The porcine trophoblastic interferon-c, secreted by a polarized epithelium, has specific structural and biochemical properties Avrelija Cencic ˇ 1,2 ,Ce ´ line Henry 3 , Franc¸ois Lefe ` vre 1 , Jean-Claude Huet 3 , Srecko Koren 4 and Claude La Bonnardie ` re 1 1 Unite ´ de Virologie et d’Immunologie Mole ´ culaires, INRA, Jouy en Josas, France; 2 Faculty of Agriculture, University of Maribor, Slovenia; 3 Unite ´ de Biochimie des Prote ´ ines, INRA, Jouy-en-Josas, France; 4 Institute of Microbiology and Immunology, Medical Faculty, University of Ljubljana, Slovenia At the time of implantation in the maternal uterus, the trophectoderm of the pig blastocyst is the source of a massive secretion of interferon-gamma (IFN-c), together with lesser amounts of IFN-d, a unique species of type I IFN. This trophoblastic IFN-c (TrIFN-c) is an unprecedented exam- ple of IFN-c being produced spontaneously by an epithe- lium. We therefore studied some of its structural and biochemical properties, by comparison with pig IFN-c from other sources, either natural LeIFN-c (from adult leuco- cytes), or recombinant. Biologically active TrIFN-c is a dimeric molecule, of which monomers are mainly composed of a truncated polypeptide chain with two glycotypes, unlike LeIFN-c which is formed of at least two polypeptide chains and four glycotypes. TrIFN-c collected in the uterus lumen was enzymatically deglycosylated and analysed by mass spectrometry (MALDI-TOF). The data revealed that the more abundant polypeptide has a mass of 14.74 kDa, cor- responding to a C-terminal cleavage of 17 residues from the expected 143-residue long mature sequence. A minor polypeptide, with a mass of 12.63 kDa, corresponds to a C-terminal truncation of 36 amino acids. MALDI-TOF analysis of tryptic peptides from the glycosylated molecule(s) identifies a single branched carbohydrate motif, with six N-acetylgalactosamines, and no sialic acid. The only glycan microheterogeneity seems to reside in the number of L -fucose residues (one to three). The lack of the C-terminal cluster of basic residues, and the presence of nonsialylated glycans, result in a very low net charge of TrIFN-c molecule. How- ever, the 17-residue truncation does not affect the antipro- liferative activity of TrIFN-c on different cells, among which is a porcine uterine epithelial cell line. It is suggested that these specific properties might confer on TrIFN-c apartic- ular ability to invade the uterine mucosa and exert biological functions beyond the endometrial epithelium. Keywords: interferon-c; epithelium; mass spectrometry; truncated protein; N-glycosylation. Interferons (IFNs) are proteins or glycoproteins belonging to an extended family of cytokines. IFNs exert a broad spectrum of biological activities, such as eliciting an Ôantiviral stateÕ in target cells, which provides transient resistance to infection by numerous viruses [1]. Two types of IFNs have been described, which share no sequence homology: type I IFNs (a, b, x) include those produced mainly in response to a variety of viruses, while type II IFN has only one member, IFN-c, which in mammals is produced by activated T lymphocytes 1 and natural killer (NK) cells, and exerts various modulating effects on the immune response [2–4]. In pigs, from days 12–20 of development (i.e. around the time of implantation), the extra-embryonic trophectoderm secretes huge amounts (up to 250 lg per uterine horn) of IFN-c into the uterine lumen [5,6].This porcine tropho- blastic IFN-c (TrIFN-c) appears to constitute a unique case of Ôimmune IFNÕ being produced by a nonlymphoid cell. Moreover, the trophoblast is a polarized epithelium, without any tissue or functional relationship with leuko- cytes. In addition, this trophectoderm-derived IFN-c is produced in amounts that are far higher than those found in adult tissues during the immune response. The pig tropho- blast, which can reach 1 m in length, is made up of numerous trophectoderm cells, all of which are involved in a polarized (apical) IFN-c secretion through an unusual transcription of the single IFN-c gene, at around days 14–16 [7]. To date, the mechanism involved in TrIFN-c secretion has remained unclear, as has whether the epithelial origin of producing cells affects the structure and biological activity of this embryonic IFN. At the same time, the porcine trophoblast secretes another IFN, called IFN-d,whichwas found to be a novel type I IFN, as yet known only in the pig species, and which plays an unknown role in pregnancy [8]. In all known species, IFN-c is encoded by a single gene, and the protein produced by leukocytes is well characterized [9–11]. It consists of a dimer of variant glycotypes derived from the single polypeptide chain, which in man, mouse and pig contains two N-glycosylatable Asn residues [10,12,13]. Correspondence to A. Cencic ˇ , Faculty of Agriculture, University of Maribor Vrbanska 30, 2000 Maribor, Slovenia. Fax: + 386 2 22 96 071, Tel.: + 386 2 25 05 800 Abbreviations:IFN-c, interferon-gamma; TrIFN-c, trophoblastic interferon-gamma; rGIFN-c, glycosylated recombinant IFN-c; LeIFN-c, leucocytic IFN-c; rIFN-c, recombinant bacterial IFN-c; IPTG, isopropyl thio-b- D -galactoside; TMB, 3¢,3¢,5¢,5¢-tetra- methylbenzidine; VSV, vesicular stomatitis virus; MDBK, Madin-Darby bovine kidney; TBA, trophoblastic cell line; EL, endometrial glandular cell line; ST, swine testis; DMEM, Dulbecco’s modified Eagle’s medium; APA, antiproliferative activity. (Received 11 January 2002, revised 4 April 2002, accepted 22 April 2002) Eur. J. Biochem. 269, 2772–2781 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02950.x Full-length IFN-c has a basic net charge, most probably due to a near C-terminal cluster of Arg and Lys residues [14–16]. However, various forms of C-terminal truncations have been found to be associated with native IFN-c (reviewed in [17]). The fact that trophoblastic IFN-c is translated and secreted by an epithelial cell suggests that there may be some differences in the molecular structure and/or biochemical characteristics of TrIFN-c, when com- pared with leucocyte IFN-c. Consequently, the bioavaila- bility or biological activity of TrIFN-c might be changed. This paper analyses some of the structural, biochemical and functional properties peculiar to trophoblastic porcine IFN-c, by comparison with a natural IFN-c produced by activated porcine leukocytes (LeIFN-c) and a nonglycosyl- ated, recombinant porcine IFN-c expressed in Escherchia coli (rIFN-c). MATERIALS AND METHODS Source of porcine IFN-c Trophoblastic IFN-c (TrIFN-c). Pregnant gilts from the Chinese Meishan breed were anaesthetized by electric shock then normally slaughtered on day 15 of pregnancy. The entire reproductive tract was removed, and each uterine horn was flushed with 50 mL of medium 199 (Life Technology, Paisley, UK) containing penicillin G (100 UÆmL )1 ), Streptomycin (50 lgÆmL )1 ), and an antipro- tease cocktail of Trazylol, pepstatin and aprotinin. The flushed fluid was clarified by centrifugation at 2000 g,and frozen at )20 °C. Alternatively, for [ 35 S]Met labelling, TrIFN-c was collected in the supernatant of conceptus tissue maintained in culture in DMEM for 24 h at 38 °C with gentle shaking. Leucocytic IFN-c (LeIFN-c). LeIFN-c was obtained from pig peripheral blood leukocytes (PBL) stimulated with 4b-phorbol 12-myristate 13-acetate and phytohaemaggluti- nin according to a previously published protocol [18]. The supernatant containing natural LeIFN-c was collected 48 h after induction. Recombinant bacterial IFN-c (rIFN-c). The full-length porcine IFN-c cDNA, encoding the preinterferon sequence was obtained from a day 15 trophoblastic cDNA library (unpublished). From this cDNA, a translatable mature IFN-c sequence was constructed by use of PCR amplifica- tion, driven by primers designed to insert: (a) an ATG upstream of the nucleotide sequence encoding the mature protein (starting with a Gln residue); (b) two restriction sites, namely EcoRI and HindIII, in 3¢ and 5¢ ends of the coding sequence, respectively. The amplified fragment was digested with EcoRI and HindIII, and subcloned into pBS+ vector (Stratagene). The EcoRI–HindIII 456 bp fragment of one clone with the correct sequence was inserted into the expression vectors pET14 and pET22 (Novagen). The resulting plasmids pET14 metPoIFN-c and pET22met- PoIFN-c were used to transform E.Coli strain BL21 (DE3), which contains the T7 RNA polymerase under the control of the lac promoter [19]. Bacteria bearing metPoIFN-c were grown in Luria– Bertani medium supplemented with 1 m M MgCl 2 at 37 °C until D 600 ¼ 1.0. INF-c expression was induced by the addition of 1 m M isopropyl thio-b- D -galactoside (IPTG). After incubation for a further 4 h, bacteria were harvested by centrifugation at 3500 g 2 andstoredat)20 °C. The crude extractofrIFN-c was obtained essentially following the protocol developed by Vandenbroeck et al. [20]. Glycosylated recombinant IFN-c (rGIFN-c). RGIFN-c was obtained by constructing a tetracyclin-inducible expres- sion system in the RK13 cell line, as previously described [18]. Interferon assays ELISA. Coating was carried out with mAb G47 (INRA, Jouy-en-Josas) raised against porcine rIFN-c (CIBA-Geigy) in NaCl/P i (1 : 200 dilution). After overnight incubation, samples of IFN-c were diluted in assay buffer (fivefold dilutions in 0.05% Tween/NaCl/P i ). After a 1-h incubation at 37 °C, rabbit rIFN-c antiserum was added (1 : 500 dilution in NaCl/P i /0.05% Tween), and the plate was again incubated at 37 °C for one hour. Finally, 1 : 4000 diluted horseradish peroxidase-conjugated goat anti-(rabbit IgG) Ig (Biosys, France) was added. After a further 1-h incubation at 37 °C, staining was revealed with 3¢,3¢,5¢,5¢-tetra- methylbenzidine (TMB) at a concentration of 0.4 gÆL )1 in an organic base and 0.02% H 2 O 2 in a citric acid buffer according to the instructions of the supplier (Kirkegaard & Perry Laboratories Inc., or Sigma-Aldrich, USA). As a standard, porcine rIFN-c (CIBA-Geigy) was used at a concentration of 10 lgÆmL )1 . Antiviral activity. Antiviral activity was assayed by inhibi- tion of the vesicular stomatitis virus (VSV) cytopathic effect on the Madin–Darby bovine kidney (MDBK) cell line as described previously [21]. Titers were expressed in antiviral IU equivalents by a comparison with a calibrated porcine IFN-a laboratory standard. The amount of IFN-c (mg) was determined by ELISA. Specific antiviral activity was expressedinIUÆmg )1 . Growth inhibition test. The antiproliferative effect of purified TrIFN-c was measured by comparison to rGIFN-c and rIFN-c on several porcine epithelial cell lines and bovine MDBK cells. The trophoblastic cell line (TBA) was isolated from a 15-day-old pig conceptus and the endometrial glandular cell line (EL) from a cyclic uterus from Large White gilt. Both lines were developed at the Unite de Virologie et Immunologie Moleculaires, INRA, France. Swine testis (ST) is a previously published cell line [22]. In 96-well plates, quadruplicate threefold dilutions of each purified IFN (initial concentration 1 lgÆmL )1 ) were applied to monolayers of 1 · 10 5 cells (MDBK, ST) or 5 · 10 5 cells (EL, TBA) in Dulbecco’s modified Eagle’s medium (DMEM)/10% fetal bovine serum. Incubation was performed at 37 °C in an humidified incubator for 3 days. The plates were stained with Crystal Violet in ethanol, rinsed with water, and destained with 10% (v/v) acetic acid. The A 590 was measured, and the results were expressed, for each dilution, by the mean ratios (%, ± SD) of absor- bances in IFN-treated wells (n ¼ 4) to those in control wells (n ¼ 6). On ST cells, only TrIFN-c was assayed, but the effect of sheep antiserum 166 to type I IFN (a gift of C. Chany 4 , INSERM, Paris), known to neutralize IFN-d, Ó FEBS 2002 Structure of trophoblastic interferon-c (Eur. J. Biochem. 269) 2773 was tested to assess if trace amounts of IFN-d could partly account for the antiproliferative effect. By precaution, all other tests were performed in the presence of antiserum 166. IFN-c purification LPC-Hi Trap Heparin purification. Crude clarified cell culture supernatant containing rGIFN-c or bacterial crude clarified lysate were applied to a 5-mL Hi-Trap heparin column (Pharmacia, Sweden) with a flow rate of 1.5 mLÆ min )1 . After extensive washing (A 280 ¼ 0) with a Tris/HCl buffer, pH 8.0 (0.05 molÆL )1 ) and NaCl (0.5 mol L )1 ), IFN-c was eluted with a linear salt concentration gradient (0.05–1 molÆL )1 NaCl in Tris/HCl, pH 8.0) at a flow rate of 1mLÆmin )1 . Fractions positive for IFN-c were pooled and processed for further purification. Immunoaffinity chromatography. Partially purified rGIFN-c, rIFN-c or preclarified uterine flushes containing TrIFN-c were applied to a CNBr-activated Sepharose 4B (Pharmacia, Sweden) coupled with monoclonal anti-(por- cine IFN-c) Ig (C5). Unbound impurities were extensively washed off the column with NaCl/P i at pH 7.4. IFN-c was eluted with glycine/HCl buffer (0.2 molÆL )1 ),pH3.0,at which pH porcine IFN-c proved to be stable [19]. Eluted fractions were immediately raised to pH 6.0 by use of 1 M Tris base. Analytical procedures Gel filtration. Crude IFNs were applied to a 1.5 · 45 cm column packed with Sephadex G75 superfine (Pharmacia, Uppsala, Sweden). The column was equilibrated with 20 m M phosphate buffer, 0.5 M NaCl at pH 7.4. The flow rate was adjusted to 9.5 mLÆh )1 .IFN-c was assayed in every 1.5 mL fraction by ELISA and by antiviral assay on MDBK cells. Molecular mass marker proteins were bovine serum albumin (M r 66 000), ovalbumin (M r 43 000) and cytochrome c (M r 12 400). The void volume of the column was measured by use of Blue Dextran (M r 2000 000). 35 S-Labelling of natural IFN-c. For LeIFN-c,pigPBL were washed and suspended in methionine-free medium, then induced by the sequential addition of 4b-phorbol 12-myristate 13-acetate-phytohaemagglutinin, as described previously [18]. One hundred lCi per mL of a [ 35 S]Met-Cys mix (Amersham Pharmacia Biotech, Saclay, France) was added. The next day, fresh RPMI containing unlabeled methionine was added to the culture (1 : 20 dilution). Metabolically labelled LeIFN-c was harvested after 48 h of incubation. TrIFN-c was produced in the supernatant of freshly collected day 15 conceptuses as described above, except that methionine-free MEM and [ 35 S]Met-Cys (100 lCiÆmL )1 ) were used. Immunoprecipitation and deglycosylation of IFN-c. The 35 S-labelled IFN-c were concentrated against poly(ethylene glycol) (M r 20 000) to 2 mL and processed for immuno- precipitation by sheep anti-(mouse IgG) Ig (Biosys, Compie ` gne, France) coupled to Protein A–Sepharose, as previously described [18]. After final washes, the beads were resuspended in 30 lL of Laemmli buffer (glycosylated control), or in deglycosylation buffer: 30 lLof100m M Tris/HCl, pH 7.4, 1% SDS and 2% 2-mercaptoethanol (deglycosylated sample), and immediately boiled for 5 min to dissociate IFN-c from the beads. Samples of immuno- precipitated rGIFN-c, TrIFN-c and LeIFN-c in deglyco- sylation buffer were diluted 1 : 5 with 50 m M Tris/HCl, 1% Nonidet P40; recombinant N-glycosidase F (EC 3.5.1.52, from E. Coli, Boehringer, Mannheim, Germany) was added to a final concentration of 10 UÆmL )1 . The enzymatic reaction was carried out overnight at 37 °C. Deglycosylated IFN-c were precipitated with 4 vol. acetone. Washed pellets were resuspended in Laemmli buffer, then electrophoresed together with the glycosylated controls on a 15% acryla- mide gel [23]. The dessicated gel was exposed to autoradi- ography for 48 h at )70 °C. When necessary, the gels were re-exposed in a radioisotope imager (Phosphorimager, Molecular Dynamics). N-Terminal microsequence. Immunopurified TrIFN-c, obtained from uterine flushes, was subjected to electro- phoresis in SDS/PAGE, then electro-transferred on a ProBlott membrane, which was stained with Coomassie Blue R 250. The two main bands (M r 22 500 and 18 000) were cut out, and analysed for the N-terminal microse- quence. Digestion with Pyroglutamate aminopeptidase (EC 3.4.19.3, Sigma–Aldrich) was performed according to the enzyme supplier’s instructions. Automated Edman sequencing was performed using a PE Biosystems Procise 494 HT sequencer, with the reagents and methods des- cribed by the manufacturer. Mass spectrometry of proteins by MALDI-MS. Immuno- affinity-purified trophoblastic IFN-c, obtained by flushing pregnant uteri, was subjected to SDS/PAGE after treatment or mock-treatment with N-glycosydase F. After staining the gel with Coomassie blue, bands of interest were cut out and dried. Samples were transferred onto a poly(vinylidine fluoride) membrane by passive absorption as described previously [24]; the gel plugs were dried in a Speed Vac concentrator (Savant) for 30 min, then re-swollen in 50 lL 0.2 M Tris/HCl pH 8.5, 2%SDS, for 30 min. After swelling, 200 lL of HPLC water was added and then a 4 · 4mm piece of prewet 5 (methanol) PVDF membrane (Problott) was added to the solution. The procedure required 2 days at room temperature (23 °C) with gentle vortexing. At the end, the gel pieces and the solution were clear, and the membrane was blue. The membrane was washed five times with 1 mL 10% methanol with vortexing. Protein extraction was carried out by adding 40 lL of trifluoroacetic acid 5% plus CH 3 CN 50% and by gentle vortexing for 15 min. A second extraction was made, and the two extracts were pooled, then concentrated to 10 lL 6 on SpeedVac. One microliter of interferon was mixed on the stainless steel MALDI plate with 1 lL of DHB 7 (Aldrich) (10 mgÆmL )1 in 50% CH 3 CN, 0.15% v/v trifluoroacetic acid) and dried at room temperature. Mass spectra were acquired on a Voyager DE-STR + time-of-flight mass spectrometer (Applied Biosystems, Framingham, MA, USA) equipped with a nitrogen laser emitting at 337 nm. Spectra were recorded in positive linear mode with 25 kV as accelerating voltage, a delayed extraction time of 1200 ns and a 94% grid voltage. The spectra were externally calibrated using a mix composed by horse heart cyto- 2774 A. Cencic ˇ et al. (Eur. J. Biochem. 269) Ó FEBS 2002 chrome c (M + H) + ¼ 12 361.1 Da, horse apomyoglobin (M + H) + ¼ 16 952.6 Da and bovine carbonic anhydrase (M + H) + ¼ 29 024 Da. Tryptic peptide analysis by MALDI-TOF. Tryptic diges- tions of glycosylated IFNs were achieved directly in the gel matrix. The excised gel plugs were washed in 50% CH 3 CN in 50 m M NH 4 CO 3 (v/v) and then transferred to Eppendorf tubes. After dessication of the gel in SpeedVac for 30 min, the digestion was performed in 25 lLof50 m M ammonium bicarbonate pH 8.0 and 0.5 lgofmodifiedtrypsin (Promega, sequencing grade) for 18 h in a thermomixer (Eppendorf) at 37 °C with vortexing at 500 r.p.m. 8 A0.5-lL aliquot of sample was spotted directly onto the stainless steel MALDI plate. The sample was then allowed to dry at room temperature before addition of a 0.5-lL aliquot of the matrix solution. This dried-droplet sampling method was employed using a freshly prepared solution at 3mgÆmL )1 of a-cyano-4-hydroxycinnamic acid matrix in 50% (v/v) acetonitrile and 0.1% (v/v) trifluororacetic acid. For acquisition, the accelerating voltage used was 20 kV. Peptide spectra were recorded in positive reflector mode and with a delayed extraction of 130 ns and a 62% grid voltage. To analyse some peptides, spectra were recorded by the positive linear method with a delayed extraction of 160 ns and a 62% grid voltage. The spectra were calibrated using an external calibration which was composed of: Des-Arg 1 Bradykinin (M + H) + ¼ 904.468 Da, human angiotensin I (M + H) + ¼ 1296.685 Da, neurotensin (M + H) + ¼ 1672.917 Da, melittin from bee venom (M + H) + ¼ 2845.762 Da and bovine insulin B chain disulfonate (M + H) + ¼ 3494.651 Da. Samples digest with trypsin were calibrated using internal calibration with autolysis of trypsin: (M + H) + ¼ 2211.104 and 842.509 Da. RESULTS Active trophoblastic IFN-c is a dimer In order to determine the form in which TrIFN-c is present in the uterine lumen and therefore available to the endometrium, the M r of native TrIFN-c was measured by gel-filtration, in comparison with those of crude LeIFN-c and unglycosylated rIFN-c.Eachcolumnfractionwas tested by antiviral assay and by IFN-c specific ELISA. Elution profiles (Fig. 1) show that the antiviral activity eluted mostly as a single peak, around an M r of 43 000 for TrIFN-c (Fig. 1A), 50 000 for LeIFN-c (1B), and 34 000 for nonglycosylated rIFN-c. The scheme with TrIFN-c (Fig. 1A) was however, more complex; in ELISA, a single peak eluted at around 43 000, while the antiviral assay detected two peaks, one at 43 000 and slightly above, and one around 17–19 000. This second peak was most prob- ably due to the presence of IFN-d in the crude uterine flush, which had previously been shown to be monomeric, with an M r around 19 000 [25]. This peak was not detected by ELISA. Unexpectedly, for leucocytic IFN-c,andtoalesserextent for rIFN-c, the maximum ELISA score was delayed by one and two fractions with regard to antiviral activity. One possibility is that our ELISA is more specific for shortened IFN-c molecules (see below). TrIFN-c therefore appears to be essentially, if not entirely, dimeric, similar to natural LeIFN-c and unglycos- ylated recombinant IFN-c. Monomers of TrIFN-c are glycosylated and have shorter polypeptide chains Both TrIFN-c and LeIFN-c were metabolically radio- labelled with [ 35 S]Met, then immunoprecipitated with rabbit antiserum as described in Materials and methods. The Fig. 1. Sephadex G-75 gel filtration profiles of three preparations of crude porcine IFN-c. (A) TrIFN-c derived from uterine flushes of a day-15 pregnant gilt. (B) crude natural LeIFN-c.(C)recombinant IFN-c (E.coli).IneachfractionIFN-c concentration (d) was meas- ured by ELISA, and antiviral activity (h) was determined by antiviral assay on MDBK cells. Molecular weight standards and void volume (V 0 ) are indicated by arrows, and the black rectangle designates the elution area of expected IFN-c monomers. Ó FEBS 2002 Structure of trophoblastic interferon-c (Eur. J. Biochem. 269) 2775 monomers were analysed by denaturing SDS/PAGE (Fig. 2). The results were clearly contrasted: in the immu- noprecipitate from leukocytes, LeIFN-c consisted of four major bands (lane 1: M r 24 800; 22 000; 19 800; 17 500), the M r 24 800 band being slightly more pronounced. These four bands resolved into two bands on deglycosylation (lane 2: 16 000 and 14 000). As for TrIFN-c, only two main bands were visible at 22 500 and 18 000 (lane 3), which yielded one main band with an M r of 14 400 after N-glycosidase F treatment (lane 4), suggesting a single major polypeptide chain, but macroheterogeneity at the two potential glyco- sylation sites present on the IFN-c polypeptide core. They could differ in the rate and site of glycosylation, considering the 22 500-Da band as bi-glycosylated and the 18 000-Da band as monoglycosylated (Fig. 2). The deglycosylated 14 400-Da band may correspond to the truncation of about 20 amino acids in the embryonic IFN-c molecule, as the expected mass of full-length porcine IFN-c polypeptide is around 16 780 Da. In order to check if a full-length TrIFN-c form could be found in the trophoblast cells, which would be indicative of extracellular degradation, the same immunoprecipitation was performed on the conceptus cell lysate in parallel with the supernatant (Fig. 3). SDS/PAGE revealed only one major band in the cell lysate, at an apparent M r of 23 000– 24 000 (lane 2), which was slightly higher than the largest of the major monomers found in the supernatant (lane 1). Because of the scarcity of intracellular material, it was not possible to analyse the effect of N-Glycosydase F on this band, which casts a doubt on its glycosylation status. However, no equivalent of the largest LeIFN-c species (24 800) was found. Furthermore, the larger amount of TrIFN-c found, when compared to the previous experi- ment (Fig. 2), revealed that the 22 500 and 18 000 bands were the major but not the only components of TrIFN-c; two minor bands at M r 24 000 and 20 500 were also visible. These band most probably correspond to the diglycosylated and monoglycosylated forms of the minor polypeptide of M r 16 000 obtained after N-glycosidase F treatment (lane 3). TrIFN-c has an intact N-terminus, and a truncated C-terminus Immunoaffinity-purified trophoblastic IFN-c, obtained from pregnant uterine flushes, was electrophoresed in SDS/PAGE, after treatment (or mock treatment) with N-glycosydase F (Fig. 4A). Again, two major bands were found at M r 22 500 and 18 000 (lane 1), and upon deglycosylation, one major band at M r 14 400 was seen. But unlike TrIFN-c collected in the supernatant of cultured conceptuses, a minor deglycosylated band was obtained at M r 12 000 (lane 2). The two major TrIFN-c polypeptides yielded no residue by Edman microsequencing, a result compatible with a blocked pyroglutamate N-terminus (the expected mature sequence is Q-A-P-F-F-K-E-I-T-I-L-K-). Immunopurified TrIFN-c was then treated with pyroglu- Fig. 2. SDS/PAGE profiles of [ 35 S]Met metabolically labelled native TrIFN-c and LeTrIFN-c. Lane 1, control LeTrIFN-c.Lane2, N-glycosidase F treated LeIFN-c.Lane3,controlTrIFN-c.Lane4, N-glycosidase F-treated TrIFN-c. Fig. 3. SDS/PAGE profiles of [ 35 S]Met-labelled TrIFN-c after immu- noprecipitation by rabbit anti-(porcine IFN-c)Ig.Lane 1, glycosylated conceptus IFN secreted in the supernatant. Lane 2, intracellular TrIFN-c.Lane3,conceptussecretedIFN-c treated with N-Glycosi- dase F. Fig. 4. Mass determination of deglycosylated TrIFN-c species. (A) SDS/PAGE profiles of native TrIFN-c obtained in flushings of Day-15 pregnant uterus, control (lane 1) and N-glycosidase F treated (lane 2). (B) Mass spectrum obtained by MALDI-TOF of the M r 14 400 polypeptide. (C) Mass spectrum of the M r 12 000 polypeptide. 2776 A. Cencic ˇ et al. (Eur. J. Biochem. 269) Ó FEBS 2002 tamate aminopeptidase, and re-submitted to the microseq- uencing process. In the largest band, the A-P-F-F-K sequence appeared with a moderate yield, thus confirming that the native N-terminus was intact. In a second step, the mass analysis of the two deglycos- ylated bands was performed by MALDI-TOF. The 14 400- Da species yielded a main peak at (M + H) + 14 742.0 Da (Fig. 4B). This measured mass is compatible with a nonglycosylated polypeptide starting with an N-terminal pyroglutamate and ending at C-terminal L126. Indeed such a peptide has a theoretical sequence mass (average) of 14 712.0 Da, to which 17 Da must be substracted for N-terminal pyroglutamate, and 2 Da must be added for two N-glycosidase F-induced N fi D transitions, and 48 Da added for oxidation of three residues (probably the three M), respectively. The calculated (M + H) + obtained is then 14 746.0 Da, a value which differs by 4 Da from the observed mass. The MALDI-MS analysis of the minor peak with an M r of 12 000 yielded an observed (M + H) + of 12 635.0 Da (Fig. 4C). This is compatible with a deglycosylated poly- peptide with R 107 as C-terminus. Indeed such a 1–107 polypeptide with N-terminal pyroglutamate, three oxidized residues and two Asn/Asp transitions gives a calculated (M + H) + of 12 635.5, that is a 0.5-Da difference with the measured value. The second peak of the MALDI spectrum was measured at 12 762.4 Da (D mass ¼ 127.4 Da), a mass compatible with a peptide cleaved behind R107. Therefore, it is most probable that TrIFN-c is mostly composed of a polypeptide in which the C-terminus is cleaved after L126 (a lack of 17 residues), and of a minor polypeptide which is further cleaved, that is after R107 (a lack of 36 residues). TrIFN-c N-glycans contain no sialic acid, and have limited heterogeneity The tryptic peptide analysis of the four main bands obtained in PAGE were performed. We chose to point to data obtained for the M r 22 500 species. Figure 5 shows the complete sequence of pig IFN-c [26] with its theoretical trypsin cleavage sites, the peptides found upon MALDI analysis (underlined), and the deduced C-termini of each 14.74 and 12.63 kDa species (arrows). The coverage of the molecule was rather high as peptide analysis amounted to 87.3% of the sequence Q1-L126. Table 1 shows the comparison between theoretical and measured masses of tryptic peptides as provided by MALDI. Three main conclusions could be drawn: (a) on the peptide 1–6, the Fig. 5. Complete amino-acid sequence of mature porcine IFN-c [26]. Being 143 residues long, it includes two glycosylatable Asn at positions 16 and 83 (in grey). Expected trypsin cleavages are marked by slashes, and peptides analysed by MALDI-TOF are underlined. The two arrows point to the inferred C-termini of each 14.74 and 12.63 kDa species. Table 1. MALDI-TOF tryptic peptide analysis of the M r 22 500 TrIFN-c species. Peptide start Peptide end Sequence Theoretical (M + H)+ Measured (M + H) + Dmass (meas – theor) Remarks 1 6 QAPFFK 737.400 720.372 )17.028 N-term. pyroglut. 7 12 EITILK 716.455 716.452 )0.003 13 34 DYF…ILK 2397.198 4540.260 4394.224 4250.802 2143.062 1997.026 1853.604 Glycosylation Glycosylation Glycosylation 44 55 IIQSQIVSFYFK 1472.814 1472.847 0.033 56 61 FFEIFK 830.444 830.459 0.015 62 68 DNQAIQR 844.427 844.452 0.025 69 74 SMDVIK 692.364 692.348 708.351 )0.016 15.987 Oxidized Met 75 80 QDMFQR 824.372 824.395 840.379 0.023 16.007 Oxidized Met 81 88 FLNGSSGK 809.416 2805.871 2951.970 1996.455 2142.554 Glycosylation Glycosylation 98 107 IPVDNLQIQR 1195.679 1195.762 0.083 89 94 LNDFEK 765.377 765.383 0.006 109 115 AISELIK 773.476 773.480 0.004 116 123 VMNDLSPR 931.466 931.502 947.480 0.036 16.014 Oxidized Met Ó FEBS 2002 Structure of trophoblastic interferon-c (Eur. J. Biochem. 269) 2777 presence of N-terminal pyroglutamate is confirmed. (b) On peptides 69–74, 75–80 and 116–123, the three Met residues are oxidized. (c) On peptides 13–34 and 81–88, a mass excess of 2143 Da is the most probable signature of the same glycan conjugate. This mass is compatible with a sugar moiety made of three fucoses plus six N-acetylglucosamines plus three hexoses (monoisotopic mass ¼ 2142.80 Da). Table 1 and Fig. 6 show that other peaks differed from the major one by 146 Da, corresponding to the mass of fucose. Therefore, a microheterogeneity exists with at least three glycoform variants at each site, depending on the fucose content (one, two or three). Whether this is the real situation on the native molecule, or a consequence of laser-induced cleavage of fucose in the course of MALDI analysis is not known. The same analysis performed on the band at M r 18 500 (data not shown) indicated that the same glycan motif was present on the tryptic peptide 13–34, but absent on the peptide 81–88, for which the measured value was 809.42 Da [theoretical (M + H) + value is 809.41 Da]. We can there- fore conclude that if the main deglycosylated peptide is indeed 14.74 kDa, then the two main species of natural TrIFN-c found in uterine flushings have molecular masses of 19.03 kDa and 16.88 kDa, corresponding to diglycosyl- ated and monoglycosylated isoforms, respectively, the latter isoform being glycosylated on N16. As expected, the correspondance between the measured masses and observed M r in SDS/PAGE is quite good for nonglycosylated proteins, but not for glycosylated ones, as the latter have lowered electrophoretic mobility. Specific antiVSV activity of TrIFN-c is reduced Table 2 shows results concerning the antiviral activity of TrIFN-c, in comparison with LeIFN-c and two species of recombinant IFN-c, including the glycosylated rGIFN-c produced in transfected RK13 cells [18]. The specific activity of TrIFN-c on MDBK cells challenged with VSV was 1–5 · 10 5 UÆmg )1 of IFN-c (ELISA reactive), i.e. approxi- mately 10 times lower than that of its ÔadultÕ equivalent (LeIFN-c), and 20–50 times less than the two recombinant forms. TrIFN-c has an antiproliferative activity (APA) Immunoaffinity-purified IFN-c from uterine flushes did exert an APA on different cells. We first checked on pig swine testis cells that possible residual IFN-d was not a Fig. 7. Compared antiproliferative effect of TrIFN-c and other porcine IFN-c on several cell lines. Dilutions 1–6 represent serial threefold dilutions of purified IFNs, all of them being adjusted before assay to 1 lgÆmL )1 (A) ST cells treated with TrIFN-c in the absence (hatched bars) or presence (black bars) of rabbit antiserum 652 to type I IFN. (B,C,D): EL cells, TBA cells and MDBK cells, respectively, treated with serial dilutions of TrIFN-c (black bars), rGIFN-c produced in RK13 cells (grey bars) and rIFN-c expressed in E. coli (white bars). Values are means of four replicate assays per dilution, and the errors bars give the positive value of the SEM. Fig. 6. MALDI-TOF analysis of tryptic peptide 13–34 from the 22 500 IFN-c species. The area shown is an enlargement of the total mass spectrum. The main peak at an (M + H) + of 4540.26 is compatible with a N-glycosylation on N16 having the proposed structure drawn above the peak, including three fucose residues. Two other peaks on the left, with Dmass of 146.03 and 146.42 Da with each other, are compatible with masses of the peptide 13–34 with 2 and 1 fucose, respectively. (monoisotopic mass of fucose: 146.04). The two peaks marked with an asterisk represent the Na adducts of the two main peaks. The unmarked peak could not be identified. Schematic structure of the glycan conjugate was inferred by analogy with data obtained for human IFN-c [27].N-acetylglucosamine (j);hexose(d); fucose ( fi ). Table 2. Specific antiviral activity of TrIFN-c by comparison with other natural and recombinant IFN-c. Cell line IFN-c origin LeIFN-c rGIFN-c rIFN-c TrIFN-c Specific activity MDBK 1–5 · 10 5 1–5 · 10 6 2–3.5 · 10 6 5–10 · 10 6 (UÆmg )1 IFN-c) 2778 A. Cencic ˇ et al. (Eur. J. Biochem. 269) Ó FEBS 2002 significant effector of any APA by comparing the effect of dilutions from 300 ngÆmL )1 to 1.2 ngÆmL )1 in the absence or presence of antiserum to porcine type IFN (Fig. 7A), known to neutralize IFN-d [8]. Other cells were tested for their proliferation in the presence of TrIFN-c,andtwo purified recombinant proteins, one glycosylated (rGIFN-c produced in eucaryotic cells), the other free of carbohydrate chains (rIFN-c produced in E. coli). Figure 7B–D shows that, with cell-related differences, trophoblastic IFN-c exerted the same (in MDBK cells) or even more pronounced APA (in endometrial cells and trophoblast cell line TBA) than its recombinant counterparts. On pig EL and TBA cells, TrIFN-c was the most active on cell growth inhibition, especially in the first four dilutions, that is in the range of 300–11 ngÆmL )1 . On these same cell lines, recombinant E. coli-derived IFN-c was found the least active, which suggests that the glycosylation status is important for cell growth inhibition. DISCUSSION Embryonic TrIFN-c is the only IFN-c secreted by a nonlymphoid tissue. It is also a unique case among all IFN species, because it is intensely induced under physiological conditions (at the time of trophoblast implantation). TrIFN-c is secreted in substantial amount, simultaneously with IFN-d, in a polarized manner, by the trophectoderm. The precise structure and function of this embryonic, epithelial IFN-c has not been clarified to date. In this work, we have demonstrated that structural, biochemical and biological differences exist between TrIFN-c and LeIFN-c. As shown by gel filtration chromatography, TrIFN-c is accessible to the uterine lumen in the form of relatively heterogeneous glycosylated dimer with an apparent M r of 43 000. A shift towards a lower M r was noted for TrIFN-c, when compared to LeIFN-c, which eluted as a major heterogeneous peak at a M r between 50 000 and 60 000. On the other hand, rIFN-c exhibits no macroheterogeneity, as it elutes as one homogeneous peak at around 34 000, a size compatible with the correct predicted size of a biologically active dimeric protein. We can therefore conclude that functional embryonic IFN-c (TrIFN-c), like LeIFN-c,isa dimer. The weak antiviral activity found in fractions corresponding to monomers is certainly that of IFN-d, with an M r around 19 000, which is also present in uterine flushes [25]. As revealed by the electrophoretic profiles of 35 S-labelled TrIFN-c and LeIFN-c immunoprecipitates, TrIFN-c monomers differ from the LeIFN-c in terms of their polypeptide length and glycosylation pattern. Electropho- retic profile of TrIFN-c exhibits two major bands that are equimolar, with an apparent M r values of 22 500 and 18 000, thus suggesting that dimers are composed of equal proportions of mono glycosylated and biglycosylated monomers. The two glycoforms resolve into a major M r 14 000 band upon enzymatic deglycosylation with N-glycosidase F. The fact that TrIFN-c secreted in the supernatant of conceptus in culture presents with the same truncation as TrIFN-c collected in uterine fluids suggests that the cleavage of natural TrIFN-c is not due to endometrial peptidases. The pig trophectoderm has been shown to express various proteases, among which plasmi- nogen activator and different matrix metalloproteinases, which could, directly or by activation of protease cascades, lead to the observed cleavage of TrIFN-c [27,28]. In addition, in the flushed fluids, where TrIFN-c is supposed to be present in its bioavailable form, a minor polypeptide variant was observed after treatment with N-glycosidase F, with an apparent M r of 12,500, corresponding to a still more cleaved polypeptide. The MALDI-TOF resolution of deglycosylated TrIFN-c monomers, obtained from the uterine flush, confirmed results obtained by SDS/PAGE electrophoresis. The major form of TrIFN-c molecule is a polypeptide with a mass of 14 741 Da and a minor one with a mass of 12 634 Da. As confirmed by MALDI analysis, the two polypeptides found in TrIFN-c are truncated at the C-terminus. The major polypeptide lacks 17 C-terminal amino acids, as compared to the full length sequence, and a minor one is further truncated by 36 residues. Porcine LeIFN-c and TrIFN-c monomers are glycosylated, unlike human IFN-c,where nonglycosylated forms have also been found in crude preparations [29]. In the TrIFN-c molecule, little variability in the glycan structure are observed. Only three variants in glycan composition were found at both N-glycosylation sites, which differ only by the number of bound L -fucose molecules. Surprisingly, TrIFN-c glycans terminate with N-acetylglucosamine and not with sialic acid like for human IFN-c. Indeed, post-translational modifications, including glycosylation, are strongly dependant upon the type and physiological status of producing cells, and may signifi- cantly influence the characteristics of a glycoprotein [16,17,29]. From this point of view, no direct comparison has been possible with the glycan structure and heterogen- eity of porcine LeIFN-c, for which low amounts obtained in phytohaemagglutinin-activated pig PBL did not allow the same mass spectroscopy analysis. As a consequence of the C-terminal truncation, the native TrIFN-c lacks seven basic residues, in particular the R-K- R-K-R cluster (residues 127–131). It is therefore expected to be less positively charged than LeIFN-c or rIFN-c,which comprise full-length molecules. Indeed, unlike the two other IFNs, TrIFN-c, when analyzed by chromatofocusing, did not yield a readable pI, as it did not bind to a Mono-P column. In addition, attempts at binding TrIFN-c onto CM-cellulose columns at neutral pH were unsuccessful (data not shown). Although the calculated pI is 10.66 for the full length IFN-c molecule and 9.66 for the 1–126 polypep- tide, TrIFN-c behaves as a molecule without measurable net charge. Concerning biological activities, we found divergent results for antiviral and APA. The data shown in Table 2 suggest that TrIFN-c is much less antiviral than LeIFN-c and rIFN-c, as far as VSV challenge is concerned. It is possible however, that the relative values for TrIFN-c specific activity are underestimated, if it happened that the specificity of our ELISA test was slightly or significantly better for truncated molecules. In any case, the C-terminal truncation of TrIFN-c is most probably not the only reason for the reduced antiviral activity of TrIFN-c on MDBK cells (Table 2), such as that previously described for HuIFN-c [14]. The specific glycan composition that we found for TrIFN-c might also contribute to this reduced antiviral activity. On the other hand, we observed that TrIFN-c exhibits an APA on homologous (ST, TBA, EL) and heterologous Ó FEBS 2002 Structure of trophoblastic interferon-c (Eur. J. Biochem. 269) 2779 (MDBK) cells, that does not significantly differ from intact nonglycosylated rIFN-c and intact glycosylated porcine rGIFN-c (Fig. 7). Moreover, the APA of TrIFN-c on EL and TBA cells was even higher as compared to the intact porcine IFN-c. It seems that, especially in homologous cell lines, an intact IFN-c C-terminus is not essential for its biological function, as was shown for human IFN-c [30]. Our results also confirm previous data, namely that IFN-c antiviral and APAs can be dissociated [31–33]. Our results shed some light on the specific structure and properties of this atypical porcine trophoblastic IFN-c, produced by a polarized epithelium. It is probable that the structural and chemical characteristics of TrIFN-c affects its bioavailability and biological effect(s) on the maternal uterus. In particular, this shortened version of IFN-c, lacking a well known ECM-binding sequence and with very weak net charge, could be more prone than full-length IFN- c to cross the endometrial epithelium, and to reach cellular targets located in the uterine mucosa (e.g. lymphoid or endothelial cells). It is possible that the very particular context 9 in which this embryonic IFN-c is produced, namely between two opposite epithelia, has favoured the selection of a functionally adapted molecule, which differs from adult lymphoid IFN-c more by its bioavailability in this particular context than by its biological activity. ACKNOWLEDGEMENTS We would like to thank Christiane De Vaureix for her technical help. This work was supported by grants from the Slovenian Scientific Foundation and from the French Ministry of Foreign Affairs. REFERENCES 1. De Maeyer, E. & De Maeyer-Guignard, J. (1988) Interferons and Other Regulatory Cytokines.JohnWiley&sons,NewYork. 2. Ijzermans, J.N.M. & Marquet, R.L. (1989) Interferon-gamma: a Review. Immunobiol. 179, 456–473. 3. Baron, S., Tyring, S.K., Fleischmann Jr, R.W., Coppenhaver, D.H., Niesel, D.W., Klimpel, G.R., Stanton, J. & Hughes, T.K. (1991) The interferons, Mechanisms of action and clinical appli- cations. J. Am. Med Assoc. 266, 1375–1383. 4. Degre, M. (1996) Interferons and other cytokines in bacterial infections. J. Interferon Cytokine Res. 16, 417–426. 5. Cross, J.C. & Roberts, R.M. (1989) Porcine conceptuses secrete an interferon during the pre-attachment period of early pregnancy. Biol. Reprod. 40, 1109–1118. 6. La Bonnardie ` re, C., Martinat-Botte ´ ,F.,Terqui,M.,Lefe ` vre, F., Zouari, K., Martal, J. & Bazer, F.W. (1991) Production of two species of interferon by Large White and Meishan pig con- ceptuses during the peri-attachment period. J. Reprod. Fertil. 91, 469–478. 7. Lefe ` vre, F., Martinat-Botte ´ , F., Guillomot, M., Zouari, K., Charley, B. & La Bonnardie ` re, C. (1990) Interferon-gamma gene and protein are spontaneously expressed by the porcine tropho- ectoderm early in gestation. Eur. J. Immunol. 20, 2485–2490. 8. Lefe ` vre, F. & Boulay, V. (1993) A novel and atypical type one interferon gene expressed by trophoblast during early pregnancy. J. Biol. Chem. 268, 19760–19768. 9. Arakawa, T., Hsu, Y.R., Chang, D., Stebbing, N. & Altrock, B. (1986) Structure and activity of glycosylated human interferon-c. J. Interferon Res. 6, 687–695. 10. Ealick, S.E., Cook, W.J., Kumar, S.V., Carson, M., Nagabhushan, T.L., Trotta, P.P. & Bugg, C.E. (1991) Three-dimensional structure of recombinant human interferon-c. Science 252, 689–701. 11. Denesyuk, A.I., Zavyalov, V.P. & Korpela, T. (1994) Common structural patterns of cytokine outer surfaces. Bioch. Biophy. Res. Comm. 201, 1396–1400. 12. Dorner, F., Scriba, M. & Weil, R. (1973) Interferon: evidence for its glycoprotein nature. Proc. Natl Acad. Sci. USA 70, 1981–1985. 13. Sareneva, T., Pirhonen, J., Cantell, K., Kalkkinen, N. & Julkunen, I. (1994) Role of N-glycosylation in the synthesis, dimerization and secretion of human interferon-c. Biochem. J. 303, 1–10. 14. Arakawa, T., Narachi, M.A., Hsu, Y.R., Everett, R.R., Lai, P.H. & Fish, E.N. (1989) The effect of C-terminal processing on the activity of human interferon-c. Drug Des. Del. 4, 217–225. 15. Honda, S., Asano, T., Kajio, T., Nakagawa, S., Ikeyama, S. & Ichimori, Y. (1987) Differential purification by immunoaffinity chromatography of two carboxy-terminal portion-deleted deriv- atives of recombinant human interferon-c from Escherichia coli. J. Interferon Res. 7, 145–154. 16. James, D.C., Goldman, M.H., Hoare, M., Jenkins, N., Oliver, R.W.A., Green, B.N. & Freedman, R.B. (1996) Posttranslational processing of recombinant human IFN-c in animal expression systems. Protein Sci. 5, 331–340. 17. Hooker, A. & James, D. (1998) The glycosylation heterogeneity of recombinant human IFN-c. J. Interferon Cytokine Res. 18,287– 295. 18. Cencic, A., Lefe ` vre, F., Koren, S. & La Bonnardie ` re, C. (1999) Tetracycline-controlled expression of glycosylated porcine inter- feron-c in mammalian cells. Anim. Biotechnol. 10, 63–79. 19. Cencic, A. (1995) Porcine immune interferon. M.Phil. Thesis. Biotechnical faculty, University of Ljubljana. 20. Vandenbroeck, K., Willems, L., Billiau, A., Opdenakker, G. & Huybrechts, R. (1994) Glycoform heterogeneity of porcine inter- feron-c expressed in Sf9 cells. Lymphok. Cytok. Res. 13, 253–258. 21. La Bonnardie ` re, C. & Laude, H. (1981) High interferon titer in newborn pig intestine during experimentally induced viral enteritis. Infect. Immun. 32, 28–31. 22. McClurkin, A.W. & Norman, J.O. (1966) Studies on Transmis- sible gastroenteritis of swine II. selected characteristics of a cyto- pathogenic virus common to five isolates from transmissible gastroenteritis. Can. J. Comp. Med. Vet. Sci. 30, 190–198. 23. Laemmli, U.K. (1970) Cleavage of structural proteins during theassemblyoftheheadofthebacteriophageT4.Nature 227, 680–685. 24. Messer, M., Griffiths, M., Rismiller, P.D. & Shaw, D.C. (1997) Lactose synthesis in a monotreme, the echidna (Tachyglossus aculeatus): isolation and amino acid sequence of echidna alpha- lactalbumin. Comp.Biochem.Physiol.188B, 403–410. 25. Niu, P.D., Lefe ` vre, F., Me ` ge, D. & La Bonnardie ` re, C. (1995) Atypical porcine type I interferon. Biochemical and biological characterization of the recombinant protein expressed in insect cells. Eur. J. Biochem. 230, 200–206. 26. Dijkmans, R., Vandenbroeck, K., Beuken, E. & Billiau, A. (1990) Sequence of the porcine interferon-gamma (IFN-gamma) gene. Nucleic Acids Res. 25, 4259. 27. Geisert, R.D. & Yelich, J.V. (1997) Regulation of conceptus development and attachment in pigs. J. Reprod Fertil. 52 (supplement), 133–149. 28. Menino, A.R., Hogan, A., Schultz, G.A., Novak, S., Dixon, W. & Foxcroft, G.H. (1997) Expression of proteinases and proteinase inhibitors during embryo-uterine contact in the pig. Dev. Genet. 21, 68–74. 29. Mortz, E., Sareneva, T., Julkunen, I. & Roepstorff, P. (1996) Does matrix-assisted laser desorption/ionization mass spectrometry allow analysis of carbohydrate heterogeneity in glycoproteins? A study of natural human interferon-c. J. Mass Spectr. 31, 1109– 1118. 30. Luk, S.K., Jay, E. & Jay, F.T. (1990) Structure-function analysis of the human interferon-gamma. The COOH terminus is not essential for functional activity. J. Biol. Chem. 265, 13314–13319. 2780 A. Cencic ˇ et al. (Eur. J. Biochem. 269) Ó FEBS 2002 31. Carter, W.A., Swartz, H. & Gillespie, D.H. (1985) Independent evolution of antiviral and growth-modulating activities of inter- feron. J. Biol. Resp. Mod. 4, 447–459. 32. Goldberg, M., Belkowski, L.S. & Bloom, B.R. (1989) Regulation of macrophage growth and antiviral activity by interferon-gamma. J. Cell Biol. 109, 1331–1340. 33. Caruso, A., Tiberio, L., De Rango, C., Bonfanti, C., Flamminio, G.,Gribaudo,G.,Monti,E.,Viani,E.,Manca,N.,Garotta,G., Landolfo, S. & Turano, A. (1993) A monoclonal antibody to the NH2-terminal segment of human IFN-c selectively interferes with the antiproliferative activity of the lymphokine. J. Immunol. 150, 1029–1035. Ó FEBS 2002 Structure of trophoblastic interferon-c (Eur. J. Biochem. 269) 2781 . antiviral activity (h) was determined by antiviral assay on MDBK cells. Molecular weight standards and void volume (V 0 ) are indicated by arrows, and the black. bi-glycosylated and the 18 000-Da band as monoglycosylated (Fig. 2). The deglycosylated 14 400-Da band may correspond to the truncation of about 20 amino acids