| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 Society of Systematic Biologists
The Effect of Paralogous Lineages on the Application of Reconciliation Analysis by Cophylogeny Mapping
Edited by Adrian Paterson: Assiciate Editor
Department of Zoology, University of Oxford South Parks Road, Oxford, OX1 3PS, United Kingdom E-mail: andrew.jackson{at}zoo.ox.ac.uk
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionConclusionReferences |
|---|
Paralogy defines similarity caused by duplication rather than common descent and is well known in the case of paralogous gene copies within a single genome. The term is here extended to paralogous lineages of associates within a single host. The phylogenies of four genera within the Herpesviridae were reconciled with host phylogenies using cophylogenetic mapping. The observed correspondence for each pair of phylogenies was evaluated through randomization of the viral phylogeny and demonstrated to be greater than expected by chance. A simulation study was then carried out to assess the influence of paralogous lineages on the efficacy of reconciliation analysis. Combining viral taxa from different genera that infected common hosts introduced incongruence into the cophylogenies and reduced both the minimum and maximum observed number of codivergence events relative to the initial analysis of orthologous clades. However, at an average sample size this did not alter the fundamental significance of observed correspondence. With smaller sample sizes, the number of orthologous taxa selected at random from the pool of taxa was reduced. False-negative results then increased in proportion from 0.02 to 0.33. These results demonstrated that reconciliation analysis is robust under conditions of paralogy at "normal" sample sizes but is adversely affected by a combination of paralogy and low sample size. Consideration of phylogenies for Papillomavirus, Atadenovirus, and Mastadenovirus suggest that paralogous lineages may be a widespread phenomenon among DNA viruses and that duplication irrespective of host speciation is an important cause of viral diversification.
Keywords: Cophylogeny; cospeciation; herpes; Herpesviridae; mapping; Papillomavirus; paralogy; reconciliation analysis
Received March 18, 2004; Accepted June 16, 2004
Fitch (1970) first used the terms orthology and paralogy to define genes that were homologous either because they were the same in different organisms or were duplicates in the same or different organisms. It is now understood that, of all the copies of a gene in two organisms, some may be related by orthology, derived by descent and so closest relatives, whereas others may be paralogous, derived from past duplication events and so separate lineages. In drawing parallels between the three spatial levels of historical biogeography, host–parasite interactions, and gene–organism relations, Page (1993) recognized that paralogy existed not only at the gene level but also in terms of associate lineages within a host and of populations inhabiting an area. The conceptual equivalence between these problems has meant all can be addressed by phylogenetic reconciliation analysis (Page and Charleston, 1997, 1998).
Phylogenetic reconciliation describes the procedure for comparing two phylogenies that are associated through time but which are partially incongruent nonetheless. The incongruence between two trees is reconciled by invoking evolutionary events from a prescribed four-event cophylogeny model, shown in Figure 1. The "associate" is assumed to track the "host" through time and to evolve through: "codivergence" (synchronous cladogenesis or phylogenetic tracking), "duplication" (unilateral duplication of an associate), "loss" (the unilateral extinction or disappearance of an associate), and "switching" (the unilateral transfer of an associate to another host) (Page, 1990, 1994). The type of reconciliation analysis used in this study is cophylogeny mapping (Page, 1993, 1994; Charleston, 1998; Paterson and Banks, 2001). The associate tree is embedded within the host tree and associates nodes are mapped to optimal positions within the host topology, based on known associations.
|
). (b) Phylogenies for six hosts (A to F) and associates (a to f) are compared and differ in a single respect: associate "c" is misplaced. (c) Application of the evolutionary model to explain the position of "c" has 847 possible solutions, of which 7 are potentially optimal under some set of event costs.The result is a selection of possible combinations of events that can describe the "reconciled tree," which are guaranteed to be optimal (i.e., require the fewest events) under some set of event costs (see Materials and Methods). These solutions can be used to estimate the overall significance of observed correspondence between two trees as well as derive hypotheses about specific transitions. However, the results are obviously contingent on the validity of the initial phylogenies and of the designated associations. This study will address the influence of paralogous lineages on reconciliation analysis; Figure 2 shows that these may have the capacity to mislead reconciliation analysis if the sampling strategy is poor and fails to recognize them as such. Paralogy may result in both a reduction in overall correspondence (measured in codivergent nodes) and spurious transitions.
|
Herpesviruses and their hosts were the model system used to investigate the effect of paralogy on reconciliation analysis. The Herpesviridae are double-stranded DNA viruses with large genomes between 125 and 230 kbp in length. Transmitted in fluids, these viruses replicate and transcribe within the host nucleoplasm, typically form persistent and unapparent infections, and are highly prevalent in the host population (Umene and Sakaoka, 1999). This persistent infection is often facilitated by the viruses invading immune system cells, as in the case of Epstein-Barr virus, which colonizes antibody-producing B lymphocytes and promotes their proliferation; for this reason the virus is associated with malignancies of the immune system (Crawford, 2001).
Herpesviral genomes have evolved through a series of gene acquisitions from both host cell and other viruses (Raftery et al., 2000; Holzerlandt et al., 2002); for example, the rep gene has apparently been acquired by human herpesvirus 6 from the adeno-associated 2 virus (Umene and Sakaoka, 1999). Acquisition of host genes responsible for cell-cycle control and immunity allow herpesviruses to manipulate host cell division and account for their oncogenic capacity. Other acquisitions facilitate inhibition of the immune response and cytotoxic reactions to their infection (Umene and Sakaoka, 1999). Such molecular adaptations reduce the host's ability to clear the virus rather than increase virulence and the evolution of a largely benign, yet abundant and persistent, infection has been taken to reflect a long coevolutionary episode (Shadan and Villarreal, 1993; Mayo and Pringle, 1998).
The origin of all mammalian and avian herpesviruses was between 200 and 230 Mya (McGeoch et al., 1995). The phylogeny of the clade is well defined and breaks down into three independent lineages: the 
, β, and 
subfamilies (Roizman et al., 1992); the origin of each of these is 80 to 120 million years ago (McGeoch et al., 1995). Coevolution at the microevolutionary level is expected to mirror a long-term interaction throughout evolutionary time. For Umene and Sakaoka (1999), the host specificity and lack of interspecific transmission by herpesviruses suggests that they have diversified in concert with their hosts. Although consistent, such qualities do not necessitate cospeciation. However, correspondence between viral phylogenies and host taxonomy was noted by McGeoch et al. (1995, 2000) and this, in combination with a positive correlation between viral genetic distances and host divergence times, strongly suggested a pattern of codivergence throughout the history of the Herpesviridae. The argument is compounded by the apparent utility of herpesvirus as a marker for divergence of human populations (Rojas et al., 1993).
Hence, the antiquity, phylogeny, and life strategy of herpesviruses make it reasonable to suggest that the ancestor of mammals had herpes. Reconciliation analysis identifies if all (or any) lineages display a codivergent dynamic with their hosts and if this is the only transition to have affected the association. Knowing that there are independent lineages among these viruses that infect the same hosts, this group is suitable for evaluating the effect of paralogy and sampling on reconciliation analysis. One third of all herpesvirus sequences in GenBank are partially or entirely uncharacterized; this, and the high prevalence of multiple "types" within a single host, all belonging to a single recognized lineage, indicates that there is a real danger of producing false negatives when using reconciliation analysis by inadequately defining associates. The objective here was to fully characterize herpesvirus cophylogeny using cophylogenetic mapping with TreeMap v2.0 (ttp://taxonomy.zoology.gla.ac.uk/% 7emac/treemap/ index.html), for , β, and clades within the Herpesviridae. This established the characteristic dynamics within clades of orthologous lineages as a standard to which paralogous taxon sets could be compared. The effect of paralogy was assessed by sampling randomly from herpesviruses from all lineages. This simulated a sampling strategy lacking knowledge of homology. The effect of this ignorance and of decreasing sampling effort were evaluated to quantify the chances of obtaining a false-negative result and to assess how the "diagnostic" function of reconciliation analysis was affected; i.e., its ability to infer evolutionary scenarios for resolving incongruence.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Sampling and Phylogeny Estimation
Molecular sequences were retrieved from GenBank for four viral genera in the Herpesviridae: Varicellovirus (VCV,
-Herpesviridae), Cytomegalovirus (CMV, β-Herpesviridae), Lymphocryptovirus (LCV, -Herpesviridae), and Rhadinovirus (RV, -Herpesviridae). Generally, sample size was maximized to enable the most comprehensive analysis possible and this dictated the size and identity of the marker (a catalogue of all sequences are available from the author [http://www. cophylogeny.com]; for details of associations see Fig. 3a to 6a):- VCV: 14 associations; phylogeny estimated from glycoprotein B protein sequences (1036 characters). Sequences were obtained from diverse origins: Borchers et al. (1991), Telford et al. (1992), Spatz and Maes (1993), Harder et al. (1996), King et al. (1997), Ros and Belak (1999), Reubel et al. (2002), Wagenaar et al. (2003), Re et al. (unpublished, 2003).
- CMV: 11 associations with primates; phylogeny estimated from the DNA polymerase nucleotide sequence (700 bp). CMV from Colobus badius and Aotus sp. were placed with shorter sequences (179 bp). Sequences were obtained from Stetter et al. (1995), van Devanter et al. (1996), Swanson et al. (1998), Blewett et al. (2003), and Ehlers et al. (2003). Additional CMV are available for nonprimate hosts but these excluded because their sampling was too sparse.
- LCV: 19 associations; phylogeny estimated from DNA polymerase nucleotide sequences (433 bp). The majority of sequences were taken from Ehlers et al. (2003); additional sources were Lacoste et al. (2001) and de Thoisy et al. (2003).
- RV: 13 associations (with 10 ruminant hosts); phylogeny estimated from DNA polymerase nucleotide sequences (174 bp). Sequences were taken from: Rovnak et al. (1998), Li et al. (2000, 2001, 2003), Chmielewicz et al. (2001) and Klieforth et al. (2002). Polymerase sequences are also available for certain primates and other mammals but these were excluded because they were too distant from ungulate RV and would only serve to inflate the level of codivergence by providing an outgroup.
|
), loss (•), and switching (
). (c) A consensus of POpt; each of the 36 solutions comprise a combination of the four event types and is drawn as a quadrangle by plotting the number of codivergence (CO), duplication (DU), loss (LO), and switching (SW) events on the axes. Circled values refer to multiple solutions with the same combination of events; shaded quadrangles indicate solutions with significant codivergence. (d) A consensus of associate origins. The frequency with which each associate lineage originated through either codivergence (top bar) or switching (bottom bar); the frequency of each is recorded for each associate node and then all nodes are then ranked in terms of consistency from left to right. Squared values refer to the number of duplications also implicated in the origin of the lineage.
|
|
|
Sequences were aligned by eye after initial multiple alignment in ClustalX (Thompson et al., 1997). Phylogenies were estimated with PAUP* v4.0b (Swofford 1998) and using both maximum parsimony (MP) and maximum likelihood (ML). For MP searches, each phylogeny was reconstructed using the general program settings: TBR swapping algorithm, initial tree obtained by "simple" stepwise addition, gaps were treated as missing data, ACCTRAN character state optimization was employed, and multiple states considered uncertain. Confidence intervals for each estimate were obtained using 500 nonparametric bootstrap replicates (Felsenstein, 1985). To both assess the effects of bias on MP estimation and the effect of phylogenetic error in the tree generally, the tree was also estimated using ML with a full general time-reversible model (Yang, 1994), estimated in PAUP from the data. Host trees were taken from the published estimates of mammal (Amrine-Madsen et al., 2003; Hudelot et al., 2003), primate (Eizirik et al., 2001; Arnason et al., 2002), and artiodactyl (Hassanin and Douzery, 2003) phylogenies.
Phylogenetic Reconciliation
Reconciliation analysis was carried out using TreeMap v2.0 on both the MP bootstrap consensus and ML heuristic topologies for each pair of phylogenies. When using TreeMap v2.0 it is possible to assign differential costs to each evolutionary event and use total events to assess the overall optimality of a solution. However, making these assumptions requires knowledge of the relative probabilities of different events; a more reserved approach is to entertain all possibilities during the analysis and then evaluate the solutions a posteriori. Hence, costs were set to 0 for codivergence and 1 for all noncodivergence events (duplication, loss, and switching); this is required as a pragmatic restriction on optimality. The only effect of the cost structure was to favor solutions with minimal requirements for noncodivergence events and so ensure logical consistency (sensu Ronquist, 2003). If two sibling hosts harbor sibling associates, a consistent method will return a solution with a single codivergence, rather than a series of switches or duplications and losses. The costs, or total cost, were not used as a general parsimony criterion for selecting an optimal reconciled tree. Rather, all solutions were filtered to remove guaranteeably suboptimal solutions (e.g., those with extraneous noncodivergence events), leaving a set of solutions potentially optimal under some set of costs, referred to as POpt.
Significance Testing
POpt consists of solutions with a variable number of codivergence events (NCE). The significance of each value was obtained through randomization tests, as first described by Rosen (1978) and specified for TreeMap by Page (1990). Randomization was carried out using the significance option in TreeMap v2.0, building 100 randomly resampled jungles, randomizing the associate tree only. It is appropriate that the associate tree is randomized because the evolutionary model is fundamentally asymmetric (the associate tracks the host) and the null hypothesis specifically refers to the congruence shown by associate tree relative to random trees. The null hypothesis that the level of similarity was due to chance alone was rejected if the given number of codivergences was seen in no more than five randomized jungles. Where it was not possible to obtain a significance value, the significance value for the nearest event number possible was obtained and the missing value declared equal to or less than this.
Resampling and Reconciliation of Paralogous Taxon Sets
The effect of paralogous lineages was tested by placing all 57 associations in the four orthologous taxon sets previously reconciled into a single pool. One hundred replicate tanglegrams were created by selecting associations at random from this pool. However, given that the four viruses often infect the same hosts, the sampling was stratified such that hosts were placed in categories, each category providing a set proportion of the total. Hence, each replicate comprised four associations from cetartiodactylans, one association from carnivores, and three associations from each of cercopithecines, hominoids, and platyrrhines. These numbers were proportional to the total contribution of each category to the pool. Associations were numbered and selected by generating a list of randomized numbers in Microsoft Excel. Sampling in this stratified random manner ensured that replicates were not artefactually incongruent due to a preponderance of associations involving the same hosts.
To simulate sample sizes typical of cophylogeny studies, the size of taxon sets was determined by considering 54 cophylogeny studies that have presented a tanglegram; i.e., two phylogenies articulated through the associations between hosts and associates, since 1996. The number of associates included has been as large as 41 (pinworms of primates; Hugot, 1999) and as small as 5 (GB viruses of primates; Charrel et al. 1999). For hosts, the number ranges between 4 and 36. The average number of associates and hosts (±SD) is 14.6 (±7.8) and 13.6 (±7.1), respectively. Given this, each randomized replicate contained 14 associations; this number also approximates to the greatest number of associations that can be analyzed easily. Each replicate was reconciled and significance tests were performed on the minimum and maximum number of codivergence events (NminCE and NmaxCE) as described above.
Sample size itself can alter the results of reconciliation analysis, especially where terminal and internal nodes present different dynamics (i.e., basal nodes become proportionally more influential in small trees). Given that tanglegrams are routinely presented with fewer than 14 associations, the effect of sample size in conjunction with paralogous lineages was assessed by, firstly, reconciling 100 randomized replicates with 12 taxa and, secondly, with 10 taxa.
Preliminary Cophylogenetic Comparisons for Papillomavirus and Adenoviridae
Few viruses have received the scrutiny that has been applied to the Herpesviridae. Consequently, sampling effort is generally inadequate for a thorough reconciliation analysis. However, enough sequences are available to give an indication of cophylogeny for Papillomavirusand adenoviruses, which might indicate whether codivergence and paralogous lineages are widespread in other DNA viruses. MP and ML phylogenies were estimated, as described above, for 24 Papillomavirus from various mammals, 8 Atadenovirus infecting vertebrates, and 28 Mastadenovirus also infecting mammals, using the following data (details of sequences are available from the author at www.cophylogeny.com):
- Papillomavirus: the majority of strains were estimated using complete genome sequences (7195 bp after alignment). Additional taxa were placed using sections of the L1 gene: Colobus guereza PV2 (3611 bp), Macaca fascicularis PV, Colobus guereza PV1, Trichechus manatus PV (512 bp), and Gorilla gorilla PV (479 bp). In the absence of effective outgroups, the trees were midpoint rooted.
- Atadenovirus: the MP and ML phylogenies were estimated from the hexon gene (2725 bp) and rooted using Aviadenovirus as an outgroup.
- Mastadenovirus: the MP phylogeny was estimated from the hexon gene (2770 bp) and rooted using Aviadenovirus as an outgroup.
Reconciliation analysis was carried out, as described above, for Atadenovirus, using the accepted vertebrate and ruminant phylogenies (Hassanin and Douzery, 2003). Papillomavirus was also reconciled with its mammalian hosts, but after the removal of the widespread human and bovine strains, leaving only single representatives in orthologous positions (i.e., Homo sapiens PV type 6 and Bos taurus PV type 2, respectively). This was an attempt to identify any underlying congruence in the absence of paralogous lineages. The effect of these removals could be gauged through comparison with the full MP tree.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Phylogenetic Reconciliation of Herpesviridae Phylogenies with Hosts
The phylogenies estimated for each of the four herpes subfamilies were robust using MP, as evidenced by the high bootstrap values shown in Figures 3a to 6a (generally above 80%). Also, importantly for the use of cophylogenetic mapping, the trees were well resolved. ML estimates were less robust and each bootstrap consensus was partially unresolved but the topologies of ML heuristic estimates were not substantially different from MP. In the case of LCV, certain taxa had to be removed to ensure complete resolution; these were viruses from the catarrhines Semnopithecus entellus, Macaca fuscata, and Cercocebus aterrimus. Removal of these species is unlikely to have altered the fundamental nature of the reconciliation analysis because they fell into positions within the tree consistent with codivergence with hosts. For example, M. fuscata LCV was removed because it caused a tritomy with viruses from M. fascicularis and M. silenus. Similarly, LCV from the mangabey C. aterrimus formed a tritomy with viruses Papio hamadryas and Mandrillus sphinx, mirroring the clade formed between baboons, mandrills, and mangabeys (Purvis, 1995).
Once reconciled, all four herpesviruses display significant codivergence with their host trees, if it is assumed that they are corooted (i.e., the ancestral host was infected by the ancestral virus[es]). Table 1 shows that this significance is greater for VCV, CMV, and LCV (all P < 0.01) than RV (P = 0.03) and that the fundamental result did not differ when using ML trees. The results of reconciliation analysis are presented for each virus in Figures 3 to 6. Each figure comprises a tanglegram, showing the original input trees; a single solution from POpt with NmaxCE; a consensus of all solutions in POpt and a consensus of associate lineages by origin (whether through codivergence or switching). The consensus of POpt plots each solution as a quadrangle on four axes, which represent the four event types. For example, in Figure 3b, a reconciled tree is shown with 20 codivergences, 6 duplications, 6 losses, and 1 switch (note that codivergences and duplications are counted by lineage); this solution represents a single quadrangle in Figure 3c, which is shaded because 20 codivergence events are significantly congruent. The result is a series of overlapping quadrangles showing the total variation in possible solutions to the reconciliation problem. This information reflects the nature of POpt; among any one set of solutions, each of which could be optimal under a certain set of costs, there are those that explain incongruence in terms of switches and those that invoke duplications and losses. The consensus diagrams (Fig. 3c to 6c) illustrate these extremes and all the combinations of switches and losses in between. They also show that each analysis produces solutions with significant codivergence but also several with many more switches and, consequently, fewer codivergences. These solutions might be optimal if switches were found to be very "cheap," that is, likely events. However, low-codivergence solutions can be discounted if host and associate trees are assumed to be corooted, leaving a subset for which Figure 3b to 6b are representative. Codivergence is the characteristic dynamic for these systems, although even among highly codivergent solutions, individual incidences might have several different explanations.
|
Looking at the putative viral origins, Figures 3d to 6d show that many associate lineages (both terminal and internal) originate through codivergence across the whole of POpt. There are then other, mostly internal, lineages that are inconsistent in their origins, to various degrees. However, these lineages with equivocal origins are associated with the low-codivergence solutions that were discounted above because they did not invoke corooted host and associate trees. It is because these solutions invoke derived origins of the association that they invoke several associate switches back to more basal host lineages. Hence, frequent switching is reflected in the consensus of associate origins; even so, it is notable how many lineages are consistent even as part of low codivergence solutions.
The slightly lower significance of codivergence in RV (Figure 6) is reflected in a wider distribution of switch origins in Figure 5d than in the other systems. Here, more lineages originate by switching in a proportion of solutions. Much of this incongruence is reflected as multiple interactions in the tanglegram; for example, Oryx gazella and Ovis aries each possess two unrelated viruses. The level of duplication also varies between the four viruses. Many of the lineages originating through codivergence also required duplication; these are marked with white boxes in Figure 3d to Figure 6d. RV required more duplications, and across more lineages, to be reconciled with their host tree than the other viruses because of the multiple interactions described by the tanglegram.
Influence of Paralogous Lineages on Reconciled Trees
Reconciliation of host-associate cophylogenies in which the associate tree contains a proportion of randomly selected paralogous lineages results in a reduction in NCE relative to orthologous taxon sets, such as those described in the previous section. Figure 7 demonstrates this effect; both the minimum and maximum values for NCE are reduced compared to an analysis of LCV with the same sample size. The reduction in NCE was largely due to the introduction of switches into those replicates with high proportions of paralogous taxa. Where paralogy was pervasive, reconciliation analysis invoked switches and thereby lowered the number of possible codivergence events; this happens because switches circumvent nodes, possibly leaving them without any associate, and therefore preclude their contribution to NCE.
|
However, although NCE was reduced by introducing paralogy, this was not to the extent that observed congruence became nonsignificant; i.e., no greater than chance. Only 2 of 100 replicates were nonsignificant when sampling 14 associations. Significance was largely maintained by basal codivergences in each replicate. Figure 8 describes how the presence of paralogous lineages was reconciled by invoking one or two basal duplications; these were followed by codivergence events associated with the most basal divergences in the host tree. In addition to codivergences at the tips where relationships remained orthologous, these ensured that almost all replicates retained a significant NCE. The other effect of these duplications was to increase the number of associate lineages required to have existed; the descendent branches of these duplicate lineages were not represented and, therefore, several loss events were required to account for their absence. Hence, the effect of incorporating paralogy was to increase the level of loss, which can be taken to indicate the number of associates currently unsampled.
|
Effect of Decreasing Sampling Effort on the Influence of Paralogous Lineages
The reciprocal monophyly ensured by the presence of paralogous associate clades resulted in basal codivergence events. This was still evident when sampling 12 associations, but the number of orthologous clades available at the tips was reduced. This meant that significance was lost on 10 occasions, raising the chance of a false negative from a reasonably low 0.02 to 0.1, and was especially so when multiple interactions were included in the sample, since these remove one or more hosts and preclude their contribution to NCE. The effect was compounded at yet lower sample size, where the chance of a false negative rises from 0.1 to 0.33 with 10 associations.
Preliminary Reconciliation Analysis of Other DNA Viruses
Estimation of the Papillomavirus phylogeny (80.4% parsimony informative characters (p.i.); Fig. 9a) produced a single most parsimonious tree that was generally robust towards the tips, but was more ambiguous at certain basal nodes relating to the placement of the FPV/CPV clade. The tree shows that the various human types are widespread; HPV14 is more closely related to ungulate papillomaviruses than other primate viruses, HPV4 groups with rabbit oral Papillomavirus, whereas HPV18 is most closely related to cat and dog viruses. Within primates, HPV6 seems most closely related to chimp viruses, whereas other human viruses (HPV3, HPV10, and HPV83) group with macaque virus (MmPV). Bovine and rabbit viruses are also widespread. The widespread placement of human viruses is consistent with the two unrelated "supergroups" identified by Chan et al. (1995). The presence of multiple widespread associates from single host species will preclude significant correspondence even if it exists. Placing a large number of associates on a single host inflates the number of associate nodes without providing opportunities for codivergence; therefore, finding randomized trees with a significant number of codivergences for the given number of associates is unlikely. It is difficult to understand how switching could explain the affinities of the various human or bovine papillomaviruses, but the alternative explanation is that these widespread taxa belong to distinct clades that have been poorly sampled. Other representatives of these clades may exist in nonhuman hosts but, without these, the putative paralogous lineages complicate the analysis and may obscure genuine codivergent dynamics among orthologous components of the data set.
|
After removing these widespread multiple types from the tree, the tanglegram in Figure 9b shows that there is reasonable agreement between the phylogenies of Papillomavirus and its hosts. This is especially so within, and largely due to, the primate and cetartiodactyl clades. The addition of other taxa, based on shorter sequences from the L1 gene, further supports congruence. For example, viruses from Macaca fascicularis (MfPV) and Colobus guereza (CgPV1) augment the congruence established by the relationships of MmPV, PtPV, PpPV, and HPV6. However, other additional taxa do not conform to this pattern; a gorilla virus (GPV) groups within the cetartiodactyl clade, a second virus from Colobus (CgPV2) groups closely with FPV and CPV, and a virus from manatees (TmPV) groups with that of a porpoise (PsPV). Despite these exceptions, NmaxCE = 26 when switching is permitted and 24 without switching; both values are significantly congruent (P < 0.01).
The few sequences that currently exist for Atadenovirus suggest a pattern of significant codivergence (NmaxCE = 12, P < 0.01). The viral phylogeny (64.3% p.i.) was completely resolved and robust, with all but one node supported with a bootstrap value greater than 90%, as shown in Figure 10. The single ambiguous node relates to the placement of a possum virus (Trichosurus vulpecula ATAV), which is provisionally placed outside the remaining mammal and bird viruses. The Mastadenovirus phylogeny (68.6% p.i.; Fig. 11) was well resolved and generally robust, the exceptions relating to certain basal nodes. The tree shows that human and simian viruses are paraphyletic with a virus from cats but are otherwise placed apart from viruses in other mammals. The multiple types of Mastadenovirus from pigs, sheep, and cows do not group by host but instead appear to form clusters with viruses from other hosts. S. scrofa MSAV 5 forms a clade with B. taurus MSAV 2 and O. aries MSAV A, and S. scrofa MSAV A groups with B. taurus MSAV B, whereas B. taurus MSAV 4 and O. aries MSAV 7 form a basal clade.
|
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Herpes Cophylogeny
The observations that herpesviruses are highly host specific and adapted for persistent infection (Umene and Sakaoka, 1999; Crawford, 2001) is reflected in broadly codivergent cophylogenies. This, in itself, is not surprising because host–virus codivergence is the consensual view of herpes evolution (Umene and Sakaoka, 1999; McGeoch and Davison, 1999), and reconciled trees containing numerous switches between dissimilar hosts may be discounted. However, this dynamic is only evident because of the advanced nature of herpesvirus taxonomy, which recognizes the different subfamilies, genera, and even clades within genera. Without this information, herpes cophylogeny would probably be misinterpreted as a case of frequent switching. There are several distinct clades infecting the same set of hosts and each of these are codivergent; this indicates a series of duplications during the evolution of this group; that is, unilateral cladogenesis by the virus irrespective of host speciation. So much is obvious with regard to the
, β, and clades, but this extends into the subfamilies, even into genera. Each genus of herpesvirus displayed an equal degree of phylogenetic correspondence (P < 0.01), with the exception of RV. This virus was still significantly congruent (P = 0.03), reflecting the codivergence of RV in sheep and goats, for example, but required many more duplication and loss events to be reconciled (see Fig. 5); fewer taxa in this set had unambiguous origins through codivergence than VCV, CMV, or LCV. The greater incongruence seen between RV and its hosts is a result of multiple interactions; i.e., hosts possessing multiple and unrelated viruses. This is explained in Figure 6 using a series of duplications, generating the hypothesis that paralogous lineages exist within Rhadinovirus which have each diversified with the ruminant hosts, resulting in O. aries, C. hircus, and Or. gazella, each having two viruses amongst those sampled. The number of loss events also required indicates the proportion of sequences (excluding extinction) not yet sampled.
Duplication events are also invoked to reconcile minimal, but robust, incongruence involving VCV from C. hircus and CMV from Cer. aethiops. In Figure 5b, the existence of a group of LCV infecting G. gorilla, P. panicus, and P. troglodytes is explained with a host switch. However, this group could equally, and perhaps better, be explained with a basal duplication, leading to two paralogous groups of LCV infecting apes (of which LCV in H. leucogenys, H. lar, G. gorilla, and P. pygmaeus comprise the other clade). Therefore, although more data and wider sampling may yet revise these estimates, current phylogenies present evidence for recent duplication of viral lineages. Hence, each genus of herpesvirus supports the view that herpes has codiverged with its hosts because mammals first began to diversify, consistent with the origin of all herpes predating mammals and each genus having originated around the time of mammalian radiation (
120 million years ago; McGeoch et al., 2000). However, from the natural taxonomy and distribution of herpes subfamilies, through to the minutiae of terminal clades in each genus, there is evidence for paralogous clades, consistent with the view that viral diversification has not only been characterized by codivergence but also duplication.
The Effect of Paralogous Lineages on Reconciliation Analysis
The simulations carried out here have shown that failure to recognize paralogy within a group of taxa can cause reconciliation analysis to falsely reject significant codivergence. At the average sample size (14 associations) this affects the diagnostic function of reconciliation analysis. Both NminCE and NmaxCE are reduced relative to analysis of an orthologous clade, reflecting the lower number of guaranteed codivergence events in optimal solutions. Furthermore, as Figure 8 demonstrates, paralogous lineages necessitate basal duplications and subsequent losses. In a sense, this is reassuring because the analysis correctly infers the existence of multiple, distinct lineages; but without knowledge of sampling effort, it also invokes a high frequency of extinction. However, in such a situation, it is likely that one would reasonably conclude from reconciliation analysis that multiple lineages existed. Significance testing in TreeMap is also robust under the influence of paralogy; the chance of a false negative with 14 associations was 0.02. A combination of the basal codivergence events invoked due to paralogy and the codivergence resulting from orthologous relationships at the tips was enough to ensure significance in almost all replicates.
At smaller sample sizes, the efficacy of reconciliation analysis is clearly compromised. The type 2 error rate was unacceptable for both 10 (0.1) and 12 (0.33) associations. Loss of significance reflected the reduced orthology at the tips; indeed, the essential paralogy included in the simulations was not problematic in itself. Paralogy was solved at the base, resulting in codivergence events that maintained significance in combination with those at the tips. These tipward codivergences occurred with lower frequency at lower sample sizes, when the chances of sampling viruses from orthologous groups were reduced. The loss of orthology was compounded by the frequency of multiple interactions. In summary, reconciliation analysis applied using TreeMap is robust under paralogous conditions, except at the lower end of the sample size range. And so avoiding paralogy becomes one of many reasons for attempting to sample as widely as possible.
The Potential for Duplication in Viral Cophylogeny
To what extent might unilateral cladogenesis be an important diversifying force in viral phylogeny? Moreover, to what extent have phylogenies failed to identify paralogy? The phylogeny of Papillomavirus might illuminate these questions. Papillomaviruses are small DNA viruses found in mammals and birds (Howley, 1996). They specialize in forming persistent and largely benign infections within epithelial keratinocytes, whether they be external (cutaneous) or mucosal (Howley, 1996). Papillomaviruses possess genes for the manipulation of host cell cycle control and accidental integration of the viral genome results in overexpression of these genes and, consequently, cell immortalization (Munger, 2002); this makes certain human papillomaviruses the most common cause of gynecological cancers (Dürst et al., 1983). They are transmitted in epithelial cells sloughed off the host surface (Bryan and Brown, 2001), which occurs during repeated contact, for instance, during sex or birth (Kaye et al., 1996; Dos Santos et al., 1998), resulting in conspecific or vertical transmission. Like herpesviruses, they are highly host specific (Howley, 1996) and adapted for manipulating the host immune response (Frazer et al., 1999) by reducing their own immunogenicity (O'Brien and Campo, 2002). And like herpesviruses, their biology and their utility as a marker of human populations (Chan et al., 1992; Ong et al., 1993; Ho et al., 1993) suggest that Papillomavirus have diversified in concert with their hosts (Bernard, 1994).
The tanglegram shown in Figure 9b reflects the significant phylogenetic correspondence between viral and host phylogenies when widespread human and bovine viruses are removed. Interpreting this congruence as codivergence is supported by the genetic distances between viral strains. Figure 12 shows a correlation of genetic distances and host divergence times for those associations involved in codivergence events in some or all of the solutions within POpt. The genetic distances were estimated using a fully parameterized ML model. Published divergence times were used for primates (Yoder and Yang, 2000; Glazko and Nei, 2003), artiodactyls (Hassanin and Douzery, 2003), and mammalian ancestors (Springer et al., 2002). Divergence times for Macaca and Colobus and between deer and elk were estimated using cytochrome b gene and mitochondrial d-loop sequences, respectively; phylogenies containing these and related organisms were calibrated using published divergence times to yield rate estimates (0.031 and 0.016 substitutions per site per year for cytochrome b and d-loop sequences, respectively). These rates were then used to calculate divergence times for the species lacking published dates. As in the case of herpesviruses (McGeoch et al., 2000), the correlation provides strong evidence for synchronous cladogenesis, with a slope approximately 1:1.
|
However, this cannot be the whole story because Figure 9a demonstrates that multiple types from humans and cattle are phylogenetically widespread, as observed previously (Chan et al., 1995). The polyphyly of human and bovine viruses confirms that many are most closely related to viruses in unrelated hosts. Unless humans have contracted Papillomavirus, which is seldom observed crossing species barriers (Howley, 1996), from a wide range of other animals, each human type potentially represents a member of an orthologous clade formed after a duplication event. Depending on its age, that duplication event might also have derived viruses in many other hosts, orthologous to the human papillomavirus, which are yet to be sampled. On the basis of congruence among viruses from P. panicus, P. troglodytes, and H. sapiens, HPV6 would seem to be the ortholog to the other sampled ape viruses. Similarly, HPV3, HPV10, and HPV83 would seem to be most closely related to viruses from Old World monkeys (M. macaca, M. fascicularis, and C. guereza).
Sampling effort has typically been focused on identifying human papillomaviruses and this has yielded over 70 recognized "species" (Howley, 1996). Some may have arisen since the origin of humans but this value gives some indication of the potential for duplication during viral evolution. Were each of these human papillomaviruses one representative of a clade spanning many mammalian hosts, which have so far gone unsampled, their number would indicate both a record of frequent duplication and a desperately poor knowledge of papillomavirus diversity.
Seminal phylogenetic studies of Papillomavirus identified clades corresponding to known differences in infection biology, such as those primarily infecting urogenital mucosa, those infecting oral epithelia, those causing epidermodysplasia verruciformis and other cutaneous lesions, and those causing fibropapillomas (Chan et al., 1992, 1995). Hence, the biological distinctions between the putative paralogs are partially understood and suggest that ecological forces may have caused their initial specialization. Improved sampling of nonhuman hosts would be required to test the idea that each host possesses a community of papillomaviruses occupying specialized niches and paralogous to one another.
Cases Where the Scale of Paralogy is Unknown
The phylogenies of herpesviruses and papillomaviruses indicate the presence of paralogous lineages. For many viruses, however, sampling of nonhuman viruses is so poor that they would not be apparent if they existed. This might prevent any codivergent pattern from emerging. It is not certain whether codivergence and duplication affect the phylogeny of adenoviruses. Sampling of Mastadenovirus is adequate for reconciliation, but the analysis reveals little suggestive of codivergence with mammals. The Mastadenovirus phylogeny (Fig. 11) suggests that primate viruses form a clade, whereas those from other mammals are interspersed. Interestingly, however, viruses from cattle, sheep, and pigs group together on three occasions in different parts of the tree. This would be consistent if they were paralogous clades, each codivergent. However, there is reason to believe codivergence is possible because the related genus Atadenovirus is highly codivergent, despite there being few sequences available. Although significant (P < 0.01), this analysis falls below the number of associations for which randomization can reliably provide statistical treatment (i.e., 10) and is only provisional. Furthermore, like Papillomavirus, Mastadenovirus phylogeny indicates that distinct clades correspond to ecological specificities, like gastroenteric or respiratory infections, suggesting an evolutionary mechanism for duplication within a host (Bailey and Mautner, 1994). Further phylogenetic studies are required to test the idea that related strains from different hosts group by pathology or infection type.
The accepted taxonomy of adenoviruses suggests that each vertebrate class possesses a clade of specialist adenoviruses (Davison et al., 2003); therefore, a model of ancient codivergence has been suggested to account for this distribution (Benko and Harrach, 2003). Therefore, codivergence, or at least the evolution of host specificity, has been given a prominent role in adenoviral phylogeny. However, exceptions have appeared that require ad hoc host-switching events for explanation, weakening the codivergence model. For example, Siadenovirus, the putative amphibian adenovirus clade, also infects birds (Benko and Harrach, 2003). The wide distribution of hosts within Atadenovirus has been explained by rampant host switching, in order to defend the ancient codivergence model, in which Atadenovirus is the "reptilian" adenovirus clade (Harrach, 2000). Certainly, this model has substantial difficulties and is based on inadequate sampling of nonhuman adenoviruses. If adenoviruses were host specific enough to have codiverged during the radiation of vertebrates, why does Mastadenovirus not show evidence of codivergence during the radiation of mammals? Even if adenovirus lineages really did codiverge with vertebrates, it is clear that a codivergence model is not sufficient to explain diversification. Duplication prior to and following vertebrate diversification needs to be considered in addition. Atadenovirus may represent a codivergent radiation of viruses and vertebrates, but this would require its origin to predate vertebrate radiation. This origin is possible given the presumed age of adenoviruses and the large evolutionary distance between Atadenovirus and other genera (Benko and Harrach, 2003). The origin of Atadenovirus would be clarified given a reliable rooting of the adenovirus phylogeny, which to date have been unrooted or rooted under the assumption of codivergence (Davison et al., 2003; Benko and Harrach, 2003). In short, there are indications that codivergence may have affected adenoviral phylogeny, although there are multiple difficulties with the established model relating to vertebrate classes. If a codivergence model is to be extended to Mastadenovirus, it can only be applied in conjunction with duplication events.
|
Conclusion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Codivergence between herpesviruses, and perhaps other DNA viruses, and their hosts seems to have been a ubiquitous and characteristic dynamic. This is compatible with long-term coevolution and the strategy of a persistent infection and adaptation for immune evasion, coupled with specializations allowing transmission to conspecifics when virulence is low and infection benign. However, duplication is an apparent, but unquantified, force in DNA virus diversification and is perhaps the main source of innovation. Where human viruses are distinguished by important infection specificities, relatives in nonhumans must be identified to confirm that the same specificity exists and, therefore, strengthens the argument that the mechanism for unilateral cladogenesis of viruses is effectively niche differentiation. In herpes and perhaps Papillomavirus and Adenoviridae, duplication has produced paralogous lineages, which have diversified to form paralogous clades with overlapping host distributions. This has the capacity to adversely affect the function of reconciliation analysis and our understanding of cophylogeny. It may account for a lack of codivergence amongst Mastadenovirus. The results from reconciliation analysis of herpes phylogeny suggest that a small group of related hosts—apes, for example—should be sampled exhaustively to appreciate the full diversity of viruses within a single host and the role of duplication in DNA virus evolution.
|
Acknowledgements |
|---|
This manuscript benefited from the comments of Michael Charleston, Rod Page, Eddie Holmes, Adrian Paterson, and two anonymous referees. This work is supported by a UK Medical Research Council studentship.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
-
Amrine-Madsen H., Koepfli K. P., Wayne R. K., Springer M. S. A new phylogenetic marker, apolipoprotein B, provides compelling evidence for eutherian relationships. Mol. Phylogenet. Evol. (2003) 28:225–240.[CrossRef][Web of Science][Medline]
Arnason U., Adegoke J. A., Bodin K., Born E. W., Esa Y. B., Gullberg A., Nilsson M., Short R. V., Xu X. F., Janke A. Mammalian mitogenomic relationships and the root of the eutherian tree. Proc. Natl. Acad. Sci. USA (2002) 99:8151–8156.
Bailey A., Mautner V. Phylogenetic relationships among adenovirus serotypes. Virology (1994) 205:438–452.[CrossRef][Web of Science][Medline]
Benko M., Harrach B. Molecular evolution of adenoviruses. Curr. Top. Microbiol. (2003) 272:3–35.
Bernard H. U. Coevolution of human papillomavirus with huma. populations. Trend. Microbiol. (1994) 2:140–142.[CrossRef][Medline]
Blewett E. L., Lewis J., Gadsby E. L., Neubauer S. R., Eberle R. Isolation of cytomegalovirus and foamy virus from the drill monkey (Mandrillus leucophaeus) and prevalence of antibodies to these viruses amongst wild-born and captive-bre. individuals. Arch. Virol. (2003) 148:423–433.[CrossRef][Web of Science][Medline]
Borchers K., Weigelt W., Buhk H. J., Ludwig H., Mankertz J. Conserved domains of glycoprotein B (gB) of the monkey virus, simian agent 8, identified by comparison with herpesvirus gBs. J. Gen. Virol. (1991) 72:2299–2304.
Bryan J. T., Brown D. R. Transmission of human papillomavirus type 11 infection by desquamated cornified cells. Virology (2001) 281:35–42.[CrossRef][Web of Science][Medline]
Chan S. Y., Bernard H. U., Ong C. K., Chan S. P., Hofmann B., Delius H. Phylogenetic analysis of 48 papillomavirus types and 28 subtypes and variants—a showcase for the molecular evolution of DNA viruses. J. Virol. (1992a) 66:5714–5725.
Chan S. Y., Delius H., Halpern A. L., Bernard H. U. Analysis of genomic sequences of 95 papillomavirus types—uniting typing, phylogeny, an. taxonomy. J. Virol. (1995) 69:3074–3083.
Chan S. Y., Ho L., Ong C. K., Chow V., Drescher B., Durst M., Termeulen J., Villa L., Luande J., Mgaya H. N., Bernard H. U. Molecular variants of human papillomavirus type-16 from 4 continents suggest ancient pandemic spread of the virus and its coevolution with humankind. J. Virol. (1992b) 66:2057–2066.
Charrel R. N., De Micco P., de Lamballerie X. Phylogenetic analysis of GB viruses A and C: Evidence for cospeciation between virus isolates and their primate hosts. J. Gen. Virol. (1999) 80:2329–2335.
Chmielewicz B., Goltz M., Ehlers B. Detection and multigenic characterization of a novel gammaherpesvirus i. goats. Viru. Res. (2001) 75:87–94.[CrossRef]
Crawford D. H. Biology and disease associations of Epstein-Barr virus. Philos. T. Roy. Soc. B (2001) 356:461–473.[CrossRef]
Davison A. J., Benko M., Harrach B. Genetic content and evolution of adenoviruses. J. Gen. Virol. (2003) 84:2895–2908.
de Thoisy B., Pouliquen J. F., Lacoste V., Gessain A., Kazanji M. Novel gamma-1 herpesviruses identified in free-ranging New World monkeys (Golden-handed Tamarin [Saguinus midas], Squirrel Monkey [Saimiri sciureus], and white-faced Saki [Pithecia pithecia]) in Frenc. Guiana. J. Virol. (2003) 77:9099–9105.
Dos Santos R. C. S., Lindsey C. J., Ferraz O. P., Pinto J. R., Mirandola R. S., Benesi F. J., Birgel E. H., Pereira C. A. B., Becak W. Bovine papillomavirus transmission and chromosomal aberrations: An experimental model. J. Gen. Virol. (1998) 79:2127–2135.[Abstract]
Ehlers B., Ochs A., Leendertz F., Goltz M., Boesch C., Matz-Rensing K. Novel Simian homologues of Epstein-Bar. virus. J. Virol. (2003) 77:10695–10699.
Eizirik E., Murphy W. J., O'Brien S. J. Molecular dating and biogeography of the early placental mammal radiation. J. Hered. (2001) 92:212–219.
Felsenstein J. Confidence-limits on phylogenies—An approach using the bootstrap. Evolution (1985) 39:783–791.[CrossRef][Web of Science]
Fitch W. M. Distinguishing homologous from analogou. proteins. Syst. Zool. (1970) 19:99–113.
Frazer I. H., Thomas R., Zhou J. A., Leggatt G. R., Dunn L., McMillan N., Tindle R. W., Filgueira L., Manders P., Barnard P., Sharkey M. Potential strategies utilised by papillomavirus to evade hos. immunity. Immunol. Rev. (1999) 168:131–142.[CrossRef][Web of Science][Medline]
Glazko G. V., Nei M. Estimation of divergence times for major lineages of primate species. Mol. Biol. Evol. (2003) 20:424–434.
Harder T. C., Harder M., Vos H., Kulonen K., Kennedy-Stoskopf S., Liess B., Appel M. J., Osterhaus A. D. Characterization of phocid herpesvirus-1 and -2 as putative alpha- and gammaherpesviruses of North American and European pinnipeds. J. Gen. Virol. (1996) 77:27–35.
Harrach B. Reptile adenoviruses in cattle. Act. Vet. Hung. (2000) 48:485–490.
Hassanin A., Douzery E. J. P. Molecular and morphological phylogenies of Ruminantia and the alternative position of th. Moschidae. Syst. Biol. (2003) 52:206–228.
Ho L., Chan S. Y., Burk R. D., Das B. C., Fujinaga K., Icenogle J. P., Kahn T., Kiviat N., Lancaster W., Mavromaranazos P., Labropoulou V., Mitranirosenbaum S., Norrild B., Pillai M. R., Stoerker J., Syrjaenen K., Syrjaenen S., Tay S. K., Villa L. L., Wheeler C. M., Williamson A. L., Bernard H. U. The genetic drift of human papillomavirus type-16 is a means of reconstructing prehistoric viral spread and the movement of ancient huma. populations. J. Virol. (1993) 67:6413–6423.
Holzerlandt R., Orengo C., Kellam P., Alba M. M. Identification of new herpesvirus gene homologs in the huma. genome. Genom. Res. (2001) 12:1739–1748.[CrossRef][Web of Science]
Howley P. M. Papillomavirinae: The viruses and their replication. In: Fundamental virology—Fields B. N., Knipe D. M., Howley P. M., eds. (1996) 3rd edition. Philadelphia: Lippincott-Raven. 763–841.
Hudelot C., Gowri-Shankar V., Jow H., Rattray M., Higgs P. G. RNA-based phylogenetic methods: Application to mammalian mitochondrial RNA sequences. Mol. Phylogenet. Evol. (2003) 28:241–252.[CrossRef][Web of Science][Medline]
Hugot J. P. Primates and their pinworm parasites: The Cameron hypothesi. revisited. Syst. Biol. (1999) 48:523–546.
Kaye J. N., Starkey W. G., Kell B., Biswas C., Raju K. S., Best J. M., Cason J. Human papillomavirus type 16 in infants: Use of DNA sequence analyses to determine the source of infection. J. Gen. Virol. (1996) 77:1139–1143.
King D. P., Parselles R., Gulland F. M., LaPointe J. M., Lowenstine L. J., Ferrick D. A., Stott J. L. Antigenic and nucleotide characterization of a herpesvirus isolated from Pacific harbor seals (Phoca vitulina richardsii). Arch. Virol. (1997) 143:2021–2027.[Web of Science]
Klieforth R., Maalouf G., Stalis I., Terio K., Janssen D., Schrenzel M. Malignant catarrhal fever-like disease in barbary red deer (Cervus elaphus barbarus) naturally infected with a virus resembling alcelaphine herpesvirus 2. J. Clin. Microbiol. (2002) 40:3381–3390.
Lacoste V., Mauclere P., Dubreuil G., Lewis J., Georges-Courbot M. C., Gessain A. A novel gamma 2-herpesvirus of the rhadinovirus 2 lineage i. chimpanzees. Genom. Res. (2001) 11:1511–1519.[CrossRef]
Li H., Dyer N., Keller J., Crawford T. B. Newly recognized herpesvirus causing malignant catarrhal fever in white-tailed deer (Odocoileus virginianus). J. Clin. Microbiol. (2000) 38:1313–1318.
Li H., Gailbreath K., Bender L. C., West K., Keller J., Crawford T. B. Evidence of three new members of malignant catarrhal fever virus group in muskox (Ovibos moschatus), Nubian ibex (Capra nubiana), and gemsbok (Oryx gazella). J. Wildlif. Dis. (2003) 39:875–880.
Li H., Keller J., Knowles D. P., Crawford T. B. Recognition of another member of the malignant catarrhal fever virus group: An endemic gammaherpesvirus in domestic goats. J. Gen. Virol. (2001) 82:227–232.
Mayo M. A., Pringle C. R. Virus taxonomy—1997. J. Gen. Virol. (1998) 79:649–657.[Web of Science][Medline]
McGeoch D. J., Cook S., Dolan A., Jamieson F. E., Telford E. A. R. Molecular phylogeny and evolutionary timescale for the family of mammalian herpesviruses. J. Mol. Biol. (1995) 247:443–458.[CrossRef][Web of Science][Medline]
McGeoch D. J., Davison A. J. The descent of human herpesvirus 8. Semin. Cance. Biol. (1999) 9:201–209.
McGeoch D. J., Dolan A., Ralph A. C. Toward a comprehensive phylogeny for mammalian and avia. herpesviruses. J. Virol. (2000) 74:10401–10406.
Munger K. The role of human papillomaviruses in human cancers. Front. Biosci. (2002) 7:D641–D649.[CrossRef][Web of Science][Medline]
O'Brien P. M., Campo A. S. Evasion of host immunity directed by papillomavirus-encode. proteins. Viru. Res. (2002) 88:103–117.[CrossRef]
Ong C. K., Chan S. Y., Campo M. S., Fujinaga K., Mavromaranazos P., Labropoulou V., Pfister H., Tay S. K., Termeulen J., Villa L. L., Bernard H. U. Evolution of human papillomavirus type-18—An ancient phylogenetic root in Africa and intratype diversity reflect coevolution with human ethnic-groups. J. Virol. (1993) 67:6424–6431.
Page R. D. M. Component analysis: A valiant failure. Cladistics (1990) 6:119–136.[CrossRef][Web of Science]
Page R. D. M. Genes, organisms and areas: The problem of multipl. lineages. Syst. Biol. (1993) 42:77–84.
Page R. D. M. Maps between trees and cladistic analysis of historical associations among genes, organisms an. areas. Syst. Biol. (1994) 43:58–77.
Page R. D. M., Charleston M. A. From gene to organismal phylogeny: Reconciled trees and the gene tree species tree problem. Mol. Phylogenet. Evol. (1997) 7:231–240.[CrossRef][Web of Science][Medline]
Page R. D. M., Charleston M. A. Trees within trees: Phylogeny and historical associations. Trend. Ecol. Evol. (1998) 13:356–359.[CrossRef]
Paterson A. M., Banks J. Analytical approaches to measuring cospeciation of host and parasites: Through a glass, darkly. Int. J. Parasitol. (2001) 31:1012–1022.[CrossRef][Web of Science][Medline]
Raftery M., Muller A., Schonrich G. Herpesvirus homologues of cellular genes. Viru. Genes (2000) 21:65–75.[CrossRef]
Reubel G. H., Pekin J., Webb-Wagg K., Hardy C. M. Nucleotide sequence of glycoprotein genes B, C, D, G, H and I, the thymidine kinase and protein kinase genes and gene homologue UL24 of an Australian isolate of canine herpesvirus. Viru. Genes (2002) 25:195–200.[Web of Science]
Roizman B., Desrosiers R. C., Fleckenstein B., Lopez C., Minson A. C., Studdert M. J. The family Herpesviridae: An update. Arch. Virol. (1992) 123:425–449.[CrossRef][Web of Science][Medline]
Rojas J. M., Dopazo J., Martin-Blanco E., Lopez-Galindez C., Tabares E. Analysis of genetic variability of populations of herpes-simple. viruses. Viru. Res. (1993) 28:249–261.[CrossRef]
Ronquist F. Parsimony analysis of coevolving species associations. In: Tangled trees: Phylogeny, cospeciation and coevolution—Page R. D. M., ed. (2003) Chicago: Chicago University Press. 22–64.
Ros C., Belak S. Studies of genetic relationships between bovine, caprine, cervine, and rangiferine alphaherpesviruses and improved molecular methods for virus detection and identification. J. Clin. Microbiol. (1999) 37:1247–1253.
Rosen D. E. Vicariant patterns and historical explanation i. biogeography. Syst. Zool. (1978) 27:159–188.
Rovnak J., Quackenbush S. L., Reyes S. A., Baines J. D., Parrish C. R., Casey J. W. Detection of a novel bovine lymphotropi. herpesvirus. J. Virol. (1998) 72:4237–4242.
Shadan F. F., Villarreal L. P. Coevolution of persistently infecting small DNA viruses and their hosts linked to host-interactive regulatory domains. P. Natl. Acad. Sci. USA (1993) 90:4117–4121.
Spatz S. J., Maes R. K. Immunological characterization of the feline herpesvirus-1 glycoprotein B and analysis of its deduced amino acid sequence. Virology (1993) 197:125–136.[CrossRef][Web of Science][Medline]
Springer M. S., Murphy W. J., Eizirik E., O'Brien S. J. Placental mammal diversification and the Cretaceous-Tertiary boundary. Proc. Natl. Acad. Sci. USA (2002) 100:1056–1061.[CrossRef][Web of Science]
Stetter M. D., Worley M. B., Ruiz B. Herpesvirus-associated malignant-lymphoma in a slow loris (Nycticebus coucang). J. Zoo. Wildlif. Med. (1995) 26:155–160.
Swanson R., Bergquam E., Wong S. W. Characterization of rhesus cytomegalovirus genes associated with anti-viral susceptibility. Virology (1998) 240:338–348.[CrossRef][Web of Science][Medline]
Swofford D. W. PAUP* Phylogenetic Reconstruction Using Parsimony and other methods (1998) Sunderland, Massachusetts: Sinauer Associates.
Telford E. A., Watson M. S., McBride K., Davison A. J. The DNA sequence of equin. herpesvirus-1. Virology (1992) 189:304–316.[CrossRef][Web of Science][Medline]
Thompson J. D., Gibson T. J., Plewniak F., Jeanmougin F., Higgins D. G. The ClustalX windows interface: Flexible strategies for multiple sequence alignment aided by quality analysis tools. Nuclei. Acid. Res. (1997) 24:4876–4882.
Umene K., Sakaoka H. Evolution of herpes simplex virus type 1 under herpesviral evolutionar. processes. Arch. Virol. (1999) 144:637–656.[CrossRef][Web of Science][Medline]
Van Devanter D. R., Warrener P., Bennett L., Schultz E. R., Coulter S., Garber R. L., Rose T. M. Detection and analysis of diverse herpesviral species by consensus primer PCR. J. Clin. Microbiol. (1996) 34:1666–1671.
Wagenaar T. R., Chow V. T., Buranathai C., Thawatsupha P., Grose C. The out of Africa model of varicella-zoster virus evolution: Single nucleotide polymorphisms and private alleles distinguish Asian clades from European/North American clades. Vaccine (2003) 21:1072–1081.[CrossRef][Web of Science][Medline]
Yang Z. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: Approximate methods. J. Mol. Evol. (1994) 39:306–314.[CrossRef][Web of Science][Medline]
Yoder A. D., Yang Z. Y. Estimation of primate speciation dates using local molecular clocks. Mol. Biol. Evol. (2000) 17:1081–1090.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  E. Schulz, M. Gottschling, G. Wibbelt, E. Stockfleth, and I. Nindl Isolation and genomic characterization of the first Norway rat (Rattus norvegicus) papillomavirus and its phylogenetic position within Pipapillomavirus, primarily infecting rodents J. Gen. Virol., November 1, 2009; 90(11): 2609 - 2614. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Schulz, M. Gottschling, I. G. Bravo, U. Wittstatt, E. Stockfleth, and I. Nindl Genomic characterization of the first insectivoran papillomavirus reveals an unusually long, second non-coding region and indicates a close relationship to Betapapillomavirus J. Gen. Virol., March 1, 2009; 90(3): 626 - 633. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||