INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Among the morbilliviruses, measles virus (MV) is associated with an intermediate capacity to cause neurological complications. These include the acute postinfectious measles encephalitis, which develops 2 to 4 weeks after infection, or the late complications, measles inclusion body encephalitis in immunocompromised patients and subacute sclerosing panencephalitis (SSPE), which develops months to years after the initial infection, based on a persistent MV infection (reviewed in reference 3). In late stages of SSPE, massive amounts of MV antigen can be detected in inclusion bodies in various neural cell types (1). SSPE is characterized by a restriction of the viral envelope protein expression as a consequence of mutational, transcriptional, and translational alterations (1, 5). An additional constraint is exerted by the high concentration of antiviral antibodies present in the cerebrospinal fluid of SSPE patients. Tissue culture experiments demonstrated that virus-neutralizing antibodies downregulate not only viral gene expression but also transcription and can completely suppress viral replication (2, 39). Similar results have been obtained in vivo using Lewis rats (22, 38).

Suckling rodents have successfully been used as animal models (predominantly mice and rats) for different forms of MV-induced encephalitis (21, 23, 38). Transgenic mice which express CD46, one of the MV receptors (7, 27), have also been used to induce MV-induced encephalitis (15, 26, 31). However, for development of the acute encephalitis following infection of suckling rats with the rodent-adapted MV strain CAM/RB, or mice with the HNT (hamster neurotropic) strain, the transgenic expression of receptors such as CD46 appears not to be necessary (23, 24, 32, 35). After intracerebral infection with CAM/RB (RB indicates passage in rat brain), 1- to 14-day-old Lewis rats develop a lethal acute measles encephalitis, whereas older animals develop a subacute measles encephalitis (23).

Antiviral antibodies may lead to a restriction of the viral gene expression but also to the selection of escape variants. When monoclonal antibodies (MAbs) are used experimentally to select escape variants, resulting viruses with altered hemagglutinin (H) protein structures might induce differential pathogenicity in animals. This was observed with escape variants selected in the presence of the MAbs L77, Nc32, K71, and K29 recognizing four different epitopes on MV H (20). Variant CAM/RB viruses escaping the MAbs L77 and Nc32 were neurovirulent, whereas viruses escaping the MAbs K29 and K71 appeared to have lost neurovirulence. The H genes of these viruses have been sequenced elsewhere (20). However, because of the number of amino acid changes in this gene and the possibility that changes in other genes also affect the specific phenotype, the molecular basis of the antibody escape and neurovirulence could not be unequivocally determined in earlier experiments. The generation of recombinant MVs has opened the way to make definitive linkages between mutations introduced experimentally into the viral genome and specific phenotypes (30). We therefore assessed, using recombinant MVs, the influence of directed mutations in the H gene on antibody escape and neurovirulence. After intracerebral injection into suckling C57BL/6 mice, a recombinant virus, expressing the H gene of CAM/RB (EdtagCAMH), induced neurological disease, and MV antigen was found in neurons and neuronal processes of the hippocampus, frontal and olfactory cortices, and neostriatum (9). However, the neurovirulence of EdtagCAMH was partially reduced compared to that of the natural strain CAM/RB. Thus, the results indicated that the H protein, albeit an important determinant of neurovirulence, is not the sole determinant and that other viral genes contribute to the observed virus-induced central nervous system disease.

In this study, we generated and tested recombinant viruses expressing single and combined mutations putatively mediating escape from the anti-H MAbs, which were suggested to be associated with neurovirulence (20). We successfully proved the role of mutations in mediating escape from four anti-H MAbs but found, surprisingly, that antibody escape and neurovirulence in suckling Lewis rats are associated with different alterations in the H protein and that phenotypes are not linked.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Antibodies, cells, and propagation of viruses. The MAbs L77, Nc32, K71, and K29 (anti-MV H [20]); anti-MV fusion (F) protein (A504); and anti-MV nucleocapsid protein (F227) were produced from hybridomas using RPMI medium containing 10% FCS and purified over protein G-Sepharose in our laboratory. The fluorescein isothiocyanate (FITC)-conjugated and phycoerythrin (PE)-conjugated rabbit anti-mouse immunoglobulin antibodies and streptavidin-FITC were purchased from DAKO GmbH, Hamburg, Germany. Vero cells were cultured in minimum essential medium containing 5% FCS, and BJAB cells were cultured in RPMI medium containing 10% FCS. CAM/RB, the rodent-adapted CAM strain of MV, was passaged by intracerebral infection of brains of 1-day-old rats and reisolation of virus after 4 days postinfection (p.i.). MV strain Edmonston; Edtag-based recombinants; CAM/Vero (Vero cell-passaged CAM strain); and tissue culture-selected escape mutants CAM/L77, CAM/Nc32, CAM/K71, and CAM/K29 escaping the MAbs L77, Nc32, K71, and K29, respectively, were propagated on Vero cells. For virus production, cells were infected at a multiplicity of infection (MOI) of 0.01, and virus was harvested when maximum giant cell formation was observed by one cycle of freezing-thawing and two centrifugations to pellet cell debris. Supernatants were stored at 80°C.

Construction and rescue of recombinant viruses. All full-length, mutagenized constructs were assembled in an H gene insertion vector (pMVins-H2), the construction of which has been previously described (8a, 9). Briefly, this vector contains two unique restriction sites (PacI and AatII) which permit the directional cloning of complete H genes obtained by PCR amplification using the H-specific, PacI- or Aat II-containing oligonucleotides uniH+ and uniH2. We have used this vector to generate plasmid p(+)MVCAMH, which contains the H gene of CAM/RB in the Edmonston background. The p(+)MVCAMH plasmid was used as the template to construct a set of full-length MV plasmids containing the nucleotide changes specific to the MAb escape mutations previously identified (20). Overlapping oligonucleotides (24-mers) were synthesized with the appropriate point mutations. Two separate PCRs were performed using the proofreading DNA polymerase Pwo (Boehringer). The 5' portion of the H gene was amplified using uniH+ and the mutagenic (antigenome-sense) oligonucleotide. The 3' portion of the gene was amplified using the mutagenic (genome-sense) oligonucleotide and uniH2. PCR products were gel purified from 1% agarose blocks using a dialysis membrane, and the DNA was phenol extracted and ethanol precipitated. The DNA fragments were mixed in equimolar amounts, and 100 ng of the mixture was added to a standard PCR in the absence of primers. The thermocycler was paused at the beginning of cycle 11 (94°C), during which time primers uniH+ and uniH2 were added. Full-length mutagenized H genes were amplified during the remaining 25 cycles. Following restriction with PacI and AatII, the H genes were inserted into similarly cleaved pMVins-H2 to generate the full-length MV constructs. Alternatively, full-length H genes were amplified by reverse transcription-PCR (RT-PCR) from total RNA isolated from infected Vero cells as described previously (9). Plasmids were sequenced by dideoxynucleotide chain termination (ABI Prism) using MV H-specific primers to verify that the mutagenesis was successful. This also confirmed that the PCR had not introduced nonspecific mutations into the H genes. Recombinant viruses were rescued from these constructs following Lipofectin-mediated transfection of MVA-T7-infected HeLa cells as previously described (9).

Sequencing of the H genes from recombinant viruses. To determine whether the nucleotide exchanges which were introduced in the recombinants are correct and whether there are additional exchanges in the H genes of the escape mutants, the H genes were sequenced. Total RNA from infected BJAB cells (48 h) was prepared with Qiagen Spin columns as recommended by the manufacturer. The RNA was reverse transcribed with SuperScript II (GIBCO/BRL). The H gene of the virus was amplified by PCR using the primers H62 (upper) (5'-ATCCACAATGTCACCACAACG-3') and H26 (lower) (5'-TATGCCTGATGTCTGGGTGA-3'). The amplified cDNAs of the H genes were used as templates in the sequencing reaction (ABI-PRISM Bigdye Terminator Cycle Sequencing Ready Reaction kit; Applied Biosystems) using further H-specific primers (H30, 5'-GATAGGGAGTACGACTTCAG-3'; H31, 5'-TTGAAGTAGGTGTTATCAGAA-3'; H32, 5'-CACCATTGAAGGATAACAGGA-3'; H33, 5'-AGGTGGATGGTGATGTCAAA-3'; H25, 5'-GAAGTATCGTAGGTTGCCA-3'; H24K, 5'-CCACTCGGGATTCTCGCAGAG-3'; H23K, 5'-GGATTTCTGATAACACCTA-3'; H20, 5'-CCTTGACCTGATGCTCGATTG-3'; and H27, 5'-TGACATCATGTGATTGGTTC-3'). The analysis was performed using an ABI PRISM 310 Genetic Analyzer (Perkin-Elmer).

Antibody escape assays. To quantify the antibody escape by flow cytometry, BJAB cells were infected with the natural variants or recombinant viruses and analyzed by double staining with the various MAbs to H and a MAb to the fusion protein (F) as an internal control of expression of the envelope proteins. BJAB cells were infected at an MOI of 1 and incubated for 2 to 3 days at 37°C. The cells were washed with fluorescence-activated cell scanner (FACS) buffer (calcium- and magnesium-free phosphate-buffered saline containing 0.4% bovine serum albumin and 0.01 M NaN₃) and incubated for 1 h with the primary antibodies recognizing H (K29, K71, Nc32, and L77), washed twice and treated for 1 h with secondary antibody (PE-conjugated anti-mouse antibody), washed, blocked for 10 min with mouse serum and then treated for 1 h with the biotinylated anti-F MAb A504, washed, and incubated for 1 h with streptavidin-FITC. The fluorescence signals were determined by flow cytometry using a FACScan (Becton Dickinson) cell sorter. The F-specific signals of the cells infected with various virus variants were used as standards for the expression of the viral envelope proteins and equalized for comparison of the H-specific signals obtained with different MAbs. The H-specific signals of EdtagCAMH-infected cells for each H antibody were set to 100%, and the signals of cells infected with virus variants were then expressed in relation to this.

Animal infections and histology. Timed pregnant Lewis rats were purchased from Harlan-Winkelmann (Borchem, Germany). One- to two-day-old pups were infected intracerebrally in the left hemisphere with 20 µl of virus suspension (1 × 10⁴ to 5 × 10⁴ PFU). Body weights of animals were measured at days 0, 3, 5, 7, and 10 p.i. Animals were infected for no longer than 10 days, anesthetized, and sacrificed by decapitation. The brains were removed and fixed in 4% paraformaldehyde in 0.2 M phosphate buffer (pH 7.4) for at least 2 days before frontal brain sections were embedded in paraffin. Tissue sections were routinely stained for hematoxylin-eosin and Luxol-fast blue. In brains infected with a neurovirulent recombinant, cortical and infracortical necrosis occurred in the left hemisphere and frequently also in the right hemisphere, revealing numerous pycnotic nuclei and a perifocal edema.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning and rescue of recombinant MVs expressing H antibody escape mutations. In order to study the potential linkage between antibody escape and neurovirulence, a set of recombinant viruses (summarized in Table 1) was generated based on EdtagCAMH (9) and the sequences of MAb escape mutants selected from CAM/RB (20). Mutations were introduced into the CAM/RB H gene by overlap PCR using mutagenic primers. All mutagenized H genes were inserted into the plasmid vector (pMVins-H2) to generate full-length recombinant constructs (Fig. 1). These were transfected into HeLa cells using the host-range-adapted helper virus MVA-T7 system (9). Recombinant viruses were rescued 6 to 10 days posttransfection. Total RNA was isolated from virus-infected cells, and H genes were amplified by RT-PCR. The mutagenized portions were sequenced to ensure that the mutations were stably retained in the recombinant viruses, and no spurious mutations were identified. Recovery of the recombinant viruses was reproducible in three independent transfections, they all grew to equivalent titers (1 × 10⁵ to 5 × 10⁵ PFU/ml/10⁶ cells) at similar rates, and no differences in plaque size were observed for Vero cells.

View larger version (45K): [in this window] [in a new window]   FIG. 1.   Location of antibody-selected escape mutations in the H protein of CAM/RB and construction of recombinant viruses. The H genes of the rodent-adapted MC CAM/RB (a) escape mutants were sequenced, and the amino acid alterations were identified (b). Plasmid p(+)MVCAMH, which contains the wild-type CAM/RB H gene (c), was used as the template to introduce back mutations into the gene by overlap PCR. The 3' fragment of the gene was amplified using uniH2 and the required mutagenic primer (d). The 5' fragment was amplified using uniH+ and the complementary mutagenic primer (e). The PCR products were joined by overlap PCR (f) to generate a full-length H gene which was restricted with PacI and AatII and inserted into the insertion vector p(+)MVins-H2 (g). Steps d and e were omitted when full-length H genes were amplified directly from total RNA by RT-PCR.

Determination of antibody escape. Immunofluorescence was used to examine H protein expression in Vero cells infected with the recombinant viruses using the four anti-H MAbs L77, Nc32, K71, and K29. Representative micrographs are shown in Fig. 2. Recombinant viruses 1 and 2 were still recognized by L77 (Fig. 2B and D); however, introduction of the double mutation in recombinant 3 ablated antibody binding (Fig. 2F). These observations were identical when EdtagCAMH mutants 7 to 9 were examined (data not shown). Recombinants 4 and 10 no longer bound MAb Nc32 (Fig. 2H and data not shown), and recombinants 12 and 13 no longer bound MAbs K71 and K29, respectively (Fig. 2J and L).

View larger version (123K): [in this window] [in a new window]   FIG. 2.   Analysis of recombinant escape mutants by confocal microscopy. Vero cells were infected with the parental and recombinant viruses at an MOI of 0.01. Cells were fixed and examined by confocal laser scanning microscopy for immunoreactivity, and micrographs represent 8- to 10-µm-deep composite optical sections. The viral H proteins were detected with different MAbs and secondary Cy3-conjugated antibodies (green), and nuclei were counterstained with propidium iodide (red); all images were obtained in double excitation mode. EdtagCAMH mutants were mutant 1 (K71 [A] and L77 [B]), recombinant 2 (K71 [C] and L77 [D]), recombinant 3 (K71 [E] and L77 [F]), recombinant 4 (L77 [G] and Nc32 [H], recombinant 12 (Nc32 [I] and K71 [J]), and recombinant 13 (K71 [K] and K29 [L]). Nonselecting control antibodies were used to demonstrate H protein expression in the recombinant viruses (A, C, E, G, I, and K), and the selecting antibodies were used to determine if the mutated H proteins were no longer bound (B, D, F, H, J, and L). Expression of the H gene in the parental viruses was examined as controls: Edtag (Nc32 [M]), CAM/RB (Nc32 [N]), EdtagCAMH (Nc32 [O]), and EdtagCAMH/Vero (L77 [P]). Magnification, ×160.

View larger version (59K): [in this window] [in a new window]   FIG. 3.   FACS analysis of natural escape mutants. The relative expression of the H protein on the surface of BJAB cells infected with CAM/RB and the natural escape variants CAM/Nc32, CAM/K71, CAM/L77, and CAM/K29 was determined by double staining and flow cytometric analysis. The relative signals obtained after staining of the cells with MAbs L77, Nc32, K71, and K29 are shown. Signals detected by anti-F antibody A504 were taken as standards for the glycoprotein expression, and signals obtained by the four anti-H antibodies were set in relation to the signals of CAM-infected cells (=100%). The MAbs do not recognize cells infected with their respective escape variants.

View larger version (41K): [in this window] [in a new window]   FIG. 4.   FACS analysis of recombinant escape mutants. The expression of the H and F proteins on the surface of BJAB cells infected with the EdtagCAMH mutants 1 to 13 (bars 1 to 13, respectively) and EdtagCAMH (bar 14) was determined by double staining with anti-H (L77, Nc32, K71, and K29) and anti-F MAbs and flow cytometric analysis (Fig. 2). The signal intensity of EdtagCAMH-infected cells was set to 100% (bar 14), and the intensities of the other H signals are presented in relation to this signal.

Analysis of the neurovirulence of recombinants. In order to analyze the neurovirulence of the recombinants, we infected 1- to 2-day-old suckling Lewis rats intracerebrally with approximately 2 × 10⁴ PFU of the viruses. As a disease parameter, we measured the body weight of the infected animals, which closely reflects the degree of infection. Infection with Edtag did not induce a detectable disease in the animals, whereas infection with EdtagCAMH caused strong signs of disease leading to death within approximately 5 days (Fig. 5A). Thus, as found with mice (9), replacement of Edmonston H by the CAMH gene alone is sufficient to cause neurovirulence in suckling rats. The EdtagCAMH mutants 1 to 5 caused a disease similarly as strong as that caused by EdtagCAMH, leading to death between days 5 and 7 p.i. In contrast, recombinants 6 to 13 were not neurovirulent, and animals survived and were apparently healthy (Fig. 5A). Thus, unexpectedly, the mutations mediating the antibody escape did not cause differential neurovirulence, whereas the presence of amino acids G and S at positions 195 and 200, respectively, in EdtagCAMH and the recombinants 1 to 5 correlated with the induction of the lethal brain disease.

View larger version (23K): [in this window] [in a new window]   FIG. 5.   Analysis of the body weight of cerebrally infected rats. One- to two-day-old pups were infected intracerebrally in the left hemisphere with 20 µl of virus suspension. Body weights of animals were measured at days 0, 3, 5, 7, and 10 p.i. Data for mock infection (medium) and infection with the recombinants Edtag, EdtagCAMH, and mutants 1 to 13 (approximately 2 × 104 PFU) are shown in panel A. The body weight development of rodent-adapted CAM/RB (2 × 103 PFU)- and cell culture-adapted CAM/Vero (2 × 104 PFU)-infected suckling rats is shown in panel B. The animals infected with the neurovirulent virus CAM/RB and the recombinants EdtagCAMH and mutants 1 to 5 died between days 5 and 7. dpi, day postinfection.

View larger version (80K): [in this window] [in a new window]   FIG. 6.   Immunohistological analysis of brain infections with a neurovirulent and a nonneurovirulent MV recombinant. Staining patterns of MV nucleocapsids in frontal brain sections of rats 5 days p.i. with EdtagCAMH (A and C) and at day 10 p.i. with recombinant 6 (B and D) are shown. Schematic drawings of immunolabeled brain sections (A and B) demonstrate immunoreactivity with MAb F227 against MV N for EdtagCAMH (C) and recombinant 6 (D) as seen with lower magnification. Higher magnifications prove that the majority of neurons contain MV nucleocapsids in a brain infected with EdtagCAMH (C), whereas only a few scattered neurons are immunolabeled in a brain infected with recombinant 6 (D) (bar = 50 µm).

Analysis of neurovirulence and H gene sequence of Vero cell-adapted CAM. The sequence analysis of the H genes of the natural escape variants detected changes at amino acid positions 195 and 200 in all variants, although they varied in their neurovirulent capacity and therefore this mutation was not suggested to be responsible for the neurovirulent capacity of the variants (20). This is in contrast to our findings with the MV recombinants, where mutations at these positions determined neurovirulence. We therefore compared the neurovirulence and the H gene sequences of CAM/Vero, a CAM strain adapted by at least 30 passages on Vero cells, with CAM/RB propagated in rat brain. CAM/Vero was not neurovirulent for suckling rats (Fig. 5B), and its H protein is recognized by all four anti-H MAbs, L77, Nc32, K71, and K29 (data not shown). In CAM/Vero, we found only two nucleotide changes in the H gene, at positions 603 (GA) and 619 (GA), resulting in amino acid exchanges at positions 195 (GR) and 200 (SN), as found in the natural escape variants and which were introduced in the recombinants 6 to 13 (Table 1). The CAM/Vero H gene sequence is thus identical to that of recombinant 6. This confirmed the results using the recombinant viruses 6 to 13 and demonstrated that amino acids glycine (G) at position 195 and serine (S) at position 200 are decisive for the neurovirulence of the CAM/RB and the recombinant MVs. When we compared the sequences of all known MV H proteins at positions 195 and 200, we found that the neurovirulent rodent-adapted strain CAM/RB is unique in these positions, expressing G and S, respectively.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Interaction of the viral envelope proteins with cellular receptors is not only a prerequisite for viral entry but also has great impact on the pathogenesis of virus-induced diseases (37). Single amino acid changes in the envelope proteins of several viruses, including mumps virus (18) and MV (20), were found to have a profound and differential influence on pathogenesis. Alteration of viral envelope proteins may have a variety of pathogenic consequences: single mutations affect neuroinvasiveness of tick-borne encephalitis virus (14, 33, 41), neurovirulence of dengue virus (13) and Japanese encephalitis virus (28), viral persistence in the case of lymphocytic choriomeningitis virus (25), cell-to-cell spread of rabies virus (6, 40), and virus replication efficiency of Sindbis virus (8). Using recombinant MV with a different H protein, we found earlier that replacement of the Edmonston H by CAM/RB H is sufficient to induce neurovirulence in mice (9). As described here, the same holds true for intracerebral infection of suckling Lewis rats. The H protein of CAM/RB is sufficient to confer neurovirulence on recombinant MV based on the viral backbone of the nonneurovirulent strain Edmonston. The exact role of single mutations in antibody escape or neurovirulence could not be determined earlier, because there were several mutations in the H proteins and mutations in other viral genes could not be ruled out as having modulating effects (20). Here, we succeeded in identifying single amino acid mutations which mediate the antibody escape and we could separate them from those mediating neurovirulence. Mutations mediating antibody escape are situated on the outer propellers of the MV H protein according to the Langedijk model of paramyxovirus attachment proteins (19). This part of the molecule is supposed to interact not only with neutralizing antibodies (12, 17, 29) but also with the cellular receptors for MV, CD46 (7, 27) and CD150 (11, 16, 42). Since the escape mutations did not affect the replication of the viruses in Vero cells, the interaction with the receptor CD46 appears not to be altered.

In contrast to the mutations mediating antibody escape, the mutations associated with neurovirulence are situated in the stem 2 region of the H protein. The surface of this area is proposed to be parallel to the vertical axis of the H molecule and could affect interactions with molecules in the same membrane as H (e.g., F). It is not clear what structural consequences the mutation at position 195 (GR) may have, except that it potentially introduces a new positive charge on the surface of the molecule. The change at position 200 (SN) generates an additional potential glycosylation site in the CAM/RB H molecule. This sixth potential glycosylation site in the H protein may well cause a structural alteration or mask binding sites, leading to the loss of neurovirulence. The structural alteration might affect the interaction with the unknown cellular attachment receptor in the rat or the fusion helper function of H and, thus, virus entry or cell-to-cell spread. The H protein interaction with further viral proteins such as the matrix-fusion protein complex (M-F) might also influence the viral spread (3, 4). An infectious matrix protein deletion mutant MV exhibited a higher fusogenic capacity than did standard virus and penetrated more deeply into the brain parenchyma in mouse brains (3). Similar results concerning the spread of virus were found with recombinant viruses lacking the cytoplasmic tail of the fusion (F) or hemagglutinin (H) protein, suggesting that the interaction of the M protein with the cytoplasmic parts of these proteins is involved in the regulation of virus-induced cell fusion (4). Interestingly, mutations in the matrix protein and the cytoplasmic domain of the fusion protein have been found in several SSPE brains (1, 5, 36). To prove whether virus entry or cell-to-cell spread is indeed affected by mutation of the glycosylation site at position 200 requires further studies using single mutations at positions 195, 200, and 202 as well as positions 157 and 175 in the CAM/RB background. The latter two must be considered because the mutations in CAM/RB that confer neurovirulence are unique and not present in any other MV lineage or in the other rodent-neurovirulent strain HNT (34).

The sequence analysis of the H genes of the natural escape variants detected changes at amino acid positions 195 and 200 in all variants, in neurovirulent as well as nonneurovirulent viruses (20). Therefore, this mutation initially was not suggested to be responsible for the differential neurovirulent capacity of the variants. Since the variants described above were amplified using Vero cells before RNA was isolated for sequencing, this mutation could have been introduced or selected during the growth in tissue culture as an adaptation to Vero cells. It is known that the CAM/RB strain loses its neurovirulence when propagated several times in tissue culture. The finding that the nonneurovirulent virus CAM/Vero bears only these two mutations at positions 195 and 200 in its H gene in comparison to CAM/RB strongly supports the suggestion that these mutations are selected by Vero cell passage and not for neurovirulence. Our findings using recombinant viruses suggest that the differences in the neurovirulence of the natural escape variants have apparently not been induced by the antibodies used for selecting the escape variants but rather are a consequence of the necessary propagation of the viruses in tissue culture.

On the basis of recombinant viruses, it is clear that these two mutations are the only difference between the recombinant EdtagCAMH, which is neurovirulent, and mutant 6, which is nonneurovirulent, and that antibody escape and neurovirulence are not linked in this model.