Molecular characterisation of virulence graded field isolates of myxoma virus

Myxoma virus (MV) causes myxomatosis in the European rabbit (Oryctolagus cuniculus). The disease is a recurrent problem in rabbit farms and in wild rabbit populations throughout Europe [1‐4]. Due to the unique manner of its introduction into the European rabbit population, MV provides an excellent model for studying the coevolution of a virus and its host. The complex virus/host relationship has been shown to lead to the emergence of more transmissible/attenuated virus strains and more resistant hosts (for review see [5]). Such studies have been carried out using experimental rabbit infections, however, few studies have characterised the sequence changes incurred by the virus during adaptation to its new host [1, 6‐10].

MV strains are classified into 5 virulence grades (I to V, I being most virulent and V the most attenuated) based on the mean survival time of rabbits after infection [10]. Due to the large size of the viral genome (161.8 kb) restriction fragment length polymorphism (RFLP) analysis of purified viral genomes has been the traditional method for the molecular differentiation of MV strains [6, 11]. However, no correlation between RFLPs and attenuation has been demonstrated [6].

Recent studies of field strains of MV have focussed on sequencing genes or gene fragments, however, the majority of these strains have not been characterised for virulence in rabbits [1, 9]. Epidemiological studies on large numbers of virus isolates are essential to differentiate the types of circulating MV-isolates and to identify new emergent strains which might have different degrees of virulence.

Hampered by the large size of the viral genome, to date only two full-length genome sequences are available, strain Lausanne (Lu) [12] and strain 6918 [8]. These data allowed the first direct comparison between two full-length MV genomes, one (Lu) being wild-type (wt) and the other (6918) being attenuated [8].

Strain 6918 was first isolated during field sampling of MV in Spain [13]. In that study twenty isolates, collected from wide ranging locations between 1992 and 1995, were further characterized for their virulence grade and horizontal transmission in experimental infections of rabbits. These MV-isolates included a representative example of all five virulence grades providing an excellent starting biological material for molecular studies on MV strain differentiation and for the investigation of the molecular basis of virus attenuation in the field.

The aims of the study presented here were to molecularly characterise nine MV isolates [13] and to identify mutations that could be potential causes for virus attenuation. In carrying out this study we also aimed to identify genetic markers for the differentiation of MV isolates and to identify variable hotspots in the MV genomes therefore providing valuable tools for future epidemiological studies.

To accomplish these objectives preliminary RFLP analysis of purified full-length virus genomes and PCR-amplified terminal inverted repeats (TIR) regions were coupled with a comprehensive sequencing study of 16 genomic regions encompassing 20 genes (14 completely sequenced and 6 partially sequenced) and 7 intergenic regions. The sequenced regions covered approximately 10% of the virus genomes. The results of these analyses, together with the results from the published full sequence of the attenuated 6918 strain, are discussed with regard to MV strain differentiation, and potential causes for virus attenuation.

In order to identify regions of variability in the MV genome to be used for MV strain differentiation, phylogenetic studies and to investigate possible reasons for virus attenuation we endeavoured to molecularly characterise nine representative strains of MV from the 20 originally isolated in Spain between the years 1992 and 1995 [13] from wide ranging geographic locations (see Additional file 1). The virus strains had previously been characterised for their transmissibility and virulence in rabbits, with each strain given a virulence grade A-E [13], these grades being exactly equivalent to the grades I-V previously designated by Fenner and Marshall (1957). The selected strains were representative of different levels of virulence: 87, 466, 2012 and 86 (grade A); 2788, 7514 and 1312 (grade B); 7411 (grade C) and strain 4604 (grade E), which corresponded to a naturally attenuated MV isolate (See Additional file 2). It should be mentioned that although strain 6918 (grade E) was not directly analyzed in the present study the published sequence data [8] from the relevant regions of this strain were analyzed together with those from the 9 isolates studied. This was done because strain 6918 was originally isolated and characterized, in terms of virulence and transmissibility, under the same conditions [13] as the strains studied in this paper.

We selected 16 genome regions for sequence analysis (Table 1) based on the following criteria; 1) previous publications of MV strain characterisation [3, 4, 6‐8], 2) publications identifying MV virulence factors [14‐23], 3) RFLP analysis of genomic viral DNA and TIRs obtained using long-range PCR [24] (data not shown) and 4) proposed ORF functions [12].

The sequences of the investigated genomic regions included 7 intergenic sequences and 20 complete genes or gene fragments, representing approximately 10% of the viral genome.

The sequence data from the 9 MV field strains indicated that two of them (466 and 2012) although isolated in very distant places in Spain (Additional file 1) could not be differentiated from one another and should be considered, with the available sequence data, as a single virus haplotype. All the remaining strains could be distinguised from each other using the obtained sequence data. The MV field strain sequences corresponding to the 16 amplified regions were aligned and compared to the published 6918 sequences (also isolated in Spain) and to those from reference strain Lausanne (Lu). Five of the regions analysed (corresponding to genes or gene fragments M034, M069, M071, M130 and M135) were identical (data not shown) in all the analyzed MV genomes.

In order to facilitate the description and analysis of the observed mutations in different MV isolates we have grouped the changes into three categories: those occurring within intergenic regions (Table 2), in the TIR (Table 3) or in central regions of the MV genome (Tables 4 and 5). The mutations are presented using the format; XpositionY, where X indicates the original Lu residue (GenBank accession number AF170726), position refers to the nucleotide number of the Lu sequence and Y indicates the mutation detected. For clarity the observed changes are described in separate sections.

Since the deliberate release of MV in Australia in 1950 and France in 1952 the emergence of more resistant hosts and the adaptation of virus strains has been well documented [25, 26]. However, few studies have addressed the sequence changes that have occurred in the MV genome during this period [1, 6, 8, 9] or more specifically the sequence changes involved in virus attenuation. Analysis of the molecular biology of MV in Australia have identified mutations in field strains [6] which have been used to study archived MV isolates and to track virus spread in the field [6, 11]. Although attenuated strains were detected no correlation between the observed RFLPs and virulence was described. With the exception of the sequencing of the avirulent strain 6918 [8] and recent analyses of virulent MV circulating in Portugal and Spain [1, 9] extensive studies on the molecular biology of European strains of MV are lacking, with sequence analysis concentrated on a single short fragment (491 nts) of the M022 envelope protein gene [3, 4].

In order to bridge this gap in our understanding of the biology of MV we set out to molecularly characterise nine strains of MV isolated in Spain between the years 1992 and 1995. These strains were from wide ranging geographic locations from throughout Spain and had been previously graded for virulence by experimental infection of rabbits, and were chosen because they represent viruses with markedly different levels of attenuation [13].

To carry out epidemiological studies of MV it is essential to be able to differentiate between virus strains. Experimental infection of rabbits is not a suitable method for large scale continuous MV monitoring, therefore molecular characterisation of strains is preferable.

Using the mutations detected in this study the majority of the virus strains could be differentiated from one another. Despite a very high degree of sequence conservation, variable hotspots (regions containing M004L/R, M036L and M148R) in the genome were detected that could be used to provide phylogenetic data. We have not included a phylogenetic analysis using these fragments as the number of samples analysed was small (n = 9). However, the genetic relatedness of several strains could be inferred from shared mutations. In particular strains 466 and 2012, both virulence grade A [13] could not be differentiated from each other and shared 7 mutations with respect to the Lu sequence. Several strains also shared multiple mutations e.g. strains 87 and 86, and strains 2788 and 4604. These results may indicate that particular strains are derived from others or that the sites and type of mutations observed occur frequently. It is likely that some of the mutations arise frequently and revert showing the dynamic nature of the virus in the field. Further work will be required to determine if the mutations described are stable at the population level. Interestingly, the strains most related to each other originate in regions separated by between 500 and 1000 km (see Additional file 1). This observation is likely the consequece of movement of wild rabbits for repopulation of areas [27] rather than examples of convergent evolution.

The grossest changes were observed in the intergenic regions analysed. Deletions and insertions of tandem repeat sequences were common and have been reported in the Australian field studies [6]. These mutations are likely to be too frequent to be useful in phylogenetic studies.

With the exception of M036L, the highest numbers of mutations were observed in the terminal regions. Strains 87, 466, 2012 and 86 (all virulence grade A) contained numerous mutations in the TIRs and immediately flanking regions. Only strains 87 and 86 shared a further synonymous mutation in the central genome located in the M141R gene. However, the viruses in which we observed the highest numbers of mutations in the central regions were viruses that had reduced virulence grades when compared to Lu [13]. Strains 2788, 7514, 1312 (all grade B), 7411 (grade C) and 4604 (grade E) all contained mutations to genes in the central more conserved region of the genome. The presence of these mutations is surely indicative of the presence of more mutations yet to be detected, the cumulative effect of which may explain the attenuation of these strains.

In strain 6918 eighty mutations with respect to Lu were observed [8] however, it is not yet known which are responsible for the virus attenuation. Five 6918 genes carried mutations that disrupted ORFs (M009L, M036L, M069R, M135R and M148R). When the mutations found in the nine strains studied here and the strain 6918 are compared none were common. However, our data may provide insights into the reasons for virus attenuation of some of the nine strains analysed. Four genes contained mutations that disrupted ORFs, M002L/R, M009L, M036L and M017L.

M002 (previously termed MT-2) is a tumor necrosis factor receptor homolog with immunomodulatory functions [28]; reviewed by [29]. Inactivation studies have shown that M002 is a virulence factor [28]. Strain 2788 contained an insertion that disrupted the M002 ORF, and was somewhat but not completely attenuated with respect to Lu (strain 2788 is virulence grade B). While the analysis of individual genes may give us clues to the virulence of particular strains it is clear that factors such as compensatory mutations complicate the study of attenuation of viruses with large genomes.

While the function of M036 is unknown, disruptions to the M036 ORF occur in virulent ([6] and this work) and attenuated strains ([8] and this work) so it is unlikely that this mutation has an effect on virulence. It is clear that although the full gene product is not necessary for virus replication, it is maintained in the virus genome and therefore may offer variable regions to target to obtain phylogenetic data.

The M009L gene forms part of the interesting family of Kelch-like proteins represented in MV Lu genome by five genes (M006L/R, M008L/R, M009L, M014L and M140R) 3 as single copies and 2 duplicated at the left and right TIR. The M009 ORF is disrupted in the attenuated strain 6918 [8] and strain 1312. Morales et al., (2008) argue that this mutation is unlikely to be a critical virulence factor. In the study presented here, M009L was observed to contain a dinucleotide insertion that disrupted the ORF in strain 1312 and in virulence studies strain 1312 showed a somewhat reduced virulence (grade B, [13]) but was lethal in all rabbits infected. This could confirm that this ORF does not act as a critical virulence factor, however, until the appropriate knockout viruses are constructed and tested it cannot be ruled out that this protein plays some role in virulence.

The function of M017L is unknown. Nevertheless, considering the truncation of this ORF in strains 2788 and 4604 which showed reduced or complete lack of virulence respectively it is tempting to speculate that the truncation of this ORF, when coupled with the other mutations observed, may have an effect on virulence. Further work will be required to obtain the full-genome sequences and elucidate the exact role of the M017 protein in the virus life cycle.

It is unlikely that we have identified all the mutations responsible for the attenuation of the strains 7411 and 4604, virulence grades C and E respectively. Preliminary RFLP analysis of full-length genomes and TIRs amplified by long range PCR showed that no large deletions in the genomes of these viruses had occurred (data not shown). It is therefore likely that attenuation is due to the accumulation of point mutations and only full genome sequencing coupled with mutational analysis will show why these strains are less virulent.

Interestingly the most attenuated strain used in the study was strain 4604 which was isolated from the same region as strain 87, the most virulent isolate studied. No common mutations were identified between these strains in this study indicating that there are likely to be genetically distinct strains circulating in any one region. Repeated sampling will be required to show which strains predominate at present.

One of the most striking questions about the biology of MV is why should attenuated MV strains arise? The attenuation of some of the Spanish isolates may reflect a necessity for a tendency toward attenuation in geographical locations that had low density rabbit populations at the time of virus isolation [30]. Strains 7514, 1312 and 7411 all come from regions with low density wild rabbit populations. As MV has no other natural reservoir other than the European rabbit longer survival times in infected rabbits will increase the likelihood of encountering a new host or vectors in smaller populations. This is one aspect of virus/host evolution that could be examined using phylogenetic studies of MV isolates in Spain.

In summary, the data included in this study provide an insight into the mechanisms of attenuation of MV strains and indicates useful targets for use in phylogenetic and epidemiological studies. The genes M004L/R, M036L and M148R were shown to contain variable sequences and can be considered possible targets for genetic relatedness studies. Using the type of data obtained in this study it will be possible to identify the types of virus circulating in an epidemic or outbreak, determine if new virus types are emerging or if one virus type is predominating in a particular area and be able to track the spread of these virus types.