Heterophile binding of human antibodies to glycoproteins of retroviruses.

The binding of human immunoglobulin to Type C viruses has been analysed by radioimmunoassay. The assay is a double-antibody, solid-phase RIA, which has been optimized and calibrated using rabbit and human anti-MuLV sera. It detects varying concentrations of IgG binding to HL-23-V-1, a human Type C virus isolate, in all of a large number of human sera tested. As judged by inhibition with nonspecific glycoproteins, heterophile antigens and pure saccharides, this binding is to the glycoside moiety of the virus-envelope glycoproteins, in agreement with other recent reports. Nonspecific binding of this type stringently restricts the interpretation which can be placed on these and earlier data in man concerning antibodies to Type C viruses. It does not however exclude the possibility that Type C viruses do occur in man and do elicit antibody therein.

SERO -EPIDEMIOLOGY provides important clues to the possible activity of oncogenic viruses in man. It is largely on this basis that Epstein-Barr virus has been identified, not only as the cause of infectious mononucleosis (Niederman et al., 1970) but also as a likely cause of Burkitt's lymphoma (de The et al., 1978) and nasopharyngeal carcinoma (Henle & Henle, 1976). This type of evidence has also been important in detecting endogenous and exogenous Type C virus activity in animals (Charman et al., 1975;Jhle et al., 1976;Nowinski & Kaehler, 1974). Any claim for serological evidence of Type C virus in man (Kurth et al., 1977) therefore deserves careful scrutiny.
The widespread presence of anti-Type C virus antibodies in the human population (healthy and cancer patients) has been described (Aoki et al., 1976;Caldwell et al., 1975: Kurth et al., 1977Kurth & Mikschy, 1978;Loui et al., 1976;Snyder et al., 1976), but also questioned (Gardner et al., 1977;Krakower & Aaronson, 1978: Stephenson & Aaronson, 1976. In the absence of a Type C virus of indisputable origin, proteins from well characterized mammalian Type C viruses have been used in the above studies. As most of these viruses are immunologically and biochemically similar, it was hoped that a putative human virus would follow the same pattern. The discrepancies between the reports have been ascribed to the use of different techniques, viral antigens and serum samples (Kurth et al., 1977;Kurth & Mikschy, 1978). Moreover, the specificity of these antibodies has been questioned (Hogg et al., 1979;Snyder et al., 1976).
A further investigation of the occurrence and specificity of antibodies to these viruses in man has accordingly been carried out. This started with the aim of carrying out sero-epidemiology, but developed into a critical study of the specificity of the antibodies involved. The virus chosen as the basis for the assay was HL-23V-1 virus, isolated from a human leukaemic cell line  and known to be similar to nonhuman primate viruses (Chan et al., 1976;Okabe et al., 1976;Teich et al., 1975). Sera were also tested for reactivity to murine Moloney leukaemia virus (M-MuLV) and to avian virus, the Prague strain of Rous sarcoma virus (RSV) Subgroup A. A simplified double-antibody solid-phase radioimmunoassay was developed and standardized. Binding activity was detected in a large number of normal human sera. Inhibition studies indicate that this binding was directed at the carbohydrates of viral glycoproteins, and was nonspecific in character. These findings have been reported in abstract (Russell et al., 1979). Similar conclusions have been reached in recent studies using radioimmuno-precipitation rather than solidphase assays (Barbacid et al., 1980;Snyder & Fleissner, 1980). METHODS The following cell lines were used: NIH/ 3T3, mouse fibroblast; NRK, rat kidney; KNRK, NRK transformed by murine sarcoma virus (MSV) Kirsten strain; 1OK, KNRK infected with HL-23V-1 (Teich et al., 1975); 7605L, human diploid fibroblasts; XC (Svoboda, 1960;Rowe et al., 1970). Cells were routinely grown in Dulbecco's modification of Eagle's medium supplemented with 10% foetal calf serum (FCS).
HL-23V-1 and the Moloney strain of mouse leukaemia virus (M-MuLV) were pelleted respectively from supernatants of 1OK and NIH/3T3 (M-MuLV-infected) cell lines and then purified in sucrose-density gradients. The-Prague strain of Rous sarcoma virus Subgroup A (RSV) was purified from supernatant of virus-infected chicken embryo fibroblast culture. This medium was supplemented with 10% tryptose phosphate broth, 1% chicken serum and 1% FCS. Feline leukaemia virus (FeLV-A/F422) produced from a cell line derived from a cat lymphoma, was received from Dr 0. Jarrett (Glasgow).
After dialysis against phosphate-buffered saline (PBS), viral protein concentration was determined by the Lowry method and the viruses kept in aliquots at -70°C.
Human sera were collected from normal volunteers or supplied by Dr R. Kurth (Friedrich -Miescher -Laboratorium, Max-Planck-Institute, Tubingen, West Germany) and Dr F. Katz (St Bartholomew's Hospital, London). Sera from terminal cancer patients immunized against murine Rauscher leukaemia virus (R-MuLV) and respective preimmune samples were kindly supplied by Dr E. M. Hersh (M.D. Anderson Hospital, Houston, Texas) (Hersh et al., 1974).
Rabbit anti-HL-23V-1 and anti-M-MuLV were prepared by s.c. injection of 400 ,ug of virus protein in complete Freund's adjuvant, followed by 200 ,ug of protein in incomplete Freund's adjuvant at 3 weekly intervals. Rabbit anti-Ig antibodies were purified by affinity chromatography. Antibodies and viruses were labelled by the chloramine T method (Greenwood et al., 1963).
In the binding assay, virus was first disrupted by freezing and thawing x 10. To each well of a flexible polyvinyl chloride "U" microtitration plate (Cooke) 1 jtg of viral protein in 50 Jul of PBS was added. After adsorbing overnight at room temperature the plates were washed in PBS and used or stored at -70°C. Before use, to minimize nonspecific binding, wells were filled with 100 jil of 4% human serum albumin (HSA, chosen because this protein does not competitively inhibit binding) in PBS and incubated for 1 h at 37°C. The plates were washed again and 50 ,ul of the antiserum dilutions in 2% HSA were added in triplicate and incubated for 1 h at 37°C, or overnight at 4°C, as indicated. After further washing, the plates were incubated for 1 h at 37°C with 50 ,ul/well of 1251-labelled anti-Ig antibody (16 ng) in 2% HSA. After final washing, wells were cut out and counted in a gamma counter. The figures show specific binding, given by subtracting the background binding to wells not coated with antigen from the binding in the presence of antigen. Titres are given by the inverse of the dilution giving 50% of the maximumIi binidiuig of the labelled second antibody. At the 5000 end point the assay detects 150 ng of purified antibody/ml of serum. In order to test for cross-reactivity between positive human sera and rabbit anti-viral sera, virus-coated plates w,ere preincubated for 1 h either wNith first human then rabbit serum, or vice versa. Competition was carried out by pre-incubating sera at their 5000 endpoint with increasing dilutions of antigen or monosaccharides. Incubation was carried out in microtubes at 4°C for 3 h, or as indicated, and NAas slightly higher at this than at room temnperature. After centrifugation the standard binding assay was performed. Maximunm binding was given by the unabsorbed samples. WA'hen human sera were tested for the presence of antigen(s) Ig was removed from sera by affinity chromatography.
Neutralization of HL-23V-1 infectivity was measured by the XC cell cytopathogenicity assay (Rowe et al., 1970)  Representative titration curves are shown in Fig. 1 for the binding of various human immunoglobulins to HL-23V-1. The curves are in parallel for (i) serum from a representative normal donor, (ii) pooled normal IgG, and (iii) anti-R-MuLV serum of human origin. All 97 human sera tested were positive, with a range of titre from 1: 8 to 1: 588. Titres in about the same range were obtained for binding to M-MuLV (data not shown). The titres could not be related to the condition of the serum donors. In the neutralization assay, none of the human sera neutralized I.
HL-23V-1 at a dilution of 1: 2, while a control rabbit antiserum to the virus neutralized at I :8000. Free HL-23V-1 competitively inhibited binding of normal human Ig, with minor variations between individual sera (Fig.  2a). So did M-MuLV with selected normal human serum, and to a lesser e.xtent so did RSV, but not a control protein, tetanus toxoid (Fig. 2b). A variety of glycoproteins and sera were then tested for competitive activity with selected normal human sera: BSA, fetuin, a2-macroglobulin and purified gp7O of HL-23V-1 proved active (e.g. 2%o BSA gave 18%o inhibition) and only HSA was inactive, as would be expected from the design of the assay. Other non-glycosylated proteins, besides tetanus toxoid and ribonucleases A and B, proved inactive, as did purified p30 of HL-23V-1. FCS and normal rabbit serum were both active, inhibiting by 44-5900 at a concentration of 500o. Reversing the assay, HL-23V-1 proved able to block the binding of human Igs to x2-macroglobulin adsorbed on to plates.
Intact cells were next examined for competitive activity. The mouse and rat cell lines could all absorb activity from selected normal human Igs, but not from rabbit anti-HL-23V-1 serum. In order to evaluate the importance of material picked up from the medium, cells were grown in media containing either FCS or in human serum selected for low reactivity towards HL-23V-1. KNRK cells grown under either condition removed all binding activity, whilst human 7605L cells, whether grown in FCS, human serum, or infected with HL-23V-1, did not remove the activity significantly (Fig. 3). KNRK cells and uninfected 7605L cells removed little activity from rabbit anti-HL-23V-1; 7605L cells infected with HL-23V-1 did absorb, as expected. Thus cell membranes bear intrinsic antigens which can absorb the binding activity from normal human Ig but not from anti-viral antibody. This clearly distinguishes between the specificity of the 2 types of Ig, and suggests that the normal human Igs are binding to heterophile antigens. In conformity with the heterophile-binding hypothesis, human (Blood Group A, AB, 0), hamster, sheep, chicken and rabbit, but not human or chicken erythrocytes could absorb to vary- serum in competition with human serum for viral antigenic sites. 0 Virus-adsorbed plates were incubated with the rabbit immune serum (1: 40) for 60 min at 37°C. After washing, human serum (sample 60) was titrated. * Human serum titration curve in the absence of preincubation with anti-HL-23V-1 serum.
OZ2-rhamnose) were not. In one experiment performed with U2-macroglobuhin-adsorbed plates, and the same serum end-point dilution, a mixture of the 3 abovementioned sugars showed an adsorption of 80% at a concentration of 2 mm.
In further confirmation of the distinction in specificity between normal human serum and rabbit anti-HL-2VI serum, the 2 types of serum did not compete with one another in binding to virus (Fig. 4).
In an attempt to identify further the viral components to which normal human Ig bind, virus eluted from sepharosebound pooled human Ig were run in 10% acrylamide gel. Only one band co-migrated gel and 2 with mol. wts of 64,000 and 60,000 respectively.
Several bleedings of 2 patients immun-. ized with R-MuLV (Hersh et al., 1974;Charman et al., 1975) were tested in the < assay, with similar results (Fig. 6). The reactivity of the immune serum at the 50% end-point when tested on M-MuLV plates, is reduced to 86% after absorption U' with rabbit red cells and by a further 700 to 7900 after absorption with KNRK cells. When tested using HL-23V-1 plates the reactivity at the 50% end-point was re-K duced to 3000 after absorptions with KNRK or rabbit red cells. The residual reactivity may reflect cross-reactivity # between HL-23V-1 and R-MuLV due to interspecific determinants. Alternatively, it may be a consequence of the presence of Kirsten murine sarcoma virus in the HL-23V-1 stock (Teich et al., 1975). The binding of the pre-immune serum virtually disappears in both cases after asbsorption.

DISC USSION
A solid-phase radioimmunoassay has been developed for the analysis of antibodies to Type C virusus. The assay is easy to perform, as it avoids the handling of large numbers of tubes. It proved to be sensitive, precise and specific.
Using this assay, Ig binding to the human retrovirus HL-23V-1 were detected in all of a large series of human sera. As judged by inhibition with isolated viral proteins, and by analysis of viral proteins binding to immobilized Ig, the binding is directed mainly at the envelope glycoprotein gp7O. In this respect our findings confirm earlier reports (Kurth et al., 1977;Kurth & Mikschy, 1978). Our interpretation of this binding is, however, very difficult, and conforms with that offered in more recent work (Barbacid et al., 1980;Snyder & Fleissner, 1980). As judged by competitive inhibition in the assay, binding activity has the following distribution: (i) diverse glycoproteins (fetuin, Ac2-macroglobulin, BSA, as well as _. v _ , .* w.i. _. .. :"w tive, but not non-glycosylated proteins (tetanus toxoid, ribonuclease A and B); (ii) mouse and rat cell lines are positive, but not as a result of picking up glycoproteins from their growth medium; (iii) various mammalian erythrocytes are positive, but not avian ones; (iv) human serum proteins and a human cell line are negative, as would be expected using Ig of human origin; and (v) certain monosaccharides at high concentrations are positive. This is precisely the distribution expected of Ig showing heterophile binding, i.e. binding to various carbohydrate groups present on cell membranes (Burger, 1971) and probably mainly as a result of immunization with bacterial cell-wall antigens. This hypothesis receives further support from our findings, again based on inhibition, that the specificities in normal human sera and immune anti-viral sera are distinct. This is not the first time that serologists have been misled by cross-reactions by carbohydrate-binding antibodies. Rabbit antisera to fish Igs were thought initially to detect a T-cell receptor, but upon further analysis turned out to be directed at carbohydrate determinants (Yamaga et al., 1977).
Our conclusions agree with those of Barbacid et al. (1980) and Snyder & Fleissner (1980). We think it important to provide data based on a solid-phase assay, which avoids the criteria devolved at precipitation assays. The problem of trapping irrelevant proteins in the immune complexes at high serum concentrations (Charman & Gilden, 1978) does not apply here. Thus, solid-phase assays should be useful in any futuire search for truly virusspecific antibody.
The possibility that man is a nonresponder to Type C viral proteins can be eliminated. In this study sera obtained from individuals before and after vaccination with Rauscher-MuLV were tested for reactivity to Moloney MuLV, which shares antigenicity with R-MuLV7. Both pre-and post-vaccination sera contained antibodies of the type detected in the rest of this report. However, after removal of this reactivity by absorption with either KNRK cells or rabbit red cells, detectable anti-Moloney viral antibodies remained in post-vaccination sera but not in the prevaccination sera (Fig. 6). Thus, man is able to respond specifically to Type C virus if brought into contact with it, and this humoral response can be detected in the radioimmunoassay used in this study.
We therefore conclude that the immunoglobulin present in human serum which binds to retroviruses is anti-carbohydrate in nature, poorly specific, and cannot be taken as evidence of retrovirus infection. This work was in part supported by a scholar.ship from the Lady Tata AMemorial Truist.