INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

To characterize the cell tropism of measles virus (MV) in molecular terms, the interaction of MV envelope proteins with surface molecules of target cells has been recently analyzed (3, 16, 39). It is well established that vaccine strains of MV and certain wild-type strains adapted to Vero cells use CD46 (membrane cofactor protein) as the major viral receptor (10, 23). The hemagglutinin (H) protein alone binds CD46, and binding is associated with a downregulation of CD46 from the cell surface (24, 33). In contrast, recent evidence suggests that wild-type MVs that have been isolated on human or monkey B-cell lines show either no or extremely weak binding to CD46 and that their infection cannot be inhibited by a number of monoclonal antibodies (MAbs) specific for CD46 (3, 16). This has led to the suggestion that these wild-type MVs may use a cellular receptor other than CD46 which has not yet been identified (3, 16, 39). Most of these strains are lymphotropic, grow poorly on adherent cells such as HeLa, and do not downregulate human CD46 (33). The amino acids responsible for the different phenotypes of wild-type and vaccine strain CD46 modulation have been thoroughly characterized by mutational analysis of the H protein (2, 21).

The role of the MV H protein in the promotion of virus-cell and cell-cell fusions is not well understood. In some fusion studies of related paramyxoviruses, the expression of the fusion (F) protein alone in a cell has been sufficient to induce syncytium formation (1, 15). However, most studies suggest that the H or hemagglutinin-neuraminidase (HN) protein is also required, either (i) to provide effective binding of the two membranes to be fused (presumably through interaction with a specific receptor), (ii) to provide a supporting role for the F protein to allow it to assume the correct conformation to form a fusion pore, or (iii) to provide a combination of the two (reviewed in reference 20). A number of studies have shown that efficient F/H- or F/HN-mediated fusion occurs only when F and H/HN from the same paramyxovirus strain are coexpressed in the same cell, suggesting that a type-specific interaction between the F and H/HN proteins is required for successful fusion (8, 14, 17, 44). Therefore, chimeric protein approaches, i.e., swapping protein sequences between different paramyxoviruses, have been used to map the domains important for the fusion-promoting function of the HN protein (9, 38, 41). The membrane-proximal end of the HN ectodomain was found to be essential. A similar approach has been used to map domains in the F proteins that are important for fusion. Again membrane-proximal domains were found to be involved (5, 40, 42). These contain a cysteine-rich region and a leucine zipper motif which could interact with the domain in the H/HN protein during the fusion event. Indeed, peptides corresponding to the leucine zipper sequence can effectively inhibit F-mediated cell fusion, but only when the peptide has the same sequence as the F protein (43).

It is also not known what role the MV proteins other than F and H play in tropism and attenuation, but from examples in other virus systems such as human parainfluenza virus (36) or poliovirus (27) it seems likely that modifications in transcription, replication, or translation, possibly involving interaction with specific host factors, play a role in attenuation. In support of this postulate, it has recently been shown (37) that a short passage adaptation of a monkey B-cell MV isolate to Vero cells leads to a number of alterations in the P and L genes, and the authors suggest that these changes affect the virus replication in a cell type-specific manner.

In this study we analyzed the syncytium-inducing ability of MV F and H proteins of vaccine and wild-type strains in different combinations in transfected cells. We also constructed a number of viable recombinant vaccine MVs. It was observed that wild-type and vaccine strain glycoproteins can interact functionally. The possession of a wild-type envelope affects the cell entry and cell-to-cell spread of a vaccine strain of MV in adherent cell cultures. However, differences in the cytopathogenicity compared to the wild-type virus infection indicate also that other cellular factors are involved in determining virus cell tropism.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Antibodies, cells, and viruses. The Epstein-Barr virus-negative human lymphoblastoid B-cell line BJAB and the Epstein-Barr virus-transformed marmoset adherent B-lymphocytic cell line B95a were grown in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS). Vero cells (derived from African green monkey kidney) were grown in minimal essential medium supplemented with 10% FCS. HeLa cells were cultured in RPMI 1640-2.5% FCS. All cells were grown at 37°C in 5% CO₂.

Plasmid constructions. Plasmids expressing the Edmonston strain F and H glycoproteins under the control of a cytomegalovirus promoter (pCG-EdF and pCG-EdH) and the empty vector (pCG) were a kind gift of R. Cattaneo, Zürich, Switzerland. To generate plasmids expressing the WTF glycoproteins, RNA was first extracted from WTF-infected BJAB cells by using an RNeasy RNA purification kit (Qiagen, Hilden, Germany). Regions spanning the F and H genes were then reverse transcribed from the viral genomic RNA strand by using Superscript reverse transcriptase (RT) (Gibco BRL, Eggenstein, Germany) and the primers 5'-CCTACAAGCTTGAAACACAAATGTCCCACAAGT-3' (H; binds nucleotides 7123 to 7147 of the MV plus-sense genome, between the F and H open reading frames [ORFs]) and 5'-CATGGAATTCCTCAACACAAGAACTCCACAACC-3' (F; binds nucleotides 4762 to 4787 of the MV plus-sense genome, between M and F ORFs) (EcoRI and HindIII recognition sites are underlined). The RT products were then amplified by PCR with the proofreading Pfu polymerase (Stratagene, Heidelberg, Germany), the same primers, and in addition primers 5'-CAAGGAATTCAGGGTATAAGATCTGGTTGACAG-3' (H; binds nucleotides 9272 to 9247 of the MV plus-sense genome at the beginning of the L ORF) and 5'-CCTACAAGCTTGGGATGGGGGTTATCTTTGT-3' (F;binds nucleotides 7327 to 7305 of the MV plus-sense genome between F and H ORFs). EcoRI-HindIII-digested PCR products were then cloned into EcoRI-HindIII-digested pBluescript II KS() vector to give pBS-WTFBH and pBS-WTFBF. These constructs were then cycle sequenced on both strands with overlapping primers by using an ABI310 sequencer (Perkin-Elmer Applied Biosystems), and this sequence was compared to that of directly sequenced PCR products. The expression plasmid pCG-WTFBH was constructed by the ligation of a PacI-SpeI fragment from pBS-WTFBH to PacI-SpeI-cleaved pCG. Plasmid pCG-WTFBF was constructed in a two-step process to remove the long noncoding region between the M and F genes to produce a plasmid with the same upstream sequences as pCG-EdF. First a 368-bp fragment at the 5' end of the F gene was amplified by PCR with the primers 5'-CGCGGATCCAATGTCCATCATGGGTCTC-3' (F protein start site; binds nucleotides 5444 to 5466 of the MV plus-sense genome) and 5'-ACTACTCCCGCAAATCTCTT-3' (within the F ORF, binds nucleotides 5807 to 5788 of the MV plus-sense genome). The BamHI recognition site is underlined. This fragment was then cleaved with BamHI and at an internal site with HpaI (nucleotide 5502 of the MV plus-sense genome) and ligated with a HpaI-PacI fragment from pBS-WTFBF and BamHI-PacI-cleaved pCG vector.

Cell fusion assays. HeLa or Vero cells were seeded in 35-mm-diameter wells to reach 60 to 80% confluence 1 day after plating. Equal amounts (1.5 µg) of each of the MV glycoprotein-expressing plasmids pCG-EdF, pCG-EdH, pCG-WTFBF, and pCG-WTFBH were cotransfected in different combinations in duplicate by using Superfect transfection reagent (Qiagen) according to the manufacturer's protocol. Following transfection, one of each duplicate sample was held in a medium containing 50 µg of fusion inhibitory peptide (FIP; Sigma Aldrich, Deisenhofen, Germany) per ml to inhibit the formation of syncytia (29) and allow the accurate quantification of surface protein expression by fluorescence-activated cell sorter (FACS) analysis. Syncytium formation was quantified 48 h posttransfection as described previously (41). Briefly, three photographs were taken randomly of each transfected sample. These were then digitized with an Agfa scanner and analyzed with National Institutes of Health Image graphics software. The extent of cell fusion was calculated as the area of cells contained in syncytia as a percentage of the total cell area photographed, which amounted to approximately 25% of the plated cells.

Wild-type glycoprotein function assay. HeLa or Vero cells were transfected with MV glycoproteins and maintained in a medium containing 50 µg of FIP per ml for 48 h. The cells were then harvested in calcium- or magnesium-free phosphate-buffered saline containing 1 mM EDTA and washed in the medium, and 2 × 10⁵ cells were then plated onto a monolayer of B95a cells in a 35-mm-diameter dish. Syncytia could be observed within 6 h of cocultivation.

Recombinant MV rescue. The recombinant MVs were rescued in HeLa cells using the attenuated vaccinia virus expressing T7 polymerase (MVA-T7), essentially as described previously (32). HeLa cells were seeded in 35-mm-diameter wells at 50 to 60% confluency 1 day before transfection. The cells were infected with MVA-T7 in OptiMEM (Gibco BRL) at a multiplicity of infection (MOI) of 0.8 for 1 h. After washing of the cell monolayer, the cells were transfected with 0.5 µg of pEMC-La (encoding the MV polymerase), 1.5 µg of pEMC-P, 1.5 µg of pEMC-N, and 5 µg of each of the MV antigenomic plasmid constructs by using Lipofectin transfection reagent (Gibco BRL). Two days after transfection, syncytia could be seen in the cells transfected with the p(+)MVNSe control. In the case of the recombinant viruses, 1 day posttransfection 2 × 10⁵ BJAB cells were added to the transfected HeLa cells to enable the further replication of successfully rescued virus expressing wild-type glycoproteins. Five days posttransfection a crude virus cell lysate was made by freeze-thawing the cells, and the clarified supernatant was used to infect Vero cells (Ed-tag, MV[WTF F]Ed) or BJAB cells (MV[WTF H]Ed and MV[WTF F/WTF H]Ed) to generate high-titer viral stocks. The identity of the rescued viruses was confirmed by RT-PCR followed by restriction digest analysis and direct cycle sequencing.

One-step viral kinetics. BJAB cells (3 × 10⁵) were infected at an MOI of 3 for 2 h at 37°C. The cells were washed twice with medium and were then plated out at 5 × 10⁴ cells per well of a 96-well plate in 200 µl of medium and further incubated at 37°C. Total samples were collected at different time points and were stored at 80°C. The virus production was then assayed by 50% tissue culture infective dose (TCID₅₀) titration on B95a cells.

Multistep viral kinetics. The growth rates of recombinant and parental viruses were compared in BJAB cells by FACS analysis. BJAB cells (2 × 10⁶) were infected at an MOI of 0.1 for 2 h at 37°C. The cells were washed twice with medium and were then plated out at 4 × 10⁵ cells per well of a 48-well plate in 500 µl of medium and further incubated at 37°C. The cells were harvested at 24-h intervals, and viral N protein expression was measured by FACS analysis.

Nucleotide sequence accession number. The sequence for the WTFB F gene is deposited in the EMBL data bank under accession no. AJ133108.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Syncytium formation following transfection of adherent cells refractive to wild-type MV infection. To investigate the role of the MV glycoproteins in receptor specificity and cell-cell fusion events, expression plasmids encoding the F and H proteins from the WTF wild-type strain and Edmonston vaccine strain were used in transfection studies on Vero and HeLa cells. The WTF plasmids were constructed as detailed in Materials and Methods by RT-PCR with MV-specific primers from WTF-infected BJAB cells. The amino acid sequence differences between the WTF and Edmonston glycoprotein genes are shown in Table 1. Within 8 to 10 h, as soon as the H and F proteins could be detected on the surface of the transfected cells, the first syncytia began to appear in the EdH/EdF double-transfected cells (data not shown). Within 48 h of transfection, almost all Vero (data not shown) and HeLa cells were present in syncytia, many of which had detached from the plate (Fig. 1A). In the case of WTFH/WTFF double-transfected cells, very little evidence of syncytium formation could be seen (Fig. 1B), even 72 h after transfection. In a parallel transfection, the cells were cultured in the presence of 50 µg of FIP per ml to inhibit syncytium formation (29) and to allow the measurement of F and H glycoprotein expression on single transfected cells by immunostaining and flow cytometry (Table 2). All constructs led to a cell surface expression of the glycoproteins on HeLa and Vero cells (Table 2). To test the functionality of the F/H complexes, a proportion of these FIP-treated, transfected cells were cocultivated in the absence of FIP with a monolayer of B95a cells, a monkey B-lymphocytic cell line that is susceptible to wild-type MV infection. Within 8 h, cells transfected with the vaccine or wild-type proteins had both recruited many B95a cells into syncytia, indicating that in both cases a functional complex was formed on the cell surface (Fig. 1G and H; Table 2).

View larger version (292K): [in this window] [in a new window]   FIG. 1.   Cell-cell fusion induced following transfection of HeLa cells with plasmids expressing vaccine or wild-type glycoproteins. HeLa cells were transfected in duplicate wells as detailed in Materials and Methods with EdH plus EdF (A and G), WTFF plus WTFH (B and H), WTFF plus EdH (C and I), EdF plus WTFH (D and J), and pCG vector (E). (F) Vero cells transfected with EdF + WTFH. Transfected HeLa cells (A to E) or transfected Vero cells (F) were photographed 48 h posttransfection. (G to J) A total of 2 × 105 transfected HeLa cells were plated onto a monolayer of B95a cells 48 h posttransfection. Following transfection, the HeLa cells had been treated with 50 µg of FIP per ml, which completely inhibited syncytium formation (data not shown). Cells were photographed 8 h after cocultivation (G to J). Bar in panel F, 100 µm (A to F); bar in panel I, 50 µm (G to J).

Construction and rescue of recombinant MVs containing wild-type glycoproteins. As the heterologous glycoproteins were able to induce cell-cell fusion, we decided to investigate cell tropism further by constructing recombinant MVs containing single or double glycoprotein exchanges between wild-type and vaccine strains. The full-length cDNAs were cloned as shown in Fig. 2 and described in Materials and Methods by using unique sites present in a recombinant MV clone (28, 35).

View larger version (29K): [in this window] [in a new window]   FIG. 2.   Schematic representation of the MV genomic plasmid construct p(+)MVNSe and the wild-type recombinant constructs. The unique restriction sites used for the introduction of the wild-type glycoprotein sequences are indicated. The F and H glycoprotein genes of WTF were cloned by RT-PCR as detailed in Materials and Methods. The WTF sequence is indicated by the shaded boxes.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   Characterization of recombinant MVs with an antibody that distinguishes between Ed and WTF H glycoproteins. BJAB cells were infected with an MOI of 0.1 and 3 days postinfection were labelled for FACS analysis by using, as a control, a standard isotype control antibody or the mouse MAbs specific for the MV-F protein (A504) and MV-H protein (L77 and K53). MAb K53 is unable to bind to the WTFH protein, whereas L77 recognizes both EdH and WTFH efficiently.

Growth of the recombinant viruses in lymphoid cells. Both vaccine and wild-type viruses grow in human and monkey B-cell lines. We therefore chose the adherent monkey B-cell line B95a to titrate the recombinant MVs for subsequent infection comparisons. First, we carried out a one-step growth analysis of the recombinant viruses in a human B-cell line, BJAB. The viruses showed almost identical infection kinetics (Fig. 4A), indicating that the viral RNA replication, production of viral proteins, and assembly of infectious virions are not greatly influenced by changes in the viral glycoproteins. The only exception was the MV(WTF F)Ed virus that produced a consistently higher titer at early time points after infection. Similarly, the viral kinetics over a number of replication cycles were also similar as detected by FACS analysis, although MV(WTF F)Ed again showed the most rapid spread (Fig. 4B). In this particular experiment the high initial rate of infection of BJAB cells with MV(WTF H)Ed was due to the use of a higher MOI (Fig. 4B). As expected, only Ed-tag and MV(WTF F)Ed were able to downregulate CD46 (data not shown).

View larger version (27K): [in this window] [in a new window]   FIG. 4.   Growth of recombinant viruses in human lymphoid cells. (A) BJAB cells were infected at an MOI of 3, and 5 × 104 cells were then incubated at 37°C in 200 µl of medium. At various time points after infection samples were removed and stored at 80°C. Cell lysates were then titrated on B95a cells, and virus titers were determined as log10 TCID50/ml. The means from two experiments are shown. (B) BJAB cells were infected at an MOI of 0.1 and were analyzed at 24-h intervals, beginning at day 0 (0d), following infection for the expression of the MV-N protein by FACS analysis. In this particular experiment a higher MOI for the MV(WTF H)Ed virus (0.3 MOI) was used.

Growth of recombinant virus in cells refractive to wild-type MV. As the viruses expressing WTF H could be titrated only on lymphocytic cell lines, it was decided to compare the infection kinetics of the recombinant MVs, Ed-tag, and WTF by using viruses titrated on B95a cells. Ed-tag and MV(WTF F)Ed have a TCID₅₀ on Vero cells that is 10-fold higher than that on B95a cells (data not shown). Therefore, an inoculum 10-fold larger than those of the other viruses was required to infect B95a, HeLa, and Vero cells at an equal MOI of 0.1 (calculated on B95a cells). Virus propagation was monitored by CPE development, immunohistochemistry, and the presence of cell-associated or cell-free infectious virus. All viruses grew well on B95a cells with similar CPE (Fig. 5A to E) and similar kinetics of viral production (Fig. 6A to E). The exceptions were Ed-tag and MV(WTF F)Ed, which showed a higher initial titer of cell-associated virus, presumably due to the larger virus inoculum (Fig. 6A and E). In HeLa cells, the viruses containing the wild-type H glycoprotein (WTF, MV[WTF H]Ed and MV[WTF F/WTF H]Ed) were fully restricted in their ability to spread by cell-cell fusion and only isolated single cells were positive for viral antigen at 2 days postinfection (Fig. 5O to Q). By 4 days postinfection, the viruses were still highly restricted (Fig. 5S to U), but in all cases a few foci of infection showed a limited spread to neighboring cells, an example of which is shown for MV(WTF H)Ed (Fig. 5T). All three viruses also produced only low titers of infectious virus with identical kinetics not exceeding 10³ TCID₅₀ (Fig. 6L to N). With the Ed-tag control, almost all cells were present in syncytia within 2 days postinfection (Fig. 5N). Due to the efficient cell-cell fusion and the larger virus inoculum, virus production was suboptimal, also reaching only 10³ TCID₅₀ (Fig. 6K). Remarkably, the MV(WTF F)Ed virus showed very good cell-to-cell spread but caused almost no cell-cell fusion (Fig. 5R) and produced large amounts of virus (Fig. 6O).

View larger version (56K): [in this window] [in a new window]   FIG. 5.   Spread and cytopathic effect of the recombinant viruses in cell culture. B95a cells (a to e), Vero cells (f to m), and HeLa cells (n to u) were infected with Ed-tag (a, f, and n), WTF (b, g, k, o, and s), MV(WTF H)Ed (c, h, l, p, and t), MV(WTF F/WTF H)Ed (d, i, m, q, and u), or MV(WTF F)Ed (e, j, and r) at an MOI of 0.1 in suspension and were then plated out on glass coverslips. At 1, 2, and 4 days postinfection (dpi) the cells were fixed and permeabilized, and the expression of the MV-H protein was studied by immunohistochemistry. Magnification, ×80 (all panels).

View larger version (22K): [in this window] [in a new window]   FIG. 6.   Growth and release of recombinant virus in cell culture. B95a cells (A to E), Vero cells (F to J), and HeLa cells (K to O) were infected with Ed-tag (A, F, and K), WTF (B, G, and L), MV(WTF H)Ed (C, H, and M), MV(WTF F/WTF H)Ed (D, I, and N), or MV(WTF F)Ed (E, J, and O) at an MOI of 0.1 in suspension. A total of 105 cells were then plated out in 48-well plates in 0.5 ml of medium, and cell-associated (squares) and cell-free (circles) virus were titrated on B95a cells at various time points postinfection as detailed in Materials and Methods. Virus titers were calculated as log10 TCID50/ml.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The functional interactions of the F and H/HN glycoproteins of the paramyxoviruses have almost always been assayed by transfection assays carried out in tissue cultures (Fig. 1 [8, 26, 40]). In the H/HN protein, the membrane-proximal ectodomain is required for fusion helper function (38, 41). The most important regions that have been identified in the F proteins include a cysteine-rich region in the F1 ectodomain and a leucine zipper motif adjacent to the membrane (40, 42). In addition, a second heptad repeat sequence is also present adjacent to the fusion peptide in F1. It seems likely that the F protein assumes a complex structure, perhaps involving an association of these different elements, while an additional interaction with the homotypic H protein is required for cell fusion to occur. Active virus-cell fusion most probably requires a receptor-triggered conformational change in the F-H or F-HN complex which would lead to the insertion of the hydrophobic fusion peptide into the cell membrane (20, 40). The interaction between the MV F and H proteins appears to be weak, as it can only be demonstrated directly following cross-linking and coimmunoprecipitation (22) or indirectly by cocapping studies (18). Other evidence that supports the notion that F and H must interact for efficient fusion to occur is that antibodies specific for the MV H protein that do not inhibit receptor binding do inhibit fusion (13).

We carried out transfection experiments similar to those described by others with plasmids encoding the WTF and Ed F and H proteins. Using homotypic glycoprotein pairings, these transfection experiments presented in Fig. 1 produced the results expected from our current knowledge of MV wild-type and vaccine strain receptor usage. CD46, the vaccine strain receptor, is present on both HeLa and Vero cells and allows the efficient binding of Edmonston H protein (3). When the association of the H protein with the F protein is good and the cellular receptor is present, then efficient cell-cell fusion occurs (Fig. 1A; EdF/EdH). When the glycoproteins from the lymphotropic wild-type strain WTF were transfected, in the absence of a known high-affinity receptor, no cell-cell fusion was seen (Fig. 1B). In contrast, in the presence of a receptor for WTF on B95a cells efficient cell-cell fusion occurred, proving that the F/H complex was active (Fig. 1H). The use of heterotypic glycoprotein pairings, however, produced contrasting results. Cotransfection of the EdH and WTFF constructs resulted in the formation of large plaques involving all the cells, similar to the homotypic Ed pairing. Therefore, the WTFF protein can functionally replace the EdF protein when paired with a receptor-binding EdH protein. A reproducible slight delay in the kinetics of plaque spread (data not shown) could indicate that the heterotypic F/H complex formation is perhaps slightly suboptimal, although no differences were noted on B95a cell coculture. However, a surprising result was that the substitution of the EdF protein for the WTFF protein could to a limited extent rescue the fusion deficit of the homotypic wild-type complex in HeLa and Vero cells (Fig. 1D and F). Although fewer plaques were present (Table 2), this indicates that the F protein can also play a role in determining cell tropism.

What mechanism could explain the different fusion characteristics of the WTF/Ed heterotypic and homotypic complexes? Previous studies have indicated that efficient fusion events require a very specific interaction between homotypic paramyxovirus glycoproteins. One possible explanation is that the presence of sequence differences between the F and H/HN proteins of different strains does not allow the efficient formation of a functional F/H or F/HN multimer (17, 34). However, in the case of the heterotypic MV complexes, the complex was still active when B95a cells were used as targets for cell fusion. In addition, the sequence differences between Ed and WTF F and H proteins are not in the membrane-proximal domains important for F/H interaction. Another possibility is that the heterotypic proteins interact but that the active insertion of the fusion peptide into the target cell membrane is in some way affected. Using mutated forms of CD46, it has been clearly demonstrated how changing the length of the CD46 molecule can adversely affect cell-cell fusion (4). When the molecule is too long, the fusion peptide is unable to insert itself into the membrane. When it is too short, binding can no longer occur. In the case of WTF cell binding and entry, the receptor could have a morphology quite different from that of CD46, meaning that when this receptor is not present (e.g., HeLa and Vero cells) structural constraints could prevent the active insertion of the fusion peptide. However, when the heterotypic F and H proteins interact, these structural constraints could be overcome, resulting in active fusion both with cells expressing or not expressing CD46. A further possibility would be the requirement for a strain-specific fusion protein receptor or coreceptor, where only one of the two receptors for F or H would be required for cell-cell fusion to occur. Strain-specific neutralizing antigenic sites have been mapped in the F2 region of MV F, indicating that F proteins from different MV strains can be distinguished structurally (12). The only sequence differences between WTFF and EdF are also in the F2 region.

The creation of recombinant MVs based on the Edmonston vaccine strain but containing exchanges in the glycoproteins with the lymphotropic wild-type strain WTF allowed us to assess the role of the other viral gene products and cellular factors in MV cell tropism. After infection of HeLa cells, the data appeared similar to those after transfection. The viruses containing the WTFH protein formed no syncytia (no cell-cell fusion), and only isolated positive cells were seen, while the control Ed-tag infection produced a normal vaccine-type CPE with characteristic extended syncytia. An unexpected result was with the MV(WTF F)Ed recombinant that showed Ed-tag-type spread and infection of all the cells, indicating that virus-cell fusion occurs efficiently, but showed almost no cell-cell fusion. This contradicts the transfection findings when EdH and WTFF could cooperate well to form syncytia in HeLa cells. A recent paper (11) has shown that the surface density of paramyxovirus glycoproteins plays an important role in the speed and extent of cell-cell fusion. While the simian virus 5 F protein surface density was directly related to fusion efficiency, human parainfluenza virus 3 fusion was dependent on the densities of both the F and HN proteins. However, in the case of our recombinant virus MV(WTF F)Ed, infected cells showed much higher mean fluorescence intensities than transfected cells (data not shown), so a low surface density is unlikely to be the factor responsible for a lack of syncytium induction by this recombinant. A more likely explanation for the differences seen following the expression of MV glycoproteins in isolation in cell culture and the use of infectious viruses is the presence of the M protein. M has been shown to play an important role in regulating cell-cell fusion and virus spread via interactions with the cytoplasmic tails of the F and H proteins (6, 7). It appears that M inhibits cell-cell fusion and promotes the assembly of virus particles. However, as the cytoplasmic tails of the Edmonston and WTF glycoproteins are identical in sequence, EdM would not be expected to play a differentiating role in the assembly and budding of the recombinant viruses. Indeed, in B95a cells all recombinant viruses grew similarly (Fig. 5). However, it may be that M inhibits the cell-to-cell membrane fusion by a heterotypic F/H complex with weak fusogenic potential more effectively than the homotypic complexes while not adversely affecting virus-cell fusion. It is also possible that other molecules are present in the cell membrane that play a role in MV cell entry or fusion and that could play a differential role in these processes. Molecules promoting or hindering virus cell entry or fusion have been described for Sendai virus and Newcastle disease virus (19, 25), and it is possible that molecules such as these could be expressed in a cell-specific manner. Cell-specific differences in the membrane lipid composition can also lead to changes in fusogenic activity (31). This could also explain the altered fusion characteristics seen between HeLa and Vero cells after infection with the recombinant viruses (for example, the ability of MV[WTF F]Ed to induce small syncytia in Vero cells but not in HeLa cells [Fig. 5J and R]).

The roles of the glycoproteins and cellular membrane proteins in virus cell entry and cell-cell fusion appear to be very complex and play an important role in determining virus tropism. However, from these studies it is also clear from the rapid spread of MV(WTF F/WTF H)Ed in comparison to WTF in Vero cells that the other MV proteins also play a role in cell tropism. In this case it is most likely that the Ed-tag, Vero-adapted replicative machinery produced such a high viral load of MV(WTF F/WTF H)Ed that this virus (and MV[WTF H]Ed) could enter cells via a low-affinity route. Although WTF could also enter Vero cells via this route, inefficient replication due to its lymphocyte-adapted replication proteins would result in a low-level production of infectious virus. A recent paper (37) has shown that the adaptation of a wild-type virus isolated on B95a cells to Vero cells involves changes in the P and L genes that appear to optimize viral transcription in this cell type. However, after adaptation to Vero cells the virus grew less efficiently in B95a cells than the B95a-isolated virus, with similar titers being reached only when 10-fold more Vero-adapted virus was used as an inoculum. We saw a similar effect when we compared the growth of Ed-tag and MV(WTF F)Ed to WTF in B95a cells, but this reduced titer could be rescued when the WTF H protein was expressed by the virus (e.g., MV[WTF H]Ed), arguing that virus cell entry or cell-to-cell spread of the virus also plays a role in our system. Takeda et al. also reported no functional differences in the glycoproteins from the two viruses (B95a and Vero adapted) when compared in a cotransfection assay with the F glycoprotein in B95a cells. They used a cotransfection system also requiring vaccinia virus coinfection, and it is possible that vaccinia virus structural proteins could have affected the fusion efficiency as seen previously (20). We, however, tested the functions of the WTF and Ed glycoproteins directly in two different cell types (HeLa and Vero) and indirectly in B95a cells and saw differences in all three cases (although the fusion with B95a cells showed the smallest difference). It is also possible that these differing results are due to the different isolation histories of the viruses. Our wild-type virus has been isolated and passaged on a human B-lymphocytic cell line (BJAB) and not a monkey cell line, and this was compared with a vaccine strain virus that has a very varied passage history. Takeda et al. compared two extremely similar viruses differing only by 10 passages in Vero cells. However, in spite of these differences, when we previously carried out a similar experiment and adapted our WTF virus to Vero cells, we also saw functional changes in the H glycoprotein, in that it acquired the ability to bind to CD46 while our BJAB-isolated WTF could not (3). It therefore seems that both changes in the transcriptional machinery and receptor usage are important determinants of MV tropism.

It should now be possible to use these recombinant viruses and WTF virus that has been adapted to Vero cells to map the residues important for Vero cell adaptation and vaccine strain attenuation and to search for wild-type MV attachment receptors and fusion-promoting or -inhibiting proteins.