Short regions of sequence identity between the genomes of human and rodent parvoviruses and their respective hosts occur within host genes for the cytoskeleton, cell adhesion and Wnt signalling

Parvoviruses are small, single-stranded DNA viruses that cause a broad spectrum of disease in their host species. Disease may be caused by direct infection and lysis of rapidly dividing, target host cells or indirectly through the action of a specific immune response (Bloom & Kerr, 2006). Either of these aspects may be prolonged following the acute phase. In addition, parvoviral modification of host-cell functions through epigenetic phenomena has been demonstrated (Iseki et al., 2005). This may account for the differential expression of six human genes in parvovirus B19 (B19V)-seropositive versus -seronegative persons (Kerr et al., 2005). The existence of such phenomena has been suggested previously for persistent virus infections (de la Torre et al., 1991; Oldstone, 1998).

Parvoviruses persist following acute infection (Lefrère et al., 2005), but the precise mechanism(s) are not understood. Adeno-associated virus (AAV) is known to integrate in a site-specific manner into chromosome 19 in human cells in vitro (Kotin et al., 1990, 1992) and minute virus of mice (MVM) has been shown to integrate into episomes in a site-specific manner in vitro (Corsini et al., 1997). However, the general means by which human and animal parvoviruses persist in vivo are incompletely understood.

It is possible that sequence identity may exist between particular parvovirus species and their respective host species and may be biologically relevant to the pathogenesis and/or persistence of the parvoviruses. Recently, several reports (Bennasser et al., 2005; Cai et al., 2005; Grey et al., 2005; Pfeffer et al., 2004; Sullivan et al., 2005) have highlighted the importance of host-gene modulation by viral microRNAs (miRNAs) in order to achieve optimal conditions for virus replication by RNA silencing. Viruses with small genomes, such as parvoviruses, are limited by the very small number of proteins they produce and therefore interfering RNAs may be important in their life cycle.

With the recent availability of complete genome sequences of human, rat and mouse, and the ability to search these sequences and precisely pinpoint matching regions within the sequences, it is possible to characterize this phenomenon precisely. In this study, we used the resources of the NCBI and the Celera Discovery System (CDS) to demonstrate short (1726 nt) regions of identity between several human and rodent parvoviruses and their respective host species, noted the metabolic pathways involved in each virushost interaction and demonstrated that these sequences appear to be conserved in each parvovirus species studied.

Initial BLAST searches of various representative rodent and human parvovirus genomes against the genomes of their various host species were performed using the resources of the NCBI (). Details of the observed homologies were catalogued further using CDS; criteria for inclusion and further study were that the sequence had 100 % identity with at least 17 nucleotides, as this is accepted to be the minimum number of nucleotides required for unique recognition in eukaryotes. Parvovirus genomes examined at this stage comprised AAV type 2 (AAV-2) (GenBank accession no. J01901[GenBank] ); B19V, strain Au (B19-Au) (GenBank accession no. M13178[GenBank] ); mouse parvovirus type 1 (MPV-1) (GenBank accession nos U12469[GenBank] ); minute virus of mice, prototype strain (MVMp) (GenBank accession no. J02275[GenBank] ); minute virus of mice, immunosuppressive variant (MVMi) (GenBank accession no. X02481[GenBank] ); Kilham rat virus (KRV) (GenBank accession no. U79033[GenBank] ); and rat parvovirus type 1 (RPV-1) (GenBank accession no. AF036710[GenBank] ). Host genomes examined were those of Homo sapiens, Mus musculus and Rattus norvegicus. For each parvovirus, each identical region was noted and numbered. A BLAST search was performed in CDS for each identical region against the respective host genome in which it had been found and the particular gene hits were recorded for each. NCBI and Celera default parameters were used in the searches and any potential hits were recorded manually. Each parvovirus sequence was also used to BLAST search non-respective host genomes.

The degree of conservation of the identical sequences within each parvovirus species was assessed using all of the available sequences for that parvovirus species on NCBI. These were then aligned using CLUSTAL W version 1.83 (), and the percentage identity within these regions was recorded. Parvovirus species sequences included in this analysis were as follows: AAV-2 (GenBank accession nos J01901[GenBank] , NC_001401[GenBank] and AF043303[GenBank] ), B19V (GenBank accession nos M13178[GenBank] , AY386330[GenBank] , Z68146[GenBank] , NC_000883[GenBank] , AF162273[GenBank] , AB126270[GenBank] , AB126271[GenBank] , AY504945[GenBank] , AB126269[GenBank] , AB126262[GenBank] , M24682[GenBank] and AB126268[GenBank] ), MPV (GenBank accession nos U12469[GenBank] , U34253[GenBank] , U34254[GenBank] and NC_001630[GenBank] ), MVMp (GenBank accession nos J02275[GenBank] and NC_001510[GenBank] ), MVMi (GenBank accession no. M12032[GenBank] ), KRV (GenBank accession nos AF321230[GenBank] , U79033[GenBank] , AF036711[GenBank] and AF317513[GenBank] ), RPV (GenBank accession nos AF036710[GenBank] and AF317513[GenBank] ).

For each interaction examined, gene ontology information was obtained for all host genes containing the respective identical sequence using the Applied Biosystems PANTHER resource (). Those pathways that were implicated in parvovirushost interactions were examined further in terms of the known pathogenesis of the various parvovirus infections and the gene homologues between different host species noted. A binomial statistics tool was then used to compare classifications of multiple clusters of gene lists with a host gene reference list to determine statistically over- or underrepresentation of gene classification categories. Each list was compared with the reference list using the binomial test (Cho & Campbell, 2000) for each pathway term in the PANTHER database. This was performed for the list of genes containing identical regions for each parvovirus species examined within each parvovirushost system.

Homologous sequences for each virushost system examined were searched using BLAST against known miRNA sequences using the miRBase site (). Each homologous sequence was used to search using BLAST against both mature mirnas and stemloop sequences using the SSEARCH algorithm, which is useful for finding short sequences within the library, using the default e-value cut-off of 10. The percentage identity to known miRNAs was noted.

Short regions of sequence identity were found between all parvovirus genomes and their respective host species (Fig. 1 and Table 1). These occurred in different locations within the parvovirus genome and ranged from 17 to 26 nt. They were found in either the plus or minus orientation in the respective host genes. Each of these sequences had a variable number of gene hits within the respective host genome, ranging from one to 50 (Table 1). Almost all were non-coding sequences, but an occasional identical sequence was found within a host gene-coding sequence. For example, in the case of the B19Vhuman interaction, two out of 72 identical sequences occurred within gene-coding regions. There was some similarity between parvovirus species and non-respective hosts, but this was minimal and there was a marked contrast between sequence identity in respective versus non-respective hosts (data not shown).

Table 1. Regions of sequence similarity have been shaded to indicate the degree of conservation in the viruses (closed boxes, 100 % conservation; open boxes, less than 100 % conservation; shaded boxes, only one sequence available).

Table 1. Virus and host sequences, showing regions of sequence identity (17 nt) The number of gene hits in the respective host genome, the location of the sequence in each parvovirus genome and the percentage conservation in each parvovirus are shown.

Regions of sequence identity occurred in various locations within the parvovirus genome, including non-structural genes (n=14), capsid genes (n=17) and non-coding regions (n=5) (Table 1). Alignment of multiple sequences for particular parvovirus species revealed a significant degree of conservation of the most similar sequences; in 23 cases, there were two or more genome sequences for alignment, among which 16 were 100 % conserved (Table 1). These sequences involved no recognized protein domain in any parvovirus studied.

It is interesting that the number of MVMp and MVMi sequences in the rat was much greater than that in the natural mouse host. It is difficult to explain this, as there was no apparent difference in host-gene theme(s) involved in the rat compared with the mouse.

Table 2 shows the predominant pathways of genes containing regions of sequence identity for each virushost interaction and highlights the importance of the cytoskeleton, cell adhesion and the Wnt pathway. Pathways that show statistically significant over- or underrepresentation are highlighted.

Table 2. Host genes containing regions of parvovirushost sequence identity for each parvovirushost system and the metabolic pathways within which these genes function The respective parvovirus gene within which each sequence occurs is shown in superscript. Shaded boxes indicate significantly over- and underrepresented pathways compared with all genes in that particular host (P0.05).

Regions of sequence identity were found to be quite unrelated to both mature miRNA sequences (% identity: range 3968 %; mean 50.1 %) and stemloop sequences (% identity: range 5282 %; mean 67.8 %) (Table 3). None of these was an exact or nearly exact match.

Table 3. Percentage sequence similarity between parvovirushost homologous regions and known mature and stemloop microRNA genes in the respective host genome

Table 4 shows all 72 gene hits for the five regions of sequence identity in B19V, illustrating the occurrence of multiple copies of each sequence within the host genome and the fact that such regions are distributed widely.

Table 4. Regions of sequence similarity for human parvovirus B19 within the human genome: location in B19 genome, human genes and identifiers, numerical location, length of sequence, percentage similarity, coding capacity, GenBank numbering and the Celera exon name, if applicable

This study was undertaken to investigate the possibility of sequence identity between the genomes of the parvoviruses and their respective host species, as has been shown for certain members of the herpesviridae (Holzerlandt et al., 2002), and to make a preliminary assessment of the possible biological relevance of this phenomenon in terms of parvovirus pathogenesis and persistence.

Short regions of sequence identity were found for each parvovirushost system studied. In each parvovirus species, these regions were highly conserved. In the host, these regions were generally non-coding. However, this does not rule out a possible role in promotion of virus persistence through homologous recombination or regulatory effects on host transcription, possibly via mRNA splicing or RNA interference. Viruses tune the metabolism of their hosts for their own purposes. This may be done with interfering proteins (inhibitory proteins and transcription factors), but viruses with small genomes may use their transcribed RNAs to silence host genes. Depending on the grade of complementarity, the host RNA may be cleaved or translation may be downregulated. In each host species, these regions clustered within several functional categories of which some are known to be important in parvovirus infections, such as the cytoskeleton (Kerr et al., 2005; Suikkanen et al., 2003) and cell adhesion (Bantel-Schaal & Stohr, 1992; Kerr et al., 2005; Ueno et al., 2001; Weigel-Kelley et al., 2003).

Parvoviruses have been shown to be oncosuppressive in vivo and this has been demonstrated by the prevention of tumour establishment, reduction or arrest of tumour growth, regression of established tumours, diminished uptake of transplantable tumour cells and prolongation of the life of tumour-bearing animals (Cornelis et al., 2006). Thus, parvoviruses are being investigated as anti-cancer agents. One approach makes use of the fact that the Wnt pathway is constitutively activated in colon cancer cells through certain mutations. Hybrid MVM/H-1 parvoviruses have been engineered to contain a modified MVM P4 promoter to confer sensitivity to β-catenin/Tcf complexes and it has been shown that replication of these mutant viruses only occurs in the cancer cells; these vectors are being developed for colon cancer therapy (Malerba et al., 2003, 2006). It is therefore interesting that the Wnt pathway was highlighted in the present study and, in the case of the MVM immunosuppressive parvovirusrat system, even involved the rat Tcf3 gene (Table 2).

The possibility that these homologous sequences may represent miRNA molecules seems unlikely, as they showed limited sequence homology to both mature and stemloop (precursor) miRNA molecules and did not fulfil the structural requirements of miRNAs. However, as newly discovered PIWI-interacting RNAs (piRNAs) also do not fulfil these requirements, a possible regulatory role cannot be ruled out (Cox et al., 1998; Sharma et al., 2001). It is interesting that the homologous sequence, B19-human-1, is contained in the human PIWIL1 gene (GenBank accession no. NM_004764[GenBank] ), which encodes a member of the PIWI subfamily of Argonaute proteins that contains both PAZ and PIWI motifs and plays important regulatory roles in stem-cell renewal, RNA silencing and translational regulation (Girard et al., 2006).

These findings may indicate a potential for RNA silencing by parvoviruses and may also facilitate persistence in some way. However, both of these possibilities remain to be confirmed. Small DNA viruses (polyomaviruses, papillomaviruses and parvoviruses) are species-specific and appear to have evolved with their host species. Their genetic stability has been attributed to their ability to maintain a benign persistent state in vivo, which involves a cell-cycle-regulated episomal state. To achieve benign inapparent viral persistence, small DNA viruses may circumvent the host acute-phase reaction by mechanisms that are evolutionarily adapted to the immune system and the related cytokine communication networks (Shadan & Villareal, 1995). Therefore, these homologous sequences may also have relevance for benign persistence. It is interesting in this regard that such sequence homology has also been documented for the herpesviruses (Holzerlandt et al., 2002).

In conclusion, we have described short regions of virushost homology that occur in clusters of host genes within particular pathways of known importance in parvovirus infections of humans and rodents. The precise significance of this finding for parvovirus pathogenesis and persistence and the possible mechanisms involved remain to be determined.

The authors would like to acknowledge the helpful advice of Professor Luc Montagnier.