Molecular Phylogenetics and Evolution 83 (2015) 137–142
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev
Divergence and codon usage bias of Betanodavirus, a neurotropic pathogen in fish Mei He, Chun-Bo Teng ⇑ College of Life Science, Northeast Forestry University, Harbin 150040, China
a r t i c l e
i n f o
Article history: Received 4 September 2014 Revised 25 November 2014 Accepted 30 November 2014 Available online 10 December 2014 Keywords: Betanodavirus Divergence Codon usage bias Viral nervous necrosis
a b s t r a c t Betanodavirus is a small bipartite RNA virus of global economical significance that can cause severe neurological disorders to an increasing number of marine fish species. Herein, to further the understanding of the evolution of betanodavirus, Bayesian coalescent analyses were conducted to the time-stamped entire coding sequences of their RNA polymerase and coat protein genes. Similar moderate nucleotide substitution rates were then estimated for the two genes. According to age calculations, the divergence of the two genes into the four genotypes initiated nearly simultaneously at 700 years ago, despite the different scenarios, whereas the seven analyzed chimeric isolates might be the outcomes of a single genetic reassortment event taking place in the early 1980s in Southern Europe. Furthermore, codon usage bias analyses indicated that each gene had influences in addition to mutational bias and codon choice of betanodavirus was not completely complied with that of fish host. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Viral nervous necrosis (VNN), also termed as viral encephalopathy and retinopathy (VER), is a highly infectious disease that can cause serious damage typical of extensive vacuolation to the piscine central nervous system (Glazebrook et al., 1990; Yoshikoshi and Inoue, 1990). To date, more than 50 marine teleosts, as well as several freshwater species, have been found susceptible (Bandín and Dopazo, 2011; Vendramin et al., 2014). Infected fish, larvae and juveniles in particular, exhibit various clinical signs such as anorexia, darkened coloration and erratic swimming behavior, and suffer mass mortalities that may reach up to 100% (Munday et al., 2002). Since its first emergence in the 1980s, VNN has been a major impediment to mariculture, outbreaks of which have been reported from all over the world accompanied with significant economic losses (Nakai et al., 2009). The aetiological agent of VNN is a small naked RNA virus assigned to the genus Betanodavirus in the family Nodaviridae, distantly related to insect-infecting Alphanodavirus (Ball et al., 2000). It has a single-stranded RNA genome composed of two genetic molecules designated as RNA1 and RNA2 encapsulated by the icosahedral capsid. Being positive-sense, both of them possess the methylated caps but lack the poly (A) tails (Mori et al., 1992). RNA1, the larger segment (3.1 kb), encodes the RNA-dependent ⇑ Corresponding author. Fax: +86 451 8219 1784. E-mail address: [email protected] (C.-B. Teng). http://dx.doi.org/10.1016/j.ympev.2014.11.016 1055-7903/Ó 2014 Elsevier Inc. All rights reserved.
RNA polymerase (RdRp, or protein A) (Nagai and Nishizawa, 1999), while RNA2, the smaller one (1.4 kb), encodes the coat protein (Cp) (Nishizawa et al., 1995). During viral replication, a subgenomic RNA3 is synthesized from the 30 terminus of RNA1 and expresses two small proteins, B1 and B2, with regulatory functions (Chen et al., 2009; Fenner et al., 2006; Iwamoto et al., 2005; Su et al., 2009). Based on phylogenetic analysis of the variable T4 region of the Cp gene, betanodaviruses can be categorized into four genotypes (Nishizawa et al., 1997): striped jack nervous necrosis virus (SJNNV), tiger puffer nervous necrosis virus (TPNNV), barfin flounder nervous necrosis virus (BFNNV) and red-spotted grouper nervous necrosis virus (RGNNV), with SJNNV being the type species. This taxonomic classification holds true for the RdRp gene, despite the difference in genetic relationships (Toffolo et al., 2007). Moreover, a single turbot nodavirus (TNV) has been proposed by Johansen et al. (2004) to be the fifth genotype. The four major genotypes differ in various aspects including host range, geographic distribution and optimum temperature (Nakai et al., 2009). SJNNV is endemic to Japan affecting certain fish species; however, it also exists (mostly as reassortants) in Southern Europe (Panzarin et al., 2012). TPNNV is so far only found in Japan from Tiger puffer (Takifugu rubripes) and Japanese flounder (Paralichthys olivaceus) (Nishizawa et al., 1997). BFNNV prefers coldwater teleosts in Japan and the North Atlantic. In contrast, RGNNV possesses the broadest host spectrum covering various kinds of warm-water finfish and the widest geographical range involving
138
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
Asia, Australia, North America, and the Mediterranean (Bandín and Dopazo, 2011). As is common in multipartite viruses, reassortment is employed by betanodaviruses to drive their evolution. Interestingly, all reassortants identified to date are generated from either combination of genomic segments between SJNNV and RGNNV taking place in the Mediterranean basin and the connected Iberian Atlantic region (Olveira et al., 2009; Panzarin et al., 2012; Toffolo et al., 2007; Vendramin et al., 2014). Moreover, recently, based on the dated partial sequences, the rates of nucleotide substitution and the times to the most recent common ancestor (TMRCAs) of the RGNNV group in Southern Europe have been estimated, which revealed a higher rate and a younger age for the RdRp gene there. However, the differences were not significant as the 95% highest probability density (HPD) ranges for the two segments overlapped (Panzarin et al., 2012). Here, to broaden the knowledge of the molecular epidemiology and evolutionary dynamics of betanodavirus, Bayesian coalescent method was applied to the time-stamped complete coding sequences of the RdRp and Cp genes sampled worldwide. Moreover, to better understand the processes governing their evolution, codon usage bias of each gene was also examined.
2. Materials and methods Full-length coding sequences of the RdRp and Cp genes were retrieved from GenBank and aligned with CLUSTAL W (Thompson et al., 1997). Dataset compilation, Bayesian estimates, and codon usage bias analyses were conducted as described previously (He et al., 2013). Details of the betanodavirus isolates analyzed, including collection dates supplemented via literature and Genbank Accession numbers, were listed in Table S1. When the Bayesian Markov chain Monte Carlo (MCMC) method in BEAST v1.7.4 (Drummond et al., 2012) was applied to each panel, the four clock models (strict, exponential, lognormal and random local) as well as the five demographic models (constant, exponential, expansion, logistic and Bayesian skyline plot) were compared (He et al., 2014a). Specifically, the strict clock model was evaluated by the coefficient of variation (CoV), a scale of the degree of clocklikeness of the data (Drummond et al., 2006). Thus, it was rejected for RdRp in that CoV did not encompass zero. On the contrary, it was accepted for Cp since the lower 95% HPD of CoV approached zero. For RdRp, among the three relaxed clock models, the lognormal one was utilized as the other two failed to describe the evolutionary dynamics. Moreover, as most demographic models yielded similar results for Cp, the logistic growth one was chosen owing to its best reliability and convergence for RdRp. Then, independent analyses for 10–25 million MCMC iterations were combined (10% burn-in) to assure effective sample size (>200). Isolate information (name/year/origin/accession) was also included in each maximum clade credibility (MCC) tree. Two indices of codon usage bias, the effective number of codons used by a gene (Nc) and the frequency of G + C at the synonymous 3rd codon position (GC3S) were examined for RdRp and Cp. The overall variation in the codon usage was reflected by the relative synonymous codon usage (RSCU), the value of which higher than 1.0 indicates that the corresponding codon is adopted more frequently than expected, and vice versa. Besides, correspondence analysis, a multivariate statistical technique that yields orthogonal axes to account for variations, was conducted to identify major trends in amino acid usage. All aforementioned analyses were done by CodonW 1.4.4 (http://codonw.sourceforge.net). Moreover, an alphanodavirus Flock house virus (FHV) and a fish host Dicentrarchus labrax with more codons available (http://www.kazusa.or.jp/ codon/) were selected for comparisons of RSCU and GC content.
In addition, to infer the genetic relationships of the chimeras (30 RG/SJ and 1 SJ/RG), alignments of their partial RdRp (867 nt, referred to CDS positions 154-1020 of SJNNV/AB056571) and Cp (504 nt, CDS positions 357-860 of SJNNV/AB056572) sequences were created for phylogenetic analyses, respectively. Each Maximum Likelihood (ML) tree was drawn by MEGA 5.1 (Tamura et al., 2011) with 1000 bootstrap replicates under the best-fit nucleotide substitution model (TN93 + G for RdRp and T92 + G for Cp). 3. Results 3.1. Nucleotide substitution rates of the RdRp and Cp genes As listed in Table 1, when entire coding sequences of the RdRp gene from 49 betanodavirus isolates spanning 19 years, consisting of 32 RG, 14 BF, 1 TP and 2 SJ-type viruses, were subjected to Bayesian analysis, the average rate of nucleotide substitution was estimated to be 3.60 104 subs/site/year, with the 95% HPD values varying from 1.40 104 to 6.18 104. A similar mean rate at 3.69 104 (2.31 104–5.06 104) subs/site/year was obtained for the Cp gene based on 73 isolates spanning 21 years, composed of 54 RG, 8 BF, 1 TP and 10 SJ-type viruses. Substitution rates were also calculated for the two genes of RGNNV, the largest genotype with more complete sequence data available. The reassortants and MnNNV-12-06 (Fig. 1A) being excluded, the average rate of RdRp from 24 isolates was 4.28 104 (3.35 105–8.79 104) subs/site/year, which was insignificantly higher than that of Cp from 52 isolates at 3.79 104 (2.01 104–5.60 104) subs/site/year. In fact, rate difference was still minor when the Cp panel was truncated to be identical to the RdRp panel (data not shown). 3.2. Phylogenetic relationships of betanodaviruses The MCC trees (Fig. 1) calculated for the RdRp and Cp genes of betanodavirus isolates collected worldwide confirmed the categorization of four major genotypes (SJ, TP, BF and RG), the difference in genetic relationships involving BFNNV, as well as the reassortments between SJNNV and RGNNV. Notably, the Atlantic cod nervous necrosis virus (ACNNV) circulating in Canada described by Gagnè et al. (2004) could be classified as a subtype of BFNNV (Fig. 1B). Within the large RGNNV group, the isolates analyzed here clustered into 3 (a–c, Fig. 1A) and 6 (a–e, Fig. 1B) well-supported subclades (posterior probability value >0.9) based on their RdRp and Cp sequences, respectively. The subtype a of Cp, for example, could be further divided as well (1–7, Fig. 1B). According to the arbitrary classification, incongruent topologies were observed for GPNNV1108-P.Langkawi-1 sampled from Malaysia. In the RdRp
Table 1 Details of datasets and estimates of the two betanodaviral genes. Parametera
RdRp
Cp
No. of sequences Time span Substitution model Molecular clock Demographic model Mean substitution rate (104) 95% HPD rate (104) Mean TMRCA 95% HPD year
49 1993–2012
73 1991–2012 GTR + G + I
Lognormal
Strict Logistic
3.60 1.40–6.18 695 267–1238
3.69 2.31–5.06 704 446–990
a HPD: highest probability density; TMRCA: time to the most recent common ancestor.
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
139
Fig. 1. Maximum clade credibility phylogenies of the RdRp (A) and Cp (B) genes of betanodavirus. The trees are scaled to time generated under the better molecular clock and demographic models (Table 1). Nodes correspond to mean TMRCAs. Age estimates with the 95% highest probability density values are given for some nodes. Isolate information (name/year/origin/accession) and genotype classification are shown on the right. Subtypes of RGNNV designated here with lowercase letters are labeled above their nodes. In the Cp tree, the seven clusters within the subtype a supported by >0.9 posterior probability value are also indicated on the right with Arabic numerals. Two special isolates cited in the discussion were marked.
140
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
tree, it was clustered with three Malaysian isolates in the b clade; however, in the Cp tree, it fell into the a1 clade distinct from the c clade formed by the three Malaysian isolates. This might be a reflection of an intra-genotype reassortment event. 3.3. Divergence of the RdRp and Cp genes The mean TMRCA of the RdRp gene was 695 (95% HPD: 267– 1238) years, which was nearly simultaneous with that of the Cp gene at 704 (446–990) years. Therefore, both of the two initial bifurcation events of the analyzed betanodavirus isolates might have taken place at the onset of the 14th century. However, the divergence scenarios of the two genes into the four genotypes were quite different, as revealed by the time-scaled MCC phylogenies (Fig. 1). According to the RdRp tree (Fig. 1A), the first event occurred between the singular progenitor of RGNNV and the common ancestor shared by the other three genotypes. Only four decades later, the SJ-type virus diverged out; however, subsequently, it took the precursor of TPNNV about four and a half centuries to branch off from that of BFNNV. In contrast, seen from the Cp phylogram (Fig. 1B), the primary event was that the ancestor of RGNNV and BFNNV separated from that of SJNNV and TPNNV. Coincidentally, the two secondary events resulting in the four genotypes both happened 140 years later. When it came to the divergence within each genotype (except TPNNV which has only one isolate fully sequenced so far), differences were also observed for RGNNV, as reflected by the dissimilar TMRCA estimates for the two genes (Fig. 1). Even without regard to DL-389-I96, a SJ/RG type chimera (Fig. S1; Vendramin et al., 2014) more divergent from the other RGNNV isolates, there was still a 64-year age gap between the RdRp and Cp MRCAs (85 vs. 149), despite the partially overlapping HPD intervals (36–147 vs. 92– 213). Interestingly, if MnNNV-12-06 was excluded, which was clustered with the reassortants and did not diverged out until 40 years ago in the RdRp tree, the age estimates were much similar: 85 (36–147) vs. 91 (57–129). For the other two genotypes, timing results based on the two genes were also similar. The last common ancestor of SJNNV was just in its sixties, whereas that of the BFNNV subgroup composed of the Japanese and Norwegian isolates was around forty years old (Fig. 1). Currently, as no RdRp sequence was available for the Canadian BFNNV isolates, it was unknown whether their RdRp ancestor has diverged out nearly one and a half century ago as the Cp one (Fig. 1B). In addition, TMRCAs were also estimated for the seven RG/SJ type reassortants, which suggested that their divergence began in the 1980s (Fig. 1). 3.4. Codon usage bias of each betanodaviral gene To measure codon usage bias of each betanodaviral gene, Nc and GC3S were calculated. As Nc can take values from 20 for extreme bias when only a single synonymous codon is adopted, to 61 for no bias when all synonymous codons are equally used (Wright, 1990), the high Nc values here (>50, Fig. 2) indicated that the degree of codon usage bias was slight in each gene. When Nc was plotted against GC3S (Fig. 2A and B), the genotypes/subtypes exhibited different distribution patterns. This was more pronounced in the Cp plot (Fig. 2B), suggesting different selection constraints on the genotypes/subtypes, particularly on their RNA2 segments. Interestingly, although the reassortants possessed SJ-type Cp genes, their points were all clustered with the RG points but not the other SJ points (Fig. 2B), supporting the observation based on sequence comparison that the acquired SJ segments have changed to some extent toward the replaced RG segments (Olveira et al., 2009).
Fig. 2. Nc-plots of the RdRp (A) and Cp (B) genes of betanodavirus. Nc (the effective number of codons) vs. GC3S (the GC content at the synonymous third codon position) is plotted for each betanodaviral isolate. The four genotypes are distinguished with different forms and colors as indicated on the right. The continuous curve represents the expected Nc values under assumptions of no selection other than GC composition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Moreover, the Cp points of the BFNNV subgroup circulating in Japan and Norway were clustered just below the curve of the expected values, reflecting no selection other than GC composition. In contrast, other points were located away from the curve, implying influences aside from mutational bias (He et al., 2014b). Indeed, Nc of RdRp was in significant correlation with GRAVY, an index of protein hydrophobicity (R = 0.64, P < 0.01), so was Nc of Cp (R = 0.78, P < 0.01). This factor was also recommended by the correspondence analyses to be a major contributor to the amino acid usage variation in that the first axis, the principal trend, was significantly (P < 0.01) correlated with GRAVY (R = 0.62 for RdRp and R = 0.70 for Cp). In addition, RSCU was calculated to determine the overall variation in the codon usage. From the values, a predominance of C over U was observed in many synonymous 3rd codon positions (Table S2). Interestingly, in respect of codon preference and 3rd letter GC content, it looked like that betanodavirus (GC3rd 0.55) was in a transitional stage between a kin insect-infecting alphanodavirus (FHV, 0.47) and a host fish species (D. labrax, 0.63) (http://www.kazusa.or.jp/codon/), with more resemblance to the latter (Table S2).
4. Discussion According to Bayesian coalescent analyses applied to the dated entire coding sequences of betanodavirus isolates collected worldwide, the divergence of the RdRp and Cp genes into the four genotypes not only initiated nearly simultaneously but progressed at a similar moderate overall rate. However, it should be pointed out that, as reflected by the wider 95% HPD values (Fig. 1), age estimates for the long branches were less reliable than those for the short branches, especially when the two genes were subjected to strong purifying selection (Panzarin et al., 2012; Toffolo et al.,
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
2007) that could result in underestimation of the lengths of the long branches (Wertheim et al., 2013). Although there are expected to be some variations in the timing results of the primary bifurcation events when more complete sequence data are analyzed, it could be safely concluded that betanodavirus, even the four genotypes, had emerged long before the recognition of VNN. The nucleotide substitution rates and the TMRCAs of the RGNNV strains from the Indo-Pacific and Southern Europe were also calculated, which differed from those estimated for the RGNNV group in Southern Europe (Panzarin et al., 2012), especially the mean rate for the Cp gene (3.79 104 vs. 2.42 104) and the average age for the RdRp gene (85 vs. 188). Although not significant given the overlap between the HPD intervals (2.01 104–5.60 104 vs.1.07 104–3.99 104 as well as 36–147 vs. 79–328), these differences could be due to panel composition. As evolutionary rates can vary among genotypes/subtypes, which has been demonstrated previously (He et al., 2013, 2014b), it was possible that Cp of the Southern European RGNNV group evolved relatively slower, whereas since the IX and X clusters in their RdRp tree, which were outgroup to the analyzed subtypes here (a–c, Fig. 1A), were not covered in our dataset, it was certain that a much older age was obtained for RdRp in their analysis. In general, segmented RNA viruses achieve their rapid evolution via processes including mutation, reassortment and recombination (Worobey and Holmes, 1999). Currently, no significant signal for recombination was detected in the available betanodaviral sequences by the multiple approaches as applied before (He et al., 2012). It seemed that betanodaviruses preferred the former two ways to generate their genetic diversity and adapt to a new environment. Intriguingly, all inter-genotype reassortants so far were identified in Southern Europe generated from the exchange of genomic segments between SJNNV and RGNNV, reflecting a special niche there. Of note, it did not require frequent exchange to establish the large RG/SJ population. According to the MCC trees (Fig. 1), the seven chimeras might be the outcomes of a single reassortment event taking place in the early 1980s. In fact, inferred from the phylogenetic relationships based on the partial sequences (Fig. S1), all but one RG/SJ reassortants sampled to date (29 of 30, except 28.2005), despite their diverse origins, were likely to be derived from this event. Moreover, intra-genotype reassortment might occur as well, as exemplified by GPNNV1108-P.Langkawi-1 whose RdRp and Cp genes belonged to distinct RGNNV subtypes (Fig. 1). Certainly, reassortment complicated genetic relationships and genotype classification of betanodaviruses, which might render an explanation for the observed differences in the dated divergence among genotypes as well as within RGNNV. It was possible that BFNNV, which exhibited incongruent topologies in the MCC trees (Fig. 1), was a SJ/RG type chimera born in an early event that occurred a couple of centuries ago. Interestingly, BFNNV was once isolated in France during an outbreak at low temperature (Thiéry et al., 2004) from European sea bass (D. labrax) that is susceptible to both RG/SJ and SJ/RG reassortants (Panzarin et al., 2012; Vendramin et al., 2014). Moreover, MnNNV-12-06 possessed an old type Cp gene but a young type RdRp gene (Fig. 1), which might be due to an intra-genotype reassortment event. Another intricacy was the correlation between genetic relationships and geographical location, particularly within the RGNNV genotype. As could be seen in Fig. 1B, isolates sampled from different areas could fall into the same subcluster (a1, a2, a4, and a5), whereas distinct groups were circulating in the same country (e.g., a1, a7 and c in Malaysia). This might be the outcome of multiple factors such as commerce of infected stocks, transfer of wild carriers and parallel/convergent evolution (Dalla Valle et al., 2001; Toffolo et al., 2007). A definite example of the first factor
141
was the Israeli isolate LC121100-IL. It was identified in Asian sea bass (Lates calcarifer) imported from Thailand as larvae (Ucko et al., 2004) and was indeed clustered within a Southeast Asian group (in a7). Of note, betanodaviruses, including reassortants, were also found in marine invertebrates (Gomez et al., 2008; Panzarin et al., 2012), which might further perplex the origin and spread of the virus. As mentioned above, the four genotypes differ in optimum growth temperature. In cultured cells, RG, SJ, TP and BF exhibited higher titers at 25–30 °C, 20–25 °C, 20 °C and 15–20 °C, respectively (Iwamoto et al., 2000). An interesting correlation was then observed between the optimal temperature and the divergence of the RdRp gene (Fig. 1A): seen from the line of BFNNV, the earlier the genotype branched off, the warmer the preferred temperature was. Indeed, experiments using chimeric viruses (Hata et al., 2010; Panzarin et al., 2014) have demonstrated the major role RNA1 plays in determining temperate sensitivity via the region 84– 1419 encoding the first 445 amino acid residues of RdRp, particularly the region 1088–1419. Actually, the secondary structures of the amino acid region 335–445 of RdRp predicted for the four genotypes were different (data not shown). Furthermore, codon usage bias, an important indicator of the forces shaping the evolution of RNA viruses (Jenkins and Holmes, 2003), was examined for each betanodaviral gene. From the comparison between the actual distribution and the expected one in the Nc-plot, it was visible that each gene had additional codon usage bias independent of mutation pressure. Perhaps, as suggested by the statistical analyses, one driving force of the nucleotide changes was the hydropathy level of the amino acid sequence, on which the stability of the protein is largely dependent. This might be of consequence in the evolution of betanodavirus, for instance, adaptation to low temperature. Interestingly, from another comparison between virus and host in terms of codon choice, it looked like that betanodavirus had not fully conformed to fish yet, suggesting that its codon usage bias might be less affected by the host’s tRNA abundance. At the expense of high efficiency and accuracy in translation, such disobedience to the cellular codon preference may be subtly utilized to fine-tune translation kinetics to assure proper protein folding, as may happen in the hepatitis A virus (Pintó et al., 2012) and the vesicular stomatitis viruses (Liang et al., 2014). Indeed, in the middle of the amino acid region 83–216 that probably folds into the inner b-sandwich shell of the T = 3 capsid of betanodavirus (Tang et al., 2002), the eight residues ‘AL(I/F)QATRGA’ from positions 140 to 147 associated with an a-helix structure are encoded by non-abundant cellular codons (Fig. S1). Perhaps, this cluster is strategically located there to slow down ribosome traffic rate to avert interference with the folding of the nascent peptide of Cp. As noted previously (He et al., 2014a), codon usage information may have practical implications for viral gene expression control and vaccine design. Since in several amino acids, such as Arg and Val, betanodaviruses and teleosts exhibited different codon preferences (Table S2), optimizing the favor of the virus to that of the host may be able to elevate gene expression level thus achieve sufficient viral proteins for immunity generation (Haas et al., 1996). On the contrary, deoptimizing the synonymous codons with those non-preferred by both of them, such as codons ended with ‘UA’ (Table S2), may be able to impair gene expression and virulence thus be used to develop attenuated vaccines (Mueller et al., 2006).
5. Conclusion In our study, tip calibration revealed that in the divergence of the betanodaviral RdRp and Cp genes, the initiation dates and the
142
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
overall rates were similar but the diversifying scenarios were different, while codon usage bias analyses suggested that codon choice of betanodavirus was not fully complied with that of fish host, which might be utilized to regulate translation kinetics. Our findings thus further the understanding of this severe neurotropic pathogen that has become an obstacle of marine aquaculture and provide useful information for future studies such as vaccine design. Acknowledgments We are deeply grateful to the reviewers for their helpful comments for improving our manuscript. This work was supported by the Fundamental Research Funds for the Central Universities of China (No. DL13EA06). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ympev.2014. 11.016. References Ball, L.A., Hendry, D.A., Jonshon, J.E., Ruckert, R.R., Scotti, P.D., 2000. Family Nodaviridae. In: Van Rengenmortel, M.H.V., Fauquet, C.M., Bishop, D.H.L., Carstens, E.B., Estes, M.K., Lemon, S.M., Maniloff, J., Mayo, M.A., McGeoch, D.J., Pringle, C.R., Wickner, R.B. (Eds.), Virus Taxonomy: Seventh Report of the International Committee on Taxonomy of Viruses. Academic Press, San Diego, pp. 747–755. Bandín, I., Dopazo, C.P., 2011. Host range, host specificity and hypothesized host shift events among viruses of lower vertebrates. Vet. Res. 42, 67. Chen, L.J., Su, Y.C., Hong, J.R., 2009. Betanodavirus non-structural protein B1: a novel anti-necrotic death factor that modulates cell death in early replication cycle in fish cells. Virology 385, 444–454. Dalla Valle, L., Negrisolo, E., Patarnello, P., Zanella, L., Maltese, C., Bovo, G., Colombo, L., 2001. Sequence comparison and phylogenetic analysis of fish nodaviruses based on the coat protein gene. Arch. Virol. 146, 1125–1137. Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88. http://dx.doi.org/10.1371/ journal.pbio.0040088. Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. Fenner, B.J., Thiagarajan, R., Chua, H.K., Kwang, J., 2006. Betanodavirus B2 is an RNA interference antagonist that facilitates intracellular viral RNA accumulation. J. Virol. 80, 85–94. Gagné, N., Johnson, S.C., Cook-Versloot, M., MacKinnon, A.M., Olivier, G., 2004. Molecular detection and characterization of nodavirus in several marine fish species from the northeastern Atlantic. Dis Aquat Organ 62, 181–189. Glazebrook, J.S., Heasman, M.P., de Beer, S.W., 1990. Picorna-like viral particles associated with mass mortalities in larval barramundi, Lates calcarifer Bloch. J. Fish Dis. 13, 245–249. Gomez, D.K., Baeck, G.W., Kim, J.H., Choresca Jr., C.H., Park, S.C., 2008. Molecular detection of betanodaviruses from apparently healthy wild marine invertebrates. J. Invertebr. Pathol. 97, 197–202. Haas, J., Park, E.C., Seed, B., 1996. Codon usage limitation in the expression of HIV-1 envelope glycoprotein. Curr. Biol. 6, 315–324. Hata, N., Okinaka, Y., Iwamoto, T., Kawato, Y., Mori, K., Nakai, T., 2010. Identification of RNA regions that determine temperature sensitivities in betanodaviruses. Arch. Virol. 155, 1597–1606. He, M., He, C.Q., Ding, N.Z., 2012. Natural recombination between tobacco and tomato mosaic viruses. Virus Res. 163, 374–379. He, M., Ding, N.Z., He, C.Q., Yan, X.C., Teng, C.B., 2013. Dating the divergence of the infectious hematopoietic necrosis virus. Infect. Gen. Evol. 18, 145–150. He, M., An, T.Z., Teng, C.B., 2014a. Evolution of mammalian and avian bornaviruses. Mol. Phylogenet. Evol. 79, 385–391. He, M., Yan, X.C., Liang, Y., Sun, X.W., Teng, C.B., 2014b. Evolution of the viral hemorrhagic septicemia virus: divergence, selection and origin. Mol. Phylogenet. Evol. 77, 34–40. Iwamoto, T., Nakai, T., Mori, K., Arimoto, M., Furusawa, I., 2000. Cloning of the fish cell line SSN-1 for piscine nodaviruses. Dis. Aquat. Organ. 41, 81–89. Iwamoto, T., Mise, K., Takeda, A., Okinaka, Y., Mori, K., Arimoto, M., Okuno, T., Nakai, T., 2005. Characterization of Striped jack nervous necrosis virus subgenomic
RNA3 and biological activities of its encoded protein B2. J. Gen. Virol. 86, 2807– 2816. Jenkins, G.M., Holmes, E.C., 2003. The extent of codon usage bias in human RNA viruses and its evolutionary origin. Virus Res. 92, 1–7. Johansen, R., Sommerset, I., Torud, B., Korsnes, K., Hjortaas, M.J., Nilsen, F., Nerland, A.H., Dannevig, B.H., 2004. Characterization of nodavirus and viral encephalopathy and retinopathy in farmed turbot, Scophthalmus maximus (L.). J. Fish Dis. 27, 591–601. Liang, Y., He, M., Teng, C.B., 2014. Evolution of the vesicular stomatitis viruses: divergence and codon usage bias. Virus Res. 192, 46–51. Mori, K., Nakai, T., Muroga, K., Arimoto, M., Mushiake, K., Furusawa, I., 1992. Properties of a new virus belonging to nodaviridae found in larval striped jack (Pseudocaranx dentex) with nervous necrosis. Virology 187, 368–371. Mueller, S., Papamichail, D., Coleman, J.R., Skiena, S., Wimmer, E., 2006. Reduction of the rate of poliovirus protein synthesis through large-scale codon deoptimization causes attenuation of viral virulence by lowering specific infectivity. J. Virol. 80, 9687–9696. Munday, B.L., Kwang, J., Moody, N., 2002. Betanodavirus infections of teleost fish: a review. J. Fish Dis. 25, 127–142. Nagai, T., Nishizawa, T., 1999. Sequence of the non-structural protein gene encoded by RNA1 of striped jack nervous necrosis virus. J. Gen. Virol. 80 (Pt 11), 3019– 3022. Nakai, T., Mori, K., Sugaya, T., Nishioka, T., Mushiake, K., Yamashita, H., 2009. Current knowledge on Viral Nervous Necrosis (VNN) and its causative betanodaviruses. Isr. J. Aquacult.-Bamid. 61, 199–207. Nishizawa, T., Mori, K., Furuhashi, M., Nakai, T., Furusawa, I., Muroga, K., 1995. Comparison of the coat protein genes of five fish nodaviruses, the causative agents of viral nervous necrosis in marine fish. J. Gen. Virol. 76 (Pt 7), 1563– 1569. Nishizawa, T., Furuhashi, M., Nagai, T., Nakai, T., Muroga, K., 1997. Genomic classification of fish nodaviruses by molecular phylogenetic analysis of the coat protein gene. Appl. Environ. Microbiol. 63, 1633–1636. Olveira, J.G., Souto, S., Dopazo, C.P., Thiéry, R., Barja, J.L., Bandín, I., 2009. Comparative analysis of both genomic segments of betanodaviruses isolated from epizootic outbreaks in farmed fish species provides evidence for genetic reassortment. J. Gen. Virol. 90, 2940–2951. Panzarin, V., Fusaro, A., Monne, I., Cappellozza, E., Patarnello, P., Bovo, G., Capua, I., Holmes, E.C., Cattoli, G., 2012. Molecular epidemiology and evolutionary dynamics of betanodavirus in southern Europe. Infect. Gen. Evol. 12, 63–70. Panzarin, V., Cappellozza, E., Mancin, M., Milani, A., Toffan, A., Terregino, C., Cattoli, G., 2014. In vitro study of the replication capacity of the RGNNV and the SJNNV betanodavirus genotypes and their natural reassortants in response to temperature. Vet. Res. 45, 56. Pintó, R.M., D’Andrea, L., Pérez-Rodriguez, F.J., Costafreda, M.I., Ribes, E., Guix, S., Bosch, A., 2012. Hepatitis A virus evolution and the potential emergence of new variants escaping the presently available vaccines. Future Microbiol. 7, 331– 346. Su, Y.C., Wu, J.L., Hong, J.R., 2009. Betanodavirus non-structural protein B2: a novel necrotic death factor that induces mitochondria-mediated cell death in fish cells. Virology 385, 143–154. Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5: Molecular Evolutionary Genetics Analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739. Tang, L., Lin, C.S., Krishna, N.K., Yeager, M., Schneemann, A., Johnson, J.E., 2002. Virus-like particles of a fish nodavirus display a capsid subunit domain organization different from that of insect nodaviruses. J. Virol. 76, 6370–6375. Thiéry, R., Cozien, J., de Boisséson, C., Kerbart-Boscher, S., Névarez, L., 2004. Genomic classification of new betanodavirus isolates by phylogenetic analysis of the coat protein gene suggests a low host-fish species specificity. J. Gen. Virol. 85, 3079– 3087. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 25, 4876–4882. Toffolo, V., Negrisolo, E., Maltese, C., Bovo, G., Belvedere, P., Colombo, L., Valle, L.D., 2007. Phylogeny of betanodaviruses and molecular evolution of their RNA polymerase and coat proteins. Mol. Phylogenet. Evol. 43, 298–308. Ucko, M., Colorni, A., Diamant, A., 2004. Nodavirus infections in Israeli mariculture. J. Fish Dis. 27, 459–469. Vendramin, N., Toffan, A., Mancin, M., Cappellozza, E., Panzarin, V., Bovo, G., Cattoli, G., Capua, I., Terregino, C., 2014. Comparative pathogenicity study of ten different betanodavirus strains in experimentally infected European sea bass, Dicentrarchus labrax (L.). J. Fish Dis. 37, 371–383. Wertheim, J.O., Chu, D.K., Peiris, J.S., Kosakovsky Pond, S.L., Poon, L.L., 2013. A case for the ancient origin of coronaviruses. J. Virol. 87, 7039–7045. Worobey, M., Holmes, E.C., 1999. Evolutionary aspects of recombination in RNA viruses. J. Gen. Virol. 80 (Pt 10), 2535–2543. Wright, F., 1990. The ‘effective number of codons’ used in a gene. Gene 87, 23–29. Yoshikoshi, K., Inoue, K., 1990. Viral nervous necrosis in hatchery-reared larvae and juveniles of Japanese parrotfish, Oplegnathus fasciatus (Temminck & Schlegel). J. Fish Dis. 13, 69–77.
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev
Divergence and codon usage bias of Betanodavirus, a neurotropic pathogen in fish Mei He, Chun-Bo Teng ⇑ College of Life Science, Northeast Forestry University, Harbin 150040, China
a r t i c l e
i n f o
Article history: Received 4 September 2014 Revised 25 November 2014 Accepted 30 November 2014 Available online 10 December 2014 Keywords: Betanodavirus Divergence Codon usage bias Viral nervous necrosis
a b s t r a c t Betanodavirus is a small bipartite RNA virus of global economical significance that can cause severe neurological disorders to an increasing number of marine fish species. Herein, to further the understanding of the evolution of betanodavirus, Bayesian coalescent analyses were conducted to the time-stamped entire coding sequences of their RNA polymerase and coat protein genes. Similar moderate nucleotide substitution rates were then estimated for the two genes. According to age calculations, the divergence of the two genes into the four genotypes initiated nearly simultaneously at 700 years ago, despite the different scenarios, whereas the seven analyzed chimeric isolates might be the outcomes of a single genetic reassortment event taking place in the early 1980s in Southern Europe. Furthermore, codon usage bias analyses indicated that each gene had influences in addition to mutational bias and codon choice of betanodavirus was not completely complied with that of fish host. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Viral nervous necrosis (VNN), also termed as viral encephalopathy and retinopathy (VER), is a highly infectious disease that can cause serious damage typical of extensive vacuolation to the piscine central nervous system (Glazebrook et al., 1990; Yoshikoshi and Inoue, 1990). To date, more than 50 marine teleosts, as well as several freshwater species, have been found susceptible (Bandín and Dopazo, 2011; Vendramin et al., 2014). Infected fish, larvae and juveniles in particular, exhibit various clinical signs such as anorexia, darkened coloration and erratic swimming behavior, and suffer mass mortalities that may reach up to 100% (Munday et al., 2002). Since its first emergence in the 1980s, VNN has been a major impediment to mariculture, outbreaks of which have been reported from all over the world accompanied with significant economic losses (Nakai et al., 2009). The aetiological agent of VNN is a small naked RNA virus assigned to the genus Betanodavirus in the family Nodaviridae, distantly related to insect-infecting Alphanodavirus (Ball et al., 2000). It has a single-stranded RNA genome composed of two genetic molecules designated as RNA1 and RNA2 encapsulated by the icosahedral capsid. Being positive-sense, both of them possess the methylated caps but lack the poly (A) tails (Mori et al., 1992). RNA1, the larger segment (3.1 kb), encodes the RNA-dependent ⇑ Corresponding author. Fax: +86 451 8219 1784. E-mail address: [email protected] (C.-B. Teng). http://dx.doi.org/10.1016/j.ympev.2014.11.016 1055-7903/Ó 2014 Elsevier Inc. All rights reserved.
RNA polymerase (RdRp, or protein A) (Nagai and Nishizawa, 1999), while RNA2, the smaller one (1.4 kb), encodes the coat protein (Cp) (Nishizawa et al., 1995). During viral replication, a subgenomic RNA3 is synthesized from the 30 terminus of RNA1 and expresses two small proteins, B1 and B2, with regulatory functions (Chen et al., 2009; Fenner et al., 2006; Iwamoto et al., 2005; Su et al., 2009). Based on phylogenetic analysis of the variable T4 region of the Cp gene, betanodaviruses can be categorized into four genotypes (Nishizawa et al., 1997): striped jack nervous necrosis virus (SJNNV), tiger puffer nervous necrosis virus (TPNNV), barfin flounder nervous necrosis virus (BFNNV) and red-spotted grouper nervous necrosis virus (RGNNV), with SJNNV being the type species. This taxonomic classification holds true for the RdRp gene, despite the difference in genetic relationships (Toffolo et al., 2007). Moreover, a single turbot nodavirus (TNV) has been proposed by Johansen et al. (2004) to be the fifth genotype. The four major genotypes differ in various aspects including host range, geographic distribution and optimum temperature (Nakai et al., 2009). SJNNV is endemic to Japan affecting certain fish species; however, it also exists (mostly as reassortants) in Southern Europe (Panzarin et al., 2012). TPNNV is so far only found in Japan from Tiger puffer (Takifugu rubripes) and Japanese flounder (Paralichthys olivaceus) (Nishizawa et al., 1997). BFNNV prefers coldwater teleosts in Japan and the North Atlantic. In contrast, RGNNV possesses the broadest host spectrum covering various kinds of warm-water finfish and the widest geographical range involving
138
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
Asia, Australia, North America, and the Mediterranean (Bandín and Dopazo, 2011). As is common in multipartite viruses, reassortment is employed by betanodaviruses to drive their evolution. Interestingly, all reassortants identified to date are generated from either combination of genomic segments between SJNNV and RGNNV taking place in the Mediterranean basin and the connected Iberian Atlantic region (Olveira et al., 2009; Panzarin et al., 2012; Toffolo et al., 2007; Vendramin et al., 2014). Moreover, recently, based on the dated partial sequences, the rates of nucleotide substitution and the times to the most recent common ancestor (TMRCAs) of the RGNNV group in Southern Europe have been estimated, which revealed a higher rate and a younger age for the RdRp gene there. However, the differences were not significant as the 95% highest probability density (HPD) ranges for the two segments overlapped (Panzarin et al., 2012). Here, to broaden the knowledge of the molecular epidemiology and evolutionary dynamics of betanodavirus, Bayesian coalescent method was applied to the time-stamped complete coding sequences of the RdRp and Cp genes sampled worldwide. Moreover, to better understand the processes governing their evolution, codon usage bias of each gene was also examined.
2. Materials and methods Full-length coding sequences of the RdRp and Cp genes were retrieved from GenBank and aligned with CLUSTAL W (Thompson et al., 1997). Dataset compilation, Bayesian estimates, and codon usage bias analyses were conducted as described previously (He et al., 2013). Details of the betanodavirus isolates analyzed, including collection dates supplemented via literature and Genbank Accession numbers, were listed in Table S1. When the Bayesian Markov chain Monte Carlo (MCMC) method in BEAST v1.7.4 (Drummond et al., 2012) was applied to each panel, the four clock models (strict, exponential, lognormal and random local) as well as the five demographic models (constant, exponential, expansion, logistic and Bayesian skyline plot) were compared (He et al., 2014a). Specifically, the strict clock model was evaluated by the coefficient of variation (CoV), a scale of the degree of clocklikeness of the data (Drummond et al., 2006). Thus, it was rejected for RdRp in that CoV did not encompass zero. On the contrary, it was accepted for Cp since the lower 95% HPD of CoV approached zero. For RdRp, among the three relaxed clock models, the lognormal one was utilized as the other two failed to describe the evolutionary dynamics. Moreover, as most demographic models yielded similar results for Cp, the logistic growth one was chosen owing to its best reliability and convergence for RdRp. Then, independent analyses for 10–25 million MCMC iterations were combined (10% burn-in) to assure effective sample size (>200). Isolate information (name/year/origin/accession) was also included in each maximum clade credibility (MCC) tree. Two indices of codon usage bias, the effective number of codons used by a gene (Nc) and the frequency of G + C at the synonymous 3rd codon position (GC3S) were examined for RdRp and Cp. The overall variation in the codon usage was reflected by the relative synonymous codon usage (RSCU), the value of which higher than 1.0 indicates that the corresponding codon is adopted more frequently than expected, and vice versa. Besides, correspondence analysis, a multivariate statistical technique that yields orthogonal axes to account for variations, was conducted to identify major trends in amino acid usage. All aforementioned analyses were done by CodonW 1.4.4 (http://codonw.sourceforge.net). Moreover, an alphanodavirus Flock house virus (FHV) and a fish host Dicentrarchus labrax with more codons available (http://www.kazusa.or.jp/ codon/) were selected for comparisons of RSCU and GC content.
In addition, to infer the genetic relationships of the chimeras (30 RG/SJ and 1 SJ/RG), alignments of their partial RdRp (867 nt, referred to CDS positions 154-1020 of SJNNV/AB056571) and Cp (504 nt, CDS positions 357-860 of SJNNV/AB056572) sequences were created for phylogenetic analyses, respectively. Each Maximum Likelihood (ML) tree was drawn by MEGA 5.1 (Tamura et al., 2011) with 1000 bootstrap replicates under the best-fit nucleotide substitution model (TN93 + G for RdRp and T92 + G for Cp). 3. Results 3.1. Nucleotide substitution rates of the RdRp and Cp genes As listed in Table 1, when entire coding sequences of the RdRp gene from 49 betanodavirus isolates spanning 19 years, consisting of 32 RG, 14 BF, 1 TP and 2 SJ-type viruses, were subjected to Bayesian analysis, the average rate of nucleotide substitution was estimated to be 3.60 104 subs/site/year, with the 95% HPD values varying from 1.40 104 to 6.18 104. A similar mean rate at 3.69 104 (2.31 104–5.06 104) subs/site/year was obtained for the Cp gene based on 73 isolates spanning 21 years, composed of 54 RG, 8 BF, 1 TP and 10 SJ-type viruses. Substitution rates were also calculated for the two genes of RGNNV, the largest genotype with more complete sequence data available. The reassortants and MnNNV-12-06 (Fig. 1A) being excluded, the average rate of RdRp from 24 isolates was 4.28 104 (3.35 105–8.79 104) subs/site/year, which was insignificantly higher than that of Cp from 52 isolates at 3.79 104 (2.01 104–5.60 104) subs/site/year. In fact, rate difference was still minor when the Cp panel was truncated to be identical to the RdRp panel (data not shown). 3.2. Phylogenetic relationships of betanodaviruses The MCC trees (Fig. 1) calculated for the RdRp and Cp genes of betanodavirus isolates collected worldwide confirmed the categorization of four major genotypes (SJ, TP, BF and RG), the difference in genetic relationships involving BFNNV, as well as the reassortments between SJNNV and RGNNV. Notably, the Atlantic cod nervous necrosis virus (ACNNV) circulating in Canada described by Gagnè et al. (2004) could be classified as a subtype of BFNNV (Fig. 1B). Within the large RGNNV group, the isolates analyzed here clustered into 3 (a–c, Fig. 1A) and 6 (a–e, Fig. 1B) well-supported subclades (posterior probability value >0.9) based on their RdRp and Cp sequences, respectively. The subtype a of Cp, for example, could be further divided as well (1–7, Fig. 1B). According to the arbitrary classification, incongruent topologies were observed for GPNNV1108-P.Langkawi-1 sampled from Malaysia. In the RdRp
Table 1 Details of datasets and estimates of the two betanodaviral genes. Parametera
RdRp
Cp
No. of sequences Time span Substitution model Molecular clock Demographic model Mean substitution rate (104) 95% HPD rate (104) Mean TMRCA 95% HPD year
49 1993–2012
73 1991–2012 GTR + G + I
Lognormal
Strict Logistic
3.60 1.40–6.18 695 267–1238
3.69 2.31–5.06 704 446–990
a HPD: highest probability density; TMRCA: time to the most recent common ancestor.
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
139
Fig. 1. Maximum clade credibility phylogenies of the RdRp (A) and Cp (B) genes of betanodavirus. The trees are scaled to time generated under the better molecular clock and demographic models (Table 1). Nodes correspond to mean TMRCAs. Age estimates with the 95% highest probability density values are given for some nodes. Isolate information (name/year/origin/accession) and genotype classification are shown on the right. Subtypes of RGNNV designated here with lowercase letters are labeled above their nodes. In the Cp tree, the seven clusters within the subtype a supported by >0.9 posterior probability value are also indicated on the right with Arabic numerals. Two special isolates cited in the discussion were marked.
140
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
tree, it was clustered with three Malaysian isolates in the b clade; however, in the Cp tree, it fell into the a1 clade distinct from the c clade formed by the three Malaysian isolates. This might be a reflection of an intra-genotype reassortment event. 3.3. Divergence of the RdRp and Cp genes The mean TMRCA of the RdRp gene was 695 (95% HPD: 267– 1238) years, which was nearly simultaneous with that of the Cp gene at 704 (446–990) years. Therefore, both of the two initial bifurcation events of the analyzed betanodavirus isolates might have taken place at the onset of the 14th century. However, the divergence scenarios of the two genes into the four genotypes were quite different, as revealed by the time-scaled MCC phylogenies (Fig. 1). According to the RdRp tree (Fig. 1A), the first event occurred between the singular progenitor of RGNNV and the common ancestor shared by the other three genotypes. Only four decades later, the SJ-type virus diverged out; however, subsequently, it took the precursor of TPNNV about four and a half centuries to branch off from that of BFNNV. In contrast, seen from the Cp phylogram (Fig. 1B), the primary event was that the ancestor of RGNNV and BFNNV separated from that of SJNNV and TPNNV. Coincidentally, the two secondary events resulting in the four genotypes both happened 140 years later. When it came to the divergence within each genotype (except TPNNV which has only one isolate fully sequenced so far), differences were also observed for RGNNV, as reflected by the dissimilar TMRCA estimates for the two genes (Fig. 1). Even without regard to DL-389-I96, a SJ/RG type chimera (Fig. S1; Vendramin et al., 2014) more divergent from the other RGNNV isolates, there was still a 64-year age gap between the RdRp and Cp MRCAs (85 vs. 149), despite the partially overlapping HPD intervals (36–147 vs. 92– 213). Interestingly, if MnNNV-12-06 was excluded, which was clustered with the reassortants and did not diverged out until 40 years ago in the RdRp tree, the age estimates were much similar: 85 (36–147) vs. 91 (57–129). For the other two genotypes, timing results based on the two genes were also similar. The last common ancestor of SJNNV was just in its sixties, whereas that of the BFNNV subgroup composed of the Japanese and Norwegian isolates was around forty years old (Fig. 1). Currently, as no RdRp sequence was available for the Canadian BFNNV isolates, it was unknown whether their RdRp ancestor has diverged out nearly one and a half century ago as the Cp one (Fig. 1B). In addition, TMRCAs were also estimated for the seven RG/SJ type reassortants, which suggested that their divergence began in the 1980s (Fig. 1). 3.4. Codon usage bias of each betanodaviral gene To measure codon usage bias of each betanodaviral gene, Nc and GC3S were calculated. As Nc can take values from 20 for extreme bias when only a single synonymous codon is adopted, to 61 for no bias when all synonymous codons are equally used (Wright, 1990), the high Nc values here (>50, Fig. 2) indicated that the degree of codon usage bias was slight in each gene. When Nc was plotted against GC3S (Fig. 2A and B), the genotypes/subtypes exhibited different distribution patterns. This was more pronounced in the Cp plot (Fig. 2B), suggesting different selection constraints on the genotypes/subtypes, particularly on their RNA2 segments. Interestingly, although the reassortants possessed SJ-type Cp genes, their points were all clustered with the RG points but not the other SJ points (Fig. 2B), supporting the observation based on sequence comparison that the acquired SJ segments have changed to some extent toward the replaced RG segments (Olveira et al., 2009).
Fig. 2. Nc-plots of the RdRp (A) and Cp (B) genes of betanodavirus. Nc (the effective number of codons) vs. GC3S (the GC content at the synonymous third codon position) is plotted for each betanodaviral isolate. The four genotypes are distinguished with different forms and colors as indicated on the right. The continuous curve represents the expected Nc values under assumptions of no selection other than GC composition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Moreover, the Cp points of the BFNNV subgroup circulating in Japan and Norway were clustered just below the curve of the expected values, reflecting no selection other than GC composition. In contrast, other points were located away from the curve, implying influences aside from mutational bias (He et al., 2014b). Indeed, Nc of RdRp was in significant correlation with GRAVY, an index of protein hydrophobicity (R = 0.64, P < 0.01), so was Nc of Cp (R = 0.78, P < 0.01). This factor was also recommended by the correspondence analyses to be a major contributor to the amino acid usage variation in that the first axis, the principal trend, was significantly (P < 0.01) correlated with GRAVY (R = 0.62 for RdRp and R = 0.70 for Cp). In addition, RSCU was calculated to determine the overall variation in the codon usage. From the values, a predominance of C over U was observed in many synonymous 3rd codon positions (Table S2). Interestingly, in respect of codon preference and 3rd letter GC content, it looked like that betanodavirus (GC3rd 0.55) was in a transitional stage between a kin insect-infecting alphanodavirus (FHV, 0.47) and a host fish species (D. labrax, 0.63) (http://www.kazusa.or.jp/codon/), with more resemblance to the latter (Table S2).
4. Discussion According to Bayesian coalescent analyses applied to the dated entire coding sequences of betanodavirus isolates collected worldwide, the divergence of the RdRp and Cp genes into the four genotypes not only initiated nearly simultaneously but progressed at a similar moderate overall rate. However, it should be pointed out that, as reflected by the wider 95% HPD values (Fig. 1), age estimates for the long branches were less reliable than those for the short branches, especially when the two genes were subjected to strong purifying selection (Panzarin et al., 2012; Toffolo et al.,
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
2007) that could result in underestimation of the lengths of the long branches (Wertheim et al., 2013). Although there are expected to be some variations in the timing results of the primary bifurcation events when more complete sequence data are analyzed, it could be safely concluded that betanodavirus, even the four genotypes, had emerged long before the recognition of VNN. The nucleotide substitution rates and the TMRCAs of the RGNNV strains from the Indo-Pacific and Southern Europe were also calculated, which differed from those estimated for the RGNNV group in Southern Europe (Panzarin et al., 2012), especially the mean rate for the Cp gene (3.79 104 vs. 2.42 104) and the average age for the RdRp gene (85 vs. 188). Although not significant given the overlap between the HPD intervals (2.01 104–5.60 104 vs.1.07 104–3.99 104 as well as 36–147 vs. 79–328), these differences could be due to panel composition. As evolutionary rates can vary among genotypes/subtypes, which has been demonstrated previously (He et al., 2013, 2014b), it was possible that Cp of the Southern European RGNNV group evolved relatively slower, whereas since the IX and X clusters in their RdRp tree, which were outgroup to the analyzed subtypes here (a–c, Fig. 1A), were not covered in our dataset, it was certain that a much older age was obtained for RdRp in their analysis. In general, segmented RNA viruses achieve their rapid evolution via processes including mutation, reassortment and recombination (Worobey and Holmes, 1999). Currently, no significant signal for recombination was detected in the available betanodaviral sequences by the multiple approaches as applied before (He et al., 2012). It seemed that betanodaviruses preferred the former two ways to generate their genetic diversity and adapt to a new environment. Intriguingly, all inter-genotype reassortants so far were identified in Southern Europe generated from the exchange of genomic segments between SJNNV and RGNNV, reflecting a special niche there. Of note, it did not require frequent exchange to establish the large RG/SJ population. According to the MCC trees (Fig. 1), the seven chimeras might be the outcomes of a single reassortment event taking place in the early 1980s. In fact, inferred from the phylogenetic relationships based on the partial sequences (Fig. S1), all but one RG/SJ reassortants sampled to date (29 of 30, except 28.2005), despite their diverse origins, were likely to be derived from this event. Moreover, intra-genotype reassortment might occur as well, as exemplified by GPNNV1108-P.Langkawi-1 whose RdRp and Cp genes belonged to distinct RGNNV subtypes (Fig. 1). Certainly, reassortment complicated genetic relationships and genotype classification of betanodaviruses, which might render an explanation for the observed differences in the dated divergence among genotypes as well as within RGNNV. It was possible that BFNNV, which exhibited incongruent topologies in the MCC trees (Fig. 1), was a SJ/RG type chimera born in an early event that occurred a couple of centuries ago. Interestingly, BFNNV was once isolated in France during an outbreak at low temperature (Thiéry et al., 2004) from European sea bass (D. labrax) that is susceptible to both RG/SJ and SJ/RG reassortants (Panzarin et al., 2012; Vendramin et al., 2014). Moreover, MnNNV-12-06 possessed an old type Cp gene but a young type RdRp gene (Fig. 1), which might be due to an intra-genotype reassortment event. Another intricacy was the correlation between genetic relationships and geographical location, particularly within the RGNNV genotype. As could be seen in Fig. 1B, isolates sampled from different areas could fall into the same subcluster (a1, a2, a4, and a5), whereas distinct groups were circulating in the same country (e.g., a1, a7 and c in Malaysia). This might be the outcome of multiple factors such as commerce of infected stocks, transfer of wild carriers and parallel/convergent evolution (Dalla Valle et al., 2001; Toffolo et al., 2007). A definite example of the first factor
141
was the Israeli isolate LC121100-IL. It was identified in Asian sea bass (Lates calcarifer) imported from Thailand as larvae (Ucko et al., 2004) and was indeed clustered within a Southeast Asian group (in a7). Of note, betanodaviruses, including reassortants, were also found in marine invertebrates (Gomez et al., 2008; Panzarin et al., 2012), which might further perplex the origin and spread of the virus. As mentioned above, the four genotypes differ in optimum growth temperature. In cultured cells, RG, SJ, TP and BF exhibited higher titers at 25–30 °C, 20–25 °C, 20 °C and 15–20 °C, respectively (Iwamoto et al., 2000). An interesting correlation was then observed between the optimal temperature and the divergence of the RdRp gene (Fig. 1A): seen from the line of BFNNV, the earlier the genotype branched off, the warmer the preferred temperature was. Indeed, experiments using chimeric viruses (Hata et al., 2010; Panzarin et al., 2014) have demonstrated the major role RNA1 plays in determining temperate sensitivity via the region 84– 1419 encoding the first 445 amino acid residues of RdRp, particularly the region 1088–1419. Actually, the secondary structures of the amino acid region 335–445 of RdRp predicted for the four genotypes were different (data not shown). Furthermore, codon usage bias, an important indicator of the forces shaping the evolution of RNA viruses (Jenkins and Holmes, 2003), was examined for each betanodaviral gene. From the comparison between the actual distribution and the expected one in the Nc-plot, it was visible that each gene had additional codon usage bias independent of mutation pressure. Perhaps, as suggested by the statistical analyses, one driving force of the nucleotide changes was the hydropathy level of the amino acid sequence, on which the stability of the protein is largely dependent. This might be of consequence in the evolution of betanodavirus, for instance, adaptation to low temperature. Interestingly, from another comparison between virus and host in terms of codon choice, it looked like that betanodavirus had not fully conformed to fish yet, suggesting that its codon usage bias might be less affected by the host’s tRNA abundance. At the expense of high efficiency and accuracy in translation, such disobedience to the cellular codon preference may be subtly utilized to fine-tune translation kinetics to assure proper protein folding, as may happen in the hepatitis A virus (Pintó et al., 2012) and the vesicular stomatitis viruses (Liang et al., 2014). Indeed, in the middle of the amino acid region 83–216 that probably folds into the inner b-sandwich shell of the T = 3 capsid of betanodavirus (Tang et al., 2002), the eight residues ‘AL(I/F)QATRGA’ from positions 140 to 147 associated with an a-helix structure are encoded by non-abundant cellular codons (Fig. S1). Perhaps, this cluster is strategically located there to slow down ribosome traffic rate to avert interference with the folding of the nascent peptide of Cp. As noted previously (He et al., 2014a), codon usage information may have practical implications for viral gene expression control and vaccine design. Since in several amino acids, such as Arg and Val, betanodaviruses and teleosts exhibited different codon preferences (Table S2), optimizing the favor of the virus to that of the host may be able to elevate gene expression level thus achieve sufficient viral proteins for immunity generation (Haas et al., 1996). On the contrary, deoptimizing the synonymous codons with those non-preferred by both of them, such as codons ended with ‘UA’ (Table S2), may be able to impair gene expression and virulence thus be used to develop attenuated vaccines (Mueller et al., 2006).
5. Conclusion In our study, tip calibration revealed that in the divergence of the betanodaviral RdRp and Cp genes, the initiation dates and the
142
M. He, C.-B. Teng / Molecular Phylogenetics and Evolution 83 (2015) 137–142
overall rates were similar but the diversifying scenarios were different, while codon usage bias analyses suggested that codon choice of betanodavirus was not fully complied with that of fish host, which might be utilized to regulate translation kinetics. Our findings thus further the understanding of this severe neurotropic pathogen that has become an obstacle of marine aquaculture and provide useful information for future studies such as vaccine design. Acknowledgments We are deeply grateful to the reviewers for their helpful comments for improving our manuscript. This work was supported by the Fundamental Research Funds for the Central Universities of China (No. DL13EA06). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ympev.2014. 11.016. References Ball, L.A., Hendry, D.A., Jonshon, J.E., Ruckert, R.R., Scotti, P.D., 2000. Family Nodaviridae. In: Van Rengenmortel, M.H.V., Fauquet, C.M., Bishop, D.H.L., Carstens, E.B., Estes, M.K., Lemon, S.M., Maniloff, J., Mayo, M.A., McGeoch, D.J., Pringle, C.R., Wickner, R.B. (Eds.), Virus Taxonomy: Seventh Report of the International Committee on Taxonomy of Viruses. Academic Press, San Diego, pp. 747–755. Bandín, I., Dopazo, C.P., 2011. Host range, host specificity and hypothesized host shift events among viruses of lower vertebrates. Vet. Res. 42, 67. Chen, L.J., Su, Y.C., Hong, J.R., 2009. Betanodavirus non-structural protein B1: a novel anti-necrotic death factor that modulates cell death in early replication cycle in fish cells. Virology 385, 444–454. Dalla Valle, L., Negrisolo, E., Patarnello, P., Zanella, L., Maltese, C., Bovo, G., Colombo, L., 2001. Sequence comparison and phylogenetic analysis of fish nodaviruses based on the coat protein gene. Arch. Virol. 146, 1125–1137. Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics and dating with confidence. PLoS Biol. 4, e88. http://dx.doi.org/10.1371/ journal.pbio.0040088. Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. Fenner, B.J., Thiagarajan, R., Chua, H.K., Kwang, J., 2006. Betanodavirus B2 is an RNA interference antagonist that facilitates intracellular viral RNA accumulation. J. Virol. 80, 85–94. Gagné, N., Johnson, S.C., Cook-Versloot, M., MacKinnon, A.M., Olivier, G., 2004. Molecular detection and characterization of nodavirus in several marine fish species from the northeastern Atlantic. Dis Aquat Organ 62, 181–189. Glazebrook, J.S., Heasman, M.P., de Beer, S.W., 1990. Picorna-like viral particles associated with mass mortalities in larval barramundi, Lates calcarifer Bloch. J. Fish Dis. 13, 245–249. Gomez, D.K., Baeck, G.W., Kim, J.H., Choresca Jr., C.H., Park, S.C., 2008. Molecular detection of betanodaviruses from apparently healthy wild marine invertebrates. J. Invertebr. Pathol. 97, 197–202. Haas, J., Park, E.C., Seed, B., 1996. Codon usage limitation in the expression of HIV-1 envelope glycoprotein. Curr. Biol. 6, 315–324. Hata, N., Okinaka, Y., Iwamoto, T., Kawato, Y., Mori, K., Nakai, T., 2010. Identification of RNA regions that determine temperature sensitivities in betanodaviruses. Arch. Virol. 155, 1597–1606. He, M., He, C.Q., Ding, N.Z., 2012. Natural recombination between tobacco and tomato mosaic viruses. Virus Res. 163, 374–379. He, M., Ding, N.Z., He, C.Q., Yan, X.C., Teng, C.B., 2013. Dating the divergence of the infectious hematopoietic necrosis virus. Infect. Gen. Evol. 18, 145–150. He, M., An, T.Z., Teng, C.B., 2014a. Evolution of mammalian and avian bornaviruses. Mol. Phylogenet. Evol. 79, 385–391. He, M., Yan, X.C., Liang, Y., Sun, X.W., Teng, C.B., 2014b. Evolution of the viral hemorrhagic septicemia virus: divergence, selection and origin. Mol. Phylogenet. Evol. 77, 34–40. Iwamoto, T., Nakai, T., Mori, K., Arimoto, M., Furusawa, I., 2000. Cloning of the fish cell line SSN-1 for piscine nodaviruses. Dis. Aquat. Organ. 41, 81–89. Iwamoto, T., Mise, K., Takeda, A., Okinaka, Y., Mori, K., Arimoto, M., Okuno, T., Nakai, T., 2005. Characterization of Striped jack nervous necrosis virus subgenomic
RNA3 and biological activities of its encoded protein B2. J. Gen. Virol. 86, 2807– 2816. Jenkins, G.M., Holmes, E.C., 2003. The extent of codon usage bias in human RNA viruses and its evolutionary origin. Virus Res. 92, 1–7. Johansen, R., Sommerset, I., Torud, B., Korsnes, K., Hjortaas, M.J., Nilsen, F., Nerland, A.H., Dannevig, B.H., 2004. Characterization of nodavirus and viral encephalopathy and retinopathy in farmed turbot, Scophthalmus maximus (L.). J. Fish Dis. 27, 591–601. Liang, Y., He, M., Teng, C.B., 2014. Evolution of the vesicular stomatitis viruses: divergence and codon usage bias. Virus Res. 192, 46–51. Mori, K., Nakai, T., Muroga, K., Arimoto, M., Mushiake, K., Furusawa, I., 1992. Properties of a new virus belonging to nodaviridae found in larval striped jack (Pseudocaranx dentex) with nervous necrosis. Virology 187, 368–371. Mueller, S., Papamichail, D., Coleman, J.R., Skiena, S., Wimmer, E., 2006. Reduction of the rate of poliovirus protein synthesis through large-scale codon deoptimization causes attenuation of viral virulence by lowering specific infectivity. J. Virol. 80, 9687–9696. Munday, B.L., Kwang, J., Moody, N., 2002. Betanodavirus infections of teleost fish: a review. J. Fish Dis. 25, 127–142. Nagai, T., Nishizawa, T., 1999. Sequence of the non-structural protein gene encoded by RNA1 of striped jack nervous necrosis virus. J. Gen. Virol. 80 (Pt 11), 3019– 3022. Nakai, T., Mori, K., Sugaya, T., Nishioka, T., Mushiake, K., Yamashita, H., 2009. Current knowledge on Viral Nervous Necrosis (VNN) and its causative betanodaviruses. Isr. J. Aquacult.-Bamid. 61, 199–207. Nishizawa, T., Mori, K., Furuhashi, M., Nakai, T., Furusawa, I., Muroga, K., 1995. Comparison of the coat protein genes of five fish nodaviruses, the causative agents of viral nervous necrosis in marine fish. J. Gen. Virol. 76 (Pt 7), 1563– 1569. Nishizawa, T., Furuhashi, M., Nagai, T., Nakai, T., Muroga, K., 1997. Genomic classification of fish nodaviruses by molecular phylogenetic analysis of the coat protein gene. Appl. Environ. Microbiol. 63, 1633–1636. Olveira, J.G., Souto, S., Dopazo, C.P., Thiéry, R., Barja, J.L., Bandín, I., 2009. Comparative analysis of both genomic segments of betanodaviruses isolated from epizootic outbreaks in farmed fish species provides evidence for genetic reassortment. J. Gen. Virol. 90, 2940–2951. Panzarin, V., Fusaro, A., Monne, I., Cappellozza, E., Patarnello, P., Bovo, G., Capua, I., Holmes, E.C., Cattoli, G., 2012. Molecular epidemiology and evolutionary dynamics of betanodavirus in southern Europe. Infect. Gen. Evol. 12, 63–70. Panzarin, V., Cappellozza, E., Mancin, M., Milani, A., Toffan, A., Terregino, C., Cattoli, G., 2014. In vitro study of the replication capacity of the RGNNV and the SJNNV betanodavirus genotypes and their natural reassortants in response to temperature. Vet. Res. 45, 56. Pintó, R.M., D’Andrea, L., Pérez-Rodriguez, F.J., Costafreda, M.I., Ribes, E., Guix, S., Bosch, A., 2012. Hepatitis A virus evolution and the potential emergence of new variants escaping the presently available vaccines. Future Microbiol. 7, 331– 346. Su, Y.C., Wu, J.L., Hong, J.R., 2009. Betanodavirus non-structural protein B2: a novel necrotic death factor that induces mitochondria-mediated cell death in fish cells. Virology 385, 143–154. Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5: Molecular Evolutionary Genetics Analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28, 2731–2739. Tang, L., Lin, C.S., Krishna, N.K., Yeager, M., Schneemann, A., Johnson, J.E., 2002. Virus-like particles of a fish nodavirus display a capsid subunit domain organization different from that of insect nodaviruses. J. Virol. 76, 6370–6375. Thiéry, R., Cozien, J., de Boisséson, C., Kerbart-Boscher, S., Névarez, L., 2004. Genomic classification of new betanodavirus isolates by phylogenetic analysis of the coat protein gene suggests a low host-fish species specificity. J. Gen. Virol. 85, 3079– 3087. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 25, 4876–4882. Toffolo, V., Negrisolo, E., Maltese, C., Bovo, G., Belvedere, P., Colombo, L., Valle, L.D., 2007. Phylogeny of betanodaviruses and molecular evolution of their RNA polymerase and coat proteins. Mol. Phylogenet. Evol. 43, 298–308. Ucko, M., Colorni, A., Diamant, A., 2004. Nodavirus infections in Israeli mariculture. J. Fish Dis. 27, 459–469. Vendramin, N., Toffan, A., Mancin, M., Cappellozza, E., Panzarin, V., Bovo, G., Cattoli, G., Capua, I., Terregino, C., 2014. Comparative pathogenicity study of ten different betanodavirus strains in experimentally infected European sea bass, Dicentrarchus labrax (L.). J. Fish Dis. 37, 371–383. Wertheim, J.O., Chu, D.K., Peiris, J.S., Kosakovsky Pond, S.L., Poon, L.L., 2013. A case for the ancient origin of coronaviruses. J. Virol. 87, 7039–7045. Worobey, M., Holmes, E.C., 1999. Evolutionary aspects of recombination in RNA viruses. J. Gen. Virol. 80 (Pt 10), 2535–2543. Wright, F., 1990. The ‘effective number of codons’ used in a gene. Gene 87, 23–29. Yoshikoshi, K., Inoue, K., 1990. Viral nervous necrosis in hatchery-reared larvae and juveniles of Japanese parrotfish, Oplegnathus fasciatus (Temminck & Schlegel). J. Fish Dis. 13, 69–77.
