Vol. 83,pp. 3146-3150, May 1986. Biochemistry. Biochemical and functional similarities between human eosinophil-derived neurotoxin and eosinophil cationic.
Proc. Natl. Acad. Sci. USA Vol. 83, pp. 3146-3150, May 1986 Biochemistry
Biochemical and functional similarities between human eosinophil-derived neurotoxin and eosinophil cationic protein: Homology with ribonuclease (basic proteins/eosinophil granules/glycoproteins)
GERALD J. GLEICH, DAVID A. LOEGERING, MICHAEL P. BELL, JAMES L. CHECKEL, STEVEN J. ACKERMAN, AND DAVID J. MCKEAN Department of Immunology, Mayo Clinic and Mayo Foundation, Rochester, MN 55905
Communicated by Paul B. Beeson, December 30, 1985
ABSTRACT Eosinophil-derived neurotoxin (EDN) and eosinophil cationic protein (ECP) were isolated from lysates of human eosinophil granules by gel filtration and ion exchange chromatography on heparin-Sepharose. Radioimmunoassay, using monoclonal antibodies, of fractions from the heparinSepharose chromatography showed one peak of EDN activity and two peaks of ECP activity (termed ECP-1 and ECP-2). EDN, ECP-1, and ECP-2 each exhibited heterogeneity in charge and molecular weight when analyzed by two-dimensional nonequilibrium pH gradient electrophoresis and NaDodSO4/PAGE. Digestion of EDN with endoglycosidase F (endo F) decreased its molecular weight and charge heterogeneity. Thus, EDN likely contains a single complex oligosaccharide. Endo F digestion of ECP-1 and ECP-2 decreased the molecular weight of both polypeptides, indicating that both likely contain at least one complex oligosaccharide. Amino acid sequence analyses showed that ECP-1 and ECP-2 are identical from residue 1 through residue 59 and that the sequences of EDN and ECP are highly homologous (37 of 55 residues identical). Both EDN and ECP NH2-termnu*al sequences showed significant homology to RNase, especially in regions of the RNase molecule involved in ligand binding. EDN, ECP-1, and ECP-2 had neurotoxic activity, causing the Gordon phenomenon at doses down to 0.15 ,ug when injected into the cisterna magna; the proteins were comparable in their activities. These results indicate that EDN and ECP are related proteins and suggest that they derived from genes associated with the RNase family.
homology was found between partial amino acid sequences of EDN and ECP, and both showed striking homology with RNase.
MATERIALS AND METHODS Eosinophil Granules. From nine patients with marked peripheral blood eosinophilia, eosinophils were collected by cytapheresis with hydroxyethyl-starch (15). The mean percentage of eosinophils in the concentrate was 79% (range, 64-94%); between 2 x 1010 and 2 x 1011 eosinophils were collected from a single patient. For purification of the eosinophil granules from the large numbers of cells obtained, we used procedures described previously (1-3, 7, 16). The enriched granule fractions were transferred to 2-ml freezing vials (Nunc) and stored in liquid nitrogen for up to 3 years. EDN and ECP Isolation. Frozen granule pellets were thawed, adjusted to pH 3 with 0.1 M HCl, and centrifuged at 40,000 x g for 20 min. Supernatants from the granule extracts were applied to a 1.2 x 48 cm Sephadex G-50 column equilibrated with 0.025 M NaOAc/0.15 M NaCl, pH 4.2, as described (3, 7, 16). Fractions containing EDN and ECP in the second peak were pooled, dialyzed against 5 mM Na2HPO4/NaH2PO4/75 mM NaCl, pH 7.4, and applied to a 1.2 x 4 cm column of heparin-Sepharose (Pharmacia Fine Chemicals) equilibrated with the same buffer. The column was washed with starting buffer, and then a linear NaCl gradient was applied with a limit concentration of 1.5 M in 5 mM Na2HPO4/NaH2PO4, pH 7.4. Protein-containing fractions were dialyzed and lyophilized. Radioimmunoassay for EDN and ECP. EDN and ECP were measured by a competitive binding radioimmunoassay using murine monoclonal antibodies. Solutions (10 ,ug/ml) of purified monoclonal antibodies to EDN (J18-1A2) and to ECP (J18-1B6) (unpublished method) were coated on plastic prongs (Falcon F.A.S.T.; Becton Dickinson Labware, Oxnard, CA) and these prongs were incubated for 2 hr at room temperature with rocking. Nonspecific binding to the prongs was blocked by incubation with 0.1 M K2HPO4/KH2P04, pH 7.4/0.5% fetal calf serum/0.1% protamine sulfate for 2 hr at room temperature with rocking. After a washing step, the antibody-coated prongs were incubated (with rocking) with 0.1 ml of EDN or ECP as a competitive inhibitor for 30 min before addition of 0.1 ml containing 0.5 ng of radioiodinated EDN or ECP (specific activities: EDN, 45 mCi/mg; ECP, 51 mCi/mg; 1 Ci = 37 GBq). Plates were incubated overnight at 4°C with rocking, washed with
The human eosinophil granule contains several cationic proteins including the major basic protein (1-3), the eosinophil cationic protein (ECP) (4, 5), the eosinophilderived neurotoxin (EDN) (6, 7), and the eosinophil peroxidase (EPO) (8). Major basic protein and ECP are potent helminthotoxins (9-11), and major basic protein is toxic to mammalian cells (9, 12). When injected intrathecally into rabbits or guinea pigs, both EDN and ECP, but not major basic protein or EPO, produce the Gordon phenomenon, a neurologic syndrome characterized in the rabbit by stiffness, ataxia, muscle weakness, and muscle wasting (6, 7, 13). Recently, Fredens and his colleagues isolated a protein from human eosinophil granules, termed protein X (13, 14), that had neurotoxic activity (13) and may be the same as EDN. Here we report the isolation and biochemical characterization of EDN and ECP from eosinophil granules. Purified EDN and ECP displayed both charge and molecular weight heterogeneity when analyzed by nonequilibrium pH gradient electrophoresis (NEPHGE) and Na]DodSO4/PAGE, yet each gave a single NH2-terminal amino acid sequence. Marked
Abbreviations: ECP, eosinophil cationic protein; endo F, endoglycosidase F; endo H, endoglycosidase H; EDN, eosinophilderived neurotoxin; EPO, eosinophil peroxidase; NEPHGE, nonequilibrium pH gradient electrophoresis; Pi/NaCl, Dulbecco's phosphate-buffered saline.
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
3146
Biochemistry: Gleich et al. Dulbecco's phosphate-buffered saline (Pi/NaCl) containing 0.05% Tween 20, and blotted. The microsticks were cut, the pieces were placed in tubes, and the radioactivity was measured in a gamma scintillation counter. The standard curves for the radioimmunoassays ranged from 2.5 to 100 ng/ml for EDN and from 7.5 to 150 ng/ml for ECP. Results were analyzed by nonweighted logit-log transformation and linear regression using a Hewlett-Packard 9845A computer and radioimmunoassay program 09845-14254. Analytical Electrophoresis. Proteins were analyzed by onedimensional NaDodSO4/PAGE as described by Laemmli (17) except that reduction by dithiothreitol was conducted at pH 8.0 under the conditions described by Fairbanks et al. (18). EDN and ECP were also analyzed on two-dimensional gels: NEPHGE was run at 500 V for 100 min in the first (horizontal) dimension, and 12.5% NaDodSO4/PAGE (slab) was run in the second (vertical) dimension (19). Glycosidase Digestions. EDN and ECP preparations were boiled in 10% NaDodSO4 solution, precipitated with acetone, and dissolved in the appropriate buffers prior to digestion with endoglycosidases F and H (endo F and endo H) from New England Nuclear. Endo F and endo H digestions were done as described by Borst et al. (20). Amino Acid Sequences. EDN and ECP samples for sequence analysis were isolated on heparin-Sepharose, reduced with 0.01 M dithiothreitol in 7 M guanidinium chloride/0.02 M TrisHCl, pH 8.6, for 90 min at 370C, alkylated with 0.02 M iodoacetamide for 30 min at 220C, dialyzed against 0.1 M acetic acid, and lyophilized. The analysis was performed on a Beckman model 890D sequencer with a 0.1 M Quadrol program modified from Beckman 030176. Heptafluorobutyric acid cleavage was not included in the initial sequencer cycle to remove contaminating NaDodSO4. Polybrene (1 mg) was used to minimize peptide extraction from the cup (21). Phenylthiohydantoins were identified by HPLC on a Beckman Ultrasphere ODS reversephase column with a program modified from Tarr (22). To confirm the position of cysteine residues, proteins were reduced with dithiothreitol in 0.2 M Tris HCl, pH 8.6, and alkylated with iodo[14C]acetic acid (21). The 14C-labeled alkyl derivative proteins were subjected to sequence analysis, and the radioactivity associated with each residue was determined by liquid scintillation counting. Analysis of sequence data and homology search were performed by computer analysis with the DNASTAR program (DNASTAR, Madison, WI) and an IBM-XT computer. Sequence homologies were evaluated statistically by the method of Lipman and Pearson (23). Bioassay for Neurotoxic Activity. Neurotoxic activity was assayed by intrathecal injection of eosinophil proteins into New Zealand White rabbits as described (7). Test animals received eosinophil proteins in a volume of 0.2 ml and control animals received 0.2 ml of Pi/NaCl. Rabbits were observed daily for signs of the neurotoxic reaction by an observer who did not know what substance had been administered. RESULTS Purification of EDN and ECP. The results of a typical fractionation of EDN and ECP on heparin-Sepharose are shown in Fig. 1. A symmetric peak emerged at about 20,000 ,umho, and this material reacted with monoclonal antibodies to EDN. Two broad peaks followed, one at about 40,000 ,umho (ECP-1) and the other at about 50,000 ,umho (ECP-2), and material from both reacted with monoclonal antibodies to ECP. The pattern of immunoreactivity for ECP was the same whether the early-eluting (ECP-1) or later-eluting (ECP-2) ECP was used as the radiolabeled antigen in the assay. The NaDodSO4/PAGE fractions rich in EDN yielded a major band migrating at 18.6 kDa with a second minor band at 20.1
Proc. Natl. Acad. Sci. USA 83 (1986) 0.3 oE
cC
O cm 0.1
(0Q
0
co
E
>eM
60 = o 40 c, x 20 'D ._
0.2
_ r
-
.. \L I.'-* I*...
-N I....
0 0.2
E o.1
z u -
E
3147
0
0 E
AONO
0
I
-118
---.I
---0
so
0.2 0
A\
a
E o.1
J -01,
0 w
-t-.d
0
120
40
IIW
-,
seseeseesel---------L 80
.-Oew 160
200, 240
ml FIG. 1. Purification of EDN and ECP on heparin-Sepharose. Fractions from the second peak of eosinophil proteins from the Sephadex G-50 fractionation were pooled, dialyzed against 5 mM Na2HPO4/NaH2PO4/75 mM NaCl, pH 7.4, and applied to heparinSepharose. A linear NaCl gradient was applied with a 1.5 M NaCl limit. The radioimmunoassay for ECP used radiolabeled ECP-1.
kDa (Fig. 2). ECP-1 produced a major band migrating at 19.3 kDa and a minor band at 21.0 kDa. ECP-2 produced three bands: a doublet at about 18 kDa and another band at 18.6 kDa. Fractions rich in EDN or ECP activity were pooled and analyzed by the biuret method (24) and absorbance at 280 nm. For EDN, A1Ym = 15.5 0.02 (mean SD); for ECP-2, A1%m = 16.4 0.04. Two-dimensional analysis (NEPHGE and NaDodSO4/ PAGE) of EDN (Fig. 3) showed that EDN migrated as two major spots that differed in net charge but had the same apparent molecular weight. Two additional minor spots with slightly higher molecular weights were observed directly above each major spot. ECP-1 migrated as a single major spot with a heterogeneous tail in the acidic direction. ECP-2 migrated as a major spot that was more acidic than ECP-1 but also exhibited a heterogeneous tail that extended toward the basic end ofthe gel. A minor band at slightly higher molecular weight with charge heterogeneity comparable to that of the ±
±
±
Mr
x10-3 68_
24 am
18.4 12.3
VW
ri 76
60
f
68
92 84
132
164
148
140
156
196
180
172
188
204
FIG. 2. NaDodSO4/PAGE of EDN and ECP in selected fractions from Fig. 1. The marker proteins are bovine serum albumin, ovalbumin, trypsinogen, (-lactoalbumin, and cytochrome c. The lanes are identified by elution volume, corresponding to fractions in Fig. 1.
, blip@. ~ ~ ~.
.:
3148
Biochemistry: Gleich et al.
Proc. Natl. Acad. Sci. USA 83 (1986)
i -:
ha A*EDN Z
ha A-. ,.
A.
::
::.:
ffi
*:
.
