Identification of the 3' and 5' terminal sequences of the 8 rna genome segments of european and north american genotypes of infectious salmon anemia virus (an orthomyxovirus) and evidence for quasispecies based on the non-coding sequences of transcripts

Infectious salmon anemia (ISA) virus (ISAV) is a pathogen of marine-farmed Atlantic salmon (Salmo salar); a disease first diagnosed in Norway in 1984 [1]. It has continued to cause major disease outbreaks in marine fish [2, 3] with the clinical signs of severe anaemia, congestion of the liver and spleen along with haemorrhagic liver necrosis [4]. ISA is an OIE [Office International des Epizooties] listed disease [1]. The ISA virus was first propagated in cell culture in 1995 [5], which allowed its molecular characterization [6] and subsequent taxonomic classification to the family Orthomyxoviridae, genus, Isavirus[7].

The ISAV particles are enveloped (90-140 nm in diameter) and contain a genome of eight single-stranded (ss)RNA segments of negative polarity [1, 6]. Like in other orthomyxoviruses such as influenza A virus, each RNA segment contains one to three open reading frames (ORFs) flanked by the 5' and 3' non-coding regions (NCRs). In influenza A virus, the first 12 nucleotides at the 3' end and the first 13 nucleotides at the 5' end of NCR in all the viral RNA segments are highly conserved [8‐13]. These partially complementary termini base pair to form terminal panhandle structures [14], which function as promoters by interacting with the viral polymerase complex during replication and transcription of viral RNA [8, 15‐20]. Moreover, the segment specific NCR sequences may play important roles in virus virulence [21], and in the rescue of influenza virus using the reverse genetics system [8, 22]. Although the terminal sequences of members of the family Orthomyxoviridae such as Influenzavirus A have been analyzed extensively and subsequently used in engineering recombinant viruses, those of Isavirus remain largely unknown, and the few reported are from different ISAV strains and on different ends of the different RNA segments [23‐30]. Moreover, others have reported unique 5' terminal sequences for ISAV RNA segments 2 [31], 3 [26], 5 [27], and 8 [7], indicating a variation in the RNA templates used. Furthermore, the focussed sequencing of cDNA to ISAV mRNA to date has meant that all 3' sequences of ISAV found in the GenBank Database [32] till now (except for genome segments 6 [27, 28] and 7 [23, 33]) are incomplete since the 3' terminal sequences downstream of the orthomyxoviral polyadenylation signal are removed during polyadnyelation [34].

Sequence analysis of several ISAV isolates in the ORFs on the eight genomic RNA segments consistently reveals two genotypes that are designated with respect to their geographic origin, European and North American; the two show 15-19% difference in their amino acid sequences of the surface glycoproteins, fusion (F) protein and haemagglutinin-esterase (HE) protein [35]. It has been proposed to designate the European genotype as Genotype I and the North American genotype as Genotype II because the virus has now been reported in Europe, North America, and South America [3]. ISAV isolates can be further differentiated on the basis of insertion/deletions in a highly polymorphic region (HPR) spanning residues ³³⁷V to M³⁷² in the stem of the HE protein, adjacent to the transmembrane region [36], but the HPR is vaguely defined [37], and has been rejected in epidemiological investigations because HPR groups vary significantly and are not suited as an indicator of relatedness between virus isolates [38]. There is a lack of information on the genetic changes in NCRs of any RNA segment of ISAV although these sequences are known to play a vital role in replication of orthomyxoviruses [8, 15].

ISAV being an orthomyxovirus is characterized by abundant genetic variation. Orthomyxoviruses such as influenza viruses have high mutation rates because the viral RNA-dependent RNA polymerases have a high misincorporation frequency and have no proofreading-repair mechanisms [39]. Moreover, mismatch repair mechanisms are unlikely to operate on replicating RNA [40] and cannot operate on ssRNA progeny genomes [39]. In addition, RNA viruses generally have very short replication times and generate very high virus yields [41], two characteristics that strongly accelerate evolution of RNA viruses. As a consequence, viral quasispecies populations [39, 40] are present wherever orthomyxoviruses multiply in a host [42, 43], and are connected with a high potential for rapid evolution [44, 45] since the multiple variants are subjected to continuous adaptation pressure [46]. Quasispecies in influenza A virus isolates and samples have been readily identified using the highly discriminating methodologies of Mass Spectrometry coupled to RT-PCR [47], and high-resolution genome sequencing [43]. There is very limited information on the existence of ISAV quasispecies populations. In the report on ISAV terminal sequences of segments 7 and 8 [23], the mRNA transcripts of these segments showed heterogeneous 5' ends suggestive of a quasispecies, however this was not further pursued as the heterogeneity was attributed only to cap-stealing that is characteristic of influenza virus mRNA synthesis [24, 48]. Kibenge et al. [3] used sequence analysis of RT-PCR products of ISAV segment 6 ORF obtained directly from fish tissue and found 24 distinct HPR variants associated with the 2007-2009 ISA epizootic in Chile, but only 7 distinct ISAV strains based on segments 5 and 6 phylogenetic analyses. The appearance of multiple HPR groups in such a short time in tissues from the same or different fish originating from the same or different fish farms indicated that the ISAV HPR groups existed as quasispecies populations [3]. To examine further the genetic diversity of ISAV transcripts and the intra-segment viral evolution of ISAV, the terminal sequences of an RT-PCR product from each end of the eight RNA segments of ISAV strain ADL-ISAV-07 were determined. Because of the vital role they play in virus replication in orthomyxoviruses [9], it was considered that the diversity of terminal sequences in a population of viruses present in an infected cell lysate as represented by RT-PCR would be a true indication of ISAV quasispecies. In addition, the 3' and 5' NCR sequences of each of the eight segments of the ISAV genome were determined from vRNA extracted from purified virus particles of ISAV strain NBISA01. The two different ISAV strains belong to the two genotypes of ISAV (ADL-ISAV-07 is of European genotype and NBISA01 is of North American genotype) and each was a source of a different RNA type (mRNA/cRNA versus vRNA) such that the NCR sequences obtained were complementary, allowing the simultaneous identification and confirmation of the 3' and 5' NCR sequences of the 8 RNA genome segments of both genotypes of ISAV. The experimental design of this study is illustrated in Figure 1.

The present study focussed on the quasispecies distribution found in the NCRs of mRNA/cRNA of each genome segment of ISAV. Comparison of 3' UTR sequences of mRNA transcripts from the different segments revealed heterogeneity among the five different clones of the same segment for all RNA segments. The 5' and 3' NCRs of the cRNA from the different segments were also variable; the NCRs of segment 5 cRNA were the least variable with all five clones showing identical sequence, whereas the 3' NCR of segment 3 cRNA had the highest HI (0.8) of all NCRs of ISAV cRNA, with all five clones having different sequences (Table 5). In a previous report on ISAV terminal sequences of segments 7 and 8 [23], two different sets of sequences, with and without 5'-end heterogeneous extensions were found; the heterogeneity was attributed to cap-stealing that is characteristic of influenza virus mRNA synthesis [48], whereas the sequences without heterogenous extensions were attributed to viral cRNA. In the present study, we were not able to obtain any sequences from the 5' ends of viral mRNA because our 5' RACE protocol did not work with the total RNA preparations, and the RNA ligations worked only for viral cRNA. Therefore, the heterogeneity demonstrated in the present study cannot be explained by cap-stealing, as it consisted mostly of deletions in the 5' NCR sequences of the cRNA (Figures 4 and 5). Moreover, the heterogeneity was also found in the 3' NCR sequences of cRNA (Figures 4 and 5) and in 3' UTR sequences of viral mRNA (Figure 2). It is our considered opinion that this sequence variation in terminal sequences of the same segment end is suggestive of intra-segment ISAV quasispecies. It is interesting that on one hand both 5' and 3' NCRs of segment 5 (F gene) cRNA showed no variation in the five clones while on the other extreme all five clones in each of the 3' NCR of segment 3 (NP gene) cRNA, and the 3' UTR sequences of mRNA of segments 1, 3, and 6 (PB1, NP, and HE genes, respectively) had different sequences.

The biological significance of quasispecies in the terminal NCR sequences of ISAV is not known at this time. We speculate that the quasispecies detected may play an important role in ISAV replication. For replication in influenza virus, it is known that viral RNA dependent RNA polymerase initiates the RNA synthesis on viral RNA by binding to the panhandle structure formed as a result of partial complementarity of 3' and 5' non-coding sequences of viral RNA [23]. Therefore, any alteration in the sequence of non-coding region may affect the complementarity of 3' and 5' non-coding sequences, thereby affecting the formation of panhandle structure and viral replication as evidenced in the present study in which the deletion of nucleotides (A and G) from the 3' non-coding region of segment 6 (of ADL-ISAV-07) affected the formation of panhandle structure (Figure 9). Moreover, quasispecies may also play a role at the level of protein expression. Wang and Lee [10] observed an alteration in the protein expression level as a result of induction of mutation in the non-coding regions of PB1 and PA genes of Influenza A virus. On the basis of that study, it is possible that any change in the sequence of non-coding regions in ISAV may affect protein expression as well. The transcripts of ISAV were analyzed at 15°C since this is the optimal growth temperature for ISAV. The transcription and replication of ISAV is based on the influenza virus model system and the predicted secondary structures of ISAV segments appear to be analogous to the panhandle structures for influenza virus. Therefore, the stability of ISAV panhandle structures at 15°C was compared to that of the influenza virus panhandle formation at 37°C (Figures 8 and 9).

We report for the first time a comprehensive unambiguous analysis of NCR sequences of all eight RNA genome segments from two strains of ISAV belonging to the two ISAV genotypes. The experimental design used (Figure1), whereby the viral mRNA/cRNA was from ISAV strain ADL-ISAV-07 of European genotype and the vRNA was from ISAV strain NBISA01 of North American genotype, such that the NCR sequences obtained were complementary, allowed the simultaneous identification and confirmation of the 3' and 5' NCR sequences of the 8 RNA genome segments of both ISAV genotypes. The terminal sequences of the 8 RNA genome segments are highly conserved among the two ISAV genotypes. The 3' NCRs of each ISAV RNA gene segment are of variable lengths and the orthomyxoviral consensus sequence 5'-AGCAAAGA (in the message sense 5'-3') is present in all segments except for segment 5 at position 3 with a C→T mutation, segments 1, 4, 5, and 7 at position 4 with A→T mutation, and segment 5 at position 7 with a G→A mutation. Therefore, consistent with other reports of the ISAV 5' NCR sequences [25‐30], we have confirmed the identity of the ISAV consensus 5' end sequences (Figure 6). Therefore, the first three nucleotides at the 3' end of vRNA in all members of Orthomyxoviridae are GCU-3' (except in ISAV segment 5 with ACU-3').

However, a BLAST search [56] of the GenBank Database [32] revealed additional unique 5' terminal sequences present on ISAV RNA segments 2 [31], 3 [26], 5 [27], and 8 [57, 58]. These sequences, which range from 1 to 12 additional nucleotides at the 5' ends (Table 10), are probably due to sequencing mRNA or circularized cRNA templates resulting in appearance of heterogenous sequences at the 5' ends. The 5' NCRs in the present study were also of variable lengths on the different RNA segments, and were also characterized by an orthomyxoviral consensus sequence CA^T/_ATTTTTACT-3' (in the message sense 5'-3') in all segments except for segment 3 at position 4 with a T→A mutation, segments 3, 4, 5, and 7 at position 9 with T→A mutation, segment 8 at position 10 with T→A mutation, and segment 7 at position 11 with a C→G mutation. A BLAST search [56] of the GenBank Database [32] revealed that all ISAV 3' terminal sequences reported to date are incomplete (i.e., they lack the consensus sequence) except for genome segments 6 [27, 28] and 7 [23, 33]. This is because most ISAV sequences reported are of cDNA to mRNA in which the 3' end sequences are removed during polyadenylation of the mRNA transcripts [34]. The sequences missing at the 3' ends of the positive strand for different RNA segments reported in the GenBank Database [32] ranged from 6 to 21 nucleotides (Table 10). In the present study, the 3' sequences of the mRNA transcripts of ADL-ISAV-07 terminated 13-18 nucleotides from the full 3' terminus of cRNA, continuing as a poly(A) tail, which corresponded with the location of the polyadenylation signal (Figure 2). Thus, this is the first report to unambiguously identify the ISAV consensus 3' end sequences (Figure 7). Therefore, the first three nucleotides at the 5' end are 5'-AGU (or 5'-TCA in the cDNA), with the putative polyadenylation signal 13-15 nucleotides downstream of the 5' end terminus of the vRNA. This is also true for other members of Orthomyxoviridae[8, 15]. Thus the changes detected in the quasispecies may affect the polyadenylation of RNA transcripts of segments 4 and 8 of ISAV strain ADL-ISAV-07. Other features of the ISAV genome, like promoter sequences have not been reported yet. However, by analogy with influenza virus, it is possible to annotate the location of promoters in ISAV. Various studies on influenza virus have reported that the first 12 nucleotides at the 3' end of non-coding regions and the first 13 nucleotides at the 5' end of non-coding regions of all the viral RNA segments form the promoter responsible for replication and transcription [10, 55, 59, 60]. Based on the analogy, it appears that the first 7 nucleotides at 3' end of non-coding region and the first 8 nucleotides at 5' end of non-coding regions of all the ISAV RNA segments may comprise the ISAV promoter (Figure 10). Thus the changes detected in the quasispecies may affect the promoter activity in all the segments of ISAV (ADL-ISAV-07) except segment 5.

It is interesting that in the quasispecies study, all five clones of the 3' NCR of segment 5 cRNA of ADL-ISAV-07 had identical sequence (HI = 0.0), and yet it was the most divergent between the two ISAV genotypes at 77.2% sequence identity. This is even more divergent than at the amino acid level [3]. Conversely, the 3' NCR sequence of segment 3 cRNA of ADL-ISAV-07 had a heterogeneity index of 0.8 and yet it was 100% identical between the two ISAV genotypes. Thus the 5' NCR sequences of segments 1, 2, 3, 5, and 7, and the 3' NCR sequences of segments 3 and 4 cRNA were 100% identical in the two genotypes. These identical sequences were also present in ISAV strains of both genotypes for sequence deposited in the GenBank database [32]. Clearly, the biological significance of ISAV quasispecies warrants further study.

In conclusion, using sequence analysis of multiple clones derived from one RT-PCR product for each mRNA/cRNA transcript end for the 8 RNA genome segments, we present evidence of intra-segment ISAV quasispecies, with some RNA segments being more prone to genetic changes in their transcripts. The 3' NCR sequence of segment 5 was the least divergent within the viral population but was the most divergent between the two ISAV genotypes. Conversely, the viral population of 3' NCR sequence of segment 3 cRNA transcripts was the most heterogenous but the consensus sequence was identical between the two ISAV genotypes. Moreover we report for the first time the comprehensive unambiguous identification of the 5' and 3' terminal sequences of the 8 RNA genome segments from two strains of ISAV representing the two genotypes of ISAV. Because most ISAV sequences are of cDNA to mRNA, they do not contain the 3' end sequences, which are removed during polyadenylation of the mRNA transcripts. We report for the first time the ISAV consensus sequence CA^T/_ATTTTTACT-3' (in the message sense 5'-3') in all segments of both ISAV genotypes.