The Receptor Binding Site of Feline Leukemia Virus Surface Glycoprotein Is Distinct from the Site Involved in Virus Neutralization

Abstract

The external surface glycoprotein (SU) of feline leukemia virus (FeLV) contains sites which define the viral subgroup and induce virus-neutralizing antibodies. The subgroup phenotypic determinants have been located to a small variable region, VR1, towards the amino terminus of SU. The sites which function as neutralizing epitopes in vivo are unknown. Recombinant SU proteins were produced by using baculoviruses that contained sequences encoding the SUs of FeLV subgroup A (FeLV-A), FeLV-C, and two chimeric FeLVs (FeLV-215 and FeLV-VC) in which the VR1 domain of FeLV-A had been replaced by the corresponding regions of FeLV-C isolates. The recombinant glycoproteins, designated Bgp70-A, -C, -215, and -VC, respectively, were similar to their wild-type counterparts in several immunoblots and inhibited infection of susceptible cell lines in a subgroup-specific manner. Thus, Bgp70-A interfered with infection by FeLV-A, whereas Bgp70-C, -VC, and -215 did not. Conversely, Bgp70-C, -VC, and -215 blocked infection with FeLV-C, while Bgp70-A had no effect. These results indicate that the site on SU which binds to the FeLV cell surface receptor was preserved in the recombinant glycoproteins. It was also found that the recombinant proteins were able to bind naturally occurring neutralizing antibodies. Bgp70-A, -VC, and -215 interfered with the action of anti-FeLV-A neutralizing antibodies, whereas Bgp70-C did not. Furthermore, Bgp70-C interfered with the action of anti-FeLV-C neutralizing antibodies, while the other proteins did not. These results indicate that the neutralizing epitope(s) of FeLV SU lies outside the subgroup-determining VR1 domain.

Feline leukemia virus (FeLV) is a major cause of degenerative and malignant diseases in domestic cats. The envelope of FeLV, which is a typical type C retrovirus, is studded with a transmembrane protein (TM) which anchors an external surface glycoprotein (SU). The FeLV SU (gp70) is the target of virus-neutralizing antibodies and the site of the initial virus-host cell interaction (14, 19, 35, 37, 39).

Three subgroups of FeLV (A, B, and C) have been defined on the basis of interference with superinfection (39). FeLV subgroup A (FeLV-A) is found in all cases of FeLV infection (13). It is antigenically monotypic (36) and is believed to be responsible for interhost transmission. FeLV-B is found in approximately 33% of FeLV-positive cats that are otherwise healthy (13). Available sequence data and experimental evidence suggest that FeLV-B arises by recombination with endogenous FeLV (22, 24, 26, 41). FeLV-C isolates are rare and occur only in association with FeLV-A or with FeLV-A and FeLV-B (13). FeLV-C isolates are uniquely associated with the development of pure erythrocyte aplasia, which is one of the most acute degenerative retroviral diseases known (20, 25). FeLV-C is thought to arise by mutation from FeLV-A (22), although the prototypic FeLV-C strain FeLV-C/Sarma contains additional sequences, probably derived from endogenous FeLV. Rigby (32) demonstrated by site-directed mutagenesis that the determinants of the superinfection interference phenotype of FeLV-C were located within variable region 1 (VR1) (Fig. 1). The disease-causing and infectivity phenotypes are, however, not completely associated with this small region, and other sequence differences, possibly near VR5, may play a role in the generation of these phenotypes (33).

FIG. 1.

Variable regions of FeLV and recombinant surface glycoproteins. This diagram shows the predicted peptide sequences of the four recombinant proteins produced during this study. Above the peptide sequences is a single line representing the FeLV-A/Glasgow-1 sequence and showing the positions of the variable regions (VR) of FeLV SU (as defined by Rigby [32]) as shaded areas. The unshaded box represents the neutralizing epitope that is recognized by the monoclonal antibodies C11D8 (6, 8) and 3-17 (40a, 47). This epitope is common to all FeLV sequences used in this diagram. The numbers refer to the predicted amino acid residues of the envelope polyprotein precursor of FeLV-A/Glasgow-1 (41). Note that the C-terminal end of the transmembrane protein (the anchoring domain) is absent in the recombinant proteins. L, leader peptide.

The existence of three FeLV subgroups with different host ranges has traditionally been taken as evidence of the presence of three cellular receptors, one for each subgroup (4, 11, 30, 33, 39). Recently, however, some doubt has been cast on this assumption (28). The evidence for three subgroups was originally augmented by neutralization data (39). Subsequently, others found that a degree of cross-neutralization occurred between the subgroups, with several isolates of FeLV-C being indistinguishable from FeLV-A in neutralization assays (36). However, FeLV-C/Sarma can be distinguished from FeLV-A, although not from an isolate of FeLV-B (FeLV-B/ST), by this technique. These results suggested that the neutralization epitope(s) of FeLV SU may be distinct from the subgroup-determining regions.

This paper describes the production of recombinant surface glycoproteins of FeLV by using the baculovirus expression vector system (18). In vitro experiments with these proteins demonstrated that they were able to block virus replication and inhibit the activity of neutralizing antibodies. Using chimeric FeLV-A/C proteins, we demonstrate that the neutralizing epitopes of FeLV are distinct from the subgroup-determining regions.

MATERIALS AND METHODS

Virus strains, plasmid vectors, and cell lines.

The FeLV strains used were FeLV-A/Glasgow 1 (15), FeLV-B/Snyder-Theilen (ST) (39), FeLV-C/Sarma (39), and FeLV-C/FA27C (25). The proviral clones of FeLV-A and FeLV-C, called pFGA and pFSC, respectively, and proviral clones of chimeric A/C FeLVs created by site-directed mutagenesis, called pVC and pV215, were obtained from M. Rigby. Both of these chimeric viruses have an FeLV-A backbone but contain a small region near the 5′ end of an FeLV-C specific sequence which confers many of the properties of FeLV-C (33). The sequences of the two FeLV-C isolates used in the production of these chimeras are very similar in this small 5′ region, differing by only one amino acid residue (isoleucine for methionine in pV215).

Autographa californica multiple nuclear polyhedrosis virus (AcMNPV) was obtained from N. Stow (Institute of Virology, Glasgow, United Kingdom). Viral DNA was obtained as described elsewhere (18). The baculovirus transfer vector pAcLacZ+ was provided by R. Possee (NERC Institute of Virology, Oxford, United Kingdom). This vector contains 5′ and 3′ polyhedrin flanking sequences, between which is inserted a BamHI cloning site and polyhedrin promoter as well as the lacZ gene under the control of the baculovirus p10 promoter. Recombinant viruses produced by using this vector express β-galactosidase. The vector was modified by the insertion of a multiple cloning site containing unique PstI, SmaI and XbaI recognition sequences into the BamHI cloning site of the original vector. The original BamHI cloning site was also preserved, and the resulting vector was called pAcLacMCS.

The QN10S feline cell line, used for the assay of FeLV, was maintained as described elsewhere (12, 27). Infections of QN10S cells with FeLV were performed by using 12-well tissue culture trays containing 22-mm-diameter wells. The cells were subcultured the previous day and seeded at a density of 4 × 10⁴ per well. The virus was inoculated into 1 ml of cell medium containing 4 μg of Polybrene per ml. The cells were incubated at 37°C for 2 h with occasional agitation, and the inoculum was then replaced with fresh medium. Foci of transformed cells were counted at 6 days postinfection.

The F422 (29) and FL74 (17) cell lines were used as sources of purified FeLV for immunological assays. Large quantities of the cells were grown in roller flasks, and the viral particles from the culture fluid were concentrated by ultrafiltration followed by centrifugation in a 20 to 50% sucrose gradient. The preparations of purified virus were lysed with 1% Triton X-100 on ice for 30 min, diluted with an equal volume of Tris-buffered saline (TBS), and stored at −20°C until used.

The Sf9 cell line was used for the production of recombinant baculoviruses and proteins. The cells were maintained without CO₂ supplementation at 28°C in TC100 medium (Gibco) with 2 mM glutamine, 100 IU of penicillin per ml, 100 μg of streptomycin per ml, and 10% fetal calf serum. For the production of recombinant proteins, the cells were maintained in a serum-free insect cell medium (SF900 II; Gibco) after infection with the recombinant virus.

Sera and antibodies.

Rabbit 709 serum was provided by G. Reid. The serum was produced by immunization of a rabbit with FeLV glycoproteins purified by lentil lectin affinity chromatography. The serum reacts well with native FeLV SU in Western blots.

Immune cat sera P11 and P12 were derived from cats that had recovered from natural contact infection with FeLV-A/Glasgow 1 in the experiments of Madewell and Jarrett (21). P11 serum had an FeLV-neutralizing titer of 128 against FeLV-A and reacted specifically with FeLV-A SU, but not with FeLV-B/ST SU or FeLV-C/Sarma SU, in enzyme-linked immunosorbent assays (ELISAs). P12 serum is similar to P11 except that its FeLV-A-neutralizing titer is slightly higher (256). Cat serum 31 was derived from an experimental infection with biologically cloned FeLV-C/Sarma. The neutralizing-antibody titers of this serum were 16 against FeLV-A/Glasgow-1, 16 against FeLV-B/ST, and 1,024 against FeLV-C/Sarma.

The monoclonal antibody 3-17 was donated by K. Weijer (European Veterinary Laboratory). This antibody neutralizes all FeLV subgroups in vitro (45). One of us (N.S.) demonstrated that the antibody recognizes a region covering amino acid residues 204 to 208 of FeLV-A SU, which has been shown to be a neutralizing epitope present in all FeLV subgroups (6). Peptides of this region induce virus-neutralizing antibodies in rabbits but not in cats (23, 46). The antibody was conjugated to alkaline phosphatase (AP) to give antibody 3-17AP.

Immunostaining.

Polyacrylamide gel electrophoresis was performed under reducing conditions on 7.5% or 5 to 20% gradient gels. The proteins were electroblotted onto nylon or nitrocellulose membranes which were then incubated overnight in a blocking solution (TBS plus 3% skim milk powder plus 10% goat serum) and then washed several times in phosphate-buffered saline (PBS) with 0.05% Tween 20 (PBS-Tween). The membrane was incubated for 2 h with fresh blocking solution (to which the serum or monoclonal antibody, 0.2% [vol/vol] 0.5 M EDTA, and 0.5% [vol/vol] Tween 20 had been added). If a second antibody was to be used, then the membrane was briefly washed in PBS-Tween before being incubated with a similarly prepared solution for a further hour. All such antibodies were conjugated with AP. The membrane was then washed several times with PBS-Tween and was finally washed in AP buffer (200 mM NaCl, 10 mM MgCl₂, 100 mM Tris [pH 9.5]). The blots were developed by incubation with a substrate solution containing 1.65 μg of bromochloroindolylphosphate (BCIP) and 3.3 μg of nitroblue tetrazolium (NBT) in 10 ml of AP buffer.

ELISA for FeLV SU.

An ELISA was used for the rapid detection of recombinant SU as described elsewhere (27). Briefly, a microtiter plate was coated with a 1:1,000 dilution of rabbit 709 serum. The plate was then washed twice with TBS plus 0.05% Tween 20 (TBS-Tween) and blocked for 1 h with blocking solution. The antigens to be tested were diluted with 1% Empigen in TBS, added to the plate, and left for 3 h. A further two washes were followed by the addition of a 1:1,000 dilution of 3-17AP in blocking solution. The plate was then washed thoroughly with TBS-Tween before addition of 50 μl of a 1-mg/ml solution of 4-nitrophenyl phosphate in AP buffer to each well and incubation for approximately 4 h. The reaction was stopped with 50 μl of 0.4 M NaOH, and the result was read at 405 nm.

Production of recombinant baculoviruses.

All DNA manipulations were performed as described elsewhere (38). The oligonucleotide primers CCCGGATCCCTGCAGGACCAACCACC and AGGAGGTAATACCCGGTAGTGGTAGTGGTAGTGACTGGGCCCGCG were used to generate PCR fragments of the env genes of the proviral clones pFGA, pV215, and pVC. A similar fragment was amplified from pFSC by using a different 3′ primer (AGGAGGTAGTACCCGGTAGTGGTAGTGGTAGTGACTGGGCCCGCG). These fragments included all of the SU and the TM sequences as far as the ApaI site in FeLV-A env (nucleotide 1838) (41), which code for the first 147 amino acids at the amino terminus, but did not include the anchoring transmembrane region. The fragments had six histidine codons at their carboxy termini that were intended to allow the use of a nickel-nitrilotriacetic acid adsorbent to affinity purify the proteins from the supernatants of infected cells (8). Following BamHI/SmaI digestion, the fragments were all cloned directly into a preparation of BamHI/SmaI-digested pAcLacMCS which had been purified by agarose gel electrophoresis followed by electroelution. The recombinant plasmids were purified on a cesium chloride gradient. To distinguish between the recombinant transfer vectors, each was digested with EcoRI, HindIII, AccI, ClaI, ApaI, BglII, and KpnI. The VR1 regions of the transfer vectors and parent plasmids were also directly sequenced (Sequenase; Stratagene). The results of these experiments were in accordance with predictions based on available sequence data. Figure 1 illustrates the predicted peptide sequences of the recombinant proteins that were produced from these cloned PCR fragments.

The recombinant pAcLacMCS transfer vectors were cotransfected by calcium phosphate precipitation with wild-type baculovirus DNA into Sf9 insect cells (18). Once numerous polyhedrin crystals could be seen, the medium was harvested and centrifuged at 3,500 rpm for 10 min, and the supernatant fluid was stored at 4°C. To confirm the success of the cotransfection, the cells were fixed with 2% formaldehyde and 0.2% glutaraldehyde in PBS for 10 min at 4°C. They were then stained for β-galactosidase activity with 50 mM potassium ferricyanide–50 mM potassium ferrocyanide–20 mM MgCl₂–0.4 mg of X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) per ml in PBS.

The recombinant viruses were purified from the wild-type viruses by a plaque purification technique as described elsewhere (18, 27). To achieve complete purification, four or five rounds of purification were required. Once purified stocks were obtained, large volumes of high-titer stocks were generated by infecting Sf9 cells in early logarithmic growth phase with 0.1 PFU/cell. The recombinant viruses were termed AcAHS, AcCHS, AcVCHS, and Ac215HS. Restriction enzyme digests (HindIII and BglII) of the FeLV VR1 regions obtained as PCR fragments from the recombinant viruses were compared with digests of fragments obtained from the original proviral clones.

Production and purification of recombinant proteins.

Samples of Sf9 cells, previously infected with the recombinant viruses, and their culture media were analyzed by immunoblotting with 3-17AP. Most of the recombinant proteins were released into the supernatant. The optimum time at which to harvest the supernatant was determined by detection of SU by ELISA in aliquots of medium taken from cells infected with several different doses of the recombinant AcCHS virus. The results indicated that 1 to 3 PFU/cell produced optimum protein levels at 67 to 76 h postinfection.

To prepare large quantities of each protein, 10 flasks (225 cm²) were seeded with Sf9 cells at a density of 2 × 10⁷ cells per flask. The next day the cells were incubated for 2 h with an inoculum containing 2 PFU of one of the recombinant baculoviruses per cell. The cells were then washed once with SF900 II serum-free medium and were maintained in SF900 II medium for 67 h. After this period, the supernatant was harvested, clarified by centrifugation at 5,000 rpm, and then concentrated by ultrafiltration with a cellulose triacetate filter with a nominal molecular weight cutoff of 20,000 (Sartorius) in a model 8200 stirred cell (Amicon). The concentrated supernatant was then purified by lentil lectin affinity chromatography. A mixture of 0.5 M methyl-α-glucopyranoside together with 0.5 M methyl-α-mannopyranoside in TBS was used to elute the recombinant protein. The resulting solution was dialyzed extensively against TBS at 4°C. The total quantity of protein in each preparation was assayed by using Bradford’s reagent (38). The proteins were stored in aliquots at −70°C until used. The recombinant SU proteins produced were designated Bgp70-A, -C, -VC, and -215 (Fig. 1).

A control protein solution was prepared from an uninfected Sf9 culture. The culture medium was harvested, concentrated, and purified as described above to form a preparation designated C5L. Similarly, a control solution was prepared from a culture infected with wild-type AcMNPV (2 PFU/cell). The supernatant was concentrated and purified to form a preparation designated C6L. A recombinant feline immunodeficiency virus (FIV) SU protein, Bgp100, was also prepared for use as a control preparation (27).

Infection interference assay.

Infection interference assays were performed to investigate the ability of the recombinant proteins to block FeLV infection of susceptible cells. The ability of detergent-disrupted FeLV particles to interfere with infection of susceptible cells had previously been demonstrated (data not shown). In preliminary experiments it was found that optimum interference by the recombinant FeLV-A surface glycoprotein, Bgp70-A, with FeLV-A infection was achieved when the protein was present both during and after the virus adsorption phase of infection. Mixtures of virus and recombinant protein were made by adding various aliquots of the recombinant (or control) proteins to 1-ml aliquots of fresh medium containing approximately 80 focus-forming units (FFU) of FeLV-A, FeLV-B, or FeLV-C. The culture fluid was removed from QN10S cells seeded the day previously as described above and was replaced with 0.5-ml aliquots of the virus-protein mixtures. Each assay was performed in duplicate. The cells were incubated for 2 h, after which time the mixtures were removed and replaced with 0.5-ml aliquots of culture medium to which had been added recombinant protein alone. The cells were incubated for a further 2 days, and another 1-ml aliquot of fresh medium was added. The cells were maintained for a further 4 days, and the foci on each plate were counted.

Neutralization inhibition assay.

The neutralization inhibition assay measured the ability of recombinant proteins to block the action of virus-neutralizing antibodies. Aliquots of 50 μl of dilutions of recombinant proteins and immune cat serum were mixed in a checkerboard fashion in a round-bottom microtiter tray. FeLV was diluted with medium until a titer of 2.4 × 10³ FFU/ml was obtained. From this diluted stock, further volumes of 50 μl containing 120 FFU were added to the reaction mixtures. The reaction mixtures were then incubated for 2 h at 37°C in a humidified incubator. Three samples of 25 μl were taken from each reaction mixture and were used to infect three wells of QN10S cells. The reaction inocula were incubated with the cells for a further 90 min. The overlying medium was then replaced with fresh medium, and the cells were incubated for a further 6 days, when the foci were counted microscopically. The mean number of foci was calculated from these data.

As these assays were potentially susceptible to a wide range of nonspecific or spurious reactions, a number of controls were included. To control for direct interaction of the virus and the antigen, some reactions were performed without serum. To demonstrate that the immune cat serum contained virus-neutralizing antibodies, other reactions were performed without recombinant protein. To establish that the recombinant proteins were not acting with nonspecific components of the immune cat serum to enhance viral entry, some reactions were performed with specific-pathogen-free (SPF) cat serum. To control for nonspecific cytotoxicity of the antigen, some wells had only the proteins added. Nonspecific inhibition of neutralizing activity by baculovirus or insect cell proteins was controlled by inclusion of reactions with a mixture of the two negative controls (C5L and C6L) in place of the recombinant protein. When one component of the reaction was omitted, fresh medium was used in its place. In addition, each assay was performed at least twice on separate days to reduce the likelihood of spurious reactions.

RESULTS

Recombinant FeLV surface glycoproteins are similar to their wild-type counterparts.

Four baculovirus transfer vectors were produced, containing fragments of FeLV env coding for SU and part of TM of FeLV-A/Glasgow 1, FeLV-C/Sarma, and two VR1 chimeric viruses, FeLV-A/C-FZ215 and FeLV-A/C-Sarma (32). By using electroblotting and subsequent immunostaining with a monoclonal antibody (3-17AP), it was demonstrated that infection of Sf9 cells with the recombinant baculoviruses resulted in the expression of the truncated env genes. The four recombinant proteins (Bgp70-A, -C, -215, and -VC) were found to be exported into the culture supernatant of the infected Sf9 cells. The proteins were concentrated by ultrafiltration and purified by lentil lectin affinity chromatography. The eluates from the lentil lectin columns all contained at least one major impurity, a species of about 65 kDa which was probably the major baculovirus envelope glycoprotein, gp64 (44). The major product in the eluate which reacted with the 3-17AP monoclonal antibody was slightly larger than the SU from F422. The difference in size was probably partly due to the presence of the nonanchoring portion of TM. Deglycosylation of the recombinant protein with PNGase F (Oxford Glycosystems) demonstrated that the proteins were glycosylated but not to the same extent as their native counterparts (27).

Western blotting was performed to analyze the binding of five sera to the recombinant proteins. The results of these analyses are shown in Fig. 2. The monoclonal antibody 3-17AP, the rabbit anti-gp70 serum, and the immune cat serum, but not the SPF cat serum or the normal rabbit serum, recognized all of the recombinant proteins. The pattern of binding to the recombinant proteins observed with the rabbit anti-gp70 serum was similar to that observed with the monoclonal antibody 3-17AP. In contrast, the pattern of binding with the recovered-cat serum was different. Both the 709 rabbit serum and 3-17AP bound to a 60-kDa species as well as to the major 75-kDa species. The 147-amino-acid-residue fragment of TM that was included in the recombinant constructs was calculated to have a mass of 15 kDa. This suggests that the 60-kDa band represented cleavage of the SU and TM portions of the proteins. The P11 serum did not bind to the smaller species and bound only to a component of the broad 75-kDa band. Similar differences were also seen with the native FeLV positive controls, confirming a species-dependent difference in immunological reactivity to both native and recombinant FeLV SU. The results obtained from the immunostained blots demonstrated that the baculovirus expression system had produced Bgp70s that were immunologically similar to the native proteins.

FIG. 2.

Analysis of recombinant FeLV SUs. To analyze the four recombinant proteins and compare their immunoreactivities with those of their wild-type counterparts, several sodium dodecyl sulfate–7.5% polyacrylamide gels were run, and immunoblotting was performed. (a) Nitrocellulose membrane probed with a 1:600 dilution of 3-17AP. (b) Similar membrane probed with rabbit 709 anti-gp70 serum diluted 1:500. The membrane was incubated with a goat anti-rabbit immunoglobulin G–AP conjugate (Bio-Rad). (c) SPF rabbit serum used as a control. (d) Similar membrane probed with a 1:100 dilution of the immune cat serum P11. A goat anti-cat immunoglobulin G–AP conjugate was used to detect binding. (e) SPF cat serum used as a control. Control protein preparations of wild-type-infected and uninfected supernatants (C61 and C5L respectively), as well as two FeLV-positive controls, were included in all blots. Lanes: 1, FL74; 2, F422; 3, C6L; 4, C5L; 5, 6, 7, and 8, recombinant FeLV SUs Bgp70-215, -VC, -C, and -A, respectively; 9, molecular mass markers.

Recombinant FeLV surface glycoproteins interfere with virus infection according to the subgroup identity of their VR1 sequences.

The capacity of Bgp70-A to interfere with the infection of susceptible cells by all three FeLV subgroups was investigated. The recombinant FIV SU Bgp100 was used as a control in this experiment. The results are shown in Fig. 3. A reduction in the number of foci was observed when Bgp70-A was included in the culture fluid during and after FeLV-A infection of QN10S cells, indicating that Bgp70-A interfered with FeLV-A infection. The recombinant FIV SU protein, Bgp100, did not interfere with FeLV-A, FeLV-B, or FeLV-C infections. Bgp70-A did not interfere with FeLV-B or FeLV-C infections, indicating that the interference observed between Bgp70-A and FeLV-A was subgroup specific. In the control reactions a slight increase in the number of foci was observed with increasing volume of recombinant protein added. The rise was constant for the various controls and was not dependent on the FeLV subgroup used. The slight decrease in serum protein concentration, caused by adding the recombinant protein solution to a maximum volume of 50 μl to the original 0.5-ml volume of viral inoculum, might be responsible. Alternatively, a nonspecific interaction between FeLV and baculovirus proteins may account for this observation.

FIG. 3.

Infection interference assay: titration of protein. The following experiment was performed to determine whether Bgp70-A could interfere with viral infection of susceptible cells and whether such interference was subgroup specific. QN10S cells were infected for 2 h with a 0.5-ml aliquot of either virus alone or virus with recombinant protein added. The inoculum was removed and replaced with a 0.5-ml aliquot of either medium alone or medium with recombinant protein added. Two days later, a further 1 ml of medium was added. Each assay was performed in duplicate. The plaques were counted at 6 days postinfection. The average number of plaques was expressed as a percentage of the average of number of plaques counted in controls with no protein. Symbols: ⧫, Bgp70-A plus FeLV-A; ▴, Bgp70-A plus FeLV-B; ▪, Bgp70-A plus FeLV-C; ∗, Bgp100 plus FeLV-A (control); ×, Bgp100 plus FeLV-B (control); +, Bgp100 plus FeLV-C (control).

To extend these observations, another infection interference assay was performed with a larger range of proteins. QN10S cells were infected with FeLV-A inocula to which had been added various concentrations of one of the recombinant proteins, Bgp70-A, Bgp70-C, Bgp70-215, or Bgp100. The C5L preparation was included as an additional control. The results are shown in Fig. 4a. The specific interference with FeLV-A infection by Bgp70-A observed in Fig. 3 was again demonstrated. Neither Bgp100 nor C5L interfered with the infection of susceptible cells. Significantly, Bgp70-C and Bgp70-215 failed to inhibit the infection of the cells by FeLV-A. This result confirmed the subgroup-specific nature of the infection interference caused by the recombinant proteins. The infection of the cells was completely inhibited at the highest dose of Bgp70-A. The assay was repeated with different batches of protein with similar results (Fig. 4b). A mixture of equal proportions of the control preparations, C5L and C6L, was also used in this assay, and a small increase in the number of foci produced by the control reactions was seen again.

FIG. 4.

Infection interference assay: inhibition of FeLV-A and FeLV-C. (a) The following experiment was performed to investigate the ability of the recombinant proteins to interfere with the infection of susceptible cells by FeLV-A/Glasgow-1. QN10S cells, seeded in a 12-well tray the previous day, were incubated with a 0.5-ml viral inoculum containing a diluted recombinant protein for 2 h. After this period, the inoculum was removed and replaced with 0.5 ml of fresh medium containing only the protein. After 2 days, a further 1 ml of fresh medium was added, and the cells were incubated for a further 5 days. Each assay was performed in duplicate. The plaques were counted at 7 days postinfection. The average number of plaques was expressed as a percentage of the average of number of plaques counted in controls with no protein. (b) The experiment was repeated with different batches of protein. (c) Another experiment was performed to investigate the ability of the same recombinant proteins to interfere with the infection of susceptible cells by FeLV-C/FA27C. Symbols: ▾, Bgp70-A; ▪, Bgp70-C; ☼, Bgp70-215; ○, Bgp70-VC; ×, C5L (a) or C5L and C6L (b and c) (control); ⌂, Bgp100 (control).

To determine if FeLV-C infection could be inhibited by the recombinant chimeric FeLV-A/C protein, Bgp70-215, an infection interference assay was performed with FeLV-C/FA27C as the infecting virus and the same range of recombinant proteins used previously. In this assay an attempt was made to keep the volume of the protein inoculum constant by the addition of appropriate volumes of PBS. The results are presented in Fig. 4c. Inhibition of focus formation was observed in those wells that contained either Bgp70-C or the FeLV-A/C chimeric protein, Bgp70-215. This result extended previous observations on the ability of Bgp70-A to interfere with FeLV-A infection by demonstrating that Bgp70-C interfered specifically with FeLV-C. The ability of Bgp70-215 to interfere with FeLV-C, but not FeLV-A, demonstrates the functional similarity between the A/C subgroup-determining VR1s of the recombinant proteins and native FeLVs (33). Interestingly, the slight increase in the number of foci with increasing volumes of control protein preparations identified previously was not observed in this experiment. It may be relevant that decreases in protein concentration increase absorption of FeLV (34).

Recombinant FeLV surface glycoproteins inhibit the action of neutralizing antibodies independently of subgroup.

Neutralization inhibition assays were used to investigate the ability of neutralizing antibodies to bind to the recombinant proteins. Initially, the amount of Bgp70-A required to block the action of neutralizing antibodies was titrated. A significant, specific inhibition of neutralizing activity was demonstrated (Fig. 5). The negative protein control (C5L and C6L) did not block the action of neutralizing antibodies. In the absence of serum, increasing quantities of recombinant protein appeared to cause a decrease in the number of foci, albeit that the number of foci obtained with the serum from the SPF cats was always lower. This result suggested that the protein might be directly inhibiting the replication of the virus, i.e., infection interference. As this particular result was not obtained in other neutralization inhibition assays (see below) and the quantities of protein required for these assays were much smaller than those used in the infection interference assays described above, normal experimental variation was probably responsible for this observation.

FIG. 5.

Neutralization inhibition assay: titration of recombinant protein Bgp70-A. To investigate the ability of the recombinant Bgp70-A protein to specifically block the action of neutralizing serum, dilutions of the P11 immune cat serum were incubated with dilutions of the protein and a fixed dose of virus for 2 h. Aliquots from this reaction mixture were then used to infect QN10S cells. The plaques produced were counted 7 days later. Controls included no protein (None), a mixture of wild-type and uninfected supernatants prepared in a fashion similar to that for Bgp70-A (C5L/C6L), and an SPF serum. The greater the number of plaques, the greater is the inhibition of the neutralizing antibodies.

A second neutralization inhibition assay investigated the ability of the other Bgp70s to block the action of neutralizing antibodies. Several different recombinant protein preparations were used in place of the dilutions of Bgp70-A. The results are presented in Fig. 6a. Inhibition of neutralizing activity was observed with Bgp70-A, Bgp70-VC, and Bgp70-215. No inhibition was observed with Bgp70-C or the negative controls. This assay was repeated with the P12 serum, with similar results (Fig. 6b), implying that the P11 serum was not unusual in its specificity for FeLV-A. As both the P11 and P12 sera were derived from cats that had recovered from natural infections with FeLV, it follows that their specificity for FeLV-A may be common in naturally recovered cats.

FIG. 6.

Neutralization inhibition assay: reaction of recombinant proteins with neutralizing sera P11, P12, and cat 31. (a) To investigate the ability of the recombinant Bgp70 proteins to specifically block the action of a neutralizing serum, dilutions of the P11 immune cat serum were incubated with 2.25 μg of the proteins and a fixed dose of virus for 2 h. Aliquots from these reaction mixtures were then used to infect QN10S cells. The plaques produced were counted 7 days later. Controls included no protein (None), a mixture of wild-type and uninfected supernatants prepared in a fashion similar to that for Bgp70-A (C5L/C6L), and an SPF cat serum. (b) A similar experiment was performed with P12 immune cat serum. (c) The reciprocal experiment of that for panels a and b, that is, determination of the ability of the recombinant Bgp70 proteins to specifically block the action of an FeLV-C-specific neutralizing serum, was also performed. Dilutions of cat 31 serum were incubated with 20 μl of the proteins and a fixed dose of FeLV-C/Sarma for 2 h. Aliquots from these reaction mixtures were then used to infect QN10S cells. The plaques produced were counted 7 days later. Controls included no protein (None), a mixture of wild-type and uninfected supernatants prepared in a fashion similar to that for Bgp70-C (C5L/C6L), a recombinant FIV SU protein (Bgp100), and an SPF serum. (d) Inability of a denatured preparation of recombinant Bgp70-A protein to specifically block the action of an FeLV-A-neutralizing serum. The protein was denatured by heating to 90°C for 10 min in the presence of 6 M urea. A mixture of wild-type and uninfected supernatants (C5L/C6L) was treated in a fashion similar to that for the Bgp70-A preparation. Dilutions of the P11 immune cat serum were incubated with 25-μl aliquots of the protein samples and a fixed dose of FeLV-A for 2 h. Aliquots from these reactions were then used to infect QN10S cells. The plaques produced were counted 7 days later. A no-protein control was included (None). An SPF serum was also used. In all of these experiments, the greater the number of plaques, the greater is the inhibition of the neutralizing antibodies.

A third assay investigated the ability of the recombinant proteins to block the neutralizing activity of cat 31 serum, which was specific for FeLV-C/Sarma. This experiment was performed as described above with the exception that FeLV-C/Sarma was used in place of FeLV-A and cat 31 serum was used in place of cat serum P11 or P12. The results are presented in Fig. 6c. Inhibition of neutralizing activity was observed only with the Bgp70-C protein. Significantly, no inhibition was observed with the chimeric Bgp70-215 protein. This result showed that the neutralizing antibodies present in the cat 31 serum did not recognize the region corresponding to VR1 of FeLV-C in an FeLV-A backbone and suggested either that the conformation of the region is affected in the chimera or that the neutralizing antibodies do not bind to VR1.

To investigate the importance of conformation on the neutralizing epitope(s) of FeLV, a fourth assay was performed. An aliquot of the Bgp70-A protein was denatured by being heated to 90°C for 10 min in the presence of 6 M urea and then was dialyzed overnight against TBS. A control preparation containing equal proportions of C5L and C6L was treated in a similar fashion. Untreated controls were assayed concurrently. The results are presented in Fig. 6d. As in previous experiments, the untreated Bgp70-A protein inhibited neutralization while the control preparations did not. The denatured Bgp70-A preparation did not inhibit neutralization, suggesting that the neutralizing epitope(s) of FeLV-A is conformational.

DISCUSSION

This paper describes the production, purification, and analysis of recombinant FeLV surface glycoproteins. Other workers have expressed the entire FeLV env gene in the baculovirus expression vector system and found that their product was not secreted or processed into gp70 and p15(E) (42). Similarly human immunodeficiency virus, human T-cell leukemia virus type 1, and FIV env gene products are not cleaved when expressed in insect cells (1, 9, 43). However, the proteins in this study were produced in the baculovirus expression system from truncated env genes of FeLV-A, FeLV-C/Sarma, and two chimeric FeLV-A/C chimeras produced by Rigby (32). Truncated env genes were used, as it is known that retrovirus surface glycoproteins are secreted in quantity only if the transmembrane portion of the TM protein is deleted and the amino terminus of TM is left intact (5). As expected, the recombinant proteins were efficiently exported and could be detected in the culture fluid of infected cells. The proteins were purified by lentil lectin affinity chromatography and were demonstrated to be immunologically similar to their native counterparts in ELISAs and immunoblots. The purified recombinant proteins were found to be approximately 75 kDa in size and were aberrantly glycosylated. A minor degree of cleavage of the proteins was observed in immunoblots probed with the 3-17AP or rabbit 709 antibody. It is not obvious why cat serum P11 did not bind to the lower species seen in the other blots.

The recombinant proteins interfered with the infection of susceptible cells in a subgroup-specific manner. A chimeric protein containing VR1 of FeLV-C in an FeLV-A backbone (Bgp70-215) blocked the infection of FeLV-C but not FeLV-A. This result supports data demonstrating that VR1 of FeLV determines the subgroup phenotype of FeLV-A and FeLV-C (33). The existence of three FeLV subgroups has often been suggested to imply that there are three distinct subgroup-specific cell receptors for FeLV (2, 31, 33, 39). The ability of the recombinant glycoproteins to block viral replication may be useful in the identification of the FeLV receptor(s). Preliminary studies on a putative FeLV-A receptor have been performed by others (7).

It has been demonstrated that the neutralizing antibodies induced in cats by FeLV infection are only partially subgroup specific (36). If the FeLV-A/C phenotype is determined mostly by VR1 (33), then it is logical to suggest that VR1 may be involved in the binding of neutralizing antibodies. Neutralization inhibition assays demonstrated that the recombinant protein Bgp70-A, but not Bgp70-C, was able to block the action of neutralizing antibodies to FeLV-A. However, the two FeLV-A/C chimeric proteins, Bgp70-VC and Bgp70-215, were also able to block the neutralizing activities of these sera. Conversely, the chimeric proteins and Bgp70-A were unable to block the virus-neutralizing activity of a serum derived from a cat infected with FeLV-C/Sarma. These results indicated that most of the neutralizing activity of these sera was directed at regions other than that which determined the subgroup phenotype. This provided direct evidence that the subgroup-determining and neutralizing-antibody binding sites of FeLV are distinct (36). Indeed, the subgroup-determining VR1 region appears to play little or no part in the binding of neutralizing antibodies.

The nature of viral epitopes involved in neutralization is known for only some viruses, including the V3 loop of human immunodeficiency virus (16, 40) and a loop of influenza virus hemagglutinin (47). Insofar as generalizations can be made from these examples, it appears that neutralizing epitopes are composed of contiguous amino acids, albeit often arranged in a particular conformation. Other regions of the molecule may play a peripheral role in binding. Our neutralization inhibition data suggest that a conformational epitope is present on FeLV. In the future it may be possible to identify a region or regions that are present in Bgp70-A, but that are altered in Bgp70-C, which may serve as binding sites for neutralizing antibodies. If VR1 is not directly involved in the binding of neutralizing antibodies, then by comparison of sequence data (Fig. 1), it follows that the neutralizing-antibody binding site(s) is likely to be VR4 and/or VR5. Both regions are hydrophilic and possess moderately high surface probabilities when examined by using the PLOTSTRUCTURE program, which is part of the Genetics Computer Group package for the VAX network (3, 10). The production of FeLV-A/C VR4 and VR5 chimeric SU proteins would allow these hypotheses to be tested. Specifically, by mutation of nucleotide residues 811 and 812 (cytidine to thymidine and cytidine to guanosine, respectively) (41), a convenient XbaI site could be introduced. Similarly, alteration of nucleotide residue 1019 (cytidine to guanosine) introduces an XhoI site. None of these mutations would result in a change to the amino acid sequence.

All of our results agree with previously published work on the specificity of neutralizing antibodies in relation to subgroup phenotype (35, 36). By utilizing subgroup chimeras and isolated recombinant SUs, we have confirmed that the regions of SU responsible for subgroup specification are distinct from the regions that are recognized by neutralizing antibodies. The ability to produce recombinant FeLV SUs which will block the neutralizing activity of sera from cats that have recovered from a natural challenge represents a useful tool for further studies directed at identifying those epitopes that are recognized by virus-neutralizing antibodies.