Abstract
Mutation of the daf-2 insulin/IGF-1 receptor activates the DAF-16/Foxo transcription factor to suppress the transgenerational sterility phenotype of prg-1/piRNA mutants that are deficient for piRNA-mediated genome silencing. As with PRG-1/piRNAs, mutations in the nuclear RNA interference gene nrde-1 compromised germ cell immortality, but deficiency for daf-2 did not suppress the transgenerational sterility of nrde-1 or nrde-4 single mutants or of prg-1; nrde-4 or prg-1; hrde-1 double mutants. NRDE-1 and NRDE-4 promote transcriptional silencing in somatic cells via the nuclear Argonaute protein NRDE-3, which was dispensable for germ cell immortality. However, daf-2 deficiency failed to promote germ cell immortality in prg-1; nrde-3 mutants. Consistently, we found that DAF-16 activity in somatic cells suppressed the transgenerational sterility of prg-1 mutants via the SID-1 dsRNA transmembrane channel that promotes systemic RNAi as well as Dicer, the dsRNA binding protein RDE-4 and the RDRP RRF-3. We conclude that DAF-16 activates a cell-non-autonomous systemic RNAi pathway that promotes small RNA-mediated genome silencing in germ cells to suppress loss of the genomic immune surveillance factor Piwi/PRG-1.
Introduction
Small RNAs are key regulators of the transcriptome in eukaryotes, regulating RNA stability or translation in the cytoplasm and transcription in the nucleus (1). Small RNA-mediated transcriptional silencing in germ cells of metazoans is carried out by the Piwi Argonaute protein and Piwi-associated small RNAs termed piRNAs, which suppress expression of transposable elements and other genomic parasites (2). This has several consequences. First, Piwi promotes genome stability by suppressing the potentially mutagenic effects of transposition or viral integration, which could have toxic, neutral or beneficial effects (3–5). Second, it promotes heterochromatin formation at pericentromeres, which are transposon-rich segments of the genome that create large blocks of constitutive heterochromatin in order to promote accurate chromosome segregation during mitosis (6). Aside from its silencing properties, Drosophila Piwi can also promote transcriptional activation, as can the C. elegans anti-silencing Argonaute protein CSR-1 (7, 8). Drosophila Piwi can also regulate HSP90 activity and can promote mitotic stem cell function and meiosis (7, 9).
The C. elegans Piwi ortholog PRG-1 interacts with piRNAs to promote silencing of transposons and some genes. prg-1 mutants display activation of the Tc3 DNA transposon (10, 11), but the overall frequency of de novo transposition events in prg-1 mutants is very low (12). This is because C. elegans PRG-1/Piwi can act to initiate silencing of transposons or transgenes that enter the germ line, but genomic silencing is then largely maintained by Mutator proteins that utilize RNA dependent RNA polymerases (RDRPs) to promote the biogenesis of endogenous siRNAs which are 22 nucleotides long and have a G as their first nucleotide, referred to as 22G-RNAs (13, 14). Loss of Mutator-mediated 22G RNA biogenesis leads to significant levels of de novo transposition, termed a Mutator phenotype (15–17). Therefore, although PRG-1/piRNAs act in perinuclear P-granules in conjunction with neighbouring Mutator bodies to promote the biogenesis of 22G RNAs that promote genomic silencing in response to foreign nucleic acids (18), this silencing is often maintained in a manner that is independent of PRG-1/piRNAs or only partially dependent on PRG-1 (19). 22G-RNAs that are created in response to piRNAs can associate with the germline-specific nuclear Argonaute protein HRDE-1 to promote silencing of foreign transgenes (13, 14, 20). HRDE-1 and associated 22G RNAs promote genome silencing in germ cells via the Nuclear RNAi Defective (NRDE) proteins NRDE-1, NRDE-2 and NRDE-4 that stimulate the activities of histone modifying enzymes to cause genomic silencing in response to small RNAs (20–22). The nuclear RNAi pathway can target foreign transgenes for genomic silencing. Once established, this silent state can persist for many generations in the absence of piRNAs, termed RNAi inheritance (13, 14, 20, 23, 24).
Piwi mutants in many species are immediately sterile (7), whereas C. elegans prg-1 mutants are fertile but have been observed to display reduced levels of fertility at elevated temperatures (10, 11, 25). Outcrossed prg-1 mutants initially display robust levels of progeny but after many generations will become completely sterile, indicating that PRG-1/Piwi promotes germ cell immortality (12). It is unlikely that the very modest levels of transposition in prg-1 mutants (10, 11) are responsible their defect in germ cell immortality, as mutator mutants that display significant levels of transposon activity in germ cells remain fertile indefinitely if grown at low temperatures (12).
dsRNA can be administered to C. elegans either through uptake from the gut by E. coli that they feed on, direct injection, soaking, or expression of hairpin-forming transgenes or through uptake from the gut by E. coli that they feed on (26). This exogenous RNAi pathway requires the Dicer nuclease DCR-1 and its dsRNA binding cofactor RDE-4 (1). RNAi-induced in response to exogenous dsRNA triggers can spread throughout the animal and be passed to the next generation via germ cells (27). Similar to the transgene silencing above, RNAi-induced silencing of endogenous loci can persist for many generations (28). Transmission of RNAi between cells depends on the Systemic Interference Defective (Sid) genes, which include the SID-1 transmembrane protein that promotes import of short dsRNA fragments into cells (29, 30). The systemic RNAi pathway can respond to dsRNAs expressed in somatic cells to induce a persistent heterochromatic genome silencing over many generations (31). Endogenous functions of this systemic RNAi pathway are not well understood but are likely to include viral immunity (32, 33).
A second Dicer complex, the Enhanced RNAi (ERI) complex, is required for silencing of endogenous loci, including duplicated genes and transposable elements, which are silenced independently of the piRNA pathway (34–36). The ERI complex generates a cohort of primary small RNAs termed 26G RNAs, utilising the RNA dependent RNA polymerase RRF-3 (37, 38). Both classes of primary small RNAs generated by Dicer are amplified via RDRPs into effector 22G-RNA populations that bear perfect homology to their targets to promote either transcriptional or post-transcriptional silencing (16). In the absence of the 26G RNA Eri pathway, the response to exogenous RNAi is enhanced (39), likely due to release of limiting amounts of proteins such as DCR-1 and RDE-4, which are components of both exogenous RNAi and endogenous 26G pathways (37, 38).
Although deficiency for prg-1/piRNAs leads to complete sterility after growth for many generations, bouts of starvation can delay the onset of sterility (12). Intriguingly, mutation of the daf-2 insulin/IGF1 receptor (40), whose activity is suppressed by starvation, strongly suppresses the transgenerational fertility defect of prg-1 mutants (12). The DAF-16 transcription factor that functions downstream of daf-2 to promote longevity and stress resistance is required for the ability of starvation and daf-2 mutation to ameliorate the fertility defects of prg-1 mutants (12). The exact cause of the transgenerational sterility of prg-1 mutants and how this is suppressed by daf-2 mutations are unknown. However, the suppression of sterility by daf-2 mutation requires the activity of the endogenous RNAi pathway as it requires 22G-RNAs and H3K4 demethylases that promote genomic silencing (12). These results suggest that daf-2 acts to compensate for prg-1 deficiency by upregulating an alternative small RNA pathway that silences repetitive regions of the genome.
Here we demonstrate that DAF-16 suppresses the fertility defects of prg-1 by activating a systemic small RNA pathway that requires somatic and germline nuclear Argonaute proteins, as well as nuclear RNAi factors that function downstream of PRG-1/piRNAs to promote germline immortality. Therefore, an endogenous function of systemic RNAi in C. elegans is to promote genomic silencing in the germline if the Piwi/piRNA germline immune response system is compromised.
Results
A nuclear silencing process promotes germ cell immortality
tir-1 encodes a toll-like receptor and is required for innate immunity (41–43). Strains containing the tir-1(ky388) mutation displayed sterility after growth for a few generations but can be rejuvenated after outcrosses with N2 wildtype (C. Bargmann, personal communication). This suggested a defect in germ cell immortality, but strains deficient for independent mutations in tir-1, tm1111 and tm3036, failed to become sterile (n=2 strains per allele, 30 generations of growth), suggesting that tir-1(ky388) contained a linked mutation responsible for its Mortal Germline phenotype. Three-factor crosses with + tir-1(ky388) + / unc-93 + dpy-17 heterozygotes revealed a novel mortal germline mutation yp4 located just to the left of tir-1 (Fig 1A,C-D, S1 Table).
Analysis of ethylmethane sulphonate-induced mrt mutants revealed one mutation, yp5, with a map position similar to that of yp4. unc-93 yp4 + / + yp5 dpy-17 but not unc-93 + + / + yp5 dpy-17 or unc-93 yp4 + / + + dpy-17 trans-heterozygotes resulted in progressive sterility (n=4, 5 and 6 strains propagated, respectively) (Fig 1A, S1 Table), indicating that yp5 fails to complement yp4 and that their Mrt phenotypes are likely to be caused by mutations in the same gene. Sequencing of ~120 kb in the genetic map interval of yp4 revealed 4 mutations in yp4 mutant genomic DNA, one of which caused a stop codon at amino acid Q692 in the predicted protein product of C14B1.6 (Fig 1B). Genomic DNA of yp5 possessed a G to T point mutation that in C14B1.6 is predicted to cause a G650D substitution in an amino acid that is conserved in C14B1.6 homologs in closely related nematodes (Fig 1B). BLAST analysis failed to reveal homologs of this protein in more distantly related organisms. While this work was in progress, C14B1.6 was defined to encode the nuclear RNA interference gene nrde-1, a nuclear RNAi protein that promotes histone methylation at loci targeted for silencing by exogenous dsRNA triggers (21).
Most nuclear RNAi proteins are likely to be ubiquitously expressed in C. elegans tissues and function in both germ and somatic cells (20–22). Independent single copy insertions of Ppgl-3::nrde-1::nrde-1 3’UTR (44), a nrde-1 transgene driven by the germline-specific promoter of pgl-3 (45), rescued the Mortal Germline phenotype of nrde-1(yp4), indicating that NRDE-1 functions in germ cells to promote germ cell immortality (Fig 1C-E).
To further understand the role of nuclear RNAi in germline immortality, we outcrossed mutations in the remaining components of the Nrde pathway, which promote transcriptional silencing in somatic cells in response to exogenous dsRNAs, and tested these for defects in germ cell immortality. Like NRDE-1, deficiency for a second nuclear RNAi protein with no homologs outside of closely related nematodes, NRDE-4 (21), resulted in progressive sterility at 20°C (average 50% transgenerational lifespan of 10.22 generations n=4 experiments) (Fig 1C, 2A-B, S1 Table). nrde-2 mutants remained fertile indefinitely at 20°C but became progressively sterile if propagated at 25°C (Fig 1C-D, S1 Table) (22), whereas deficiency for the somatic Argonaute protein required for nuclear RNAi, NRDE-3 (46), resulted in strains that could be propagated indefinitely at any temperature (Fig 1C-D, S1 Table). However, deficiency for the germline nuclear Argonaute protein HRDE-1 (20) led to progressive sterility only at 25°C (Fig 2C, S1 Table). In conclusion, the germline components of the nuclear RNAi pathway, NRDE-1, NRDE-2, NRDE-4 and HRDE-1, promote germ cell immortality. While this study was in progress, analogous conditional and non-conditional defects in germ cell immortality were reported for strains with defects in the nuclear silencing proteins HRDE-1, NRDE-1, NRDE-2 and NRDE-4 mutants (20).
nrde factors are required for daf-2 to suppress the fertility defect of prg-1
The germline nuclear Argonaute protein HRDE-1, and the nuclear RNAi factors NRDE-1, NRDE-2 and NRDE-4 act downstream of PRG-1/piRNAs to promote dsRNA-induced inheritance of transgene silencing in germ cells over many generations. Therefore, we asked if the Mortal Germline phenotypes of prg-1 and nrde-1 or nrde-4 are additive. prg-1 single mutants had a significantly longer transgenerational lifespan when compared to either nrde-1 or nrde-4 (p<0.003 all comparisons, Fig 2A, S2 Table). When prg-1 was combined with nrde-4, the transgenerational sterility phenotypes of these double mutants were not significantly shorter than nrde-4 single mutant controls (p=1, Fig 2A, S2 Table). Additionally, when prg-1 was combined with nrde-1, the transgenerational lifespan was not like prg-1, but short like nrde-1, often with prg-1; nrde-1 double mutants having shorter transgenerational lifespans than nrde-1 or nrde-4 single mutants (Fig 2A, S1 Table, S2 Table). We conclude that although prg-1 and the nrde genes promote epigenetic silencing in germ cells and germ cell immortality, that nrde genes may function at least partially in parallel with prg-1 in maintaining germ cell immortality. Consistently, PRG-1/Piwi and the NRDE genes have recently been suggested to silence partially overlapping segments of the genome, for example transposon classes, possibly due to a form of redundancy that has been built into the C. elegans genome silencing system (47).
Given that prg-1 deficiency can be suppressed by daf-2, we combined daf-2 with nrde-1 or nrde-4, to determine if daf-2 can suppress the fertility defect of the nrde mutants. However, daf-2; nrde double mutants still became sterile similar to nrde single mutant controls, as did prg-1; daf-2; nrde-4 triple mutants (Fig 2B, Fig S1, S1 Table, S2 Table). We also tested the germline nuclear Argonaute protein that functions upstream of NRDE-1 and −4 to promote transgene silencing in the germline, HRDE-1 (13, 14, 20, 24). Given that hrde-1 mutants do not become sterile at 20°C (Fig 2C) and given that daf-2 mutants arrest as dauer larvae at 25°C (48), we created prg-1; daf-2 hrde-1 triple mutants and found that these became sterile (Fig 2C, S1 Table). Therefore, the germline silencing function of the nuclear RNAi pathway is required for daf-2 to suppress the fertility defects of prg-1 mutants. We next tested the somatic branch of the nrde pathway using NRDE-3, a nuclear Argonaute protein that is specific to somatic cells and is not required for germ cell immortality (Fig 1C-D, S1 Table). We found that prg-1; daf-2; nrde-3 triple mutants displayed transgenerational sterility similar to that of prg-1; nrde-3 (p=1, Fig 2C, S1 Table, S2 Table). Therefore, both somatic and germline branches of the nuclear RNAi pathway are required for daf-2 deficiency to suppress the fertility defects of prg-1. We further analysed this somatic RNAi pathway by characterizing GFP::NRDE-3 expression (46) in embryos of wild-type, prg-1, daf-2 and prg-1; daf-2 mutants. While all wild-type and prg-1 mutants displayed GFP::NRDE-3 mainly in the nucleus, daf-2 and prg-1; daf-2 mutants showed a more frequent localization of GFP::NRDE-3 to the cytoplasm and an absence of clear nuclear localization for the same embryonic stages (Fig 3A-B). Given that NRDE-3::GFP becomes more nuclear in response to exogenous dsRNA triggers that induce nuclear silencing (46), this suggests that a compromised DAF-2/insulin/IGF1 pathway may influence endogenous levels of nuclear small RNA silencing in the soma.
DAF-16 acts in somatic cells to suppress prg-1
When daf-2 is deficient, the DAF-16 transcription factor functions in a cell-non-autonomous manner in somatic cells to promote longevity (49, 50). To better characterize the mechanism whereby the DAF-16 pathway suppresses the fertility defect of prg-1 mutants, we tested independent repetitive DAF-16 transgenes that rescue the longevity defect caused by deficiency for daf-16, muIs71 [GFP::daf-16a (bKO)], which was constructed from a large genomic clone of the daf-16 locus (50), and a second transgene zIs356 [daf-16a::GFP] that contains the cDNA for daf-16a fused to GFP (51). We found that 4 combinations of prg-1 daf-16; daf-2 strains containing either daf-16 transgene could be propagated indefinitely, as observed for prg-1; daf-2 mutants (6 allele combinations, p=1 all comparisons, Table S2) and in stark contrast to the progressive sterility phenotype of prg-1 daf-16; daf-2 strains (8 allele combinations, p<0.0002 all comparisons, Table S2) (Fig 4A, S1 Table, S2 Table).
muIs71 and zIs356 are repetitive transgenes that are commonly silenced in germ cells via ‘cosuppression’ (52, 53), a small RNA-based phenomenon that silences both repetitive transgenes and as well as endogenous loci that are homologous to the transgene and foreshadowed the discovery of RNA interference (54, 55). PRG-1/piRNAs play roles in initiation and in some cases maintenance of transgene silencing (13, 14, 19, 24). We therefore sought to confirm the model that somatic DAF-16 is responsible for the mechanism by which daf-2 mutation suppresses the transgenerational fertility defect of prg-1. We found that animals that solely contain the zIs356 DAF-16 transgene fail to express DAF-16::GFP in germ cells of otherwise wildtype animals or of prg-1 daf-16 and prg-1 daf-16, daf-2 mutant animals (Fig 5A-F). These results imply that daf-16 acts cell-non-autonomously in somatic cells in order to mediate the suppression of prg-1. However, we were unable to rescue prg-1 daf-16 daf-2 sterility with daf-16 transgenes expressed in the intestine (ges-10 promoter, muEx227) or the muscle (myo-3 promoter, muEx212) (Fig 4B, S1 Table) (50). Although DAF-16 has been shown to function cell-non-autonomously in the intestine or nervous system to promote somatic longevity (50), we found that expression of DAF-16 in the intestine was not capable of promoting the fertility of prg-1 mutants. This indicates that the role of DAF-16 in repressing the fertility defects of prg-1/piRNA mutants is likely to be distinct from its effects on somatic longevity, which we confirmed for daf-2; rde-4 double mutants in the accompanying manuscript (Heestand, Simon, Frenk et al).
DAF-16 activates a cell-non-autonomous RNAi pathway to suppress prg-1
We showed previously that mutation of daf-2 eliminates the fertility defects of prg-1 mutants by activating via the mutator genes mut-7 and rde-2, which create 22G-RNAs that directly promote genomic silencing (12). Given that DAF-16 acts in the soma to promote fertility of prg-1; daf-2 double mutants, we asked if DAF-16 might upregulate a systemic RNAi pathway in the soma to promote the fertility of prg-1 mutants. The SID-1 transmembrane protein promotes systemic RNAi by importing short dsRNA fragments that are intermediates in systemic siRNA responses (29, 30, 56). We therefore constructed prg-1; daf-2; sid-1 triple mutants, which displayed a transgenerational sterility phenotype similar to prg-1 single mutants (Fig 4C, Table S1). These results indicate that DAF-16 acts in somatic cells to upregulate a cell-non-autonomous RNA interference pathway that suppresses the sterility of prg-1 mutants.
The 26G-RNA pathway is an endogenous RNA interference pathway that relies on 26 nt primary siRNAs that possess a 5’ guanine (37, 38). Given the role of sid-1 in suppression of the Mortal Germline (Mrt) phenotype of prg-1, we asked if genes that promote the 26G-RNA RNAi pathway were required for DAF-16 to promote fertility of prg-1 mutants, typically by testing independent mutations per 26G-RNA pathway gene. We found that the exonuclease ERI-1 and the Argonaute protein ERGO-1, which are required for 26G-RNA biogenesis, were dispensable for suppression of prg-1 mutant fertility defects by DAF-16 (Fig 4C, Table S1). Several proteins that promote 26G RNA biogenesis were required to rescue of the transgenerational fertility defect of prg-1 mutants, including the C. elegans Dicer orthologue DCR-1, its interacting partner the dsRNA binding protein RDE-4 and the RRF-3 RNA-dependent RNA polymerase that creates primary siRNAs for the endogenous 26G RNAi pathway (Fig 6A, S1 Table) (37, 38). The exogenous RNAi pathway requires the Argonaute protein RDE-1, which promotes biogenesis of primary siRNAs from exogenous dsRNA (1), was not required to suppress the fertility defects of prg-1 mutants (Fig 4C, S1 Table). Therefore, some but not all genes that promote the biogenesis of 26G siRNAs are required for DAF-16 to suppress the fertility defect of prg-1 mutants.
Analysis of dsRNA and siRNA levels in response to daf-2 deficiency
The requirement for DCR-1 and RDE-4 in the suppression of sterility in prg-1; daf-2 double mutants suggests the involvement of a dsRNA intermediate, because RDE-4 interacts with its targets via a dsRNA interaction domain (57). Adenosine deaminases modify double stranded RNAs by converting adenine to inosine, which results in dsRNA destruction (58) and therefore precludes dsRNA processing into 1° siRNAs by RDE-4/DCR-1 and subsequent creation of 2° 22G RNAs (59). To investigate potential daf-2-dependent changes in dsRNA processing, we re-analysed previously published daf-2 RNA-seq data (60) and lists of previously defined daf-16-dependent loci based on RNA microarrays (61). Loss of adenosine deaminase activity results in targeting of endogenous dsRNA by the RNAi machinery (59), which is one possible mechanism by which loss of daf-2 could drive changes in dsRNA processing. However, we found that the C. elegans adenosine deaminases adr-1 and adr-2 were unchanged in daf-2 mutants when compared to wildtype (S3 Table), indicating that transcriptional repression of these genes is not responsible for altered dsRNA metabolism in daf-2 mutants. We confirmed this by asking if small RNAs derived from dsRNA loci are generally upregulated by mutation of daf-2. We studied previously published small RNA datasets for prg-1 mutants (generations F4, F8 and F12) in comparison with prg-1; daf-2 double mutants (12) for changes in 22G RNA abundance at 1523 double-stranded RNA loci in the C. elegans genome (59). We found 33 dsRNA loci for which abundance of 22G RNAs is increased at least twofold in prg-1 daf-2 animals compared with prg-1 in at least one generation. 10 of these loci showed consistent 22G RNA increases for all generations (Fig. 6A, 6B). We also found 18 loci that displayed at least a twofold decrease in the levels of 22G RNAs in prg-1 daf-2 versus prg-1 in at least one generation, and eight loci with consistently decreased 22G RNA levels across all generations (Fig. 6A). Most dsRNAs were found to occur in introns of protein-coding genes (59). The observed differences in siRNA abundance upon loss of daf-2 could lead to alterations in the expression of the respective genes. We looked for concordance between siRNA abundance and full-length transcript abundance by dsRNA comparing loci with differentially regulated 22Gs with a list of genes that were found to be significantly upregulated in late-generation prg-1 but not in late-generation prg-1 daf-2 by tiling microarray (12). Only two out of the 206 identified genes overlapped with dsRNA loci (Y69A2AL.2 and Y105C5A.13), and neither locus displayed consistently upregulated siRNAs in prg-1 daf-2 double mutants.
Dicer has been shown to passively bind but not process some endogenous RNA targets in C. elegans and in human cells (62). Of the 2508 C. elegans DCR1-bound loci detected, 367 (~15%) overlapped with dsRNA loci. Only three of these DCR-1-associated dsRNA loci (F45H10.1, F52B11.1 and H24K24.4) showed consistently increased siRNAs in prg-1 daf-2 double mutants (Fig. 6B), indicating that dsRNAs that are passively bound to DCR-1 do not become actively processed in prg-1 mutants that are deficient for daf-2.
daf-2 mutants have been previously shown to be hypersensitive to exogenous RNAi, and our work with prg-1 has defined an endogenous RNAi pathway that daf-2 stimulates. We looked for daf-2 dependent expression changes of genes associated with RNAi or transcriptional silencing in the daf-2 reference data described above.
Analysis of the RNA-seq data revealed a greater than two-fold downregulation of the putitative H3K79 methyltransferase dot-1 and a greater than twofold upregulation for three histone genes (his-12, his-16, his-43), the putative membrane transporter pgp-4 and T23G4.2, a homologue of the human mitochondrial lon pepitdase 1. However, these genes were not previously identified as DAF-16 targets (60, 61). We therefore conclude that loss of daf-2 does not trigger strong changes to transcription of DAF-16 targets known to be involved in RNAi or transcriptional silencing pathways.
Discussion
We found that although deficiency for daf-2 can suppress the fertility defects of prg-1 mutants, it was not able to suppress the transgenerational sterility defect of nrde-1 or nrde-4 mutants, nor of prg-1; nrde double mutants. This indicates that DAF-16 suppresses the sterility of prg-1 mutants via a small RNA-mediated genome silencing response within the nucleus. This is consistent with previous results that showed that daf-2 mutation requires secondary 22G small RNAs generated by mut-7 and the H3K4 histone demethylase spr-5 that promote genomic silencing (12). We studied this transcriptional silencing response to reduced insulin signalling and found that a somatic function of the NRDE pathway was required for its ability to suppress the sterility of PRG-1/Piwi. Two C. elegans somatic siRNA pathways have been previously described: the 26G and the anti-viral/exogenous RNAi response pathways (1). Our results indicate that neither of these pathways is responsible for suppressing the fertility defect of prg-1 mutants. However, given that RDE-4 and DCR-1 were required, we suggest that the dsRNA-binding properties of RDE-4 mean that an endogenous form of dsRNA is the initial trigger in this DAF-16 endogenous RNAi pathway.
Based on our data, we propose the following model (Fig 7): DAF-16 activates expression of a limiting factor in an endogenous RNAi pathway that is not typically systemically activated; this could correspond to transcription or processing of a dsRNA intermediate that triggers a systemic RNAi response, or possibly to the activation or expression of a protein that promotes small RNA-mediated genome silencing in somatic cells. RRF-3 likely acts in somatic cells to create 1° siRNAs from dsRNA that associates with RDE-4/DCR-1. These primary siRNAs associate with target RNAs and the RDRP RRF-1 is recruited to generate de novo secondary 22G RNAs that associate with somatic NRDE-3, which may enter the nucleus to promote a nuclear RNAi response that could result in amplification of siRNAs (6) to generate siRNA-related dsRNA intermediates that are imported into germ cells by SID-1, thereby triggering a second round of 22G siRNA biogenesis that we suggest might be tertiary (30) siRNAs (Fig 7C), which have been recently observed in another context (63). However, we found that NRDE-3::GFP displayed a diffuse homogenous cytoplasmic and nuclear localization in response to mutation of daf-2, such that more embryos had mainly cytoplasmic or no distinct nuclear GFP::NRDE-3 expression. In contrast, loss of the dsRNA degrading adenosine deaminases adr-1 and adr-2 results to high levels of endogenous dsRNAs that, in rrf-3 or eri-1 mutant backgrounds that eliminate the endogenous 26G RNA pathway and nuclear NRDE-3 (46), are converted into siRNAs that promote strong nuclear localization of NRDE-3 (Fig. 7B) (59). Nuclear NRDE-3 can be similarly induced in response to exogenous dsRNA triggers (46). Consistent with lack of strong nuclear NRDE-3 in either daf-2 or prg-1; daf-2 mutants, we failed to see strong decreases in transcription of adr-1 or adr-2 in response to mutation of daf-2, nor did we observe consistent induction of siRNA levels in prg-1; daf-2 mutants for loci known to produce dsRNA (Fig. 6). This indicates that a general alteration to dsRNA production or processing is unlikely to explain the suppression of prg-1 sterility by daf-2 mutation. We also found that processing of the pool of dsRNAs that are passively (constitutively) bound to Dicer (62) was not generally upregulated and we failed to identify RNAi or transcriptional silencing genes whose expression was significantly altered by mutation of daf-2. We suggest that one or more dsRNA loci expressed in somatic cells are likely processed and transported to the germline to become a source of siRNAs that might substitute for the lack of piRNA-mediated genome silencing in prg-1 mutants. Interestingly, one of the most upregulated transcripts in daf-2 mutants is the long non-coding RNA tts-1, which contributes to lifespan extension by binding to ribosomes and inhibiting translation (64, 65). Though siRNAs to tts-1 were not significantly altered in prg-1; daf-2 double mutants in comparison to prg-1 mutants, we suggest that a similar non-coding RNA, whose transcription or processing into siRNAs may be upregulated by DAF-16, may promote germ cell immortality in prg-1; daf-2 mutants.
Although we addressed several plausible mechanisms by which dsRNA metabolism might be altered in response to mutation of daf-2, based on known properties of the endogenous DAF-16 small RNA silencing pathway (Fig. 7C), it remains to be understood: 1) which tissue DAF-16 acts in to promote this RNAi pathway, 2) which locus or loci in somatic cells generates small RNAs that are relevant to this pathway, 3) which locus or loci in germ cells do these dsRNAs target to promote the fertility of prg-1 mutants, and 4) if upregulation of the endogenous RNAi pathway by DAF-16 is regulated by the same mechanism that enhances the exogenous RNAi pathway in daf-2 mutants, which protects the genome against exogenous parasites such as transposons that are transmitted by horizontal gene transfer or viruses (Fig. 7A) (66).
Systemic RNAi has been very useful as an experimental tool for suppressing gene expression, and it can suppress viral infections in plants and has the potential to suppress viral infections in C. elegans for multiple generations, possibly allowing small RNA-mediated transgenerational responses to both pathogenic and non-pathogenic environmental stimuli in a manner that Jean-Baptiste Lamarck and Charles Darwin envisioned might be an illustrious tool in the crucible of evolution (5, 67). Here we define a natural function of systemic RNAi (Fig 7C), where RNAi intermediates imported by SID-1 dsRNA transporter can act to suppress the transcriptional silencing defect that compromises germ cell immortality of prg-1/Piwi mutants. This scenario may be realistic because the PRG-1/Piwi silencing system represents an innate immune system of the germline, which might be a target of viral or transposon genomic parasites that seek to suppress the endogenous defences against their expression and replication. Viruses are well known to induce stress responses, as observed for mammalian cells that are exposed to dsRNAs (68). We posit that it is such attacks, possibly common in the arms race between host and parasite, might make the stress resistance function of the DAF-16/Foxo pathway a logical mechanism by which to activate a response to silencing of PRG-1 in a manner that ensures immortality of the germ line. In this sense, somatic diapause and germ cell immortality, which can both be promoted by high DAF-16/Foxo and low DAF-2/insulin/IGF1 receptor signalling, may represent two faces of the same coin, whose common goal is to ensure evolutionary success by strengthening the soma and the germ line in times of stress.
Materials and Methods
Strains
All strains were cultured at 20°C on Nematode Growth Medium (NGM) plates seeded with E. coli O50 (69). Strains used include Bristol N2 wild type, prg-1(n4357) nrde-3(gg66), prg-1(tm872) nrde-3(gg66), prg-1(n4357) I, prg-1(tm872) I, nrde-1(yp4), nrde-1(yp5), nrde-4(gg131), prg-1(n4357) nrde-1(yp4), prg-1(n4357) nrde-1(yp5), prg-1(tm872) nrde-1(yp4), prg-1(tm872) nrde-1(yp5), prg-1(n4357) nrde-4(gg131), daf-2(e1368), daf-2(e1368) nrde-1(yp4), daf-2(e1368) nrde-1(yp5), daf-2(e1368) nrde-4(gg131), hrde-1(tm1200), prg-1(tm872) hrde-1(tm1200) daf-2(e1368), prg-1(tm872) nrde-3(gg66), prg-1(tm872) daf-2(m41) nrde-3 (gg66), prg-1(n4357) daf-2(e1370), prg-1(tm872) daf-2(e1370), prg-1(n4357) daf-16(mu86) daf-2(e1370), prg-1(tm872) daf-16(mu86) daf-2(e1370), prg-1(n4357) daf-16(mu86) daf-2(e1370) muls71, prg-1(n4357) daf-16(mu86) daf-2(e1370) zls356, prg-1(tm872) daf-16(mu86) daf-2(e1370) muls71, prg-1(tm872) daf-16(mu86) daf-2(e1370) zls356, prg-1(n4357) daf-16(mu86) daf-2(e1370) muEx211, prg-1(n4357) daf-16(mu86) daf-2(e1370) muEx212, prg-1(n4357) daf-16(mu86) daf-2(e1370) muEx227, prg-1(n4357) daf-2(e1368) sid-1(qt9), prg-1(tm872) daf-2(e1368) sid-1(qt9), prg-1(tm872) daf-2(e1370) eri-1(mg366), prg-1(n4357) daf-2(e1370) ergo-1(gg98), prg-1(tm872) daf-2(e1370) ergo-1(gg98), prg-1(n4357) daf-2(e1368) rde-1(ne217), prg-1(tm872) daf-2(e1368) rde-1(ne217), prg-1(n4357) daf-2(e1368) rde-4(ne301), prg-1(tm872) daf-2(e1368) rde-4(ne301), prg-1(tm872) rrf-3(pk1426) daf-2(e1368), prg-1(n4357) daf-2(e1368) dcr-1 please check, prg-1(tm872) daf-2(e1368) dcr-1(mg375), gfp::nrde-3, daf-2(e1368) gfp::nrde-3, prg-1(n4357) daf-2(e1368) gfp::nrde-3, prg-1(n4357) gfp::nrde-3.
Imaging and DAPI staining
DAF-16::GFP worms were dissected and fixed in 1%PFA before DAPI staining. Images were taken using immunofluorescence on Nikon Eclipse E800 microscope with a NIS Elements software. Some NRDE-3::GFP embryos were visualized as z-stacks with a LSM 710 laser scanning confocal in order to determine the total number of nuclei per embryo that was then applied to calculate the total number of embryos in single stack images. Embryos between 110-190 cell stage were selected for analysis. Images were processed using ImageJ.
Statistical Analysis
For GFP::NRDE-3 embryo phenotypes, contingency tables were constructed and pairwise Chi Square tests were used to determine significant differences in germline phenotype distributions. For trans-generational lifespan assays, pairwise log rank tests were performed. All reported p-values were adjusted using Bonferroni correction when multiple comparisons were performed.
Sterility Assays
Worms were grown by transferring six L1s to fresh NGM plates every week. They were defined as sterile when no more than six worms were found on the plate of the date of transfer.
RNA-seq analysis
The following publicly available RNA-seq datasets were download from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/): GSE40572 (prg-1 and prg-1; daf-2 small RNAs) and GSE93724 (daf-2 and wild-type mRNA). For mRNA-seq, adapter trimming was performed as required using the bbduk.sh script from the BBmap suite (70) and custom scripts. Reads were then mapped to the C. elegans genome (WS251) using hisat2 (71) with default settings and read counts were assigned to protein-coding genes using the featureCounts utility from the Subread package (72). 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 small RNA-seq data, fasta files were filtered for 22G reads using a custom python script. Reads were then converted to fasta format and mapped to the C. elegans genome (WS251) using bowtie (73) with the following options: −M 1 −v 0 - best -strata. Reads mapping to dsRNA defined in (59) were extracted using bedtools (74). Read counts were normalized by total number of mappable reads in each library and quantified using custom R scripts. A pseudocount of 1 was added after normalization to avoid division-by-zero errors when calculating log2-fold change.
To find detect changes in genes involved in RNAi or gene silencing based on the daf-2 RNA-seq data, a list of all GO terms was downladed from geneontology.org and filtered for GO terms matching the following regular expression: “RNAi|(RNA interference)|(siRNA)|silenc|piRNA”. C. elegans genes matching any of the GO terms in this set were obtained from Biomart (75) and searched against the DESeq2 output from the daf-2 mutant RNA-seq data.
Competing interests: Authors declare no competing interests.
Author contributions: M.S., M.S., A.H., M.G, A.N.S., A.S. performed experiments. S.F. performed RNA analysis and statistical analysis of the data. B.H., M.S., S.F. and S.A. wrote the manuscript. S.A. and A.S. supervised research.
Acknowledgements
We thank J. Mitchell for assistance with experiments concerning nrde-1 and members of the Ahmed lab for critical reading of the manuscript. 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 grant RO1 GM083048 (S.A).