INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Rabies virus (RV) is a negative-strand RNA virus belonging to the rhabdovirus family, of which vesicular stomatitis virus (VSV) is the prototype. The RV glycoprotein (G) is organized in trimers protruding from the viral envelope (17, 57). Although RV multiplies transiently in muscle cells at the site of a bite and is found late in infection in secretory tissues such as the salivary glands, its growth is essentially restricted to neurons (8). Massive amounts of RV nucleocapsids accumulate in Negri bodies (38). RV inoculation of mice is generally lethal. Virulence is determined by the amino acid present at position 333 of G (15, 49, 55). For example, an R333Q mutant (i.e., whose arginine at position 333 is changed to glutamine) can infect olfactory receptor cells (27) and peripheral sensory and motor neurons (11, 14) but does not propagate to the central nervous system. Entry of motoneurons by the K330N+R333M double mutant is completely blocked, and infection of sensory neurons is very inefficient (11a).

It has been postulated that the nicotinic acetylcholine receptor (nAChR) serves as a receptor for RV (29). RV binds to the neuromuscular junction (6, 7, 28). G has sequence homology to the binding sites of some snake neurotoxins which bind to the  subunit of the muscular nAChR (30); an anti-idiotypic monoclonal antibody (MAb) raised to anti-G MAb recognizing purified nAChR binds also to several brain structures (20), and purified RV binds to the  subunit in an overlay protein binding assay (16). However, it is still unclear whether this interaction is able to mediate virus entry. In addition, RV can infect neurons which do not express nAChR (33); therefore, further molecules must act as a viral receptor(s).

In contrast to RV neuronal tropism in vivo, laboratory-passaged or fixed strains such as CVS, PV, and ERA bind in vitro to every neuronal or nonneuronal cell line tested so far, regardless of its species of origin (21, 31, 45, 48, 50, 59), making it impossible to use strategies successfully designed for cloning the receptor(s) for other viruses. These strategies rely on the existence of cell lines nonpermissive for infection and/or virus binding. In the cases of measles virus and of the transmissible gastroenteritis coronavirus, MAbs directed against cell surface proteins inhibiting virus binding have been isolated (37). Immunoaffinity purification of the viral receptor(s) was performed, and the molecule(s) was further characterized (13, 35, 60). Another strategy used for retroviruses (1, 3) and poliovirus (34), as well as for rhinovirus (19, 53) and echovirus (56), was based on transfection of a nonpermissive cell line with genes from permissive cells.

In this study, we have expressed RV G by using a recombinant baculovirus at high levels on the surfaces of insect cells. We can demonstrate specific binding of G to neuronal cell lines. This binding depends on the presence of the arginine 333 and lysine 330 and, like RV infection in vivo, is specific for neurons, suggesting that these neuronal cell lines express a specific RV receptor(s).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells. Cells from Spodoptera frugiperda (Sf21) were grown in TC100 medium plus 10% fetal bovine serum (FBS) at 28°C.

Viruses and MAbs. The CVS and PV strains of RV as well as antigenic mutant K4-5 (11a) were propagated on BSR cells as previously described (44).

Recombinant baculoviruses. Recombinant baculovirus expressing the CVS RV G was obtained from Préhaud and colleagues (43). To generate recombinant baculoviruses expressing the mutated form of RV G, BSR cells were infected with the following RV antigenic mutants: the R333Q mutant or AvO1, the R333M mutant or RL1, the K330N mutant or RK4, and the K330N+R333M mutant or K4-5 (11a). Total RNA was isolated 20 h later, and the RV G mRNA was amplified by reverse transcription-PCR with oligodeoxynucleotides 5'GCCGGAGATGACCGGCCTTCA and 5'CGGGGGATCCAGACTTAAGGAAAGATGGTTCC hybridizing to the 3' end and to the complementary sequence of the 5' end of the G mRNA, respectively. The amplified DNA fragment was digested by BamHI and inserted into the transfer vector pAcYM1 (32). Recombinant plasmids were isolated, and the sequence for each was confirmed. These recombinant plasmids were used for cotransfection of Sf21 cells with linearized lacZ baculovirus DNA from the Autographa californica nuclear polyhedrosis virus. Recombinant baculoviruses were isolated by two rounds of plaque purification with Sf21 cells. Stocks were then prepared, and the production of G upon infection of insect cells by the recombinant baculoviruses was monitored by Western blotting with specific anti-G MAb 17D2.

Binding assays. (i) With fluorescent Sf21 cells. Monolayers of mammalian cells were grown on poly-L-lysine (Sigma)-treated petri dishes (60-mm diameter) for 2 to 3 days. Noninfected Sf21 cells or Sf21 cells infected with baculoviruses at MOIs of 3 to 5 for 26 to 30 h were collected, spun, and resuspended in DMEM (10⁶ cells/ml); 5 µl of 5' carboxyfluoresceindiacetate (5'-CFD) (Calbiochem) per ml was added to fluorescently label the cells, which were incubated for 20 min at 28°C, diluted with DMEM, spun at 1,000 × g for 3 min, washed once with DMEM, and spun again. The cells were finally resuspended in DMEM plus 5 mM EDTA (1 × 10⁶ to 2 × 10⁶ cells/ml). Two milliliters of the cell suspension was added dropwise to the mammalian cell monolayers and incubated for 20 min at room temperature. Unbound Sf21 cells were removed by three washes with phosphate-buffered saline solution (PBS), pH 7.4. Monolayers were then observed with an Olympus BX40 fluorescent microscope. In the cases of IMR-32, SK-N-SH, and HeLa cells, the cells were grown on untreated petri dishes and the binding was performed in the absence of EDTA.

(ii) With radiolabeled Sf21 cells. In some cases, the assay was performed as described above except that the Sf21 cells were radiolabeled with 20 to 40 µCi of ³⁵S (Express ³⁵S³⁵S, 1,000 Ci/mmol; NEN) per 10⁷ cells for 20 h before use. The cells were treated as described in the preceding paragraph. Before they were used, the radioactivity of an aliquot of the cell suspension was measured in a -scintillation counter (Rackbeta 1211; LKB). After the cells were washed, the bound Sf21 cells were scraped into 1 ml of TD buffer (137 mM NaCl, 25 mM Tris-HCl [pH 7.4], 0.7 mM Na₂HPO₄, 5 mM KCl) plus 10 mM EDTA and the radioactivity of a 200-µl aliquot was measured. The binding efficiency was expressed as the ratio of the number of bound insect cells to the total number of insect cells added to the monolayer.

(iii) On paraformaldehyde-fixed Sf21 cells. Monolayers of 3 × 10⁶ Sf21 cells were infected with recombinant baculoviruses at an MOI of 3. At 30 h postinfection, the cells were fixed with cold 4% paraformaldehyde in PBS (pH 7.4) for 5 min, and the monolayers were washed three times with PBS buffer and then washed once with DMEM.

Surface immunoprecipitation of G expressed by insect cells. Monolayers of 3 × 10⁶ Sf21 cells were infected with different recombinant baculoviruses at an MOI of 3. They were then labeled for 18 h with 50 µCi of [³⁵S]methionine (Express ³⁵S³⁵S, 1,000 Ci/mmol; NEN) in 2.5 ml of a mixture containing four parts methionine-free SF900 II medium (GIBCO BRL) and one part TC100 medium. Twenty-four hours after infection, cell monolayers were washed once with TD buffer and then incubated with 2 µl of site II-specific anti-G MAb (30AA5) in 1 ml of TD buffer for 1 h on ice. Cell monolayers were then washed three times with cold TD buffer to remove unbound antibodies, and the cells were lysed in 1 ml of TD buffer plus 1% CHAPS {3-[(3-cholamidopropyl)-dimethylammonio]-1-propanosulfonate} and protease inhibitors (2 µg each of leupeptin, antipain, pepstatin, and chymostatin per ml and 16 µg of aprotinin per ml) for 30 min on ice, and the nuclei were removed by centrifugation at 12,000 rpm in an Eppendorf centrifuge for 5 min. The supernatants were incubated with 15 µl of protein A-Sepharose (Sigma) for 1 h at 4°C under agitation, and the beads were washed three times with TD buffer plus 1% CHAPS. Total immunoprecipitated proteins were run on a 10% polyacrylamide gel.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

NG108-15 cells bind to Sf21 cells expressing RV G (Gcvs-Sf21 cells). We have chosen a baculovirus expression system because G is expressed at high levels at the cell surface in its native configuration (43). The NG108-15 line is a hybrid of rat glioma C6 cells and mouse neuroblastoma N18 cells and displays motoneuron-like properties (39, 40). We monitored the binding of RV G by using rosetting assays with lepidopteran Sf21 (S. frugiperda) cells expressing RV G of the CVS strain (Gcvs-Sf21 cells) and mouse neuroblastoma NG108-15 cells. As shown in Fig. 1A, insect cells expressing RV G and labeled by the cytoplasmic fluorescent dye 5'-CFD attached to NG108-15 cells. They did not bind to the petri dish in regions without neuronal cells. About 20% of the NG108-15 cells bound to insect cells. This may reflect the heterogeneity of the neuronal cell population. Sf21 cells not expressing G did not bind to NG108-15 cell monolayers (Fig. 1B).

View larger version (141K): [in this window] [in a new window]   FIG. 1.   Binding of 5'-CFD-labeled Gcvs-Sf21 (A) or Sf21 (B) cells to NG108-15 cell monolayers. Bound cells were visualized with a fluorescent microscope. Magnification, ×291.

Quantitative analysis of binding between Sf21 cells and the BSR or NG108-15 cell line. In order to quantitate binding, we infected insect cells with recombinant baculovirus at an MOI of 3 to 5. This MOI does not induce strong cytopathic effects before 36 h; thus, the integrity of the membrane is maintained. As shown in Fig. 2, about 30% of radiolabeled Gcvs-Sf21 cells bound to NG108-15 cell monolayers. Noninfected Sf21 cells or those infected with a recombinant baculovirus expressing -galactosidase did not bind (<1%) to NG108-15 cell monolayers. Expression of RV G at the surfaces of insect cells does not enhance their binding to fibroblastic BSR cells, showing that the binding is specific for neuronal cells.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Binding of noninfected Sf21 (ni-Sf21) or Sf21 cells expressing either RV G (Gcvs-Sf21) or -galactosidase (LacZ-Sf21) to NG108-15 or BSR cell monolayers. A total of 106 radiolabeled insect cells were added dropwise to 2 × 106 NG108-15 cells or 3 × 106 BSR cells, and the percentages of bound cells were determined. Each bar represents the average of two samples.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   Binding of Sf21 () or Gcvs-Sf21 () cells to NG108-15 cell monolayers. Different amounts of radiolabeled insect cells (from 0.2 × 106 to 5 × 106 cells) were added dropwise to 2 × 106 NG108-15 cells. The amount of radioactivity bound to the NG108-15 cells at the end of the assay was measured. Each point represents the average of three determinations.

Preincubation of the insect cells expressing RV G with anti-G MAbs inhibits binding to neuronal cells. To ascertain whether the specificity of binding to NG108-15 cells is mediated by RV G, we examined attachment in the presence of anti-G antibodies. Because anti-G antibodies caused aggregation of Gcvs-Sf21 cells, we were obliged to change our experimental procedure. The binding assay was performed by the addition of dissociated, labeled NG108-15 cells to paraformaldehyde-fixed Sf21 cells. As shown in Fig. 4, under these conditions, 14% of the NG108-15 cells that were added bound to Gcvs-Sf21 cells, whereas 1.5% bound to noninfected Sf21 cells. Preincubation of the insect cell monolayers with different anti-G MAbs decreased the capacity of NG108-15 cells to bind to Gcvs-Sf21 cells by up to 90% (Fig. 4). MAb 8D2, directed against RV nucleoprotein, had no effect on the binding.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Inhibition of binding by MAb treatment. Sf21 cells that were not infected (ni) or Gcvs-Sf21 cells were fixed to culture dishes with paraformaldehyde and treated with different antibodies (100-fold dilution of ascitic fluids). , no MAb. 8D2 is an antinucleoprotein MAb, while 30AA5, 18B5, 40EB1, 41BC2, 17D2, and 8C3 are anti-G MAbs. Radiolabeled NG108-15 cells were collected, dissociated, and added dropwise to the fixed Sf21 cells. After removal of unbound cells, the specific binding was measured as described in Materials and Methods. Each assay was done in triplicate. The standard deviation was less than 0.5%.

Neuronal cell lines of various origins contain specific binding sites for RV G. Gcvs-Sf21 cells were found to bind not only to NG108-15 cells but also to the mouse neuronal cell lines F11, NIE-115, and NS20Y, to Neuro 2a cells, to cells of the human neuronal cell line IMR-32, and to cells of the rat adrenal pheochromocytoma cell line PC-12 (Table 1). No specific binding was found with mouse fibroblastic 3T3 cells, differentiated or nondifferentiated myoblast C2C12 cells, human SK-N-SH cells, or epithelial HeLa and A431 cells. Interestingly, the hamster cell line BSR used to propagate RV in vitro did not mediate the binding of Gcvs-Sf21 cells. The simian CV1 and Cos 7 cells and the canine MDCK-II cell line were also negative for mediation of binding (Table 1).

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Comparison of the binding activities of different neuronal cells for Sf21 () or Gcvs-Sf21 (G) cells. Lepidopteran cells were fixed with 4% paraformaldehyde before use. Radiolabeled neuronal cells were collected, dissociated, and added dropwise to insect cell monolayers. Results were expressed as percentages of cell-associated radioactivity bound to the monolayers. Each value is the average of four determinations (except for the F11 cell line, for which experiments were done only in duplicate).

Binding activity of NG108-15 cells with mutated forms of G. Previous results have demonstrated that mutant viruses with a substitution for arginine 333 in RV G are avirulent and impaired in their capability to enter certain nerve endings, suggesting that this region is critical for binding to a cell surface receptor(s). Therefore, we generated baculovirus recombinants expressing G with various mutations: R333Q, R333M, K330N, and K330N+R333M. The expression of G at the surfaces of Sf21 cells was analyzed 24 h after infection with the different baculovirus recombinants. Mutated G is present in equivalent amounts at the surfaces of insect cells, as determined by surface immunoprecipitation with an MAb raised to antigenic site II (30AA5) (Fig. 6) and by the ability of Gcvs-Sf21 cells to generate syncytia at pHs below 5.9 (data not shown). The insect cells expressing mutated RV G were then used in binding assays on NG108-15 cell monolayers. Figure 7 shows that mutations R333Q, R333M, and K330N had no effect on the binding properties of G for NG108-15 cells. The combined presence of mutations K330N and R333M in G suppressed binding to NG108-15 cells.

View larger version (53K): [in this window] [in a new window]   FIG. 6.   Surface expression of mutated RV G by insect cells. Sf21 cells were infected by different recombinant baculoviruses expressing mutated G. The cells were radiolabeled, and surface immunoprecipitations were performed with MAb 30AA5 as described in Materials and Methods. V, purified radiolabeled RV (PV strain); Gv, G in RV; Gb, RV G in Gcvs-Sf21 cells; ni, noninfected Sf21 cells.

View larger version (31K): [in this window] [in a new window]   FIG. 7.   Binding of Sf21 cells expressing G with different mutations to NG108-15 cell monolayers. The experiment was performed as described for Fig. 2. Each bar represents the average of three determinations. Error bars indicate standard deviations.

Purified RV can compete for Gcvs-Sf21 cell binding sites on the surfaces of NG108-15 cells. To exclude the possibility that binding of Gcvs-Sf21 cells to neuronal cell lines is not due to particular modifications of G in insect cells, we performed a competition assay with purified RV. Figure 8 shows that preincubation of NG108-15 cells with 30 µg of purified RV (CVS strain) reduced the subsequent binding of Gcvs-Sf21 cells to these cells by 40%; no inhibition was observed if the cells were pretreated with the double mutant (K4-5 [K330N+R333M]). Stronger inhibition (65%) was observed with 60 µg of purified RV (PV strain). Similar inhibition was obtained if NG108-15 cells were pretreated with purified RV G rosettes (data not shown).

View larger version (25K): [in this window] [in a new window]   FIG. 8.   Binding of Gcvs-Sf21 cells to NG108-15 cells pretreated with purified RV. Monolayers of NG108-15 cells were incubated with concentrated RV (30 µg of CVS or K4-5 or 60 µg of PV for 2 h at room temperature). The assay was then performed as described for Fig. 2. Each assay was done in triplicate; the standard deviation was less than 0.5%.

Following infection by the CVS strain of RV, NG108-15 cells greatly reduce binding activities for Gcvs-Sf21 cells. Infection by a virus can inhibit superinfection by the same virus. In some cases, infection may result in internalization of its receptor or may lead to saturation of the receptor(s) at the cell surface. In order to examine the effect of preinfection of neuroblastoma cells by RV upon binding activity, we infected NG108-15 cells with the CVS strain of RV for 24 h. Figure 9 shows that RV-infected NG108-15 cells were no longer able to bind to Gcvs-Sf21 cells. However, VSV-infected NG108-15 cells still showed significant binding activity. The 50% reduction in binding to VSV-infected NG108-15 cells is likely due to the severe cytopathic effect of the virus. Following infection by RV, a decrease in the binding of Gcvs-Sf21 cells to PC-12 or Neuro 2a cells was also observed (data not shown).

View larger version (28K): [in this window] [in a new window]   FIG. 9.   Binding of Gcvs-Sf21 cells to RV- or VSV-infected NG108-15 cells. Monolayers of NG108-15 cells were infected by the parental RV strain CVS for 30 h or by VSV for 6 h. The assay was then performed as described for Fig. 2. The assay was performed in triplicate. Error bars indicate standard deviations.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Here we were able to demonstrate that RV G specifically binds to neuronal cells. We did so by development of a rosetting assay based on expression of RV G by a recombinant baculovirus on the surfaces of insect cells.

Neuronal cell lines from mice (NG108-15, F11, NIE-115, NS20Y, and Neuro 2a), humans (IMR-32), and rats (PC-12) contain specific attachment sites for RV G, whereas no receptor sites are found on epithelial (HeLa and A431), fibroblast (3T3 and BSR), glioma (C6), or myoblast (C2C12) cells (Table 1). Treatment of Gcvs-Sf21 cells with several anti-G MAbs recognizing different epitopes abolishes binding to neuronal cell lines by steric interference. Insect cells expressing mutated G carrying the double substitution K330N+R333M bind very inefficiently to NG108-15 cells.

These results are consistent with the neurotropism of RV in vivo. Several categories of neurons of mammals having different neurotransmitters are permissive for RV. Both cholinergic (NG108-15, F11, NS20Y, and IMR-32) and adrenergic (NIE-115) neuroblastoma cells express binding sites for Gcvs-Sf21 cells, suggesting that there is no relation between neurotransmitter secretion and expression of RV-specific binding sites.

Single mutations (R333Q, R333M, or K330N) in G neither abolish the binding of Gcvs-Sf21 cells to neuronal cell lines nor impair RV infection of sensory and motoneurons after intramuscular inoculation. On the other hand, doubly mutated G (K330N+R333Q) greatly reduces the affinity of Gcvs-Sf21 cells for NG108-15 cells and also blocks infection of sensory and motoneurons (11a). The neuronal cell line NG108-15 is a derivative of a neural crest tumor and has motoneuron-like properties after differentiation (39, 40).

The RV CVS strain and its avirulent mutants can infect fibroblast cells in vitro (4a, 14, 48, 50, 54a, 59), in contrast to the pronounced neurospecificity of the G interaction observed here and in RV-infected animals. The existence of two types of receptors for RV, as postulated by others (20), may explain this apparent contradiction: a high-affinity receptor might be expressed only by neuronal cells, enabling infection of neurons in an animal (11, 14, 27) and allowing for rosette formation between Gcvs-Sf21 cells and neuroblastoma cells. We can exclude the possibility that nAChR is a high-affinity receptor capable of mediating Gcvs-Sf21 cell binding, since saturation of nAChR in PC-12 cells (20) by an MAb raised to the  subunit of the nAChR does not abrogate binding of PC-12 cells to Gcvs-Sf21 cells (data not shown). Further receptors, presumably of lower affinities, could be expressed by a variety of cells of neuronal and nonneuronal origins (50, 59). These ubiquitous low-affinity receptors could be present at high densities on the cell surface and could include the saturable binding sites on BHK cells (59). Phospholipids (51), glycolipids (10), and gangliosides (9, 52) as well as proteins (5) have been reported to serve as receptors for RV. So the nature of the high-affinity receptor on neurons remains elusive, in large part due to the difficulty in generating an in vitro assay system whose results are not made equivocal by the presence of low-affinity binding sites.

Gcvs-Sf21 cells apparently do not bind to the low-affinity receptors present on nonneuronal cells. Differences in protein modification between G's inserted into the viral and insect cell membranes may account for this observation. G expressed by lepidopteran cells lacks sialic acid and has less complex sugar chains than G present in the viral membrane (23, 43, 54), as illustrated by the difference in their migrations in a polyacrylamide gel (Fig. 6). Alternatively, the density of G at the insect cell surface might be insufficient to mediate binding to low-affinity receptors. Binding of Gcvs-Sf21 cells to neuronal high-affinity receptors is not due to particular modifications of G in insect cells; purified RV (PV or CVS strain but not the double mutant K4-5) is able to bind to these receptors and to compete with Gcvs-Sf21 cell binding.

NG108-15 cells infected by the parental RV CVS strain do not bind to Gcvs-Sf21 cells, suggesting that after infection, the RV receptor is no longer available at the surface. This is compatible with observations for other enveloped viruses which have developed mechanisms to prevent binding of released viruses to infected cells. In the case of influenza virus infection, the neuraminidase cleaves the sialic acid at the cell surface, thus preventing docking of the virus on the infected cell (24, 41). Measles virus and human immunodeficiency virus type 1 both down-regulate their receptors (CD46 and CD4, respectively). CD4 specifically interacts with gp120 inside the rough endoplasmic reticulum (12), inhibiting further its transport (22) and leading to its degradation (58). A different mechanism is responsible for CD46 disappearance (36) at the cell surface. Contact between cells expressing hemagglutinin and cells bearing CD46 is sufficient to trigger CD46 down-regulation in the absence of viral infection (25, 46). In the case of RV, sequestration of the receptor in infected NG108-15 cells could be responsible for the disappearance of binding activity, supporting our hypothesis that Gcvs-Sf21 cell binding sites at the surfaces of neuronal cells are indeed the specific RV receptors.

We believe that there is a broad applicability of this approach to the identification of binding sites for viral G. RV is one of many viruses for which specific binding sites have remained elusive. Some viruses, such as hepadnavirus, do not replicate in culture cells; others, such as Ebola virus, are too dangerous. Some, such as herpesviruses, have several external proteins which make it difficult to determine which are responsible for neuron infection. Viral G, normally membrane-anchored proteins, can be difficult to purify. The Sf21 cells do not self-aggregate or bind to any cell line tested (Fig. 1 to 3 and Table 1) and can be used for analyzing virus-cell interactions.