Human eosinophil cytotoxicity-enhancing factor. Eosinophil-stimulating and dithiol reductase activities of biosynthetic (recombinant) species with COOH-terminal deletions.

U937 cells produce eosinophil cytotoxicity-enhancing factor (ECEF) polypeptides of 14 and 10 kDa that have identical NH2-terminal amino acid sequences. The 10-kDa form has greater eosinophil-stimulating activity (half-maximal at > 20-fold lower concentration). We considered the hypothesis that there is a precursor-product relationship between the 14- and 10-kDa species. Recombinant 14-kDa 104-amino acid ECEF (rECEF-104) had a slight stimulatory effect on eosinophil cytotoxic function at concentrations of 160 nM and above. In contrast, two species, rECEF-80 and rECEF-84, representing cleavage products of approximately 10 kDa had substantial statistically significant cytotoxicity-enhancing activity at concentrations as low as 10 pM. This evidence demonstrates the potential to generate the high-activity ECEF species by proteolytic cleavage of the 104-amino acid species. Another feature of this cytokine is the sequence from amino acids 31 to 34, which constitutes the conserved and active site of the enzyme thioredoxin. When tested for dithiol reductase enzymatic activity, rECEF-104 was active in a dose- and time-dependent manner, whereas the truncated forms of the molecule had no dithiol reductase activity. Thus the eosinophil-stimulating functions of the molecule do not correlate with its enzymatic activity. The evidence shows that the enzymatic activity is not essential for the initial interaction of ECEF with the eosinophil, and it suggests that the ECEF molecule functions by means of two discrete mechanisms.

U937 cells produce eosinophil cytotoxicity-enhancing factor (ECEF) polypeptides of 14 and 10 kDa that have identical NH,-terminal amino acid sequences. The 10-kDa form has greater eosinophil-stimulating activity (half-maximal at >2O-fold lower concentration). We considered the hypothesis that there is a precursor-product relationship between the 14-and 10-kDa species. Recombinant 14-kDa 104-amino acid ECEF (rECEF-104) had a slight stimulatory effect on eosinophil cytotoxic function at concentrations of 160 nM and above. In contrast, two species, rECEF-80 and rECEF-84, representing cleavage products of approximately 10 kDa had substantial statistically significant cytotoxicity-enhancing activity at concentrations as low as 10 p~. This evidence demonstrates the potential to generate the high-activity ECEF species by proteolytic cleavage of the 104-amino acid species. Another feature of this cytokine is the sequence from amino acids 31 to 34, which constitutes the conserved and active site of the enzyme thioredoxin. When tested for dithiol reductase enzymatic activity, rECEF-104 was active in a dose-and time-dependent manner, whereas the truncated forms of the molecule had no dithiol reductase activity. Thus the eosinophil-stimulating functions of the molecule do not correlate with its enzymatic activity. The evidence shows that the enzymatic activity is not essential for the intial interaction of ECEF with the eosinophil, and it suggests that the ECEF molecule functions by means of two discrete mechanisms.
The proinflammatory functions of human eosinophils are subject to regulation by cytokines. In terms of units of activity produced in uitro by monocytes or U937 cells, the predominant cytokine affecting eosinophil functions is an acidic heat-stable polypeptide, designated ECEF.' This material augments both * This work was supported by National Institutes of Health Grants AI-28525 and AI-22532 and by an award from the Milton Fund. This report represents the third in a series, following Refs. 3 and 5. 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. eosinophil cytotoxic function (1)(2)(3) and leukotriene synthesis (4).
ECEF derived from U937 cells consists of at least two polypeptide species, including a more abundant 14-kDa form and a less abundant 10-kDa form. The 14-kDa form is preferentially released into surrounding culture medium, whereas the 10-kDa form is preferentially retained in the cell membrane, externally oriented. With respect to eosinophil cytotoxic function, the 10-kDa form has >20 times higher activity. Both 14-and 10-kDa species have the same NH2-terminal amino acid sequence, and neither is glycosylated (5).
14-kDa ECEF appears to be identical to the cytokine/ enzyme known as "thioredoxin" or "adult T cell leukemiaderived factor" that has been studied (and its cDNA cloned and expressed) for its ability to regulate the growth or development of certain lymphocyte cell lines (6-9) and to protect U937 cells from the toxic effects of tumor necrosis factor (10). An interesting structural feature of this cytokine/enzyme is the presence of the conserved and active dithiol reductase moiety from the enzyme thioredoxin, which may be involved in the mechanism of signal transduction. 14-kDa ECEF consists of a single polypeptide chain of 104 amino acids (8, 9). Except for the 20 NH,-terminal amino acids, the structure of 10-kDa ECEF is unknown.
One model to account for these observations would be that there is a precursor-product relationship between the ECEF species, such that the 10-kDa polypeptide is derived by proteolytic cleavage of the 14-kDa precursor. Such a processing event has been demonstrated for many procytokines and prohormones (see Refs. 11 and 12) and, in U937 cells, for the specific examples of the cytokine oncostatin M (13). If this is the case for the ECEF species, then it should be possible to prepare a 10-kDa ECEF polypeptide equivalent to a truncated 14-kDa species and to demonstrate its enhanced activity. Our approach was to prepare recombinant ECEF polypeptides representing the full-length 14 kDa molecule and two (approximately 10 kDa) species truncated at the COOH terminus. We tested the activity of these substances in the assay of eosinophil cytotoxic function. Since 14-kDa ECEF (thioredoxin) is known to possess dithiol reductase activity (8), we evaluated the catalytic activity of the truncated rECEF polypeptides.

MATERIALS AND METHODS
cDNA Encoding ECEF-104"cDNA clones encoding ECEF-104 (full-length) were isolated from a PMA-stimulated U937 XgtlO cDNA library (Clontech, Palo Alto, CA). The probe for screening of the library was generated by polymerase chain reaction, using sequence information from Ref. 8. The primers 5"GTGAAGCAGATCGA GAGCAAG-3' (+strand NH2 terminus) and 5"GACTAATTCAT TAATGGTGGC-3' (-strand, COOH terminus) were used to amplify a sequence from total amplified library phage DNA (0.5 pg), resulting in a single amplified species of the appropriate size (312 base pairs).
This fragment was labeled with 32P by random priming (kit from Boehringer Mannheim) and used to screen the same U937 cell cDNA library.
From a total of approximately 400,000 phage clones, approximately 200 were positive on initial screening. Six of these were selected for further study and subcloned. The DNA sequence of the inserts was determined, and for all six, the sequence of the coding region was identical to that described in Ref. 9. Phage DNA was purified by phenol/chloroform extraction of SDS/EDTA-treated polyethylene glycol-precipitated phage from plaque lysates (14).
Production of rECEF Species-DNA species were prepared for insertion in the p-MALc expression vector (New England Biolabs, Beverly, M A except as specified, methods were according to the manufacturer's instructions) by polymerase chain reaction, using purified ECEF-104 cDNA clone as a template. The primers were designed to incorporate the coding region of the desired length, beginning with the NH2-terminal valine, and followed by a stop codon and an XbaI site. Thus, the following oligonucleotide primers were employed for all NH, termini, 5"GTGAAGCAGATCGA-GAGCAAG-3'; for the COOH terminus of rECEF-104, 5'-GGG TCTAGATTAGACTAATTCATTAATGGTGG-3'; rECEF-84, 5'-GGGTCTAGACTACTTTTGTCCCTI'CTI'AAAAAAGTGC-3'; and rECEF-80,

' -G G G T C T A G A C T A C T T A C T G G A A T G T T
GGCG-3'. The polymerase chain reaction fragments generated in the respective reactions were purified on SpinBind columns (FMC Bio-Products, Rockland, ME), blunted with Klenow enzyme, digested with XbaI, purified a second time on SpinBind columns, and ligated to StuI/ XbaI-digested pMAL-c plasmid. The constructs were used to transform TB1 cells. This permitted expression of a fusion protein consisting of Escherichia coli maltose-binding protein, the factor Xa cleavage site, and the ECEF species without additional amino acids at the NH, terminus. Several resulting colonies from each construct were selected for characterization. Initially, colonies were selected for production of an isopropyl-I-thio-P-D-galactopyranoside-induced fusion protein of the appropriate size. Subsequently, the production of the desired rECEF structure was confirmed by DNA sequence analysis of the plasmids and SDS-PAGE analysis of the factor Xa cleaved rECEF species. Maltose-binding protein produced by the pMAL-c plasmid with no insert was used for control studies.
Fusion proteins were cleaved with factor Xa (Boehringer Mannheim) according to instructions (New England Biolabs), except that 1% enzyme:protein (w/w) was used at 37 "C overnight.
Concentrations of rECEF species were determined spectrophotometrically, based on the absorbance of cleaved fusion proteins at 280 nm. For conversion of absorbance values to milligrams of protein, an extinction coefficient of 1.27 (=I mg/ml) was used. This coefficient was calculated for the cleaved MBP-rECEF-104 fusion protein, using the Bradford assay against a bovine serum albumin standard. The 1% contribution of factor Xa to the protein concentration was ignored. The major contributor to the absorbance was the maltose-binding protein, which was present at equimolar concentrations with the rECEF species. This method of assay was chosen, because the ECEF protein itself absorbs poorly at 280 nm (having only 1 tyrosine) and because of the possible effect that removal of short lysine-rich sequences at the COOH terminus might have on the Bradford assay.
DNA Sequence Analysis-Sequence analysis was performed by the dideoxy termination method (15), using the Sequenase enzyme and kit (United States Biochemical Corp.). Generally, the template was either the pMAL-c or Bluescript (Stratagene, La Jolla, CA) plasmid containing the insert of interest. In some experiments, sequence analysis was performed directly on products of a polymerase chain reaction (16). For example, insert regions from phage XgtlO clones were amplified by sets of primers complementary to regions flanking the insert. One of each set of primers was kinased with a 5"terminal phosphate. After the amplification, the double-stranded product was treated with h exonuclease to digest the strand with the 5"terminal phosphate. Hopp-Woods hydrophilicity analysis was performed using software of the Molecular Biology Computer Research Resource (Dana-Farber Cancer Institute and Harvard School of Public Health, Boston, MA).
SDS-PAGE Analysis-This method was carried out using Laemmli system buffers (17) and 15% polyacrylamide gels. For certain experiments, rECEF species were recovered from the polyacrylamide gel by electroelution in 0.1 M ammonium bicarbonate, using a dialysis membrane with a molecular weight cut-off of 3500 (model 1750 electrophoretic concentrator and sample cups, ISCO, Lincoln, NE). Eluted material was recovered in the 200-500 r l of solution closest to the membrane, lyophilized, and redissolved in assay medium (see below).
Eosinophil Purification and Assay of Eosinophil Cytotoxic Fumtwn-These procedures were carried out as described previously (2,3), except for the use of pooled monoclonal antibodies to mediate eosinophil antibody-dependent cellular toxicity. Briefly, human eosinophils were purified from the blood of normal or mildly eosinophilic subjects by dextran sedimentation and discontinuous metrizamide gradient centrifugation (18). Erythrocytes were removed by hypotonic lysis.
For cytotoxicity assays, IO5 purified eosinophils were cultured in roundbottomed microtiter wells in the presence of schistosomula-specific antibody (6.25% each of cell line E. 1,E.3,E.4,and M.4 culture supernatants;see Ref. 19), 100 mechanically transformed schistosomula (20), and a control or test source of ECEF. All components were dialyzed, dissolved, or suspended in assay medium (minimal essential medium with Earle's salts, supplemented with 25 mM HEPES, 2 mM glutamine, 100 units/ml penicillin G, 100 pg/ml streptomycin, and 10% fetal calf serum, all from GIBCO). Samples of ECEF had no direct toxicity to schistosomula targets.
Assay of Dithkl Reductase Catalytic Actiuity-The method of Holmgren (21) was used with some modifications. Test samples were added to wells of a microtiter plate (Costar no. 3596, Costar Corp., Cambridge, MA) in 0.1 M potassium phosphate buffer, pH 7.0, containing 2 mM EDTA and 1 mg/ml of insulin (Hybrimax no. 14011, Sigma). The reaction was initiated by the addition of dithiothreitol to a concentration of 1 mM. The final volume was 230 p1. Free insulin P-chain precipitates from solution under these conditions, so the increase in turbidity of the reaction mixture correlated with dithiol reductase activity. Absorbance was monitored over time at room temperature at 630 nm, using an automated enzyme-linked immunosorbent assay plate reader (model MR 600, Dynatech Laboratories, Inc., Alexandria, VA).

RESULTS
The molecule known as 14-kDa ECEF, thioredoxin, or adult T cell leukemia-derived factor is a 104-amino acid polypeptide, with a calculated molecular weight of 12,744 (9), so that a cleavage event at the COOH terminus yielding a 10-kDa polypeptide would occur in the vicinity of amino acid 82. Visual inspection and Hopp-Woods hydrophilicity analysis of the sequence in this region reveal that amino acids 72-79 are uncharged, whereas the sequence of amino acids from positions 80 through 84 (KKGQK) is highly charged, containing 3 lysines.
This lysine-rich region would appear to be a favorable site for a cleavage event, but it does not contain any of the recognized cleavage sequences employed in eukaryotic cells for the processing of prohormones or procytokines. Most notably, it contains no arginines (for review, see Ref. 11). Thus, there was no clear candidate for the position of the cleavage site, and it was decided to construct two truncated recombinant ECEF polypeptides: the first terminating at amino acid 80 (rECEF-80) and the second terminating at amino acid 84 (rECEF-84). The full-length 104-amino acid polypeptide (rECEF-104) was also prepared for comparison of eosinophil-stimulating and dithiol reductase activities (see Ref. 9 for complete sequence information).
The rECEF species were generated as fusion proteins with E. coli maltose-binding protein partner and an intervening factor Xa cleavage site. The construction was such that cleavage with factor Xa generated the ECEF NH, terminus, without any additional amino acids. After purification of the fusion protein by amylose resin affinity chromatography, the fusion proteins were cleaved with factor Xa. Analysis of the cleaved rECEF polypeptides showed that they were of the expected size (Fig. 1). Attempts to separate the free rECEF species from the maltose-binding protein by size exclusion or anion exchange high performance liquid chromatography or by an ad-  Eosinophil-stimulating activity of rECEF epecies. The cytotoxic function of eosinophils was evaluated in the standard 40-h antibody-dependent assay using Schktosoma mansoni larvae as targets. rECEF species were tested at the indicated concentrations (serial 5fold from 800 nM at the Ieft to 82 fM at the right). None of these substances had any direct toxicity to the targets. A representative experiment is shown. Data represent the means of triplicate determinations. Maltose-binding protein ( M B P ) treated with factor Xa and factor Xa alone (added to buffer, incubated, and diluted in a similar manner) were tested for activity at the three highest concentrations only. Increases in cytotoxic function were statistically significant (twotailed t test for independent measurements) with respect to the "no ECEF" control for rECEF-104 a t ROO and 160 nM only and for rECEF-84 and rECEF-80 at 10 p~ and higher. ditional passage over the amylose resin were unsuccessful, with rECEF and maltose-binding protein co-eluting in all fractions.
When rECEF species were separated by preparative SDS-PAGE, the polypeptides tended to precipitate from solution in the absence of detergent. This did not occur when the rECEF species were left in the presence of the maltose binding protein (data not shown). Therefore, preparations of cleaved but unseparated fusion proteins were used for most experiments, with controls for possible effects of the maltose-binding protein and factor Xa.
In the assay of eosinophil cytotoxic function, concentrations of 160 and 800 nM rECEF-104 had a slight, if any, stimulatory effect (Fig. 2). In separate experiments, higher concentrations of rECEF-104 did not have a greater effect on cytotoxic function (data not shown). By comparison, rECEF-80 and rECEF-84 enhanced eosinophil cytotoxic function substantially, at concentrations from 10 PM to 800 nM (Fig. 2). The effect of rECEF-80 was significant in three experiments at concentrations of 10 PM and higher, the effect of rECEF-84 was similar in two experiments and significant at 1.2 nM and higher in a third ( p < 0.05 by the two-tailed t test for independent samples).
None of the rECEF preparations had direct toxicity to schistosomula targets. In parallel, two control samples (the maltose-binding protein as produced by the pMAL-c plasmid without an insert, treated with factor Xa and buffer alone treated with factor Xa, prepared by the same methods as the fusion proteins, and tested at the three highest concentrations) had no effect on eosinophil cytotoxic function (Fig. 2).
The rECEF preparations were tested for dithiol reductase catalytic activity (the ability to catalyze the reduction of the intrachain disulfide bonds of insulin in the presence of a suboptimal concentration (1 mM) of dithiothreitol). Of the three rECEF species, only rECEF-104 had the catalytic activity. This property was detectable at concentrations as low as 160 nM (Fig. 3A) and was time-dependent (Fig. 3B). Control preparations of factor Xa-treated maltose-binding protein and Xa-treated buffer had no enzymatic activity (Fig. 3A).
Two further experiments were carried out to address the possible contribution of maltose-binding protein or factor Xa to the activities that were observed. The first of these involved the separation of rECEF-80 from maltose-binding protein and factor Xa by preparative SDS-PAGE. Material prepared in this manner enhanced eosinophil cytotoxic function over a range of concentrations from approximately 2 to 200 nM, whereas mock-eluted material from a different region of the same gel did not (Fig. 4). However, as mentioned above, rECEF-80 prepared in this manner tended to precipitate from solution, an occurrence that interfered with accurate measurement of the effective dose.
The activity of factor Xa in these preparations was addressed specifically by assay with the chromogenic substrate Chromozym X. It was found that substantial factor Xa activity does persist in these preparations for over a month at 4 "C. However, this activity was almost completely destroyed by boiling of the sample for 5 min. Eosinophil-stimulating and dithiol reductase activities were not affected by boiling and, thus, were not dependent on the proteolytic activity of factor Xa (Fig. 5 ) .

DISCUSSION
The preparations of the rECEFs contained an equimolar concentration E. coli maltose-binding protein and 1% (in protein mass) of factor Xa. Neither maltose-binding protein, factor Xa, nor possible other undetected contaminants from E. coli or the isolation procedure had any independent activity in the assays of ECEF function. Taken together with the differential activities of the rECEF species, the heat sensitivity of factor Xa as compared with ECEF, and the activity of SDS-PAGEpurified ECEF, this information argues that the observed effects were due only to the rECEF species, not to other components of the preparation.
The activity of rECEF-80 and -84 was similar to that determined for purified natural 10-kDa ECEF (half-maximal between 16 and 400 p~) , as evaluated by the same biological assay (3). This finding demonstrates the potential to generate an active 10-kDa ECEF species by proteolytic cleavage near the COOH terminus of 14-kDa ECEF. To date, it has not been possible either to demonstrate a precursor-product relationship between 14-and 10-kDa ECEF species by pulse-chase-type experiments ( 5 ) or to determine the COOH-terminal amino acid sequence of 10-kDa ECEF. The chief difficulty in these studies has been that 10-kDa ECEF is produced only in trace quantities, such that hours of metabolic radiolabeling are required to detect it (5).
The loss of dithiol reductase activity by the truncation of ECEF shows that at least part of the sequence from amino acid position 85 to the COOH terminus is essential to the dithiol reductase catalytic activity in solution, even though the conserved catalytic moiety (amino acids 31-34) was present in all rECEF species.
Thus, truncation of the 14-kDa ECEF COOH terminus increases the molecule's eosinophil-stimulating activity but abolishes the dithiol reductase activity. This evidence appears to argue against the involvement of the enzymatic activity in the mechanism of signal transduction in the eosinophil. Clearly, the dithiol reductase activity is not required for the initial interaction of rECEF-80 or -84 with the eosinophils. However, it is not possible to rule out a subsequent role of the dithiol reductase, because its activity might be reconstituted by a FIG. 5. Differential sensitivity to heating of rECEF and factor Xa activities. Aliquots of rECEF preparations were boiled for 5 min and assayed as described previously. Factor Xa activity was measured using a 20 PM concentration of rECEF-104 for heated and unheated samples, with spectrophotometric values (405 nm, background subtracted) to be read on the left a i s . Dithiol reductase activity was measured using a 640 nM concentration of rECEF-104, with the spectrophotometric values (630 nM, background subtracted) to be read on the left a i s . Eosinophil cytotoxic function was measured using a 1.2 nM concentration of rECEF-80, with the percentage of targets killed to be read on the right axis (compared with 28% with no ECEF).
conformational change or association with other molecules after the initial interaction with the eosinophil.
Studies involving the full-length ECEF (thioredoxin/adult T cell leukemia-derived factor) molecule suggest that it is pleiotropic in its activities, with documented effects on several cell types. In all of these studies, the effective dose was 100 nM or higher, a concentration uncharacteristically high for a cytokine. Our finding that rECEF-80 or -84 has stronger activity at doses as low as 10 PM suggests that a truncated form of ECEF may be the physiologically active molecule and that rECEF-80 or -84 should be used to study the biology of this regulatory factor.