Pseudogenization of the MCP-2/CCL8 chemokine gene in European rabbit (genus Oryctolagus), but not in species of Cottontail rabbit (Sylvilagus) and Hare (Lepus)

Background Recent studies in human have highlighted the importance of the monocyte chemotactic proteins (MCP) in leukocyte trafficking and their effects in inflammatory processes, tumor progression, and HIV-1 infection. In European rabbit (Oryctolagus cuniculus) one of the prime MCP targets, the chemokine receptor CCR5 underwent a unique structural alteration. Until now, no homologue of MCP-2/CCL8a, MCP-3/CCL7 or MCP-4/CCL13 genes have been reported for this species. This is interesting, because at least the first two genes are expressed in most, if not all, mammals studied, and appear to be implicated in a variety of important chemokine ligand-receptor interactions. By assessing the Rabbit Whole Genome Sequence (WGS) data we have searched for orthologs of the mammalian genes of the MCP-Eotaxin cluster. Results We have localized the orthologs of these chemokine genes in the genome of European rabbit and compared them to those of leporid genera which do (i.e. Oryctolagus and Bunolagus) or do not share the CCR5 alteration with European rabbit (i.e. Lepus and Sylvilagus). Of the Rabbit orthologs of the CCL8, CCL7, and CCL13 genes only the last two were potentially functional, although showing some structural anomalies at the protein level. The ortholog of MCP-2/CCL8 appeared to be pseudogenized by deleterious nucleotide substitutions affecting exon1 and exon2. By analyzing both genomic and cDNA products, these studies were extended to wild specimens of four genera of the Leporidae family: Oryctolagus, Bunolagus, Lepus, and Sylvilagus. It appeared that the anomalies of the MCP-3/CCL7 and MCP-4/CCL13 proteins are shared among the different species of leporids. In contrast, whereas MCP-2/CCL8 was pseudogenized in every studied specimen of the Oryctolagus - Bunolagus lineage, this gene was intact in species of the Lepus - Sylvilagus lineage, and was, at least in Lepus, correctly transcribed. Conclusion The biological function of a gene was often revealed in situations of dysfunction or gene loss. Infections with Myxoma virus (MYXV) tend to be fatal in European rabbit (genus Oryctolagus), while being harmless in Hares (genus Lepus) and benign in Cottontail rabbit (genus Sylvilagus), the natural hosts of the virus. This communication should stimulate research on a possible role of MCP-2/CCL8 in poxvirus related pathogenicity.


Background
The Lagomorph family of Leporidae (leporids) originated in the New World (Neoartics/Americas), an area which still is home to the most successful of living leporids i.e. species of Sylvilagus and Lepus. The genera Sylvilagus (cottontail rabbits) and Lepus (jack rabbits or hares) comprise numerous species, with Lepus having conquered also the Old World (Paleoartics/Afro-Eurasia) [1]. In contrast, typical Old World leporid genera tend to be monotypic, inhabiting isolated areas where many of them are listed as endangered [2]. The recent world-wide success of the European rabbit (Oryctolagus cuniculus), which in prehistoric times was confined to the Southwestern parts of the Iberian Peninsula, was largely, if not entirely, due to human activity [3][4][5][6].
The introduction of Myxoma virus (MYXV) during the midst of last century as a method of rabbit pest control had devastating effects on populations of European rabbit with reported mortality rates approaching 100% in Europe and Australia [7]. This was in sharp contrast to the very mild pathology caused by the virus in its natural host and reservoir, i.e. species of the genus Sylvilagus [8,9]. For Lepus species, only few cases of MYXV infections were reported and experiments in France have shown that most individuals are innately resistant, reviewed in [7]. In nature, infection with MYXV occurs through bites by flying or jumping insects. Replication of virus starts in MHC-II positive dendritic-like cells at the bite lesions and is passed on to T cells of lymph nodes draining the inoculation site [10]. The pathogenesis of MYXV infection apparently depends upon the aptitude of avoiding the spreading of infected cells throughout the lymphatic system. Whereas in cottontail rabbits MYXV infection remains localized, in 'naïve' European rabbits (below "Rabbit") b , the MYXV infected cells rapidly spread to distal nodes. This results in a generalized leukocyte depletion, particular of CD4+ T cells, which leads to a systemic immunodepression with fatal outcome i.e. myxomatosis [11,12].
Leukocyte migration and trafficking are mainly governed through interactions of a variety of chemokines with their cellular receptors [13,14]. Insights in the parasite strategies of immune evasion offer major gateways for identifying genetic components of pathways allowing Cottontail rabbit to cope with MYXV infection. Studies of different research teams have shown that this virus encodes a number of proteins that manipulate factors of the innate immune system of the host, among them proteins interfering directly with chemo-attractive functions of the CC chemokines [15,16]. It shows that these proteins have played a role during the process of coadaptation between virus and host, and most likely still do. These findings have been of cardinal guidance in the search for host genes (candidate genes) that could make the difference between susceptible vs. resistant species.
MYXV is a large double-stranded DNA virus of the poxvirus family (genus Leporipoxvirus). There have been indications that the CCR5 receptor might play a crucial role during MYXV infection, as it is the case by HIV infection in human [17,18], although the experimental evidence for this has been disputed [19]. However as already mentioned, the variation of pathogenicity of the MYXV among leporid species does not depend upon the fact whether or not the virus can enter and replicate in the host cell, but more likely on a constellation of endogenous factors preventing or permitting the dissemination of infected cells throughout the lymphatic system [11,12]. Studies of pathways underlying the contrasting outcomes of MYXV infection may therefore contribute to a more general understanding of pathogenesis due to large DNA viruses in mammals, inclusive humans. In view of the importance of CCR2 and CCR5 receptors in HIV infection, genes controlling these receptors and their ligands might be prefigurative of such 'candidate genes'. This led to the discovery of a gene conversion that altered the second external loop of Rabbit CCR5. This mutation occurred in the ancestral lineage of the Old World genera including Oryctolagus and Bunolagus, but not in the lineages of Sylvilagus and Lepus species [20,21]. Although these differences at CCR5 obviously do not arbitrate the entry of MYXV for lymphocytes, they might affect CCR5 related pathways of signal transduction [17][18][19]. Note that Bunolagus species being highly endangered, studying their susceptibility to myxomatosis proved impracticable [2].
We therefore have taken a closer look at the main ligands of the Rabbit CCR2 and CCR5 receptors which are the 'macrophage inflammatory proteins' chemokines (MIP's) and the 'monocyte chemotactic proteins' (MCP's). The excellent recent review of the gene organization of mammalian chemokines by Nomiyama and coworkers [22], while comprehensive by extending to non-eutherian mammals (Metatheria and Monotremata), did not include Lagomorpha (Rabbits and Hares). Indeed, chemokine data on Rabbit are incomplete and sometimes erratic (see below). The Rabbit Genome Project being recently completed at the Broad Institute at 7x coverage [23], we have assessed the Rabbit Whole Genome Sequence (WGS) data for orthologs of the mammalian genes of MIP-RANTES and MCP-Eotaxin. Our analyses based on nucleotide sequence similarity revealed that Rabbit possesses proper orthologs of three MCP encoding genes (CCL7, CCL8, and CCL13) which are not identified by gene finder methods used by GenBank. The non-annotation can indicate that in Rabbit these genes may have acquired singularities hampering transcription or disqualifying them as functional proteins. We have searched for such traits and, at the event, verified their presence or absence in species of the leporid genera Oryctolagus, Bunolagus, Sylvilagus and Lepus.

*)
: pos 1 corresponds to position 32582070 of NC_000017.1. The Rabbit genome fragment R-MCPgb containing the CCL2 and CCL1 genes is compared to its Human correlate according to the alignment in Additional file 5. The identification of the Rabbit orthologs of the Human CCL genes based on sequence similarity was checked for consistency using Genscan [29], which furthermore provided estimates of the quality of intron splice sites (values varying between 0 and 1 are shown under the heading "P"). Positions of transcription initiator and termination sites were estimated by homology with Human. The exons annotated in the GenBank feature file of the Rabbit sequence (Additional file 1) are marked "v" under "GenB".
in most, if not all mammal species studied [22], and in Human and Mouse are subject of intense investigations due to their importance in regulating inflammatory and anti-tumoral effects and for their role in HIV infection [24][25][26][27][28]. We will adopt the CCL nomenclature unless we are dealing with proteins. For most mammals, MCP encoding genes are organized as a syntenic group composed of the chemokine genes CCL2-CCL7-CCL11-CCL8-CCL13-CCL1 [22] (in that order; CCL1 serves here only as a syntenic marker). In Human the CCL2-CCL1 encompassing syntenic group is located on chromosome 17 [GenBank: NC_000017.1_ REGION:32582070..32690817]. The fragment of Rabbit chromosome 19 is homologous to this region [Genbank: NC_013687_REGION:23720000..23798000]. We will refer to it as "R-MCPgb".
The pronounced interspecies similarity between the coding sequences allowed localizing the orthologs of the CCL2, -7, -11, -8, and -13 genes in the Rabbit genome. The exons positions are listed side by side for Rabbit and Human in Table 1. The sequence alignments can be consulted in Additional files, with annotations for the undocumented genes (CCL7: Additional file 2, CCL8: Additional file 3, CCL13: Additional file 4) or for the entire MCPgb regions as a 'blunt' 108 kb alignment in FASTA format (Additional file 5).
Orthologs of CCL7 and CCL13 exist and are similar among leporid species Of the two isoforms of the "Rabbit A11 chemokine" annotated in the GenBank, "isoform 2" is orthologous to mammalian CCL11, whereas the "isoform 1" differs        Figure 1 The rabbit ortholog of human CCL7 gene exists and is transcribed. The "A11 isoform 1" predicted by GenBank suggests that exon1 of the Rabbit ortholog of CCL7 uses preferentially exon2 and exon3 of CCL11 during transcription (Additional file 2). We show that cDNA of the leporid species (Oryctolagus cuniculus, Lepus granatensis and Lepus europaeus) contain a transcript uniting three exons orthologous to those of human CCL7 gene. The position of missing MCP-3 characteristic N-glycosylation site is underlined (N X S in other mammals). Variable amino acid residues are highlighted in gray. from CCL11 only by using as initiating exon a sequence located upstream between CCL2 and CCL11. Sequence alignments make it clear that this exon is orthologous to the initiating exon of mammalian CCL7 and at the same time reveal potential exons that are orthologs of the mammalian CCL7 exons 2 and 3. The mRNA transcripts of both CCL11 isoforms are also reported as 'transcriptional variants 2' [GenBank: XM_002719227.1] and 'transcriptional variant 1' [GenBank: XM_002719226.1].

S Y C P R E A V I F K T K Q H R E V C A D P K W P W A Q N A
We have evaluated the putative functionality of the CCL7, -11, and -13 genes, by submitting the R-MCP fragment to Gene Finder software, which did report three exons for each of five genes (CCL2, -7, -11, -13, and -1). All exons were localized exactly as previously inferred by sequence similarity (Table 1). The predicted genes were confirmed by testing specimens of Oryctolagus and Lepus for correct transcription of the CCL7 gene by PCR amplification of cDNA. The CCL7 gene appeared to be transcribed as predicted in Table 1 ( Figure 1). Although they are not identified as such, Ensembl.org reports a Rabbit sequence [Ensembl: ENSOCUG00000013412] and its translation [Ensembl: ENSOCUT00000013408] which correspond to transcripts of the CCL13 ortholog as defined in Table 1 ( Figure 2; for detailed descriptions see Additional file 6). The fact this sequence was derived from cDNA implies that also the CCL13 gene is transcribed. We note the minor sequencing differences at the 3' end of the sequence which might have to do with proximity of the reverse primer used for cDNA amplification.
Transcription alone does not necessarily warrant a functional gene product. The proteins deduced from Rabbit CCL7/MCP-3 and CCL13/MCP-4 CDS sequences show indeed structural anomalies which could disqualify them as functional MCP chemokines. Rabbit MCP-3/CCL7 misses an N-glycosylation site which is present in all known MCP-3 sequences [30]. As glycosylation of MCP-3 may affect its biological activity [31], the loss of the AsnXSer site (underlined in Figure 1) might impair normal chemokine function.
The situation is more problematic with MCP-4/ CCL13. Mature MCP chemokines are derived from the precursor sequence after cyclization of the glutamine Gln24 residue, which is encoded by the 3' end of exon1. In reports on the biological activity of MCP's, the resulting N-terminal pyroglutamic acid (pGlu) is therefore referred to as pGlu1 rather than Gln24. Indeed, most reported MCP protein precursors (CCL2, -7, -8, and -13) are characterized by a 24GlnPro25 motif which after cyclization can further be modified by different types of metalloproteinases [32][33][34]. At the same time, pGlu blocks the action of serine protease peptidases, which in non-MCP chemokines recognize the ubiquitous Pro residue at position 2 of the NH 2 -terminus of the mature protein, by this way fine tuning their function [24,35]. The MCP-4/CCL13 precursor protein inferred from the R-MCPgb sequence is particular by showing a 24GlnThr25 motif instead of the canonical 24GlnPro25. Whereas a Gln24 residue is not a prerequisite of chemokine maturation (e.g. in Rodent MCP-2/CCL8 and Human Eotaxin/CCL11 it is replaced by Gly), the absence of a 24X-Pro25 motif is liable to prevent normal posttranslational processing. However, we found by PCR of gDNA that these singularities encoded by the Rabbit CCL7 and CCL13 genes are shared among the different leporid genera studied, including Sylvilagus (Figures 1 and 2), making a contribution to species-specific variation in disease resistance highly unlikely.
Interestingly, the two Ensembl ENSOCU sequences mentioned were either designated as RABIT CCL8 or as Rabbit CCL7 (Additional file 6). The confusion about  identifying the Rabbit CCL13 and CCL7 orthologs is probably due to the relatively large protein distances separating them from their mammalian correlates ( Figure 2). Different methods of phylogenetic reconstruction produced nevertheless trees in which the paralogous genes did cluster according to orthology, inclusive the Rabbit genes ( Figure 3). Bootstrap values were however very low. Incidentally, the branch lengths of Rabbit CCL7 and -13 nodes were about two times larger in comparison to the average branch length of the Rabbit CCL2, -8, and -11 nodes. Note that Leporid CCL8 is relatively well conserved ( Figure 3). We conclude that Leporid orthologs of CCL7 and CCL13 exist and are transcribed but that their contribution to species-specific disease resistance is unlikely.
The Rabbit ortholog of CCL8 exists but is pseudogenized Sequence alignments of mammalian CCL8 mRNA (at least for swine, horse, cow, panda, human and dolphin) identify clearly a single region of outstanding CCL8 CCL11mamu  Figure 3 Gene orthology of CCL2, -7, -11, -8, and -13 of placental mammals supported by phylogeny. The evolutionary history was inferred by using the Maximum Likelihood method conducted in MEGA5 [36]. The tree with the highest log likelihood (-4585.9908) is shown. Bootstrap values are placed next to the branches. Codon positions included were 1 st + 2 nd + 3 rd and were limited to those encoding the mature chemokine. There were 231 positions in the final data set. Positions of leporid branches are highlighted by triangles ▲ for functional, or by r for pseudogenes. Taxon names of sequences obtained in this study are extended by 'gDNA' when derived from genomic DNA or by 'cDNA' when (also) derived from cDNA. All other sequences represent a sample of mRNA sequences listed in Additional file 9. Mouse CCL12 and CCL8 are located between the CCL11 and CCL1 genes and are likely duplicates of the ortholog of murid CCL8 [22]. homology within R-MCPgb. It is located between the CCL11 and CCL13 genes, as is the case for CCL8 in most, if not all mammals for which this has been checked [22]. It is interesting that gene orthology was much better revealed when the untranslated regions (UTR's) were included. This is illustrated in Figure 4. Whereas the coding regions of CCL8 show pronounced cross-paralog similarity with at least CCL2, -7, and -11, the UTR's are highly gene specific.
The comparison of the Rabbit CCL8 sequence with its mammalian counterparts ( Figure 5) reveals several deleterious mutations at exon1 and exon2, as well as at in-tron2. At exon1 the initiating methionine codon (ATG) was mutated into an isoleucine codon (ATA) and the CAG codon of the canonical GlnPro motif (CAG.CCA) at the 3' end of the exon was changed into TAG. This premature stop codon is however put out of frame by a 10 base pair (bp) deletion, which incidentally transforms the in-frame LeuSer codons (CTG.AGC) into a premature stop codon (c.TGA.gc). The 3' part of exon2 is corrupted by a 1 bp insert, while the GT donor site of intron2 is mutated into AT. In addition, the codon of the third of the four characteristic cysteine residues (TGT) was altered into arginine (CGT), or histidine (CAT) in some wild rabbits (Figure 6). At this position, a cysteine residue is mandatory for the formation of a disulfide bound with the first cysteine of the characteristic cysteine pair [37] and present in all CC chemokines.
A first question was in how far the WGS CCL8 sequence is representative of the species, and if so, whether this situation is limited to the genus Oryctolagus. In order to assess the distribution and history of this apparent gene loss, we designed primers for each exon of the Rabbit CCL8 ortholog (see Methods). By PCR we obtained the CCL8 pseudogene (CCL8ps) DNA sequences of wild rabbits of both subspecies Oryctolagus c. cuniculus and Oryctolagus c. algirus. These rabbits were collected in the original distribution range of the genus (i.e. the Iberian Peninsula), where the gene diversity is much greater than in domestic and wild rabbits of the more recent areas of Rabbit colonization [38,39]. The PCR products confirmed that the CCL8 ortholog is pseudogenized in all Rabbit genomes studied. Individual variation was observed. Only one rabbit showed a sequence identical with the CCL8 sequence of R-MCPgb. It is interesting that for a majority of wild rabbits, the CCL8 genes appeared as "less" derived compared to the Thorbecke rabbit of the WGS files ( Figure 6). While all genomes studied showed the initiation site mutation ATG->ATA, many of them did not show the 10 bp deletion at exon1, and the vast majority did feature the canonic 24GlnPro25 codons instead of the in-frame stop codon of the WGS sequence. Moreover they disposed of one or even two Met(ATG) inframe codons, which could possibly provide a rescue initiation site ( Figure 6).  Figure 4 Fishing the Rabbit CCL8 ortholog using NCBI Blast-Align. Human CCL8 mRNA is compared to the Rabbit genomic fragment containing both the CCL2 and the CCL1 region (R-MCPgb). The graphical representation visualizes the outstanding similarity of one genomic region covering the human RNA transcript almost entirely. At the same time it localizes other structurally related regions and reveals that compared to the coding regions, untranslated region (UTR's) might be more gene-specific. The graph was produced using the online facilities offered by NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi; option: Align).
The situation is more clear-cut at exon2. The three deleterious mutations were present in all Rabbit haplotypes: (1) the altered third mandatory cysteine codon, (2) the 1 bp insert, and (3) the donor site alteration. On the other hand, at least two genomes showed the loss of the characteristic CysCys motif of exon2 (TGCTGC->TGCTCC), a deleterious mutation not present in the WGS Rabbit. The alignment with wild rabbit sequences furthermore revealed an interesting 33 bp insertion in the WGS sequences, which went unnoticed in previous alignments with non-leporid sequences ( Figure 5). This insert was present in both subspecies, although not in all specimens. It results from an exact direct repeat at the junction exon2-intron2, spanning the 1 bp insert and the donor site mutation. Because both of these disabling mutations were present in all rabbit haplotypes, and repeated in the insert, the duplication is most likely more recent. This also applies to the 10 bp deletion at exon1 (you can't delete nor duplicate what didn't already exist). It indicates that the latter are consequences of a prior loss of functionality, and are part of a process of pseudogenization.
The terminating exon3 was found to be potentially "functional" in all rabbits studied.
The Rabbit CCL8 has been pseudogenized for more than 4 million years These deleterious mutations are shared among both Rabbit subspecies O. c. cuniculus and O. c. algirus. These subspecies did separate about 2 My ago [40,41], implying that the pseudogenization of CCL8 must be relatively old and possibly older than the genus. This was corroborated by the CCL8 sequence obtained with one Riverine Rabbit, Bunolagus monticularis, which showed all four deleterious mutations shared among Oryctolagus. The pseudogenization must precede the genus split of Bunolagus and Oryctolagus which occurred an estimated 4 My ago [40,42].
We note in both haplotypes of the Bunolagus specimen the absence of the 10 bp deletion at the 5' region of  (Figure 6). It further supports the viewpoint that the two occasional indels occurred after the initial pseudogenization. The Bunolagus CCL8 pseudogene differs however from that of Oryctolagus by the absence of a theoretical 'rescue' initiator codon (ATG->ACG) in exon1, and the loss of the TATA box (TATAAA->TgTgAA), which Oryctolagus shares with many mammalian CCL8 genes.

MCP-2/CCL8 appears to be functional in Lepus and
Sylvilagus species The next question was whether the CCL8 gene is also pseudogenized in species of leporid lineages that do not share the CCR5 mutation with rabbit. As mentioned before, these species are also those that are resistant to MYXV. Using the primer sets designed for amplifying the three Rabbit-CCL8ps exons, we found that all haplotypes obtained with two Lepus specimens and three Sylvilagus specimens possess a CCL8 ortholog which does not show any disabling mutations. The CDS consensus sequences are shown in alignment with Rabbit sequences (see Figure 6). The same alignment including the complete UTR's and intron regions can be found in Additional file 7. The usual amino acid numbering of mature MCP-2 protein is shown in the alignment presented in Figure 7. The functionality of the CCL8 gene was further evaluated by analyzing cDNA obtained with specimens of species Oryctolagus and Lepus. Although the primers were designed according to the Oryctolagus sequence, the expected CCL8 CDS sequences were obtained with both specimens of Lepus but not with cDNA of Oryctolagus. This failure was not due to the quality of the Rabbit cDNA, because successful PCR amplification of CCL7 was obtained using the same cDNA preparation with the appropriate primers (Figure 1). Tissue samples of Sylvilagus specimens were not suited for RNA extraction, but the extensive sequence similarity with Lepus CCL8 predicts correct splicing and gene transcription, which was also confirmed by gene finder software applied to the genomic sequences of both genera.
The Rabbit genome contains only one CCL8-like gene A last question was whether functional CCL8-like genes might exist outside or within the R-MCPgb region studied. At least in mouse, cow and elephant the CCL8 genes are indeed duplicated, which, parenthetically, might be a further indication of their relative importance (in mouse they are named CCL12 and CCL8, although CCL12 is more similar to mammalian CCL8; Figure 3). By blasting the (entire) Rabbit WGS database (inclusive the Trace File Archives) with the Rabbit-CCL8 sequences here presented, at the exception of full identity with the actual query, highest similarities were obtained with sequences of the CCL11 and CCL7 genes embedded in the R-MCPgb fragment. It strongly suggests that there is no other CCL8-like gene in the entire genome of the specimen studied in the Rabbit Genome Project.
We can therefore ascertain that the unique ortholog of the mammalian CCL8 gene is pseudogenized in the Old Word leporid genera Oryctolagus and Bunolagus, while potentially functional in Sylvilagus and Lepus species and, at least in Lepus, being correctly transcribed.

Discussion
We show that both MCP-2/CCL8 and CCR5 were altered in lineage of Oryctolagus and Bunolagus due to apomorphic mutations which did not occur in the lineage comprising species of Sylvilagus and Lepus. It implies that the species known to be reservoirs of MYXV, i.e. S. brasiliensis and S. bachmani, dispose most likely of normal (plesiomorphic) CCR5 and CCL8 genes. Although such deductions are in line with current phylogenetic inference, they might be worthwhile to be verified, as falsification would imply the independent alteration of one or both of these genes in lineages separated in space and time for millions of years. Meanwhile we will assume that the CCR5 and CCL8 genes of the different Sylvilagus species do not differ significantly from those shared by the New World rabbits of this study.
The situation revealed by the presented data might orientate research towards the role of chemokine and their receptors in host species of MYXV and could lead to new insights in processes of parasite-host coadaptation. In mice MCP-2/CCL8 and CXCL12 were found to cooperate to attract hematopoietic progenitors of immune-regulatory dendritic cells [43] while Islam and coworkers [44] describe mouse MCP-2/CCL8 as crucial (See figure on previous page.) Figure 6 Gene variation at the CCL8 locus within and among leporids. PCR products of leporid specimens were obtained using either gDNA or cDNA and are aligned with the CCL8ps sequence of rabbit WGS. Potential Stop and Translation Initiating codons are underlined. The positions of the cDNA primers are bold underlined. 'Orcu CCL8ps' represents the consensus of 11 haplotypes of Oryctolagus cuniculus specimens of both subspecies (O. c. cuniculus and O. c. algirus), after excluding one haplotype that was identical to CCL8ps of WGS (OccTar109allele1) and ignoring singletons (i.e. nucleotide differences observed only once). Occ: Oryctolagus cuniculus; Bumo: Bunolagus monticularis; Syfl: Sylvilagus floridanus; Leti: Lepus timidus; Legr: Lepus granatensis; Leeu: Lepus europaeus. '.': identical to leader sequence; '-': indel. The data shown are limited to the exon containing fractions. An alignment of the entire 1.6Kb gene region is presented in Additional file 7.
regulator of T(h)2 cell homing. In Human, MCP-2 is known to bind to chemokine receptors CCR1, CCR2 and CCR5 and can act as a potent inhibitor of HIV-1 infectivity [44][45][46]. More in particular, compared to other chemokines, MCP-2 was found the most efficient inhibitor of the HIV protein gp120 for CCR5 receptor binding [47,48]. Studies of the role of MCP chemokines in Human and Mouse revealed that inflammation is regulated by feedback mechanisms where proteases play an important role [24,[33][34][35]. They act by removing the Nterminal tetra or penta peptide of MCP's which can transform them into MCP antagonists. Different research groups reported that natural occurring posttranslational modified MCP-2/CCL8 products can completely (sic) block the chemotactic effects of intact MCP's and of RANTES/CCL5, and have identified natural MCP-2(6-76) (cf. Figure 7) as a potent and functional CC chemokine inhibitor [28,33,34]. These studies have highlighted the role played by MCP-2 products in the subtle agonistantagonist interplay with CC chemokine receptors, including CCR5.
In this context our data might fuel speculations about possible reasons underlying the permanent loss in Old World rabbits of an important gene function,  which has been well preserved throughout mammalian evolution (cf. Figures 3 and 4; a CCL8-like gene has also been reported in bony fish [GenBank: BT048349]). One explanation could be that functions of CCL products can be redundant or interchangeable, at least in leporids, which would be at odds with the evolutionary perpetuation of CCL gene identity. More interesting is the hypothesis of a causal link between the appearance within the lineage of Old World rabbits of the alteration at the second external loop of CCR5 on one hand, and the pseudogenization of one of its prime ligands on the other. Both events are highly unusual and none of them did occur in the Sylvilagus-Lepus lineage, nor in any other studied species. Although the argument is somewhat circular -we looked at the CCL8 gene precisely because of the CCR5 alterationit can offer a plausible explication for the knock-out of an otherwise prominent gene function over a period of more than 4 My (cf. Figure 3). If "lost by accident", the CCL8 gene could during this period have been repaired by back-mutations or by gene conversion with one of its neighbors. One might consider that receptor alteration occurred first, making CCL8 either useless or even detrimental, allowing or forcing its permanent pseudogenization. Or on the contrary, the CCL8 gene knock-out, and the consecutive perturbation of CCR5dependent signaling pathways (e.g. due to the loss of a potential (ant)agonist of other CCR5 ligands such as Rantes/CCL5), may have favored structural change at its orphaned target (i.e. at the second external loop of CCR5). Regardless of the scenario, we could be facing an irreversible situation where a "gene knock-out" resulted in a gene "lock-out". Indeed gene repair would not be favored by selection if the recovered ligand can F1 R1 F2 ->-------------------------------ps: 1269-1634  ------------------------------------------------------------------------------------------------------------------------------------  A further question beyond our competence is whether, when, and in which cellular environment CCL8 is expressed in Sylvilagus during MYXV infection and how, at the event, it contributes to the clearance of infected lymphocytes. In this context it might be interesting to mention that MCP-2 expression was down regulated in human HIV infected brain cells by miRNA146a [49], a microRNA which we found to be present in Rabbit (the miRNA146a sequences are identical among Human and Rabbit and are localized in same chromosomal region; WvdL unpublished observations).

Conclusions
The large number of host-specific immunodulatory proteins encoded by MYXV implies multiple levels of elaborate interactions between the virus and its natural host which can be the outcome of thousands of years of co-adaptive evolution. Identifying the constituents of this interplay remains a huge challenge, as host factors involved can be even more numerous. It might therefore be worthwhile to consider that the knock-out of a single host factor could severely affect this virus-host equilibrium. Given that monocyte chemotactic proteins control patterns of leukocyte migration, which in turn govern the outcome of MYXV infection in rabbits, the observation of a factual correlation between the near absence of MYXV virulence and the concurring presence of a functional MCP-2/CCL8 and an 'intact' CCR5 gene, could promote studies on the role played by this particular chemokine ligand-receptor interaction in keeping Myxoma virus under control.

Methods
Tissue samples specimens of leporid species belonging to genera Oryctolagus, Lepus and Sylvilagus were provided by CIBIO Lagomorph Tissue Collection maintained by Paulo C. Alves (CIBIO, Vairão, Portugal; pcalves@mail.icav.up.pt). Species and sample names are listed in Table 2. All samples are from wild populations.
gDNA For the genomic DNA extraction of Oryctolagus (9 specimens) and Lepus (5 specimen) we used liver tissues preserved at -20°C in RNA stabilizing medium. We were privileged by the generous gift of Bunolagus gDNA prepared by Conrad Mathee. For Sylvilagus we only disposed of blood clots with mRNA not suitable for cDNA synthesis. Genomic DNA was extracted using the Easy-Spin Genomic DNA Minipreps Tissue Kit (Citomed) according to manufacturer's instructions.
cDNA Total RNA was prepared for one wild rabbit (Oryctolagus cuniculus algirus: Oca32), and four hares (genus Lepus). The hare specimens represent two species: Iberian hare (Lepus granatensis: Legr2, Legr6) and European brown hare (Lepus europaeus: Leeu1, Leeu2). RNA was extracted using the guanidinium thiocyanate-phenol-chloroform extraction method (TRIzol) according to manufacturer's instructions (Molecular Research Center, Inc., Cincinnati, OH, USA). Next, first strand cDNA was prepared from 5 μg of RNA and synthesized using oligo (dT) primers [50]. The putative CCL8 and CCL7 transcripts were PCRamplified using a primer set located in the UTR regions.

PCR
Primers were designed according to the R-MCPgb fragment of the Rabbit chromosome 19 [GenBank: NC_013687.1_REGION:23720000..23798000], using the online software Primer-Blast provided by NCBI [51]. For amplification of genomic CCL8, primers were designed separately for each of the three exons, in a way covering all coding and intron regions (see Figure 8). Primer pairs are listed in Table 3. PCR methods were standard. Details are given in Additional file 8. For the sequencing reactions we used ABI PRISM BigDye Terminator v3.1 Cycle Sequencing Kit and protocols were followed according to the manufactures. The sequencing reactions were cleaned with Sephadex™ from GE Healthcare Life Sciences. Sequencing was performed on an ABI PRISM 310 Genetic Analyser (PE Applied Biosystems). PCR products were sequenced in both directions.

Source of data
All sequence data except those produced in this study were obtained from GenBank database of the NCBI platform [52] or Ensembl [53]. The GenBank accession numbers of nucleotide sequences used in this study are listed in Additional file 9. Sequences produced in this study were submitted to GenBank and the accession numbers are listed in Table 4.

Sequence analysis
Alignments were done using the online software "Align" provided by the NCBI site in combination with online software Dialign [54] and Clustal W as incorporated in the MEGA5 package [36] and improved by visual corrections using BioEdit [55]. Phylogenetic analysis shown was conducted using Maximum Likelihood method provided in MEGA5 [36]. The probability of gene transcription of undocumented CCL orthologs was evaluated with GenScan [29].
Endnotes a) MCP-2/CCL8 and similar. Read: either the Monocyte chemotactic protein type 2 encoded by the gene CCL8, or the CCL8 gene encoding the MCP-2 protein, depending on context. Maintaining the MCP nomenclature for proteins is preferred because used in studies of chemokines function. b) Rabbit and rabbit: Species name are capitalized when used to avoid irrelevant repetitions of scientific names (European rabbit or Oryctolagus cuniculus). Thus "Rabbit genome" or "Rabbit