Selfish conflict underlies RNA-mediated parent-of-origin effects

Pliota, Pinelopi; Marvanova, Hana; Koreshova, Alevtina; Kaufman, Yotam; Tikanova, Polina; Krogull, Daniel; Hagmüller, Andreas; Widen, Sonya A.; Handler, Dominik; Gokcezade, Joseph; Duchek, Peter; Brennecke, Julius; Ben-David, Eyal; Burga, Alejandro

doi:10.1038/s41586-024-07155-z

Download PDF

Article
Open access
Published: 06 March 2024

Selfish conflict underlies RNA-mediated parent-of-origin effects

Nature volume 628, pages 122–129 (2024)Cite this article

9284 Accesses
182 Altmetric
Metrics details

Subjects

Abstract

Genomic imprinting—the non-equivalence of maternal and paternal genomes—is a critical process that has evolved independently in many plant and mammalian species^1,2. According to kinship theory, imprinting is the inevitable consequence of conflictive selective forces acting on differentially expressed parental alleles^3,4. Yet, how these epigenetic differences evolve in the first place is poorly understood^3,5,6. Here we report the identification and molecular dissection of a parent-of-origin effect on gene expression that might help to clarify this fundamental question. Toxin-antidote elements (TAs) are selfish elements that spread in populations by poisoning non-carrier individuals^7,8,9. In reciprocal crosses between two Caenorhabditis tropicalis wild isolates, we found that the slow-1/grow-1 TA is specifically inactive when paternally inherited. This parent-of-origin effect stems from transcriptional repression of the slow-1 toxin by the PIWI-interacting RNA (piRNA) host defence pathway. The repression requires PIWI Argonaute and SET-32 histone methyltransferase activities and is transgenerationally inherited via small RNAs. Remarkably, when slow-1/grow-1 is maternally inherited, slow-1 repression is halted by a translation-independent role of its maternal mRNA. That is, slow-1 transcripts loaded into eggs—but not SLOW-1 protein—are necessary and sufficient to counteract piRNA-mediated repression. Our findings show that parent-of-origin effects can evolve by co-option of the piRNA pathway and hinder the spread of selfish genes that require sex for their propagation.

Mating can initiate stable RNA silencing that overcomes epigenetic recovery

Article Open access 09 July 2021

Satellite DNA-mediated diversification of a sex-ratio meiotic drive gene family in Drosophila

Article 06 September 2021

Unistrand piRNA clusters are an evolutionarily conserved mechanism to suppress endogenous retroviruses across the Drosophila genus

Article Open access 13 November 2023

Main

Diploid organisms carry two copies of each gene: one inherited from their mother and the other one from their father. Typically, these copies are functionally interchangeable. Imprinted genes are the exception to this rule. They keep an epigenetic memory of their gametic provenance, making maternal and paternal genomes non-equivalent, which has a large effect on embryonic development, species hybridization and human disease¹⁰. Multiple theories have been put forward to explain the evolution of imprinting. The most accepted theory—kinship conflict—states that imprinting arises when there are conflicting interests between maternal and paternal genomes owing to differential investment in their offspring^3,4. Notably, this theory presupposes the existence of mechanisms that establish differences in the expression of maternal and paternal alleles—otherwise, there would be nothing to select on³. This raises the critical question of how parent-of-origin effects on gene expression evolve in the first place.

The discovery of the first imprinted loci in mammals led to the hypothesis that imprinting evolved from host defence mechanisms that use DNA methylation to keep viruses and parasitic genes at bay^11,12. This is in line with the close proximity of many imprinted loci to transposable elements in plants^13,14 and piRNA-induced DNA methylation of a retrotransposon being critical for the paternal imprinting of mouse Rasgrf1 (ref. ¹⁵). However, the evolutionary origins of imprinting remain poorly understood at the molecular level. More recently, histone modifications, such as H3K27me3, have been reported to act as imprinting marks independently of DNA methylation in mice¹⁶. These observations have raised the possibility that a link between parent-of-origin-dependent gene expression and host defence mechanisms can also be found in organisms that lack DNA methylation but are rich in small regulatory RNAs, such as Caenorhabditis elegans and related nematodes¹⁷. Here we dissect the mechanism behind a parent-of-origin effect on gene expression and provide a physiological context for the emergence of imprinting.

A TA with a parent-of-origin effect

C. tropicalis is a hermaphroditic nematode that—unlike its more widely distributed relative C. elegans—inhabits exclusively equatorial regions¹⁸. While studying genetic incompatibilities between the two C. tropicalis wild isolates NIC203 (Guadeloupe, France) and EG6180 (Puerto Rico, USA), we uncovered a maternal-effect TA, which we named slow-1/grow-1 (ref. ⁹). This selfish element is located in NIC203 chromosome III and comprises three tightly linked genes: a maternally expressed toxin, slow-1, and two identical and redundant antidotes, grow-1.1 and grow-1.2, which are expressed zygotically. For simplicity, we will refer to the two antidotes collectively as grow-1 unless specifically noted (Extended Data Fig. 1a and Supplementary Discussion). Slow-1 transcripts are maternally loaded into eggs prior to fertilization and remain stable in embryos, at least until the 20-cell stage. However, from the comma stage until hatching, slow-1 transcripts are found only in the germline precursor cells⁹. SLOW-1 is homologous to nuclear hormone receptors, whereas the antidote GROW-1 has no homology to known proteins. In crosses between TA carrier and non-carrier strains, heterozygous mothers poison all their eggs but only progeny that inherit the TA can counteract the toxin by zygotically expressing its antidote (Extended Data Fig. 1b). Whereas wild-type worms typically take two days to develop from the L1 stage to the onset of egg laying, embryos poisoned by maternal SLOW-1 take on average four days. This developmental delay imposes a high fitness cost and favours the spread of the selfish element in the population⁹.

To study the inheritance of slow-1/grow-1 TA, we previously generated a near-isogenic line strain (hereafter referred to as ‘NIL’) containing the slow-1/grow-1 NIC203 chromosome III locus in an otherwise EG6180 background⁹. As expected, slow-1 mRNA was detected in the NIL but not in EG6180 (Extended Data Fig. 1c). As previously reported, in crosses between NIL hermaphrodites and EG6180 males, the toxin induced developmental delay in all the F₂ homozygous non-carrier (EG/EG) individuals⁹ (100% delay, n = 34; Fig. 1a). However, we noticed an unexpected pattern of inheritance when performing the reciprocal cross. If EG6180 hermaphrodites were mated to slow-1/grow-1 NIL males, most of their F₂ EG/EG progeny were not developmentally delayed but phenotypically wild type (9.4% delay, n = 53; P ≤ 0.0001; Fig. 1a,b). This was surprising, because known TAs—including C. elegans peel-1/zeel-1 and sup-35/pha-1, the Medea locus in Tribolium, and the mouse homogeneously staining region (HSR) locus—affect non-carrier individuals regardless of whether the element is inherited from the maternal or paternal lineage^8,19,20,21 (Extended Data Fig. 1d).

**Fig. 1: *slow-1/grow-1*, a selfish element with a parent-of-origin effect.**

We also investigated the inheritance pattern of two recently discovered maternal-effect TAs in C. tropicalis and C. briggsae that cause developmental delay^9,22. However, we found no evidence of a parent-of-origin effect, indicating that this is not a general feature of non-lethal toxins (Fig. 1b and Extended Data Fig. 1e). Mito-nuclear incompatibilities could not explain the observed pattern because both parental lines carry the same mito-genotype (Extended Data Fig. 1f). Moreover, C. tropicalis, like all nematodes of the Rhabditida group, lacks de novo methyltransferases, making the involvement of mammalian-like epigenetic imprinting unlikely²³. Because parent-of-origin effects are extremely rare in nematodes and all reported cases involve transgenic reporters, we set out to investigate this phenomenon^24,25.

Reduced dosage of the SLOW-1 toxin

Maternally expressed slow-1 causes the slow-1/grow-1 TA delay phenotype. Thus, we reasoned that the parent-of-origin effect could stem from reduced expression of the paternally inherited toxin in the germline of F₁ heterozygous mothers. To test this idea, we performed reciprocal crosses between EG6180 and the slow-1/grow-1 NIL strains, followed by RNA sequencing (RNA-seq) of F₁ heterozygous young adult hermaphrodites. In agreement with our hypothesis, slow-1 mRNA levels were significantly lower in F₁ mothers when slow-1/grow-1 was paternally inherited (2.4-fold decrease, P = 0.0092; Fig. 1c). The slow-1 parent-of-origin effect was not exclusive to the recombinant NIL strain, as we observed the same difference in slow-1 gene expression when performing reciprocal crosses between NIC203 and EG6180 parental strains (Extended Data Fig. 1g).

To independently validate the parent-of-origin effect on gene expression at the protein level, we first tagged the endogenous slow-1 locus with mScarlet on its N terminus. In agreement with its maternal-effect, SLOW-1 was present in the gonads of hermaphrodites and loaded into eggs prior to fertilization (Extended Data Fig. 1h,i). Next, we performed reciprocal crosses between mScarlet::slow-1 in the NIL background and EG6180 strains and quantified the fluorescence signal in the germline of their F₁ progeny. In agreement with both our genetic crosses and RNA-seq experiments (Fig. 1b,c), when slow-1/grow-1 was paternally inherited, SLOW-1 protein levels were significantly lower in the germline of F₁ individuals (Fig. 2a,b), as well as in F₂ 2-cell stage embryos (Fig. 2a,c). To test whether the SLOW-1 dosage correlated with the severity of the phenotype, we impaired the antidote function in the parental NIL strain, which expresses twice as much slow-1 mRNA compared to heterozygous worms (Extended Data Fig. 1a). We found that grow-1.1(+/−); grow-1.2(−/−) worms were viable but we could not retrieve any viable grow-1.1(−/−); grow-1.2(−/−) individuals among their progeny. The double homozygous mutants arrested as larvae and died before laying eggs, indicating that slow-1 is dosage-sensitive (Extended Data Fig. 1a). These results show that the lack of activity of slow-1/grow-1 following its paternal inheritance stems from a reduction in slow-1 mRNA levels in the germline of F₁ hermaphrodites and, consequently, a reduced dosage of the toxin in F₂ embryos.

**Fig. 2: Paternal inheritance of *slow-1/grow-1* leads to decreased SLOW-1 dosage.**

slow-1 is transgenerationally repressed

In C. elegans, silencing of transgenes can result in the inheritance of the repressed state for multiple generations^26,27. Typically, this transgenerational effect is mediated by small RNAs (sRNAs) in response to external or internal cues. To test whether sRNAs could underlie the impaired expression of the paternally inherited slow-1/grow-1 allele, we explored whether the inheritance of slow-1/grow-1 could compromise its toxicity in subsequent generations. To test this, we first crossed EG6180 hermaphrodites to NIL males and singled their F₂ progeny. Then, we identified F₂ homozygous slow-1/grow-1 hermaphrodites, allowed them to self-fertilize, collected their progeny (F₃), and crossed them back to EG6180 males. Finally, we collected the F₄ heterozygous offspring, allowed them to self-fertilize, and inspected their progeny (F₅) (Fig. 3a). In this way, the impaired slow-1/grow-1 allele was reintroduced into the maternal lineage, which enabled us to probe whether slow-1 could delay its progeny once again. We found that 97% (n = 34) of F₅ homozygous slow-1/grow-1 (EG/EG) individuals were phenotypically wild type, indicating that slow-1/grow-1 activity was largely impaired 3 generations after paternal inheritance (Fig. 3b). Additional crosses revealed that slow-1/grow-1 regained its activity 9 generations after paternal inheritance (22.2% (n = 27) of EG/EG individuals were phenotypically wild type), indicating that the slow-1 repressed state can be spontaneously reversed^26,28 (Fig. 3b).

**Fig. 3: *slow-1* is targeted by the piRNA pathway.**

piRNAs target slow-1

Since the transgenerational repression of slow-1/grow-1 does not stem from an external trigger, we reasoned that endogenous piRNAs could mediate this effect. PRG-1, the C. elegans orthologue of Drosophila PIWI-clade proteins, binds piRNAs and is essential for their function^29,30. To study the role of PIWI and other Argonaute proteins in slow-1 repression, we first identified homologues and built a comprehensive Argonaute phylogeny (Extended Data Fig. 2). The C. tropicalis genome encodes two PRG-1 orthologues on chromosome I, which we named PRG-1.1 and PRG-1.2, both of which are maternally loaded into eggs (Fig. 3c). They share 87.7% protein sequence identity and are probably the result of a recent gene duplication event (Extended Data Figs. 2 and 3a–c). To test whether the repression of slow-1/grow-1 was dependent on piRNA activity, we generated prg-1.1 and prg-1.2 null alleles in an EG6180 background. Both prg-1.1 and prg-1.2 mutant lines were viable and did not show any obvious signs of developmental delay or larval arrest (0%, n = 118 and 0%, n = 95, respectively). However, prg-1.1;prg-1.2 double mutants were fully sterile indicating significant redundancy between these two genes (Extended Data Fig. 3d).

Next, we set up crosses between EG6180 hermaphrodites and NIL males, in which both parents carried null alleles of either prg-1.1 or prg-1.2. Loss of prg-1.2 impaired slow-1/grow-1 repression when inherited through the paternal lineage: 69.2% (n = 78) of F₂ homozygous EG/EG individuals were developmentally delayed (Fig. 3d). By contrast, loss of prg-1.1 had only a minor effect on the activity of the TA (6.25% of EG/EG were delayed, n = 16) (Fig. 3d). Additional crosses revealed that maternally provisioned prg-1.2 is necessary for slow-1 repression (Extended Data Fig. 3e) and that prg-1.2 is necessary for the initiation but not the maintenance of the repression^31,32 (Extended Data Fig. 3f).

To identify the specific piRNAs responsible for slow-1 repression, we first annotated 27,445 piRNAs in C. tropicalis with a mean abundance of 0.1 ppm or higher (Methods and Supplementary Data 1). As in C. elegans and C. briggsae, piRNAs were found almost exclusively in chromosome IV³³ (96.9%; Fig. 3e). Next, we leveraged known targeting rules, predicted piRNA-target binding energies and overall complementarity to define a list of top candidates (Methods and Supplementary Data 2). We observed that two of our top piRNA candidate loci were in tight genetic linkage: Ctr-21ur-06949 and Ctr-21ur-06917 were only 4.6 kb apart on chromosome IV and both were predicted to bind the 3′ untranslated region (UTR) of slow-1 (Fig. 3f). To test their role in slow-1 repression, we deleted these piRNAs and performed paternal crosses. Of note, whereas deletion of individual piRNAs had no effect on slow-1/grow-1 repression, simultaneous loss of both piRNAs hindered slow-1/grow-1 repression, phenocopying the prg-1.2 loss of function mutation (Fig. 3d,f). As a control, only background levels of delay were observed in the parental single and double piRNA mutant strains (Ctr-21ur-06949(Δ): 2%, n = 100; Ctr-21ur-06917(Δ): 1.25%, n = 80; Ctr-21ur-06949(Δ); Ctr-21ur-06917(Δ): 2.2%, n = 180). These results show that piRNAs repress slow-1 following its paternal inheritance and that their activity is epistatic.

PRG-1s are redundant but non-equivalent

Given the epistatic nature of the slow-1 piRNA-mediated repression (Fig. 3f) and the synthetic lethality observed in prg-1.1 and prg-1.2 double mutants (Extended Data Fig. 3d), we hypothesized that any role of prg-1.1 in the repression of slow-1 might be masked by genetic redundancy. To test this idea, we generated triple mutant worms carrying the prg-1.1 mutant allele along with the double piRNA deletion and performed paternal inheritance crosses of the TA. In contrast to the partial de-repression of slow-1/grow-1 observed in either prg-1.2 or double piRNA mutants (Fig. 3f), de-repression of the TA was almost complete when prg-1.1 and the two piRNAs were mutated (Fig. 3f). As a control, no developmental defects were observed in the triple mutant parental line (0%, n = 89). Immunoprecipitation of PRG-1.1 and PRG-1.2 followed by sRNA sequencing (sRNA-seq) revealed that these Argonautes bind at large the same piRNA population—including both Ctr-21ur-06949 and Ctr-21ur-06917 (Extended Data Fig. 4 and Supplementary Data 2). However, their binding preference are not entirely equivalent, probably contributing to their differential effects on slow-1 repression (Extended Data Fig. 4 and Supplementary Note 1).

slow-1 is epigenetically repressed

In C. elegans, piRNAs trigger the production of secondary 22G-RNAs that are complementary to the target mRNA. These 22G-RNAs are bound by nuclear Argonautes HRDE-1 and WAGO-10, which in turn recruit chromatin-modifying enzymes to the target locus and mediate its epigenetic repression^34,35 (Extended Data Fig. 5a). Two putative histone methyltransferases, SET-25 (H3K9me3) and SET-32 (H3K23me3), have a crucial role in this process^35,36. To test whether effectors of the piRNA pathway mediate slow-1 repression, we generated putative null alleles of several known factors (Fig. 3g and Extended Data Fig. 5a,d). Ctr-hrde-1 (the closest homologue of hrde-1 and wago-10 in C. elegans), Ctr-simr-1 and Ctr-mut-16 were essential for fertility, preventing further characterization (Extended Data Fig. 5a–d). However, Ctr-rrf-1, Ctr-wago-15 (a close paralogue of Ctr-hrde-1), Ctr-set-25 and Ctr-set-32 mutants were viable and fertile. We set up paternal crosses using these four mutants and found that loss of Ctr-set-32 was sufficient to impair slow-1/grow-1 repression when inherited through the paternal lineage, phenocopying loss of piRNAs (Fig. 3g and Extended Data Fig. 5e). The involvement of a histone methyltransferase in the parent-of-origin effect strongly suggested that slow-1 repression occurs at the transcriptional level. In agreement with this model, we performed quantitative PCR with reverse transcription (RT–qPCR) and found that both slow-1 mRNA and pre-mRNA levels were markedly reduced in the F₁ generation following paternal inheritance of the TA (15.2% and 20.2% of levels following maternal inheritance, respectively; Fig. 3h).

Next, we asked whether paternal inheritance of slow-1/grow-1 leads to the accumulation of 22G-RNAs targeting slow-1. To test this, we leveraged the transgenerational inheritance of the slow-1/grow-1 repressed state (Fig. 3a). First, we carried a slow-1/grow-1 paternal cross, we then isolated F₂ homozygous TA carriers, propagated and expanded the population for two generations, and finally sequenced the sRNA pool of the F₄ young hermaphrodites (Extended Data Fig. 5f). Paternal inheritance of slow-1/grow-1 resulted in a marked 33.7-fold up-regulation of 22G-RNAs complementary to slow-1 compared to the control line, in which the TA is active (Fig. 3i,j; P = 0.0013). We observed a local peak of 22G-RNA biogenesis within the predicted piRNA recognition sites (Extended Data Fig. 5g); however, most 22G-RNAs were derived from the 5′ of the transcript (Fig. 3i). We also identified 26G-RNAs complementary to slow-1 in the repressed state. These 26G-RNAs were significantly less abundant than 22G-RNAs but were almost completely absent from control samples (Fig. 3j). Since 22G-RNAs were readily detectable in the great-great-granddaughters of the original male TA carriers and most of them were not derived from predicted piRNA binding sites (Fig. 3i), our results suggest that these sRNAs mediate the inheritance of the slow-1 epigenetic state³⁷. Next, given that set-25 and set-32 are jointly required for the deposition of H3K9me3 in C. elegans³⁸, we investigated whether paternal inheritance of the TA could lead to the accumulation of this repressive histone mark in slow-1 (refs. ^32,39). To test this, we performed H3K9me3 chromatin immunoprecipitation with sequencing (ChIP–seq) in F₄ individuals following paternal inheritance of the TA, as well as in the parental NIL control. We did not observe any significant H3K9me3 enrichment in slow-1, even though we detected H3K9me3 enrichment in other loci (Fig. 3k). Although we cannot rule out potential limitations of our assay such as dilution of the germline signal or unspecific binding of the antibody, this result suggests that H3K9me3 may not be required for the maintenance of silencing, in line with recent findings^40,41,42.

Maternal slow-1 mRNAs counter piRNAs

The repressive action of piRNAs accounts for the low levels of slow-1 following paternal inheritance; however, piRNAs alone cannot explain the parent-of-origin effect. Thus, we investigated whether mechanisms that are known to facilitate the expression of genes in the C. elegans germline—periodic 10-bp motif of A_n/T_n clusters (PATCs) and CSR-1—might prevent slow-1 repression. PATCs are typically found in the introns of germline-expressed genes and can promote the expression of transgenes in the germline^43,44. However, slow-1 introns exhibited very low PATC scores (Extended Data Fig. 6a). CSR-1 is the only Argonaute that can activate transgenes silenced by piRNAs; however, loss of maternal CSR-1 did not impair slow-1 activity, suggesting that CSR-1 is not responsible for the parent-of-origin effect (Extended Data Fig. 6 and Supplementary Note 2).

While performing reciprocal crosses between the wild-type NIL and a slow-1(−)/grow-1(−) double mutant NIL strain, we made an intriguing observation. Analogous to crosses between NIL hermaphrodites and EG6180 males (Fig. 1b), when wild-type NIL hermaphrodites were mated to the double mutant males in which both toxin and antidote carry null frameshift mutations, 28.9% (n = 190) of the F₂ progeny were delayed and all homozygous double mutant individuals were delayed (100%, n = 31; Fig. 4a). However, we observed the same inheritance pattern in the reciprocal cross. 22.1% (n = 140) of the F₂ progeny were delayed and delayed individuals were homozygous double mutants (95.8%; n = 24), indicating that slow-1/grow-1 was fully active when inherited via the paternal lineage (Fig. 4a). These results indicated that the slow-1/grow-1 double mutant and EG6180 haplotypes were not equivalent, and that maternal inheritance of a null slow-1 allele could somehow prevent its piRNA-mediated repression. Furthermore, given that slow-1 was able to protect the paternal allele from repression despite carrying a frameshift null mutation, we hypothesized that slow-1 mRNA, but not SLOW-1 protein, is necessary for this phenomenon.

**Fig. 4: Maternal *slow-1* transcripts inhibit piRNA-mediated repression.**

To test this hypothesis, we used CRISPR–Cas9 to delete the full coding region of slow-1 in an otherwise identical genetic background to the double mutant NIL strain carrying slow-1(−)/grow-1(−). In contrast to the frameshift allele, the deletion allele (slow-1Δ) removes the entirety of the slow-1 transcript. Then, we performed reciprocal crosses between the slow-1(Δ)/grow-1(−) NIL strain and the wild-type NIL and inspected their F₂ progeny (Fig. 4b). When NIL hermaphrodites were crossed to slow-1(Δ)/grow-1(−) NIL males, we observed 26.8% (n = 190) delay among the F₂ offspring, whereas all genotyped worms homozygous for the mutant allele were delayed (100%, n = 18; Fig. 4b). By contrast, when slow-1(Δ)/grow-1(−) NIL hermaphrodites were crossed to NIL males, we observed baseline delay among their F₂ progeny (3.8%, n = 129) (Fig. 4b). These results indicate that the slow-1(−) allele but not slow-1(Δ) is able to protect a paternally inherited slow-1/grow-1 TA from piRNA repression and identify slow-1 mRNA as the ‘licensing’ signal.

To test whether slow-1 mRNA is sufficient for licensing, we transcribed slow-1 RNA in vitro and injected it into the gonads of 16 slow-1(Δ)/grow-1(−) NIL hermaphrodites, mated those to slow-1/grow-1 NIL males, and inspected their F₂ progeny. Critically, we mutated the start codon of the slow-1 cDNA that served as a transcription template, resulting in RNA that cannot be translated into SLOW-1 protein. Following injection of noncoding slow-1 RNA into the gonads of slow-1(Δ)/grow-1(−) mothers, there was a significant increase in the proportion of delayed individuals among the F₂ compared to a control injection (13.8% delayed, n = 650 and 4% delayed, n = 296 respectively, P < 0.0001; Fig. 4c). Most (84.7%, n = 61) of genotyped delayed individuals were homozygous for the slow-1(Δ)/grow-1(−) allele, showing that the effect was highly specific. Overall, injection of slow-1 RNA increased the proportion of delayed F₂ individuals among double mutants from 9.6% (n = 62) in the control cross to 47.6% (n = 128) in the RNA-injected animals (P < 0.0001; Fig. 4f). The partial rescue of zygotic slow-1 expression probably reflects technical limitations in the injection protocol, as we observed a wide range of rescue depending on the injected mother (Extended Data Fig. 7a). These results show that slow-1 RNA is sufficient to license a paternally inherited slow-1/grow-1 allele and that this effect does not depend on SLOW-1 protein. Epigenetic licensing by maternal transcripts has only been described for one gene to date: the C. elegans sex-determining gene fem-1 (ref. ⁴⁵) (Supplementary Discussion). Because licensing could be a common mechanism in nematodes and offers a physiological framework to better understand other epigenetic phenomena, we sought to study its requirements^25,46

Molecular requirements of licensing

First, we studied the effect of maternal slow-1 dosage on licensing. To do so, we deleted 620 bp upstream of the slow-1 coding region and then proceeded to knock out the antidote grow-1. RNA-seq of the resulting promoter deletion strain revealed a 176-fold decrease in slow-1 mRNA levels, whereas neighbouring genes were unaffected (Extended Data Fig. 7b). We observed only limited abnormal phenotypes in the double mutants, even though they lacked the grow-1 antidote (8.71% delay, n = 70), suggesting that the amount of SLOW-1 toxin made in the promoter deletion line was insufficient to poison embryos. However, when we crossed slow-1(Δprom)/grow-1(−) hermaphrodites to NIL males, the paternal allele was fully active: 27.7% of F₂ individuals were delayed (Extended Data Fig. 7c). These results indicate that a 176-fold reduction in slow-1 maternal mRNA abundance abolishes its toxicity but not its licensing activity, suggesting that licensing does not rely on a sponge-like mechanism but probably involves a catalytic step (Extended Data Fig. 7c).

Sequence similarity between slow-1 maternal transcripts and their zygotic counterparts is probably key for the establishment of licensing. To explore whether this requirement is an intrinsic property of slow-1 or a general feature, we asked whether sequence similarity to a foreign sequence could also license slow-1. To do so, we took advantage of the mScarlet::slow-1 fusion strain (Fig. 2a). Importantly, tagging of SLOW-1 with an N-terminal mScarlet reporter did not interfere with its toxicity or the parent-of-origin effect (Fig. 4d). To emulate the licensing signal, we first generated a strain carrying mScarlet in a germline-permissive site (chromosome V) in an otherwise EG6180 background. We then crossed hermaphrodites expressing maternal mScarlet to mScarlet::slow-1 males and scored their F₂ progeny. Notably, maternal mScarlet transcripts fully licensed endogenous tagged mScarlet::slow-1 (Fig. 4d). These results indicate that sequence similarity to a foreign maternal transcript is sufficient for epigenetic licensing. Moreover, they suggest that the licensing signal can spread through the zygotic transcript, as maternal mScarlet countered piRNAs targeting slow-1 despite the lack of sequence similarity between the two genes.

Finally, we set out to investigate at what step of the piRNA pathway licensing countered repression: target recognition or transcriptional silencing. Target recognition depends on complementarity to the mature mRNA, whereas transcriptional silencing relies on complementarity to the nascent transcript, which guides the repression machinery to the target locus (Fig. 4e). We reasoned that we could distinguish between these possibilities by testing whether maternal mScarlet could license slow-1 in the context of a polycistronic operon⁴⁷. To do this, we inserted the 256-bp intergenic region from the C. tropicalis gpd-2::gdp-3 operon in between mScarlet and slow-1 using CRISPR–Cas9. This intergenic sequence (hereafter termed SL2) contains the 3′ acceptor site for the SL2 RNA trans-splicing leader⁴⁸. The resulting operon, mScarlet::SL2::slow-1, is under the control of the native slow-1 promoter—mScarlet and slow-1 are transcribed as a single polycistronic pre-mRNA in the germline and later trans-spliced into two independent mRNAs (Fig. 4f). As expected, we detected mScarlet in the germline of these worms and their early embryos (Extended Data Fig. 7d).

To validate our approach, we performed reciprocal crosses between mScarlet::SL2::slow-1 worms and the EG6180 parental strain and found that slow-1 was active only when maternally inherited, indicating that the operon architecture did not interfere with the parent-of-origin effect (Fig. 4g). Furthermore, lack of maternal slow-1 led to co-repression of mScarlet when the operon was paternally inherited, in agreement with silencing being guided by the nascent transcript (Extended Data Fig. 7e). We then crossed hermaphrodites expressing maternal mScarlet mRNA to males carrying the mScarlet::SL2::slow-1 operon and scored their F₂ progeny. We observed no delayed EG/EG F₂ individuals, indicating that homology to mScarlet was not sufficient to license slow-1, despite being part of the same pre-mRNA molecule. Given that maternal mScarlet mRNA efficiently licensed slow-1 when both genes were part of a monocistronic transcript, our results indicate that zygotic slow-1 is licensed post-transcriptionally. For instance, licensing could hinder the binding of piRNAs to their target or the subsequent amplification of 22G-RNAs in the perinuclear nuage. One implication of this model is that licensing should be incapable of countering transcriptional silencing mediated by pre-existing 22G-RNAs. Supporting this idea, maternal slow-1 transcripts originating from a repressed allele lost their ability to license a naïve paternal allele (Extended Data Fig. 7f), presumably because repressive 22G-RNAs that are loaded into eggs alongside maternal transcripts⁴⁹ can effectively by-pass the licensing signal.

From piRNAs to parent-of-origin effects

Haig’s kinship theory explains why natural selection favours different levels of expression of maternally and paternally inherited alleles. However, it does not address how these epigenetic differences evolve in the first place. Here we show that in the nematode C. tropicalis, parent-specific expression originates by co-option of the piRNA pathway, which in worms is essential to distinguish self from non-self^32,35,50 (Fig. 5). Similar to classical imprinting, slow-1 expression levels depend on whether the TA is maternally or paternally inherited. However, there are two important differences: (1) the slow-1 parent-of-origin effect is not acquired by gametic identity but specifically triggered by outcrossing; and (2) imprinted loci reset in the germline every generation, whereas slow-1 repression resets only after multiple generations of selfing. We propose that this parent-of-origin effect could represent an intermediate evolutionary state, which we refer to as proto-imprinting.

**Fig. 5: Model illustrating the *slow-1/grow-1* parent-of-origin effect.**

Our results also indicate that parent-of-origin effects could provide a selective advantage to the host. Repression of slow-1 following paternal inheritance of the TA hinders its gene drive activity for multiple generations and decreases the incidence of intraspecific genetic incompatibilities. Remarkably, an evolutionary related but highly divergent TA, slow-2/grow-2, does not show a parent-of-origin effect, suggesting that this trait can evolve quickly in nature (Extended Data Figs. 8 and 9 and Supplementary Note 3). Because TAs and analogous maternal-zygotic lethal factors are not only present in nematodes but also segregate in wild insect, plant, and mouse populations^{7,20,21,51,52}, we propose that co-option of sRNA-mediated defence systems originating from selfish conflict might be a recurrent event facilitating the evolution of imprinting.

Methods

Maintenance of worm strains

Nematodes were grown on modified nematode growth medium (NGM) plates with 1% agar/0.7% agarose to prevent C. tropicalis burrowing. Experiments were conducted at either 25 °C (C. tropicalis) or 20 °C (C. elegans). csr-1(+/−) strains were cultured on 6-cm NGM plates supplemented with 500 μl of G418 (25 mg ml⁻¹) for selecting heterozygous null individuals. Supplementary Table 2 lists all study strains, some of which were provided by the Caenorhabditis Genetics Centre, funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).

Phenotyping and genotyping of crosses

For crosses, 4–5 L4 hermaphrodites were mated with 30–40 males in a 12-well plate with modified NGM. After 2 days, 10 L4 F₁ progeny were transferred to separate plates, genotyped by PCR, and at least 10 embryos per F₁ hermaphrodite were singled into 6-cm NGM plates. Each F2 individual was visually inspected daily for up to 7 days, classified for developmental stage, and any phenotypic abnormalities. Embryonic lethality, arrested development, and delayed reproduction were assessed. Sterility was noted for adults not producing progeny. After 7 days, worms were lysed and genotyped. A list of primers used for genotyping can be found in Supplementary Table 3. Crosses involving csr-1(−); slow-1/grow-1 hermaphrodites vs EG6180 males or injected hermaphrodites vs NIL males were selected based on a pmyo-2::mScarlet reporter.

Generation of C. tropicalis transgenic lines

For CRISPR–Cas gene editing, we adapted previous protocols⁵³. In brief, 250 ng µl⁻¹ Cas9 or Cas12a proteins were incubated with 200 ng µl⁻¹ CRISPR RNA (crRNA) and 333 ng µl⁻¹ trans-activating crRNA (tracrRNA) before adding 2.5 ng µl⁻¹ co-injection marker plasmid (pCFJ90-mScarlet-I). For HDR, donor oligos (IDT) or biotinylated and melted PCR products were added at a final concentration of 200 ng µl⁻¹ or 100 ng µl⁻¹, respectively. Following injections into young hermaphrodites, mScarlet-positive F1 were singled, and their offspring screened by PCR and Sanger sequencing to detect successful editing. To clone the mScarlet::SLOW-1 donor, we added ~300-bp homology arms amplified from QX2345 genomic DNA to mScarlet-I (from pMS050) in pBluescript via Gibson assembly. Because csr-1 is essential for viability in C. elegans, we first devised a strategy to stably propagate a csr-1 heterozygous line in the absence of classical genetic balancers. To do so, we used CRISPR–Cas9 to introduce a premature stop mutation in the endogenous csr-1 locus followed by a neoR cassette, which confers resistance to the G418 antibiotic (Extended Data Fig. 6d). For the csr-1::neoR donor, we first replaced the C. elegans rps-27 promoter and unc-54 3′ UTR in pCFJ910 with 500 bp upstream and 250 bp downstream of the C. tropicalis rps-20 gene. This rps-20::neoR cassette was then flanked with ~550-bp homology arms amplified from EG6180 worms and inserted into pBluescript. Correct targeting introduces a stop codon after residue L337 of CSR-1 followed by a ubiquitously expressed neomycin resistance. We propagated the mutant line in plates containing G418 and thus actively selecting for heterozygous csr-1(−) null individuals. Upon drug removal, most homozygous csr-1(−) individuals derived from heterozygous mothers developed into adulthood but were either sterile or laid mostly dead embryos. However, a small fraction of null mutants was partially fertile and homozygous csr-1(−) lines could be stably propagated for multiple generations despite extensive embryonic lethality in the population (Extended Data Fig. 6d). All gRNAs and HDR templates are available on Supplementary Tables 4 and 5.

In vitro RNA transcription and injection

The slow-1 cDNA was cloned into pGEM-T Easy (Promega, A1360), with a 5′ T7 RNA polymerase site and the start codon mutated RNA-only transcription (ATG>TTG). The plasmid was digested with NotI to release the insert (NEB, R0189), which was subsequently purified by gel-extraction and used as template for RNA synthesis. RNA was prepared using the HiScribe T7 Quick High Yield kit (NEB, E2050) with the following modifications: addition of 3 µl of 10 mM DTT and 1 µl of RNaseOUT (Thermo, 10777019). After overnight transcription, the reaction was diluted, treated with RNase-free DNase I (NEB, M0303S), bead-purified (Vienna Biocenter MBS 5001111, High Performance RNA Bead Isolation), quantified (Thermo, Q32852), and stored at −80 °C. Injections were repeated twice using independently transcribed RNA at concentrations: 150 nM and 400 nM yielding identical results.

Reciprocal crosses with the mScarlet::slow-1 reporter line

To assess SLOW-1 expression in F₁ progeny from reciprocal crosses between mScarlet::SLOW-1 NIL and EG6180 strains, we conducted 2 sets of crosses: (1) SLOW-1::mScarlet dpy (INK461) hermaphrodites to EG6180 males for maternal inheritance; and (2) EG6180 dpy (QX2355) hermaphrodites to mScarlet::SLOW-1 NIL males (INK459) for paternal inheritance. Wild-type young adult F₁ progeny were immobilized in NemaGel on a glass slide and imaged using an Axio Imager.Z2 (Carl Zeiss) widefield microscope with a Hamamatsu Orca Flash 4 camera, (excitation 545/30 nm filter). The analysis was performed in FIJI, by tracing the germline in the DIC channel and measuring mean fluorescence, including gut autofluorescence.

Sequencing and genome assembly of EG6180

We extracted high molecular weight genomic DNA using the Masterpure Complete DNA and RNA purification kit (tissue sample protocol, Lucigen). We prepared 8 kb, 20 kb and unfragmented sequencing libraries using the 1D Ligation Sequencing Kit (Oxford Nanopore SQK-LSK109). The 8 kb fragmentation was done using g-TUBE (Covaris). Library was loaded on a MinION MK1B device (Oxford Nanopore). Read calling was done using MinKNOW software. We performed a hybrid assembly, incorporating Illumina sequencing reads of EG6180 with some modifications as detailed below⁹. We used assembled Illumina reads to correct raw Nanopore reads, which were assembled using Flye Assembler⁵⁴. The preliminary assembly included 119 contigs in 107 scaffolds (Scaffold N50 was 1,489,504 bp). We derived synteny blocks between the provisional assembly and our chromosome-level NIC203 assembly using Sibelia⁵⁵ and used the synteny blocks to scaffold the contigs to chromosome level using Ragout⁵⁶.

Identification of C. tropicalis Argonaute proteins and piRNA pathway effectors

We annotated functional domains in C. tropicalis NIC203 using Interproscan 5 as part of our previous NIC203 genome assembly⁹. We identified Argonaute proteins with PFAM domains, including Piwi (PF02171), PAZ (PF02170), N-terminal domain of Argonaute (PF16486), Argonaute linker 1 (PF08699), Mid domain of Argonaute (PF16487) and Argonaute linker 2 (PF16488) domains. We excluded a protein with low molecular weight (41 kDa) as unlikely to be an Argonaute and the orthologue of C. elegans Dicer that represented an outgroup to the rest of the proteins. After aligning those sequences to C. elegans Argonautes identified in a previous study⁵⁷ using Clustal Omega we conducted phylogenetic analysis using iqtree2 (ref. ⁵⁸), with 1,000 replicates of the approximate likelihood-ratio test (--alrt 1000) and 1,000 boostraps (-b 1000). iqtree2 carries out an initial model selection step, and a substitution model with the general Q matrix, empirical codon frequencies, a proportion of invariable sites and a free rate heterogeneity (Q.pfam+F + I + R4) was selected. Additional orthologues of C. elegans piRNA effector genes were identified through reciprocal blastp searches, synteny conservation, and gene trees from Wormbase Parasite⁵⁹. C. elegans mut-16, rrf-1, and simr-1 have 1:1 orthologues in C. tropicalis. The evolutionary history of SET proteins is complex due to their propensity to gain and lose paralogues within Caenorhabditis. The gene annotated gene as C. tropicalis set-25, is the closest among six paralogues in its genome. Thus, the absence of a phenotype in the mutant may be attributed to genetic redundancy. The gene annotated as C. tropicalis set-32 is a close orthologue of two C. elegans genes: set-21 and set-32. The SET domains of C.tr-SET-32 and C.el-SET-32 are ~48% identical at the protein level. Additionally, using Alphafold2 (ref. ⁶⁰) we found that these two proteins have high structural similarity (root mean square deviation = 0.962) and using the predicted structure of C.tr-SET-32 as a query retrieved C.el-SET-32 as the top hit in C. elegans (Foldseek)⁶¹.

Transgenerational silencing of slow-1/grow-1

In the transgenerational inheritance experiments, EG6180 hermaphrodites were crossed to NIL (QX2345) males. F₁ individuals were genotyped after laying embryos to distinguish between self-progeny from cross-progeny. F₂ embryos from cross-progeny mothers were singled, allowed to lay eggs and genotyped. F₃ homozygous carriers for slow-1/grow-1 propagated for multiple generations and mated to EG6180 males. The slow-1/grow-1 TA activity was assessed by determining the proportion of delayed EG/EG non-carriers.

Single molecule in situ hybridization

Stellaris FISH Probes targeting slow-1, slow-2 and pgl-1 were designed using the Stellaris RNA FISH Probe Designer (Biosearch Technologies). The probes were labelled with Quasar 570, CAL Fluor Red 610 or Quasar 670, respectively (Biosearch Technologies). The protocol was adapted from Raj et al.⁶² and described in ref. ⁹. For imaging, an Axio Imager.Z2 (Carl Zeiss) widefield microscope with a Hamamatsu Orca Flash 4 camera and a 63×/1.4 plan-apochromat Oil DIC objective was used. Filters used were: DAPI excitation 406/15 nm, emission 457/50 nm and Quasar 570 excitation 545/30 nm, emission 610/75 nm. z-stack images with 40 slices (step size 0.2 µm) were acquired. Image analysis was performed with the FIJI plugin RS-FISH⁶³ with parameters set at Sigma 1.44, and threshold 0.0062.

RNA extraction and RNA-seq

Total RNA was extracted from approximately 100 young adult hermaphrodites and F₁ progeny, with the later using recessive mutations to visually discriminate cross-progeny from self-progeny. Reciprocal crosses were set up between parental strains for maternal or paternal inheritance of slow-1/grow-1 by mating INK531 hermaphrodites (uncoordinated worms in NIC203 background) to EG6180 males and QX2355 hermaphrodites (dumpy worms in EG6180 background) to NIC203 males and selecting phenotypically wild-type progeny for RNA extraction. Reciprocal crosses between NIL and EG6180 strains were performed analogously (INK255 hermaphrodites (dumpy worms in NIL background) to EG6180 males and QX2355 hermaphrodites (dumpy worms in EG6180 background) to QX2345 NIL males). Total RNA was extracted following a modified version of the protocol in⁶⁴ including multiple M9 washes, TRizol and chloroform incubation, phase-separation, isopropanol precipitation and resuspension in RNase-free water. Samples with RNA integrity number (RIN) > 8 were used for library preparation using the NEBNext Poly(A) kit and sequenced on NextSeq2000 P2 SR100 or NovaSeq S1 PE100 at the Vienna Biocenter NGS facility. To reduce reference bias, raw reads were aligned to a concatenated NIC203 + EG6180 genome/transcriptome assembly using STAR and bcbio-nextgen (https://github.com/bcbio/bcbio-nextgen). Transcript quantification and normalization were performed with tximport and Deseq2 (ref. ⁶⁵). We used Deseq2 to fit a model for the normalized counts using the strain identity of the mother and sequencing batch (Nextseq vs NovaSeq libraries) as fixed effects and compared the model to a null model that included only batch using a likelihood-ratio test. Despite identifying an outlier in the slow-1/grow-1 paternal inheritance samples (Fig. 1d), no obvious difference between the outlier and the other samples in terms of RNA quality and mRNA-seq quality control were identified. However, since each library was derived from an independent genetic cross, we cannot discard a human error, and therefore decided that it would be best practice to keep the outlier in the final analysis.

RT–qPCR

RNA was extracted from adult worms (50 males or 100 hermaphrodites per biological replicate) using TRIzol-chloroform extraction, followed by Dnase I digestion⁶⁶ and then RNA concentrations were measured using the Qubit High-Sensitivity RNA fluorescence kit (Thermo). cDNA was prepared with SuperScript III reverse transcriptase (Thermo) using random hexamers. Intron-spanning primers were validated with standard curves from QX2345 cDNA to ensure amplification efficiency and an r² value above 0.95. The following primers were used: FW-slow-1-mRNA: 5′-GAGCTACCGGAACTGGATAAAG-3′, RV-slow-1-mRNA: 5′-CAGAGTTCTCGGAAGTCTCCTC-3′, FW-slow-1-pre-mRNA: 5′-CGGACTGGATGAAACATTTAGC-3′, RV-slow-1-pre-mRNA: 5′-GAGCGGTGTTGACctgaatc-3′, FW-cdc-42: 5′-CGATTAAATGTGTCGTCGTAGG-3′, and RV-cdc-42: 5′-ACCGATCGTAATCTTCTTGTCC-3′. All samples had at least 3 biological replicates. We used the ∆∆C_t method to calculate relative fold change and chose cdc-42 as a housekeeping gene^67,68. Cdc-42 expression showed a low coefficient of variation in our RNA-seq datasets suggesting its validity as a housekeeping gene. All RT–qPCR reactions were prepared with the Luna Universal qPCR and RT–qPCR kit (NEB) and run with an annealing temperature of 58 °C. All biological replicates were run in technical quadruplicate and any reactions with abnormal amplification curves or melting temperatures were omitted before analysis (distinct from reactions for which we observed no amplification, which were not omitted). Representative samples from each condition were Sanger sequenced. We confirmed the absence of genomic DNA contamination in RNA samples by performing PCRs with gDNA-specific primers using the RNA as template and observed no amplification after 40 cycles. RT–qPCR indicated specific amplification of slow-1 in both hermaphrodites and males. However, the higher C_t values for males (34.27 versus 28.31 on average) and greater variability (s.d. of 1.55 versus 0.65 in the NIL) suggest much lower expression levels in males. This variability hinders a reliable estimate of abundance and assessment of the parent-of-origin effect in males.

Small RNA library preparation and sequencing

We isolated sRNAs, using the TraPR protocol⁶⁹. In brief, frozen worm pellets (2,000 worms per parental line) were supplemented with 350 µl lysis buffer, (20 mM HEPES-KOH, pH 7.9, 10% (v/v) glycerol, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM DTT, 0.1% v/v Triton X-100). Samples were mechanically disintegrated and subjected to 4 freeze–thaw cycles in liquid nitrogen. The resulting lysates were cleared by centrifugation and the sRNA fraction was isolated using the TraPR Small RNA Isolation Kit (135.24, LEXOGEN). Isolated sRNA was treated with RppH (M0356S, BioLabs), to ensure 5′ monophosphate-independent capturing of small RNAs⁷⁰, following purification with Agencourt RNA Clean XP magnetic beads (BECKMAN COULTER). The sRNA was ligated to a 32-nt 3′ adapter with unique barcodes (sRBC, Supplementary Table 6, IDT) using truncated T4 RNA ligase 2 (M0373L, NEB). The resulting RNA was run on 12% SequaGel–UreaGel (National Diagnostics) and purified with ZR small-RNA PAGE Recovery Kit (R1070, ZYMO RESEARCH). The 37-nt-long 5′ adapter was ligated to the sRNAs using T4 RNA ligase (M0204S, NEB). The resulting RNA was cleaned up (R1015, ZYMO RESEARCH), reverse-transcribed, and PCR amplified. The cDNA fragments (160–190 nt) were extracted and gel purified (D4008, ZYMO RESEARCH). Small RNA Libraries were sequenced in triplicates on a NovaSeq S1 SR100 mode (Illumina) at the Vienna Biocenter NGS facility. All sequencing libraries generated for this project are listed in Supplementary Table 7.

sRNA immunoprecipitation

To study piRNA binding preferences of PRG-1.1 and PRG-1.2, we performed sRNA immunoprecipitation of N-terminally Flag-tagged PRG-1.1 (INK775) and PRG-1.2 (INK735) followed by sRNA-seq. For each of the 3 biological replicates (50,000 worms each), 18 worm plates (9 cm) were bleached to synchronize the population. Young adults were collected, frozen at −70 °C, thawed and washed with RIP buffer (50 mM Hepes pH 7.2, 150 mM NaCl, 0.01% NP-40). For lysis, RIP buffer and Benzonase were added and sonicated in a Diagenode Bioruptor followed by cleaning via centrifugation. For immunoprecipitation, 200 µl of Anti-Flag M2 Magnetic Beads (Millipore) were used (4 °C, overnight). The bound proteins were eluted in 500 µl 0.1 M GlycinHCl pH 2.7 for 5 min at room temperature. And transferred into a vial with 50 µl 1 M Tris-HCl pH 8. The proteins were digested with Proteinase K (0.7 mg ml⁻¹), and denatured proteins were removed by centrifugation following proteinase K inactivation. Samples were stored at −70 °C until library preparation.

Small RNA analysis

Sequencing adapters were trimmed from 5′ and 3′ ends using Cutadapt v1.18 (ref. ⁷¹). Extracted 21U and 22G reads aligned to the genome using hisat2 v2.1 (ref. ⁷²). For 22 G, only reads mapped to the coding sequences were analysed; for 21U, reads mapped to coding sequences, tRNAs and rRNAs were excluded using seqkit v0.13 and samtools v1.10. 22 G reads were quantified using featureCounts (Rsubread, R), normalized by the total number of 22 G per replicate, and visualized using the Gviz R package⁶². Candidate 21U-RNAs were identified based on perfect mapping and abundance criteria (>0.1 ppm). A custom script quantified 21U-RNAs and reads were normalized to miRNAs predicted based on homology to C. elegans miRNAs. To identify potential 21U-RNAs slow-1 candidates we used known targeting rules in C. elegans and binding energies. First, putative binding sites and energies for all 21U-RNAs against slow-1 mRNA were predicted with RNAduplex (ViennaRNA Package v2.0.58)⁶³, of which five best duplexes for every piRNA were taken. Candidate piRNAs without bubbles during binding and no more than 4 mismatches outside the seed region were extracted and ranked by binding energy (Supplementary Data 1). The second candidate list was generated considering the overall level of binding continuity by using Nucleotide blast v2.2.26 in blastn-short mode. Only 21U-RNAs with no mismatches or gaps in the seed region were selected for further analysis. Finally, we ranked 21U-RNAs by the total length of the ungapped alignment to slow-1 (Supplementary Data 1).

Chromatin immunoprecipitation

For chromatin immunoprecipitation, we collected an F₄ population of homozygous carriers for the repressed slow-1 allele after paternal inheritance, which was highly enriched in s22G-RNA complementary to slow-1 (Fig. 3i,j). First, we crossed EG6180 hermaphrodites to NIL males. The F₂ were genotyped to identify repressed slow-1/grow-1 (NIC/NIC) worms which were expanded for two generations (F₄) and collected as young adults. Each ChIP sample represents an independent genetic cross. Worms (200 µl) were collected, washed and incubated to minimize bacterial content and frozen in liquid nitrogen. For ChIP, we used the protocol described⁶⁴. Shortly the frozen worm pellet was pulverized by grinding in mortar with liquid nitrogen and the powder was crosslinked in 1 ml ice-cold RIPA buffer supplemented with 2% formaldehyde to crosslink (10 min, 4 °C). After quenching by addition of 100 µl 1 M Tris-HCl (pH 7.5), the sample was sonicated using Covaris for 600 s to achieve chromatin fragments of 200–500 bp. Fifty microlitres of the lysate was saved as an input fraction. Chromatin was immunoprecipitated using anti-H3K9me3 antibody (Ab8898, Abcam). The immunoprecipitation product was incubated with Protein A Dynabeads (Thermofisher scientific) and washed with LiCl. The immunoprecipitation product was eluted from beads and DNA was purified using ChIP DNA Clean and Concentrator kit (Zymo Research). Input control fractions were treated similarly to immunoprecipitation samples. DNA libraries were prepared with NEBNext Ultra II DNA Library Prep Kit (Illumina), deduplicated using bbmap v38.26, aligned using bwa mem v0.7.17 (ref. ⁶⁵), and normalized by the number of reads that mapped to the genome with samtools v1.10 (ref. ⁷³). Peaks were called by macs2 v2.2.5 with –broad and –mfold 1 50 options⁷⁴. Quality control plots were made using deeptools v3.3.1 (ref. ⁷⁵). H3K9me3 signal was calculated as read counts per genomic position in the ChIP sample normalized by counts in the corresponding input sample using bedtools v2.27 (ref. ⁷⁶) and custom R (v4.3) script.

Immunohistochemistry

Gravid nematodes were washed from plates, and embryos were extracted using bleach solution. The embryo suspension was applied to prepared poly-l-lysine slides (Sigma-Aldrich, P8920), and immersed into liquid nitrogen, fixed in ice-cold methanol (10 min) followed by acetone (10 min), and rehydrated in descending ethanol concentrations (95%, 70%, 50% and 30% ethanol). Fixed embryos were blocked in 3% BSA (VWR Life Science, 422351 S), followed by incubation with anti-Flag M2 primary antibody (Sigma-Aldrich, F3165, diluted 1:3,000). After washing, a secondary antibody Alexa Fluor A568 (ThermoFisher Scientific, A-11031, diluted 1:3,000) was applied, followed by additional washes. The final wash contained DAPI (Merck, D9542, 5 ng ml⁻¹). Processed embryos were mounted with Fluoroshield (Sigma-Aldrich, F6182) and imaged at Axio Imager 2 (ZEISS).

Fluorescence intensity quantification

Twenty-four-bit raw images were analysed in Fiji (v1.53r)⁷⁷. Embryos were selected by freehand tool and the same selection mask was used to capture background fluorescence intensity for each embryo. To compare fluorescence intensities between strains we used corrected total cell fluorescence (CTCF) parameter (CTCF = integrated density − (area of selected cell × mean fluorescence of background readings)). At least 23 embryos were used for quantification.

Worm protein lysate preparation and western blot

Gravid adult worms were collected, washed, and flash-frozen in the liquid nitrogen. Worm pellets were resuspended in ice-cold lysis buffer (30 mM HEPES pH 7.4, 100 mM KCl, 2 mM MgCl2, 0.05% IGEPAL, 10% glycerol and 1 tablet of protease inhibitors (Roche, 11836153001)) and lysed by sonication in Bioruptor (UCD-200, Diagenode) followed by centrifugation to obtain the supernatant. After protein quantification by Bradford assay (Thermo Scientific, 23238), samples were diluted, resuspended in SDS loading buffer, and loaded onto NuPAGE gels (Invitrogen). Samples were transferred to 0.45 µm PVDF membrane (Thermo Scientific, 88518) and blocked with 4% non-fat milk in TBS-T. Membranes were incubated with anti-Flag M2 (mouse, 1:2,000, Sigma-Aldrich, F3165) or anti-actin (rabbit, 1:3,000, Abcam, ab13772) primary antibody overnight followed by incubation with HRP-conjugated anti-mouse (1:10,000, Invitrogen, G-21040) or anti-rabbit (1:10,000, Jackson Immuno, 111-035-045) secondary antibody. Detection was performed using ECL reagent (Cytiva, RPN2106) and imaged with ChemiDoc MP (Bio-Rad). Membranes were stripped before reprobing (Thermo Scientific, 21059).

Live imaging of mScarlet::SLOW-1

Approximately 20 gravid adults were dissected in M9 medium under a stereo microscope. Embryos were transferred to individual wells in a Thermo Scientific Nunc MicroWell 384-Well Optical-Bottom Plate (Thermo Scientific). Embryos were imaged using an Olympus spinning disk confocal based on an Olympus IX3 Series (IX83) inverted microscope, equipped with a dual-camera Yokogawa W1 spinning disk (Yokogawa Electric Corporation) and two ORCA-Flash 4.0 V3 Digital CMOS cameras (Hamamatsu). Each field was imaged using a 40×/0.75 NA (air) objective, 16 z-sections at 2 µm and conditions were as follows: bright-field (100% power 30 ms) 568 nm, (100% power, 500 ms). Image acquisition was performed using CellSense software (Olympus). Image processing and montages were created using Fiji and embryoCropUI⁷⁸.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

All sequencing data generated in this study are available under NCBI Project accession PRJNA850171.

References

Barlow, D. P. & Bartolomei, M. S. Genomic imprinting in mammals. Cold Spring Harb. Perspect. Biol. 6, a018382 (2014).
PubMed PubMed Central Google Scholar
Reik, W. & Walter, J. Genomic imprinting: parental influence on the genome. Nat. Rev. Genet. 2, 21–32 (2001).
CAS PubMed Google Scholar
Haig, D. The kinship theory of genomic imprinting. Annu. Rev. Ecol. Syst. 31, 9–32 (2000).
Google Scholar
Moore, T. & Haig, D. Genomic imprinting in mammalian development: a parental tug-of-war. Trends Genet. 7, 45–49 (1991).
CAS PubMed Google Scholar
Ondičová, M., Oakey, R. J. & Walsh, C. P. Is imprinting the result of “friendly fire” by the host defense system? PLoS Genet. 16, e1008599 (2020).
PubMed PubMed Central Google Scholar
Burt, A. & Trivers, R. Genes in Conflict: The Biology of Selfish Genetic Elements (Harvard University Press, 2006).
Burga, A., Ben-David, E. & Kruglyak, L. Toxin–antidote elements across the Tree of Life. Annu. Rev. Genet. 54, 387–415 (2020).
CAS PubMed Google Scholar
Ben-David, E., Burga, A. & Kruglyak, L. A maternal-effect selfish genetic element in Caenorhabditis elegans. Science 356, 1051–1055 (2017).
CAS PubMed PubMed Central ADS Google Scholar
Ben-David, E. et al. Ubiquitous selfish toxin–antidote elements in Caenorhabditis species. Curr. Biol. 31, 990–1001.e5 (2021).
CAS PubMed Google Scholar
Lafon-Placette, C. et al. Paternally expressed imprinted genes associate with hybridization barriers in Capsella. Nat. Plants 4, 352–357 (2018).
CAS PubMed Google Scholar
Barlow, D. P. Methylation and imprinting: from host defense to gene regulation? Science 260, 309–310 (1993).
CAS PubMed ADS Google Scholar
Li, E., Beard, C. & Jaenisch, R. Role for DNA methylation in genomic imprinting. Nature 366, 362–365 (1993).
CAS PubMed ADS Google Scholar
Pignatta, D. et al. Natural epigenetic polymorphisms lead to intraspecific variation in Arabidopsis gene imprinting. eLife 3, e03198 (2014).
PubMed PubMed Central Google Scholar
Gehring, M. Genomic imprinting: insights from plants. Annu. Rev. Genet. 47, 187–208 (2013).
CAS PubMed Google Scholar
Watanabe, T. et al. Role for piRNAs and noncoding RNA in de novo DNA methylation of the imprinted mouse Rasgrf1 locus. Science 332, 848–852 (2011).
CAS PubMed PubMed Central ADS Google Scholar
Inoue, A., Jiang, L., Lu, F., Suzuki, T. & Zhang, Y. Maternal H3K27me3 controls DNA methylation-independent imprinting. Nature 547, 419–424 (2017).
CAS PubMed PubMed Central Google Scholar
Youngman, E. M. & Claycomb, J. M. From early lessons to new frontiers: the worm as a treasure trove of small RNA biology. Front. Genet. 5, 416 (2014).
PubMed PubMed Central Google Scholar
Kiontke, K. C. et al. A phylogeny and molecular barcodes for Caenorhabditis, with numerous new species from rotting fruits. BMC Evol. Biol. 11, 339 (2011).
CAS PubMed PubMed Central Google Scholar
Seidel, H. S., Rockman, M. V. & Kruglyak, L. Widespread genetic incompatibility in C. elegans maintained by balancing selection. Science 319, 589–594 (2008).
CAS PubMed PubMed Central Google Scholar
Beeman, R. W., Friesen, K. S. & Denell, R. E. Maternal-effect selfish genes in flour beetles. Science 256, 89–92 (1992).
CAS PubMed ADS Google Scholar
Weichenhan, D., Traut, W., Kunze, B. & Winking, H. Distortion of Mendelian recovery ratio for a mouse HSR is caused by maternal and zygotic effects. Genet. Res. 68, 125–129 (1996).
CAS PubMed Google Scholar
Widen, S. A. et al. Virus-like transposons cross the species barrier and drive the evolution of genetic incompatibilities. Science 380, eade0705 (2023).
CAS PubMed ADS Google Scholar
Rošić, S. et al. Evolutionary analysis indicates that DNA alkylation damage is a byproduct of cytosine DNA methyltransferase activity. Nat. Genet. 50, 452–459 (2018).
PubMed PubMed Central Google Scholar
Sha, K. & Fire, A. Imprinting capacity of gamete lineages in Caenorhabditis elegans. Genetics 170, 1633–1652 (2005).
CAS PubMed PubMed Central Google Scholar
Devanapally, S. et al. Mating can initiate stable RNA silencing that overcomes epigenetic recovery. Nat. Commun. 12, 4239 (2021).
CAS PubMed PubMed Central ADS Google Scholar
Alcazar, R. M., Lin, R. & Fire, A. Z. Transmission dynamics of heritable silencing induced by double-stranded RNA in Caenorhabditis elegans. Genetics 180, 1275–1288 (2008).
CAS PubMed PubMed Central Google Scholar
Houri-Zeevi, L., Korem Kohanim, Y., Antonova, O. & Rechavi, O. Three rules explain transgenerational small rna inheritance in C. elegans. Cell 182, 1186–1197.e12 (2020).
CAS PubMed PubMed Central Google Scholar
Moore, R. S., Kaletsky, R. & Murphy, C. T. Piwi/PRG-1 Argonaute and TGF-β mediate transgenerational learned pathogenic avoidance. Cell 177, 1827–1841.e12 (2019).
CAS PubMed PubMed Central Google Scholar
Bagijn, M. P. et al. Function, targets, and evolution of Caenorhabditis elegans piRNAs. Science 337, 574–578 (2012).
CAS PubMed PubMed Central ADS Google Scholar
Batista, P. J. et al. PRG-1 and 21U-RNAs interact to form the piRNA complex required for fertility in C. elegans. Mol. Cell 31, 67–78 (2008).
CAS PubMed PubMed Central Google Scholar
Luteijn, M. J. et al. Extremely stable Piwi-induced gene silencing in Caenorhabditis elegans. EMBO J. 31, 3422–3430 (2012).
CAS PubMed PubMed Central Google Scholar
Shirayama, M. et al. piRNAs initiate an epigenetic memory of nonself RNA in the C. elegans germline. Cell 150, 65–77 (2012).
CAS PubMed PubMed Central Google Scholar
Ruby, J. G. et al. Large-scale sequencing reveals 21U-RNAs and additional microRNAs and endogenous siRNAs in C. elegans. Cell 127, 1193–1207 (2006).
CAS PubMed Google Scholar
Buckley, B. A. et al. A nuclear Argonaute promotes multigenerational epigenetic inheritance and germline immortality. Nature 489, 447–451 (2012).
CAS PubMed PubMed Central ADS Google Scholar
Ashe, A. et al. piRNAs can trigger a multigenerational epigenetic memory in the germline of C. elegans. Cell 150, 88–99 (2012).
CAS PubMed PubMed Central Google Scholar
Schwartz-Orbach, L. et al. Caenorhabditis elegans nuclear RNAi factor SET-32 deposits the transgenerational histone modification, H3K23me3. eLife 9, e54309 (2020).
CAS PubMed PubMed Central Google Scholar
Sapetschnig, A., Sarkies, P., Lehrbach, N. J. & Miska, E. A. Tertiary siRNAs mediate paramutation in C. elegans. PLoS Genet. 11, e1005078 (2015).
PubMed PubMed Central Google Scholar
Kalinava, N., Ni, J. Z., Peterman, K., Chen, E. & Gu, S. G. Decoupling the downstream effects of germline nuclear RNAi reveals that H3K9me3 is dispensable for heritable RNAi and the maintenance of endogenous siRNA-mediated transcriptional silencing in Caenorhabditis elegans. Epigenet. Chromatin 10, 6 (2017).
Google Scholar
Gu, S. G. et al. Amplification of siRNA in Caenorhabditis elegans generates a transgenerational sequence-targeted histone H3 lysine 9 methylation footprint. Nat. Genet. 44, 157–164 (2012).
CAS PubMed PubMed Central Google Scholar
Kalinava, N. et al. C. elegans heterochromatin factor SET-32 plays an essential role in transgenerational establishment of nuclear RNAi-mediated epigenetic silencing. Cell Rep. 25, 2273–2284.e3 (2018).
CAS PubMed PubMed Central Google Scholar
Woodhouse, R. M. et al. Chromatin modifiers SET-25 and SET-32 are required for establishment but not long-term maintenance of transgenerational epigenetic inheritance. Cell Rep. 25, 2259–2272.e5 (2018).
CAS PubMed Google Scholar
Lev, I., Gingold, H. & Rechavi, O. H3K9me3 is required for inheritance of small RNAs that target a unique subset of newly evolved genes. eLife 8, e40448 (2019).
PubMed PubMed Central Google Scholar
Fire, A., Alcazar, R. & Tan, F. Unusual DNA structures associated with germline genetic activity in Caenorhabditis elegans. Genetics 173, 1259–1273 (2006).
CAS PubMed PubMed Central Google Scholar
Frøkjær-Jensen, C. et al. An abundant class of non-coding DNA can prevent stochastic gene silencing in the C. elegans germline. Cell 166, 343–357 (2016).
PubMed PubMed Central Google Scholar
Johnson, C. L. & Spence, A. M. Epigenetic licensing of germline gene expression by maternal RNA in C. elegans. Science 333, 1311–1314 (2011).
CAS PubMed ADS Google Scholar
Gajic, Z. et al. Target-dependent suppression of siRNA production modulates the levels of endogenous siRNAs in the Caenorhabditis elegans germline. Development 149, dev200692 (2022).
CAS PubMed PubMed Central Google Scholar
Blumenthal, T. & Gleason, K. S. Caenorhabditis elegans operons: form and function. Nat. Rev. Genet. 4, 110–118 (2003).
Google Scholar
Spieth, J., Brooke, G., Kuersten, S., Lea, K. & Blumenthal, T. Operons in C. elegans: polycistronic mRNA precursors are processed by trans-splicing of SL2 to downstream coding regions. Cell 73, 521–532 (1993).
CAS PubMed Google Scholar
Gu, W. et al. Distinct Argonaute-mediated 22G-RNA pathways direct genome surveillance in the C. elegans germline. Mol. Cell 36, 231–244 (2009).
CAS PubMed PubMed Central Google Scholar
Conine, C. C. et al. Argonautes promote male fertility and provide a paternal memory of germline gene expression in C. elegans. Cell 155, 1532–1544 (2013).
CAS PubMed PubMed Central Google Scholar
Yu, X. et al. A selfish genetic element confers non-Mendelian inheritance in rice. Science 360, 1130–1132 (2018).
CAS PubMed ADS Google Scholar
Chen, J. et al. A triallelic system of S5 is a major regulator of the reproductive barrier and compatibility of indica–japonica hybrids in rice. Proc. Natl Acad. Sci. USA 105, 11436–11441 (2008).
CAS PubMed PubMed Central ADS Google Scholar
Ghanta, K. S. & Mello, C. C. Melting dsDNA donor molecules greatly improves precision genome editing in Caenorhabditis elegans. Genetics 216, 643–650 (2020).
CAS PubMed PubMed Central Google Scholar
Kolmogorov, M., Yuan, J., Lin, Y. & Pevzner, P. A. Assembly of long, error-prone reads using repeat graphs. Nat. Biotechnol. 37, 540–546 (2019).
CAS PubMed Google Scholar
Minkin, I., Patel, A., Kolmogorov, M., Vyahhi, N. & Pham, S. in Algorithms in Bioinformatics (eds Darling, A. & Stoye, J.) 215–229 (Springer, 2013).
Kolmogorov, M., Raney, B., Paten, B. & Pham, S. Ragout—a reference-assisted assembly tool for bacterial genomes. Bioinformatics 30, i302–i309 (2014).
CAS PubMed PubMed Central Google Scholar
Seroussi, U. et al. A comprehensive survey of C. elegans argonaute proteins reveals organism-wide gene regulatory networks and functions. eLife 12, e83853 (2023).
CAS PubMed PubMed Central Google Scholar
Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).
CAS PubMed PubMed Central Google Scholar
Howe, K. L., Bolt, B. J., Shafie, M., Kersey, P. & Berriman, M. WormBase ParaSite—a comprehensive resource for helminth genomics. Mol. Biochem. Parasitol. 215, 2–10 (2017).
CAS PubMed PubMed Central Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
CAS PubMed PubMed Central ADS Google Scholar
van Kempen, M. et al. Fast and accurate protein structure search with Foldseek. Nat. Biotechnol. https://doi.org/10.1038/s41587-023-01773-0 (2023).
Hahne, F. & Ivanek, R. Visualizing genomic data using Gviz and Bioconductor. Stat. Genomics 1418, 335–351 (2016).
Google Scholar
Lorenz, R. et al. ViennaRNA package 2.0. Algorithms for Mol. Biol. 6, 26 (2011).
Google Scholar
Ni, J. Z., Chen, E. & Gu, S. G. Complex coding of endogenous siRNA, transcriptional silencing and H3K9 methylation on native targets of germline nuclear RNAi in C. elegans. BMC Genomics 15, 1157 (2014).
PubMed PubMed Central Google Scholar
Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Preprint at https://arxiv.org/abs/1303.3997 (2013).
Green, M. R. & Sambrook, J. Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Laboratory Press, 2012).
Hoogewijs, D., Houthoofd, K., Matthijssens, F., Vandesompele, J. & Vanfleteren, J. R. Selection and validation of a set of reliable reference genes for quantitative sod gene expression analysis in C. elegans. BMC Mol. Biol. 9, 9 (2008).
PubMed PubMed Central Google Scholar
Serobyan, V. et al. Transcriptional adaptation in Caenorhabditis elegans. eLife 9, e50014 (2020).
CAS PubMed PubMed Central Google Scholar
Grentzinger, T. et al. A universal method for the rapid isolation of all known classes of functional silencing small RNAs. Nucleic Acids Res. 48, e79 (2020).
CAS PubMed PubMed Central Google Scholar
Almeida, M. V., de Jesus Domingues, A. M., Lukas, H., Mendez-Lago, M. & Ketting, R. F. RppH can faithfully replace TAP to allow cloning of 5′-triphosphate carrying small RNAs. MethodsX 6, 265–272 (2019).
PubMed PubMed Central Google Scholar
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10–12 (2011).
Google Scholar
Kim, D., Paggi, J. M., Park, C., Bennett, C. & Salzberg, S. L. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37, 907–915 (2019).
CAS PubMed PubMed Central Google Scholar
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
PubMed PubMed Central Google Scholar
Zhang, Y. et al. Model-based Analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008).
PubMed PubMed Central Google Scholar
Ramírez, F. et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 44, W160–W165 (2016).
PubMed PubMed Central Google Scholar
Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010).
CAS PubMed PubMed Central Google Scholar
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
CAS PubMed Google Scholar
Wang, S. et al. A high-content imaging approach to profile C. elegans embryonic development. Development 146, dev174029 (2019).
CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

The authors thank members of the Burga laboratory and J. Mueller for critical reading of the manuscript. Research in the Burga laboratory is supported by the Austrian Academy of Sciences, the city of Vienna, and a European Research Council (ERC) Starting Grant under the European 20 Union’s Horizon 2020 programme (ERC-2019-StG-851470). A.K. is supported by a Boehringer Ingelheim Fonds (BIF) PhD Fellowship. S.A.W. is supported by the European Union’s Framework Programme for Research and Innovation Horizon 2020 (2014‐2020) under the Marie Curie Skłodowska Grant Agreement Nr. 847548. Work by the Ben-David laboratory for this study was supported by the Israel Science Foundation (grant no. 2023/20). We thank Life Science Editors for editing services.

Author information

Eyal Ben-David
Present address: Illumina Artificial Intelligence Laboratory, Illumina, San Diego, CA, USA
These authors jointly supervised this work: Eyal Ben-David, Alejandro Burga

Authors and Affiliations

Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA), Vienna BioCenter (VBC), Vienna, Austria
Pinelopi Pliota, Hana Marvanova, Alevtina Koreshova, Polina Tikanova, Daniel Krogull, Andreas Hagmüller, Sonya A. Widen, Dominik Handler, Joseph Gokcezade, Peter Duchek, Julius Brennecke & Alejandro Burga
Vienna BioCenter PhD Program, Doctoral School of the University of Vienna and Medical University of Vienna, Vienna, Austria
Hana Marvanova, Alevtina Koreshova, Polina Tikanova & Daniel Krogull
Department of Biochemistry and Molecular Biology, Institute for Medical Research Israel–Canada, The Hebrew University of Jerusalem, Jerusalem, Israel
Yotam Kaufman & Eyal Ben-David

Authors

Pinelopi Pliota
View author publications
You can also search for this author in PubMed Google Scholar
Hana Marvanova
View author publications
You can also search for this author in PubMed Google Scholar
Alevtina Koreshova
View author publications
You can also search for this author in PubMed Google Scholar
Yotam Kaufman
View author publications
You can also search for this author in PubMed Google Scholar
Polina Tikanova
View author publications
You can also search for this author in PubMed Google Scholar
Daniel Krogull
View author publications
You can also search for this author in PubMed Google Scholar
Andreas Hagmüller
View author publications
You can also search for this author in PubMed Google Scholar
Sonya A. Widen
View author publications
You can also search for this author in PubMed Google Scholar
Dominik Handler
View author publications
You can also search for this author in PubMed Google Scholar
Joseph Gokcezade
View author publications
You can also search for this author in PubMed Google Scholar
Peter Duchek
View author publications
You can also search for this author in PubMed Google Scholar
Julius Brennecke
View author publications
You can also search for this author in PubMed Google Scholar
Eyal Ben-David
View author publications
You can also search for this author in PubMed Google Scholar
Alejandro Burga
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

P.P. conducted the genetic analyses for this study under the supervision of E.B.-D. and A.B. sRNA-seq, ChIP–seq and other molecular assays were led by H.M. with the support of P.P., D.H., A.K., P.T., D.K., A.H., Y.K. and S.A.W. under the supervision of J.B. and A.B. Genome assemblies and all bioinformatic analyses were carried out by A.K. and supported by E.B.-D. Design and generation of worm transgenic lines was performed by J.G. and P.D. A.B. conceptualized the study, designed experiments together with P.P., and wrote the manuscript.

Corresponding author

Correspondence to Alejandro Burga.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature thanks Robert Feil, Mofang Liu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer review reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 slow-1 has a parent-of-origin effect and grow-1.1 and grow-1.2 are redundant antidotes.

a, Corrected NIC203 genome assembly showing segmental duplication of the grow-1 antidote (top). Selfing of grow-1.1(+/−); grow-1.2(−/−) strain. All grow-1.1(−/−); grow-1.2(−/−) individuals were developmentally arrested during larval development and did not produce any viable offspring. Thus, grow-1.1 and grow-1.2 are genetically redundant (bottom). Data are presented as mean values +/−95% CI. b, Model illustrating the mechanism of action of the slow-1/grow-1 TA. In crosses between the carrier strain (slow-1/grow-1 Chr. III NIL) and non-carrier strain (EG6180), 25% of the F₂ progeny is developmentally delayed because EG/EG homozygotes did not inherit the TA and cannot express the antidote to counteract the maternally deposited toxin. c, Quantification of slow-1 mRNA expression by smFISH (N = 3, n = 16–30 per repeat, two-sided unpaired t-test, p = 0.0006, mean +/−SEM) and RNA-seq in both NIL and EG6180 parental strains (two-sided unpaired t-test, p < 0.0001, mean +/−SEM). d, Previously, we showed that sup-35/pha-1 is active when maternally inherited (Ben-David, et al. Science 2017). To test whether sup-35/pha-1 is active when paternally inherited, we crossed DL238 hermaphrodites and N2 males. N2 carries two TAs, peel-1/zeel-1 and sup-35/pha-1. We observed 46.7% embryonic lethality among their F₂ progeny (n = 340, mean +/−95% CI is shown), as expected from the activity of two TAs segregating independently. To confirm the activity of both TAs, we genotyped wild-type F₂ progeny for both peel-1/zeel-1 (Chr. I) and sup-35/pha-1 (Chr. III) and found that the vast majority of WT progeny were either homozygous or heterozygous carriers, indicating that both peel-1/zeel-1 and sup-35/pha-1 non-carrier individuals died as embryos. e, Activity of the C. briggsae HK104 Chr. III msft-1 TA in reciprocal crosses. Penetrance of the toxin, the percentage of F₂ non-carrier individuals that are phenotypically affected, is used as a proxy for TA activity (HK Chr. III TA: n_M = 13, n_P = 35, p = 0.42; two-sided Fisher’s exact test; mean +/−95% CI is shown). f, Illumina short-reads from NIC203, EG6180, and Chr. III NIL DNA libraries were aligned against the NIC203 mitochondrial genome. Each dot represents a SNP. As expected from our cross scheme, the Chr. III NIL has the EG6180 mitochondrial genotype. Those SNPs shared by all strains likely reflect an error in the original NIC203 assembly. g, The slow-1 parent-of-origin effect is also present in a reciprocal cross between NIC203 and EG6180 parental lines. The abundance of slow-1 transcripts is higher when the slow-1 locus is maternally inherited (two-way ANOVA, interaction p = 0.0005, Holm-Sidak post hoc test, p_slow-1 < 0.0001, mean +/−SEM is shown). h, Expression pattern of mScarlet::SLOW-1 during embryonic development. In early embryos, SLOW-1 appeared to be associated with the nuclear envelope and was quickly degraded during embryogenesis. SLOW-1 was not detectable in the soma by the time embryos reached the comma-stage (N = 2, number of embryos imaged = 20). i, Expression pattern of mScarlet::SLOW-1 in the hermaphroditic gonad. Quantification of signal intensity in gonads of NIL and mScarlet::SLOW-1 strains (two-sided unpaired t-test, p < 0.0001, mean +/−SEM is shown).

Extended Data Fig. 2 Phylogenetic tree of C. tropicalis and C. elegans Argonaute proteins.

C. elegans (yellow) and C. tropicalis (blue) Argonaute proteins. Putative pseudogene in gray. Also included A. thaliana AGO1 and D. melanogaster PIWI. Red circles denote bootstrap values > 95%.

Extended Data Fig. 3 C. tropicalis PRG-1.1 and PRG-1.2 are redundant paralogs and PRG-1.2 acts maternally.

a, Protein alignment of PRG-1.1 and PRG-1.2. The two proteins share 87% amino-acid pairwise identity with 722/828 identical sites. Conserved PAZ and PIWI domains are highlighted. b, Detection of endogenously tagged PRG-1.1::FLAG and PRG-1.2::FLAG by western blot. Black arrow indicates the expected MW. EG6180 was used as a negative control. Western blot against β-Actin serves as a sample processing control. Uncropped gels available in Supplementary Fig. 1. c, Quantification of FLAG immunofluorescence quantification of C. tropicalis PRG-1.1 and PRG-1.2 expression from embryos, (n_EG6180 = 53, n_prg-1.1 = 34, n_prg-1.2 = 35, one-way ANOVA, p < 0.0001, Tukey post hoc test, p_{EG vs prg-1.1} < 0.0001, p_{EG vs prg-1.2} < 0.0001, p_{prg-1.1 vs prg-1.2} < 0.0001, mean +/−SEM is shown). d, Selfing of prg-1.1(−); prg-1.2(+/−) strain. All prg-1.1(−); prg-1.2(−) individuals were sterile, therefore the line couldn’t be propagated (mean +/−95% CI is shown). e, Cross of prg-1.2(−) hermaphrodites to NIL males (top) and NIL hermaphrodites to prg-1.2(−) males (bottom) indicates that paternal slow-1 is repressed only when prg-1.2 is maternally inherited. Percentage of F₂ EG/EG delayed progeny (right) when prg-1.2 is absent from the mother or the father compared to the WT cross. When prg-1.2 null mutant mothers were crossed to wild-type NIL males, we observed that 52.1% of F₂ homozygous EG/EG individuals were delayed. In contrast, when EG6180 mothers were crossed to prg-1.2 null mutant NIL males, F₂ homozygous EG/EG individuals were mostly wild-type (13.8% delay), indicating that maternal prg-1.2 is necessary for slow-1 repression (n_p = 53, n_{mother prg-1.2(−)} = 23, n_{father prg-1.2(−)} = 29, two-sided Fisher’s exact test compared to WT, p_{mother prg-1.2(−)} = 0.0001, p_{father prg-1.2(−)} = 0.71, mean +/−95% CI is shown). f, PRG-1.2 is not necessary for the maintenance of slow-1 repression. In this crossing scheme mothers provide PRG-1.2 to their F₁ progeny, which is sufficient for piRNA-mediated repression, and the slow-1/grow-1 is paternally inherited. Since EG6180 hermaphrodites do not provide maternal slow-1 transcripts, then the slow-1 paternal allele is epigenetically repressed in the F₁. Then F₂ progeny are singled, and their offspring genotyped for both the TA and the prg-1.2 locus. Hermaphrodites that are homozygous carriers for both the TA and the prg-1.2(−/−) null allele are identified (6.25% the F₂ progeny) and propagated for 1 or 7 generations. Then these hermaphrodites are crossed to EG6180 prg-1.2 (−/−) males to test whether the TA is active. Inset shows the observed activity of the TA measured as percentage of delayed (EG/EG) individuals (n_M = 34, n_{prg-1.2_3_gen} = 62, two-sided Fisher’s exact test, p < 0.0001, mean +/−95% CI is shown).

Extended Data Fig. 4 Characterization of the redundant role of PRG-1.1 and PRG-1.2 in C. tropicalis.

a, Schematic of the sRNA immunoprecipitation (RIP) protocol. RIP was performed in biological triplicates on N-terminally FLAG-tagged PRG-1.1 and PRG-1.2 strains (tagging the endogenous locus) followed by sRNA-seq. b, Class enrichment among sRNAs bound either to PRG1-1.1 and PRG-1.2 compared to a total sRNA-seq protocol (n_PRG-1.1 = 3, n_PRG-1.2 = 3). c, Differences in the abundance of piRNAs bounds by each PRG-1 paralog. Normalized abundance (transcript per million) of PRG-1.1 (left) and PRG-1.2 (right) bound piRNAs (21U-RNAs) are plotted on the Y-axis and genomic coordinate of the respective piRNA loci on the X-axis. Differentially bound piRNAs are shown with different thresholds of significance and are enriched within specific genomic clusters (moderated t-test, p-value adjusted with Benjamini & Hochberg correction).

Extended Data Fig. 5 Characterization of mutants in the in C. tropicalis piRNA pathway and the repressive signal.

a, Schematic of the piRNA pathway in C. elegans (left). C. tropicalis mutants generated in this study, their fertility status, and evolutionary relationship to C. elegans proteins (right). b, C. tropicalis mut-16 mutant alleles generated with CRISRP/Cas9. Homozygous mutants were sterile. c, C. tropicalis hrde-1 mutant allele generated with CRISRP/Cas9. Homozygous mutants were sterile. d, C. tropicalis simr-1 mutant allele generated with CRISRP/Cas9. Homozygous mutants were sterile. e, Alphafold2 predictions of C. elegans SET-32 and C. tropicalis SET-32 proteins and their structural alignment (PYMOL). Both proteins are highly similar (RMSD = 0.962). f, Mating scheme to generate an F₄ homozygous population for the repressed slow-1/grow-1 TA following its paternal inheritance. This F₄ population was subjected to sRNA-seq in biological quadruplicates (each time starting from an independent initial parental cross). As a control for an active or licensed TA, we performed sRNA-seq in the NIL parental line in biological quadruplicates. Red asterisk denotes slow-1 repressed state. g, Zoom-in into the 21ur-06949 and 21ur-06917 predicted binding sites in slow-1 and the 22G-RNAs mapping to these regions. The 22G-RNAs are derived from the F₄ “slow-1 repressed” population. Each track is one of four biological quadruplicates. A modest but clear peak is observed within the predicted binding region.

Extended Data Fig. 6 Examining the role of PATCs and CSR-1 in slow-1 licensing.

a, slow-1 lacks PATCs in its intronic sequences. PATC periodicity analysis for positive control C. elegans smu-1 and C. tropicalis slow-1. Highest periodicity found at 10.5 bp and 20 bp respectively. Summary table of PATC analysis. Phasing threshold was set to 60 (~1% Phasing in random DNA). Analysis was run on https://wormbuilder.org/patc/. b, Detection of endogenously tagged FLAG::CSR-1a and FLAG::CSR-1a+b by western blot. Black arrows indicate the expected MW of each isoform. Western blot against β-Actin serves as a sample processing control. Uncropped gels available in Supplementary Fig. 1. c, Representative immunostaining images against FLAG::CSR-1a and FLAG::CSR-1a+b line in 2-cell stage embryos. EG6180 serves as a negative control (top). FLAG immunofluorescence quantification of C. tropicalis CSR-1 expression from 2-cell stage embryos of EG6180 (negative control), FLAG::CSR-1a, and FLAG::CSR-1a+b (2 repeats, n_EG6180 = 44, n_CSR1a = 40, n_CSR1a+b = 105, one-way ANOVA, p < 0.0001, Tukey post hoc test, p_{EG vs CSR1a} = 0.41, p_{EG vs CSR1a+b} < 0.0001, p_{CSR1a vs CSR1a+b} < 0.0001, mean +/−SEM is shown). d, Generation of balanced csr-1 null strain by inserting NeoR in csr-1 and growing worms in G418 antibiotic. e, Representative images of wild-type (left), and csr-1(−) (right) embryos. Anaphase bridging events were observed in 14.75% of csr-1(−) early embryos (n = 61) compared to 0% (n = 36) of EG6180 WT. f, Crosses between csr-1(−); slow-1/grow-1 hermaphrodites and EG6180 males indicate that slow-1/grow-1 is fully active when maternally inherited. csr-1(−) hermaphrodites were derived from selfing of csr-1(+/−) mothers. (n_M = 34, n_{csr-1(−/−)} = 32, two-sided Fisher’s exact test, p = 0.4848, mean +/−95% CI is shown).

Extended Data Fig. 7 Characterization of the licensing signal.

a, Variability in the rescue of slow-1/grow-1 activity across injected hermaphrodites following paternal inheritance. Sixteen slow-1(Δ)/grow-1(−) NIL hermaphrodites (A to P) were injected in both gonad arms with in vitro transcribed slow-1 RNA (mutated start codon) and later mated to slow-1/grow-1 NIL males. The total number of homozygous slow-1(Δ)/grow-1(−) F₂ progeny from each hermaphrodite is shown on top of each bar. The sample labeled as “random” represents embryos randomly picked from different injected mothers (mean +/−95% CI is shown). b, Quantification of slow-1 transcripts following deletion of a 620 bp region upstream of slow-1. Comparison between slow-1(+)/ grow-1.1(+)/grow-1.2(−) NIL and slow-1(Δpr)/grow-1.1(+)/grow-1.2(−) NIL strains by RNA-seq. Deletion of the slow-1 promoter causes a 176-fold decrease in slow-1 transcript levels (two-sided unpaired t-test, p < 0.0001, mean +/−SEM is shown). Quantification of the genes immediately upstream of slow-1 and downstream of grow-1. Deletion of the slow-1 promoter does not change the transcript level of the two neighboring genes (two-way ANOVA, interaction p = 0.81, mean +/−SEM is shown). c, Cross between slow-1(Δpr)/grow-1(−) NIL hermaphrodites and NIL males. All homozygous slow-1(Δpr)/grow-1(−) NIL F₂ progeny are delayed (100%, n = 21, mean +/−95% CI is shown) (bottom). d, Expression pattern of the mScarlet::SL2::slow-1 operon in a gravid hermaphrodite. mScarlet fluorescence is ubiquitous in the germline (including oocytes) an also present in early embryos, as expected from being driven by the endogenous slow-1 promoter. GFP channel shows the autofluorescence of the gut as a reference (n = 35). e, Quantification of mScarlet fluorescence in the F₁ progeny of reciprocal crosses between the mScarlet::SL2::slow-1 operon and EG6180 strains. Each data point represents one individual. Expression levels of mScarlet are lower when the operon is paternally inherited likely through piRNA-mediated co-repression due to lack of slow-1 licensing (n_m = 14, n_p = 23_, two-sided unpaired t-test, p < 0.0001, mean +/−SEM is shown). f, Schematic of a cross designed to test whether an epigenetically repressed slow-1 allele can license a naïve zygotic one. To this end, we took advantage of the slow-1(fs)/grow-1(−) double mutant line that carries a slow-1 frameshift mutation. This mutation renders the slow-1 toxin inactive, but it doesn’t affect the ability of its maternal mRNA to license a paternally inherited slow-1 copy (Fig. 4a). We first crossed EG6180 hermaphrodites to slow-1(fs) grow-1(−) males and collected F₂ homozygous carriers of the repressed slow-1(fs) allele. We let these hermaphrodites self-fertilize for one generation, collected F₃ hermaphrodites and crossed them to NIL males carrying the wild-type TA. Finally, we phenotyped and genotyped their granddaughters (F₅ generation). If the wild-type TA were active, i.e. licensed, then progeny homozygous for the slow-1(fs)/grow-1(−) allele should be developmentally delayed. We found no significant delay among the progeny, and all homozygous mutants were wild-type indicating that licensing had been compromised.

Extended Data Fig. 8 A related but divergent TA, slow-2/grow-2, is active when paternally inherited.

a, Alignment of Illumina short-reads to the EG6180 de novo assembly. Highlighted region is Chr. III EG6180 region homologous to NIC203 slow-1/grow-1. b, slow-2 expression in EG6180 2-cell stage embryos by smFISH. pgl-1 serves as a positive control (quantification in Extended Data Fig. 9a). c, Mating of slow-2(−)/grow-2(−) double mutant NIL (hermaphrodites) to the parental EG6180 line causes developmental delay of homozygous double-mutant F₂ individuals indicating that slow-2/grow-2 is a TA, (mean +/−95% CI is shown). d, Two TAs are active in crosses between the NIL and EG6180: slow-1/grow-1 and slow-2/grow-2, respectively (Fig. 1b). The penetrance of the slow-2/grow-2 TA is incomplete; however, the toxin is equally active when maternally or paternally inherited (n_M = 40, n_P = 67, two-sided Fisher’s exact test, p = 0.31, mean +/−95% CI is shown). e, Reciprocal crosses between the NIL and EG6180 followed by RNA-seq of their F₁ progeny indicates that slow-2 transcripts are equally abundant when maternally or paternally inherited (two-sided unpaired t-test, p = 0.6451, mean +/−SEM is shown).

Extended Data Fig. 9 Characterization of slow-2 and grow-2 null alleles and comparison to slow-1.

a, Quantification of slow-2 expression in EG6180 embryos by smFISH (N = 3, n = 16–30 per repeat, two-sided unpaired t-test, p = 0.0013, mean +/−SEM is shown). slow-2 expression in EG6180 and absence in the NIL was also confirmed by RNA-seq data (n = 3, two-sided unpaired t-test, p < 0.0001, mean +/−SEM is shown). b, The slow-2 null allele generates a three base pair deletion in the first exon, which creates a premature stop codon. This results in a shorter peptide (54 aa compared to 357 aa). Representative Sanger sequences of WT and mutant alleles. c, The grow-2 null allele generates a deletion and frameshift in the first exon which creates a premature stop codon. This results in a shorter peptide (6 aa compared to 124 aa). Representative Sanger sequences of WT and mutant alleles. d, Protein alignment of SLOW-1 and SLOW-2 (top), and protein alignment of GROW-1 and GROW-2 (bottom). e, Quantification of total 22G-RNAs derived from slow-1 and slow-2 (top left, two-sided unpaired t-test p = 0.042, mean +/−SEM is shown). Distribution of sRNA per parental strain (top right, mean + SEM is shown). RNA-seq and 22G-sRNA short-reads aligned against slow-1 and slow-2. sRNA libraries were generated in biological triplicates. Notice the differences in the y-axis scale between slow-1 and slow-2.

Supplementary information

Supplementary Information

Supplementary Methods, Supplementary Fig. 1, Supplementary Tables 1–7, Supplementary Discussion, Supplementary Notes 1–3 and references.

Reporting Summary

Supplementary Data 1

Annotation of C. tropicalis piRNAs and ranking.

Supplementary Data 2

Summary of PRG-1.1 and PRG-1.1 sRNA data.

Peer Review File

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Pliota, P., Marvanova, H., Koreshova, A. et al. Selfish conflict underlies RNA-mediated parent-of-origin effects. Nature 628, 122–129 (2024). https://doi.org/10.1038/s41586-024-07155-z

Download citation

Received: 21 November 2022
Accepted: 02 February 2024
Published: 06 March 2024
Issue Date: 04 April 2024
DOI: https://doi.org/10.1038/s41586-024-07155-z

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.