Identification of Host-Specificity Determinants in Betanodaviruses by Using Reassortants between Striped Jack Nervous Necrosis Virus and Sevenband Grouper Nervous Necrosis Virus

Abstract

Betanodaviruses, the causal agents of viral nervous necrosis in marine fish, have bipartite positive-sense RNAs as genomes. The larger genomic segment, RNA1 (3.1 kb), encodes an RNA-dependent RNA polymerase, and the smaller genomic segment, RNA2 (1.4 kb), codes for the coat protein. Betanodaviruses have marked host specificity, although the primary structures of the viral RNAs and encoded proteins are similar among betanodaviruses. However, no mechanism underlying the host specificity has yet been reported. To evaluate viral factors that control host specificity, we first constructed a cDNA-mediated infectious RNA transcription system for sevenband grouper nervous necrosis virus (SGNNV) in addition to that for striped jack nervous necrosis virus (SJNNV), which was previously established by us. We then tested two reassortants between SJNNV and SGNNV for infectivity in the host fish from which they originated. When striped jack and sevenband grouper larvae were bath challenged with the reassortant virus comprising SJNNV RNA1 and SGNNV RNA2, sevenband groupers were killed exclusively, similar to inoculation with SGNNV. Conversely, inoculations with the reassortant virus comprising SGNNV RNA1 and SJNNV RNA2 killed striped jacks but did not affect sevenband groupers. Immunofluorescence microscopic studies using anti-SJNNV polyclonal antibodies revealed that both of the reassortants multiplied in the brains, spinal cords, and retinas of infected fish, similar to infections with parental virus inoculations. These results indicate that viral RNA2 and/or encoded coat protein controls host specificity in SJNNV and SGNNV.

Betanodaviruses, members of the family Nodaviridae, are the causal agents of a highly destructive disease in hatchery-reared larvae and juveniles of a variety of marine fish. The disease was named viral nervous necrosis (VNN) when it was first described in 1990 (26) and is also known as viral encephalopathy and retinopathy (21). VNN disease has spread to 30 or more marine fish species of 14 families in the Indo-Pacific region, the Mediterranean region, Scandinavia, and North America (16). Recently, adult and mature fish have also suffered from the disease (16). Affected fish exhibit a range of neurological abnormalities, which are characterized as vacuolization and cellular necrosis in the central nervous system and retina.

Betanodaviruses are nonenveloped, spherical viruses with a bipartite positive-sense RNA genome (4, 11, 23). The larger genomic segment, RNA1 (3.1 kb), encodes an RNA-dependent RNA polymerase (so-called protein A). The smaller genomic segment, RNA2 (1.4 kb), encodes the coat protein (CP). Very recently, we characterized a subgenomic RNA3 (378 nucleotides [nt]) that encodes protein B, of unknown function(s). RNA3 is synthesized from the 3′ terminus of RNA1 during RNA replication. RNA3 is involved in viral RNA replication as well as encapsidation (T. Iwamoto, K. Mise, K.-I. Mori, M. Arimoto, T. Nakai, and T. Okuno, submitted). These characteristics are similar to those of alphanodaviruses (2), which are the other members of the family Nodaviridae. In contrast to betanodaviruses, alphanodaviruses, especially flock house virus (FHV), are well studied in many aspects, including virus structure, encapsidation, viral RNA replication, and gene silencing (2, 14), and provide attractive model systems for molecular virological study. Although alphanodaviruses and betanodaviruses seem to share many virological characteristics, there are critical differences between the two viruses. An interesting example is that alphanodaviruses, but not betanodaviruses, require the conversion of provirions into mature infectious virions by the processing of CP (protein α) (2). Another example is that alphanodaviruses can multiply in a wide range of cultured cells derived from insects and mammals as well as in yeasts, whereas betanodaviruses exclusively infect specific fish cell lines (3, 9).

Betanodaviruses can be classified into four types, designated striped jack nervous necrosis virus (SJNNV), berfin flounder nervous necrosis virus (BFNNV), tiger puffer nervous necrosis virus (TPNNV), and redspotted grouper nervous necrosis virus (RGNNV), based on the similarity of the partial RNA2 sequences encoding the C-terminal halves of the CPs (20). Among the four types of betanodavirus isolates, the partial RNA2 sequences, bearing almost entire open reading frames (approximately 870 bp), share 76 to 82% homology at the nucleotide level and 81 to 88% homology at the amino acid level (19). The RNA1 sequence similarities between SJNNV and greasy grouper nervous necrosis virus (GGNNV), the latter of which is closely related to RGNNV, were 80% at the nucleotide level and 87% at the amino acid level (23). However, the RNA2 sequence similarities between SJNNV and four alphanodaviruses (FHV, black beetle virus, boolarra virus [BoV], and nodamura virus) are only 29% or less at the nucleotide level and 11% or less at the amino acid level (19). Similarly, the SJNNV protein A sequence also shares weak homology with those of FHV (28%), black beetle virus (28%), BoV (27%), and nodamura virus (29%) (12).

In spite of the nucleotide and amino acid sequence similarities among betanodaviruses, SJNNV and TPNNV cause diseases only in the striped jack Pseudocaranx dentex and the tiger puffer Takifugu rubripes, respectively, whereas BFNNV has been isolated from some cold-water species, such as the berfin flounder Verasper moseri, the turbot Scophthalmus maximus, and the Atlantic halibut Hippoglossus hippoglossus. Interestingly, RGNNV has a broad host range and causes diseases for a variety of warm-water fish species, particularly groupers and sea bass (16, 20). However, the mechanisms underlying the host specificity in betanodaviruses is unknown. Similarly, for alphanodaviruses nothing is known about the viral or host components that determine host specificity (2).

We recently established a reverse genetics system for SJNNV (11) with the help of cultured cell lines that are infected by all of the betanodaviruses identified so far (8-10). For this study, to identify the viral components that control host specificity in betanodaviruses, we established a reverse genetics system for sevenband grouper nervous necrosis virus (SGNNV), a member of the RGNNV type, and produced reassortant viruses between SJNNV and SGNNV. We evaluated the infectivity of the reassortants in striped jacks and the sevenband grouper Epinephelus septemfasciatus.

MATERIALS AND METHODS

Viruses and cells.

SJNNV (SJNag93 strain) (11) and SGNNV (SGWak97 strain) (9) were used for this study. The E-11 cell line (10) was grown at 25°C in Leibovitz's L-15 medium (Invitrogen, Carlsbad, Calif.) supplemented with 5% fetal bovine serum.

SGNNV purification.

Brains and eyes recovered from infected sevenband grouper juveniles were homogenized. The homogenate was inoculated into E-11 cells and the cells were incubated at 25°C. The culture supernatants of the infected cells, containing progeny virus, were harvested and centrifuged at 10,000 × g for 10 min at 4°C. The resulting supernatants were layered onto 10 to 40% (wt/vol) sucrose gradients and centrifuged at 80,000 × g for 2 h at 16°C. Each fraction was collected, and its virus content was analyzed by Western blot analysis as described below. Positive fractions were concentrated in a centrifugal filter unit (Centricon; Amicon, Beverly, Mass.) according to the manufacturer's instructions to yield purified virions.

Determination of 5′ and 3′ ends of the SGNNV genome.

SGNNV RNA1 and RNA2 were extracted from the purified virions by using ISOGEN-LS (Nippon Gene, Tokyo, Japan) according to the manufacturer's instructions and were used as the templates for cDNA synthesis. To obtain initial viral cDNA clones, we synthesized double-stranded cDNAs from the extracted RNAs by using the TimeSaver cDNA synthesis kit (Amersham, Uppsala, Sweden) and random hexamer oligonucleotide primers according to the supplier's instructions. cDNAs thus obtained were cloned into pBluescript SK(−) (Stratagene, La Jolla, Calif.), and large cDNA clones for SGNNV RNA1 and RNA2 were selected by PCR using M4 and RV primers (Table 1), which amplify a cloned DNA fragment from this vector. Since there was a possibility that these large cDNA clones still lacked 5′ and 3′ end sequences, terminal sequences were determined by the rapid amplification of cDNA ends (RACE) method, as described previously (11). To obtain full-length viral cDNAs, we synthesized two sets of oligonucleotide primers, SG1-5ST7 and SG1-3Ec for RNA1 and SG2-5ST7 and SG2-3Ec for RNA2 (Table 1), based on the RACE results. Reverse transcriptase (RT)-PCR was performed with these primers and with SGNNV virion RNAs as templates. SG1-5ST7 and SG2-5ST7 have a T7 promoter sequence and a SalI restriction site, and SG1-3Ec and SG2-3Ec have an EcoRI site. Amplified fragments were digested with SalI and EcoRI and were cloned into the SalI and EcoRI sites of pUC119 (Takara, Otsu, Japan) according to a standard protocol (22). Full-length viral cDNAs were verified by sequencing, and the plasmids bearing correct full-length clones were named pSG1TK5 (for RNA1) and pSG2TK13 (for RNA2).

TABLE 1.

Oligonucleotide primers used in this study

Primer name	Sequence (5′-3′)a	Use
SG1-5ST7	CCCCGTCGACtaatacgactcactatagTAACATCACCTTCTTGCTCT	Full-length SGNNV RNA1 cDNA cloning
SG1-3Ec	ACCGGAATTCGCCGAAGCGTAGGACAGCA	Full-length SGNNV RNA1 cDNA cloning
SG2-5ST7	CCCCGTCGACtaatacgactcactatagTAATCCATCACCGCTTTGCA	Full-length SGNNV RNA2 cDNA cloning
SG2-3Ec	ACCGGAATTCGCCGAGTTGAGAAGCGATC	Full-length SGNNV RNA2 cDNA cloning
SJ1-Fid	CTACCAAGCTCTTGGTGACA	Detection of SJNNV RNA1
SJ1-Rid	GGCGTGCAGCTCCTCCTCTC	Detection of SJNNV RNA1
SJ2-Fid	CAAACGCTGTCTTTGTCACT	Detection of SJNNV RNA2
SJ2-Rid	ACAGGTTCGGCGAGGTAAGC	Detection of SJNNV RNA2
SG1-Fid2	AACGCGTCATCGCTGAGAAG	Detection of SGNNV RNA1
SG1-Rid2	ATCAGTGTCGTAGGCACT	Detection of SGNNV RNA1
SG2-Fid	ATGGAGCAGTCTTCCAGCTG	Detection of SGNNV RNA2
SG2-Rid	GCGCTTCCAGCCGTGTATAG	Detection of SGNNV RNA2
RV	CAGGAAACAGCTATGAC	Amplification of a clone from pBluescript SK(−)
M4	GTTTTCCCAGTCACGAC	Amplification of a clone from pBluescript SK(−)

^aThe SGNNV-specific sequences determined by RACE are underlined. The T7 promoter sequence is indicated in lowercase. Restriction enzyme sites are italicized.

Transfection of cultured fish cells and infection assay.

Synthesis of capped transcripts from EcoRI-linearized full-length cDNA plasmids and inoculation of E-11 cells with transcripts were performed as described previously (11). At 24 h posttransfection, the infectivity of the transcripts and their progeny was examined by immunofluorescence staining with anti-SJNNV rabbit polyclonal antibody and fluorescein isothiocyanate-conjugated swine immunoglobulin raised against rabbit immunoglobulin (Dako, Copenhagen, Denmark), as described previously (17).

Western blot analysis.

Transfected E-11 cells were suspended in Laemmli's sample buffer (13) and subjected to sodium dodecyl sulfate-12.5% polyacrylamide gel electrophoresis. Western blot analysis was carried out as described previously (5), using an Immobilon-P transfer membrane (Millipore, Bedford, Mass.). SJNNV and SGNNV CPs were detected with anti-SJNNV rabbit polyclonal antibody and alkaline phosphatase-conjugated goat anti-rabbit secondary antibody (Bio-Rad, Hercules, Calif.), followed by incubation with 5-bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium for color development. The anti-SJNNV polyclonal antibody was applicable for the detection of SGNNV CP because there was a high degree of cross-reactivity between them.

Virulence assay for fish larvae.

One-day-old striped jack and sevenband grouper larvae, reared at Kamiura Station of the Japan Sea-Farming Association, were used for virulence assays of reassortants and parental viruses. Larvae were kept at 23°C without feeding in glass beakers containing 1 liter of seawater with 5 μg of kanamycin/ml. For striped jacks and sevenband groupers, 250 and 100 larvae, respectively, were used for each treatment. Prior to infection experiments, these larvae were confirmed to be free of SJNNV and SGNNV by RT-PCR, as described previously (18). At 5 days posttransfection with in vitro transcripts, culture supernatants (100 μl) of E-11 cells were applied into the beakers, and resulting dead or moribund fish were collected daily for 5 days. The virus titers (the 50% tissue culture infectious dose [TCID₅₀]/ml) of the culture supernatants, determined as described previously (10), were 10^6.6 TCID₅₀/ml for striped jacks and 10^8.6 TCID₅₀/ml for sevenband groupers. Collected dead or moribund fish were fixed with 10% formalin and embedded in paraffin. Sections were subjected to immunofluorescence staining as described above.

Viral segment-specific RT-PCR.

The total RNA was extracted from the fish larvae challenged with reassortants or parental viruses by using ISOGEN (Nippon Gene) and was used as a template for RT-PCR. To amplify SJNNV- or SGNNV-specific segments, oligonucleotide primers were designed (Table 1) and used for RT-PCR. After reverse transcription with SuperScript II (Invitrogen) at 42°C for 50 min, PCR was performed with Ex Taq polymerase (Takara) for 30 cycles of denaturation at 94°C for 40 s, annealing at 65°C for 60 s, and extension at 72°C for 90 s for RNA1. Similarly, for RNA2, PCR was done for 30 cycles of denaturation at 94°C for 40 s, annealing at 55°C for 60 s, and extension at 72°C for 60 s.

Nucleotide sequence accession numbers.

The GenBank accession numbers of the sequences reported in this paper are AY324869 and AY324870.

RESULTS

Cloning of full-length SGNNV cDNAs.

We synthesized cDNA clones of SGNNV genomic RNAs by using purified virion RNA samples and random hexamer oligonucleotide primers. Since initial cDNA clones likely lacked the 5′ and 3′ end sequences of the SGNNV genome, unknown terminal sequences were determined by the RACE method, as described previously (11). To obtain full-length cDNA clones of SGNNV RNA1 and RNA2, we designed two sets of oligonucleotide primers (Table 1) based on the 5′- and 3′-terminal sequences determined by RACE. Full-length viral cDNAs were amplified by RT-PCR using the oligonucleotide primers and were individually cloned into a pUC119 vector. Full-length clones (pSG1TK5 for RNA1 [3,105 nt] and pSG2TK13 for RNA2 [1,434 nt]) were verified by DNA sequencing and were used for further studies. Protein A, protein B, and CP sequences were deduced from the full-length cDNAs by open reading frame analysis (data not shown).

Structural similarities between SGNNV and other nodaviruses.

BLASTN searches for SGNNV RNA1 homologs hit GGNNV RNA1 (23) with the best matching score (97%). When SGNNV protein A and protein B were submitted for BLASTP searches, GGNNV protein A (98%) and protein B (97%) (23) were identified as the best matching homologs (data not shown). Pair-wise alignments of SGNNV RNA1, protein A, and protein B sequences with those of GGNNV revealed that SGNNV RNA1 is 2 nt longer than that of GGNNV, although the encoded protein A and protein B were the same size as those of GGNNV (982 and 75 amino acids, respectively). The two extra nucleotides were located in the 3′ untranslated region of SGNNV RNA1. Similarly, when we performed BLASTN and BLASTP searches for SGNNV RNA2 homologs, GGNNV RNA2 (23) showed the highest score at the nucleotide level (99%) and showed the second highest score at the amino acid level (98%). The Epinephelus coioides nervous necrosis virus CP (protein ID AAN04042.2) gave the highest score (99%) in the last search. SGNNV RNA2 was 1 nt longer than that of GGNNV, but they encoded proteins of the same size (338 amino acid residues). Thus, SGNNV and GGNNV are closely related to each other and belong to the RGNNV type (20, 23). In contrast, the sequence similarity between SGNNV and SJNNV (11; Iwamoto et al., submitted) was not comparatively high at either the nucleotide level (82% for RNA1 and 80% for RNA2) or the amino acid level (87% for protein A, 75% for protein B, and 81% for CP) (data not shown).

Surface probability plots (7) of SJNNV and SGNNV CP sequences (Fig. 1) showed that there were several positions that were possibly exposed on the surface of the CP molecule in each virus. In addition, the pair-wise plot elucidated that some of these positions had marked differences in surface probability scores between SJNNV and SGNNV. Such positions were mainly located in the N-terminal (positions 10 to 42) and C-terminal (positions 236 to 338 in SGNNV; this region corresponds to positions 238 to 340 in SJNNV) regions of CP. This indicates the presence of structural variations in the CPs in these regions. In contrast, the middle part of CP, especially from positions 86 to 219, is nearly identical in surface probability scores for SJNNV and SGNNV, suggesting that both of the CPs have similar structures in this region.

FIG. 1.

Pair-wise surface probability plot for SJNNV and SGNNV CPs. The surface probabilities for SJNNV and SGNNV CPs (340 and 338 amino acids, respectively) were analyzed by the method of Emini et al. (7). The amino acid positions were adjusted according to the ClustalW analysis data for the two CP sequences (data not shown), in which amino acid residues 236 and 237 in SJNNV CP were skipped in SGNNV CP. The SJNNV plot (black) was layered onto the SGNNV plot (red). Thus, some parts of the SGNNV plot were hidden by the SJNNV plot, indicative of identity in surface probability value.

Infectivity of the reassortant viruses in cultured fish cells.

For testing of the infectivity of transcripts from pSG1TK5 and pSG2TK13, E-11 cells were inoculated with equimolar mixtures of the transcripts, and virus infection was evaluated by immunofluorescence staining with anti-SJNNV antibody. After 24 h of incubation at 25°C, strong fluorescent signals were observed, indicative of virus multiplication in the E-11 cells, although the rate of stained cells was low (Fig. 2A, panel b). More stained cells were obtained when E-11 cells were transfected with viral RNAs purified from SGNNV virions (data not shown). This indicates that the in vitro transcripts used were not as infectious as virion RNAs. The reasons may be that the 3′ ends of the in vitro transcripts were not protected by the unknown factor that has been detected in virion RNAs (11) and/or that the in vitro transcripts had four extra nonviral nucleotides (AAUU) at the 3′ termini which were derived from the plasmid vectors.

FIG. 2.

Multiplication of parental and reassortant viruses in E-11 cells. (A) E-11 cells were transfected with a mixture of in vitro transcripts from pSJ1TK19 and pSJ2TK30 (a), pSG1TK5 and pSG2TK13 (b), pSJ1TK19 and pSG2TK13 (c), or pSG1TK5 and pSJ2TK30 (d). At 24 h posttransfection, viruses were detected by immunofluorescence staining with anti-SJNNV rabbit polyclonal antibody. (e) Mock-inoculated cells. Bar, 50 μm. (B) For each treatment, proteins were extracted from the E-11 cells (approximately 30,000 cells) at 24 h posttransfection with in vitro transcripts. Extracted proteins were separated by sodium dodecyl sulfate-12.5% polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane. SJNNV and SGNNV CPs in the transfected cells were detected by Western blot analysis using anti-SJNNV rabbit polyclonal antibody. Authentic SJNNV and SGNNV virion samples were also used as positive controls.

Western blotting analysis of the proteins extracted from the inoculated cells showed that the anti-SJNNV antibody cross-reacted with a protein showing similar mobility to that of CP from purified SGNNV virions as well as to that of SJNNV CP (Fig. 2B). These data support the idea that the anti-SJNNV antibody detected SGNNV CP as fluorescent signals, as shown in Fig. 2A, panel b. Interestingly, when transcripts from pSJ1TK19 (11) and pSG2TK13 were inoculated into E-11 cells, a high level of virus multiplication was also observed (Fig. 2A, panel c), like that from the inoculation of transcripts from pSJ1TK19 and pSJ2TK30 (Fig. 2A, panel a) (11) or from pSG1TK5 and pSG2TK13 (Fig. 2A, panel b). Similar virus multiplication was also obtained with transcripts from pSG1TK5 and pSJ2TK30 (Fig. 2A, panel d). When the supernatants of cell cultures infected with the reassortant viruses were treated with S7 nuclease, nuclease-resistant viral RNAs were detected by RT-PCR. In contrast, viral RNAs were not amplified by RT-PCR using S7 nuclease-treated RNA transcripts (data not shown). These results suggest that the reassortant viruses have encapsidation competence in E-11 cells. For convenience, progeny viruses derived from the in vitro transcripts of pSJ1TK19/pSJ2TK30, pSG1TK5/pSG2TK13, pSJ1TK19/pSG2TK13, and pSG1TK5/pSJ2TK30 were named SJ1/SJ2, SG1/SG2, SJ1/SG2, and SG1/SJ2, respectively.

Infectivity of the reassortant viruses in fish.

For examination of putative host-specificity determinants of SJNNV and SGNNV, the two kinds of reassortant viruses, SJ1/SG2 and SG1/SJ2, were inoculated into the fish from which they originated, and the infectivity was compared with that of parental viruses SJ1/SJ2 and SG1/SG2. Supernatants from the E-11 cell cultures infected with each of the four viruses were exposed to 1-day-old striped jack and sevenband grouper larvae. In striped jacks, a trace amount of fish died after mock inoculation and inoculation with either SG1/SG2 or SJ1/SG2 through 5 days postinoculation (Fig. 3). In contrast, SJ1/SJ2 and SG1/SJ2 started to kill striped jack larvae by 3 days after inoculation and killed all the fish in 5 days (Fig. 3). In the striped jack larvae killed by SG1/SJ2, viruses were localized in the brains, spinal cords, and retinas, which was indistinguishable from the case with SJ1/SJ2 infection (Fig. 4). No evidence for virus multiplication was obtained from the striped jacks inoculated with SG1/SG2 or SJ1/SG2 (Fig. 4). In the case of viral inoculation into sevenband grouper larvae, dead fish appeared at 2 days postinoculation when inoculated with SG1/SG2 or SJ1/SG2. However, SJ1/SJ2 or SG1/SJ2 inoculation did not affect the mortality of sevenband grouper larvae compared with mock inoculation (Fig. 3). SG1/SG2 and SJ1/SG2 viruses were similarly localized in the brains, spinal cords, and retinas of fish (Fig. 4). Virus multiplication in the larvae was also evaluated by RT-PCR. The total RNA was isolated from the inoculated larvae and used as a template for RT-PCR. As shown in Fig. 5, both RNA1 and RNA2 were amplified from the larvae that were affected by virus inoculation (Fig. 3 and 4). Viral RNAs were not detected in the larvae that showed a similar mortality rate to that of mock-inoculated fish. Collectively, these results indicate that SJ1/SJ2 and SG1/SJ2 specifically infected striped jack larvae and multiplied in the brains, spinal cords, and retinas, giving typical nervous necrotic phenotypes. In contrast, SG1/SG2 and SJ1/SG2 multiplied exclusively in sevenband groupers, with characteristics of viral nervous necrosis disease.

FIG. 3.

Cumulative mortality of striped jack and sevenband grouper larvae exposed to parental or reassortant viruses. Striped jack and sevenband grouper larvae were bath challenged with the culture supernatant of E-11 cells infected with SJ1/SJ2, SG1/SG2, SJ1/SG2, or SG1/SJ2, and dead fish were counted daily for 5 days. For striped jacks and sevenband groupers, data are represented as means ± standard deviations from three and two independent experiments, respectively.

FIG. 4.

Infectivity and localization of parental or reassortant viruses in striped jacks and sevenband groupers. Striped jack and sevenband grouper larvae were bath challenged with the culture supernatant of E-11 cells infected with SJ1/SJ2, SG1/SG2, SJ1/SG2, or SG1/SJ2. At 3 to 4 days postinoculation, 10 dead larvae were collected and sectioned. Viruses were detected by immunofluorescence staining with anti-SJNNV polyclonal antibody. Representative data for each treatment are shown here. In the case of striped jacks inoculated with SG1/SG2 or SJ1/SG2, living fish were used for immunofluorescence staining because all the inoculated larvae survived during the test periods.

FIG. 5.

Detection by RT-PCR of viral RNA segments in the striped jack (A) and sevenband grouper (B) larvae inoculated with parental or reassortant viruses. Total RNA was isolated from the striped jack or sevenband grouper larvae at 3 days postinoculation with SJ1/SJ2, SG1/SG2, SJ1/SG2, or SG1/SJ2. RT-PCR was performed with viral segment-specific primers (Table 1) and with extracted RNA as the template. Viral RNA segments were detected as the following PCR products: SJNNV RNA1 (1,748 bp) and RNA2 (483 bp) and SGNNV RNA1 (1,094 bp) and RNA2 (503 bp). M, molecular size markers; a lambda DNA-StyI digest and a φX174 DNA-HaeIII digest were electrophoresed beside the RNA1- and RNA2-specific products, respectively.

DISCUSSION

We have successfully constructed a genetic engineering system in SGNNV and tested the infectivity of reassortant viruses between SGNNV and SJNNV. The reassortant SG1/SJ2 infected striped jacks as the parental SJ1/SJ2 did. Conversely, SJ1/SG2 infected sevenband groupers, similar to the case with SG1/SG2. For both reassortants, no difference in symptom phenotypes was observed when compared with the parental viruses. These results indicate that, for SJNNV and SGNNV, host specificity is determined by RNA2 and/or its encoded CP, even though their RNA2 and CP sequences share as much as 80 and 81% homology, respectively. This is the first report to identify the viral factors that determine host specificity for members of the family Nodaviridae.

The pair-wise surface probability plot for SJNNV and SGNNV CPs (Fig. 1) showed that surface probability scores were significantly different in the N- and C-terminal regions of CP, indicative of structural differences in these regions. Interestingly, similar results were obtained when CP sequences from the four types of betanodaviruses (20), SJNNV, SGNNV, BFNNV, and TPNNV, were used for a pair-wise surface probability plot (unpublished data). The N-terminal regions of SJNNV and SGNNV CPs contained many positively charged amino acid residues, and these basic residues seemed to contribute to the good surface probability scores. As demonstrated for alphanodaviruses, such basic residues may play a role in neutralizing the negative charge of viral RNAs in virions and also may be involved in the encapsidation process (15, 24). Thus, the putative protruding positions in the C-terminal region, which had different surface probability scores for SJNNV and SGNNV, may be good candidates for host specificity determinants. Cryo-electron microscopy analysis and folding motif analysis of a betanodavirus, malabaricus grouper nervous necrotic virus, suggested that the C-terminal 122 residues (Glu²¹⁷ to Asn³³⁸) of CP are displayed on the surface of the virion (25). These 122 residues cover the CP region (Asn²³⁶ to Asn³³⁸ in SGNNV and Thr²³⁸ to Asn³⁴⁰ in SJNNV) that contained the putative protruding positions with different surface probability scores for SJNNV and SGNNV (Fig. 1). Therefore, the variable regions in the C termini of SJNNV and SGNNV CPs probably are displayed on the surfaces of the virions as well as on the CP molecules. Another notable report showed that the partial nucleotide sequence of betanodavirus RNA2, encoding the C-terminal half of CP, is suitable for the classification of betanodaviruses (20). With this nucleotide region, betanodaviruses can be classified into the four types, namely SJNNV, RGNNV, BFNNV, and TPNNV. Interestingly, these four types of viruses have marked host specificities (20). Collectively, these direct and indirect lines of evidence suggest that some of the C-terminal protruding domains of a betanodavirus CP distinguish between host fish. The mechanisms by which these domains direct host specificity are still unknown. However, one possible scenario is a direct interaction of the protruding domain with a putative host-specific receptor or with other host factors.

We could not rule out the possibility that RNA2 and/or CP might determine host specificity by controlling viral RNA synthesis. In FHV-infected cultured Drosophila cells, RNA3 accumulates predominantly early during infection, and at later times its synthesis is inhibited by the replication of RNA2. Furthermore, RNA2 replication depends on RNA3 but not on either of its translation products (6, 27). Generally, optimal virus propagation requires coordinated replication among the segments, since a full complement of genomic segments is needed for viral infectivity. So far, not much is known about segment synthesis in betanodaviruses. Nevertheless, mutual transcriptional regulation between RNAs 2 and 3 could also occur in betanodaviruses, and coordinated viral RNA replication might be disturbed specifically in fish (not in cultured cells) by RNA2-host incompatibility (Fig. 3 to 5). Thus, in this model, RNA2 should play such a role with the help of a host-specific factor.

Finally, a protein B function could contribute to host specificity determination by RNA2 in betanodaviruses. Protein B was immunologically detected in SJNNV-infected E-11 cells (Iwamoto et al., submitted). In FHV, protein B suppresses posttranscriptional gene silencing, which is believed to be an important defense system of hosts against virus infection (14). As discussed above, RNA3 synthesis in betanodaviruses might be regulated by RNA2, as in alphanodaviruses. Hence, RNA2-host incompatibility possibly disturbs protein B synthesis or its timing in fish, which prevents viruses from producing a host defense suppressor at appropriate times.

Further investigations are required to elucidate the most probable mechanisms for host specificity determination by RNA2. One useful experiment is a time-course Northern blot analysis that uses RNA1 and -2 probes on the RNA samples prepared from the fish larvae inoculated with reassortants or parental viruses. Detection (or missing) of RNAs 1 to 3 or homodimers or heterodimers of viral RNAs (1) may give some insight into the viral multiplication step that is affected by segment substitution.

We demonstrated that RNA2 and/or CP are host specificity determinants for SJNNV and SGNNV. However, it cannot be ruled out that RNA1, protein A, and/or protein B might also determine host specificity under other experimental conditions. In this study, striped jacks and sevenband groupers were kept at 23°C after virus inoculation to verify infectivity. In cultured fish cells, SJNNV gave good virus yields at 20 and 25°C, but showed poor yields at 15 and 30°C. In the case of SGNNV, good viral multiplication occurred at 30°C as well as at 20 and 25°C (10). The mechanism by which this temperature sensitivity occurs remains to be addressed. However, when this temperature sensitivity is accounted for by the enzyme activity of viral RNA-dependent RNA polymerase at those temperatures, RNA1 might also affect host specificity at 30°C in addition to the major contribution of RNA2. That is, at 30°C the SJ1/SG2 reassortant could have no or a low level of infectivity in sevenband groupers because of the possible poor RNA-dependent RNA polymerase activity of SJNNV protein A at this high temperature.