piRNAs can trigger a multigenerational epigenetic memory in the germline of C. elegans

Abstract

Transgenerational effects have wide-ranging implications for human health, biological adaptation, and evolution; however, their mechanisms and biology remain poorly understood. Here, we demonstrate that a germline nuclear small RNA/chromatin pathway can maintain stable inheritance for many generations when triggered by a piRNA-dependent foreign RNA response in C. elegans. Using forward genetic screens and candidate approaches, we find that a core set of nuclear RNAi and chromatin factors is required for multigenerational inheritance of environmental RNAi and piRNA silencing. These include a germline-specific nuclear Argonaute HRDE1/WAGO-9, a HP1 ortholog HPL-2, and two putative histone methyltransferases, SET-25 and SET-32. piRNAs can trigger highly stable long-term silencing lasting at least 20 generations. Once established, this long-term memory becomes independent of the piRNA trigger but remains dependent on the nuclear RNAi/chromatin pathway. Our data present a multigenerational epigenetic inheritance mechanism induced by piRNAs.

Graphical abstract

Figure 1

A Novel Inheritance Paradigm Demonstrates that Transgenerational Inheritance Is Associated with Continued Small RNA Production (A) A diagram of the Hrde sensor inheritance paradigm. Green animals illustrate the germline-expressed GFP sensor, whereas black worms represent silenced animals. (B) Representative images showing the germline-expressed transgene. The left panel shows the germline of an animal fed control vector, whereas the right panel shows the germline of an animal whose parent was treated with dsRNA targeting the GFP transgene. Arrows show the developing oocytes. (C) Graph showing the percentage of GFP-silenced animals following exposure to GFP RNAi. Wild-type worms do not contain the hrde sensor; P0-F4 animals carry the sensor and differ only in their exposure to dsRNA. GFP fluorescence of the transgene and the percentage of silenced animals per plate were determined using a large particle biosorter and FlowJo. Ten silenced worms were selected from each plate to produce the next generation. At least 1,000 worms were analyzed per plate with the following number of replicates: P0 (GFP vector) n = 3, F1 n = 18, F2 n = 11, F3 n = 8, F4 n = 5. Error bars represent the SEM. Silencing was normalized to wild-type to account for autofluorescence of the intestine. (D) qRT-PCR showing levels of nascent, unspliced pre-mRNA and mRNA for the GFP transgene in silenced, GFP RNAi treated F2 and control worms. Fold change is shown relative to control and normalized to ama-1 expression. n = 4, 3, 4, 4 for pre-mRNA control, gfp, mRNA control and gfp, respectively. (E) Small RNA reads with unique perfect match in the transgene construct and no perfect match in the reference genome are shown for P0 and F4 L4 stage animals. F4 animals are the offspring of silenced animals in previous generations. Antisense and sense reads are shown in red and blue, respectively. Profiles indicate number of reads per million. Schematic indicates the transgene structure. Blue bars are pie-1 genomic DNA, 5′ and 3′ UTR, and exon (thin, medium, thick), respectively. Thick green/yellow bars represent GFP/his-58, respectively. (F) Size distribution of small RNA reads in (E). For each size, the relative contribution of small RNAs with a particular 5′ nucleotide is represented in colors as indicated. Error bars represent SEM. See also Figures S1 and S2.

Figure 2

Transgenerational Inheritance Requires NRDE-2 and the Germline-Specific Nuclear Argonaute HRDE-1/WAGO-9 (A) Biosorter analysis of WT, nrde-2, and hrde-1/wago-9 animals showing the failure of heritable silencing in these mutant strains. GFP fluorescence of the transgene and the percentage of silenced animals per plate were determined using a large particle biosorter and FlowJo. Ten silenced worms were selected where possible from each plate to produce the next generation. At least 500 worms were analyzed per plate with the following number of replicates (empty vector, GFP RNAi, F1, F2, respectively). WT, n = 7, 6, 4, 4; nrde-2, n = 3, 3, 3, 3; hrde-1/wago-9, n = 12, 6, 6, 6. Error bars represent the SEM. (B) HRDE-/WAGO-9 is expressed in the germline. Wild-type dissected germlines (adults) were stained with anti-HRDE-1/WAGO-9 (green) and a P-granule-specific antibody (OIC1D4, red). DNA was costained with DAPI (blue). Images on the right are merged from all three channels. (C and D) HRDE-/WAGO-9 is a nuclear protein. Immunostainings were performed on dissected gonads from adult wild-type (N2) or hrde-1/wago-9 (tm1200) animals using anti-HRDE-1/WAGO-9 (green) and anti-α-tubulin antibodies (red). DNA was costained with DAPI (blue). Images on the right are merged from all three channels. Images shown are germ cells in the transition zone/pachytene region (C) and oocytes (D).

Figure 3

The Germline Nuclear RNAi/Chromatin Pathway Acts Downstream of Small RNA Production and Stability (A) Size distribution of small RNA reads with unique perfect match in the transgene construct and no perfect match in the reference genome are shown for wild-type and nrde-2 F2 animals. For each size, the relative contribution of small RNAs with a particular 5′ nucleotide is represented in colors as indicated. (B) The heterochromatin protein HPL-2 is required for piRNA sensor silencing. DIC and fluorescence microscopy of piRNA sensor germlines in indicated mutant genotypes. Scale bars, 20 μm. (C) HPL-2 acts downstream of 22G-RNA biogenesis. Northern blot of total RNA from control sensor and indicated piRNA sensor strains. Probes were against piRNA 21UR-1, a piRNA sensor-specific 22G-RNA, and the Piwi-independent endo-siRNA siR26-263. For oligonucleotide sequences, see Bagijn et al., 2012. (D and E) Antisense 22G-RNA profiles are shown for selected elements. Profiles indicate number of reads per million. piRNA target sites are indicated above each profile as explained in the color key. See also Figure S3.

Figure 4

The Germline Nuclear RNAi/Chromatin Pathway Acts in trans but Cannot Exit the Germline (A and B) trans-heterozygous animals were generated by crossing SX1866 hermaphrodites with piRNA sensor males. Strain SX1866 expressing H2B-GFP under control of the ubiquitous dpy-30 promoter was generated by MosSCI into ttTi5606 on chromosome II (mjSi1[dpy-30::his-58::gfp::tbb-2]). DIC and fluorescence microscopy of animals from the parental line (A) or of trans-heterozygous animals from the cross (B). Note that the parental line expresses H2B-GFP from two copies in the genome and is therefore brighter. Yellow arrowheads indicate germ cell nuclei; red arrowheads indicate somatic (intestinal) cell nuclei.

Figure 5

piRNAs Can Induce Stable Multigenerational Inheritance that Does Not Require PRG-1 for Maintenance (A) Schematic showing generation of prg-1; piRNA sensor strain, which has lost the requirement for PRG-1 to maintain transgene silencing (for details of previous outcrosses, see Experimental Procedures). Analogous crosses were performed for nuclear RNAi factors, with nrde-2 requiring further intermediate steps. (B) Following a number of crosses in a prg-1-sufficient background, the piRNA reporter is desilenced in nrde-1 and nrde-2, but not prg-1, mutant backgrounds. Differential interference contrast or GFP epifluorescence photos are shown. White bars correspond to 100 μm. (C) Outcrossed piRNA sensors fail to express GFP in rsd-2 or rsd-6 mutant backgrounds. Higher autofluorescence is observed for these strains, which were raised at 25°C, than for those in (B), which were raised at 20°C. See also Figures S4, S5, and S6.

Figure 6

A Model of Transgenerational Silencing in the Germline of C. elegans Triggers such as environmental RNAi and endogenous piRNAs lead to the establishment of a nuclear RNAi/chromatin pathway. Maintenance of silencing requires nuclear RNAi factors, including the germline-specific nuclear Argonaute HRDE-1/WAGO-9 and chromatin proteins such as the HP1 ortholog HPL-2 and the putative histone methyltransferases SET-25 and SET-32. Silencing can be maintained into the F1 for multiple generations (F1–F5) or can become epi-allelic with multigenerational, nonstochastic inheritance. Silencing might be suppressed by a germline licensing pathway that recognizes bona fide germline transcripts (CSR-1 22G-RNA pathway) or might be enforced through the recognition of unpaired DNA during meiosis.

Figure S1

The Hrde Sensor, Related to Figure 1A Cartoon showing the components of the Hrde sensor transgene used in the transgenerational inheritance assays (top panel) and the piRNA sensor (bottom panel). Also indicated are the locations of the primers used for qRT-PCR analysis, the location of the dsRNA feeding trigger and the location of the piRNA 21U trigger.

Figure S2

Quantification of GFP Fluorescence in the Hrde Sensor Assay Using Flow Cytometry, Related to Figure 1C Representative, smoothed histograms showing relative GFP fluorescence of control, GFP RNAi, F1 and F2 wild-type adult animals. Raw data is gated to exclude larval animals, which cannot express the transgene. GFP fluorescence is on an arbitrary scale.

Figure S3

A Nuclear RNAi and Chromatin Factor Are Required for piRNA Sensor Silencing, Related to Figure 3B DIC and fluorescence microscopy of piRNA sensor germlines in wild-type and indicated mutant genotypes. Scale bars are 20 μm.

Figure S4

piRNA Sensor Silencing Can be PRG-1 Dependent, Related to Figure 5 Schematic exemplifying a cross where introduction of the piRNA sensor into a prg-1 mutant background leads to sensor activation as described previously (Bagijn et al., 2012). Subsequent re-introduction of a wild-type prg-1 allele is not sufficient to induce inheritance of silencing in the prg-1 mutant offspring.

Figure S5

Multigenerational Inheritance of Silencing of the piRNA mCherry Sensor, Related to Figure 5 We crossed prg-1(n4357) I mutant hermaphrodites to males carrying one allele of our piRNA mCherry sensor and PCR-selected for mCherry sensor heterozygous F1 offspring. In the F2 generation we used one line each of worms homozygous for one of the desired alleles and heterozygous for the respective other allele and re-individualized F3s to obtain both alleles as homozygotes. In the F3 generation we obtained no double mutants from the prg-1(n4357) homozygous F2 but observed mCherry sensor expression in a line heterozygous for the mCherry sensor. Offspring of this F3 was individualized again and we obtained one line mutant for both alleles. This line stably expresses mCherry in the germline. Conversely, we obtained 2 F3 lines of double mutants coming from the mCherry sensor homozygous F2, both of which did not reactivate mCherry expression.

Figure S6

RSD-6 Is Required in Germ Cells for Environmental dsRNA, Related to Figure 5C Bacteria expressing dsRNA targeting pop-1 or par-6, which are expressed in the maternal germline to promote embryogenesis (Lin et al., 1995; Watts et al., 1996), cause highly penetrant embryonic lethality when fed to wild-type but not to rsd-6 mutant hermaphrodites. Single copy transgenes that express rsd-6 ubiquitously (Pdpy-30::rsd-6) or in the germline (Ppgl-3::rsd-6) both rescue the responses of an rsd-6 mutation to pop-1 or par-6 dsRNA.