Interleukin-5, Interleukin-3, and Granulocyte-Macrophage Colony-stimulating Factor Cross-compete for Binding to Cell Surface Receptors on Human Eosinophils*

Human interleukin (1L)-5 receptors were character- ized by means of binding studies using bioactive “‘I-labeled IL-5. Of purified primary myeloid cells, eosin- ophils and basophils but not neutrophils or monocytes expressed surface receptors for IL-5. Binding studies showed that eosinophils expressed a single class of high affinity receptors (K. = 1.2 x lo1’ M - ~ ) with the num- ber of receptors being small (< 1000 receptors/cell) and varying between individuals. Among several cell lines examined only HL-60 cells showed detectable IL-5 receptors which were small in numbers (200 receptors/ cell) and also bound lz61-IL-5 with high affinity. The binding of IL-5 was rapid at 37 “C while requiring several hours to reach equilibrium at 4 “C. Specificity studies revealed that the two other human eosinophil- opoietic cytokines IL-3 and granulocyte-macrophage colony-stimulating factor (GM-CSF) inhibited the binding of ‘2‘I-IL-5 to eosinophils. No competition was observed by other eosinophil activating or nonactivat- ing cytokines. The inhibition of “‘I-IL-5

Human interleukin (1L)-5 receptors were characterized by means of binding studies using bioactive "'Ilabeled IL-5. Of purified primary myeloid cells, eosinophils and basophils but not neutrophils or monocytes expressed surface receptors for IL-5. Binding studies showed that eosinophils expressed a single class of high affinity receptors ( K . = 1.2 x lo1' M -~) with the number of receptors being small (< 1000 receptors/cell) and varying between individuals. Among several cell lines examined only HL-60 cells showed detectable IL-5 receptors which were small in numbers (200 receptors/ cell) and also bound lz61-IL-5 with high affinity. The binding of IL-5 was rapid at 37 "C while requiring several hours to reach equilibrium at 4 "C. Specificity studies revealed that the two other human eosinophilopoietic cytokines IL-3 and granulocyte-macrophage colony-stimulating factor (GM-CSF) inhibited the binding of '2'I-IL-5 to eosinophils. The competition between IL-5, IL-3, and GM-CSF on the surface of mature eosinophils may represent a unifying mechanism that may help explain the common biological effects of these three eosinophilopoietic cytokines on eosinophil function. This unique pattern of competition may also be beneficial to the host by preventing excessive eosinophil stimulation.
Human interleukin-5 (ILL-5) is a T cell-derived cytokine * This work was supported by grants from the National Health and Medical Research Council (Australia) and by Grant 1 R 0 1 CA45 822-01 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Although stimulating preferentially the eosinophil lineage, IL-5 has been so far shown to stimulate the same pattern of functional responses as IL-3 and GM-CSF.
For example, mouse (m) and human (h) IL-5, like hIL-3 and hGM-CSF, enhance h-eosinophil antibody-dependent killing of tumor cells (2, 7-9) and helminths (10-E), stimulate phagocytosis of yeast (2,(7)(8)(9), increase the expression of the complement receptor type 3 (7, B), prolong eosinophil survival (10)(11)(12)(13), increase the biosynthesis of proteoglycans (14), and facilitate the conversion of normodense eosinophils to the hypodense phenotype (10)(11)(12). Importantly, not only the spectrum of functions but also the magnitude of the stimulation of each eosinophil function by IL-5, IL-3, and GM-CSF is very similar. These results raised the possibility that all three eosinophil growth factors have a common step in their signaling mechanism. In addition, the cross-reactivity of mIL-5 with human eosinophils suggested that this cytokine recognised the human IL-5 receptor.
We have investigated the distribution of the human IL-5 receptor on several primary cell types and cell lines, the kinetics of IL-5 binding, and the molecular species involved in binding to IL-5. In particular we examined the relationship between human eosinophil IL-5 receptors and those for IL-3 and GM-CSF. Significantly the eosinophil IL-5 receptor interacted with IL-3 and GM-CSF but not with other cytokines. It is proposed that this interaction is an important determinant in unifying signaling on human eosinophils and in preventing their excessive stimulation in vivo.

EXPERIMENTAL PROCEDURES
Purification of Human Eosinophils and Other Myeloid Cells-Eosinophils were obtained from the peripheral blood of eosinophilic (9-34%) individuals or from pleural effusions (6045% eosinophils) and purified by centrifugation on a hypertonic gradient of metrizamide as described (15). The eosinophil preparations were always >93% pure. Highly enriched human basophils were obtained from a patient with end-stage Philadelphia chromosome-positive chronic granulomatous leukemia undergoing basophilic differentiation. After dextran sedimentation, basophils were purified by centrifugation on a hypertonic gradient of metrizamide. The 20% metrizamide fractions containing >90% basophils, as judged by Alcian Blue staining, were used for binding studies. Neutrophils were purified from normal volunteers by centrifugation on a hypertonic gradient of metrizamide as described (15). Monocytes were purified by countercurrent elutriation. Mononuclear cells were washed twice in RPMI 1640 a t 150 X g and resuspended in medium containing 0.1% heat-inactivated human AB serum. The cells were separated in a Beckman J-GM/E elutriator 24741 (I'alo Alto, CA) using the Sanderson chamher, with a rotor speed of 2050 rpm and a flow rate of 11.8 ml/min. Cells remaining in the chamher after 30 min were collected, washed twice in KI'MI, and used immediately. This method resulted in a monocvte purity hy Wright's-Giemsa staining of 91% with >W'; of elutriated cells heing phagocytic for opsonized zymosan. Hrcomhinnnt Human 11,-5 nnd O i h r r Cylokinrs-Purified rhlL-5, rhlL4, and rhGM-CSF produced in yeast were used for radiolaheling. These cytokines were constructed as fusion proteins containing an additional octapeptide at the NH, terminus, expressed in yeast, and purified as previously described (16. 17). A tyrosine present in the octapeptide facilitates radiolaheling to high specific activity (17,18). As competitors the following purified rh-cvtokines were used: rh1L 5, rhlL-2, rhll,-4, rhI1,-3, rhGM-CSF, rhll,-l(y, and rhIL-l{j (produced in Iischrrichin coli), rhlL-3 (produced in ! ; . coli, gift from Dr. S. Clark, Cenetics Institute, Camhridge, MA), rhGM-CSF (produced in COS cells, gift from Dr. S. Clark, Genetics Institute). rh-tumor necrosis factor CY (produced in E;. coli, gift from Genentech. S a n Francisco, CA), rhIL-6 (produced in E;. coli, gift from Dr. S. Clark, Genetics Institute), and leukemia inhibitory factor (produced in E . coli, gift from I k . N. Nicola, WEHI, Melhourne, Victoria). Human and mouse IL-5 in the form of COS cell supernatants were also used.
Hindinx Assays-For 4 "C hinding assays 2 X 10" cells were incubated in 0.15 ml of hinding medium (RI'MI 1640, pH 7.4, supplemented with 20 mM Hepes, 0.5"; hovine serum alhumin, and 0.1% NaN:,) containing radioligand and medium or unlabeled cvtokines in siliconized glass tuhes on a rotating tahle for If h. Cell suspensions were then resuspended, overlaved onto 0.2 ml of fetal calf serum and centrifuged for :10 s at maximum speed in a Beckman Microfuge 12. T h e visil)le cell pellet was removed hv cutting, and the radioactivity in the pellet was determined in a I'ackard Auto-Gamma 5650. Nonspecific hinding was determined hy incuhation in the presence of a 100-fold or greater molar excess of unlabeled cvtokine. Sperific hinding was calculated hy suhtracting nonspecific hinding from total binding. Free radioligand was the difference hetween total and specifically hound radioactivity (23).
To determine the amount of receptor-ligand internalization under these conditions, acid-resistant radioactivity was measured at the conclusion of the assay as descril)ed (24, 25). Briefly, cell suspensions were centrifuged through 0.5 ml of ice-cold fetal calf serum; the cell pellet was resuspended in hinding medium and divided into two identical samples. Kach aliqr~ot was diluted in two volumes of 0.5 M acetic acid, 0.5 M NaCl, pH 2.0. or hinding medium, pH 7.4, for 5 min o n ice and then centrifuged through fetal calf serum. Dissociahle radioactivity was defined as the dil'ference hetween total and acidresistant specific radioactivity hound. Binding data were analyzed I y the ERDA and LIGAND computer programs (IXvvier-HIOSOF'T, Camhridge, United Kingdom). For some experiments, hinding assays were performed at : E "(: and analvzed as previously tlescril)ed ( 2 2 ) .
Association experiments and curve fitting were carried out as descrihed elsewhere (22). glycerol) with o r without 5 % 2-mercaptoethanol and analyzed on gels hy the procedurr of Laemmli (26). After electrophoresis, gels were stained with Coomassie exposed to Kodak X-Omat AH film at -70 -C.
A low molecular weight hand was also seen prohahlv representing free iodine at the dye front. Functional testing of '.'.'I-IL-5 laheled hv the iodine monochloride method in an eosinophil polarization assay ( 7 ) showed little or no loss of hiological activity (Fig. 1B). The other two laheling methods produced a similar retention of hiological activitv (data n o t shown).
Charactprization of " ' I -I L -5 Binding-The kinetics of "'I-IL-5 hinding at different temperatures w a s estahlished using a clone of t h e myeloid cell line HI,-60 which had heen selected for higher expression of IL-5 receptors. Fig. 2A shows hinding of '"'I-IL-5 to HL-60 cells at 4, 25, and 37 "C. At 25 and 37 "(: the hinding was rapid and saturahle, reaching apparent equilibrium hy 1-2 h. Binding at 4 "C was much slower, requiring >3 h to reach equilihrium. Fig. 2B illustrates the association kinetics at 37 "C of ""I-IL-5 hinding to HI,-60 cells, over a broader concent,ration range. Rased on these data, 60 min w a s selected as the optimal time period for hinding experiments at 37 "C to hoth allow achievement of stable maximum hinding while minimizing the length of incuhation. Sodium azide is included in all incuhations at 37 "C to minimize internalization of ligand, however, at least with the mveloid cell line HL-60, some internalization does occur over this time period. In this case, the conditions for true hinding equilihrium have not heen met and the hinding parameters ohtained represent values for the case in which 11,-5 hinding has reached an experimentally determined stahle maximum. IL-5 to eosinophils at 4 "C (panel A ) and binding to HL-60 cells at 37 "C in uitro under the conditions described above (panel B). In both cases Scatchard analysis of the data yielded a straight line, indicating a single class of high affinity binding sites for IL-5 on both cell types. The data in Fig. 3A show that eosinophils bound to '2s1-IL-5 with a calculated KO of 1.2 X 10" M" and expressed 218 specific binding sites/cell. Similar results were obtained in two other experiments (data not shown). For HL-60 cells, the apparent calculated K , was 6.2 Nsinophil Growth Factors 24743 k 1.7 X lo9 M" with 180 -t 70 specific binding sites/cell (average from 14 binding experiments) using IL-5 radiolabeled with Bolton-Hunter reagent, and 6.6 f 3.0 X lo9 M" with 200 f 90 specific binding sites/cell (average from four binding experiments) using IL-5 radiolabeled with the Enzymobead reagent. These results verify that IL-5 radiolabeled on either tyrosine or lysine residues exhibits essentially identical binding characteristics.
Cellular Distribution of the IL-5 Receptor-Human eosinophils and basophils, but not neutrophils, monocytes, or lymphocytes purified from peripheral blood, bound I2'1-IL-5. (Table I). Tissue eosinophils obtained from two patients with pleural effusions also expressed IL-5 receptors although these were much fewer in numbers than blood eosinophils (Table  I). Basophils from a patient with chronic myeloid leukemia undergoing basophilic differentiation expressed similar numbers of receptors to eosinophils. Examination of several cell lines under high affinity conditions demonstrated that only the myeloid cell line HL-60 expressed IL-5 receptors while in all other myeloid cell lines tested IL-5 receptors were undetected (Table I). Non-myeloid cell lines including cells of T cell, B cell, and pre-B cell origin also lacked high affinity IL-5 receptors.
Specificity of IL-5 Binding-In order to ascertain the specificity of 1251-IL-5 binding, different cytokines at 100-fold or greater molar excess concentration were incubated with either eosinophils or HL-60 cells and a subsaturating concentration of '2s1-IL-5. The results showed that not only IL-5 but also GM-CSF and IL-3 inhibited the binding of '*'I-IL-5 to eosinophils (Table 11). Similarly, GM-CSF was also able to inhibit the binding of lZ5I-IL-5 to HL-60 cells, while IL-3, which does not bind to HL-60 cells, was ineffective. In contrast, a number of other cytokines did not inhibit binding to either cell type. Mouse IL-5, like human IL-5, fully competed for l2'I-IL-5 binding on eosinophils.

TABLE I1
Specificity of the human ZL-5 receptor Eosinophils were incubated for 16 h at 4 "C with 50-80 PM lz5I-IL-5 in the presence of different purified competitors a t 10" M except for COS cell supernatants which were tested at 1:13 dilution. Total specific binding ranged from 1424 f 29 to 4538 f 102 cpm. The values obtained are the means of two to three different experiments performed in triplicate +. S.D. The eosinophil purity in these experiments ranged between 93-99%. HL-60 cells were incubated for 1 h at 37 "C with 5 X 10"' M lZ5I-IL-5 and competitors at a 200-fold or greater molar excess. To ascertain that the competition for lz5I-IL-5 binding to eosinophils by GM-CSF and IL-3 took place at the cell surface, acid elution experiments were performed where after incubation for 16 h at 4 "C eosinophils were resuspended in medium pH = 7.0 or pH = 2.0. In an experiment where the competition by IL-3 was 41.4% and by GM-CSF was 44.6%, 96.5% of cell-associated radioactivity was acid-dissociable. In the presence of IL-3 and GM-CSF, the acid-dissociable counts were 79.0 and 81% of cell-associated radioactivity, respectively. In two other experiments the percent acid-dissociable specific counts in the absence of competitors or in the presence of GM-CSF and IL-3 were 91.4, 85.8, and 85.5, respectively, in experiment one, and 87.7, 100.0, and 94.5, respectively, in experiment two. The fact that the non-acid-dissociable counts were always much less than the percent competition with GM-CSF and IL-3 indicates that the competition observed occurred on the surface of eosinophils.
Given that the binding of '251-IL-5 was inhibited by GM-CSF and IL-3 it was important to determine whether this competition was unidirectional or whether all three eosinophil hemopoietic growth factors cross-competed for binding. The results using eosinophils from 19 different individuals showed that the binding of '251-IL-5 was consistently inhibited by GM-CSF and IL-3 with GM-CSF being the stronger competitor (Fig. 4). However, when IL-5 was used as a competitor it inhibited the binding of GM-CSF and IL-3 to eosinophils from only a few donors, and in general this competition was of a smaller magnitude to that seen with GM-CSF or IL-3 (Fig. 4). Similar results were found using HL-60 cells, where GM-CSF consistently inhibited IL-5 binding; however, IL-5 was much less effective at competing for '251-GM-CSF. The results from three different experiments showed that the binding of lz5I-IL-5 was inhibited 45.7 9.3% by GM-CSF and that the binding of '251-GM-CSF was inhibited 7.8 f 1.6% The competition by GM-CSF and IL-3 of '251-IL-5 binding to eosinophils was further studied by carrying out quantitative inhibition binding experiments. The results showed that the by IL-5. inhibition by GM-CSF and IL-3 was not complete up to a concentration of competitor of M (Fig. 5). Reciprocally, IL-5 only partially inhibited the binding of '251-GM-CSF and lz5I-IL-3 to these eosinophils (Fig. 5). Similar results were obtained with HL-60 cells where GM-CSF competition of lZ5I-IL-5 binding was only partial up to a concentration of competitor of low7 M (data not shown).
These complexes were not seen in the absence of cross-linker or when excess unlabeled IL-5 was added to the reaction mixture. Importantly, unlabeled GM-CSF almost completely inhibited the formation of both complexes while unlabeled IL-3 did not compete. We have been unable, so far, to obtain cross-linking data from peripheral blood eosinophils under several different experimental conditions. This may be related to the unique composition of these cells in terms of granule contents and variety of proteases.

DISCUSSION
We show here that human eosinophils express high affinity receptors for IL-5 and that the binding of IL-5 is inhibited by the two other eosinophil-active hemopoietic growth factors, IL-3 and GM-CSF. In contrast, neither tumor necrosis factor a, a cytokine that stimulates the function (27) but not the production of human eosinophils, nor IL-4, IL-6, or leukemia inhibitory factor inhibited IL-5 binding. The competition between IL-5, IL-3, and GM-CSF on the surface of eosinophils may be important for understanding the common activation properties of these cytokines. Several studies on eosinophil function have shown that IL-5 as well as IL-3 and GM-CSF enhance eosinophil superoxide production (2,8,9), antibodydependent killing of tumor cells (2,(7)(8)(9) and helminths (10)(11)(12), degranulation (28), survival in vitro (10-13), conversion from a normodense to a hypodense phenotype (10-121, phagocytosis (2,(7)(8)(9), and the biosynthesis of proteoglycans (14). Interactions with a common receptor complex may constitute one unifying mechanism. Other possibilities such as the utilization of a common signal-transduction pathway downstream from the receptor remain to be explored. The second implication of the present findings is that the competition between IL-5, IL-3, and GM-CSF may serve to limit eosinophil stimulation, thus preventing the excessive release of oxygen products, leukotrienes, and granule contents bg these cytokines with potentially harmful consequences to the host (29). Of human primary myeloid cells, only eosinophils and basophils expressed IL-5 receptors. T h e IL-5 receptors on human blood eosinophils were low in numbers (215-690/cell) a n d of high affinity ( K , = 1.2 X 10"' M"). No low affinity receptors were detected on eosinophils using "'I-IL-5 up to a concentration of 1.4 nM. In two cases where eosinophils from pleural effusions were studied the numbers of IL-5 receptors were very small (Table I). I t is not clear whether this represents a true difference between blood and tissue eosinophils or perhaps more likely that these eosinophils have been exposed to IL-5 in uiuo, leading to occupation or downregulation of IL-5 receptors. Although GM-CSF or IL-3 might also occupy IL-5 receptors, the selective accumulation of eosinophils in these two cases makes IL-5 the more likely ligand.
IL-5 also bound to basophils purified from the blood of one patient with chronic myeloid leukemia undergoing basophilic differentiation but not to neutrophils or monocytes. This is consistent with the ability of IL-5 to enhance histamine release from human basophils (30,31) and with the inability of IL-5 to stimulate or prime neutrophils for the release of superoxide anion and to stimulate monocyte adherence (data '"CGM-CSF J not shown). Of t.he human cell lines studied, IIA-.5 receptors were only detected on the promyelocvtic leukemia HL-GO. A subline of HL-60 cells has also recently heen shown to express IL-5 receptors after incubation with Na-butyrate (32). Interestingly in view of the capacity of mouse 11, -5 to stimulate and to bind to mouse R cells with high affinity (B3), high affinity binding sites for 11,-5 on human R and pre-R cell lines were not detected.
Mouse IL-5 has been demonstrated to be active on human eosinophils and in fact this cytokine has been widely used on human eosinophils on a variety of functional studies (7,12,13,27). It was, therefore, important to establish whether it bound to the same receptor as human IL-5. T h e full competition of mouse IL-5 for ""I-human IL-5 binding to eosinophils strongly suggests that this is the case. These results are also relevant for structure-function studies of the 11,-5 molecule as they suggest that the binding domains in human and mouse IL-5 are present in the conserved regions of these molecules.
The inhibition of IL-5 binding to eosinophils bv IL-3 and GM-CSF was a very consistent finding ohserved in 12/12 individuals examined. This is in contrast with a recent report mentioning lack of competition; however, no data was shown nor was the bioactivity of the "~'1-IL-5 used presented (34). In contrast to IL-3 and GM-CSF, both of which consistently inhibited the binding of ""I-11,-5 to eosinophils (Table 11, The competition for ""I-IL-5 binding to eosinophils hy 11,-3 and GM-CSF was partial even when tested at high concentrations of competitors. This may be due to the Association of some but not all IL-5 receptors with I L -3 and GM-CSF receptors or to the existence of eosinophil subpopulations expressing either IL-5-specific receptors, or IL-S-, ll,-:l-, and GM-CSF-associated receptors. Morphological and functional differences have been noted in eosinophils generated in ritro (35) and in uiuo (5, 36-38). The existence of eosinophil suh-  populations has also been suggested based on the additive effects of IL-5 and IL-3 and of IL-5 and GM-CSF in stimulating eosinophil formation (39). It would be of great interest to establish whether there are subpopulations of eosinophils capable of binding only to IL-5 or to all three eosinophil hemopoietic growth factors. Autoradiographic studies may ultimately answer this question.
The physical nature of the competition between IL-5, IL-3 and GM-CSF is not yet known, but it may reflect the relatedness of these three eosinophil growth factors and their receptors. The relatedness between IL-5, IL-3, and GM-CSF extends from the structure and close localization of their genes on the long arm of chromosome 5 (40,41) to the structure of the mature polypeptides. The IL-5, IL-3, and GM-CSF molecules have conserved features, in particular an area of hydrophilic amino acids in the COOH terminus (42,43), and their tertiary structure is predicted to be highly conserved (44). While the homology in the COOH terminus could suggest a common binding domain this appears unlikely in view of the pattern of binding and cross-reactivity in primary human myeloid cell types (Table 111). It is clear that while in eosinophils the cross-reaction is most evident (18), in monocytes IL-3 and GM-CSF cross-compete (17, 25) but IL-5 is unable to compete, and on neutrophils neither IL-5 nor IL-3 compete for GM-CSF binding (18). It is apparent therefore that for inhibition to occur the homologous receptor needs to be expressed (Table 111), and the unique pattern of competition between IL-5, IL-3, and GM-CSF is observed on eosinophils because these cells express all three receptors. This is supported by the data with HL-60 cells which express GM-CSF but not IL-3 receptors, where GM-CSF but not IL-3 inhibits '2'1-IL-5 binding (Table 11) and the formation of the M, 150,000 and 80,000 complexes (Fig. 6).
The inhibition of '251-IL-5 binding by IL-3 and GM-CSF may be explained at the receptor level by postulating either steric hindrance or a common receptor complex. In both cases a specific association of IL-5, IL-3, and GM-CSF receptors needs to be invoked. In the first case, the binding of one ligand may physically prevent the binding of a second ligand, while in the second case the receptors for IL-5, IL-3, and GM-CSF may share a polypeptide chain. In this situation IL-5, IL-3, and GM-CSF receptors may be composed of unique binding proteins which are specific for the respective ligands and which become associated with another polypeptide chain that is limiting. This postulate is supported by the recent cloning of a GM-CSF receptor from placenta which is specific for GM-CSF (45) and by the cross-linking experiments with IL-5 shown here demonstrating competition for more than one protein. In addition the recent cloning of a second GM-CSF receptor subunit that does not bind GM-CSF but associates with the binding chain of the GM-CSF receptor to provide high affinity binding (46) raises the possibility that this chain is also involved in providing high affinity IL-5 binding in human eosinophils. While the exact identity and relationship of these polypeptide chains awaits their molecular cloning the presence of a common receptor complex on human eosinophils may represent one mechanism by which IL-5, IL-3, and GM-CSF exert their common biological activity on these cells.