The transposition rate has little influence on the plateauing level of the P-element

On the origin of the P-element invasion in Drosophila simulans

The horizontal transfer (HT) of the P-element is one of the best documented cases of the HT of a transposable element. The P-element invaded natural D. melanogaster populations between 1950 and 1980 following its HT from Drosophila willistoni, a species endemic to South and Central America. Subsequently, it spread in D. simulans populations between 2006 and 2014, following a HT from D. melanogaster. The geographic region where the spread into D. simulans occurred is unclear, as both involved species are cosmopolitan. The P-element differs between these two species by a single base substitution at site 2040, where D. melanogaster carries a 'G' and D. simulans carries an 'A'. It has been hypothesized that this base substitution was a necessary adaptation that enabled the spread of the P-element in D. simulans, potentially explaining the 30-50-year lag between the invasions of D. melanogaster and D. simulans. To test this hypothesis, we monitored the invasion dynamics of P-elements with both alleles in experimental populations of D. melanogaster and D. simulans. Our results indicate that the allele at site 2040 has a minimal impact on the invasion dynamics of the P-element and, therefore, was not necessary for the invasion of D. simulans. However, we found that the host species significantly influenced the invasion dynamics, with higher P-element copy numbers accumulating in D. melanogaster than in D. simulans. Finally, based on SNPs segregating in natural D. melanogaster populations, we suggest that the horizontal transfer of the P-element from D. melanogaster to D. simulans likely occurred around Tasmania.

The impact of insertion bias into piRNA clusters on the invasion of transposable elements

In our current understanding of transposable element (TE) invasions TEs move freely until they accidentally insert into a piRNA cluster. They are then silenced by the production of piRNA cognate to the TE. Under this model, one would expect that TEs might evolve to avoid piRNA clusters. Yet empirical observations show that some TEs, such as the P-element, insert into piRNA clusters preferentially. We were thus wondering if such a bias could be beneficial for the TE, for example by minimizing harm to the host while still being able to selfishly spread in populations. We decided to model insertion bias to determine if there was ever a situation in which insertion bias was beneficial to the TE. We performed extensive forward simulations of TE invasions with differing insertion biases into piRNA clusters. We found that insertion bias significantly altered the invasion dynamics of TEs, primarily by changing the copy number of the TE in individuals prior to silencing. Insertion into a piRNA cluster reduced the deleterious effects of TEs to the host population, but we found that TEs avoiding piRNA clusters out-compete TEs with a bias towards cluster insertions. Insertion bias was only beneficial to the TE when there was negative selection against TEs and a lack of recombination. Different TEs show different insertion biases into piRNA clusters suggesting they are an attribute of the TE not the host, yet scenarios in which this is beneficial to the TE are quite limited. This opens up an interesting area of future research into the dynamics of insertion bias during TE invasions.

Tolerance thresholds underlie responses to DNA damage during germline development

Selfish DNA modules like transposable elements (TEs) are particularly active in the germline, the lineage that passes genetic information across generations. New TE insertions can disrupt genes and impair the functionality and viability of germ cells. However, we found that in P-M hybrid dysgenesis in Drosophila, a sterility syndrome triggered by the P-element DNA transposon, germ cells harbor unexpectedly few new TE insertions despite accumulating DNA double-strand breaks (DSBs) and inducing cell cycle arrest. Using an engineered CRISPR-Cas9 system, we show that generating DSBs at silenced P-elements or other noncoding sequences is sufficient to induce germ cell loss independently of gene disruption. Indeed, we demonstrate that both developing and adult mitotic germ cells are sensitive to DSBs in a dosage-dependent manner. Following the mitotic-to-meiotic transition, however, germ cells become more tolerant to DSBs, completing oogenesis regardless of the accumulated genome damage. Our findings establish DNA damage tolerance thresholds as crucial safeguards of genome integrity during germline development.

Experimentally evolving Drosophila erecta populations may fail to establish an effective piRNA-based host defense against invading P-elements

To prevent the spread of transposable elements (TEs), hosts have developed sophisticated defense mechanisms. In mammals and invertebrates, a major defense mechanism operates through PIWI-interacting RNAs (piRNAs). To investigate the establishment of the host defense, we introduced the P-element, one of the most widely studied eukaryotic transposons, into naive lines of Drosophila erecta We monitored the invasion in three replicates for more than 50 generations by sequencing the genomic DNA (using short and long reads), the small RNAs, and the transcriptome at regular intervals. A piRNA-based host defense was rapidly established in two replicates (R1, R4) but not in a third (R2), in which P-element copy numbers kept increasing for over 50 generations. We found that the ping-pong cycle could not be activated in R2, although the ping-pong cycle is fully functional against other TEs. Furthermore, R2 had both insertions in piRNA clusters and siRNAs, suggesting that neither of them is sufficient to trigger the host defense. Our work shows that control of an invading TE requires activation of the ping-pong cycle and that this activation is a stochastic event that may fail in some populations, leading to a proliferation of TEs that ultimately threaten the integrity of the host genome.

Spoink, a LTR retrotransposon, invaded D. melanogaster populations in the 1990s

During the last few centuries D. melanogaster populations were invaded by several transposable elements, the most recent of which was thought to be the P-element between 1950 and 1980. Here we describe a novel TE, which we named Spoink, that has invaded D. melanogaster. It is a 5216nt LTR retrotransposon of the Ty3/gypsy superfamily. Relying on strains sampled at different times during the last century we show that Spoink invaded worldwide D. melanogaster populations after the P-element between 1983 and 1993. This invasion was likely triggered by a horizontal transfer from the D. willistoni group, much as the P-element. Spoink is probably silenced by the piRNA pathway in natural populations and about 1/3 of the examined strains have an insertion into a canonical piRNA cluster such as 42AB. Given the degree of genetic investigation of D. melanogaster it is perhaps surprising that Spoink was able to invade unnoticed.

The impact of paramutations on the invasion dynamics of transposable elements

According to the prevailing view, the trap model, the activity of invading transposable elements (TEs) is greatly reduced when a TE copy jumps into a piRNA cluster, which triggers the emergence of piRNAs that silence the TE. One crucial component in the host defence are paramutations. Mediated by maternally deposited piRNAs, paramutations convert TE insertions into piRNA producing loci, thereby transforming selfish TEs into agents of the host defence. Despite this significant effect, the impact of paramutations on the dynamics of TE invasions remains unknown. To address this issue, we performed extensive forward simulations of TE invasions with piRNA clusters and paramutations. We found that paramutations significantly affect TE dynamics, by accelerating the silencing of TE invasions, reducing the number of insertions accumulating during the invasions and mitigating the fitness cost of TEs. We also demonstrate that piRNA production induced by paramutations, an epigenetically inherited trait, may be positively selected. Finally, we show that paramutations may account for three important open problems with the trap model. Firstly, paramutated TE insertions may compensate for the insufficient number of insertions in piRNA clusters observed in previous studies. Secondly, paramutations may explain the discrepancy between the observed and the expected abundance of different TE families in Drosophila melanogaster. Thirdly, piRNA clusters may be crucial to trigger the host defence, but paramutations render the clusters dispensable once the defence has been established. This could account for the lack of TE activation when three major piRNA clusters were deleted in a previous study.

The composition of piRNA clusters in Drosophila melanogaster deviates from expectations under the trap model

BACKGROUND

It is widely assumed that the invasion of a transposable element (TE) in mammals and invertebrates is stopped when a copy of the TE jumps into a piRNA cluster (i.e., the trap model). However, recent works, which for example showed that deletion of three major piRNA clusters has no effect on TE activity, cast doubt on the trap model.

RESULTS

Here, we test the trap model from a population genetics perspective. Our simulations show that the composition of regions that act as transposon traps (i.e., potentially piRNA clusters) ought to deviate from regions that have no effect on TE activity. We investigated TEs in five Drosophila melanogaster strains using three complementary approaches to test whether the composition of piRNA clusters matches these expectations. We found that the abundance of TE families inside and outside of piRNA clusters is highly correlated, although this is not expected under the trap model. Furthermore, the distribution of the number of TE insertions in piRNA clusters is also much broader than expected.

CONCLUSIONS

We found that the observed composition of piRNA clusters is not in agreement with expectations under the simple trap model. Dispersed piRNA producing TE insertions and temporal as well as spatial heterogeneity of piRNA clusters may account for these deviations.

A population genetics theory for piRNA-regulated transposable elements

Transposable elements (TEs) are self-reproducing selfish DNA sequences that can invade the genome of virtually all living species. Population genetics models have shown that TE copy numbers generally reach a limit, either because the transposition rate decreases with the number of copies (transposition regulation) or because TE copies are deleterious, and thus purged by natural selection. Yet, recent empirical discoveries suggest that TE regulation may mostly rely on piRNAs, which require a specific mutational event (the insertion of a TE copy in a piRNA cluster) to be activated - the so-called TE regulation "trap model". We derived new population genetics models accounting for this trap mechanism, and showed that the resulting equilibria differ substantially from previous expectations based on a transposition-selection equilibrium. We proposed three sub-models, depending on whether or not genomic TE copies and piRNA cluster TE copies are selectively neutral or deleterious, and we provide analytical expressions for maximum and equilibrium copy numbers, as well as cluster frequencies for all of them. In the full neutral model, the equilibrium is achieved when transposition is completely silenced, and this equilibrium does not depend on the transposition rate. When genomic TE copies are deleterious but not cluster TE copies, no long-term equilibrium is possible, and active TEs are eventually eliminated after an active incomplete invasion stage. When all TE copies are deleterious, a transposition-selection equilibrium exists, but the invasion dynamics is not monotonic, and the copy number peaks before decreasing. Mathematical predictions were in good agreement with numerical simulations, except when genetic drift and/or linkage disequilibrium dominates. Overall, the trap-model dynamics appeared to be substantially more stochastic and less repeatable than traditional regulation models.