Abstract
Germ cells deficient for Piwi and associated small RNA genome silencing factors transmit a form of heritable stress that induces sterility after growth for several generations. The cause of this transgenerational sterility phenotype is not understood but has been attributed to progressive deterioration of heterochromatin and associated DNA damage. Sterile small RNA genome silencing mutants displayed inconsistent increases in DNA damage signaling but consistently altered perinuclear germ granules. Germ granule dysfunction was sufficient to induce phenotypes associated with sterile small RNA genome silencing mutants, including germline atrophy, reproductive diapause and univalents in oocytes. Genes that perturb germ granule structure were not compromised in sterile small RNA mutants, suggesting a post-transcriptional reproductive arrest mechanism. We conclude that the integrity of germ granules, which are intimately associated with Piwi silencing factors, orchestrates the sterility of Piwi deficient mutants and could be generally relevant to regulation of reproductive arrest in response to stress.
Significance Statement Sterility in response to heterochromatin dysfunction was recognized in Drosophila P-M Hybrid Dysgenesis experiments carried out by Margaret Kidwell in the 1970’s and more recently in mutants deficient for Piwi/piRNA silencing. A central model of these studies is that this sterility is caused by transposon expression and associated genomic instability. We demonstrate that sterility in response to deficiency for Piwi is a form of reproductive arrest that is orchestrated by germ granule dysfunction. Germ granules promote genomic silencing of transposons and viruses, and they are likely to be frequently targeted by these parasites. We hypothesize the evolution of a reproductive arrest mechanism that responds to attacks on germ granules by disrupting their integrity.
Introduction:
Germ cells give rise to the mortal somatic tissues while maintaining themselves in a pristine condition that allows them to be propagated for an indefinite number of generations. This capacity for self-renewal is termed “germ cell immortality”, and functions to protect the germline from forms of damage that can result in sterility (Smelick and Ahmed, 2005). Defects in Caenorhabditis elegans (C. elegans) genes that promote small RNA-mediated genome silencing (Fig. 5a) can cause a Mortal Germline (mrt) phenotype, where robust fertility occurs for several generations but deteriorates in later generations, finally culminating in complete sterility. Many mrt mutants only become sterile when grown at the non-permissive temperature 25°C (Ahmed and Hodgkin, 2000), some of which are deficient for small RNA-mediated genome silencing (Ashe et al., 2012; Buckley et al., 2012; Burkhart et al., 2011; McMurchy et al., 2017; Sakaguchi et al., 2014; Smelick and Ahmed, 2005; Spracklin et al., 2017). However, the piRNA-deficient mutant prg-1 and the nuclear RNA interference (RNAi) mutants nrde-1 and nrde-4 display progressive sterility at any temperature (Batista et al., 2008; Buckley et al., 2012; Burkhart et al., 2011; Das et al., 2008; McMurchy et al., 2017; Simon et al., 2014).
Piwi is a conserved germ line Argonaute protein that interacts with thousands of small RNAs termed piRNAs to repress transposons and foreign nucleic acids that represent parasitic threats to the integrity of the genome (Aravin et al., 2007; Luteijn and Ketting, 2013; Siomi et al., 2011). Deficiency for the C. elegans Piwi ortholog prg-1 compromises germ cell immortality, and late-generation prg-1 mutant strains were recently shown to elicit a striking germ cell atrophy phenotype at the generation of sterility, as L4 larvae mature into 1 day old adults (Heestand et al., 2018). Moreover, atrophied germ lines of sterile prg-1 mutants can regrow on day 2 of adulthood, and a small fraction of constitutively sterile 5 day old prg-1 mutant adults can become fertile again if they are shifted to a distinct food source (Heestand et al., 2018). This indicates that the late-generation sterility of prg-1/Piwi mutants is a form of reproductive arrest termed Adult Reproductive Diapause, which can be induced in response to environmental stresses such as starvation (Angelo and Van Gilst, 2009; Padilla and Ladage, 2012; Tatar and Yin, 2001). We therefore proposed that late-generation sterility of prg-1/Piwi mutants may occur in response to accumulation of a heritable stress that is transmitted by germ cells (Heestand et al., 2018). Consistently, temperature-sensitive small RNA genome silencing mutants rsd-2 and rsd-6 also exhibit small and empty germlines at the generation of sterility (Sakaguchi et al., 2014).
Transposon-induced genome instability has been implicated in the acute sterility phenotype of Piwi/piRNA mutants in several species (Juliano et al., 2011; Siomi et al., 2011), and fertility defects in response to heterochromatin dysfunction have been associated with increased levels of RNA-DNA hybrids and of transposon-induced DNA damage (McMurchy et al., 2017; Zeller et al., 2016). C. elegans prg-1 mutants display progressively increased levels of germline apoptosis when grown for successive generations, and apoptosis promotes germ cell atrophy for sterile rsd-2, rsd-6 and prg-1 mutants (Heestand et al., 2018; Sakaguchi et al., 2014; Simon et al., 2014). Given that very low levels of transposition occur in late-generation prg-1 mutants (Simon et al., 2014), and given that the transgenerational sterility phenotype of small RNA-mediated genome silencing mutants is reversible (Heestand et al., 2018; Simon et al., 2014; Spracklin et al., 2017), their sterility is unlikely to be a consequence of debilitating levels of transposon-induced DNA damage. Finally, high levels of transposition occur in C. elegans mutator mutants but this is not sufficient to induce transgenerational sterility at permissive temperature (Simon et al., 2014).
Another proposed cause of sterility in C. elegans small RNA genome silencing mutants is misrouting of small RNAs that regulate the epigenome (de Albuquerque et al., 2015; Phillips et al., 2015). Some Argonaute proteins that regulate epigenomic pathways are located in perinuclear germ granules, which are conserved germline structures akin to liquid droplets that interact with nuclear pore complexes and are necessary for fertility (Updike et al., 2011). C. elegans germ granules are termed P granules based on antibodies that label the P blastomere that generates the germ cell lineage (Strome and Wood, 1982). Core P granule proteins have been shown to play redundant functions making it difficult to eliminate P granules (Updike et al., 2014). Here we study the late-generation sterility phenotype of small RNA-mediated genome silencing mutants and find that it is linked to dysfunction of P granules, a mechanism that could be generally relevant to the regulation of stress-induced reproductive arrest and to the long-standing observation that desilencing of heterochromatin can induce sterility.
Results:
Genome instability in small RNA mutants
Transgenerational sterility in C. elegans telomerase mutants that lack the ability to maintain sequences that cap chromosome ends is caused by high levels of telomere fusions and associated genome damage (Ahmed and Hodgkin, 2000). We examined prg-1 mutants grown at 20°C across several generations for the presence of the DNA damage response, as detected by an antibody to phosphorylated S/TQ, a protein modification that is created by two sensors of DNA damage, the phosphatadylisonisol-3-kinase like protein kinases ATM and ATR (Kim et al., 1999; Vermezovic et al., 2012). We observed a significant increase in the fraction of germ cells with upregulated DNA damage response for late-generation fertile prg-1 mutants (~3-fold) and sterile prg-1 mutants (~6-fold) when compared to either wild-type controls or to early-generation prg-1 mutants (Fig. 1a,b). We also tested temperature-sensitive small RNA mutants and found that the wild-type DNA damage response was upregulated in many cells by growth at 25°C but that this level did not change in sterile rsd-6, nrde-2 or hrde-1 mutant germlines (Fig. 1b,c). Therefore, although the DNA damage response is induced in late-generation prg-1 mutants, it is not consistently altered in sterile small RNA genome silencing mutants.
Small RNA mutants exhibit P- granule defects at sterility
Deficiency for PRG-1 causes delocalization of the P granule protein PGL-1 (Phillips et al., 2015), and combined loss of the epigenomic regulators SPR-5 and LET-418, or loss of H3K9 methylation modifier SET-2 and the nuclear RNAi pathway, results in immediate sterility that is associated with disruption of P granules (Käser-Pébernard et al., 2014; Robert et al., 2014). Whilst characterizing germ line defects of sterile rsd-6 mutants by immunofluorescence, we found that a P granule antibody exhibited altered staining patterns in sterile but not fertile late-generation rsd-6 mutant siblings (Fig 2d-d’). We also found that sterile late-generation hrde-1 and nrde-2 temperature-sensitive mutant germ lines displayed P granule defects (Fig 2c-c’, S1), which were absent or comparatively minor in fertile late-generation mutants that were close to the generation of sterility (Fig. 2c-c’, S1, Table 1). When compared to wild-type controls, P granule dysfunction in the sterile mutants ranged from a loss of P granules in a subset of cells to varying amounts of P granule disorganization (Fig. 2, S1). Sterile hrde-1(tm200) mutant animals displayed the most severe P granule phenotype, as the majority of sterile hrde-1 mutant germ cells exhibited little to no staining (Fig. S1). We also examined the temperature sensitive mrt mutant rbr-2(tm1231), which encodes a histone 3 lysine 4 demethylase that promotes genomic silencing and likely functions downstream of small RNAs to promote germ cell immortality (Alvares et al., 2014). Similarly to hrde-1 and nrde-2, late generation fertile rbr-2 animals did not exhibit obvious P granule defects, while loss of P granules was observed in many germ cells at sterility (Fig. 2e-e’).
Germ line atrophy occurs in some but not all prg-1 mutants as L4 larvae develop into sterile adults (Heestand et al., 2018). Large-scale apoptosis contributes to this germ cell degeneration phenotype (Heestand et al., 2018), and P granules have been shown to vanish in cells that undergo apoptosis (Min et al., 2016; Sung et al., 2017). However, we observed widespread P granule dysfunction in sterile germlines of rsd-6 mutants that were wildtype in size and had not undergone significant amounts of apoptosis (Fig. 2d).
nrde-1 and nrde-4 animals displayed a distinct phenotype, whereby only a subpopulation displayed P granule defects (60% and 22% respectively (Fig. S1, Table S1)). This suggests that loss of nrde-1 and nrde-4, which act downstream of multiple small RNA (and possibly other epigenetic) pathways, may lead to sterility via several distinct mechanisms.
In contrast to most small RNA genome silencing mutants where fertility defects are not severe until the generation of sterility, late-generation prg-1 mutants display very small brood sizes for many generations prior to sterility (Heestand et al., 2018). By immunofluorescence, we observed a wide range of P granule defects that could be identified in both sterile and fertile late-generation prg-1 mutant germ cells (Fig. 2b, S1). Many germlines exhibited strong P granule dysfunction that either resulted in a total absence of or an irregular distribution of P granules (Fig. 2b-b’, S1).
C. elegans P granules are adjacent to Mutator bodies, which house Mutator proteins that amplify primary siRNA populations into secondary effector siRNAs via RNA-dependent RNA polymerase (Phillips et al., 2012). mutator mutants are dysfunctional for of secondary siRNAs and remain fertile indefinitely at low temperatures (Simon et al., 2014), but display an immediate highly penetrant sterility phenotype at 25°C that could be caused by dysfunction of small RNA-mediated heterochromatin (Ketting et al., 1999). We examined sterile mut-14 mutants at the restrictive temperatures 25°C and observed pronounced P granule defects in 87% of animals (Fig. 2f-f’, Table 1).
P granule dysfunction is sufficient to elicit phenotypes of sterile small RNA silencing mutants
We initially addressed the significance of P granule dysfunction in sterile late-generation small RNA genome silencing mutants using an RNA interference clone that simultaneously targets four P granule subunits pgl-1, pgl-3, glh-1, glh-4 (Knutson et al., 2017; Updike et al., 2014). This resulted in many F2 sterile adults at 25°C (80%) (Fig. 3a). The germlines of F2 P granule RNAi L4 larvae were normal in size in comparison to control worms. However, a pronounced germ cell atrophy phenotype occurred as many P granule-depleted animals matured into sterile adults, resulting in a significant change in their germline profile (Fig. 3b, p=3.93E-24). Furthermore, oocyte univalents were previously observed in sterile but not fertile late-generation rsd-6 mutants (Sakaguchi et al., 2014), and 45.5% of oocytes of sterile P granule depleted adults contained 7-12 univalents (Fig. 3c, e-h).
Deficiency for pgl-1 phenocopies of reproductive arrest of sterile prg-1 mutants
We considered that the germline atrophy phenotypes observed in P granule-depleted worms might mimic the Adult Reproductive Diapause phenotype of sterile prg-1 mutant adults, which can regrow their atrophied germlines and become fertile (Heestand et al., 2018). To test this hypothesis, we used a mutant deficient for the P granule component pgl-1, which displays a pronounced sterility phenotype if shifted to the restrictive temperature of 25°C (Kawasaki et al., 1998). While most first generation F1 pgl-1 animals at 25°C displayed normal germlines at the L4 larval stage, there was a significant shift in the germline profiles with 52% of day 1 adults showing germline atrophy (Fig. 3b) (p=1.04E-35), similar to P granule RNAi knockdown and sterile prg-1 mutants (Heestand et al., 2018). Furthermore, when oocytes of sterile pgl-1 mutant adults were scored, we found that 56.7% of oocytes had univalents, in agreement with the oocyte univalents observed in response to P granule depletion by RNAi (Fig. 3c).
At 25°C, some F1 pgl-1 mutants were fertile and gave rise to maternally depleted F2 animals that were uniformly sterile. F2 L4 larvae had germline profiles that displayed more severe germ cell proliferation defects than those of F1 L4 larvae, and day 1 adults showed a striking 54% increase in empty germlines (Fig. 3d) (p=2.44E-05). Therefore, germlines of F1 progeny of pgl-1 mutant mothers that were shifted to the restrictive temperature mimicked the germ cell degeneration phenotypes of sterile small RNA genome silencing mutants as L4 larvae matured into adults, whereas F2 generation pgl-1 mutant larvae displayed more pronounced germ cell proliferation defects.
We asked if the fertility of sterile 25°C F1 pgl-1 mutants could be restored by shifting them to the permissive temperature of 20°C. Sterile F1 pgl-1 mutants that lacked embryos were shifted to 20°C on day 1 of adulthood and their germlines were scored after 48 hours, which revealed a significant change in their germline profiles (p=0.007), with a shift from atrophied germlines towards large germlines when compared to sterile F1 pgl-1 sibling controls that remained sterile at 25°C (Fig. 3i). In agreement with this amelioration of germline size, we observed that some sterile F1 pgl-1 mutant adults that were shifted to 20°C for several days had laid oocytes and dead embryos, and that a few even gave rise to progeny (Fig. 3j). In contrast, sterile pgl-1 mutant control adults that were kept at 25°C never gave rise to oocytes, dead embryos or living progeny (Fig. 3j). The frequency of fertility observed in this circumstance is very similar to sterile prg-1 mutant adults that exit Adult Reproductive Diapause in response to an alternative food source (Heestand et al., 2018). These results indicate that P granule dysfunction could induce Adult Reproductive Diapause, a transient and potentially adaptive form of reproductive arrest.
Transcriptional consequences of germ granule dysfunction and small RNA genome silencing mutant sterility
The sterility of Piwi/piRNA mutants is thought to result in upregulated expression of transposons and associated genome instability (Heestand et al., 2018). In an effort to deduce how germ granule dysfunction might occur in small RNA genome silencing mutants, we sequenced RNA from the nuclear RNAi defective mutants nrde-1, nrde-2 and nrde-4. These mutants are all defective in nuclear silencing in response to exogenous dsRNA triggers (Buckley et al., 2012; Guang et al., 2010, 2008), and at 25°C they all have Mortal Germlines (Burkhart et al., 2011). However, at 20°C nrde-1 and nrde-4 mutants become progressively sterile but nrde-2 mutants remain fertile indefinitely (Burkhart et al., 2011). We focused on nrde-1 and nrde-4 mutants because they have relatively large brood sizes at sterility at 20°C, such that large cohorts of L4 larvae that are poised to become sterile can be collected. In contrast, prg-1 mutants develop very low brood sizes in late generations (Heestand et al., 2018), and it is difficult to obtain large numbers of synchronous animals that are poised to become sterile. We therefore prepared RNA from early-generation and sterile-generation nrde-1 and nrde-4 mutants at 20°C and compared this with RNA from wildtype and nrde-2 mutant controls.
We found that sterile generation nrde-1 and nrde-4 mutant L4 larvae showed strong upregulation of similar transposon classes in comparison to early generation nrde-1 or nrde-4 mutant L4 larvae or to wildtype controls (Fig. 4a). We identified 18 transposon classes that were upregulated at least two-fold in both nrde-1 and nrde-4, 13 of which were CER retrotransposons, which is consistent with previous transgenerational analysis of hrde-1 mutants (Ni et al., 2016). However, we found that nrde-2 mutant controls that do not become sterile at 20°C displayed strong upregulation of similar transposon loci, even in early generations. Of the 18 transposons upregulated in both nrde-1 and nrde-4, 17 were also upregulated in nrde-2 compared with wildtype. As a comparison, we analyzed published RNA-seq data from prg-1(n4357) mutant animals (McMurchy et al., 2017). Previous papers have reported that distinct classes of transposons are upregulated in prg-1 and nrde-2 mutants (Das et al., 2008; McMurchy et al., 2017; Ni et al., 2016; Simon et al., 2014), and we confirmed this result by showing that MIRAGE1 was the only transposon upregulated in both nrde-1, nrde-4 and prg-1 mutants (Fig. 4a).
We compared gene expression data from sterile generation nrde-1 or nrde-4 mutant L4 larvae with late-generation prg-1 mutants (McMurchy et al., 2017) and compared this with RNA from wildtype or nrde-2 mutant control L4 larvae in order to identify common genes whose expression is specifically induced in L4 larvae that are poised to become sterile and that might explain the P granule abnormalities at sterility. Late-generation prg-1 mutant and sterile generation nrde-1 and nrde-4 mutant gene expression was very different from wildtype or nrde-2 mutant controls RNA.
Spermatogenic genes were over-represented among genes significantly upregulated in sterile generation nrde-1 and nrde-4 mutant larvae (p = 1e-15, Chi Square test for both mutants), which is consistent with what has been previously reported for spr-5 mutants that become progressively sterile (Katz et al., 2009). However, there was no general tendency of an up- or down-regulation of any gene category for prg-1, nrde-1 and nrde-4 mutants. Only 11 genes were significantly upregulated in all three mutants, and there were no commonly downregulated genes (Fig. 4c). We identified a set of 20 genes that were significantly upregulated at least fourfold in prg-1 and either nrde-1 or nrde-4. which included many pud (protein upregulated in dauer) genes that could be an indicator for genes upregulated due to a general stress response of the organism (Table S2). This may be consistent with our recent observation that DAF-16, which promotes dauer formation, is responsible for several phenotypes of sterile generation prg-1 mutant adults, including germ cell atrophy (Heestand et al., 2018).
Knockdown of a number of genes has been shown to disrupt P granule structure in a manner that could be similar to what we observe in sterile small RNA mutant adults (Updike and Strome, 2010). We compared the expression of genes known to perturb P granule structure with genes whose expression is significantly altered in late-generation prg-1 mutants or in sterile generation nrde-1 and nrde-4 mutant L4 larvae. We could find no consistent changes in P granule associated genes in any of the mutants tested.
RNA-seq data sets have been previously obtained for germlines of P granule defective L4 larvae and adults (Knutson et al., 2017). In P granule defective adults, somatic genes are overexpressed because these animals display a germline to soma transformation, but germlines of L4 larvae were not significantly different from wildtype. We compared sterile generation nrde-1 and nrde-4 RNA from L4 larvae with RNA from germlines of P granule defective L4 larvae (Knutson et al., 2017) and found no genes that were up or downregulated in both nrde-1/nrde-4 and P granule defective larvae. Moreover, only two out of the 18 transposons upregulated in nrde-1/nrde-4 were also upregulated in P granule defective L4 larvae (Fig. 4a). We conclude that L4 larvae that are poised to enter reproductive arrest as a consequence of P granule dysfunction do not display a consistent transcriptional signature, nor does P granule dysfunction cause immediate overt effects on expression of heterochromatic segments of the genome.
Our analyses indicate that prg-1 and nrde-1/nrde-4 have fundamentally different gene expression profiles at or near sterility. We conclude that the sterility phenotype of prg-1 and nrde-1/nrde-4 is physiologically similar but very different on a molecular level. Therefore, although deficiency for prg-1 and nrde-1/nrde-4 disrupts small RNA-mediated genome silencing and compromises germ cell immortality, these genes have distinct functions with regards to small RNA regulation of the genome such that we observed only a small number of overlapping transcriptional effects. These transcriptional differences could contribute to the more pronounced P granule defects in late-generation prg-1 mutant germ lines in comparison to the relatively mild effects on P granules in sterile nrde-1 and nrde-4 mutants.
Discussion:
P-M Hybrid Dysgenesis is a temperature-sensitive Drosophila fertility defect that was discovered in the 1970’s, where defects in transposon silencing were linked to reproductive failure (Kelleher, 2016; Kidwell et al., 1977). In these experiments, Drosophila M strain females were mated with males of the more recently isolated P strain to yield F1 hybrid progeny with high levels of P element transposition and temperature-sensitive sterility. The reduced fertility of P-M F1 hybrids has been attributed to lack of a piRNA response to specific transposable elements (Brennecke et al., 2008), and we note that the sterility phenotype of P-M hybrids involves gonad atrophy (Kelleher, 2016), which is reminiscent of the germ cell degeneration observed in sterile late-generation C. elegans small RNA genome silencing mutants (Heestand et al., 2018) (Fig 3). Defects in Piwi proteins lead to immediate sterility in Drosophila, zebrafish and mouse, accompanied by transposon expression, transposition and by increased levels of DNA damage, which implies that transposon-induced genome instability could cause sterility of Piwi mutants (Juliano et al., 2011; Kelleher, 2016; Siomi et al., 2011). However, analysis of effects of transposon copy number on P-M hybrid dysgenesis indicated that transposon-induced genome instability only explains a minor fraction of P-M hybrid sterility (Srivastav and Kelleher, 2017). Consistently, data from our transgenerational model of sterility in response to heterochromatin deterioration indicates that DNA damage is unlikely to be a central factor in small RNA mutant sterility.
We addressed the mechanism that evokes sterility by discovering pronounced P granule defects in sterile but not fertile late-generation animals (Updike and Strome, 2010). The common sterility phenotypes of pgl-1 mutants, P granule RNAi treated animals and small RNA genome silencing mutants imply that germ granule defects are likely to cause sterility in response to heterochromatin dysfunction. Furthermore, our data suggest that univalent chromosomes that are specifically observed in oocytes of sterile small RNA genome silencing animals are a secondary consequence of P granule dysfunction that indirectly disrupts a meiotic process (Hillers K.J. et al., 2017; Loidl, 2016). When RNA from sterile generation nrde-1 and nrde-4 mutant L4 larvae and from late-generation prg-1 mutants was examined, we failed to observe consistent depletion of any gene whose knockdown is known to disrupt P granule formation (Updike and Strome, 2010), suggesting that the mechanism by which P granules are disrupted in prg-1, nrde-1 and nrde-4 mutants may be post-transcriptional. In addition, a lack of commonly misregulated genes between sterile small RNA mutants and P granule defective animals suggests that the reproductive arrest that occurs in response to P granule dysfunction is orchestrated at a post-transcriptional level, at least at the L4 larval stage immediately prior to germ cell atrophy (Heestand et al., 2018). It is possible that RNA or protein components of P granules are released at sterility to promote reproductive arrest (Updike and Strome, 2010).
We found that some sterile pgl-1 mutants could recover at permissive temperature, resulting in growth of the germline and fertility (Fig. 3i-j), as previously observed for sterile late-generations prg-1 mutants whose germlines were in a quiescent state of Adult Reproductive Diapause (Heestand et al., 2018). Given that prg-1 mutant germ cells may transmit a form of heritable stress that occurs in response to heterochromatin dysfunction (Heestand et al., 2018), our results suggest that P granule dysfunction could represent a general mechanism for orchestrating reproductive arrest in the context of stress-induced Adult Reproductive Diapause. Possibly consistent with our observations, sterile mouse mutants with defects in piRNA biogenesis have been reported to display germ granule defects (Chuma et al., 2006; Zheng and Wang, 2012). We suggest that germ granule dysfunction could explain the sterility observed in P-M hybrid dysgenesis and the immediate sterility phenotype of Piwi mutants in several species (Juliano et al., 2011; Kelleher, 2016; Siomi et al., 2011). Environmental stresses such as starvation can also evoke reproductive arrest (Angelo and Van Gilst, 2009), and it is possible that altered germ granule integrity could represent a response mechanism to a variety of stresses. A more speculative consideration is that germ granule dysfunction might contribute to the rise in human infertility and associated fertility treatment in recent decades, which has been attributed to factors such as psychological stress, caffeine consumption and pollutants (Homan et al., 2007; Younglai et al., 2005).
Small RNA silencing proteins such as CSR-1, PRG-1 or HRDE-1 are localized to P granules, and it has been hypothesized that dysfunction of pro- and anti-silencing small RNA pathways associated with P granules could be the cause of sterility in small RNA genome silencing mutants (de Albuquerque et al., 2015; Phillips et al., 2015). Our data is generally consistent with this hypothesis, as physical disruption of P granule integrity is likely to generally perturb small RNA silencing pathways. However, we found that few transposons were mis-regulated in P granule deficient L4 larvae, indicating that the reproductive arrest phenotype that occurs in response of P granule disruption does not cause pronounced desilencing of many heterochromatic segments of the genome.
Our results establish a novel framework for considering inheritance in the context of epigenetic stress and how this can evoke transgenerational sterility. We suggest that the sterility we describe in response to epigenome dysfunction represents a developmental arrest mechanism to protect the genome or epigenome in times of stress. We note that transposons and viruses are central targets of the small RNA genome silencing system and are likely to be in a continuous arms race between parasite and host, where components of P granules that protect the host are likely to be targeted (Updike and Strome, 2010). It may therefore be reasonable to hypothesize the evolution of a reproductive arrest mechanism in the context of an attack on the P granule silencing system that is predicated on disrupting the integrity of P granules themselves.
Materials and Methods
Strains
All strains were cultured at 20°C or 25°C on Nematode Growth Medium (NGM) plates seeded with E. coli OP50. Strains used include Bristol N2 wild type, hrde-1(tm1200) III, rsd-6(yp11) I, nrde-1(yp4) III, nrde-2(gg95) II, nrde-4 (gg131) IV, prg-1 (n4357) I, prg-1 (tm872) I, mut-14 (pk730) V, and rbr-2(tm1231) IV.
RNAi Assay
Feeding RNAi plates harboring host bacteria HT115(DE3) engineered to express “quad” dsRNA (targeting P granules) were obtained from Susan Strome (Knutson et al., 2017; Updike et al., 2014). L1 larvae were placed onto freshly prepared feeding RNAi plates with dsRNA induced by 1 mM IPTG (isopropyl-β-D(-)-thiogalactopyranoside) and were transferred after 1 generation at 25°C and collected at F2 adults as described in Knutson et al., 2017 for DAPI staining, oocyte and germline analysis.
DAPI staining and Scoring
DAPI staining was performed as previously described (Ahmed and Hodgkin, 2000). Briefly, L4 larvae were selected from sibling plates and sterile adults were singled as late L4s, observed 24 hours later for confirmed sterility, and then stained 48 hours after collection. Univalents were scored by counting DAPI bodies in the −1 to −4 oocytes. Germline profiles were scored using the method outlined in Heestand et al 2018 (Heestand et al., 2018).
Statistical Analysis
Statistical analysis was performed as previously described (Heestand et al., 2018). Briefly, statistical analysis was performed using the R statistical environment (R Core Team, 2013). For germline phenotypes, contingency tables were constructed and pairwise Chi Square tests with Bonferroni correction was used to determine significant differences in in germline phenotype distributions. The significance of the DNA damage response analysis was determined by a Kruskal-Wallis test, followed by a Mann-Whitney test between individual samples. P-values were adjusted with Bonferroni correction when multiple comparisons were performed.
RNA extraction and sequencing
Animals were grown at 20°C on 60 mm NGM plates seeded with OP50 bacteria. RNA was extracted using Trizol (Ambion) followed by isopropanol precipitation. Library preparation and sequencing was performed at the UNC School of Medicine High-Throughput Sequencing Facility (HTSF). Libraries were were prepared from ribosome-depleted RNA and sequenced on an Illumina Hiseq 2500.
RNA-seq Analysis
The following publicly available RNA-seq datasets were download from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/): GSE92690 (P granule RNAi experiment) and GSE87524 (prg-1 experiment). Adapter trimming was performed as required using the bbduk.sh script from the BBmap suite (Bushnell, n.d.) and custom scripts. Reads were then mapped to the C. elegans genome (WS251) using hisat2 (Kim et al., 2013) with default settings and read counts were assigned to protein-coding genes using the featureCounts utility from the Subread package (Liao et al., 2014). For multimapping reads, each mapping locus was assigned a count of 1/n where n=number of hits. Differentially expressed genes were identified using DESeq2, and were defined as changing at least 2-fold with FDR-corrected p-value < 0.01. For analysis of transposon RNAs, reads were mapped to the C. elegans transposon consensus sequences downloaded from Repbase (http://www.girinst.org/repbase/) with bowtie (Langmead et al., 2009) using the options -M 1 -v 2. Transposons with fewer than 10 counts in each sample were excluded from further analysis. Counts were normalized to the total number of mapped reads for each library for the prg-1 dataset, or to the total number of non-ribosomal mapped reads for all other datasets. A pseudocount of 1 was added to each value to avoid division by zero errors. Analysis of sequencing data and plot creation was performed using the R statistical computing environment (R Core Team, 2013).
Accession numbers:
RNA-seq data reported in this study have been submitted to the GEO database and will be available at the time of publication.
Immunofluorescence
Adult hermaphrodites raised at 20°C or 25°C were dissected in M9 buffer and flash frozen on dry ice before fixation for 1 min in methanol at −20°C. After washing in PBS supplemented with 0.1% Tween-20 (PBST), primary antibody diluted in in PBST was used to immunostain overnight at 4 °C in a humid chamber. Primaries used were 1:50 OIC1D4 (Developmental Studies Hybridoma Bank). Secondary antibody staining was performed by using a Cy3 donkey anti-mouse or Cy-5 donkey anti-rabbit overnight at 4°C. All images were obtained using a LSM 710 laser scanning confocal and were taken using same settings as control samples. Images processed using ImageJ.
DNA Damage Assay
Worms that were close to sterility were isolated and defined as sterile if they did not have any offspring as day 3 adults at 20˚C or day 2 adults at 25˚C. Fertile siblings of sterile worms were defined as ‘close to sterility’. The presence of the DNA damage response was determined by using a phospho-specific antibody targeting the phosphorylated consensus target site of ATM and ATR kinases (pS/TQ) (Cell Signaling Technology). This antibody has only been shown to stain a DNA damage response in the germline and was used as previously described (Vermezovic et al., 2012).
Competing interests:
Authors declare no competing interests.
Author contributions:
K.B., B.H., M.S. and S.F. performed experiments. S.F. analyzed the data. K.B., B.H., M.S., S.F. and S.A. wrote manuscript.
Acknowledgments:
We thank members of the Ahmed lab for critical reading of the manuscript and Jacinth Mitchell for wildtype control, nrde-1 and nrde-2 RNA-seq data. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). This study was supported by NIH grants F32 GM120809 (K.B) and RO1 GM083048 (S.A). M.S. was supported by a DFG postdoc fellowship.