Sexual reproduction is problematic to explain due to its costs, most notably the twofold cost of sex. Yet, sex has been suggested to be favourable in the presence of proliferating intragenomic parasites given that sexual recombination provides a mechanism to confine the accumulation of deleterious mutations. Kraaijeveld et al. compared recently the accumulation of transposons in sexually and asexually reproducing lines of the same species, the parasitoid wasp Leptopilina clavipes. They discovered that within asexually reproducing wasps, the number of gypsy-like retrotransposons was increased fourfold, whereas other retrotransposons were not. Interestingly, gypsy-like retrotransposons are closely related to retroviruses. Endogenous retroviruses are retroviruses that have integrated to the germ line cells and are inherited thereafter vertically. They can also replicate within the genome similarly to retrotransposons as well as form virus particles and infect previously uninfected cells. This highlights the possibility that endogenous retroviruses could play a role in the evolution of sexual reproduction. Here, we show with an individual-based computational model that a virus epidemic within a previously parasite-free asexual population may establish a new intragenomic parasite to the population. Moreover and in contrast to other transposons, the possibility of endogenous viruses to maintain a virus epidemic and simultaneously provide resistance to individuals carrying active endogenous viruses selects for the presence of active intragenomic parasites in the population despite their deleterious effects. Our results suggest that the viral nature of certain intragenomic parasites should be taken into account when sex and its benefits are being considered.
In a recent issue of Molecular Ecology, the role of intragenomic parasites in maintaining sexual reproduction was both experimentally evaluated by Kraaijeveld et al. (2012) and discussed by Crespi and Schwander (2012). The prevalence of sex is difficult to explain, due to its costs when compared with asexual reproduction (Maynard Smith 1978; Lehtonen et al. 2012; Meirmans et al. 2012). Yet, as reviewed by Crespi and Schwander, sex can be favourable in the presence of proliferating transposons. Transposons are similar to mutations, in that their integration to non-neutral loci is likely to have deleterious effects, and sexual recombination provides a potential mechanism to confine their accumulation (Arkhipova & Meselson 2005). Kraaijeveld et al. compared, for the first time on a genome scale, the accumulation of transposons in sexually and asexually reproducing lines of the same species, the parasitoid wasp Leptopilina clavipes. They discovered that within asexually reproducing wasps, the number of gypsy-like retrotransposons was increased fourfold, whereas other retrotransposons were not. Retrotransposons differ from other transposons in their replication mechanism that involves reverse transcriptase and RNA intermediate and is in this respect similar to the replication mechanism of retroviruses. While these results give some support to the intragenomic parasite hypothesis, the role of intragenomic parasites in maintaining sex was still left without a definite answer.
Interestingly, gypsy-like retrotransposons are closely related to retroviruses (Kotnova et al. 2007). These retrotransposons have been shown to be able to form virus-like particles and infect previously uninfected cells. This highlights the possibility that endogenous retroviruses could also have an important role in the evolution of sexual reproduction. More specifically, it is possible that stable asexual variants of species like Leptopilina clavipes (which do not appear to contain active replicating transposons) can acquire over time new proliferating intragenomic parasites that increase the cost of asexual reproduction. Sex may also generate variability within the population and thus restrict the spread of parasites (Hamilton et al. 1990; Howard & Lively 1994). In the absence of sex, clonal populations may be especially vulnerable to viral epidemics and thus new intragenomic parasites.
Endogenous retroviruses are genomic elements that enter the genome of their host via germ cell infections and are thereafter inherited vertically (Feschotte & Gilbert 2012). They can replicate within the genome by retrotransposition (similarly with retrotransposons) and by reinfecting germ line cells. To illustrate their potential as intragenomic parasites, we note a recent study by Magiorkinis et al. (2012) who examined the genomes of 38 mammals and demonstrated that many virus-like retroelements had turned into active transposon-like genomic superspreaders.
In terms of fending off transposons, sex is ‘a double-edged sword’, as Crespi and Schwander put it. Sex can help get rid of transposons via recombination, but, as a drawback, transposons can also spread horizontally between organisms due to sexual recombination of chromosomes. However, as a major difference to transposons, the horizontal spread of viruses is not dependent on sex: they can spread and persist within an asexual population. Therefore, sex cannot be considered a double-edged sword in the same way as with transposons, as viruses can spread between individuals even in the absence of sexual reproduction (Table 1).
Table 1. Sexual reproduction is a double-edged sword in terms of managing transposable elements. Purifying selection is more efficient in sexual reproduction, but so is their transmission. With endogenous viruses, this bias in transmission is alleviated by the possibility of horizontal transmission even in asexuals
Transmission of transposable elements
Transmission of endogenous viruses
Vertical and horizontal
Vertical and horizontal
Vertical and horizontal
Interestingly, endogenous viruses often make their host organisms resistant to external versions of the same endogenous virus (Feschotte & Gilbert 2012). The immunity derives from the superinfection resistance of viruses that prevents the secondary infection of the same host by a similar virus. Therefore, germ line infections can be adaptive during endemic virus epidemics, if the subsequent spread of the virus within the genome can be kept under control. If they are allowed to freely proliferate within the genome, the accumulating deleterious effects can become a highly deleterious burden to their hosts (Katzourakis et al. 2005).
As Crespi and Schwander argue (Crespi & Schwander 2012), removal of transposons from deleterious locations has been suggested to be one benefit of sexual reproduction, enabling its maintenance. According to previous research, the negative effects of proliferating endogenous retroviruses are comparable to those of deleterious mutations (Katzourakis et al. 2005). As an example of the potential of endogenous viruses to reduce fitness, koalas have recently acquired a new endogenous retrovirus causing death due to leukaemia and other symptoms (Tarlington et al. 2006). In organisms such as Leptopilina clavipes, the transposons that are present in the genome at the time of the emergence of asexual lines (due to Wolbachia parasitism) may be inactive and therefore cannot proliferate even in the absence of sexual recombination. This condition may change if a virus that transforms into a new endogenous virus infects the genome. As pointed out by Katzourakis et al., without any means to control new endogenous retroviruses, the number of viral elements in genomes may increase indefinitely at an exponential rate because of their potential to reinfect and/or replicate within germ line cells (Katzourakis et al. 2005). Inspired by the results of Kraaijiveld and colleagues, we use an individual-based model to investigate whether a new intragenomic parasite can be established in a previously parasite-free asexual population via a virus epidemic.
Genome and individuals
In order to investigate the potential evolutionary effects of virus epidemics and subsequent endogenization of viruses via germ cell infections, we constructed an individual-based population model. The individuals are asexually reproducing organisms with a single linear haploid molecule as their genome. The genome contains a predefined number of loci that can contain either an inactive or an active endogenous virus, or be free of endogenous viruses. Offspring of infected individuals have a probability of carrying an endogenous version of the virus in a randomly selected locus. The other loci can subsequently become occupied by an element when the first endogenous virus replicates within the genome.
In this model, all individuals die after they have reproduced. Therefore, a single iteration of the simulation corresponds to a generation. Each individual produces R offspring. Virus infections and the number of endogenous viruses in the genome at deleterious loci can decrease this number. Within a generation, individuals reproduce in random order until the carrying capacity of the system is reached.
An endogenous virus integrates to a locus. A given fraction of loci can be set to be deleterious in terms of their effects on the reproductive fitness of the individual; the remaining fraction is neutral. If no synergistic effects are set for mutations, then the fitness of the host organism decreases linearly with the fraction of deleterious loci occupied by active or defective endogenous viruses. Integration of a virus to a neutral locus does not affect the reproductive fitness of the organism.
Whether the fitness-reducing effect actually decreases the number of offspring is randomized for each new individual separately. For example, given a reproductive rate of 4 and a 0.5 fitness cost due to the accumulation of elements in a particular genome, there is a 50% chance for each of the four progeny to not be generated. This on average would yield two offspring, but it is possible that all four or none, for that matter, are produced.
In the model, individuals in the population can be infected with external (i.e. nonendogenous) viruses. An infection decreases the reproductive fitness of the individual, but has no other effects. An infected individual can spread the virus to other individuals based on the basic reproductive number (R0) of the virus. If the virus has R0 of 2, then an infected individual will infect two randomly selected individuals of the following generation. If these selected individuals are already infected, secondary infections have no effect. If the selected individuals carry an active endogenous virus in one their loci, they are resistant to infections.
Individuals that carry active endogenous viruses within their genomes can transmit the external version of the virus to another randomly selected individual within the population. Note that this transmission results in an infection and not in a horizontal transmission of an endogenous virus.
Endogenous viruses can be either active or inactive. Active endogenous viruses protect the host against infections by external viruses, but they are also able to replicate within the genome during the reproduction of the host. A replicated active endogenous virus is inserted into a randomly selected locus within the genome. Each active endogenous virus has a chance to become inactivated during reproduction of the host. Inactive viruses cannot be reactivated, and they do not protect the host against infections. However, inactive endogenous viruses in deleterious loci still decrease host fitness.
A single iteration of the model
Each iteration begins with calculation of the number of individuals infected by viruses and the number of subsequent infections they transmit. Second, the number of infections that individuals carrying endogenous viruses transmit is added to the number of total transmission events of viruses. Third, individuals reproduce according to their particular fitness (depending on the number of deleterious elements and infection status) in random order until the system reaches the carrying capacity (if possible). Endogenous viruses can randomly replicate or become inactivated in the genomes of the new individuals. Fourth, the infection events from the first two steps are transmitted individually to randomly selected individuals. Infections are successful only if the randomly chosen individual does not carry active endogenous viruses and is not already infected.
Selected parameter values
For simulations investigating the spread of an endogenous virus within the population, the number of loci was set to 200. In the absence of reliable experimental data for horizontal transmission rates of recently acquired endogenous viruses, we chose a relatively low value of 10−2 or 10−3 depending on the simulation. In other words, an endogenous virus-harbouring individual causes an extraneous infection in another uninfected individual once every hundred or thousand generations.
In our simulations, we chose a basic reproductive number of 2 for viruses. Synergistic effect of insertions was set to be 1.0, making the mutations nonsynergistic in terms of their fitness effect. Endogenization rate was set to be 10−3, indicating that for every offspring of an infected individual there is a chance that one of thousand offspring carries an endogenous virus in its genome.
In simulations where the eradication of endogenous viruses from an asexual population due to inactivation of the intragenomic parasites was investigated, we chose relatively high inactivation rates (3 × 10−4), no chance for endogenization after infection and mild infection-associated fitness cost (5%) to demonstrate that the epidemic can maintain active endogenous viruses in the population. All the values used in the simulations are listed in Table 2.
Table 2. Simulation values used in Figures 1 and 2
Virus epidemics can establish intragenomic parasites in asexual populations
We simulated scenarios where infections were introduced into populations of asexually reproducing organisms that were initially free of endogenous viruses. The initial infections rapidly led to a population-wide epidemic (Fig. 1a). In many cases, individuals carrying an endogenous virus increased in number within the population over time. A more severe infection-associated fitness cost resulted in faster fixation of endogenous viruses in the population (Fig. 1b). The reason for this is that in the model, endogenized viruses provide resistance against new infections. Therefore, higher infection costs lead to a greater selective advantage for individuals with endogenous viruses, despite their possible deleterious mutation-like effects. Bigger populations take longer to evolve from a parasite-free population into a parasite-containing population. In very small populations, random drift can produce a population saturated with parasites even in the absence of any fitness costs. It must be noted that in our model, individuals cannot acquire immunity in any other way except through endogenization. In natural systems, the competence of individuals to combat infections (and viruses to respond accordingly) is likely to play an important role. Nevertheless, it seems possible that a previously transposable element-free asexual population can become invaded by new intragenomic parasites due to a virus epidemic.
Dolgin and Charlesworth (2006) have previously noted that large populations of asexual organisms are purged of transposable elements even if the elimination rates of active elements are well below of their replication rates. This is because individuals losing their elements are favoured in the population. In our model, the possibility of hosts harbouring endogenous viruses to occasionally infect parasite-free individuals changes this outcome. We simulated a system where the loss of functional transposable elements is selected for, and compared systems (Table 2) where the transposable element is either an endogenous virus (and thus capable of causing infections in other individuals) or a regular transposon (only capable of within-genome replication). The population was set to start with five transposable elements in each genome. Interestingly, even relative mild viruses with the average of 5% infection-associated decrease in the reproductive fitness can maintain the endogenous virus within the population practically indefinitely (Fig. 2). The outcome was the same with various population sizes (500, 800, 1200, 2000 and 5000) in ten independent repetitions. This was true even in the absence of the possibility of infections to lead into the establishment of new endogenous viruses, genuinely favouring the parasites already in the population. In contrast to endogenous viruses, populations with transposons lost all of their active elements rapidly (not shown). Therefore, the possibility of endogenous viruses to maintain an epidemic in itself selects for active resistance-conferring elements within the population.
Altogether, the viral nature of some intragenomic parasites appears to be a quality that is necessary to be taken into consideration when the role of intragenomic parasites in the evolution of genomes is being evaluated. In a recent review, Feschotte and Gilbert (2012) wrote, ‘Population genetics predict that, for every fixed endogenous virus element insertion, thousands must have occurred in germ cells but were lost from the host population’. Given the abundance of endogenous retrovirus elements in the genomes of various organisms, along with the potential of endogenous retroviruses to spread within a population and to remain functional over evolutionary times (Katzourakis et al. 2005; Kazourakis et al. 2007; Magiorkinis et al. 2012), it is possible that for asexual organisms in particular, the potential for viruses to turn into intragenomic parasites may be problematic due to the lack of recombination. This notion calls for taking retroviruses, both external and endogenous, into account when sexual reproduction and its evolutionary benefits are investigated.
M.J. designed and prepared the model. Both authors contributed to the data analysis and the writing of the manuscript.