splicing (Yokota et al., 1988). Eosinophils express both forms of FcRII (Abdelilah et al., 1998). Since those original observations, two other receptors for IgE have been identified on eosinophils and to some extent characterized. In mice, one of these, FcRI, is a tetrameric protein composed of an  chain, which binds IgE, a  chain, and two disulfide-linked  chains (Ravetch and Kinet, 1991), and is recognized by IgE with high affinity (K_d ~ 0.1 nM) (Gounni et al., 1994a,b). Interestingly, the human homolog of FcRI lacks the  subunit. The other receptor is a galactose-specific, thiol-dependent S-type lectin called Mac-2 that has a high degree of sequence homology to rat BP and carbohydrate-binding protein 35 and binds to IgE with relatively low affinity (Truong et al., 1993).

Evidence is available that eosinophils can express each variant of Fc, although the extent to which this occurs depends on whether the cells are purified from normal subjects or from individuals with eosinophilia associated with allergic inflammation or parasitosis. Further distinctions probably can be made based on the type and severity of disease. For example, the expression of FcRII is seemingly restricted to a hypodense population of eosinophils harvested from subjects with prominent eosinophilia and certain allergic disorders (Capron et al., 1986, 1989; Rumi et al., 1998), whereas little, if any, expression is detected on eosinophils from "normal" individuals (Hartnell et al., 1989; Rumi et al., 1998). Comparable data also have been reported for FcRI (Terada et al., 1995; Rajakulasingam et al., 1997, 1998; Sihra et al., 1997) and Mac-2/BP (Truong et al., 1993). In the later case, Northern blot analysis using eosinophil RNA from several eosinophilic donors probed with human Mac-2 and human BP cDNAs routinely identified a single 1.2-kb product (Truong et al., 1993). In 50% of those eosinophil preparations, complementary flow cytometry experiments identified cell surface Mac-2 expression. This was confirmed by Western immunoblotting which resulted in the labeling of a 29-kDa band, consistent with Mac-2 protein, and two smaller anonymous peptides of 20 and 15 kDa (Truong et al., 1993).

It is clear from the aforementioned discussion that there is a wide body of evidence from studies in humans that IgE can activate eosinophils and promote antibody-dependent, cell-mediated cytotoxicity through FcRI, FcRII, and Mac-2/BP. However, some controversy still surrounds the functional role of Fc on eosinophils (Kita and Gleich, 1997). In particular, allergen-antibody complexes also can activate eosinophils via FcR (Kaneko et al., 1995a). Similarly, FcR are involved in IgG-dependent cytotoxicity toward parasites (Butterworth et al., 1977). Moreover, eosinophils purified from the BAL fluid, liver granulomas, and bone marrow cultures of parasite-infected mice do not express cell surface receptors for Fc and fail to bind IgE under conditions where eosinophils can be activated following ligation of FcR by IgG (Jones et al., 1994; de Andres et al., 1997b). A major implication of those findings is that the mouse might not be a suitable species to study human immunological disease and, clearly, this requires careful investigation. Another enigmatic set of observations that currently defies explanation is the finding that antibodies raised against all three Fc receptors abolish IgE binding and antibody-dependent, cell-mediated cytotoxicity (see Kita and Gleich, 1997 and references therein).

Three functional receptors for IgG have been identified and characterized on human leukocytes. On resting human eosinophils, only one of these, FcRII (CDw32), is constitutively expressed to any extent (Hartnell et al., 1990) although murine eosinophils also express FcRIII (CD16) in reasonable numbers (de Andres et al., 1997b). IFN up-regulates FcRII expression (Valerius et al., 1990; Hartnell et al., 1992b) and is associated with enhanced IgG-induced antibody-dependent cellular cytotoxicity (Valerius et al., 1990). Those data are consistent with an earlier publication that described an increase in Fc receptor density and cytotoxicity in response to IFN (De Simone et al., 1986). Structurally, FcRII is a 40-kDa protein, for which IgG has low affinity, and is widely distributed among leukocytes.

Hartnell et al. (1992b) have reported that the expression of FcRII is up-regulated by IL-3. However, although a subsequent study (Koenderman et al., 1993) found that the addition of IL-3, IL-5, and GM-CSF to freshly prepared eosinophils produced a transient 3-fold increase in their ability to form rosettes with IgG-sensitized erythrocytes, no change in FcRII receptor expression was noted. Indeed, it was concluded that changes in FcRII signaling might reflect alterations in CR3 receptor function since the increase in rosetting activity was accompanied by a commensurate augmentation in the binding of iC3b to CR3. Further support for that contention was that an antibody against CD11b prevented the effects of IL-3, IL-5, and GM-CSF (Koenderman et al., 1993). The reason for the discrepancy between the two studies is unclear. In murine eosinophils, ligation of FcRII results in phosphatidylinositol hydrolysis which has been linked to the activation of 5-lipoxygenase (de Andres et al., 1991b).

The other Fc receptor for which IgG has low affinity is the 50- to 70-kDa FcRIII (CD16). Although not expressed constitutively by human eosinophils, high levels are found intracellularly that can be mobilized to the cell surface by mediators such as IFN, PAF, fMLP, and C5a in a cycloheximide- and dexamethasone-sensitive manner, indicating a requirement for new protein synthesis (Hartnell et al., 1992b; Zhu et al., 1998). The up-regulation of FcRIII is transient and protein is rapidly released into the medium and then is reabsorbed. The role of FcRIII on eosinophils in unknown but it is likely that the secretion of the soluble receptor might neutralize bioavailable IgG. Treatment of human eosinophils with phosphatidylinositol-specific PLC markedly reduces cell surface CD16 expression, indicating that it is a GPI-linked receptor (Zhu et al., 1998). The finding that anti-CD16 effects membrane depolarization and LTC₄ release in cytokine-treated cells (Hartnell et al., 1992b) indicates that this effect is functionally relevant. Those data are concordant with the enhanced expression of IgG receptors in general by eosinophils purified from subjects with eosinophilia (that have presumably become exposed to a host of mediators in vivo) when compared with normal individuals (Kishimoto, 1988).

The cDNA of each FcR has been cloned (Stuart et al., 1987; Simmons and Seed, 1988; Allen and Seed, 1989) which led to the discovery of two FcRIII isoforms [FcRIII-1 (CD16-1), FcRIII-2 (CD16-2)] that are encoded by distinct genes (Ravetch and Perussia, 1989; Scallon et al., 1989). FcRIII-1 is expressed predominantly by resting neutrophils and is anchored to the cell surface by a GPI linkage; however, after activation of the cell, this receptor is shed and is detected in the plasma (Huizinga et al., 1988, 1989, 1990). In contrast, FcRIII-2 is a transmembrane-spanning receptor expressed by natural killer cells (Hibbs et al., 1989; Kurosaki and Ravetch, 1989; Lanier et al., 1991). The FcRIII variant(s) expressed on eosinophils and the function(s) it specifically subserves is currently unclear, although treatment of IFN-exposed eosinophils with phosphatidylinostitol-specific PLC reduces FcRIII expression, suggesting that eosinophils express a functionally active GPI-linked form (FcRIII-1) of the receptor (Hartnell et al., 1992b).

FcRI (CD64) is a 72-kDa protein for which IgG has high affinity and is expressed almost exclusively by monocytes. However, receptors for FcRI can be induced in human eosinophils treated with IFN (Hartnell et al., 1992b).

IgG, immobilized to Sephadex beads, evokes a host of functions in eosinophils. Of significance was the demonstration in 1997 that ligation of Fc receptors on murine eosinophils provided a napoptotic signal. de Andres et al. (1997a) reported that culture of murine eosinophil precursors with a rat monoclonal antibody (2.4G2) that reacts with CD32 and CD16 promoted several hallmarks of apoptosis: chromatin condensation, annexin V binding, and induction of CD95. Since murine eosinophils express CD16 and CD32, additional experiments were performed with precursors obtained from mice in which the CD16 gene was disrupted. The results of those experiments established that apoptosis was absolutely dependent on CD32 and the activation of CD95 (de Andres et al., 1997a). Collectively, those data highlight a novel mechanism of inducing apoptosis which, if reproduced in human eosinophils, could be relevant to cell-mediated tissue injury and antibody-dependent cellular cytotoxicity (see XII.H).

Other functional effects for which IgG has been accredited include degranulation (Kita et al., 1991b,d; Tomassini et al., 1991; Kaneko et al., 1995a,b), activation of the NADPH oxidase (de Andres et al., 1997b), and the generation of LTC₄ (Shaw et al., 1985; Cromwell et al., 1988; Moqbel et al., 1990a) and PAF (Cromwell et al., 1990).

	XI. Miscellaneous

Top Previous Next References

The type I (CDw121a) and type II (CDw121b) IL-1 receptor have been identified in mice and humans. Structurally, these receptors are single transmembrane glycoproteins with approximate molecular masses of 80 kDa and 60 kDa, respectively. Both receptors bind the predominately membrane associated IL-1, IL-1 that is secreted, and IL-1RA. Studies with human eosinophils suggest that IL-1 induces the secretion of arylsulfatase and -glucuronidase whereas a combination of IL-1 and IL-1 inhibits the activation of the NADPH oxidase in response to the phorbol diester PMA (Pincus et al., 1986; Whitcomb et al., 1989). Prolonged exposure (30-180 min) of guinea pig eosinophils to IL-1 suppresses A23187-induced [³H]AA and phosphatidylcholine release by down-regulating the activity of PLA₂ (Debbaghi et al., 1992).

Sanz et al. (1995) demonstrated that intradermal administration of IL-1 to rats stimulates the accumulation of radiolabeled eosinophils to the sites of injection (Sanz et al., 1995). Moreover, persuasive evidence for a role of IL-1 in allergen-induced eosinophil migration derives from the ability of IL-1RA to prevent pulmonary eosinophilia when given by aerosol to sensitized guinea pigs immediately before challenge (Watson et al., 1993). Similarly, IL-1RA blunts the LPR and the number of hypodense eosinophils that appear in the BAL fluid of allergen-challenged sensitized guinea pigs (Okada et al., 1995). These findings have prompted clinical trials with IL-1RA in asthma, the results of which are eagerly awaited. It is noteworthy that the in vivo chemotactic activity of IL-1 is likely to be secondary to the activation of the endothelium (Lamas et al., 1988; Bochner et al., 1991a; Kyan Aung et al., 1991b; Ebisawa et al., 1992) and/or to the release of other chemoattractants such as PAF and IL-8 (Sanz et al., 1995).

The IL-2 receptor is composed of three polypeptide chains; an  chain (p55, CD25), a  chain (p 75, CD122), and a  chain (_c) that is common to several cytokine receptors. IL-2 binds to the  and  chains with low affinity but does not interact with _c. However, a high-affinity IL-2:receptor complex is achieved when the ligand ligates the _c heterotrimer; interactions of intermediate affinity also can occur with _c and _c heterodimeric forms of the receptor. In T lymphocytes, the  chain of the IL-2 receptor is essential for activation and features critical sequences within its intracellular domain necessary for effective signaling (Hatakeyama et al., 1989). The same appears to be true for _c (Zurawski and Zurawski, 1992) whereas the  chain alone cannot transduce the IL-2 signal.