Molecular Genetic Analysis of Revertants from a Poliovirus Mutant That Is Specifically Adapted to the Mouse Spinal Cord

Abstract

SA virus, a mutant of the Mahoney strain of type 1 poliovirus (PV1/Mahoney), replicates specifically in the spinal cords of mice and causes paralysis, although the PV1/Mahoney strain does not show any mouse neurovirulence (Q. Jia, S. Ohka, K. Iwasaki, K. Tohyama, and A. Nomoto, J. Virol. 73:6041–6047, 1999). The key mutation site for the mouse neurovirulence of SA was mapped to nucleotide (nt) 928 of the genome (A to G), resulting in the amino acid substitution of Met for Ile at residue 62 within the capsid protein VP4 (VP4062). A small-plaque phenotype of SA appears to be indicative of its mouse-neurovirulent phenotype. To identify additional amino acid residues involved in the host range determination of PV, a total of 14 large-plaque (LP) variants were isolated from a single point mutant, Mah/I4062M, that showed the SA phenotype. All the LP variants no longer showed any mouse neurovirulence when delivered via an intraspinal inoculation route. Of these, 11 isolates had a back mutation at nt 928 (G to A) that restored the nucleotide of the PV1/Mahoney type. The reversions of the remaining three isolates (LP8, LP9, and LP14) were mediated by a second site mutation. Molecular genetic analysis involving recombinants between Mah/I4062M and the LP variants revealed that the mere substitution of an amino acid residue at position 107 in VP1 (Val to Leu) (LP9), position 33 in VP2 (Val to Ile) (LP14), or position 231 in VP3 (Ile to Thr) (LP8) was sufficient to restore the PV1/Mahoney phenotype. These amino acid residues are located either on the surface or inside of the virus particle. Our results indicate that the mouse neurovirulence of PV is determined by the virion surface structure, which is formed by all four capsid proteins.

Poliovirus (PV), which is the causative agent of poliomyelitis, is a human enterovirus that belongs to the Picornaviridae family. The poliovirion is an icosahedral, nonenveloped particle that consists of 60 copies of each of the capsid proteins VP1, VP2, VP3, and VP4, together with a single-stranded RNA genome of positive polarity (10, 14, 21). A deep depression, called a canyon (1, 30), encircling the fivefold axes has been suggested to be an attachment site for the PV receptor (PVR) on the surface of permissive cells.

The entry of PV into cells is initiated upon virus binding to the PVR, which belongs to the immunoglobulin superfamily and is the principal host range determinant for PV infections (20, 25). Binding of PV to the PVR results in destabilization of the virion particle (8, 9, 18), which leads to the conformational changes of the viral capsid necessary for uncoating. These conformational changes include the loss of the internal capsid protein VP4 and the extrusion of the internal N terminus of VP1 (11, 12, 17, 22, 27, 32). Thus, the PVR plays a dual role in PV infection: (i) binding to PV and (ii) initiation of PV uncoating.

PV type 1 (PV1) infects only primates and multiplies exclusively in primate cell lines of human and monkey origin. Other animal species, including mice, are not susceptible to PV1 infection. This host range restriction can be overcome by introducing the human PVR gene into the mouse genome. Expression of the PVR in mouse L cells renders these normally resistant cells susceptible to multicycle PV infection (20, 25). Moreover, transgenic (Tg) mice expressing the PVR (PVR-Tg mice) have been shown to be susceptible to infection with all three serotypes of PV (19, 29).

The experiments involving virus chimeras of the PV1 Mahoney strain (PV1/Mahoney) and PV2 Lansing strain (PV2/Lansing) demonstrated that the BC loop of VP1 (amino acids 94 to 102) plays a critical role in the mouse neurovirulence of PV (4, 23, 24, 28, 33). Additional molecular determinants of the mouse adaptation were identified at amino acid residues located on the surface of as well as inside the viral capsid (2, 3, 5, 6, 15, 26). It has been suggested that both the mouse-avirulent PV1/Mahoney and mouse-adapted mutants were able to attach to the murine receptors in the central nervous system (CNS) but that only the mutant viruses could undergo the receptor-mediated conformational changes required for the subsequent replication steps (7). Furthermore, it has been revealed that viruses adapted to utilize mutant PVR retain the ability to grow in cells expressing wild-type PVR, therefore resulting in an expanded rather than changed receptor specificity (3).

We previously isolated a mouse-adapted PV1/Mahoney mutant, called SA virus, from the spinal cord of a mouse following an intracerebral inoculation with PV1/Mahoney (15). The key mutation site for SA neurovirulence has been identified as a single point mutation at nucleotide (nt) 928 (A to G) of the viral genome, resulting in the amino acid substitution of Met for Ile at residue 62 within the capsid protein VP4 (VP4062). This SA virus produced smaller plaques in cultured cells than PV1/Mahoney. One single point mutant, Mah/I4062M (previously described as Mah/SA-VP4) (15), showed the SA phenotype. To further investigate the capsid residue(s) that is involved in the host range determination of PV, 14 large-plaque (LP) variants were isolated from Mah/I4062M. The molecular genetic analysis of these LP variants revealed several amino acid residues within the viral capsid proteins that played important roles in the mouse neurovirulence determination of PV. The locations of these molecular determinants in the three-dimensional structure of the viral capsid are reported.

MATERIALS AND METHODS

Cells, viruses, and mice.

African green monkey kidney (AGMK) cells were grown in Dulbecco modified Eagle's medium (DMEM) supplemented with 5% newborn calf serum (NCS) and used for the transfection experiments with infectious clones, plaque purifications of the viruses, and plaque assays.

PV1/Mahoney was recovered from AGMK cells transfected with RNA transcribed from the infectious cDNA clone pOM (31). Preparation of Mah/I4062M virus was described in detail previously (15). Virus titers were determined by plaque assays in AGMK cells.

The PVR-Tg mouse line ICR-PVRTg21 (19) and mouse line IQI, an inbred strain of the ICR mouse, were used at the age of 6 to 10 weeks for the mouse neurovirulence tests.

Isolation of LP variants.

Monolayers of confluent AGMK cells grown on 60-mm-diameter plastic dishes were infected with Mah/I4062M virus preparations and then overlaid with 1% agarose in DMEM containing 5% NCS before being incubated for 3 days at 37°C. Viruses with the LP phenotype were picked up by using Pasteur pipettes and transferred to 500 μl of DMEM. After three cycles of freezing and thawing, the mixtures were clarified by low-speed centrifugation and the supernatants were used for three further rounds of plaque purifications. The titers of the viruses thus obtained were determined by plaque assays in AGMK cells. The viruses finally obtained were used as stock viruses for the isolation of viral RNA and for mouse neurovirulence tests.

Isolation of RNA.

Monolayers of confluent AGMK cells on 60-mm-diameter plastic dishes were infected with each virus at a multiplicity of infection (MOI) of 10. After 7 h of incubation at 37°C, the cells were washed once with phosphate-buffered saline and lysed in 1 ml of ISOGEN buffer (Nippongene, Fukuyama, Japan) for 5 min before being extracted once with chloroform. The total RNA was precipitated with isopropanol and washed in 75% ethanol. It was then resuspended in 30 μl of filtered water and used as a template for the first-strand cDNA synthesis.

RT-PCR and nucleotide sequence analysis.

The first-strand cDNA synthesis was performed with Superscript II RT (Gibco) at 37°C for 60 min using a random hexanucleotide (Amersham) as the primer. To inactivate the enzyme, the mixture was incubated at 95°C for 2 min. The product in a 2-μl reaction mixture was then used as a template for PCR. The remaining reverse transcription (RT) reaction mixture was divided into 2-μl aliquots and stored at −20°C for further use. PCR was performed with one cycle at 94°C for 3 min; 30 cycles each of 94°C for 30 s, 50°C for 1 min, and 72°C for 2 min; and one cycle at 72°C for 5 min, using Ex Taq polymerase (Takara, Kyoto, Japan).

To amplify the cDNAs encoding the 5′ noncoding region, capsid region, and noncapsid region of the viral genome, the following three pairs of primers were used, respectively: PV1(M)1>17 and PV1(M)1306<1327, PV1(M)892>912 and PV1(M)3578<3598, and PV1(M)3347>3367 and PV1(M)7208<7227. Probably due to technical difficulties, the cDNA downstream of nt 7227 was not amplified. PCR products were purified with MicroSpin S-400 HR columns (Amersham Pharmacia Biotech, Inc.) and sequenced directly according to the protocol supplied with the SequiTherm long-read cycle sequencing kit (Epicentre Technologies). Each fragment was amplified from the viral RNA of three independent PCRs, and all three were sequenced. If a mutation appeared all three times, its presence in the viral RNA was considered to be correct.

Construction of recombinant cDNAs.

Infectious recombinant cDNAs were constructed using the infectious cDNA clone of pMah/I4062M (previously described as pMah/SA-VP4 [see reference 15]) and RT-PCR fragments of the LP variants. Nucleotide sequence analysis was performed to confirm that the intended mutation was present in the PCR product and that the PCR had introduced no additional mutation(s). A fragment including a single point mutation at nt 2457 (T to C), resulting in an amino acid substitution of Thr for Ile at residue 231 within the capsid protein VP3, was amplified by RT-PCR from LP8 using primers PV1(M)892>912 and PV1(M)2788<2808. The 1.9-kb product was digested with NruI and NheI restriction enzymes and inserted into the plasmid pMah/I4062M, which had been digested with the same enzymes. The infectious clone obtained was designated pSA/I3231T. The infectious cDNA clones of pSA/I2B205F and pSA/V2033I were constructed using a similar method.

A single point mutation at nt 2798 (G to T), resulting in an amino acid substitution of Leu for Val at residue 107 in VP1, was introduced into the pMah/I4062M genome by PCR using two mutagenic primers and two nonmutagenic flanking primers. The nonmutagenic primers included a sense primer, PV1(M)2370>2389 (the nucleotide sequence of nt 2370 to 2389 of the viral RNA), and an antisense primer, PV1(M)4232<4252 (complementary sequence to nt 4232 to 4252 of the viral RNA). The mutagenic primers PV1(M)2789<2810mut1 and PV1(M)2789>2810mut1 were complementary to each other, and the nucleotide sequence of the former was the same as nt 2789 to 2810 except for a mutation site at nt 2798 (G to T). The 1.8-kb product was digested with the SpeI restriction enzyme. The SpeI fragment was then used to substitute the corresponding fragment of pMah/I4062M. The infectious cDNA clone thus obtained was designated pSA/V1107L. pSA/S1206T was constructed using a similar method. The two mutagenic primers PV1(M)3090>3110mut2 and PV1(M)3090<3110mut2 were used to introduce a mutation at nt 3095 (T to A) into the infectious clone of pMah/I4062M.

Transfection.

PvuI-linearized plasmid (4 μg) was transcribed by T7 RNA polymerase as described earlier (16). An aliquot (5 μg) of RNA synthesized was used to transfect AGMK cells (2 × 10⁶ cells) on a 60-mm-diameter plastic dish by the DEAE-dextran method (13, 31). Briefly, we prepared the RNA/DEAE-dextran solution by mixing 5 μg of RNA and DEAE-dextran (1.0 mg/ml) in HBS (280 mM NaCl, 10 mM KCl, 1.5 mM Na₂HPO₄ · 2H₂O, 12 mM dextrose, 50 mM HEPES). Monolayers of AGMK cells were then incubated with the RNA/DEAE-dextran mixture at 37°C for 30 min, after which the mixture was removed. The cells were washed twice with DMEM and overlaid with 5 ml of DMEM containing 5% NCS before being incubated for 3 to 4 days at 37°C until cytopathic effects appeared in almost all the cells. Viruses derived from plasmids pSA/I3231T, pSA/I2B205F, pSA/V1107L, pSA/S1206T, and pSA/V2033I were named SA/I3231T, SA/I2B205F, SA/V1107L, SA/S1206T, and SA/V2033I, respectively.

Mouse neurovirulence test.

Mice, at 6 to 10 weeks of age, were injected with 5 μl of virus suspension via an intraspinal inoculation route. Animals were observed for paralysis and death for up to 14 days. Those animals which showed paralysis within 24 h after injection were not considered to have virus-induced paralysis.

One-step growth curve.

Monolayers of AGMK cells grown overnight in 60-mm-diameter plastic dishes (2 × 10⁶ cells/dish) were infected with viruses at an MOI of 10 and incubated for 20 min each at room temperature and at 37°C. After the unattached viruses were removed, the cells were washed three times with DMEM. They were then overlaid with 3 ml of DMEM containing 5% NCS and incubated at 37°C in a 5% CO₂ incubator. Cells were scraped off at 0, 2, 4, 6, and 8 h postinfection and stored at −80°C. The zero time points corresponded to cells that were scraped from the plates and immediately frozen following washing. The mixture for titration was prepared by three rounds of freezing and thawing. The cell debris was removed by low-speed centrifugation. Viral titers were determined by plaque assays in AGMK cells.

Oligonucleotide primers.

The numbers in the names of the primers correspond to the nucleotide number of the PV1/Mahoney genome. The “>” and “<” symbols indicate sense and antisense, respectively. The site of substitution is shown in boldface type. Primers and their sequences are as follows: PV1(M)1>17, 5′-TTAAAACAGCTCTGGGG-3′; PV1(M)892>912, 5′-TTCCAAGTTCACCGAGCCCAT-3′; PV1(M)1306<1327, 5′-GGCGAATACCCCTAGTGCCCCC-3′; PV1(M)2370>2389, 5′-TTTCGACACCCAGAGAGATG-3′; PV1(M)2789>2810mut1, 5′-CATTTGCAATTGTGGAAGATCA-3′; PV1(M)2789<2810mut1, 5′-TGATCTTCCACAATTGCAAATG-3′; PV1(M)2788<2808, 5′-ATCTTCCACACTGCAAATAGC-3′; PV1(M) 3090<3110 mut2, 5′-CGTCGTAAAAGTGTGTATAGG-3′; PV1(M) 3090>3110 mut2, 5′-CCTATACACACTTTTACGACG-3′; PV1(M)3347>3367, 5′-ACGCTTACACCCCTCTCCACC-3′; PV1(M)3578<3598, 5′-GTAGTATTTCCTTCTAGACTC-3′; PV1(M)4232<4252, 5′-CTTATCTCTAGCTTGTGGGAT-3′; and PV1(M)7208<7227, 5′-ACGTGATCCTGAGTGTTCCT-3′.

Analysis of three-dimensional structures.

The three-dimensional structure of the PV capsid protomer was analyzed using the Protein Data Bank 3DB browser (European Bioinformatics Institute and Macromolecular Structure Database) with the ID code 2PLV (10, 14). Locations of residues were mapped onto the three-dimensional structure using a STING in the 3DB browser.

Nomenclature.

Mutations in the capsid proteins are designated as follows: the first letter is the wild-type amino acid, the first digit indicates the capsid protein, the following three digits identify the amino acid position, and the last letter is the mutant amino acid. For example, I4062M indicates that Ile at residue 62 in the capsid protein VP4 is replaced with Met. The mutation at nt 3836, which results in the substitution of Phe for Ile at residue 205 in the noncapsid protein 2B, was designated I2B205F. The recombinant viruses containing one single point mutation in the viral genome were designated “Mah/” followed by the position of the mutation. The recombinant viruses containing a new mutation in addition to the mutation I4062M were designated “SA/” followed by the position of the new mutation.

RESULTS

Isolation of LP variants.

Mah/I4062M, a single point mutant of the wild-type PV1/Mahoney with respect to nt 928 (A to G), has an SA phenotype, that is, both a small-plaque phenotype and a mouse-neurovirulent phenotype. This mutant is thought to specifically recognize a molecule on the cell surface of neurons in the mouse spinal cord as a receptor (15). To gain further insight into the molecular mechanism of host range determination in PV infection, we isolated revertants from Mah/I4062M. Since the small-plaque phenotype of the virus appears to correlate with the mouse-neurovirulent phenotype, we utilized the plaque size as a marker for isolation of the Mah/I4062M revertants which might show the LP phenotype of PV1/Mahoney in AGMK cells. LP variants were easily generated during the passages of Mah/I4062M in cultured cells. As described in Materials and Methods, 14 LP variants were isolated and named LP1, LP2, LP3, and so on. The plaques produced by the viruses of PV1/Mahoney, Mah/I4062M, and LP14 in AGMK cells are shown in Fig. 1. The plaque sizes formed by the LP variant were similar to those of PV1/Mahoney. None of the LP variants were any longer neurovirulent to mice when delivered via an intraspinal inoculation route (data not shown).

FIG. 1.

Plaque phenotypes of PV1. Monolayers of confluent AGMK cells grown in 60-mm-diameter dishes were infected with PV1/Mahoney, Mah/I4062M, or LP14. After the dishes were incubated at 37°C for 3 days, the cells were fixed and stained with 1% crystal violet.

Genotypes of LP variants.

To investigate the genotype of the LP variants, the nucleotide sequence of the genome region encoding the capsid protein VP4 was determined using a cycle sequencing method as described in Materials and Methods. The result showed that the reversion to the phenotype of PV1/Mahoney occurred in two ways (data not shown): (i) 11 isolates had a back mutation at nt 928 (G to A) that restored the wild-type nt sequence; and (ii) the remaining three isolates (LP8, LP9, and LP14) retained the SA genotype at nt 928 (G), suggesting that a suppressor mutation(s) had occurred on the genomes of these variants.

To understand the interactions between the mouse neurovirulence determinant at VP4062 and its suppressor mutation(s), we analyzed the nucleotide sequences of the genomes for LP8, LP9, and LP14.

Mutation site(s) determining the phenotypes of LP variants.

To map the possible new mutation(s) in the genomes of LP8, LP9, and LP14, cDNAs excluding the 3′-proximal portion of the viral genome were synthesized and the nucleotide sequences were determined as described in Materials and Methods. A comparison with the sequence of the viral RNA from Mah/I4062M revealed two point mutations in the genome of LP8, two point mutations in LP9, and a single point mutation in LP14 (Fig. 2). Of the two mutations in LP8, one existed at nt 2457 (T to C) and one existed at nt 3836 (A to T), resulting in an amino acid substitution of Thr for Ile at residue 231 in the capsid protein VP3 and a substitution of Phe for Ile at residue 205 in the noncapsid protein 2B, respectively. The two mutations in the genome of LP9 were located at nt 2798 (G to T) and 3095 (T to A), resulting in substitutions of Leu for Val at residue 107 and Thr for Ser at residue 206 in VP1, respectively. The single point mutation in LP14 was located at nt 1046 (G to A), resulting in an amino acid substitution of Ile for Val at residue 33 in the capsid protein VP2.

FIG. 2.

Nucleotide and amino acid substitutions in the genomes of LP variants. The PV genome is shown at the top of the panel. The genome structure of Mah/I4062M is shown as a combination of PV1/Mahoney (black box) and SA (horizontally striped box) sequences. Nucleotide and amino acid (a.a.) substitutions in the genome of Mah/I4062M and LP variants are shown as open boxes. Genome regions shown by diagonal lines were not sequenced. The 50% lethal dose (LD50 [PFU]) value of each virus obtained from the mouse neurovirulence tests following an intraspinal inoculation is shown on the right. Plaque size was measured following a 3-day incubation in AGMK cells, and the results are shown on the right. L, >3 mm; S, <1 mm.

To test whether the mutation at nt 1046 was the key suppressor mutation for the phenotype of LP14, the recombinant virus SA/V2033I was constructed and its neurovirulence was tested in IQI mice. SA/V2033I showed a phenotype similar to that of LP14 (Fig. 3). Since LP8 and LP9 contained two mutations in each of their viral genome, recombinant viruses that contained one each of the mutations derived from LP8 or LP9 were constructed as described in Materials and Methods (Fig. 3). The neurovirulence of these recombinant viruses tested in IQI mice via an intraspinal inoculation route. The recombinant viruses SA/I3231T and SA/V1107L showed plaque size and neurovirulence phenotypes similar to those of PV1/Mahoney, whereas SA/I2B205F and SA/S1206T did not. These observations indicated that the mouse neurovirulence mutation at VP4062 was suppressed by amino acid substitutions at either residue 33 in the capsid protein VP2, residue 231 in VP3, or residue 107 in VP1. Since the recombinant viruses SA/I2B205F and SA/S1206T showed a phenotype similar to that of Mah/I4062M, the amino acid substitutions at residue 205 in the noncapsid protein 2B and residue 206 in VP1 appear to have no functional interactions with the amino acid residue at VP4062 in determining mouse neurovirulence.

FIG. 3.

Recombinants between Mah/I4062M and LP variants. The genome structures of PV1 and Mah/I4062M and the amino acid (a.a.) substitutions in the genomes of the recombinants are as defined in the legend for Fig. 2. Cleavage sites of the restriction enzymes used for the allele replacement are shown by the name of the enzyme with the relevant nucleotide position. The plaque size was defined as in the legend for Fig. 2. The mouse neurovirulence of each virus was tested following an intraspinal inoculation of 10⁶ PFU of each virus. The results are shown on the right (number paralyzed/number infected).

Replication of LP variants in the CNS of mice.

The fact that LP variants did not cause disease in normal mice suggested that these viruses had lost their ability to replicate in the CNS of mice. To verify this possibility, 10⁵ PFU of LP14 was injected into normal IQI mice via an intraspinal inoculation route, and the time profile of virus replication was determined. The procedure for the recovery of virus from the spinal cord of mice followed that of our earlier report (15). Tissues were removed from the infected mice at 1, 24, 48, and 72 h postinfection and homogenized. Virus titers were determined by plaque assays. The time profiles of PV1/Mahoney, Mah/I4062M, and LP14 in the spinal cords of IQI mice are shown in Fig. 4A. In accordance with the results of the neurovirulence tests, the titers of Mah/I4062M virus in the spinal cords of normal IQI mice were increased up to 10⁶ PFU at 72 h postinfection. However, the virus titer of LP14 in the spinal cords of normal IQI mice did not increase and its time profile was similar to that of PV1/Mahoney (Fig. 4A). These results suggest that the new mutations generated in the genomes of LP variants suppressed the original mutation at VP4062, resulting in a loss of the ability to recognize the murine receptor or the capacity to undergo the receptor-mediated conformational changes required for the subsequent steps, and thus Mah/I4062M reverted from a virulent virus to an avirulent one in mice.

FIG. 4.

Kinetics of PV replication in the CNS of mice. (A) A total of 10⁵ PFU of PV1/Mahoney, Mah/I4062M, or LP14 virus was injected into non-Tg IQI mice via an intraspinal inoculation route. (B) A total of 10⁴ PFU of each virus was injected into Tg mice via an intracerebral inoculation route. The brains and spinal cords of the mice were removed at the indicated times and homogenized as described in Materials and Methods. Virus titers were determined by plaque assay.

Considering that the suppressor mutations in the genomes of LP variants might render these viruses able to use the PVR efficiently, we infected the PVR-Tg mice with 10⁴ PFU of virus suspension via an intracerebral inoculation route (Fig. 4B). The virus LP14 grew well in the brain, like PV1/Mahoney, and the virus titers of LP14 and the Mahoney virus were about 10 times higher than that of Mah/I4062M at 72 h postinfection. These results suggested that LP14 recognized PVR in the CNS of mice as efficiently as PV1/Mahoney. The suppressor mutations at residue 107 in VP1 and residue 231 in VP3 might work in a mechanism similar to that of residue 33 in VP2.

One-step growth curves of LP variants.

In our previous study (15), the titers of virus harvested from AGMK cells infected with SA virus were shown to be about 10 times lower than those of virus harvested from cells infected with PV1/Mahoney. The LP variants showed phenotypes similar to the wild-type Mahoney strain with regard to mouse neurovirulence and plaque size, which indicated that the replication process of LP variants might have reverted to the PV1/Mahoney type. To test this possibility, we performed one-step growth experiments on the recombinant LP viruses SA/I3231T, SA/V1107L, and SA/V2033I, which all contained one key suppressor mutation in the viral genome. When the AGMK cells were infected with viruses at an MOI of 10, the virus titers of SA/I3231T and SA/V1107L were approximately 10 times lower than that of PV1/Mahoney and similar to those of Mah/I4062M (Fig. 5). This suggested that some steps in viral replication of these viruses were not restored to those of PV1/Mahoney in cultured cells. However, the time profile of the virus titer for SA/V2033I was similar to that of Mahoney, suggesting that the suppressor mutation at residue 33 of VP2 facilitates virus replication in cultured cells.

FIG. 5.

One-step growth curves of LP variants. Monolayers of AGMK cells grown on 60-mm-diameter dishes were infected with each virus at an MOI of 10. Cells were scraped at the indicated times and frozen at −80°C. The mixture was prepared for titration by three rounds of freezing and thawing, and the cell debris was removed by low-speed centrifugation. Viral titers were determined by plaque assay in AGMK cells.

DISCUSSION

The characterization of the substitutions that restored the phenotype of a wild-type virus gave us an insight into the important interactions between the different capsid proteins of PV. Mah/I4062M showed the small-plaque (Fig. 1) and mouse adaptation (Fig. 2) phenotypes, which were similar to those of SA. The amino acid substitution at position VP4062 (Ile to Met) may facilitate the receptor-mediated conformational changes which are required for cell entry or uncoating subsequent to receptor attachment. It may also change the surface structure of the virion by interacting with other capsid proteins, which in turn affects the conformation of the canyon and the neighboring structure, thus rendering the virus neurovirulent in mice.

The small-plaque phenotype and the low yield of Mah/I4062M in cultured cells (Fig. 5) may be explained by a defect in viral genome replication, translation, assembly, and the low stability of the virion. The molecular genetic analysis of LP8, LP9, and LP14 revealed an interesting result in that the mere substitution of one amino acid residue at position 107 in VP1 (Val to Leu), position 33 in VP2 (Val to Ile), or position 231 in VP3 (Ile to Thr) totally restored the phenotype of PV1/Mahoney. LP variants (LP14) replicated efficiently in the brains of PVR-Tg mice (Fig. 4B), but not in the spinal cords of normal IQI mice (Fig. 4A). The one-step growth rate of SA/V2033I in AGMK cells was similar to that of PV1/Mahoney (Fig. 5). These results indicate that SA/V2033I could recognize the PVR as efficiently as PV1/Mahoney. Residue 33 in VP2 is located on the inside of the three-dimensional structure of the viral capsid and interacts with the polypeptide VP4, which is released during the receptor-mediated conformational changes that lead to 135S particles (Fig. 6). This determinant is close to a mutation at amino acid 31 (Ser to Thr) in the capsid protein VP2, which has been shown to confer the mouse-virulent phenotype to the mouse-avirulent strain PV1/Mahoney (5). The suppressor mutation at this site may allow the virus to recognize PVR as efficiently as PV1/Mahoney. This may, in turn, improve the release of VP4 from the virion during the receptor-mediated virion conformational changes by interacting with the amino acid residue at VP4062 (Met) and other capsid proteins. Thus, SA/V2033I had the phenotype of PV1/Mahoney restored. Our result is in accordance with previous reports that the internal determinants could define a functional area involved in both the early steps of the viral cycle and host range restriction in PV (5, 6, 26).

FIG. 6.

Locations of the amino acid substitutions on the three-dimensional structure of PV1/Mahoney capsid protomer. Single capsid protomer viewed from the side (left) or the outside (right) of the particle, 60 of which form the particle. VP1 is blue, VP2 is light blue, VP3 is green, and VP4 is yellow. Amino acid substitutions for LP phenotype are shown as white spheres. The amino acid residues are given (e.g., 3231, residue 231 of VP3). 5 ×, fivefold axis of symmetry.

The recombinant viruses SA/I3231T and SA/V1107L showed LP phenotypes and had lost their neurovirulence in IQI mice (Fig. 3), both of which are similar to those of PV1/Mahoney. Residue 107 in VP1 is located near the carboxyl-terminal end of the second strand of the conserved β-barrel in VP1 (residue 105 to 109), which is exposed on the surface of the virion and on the canyon wall (Fig. 6). The location of this residue may have substantial effects on the conformation of the canyon and neighboring structures, as well as the assembly of the virus particle itself. The amino acid residue at position 231 in VP3 is located on the surface of the particle (Fig. 6), near the canyon floor. The mutation at this site is a new mouse neurovirulence determinant that has not been reported before. The Ile side chain at position 231 is buried, and the residue is unlikely to play any direct role in the initial virus attachment. The nature of the amino acid residue at this location in the capsid may alter the stability of the interface between fivefold-related copies of VP1 such that it alters the ability of the virus to undergo some conformational transitions required for virus uncoating. It might also be reflected in an alteration of the ability of the virus to make a high-affinity interaction with the receptor.

Although SA/I3231T and SA/V1107L showed LP phenotypes in cultured cells, the growth rates of SA/I3231T and SA/V1107L in cultured cells were lower than that of Mahoney (Fig. 5). This is possibly because the amino acid substitution at position 231 in VP3 or at position 107 in VP1 of Mah/I4062 M may reduce the stability of the virions. Indeed, the titer of these viruses rapidly decreased after several cycles of freezing and thawing (data not shown).

The results of this study, along with those in related works (3, 5, 6, 15, 23, 26), indicate that the ability of PV1/Mahoney to infect mice can be obtained by a substitution of only one amino acid either on the surface or inside of the virion. The amino acid residues in all four capsid proteins, VP1, VP2, VP3, and VP4, that were identified as key mutation sites in this study may control the delicate balance between the need for structural stability and the requirement for the ability to undergo conformational changes mediated by the PVR and murine nerve cell receptors. The results also suggest that PVR recognition is determined by the virion surface structure, which is formed with all four capsid proteins.