Enhanced Neurovirulence of Borna Disease Virus Variants Associated with Nucleotide Changes in the Glycoprotein and L Polymerase Genes

Abstract

Borna disease virus (BDV) infection produces a variety of clinical diseases, from behavioral illnesses to classical fatal encephalitis (i.e., Borna disease [BD]). Since the genomes of most BDV isolates differ by less than 5%, host factors are believed responsible for much of the reported variability in disease expression. The contribution of BDV genomic differences to variation in BD expression is largely unexplored. Here we compared the clinical outcomes of rats infected with one of two related BDV variants, CRP3 or CRNP5. Compared to rats inoculated with CRP3, adult and newborn Lewis rats inoculated with CRNP5 had more severe and rapidly fatal neurological disease, with increased damage to the hippocampal pyramidal neurons and rapid infection of brain stem neurons. To identify possible virus-specific contributions to the observed variability in disease outcome, the genomes of CRP3 and CRNP5 were sequenced. Compared to CRP3, there were four nucleotide changes in the CRNP5 variant, two each in the G protein and in the L polymerase, resulting in four amino acid changes. These results suggest that small numbers of genomic differences between BDV variants in the G protein and/or L polymerase can contribute to the variability in BD outcomes.

Borna disease virus (BDV) is a nonsegmented negative-strand RNA virus (6) belonging to the family Bornaviridae (13, 46). BDV infection in experimental animals has been used to study the pathogenesis of virus-induced central nervous system damage and as a model for specific human diseases, e.g., autism (37). BDV encodes at least six open reading frames (ORFs) in three transcription units: nucleoprotein p40 (N), p10 (X), phosphoprotein p24 (P), matrix protein gp18 (M), membrane glycoprotein gp94 (G), and an RNA-dependent RNA polymerase p190 (L polymerase) (6, 13, 57). Virus replication and transcription take place in the nucleus, a characteristic of Bornaviridae that is unique in the order Mononegavirales (5, 8, 12, 13, 46).

BDV naturally infects a wide range of warm-blooded hosts (41). In horses and sheep, BDV causes classical Borna disease (BD), a fatal mononuclear inflammatory encephalomyelitis with severe signs of neurological disease (as reviewed in reference 41). Classical BD is in large part due to immunopathogenic damage to the nervous system by blood-borne inflammatory cells (49). However, responses to BDV infection vary according to differences in host-specific factors, e.g., species, animal strain, or age of the host at the time of infection (2, 24, 33, 48). For example, BDV-infected adult Lewis rats develop a highly immunopathogenic central nervous system disease characterized by acute, often fatal encephalitis with clinical signs of hyperactivity, aggression, and ataxia, followed by a chronic phase with resolving encephalitis and depressed behavior (19, 41). In contrast, Lewis rats infected as neonates or as immunosuppressed adults do not develop significant inflammatory infiltrates in the brain and may develop behavioral signs of disease without becoming overtly ill (24, 33, 42). Both animal strain- and age-specific host contributions to BD outcomes have also been reported in mice. For example, rodent-adapted BDV variants, but not rabbit-adapted variants or horse isolates, are able to replicate in mice (28), and a mouse-adapted BDV variant causes encephalitis and mild behavioral disease (i.e., hyperactivity) in adult MRL/+ mice but not in SJL mice (43). Infection of neonatal MRL mice with a rat-adapted BDV variant produces significant neurological disease linked to cytotoxic-T-cell-mediated immune responses restricted to H-2^k of major histocompatibility complex class I (20, 22).

In contrast to our knowledge of host-specific factors that affect BD outcomes, the relationship between BDV genomic variability and differing disease outcomes is not well characterized. Sequence comparisons of BDV isolates from diseased animals from regions where BDV is endemic and from regions where BDV is nonendemic typically reveal viral genome conservation of greater than 95% (4, 6, 47). In rare cases where a divergent subtype has been detected, for example No/98 (34), there is still no information about a change in disease phenotype associated with marked changes in genome sequence.

In order to gain a better understanding of virus-specific contributions to BD outcomes, we compared the clinical responses to our mouse-adapted BDV variant (BDV-M-P₅, now termed CRNP5) (43) and our rat-adapted prototype variant (BDV-R-P₃, now renamed CRP3), both derived from the same parent virus (BDV-R-P₂, now renamed CRP2), following inoculation into adult and neonatal Lewis rats. Here we show that adult and newborn Lewis rats inoculated with CRNP5 had more severe and rapidly fatal neurological disease with increased brain damage compared to that of rats inoculated with CRP3. The complete genomic sequencing of these related viruses revealed only four nucleotide differences associated with these significant differences in disease pathogenesis in both adult and neonatally infected Lewis rats, with two amino acid changes each in the G protein and L polymerase. Thus, these data demonstrate that BD outcome can be affected even by a small number of changes in the BDV genome sequence.

MATERIALS AND METHODS

Virus stocks.

BDV-infected MDCK cells (He/80 strain, kindly provided by R. Rott, University of Giessen, Giessen, Germany) were disrupted by ultrasonic treatment, clarified by centrifugation, inoculated intracranially (i.c.) into newborn Lewis rats (CRP1) (7), and then passaged into a second litter of Lewis rats (CRP2). CRP2 was serially passaged by i.c. inoculation into adult SJL mice an additional five times as described previously (CRNP5) (43). CRP2 was also passaged once more in neonatal Lewis rats to generate CRP3. Virus stocks were prepared as 20% (wt/vol) brain homogenate and stored at −80°C.

Virus titer determination.

Semiconfluent monolayers of C6 rat glioma cells were grown in chamber slides in Dulbecco's modified essential medium (Quality Biological, Inc., Gaithersburg, Md.) supplemented with 10% fetal calf serum (Quality Biological, Inc.) and incubated with 10-fold dilutions of virus stock as described previously (10). After 4 days at 37°C, a fluorescent focus-forming unit (FFU) assay was performed to visualize infected cells (7). The cells were fixed with cold acetone for 10 min and analyzed for immunofluorescent signal by using BDV-infected horse serum (29). Virus titers were calculated by assuming that each fluorescent focus of infected cells originated from infection with a single replication-competent virus particle.

Rats and BDV infection.

Four-week-old male (adult) inbred Lewis rats and newborn Lewis rats (Harlan, Indianapolis, Ind.) were used. Adult rats were anesthetized and either inoculated i.c. with 2 × 10³ FFU of CRP3 (n = 24) or CRNP5 (n = 20) or inoculated with an equal volume of uninfected rat or mouse brain material (n = 10 and 8, respectively). Newborn Lewis rats were inoculated i.c. with 2 × 10³ FFU of CRP3 (n = 14) or with 2 × 10² or 2 × 10³ FFU of CRNP5 (n = 12). Again, as controls, neonates were inoculated with the appropriate volume of uninfected material (n = 9). At various time points postinoculation (p.i.), under deep anesthesia, 3 to 5 rats were sacrificed in each group and the brains were removed and bisected for virus titration and histological studies. All rat experimentation conformed to the National Resource Council's guide for the care and use of laboratory animals.

Clinical Borna disease.

Infected rats were examined three times per week and scored for incidence and severity of BD. The severity of disease (SOD score) of adult infected rats was ranked on a scale of 0 to 3 as follows: 0, no disease; 1, early evidence of disease (e.g., lack of grooming or increased activity); 2, signs of neurological disease (e.g., hyperactivity, ataxia, or paresis with mobility, with rats eating and hydrated); 3, severe disease (e.g., total or near total paralysis, severe dehydration, and moribund state). The SOD scoring of neonatally infected rats from postnatal days 1 to 21 was modified due to age-specific issues of physical development: 0, no disease; 1, mild signs of neurological disease (e.g., abnormal movements); 2, significant signs of neurological disease (e.g., ataxia, paresis, or seizures); 3, severe neurological disease (e.g., prolonged seizure, paralysis, or moribund state).

Histological studies.

One half of the brain was fixed in 4% paraformaldehyde and embedded in paraffin for histological examination. Following hematoxylin and eosin (H/E) staining of 8-μm-thick sections, the severity of the encephalitic response (SOE score) was scored in a blinded fashion with a 0-to-4 scoring system as described previously (44). To examine viral antigen distribution in the brain, sections from each brain were stained by avidin-biotin immunohistochemistry (Vector, Burlingame, Calif.) by using horse anti-BDV serum (29) followed by biotinylated anti-horse immunoglobulin G (Vector) as described previously (7).

Anti-BDV antibody assay.

The anti-BDV antibody titer in the sera from infected rats was determined by indirect immunofluorescence assay with twofold dilutions of sera on cold acetone-fixed C6 cells persistently infected with BDV (C6BV cells) as described previously (10).

RNA purification and cDNA synthesis.

Total RNA derived from BDV viral stocks was extracted by using an RNA isolation kit (RNeasy Mini Kit; Qiagen, Valencia, Calif.). RNA (5.0 μg) was reverse transcribed with 200 U of SUPERSCRIPT II RNase H⁻ reverse transcriptase (Life Technologies, Rockville, Md.) and random hexamer primers as recommended by the manufacturer.

Sequencing analysis of PCR products.

The cDNAs of both BDV variants were amplified by using Pfu DNA polymerase (Stratagene, La Jolla, Calif.). Primer pairs were designed to divide the BDV genome into eight overlapping segments. The PCR products were purified and directly sequenced according to the protocol supplied with the Dye Primer cycle-sequencing kit (Applied Biosystems, Warrington, United Kingdom) by using viral specific primers. Sequencing reaction products were analyzed in an automated sequencer (model PE-ABI Prism 377; Applied Biosystems). Variant-specific primers for PCR amplification and sequencing were designed from a published BDV sequence (GenBank accession no. L27077). To confirm the results of direct sequencing, regions of interest were cloned into the pCR 2.1 vector (TA cloning kit; Invitrogen, Carlsbad, Calif.) and at least 10 distinct clones were sequenced. Nucleotide sequences were analyzed with software available online from the National Center for Biotechnology Information.

Sequencing of the 3′ and 5′ ends of the BDV genome.

In order to sequence the 3′ and 5′ ends of the BDV genome, a purified synthetic RNA transcript of known sequence was ligated to the 3′ and 5′ ends of the BDV genomic RNA with T4 RNA ligase (Life Technologies). The RNA was reverse transcribed by using random hexamer primers. The regions around the 5′ and 3′ ends were amplified by PCR with specific primers that anneal to the ligated synthetic RNA and the BDV genome sequences. These PCR products were purified and ligated into the pCR 2.1 vector (TA cloning kit; Invitrogen) and sequenced with the Dye Primer cycle-sequencing kit (Applied Biosystems) by using the M13 Reverse and M13 Forward (−40) primer.

Adventitious agent testing.

Due to the mouse passaging of CRNP5 virus (43), the absence of contaminating rodent adventitious infectious agents in CRNP5 was confirmed by using a commercial PCR amplification test at Molecular Biological Services (IMPACT Profile I and V; University of Missouri Research Animal Diagnostic and Investigative Lab). IMPACT Profile I and V tested CRNP5 for 26 infectious agents of the mouse and rat.

Statistical analysis.

The virus titer and antibody titer data were analyzed by Student's t test and are presented as means ± standard errors of the mean. A P value of <0.05 was considered the criterion for statistical significance.

Nucleotide sequence accession numbers.

The GenBank accession numbers for the CRP3 variants are AY114161 and AY114162, and that for the CRNP5 variant is AY114163. Genomic comparisons were performed with published sequences from He/80, strain V, No/98, H1, H2, and RW 98, which are reported in GenBank accession numbers L27077.2, NC_001607.1, AJ311524.1 L76237, L76238, and AF158633 respectively.

RESULTS

Adult infected Lewis rats. (i) Clinical disease.

The CRP3-inoculated rats developed signs of acute disease beginning on day 23 p.i. (median SOD score, 0.46 ± 0.16) and peaking on day 33 p.i. (2.17 ± 0.31) (Fig. 1A). After day 33 p.i., the SOD score of surviving rats decreased slightly and then plateaued as the rats recovered from acute BD and developed chronic BD. In contrast, the rats inoculated with CRNP5 showed evidence of disease as early as day 9 p.i. (0.07 ± 0.07), and two-thirds of the rats developed severe neurological signs (ataxia and paresis of hind limbs) by day 12 p.i. (2.00 ± 0.22). Neurological disease progressed rapidly, with all rats developing severe neurological disease (paralysis of hind limbs) and a moribund state between days 12 and 16 p.i. (Fig. 1A).

FIG. 1.

Clinical (A) and histological (B) expression of BD phenotype over time in rats inoculated as adults with CRP3 (○), CRNP5 (•), or normal rat or mouse brain (▿). (A) Severity of clinical BD (rated on a scale of 0 to 3); (B) severity of inflammation in the brain (encephalitis, rated on a scale of 0 to 4) in the BDV-infected adult rats. Crosses indicate when all of the remaining animals were morbidly ill and required euthanasia.

(ii) Histological analysis of the brain.

In CRP3-infected rats, mononuclear cell encephalitis was observed beginning on day 7 (in 1 of 5 rats) (Fig. 1B). The severity of the encephalitis was stable from 1 week p.i. to 3 weeks p.i. (total, 13 rats; median SOE score, 0.2 ± 0.2 to 0.4 ± 0.4). The severity of encephalitis increased and peaked at 4 weeks p.i., a time when the rats started to develop maximum neurological disease (5 rats; SOE score, 3.2 ± 0.5). In contrast, encephalitis in CRNP5-infected rats was first detected beginning on day 12 p.i. (in 5 of 5 rats after signs of disease were evident) and peaked on day 14 p.i. (2.3 ± 0.4) (Fig. 1B). Overall, the CRNP5-infected brains showed a milder degree of encephalitis than that induced by CRP3. Histological evidence of an intensive gliosis was present in both CRNP5- and CRP3-infected rats (data not shown). Rats infected with CRNP5 experienced a greater degree of damage to the pyramidal neurons in the CA3/4 region of the hippocampus than did rats infected with CRP3 (Fig. 2).

FIG. 2.

CA3/4 region pyramidal neurons of the hippocampus (arrows) of rats inoculated as adults (a to c) or neonates (d to f) with normal rat or mouse brain homogenate (Sham), CRP3, or CRNP5. These sections of brain were taken approximately 2 weeks p.i. and stained with H/E. Magnification, ×40.

(iii) BDV antibody titers.

The antibody response in rats infected with either variant was detected before the development of neurological disease. Figure 3A shows that in CRP3-infected rats, anti-BDV antibody was detected beginning 3 weeks p.i. (median titer, 1:433 ± 207) and quickly increased to 1:8,704 ± 3,197 by 4 weeks p.i. Antibody levels remained steady through the last time point evaluated (8 weeks p.i.) (data not shown). In contrast, anti-BDV antibody in CRNP5-infected rats was first detected at 1 week p.i. (Fig. 3A). The initial antibody titer was very low (1:16 ± 3) but increased to 1:1,601 ± 333 by 2 weeks p.i. Compared to CRP3-infected rats, CRNP5-infected rats had significantly higher titers of anti-BDV antibodies at 1 and 2 weeks p.i. (P < 0.001). However, at the time of disease onset in each group, BDV antibody titers in CRNP5 infection (2 weeks p.i.) were lower than those in CRP3 infection (4 weeks p.i.).

FIG. 3.

Antibody production (A) and virus replication (B) over time in rats inoculated as adults with CRP3 (○), CRNP5 (•), or normal rat or mouse brain (▿). Asterisks indicate that anti-BDV antibody in CRNP5-infected rats was detected at significantly higher titers than in CRP3-infected rats (P < 0.001). Virus titers were not significantly different between CRP3- and CRNP5-infected rats at any time point. Crosses indicate when all of the remaining CRNP5-infected animals were morbidly ill and required euthanasia.

(iv) BDV replication.

The kinetics and amount of infectious virus production were similar in both groups (Fig. 3B). BDV was detected in the brains of both CRP3- and CRNP5-inoculated rats beginning on day 7 p.i., the earliest time point checked for virus replication. Detection of infectious CRNP5 at this time coincided with the onset of BD, while disease in rats inoculated with CRP3 was not apparent until approximately 2 weeks later. The infectious virus titer measured in CRP3-infected rat brain persisted unchanged from week 4 p.i. through the last time point (8 weeks p.i.) (data not shown). None of the rats inoculated with CRNP5 survived beyond the second week p.i. The infectious virus titers were not significantly different between CRP3- and CRNP5-infected rat groups at 1 and 2 weeks p.i. (P = 0.565 and 0.089, respectively).

(v) Virus distribution.

In CRP3-infected adult rats at 2 weeks p.i., virus antigen was detected in many cells in the cortex, hippocampus, and hypothalamus (data not shown) and in a rare cell in the brain stem (Fig. 4). In CRNP5-infected brain at 2 weeks p.i., the viral antigen was detected in the same areas, but with only a few viral-antigen-positive cells in the cortex (data not shown) and a substantial number of infected cells in the brain stem (Fig. 4).

FIG. 4.

Virus antigen distribution in the representative sections of the brain stem at 2 weeks p.i. from rats inoculated with normal rat or mouse brain homogenate (Sham), CRP3, or CRNP5 as adults (a to f) or as neonates (g to l). The sections shown in panels a to c and g to i were stained with H/E. The sections shown in panels d to f and j to l were stained via immunohistochemistry for BDV antigens with no counterstain. Arrows indicate BDV antigen in infected neurons. Magnification, ×200.

Neonatally infected Lewis rats.

Due to the similarities in disease expression among neonatal rats given CRNP5 inoculum at a dose of either 2 × 10² or 2 × 10³ FFU/rat, data from both CRNP5-inoculated groups are combined for discussion below. The CRNP5 stock was negative for 26 infectious adventitious agents of the mouse and rat.

(i) Clinical disease.

As seen previously, all CRP3-infected pups survived to day 21 without obvious neurological signs of disease (Fig. 5A). In contrast, in CRNP5-infected pups, early signs of clinical disease were first detected by day 14 p.i., and by days 15 to 16 p.i. all had developed severe neurological disease, including paralysis of hind limbs, tremors, and severe ataxia (Fig. 5A), necessitating euthanasia.

FIG. 5.

Clinical expression of BD phenotype (A) and virus replication in the brain (B) over time for rats inoculated as neonates with CRNP5 (•), CRP3 (○), or normal rat or mouse brain homogenate (▿). The severity of BD was rated on a scale of 0 to 3. Crosses indicate when all of the remaining CRNP5-infected animals were morbidly ill and required euthanasia.

(ii) Histological analysis of the brain.

No evidence of inflammatory cell infiltration was observed in pups infected with CRP3 or CRNP5, but an intensive gliosis was present in CRNP5-infected rats, to a degree greater than that seen in CRP3-infected rats (data not shown). In addition, the brains of CRNP5-infected rats showed neuronal degeneration in pyramidal neurons in the hippocampus (Fig. 2) and cortex (data not shown). No gross damage to the hippocampal dentate gyrus neurons was observed in either group by days 15 to 16 p.i. (data not shown).

(iii) BDV replication.

The kinetics and amount of infectious virus production were similar in both groups (Fig. 5B). BDV was detected in the brains of both CRP3- and CRNP5-inoculated rats beginning on day 7 p.i., the earliest time point evaluated for virus replication. At day 7 and 14 p.i., the infectious virus titers were not significantly different between the CRP3- and CRNP5-infected groups (P = 0.197 and 0.313, respectively). The detection of infectious CRNP5 at day 7 p.i. preceded the onset of severe clinical signs of disease (at day 14 p.i.). None of the rats inoculated with CRNP5 survived beyond day 16 p.i.

(iv) Virus distribution.

Virus antigen was widely disseminated in CRNP5-infected neonates, with virus antigen expression in a large number of neurons of the brain stem (i.e., pons and medulla) at days 14 to 16 p.i. (Fig. 4). In contrast, in CRP3-infected neonates, viral antigen was not identified in cells of the brain stem but was largely restricted to the hippocampus, cortex, and cerebellum (data not shown). Notably, staining of the hippocampus was more prominent in CRP3-infected rats than in CRNP5-infected rats, most likely a reflection of the early degeneration of these neurons in the CRNP5-infected rats (data not shown).

Sequence of CRP3 variant.

The nucleotide sequences of published He/80 and CRP3 were similar, differing in only eight nucleotides (Table 1): nucleotide position 29 in the untranslated region prior the ORF of N, nucleotide positions 2742 and 3079 in the ORF of G, and nucleotide positions 3762, 6390, 6892, 7350, and 7777 in the ORF of L polymerase. Of the seven nucleotide changes in the ORFs of G and L polymerase, six resulted in amino acid coding changes.

TABLE 1.

Nucleotide and amino acid changes in He/80 and CRP3

Viral region	Nucleotide change	Amino acid change
Untranslated region	29 (C→A)
G protein	2742 (C→A)	19 (Gly→Gly)
	3079 (A→G)	282 (Met→Val)
L polymerase	3762 (A→T)	26 (Thr→Ser)
	6390 (G→C)	902 (Ala→Pro)
	6892 (C→A)	1069 (Thr→Lys)
	7350 (G→T)	1222 (Val→Phe)
	7777 (A→C)	1364 (Thr→Ser)

Sequence of CRNP5 variant.

Compared to the published sequence of He/80, CRNP5 had the same eight nucleotide changes as CRP3, listed above, plus four additional novel nucleotide changes compared to both He/80 and CRP3 (i.e., a total of 12 nucleotides differed between He/80 and CRNP5 and four additional nucleotides differed between CRP3 and CRNP5). These four additional nucleotide differences between CRNP5 and CRP3 all resulted in amino acid coding changes (Fig. 6). Two changes, at nucleotide positions 3608 and 3673, were located in the ORF of G at amino acid positions 458 (Phe→Ser, F458S) and 480 (Tyr→His, Y480H), respectively. The other two changes, at nucleotide positions 7936 and 8742, were located in the ORF of L polymerase at amino acid positions 1417 (Lys→Arg, K1417R) and 1686 (Gly→Arg, G1686R), respectively (Fig. 6). There were no nucleotide changes in untranslated regions.

FIG. 6.

BDV gene order, coding strategy, and amino acid positions of interest. The numbers on the bottom represent nucleotides, and those on top represent amino acid positions. ORFs in the BDV genome are depicted as boxes. The broken lines show splicing between a small upstream ORF and a downstream ORF of the L-polymerase gene. The sequence of CRP3 had heterogeneity at amino acid position 1686 with both Gly and Arg. Arrows show sites of amino acid changes between CRP3 and CRNP5. Filled circles indicate sites of nucleotide changes between He/80 and CRP3.

F458S and Y480H in the G protein were nonconservative changes. F458S caused a change from a nonpolar, hydrophobic amino acid to a polar, uncharged amino acid, and Y480H caused a change from a polar, uncharged amino acid to a positively charged amino acid. Neither change has been identified in any of the six sequenced BDV isolates from horses and humans registered in GenBank (He/80, strain V, No/98, RW 98, H1, and H2). F458S and Y480H were clustered at the C terminus of the G protein. F458S and Y480H were not located at a potential furin cleavage site or at N-glycosylation sites (17, 18, 40).

K1417R and G1686R of CRNP5 were located in the L polymerase (Fig. 6). The entire ORF of L polymerase is constructed from a small upstream ORF (nucleotides 2393 to 2409) fused to the 5′ end of the larger downstream ORF (nucleotides 3704 to 8822) (55). There were no nucleotide changes identified in the N-terminal half of CRNP5 L polymerase, which has motifs that are conserved on many viral polymerases and has a putative template recognition site that is conserved in L polymerases of Mononegavirales viruses (6, 38).

The amino acid changes in CRNP5 were located in the large downstream ORF towards the C terminus of the L polymerase (Fig. 6). K1417R was a conservative change within the group of positively charged amino acids but represents a unique nucleotide that differs from those of the three published sequences of horse isolates (He/80, strain V, and No/98).

DISCUSSION

In this study of Lewis rats, we evaluated two related BDV variants that replicated with similar kinetics and to similar viral titers in the rat brain but produced radically different disease phenotypes. CRNP5 was significantly more neuroinvasive and neurovirulent than CRP3, and since the genomic sequence differences between these two BDV variants were limited to a small number of amino acid differences in the G protein and the L polymerase, our data suggest that genomic differences in these BDV genes may contribute to the variability of neuropathogenesis of BD.

Evaluation of brain histology and encephalitis following CRP3 and CRNP5 infection of rats revealed some similarities. In adult infected Lewis rats, BD is believed to result from the cellular immune response to virus infection (32, 33, 49), and, indeed, there was some suggestion of an immunopathogenesis component in the disease phenotype in rats infected with CRNP5 as adults (e.g., onset of disease coincided with the onset of encephalitis). In addition, neonatal CRNP5 infection proceeded without evidence of significant inflammatory cell infiltrates in the brain, suggesting an immune tolerance-like effect similar to that seen following CRP3 infection (9).

Despite these similarities, other observations demonstrated substantial differences in CRNP5 infection versus CRP3 infection. Rats infected as adults with CRNP5 succumbed to infection much earlier and to a greater percentage than rats infected with CRP3. Rats infected with CRNP5 as adults were moribund at a time when their encephalitis score was significantly lower than that of CRP3-infected rats evincing milder disease. In contrast to the results of previously reported CRP3 infections of neonatal Lewis rats that largely resulted in behavioral disease and low mortality (3, 8, 14), neonatal rats infected with CRNP5 developed fatal neurological disease by approximately 2 weeks p.i. In CRP5-infected rats with or without encephalitis, rapid degeneration of pyramidal neurons in the CA3/4 region of the hippocampus was observed, a clear departure from the typically protracted hippocampal dentate gyrus damage usually observed following CRP3 infection (9). Thus, these studies demonstrate that in the Lewis rat, CRNP5 infection causes a significantly different disease phenotype than CRP3 infection, whether due to direct virus cytopathic effect, neuroimmune responses (8, 26, 36, 45), or other pathogenic processes (e.g., the inhibition of the function of amphoterin, a neurite outgrowth-promoting adhesive factor, or the indication of neurotrophins) (27, 58).

Variability in BD outcomes can be affected by the specific sites of viral replication in the brain, e.g., infected neurons in the hypothalamus of rats with BDV-induced obesity syndrome (23). Although the kinetics and titers of infectious virus in the brain were similar, CRNP5 was more widely distributed in the brain than CRP3 at the same time points. In particular, the early and intense CRNP5 antigen expression in neurons in the brain stem, at a time when few if any neurons in that region expressed virus antigen in CRP3-infected rats, clearly suggests changes in neural cell tropism and/or the ability of the CRNP5 variant to spread rapidly within the brain. As the brain stem is responsible for basic vital functions of life, including important respiratory and circulation control, the novel ability of CRNP5 to swiftly infect the brain stem neurons is consistent with the rapidly fatal outcome of CRNP5 infection, as has been suggested for the BDV-infected gerbil model (56).

CRNP5 was produced after two serial passages in rat brain (CRP2) followed by five serial passages in mouse brain (a total of seven passages in brain), while the CRP3 variant was a single brain passage in rats beyond CRP2 (a total of three passages in brain). Adaptation of viruses to growth in brain following serial brain passage (neuroadaptation) can occur via poorly defined mechanisms. However, serial passage of CRP2 for a total of seven times in rat brain (i.e., with the same total number of brain passages as CRNP5) produced a variant, CRP7, that when inoculated into neonatal rats produced a clinical picture equivalent to CRP3 with no evidence of the enhanced neurovirulence of CRNP5 (unpublished data). Thus, the high number of brain passages of CRNP5 alone does not explain the increased neuroinvasiveness and neurovirulence of this variant relative to those of CRP3. Rather, consistent with the concepts of quasispecies theory and our sequence data, it is likely that BDV contains minor variant populations, one of which may have been specifically selected and amplified during serial passage in mice.

Small numbers of virus genome changes can result in large differences in virulence. For example, a single amino acid change in (i) the envelope glycoprotein of Sindbis virus or lymphocytic choriomeningitis virus, (ii) the HA protein of avian influenza A virus, or (iii) the nsP2 replicase of Semliki Forest virus (or two amino acid changes in the envelope glycoprotein of dengue virus) changes the severity of neurovirulence outcomes (16, 21, 30, 31, 54). The specific role that each BDV gene plays in neurovirulence and BD phenotype remains poorly characterized. In fact, the reports of differences in BDV disease phenotype have tended to focus on the host, citing the high degree of genome conservation among BDV strains and variants isolated from different species. The consensus sequences of CRNP5 and CRP3 revealed only four nucleotide differences between the two virus variants, resulting in a change of only two amino acids in G protein and two amino acids in L polymerase. Given the distinct difference in disease phenotype following infection of Lewis rats with these variants, however, these data support the novel hypothesis that even a small number of changes in either the BDV G protein or L polymerase, or perhaps both, plays a significant role in BDV-associated neurovirulence. Lacking a reverse genetics system for BDV, it is impossible at this point to identify which of these four mutations, singly or in combination, was responsible for the change in BD phenotype. Notably, Enbergs et al. (15) reported that no differences in BD phenotype were identified in animals infected with the parent or related mouse-passaged variants, but their passaging of BDV in mice did not result in detection of amino acid changes (in the portion of the BDV genome sequenced).

The BDV G protein is part of the outer membrane of the virion along with the M protein, and both proteins are involved in virus entry (50). F458S and Y480H were clustered at the C terminus of the G protein. The G protein contains three hydrophobic amino acid regions, a signal sequence, the putative fusion peptide, and a transmembrane domain (TM; amino acids 468 to 492). F458S was located in front of the TM, and Y480H was located within the TM, causing a predicted change of beta sheet structure preceding the TM and a reduction in hydrophobicity of the TM by Chou-Fasman plot analysis. It is possible that the putative conformational changes in the CRNP5 G protein might increase or change the stability of the G protein complex binding to specific neural cells, possibly resulting in the apparent increase in CRNP5 cell tropism for neurons in the brain stem and/or the rapidity of dissemination throughout the brain.

The functional domains of BDV L polymerase are not well delineated, and little is known about the functions of the C-terminal half of the L polymerases of BDV and of the other nonsegmented negative-strand RNA viruses. By Chou-Fasman plot analysis, there were no identified effects on the secondary structure of the L polymerase of CRNP5 due to the amino acid changes at positions 1417 and 1689. Notably, according to PROSITE analysis, the amino acid changes were not located at predicted protein kinase phosphorylation sites (protein kinase C, tyrosine kinase, casein kinase II, and cyclic AMP- and cyclic GMP-dependent protein kinase phosphorylation sites) or at an N-glycosylation site (25).

The BDV L polymerase, which is known to associate with the BDV P protein, shares some similarity in the N-terminal half to the L polymerases from paramyxoviruses or rhabdoviruses (6, 35, 39, 55). Many RNA viruses share consecutive sequence motifs at the N-terminal half of the L polymerase. Unsegmented negative-strand RNA viruses maintain a highly conserved template recognition site close to this motif (38). There is some evidence correlating L-polymerase functions with virulence. The L polymerase is a multifunctional protein that executes all of the catalytic steps of RNA synthesis, e.g., capping and methylation, and influences viral replication, viral RNA synthesis, and host range. Mutations in the RNA polymerase of poliovirus type 1 may be one determinant of neuroattenuation in mice (52). Moreover, one of the influenza A virus polymerases, PB2, is suggested to be a determinant of cell tropism (1, 11, 51), and a single amino acid change in the PB2 protein has been shown to be associated with a change in host range and virulence in mice and monkeys (21, 51). In addition, nsP2 and nsP3 replicases of Semliki Forest virus affect neurovirulence and attenuation in mice (16, 53).

The concept of host-specific contributions to BD outcome has been well established. However, here we show, for the first time, that a small number of specific changes in the sequences of BDV G and/or L polymerase are associated with profound alterations in BDV neuropathogenesis. Thus, this work highlights a relatively novel concept in BDV pathogenesis: that the extensive genomic conservation reported among BDV strains and variants does not preclude the possibility that the relatively small number of genomic differences among BDV viruses can still be associated with significant differences in disease phenotype. These data underscore the need to elucidate both host- and virus-specific roles in the response of the nervous system to BDV infection. In the future, a reverse genetics system for studying BDV will permit detailed molecular analysis of the determinants of virulence.