Structure and Expression Strategy of the Genome of Culex pipiens Densovirus, a Mosquito Densovirus with an Ambisense Organization

Abstract

The genome of all densoviruses (DNVs) so far isolated from mosquitoes or mosquito cell lines consists of a 4-kb single-stranded DNA molecule with a monosense organization (genus Brevidensovirus, subfamily Densovirinae). We previously reported the isolation of a Culex pipiens DNV (CpDNV) that differs significantly from brevidensoviruses by (i) having a ∼6-kb genome, (ii) lacking sequence homology, and (iii) lacking antigenic cross-reactivity with Brevidensovirus capsid polypeptides. We report here the sequence organization and transcription map of this virus. The cloned genome of CpDNV is 5,759 nucleotides (nt) long, and it possesses an inverted terminal repeat (ITR) of 285 nt and an ambisense organization of its genes. The nonstructural (NS) proteins NS-1, NS-2, and NS-3 are located in the 5′ half of one strand and are organized into five open reading frames (ORFs) due to the split of both NS-1 and NS-2 into two ORFs. The ORF encoding capsid polypeptides is located in the 5′ half of the complementary strand. The expression of NS proteins is controlled by two promoters, P7 and P17, driving the transcription of a 2.4-kb mRNA encoding NS-3 and of a 1.8-kb mRNA encoding NS-1 and NS-2, respectively. The two NS mRNAs species are spliced off a 53-nt sequence. Capsid proteins are translated from an unspliced 2.3-kb mRNA driven by the P88 promoter. CpDNV thus appears as a new type of mosquito DNV, and based on the overall organization and expression modalities of its genome, it may represent the prototype of a new genus of DNV.

The family Parvoviridae encompasses nonenveloped paraspherical viruses 22 to 25 nm in diameter with an icosahedral symmetry containing a single-stranded linear DNA genome 4 to 6 kb in length. The self-priming hairpin termini of their genomes to which the nonstructural (NS) protein NS-1 binds to initiate viral DNA replication are the hallmark of this family. On the basis of their host range, the family Parvoviridae is divided into two subfamilies: the Parvovirinae infecting vertebrates and the Densovirinae pathogenic for arthropods, mainly insects (18; also http://www.ncbi.nlm.nih.gov/ICTVdb/Ictv/fr-fst-g.htm).

Whereas all vertebrate parvoviruses so far studied have a monosense genome, i.e., with NS and structural (VP) genes located on the same strand, two main types of gene organization are known among invertebrate parvoviruses. Densoviruses (DNVs) belonging to the genera Densovirus and Pefudensovirus are characterized by having ambisense genomes, i.e., with coding sequences distributed on both strands (20). As representatives of the genus Densovirus, the Junonia cœnia DNV (JcDNV) and Galleria mellonella DNV (GmDNV) possess a 6-kb genome containing in the 5′ half of one strand the major open reading frame (ORF1) encoding the four VP polypeptides, whereas three ORFs (ORF2, ORF3, and ORF4) located in 5′ half of the complementary strand encode NS polypeptides (7, 21, 22). These genomes have an inverted terminal repeat (ITR) of about 550 nucleotides (nt) able to form a Y-shaped structure at each extremity by the folding of the distal nucleotides. The genus Pefudensovirus is represented by the DNV isolated from the cockroach Periplaneta fuliginosa. DNVs isolated from the cockroach Blatella germanica and the cricket Acheta domestica (3, 13, 23) have similar physicochemical properties since their genomes are also only 5.4 kb long, the sequences encoding their structural proteins are split in two ORFs in different frames, and their short (200 to 220 nt) telomeres fold into simple fold-back hairpins. However, their sequences are quite dissimilar.

Members of the two other genera of Densovirinae, Iteravirus and Brevidensovirus, possess a monosense genome. As exemplified by Bombyx mori DNV and Casphalia extranea DNV, the genome of iteraviruses is 5.1 kb in size and contains three large ORFs, two overlapping ORFs in the left half encoding NS polypeptides and a right-half ORF encoding VP proteins. Both genomes possess an ITR of only 230 nt, and the distal 159 nt are palindromic and form a terminal J-shaped structure (8, 10). The mosquito DNVs (Aedes aegypti DNV, Aedes albopictus DNV [AalDNV], and Culex pipiens pallens DNV) and shrimp DNV (Penaeus stylirostris DNV) are representatives of the fourth genus, Brevidensovirus. The coding sequences of their small genomes (about 4 kb) are organized in a manner similar to that of the iteraviruses but with significantly smaller ORFs encoding the VP polypeptides. In contrast to other established genera, members of the Brevidensovirus genus have a unique primary sequence of about 150 nt in length at their 3′ and 5′ noncoding extremities that are able to generate a T-shaped structure by folding (1, 4, 16, 25).

We previously reported isolating from a laboratory strain of Culex pipiens a small icosahedral DNA virus that shares the main biological and biophysical properties of DNVs and that we named CpDNV (9). Preliminary analysis of this virus indicated that it has a genome of approximately 6-kb and lacks antigenic cross-reactivity with AalDNV capsid proteins and therefore that it differs significantly from other mosquito DNVs described thus far, all of which belong to the genus Brevidensovirus. We report here the sequence, organization, and expression strategy of the complete genome of the CpDNV. We show that, unlike mosquito DNVs so far described, the CpDNV genome has an ambisense gene organization and possesses ITRs. In addition, the structure of terminal hairpins, the organization and expression modalities of NS genes that imply splicing, and the presence of two promoters for their expression are features unique among DNVs and may justify a special taxonomic status for this virus.

MATERIALS AND METHODS

Virus production and DNA extraction.

Virus was purified from heavily infected C. pipiens adult mosquitoes, and viral DNA was purified as previously described (9).

Cloning of viral DNA and sequencing.

A partial restriction map of the viral DNA was established by using different restriction enzymes (data not shown), and four HindIII restriction fragments (Fig. 1A) were selected for cloning and sequencing. After the extremities were filled with Klenow enzyme, the undigested viral DNA and the four HindIII fragments were cloned at the SmaI (undigested DNA), SmaI-HindIII (5′- and 3′-terminal fragments), and HindIII (two internal fragments) sites of the multiple cloning site of pBluescript(SK+) plasmid. Clones 5, 56, 55, and 18 with inserts corresponding to the 5′-terminal (1.3 kb), the small (0.55 kb) and large (2.35 kb) internal, and the 3′-terminal (1.6 kb) HindIII fragments, respectively, and plasmid pCpDNV containing a viral insert with an estimated size of 5.8 kb were chosen for sequencing. Sequencing was performed according to the Sanger method (15), and appropriate primers were designed for bidirectional primer walking along the cloned sequences or for amplifying fragments from uncloned viral DNA.

FIG. 1.

(A) HindIII restriction map of CpDNV DNA showing the estimated sizes of the four HindIII inserts cloned in a pBluescript plasmid. (B) Schematic representation (DNA Strider program [11]) showing the ORFs of the CpDNV genome. The ORFs encoding NS proteins are located in the 5′ half of one strand, whereas the ORFs encoding capsid proteins are located in the 5′ half of the complementary strand. The positions of the putative NS and VP promoters are indicated by arrows.

Mapping of viral transcripts.

Total viral RNA was extracted from heavily infected adult mosquitoes by using Trizol reagent (Invitrogen) or the RNeasy Mini Kit Total RNA (Qiagen) according to the manufacturers' protocols. Viral mRNAs were isolated using an mRNA isolation kit (Roche Diagnostics) following the supplier's instructions. After denaturation at 65°C for 10 min in 62.5% (vol/vol) deionized formamide, 1.14 M formaldehyde, and 1.25× morpholinepropanesulfonic acid-EDTA-sodium acetate buffer, the samples were separated on 1% agarose gel in 1× TAE (40 mM Tris-acetate and 1 mM EDTA, pH 8.0) buffer. For Northern blotting, the mRNAs were transferred overnight to a nylon membrane (Amersham) by capillarity using 20× SSC ([1× SSC is 0.15 M NaCl plus 0.015 M trisodium acetate dihydrate, pH 7] plus 2H₂O) and fixed by UV light. Following treatment with prehybridization buffer (5× Denhardt's reagent, 50% [vol/vol] formamide, 0.5% [wt/vol] sodium dodecyl sulfate [SDS], 5× SSC) for 4 h at 42°C, the membranes were hybridized to ³²P-labeled probes at 42°C overnight. A 1,004-bp NS probe and a 906-bp VP probe were amplified by PCR using primers pairs Cp25/Cp22 and Cp4/Cp6 (Table 1), respectively. PCR products were purified using a High Pure PCR purification kit (Roche), and their quality was verified by electrophoresis on 1% agarose gels. Both probes were labeled by ³²P using a Megaprime DNA Labeling Kit (Amersham) and purified on Chroma Spin-30 columns (Clontech), according to the supplier's protocol. The membranes were washed three times for 20 min successively in 2× SSC-0.1% SDS, 0.2× SSC-0.1% SDS, and 2× SSC buffer solutions; membranes were dried and analyzed using an optical scanner (Storm 840; Amersham Biosciences).

TABLE 1.

Primers used for viral mRNA analysis, mapping, and splicing

Function or target	Primer	Sequence (5′→3′)	Position (nt)
NS probe	Cp22	TTAAGACGGGAGTTCTG	2595-2579
NS probe	Cp25	CCAATCTGGAATTTCACCACC	1591-1611
VP probe	Cp4	TGAGGATCTTGTGCTCC	3084-3100
VP probe	Cp6	ATAGTAATTCAAGATGG	3990-3974
NS splicing	Cp24	CACTGACATCGCGTATAGCGC	2028-2008
NS splicing	Cp25	CCAATCTGGAATTTCACCACC	1591-1611
5′ Start NS3	Cp10	TCGTATGAACAGGAAA	912-897
5′ Start NS1-NS2	Cp30	AGCCAAACATTCTGTCTCGTC	1462-1442
Poly(A) NS1-NS2	Cp26	GGACAATTAGGACAAGC	2400-2416
5′ Start VP	Cp33	CCATAACCAAGTCCAGCTAAAC	4761-4782
Poly(A) VP	Cp20	TGGTGTACAACCAGTAC	3062-3046

Analysis of the sequence.

The sequence was analyzed by DNA Strider software (11) for visualizing the putative coding sequences. The QuickFOLD program (12; also http://dinamelt.bioinfo.rpi.edu/quikfold.php) was used to analyze the secondary structure of the ITRs; and the software Splice Site Prediction by Neural Network of the Berkeley Drosophila Genome Project (14; also http://www.fruitfly.org/seq_tools/splice.html) was used for detection of splicing sites. The search for protein homologies was made by BLAST programs at the NCBI website (http://www.ncbi.nlm.nih.gov/Blast.cgi) against the nonredundant protein database (2), and alignment of protein sequences was performed using the Multalin program (5). Phylogenetic trees of VP1 and NS-1 proteins were created using MEGA4 (17).

Nucleotide sequence accession number.

The complete nucleotide sequence of CpDNV was deposited in the GenBank database under accession no. FJ810126.

RESULTS

Cloning and sequencing of the viral DNA.

Virus was purified from infected larvae, and DNA was extracted as previously reported (9). CpDNV DNA was digested with different restriction enzymes, and the HindIII restriction map was obtained (Fig. 1A) (9). The four HindIII fragments of the genome were cloned in the plasmid pBSK⁺ and sequenced. The sequence of each fragment was determined for both strands, and a size of 5,759 nt was obtained by summing their sequences. This sequence was confirmed by primer walking along the viral DNA insert of plasmid pCpDNV or uncloned DNA (see Materials and Methods).

Analysis of the sequence.

Analysis of the viral sequence using DNA Strider software (11) identified potential coding domains on both strands, indicating that, like members of the genera Densovirus and Pefudensovirus, the CpDNV genome has an ambisense organization (Fig. 1B). However, the overall organization of NS and VP putative coding sequences differs significantly from that described in these two genera. A major ORF (ORF1) occupies the 5′ half of one strand and is preceded by an upstream small ORF (ORF1′) in the same frame. BLASTP analysis of ORF1 revealed that it corresponded to the VP gene. The complementary strand contains in its 5′ half a complex of six ORFs, all clustered on a stretch of 2.3 kb and distributed in the three frames. ORF2 and ORF3′ are in frame a, ORF4 and ORF3 are in frame b, and ORF5 and ORF2′ are in frame c. As detailed below, BLASTP analyses and mapping of messenger RNAs allowed elucidation of the complex organization of this strand bearing NS protein-coding sequences.

Structure of extremities.

Alignment of 5′- and 3′-terminal sequences showed that the 254 nt of the right (3′) extremity (nt 5506 to 5759) are fully complementary to the 254-nt sequence located between nt 285 and nt 32 of the left extremity (Fig. 2). The extremities thus correspond to long ITRs, and this alignment indicates that a deletion of the 31 terminal nucleotides of the right extremity of the genome likely occurred during the process of cloning. Attempts to sequence the 3′ extremity from purified viral DNA repeatedly failed. The structure of the left extremity contains several direct repeats 6 to 8 nt long. Assuming that no nucleotides were lost during the cloning of the left extremity, the ITR is 285 nt long. The predicted secondary structure of the ITR, using the QuickFold program, yielded eight different possibilities (data not shown), but the folding of the distal 64 nt was always identical. Owing to internal palindromes, an asymmetric, imperfect J-shaped hairpin structure could be generated (Fig. 3). Although the structure of this hairpin differs from that of members of Densovirus genus, it fits the theoretical requirement for functioning as a primer for initiation of DNA replication.

FIG. 2.

Sequence of the CpDNV genome showing the beginning and end of the major ORFs present on both strands, putative promoters (TATA boxes), ATG translational initiation and stop codons (bold characters) of NS and structural (VP) proteins, and polyadenylation signals (italics). The 5′-terminal and 3′-terminal nucleotides corresponding to the ITR are in italics. Arrows indicate direct repeats in the ITR sequence. Donor (D) and acceptor (A) splicing sites within the NS transcripts are shown. The C-terminal sequences of ORF2 and ORF3 and the N-terminal sequences of ORF2′and ORF3′ eliminated by the excision of the 53-nt intronic sequence (nt 1721 to 1773) are in italics. The blue and red letters refer to NS-1 and NS-2 amino acid sequences, respectively. Rolling circle replication and ATPase motifs of NS-1 and Ca⁺⁺-binding and phospholipase A2 motifs of VP1 are underlined.

FIG. 3.

5′ extremity of the ITR of CpDNV DNA showing the hairpin structure predicted by the QuickFOLD program.

Organization of NS coding sequences.

Among members of the genus Densovirus, the three ORFs encoding NS proteins are organized as follows: the leftmost ORF encoding NS-3 is in the same frame and is separated from the NS-1 ORF by a TAA stop codon, whereas NS-2 overlaps the 5′ half of the NS-1 ORF and starts in a different reading frame using an initiation codon just a few nucleotides downstream of the NS-1 AUG codon (20). In contrast, the organization of NS genes of CpDNV among the six imbricated ORFs located in the 5′ half of the NS strand appears more complex (Fig. 1B and 2). None of these ORFs has the coding capacity necessary for encoding the two NS-1 and NS-2 proteins with sizes in the range of 60 and 30 kDa, respectively, observed for other DNVs (20).

With a size of 384 nt (nt 413 to 796) and its first ATG codon at position 437, the leftmost ORF (ORF4) has a theoretical coding capacity of a 120-amino-acid (aa) protein. The TATATAA sequence at position 385, followed by a CAGT box (nt 418) 15 nt upstream of the first ATG codon, could serve as a promoter to drive expression of this protein. A second small ORF (ORF5) starts at nt 762, with an ATG codon at position 768, and ends at nt 1103, giving a potential capacity of a 112-aa protein. The 5′ extremity of this ORF overlaps the 3′ end of ORF4. No obvious promoter sequence was present upstream of the ATG, and no polyadenylation signals were present in the close vicinity of the ORF4 and ORF 5′ extremities. The next three ORFs (ORF2, ORF3, and ORF 2′) had larger coding capacities. ORF2 spanned from nt 1012 to nt 1752, and its first ATG codon at position 1057 was the putative translational initiation codon for a 232-aa protein. The 5′ extremity of this ORF overlapped the 3′ extremity of ORF5. ORF3 started at nt 1088 and ended at nt 1834. Its first ATG codon at nt 1109 was the putative translational initiation codon of a 242-aa protein. With a size of 992 nt, ORF2′ was the largest ORF of the NS protein-coding sequences. It started at nt 1761, 6 nt downstream of the TAA stop codon of ORF2, and terminated at nt 2753. In contrast to ORF2 and ORF3, first ATG codon of ORF2′ was located far downstream, at nt 2118. The sixth ORF, ORF3′, started at nt 1756, immediately downstream and in the same frame as ORF 2, and terminated at nt 1983. This ORF, which overlapped the 3′ extremity of ORF 3 and the 5′ extremity of ORF2′, was only 227 nt long, and its first ATG codon was at nt 1837. A search for putative promoters identified a TATA-like box at position 989 (TATAAAT), 23 nt upstream of ORF2, and a CAGT box at nt 1044, 9 nt upstream of the first ATG codon. No typical TATA box could be identified upstream of ORF3, ORF2′, and ORF3′, and no polyadenylation signal was present in the close vicinity of the ORF2, ORF3, and ORF3′ extremities. The only polyadenylation signal was located at nt 2733, 15 nt upstream from the ORF2′ end.

BLASTP analysis revealed that ORF2 and ORF2′, on one hand, and ORF3 and ORF3′, on the other hand, are homologous to the N-terminal and C-terminal sequences of the NS-1 and NS-2 proteins, respectively, of members of the genus Densovirus. This analysis indicated that the coding sequences of NS-1 and NS-2 are each split into two ORFs and implied that either splicing or a frameshift had to occur to generate both proteins. A search for splicing sites revealed two putative donor sites, CTG/GTA (splicing site indicated by the slash) at nt 1619 and ACG/GTA at position 1720, and one acceptor site at position 1774 (CAG/AT) would be able to put simultaneously ORF2 and ORF3 in phase with ORF2′ and ORF3′, respectively.

Structure of the VP ORF.

Located in the 5′ half of the complementary strand, ORF1 spanned over 2253 nt from nt 5020 to nt 2768. The first ATG codon is at position 4988, i.e., 12 codons downstream of the beginning of the ORF, giving a theoretical coding capacity of 739 codons and a protein of about 80 kDa. Some of the 15 downstream ATG codons are in an environment favorable for serving as a translational initiation codon. A search for a promoter revealed two putative TATA boxes, one upstream of ORF1 (TATATAT; nt 5051) and the second (TATATT; nt 4992) only 8 nt upstream of the first ATG. A putative polyadenylation signal (AATAAA) at position 2766 shared its two first nucleotides with the TAA stop codon of ORF1, a situation similar to that observed in JcDNV and GmDNV VP-encoding sequences (7, 22). Upstream of ORF1 and in the same frame, ORF1′ started at nt 5560 and terminated at position 5117. The first putative ATG initiation codon was at position 5494, i.e., 22 codons downstream of the beginning of this ORF. With its 5′ noncoding extremity located in the ITR and lacking a typical upstream TATA box, this ORF was unlikely to be functional. Furthermore, a BLASTP analysis failed to reveal any hit with the NCBI protein database.

To further elucidate the modalities of expression of the VP and NS genes, we extracted and analyzed the viral messenger RNAs (mRNAs).

Mapping of NS and VP transcripts.

Total RNA was extracted from heavily infected adult mosquitoes, and viral mRNAs were identified by Northern blot analysis. As illustrated in Fig. 4, an NS-specific probe revealed two distinct bands, approximately 2.4 kb and 1.8 kb in size, whereas the VP-specific probe revealed a single band of about 2.3 kb. The transcription starts and polyadenylation sites of these mRNAs were determined by reverse transcription-PCR with 5′ and 3′ rapid amplification of cDNA ends (see Materials and Methods), using the primers shown in Fig. 5A and listed in Table 1. The transcription starts of the 2.4-kb and 1.8-kb NS mRNAs were mapped to positions 416 and 1045, respectively (Fig. 5B and C). The poly(A) tract of both mRNAs was mapped to nt 2758, 2 nt downstream of the TAA stop codon of ORF2′ and 19 nt downstream of the predicted polyadenylation signal at position 2733 (Fig. 5D). Mapping of the predicted splicing at the junction between ORF2 and ORF3 and between ORF2′ and ORF3′ was achieved using the set of Cp24 and Cp25 primers (Fig. 5A). These primers generated a single amplicon with a size of about 350 nt, i.e., shorter than the expected size from the viral DNA sequence of 437 nt (data not shown). Sequencing of this amplicon showed that splicing occurred between the predicted donor site at nt 1720 and the acceptor site at nt 1774, excising a 53-nt intronic sequence and putting simultaneously in frame NS-1 and NS-2 N- and C-terminal sequences (Fig. 5F). The sizes of 2,289 nt and 1,660 nt, respectively, for the large and the small NS mRNAs, regardless the poly(A) tract, are in good agreement with those of 2.4 kb and 1.8 kb estimated by Northern blotting.

FIG. 4.

Northern blot analysis of the viral mRNAs extracted from heavily infected adult mosquitoes encoding the NS and capsid (VP) proteins. After separation by agarose gel electrophoresis and transfer onto a nylon membrane, the viral mRNAs were revealed by hybridization using a ³²P-labeled NS probe and a ³²P-labeled VP probe (Table 1).

FIG. 5.

(A) Transcription map of CpDNV genome showing the primers used to determine the transcription and polyadenylation starts of NS and VP transcripts and the splicing of NS transcripts. The positions and sequences of the primers are listed in Table 1. (B) TATA box of the P7 promoter and 5′-terminal sequence of the 2.4-kb mRNA encoding the NS-3 polypeptide. (C) TATA box of the P17 promoter and 5′-terminal sequence of the 1.8-kb mRNA encoding NS-1 and NS-2. (D) 3′-Terminal sequences of the NS and VP mRNAs showing the terminal 12-nt overlap of their sequences (underlined). Poly(A) tails are in bold characters. (E) TATA box of the P87 promoter and 5′-terminal sequence of the 2.3-kb mRNA encoding VP polypeptides. Note the very short 5′-terminal untranslated sequence of this transcript. (F) Detail of the sequence of the two species of NS transcripts showing how the splicing of the 53-nt intronic sequence between nt 1720 and 1774 puts in frame the N-terminal sequences of NS-1 (blue) and NS-2 (red) with their C-terminal sequences (italics).

Surprisingly, the transcription start of the VP mRNA was not mapped upstream of the first ATG codon but at nt 4960, only 3 nt upstream of the second ATG codon (Fig. 5E). The start of the 3′-terminal poly(A) tract was mapped at position 2746, i.e., 15 nt downstream of the predicted polyadenylation signal (Fig. 5D).

The size of 2,214 nt of this mRNA, excluding the poly(A) tract, fits well with that of 2.3 kb estimated by Northern blotting. The TATA box at position 4992, 35 nt upstream of the ATG codon, is thus more likely to be part of the P89 promoter driving expression of this transcript rather than the TATA box at position 5051. As illustrated in Fig. 5D, the 3′-terminal sequences of VP and NS mRNAs are overlapping on a stretch of 12 nt, a situation similar to that previously described for other members of the genus Densovirus.

Taken together, these results indicated that the two species of NS transcripts have different transcription starts and that each is under the control of a distinct promoter, with P7 (TATA box at position 385) controlling the synthesis of the 2.4 kb transcripts and P17 (TATA box at position 989) controlling the synthesis of the 1.8-kb transcripts. The function of the 2.4-kb mRNA is probably restricted to the encoding of NS-3, a 112-aa protein. However, the possibility that it could also code for a putative ORF5 protein by a leaky scanning mechanism or a fused ORF4-ORF5 NS-3 protein by a frameshift cannot be excluded. The logical function of the 1.8-kb mRNA is to encode NS-1 and NS-2 proteins, likely by leaky scanning, as postulated for the other DNVs (20). It is worth mentioning that, contrary to the situation prevailing in DNVs of the genus Densovirus where the NS-1 and NS-2 AUG codons are separated by only a few nucleotides, here the AUG codons are 52 nt apart.

Analysis of VP and NS proteins.

All the putative coding sequences of the CpDNV genome were submitted to BLASTP analysis using the nonredundant NCBI protein database. The best E-value (2e-101) for ORF1 was with Mythimna loreyi DNV (MlDNV) VP1 sequence (34% amino acid identity; 49% positive hits) followed by the GmDNV and JcDNV VP1 sequences. Alignments of VP1 sequences using the Multalin program showed that homologies between CpDNV and these viruses are distributed along the whole sequence (data not shown), including the highly conserved GPGN Ca⁺⁺-binding site and phospholipase A2 motif (Fig. 2) (24). Phylogenetic analysis of VP1 DNV proteins (Fig. 6A) clearly showed identities of the CpDNV VP gene with the other members of the genus Densovirus. Despite this similarity, no cross-reactivity could be detected between CpDNV and JcDNV anticapsid antibodies by Western blot analysis (9). The estimated size of 90 kDa for CpDNV VP1 capsid protein by SDS-polyacrylamide gel electrophoresis (PAGE) (9) is slightly greater than the calculated size of 78,889 Da of the 730-aa residues of VP1. BLASTP analysis of ORF4 and ORF5 coding sequences failed to reveal any significant homologues within the NCBI database. In contrast, significant identities were detected between CpDNV NS proteins and NS-1 and NS-2 proteins of other DNVs. The highest E-value for NS-1 was with JcDNV (6e-151; 53% identity and 69% positive hits) and with MlDNV for NS-2 (3e-45; 40% identity and 58% positive hits). As for VP genes, alignments of NS-1 and NS-2 with their orthologs in the genus Densovirus showed that the identities spanned the whole sequence (data not shown). The homologies included the highly conserved NS-1 rolling circle replication and ATPase motifs (Fig. 2). Phylogenetic analysis of NS-1 protein confirmed the clustering of CpDNV with members of the genus Densovirus (Fig. 6B).

FIG. 6.

Neighbor-joining phylogenetic trees of NS-1 (A) and structural VP-1 (B) proteins of CpDNV and their orthologs in DNVs. For the phylogenetic trees the abbreviations of the organisms for the respective DNVs are as follows(GenBank accession numbers for the NS and VP trees are given in parentheses in respective order): Ml, Mythimna loreyi (NP_958099.1; NP_958101.1); Aal, Aedes albopictus (CAA52899.1 and, for C6/36 cells, AAM28943.1; CAA52901.1 and, for C6/36 cells, AAM28945.2); Ds, Diatraea saccharallis (O71153.1; O71155.1); Jc, Junonia coenia (Q90054.1; Q90053.1); Ce, Casphalia extranea (NP_694838.1; NP_694840.1); Cpp, Culex pipiens pallens (ABU95011.1; ABU95013.1); Dp, Dendrolimus punctatus (YP_164339.1; YP_164341.1); Gm, Galleria mellonella (NP_899650.1; Q90125.1); Bm, Bombyx mori (NP_542609.1; NP_542611.1); Ad, Acheta domestica (YP_227600.1; YP_227602.1); Bg, Blattella germanica (NP_874381.1; NP_874380.1); Mp, Myzus persicae (NP_874376.1; NP_874377.1); Pc, Planococcus citri (NP_694843.1; NP_694842.1); Pf, Periplaneta fuliginosa (NP_051020.1; BAA82963.1); Aae, Aedes aegypti (P27454.1; P27453.1). Distances were calculated using Poisson corrections. Homogeneous substitution pattern among lineages with gamma distributed rate among sites (gamma parameter, 2.25) was employed for reconstruction of the trees. The robustness of the tree was tested using bootstrap analysis (500 replicates). Numbers on the nodes indicate bootstrap values. The bona fide members of the four genera of Parvovirinae subfamily are in curly brackets.

DISCUSSION

Several DNVs infecting different species of mosquitoes or mosquito cell lines have been reported, and the complete sequences of their genomes have been published (1, 3, 25). Owing to a very similar organization of their small genomes and probably similar modes of expression, they are all considered as bona fide members of the genus Brevidensovirus (19). The data presented here about the CpDNV genome clearly demonstrate that by its size, the structure of termini, and the way its genes are organized and expressed, this genome differs totally from the brevidensovirus model and represents a new type of mosquito DNV. In contrast, CpDNV shares the main characteristics of lepidopteran DNV genomes belonging to the genus Densovirus: an ambisense organization, the presence of ITRs, and significant identities at both capsid and NS gene levels. However, CpDNV deviates from this model in at least four domains. The first concerns the terminal hairpin of the ITR. Assuming that the sequence of the 5′ extremity is complete, the double-stranded structure of the 3′ extremity spans only 19 nt. This sequence is not only shorter than that of the GmDNV or JcDNV genome but also lacks short (GAC) repeats assumed to function as an NS-1 binding site (6, 22). The second difference concerns the capsid (VP) proteins. SDS-PAGE of CpDNV revealed two major (VP2 and VP3) proteins with very close molecular sizes (64 and 57 kDa) and two minor polypeptides, VP1 and VP4, with estimated sizes of 90 and 12 kDa, respectively (9). VP4 is not sufficiently large to be one of the 60 capsid proteins and could be a proteolytic cleavage product of one of the larger VPs. The size pattern and relative amounts of the structural proteins reflect differences in the frequencies at which the ribosomes sliding down the VP mRNAs by a leaky scanning mechanism initiate the translation at the first AUG codon and downstream AUGs and in their relative strengths as initiation codons. This, in turn, indicates that the ratio of the VP proteins involved in the capsid assembly differs from that of members of the genus Densovirus (20) and could explain why no antigenic relationships were observed between CpDNV and JcDNV. The very short (3 nt) sequence upstream of VP1 is probably responsible for the small amount of VP1 in CpDNV capsid as revealed by SDS-PAGE analysis (9). The third difference concerns the split into four ORFs of the NS-1 and NS-2 coding sequences, a feature unique among DNVs. The consequence of this organization on NS-1 and NS-2 expression is unclear, but it implies a splicing mechanism. The fourth difference concerns the presence of two promoters to control the expression of NS genes, whereas in the genus Densovirus the three NS genes are under the control of a single promoter, and gene expression is regulated by alternative splicing (20). Another important difference between the NS and VP promoters of CpDNV compared to those of members of genus Densovirus is in the location of their TATA boxes. Both the leftmost (P7) and rightmost (P87) CpDNV promoters are located downstream instead of being just at the border of ITRs. Taken together, these structural differences strongly suggest that CpDNV differs sufficiently from other DNVs in both its replication and expression strategy to be considered as the prototype of a possible new genus within the subfamily Densovirinae.