The Helicase Aquarius/EMB-4 Is Required to Overcome Intronic Barriers to Allow Nuclear RNAi Pathways to Heritably Silence Transcription

Abstract

Small RNAs play a crucial role in genome defense against transposable elements and guide Argonaute proteins to nascent RNA transcripts to induce co-transcriptional gene silencing. However, the molecular basis of this process remains unknown. Here, we identify the conserved RNA helicase Aquarius/EMB-4 as a direct and essential link between small RNA pathways and the transcriptional machinery in Caenorhabditis elegans. Aquarius physically interacts with the germline Argonaute HRDE-1. Aquarius is required to initiate small-RNA-induced heritable gene silencing. HRDE-1 and Aquarius silence overlapping sets of genes and transposable elements. Surprisingly, removal of introns from a target gene abolishes the requirement for Aquarius, but not HRDE-1, for small RNA-dependent gene silencing. We conclude that Aquarius allows small RNA pathways to compete for access to nascent transcripts undergoing co-transcriptional splicing in order to detect and silence transposable elements. Thus, Aquarius and HRDE-1 act as gatekeepers coordinating gene expression and genome defense.

Keywords: C. elegans; Piwi; RNA processing; RNAi; epigenetic inheritance; nuclear RNAi; piRNA; splicing; transcription; transposable elements.

Graphical abstract

Figure 1

SILAC Proteomics Identifies Aquarius/EMB-4 as an Interactor of the Nuclear Argonaute HRDE-1 (A) SILAC labeling and IP scheme for wild-type (heavy labeled) and hrde-1(tm1200) mutants (light labeled) using anti-HRDE-1 antibodies. (B) Known protein interactions identified by the STRING database for mammalian orthologs of nuclear HRDE-1 interactors. Gray lines indicate known protein-protein interactions, red circles highlight known RNA-processing factors, blue circles highlight RNA pol II subunits, and gray circles highlight nuclear pore complex subunits. (C) Aquarius and two exon-junction complex proteins eIF4A3 and ALY are detected in HRDE-1 IPs (second column, mean log₂ fold enrichment heavy/light and number of replicates detected) and in D. melanogaster PIWI IPs (third column, number of IPs detected/total number of IPs). The effect on TE desilencing in D. melanogaster upon RNAi knockdown (fourth column, number of TEs desilenced out of four tested). (D) EMB-4 domain structure based on the mammalian homolog Aquarius and the position of emb-4 mutations used in this study. (E) Validation of protein-protein interactions between HRDE-1 and EMB-4 using anti-HRDE-1 antibodies to immunoprecipitate HRDE-1 complexes in CRISPR-tagged OLLAS-EMB-4 strain with or without RNase treatment (immunoglobulin G [IgG] negative control). (F) qRT-PCR analysis of emb-4 mRNA expression at different developmental stages (embryo to gravid adult), in animals lacking sperm (female germline fem-1(hc17)), in animals lacking oocytes (male germline fog-2(q71)), and in animals lacking a germline (soma only glp-4(bn2)). fem-1(hc17), fog-2(q71), and glp-4(bn2) are temperature-sensitive mutants and were grown at 25°C alongside the wild-type gravid adult control. Error bars represent SD. (G) Western blot analysis of EMB-4 protein levels using anti-EMB-4 antibodies for the same conditions as in Figure 2A. (H) Localization of EMB-4 protein in the germline of adult animals. See also Figures S1 and S2.

Figure 2

Aquarius/EMB-4 Is Required for the piRNA-Mediated Silencing of a Sensor Transgene (A) piRNA sensor transgene and its expression pattern in C. elegans germline. (B) Fluorescent microscope images of wild-type and mutant animal germlines with an integrated single-copy piRNA sensor transgene (germline boundaries are marked by white dotted lines). (C) qRT-PCR analysis of GFP expression in animals described in Figure 3B (two biological replicates with at least two technical replicates). See also Figure S3.

Figure 3

RNA Helicase Domain of Aquarius/EMB-4 Is Required for the Establishment of Transcriptional Gene Silencing (A) Structural alignment of the RecA1 and the pointer domains of human Aquarius (white) with C. elegans EMB-4 (red). Blown-up region shows the location of the G884R substitution found in emb-4(sa44) strain. G884R affects the loop region of the helicase domain. Green lines show possible salt bridge interactions between the amino acids. (B) Fluorescent images of wild-type and emb-4(sa44) germlines with the integrated piRNA sensor transgene (germline boundaries are marked by white dotted lines). (C and D) Scheme of genetic crosses showing the effect of emb-4(sa44) mutation during the establishment of gene silencing. (C) In the control cross, a wild-type copy of hrde-1(+) in the F1 heterozygous animals re-establishes complete piRNA sensor silencing. (D) Wild-type copy of hrde-1 fails to establish piRNA sensor silencing in the emb-4(sa44) homozygous background (number of assayed F1 progeny is indicated below each cross). See also Figure S4.

Figure 4

EMB-4 and HRDE-1 Are Required for Transcriptional Silencing of Genes and Transposable Elements (A) Log₂ fold change values of genes, which show significant expression change (adjusted p ≤ 0.05) in both hrde-1 and emb-4 mutants (151 genes in total), are plotted against each mutant background (dashed line, linear fit curve; PCC, Pearson's correlation coefficient; green dot, piRNA sensor transgene). (B) Exon-intron split analysis (EISA) for comparison of transcriptional and post-transcriptional gene expression changes. (C and D) Transcriptional and post-transcriptional gene expression changes in hrde-1 (C) and emb-4 (D) are colored as in (B). Significant gene expression changes are highlighted by larger dot size (large dots denote mRNA log₂ fold change ≥ 1, p ≤ 0.05). piRNA sensor transgene is highlighted by the green dot. (E) Model showing 22G-RNA amplification in mutator foci, which requires HRDE-1 for 22G-RNA transport and stability. (F and G) Comparison of RNA log₂ fold change in hrde-1 and emb-4 mutants with log₂ 22G-RNA density in wild-type animals (22G-RNA density = 22G-RNA count in HRDE-1 IP [wild-type]/RNA reads per kilobase per million mapped reads [wild-type]). piRNA sensor transgene is highlighted by the green dot. See also Figure S5.

Figure 5

HRDE-1 and EMB-4 Are Required for the Suppression of Multiple Transposable Element Families (A and B) Genes in Figures 4F and 4G are grouped into bins of increasing 22G-RNA density as shown on the x axis (number of genes in each bin is shown in parentheses, boxes show the lower and upper quartiles, red line shows the median, and p values of two-sample t test are shown above the box plots). Heatmap shows percent abundance of transposable elements in each 22G-RNA bin. (C) Heatmap showing RNA fold change of transposable element families in wild-type, hrde-1, and emb-4 mutants compared with mean wild-type levels. Asterisks indicate retro-element families. (D) Venn diagram summarizing the overlap of upregulated transposable element families in hrde-1 and emb-4 mutant animals as shown in (C).

Figure 6

Exemplary Upregulated Regions with mRNA, 22G-RNA, and H3K9me3 Chromatin IP Profiles in hrde-1 and emb-4 Mutants (A–F) CER9-LTR_CE/CER9-I_CE retro-elements (A), CEMUDR1 transposable element (B), Chapaev-2 transposable element (C), piRNA target gene bath-45 (D), CER15-I_CE retro-element (E), and Mirage1 transposable element (F). Repeatmasker tracks show transposable elements, and spliced EST tracks show evidence for canonical introns (UCSC genome browser). (G) H3K9me3 chromatin IP profile of piRNA sensor transgene.

Figure 7

Aquarius/EMB-4 Acts to Remove Intronic Barriers to Transcriptional Gene Silencing (A and B) 22G-RNA profiles of the three-intron (A) and the single-intron (B) piRNA sensors in wild-type, hrde-1, and emb-4 mutant animals (mean 22G-RNA abundance of three replicates) (y axis scale in wild-type animals of B is different from that of A; colors indicate different regions of the sensor transgene). (C) Silencing of the three-intron piRNA sensor (mjIs144) and the single-intron piRNA sensor (mjIs588) in wild-type and mutant animals (20 animals assayed for each condition). (D) Fluorescent microscope images of animals with the single-intron piRNA sensor (mjIs588). Germline boundaries are marked by the white dotted lines. (E and F) mRNA expression levels correlate negatively with 22G-RNA abundance and correlation increases when more exons are targeted by 22G-RNAs (gray line, all genes; black line, genes that show significant 22G-RNA change in one exon; red line, genes that show significant 22G-RNA change in all exons). The x axis shows log₂ fold change of 22G-RNAs in mutants/wild-type and the y axis shows log₂ fold change of mRNA in mutants/wild-type (correlation coefficient and p values are shown on graphs). (G) Model for Aquarius/EMB-4 function in intron-dependent transcriptional gene silencing.