MBE Advance Access originally published online on August 10, 2006
Molecular Biology and Evolution 2006 23(11):2123-2130; doi:10.1093/molbev/msl083
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
In Silico Predicted Robustness of Viroid RNA Secondary Structures. II. Interaction between Mutation Pairs
Instituto de Biología Molecular y Celular de Plantas (CSIC-UPV), València, Spain
E-mail: sfelena{at}ibmcp.upv.es.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Viroids are plant subviral pathogens whose genomes are constituted by a single-stranded and covalently closed small RNA molecule that does not encode for any protein. Most of the 29 described viroid species fold into a rodlike or quasi-rodlike structure, whereas a few of them fold as highly branched structures. In a previous study, we used RNA thermodynamic secondary structure prediction algorithms to compare the mutational robustness of all viroid species. Here we used the same approach to explore the sign and strength of epistasis among pairs of random mutations. We found that antagonistic interactions were more abundant than synergistic ones. However, despite their lower frequency, synergistic interactions tended to be more intense. Mutational robustness and the intensity of epistasis were correlated such that viroid species with large average mutational effects showed stronger antagonistic epistasis, whereas viroids with mild average mutational effects showed weaker antagonistic interactions. The strength of antagonistic epistasis decreased with genome complexity as a consequence of the gained robustness of duplicated genomes. In good agreement with our previous finding of an evolutionary trend toward increased robustness, we now found a trend toward reduced antagonistic epistasis.
Key Words: epistasis • mutational robustness • plant pathogens • RNA folding • viroids
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Epistasis, defined as the interaction between different genetic loci in determining a phenotype, is a key factor in evolutionary genetics. The existence and abundance of epistasis is important for many different evolutionary theories, including those seeking an explanation for the origin and maintenance of sexual reproduction (Kondrashov 1998
), diploidy (Kondrashov and Crow 1991), dominance (Hill 1982; Peck and Waxman 2000), reproductive isolation (Coyne 1992), and the shifting-balance process (Wright 1982; Whitlock et al. 1995; Gavrilets 1999). Much effort has been invested in finding and characterizing epistasis. Unfortunately, much of this effort has been futile because few data sets unquestionably indicate whether deleterious mutations typically have multiplicative effects or, whether in contrast, they tend to interact synergistically (i.e., on average, each new mutation has progressively more effect) or antagonistically (i.e., successive mutations are less damaging).
During the recent years, we have been witnessing an increasing interest on the topic of robustness (de Visser et al. 2003) and its relationship to epistasis (Elena et al. 2006). It has been suggested that the contribution of epistasis among deleterious mutations depends on the magnitude of mutational effects (Lenski et al. 1999; Wilke and Adami 2001; You and Yin 2002; Wilke et al. 2003) and genetic compactness (Edlund and Adami 2004). Recent work with artificial networks has shown that synergistic epistasis can arise as a by-product of selection for mutational robustness (Azevedo et al. 2006). Conversely, theory and computational work have predicted that antagonistic epistasis should appear in very compact genomes with frequent functional overlapping, a feature that should also result in low mutational robustness (Wilke and Adami 2001; Edlund and Adami 2004). In the case of RNA genomes, the abundance of secondary structures should also contribute to generating antagonistic epistasis in the form of compensatory mutations (Wilke et al. 2003; Sanjuán 2006).
Viroids are plant pathogens whose genomes are constituted by a small (246–401 nt long) single-stranded, circular, and covalently closed RNA molecule with a high degree of selfcomplementation, resulting in a compact folding. Indeed, this RNA folding represents, together with the induced symptoms in susceptible plants, the only identifiable phenotype of most viroids. Viroids, in contrast with viruses, do not encode for any protein, entirely relying on the host's translational and transcriptional factors to complete their infectious cycle. The current taxonomic classification of viroids (Flores et al. 2005) is supported by phylogenetic studies (Elena et al. 1991; Elena et al. 2001) along with structural and phenomenological properties (for a list of known viroids, Sanjuán et al. 2006, Table 1). Twenty-five out of the 29 known species of viroids belong to the family Pospiviroidae (PSTVd being the type species naming the family). Pospiviroid adopts in vitro a rodlike or quasi-rodlike secondary structure of minimal energy with 5 structural domains. The family is divided into 5 genera depending on the presence/absence of some of these domains. The other 4 viroids form the family Avsunviroidae (named after ASBVd, the type member of the family). In general, Avsunviroid RNAs do not fold into rodlike structures but into less organized, branched shapes.
|
RNA secondary structure folding is a well-studied model system. The folding of RNA sequences into secondary structures is a simple but biophysically well-grounded and powerful model for studying the mapping relationships between genotype and phenotype (Fontana 2002
). The secondary structure of an RNA molecule is taken as a predictor of the fitness of the underlying sequence (Schultes et al. 1999; Ancel and Fontana 2000; Fontana 2002; Ancel Meyers et al. 2004). However, some limitations have to be acknowledged to this approach: for example, the accuracy of the predicted folding decreases as the length of the molecule increases and tertiary structure elements (as for example, pseudoknots) are ignored. In general, in silico studies on the folding properties of RNA molecules show that the map between genotypic (sequence) and phenotypic (folding) spaces is not one-to-one because the sequence space contains extensive neutral networks connecting neighbors that fold into identical structures (Schuster et al. 1994; Schuster and Fontana 1999).
The present study is the second of a 2-series paper in which we explore mutational effects on viroid predicted secondary structure. In the first study (Sanjuán et al. 2006), we focused on the effect of single mutations and showed that 1) the distribution of single-mutation effects was highly skewed, with a high frequency of small effects and a flat tail of strong effects, with neutrality values ranging between 17% and 26%; 2) Structural robustness increased during viroid evolution; 3) Robustness was in part due to the existence of neutral neighbors folding in the same structure; 4) The Pospiviroidae rodlike structures were more robust than the Avsunviroidae branched structures; 5) Partial genome duplications contributed to increase the robustness of viroid molecules; and 6) Mutational robustness was unlikely to have evolved as a correlated response to selection for environmental robustness. In this second paper, we extend our previous findings by seeking for the sign and strength of epistatic interactions in viroid RNA secondary structure, and we assess the contribution of compensatory mutations to the overall pattern of epistasis. Next, we show that mutational robustness and epistasis are correlated such that viroid species with large average mutational effects have stronger antagonistic epistasis, whereas viroids with mild average mutational effects have weaker antagonistic epistasis. We finally explore 2 consequences of this correlation: 1) in increasing mutational robustness, partial genome duplications also decrease the frequency of antagonistic interactions; and 2) concomitant to the evolution of robustness, there is an evolutionary trend toward relaxed antagonistic epistasis.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Viroid Sequences
The sequences of each viroid species listed in table 1 were downloaded from the National Center for Biotechnology Information Web site (http://www.ncbi.nlm.nih.gov/genomes/VIRUSES/12884.html).
RNA Folding and Folding Robustness
RNA minimum free-energy secondary structures (MFESS) were computed using the algorithm implemented in the RNAfold program of the ViennaRNA Package, version 1.5 (Hofacker et al. 1994). Folding temperature was set to 25 °C. Comparison among MFESS was done using the RNAdistance program of the same package (Hofacker et al. 1994). This comparison gives a Hamming distance, di,j, between the bracket notations of both secondary structures.
For every viroid, the impact of all possible 3L (L, genome length) 1-error mutants on the predicted MFESS was evaluated by calculating the Hamming distance to the wild-type (wt) secondary structure. Mutational effects s(di,wt) were scaled to genome length: s(di,wt) = di,wt/L. For each viroid, 3L double mutants were created by randomly sampling pairs of mutations from the above set of single mutants. The standard definition of the epistasis coefficient is 
ij = Wij – WiWj, where W refers to fitness and WiWj is the expected multiplicative fitness of the double mutant (Wolf et al. 2000). However, we could not treat Hamming distances multiplicatively because there is no evidence that Hamming distances are proportional to fitness effects. Indeed, we observed that in 86.69% of all mutation pairs, the distance from the double mutant to the wild type was exactly equal to the sum of distances from each single mutant to the wild type. Therefore, because Hamming distances were mostly additive, the epistasis coefficients between pairs of mutations was computed as ij = s(di,wt) + s(dj,wt) – s(dij,wt). This definition was chosen to maximize the number of cases in which ij = 0 (for all viroid species, the mode of ij was zero). Values ij > 0 indicated antagonistic epistasis and ij < 0 indicated synergistic epistasis. The results reported here could only be directly extrapolated to fitness if Wi = exp(–Cdi,wt), C being a scaling constant.
As an alternative metric, we used the plastic Hamming distance (Ancel and Fontana 2000). Instead of finding only the MFESS, all structures within a range of energy (here, ±0.5 kcal/mol) are taken into account. For each mutant, Hamming distances to the wild type were computed for each of these structures and an average distance calculated, where the time spent on each structure, as determined by its thermodynamic stability, was used to weight the average (Ancel and Fontana 2000). Despite being sometimes fuzzier, the results were qualitatively consistent with those obtained for the simple Hamming distance. Results obtained with this plastic measure would not be reported throughout the text.
When correlations between characters were analyzed, the pairs of data could not be considered as independently drawn from the same distribution because viroid species are part of a hierarchically structured phylogeny. This nonindependence was circumvented by taking into account the phylogenetic relationships between species (Elena et al. 2001, Fig. 2) and using the independent contrast method (ICM, Felsenstein 1985).
|
PERL scripts were written for automating all the necessary computations and comparisons involving programs from the ViennaRNA package. These scripts are available upon request. All simulations were performed on a 12-node Linux cluster. Statistical analyses were done using the SPSS package v12.0 (http://www.spss.com).
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Sign and Intensity of Epistasis
All 3L pairs of random mutations were classified into 3 categories depending on whether they interacted additively, synergistically, or antagonistically (table 1). In most cases (86.69%), Hamming distances were strictly additive (
ij = 0). However, when pairs of mutations were involved in a nonadditive way, antagonistic interactions (ij > 0) were significantly more abundant than synergistic (ij < 0) ones in 28 of the viroid species (signs tests; P < 0.0001 in all cases except for PSTVd [P = 0.0022], CCCVd [P = 0.0030], and ASSVd [P = 0.0143]). CBLVd was the only case in which the number of antagonistic and synergistic interactions was approximately equal (P = 0.3963). IrVd-1 was the viroid species showing the largest additivity (91.44%), GYSVd-1 showed the largest fraction of pairs of mutations involved in synergistic interactions (11.26%), whereas PBCVd showed the largest fraction of antagonistic pairs (41.16%) and the lowest additivity (50.90%) on RNA structure.
Average ranged from slightly synergistic (–0.0128 for MPVd) to strongly antagonistic (0.1286 for PCBVd), whereas the standard deviation ranged from 0.0103 (for CCCVd) to 0.3382 (for GYSVd-1) (table 2). Despite antagonistic epistasis being clearly more abundant, the average epistasis coefficient turned out to be negative in 9 species. This occurred because synergistic interactions, despite their lower frequency, tended to be more intense. Synergistic interactions were significantly stronger than antagonistic interactions in 12 species (Mann–Whitney tests, P 
0.0219), whereas the contrary occurred in only 2 species (PLMVd and CEVd). Indeed, in 8 out of the 9 cases where the average epistasis coefficient was negative, the skewness of the distribution was significantly negative, dragging mean epistasis coefficients toward negative values (table 2).
|
To explore whether the taxonomic relationships between viroids would account for the differences observed in epistasis among random pairs of mutations, we computed a model II nested analysis of variance in which species were nested within genera and genera within families. The error term of the model was constructed by looking at differences between all 3L random pairs of mutation for each viroid species. Table 3 reports the results of this analysis. Significant differences in epistasis were detected among species belonging to the same genus, although only 4.33% of total epistasis variance could be explained by differences between species. Significant differences were detected neither among genera belonging to the same family nor among families.
|
Most viroid species fold into a rodlike or quasi-rodlike structure, whereas some do as branched conformations. The number of hairpin loops in each MFESS can be used as an indicator of how branched a structure is. Viroids folding into a rodlike conformation have 2 hairpin loops, whereas this number increases up to 11 for CChMVd, the most branched viroid. We observed that the fraction of additive interactions negatively correlated with the number of hairpin loops (ICM, r = –0.7016, 25 df, P < 0.0001). Conversely, antagonistic interactions tended to be more frequent in structures with more hairpin loops (ICM, r = 0.7057, 25 df, P < 0.0001), and though less stressed, the same qualitative pattern was observed for synergistic interactions (ICM, r = 0.4185, 25 df, P = 0.0298).
To go further into the structural details responsible for the observed pattern of epistasis, we chose 5 species with increasing numbers of structural domains (i.e., hairpin loops). PSTVd, which folds into the prototypic rodlike MFESS; CSVd, which adopts a cruciform MFESS with 4 hairpin loops; PBCVd, with the most branched folding among Pospiviroid, with 6 hairpin loops; and 2 highly branched members from the Avsunviroidae family, PLMVd (9 hairpin loops) and CChMVd. Regarding the 4 branched structures, we found that in all 4 species, epistasis was more frequent (42.12%) among mutational pairs hitting the same structural domain than among pairs hitting different domains (30.37%) (Fisher's exact test, P < 0.0001 in all 4 species). For PSTVd, at best, a single structural domain can be distinguished, the entire rod. We hence mapped the nucleotidic position of each mutational pair to test whether mutations hitting nearby positions tended to show more epistasis. Figure 1A shows the positions of antagonistic pairs along PSTVd sequence and, for the sake of comparison, for PLMVd sequence (fig. 1B). We concluded that mutation pairs hitting nearby positions in the secondary structure were more prone to be engaged in antagonistic epistasis. Additionally, antagonism occurred preferentially in short hairpins. For synergistic interactions, we did not detect a clear structural clustering.
|
The Roles of Compensatory Mutations and Multiple Hitting
It has been previously reported that compensatory mutations may produce an excess of antagonistic epistasis in RNA secondary structures (Wilke et al. 2003
). Alternatively, because viroids have extremely short genomes, it is plausible that multiple mutations may repeatedly hit the same structure. If the first mutation has seriously affected the ability to fold in a given structure, or even entirely prevented its formation, then the second mutation will more likely be neutral for this structure, and therefore, this case will be counted as antagonistic epistasis. Two categories of compensatory mutations can be distinguished. First, strictly compensatory mutations are those for which s(dij,wt) < min{s(di,wt), s(dj,wt)} such that the MFESS of the double mutant is less affected than for each single mutant, a situation expected when baseparing is restored in the double mutant. Secondly, broad-sense compensatory mutations are those for which s(dij,wt) < max{s(di,wt), s(dj,wt)}; in other words, the effect of the double mutant on MFESS is smaller than the effect of at least one of the single mutations. For each viroid species, we counted how many pairs of mutations belonged to each category. If compensatory mutations were entirely responsible for the observed pattern of epistasis, such excess of antagonistic cases would not exist anymore after removing pairs of compensatory mutations from the data set. Table 1 shows the percentages of antagonistic interactions after removing each class of compensatory pairs. Across viroid species, strictly compensatory cases represented only 0.38% of all mutation pairs. When these cases were removed from the analysis, epistasis still remained significantly antagonistic for 28 species (table 1). Broad-sense compensatory mutations represented 7.26% of all mutation pairs. Even when all broad-sense cases were excluded from the analysis, epistasis remained significantly antagonistic in 10 viroid species, whereas it was synergistic only in CBLVd (table 1, P = 0.0133). We concluded that compensatory changes are not the only factor that contributes to creating antagonistic epistasis in viroid secondary RNA structures and that, therefore, multiple hitting of the same structural element makes also a significant contribution.
Figure 1 provides some insights into the physical mapping of these 2 types of mutations. For PSTVd, the mutation pairs showing antagonistic epistasis clearly tended to group along 2 diagonals. Those pairs falling along the "direct" diagonal revels the multiple hitting effect of the same structure, whereas cases falling along the "reverse" diagonal correspond to the complementary part of the rod and, hence, can indicate multiple hitting or compensatory effects. Strictly compensatory mutations fall nearly exclusively in the "reverse" diagonal. This pattern is only visible for the largest stem of PLMVd. Synergistic cases did not follow a clear pattern (not shown).
Correlation between Robustness and Epistasis
We sought to investigate the relationship between mutational robustness (as determined by the average mutational effect) and the strength of epistatic interactions in viroid sequences. Figure 2 shows the relationship between the average mutational effects estimated for each viroid species (Sanjuán et al. 2006) and their corresponding 
values. The value of the correlation coefficient was significantly positive (ICM, r = 0.8773, 25 df, P < 0.0001). To rule out the possibility that this result was driven by differences in genome length, we repeated our calculations using the Hamming distance as a direct indicator of the mutational effect, that is, without dividing by genome length. After doing so, the above correlation remained nearly unchanged (ICM, r = 0.8593, 25 df, P < 0.0001). Therefore, we concluded that epistasis and mutational effects are not independent traits but, instead, may evolve hand-by-hand: a reduction in the magnitude of mutational effects (i.e., an increase in robustness) is associated with a reduction in the magnitude of antagonistic epistasis.
Genetic Redundancy Relaxes Antagonistic Epistasis
Taking advantage of the existence of 3 viroid species for which natural variants with partial genome duplications have been isolated, we previously observed that robustness and genetic redundancy were positively correlated (Sanjuán et al. 2006). Compared with the reference variant, CEVd C (the one that appears in tables 1 and 2), variant D-104 contains a duplication of 104 nt that affects the right terminal domain of the folded molecule. Similarly, compared with the reference variant, CCCVd fast (the one used in tables 1 and 2), CCCVd slow presents a duplication of 41 nt at the right terminal domain. These 2 cases represent paradigmatic examples of partial duplications of structural elements. The case of CbVd-1 and CbVd-3 is somehow more complex. CbVd-1 is the shortest and CbVd-3 the largest among the Coleviroid (116 nt longer). However, differences in length result both from insertions of short stretches of nucleotides in both the upper and lower strands of the rodlike structure and from the addition of terminal loops to the shorter CbVd-1 molecule. Although this case does not strictly represent a duplication event, certainly it represents another instance of extra genetic material. If duplicated genomes are less sensitive to the effect of mutations (Sanjuán et al. 2006) and the magnitude of mutational effects correlates with the intensity of antagonistic epistasis, then now we should expect antagonistic epistasis to be less common in genomes containing duplications than in unitary ones.
To test this prediction, we obtained all 3L pairs of random mutations for the 3 complex variants. As before, pairs were classified as additive, synergistic, or antagonistic. Figure 3 illustrates the effect of genetic redundancy on the percentage of pairs showing antagonistic and additive interactions. A Spearman's partial correlation coefficient, controlling for viroid species, showed that the number of antagonistic interactions significantly decreased as genome redundancy increased (
= –0.8783, 3 df, 1-tailed P = 0.0250), with a concomitant increase in the number of additive interactions between random pairs of mutations ( = 0.8783, 3 df, 1-tailed P = 0.0250). (The number of synergistic interactions did not change with increased genome redundancy [ = –0.0976, 3 df, 1-tailed P = 0.4380].)
|
In good agreement, for all 3 viroid species
was reduced. In the case of CbVd-3, the reduction was highly significant (Mann–Whitney test, P < 0.0001), and even became synergistic (table 2). However, the magnitude of the reduction was significant neither for CCCVd slow (
= –58.03%, P = 0.8281) nor in the case of CEVd D-104 ( = –20.27%, P = 0.1307). We have interpreted the above results as a consequence of redundant genomes being more robust. However, an alternative explanation exists: our metric is corrected per genome length and this might produce lower epistasis coefficients in longer genomes. To rule out this possibility, we computed a correlation coefficient between epistasis and genome length. If the alternative hypothesis was true, then a negative correlation is expected among these variables. Obviously, to avoid confounding effects, CbVd-3 was excluded from this computation. A negative, but not significant, correlation was found (ICM, r = –0.1708, 24 df, P = 0.4042), which suggests that, if any, the effect of genome length was not large enough to explain the change in epistasis observed for CCCVd slow, CEVd D-104, and CbVd-3.
An Evolutionary Trend toward Reducing Antagonistic Epistasis
As mentioned in Introduction, in the first paper of this series, we found an evolutionary trend toward increased robustness during viroid phylogenetic radiation (Sanjuán et al. 2006). Because epistasis is correlated to mutational effects, whatever evolutionary force favored the evolution of increased robustness, a concomitant evolutionary trend toward reduced antagonistic epistasis should be observed. To test this prediction, we mapped into the phylogenetic tree proposed by Elena et al. (2001). The root of this phylogenetic tree was placed by comparing viroid sequences with viroidlike (the so-called virusoids) and linear RNA satellites of viruses, with which they share certain structural properties (Diener 1989, 1991; Elena et al. 1991, 2001). Figure 4 shows the estimated for each genus (calculated as the average over all members in the genus) along with the phylogenetic tree linking the different genera. According with this tree, the split between Pelamoviroid and Avsunviroid took place earlier on during the phylogenetic radiation of viroids, whereas Hostuviroid, Pospiviroid, and Cocadviroid were the most recently divergent genera. Notice that ELVd was excluded from the analysis because its exact location in the phylogenetic tree has not been determined yet. We assigned a rank order to each node, starting from the oldest one and using tied ranks for genera derived from unresolved nodes. We found a negative correlation between the above ranks (i.e., phylogenetic deepness) and the average epistasis coefficient (r = –0.6861, 26 df, P = 0.0001). This suggests that during their evolutionary radiation, viroids changed their genomic architecture from one characterized by lower mutational robustness and stronger antagonistic epistasis (the branched structures typical of Avsunviroidae) toward another characterized by higher mutational robustness and weaker antagonistic epistasis (the rodlike conformation typical of Pospiviroidae).
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Although evolutionary biologists have been interested in epistasis for decades, this interest has recently been bolstered by the relationship between epistasis and mutational robustness, one of the hottest topics in current evolutionary research (Barkai and Leibler 1997
; Bornholdt and Sneppen 2000; von Dassow et al. 2000; de Visser et al. 2003; Wilke et al. 2003; Hermisson and Wagner 2004; Kitano 2004; Stelling et al. 2004; Wagner 2004; Azevedo et al. 2006; Elena et al. 2006). We have found a negative correlation between mutational robustness and the strength of antagonistic epistasis (fig. 2). This correlation seems to be a ubiquitous phenomenon that has been observed with digital organisms (Wilke and Adami 2001), in in silico models of bacteriophage T7 infectious cycle (You and Yin 2002) and in simulated RNA folding (Wilke et al. 2003). The dependence of epistasis on mutational effects means that both parameters cannot be evolutionarily optimized independently. If robustness evolves, then the strength of antagonistic epistasis should be relaxed. This prediction is confirmed by our comparative analyses of robustness (Sanjuán et al. 2006, Fig. 1) and epistasis (fig. 4) across viroid genera.
A causal connection between synergistic epistasis and mutational robustness can be established with the following argumentation. Robustness should occur when an organism carries several copies of the same gene (duplications or polyploidy) or through biochemical buffering mechanisms as parallel metabolic pathways, repair systems, or chaperone proteins. In such organism, mutations may accumulate without noticeably affecting the phenotype. However, after the redundant functions/buffering mechanisms have been knocked out by mutation, the organism will start manifesting the effect of the accumulated load (Elena et al. 2006). Therefore, 2 hallmarks for robust systems are 1) single mutations may have mild effects on an organism's fitness and 2) epistasis among mutations should be predominantly synergistic. By contrast, a lack of robustness is expected in haploid genomes that have no duplications, overlapping gene functions, few repair systems, and little ability to buffer mutation in general. In such systems, mutations may have a large impact on fitness, but as they accumulate, their marginal contribution to fitness diminishes (Elena et al. 2006). Therefore, 2 characteristic properties for a nonrobust system are that 1) single mutations may have large fitness effects and 2) epistasis among mutations should be predominantly antagonistic.
In excellent agreement with these predictions, it has been clearly established along the last few years that, although variance exists both in the sign and intensity of epistasis, on average, antagonistic epistasis is the rule for RNA viruses (Bonhoeffer et al. 2004; Burch and Chao 2004; Sanjuán et al. 2004b; Sanjuán 2006) and that the average fitness effect associated with point mutations in RNA genomes is larger than values estimated for DNA organisms (Sanjuán et al. 2004a; Elena et al. 2006). Here we have extended these results to the case of viroids and found evidences for an excess of antagonistic epistasis among random pairs of mutations affecting their MFESS. According to our results, antagonistic epistasis does not only arise as a consequence of compensatory mutations, as reported for simulated RNA folding (Wilke et al. 2003) but also as a consequence of mutations recurrently affecting the same structure.
Despite RNA viruses and viroids clearly adopting a nonrobust strategy compared with much more complex organisms, as eukaryotes or even bacteria, an evolutionary trend toward increased robustness and reduced antagonistic epistasis can still be observed when looking into the evolution of the viroids (Sanjuán et al. 2006, Fig. 1). This is not contradictory at all because the magnitude of the genomic changes involved in the evolution of viroids is of a much lower order than that leading to the evolution of complex organisms. A clue for understanding how viroids may have evolved increased robustness and decreased antagonistic epistasis may come from work with digital organisms (Edlund and Adami 2004; Misevic et al. 2005). It has been observed that an increase in genetic compactness (traits encoded per genomic unit) and modularity (both physical and functional) correlates with a tendency toward reducing the strength of antagonistic epistasis. Viroids rely for most of their catalytic functions on host enzymatic systems. Nonetheless, whereas Avsunviroidae still maintain the hammerhead ribozyme structure involved in selfcleavage of replication intermediates, Pospiviroidae have even lost this capability (Flores et al. 2005). The presence of the hammerhead imposes important structural constraints, likely linked with the branched structure characteristic of the Avsunviroid, and the nucleotide sites involved in correctly establishing the hammerhead are also involved in other structural elements (Flores et al. 2005). After getting rid of the hammerhead ribozyme, Pospiviroid structures would not anymore be under the functional constraints imposed by the selfcatalysis and, therefore, their structure may evolve from branched towards rodlike, increasing their structural robustness to mutation and showing relaxed antagonistic epistasis.
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
We are grateful to José A. Daròs and all the virology crew at the Instituto de Biología Molecular y Celular de Plantas for comments and suggestions. This research was supported by grants BMC2003-00066 and BFU2005-23720-E/BMC from the Spanish Ministerio de Educación y Ciencia–FEDER and by the European Molecular Biology Organization Young Investigator Program to S.F.E.
|
Footnotes |
|---|
Edward Holmes, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionAcknowledgementsReferences |
|---|
Ancel LW and Fontana W. (2000) Plasticity, evolvability, and modularity in RNA. J Exp Zool 288:242–83.[CrossRef][ISI][Medline]
Ancel Meyers LW, Lee JF, Cowperthwaite M, Ellington AD. (2004) The robustness of naturally and artificially selected nucleic acid secondary structures. J Mol Evol 58:681–91.[CrossRef][ISI][Medline]
Azevedo RB, Lohaus R, Srinivasan S, Dang KK, Burch CL. (2006) Sexual reproduction selects for robustness and negative epistasis in artificial gene networks. Nature 440:87–90.[CrossRef][Medline]
Barkai N and Leibler S. (1997) Robustness in simple biochemical networks. Nature 387:913–7.[CrossRef][Medline]
Bonhoeffer S, Chappey C, Parkin NT, Whitcomb JM, Petropoulos CJ. (2004) Evidence for positive epistasis in HIV-1. Science 306:1547–50.
Bornholdt S and Sneppen K. (2000) Robustness as an evolutionary principle. Proc R Soc Lond B 267:2281–6.[Medline]
Burch CL and Chao L. (2004) Epistasis and its relationship to canalization in the RNA virus 
6. Genetics 167:559–67.
Coyne JA. (1992) The genetics of speciation. Nature 355:511–5.[CrossRef]
de Visser JAGM, Hermisson J, Wagner GP, et al. (19 co-authors). (2003) Evolution and detection of genetic robustness. Evolution 57:1959–72.[CrossRef][ISI][Medline]
Diener TO. (1989) Circular RNAs: relics of precellular evolution? Proc Natl Acad Sci USA 86:9370–4.
Diener TO. (1991) Subviral pathogens of plants: viroids and viroidlike satellite RNAs. FASEB J 5:2808–13.[Abstract]
Edlund JA and Adami C. (2004) Evolution of robustness in digital organisms. Artif Life 10:167–79.[CrossRef][ISI][Medline]
Elena SF, Carrasco P, Daròs JA, Sanjuán R. (2006) Mechanisms of genetic robustness in RNA viruses. EMBO Rep 7:168–73.[CrossRef][ISI][Medline]
Elena SF, Dopazo J, de la Peña M, Flores R, Diener TO, Moya A. (2001) Phylogenetic analysis of viroid and viroid-like satellite RNAs from plants: a reassessment. J Mol Evol 53:155–9.[ISI][Medline]
Elena SF, Dopazo J, Flores R, Diener TO, Moya A. (1991) Phylogeny of viroids, viroidlike satellite RNAs, and the viroidlike domain of hepatitis 
virus RNA. Proc Natl Acad Sci USA 88:5631–4.
Felsenstein J. (1985) Phylogenies and the comparative method. Am Nat 125:1–15.[CrossRef][ISI]
Flores R, Hernández C, Martínez de Alba E, Daròs JA, di Serio F. (2005) Viroids and viroid-host interactions. Annu Rev Phytopathol 43:117–39.[Medline]
Fontana W. (2002) Modelling ‘evo-devo’ with RNA. BioEssays 24:116477.
Gavrilets S. (1999) A dynamical theory of speciation on holey adaptive landscapes. Am Nat 154:1–22.[CrossRef][ISI]
Hermisson J and Wagner GP. (2004) The population genetic theory of hidden variation and genetic robustness. Genetics 168:227184.
Hill WG. (1982) Dominance and epistasis as components of heterosis. Sonderdruck aus Zeitschrift für Tierzüchtungsbiologie 99:161–8.
Hofacker IL, Fontana W, Stadler PF, Bonhoeffer LS, Tacker M, Schuster P. (1994) Fast folding and comparison of RNA secondary structures. Monatsh Chem 125:167–88.[CrossRef]
Kitano H. (2004) Biological robustness. Nat Rev Genet 5:826–37.[CrossRef][ISI][Medline]
Kondrashov AS. (1998) Deleterious mutations and the evolution of sexual reproduction. Nature 336:43540.
Kondrashov AS and Crow JF. (1991) Haploidy or diploidy: which is better? . Nature 351:3145.
Lenski RE, Ofria C, Collier TC, Adami C. (1999) Genome complexity, robustness and genetic interactions in digital organisms. Nature 400:6614.[Medline]
Misevic D, Ofria C, Lenski RE. (2005) Sexual reproduction reshapes the genetic architecture of digital organisms. Proc R Soc B 273:457–64.
Peck JR and Waxman D. (2000) Mutation and sex in a competitive world. Nature 406:399404.[CrossRef][Medline]
Sanjuán R. (2006) Quantifying antagonistic epistasis in a multifunctional RNA secondary structure of the Rous sarcoma virus. J Gen Virol 87:1595–602.
Sanjuán R, Forment J, Elena SF. (2006) In silico predicted robustness of viroids RNA secondary structure. I. The effect of single mutations. Mol Biol Evol 23:1427–36.
Sanjuán R, Moya A, Elena SF. (2004a) The distribution of fitness effects caused by single-nucleotide substitutions in an RNA virus. Proc Natl Acad Sci USA 101:8396–401.
Sanjuán R, Moya A, Elena SF. (2004b) The contribution of epistasis to the architecture of fitness in an RNA virus. Proc Natl Acad Sci USA 101:15376–9.
Schultes EA, Hraber PT, LaBean TH. (1999) Estimating the contribution of selection and self-organization in RNA secondary structure. J Mol Evol 49:76–83.[CrossRef][ISI][Medline]
Schuster P and Fontana W. (1999) Chance and necessity in evolution: lessons from RNA. Physica D 133:427–52.
Schuster P, Fontana W, Stadler PF, Hofacker IL. (1994) From sequences to shapes and back: a case study in RNA secondary structures. Proc R Soc Lond B 255:279–84.[Medline]
Stelling J, Sauer U, Szallasi Z, Doyle FJ III, Doyle J. (2004) Robustness of cellular functions. Cell 118:675–85.[CrossRef][ISI][Medline]
von Dassow G, Meir E, Munro EM, Odell GM. (2000) The segment polarity network is a robust developmental module. Nature 406:16892.
Wagner A. (2004) Distributed robustness versus redundancy as causes of mutational robustness. BioEssays 27:17688.
Whitlock MC, Phillips PC, Moore FBG, Tonsor SJ. (1995) Multiple fitness peaks and epistasis. Annu Rev Ecol Syst 26:601–29.[CrossRef][ISI]
Wilke CO and Adami C. (2001) Interaction between directional epistasis and average mutational effects. Proc R Soc Lond B 268:1469–74.[Medline]
Wilke CO, Lenski RE, Adami C. (2003) Compensatory mutations cause excess of antagonistic epistasis in RNA secondary structure folding. BMC Evol Biol 3:1–14.[CrossRef][Medline]
Wolf JB, Brodie ED III, Wade MJ. (2000) Epistasis and the evolutionary process. (Oxford University Press, Oxford).
Wright S. (1982) The shifting balance theory and macroevolution. Annu Rev Genet 16:1–19.[Medline]
You L and Yin J. (2002) Dependence of epistasis on environment and mutation severity as revealed by in silico mutagenesis of phage T7. Genetics 160:127381.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Carrasco, F. de la Iglesia, and S. F. Elena Distribution of Fitness and Virulence Effects Caused by Single-Nucleotide Substitutions in Tobacco Etch Virus J. Virol., December 1, 2007; 81(23): 12979 - 12984. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Shu, X. Bo, Z. Zheng, and S. Wang RSRE: RNA structural robustness evaluator Nucleic Acids Res., July 13, 2007; 35(suppl_2): W314 - W319. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. de la Iglesia and S. F. Elena Fitness Declines in Tobacco Etch Virus upon Serial Bottleneck Transfers J. Virol., May 15, 2007; 81(10): 4941 - 4947. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||