Characterization of a Human Eosinophil Proteoglycan, and Augmentation of Its Biosynthesis and Size by Interleukin 3, Interleukin 5 , and Granulocyte/Macrophage Colony Stimulating Factor*

Human eosinophils were cultured for up to 7 days in enriched medium in the absence or presence of recombinant human interleukin (IL) 3, mouse IL 5, or recom- binant human granulocyte/macrophage colony stimulating factor (GM-CSF) and then were radiolabeled with [36S]sulfate to characterize their cell-associated proteoglycans. Freshly isolated eosinophils that were not exposed to any of these cytokines synthesized M, -80,000 Pronase-resistant ”S-labeled proteoglycans which contained M, -8,000 glycosaminoglycans. RNA blot analysis of total eosinophil RNA, probed with a cDNA that encodes a proteoglycan peptide core of the promyelocytic leukemia HL-60 cell, revealed that the mRNA which encodes the analogous molecule in eosinophils was -1.3 kilobases, like that in HL-60 cells. When eosinophils were cultured for 1 day or longer in the presence of 10 PM IL 3, 1 p~ IL 5, or 10 PM GM- CSF, the rates of [S6S]sulfate incorporation were increased -2-fold, and the cells synthesized

6-16% of these chondroitinase ABC-generated "Slabeled disaccharides were disulfated disaccharides derived from chondroitin sulfate E; the remainder were monosulfated disaccharides derived from chondroitin sulfate A. Utilizing GM-CSF as a model of the cytokines, it was demonstrated that the GM-CSFtreated cells synthesized larger glycosaminoglycans onto B-D-xyloside than the noncytokine-treated cells. Thus, IL 3, IL 5 , and GM-CSF induce human eosinophils to augment proteoglycan biosynthesis by increasing the size of the newly synthesized proteoglycans and their individual chondroitin sulfate chains.
Recently, we developed in vitro methods for maintaining the viability of human peripheral blood eosinophils for at least 7 days by culturing these cells in enriched medium (RPMI 1640 supplemented with 100 units/ml penicillin, 100 pg/ml streptomycin, 10 pg/ml gentamicin, 2 mM L-glutamine, 0.1 mM nonessential amino acids, and 10% (v/v) fetal calf serum) that contains endothelial cell-conditioned medium (1) or human recombinant granulocyte/macrophage colony stimulating factor (GM-CSF)' (2), interleukin (IL) 3 (3), or IL 5 (4). Culture in the presence of any one of these three cytokines causes the eosinophils to undergo a change in their sedimentation characteristics such that they will be recovered in a less dense region of a discontinuous metrizamide gradient. These eosinophils also exhibit an augmented capacity for killing antibody-coated Shistosoma mansoni larvae and generate more leukotriene C4 when activated with calcium ionophore than freshly isolated cells. Upon short-term exposure to these cytokines, the cells respond with the same increases in function but without an appreciable change in their density sedimentation characteristics. These postmitotic changes in the human eosinophil are induced by those cytokines (IL 3, IL 5, and GM-CSF) that cause progenitor cells to proliferate and differentiate along an eosinophil lineage (5); they are not induced by IL l,, IL 2, IL 4, tumor necrosis factor, basic fibroblast growth factor, or platelet-derived growth factor (3).
Isolation and Culture of Human Eosinophils-Human eosinophils were isolated from the peripheral blood of seven different donors, none of whom were ingesting corticosteroids, aspirin, or other nonsteroidal anti-inflammatory drugs. Two of these donors had no diagnosed clinical disorder and had normal white blood cell counts and differentials. The other five donors were diagnosed as having allergic rhinitis, allergic conjunctivitis, and/or asthma; 2-10% of their white blood cells were eosinophils. The metrizamide isolation procedure used to obtain these eosinophils was performed as described (2,13). Residual contaminating erythrocytes in the initial eosinophil preparations were eliminated by hypotonic lysis. The purity of the starting population of normodense human eosinophils was 84 & 9% (mean f S.D., n = 20) as assessed by Wright's and Giemsa staining. The eosinophils from these seven donors all had a normodense phenotype and behaved similarly not only in this study but also in previous studies of other functional parameters (1)(2)(3)(4). Neutrophils were essentially the only leukocyte contaminant. These contaminating neutrophils do not survive under the culture conditions described below (1)(2)(3), and pure populations of eosinophils were routinely obtained after 2 days of culture in the presence of IL 3, IL 5, or GM-CSF. For the RNA blot hybridization experiment described below, eosinophils from the 21/22% metrizamide interface were isolated to a purity of 299% from a patient with the idiopathic hypereosinopbilic syndrome (4).
Freshly isolated eosinophils were routinely suspended at a density of 2.5-10 X lo5 cells/ml in enriched medium (RPMI 1640 supplemented with 100 units/ml penicillin, 100 pg/ml streptomycin, 10 pg/ ml gentamicin, 2 mM L-glutamine, 0.1 mM nonessential amino acids, and 10% (v/v) fetal calf serum) in the absence or presence of 10 pM IL 3, 1.0 PM IL 5, or 10 PM GM-CSF. Cells were cultured for up to 7 days at 37 "C in a humidified atmosphere of 5% (v/v) COZ. The culture medium containing the suspension of eosinophils was aspirated every 48 h. The eosinophils were centrifuged at 250 X g for 10 min at room temperature, resuspended in fresh enriched medium containing the appropriate cytokine, and added back to the original culture dish. In one experiment, eosinophils (5 X lo5 cells) were cultured for 24 h in enriched medium supplemented with IL 1, (1-10 units/ml), IL 2 (102-104 units/ml), or IL 4 (10-1-10-4 dilutions of the Cos supernatant). Radiolabeling of Human Eosinophils and Isolation and Characterization of Their 35S-Labeled Proteoglycans-Freshly isolated and cytokine-treated eosinophils were incubated for 1-17 h with 0-200 pCi/ ml [35S]sulfate in enriched medium and in cytokine-supplemented enriched medium, respectively, in the absence or presence of 0.01-0.1 mM p-nitrophenyl-P-D-xyloside (prepared as a stock solution of 80 mg/ml in dimethyl sulfoxide). After radiolabeling, a sample of the medium was removed for analysis of released 35S-labeled proteoglycans. The cells were sedimented at 250 X g. The supernatants were removed, and samples were retained for analysis. The pelleted 35Slabeled cells were lysed (11) by incubating the cells in 50-100 pl of 0.1% (w/v) Zwittergent 3-12 containing protease inhibitors (14) for -30 s at 4 "C, followed by the addition of 1.0 ml of TSG buffer (0.1 M Tris-HCl, 0.1 M sodium sulfate, and 4 M guanidine HC1, pH 7.0). One hundred pg of heparin and 100 pg of chondroitin sulfate A glycosaminoglycan carriers were added separately, and the samples were disrupted further by sonication. Samples of the radiolabeled supernatants and the cell lysates were analyzed by Sephadex G-25/ PD-10 chromatography for the incorporation of [35S]sulfate into released and cell-associated 35S-labeled macromolecules, respectively, both of which filtered in the excluded volume of the column. The two-tailed Student's t test was used to compare differences in the incorporation of [%3]sulfate into macromolecules by freshly isolated cells and by IL 3-treated cells. The 35S-labeled macromolecules in the remainder of the cell lysates were partially purified by CsCl density gradient centrifugation (11,15). The bottom fractions were dialyzed against 0.1 M ammonium bicarbonate, lyophilized, resuspended in 0.4-0.8 ml of water, and stored at -20 "C for later analysis.
Samples of the 35S-labeled macromolecules from each preparation were diluted to 0.2-0.5 ml with TSG buffer and were applied to 1 X 110-cm Sepharose CL-GB columns that had been equilibrated in TSG buffer. To measure their hydrodynamic sizes, the void and total volumes of the columns were determined with blue dextran and [35S] sulfate, respectively; mouse bone marrow-derived mast cell chondroitin sulfate E proteoglycan (Mr -200,000) (11) and rat basophilic leukemia 1 cell chondroitin sulfate diB/heparin proteoglycan (Mr -100,000) (10) were used as reference standards. The Pronase susceptibilities of the purified =S-labeled proteoglycans were determined by incubating samples in 100 pl of Hanks' balanced salt solution containing 10 pg of Pronase for 30 min at 37 "C. As a positive control, Pronase-sensitive human foreskin fibroblast-derived 35S-labeled proteoglycan (12) was incubated in parallel with the protease. The reactions were terminated by the addition of an equal volume of 8 M guanidine HCl, and the digests were analyzed by Sepharose CL-GB chromatography for a change in the hydrodynamic sizes of their 3sSlabeled proteoglycans.
35S-Labeled glycosaminoglycans were released from the purified 35S-labeled proteoglycans via P-elimination by incubating the samples for 17 h at 4 "C in 0.5 N NaOH (18). After neutralization with acetic acid, equal volumes of 8 M guanidine HCl were added, and the hydrodynamic sizes of the 35S-labeled glycosaminoglycans were determined by the gel filtration chromatographic method of Wasteson (19).
Uptake of 2-~4C]Deoxy-D-g~ucose by Human Eosinophils-The uptake of 2-[l-14C]deoxy-~-glucose was assessed by a modification of a previously described technique (20). Triplicate assays were performed in 1.5-ml polypropylene tubes in a final volume of 0.3 ml containing 3 X lo5 eosinophils in glucose-free Dulbecco's phosphatebuffered saline containing 0.1% (w/v) fatty acid-free bovine albumin, 0.9 mM Ca2+, and 0.5 mM M e . Eosinophils were preincubated at 37 "C for 15 min in buffer lacking or containing GM-CSF (10"3-10-8 glucose (0.5 pCi) was added, and the cells were incubated for an additional 60 min. The uptake of the radiolabeled carbohydrate was stopped by the addition of 1.0 ml of 4 "C phosphate-buffered saline and centrifugation at 5000 X g for 20 s at 4 "C. One ml of 4 "c phosphate-buffered saline was added, the cells were centrifuged again, and the amounts of "C radioactivity associated with the cell pellets were quantitated by 0-scintillation counting.

2-[l-'4C]Deoxy-D-
Denhardt's buffer, 0.1% (w/v) sodium dodecyl sulfate, 1 mM EDTA, 100 pg/ml salmon sperm DNA carrier, and 10 mM sodium phosphate containing 32P04-labeled cDNA-H4 (the cDNA that encodes the proteoglycan peptide core of HL-60 cells (6)). After the blots were washed under conditions of high stringency (55 "C; 30 mM NaC1, 3 mM sodium citrate, 0.1% sodium dodecyl sulfate, 1 mM EDTA, and 10 mM sodium phosphate), autoradiography was performed with Kodak XAR film. units/ml; n = l ) , or IL 4 (a 10"-10-4 dilution of the Cos supernatant; n = 1) for 1 day was not significantly different from that of freshly isolated cells. Hydrodynamic Size and Pronase Susceptibility of Human Eosinophil 35S-Labeled Proteoglycans-After CsCl density gradient centrifugation, 78 f 9% (mean f S.D., n = 5) of the 35S-labeled macromolecules synthesized by freshly isolated eosinophils were recovered in the high density fraction, consistent with the preferential incorporation of [36S]sulfate into proteoglycans. As shown in the representative experiment in Fig. 2 A , the 35S-labeled proteoglycans synthesized by eosinophils that had not been exposed to any cytokine other than those in the fetal calf serum were smaller in hydrodynamic size than those produced by eosinophils that had been cultured in the presence of IL 3 for 7 days. In five separate experiments with cells from different donors (Fig. 3), the 35S-labeled proteoglycans synthesized by freshly isolated eosinophils that were radiolabeled for 17 h in the absence of IL 3 filtered on Sepharose CL-GB columns with a K., = 0.28 f 0.04 (mean f S.D.), whereas replicate cells radiolabeled for 17 h in the presence of IL 3 synthesized proteoglycans that possessed a K., = 0.23 f 0.01. Eosinophils from the same donors that were exposed to IL 3 for 1 day or more before being radiolabeled synthesized even larger 35S-labeled proteoglycans which filtered with a Kav = 0.15 f 0.01 (mean k S.D.) ( Figs. 2A and  3). In one experiment, the proteoglycans released into the culture medium were found to be the same size as those remaining cell-associated for both the freshly isolated eosinophils (-80,000 and -80,000, respectively) and the 7-day IL-3-treated eosinophils (-300,000 and -300,000, respectively).

Radiolabeling of
As shown in the representative experiments depicted in   The M, of the 35S-labeled proteoglycans were estimated based on their gel filtration properties on Sepharose CL-GB columns. The columns were calibrated with 35S-labeled proteoglycans from rat basophilic leukemia cells (Mr -100,000) (10) and from mouse bone marrow-derived mast cells (Mr -200,000) (11). In each experiment, the cytokine-treated eosinophils were exposed to the cytokine for 1-7 days before the 17 2, B and C, eosinophils that had been cultured for 7 days in the presence of IL 5 or GM-CSF, respectively, also synthesized substantially larger 35S-labeled proteoglycans than the freshly isolated cells. Based on the Kav value of the M, -200,000 chondroitin sulfate E proteoglycan from mouse bone marrowderived mast cells and the M , -100,000 chondroitin sulfate diB/heparin proteoglycan from rat basophilic leukemia cells, the respective average hydrodynamic sizes of the proteoglycans synthesized by the freshly isolated ( n = 7), IL 3-treated (n = 6), IL 5-treated (n = 2), and GM-CSF-treated (n = 4) eosinophils were approximately M, 80,000, 300,000, 300,000, and 300,000 (Table I).
To determine if the smaller hydrodynamic size of the freshly isolated eosinophil proteoglycan was a consequence of the initial metrizamide isolation procedure used to purify these cells from peripheral blood, eosinophils that had been cultured for 7 days in the presence of IL 3 were incubated with [35S] sulfate before and after centrifugation on metrizamide gradients ( n = 1). The 35S-labeled proteoglycans produced by freshly isolated and IL 3-cultured eosinophils filtered with respective hydrodynamic sizes of M , -80,000 and -300,000.
When replicate cultured eosinophils were recentrifuged on metrizamide gradients and then radiolabeled, their 35S-labeled proteoglycans filtered with a hydrodynamic size of M, -300,000 (data not shown).
The ability of Pronase to degrade the 35S-labeled proteoglycans synthesized by human eosinophils was assessed by Sepharose CL-GB chromatography of the digests. There was no detectable degradation of the partially purified M, -80,000 35S-labeled proteoglycans synthesized by freshly isolated eosinophils ( n = 3) or of the M, -300,000 35S-labeled proteoglycans synthesized by IL 3-treated ( n = 2), IL 5-treated ( n = l ) , or GM-CSF-treated (n = 1) eosinophils under conditions in which Pronase fully degraded human fibroblast 35S-labeled proteoglycans (data not shown).
Analysis of the 35S-Labeled Glycosaminoglycans Bound to Human Eosinophil 35S-Labeled Proteoglycans-The 35S-labeled glycosaminoglycans bound to the proteoglycans synthesized by freshly isolated eosinophils filtered on Sepharose CL-6B columns with a K., = 0.67 (Fig. 4A). Replicate eosinophils that had been exposed to IL 3 for 24 h before their 17-h radiolabeling period synthesized 35S-labeled proteoglycans containing substantially larger glycosaminoglycans that filtered with a Kav = 0.43 (Fig. 4B). In three experiments (including that depicted in Fig. 4) the 35S-labeled glycosaminoglycans produced by the starting cells filtered with a K,, of 0.64 f 0.06 (mean f S.D.), whereas those produced by eosinophils that were exposed to IL 3 for up to 7 days had a Kav of Sepharose CL-GB chromatography of the s6S-labeled macromolecules synthesized by freshly isolated human eosinophils ( A ) or eosinophils exposed to 10 p~ IL 3 for 24 h ( B ) . The %-labeled macromolecules were filtered before (0) and after (0) NaOH treatment to release the glycosaminoglycans from their proteoglycans. Vo and V, indicate the void and total volumes of the column, respectively. 0.41 k 0.02. Based on these data, the hydrodynamic sizes of the 35S-labeled glycosaminoglycans from freshly isolated, IL 3-treated, IL 5-treated, and GM-CSF-treated eosinophils were estimated to be M, -8,000, -30,000, -30,000, and -30,000, respectively (Table 11).
The 35S-labeled proteoglycans synthesized by eosinophils that had been cultured in the absence or presence of different cytokines were incubated with chondroitinase ABC, and the net percentages of the total radioactivities that were degraded to 35S-labeled disaccharides were quantitated by Sephadex G-25/PD-10 chromatography. Freshly isolated eosinophils and IL 3-treated eosinophils synthesized 3SS-labeled proteoglycans that were 91 +12% (mean k S.D., n = 3) and 94 f 1% (mean f S.D., n = 5) degraded by chondroitinase ABC, respectively. When the chondroitinase ABC-generated unsaturated 35Slabeled disaccharides from freshly isolated eosinophils (Fig.  5A) and from eosinophils cultured for 7 days in the presence of IL 3 (Fig. 5B) were analyzed by HPLC, two peaks of radioactivity were obtained which had retention times corresponding to ADi-4S and ADi-diSE. In separate experiments, freshly isolated eosinophils and eosinophils that had been exposed to IL 3 for 1, 3, or 7 days synthesized 35S-labeled chondroitin sulfate proteoglycans in which 6-9% (for untreated cells), and 9, 11, and 16% (for IL 3-treated cells) of their total chondroitinase-generated 35S-labeled disaccharides were ADi-diSE, respectively; the remainder of the 35S-labeled disaccharides in each instance were ADi-4s. In other experiments, 92 f 1% (mean f range, n = 2), 98% ( n = l), and 96% (n = 1) of the total 35S-labeled macromolecules that were produced by freshly isolated eosinophils, 7-day IL &treated eosinophils, and 7-day GM-CSF-treated eosinophils, respectively, were found to be chondroitin sulfate proteoglycans. HPLC analysis of the chondroitinase ABC digests revealed that GM-CSF-treated eosinophils synthesized chondroitin sulfate in which 12 and 88% of the disaccharides were ADi-diSE and ADi-4S, respectively (Fig. 5C).
To confirm the presence of GalNAc-4,6-diS04, the 35Slabeled proteoglycans from freshly isolated eosinophils (n = 1) and eosinophils that had been cultured for 7 days in the presence of IL 3 (n = 1) or GM-CSF ( n = 1) were incubated with chondroitinase ABC in the presence of chondro-6-sulfatase, and the samples were then analyzed by HPLC. In each case, after exposure to chondro-6-sulfatase, 100% of the disaccharides that had the retention time of authentic ADi-diSE were converted to disaccharides that had the retention time of ADi-4S (data not shown).
Effect of P-D-Xyloside on Proteoglycan and Glycosaminoglycan Biosynthesis by Eosinophils-Because IL 3, IL 5, and GM-

TABLE I1 Estimated M, of the 36S-labeled glycosaminoglycans synthesized by freshly isolated and cytokine-treated human eosinophils
The M, of the released 36S-labeled glycosaminoglycans was determined utilizing the Sepharose CL-GB chromatographic method of Wasteson (19). In each experiment, the cytokine-treated eosinophils were exposed to the cytokine for 1-7 days before the 17-h radiolabeling period. The 35S-labeled glycosaminoglycans were derived from the %-labeled proteoglycans by &elimination.  CSF similarly induced human eosinophils to increase the size of the 35S-labeled glycosaminoglycans that were bound to their proteoglycans, we arbitrarily chose to use cells that had been cultured with GM-CSF to study the effect of p-nitrophenyl-P-D-xyloside on the biosynthesis of 35S-labeled macromolecules. In a representative dose-response study with 0, 0.01, 0.033, and 0.1 mM P-D-xyloside, freshly isolated eosinophils incorporated 1.3 X IO4, 1.7 X IO4, 2.3 X lo4, and 2.4 x lo4 cpm of radioactivity into macromolecules/106 cells, respectively, whereas replicate eosinophils that were also cultured in the presence of GM-CSF for 7 days incorporated 3.1 X lo4, 3.9 X lo4, 6.4 X lo4, and 8.4 X lo4 cpm/1O6 cells, respectively. P-D-Xyloside maximally increased the incorporation of [35S]su1fate into macromolecules by 108 40% (mean f S.D., n = 3) and 152 f 47% for freshly isolated and replicate 7-day GM-CSF-treated eosinophils, respectively, compared to non-P-Dxyloside-treated cells. P-D-Xyloside treatment (0.1 mM) of freshly isolated eosinophils increased the percent of 35S-labeled macromolecules in the medium pool from 21 f 2 to 37 f 6% (mean f S.D., n = 3). After 7 days of culture in GM-CSF, P-D-xyloside increased the release of the 35S-labeled macromolecules from 19 f 16 to 34 f 9%.
To determine the size of the 35S-labeled glycosaminoglycans synthesized onto 8-D-xyloside, the cell-associated 35S-labeled macromolecules were filtered on the Sepharose CL-GB gel filtration column. In the absence of 8-D-xyloside (Fig. 6A), freshly isolated eosinophils and eosinophils that were cultured in the presence of GM-CSF for 7 days synthesized 35S-labeled proteoglycans that were M, -80,000 and -300,000, respectively. These proteoglycans contained 35S-labeled glycosaminoglycans of M , -12,000 and 32,000, respectively (data not shown). GM-CSF-treated eosinophils exposed to 0.01 mM 8-D-xyloside (Fig. 6B) and freshly isolated eosinophils exposed to 0.033 mM P-D-xyloside (Fig. 6C) synthesized M, -18,000 and -12,000 35S-labeled glycosaminoglycans onto the exogenous acceptor, respectively. At the highest dose of @-D-xyloside (0.1 mM), both populations of eosinophils synthesized M, -12,000 glycosaminoglycans onto the exogenous acceptor (data not shown). Uptake of 2-['4C]Deoxy-D-g~ue into Eosinophils-To determine if cytokine exposure increased the rate of transport of glucose into the eosinophils, freshly isolated eosinophils were preincubated with various concentrations of GM-CSF, IL 3, or IL 5 for 15 min, and the uptake of 2-['"C]deoxy-~glucose was assessed during a subsequent 60-min incubation. As shown in Fig. 7, exposure of these eosinophils to GM-CSF resulted in a dose-dependent increase in the uptake of this radiolabeled carbohydrate. In three experiments (including the one in Fig. 7), eosinophils exposed to 1 0 " ' M GM-CSF had a 241 f 151% (mean f S.D.) increase in the uptake of the radiolabeled carbohydrate compared to that by noncytokine-treated cells. Eosinophils that were exposed to incremental concentrations of IL 3 (10-~~-10-~O M; n = 1) or IL 5 (10"5-10-'1 M; n = 1) took up 3.6-and 5.8-fold more 2-["C] deoxy-D-glucose, respectively, at the maximal cytokine concentration than the freshly isolated cells.
RNA Blot Analysis-To determine if the mRNA that encodes the HL-60 cell proteoglycan peptide core is expressed in human eosinophils, total RNA was extracted from freshly isolated eosinophils (299% purity) as well as from eosinophils that had been depleted of their contaminating neutrophils by a 2-day culture with IL 3. Total RNA from two preparations of -6 X lo6 eosinophils, -1 X lo5 HL-60 cells, and -1 X 10' of RNA from human endothelial cells (Fig. 8, lane 1). Similar in those lanes that contained HL-60 cell RNA and endothelial cell RNA, no 28 S or 18 S rRNA was detected in either lane that contained human eosinophil RNA (data not shown). Nevertheless, when the RN.4 blot was probed with cDNA-H4 under conditions of high stringency, HL-60 cells (Fig. 8, lane 2), freshly isolated eosinophils (Fig. 8, lane 3), and IL 3cultured human eosinophils (Fig. 8, lane 4 ) contained similar amounts of an -1.3-kilobase mRNA that hybridized to the cDNA. In contrast, the probe failed to hybridize to any species results were obtained when total RNA was prepared from eosinophils that were cultured in GM-CSF for 4 days (data not shown).

DISCUSSION
It has been reported (23) that human eosinophils from patients with hypereosinophilia synthesize M, -60,000 35Slabeled chondroitin sulfate proteoglycans when radiolabeled in the absence of any human cytokine. We demonstrate that when freshly isolated normal eosinophils from nonatopic or mildy atopic donors are radiolabeled in the absence of any cytokine other than those in fetal calf serum, they synthesize cell-associated M, -80,000 35S-labeled proteoglycans (Figs. 2 and 3, Table I) that contain MI -8,000 35S-labeled glycosaminoglycans (Fig. 4, Table 11). More than 90% of the glycosaminoglycans bound to these 35S-labeled proteoglycans were chondroitin sulfate. As assessed by its HPLC retention time and its susceptibility to chondro-6-sulfatase, 6-9% of the unsaturated disaccharides generated by chondroitinase ABC treatment were ADi-diSE, indicating that these glycosaminoglycans were chondroitin sulfate E (Fig. 5). Inasmuch as human lung mast cells (24), basophilic leukocytes from patients with myelogenous leukemia (25), and rodent mast cells (11,26) contain chondroitin sulfate E proteoglycans in their secretory granules, it seemed likely that this unusual cellassociated proteoglycan also resided in a granule compartment in the eosinophil.
The 35S-labeled proteoglycans synthesized by human eosinophils were found to be resistant to degradation by Pronase as is also characteristic of the intragranular proteoglycans of mast cells. This finding is most likely a consequence of the unique region of the peptide core where the glycosaminoglycans are attached; this region has been shown to be rich in serine and glycine in rodent mast cells (27)(28)(29). Based on the deduced amino acid sequence of their respective cDNA, the peptide cores of the proteoglycans that are synthesized by rat L2 yolk sac tumor cells (30) and rat basophilic leukemia cells (31,32) are the same; both have a proteoglycan peptide core that contains a 49-amino acid region of alternating serine and glycine. A human analogue of this gene has been isolated from a cDNA library prepared from the promyelocytic leukemia cell line, HL-60 (6). The presence of relatively high levels of an -1.3-kilobase species of RNA that hybridized under conditions of high stringency to the cDNA that encodes this HL-60 cell proteoglycan peptide core was demonstrated in eosinophils of ~9 9 % purity (Fig. 8). Thus, although mature eosinophils contain low amounts of total RNA, they contain abundant amounts of an mRNA that encodes a specific granule-localized proteoglycan peptide core. The HL-60 cell-derived cDNA-H4 encodes a M, 17,600 proteoglycan peptide core containing an 18-amino acid region that consists primarily of alternating serine and glycine with eight possible sites for glycosaminoglycan attachment. Therefore, it is likely that all of the glycosaminoglycan attachment sites in the M, -80,000 proteoglycan that is synthesized by freshly isolated eosinophils are occupied with M , -8,000 chondroitin sulfate E chains.
Exposure of mature eosinophils to the cytokines (IL 3, IL 5, and GM-CSF) which induce hematopoietic progenitor cells to proliferate and differentiate into eosinophils (5) also causes peripheral blood-derived eosinophils to undergo postmitotic phenotypic changes (1)(2)(3)(4). We have demonstrated that upon exposure to each of these cytokines, the eosinophils altered their biosynthesis of proteoglycans. Human eosinophils that were exposed to IL 3 (Fig. l ) Table I) that contained M, -30,000 glycosaminoglycans (Fig. 4, Table 11), compatible with the utilization of the eight glycosaminoglycan attachment sites. As assessed by their susceptibility to degradation by chondroitinase ABC and by the chromatography of the chondroitinase ABC digests on HPLC columns, the cytokinetreated eosinophils synthesized 35S-labeled chondroitin sulfate E glycosaminoglycans onto their peptide cores that had a type of sulfation similar to that of the noncytokine-treated cells (Fig. 5). Thus, although the cytokine-treated cells and the freshly isolated cells synthesize proteoglycans that have a similar number of chondroitin sulfate E chains, these glycosaminoglycans are substantially larger when the cells are exposed to IL 3, IL 5, or GM-CSF.
After establishing that IL 3, IL 5, or GM-CSF each induced eosinophils to increase the size of the 35S-labeled glycosaminoglycans bound to their proteoglycans, we chose one of these cytokines (GM-CSF) to investigate its effect on the biosynthesis of 35S-labeled glycosaminoglycans onto p-nitrophenyl-P-D-xyloside. Fresh eosinophils and GM-CSF-treated eosinophils incorporated -110 and -150%, respectively, more [35S] sulfate into macromolecules when cultured in the presence of p-D-xyloside than in the absence of the exogenous glycosaminoglycan acceptor. This finding indicated that the freshly isolated eosinophils and the GM-CSF-treated cells could synthesize more glycosaminoglycans than required, most likely because the amount of peptide core that reached the Golgi was rate-limiting. Additionally, the ability of the GM-CSFtreated eosinophils to synthesize more glycosaminoglycans in the absence and also the presence of P-D-xyloside indicated that the GM-CSF-treated eosinophils had an increased biosynthetic capacity compared to noncytokine-treated cells. Noncytokine-treated cells synthesized M, -12,000 35S-labeled glycosaminoglycans onto the exogenous acceptor (Fig. 6), whereas the glycosaminoglycans of the GM-CSF-treated cells were M , -18,000. The ability of GM-CSF to induce the synthesis of larger glycosaminoglycans onto P-D-xyloside indicated that the cytokine effect on proteoglycan biosynthesis was in part independent of the amount of peptide core.
Previous biosynthetic studies with mesenchymal and epidermal cell lines (33) have revealed that the size of the glycosaminoglycans bound to their constitutively secreted proteoglycans can be increased by treatment of these cells with transforming growth factor$. In other studies on the constitutively secreted proteoglycans synthesized by chondrocytes, the length of the chondroitin sulfate chain bound to the proteoglycan has been shown to be increased -250% after cycloheximide treatment (34)(35)(36) or -30% after insulin treatment (37). In the chondrocyte studies, it was proposed that the rate of proteoglycan peptide core being translated, the speed by which the peptide core of the proteoglycan moves through the Golgi, and the available pool size of the UDPsugars all influence the length of the chondroitin sulfate side chain. Although UDP-GalNAc and UDP-GlcUA pool sizes were not measured in the present study, the finding that the uptake of 2-['4C]deoxy-~-glucose was substantially greater in the eosinophils exposed to cytokines for 60 min in phosphatebuffered saline than in the noncytokine-exposed cells (Fig. 7) suggests that the cytokine-induced endocytosis of glucose may be a factor in the regulation of the size of eosinophils proteoglycans when cells were cultured in enriched medium with fetal bovine serum and cytokine. It is also possible that cytokine treatment of eosinophils increases the pool size of phosphoadenosine-phosphosulfate by increasing the transport of sulfate and/or cysteine.
We have found that the protease-resistant cell-associated chondroitin sulfate E proteoglycan synthesized by human eosinophils can be dramatically increased in size by treatment of these cells with IL 3, IL 5, and GM-CSF; this effect is primarily due to the increased size of their glycosaminoglycans. These results provide biochemical evidence that mature human eosinophils undergo postmitotic phenotypic changes when exposed to the cytokines which also regulate their proliferative differentiation.