Nora virus, a persistent virus in Drosophila, defines a new picorna-like virus family

The basal mechanisms of innate immunity are surprisingly well conserved between humans and insects. With its well-characterized genetic system, the fruit fly, Drosophila melanogaster, has proven to be a useful model organism to study these mechanisms (Hoffmann, 2003; Hultmark, 2003; Leclerc & Reichhart, 2004; Tanji & Ip, 2005). The fruit fly faces many parasites, pathogens and other micro-organisms in its natural habitat, rotting fruit. The innate defences against bacteria and fungi in Drosophila are now relatively well understood, yet surprisingly little has been done to study how viral infections are controlled. Recent studies have shown that the JAK/STAT pathway plays a role in the response against Drosophila C virus, while the Toll signalling pathway may affect the sensitivity to Drosophila X virus infection (Dostert et al., 2005; Zambon et al., 2005). The difference in results may be explained by the fact that two different viruses were studied. Better knowledge about Drosophila viruses and their biology is needed to develop this field.

Recently, several insect viruses have been sequenced, making it possible to classify them. RNA viruses of the picorna-like superfamily are among the best studied of the insect viruses. Picorna-like viruses are further categorized on the basis of their genome organization. The vertebrate picornaviruses, the members of the family Picornaviridae, are positive-strand, single-stranded RNA viruses. They utilize their genomic RNA as an exclusive message for a single polyprotein, from which all viral proteins are produced as the result of processing (Ryan & Flint, 1997). Other viruses of the picorna-like superfamily share a conserved helicase-protease-replicase (H-P-Rep) cassette of replicative proteins with the picornaviruses (Koonin & Dolja, 1993). One heterogeneous group of picorna-like viruses found in insects and other arthropods is usually included in a floating genus, the genus Iflavirus. Their genomes are organized as in the family Picornaviridae, with a single open reading frame (ORF) where the structural proteins are encoded 5' of the H-P-Rep cassette. In contrast, in the family Dicistroviridae, the structural proteins are encoded in a second ORF in the 3' part of their genome. Similar to iflaviruses, dicistroviruses have so far only been found in insects and other arthropods.

Few viruses have been described from D. melanogaster and most have not yet been extensively studied and characterized (Ashburner et al., 2005; Brun & Plus, 1980). Three viruses have been classified as picorna-like: Drosophila A virus, Drosophila P virus and Drosophila C virus, but only the latter has been sequenced and has been shown to belong to the family Dicistroviridae (Johnson & Christian, 1998). Drosophila X virus is a double-stranded RNA virus of the family Birnaviridae (Zambon et al., 2005) and Drosophila sigma virus, a rhabdovirus with a negative-strand RNA genome, is widely spread in natural populations of D. melanogaster and is one of the best-studied viruses in Drosophila (Landès-Devauchelle et al., 1995). Furthermore, several retrotransposons have been characterized, some of which, like gypsy, can also act as independent retroviruses (Kim et al., 1994).

In this paper, we describe a picorna-like Drosophila virus, Nora virus (new in Armenian), which represents a new distinct virus family. Unlike other picorna-like viruses, its genome has four ORFs. One encodes a conserved picornavirus-like H-P-Rep cassette of replicative proteins, but the others show no obvious sequence similarity to previously described viruses. Nora virus is present as a persistent infection in several tested laboratory stocks and wild-caught flies.

Fly stocks.
All adult flies used were 35 days old and were reared at room temperature on standard yeast/agar media. The fly stocks used were as follows. Canton S, Oregon R, Nairobi and e spz² ca/TM1 were originally from the Umeå Stock Centre. Zalaszanto was isolated in 2001 by Thomas Werner from wild-caught flies in Hungary. Relish^E21, Relish^E23, Relish^E38, Relish^E20 and Relish^E26 are independent lines generated in a single P element excision experiment (Hedengren et al., 1999). Dm1 and Dm2 were recently collected from the vicinity of Umeå by Ines Anderl and Svenja Stöven. The Drosophila simulans, Drosophila erecta, Drosophila yakuba and Drosophila virilis stocks were all obtained from Jan Larsson, Umeå University.

Virus purification.
Relish^E20 flies were homogenized in 10 ml NT buffer (100 mM NaCl, 10 mM Tris/HCl, pH 7.4) and clarified by centrifugation at 4500 g for 20 min. Supernatant was extracted with an equal volume of 1,1,2-trichlorotrifluoroethane before the aqueous phase was layered over a discontinuous CsCl gradient (1.5 and 1.2 g cm³) and centrifuged at 300 000 g for 4 h in an SW60 rotor (Beckman). A band near the interface was removed by suction before passing it through a second discontinuous CsCl gradient, using the same density gradient as before. The band from the second CsCl gradient was desalted and negatively stained with 2 % phosphotungstic acid (pH 7) and examined with a Zeiss EM 900 electron microscope.

Cloning and sequencing the viral genome.
A differential display screen, comparing untreated flies with those infected for 6 or 16 h, was performed as described previously (Åsling et al., 1995). An induced PCR band was extracted from a separating gel, reamplified using the initial primer combinations and cloned by TA cloning (Invitrogen) according to the manufacturer's instructions. Using the induced PCR band as a probe, eight cDNA clones were isolated from two libraries of immunostimulated adult Canton S flies (Kylsten et al., 1990; S. K. Ekengren, 1996, unpublished). A series of 5'- and 3'-RACE reactions, using a SMART RACE cDNA Amplification kit (BD Biosciences), was conducted to obtain the remaining upstream and downstream sequences of the viral genome. For 5'-RACE, we used Relish^E20 total RNA as template and synthesized cDNA using a virus-specific oligonucleotide, as described in the user manual. A gel-purified PCR product from cDNA amplification was ligated into the pCR2.1-TOPO vector (Invitrogen). The vector was then transformed into TOPO-10-competent cells (Invitrogen) and sequenced using the Big Dye sequencing kit (Perkin Elmer Life Sciences) with either vector- or virus-specific primers. For 3'-RACE, we used an oligo(dT) primer, as described in the user manual. cDNA was amplified and sequenced as described above.

RNA preparation, Northern blot analysis and hybridization.
Relish^E20 total RNA was prepared using the Aurum total RNA kit (Bio-Rad). For Northern blots, 15 µg total RNA per lane was run on a 1 % agarose gel containing formaldehyde along with 0.249.5 kb RNA size standards (Invitrogen). Hybridization was performed under high-stringency conditions (50 % formamide, 42 °C). Virus probe was made from a 1.65 kb PCR fragment, after labelling with the Rediprime II kit (Amersham Pharmacia), according to the manufacturer's instructions. After hybridization, the filters were washed and the radioactivity monitored using a PhosphorImager (Storm; Molecular Dynamics).

DNA preparation and Southern blot analysis.
DNA was isolated from Canton S and Relish^E20 flies essentially as described by Hamilton et al. (1991). Approximately 510 µg aliquots (less for Canton S) of genomic DNA were digested with EcoRI or BamHI. The digested DNA was separated on a 0.7 % agarose gel and blotted on to a Hybond membrane (Amersham Biosciences). Probe hybridization and detection was carried out as for Northern blot analysis. Different concentrations of a control plasmid containing the probe sequence were run on the same blot as a standard.

Preparation and hybridization with ssRNA probes.
The pCR2.1-TOPO in vitro transcription system (Invitrogen) was used to synthesize radioactively labelled probes from the T3 and T7 promoters. Linearized plasmid DNA was made by cutting the end of the QRT-PCR plasmid (see below) and used as template in the transcription reaction. Briefly, a 20 µl reaction mixture containing 1 µg linearized DNA template, 1x transcription buffer (Roche), RNase inhibitor (20 U), 1 mM each ATP, GTP and CTP, T3 or T7 RNA polymerase (40 U) and 1.5 mM [³²P]UTP (10 mCi ml¹, 3000 Ci mmol¹; Amersham) was incubated at 37 °C for 2 h. After incubation, the template DNA was removed by DNase I digestion (20 U) at 37 °C for 30 min. Unincorporated nucleotides were removed on a Sepharose G-50 column (Roche) according to the manufacturer's instructions. Hybridization of Northern blots with ssRNA probes was done by pre-hybridization for 2 h at 55 °C. Hybridization was performed in the same pre-hybridization solution with the addition of the ³²P-labelled ssRNA, essentially as described by Jiang et al. (1987). After hybridization, the filters were washed and the radioactivity monitored using a PhosphorImager as above.

Treatment with RNase I.
Total RNA from Relish^E20 was treated with RNase I (BioLabs) for 30 min at 37 °C, followed by heat inactivation for 20 min at 72 °C. RNA was precipitated using 4 M LiCl and 100 % ethanol for 30 min at 80 °C, washed with 70 % ethanol and resuspended in DEPC-treated water. The presence of virus was tested by RT-PCR amplification (reverse transcription at 55 °C for 10 min, followed by 95 °C for 5 min and amplification for 40 cycles at 95 °C for 10 s, followed by 30 s of annealing/extension at 58 °C) using the same primers as described below.

Quantitative real-time RT-PCR (QRT-PCR).
QRT-PCR was performed in duplicate, using the SYBR Green detection system (Bio-Rad) in the iCycle iQ Thermal Cycler (Bio-Rad). The primers used produced a product of 141 bp and were: 5'-AACCTCGTAGCAATCCTCTCAAG-3' (forward) and 5'-TTCTTGTCCGGTGTATCCTGTATC-3' (reverse). The results were quantified by comparison with a dilution series of in vitro-transcribed RNA from a viral subclone, the QRT-PCR plasmid. This plasmid contains a 370 bp RT-PCR product from virus-specific primers [5'-TTAAGGTGTTAGAGAACAGC-3' (forward) and 5'-CGTAAACACCAACTTACTTC-3' (reverse)], subcloned into pCR2.1-TOPO as described above. Such standard curves were typically generated for each experiment and the linear range always extended below the lowest and above the highest RNA concentrations in our positive samples. The specificity of the PCR products was verified further by analysis of their melting points, which differed from the non-specific products that sometimes appeared in negative controls.

Sequence analysis and phylogenetic trees.
BLAST searches were performed against the GenBank virus database on the National Center for Biotechnology Information web server (Bethesda, USA). Transmembrane regions were predicted with TMHMM 2.0 (Krogh et al., 2001) on the Prediction Servers at the Center for Biological Sequence Analysis (Lyngby, Denmark). A search for structural motifs in the proteins was done against the structure-anchored hidden Markov models (Tångrot et al., 2006) on the UCMP FISH server at Umeå University (). Multiple sequence alignment was done with CLUSTAL_W (Jeanmougin et al., 1998) on the Baylor College of Medicine web server and phylogenetic analysis was done with a maximum-parsimony algorithm (Swofford, 1991), using PAUP 4.0b10 for Macintosh (Sinauer Associates).

A PCR fragment corresponding to Nora virus was initially found in a differential display screen for RNAs that were induced in a Drosophila Canton S strain in response to bacterial infection. Using this PCR fragment, we recovered and partially sequenced eight clones from two independent cDNA libraries, made in 1990 and 1996. The same virus could also be identified and sequenced using RNA from our current laboratory stocks. Thus, we found the virus in material isolated over a time span of more than 15 years.

General properties of Nora virus
Nora virus RNA could be detected as a single band in a Northern blot with RNA from whole flies (Fig. 1a). The same band was seen with probes from different parts of the viral genome (data not shown) and no major subgenomic RNA species were detected. We used CsCl gradient centrifugation to isolate viral particles that contained the same RNA. They were found to be non-enveloped, with a diameter of approximately 30 nm (Fig. 1b). Nora virus is a positive-strand virus, as shown by hybridization with strand-specific RNA probes (Fig. 1a). Furthermore, the viral RNA was sensitive to digestion with RNase I, confirming that it is single-stranded (Fig. 1c). No DNA form of the virus could be detected on a Southern blot, at a detection level of about 0.01 viral genomes per Drosophila genome (Fig. 1d).

(67K): Fig. 1. Characterization of Nora virus. (a) Northern blot analysis of total RNA purified from Drosophila and probed for Nora virus. Lane 1 was probed with a double-stranded probe, lane 2 is the same filter probed with a sense probe and lane 3 is a similar filter probed with an antisense probe. (b) Electron microscopy of purified virus particles. (c) Sensitivity of Nora virus RNA to single-strand-specific RNase. Control (lane 1) or RNase I-digested (lane 2) RNA was detected by RT-PCR as a 141 bp band. (d) Southern blot analysis of genomic DNA from Canton S (C) and RelishE20 (R) fly strains. The DNA was stained with ethidium bromide (upper panel) before being transferred to a membrane and probed for Nora virus (lower panel). Genomic DNA was left undigested () or digested with EcoRI or BamHI. As a positive control, different concentrations of a viral subclone were left undigested () or digested with EcoRI (+). The quantities of plasmid DNA corresponded to approximately 1, 0.1 and 0.01 molar equivalents of the RelishE20 genomic DNA.

RNA sequence and relationship to other viruses
The sequenced RNA genome of Nora virus consisted of 11879 nt, but we could not exclude the possibility that secondary structures prevented us from sequencing the entire genome. The viral RNA was followed by a poly(A) tail and was relatively A-rich (34 % A, 26 % U, 22 % G, 18 % C). The sequence contained four ORFs (Fig. 2). The second ORF from the 5' end, ORF2, was the largest, encoding 2105 aa. It encoded a picornavirus-like H-P-Rep cassette, which included an RNA-dependent RNA polymerase, an RNA helicase and a protease. The polymerase domain was the most conserved and we could identify all eight consensus motifs typical of RNA-dependent RNA polymerases of supergroup 1 (Koonin & Dolja, 1993), which includes the picorna-like viruses. A pairwise comparison shows 27 % sequence identity to the honeybee sacbrood virus (GenBank accession no. AF092924[GenBank] ) over a stretch of 407 residues. The RNA helicase domain was more similar to that of vertebrate picornaviruses such as the Ljungan virus (GenBank accession no. NC_003976[GenBank] ), with 33 % identity over 172 residues. Again, the three consensus motifs typical of picorna-related viruses (Koonin & Dolja, 1993) were all present. The protease domain was less well conserved. To date, the best hit was to the Triatoma picorna-like virus (GenBank accession no. NC_003783[GenBank] ), with 27 % sequence identity over 88 residues only. Two hydrophobic stretches near the N terminus, residues 2547 and 6890, were predicted to be transmembrane regions.

(19K): Fig. 2. Open reading frames and encoded proteins of the Nora virus genome. The upper panel shows the three sense-direction reading frames (RF13). All ORFs with a start codon are shown as arrows. The four long ORFs, ORF14, are indicated. The lower panel shows conserved domains in the four encoded (poly)proteins. TM, Two predicted transmembrane segments.

The remaining three ORFs were not closely related to any previously described proteins. ORF1 encoded a highly charged protein of 475 aa with 22 % basic (mainly lysine) and 13 % acidic residues. ORF3 was the shortest and encoded a protein of 289 aa. It was related in sequence to a large number of eukaryotic and bacterial proteins of different functions and to a few viral capsid proteins, such as the cell attachment protein σ-1 (GenBank accession no. M35963[GenBank] ) of the mammalian orthoreovirus 1 (23 % sequence identity in a region of 164 residues). A search against a structure-anchored hidden Markov model database (Tångrot et al., 2006) suggested a relationship to a family of α-helical structural motifs, which is found in spectrin and other proteins. Finally, ORF4 encoded 931 aa and was not obviously related to any known sequence.

We used the sequences of the most conserved domains, the polymerase and the helicase, to investigate the possible phylogenetic relationship between Nora virus and other picorna-like viruses. Fig. 3 showed that Nora virus is not closely related to any of the major families of picorna-like viruses. The polymerase domain sequence of Nora virus showed affinity to the Iflavirus group. On the other hand, the RNA helicase domain seemed to be more closely related to the mammalian picornaviruses. However, neither relationship had a strong bootstrap support.

(42K): Fig. 3. Phylogenetic relationship between Nora virus and other picorna-like viruses. Maximum-parsimony trees were generated for the sequences of the predicted polymerase and helicase domains, using bootstrap analysis with 500 replications. The percentage bootstrap support is indicated for branches that are supported in at least 50 % of the replications. The analysis included all arthropod viruses with significant sequence similarity to Nora virus in a BLAST search, as well as a selection of best-fitting plant and vertebrate viruses.

Ubiquitous presence of Nora virus in fly stocks
Using Northern blot analysis, we were able to detect Nora virus in a few of our D. melanogaster laboratory stocks (data not shown). However, RT-PCR experiments suggested that low levels of the virus were in fact present in most or all of our laboratory stocks and in Schneider's SL2 cells. Using QRT-PCR, we found that viral titres varied enormously, from less than 10⁴ to more than 10¹⁰ viral genomes per fly, as exemplified in Fig. 4(a). Most wild-type and mutant stocks had low titres, but high titres were found in a few mutants, including the P element excision mutant Relish^E20 (Hedengren et al., 1999). This was probably not caused by the mutation in the Relish gene, as similar titres were found in other P element excision stocks that were generated in the same study, including the precise excision line Relish^E23, which retains a wild-type Relish gene. The virus was not endemic to our laboratory, as we could also detect it in stocks from other laboratories, in some cases at high levels (results not shown), as well as in recently wild-caught animals (Dm1 and Dm2 in Fig. 4b). The wild-type Canton S stock usually had low levels of the virus, but in a few experiments flies with a high titre were found in this stock (Fig. 4c). We also detected the virus in D. simulans, a close relative of D. melanogaster, but not in the more distantly related species D. yakuba, D. erecta and D. virilis (Fig. 4b).

(13K): Fig. 4. Quantitative analysis of Nora virus RNA in different Drosophila stocks using QRT-PCR. (a) D. melanogaster from laboratory stocks. Canton S, Oregon R, Nairobi and Zalaszanto are wild-type stocks. e spz2 ca/TM1 is a recessive mutant in the Toll signalling pathway; here a mixture of heterozygous and homozygous flies was assayed. RelE20 and RelE38 are Relish deletion mutants, RelE21 and RelE26 are shorter deletions that do not affect the Relish gene and RelE23 is a wild-type precise excision stock. (b) Different Drosophila species and wild-caught D. melanogaster, Dm1 and Dm2. ND, Not detectable. (c) An example of a case when Canton S flies from two independent vials showed increased viral levels.

Here, we have reported the identification of a new virus that is persistently infecting many or possibly all stocks of D. melanogaster. Although Nora virus is clearly picorna-like, it differs radically from previously described viruses. It has four ORFs, only one of which showed significant similarity to known picorna-like viruses. Strikingly, the structural proteins of the latter did not seem to be conserved in Nora virus. Furthermore, its almost 12 kb positive-strand RNA genome is considerably larger than that of most picorna-like viruses, which typically range from 7 to 10 kb. This suggests that the virus belongs to a new, previously undescribed, family of viruses.

How the four ORFs are translated is an interesting question. There are short overlaps of 7 and 26 nt between the first three ORFs and it is therefore possible that ORF2 and ORF3 are translated after ribosomal frameshifting (Dreher & Miller, 2006). Between ORF3 and ORF4 there are 85 nt of non-coding RNA, which may act as an internal ribosome entry site for the translation of ORF4. However, neither this region nor the upstream regions of the other ORFs are obviously related to known internal ribosome entry site sequences and they have little potential to form stemloop structures.

Surprisingly, persistent Nora virus infections do not give rise to any obvious pathology and even the high-titre stocks appeared healthy. We have no explanation for the large differences in viral titres among different stocks, but they did not appear to be related to the genotype of the fly. As the virus was first detected in material from bacterially infected flies, we tested whether it could be induced by bacterial infection or other types of stress, but the results were negative (data not shown). It is possible that we are dealing with more than one viral strain. The viral sequence shown here was derived from the Relish mutant stock, which has a high viral titre. The cDNA clones we isolated from Canton S flies differed only minimally from this sequence and may also correspond to high-titre virus. In general, our Canton S wild-type stock harbours low numbers of Nora virus, but occasionally we have also found increased levels in this stock. This is probably the reason why the virus was first detected in our differential display screen and why it was well represented in the two independent cDNA libraries. We also sequenced a 113 bp fragment of a low-titre virus and did not find any differences. It would be interesting to sequence the entire genome of a low-titre virus to investigate whether there is any difference, but this is a more demanding task.

An important question is in what form Nora virus remains dormant in the flies and how it is transmitted. It is obviously not integrated into the genome, as no DNA form was detected. There are now several examples of viruses that are able to cause silent, persistent infections, but the mechanisms involved are still poorly understood (Oldstone, 2006). In fact, it has recently become apparent that this phenomenon is both common and important. For instance, human picornaviruses, such as poliovirus and coxsackieviruses, are able to cause persistent infections and this can lead to late-onset pathological complications such as myocarditis and post-polio syndrome (Julien et al., 1999; Klingel et al., 1992; Pelletier et al., 1998). The presence of a viral reservoir in the population may also have serious epidemiological consequences. The adaptations of these viruses for persistence have mainly been studied in tissue culture systems (Calvez et al., 1993; Pelletier et al., 1998). Using Nora virus, it will now be possible to take advantage of Drosophila genetics to study the interactions in vivo between a persistent virus and its host.

We would like to thank Karin Edlund for her kind help in virus isolation and Rolf Sjöberg for doing the electron microscopy, Magnus Evander, Shannon Albright and Michael Williams for fruitful discussions and/or critical comments on the manuscript, and Tobias Hainzl and Lennart Frostesjö for technical advice. Uwe Sauer and Jeanette Tångrot helped with the analysis of protein structural motifs. Jan Larsson, Thomas Werner, Ines Anderl and Svenja Stöven provided fly stocks. This research was supported by grants from the Swedish Research Council and the former Swedish Medical Research Council.

Footnotes

†Present address: Department of Botany, Stockholm University, S-106 91 Stockholm, Sweden.