A toxin-antidote selfish element increases fitness of its host

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife assessment

    This study addresses a fundamental question about the origin and evolution of selfish genetic elements, focusing on the paradoxical abundance of toxin-antidote elements in selfing Caenorhabditis species. The authors propose for the C. elegans peel-1 zeel-1 locus fitness advantages; if these the findings can be supported with additional data, they will be of considerable interest to the field due to their wider implications for the evolution of such systems.

This article has been Reviewed by the following groups

Read the full article See related articles


Selfish genetic elements can promote their transmission at the expense of individual survival, creating conflict between the element and the rest of the genome. Recently, a large number of toxin-antidote (TA) post-segregation distorters have been identified in non-obligate outcrossing nematodes. Their origin and the evolutionary forces that keep them at intermediate population frequencies are poorly understood. Here, we study a TA element in Caenorhabditis elegans called zeel-1;peel-1 . Two major haplotypes of this locus, with and without the selfish element, segregate in C. elegans . We evaluate the fitness consequences of the zeel-1;peel-1 element outside of its role in gene drive in non-outcrossing animals and demonstrate that loss of the toxin peel-1 decreased fitness of hermaphrodites and resulted in reductions in fecundity and body size. These findings suggest a biological role for peel-1 beyond toxin lethality. This work demonstrates that a TA element can provide a fitness benefit to its hosts either during their initial evolution or by being co-opted by the animals following their selfish spread. These findings guide our understanding on how TA elements can remain in a population where gene drive is minimized, helping resolve the mystery of prevalent TA elements in selfing animals.

Article activity feed

  1. eLife assessment

    This study addresses a fundamental question about the origin and evolution of selfish genetic elements, focusing on the paradoxical abundance of toxin-antidote elements in selfing Caenorhabditis species. The authors propose for the C. elegans peel-1 zeel-1 locus fitness advantages; if these the findings can be supported with additional data, they will be of considerable interest to the field due to their wider implications for the evolution of such systems.

  2. Reviewer #1 (Public Review):

    Overall, Long et al. very nicely show that the peel-1 locus gives a fitness benefit to strains independent of the zeel-1 gene. This famous TA element has been characterized solely for its role as a selfish genetic element, even though the original authors mused that it could have arisen because of a fitness benefit. This manuscript makes a valuable contribution by using both modeling and empirical results to show this point. The results have broad implications for the evolution of TA elements.

  3. Reviewer #2 (Public Review):

    In this manuscript, Long and colleagues explore a very fundamental question regarding the origin and evolution of selfish genetic elements. In particular, they focus their study on the paradoxical abundance of toxin-antidote elements in Caenorhabditis species that reproduce largely by selfing. As a model system, they study the C. elegans peel-1/zeel-1 locus, the first TA to be molecularly dissected in eukaryotes.

    Major strengths

    1. The manuscript is well-written and easy to follow.
    2. It tackles a very interesting question. Toxin-antidote elements are made of two genes, one coding for a toxin and a second one for its cognate antidote. Although these selfish genes seem relatively simple, there are two paradoxes associated with their evolutionary inception. First, what function evolved first? How can a toxin evolve in the absence of an antidote? Why would an antidote evolve in the absence of a toxin? Second, how does gene drive evolve in selfing nematodes? Toxin-antidote elements thrive under conditions that maximize their dispersal, that is, outcrossing. So, why are toxin-antidote elements so common in nematodes that mainly reproduce by selfing? The main finding of Long and colleagues, namely, that the toxin peel-1 increases the fitness of selfing hermaphrodites, has the potential to change how we think about these ubiquitous selfish elements.

    Major weaknesses

    Although the results presented by the authors are interesting and suggestive, I find the evidence largely insufficient. In particular, a lack of appropriate controls in the following experiments.

    1. The main claim of this paper boils down to a single experiment. Figure 3C and 3E. In particular the contrast between N2(marker) worms and N2 (peel-1 null; marker) strains. In essence, the authors show that peel-1 null worms lay 6% fewer embryos than WT and that they are outcompeted by N2 worms when co-cultured. However, I feel this is not properly controlled. Every time one generates a mutant worm by CRISPR (or other means) there is a chance that a secondary non-desired change is introduced. This could be due to the technique itself, for instance, CRISPR gRNA having an off-target or it could be derived from the transgenesis procedure itself. That is, the bottleneck effect associated with injecting worms and picking single progeny to establish mutant lines that could fix random mutations. Since these effects are to some degree unavoidable, careful controls must be provided. First and foremost is the generation of independent alleles. As far as I could tell, the authors only mention and do experiments with a single peel-1 null mutant. There is also no mention of backcrossing strains to the parental strain in the methods section. This is particularly troubling because at the end of the day, the authors based the whole paper on a very modest effect on fitness. Now, such a modest effect, of course, would be sufficient for natural selection to act upon in the wild, but at the same time, it could be perfectly caused by off-targeting or genetic drift of random mutations in the background. If one were to take "N2" reference strains from different labs in the world, I'm pretty sure that we would see differences in fitness, most of them with a larger effect size.

    In my opinion, a critical control missing in the study would be a "rescue" experiment by performing CRISPR editing on the peel-1 null mutant line and "fixing" the toxin allele. This should restore the phenotypes back to WT levels and would discard any secondary off-target effects. The authors could claim that the NIL experiment (Figure 2) strengthens their view because they see a similar effect as in the peel-1 null worm. However, as they also point out, these worms have >100kb introgression with multiple genes in there, thus any small effect on fitness could be perfectly due to linkage. For consistency, one would have expected also to make a peel-1 null allele in the NIL background, but that experiment was not provided either.

    In summary, there are many trivial ways in which a mutant line will have a slight decrease in fitness and none of these are controlled in this manuscript. Moreover, decreasing fitness is trivial, but increasing it, is not.

    2. The authors propose PEEL-1 increases the fitness of hermaphrodites by making them lay more eggs. Now, as far as we know, PEEL-1 is not expressed in the female gonad, only in sperm. Thus, one logical conclusion (or the only simple scenario I can think of) would be that PEEL-1 increases the total number of mature sperm in hermaphrodites. I think that further work characterizing this phenomenon would be fundamental to strengthen the claim made by the authors.

  4. Reviewer #3 (Public Review):

    This paper aimed to understand how toxin-antidote (TA) elements are spread and maintained in species, especially in species where outcrossing is infrequent and the selfish gene drive of TA elements is limited. The paper focuses on the possible fitness costs and benefits of the peel-1/zeel-1 element in the nematode C. elegans. A combination of mathematical modeling and experimental tests of fitness are presented. The authors make a surprising finding: the toxin gene peel-1 provides a fitness advantage to the host. This is a very interesting finding that challenges how we think about selfish genetic elements, demonstrating that they may not be wholly "selfish" in order to spread in a population.

    1. The authors support results found with a zeel-1 peel-1 introgressed strain by using CRISPR/Cas9 genetic engineering to precise knock-out the genes of interest. They were careful to ensure the loss-of-function of these generated alleles by using genetic crosses.

    2. Similarly, the authors are careful with controls, ensuring that genetic markers used in the fitness assays did not affect the fitness of the strain. This ensures that the genes of interest are causative for any source of fitness differences between strains, therefore making the data reliable and easily interpretable.

    3. A powerful assay for directly measuring the relative fitness of two strains is used.

    4. The authors support relative fitness data with direct measurements of fitness proximal traits such as body size (a proxy for growth rate) and fecundity, providing further support for the conclusion that peel-1 increases fitness.

    1. One major conclusion is that peel-1 increases fitness independent of zeel-1, but this claim is not well supported by the data. The data presented show that the presence of zeel-1 does not provide a fitness benefit to a peel-1(null) worm. But the experiment does not test whether zeel-1 is required for the increased fitness conferred by the presence of peel-1. Ideally, one would test whether a zeel-1(null);peel-1(+) strain is as fit as a zeel-1(+);peel-1(+) strain, but this experiment may be infeasible since a zeel-1(null);peel-1(+) strain is inviable.

    2. The CRISPR-generated peel-1 allele in the N2 background only accounts for 32% of the fitness difference of the introgressed strain. Thus, the effect of peel-1 alone on fitness appears to be rather small. Additionally, this effect of peel-1 shows only weak statistical significance (and see point 5 below). Given that this is the key experiment in the paper, the major conclusion of the paper that the presence of peel-1 provides a fitness benefit is supported only weakly. For example, it is possible that other mutations caused by off-target effects of CRISPR in this strain may contribute to its decreased fitness. It would be valuable to point out the caveats to this conclusion, or back it up more strongly with additional experiments such as rescuing the peel-1(null) fitness defect with a wild-type peel-1 allele or determining if the introduction of wild-type peel-1 into the introgressed strain is sufficient to confer a fitness benefit.

    3. The strain that introgresses the zeel-1 peel-1 region from CB4856 into the N2 background was made by a different lab. Given that N2 strains from different labs can vary considerably, it is unclear whether this introgressed strain is indeed isogenic to the N2 strain it is competing against, or whether other background mutations outside the introgressed region may contribute to the observed fitness differences.

    4. Though the CRISPR-generated null allele of peel-1 only accounts for 32% of the fitness difference of the zeel-1 peel-1 introgressed strain, these two strains have very similar fecundity and growth rates. Thus, it is unclear why this mutant does not more fully account for the fitness differences.

    5. Improper statistical tests are used. All comparisons use a t-test, but this test is inappropriate when multiple comparisons are made. Importantly, correction for multiple comparisons may decrease the already weak statistical significance of the fitness costs of the peel-1 CRISPR allele (Fig 3E), which is the key result in the paper.

    6. N2 fecundity and growth rate measurements from Fig 2B&C are reused in Fig 3C&D. This should be explicitly stated. It should also be stated whether all three strains (N2, the zeel-1 peel-1 introgressed strain, and the peel-1 CRISPR mutant) were assayed in parallel as they should be. If so, a statistical test that corrects for multiple comparisons should also be used.

    7. It appears that the same data for the controls for the fitness experiments (i.e. N2 vs. marker & N2 vs. introgressed npr-1; glb-5) may be reused in Fig 2A and 3E. If so, this should be stated. It should also be stated whether all the experiments in these panels were performed in parallel. If so, this may affect the statistical significance when correcting for multiple comparisons.

  5. Reviewer #4 (Public Review):

    In "A Toxin-Antidote Selfish Element Increases Fitness of its Host", Long et al. attempt to address an outstanding question in the evolution of toxin-antidote (TA) systems in primarily selfing species: How do TA systems escape drift and spread in a primarily selfing species? The authors use simulations to show that at outcrossing rates similar to that observed in C. elegans a TA element, like the peel-1/zeel-1 element, has a high probability of being lost to genetic drift. However, the authors show that the peel-1 gene provides a fitness advantage to strains harboring it, providing evidence for a dual role for this gene and insights into how this element might have escaped being lost to genetic drift.


    The experiments in this paper are well-framed. The authors use simulations to show that the observed frequency of the peel-1/zeel-1 TA element in the C. elegans population is highly unlikely given the inferred outcrossing rates of species.

    The authors clearly show that the 140-370kb CB4856 introgression into N2 lowers relative fitness, number of eggs laid, and animal size, relative to N2.

    The authors generated null alleles of peel-1 and zeel-1 and showed that a truncated version of PEEL-1 confers a detrimental fitness effect when compared to N2. Furthermore, the authors show that the fitness effect associated with peel-1 is independent of the antidote (zeel-1) component of this TA element.


    1. The reference N2 strain has been cultivated in the lab for decades and many different versions of this strain exist. The different versions of N2, which might have slightly different genomes, are likely to have different fitness in laboratory conditions. It is unclear whether the N2 strain used to construct QX1198 is the same N2 strain used to construct CX12311, PTM229, and PTM377 (and others derived from these). The potential difference in the N2 strain used for the construction of these strains might contribute to the large discrepancy between the relative fitness shown in Figure 2A (~0.25) and Figure 3E-F (~0.07). Alternatively, the other CB4856-specific variants present in the 140-370 kb introgression in the QX1198 strain might cause this large discrepancy.
      Regardless of the potential discrepancy among N2 strains used as the genetic background, the claim that the presence of peel-1 confers higher relative fitness is supported by Figure 3E because PTM377/409 were presumably derived from the same N2 strain.

    2. For Figures 2B and 3C, the authors report the number of eggs laid per animal. C. elegans strains can lay embryos that do not hatch and therefore fail to develop into reproductive adults. Does the difference between N2 and N2(peel-1(0)) remain when considering the number of reproductively mature progeny? Presumably, eggs laid translate to reproductive adults because a relative fitness increase is observed when peel-1 is present.

    3. The authors did not perform whole-genome sequencing of the peel-1 and zeel-1 CRISPR edited strains or mention any backcrossing done to eliminate potential off-target editing events. Therefore it is difficult to conclude whether off-target effects might influence the quantified traits presented in Figure 3. This concern is somewhat alleviated by the reciprocal competition assay presented in Figure 3E (4th boxplot), but a potential off-target editing event that lowers fitness could have segregated with the silent dpy-10 and peel-1 edits.
      The same concern is present with the zeel-1-independence experiment, however, this experiment does not have reciprocal competition experiments.

    4. In Figure 3C-D, the authors show that a homozygous truncated version of PEEL-1 confers a reduction in eggs laid per animal (proxy for brood size) and animal length (proxy for developmental speed). However, the authors do not show whether a heterozygous truncated PEEL-1 strain (N2 peel-1/peel-1(kah126)) confers the same reduction in eggs laid or animal size. Would the allele frequency dynamics derived from the simulations be affected by a fitness advantage only being conferred by the presence of two copies of peel-1?

    5. The authors show a fitness advantage associated with peel-1 in laboratory conditions. It is obviously extremely difficult to extend these observations to the wild, however, the authors do not take their observations that peel-1 confers a fitness advantage in the lab and apply their empirical observations to the simulation framework. If the laboratory fitness advantage of peel-1 did extend to the wild, one might expect the element would fix in the population in the simulation framework.

    6. It seems possible that a truncated version of the PEEL-1 protein might have unknown deleterious fitness consequences that are independent of any beneficial effect the full-length protein might have. The same is true for the truncated ZEEL-1 protein, though potentially less concerning because there are only 5 amino acids.