Parvoviruses are small, non-enveloped icosahedral viruses. These viruses are ubiquitous, and they have been isolated from many animal species. Based on their host range, viruses of the family Parvoviridae are divided into two subfamilies: Parvovirinae, which includes viruses that infect birds and mammals and Densovirinae, which includes viruses that infect insects and other arthropods [10]. The Parvovirinae and Densovirinae viruses do not share any nucleotide sequence homology. The International Committee on Taxonomy of Viruses has further classified the Parvovirinae viruses into five genera: Dependovirus, Bocavirus, Erythrovirus, Parvovirus and Amdovirus. Recently, a new genus, Hokovirus, was proposed to include a group of newly identified porcine isolates (designated PPV3 in this report) and bovine isolates (designated BPV4 in this report) that clustered with the human PARV4 and PARV5 viruses [14].

The genomes of members of the subfamily Parvovirinae are linear single-stranded DNA genomes of about 5 kilobases that contain terminal palindromic sequences. In general, the genome contains two major open reading frames (ORFs), coding for the non-structural protein(s), located at the 5′-end, and the capsid protein(s), located at the 3′-end. An additional ORF3 has been observed among viruses of the genus Bocavirus [bovine parvovirus (BPV1), canine minute virus (MVC) and human bocavirus (HBoV)], and ORF3 is located in the middle of the viral genome between ORF1 and ORF2 [2, 10]. Recently, it was reported that ORF3 of MCV is essential for viral DNA replication [20]. Parvoviruses have been proposed to replicate their genomes via a modified rolling hairpin model that involves a hairpin transfer mechanism [3, 5, 8]. In this model, two obligatory dimeric genome replicative intermediates (RF) generated via the terminal palindromic sequences are synthesized during viral DNA replication [6]. These dimeric RFs contain a sense-strand and a minus-strand, i.e., a double-stranded genome, that are covalently linked (via terminal palindromic self-priming) either at the left end or right end of the linear genome and give rise to head-to-head or tail-to-tail concatameric molecules.

In this work, we cloned a novel porcine parvovirus from a circular or a head-to-tail concatameric template. Since it is the fourth (nearly) complete DNA sequence of a porcine parvovirus described, this virus is designated PPV4. Phylogenetically, PPV4 does not cluster with the three previously identified porcine parvoviruses but exhibits limited homology to bovine parvovirus 2 (BPV2) [1]. Incidentally, a recent publication based on partial nucleotide sequence reported identification of a novel porcine boca-like virus [4] among pigs co-infected with porcine circovirus type 2 (PCV2). The ORF3 of this porcine boca-like virus and PPV4 share low nucleotide identity (45%) and amino acid identity (17.5%), which indicate that this boca-like virus and PPV4 are two distinct parvoviruses.

PPV4 was identified among swine suffering from an acute-onset disease of high mortality in North Carolina during late 2005 [7]. Case reports of this disease indicated that various viral and bacterial pathogens were identified in tissues from swine infected with PCV2. PCR analysis showed that porcine reproductive and respiratory syndrome virus, porcine respiratory coronavirus, and several different serovars of porcine enterovirus were isolated from one or more of the affected herds. PCV2 was identified in essentially all pigs from each of the herds. Tissue homogenates consisting of lung, lymph node, spleen, and heart were prepared from fresh frozen tissues of the affected swine. The tissues were triturated with sterile sea sand via mortar and pestle. Serum-free minimum essential medium was used to reconstitute a 10% solution by weight. The solution was clarified at 1,000×g for 10 min and filtered through a 0.45-micron filter. Two colostrum-deprived cesarean-derived (CDCD) pigs, pigs 753 and 754, were inoculated oronasally with 2 ml of the 10% tissue homogenates and monitored for clinical disease. Both pigs developed severe respiratory disease and had to be euthanized. Pig 754 was euthanized on day 16 post-infection and pig 753 was euthanized on day 21 post-infection.

To explore whether there were any unknown viruses present in the diseased pigs, a random shotgun sequencing experiment was conducted. RNA from a lung lavage sample of pig 753 was extracted, randomly amplified and cloned into pCR4.TOPO (Invitrogen, Carlsbad, CA, USA) as described previously [23] and sequenced using standard Sanger chemistry [17] on an ABI 3730xl sequencer. The resulting sequence reads were trimmed to remove vector and primer sequences and then assembled using Phred/Phrap [9]; (Phil Green, http://weeds.mgh.harvard.edu/goodman/doc/). Contigs with sequence similarity to known viral sequences were identified using tBLASTx. Three hundred thirty-four clones were sequenced. Alignment of the sequence reads to the GenBank NR database yielded 19 sequences with limited identity to viruses in the family Parvoviridae. Nucleotide sequences belonging to other virus families were not detected. The 19 parvovirus-related sequences were assembled by Phred/Phrap into two contigs that were experimentally verified by generating PCR amplicons the length of the contigs. Deduced amino acid sequence homology was analyzed with the translated BLAST search. Contig 1 was 1,815 base pairs (bp) long and had 34% amino acid identity to its top hit, BPV2 capsid protein. Contig 2 was 2,993 bp long and had 39% identity to the non-structural protein of BPV2. Primers at nt 3,681–3,702 and nt 4,206–4,183 (based on c17 of Fig. 1) were then designed to span the gap between the two contigs. PCR was conducted with total cellular DNA from pig 753 that was prepared using a STAT-60 DNA extraction kit purchased from Tel-Text B, Inc. (Friendswood, TX, USA). The PCR reaction was carried out with 0.1 μg of DNA in the presence of 10 mM Tris–HCl (pH 8.3), 0.2 mM each of the four deoxynucleotide triphosphates, 100 pM each of the primers, and 2.5 U of Taq polymerase. The reaction mixture was heated at 94°C for 1 min, amplified for 34 cycles at 94°C for 15 s, 55°C for 30 s, and 68°C for 5 min 15 s, and then kept at 68°C for 7 min. The resulting contig was 4,851 bp in length.

Fig. 1
figure 1

a A schematic diagram denoting the PCR oligonucleotide primers and the strategy to obtain the head-to-tail covalent linkage nucleotide sequence (indicated by dotted lines) from a circular molecule. Nested PCR was performed. The first-round PCR was conducted with primers X and Y (data not shown). Bottom panel the second-round nested-PCR was conducted using a common primer C with either primer 1, 2 or 3, as indicated at the top of each lane of the agarose gel. b Schematic diagram of plasmid p17, which contains the PPV4 c17 genome inserted into pCR4.TOPO. ORF1, ORF2 and restriction enzyme sites of KpnI and PstI are denoted. Bottom panel agarose gel analysis of plasmids p14 and p17. (i) PCR products obtained using two divergent primers, PKpnF and PKpnR, at the KpnI site of c17. (ii) The restriction enymes (KpnI and PstI) used are indicated at the top of each panel. KpnI digestion releases the entire double-stranded PPV4 genome (5.9 kb) from the cloning vector pCR4.TOPO (4.0 kb). PstI digestion yielded four fragments, labeled I–IV, and DNA fragment III, which differs between c14 and c17, is indicated by asterisk

Full size image

Based on the DNA sequence obtained from the initial random amplification and the above PCR experiment, diverging primers were designed to obtain additional nucleotide sequences upstream and downstream of the 4,851-nucleotide contig. Nested PCR was conducted with the pig 753 lung lavage DNA. In first-round PCR, primers X (tga acc cat aca ttg tac ca) and Y (gtc tac ttc ttc taa ctg ca) were used (Fig. 1a). In second-round PCR, an aliquot of the first-round products was amplified with the common primer C (ctc cgc ctg ccg aac agc ga), and one of three primers: #1 (ttt cca gat tac atg ggt tc), #2 (tga gac tcc tgg cac aga tg) or #3 (tta cat atg aaa tgg aat gg). In each case, a prominent PCR product of the expected size was obtained. The results demonstrated that a PPV4 molecule in head-to-tail configuration was present in the pig 753 DNA sample. To obtain the unit-length head-to-tail PPV4 sequence, two abutting primers of opposite polarity located at the unique KpnI site of the viral genome were used for PCR. The primers used were PKpnF-gga agt ggt acc ttc tct tac aca tca ac (forward orientation) and PKpnR-gga agt ggt acc tgt ttt cag ttg gtg at (reverse orientation), and the KpnI site is located at nt 3,342–3,348 of the clone 17 (c17) sequence. After amplification, several PCR products were visualized on an agarose gel. One of the products was approximately 6,000 bp long, and the rest were less than 4,000 bp long (Fig. 1b). The 6,000-bp PCR product was excised from the gel and cloned into the pCR4.TOPO plasmid. A miniprep plasmid DNA kit (Promega, Madison, WI, USA) was used to isolate nucleic acids from overnight TOP10 (Invitrogen, Carlsbad, CA, USA) bacterial cultures, and DNA electrophoresis was carried out in 1% agarose gel. Two plasmids (p14 and p17) containing the PPV4 DNA clones (c14 and c17), respectively, were obtained. As shown by the PstI restriction enzyme analysis, c14 is slightly smaller than c17. The DNA clones were then sequenced using primer walking. It was confirmed that c14 is 125 nucleotides (nt) shorter than c17 and the deleted sequence is located at nt 127–251. In addition, there are 15 nt differences between these two clones. The GenBank accession numbers for c14 and c17 are GQ387500 and GQ387499, respectively. PCR was also conducted with total cellular DNA purified from the lung lavage of pig 754. The partial PPV4 nucleotide sequence obtained was essentially identical to the parvovirus DNA sequence derived from pig 753. No further experimentation was conducted with the viral sequences from pig 754. To investigate whether PPV4 was present in the field cases of the 2005 outbreak, all of the pigs examined in the original study [7] were tested for PPV4 by PCR. The results showed that only two pigs (n6 and n10) of Farm 1 from North Carolina were positive for PPV4.

Phylogenetic analysis was performed on the PPV4 nucleotide sequences using full-length or nearly full-length genomes from all species of the subfamily Parvovirinae as described previously by Lau et al. [14]. Nucleotide and amino acid sequences were aligned with CLUSTAL, and phylogeny was inferred using MEGA4 [21]. The evolutionary history was inferred using the Neighbor-Joining method [16]. The evolutionary distances were computed using the Maximum Composite Likelihood method [22] and are in units of number of base substitutions per site. The bootstrap consensus tree inferred from 10,000 replicates was taken to represent the evolutionary history of the taxa analyzed [11]. Branches corresponding to partitions reproduced in fewer than 50% of the bootstrap replicates were collapsed. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (10,000 replicates) is shown next to the branches [11]. Pairwise global comparison of putative amino acid sequences coded for by ORF1, ORF2, and ORF3 were conducted by aligning the sequences for each ORF using the Needleman-Wunsch algorithm [13, 15]. Based on comparative nucleotide sequence, PPV4 did not cluster with any of the known parvoviruses, including the three previously identified porcine isolates (Fig. 2). Instead, PPV4 is most similar to BPV2. Separately, phylogenetic analyses were conducted with the deduced ORF1 or ORF2 amino acid sequences, and essentially identical results were obtained (data not shown).

Fig. 2
figure 2

Evolutionary relationships of PPV4 and other related parvoviruses. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. All positions containing gaps and missing data were eliminated only in pairwise sequence comparisons (pairwise deletion option). There were a total of 6,684 positions in the final dataset. Phylogenetic analysis was conducted using MEGA4 [21]. The porcine parvoviruses are denoted in shaded boxes

Full size image

The c17 PPV4 clone is 5,905 base pairs long. Since the nucleotide sequences of the left and right ends of the linear viral genome have not been determined, the ends of the genome are arbitrarily assigned so that the genome conforms to the general organization of parvoviruses having two major ORFs: ORF1 located at the 5′-end and ORF2 at the 3′-end. PPV4 has an additional ORF3 located between ORF1 and ORF2, which is a feature of the viruses belonging to the genus Bocavirus. Analyses were conducted to compare the ORFs of PPV4 with those of other PPVs and BPVs and members of the genus Bocavirus at the amino acid level.

  1. 1.

    ORF1: PPV4 ORF1 is 1,794 nt long and capable of coding for a protein of 588 residues. BLAST search showed that it is most closely related to the putative nonstructural protein of BPV2 and the replication proteins of adeno-associated viruses 1 (AAV-1) (45.8% identity) (GenBank accession number FJ872544) of the genus Dependovirus. It shares 33.6% amino acid residue identity with BPV2 and less with the other porcine and bovine parvoviruses. PPV4 shares 17.4–20.5% amino acid identity with the bocaviruses, while the bocaviruses share 24.8–33.5% identity among themselves.

  2. 2.

    ORF2: PPV4 ORF2 is 2,184 nt long and capable of coding for a protein of 728 residues. BLAST search showed that it is most closely related to the putative structural protein of BPV2 and the capsid proteins of various parvoviruses. It shares 24.5% amino acid residue identity with BPV2 and less with the other porcine and bovine parvoviruses. PPV4 shares 18.7–20.3% amino acid identity with the bocaviruses, while the bocaviruses share 41.4–45.6% identity among themselves.

  3. 3.

    ORF3: PPV4 ORF3 is 612 nt long and capable of coding for a protein of 204 residues. BLAST search showed that ORF3 did not have significant homology to any protein in the GenBank non-redundant protein database. With respect to coding capacity, it is similar to that of the bocaviruses. The ORF3 of BPV1, HBoV and MCV is capable of coding for a protein of 213, 219 and 188 residues, respectively. The PPV4 ORF3-encoded protein shares only 4.9–11.2% amino acid identity with the bocaviruses, while the bocaviruses share 43.3–47.0% identity among themselves. Whereas the PPV4 ORF3 does not overlap with either major ORF, MVC overlaps with ORF1, HBoV overlaps ORF2 and BPV1 overlaps with both ORF1 and ORF2.

In summary, we report the identification of a novel porcine parvovirus, PPV4. In this study, a 5,905-nt PPV4 sequence was obtained after PCR amplification using two diverging primers from the unique KpnI site of the viral genome. Thus, the templates for the generation of the genomic clones, c14 and c17, were either circular or head-to-tail concatameric molecules in which the left and right ends of the linear genome were covalently linked. Based on the genome organizataion of parvoviruses, it is expected that both the left end and the right end of the linear PPV4 genome are located between nt 1 and 848 (Fig. 1b). However, it is possible that some terminal nucleotides of the linear genome are missing in c17 and c14 during formation of the head-to-tail configuration. The genomes of parvoviruses are single-stranded linear DNA molecules; however, it has been reported that the genomes of AAV-2, a member of the genus Dependovirus, can persist in tissues as head-to-tail monomeric and concatameric circles [18, 19]. The presence of head-to-tail concatameric PPV4 genomes in tissues of infected animals suggests that this virus may be able to establish persistence in its host.

Phylogenetic analysis showed that the porcine and bovine parvoviruses exhibit high degrees of diversity. In general, the porcine or bovine isolates do not cluster phylogenetically with their respective viruses from swine or cattle. Among the porcine parvoviruses, PPV1 belongs to the genus Parvovirus, PPV3 belongs to the proposed genus Hokovirus, PPV4 and the porcine boca-like virus [4] resemble members of the genus Bocavirus, and PPV2 does not cluster with any known isolates. A similar phenomenon has also been reported for the bovine parvoviruses [14]. There are four bovine parvoviruses, but they also do not cluster together phylogenetically (Fig. 2). BPV1 belongs to the genus Bocavirus, BPV2 and BPV3 [1, 10] do not cluster with any other known isolates, and BPV4 belongs to the proposed genus Hokovirus. Interestingly, one porcine (PPV3), one bovine (BPV4) and two human parvoviruses (PARV4 and PARV5) are closely related and form the proposed genus Hokovirus [14].

Our analysis revealed several distinct characteristics of PPV4. Whereas PPV4 is most closely related to BPV2 phylogenetically, the coding capacity and genome organization of PPV4 are more similar to those of viruses of the genus Bocavirus. PPV4 is most closely related to BPV2 with respect to ORF1 and ORF2 phylogenetically; however, BPV2 does not contain a distinct ORF3 situated between the two major ORFs. Similar to the bocoviruses, PPV4 encodes an additional ORF3 located in the middle of the genome, but the deduced protein of PPV4 ORF3 is quite distinct from the ORF3-encoded proteins of the bocaviruses.

The pathogenic potential of porcine parvoviruses is not fully understood. PPV1 is ubiquitous in swine, and it causes reproductive failure in pregnant females [24]. PPV2 has been detected in swine serum but has not been incriminated in any disease [12]. The pathogenic nature of PPV3 and PPV4 remains to be determined. PPV3, a member of the proposed genus Hokovirus, has been found in tissues of healthy and sick pigs [14]. In this study, PPV4 was identified in a co-infection with porcine circovirus type 2 (PCV2) from field cases of severe PCV2-associated disease. PCV2 has been associated with multiple disease syndromes in swine and is the primary agent leading to postweaning multisystemic wasting syndrome (PMWS). It has been reported that other pathogens enhance the severity of PMWS during co-infection with PCV2. From this study, it is not clear whether PPV4 can cause disease on its own or if PPV4 contributed to the disease phenomenon in the CDCD pigs. At any event, the detection of PPV4 among diseased pigs that exhibit PCV2-associated PMWS warrants further investigation into the pathogenic nature of PPV4.