Abstract
Small regulatory RNAs, such as small interfering RNAs (siRNAs) and PIWI-interacting RNAs, regulate splicing, transcription, and genome integrity in many eukaryotes. In Caenorhabditis elegans, siRNAs bind nuclear Argonautes (AGOs), which interact with homologous premessenger RNAs to recruit downstream silencing effectors, such as NRDE-2, to direct cotranscriptional gene silencing [or nuclear RNA interference (RNAi)]. To further our understanding of the mechanism of nuclear RNAi, we conducted immunoprecipitation-mass spectrometry on C. elegans NRDE-2. The major NRDE-2 interacting protein identified was the RNA helicase MTR-4. Co-immunoprecipitation analyses confirmed a physical association between NRDE-2 and MTR-4. MTR-4 colocalizes with NRDE-2 within the nuclei of most/all C. elegans somatic and germline cells. MTR-4 is required for nuclear RNAi, and interestingly, MTR-4 is recruited to premessenger RNAs undergoing nuclear RNAi via a process requiring nuclear siRNAs, the nuclear AGO HRDE-1, and NRDE-2, indicating that MTR-4 is a component of the C. elegans nuclear RNAi machinery. Finally, we confirm previous reports showing that human (Hs)NRDE2 and HsMTR4 also physically interact. Our data show that the NRDE-2/MTR-4 interactions are evolutionarily conserved, and that, in C. elegans, the NRDE-2/MTR-4 complex contributes to siRNA-directed cotranscriptional gene silencing.
SMALL regulatory RNAs are important regulators of genome stability and gene expression in the nuclei of many eukaryotes. The biological functions for nuclear small RNAs include repressing parasitic nucleic acids such as transposons, regulating transcription, modifying chromatin states, directing DNA methylation, coordinating genome rearrangement, silencing unpaired DNA, and promoting epigenetic inheritance. The mechanisms by which small regulatory RNAs control many of these processes are poorly understood (Castel and Martienssen 2013).
In Caenorhabditis elegans, cytoplasmically produced small regulatory RNAs enter the nuclei to regulate transcription via a process referred as cotranscriptional gene silencing (cTGS) or nuclear RNA interference (RNAi; Bosher et al. 1999; Guang et al. 2008). Forward genetic screens in C. elegans identified five proteins (NRDE-1/2/3/4 and HRDE-1) that are required for nuclear RNAi (Guang et al. 2008, 2010; Buckley et al. 2012). HRDE-1 and NRDE-3 are Argonaute (AGO) proteins that direct nuclear RNAi in C. elegans germ cells and somatic cells, respectively (Guang et al. 2008; Buckley et al. 2012). According to current models, HRDE-1 and NRDE-3 Ago proteins engage small interfering RNAs (siRNAs) in the cytoplasm and escort these siRNAs into the nuclei (Guang et al. 2008; Buckley et al. 2012). Once inside the nuclei, AGO/siRNA complexes base-pair with complementary nascent RNAs emanating from elongating RNA Polymerase II (RNAP II) complexes (Guang et al. 2008; Buckley et al. 2012). HRDE-1/NRDE-3 then recruit NRDE-2, which is an evolutionarily conserved protein possessing half a tetracopeptide (HAT) repeats often found in RNA binding proteins, to pre-mRNA (Preker and Keller 1998; Guang et al. 2010). NRDE-2 recruits NRDE-4, which then recruits NRDE-1 (Burkhart et al. 2011). The hierarchical assembly of NRDE factors onto pre-mRNAs leads to inhibition of RNAP II transcription elongation (Guang et al. 2010) and directs the deposition of repressive chromatin marks, such as H3K9me3 and H3K27me3 on chromatin of genes undergoing nuclear RNAi (Guang et al. 2010; Burkhart et al. 2011; Gu et al. 2012; Mao et al. 2015). Recent studies have linked Aquarius/EMB-4, which is a conserved RNA helicase involved in pre-mRNA splicing in mammalian cells (Hirose et al. 2006; Kuraoka et al. 2008; De et al. 2015), to nuclear RNAi in C. elegans (Akay et al. 2017; Tyc et al. 2017). Aquarius/EMB-4 physically interacts with the nuclear AGO HRDE-1 and may promote nuclear RNAi by allowing the NRDE factors to compete with splicing machinery for access to nascent RNAs (Akay et al. 2017). A complete understanding of the mechanism(s) by which NRDE nuclear RNAi factors inhibit transcription, or modify chromatin states, at the direction of small nuclear RNAs is lacking.
Eukaryotic cells possess a number of RNA quality-control systems that monitor messenger RNA (mRNA) expression, processing, trafficking, and translation. One such system is mediated by the nuclear exosome, which is composed of 11 proteins that degrade or process many RNAs produced by RNA polymerases I, II, and III (Ogami et al. 2018). Nine of the 11 exosome proteins form a structural channel through which RNAs are threaded for degradation or processing. The remaining two proteins, Rrp6 and Dis3, are catalytically active and are thought to degrade or process RNA targets of the nuclear exosome (Ogami et al. 2018). The RNA degradation targets of the exosome include RNAP II transcripts harboring splicing and/or 3’ end processing defects (Hilleren et al. 2001; Milligan et al. 2005; Lemieux et al. 2011), as well as noncoding RNAs produced by constitutive and pervasive genome transcription (Preker et al. 2008; Flynn et al. 2011; Henriques et al. 2013; Szczepińska et al. 2015; Ogami et al. 2017). The RNA processing targets of the exosome include ribosomal RNAs, total RNAs, small nucleolar RNAs, and small nuclear RNAs (Lebreton et al. 2008; Schaeffer et al. 2009; Schneider et al. 2009; Zinder and Lima 2017).
Adaptor complexes couple different RNAs to the nuclear exosome. In mammals, three adaptor complexes are known and are referred to as TRAMP (Trf4-Air-Mtr4 polyadenylation) (LaCava et al. 2005; Wyers et al. 2005; Houseley et al. 2007; Reis and Campbell 2007), NEXT (Nuclear Exosome Targeting complex, Mtr4-ZCCHC8-Rbm7) (Lubas et al. 2011), and PAXT [poly(A) tail exosome targeting, Mtr4-ZFZ3H1-PABPN1] (Meola et al. 2016; Ogami et al. 2017). The DEXD box RNA helicase Mtr4 (also SKIV2L2, Dob1, or Mtrex) is a core component of each of these complexes, where it promotes RNA degradation/processing by bridging adaptor complexes and the nuclear exosome (de la Cruz et al. 1998; van Hoof et al. 2000; LaCava et al. 2005; Wyers et al. 2005; Cristodero and Clayton 2007; Houseley et al. 2007; Reis and Campbell 2007; Schilders et al. 2007; Holub and Vanacova 2012; Thoms et al. 2015).
The nuclear RNAi factor NRDE2 and MTR4 form a 1:1 complex in mammalian cells (Richard et al. 2018; Jiao et al. 2019; Wang et al. 2019). In mammalian cells, the NRDE2/MTR4 complex protects the genome from DNA damage via an unknown mechanism (Richard et al. 2018; Jiao et al. 2019), and regulates pre-mRNA levels, ensuring pre-mRNA quality by regulating exosome-based pre-mRNA degradation (Wang et al. 2019) and/or promoting pre-mRNA splicing at genes harboring poor-quality introns (Jiao et al. 2019). How the pre-mRNA quality control and genome protection functions of the NRDE2/MTR4 complex relate to each other is not known. Interestingly, a NRDE-2-like-1 protein (Nrl1) and a paralog of Mtr4 (termed Mtr4-like-1 or Mtl1) form a complex in the fission yeast Schizosaccharomyces pombe, hinting that associations between NRDE-2-like and Mtr4-like proteins are deeply conserved (Lee et al. 2013; Zhou et al. 2015; Aronica et al. 2016). The fungal NRDE-2/Mtr4-like complex is linked to the degradation of mis-spliced RNAs via the nuclear exosome (Zhou et al. 2015), and the promotion of cryptic intron splicing via a pathway that may involve small interfering RNAi-directed chromatin modification (Lee et al. 2013). Loss of the fungal NRDE-2/Mtr4-like complex in S. pombe leads to DNA damage (Aronica et al. 2016). Thus, NRDE-2/Mtr4(-like) complexes contribute to nuclear RNA surveillance and genome protection in mammalian and fungal cells. Nonetheless, the molecular mechanism(s) by which the complex surveilles pre-mRNA production in fungi, or if this process is conserved in eukaryotes, is not understood.
Here, we show that C. elegans NRDE-2 and MTR-4 interact physically, confirming that the NRDE-2/MTR-4 complex represents a deeply conserved gene regulatory module. We also show that C. elegans (ce)MTR-4 is recruited to nascent RNAs undergoing nuclear RNAi, that this recruitment is dependent upon nuclear siRNAs and NRDE-2, and that association of MTR-4 with NRDE-2 and pre-mRNA is necessary for nuclear siRNAs to trigger cTGS. These latter data establish that the NRDE-2/MTR-4 complex is a component of the C. elegans cTGS machinery.
Materials and Methods
Strains
WT N2, YY913 nrde-2 (gg518 [nrde-2::3xflag::ha]), YY1168 mtr-4 (gg588 [3xflag::gfp::mtr-4]), YY1169 mtr- 4 (gg588); nrde-2 (gg518), YY1362 nrde-2 (gg624[ ha::tagrfp::nrde-2]); mtr-4 (gg588), CA1199 ieSi38 [sun- 1p::TIR1::mRuby::sun-1 3UTR], CA1200 ieSi57 [eft-3p::TIR1::mRuby::unc-54 3UTR], YY1540 mtr-4 (gg648 [gfp::3xflag::degron::mtr-4]); ieSi38, YY1440 nrde-2 (gg91); mtr-4 (gg588),YY1539 mtr-4 (gg648); ieSi38; nrde-2 (gg518), YY1541 mtr-4 (gg648), YY1557 mtr-4 (gg648); eri-1(mg366); ieSi57, YY1561 mtr-4 (gg648); ieSi57, YY1566 npp-9 (gg654[tagrfp::SEC::3xflag::npp-9]); mtr-4 (gg588), YY1572 hrde-1(tm1200); mtr-4 (gg588), YY1585 nrde-1 (gg88); mtr-4 (gg588)C. elegans husbandry and genetics were performed as described previously (Brenner 1974). Some strains were provided by the CGC (P40 OD010440). Some strains were made by CRISPR, as described previously (Wan et al. 2018).
Mass spectrometry
We collected ∼100,000 young adults (N2 and YY913), which were then flash frozen in liquid nitrogen. Animals were ground into powder in a mortar bathed in liquid nitrogen. Powder were resuspended in 1× lysis buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 5 mM MgCl2, 1 mM EDTA, 10% glycerol, 0.25% Triton X-100, 1 mM freshly made PMSF, 1× complete protease inhibitor without EDTA; from Roche) and rotated for 45 min to 1 hr at 4°. Lysate was cleared by centrifuging at 13,000 rpm for 10 min. Supernatant was filtered with a 0.45 μm filter unit (SLHP033RS; Millipore). Filtered supernatant was incubated with FLAG M2 antibody (F1804; Sigma-Aldrich) conjugated to Dynabeads Protein G (10004D; Thermo Fisher Scientific) for 2 hr. Beads were washed four times with a 1× lysis buffer. Proteins were eluted with 500 mM NH4OH by rotating for 20 min at 37°. Then, 10% of elution was subjected to SDS-PAGE and silver staining (1610449; Bio-Rad). Remaining protein was subjected to Trichloroacetic acid precipitation and mass spectrometry on an Orbitrap liquid chromatography tandem mass spectrometry mass spectrometer.
Co-immunoprecipitation
For co-immunoprecipitation in C. elegans, ∼20,000 young adult animals were collected and flash frozen in liquid nitrogen. Animals (YY913, YY1168, and YY1169) were resuspended in 1× lysis buffer [20 mM HEPES pH 7.5, 100 mM NaCl, 5 mM MgCl2, 1 mM EDTA, 10% glycerol, 0.25% Triton X-100, 1 mM freshly made PMSF, 1× complete protease inhibitor without EDTA (Roche)] and sonicated (30 sec on, 30 sec off, 30% output for 2 min on a Qsonica Q880R sonicator, repeated four times) to lyse. Lysate was centrifuged at 13,000 rpm at 4° for 10 min and precleared with protein A agarose beads at 4° for 30 min. For immunoprecipitation of NRDE-2, supernatants were incubated with Anti-HA Affinity Matrix (clone 3F10; Roche) for 4 hr. For immunoprecipitation of MTR-4, supernatants were incubated with GFP antibody (ab290; Abcam) for 4 hr, followed by protein A for 2 hr. Beads were washed four times with 1× lysis buffer. Input and immunoprecipitated protein were separated by SDS-PAGE and detected with HA antibody (ab9110; Abcam), GFP antibody (ab290; Abcam), and Tubulin antibody (E7; Developmental Studies Hybridoma Bank). For co-immunoprecipitation in human cells, GFP::HsMTR-4 (kindly donated by the laboratory of Stephen Elledge) was transfected into 3xFLAG::NRDE-2 HEK293T cells (Jiao et al. 2019) with lipofectamine 2000, cells were resuspended in 1× lysis buffer for 30 min, followed by sonication (30 sec on, 30 sec off, 20% output for 2 min). Immunoprecipitation and Western blotting was performed as described above.
RNA immunoprecipitation
RNA immunoprecipitations were done as described previously (Guang et al. 2010). Briefly, ∼20,000 young adults or ∼300,000 embryos were collected and flash frozen in liquid nitrogen. Samples were resuspended in sonication buffer (20 mM Tris-HCl pH 7.5, 200 mM NaCl, 2.5 mM MgCl2, 10% glycerol, 0.5% NP-40, 80U/ml RNaseOUT, 1 mM DTT and 1× protease inhibitor cocktail without EDTA) and sonicated (30 sec on, 30 sec off, 30% output for 2 min on a Qsonica Q880R sonicator, repeated once). Lysates were clarified by centrifugation at 14,000 rpm for 15 min. Supernatants were precleared with protein A agarose beads and incubated with FLAG M2 agarose beads (A2220; Sigma-Aldrich) for 2 hr at 4°. Beads were washed six times with RNA immunoprecipitation buffer (20 mM Tris-HCl pH 7.5, 200 mM NaCl, 2.5 mM MgCl2, 10% glycerol, 0.5% NP-40). Protein and associated RNAs were eluted with 100 μg/ml 3xFLAG peptide (F4799; Sigma-Aldrich). RNAs were treated with Turbo DNase I for 20 min at 37° and then extracted with TRIzol reagent followed by precipitation with isopropanol. Precipitated RNAs were reverse transcribed with iScript complementary DNA (cDNA) synthesis kit and quantified with quantitative PCR (qPCR).
Auxin treatment
Auxin treatments were done as described previously (Zhang et al. 2015). Briefly, auxin indole-3-acetic acid (A10556; Alfa Aesar) was dissolved in ethanol to make a 400 mM stock (stored at 4° for up to 1 month). Auxin was added to NGM agar or RNAi agar growth plates at a final concentration of 1 mM. Auxin treatment was done by putting embryos or transferring animals of indicated developmental stages to bacteria seeded plates containing auxin. Auxin-mediated protein degradation persists for 10–24 hr after C. elegans are removed from auxin (Zhang et al. 2015). For short-term MTR-4 depletion experiments described in this work, animals were exposed to ±1 mM auxin treatment for 5 or 10 min, or 2 hr (as indicated), and then transferred to auxin-free RNAi plates seeded with control bacteria or bacteria producing oma-1/lin-15b/lir-1 double-stranded RNA (dsRNA).
RNAi assay
RNAi experiments were performed as described previously (Timmons et al. 2001). dsRNA-expressing bacteria, including lir-1 and oma-1, were obtained from Ahringer RNAi library and sequenced to confirm their identity. lin-15b dsRNA-expressing bacteria was described previously (Guang et al. 2010). The mtr-4 RNAi clone was constructed by PCR amplification of genomic DNA with primers (CGCGGTGGCGGCCGCTCTAGA TGCACAATGATGTCGGAGTTG and GTCGACGGTATCGATAAGCTT TCATCGCGTCTCTGAATTTG), and inserted into EcoRI/HindIII linearized L4440 vector by homologous recombination in Escherichia coli DH5a. For double RNAi treatments, HT115 bacteria expressing lir-1 dsRNAs were diluted with HT115 bacteria expressing nrde-2 dsRNAs (ratio of lir-1 to nrde-2 is 1:1), or mtr-4 dsRNAs (1:0.2 to 1:1, and filled with L4440 control if necessary).
H3K9me3 chromatin immunoprecipitation
Approximately 10,000 gfp::degron::mtr-4; sun-1::tir1::sun-1 3’UTR young adult animals were put on 1 mM auxin plates for 2 h, and transferred to auxin-free plates seeded with bacteria expressing control dsRNA (L4440) or oma-1 dsRNA; animals were collected 30 hr later, flash frozen in liquid nitrogen, and stored at −80°. Samples were cross-linked in 2% formaldehyde at room temperature for 30 min, the reaction was terminated with 0.125 M glycine and washed twice with M9. Cross-linked samples were resuspend in FA buffer (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 150 mM NaCl) supplemented with 1× complete protease inhibitor without EDTA (Roche), and sonicated (30 sec on, 30 sec off, 70% output for 25 min on a Qsonica Q880R sonicator). Lysates were clarified by centrifugation at 14,000 rpm for 15 min. Supernatants were precleared with protein A agarose beads and incubated with H3K9me3 antibody (07-523; Upstate (Sigma-Aldrich)) overnight. H3K9me3 antibody were precipitated with protein A agarose beads and washed sequentially with FA buffer twice, FA-500 buffer (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 500 mM NaCl) twice, LiCl buffer (0.25 M LiCl, 1% NP-40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris-HCl pH 8.0) once, and TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA) twice. Antibodies were eluted with an elution buffer (1% SDS, 0.1 M sodium bicarbonate) and H3K9me3/DNA were reverse cross-linked with 0.3 M NaCl at 65°, overnight. Precipitated DNA was purified with gel extraction kit and quantified with qPCR.
Immunofluorescence
A total of 20–40 animals in 1× egg buffer (25 mM HEPES, pH 7.3, 118 mM NaCl2, 48 mM KCl, 2 mM CaCl2, 2 mM MgCl2) were dissected on a superfrost plus slide with a needle, and then a coverslip was placed on top of the dissected animals. The slide was placed on a metal block at −80° (precooled to −80°) and the coverslip was popped off with a razor blade. The animals were treated with cold methanol for 10 min at −20°, followed by two washes with PBST (1 × PBS with 0.1% Tween20). The animals were then fixed with 100 μl of fixation buffer (4% paraformaldehyde in PBS) with a coverslip-size parafilm on top, in a humid chamber for 20 min, followed by two washes with 1× PBST. The animals were incubated in 100 μl diluted primary antibody (GFP: ab13970, Abcam; HA: 3724S, Cell Signaling Technology) with a coverslip-size parafilm on top, at room temperature overnight in a humid chamber. The next day, samples were washed with 1× PBST three times and then incubated with 100 μl diluted secondary antibody (goat anti-chicken secondary antibody Alexa Fluor 488: A11039, Thermo Fisher Scientific; goat anti-rabbit antibody Alexa Fluor 568: A11011, Thermo Fisher Scientific) with a coverslip-size parafilm on top, at room temperature for 90 min. After washing with 1× PBST three times, samples were mounted with an antifade mounting medium (H-1000; Vectashield) and sealed with nail polish.
Microscope and microscopy info
Imaging was done as described previously (Wan et al. 2018). Larval or young adult animals were immobilized with 0.1% sodium azide and mounted on glass slides before imaging. Embryos were obtained by dissecting gravid young adults in 1× egg buffer on a coverslip. Animals or embryos were imaged immediately with a Nikon Eclipse Ti microscope equipped with a W1 Yokogawa Spinning disk with 50-μm pinhole disk and an Andor Zyla 4.2 Plus sCMOS monochrome camera. Images were taken under ×60/1.4 Plan Apo Oil objective. Immunofluorescence samples were imaged with a Leica DMi8 microscope equipped with a laser scanning confocal under ×63 oil objective.
qPCR
cDNA or DNA was qualified with SYBR green according to the vendor’s instructions. Primers for qPCR can be found in Primers used in this study subsection at the end of Supplemental Material.
Data and reagent availability
Strains and plasmids are available upon request. Supplemental material available at figshare: https://doi.org/10.25386/genetics.13028186.
Results
Identification of C. elegans NRDE-2 associating proteins
Forward genetic screens have identified five components of the C. elegans nuclear RNAi machinery (NRDE-1/2/3/4 and HRDE-1) (Guang et al. 2008, 2010; Burkhart et al. 2011; Buckley et al. 2012). To identify essential and/or redundant components of the nuclear RNAi machinery, we used immunoprecipitation–mass spectrometry (IP-MS) to isolate proteins associating physically with C. elegans NRDE-2 (Guang et al. 2010). We first used CRISPR/Cas9 to introduce a 3xflag::ha epitope to the nrde-2 locus (Figure S1). The resulting fusion gene encoded a functional protein (Figure S1). We then used α-FLAG antibodies to conduct α-FLAG immunoprecipitations from extracts generated from nrde-2::3xflag::ha, or wild-type, adult stage animals. Silver staining revealed that immunoprecipitation from NRDE-2::3xFLAG::HA extracts purified proteins not present in immunoprecipitation from wild-type extracts (Figure 1A). Copurifying proteins were subjected to liquid chromatography–tandem mass spectrometry. Proteins identified by ≥20 peptides from nrde-2::3xflag::ha extracts, and zero peptides from wild-type extracts, were considered candidate NRDE-2 interacting proteins (Table S1). HRDE-1, NRDE-1, and NRDE-4, which are known to interact with NRDE-2 (Guang et al. 2010; Burkhart et al. 2011; Buckley et al. 2012), were identified as top 10 NRDE-2 interacting proteins, indicating that our NRDE-2 IP-MS was likely successful (Figure 1B). NRDE-3, which is another known NRDE-2 interacting protein, was also identified in the NRDE-2 IP-MS analysis, albeit at lower levels than other known NRDE-2 interacting proteins (Guang et al. 2010) (Table S1). Remaining NRDE-2 interacting proteins are candidate novel nuclear RNAi factors.
MTR-4 interacts with NRDE-2 in C. elegans. (A) Silver-stained gel of α-FLAG precipitated proteins from wild-type (control, N2 Bristol) or nrde-2::3xflag::ha animals. (B) Y-axis, Log10 of total peptides identified. X-axis, Molecular weight (MW) of NRDE-2 co-immunoprecipitating proteins. Proteins previously thought to interact with NRDE-2 are shown in red. Components of the U5 snRNP are shown in green. A complete list of proteins identified by more than 20 peptides in nrde-2::3xflag::ha animals and zero peptides in wild type is shown in Table S1, sheet 1. A list of all proteins identified by NRDE-2 IP-MS is shown in Table S1, sheet 2. (C) Reciprocal co-immunoprecipitation analyses of NRDE-2::3xFLAG::HA (IP: HA) and 3xFLAG::GFP::MTR-4 (IP: GFP) is shown. Co-immunoprecipitating proteins were detected by Western blot (anti-HA or anti-GFP). Tubulin was used as a loading control. Presence/absence of indicated proteins in co-immunoprecipitation extracts (input) is shown. Asterisks indicate smaller MW weight NRDE-2::FLAG::HA species. These smaller NRDE-2 species may indicate the presence of multiple nrde-2 isoforms or protein degradation occurring in vivo or in vitro.
MTR-4 is a NRDE-2 interacting protein
The DEXD RNA helicase MTR-4 was identified by over ten times more peptides than any other NRDE-2 interacting protein (Table S1). To confirm that NRDE-2 and MTR-4 interact, we introduced a 3xflag::gfp epitope immediately preceding the predicted atg start codon of the mtr-4 locus (see Materials and Methods), and then conducted genetic crosses to generate animals expressing both 3xFLAG::GFP::MTR-4 and NRDE-2::3xFLAG::HA. We then used ɑ-GFP and ɑ-HA antibodies to conduct NRDE-2 and MTR-4 co-immunoprecipitation analyses. This analysis confirmed that MTR-4 and NRDE-2 physically associate (Figure 1C). Previous studies have demonstrated an interaction between mammalian NRDE2 and MTR4 (Ogami et al. 2017; Richard et al. 2018; Jiao et al. 2019). We confirmed this interaction by conducting human (Hs)NRDE2 and (Hs)MTR4 co-immunoprecipitation analysis from extracts generated from HEK293T cells coexpressing GFP::HsMTR4 and 3xFLAG::HsNRDE2 (Figure S2). We conclude that NRDE-2 physically associates with MTR-4, and that the interaction between these two proteins is conserved from nematodes to mammals.
C. elegans MTR-4 is a ubiquitously expressed nuclear protein
We used CRISPR/Cas9 to introduce a tagrfp tag into the npp-9 gene, which encodes a component of the nuclear pore and, consequently, marks the nuclear membrane of somatic and germ cells. We then used fluorescence microscopy to visualize animals coexpressing TagRFP::NPP-9 and 3xFLAG::GFP::MTR-4. The analysis showed that MTR-4 was expressed diffusely within nuclei of all embryonic cells (Figure 2A) as well as most/all somatic cells in larval animals (Figure S3). The data show that MTR-4 is a ubiquitously or near-ubiquitously expressed, nuclear-localized protein.
MTR-4 is a nuclear protein that partly colocalizes with NRDE-2 in all/most cells. (A) Fluorescence micrographs of an ≅32 cell embryo or ∼15 pachytene stage adult germ cells expressing GFP::MTR-4 (green) and TagRFP::NPP-9 (magenta). NPP-9 is a component of the nuclear pore and marks nuclear membranes. (B and C) Fluorescence micrographs of (B) embryos or (C) adult pachytene germ cells expressing GFP::MTR-4 and HA::tagRFP::NRDE-2. Immunofluorescence using GFP and HA antibodies against animals expressing GFP::MTR-4 (green) and HA::TagRFP::NRDE-2 (magenta). (D) Inset of one nucleus from C. Bar, (A) embryos: 10 μm, germline: 5 μm; (B) 10 μm; (C) 20 μm; (D) 2 μm.
To further explore the relationship between NRDE-2 and MTR-4 in the nuclei of these cells, we conducted ɑ-HA and ɑ-GFP immunofluorescence on HA::TagRFP::NRDE-2 and 3xFLAG::GFP::MTR-4 expressing C. elegans. Immunofluorescence analysis showed that HA::TagRFP::NRDE-2 was expressed in the nuclei of all/most cells of developing embryos, a result consistent with previous reports on NRDE-2 expression patterns (Figure 2B) (Guang et al. 2010). In these cells, 3xFLAG::GFP::MTR-4 and HA::TagRFP::NRDE-2 were diffusely expressed and colocalized in nuclei (Figure 2B). We also detected 3xFLAG::GFP::MTR-4 and HA::TagRFP::NRDE-2 expression in the germline (Figure 2C). Previous reports failed to detect NRDE-2 expression in germ cells (Guang et al. 2010), a discrepancy likely due to these earlier studies using transgenesis systems prone to silencing in the germline (Mello and Fire 1995; Kelly et al. 1997). In adult C. elegans pachytene stage germ cells, MTR-4 and NRDE-2 appeared largely colocalized with both proteins concentrated near the nuclear periphery, which is the site of chromatin localization in these cells (Goldstein 1982) (Figure 2D). Finally, MTR-4, but not NRDE-2, localized to the nuclear interior of germ cells (Figure 2D, inset). We conclude that C. elegans MTR-4 is a nuclear protein expressed in most, and likely all, C. elegans cells, and that MTR-4 and NRDE-2 partially colocalize within nuclei of these cells.
MTR-4 is essential for fertility and development
Previous studies have shown that RNAi-based knockdown of MTR-4 causes sterility and larval arrest (Simmer et al. 2003). These data suggest that MTR-4 is likely to have essential functions in both the soma and the germline. We obtained a strain (RB1997) thought to harbor a deletion allele of mtr-4 (termed (ok2642)). ok2642 is predicted to delete ≅30% of the mtr-4 gene downstream of the DEAD box helicase domain, throwing all downstream sequences, which include additional conserved domains, out of frame (Figure S4A). We were surprised to find that mtr-4(ok2642) animals did not exhibit obvious fertility or developmental defects when grown under standard laboratory conditions. We wondered, therefore, if ok2642 was actually a null of mtr-4. Indeed, PCR-based genotyping analyses hinted that ok2642 was likely a complex deletion/duplication of the mtr-4 locus, possibly capable of producing full-length functional MTR-4 protein (Figure S4, B and C). Thus, for clarity, we used CRISPR/Cas9 to generate a new deletion allele of mtr-4 that lacked the highly conserved MTR-4 DEAD box helicase domain [termed mtr-4(ΔDEAD box)] (Figure S4D). We easily identified animals heterozygous for the ΔDEAD box deletion, but never animals that were homozygous for the deletion (0/60) (Figure S4E). This result supports the idea that the mtr-4 ΔDEAD box allele is homozygous lethal, that C. elegans MTR-4 is indeed essential, and that ok2642 is not a true null allele of mtr-4. Therefore, to study the function of mtr-4 and, more specifically, ask if MTR-4 is needed for nuclear RNAi, we sought to develop genetic tools that would allow for conditional depletion of MTR-4. We used CRISPR/Cas9 to introduce a degron::gfp tag to mtr-4 and then generated animals expressing degron::GFP::MTR-4 and the TIR1 E3 ubiquitin ligase in either the soma or the germline (Zhang et al. 2015). Fluorescence microscopy revealed that auxin treatment of degron::GFP::MTR-4 animals that expressed TIR1 in the germline was sufficient to deplete degron::GFP::MTR-4 from the germline but not the soma (Figure 3A). Similarly, in animals expressing TIR1 in the soma, auxin treatment triggered degron::GFP::MTR-4 depletion specifically in the soma (Figure 3B). Germline-specific depletion of MTR-4 depletion caused sterility after 1 day of auxin treatment at 20° (Figure 3C). Soma-specific depletion of MTR-4 caused larval arrest after 1 day of auxin treatment at 20° (Figure 3D). The data confirm that MTR-4 has essential functions in both the soma and germline. Because previous studies (Guang et al. 2010) have shown that nrde-2(−) animals are fertile, and do not show developmental defects, when grown at 20°, the data also show that MTR-4 is likely to have essential biological functions above and beyond that of the NRDE-2/MTR-4 complex. Such a model is consistent with the fact that MTR-4 homologs are components of multiple protein complexes (e.g., TRAMP, NEXT, and PAXT) in other eukaryotes as well as with our observation that MTR-4 and NRDE-2 partially colocalize in C. elegans germ cells (see Figure 2D).
C. elegans MTR-4 is required for fertility and development. (A and B) Overlay of fluorescent and light micrographs of an adult C. elegans expressing GFP::degron::MTR-4 and TIR1 in the germline (Psun-1::TIR1::sun-1 3’utr) (A) or soma (Peft-3::TIR1::unc-54 3’utr) (B). Animals were treated with or without 1 mM auxin for 1–2 hr. Red arrows indicate germline expression of GFP::degron::MTR-4 and white arrows indicate somatic nuclei expressing GFP::degron::MTR-4, with or without auxin treatment. (C) Adult gfp::degron::mtr-4 animals expressing TIR1 in the germline were moved to plates containing ±1 mM auxin. After 24 hr, animals were transferred to fresh plates with or without auxin, and the number of progeny produced over the next 24 hr was scored. Error bars are ± standard deviations (SD) of the mean, N = 3. (D) Larval stage 1 (L1) gfp::degron::mtr-4 animals expressing TIR1 in somatic tissues were moved to plates containing ±1 mM auxin. The percentage of animals arresting as larva are shown. Error bars are ±SD. N = 3.
MTR-4 is required for nuclear RNAi
NRDE-2 is required for nuclear RNAi (Guang et al. 2010). Four lines of evidence support the notion that MTR-4 is also required for nuclear RNAi. First, exposure of C. elegans to dsRNA induces cotranscriptional silencing of nascent RNAs exhibiting sequence homology to trigger dsRNAs (Bosher et al. 1999; Guang et al. 2008, 2010). Nuclear RNAi efficacy can be quantified in C. elegans using “operon RNAi” assays. For instance, the lir-1 and lin-26 genes are cotranscribed as part of a single transcriptional unit or operon (Dufourcq et al. 1999). RNAi-based silencing of lir-1 causes cotranscriptional silencing of the downstream gene in the operon, lin-26, which in turn causes animals to exhibit the lin-26 loss-of-function phenotype (i.e., larval arrest) (Bosher et al. 1999). The nuclear RNAi machinery (e.g., NRDE-2) is required for lir-1 RNAi to cotranscriptionally silence lin-26 and, therefore, trigger larval arrest (Guang et al. 2008, 2010) (Figure 4A). mtr-4 RNAi treatment in our hands caused sterility, but not larval arrest, allowing us to use mtr-4 RNAi to ask if knockdown of mtr-4 was sufficient to prevent a subsequent lir-1 RNAi treatment from inducing lir-1 RNAi-induced larval arrest (i.e., nuclear RNAi) (Figure 4A). Indeed, pretreatment of animals with mtr-4 RNAi inhibited lir-1 RNAi-induced larval arrest (Figure 4A), consistent with the idea that MTR-4 contributes to nuclear RNAi.
MTR-4 is required for nuclear RNAi. (A) Top panel: schematic of lir-1 nuclear RNAi assay. Bottom panel: an eri-1(mg366) genetic background was used for all conditions in this experiment because eri-1(mg366) enhances experimental RNAi and allows lir-1 RNAi-induced larval arrest to be more easily scored (Guang et al. 2008). eri-1(mg366) animals were exposed to bacteria producing lir-1 dsRNA in conjunction with bacteria producing nrde-2 dsRNA, oma-1 dsRNA, or bacteria producing mtr-4 dsRNA diluted with increasing amounts of non-dsRNA-generating bacteria L4440. Condition 9–13, eri-1(mg366) animals were exposed to bacteria producing mtr-4 dsRNA diluted with increasing amounts of non-dsRNA-generating bacteria L4440. Percentage of animals arrested as larva are indicated. Error bars are ±SD. N = 6 for first seven conditions, N = 3 for the rest of conditions. (B) Animals were exposed to bacteria producing lir-1 dsRNA as well as with or without auxin treatment for the indicated times. Percentage of animals arrested as larva are indicated. Error bars are ±SD, N = 3. (C) Adult animals were subjected treated with or without oma-1 RNAi and with or without auxin. Quantitative RT-PCR analysis was used to quantify oma-1 pre-mRNA using primers that span exon-intron junctions. oma-1 quantitative RT-PCR signal from samples not exposed (−) to oma-1 RNAi or auxin were defined as 1. All other data are expressed as a ratio of oma-1 pre-mRNA levels to the control (−) oma-1/auxin dsRNA condition. Error bars are ±SD, N = 4. P-values calculated with Student’s t-test, two-tailed. (D) H3K9me3 chromatin immunoprecipitation was conducted using H3K9me3 antibody, and precipitating DNA was quantified by quantitative RT-PCR. Data are expressed as a ratio of DNA chromatin immunoprecipitation with or without (±) oma-1 dsRNA treatment. X-axis indicates position in nucleotides (nt) along the region of chromosome IV relative to the oma-1 locus, whose ATG is defined as 0 nt. oma-1 transcriptional unit indicates the oma-1 operon. c09g9.5 and rad-26 are genes immediately 5’ or 3’ to the oma-1 operon, respectively. Error bars are ± standard error of mean (SEM). N = 3.
Second, we found that short (5–10 min) exposures of degron::GFP::MTR-4 expressing animals to auxin was sufficient to deplete somatic degron::GFP::MTR-4 (Figure S5A), but that this short-term auxin exposure was not sufficient to trigger larval arrest, as indicated by the fact that animals subjected to 5–10 min auxin treatments reached adulthood, albeit with a short delay when compared to nontreated animals (Figure S5B). Using this short-term auxin-exposure regime, we observed that short-term depletion of MTR-4 was sufficient to inhibit nuclear RNAi, as indicated by the failure of lir-1 RNAi treatments to trigger larval arrest in animals temporarily depleted for MTR-4. The data support the idea that MTR-4 contributes to nuclear RNAi (Figure 4B).
Third, we used quantitative RT-PCR to measure levels of the oma-1 pre-mRNA before or after exposure of animals to dsRNA targeting oma-1 and with or without auxin-based MTR-4 depletion. For this experiment, we treated degron::GFP::MTR-4 adult animals expressing the TIR1 E3 ubiquitin ligase in the germline with or without auxin for 2 hr and then with or without oma-1 RNAi for 16 hr (in the absence of auxin). As expected, oma-1 RNAi was able to trigger nuclear RNAi in animals not depleted for MTR-4, as indicated by the fact that oma-1 RNAi treatment reduced oma-1 pre-mRNA levels by approximately three times (Figure 4C). When animals were depleted for MTR-4, however, oma-1 RNAi failed to trigger nuclear RNAi, as indicated by oma-1 pre-mRNA levels not decreasing after oma-1 RNAi (Figure 4C). Note, MTR-4 depletion was associated with a dramatic ∼20 times increase in oma-1 pre-mRNA levels irrespective of whether or not animals were subjected to oma-1 RNAi (Figure 4C). This increase in pre-mRNA levels after MTR-4 depletion may be due to loss of the nuclear exosome RNA-degrading function of MTR-4 (see Discussion).
Fourth, nuclear RNAi causes the deposition of repressive chromatin marks such as H3K9me3 or H3K27me3 onto chromatin of genes targeted for silencing by nuclear RNAi (Guang et al. 2010; Gu et al. 2012; Mao et al. 2015). We treated degron::GFP::MTR-4 adult animals expressing the TIR1 E3 ubiquitin ligase in the germline with or without auxin for 2 hr and then with or without oma-1 RNAi for 30 hr in the absence of auxin. And after 30 hr, we conducted H3K9me3 chromatin immunoprecipitation as a readout for nuclear RNAi-mediate chromatin changes. The analysis showed that MTR-4 depletion was sufficient to prevent oma-1 RNAi from modifying chromatin states at the oma-1 locus (Figure 4D). H3K9me3 on surrounding loci (rad-26 and c09g9.5) was not affected by oma-1 RNAi and/or MTR-4 depletion (Figure 4D). Taken altogether, these four lines of evidence support the notion that MTR-4 is required for nuclear RNAi.
MTR-4 associates with pre-mRNA in response to RNAi
Because MTR-4 forms a complex with NRDE-2, we wondered if the reason MTR-4 was required for nuclear RNAi might be an indirect consequence of NRDE-2 being destabilized in the absence of MTR-4. To test this idea, we used ɑ-HA antibodies to quantify HA::TagRFP::NRDE-2 levels in animals depleted for MTR-4 via mtr-4 RNAi. As expected, mtr-4 RNAi decreased MTR-4 protein levels (Figure 5A and Figure S6). MTR-4 knockdown, however, was not accompanied by observable decreases in NRDE-2 protein levels, suggesting that the reason MTR-4 is required for nuclear RNAi is not simply because MTR-4 is required to stabilize NRDE-2 (Figure 5A and Figure S6).
MTR-4 associates with pre-mRNA in response to RNAi. (A) Western blotting was used to monitor NRDE-2::3xFLAG::HA levels and 3xFLAG::GFP::MTR-4 levels in animals treated with or without mtr-4 RNAi. Two replicates are shown. The data show that RNAi-mediated depletion of MTR-4 does not change the overall levels of NRDE-2. (B) and (C) Anti-FLAG antibodies were used for immunoprecipitation of 3xFLAG::GFP::MTR-4 from extracts derived from wild-type (WT) or nrde-2(gg91) adult animals. MTR-4 coprecipitating RNAs were converted to cDNA and quantified by quantitative RT-PCR using primers that span exon-intron junctions. Data are expressed as a ratio of MTR-4 precipitating pre-mRNA with or without oma-1 RNAi (B) or lin-15b RNAi (C). Error bars are ±SD (B) or SEM (C); N = 3. Note, loading controls showing that equal amounts of MTR-4 were immunoprecipitated from all samples in B and C are shown in Figure S7, A and B.
During nuclear RNAi, two nuclear AGOs (NRDE-3 in the soma and HRDE-1 in the germline) direct cTGS by associating with nascent RNAs and recruiting other downstream silencing factors, such as NRDE-2, to pre-mRNAs undergoing nuclear RNAi (Guang et al. 2008, 2010; Burkhart et al. 2011; Buckley et al. 2012). Once localized to pre-mRNA by AGO/siRNAs, the NRDE nuclear RNAi machinery triggers premature termination of RNAP II transcription via a currently unknown mechanism (Guang et al. 2010). RNA immunoprecipitation studies have shown that known nuclear RNAi factors (e.g., NRDE-1/2/4) associate with pre-mRNA fragments generated during nuclear RNAi (Guang et al. 2010; Burkhart et al. 2011; Buckley et al. 2012; Spracklin et al. 2017). We asked if MTR-4, like these other nuclear RNAi factors, might be recruited to pre-mRNA during nuclear RNAi. We treated 3xFLAG::GFP::MTR-4 expressing animals with or without dsRNAs targeting oma-1 or lin-15b, and then used ɑ-FLAG antibodies to pull down MTR-4 and its associated RNAs. Then, 3xFLAG::GFP::MTR-4 coprecipitating RNAs were subjected to quantitative RT-PCR, using oma-1 or lin-15b specific intron-exon primer pairs, to quantify oma-1 or lin-15b pre-mRNA associating with MTR-4 in animals that had or had not been treated with oma-1 or lin-15b RNAi. Control Western blots showed that similar levels of MTR-4 protein were immunoprecipitated in all conditions (Figure S7, A and B). However, 10–250 times more oma-1 and lin-15b pre-mRNA co-immunoprecipitated with MTR-4 after oma-1 or lin-15b RNAi, respectively, than in control non-RNAi-treated animals (Figure 5, B and C). The data show that RNAi directs MTR-4 to associate with pre-mRNA of genes undergoing cTGS. A more detailed analysis of this MTR-4-associated pre-mRNA showed that MTR-4 associates primarily with pre-mRNA encoded 5’ to sites of RNAi, suggesting that MTR-4 is likely to bind to fragments of pre-mRNA generated during nuclear RNAi (Figure 5C). The fact that MTR-4 is directed to bind pre-mRNAs of genes targeted by RNAi suggests that MTR-4 may play a direct role in the underlying mechanism of nuclear RNAi in C. elegans (see Discussion). We conclude that MTR-4 associates with pre-mRNA fragments of genes undergoing nuclear RNAi.
Hierarchical recruitment of NRDE-2 and MTR-4 to pre-mRNA during nuclear RNAi
According to current models of nuclear RNAi, the NRDE factors are recruited onto pre-mRNA in a hierarchical manner. First, the nuclear AGOs associate with pre-mRNA, presumably via Watson–Crick base pairing between AGO-associated siRNAs and pre-mRNAs (Guang et al. 2008, 2010; Buckley et al. 2012). Then nuclear AGOs recruit NRDE-2 (Guang et al. 2010). Then NRDE-2 recruits NRDE-1 (Burkhart et al. 2011). We asked where MTR-4 might fit into this assembly process. Recruitment of NRDE-2 to pre-mRNA by RNAi depends upon the nuclear AGOs HRDE-1 or NRDE-3, but not the downstream nuclear RNAi factor NRDE-1 (Guang et al. 2008, 2010; Burkhart et al. 2011). We found that the genetic requirements for MTR-4 recruitment to pre-mRNA resembled those for NRDE-2: MTR-4 failed to associate with pre-mRNA after RNAi in hrde-1(−) animals, but was able to associate with pre-mRNA after RNAi in nrde-1(−) animals (Figure 6A). Similar amounts of MTR-4 was immunoprecipitated from each sample regardless of genotype or RNAi treatment (Figure S7C).
Interdependent recruitment of NRDE-2/MTR-4 to pre-mRNA during nuclear RNAi. (A) 3xFLAG::GFP::MTR-4 coprecipitating pre-mRNA from wild-type (WT), hrde-1(tm1200) or nrde-1(gg88) adult animals. Data are expressed as a ratio of precipitating pre-mRNA with or without RNAi. Error bars are ±SEM, N = 3. (B) Animals expressing NRDE-2::3xFLAG::HA and degron::GFP::MTR-4 in the germline were treated with or without auxin for 2 hr and then transferred to plates with or without oma-1 RNAi, without auxin for 16 hr. Anti-FLAG antibodies were used for immunoprecipitation of NRDE-2::3xFLAG::HA and coprecipitating RNAs were converted to cDNA and quantified by quantitative RT-PCR using two primer pairs (primer 1 and 2) spanning oma-1 exon-intron junctions. Data are expressed as a ratio of NRDE-2 precipitating pre-mRNA with or without oma-1 RNAi. Error bars are ±SD, N = 2. (C and D) 3xFLAG::GFP::MTR-4 coprecipitating RNAs from WT, hrde-1(tm1200), nrde-1(gg88), and nrde-2(gg91) adults was converted to cDNA and bath-45 pre-mRNA was quantified by quantitative RT-PCR using primers that span bath-45 exon-intron junctions. Quantitative RT-PCR detecting the gld-2 pre-mRNA was used in C and D as a negative control, as gld-2 is not thought to be a target of nuclear RNAi (Guang et al. 2010). WT is defined as 1. Other data are expressed as fold change relative to WT. Error bars are ±SEM (C) or SD (D); N = 3.
We next asked if NRDE-2 was required for MTR-4 recruitment to pre-mRNA. MTR-4 failed to IP with the oma-1 or lin-15b pre-mRNA after oma-1 or lin-15b RNAi, respectively, in animals lacking NRDE-2 (Figure 5, B and C) despite similar levels of MTR-4 being purified under each condition (Figure S7, A and B). Previous studies have shown that oma-1 and lin-15b are expressed at near wild-type levels in animals lacking NRDE-2, indicating that loss of MTR-4/pre-mRNA binding in the absence of NRDE-2 is not due to the absence of oma-1 or lin-15b pre-mRNA in nrde-2 mutant animals (Guang et al. 2010; Buckley et al. 2012). Similarly, MTR-4 was required for NRDE-2 to be recruited to pre-mRNAs in response to RNAi: auxin-based depletion of degron::GFP::MTR-4 prevented NRDE-2 from associating with the oma-1 pre-mRNA after oma-1 RNAi (Figure 6B). As oma-1 pre-mRNA is expressed in MTR-4-depleted animals (Figure 4C), loss of NRDE-2 association with the oma-1 pre-mRNA in animals lacking MTR-4 is not the result of this loci not being expressed after MTR-4 depletion. The interdependence of NRDE-2 and MTR-4 recruitment to pre-mRNA during nuclear RNAi suggests that these proteins may be recruited together as part of a complex. The dependence of complex recruitment on HRDE-1, but not NRDE-1, suggests that NRDE-2/MTR-4 recruitment is likely a fairly early step during cTGS.
During C. elegans development, endogenously expressed siRNAs direct nuclear RNAi in the nuclei of both somatic and germline cells (Billi et al. 2014). One well-studied endogenous target of the nuclear RNAi pathway is the bath-45 gene (Buckley et al. 2012). Previous studies have shown that NRDE-2 is required for nuclear RNAi at the bath-45 locus, that NRDE-2 associates with bath-45 pre-mRNA, and that the association of NRDE-2 with the bath-45 pre-mRNA depends upon the nuclear AGO HRDE-1, but not NRDE-1 (Buckley et al. 2012). We found that MTR-4 co-immunoprecipitated with the bath-45 pre-mRNA, and that the association between MTR-4 and bath-45 pre-mRNA depended upon HRDE-1 and NRDE-2, but not NRDE-1 (Figure 6, C and D). Again, similar amounts of MTR-4 were immunoprecipitated regardless of genotype (Figure S7, D and E). Thus, MTR-4 is recruited to bath-45 pre-mRNA and the genetic requirements for this recruitment are similar to those for NRDE-2. The data are consistent with the idea that the C. elegans NRDE-2/MTR-4 complex promotes nuclear RNAi at genomic loci during normal growth and development in C. elegans.
Discussion
Here, we show that NRDE-2 and the RNA helicase MTR-4 associate with each other in C. elegans. We also show that MTR-4 is required for nuclear RNAi (cTGS) and that MTR-4 is recruited to pre-mRNA emanating from genes undergoing cTGS. MTR-4 has a well-established role in recruiting aberrant RNAs to the nuclear exosome for degradation (Thoms et al. 2015). Therefore, one obvious role for MTR-4 during cTGS might be to escort pre-mRNA fragments, generated during cTGS, to the nuclear exosome for degradation. This straightforward model is unlikely to fully explain the full function of MTR-4 during cTGS, however, for the following reason. cTGS in C. elegans causes fragmentation of RNAi-targeted pre-mRNAs, likely via premature termination of transcription (Guang et al. 2010). Being prematurely terminated, most cTGS-derived pre-mRNA fragments do not encode functional proteins (Guang et al. 2010). Therefore, further degradation of these already fragmented pre-mRNAs by MTR-4 and the nuclear exosome would not be expected to affect the abundance of functional protein produced from genes undergoing cTGS. If the role of MTR-4 in nuclear RNAi was solely to degrade fragments of pre-mRNA, then one would not expect MTR-4 to be required for nuclear RNAi, at least when gene activity is monitored at the level of protein function. And yet “operon RNAi” experiments, which monitor protein levels, indicate that MTR-4 is required for nuclear RNAi (Figure 4, A and B). Thus, if MTR-4 does help degrade cTGS-generated pre-mRNA fragments, this is unlikely to be the only function performed by MTR-4 during cTGS in C. elegans.
What might these other functions be? Homologs of C. elegans NRDE-2 and MTR-4 form a complex in both mammalian cells and in fungi (Lee et al. 2013; Zhou et al. 2015; Aronica et al. 2016; Richard et al. 2018; Jiao et al. 2019; Wang et al. 2019). Fission yeast homologs of NRDE-2 and MTR-4 (termed Nrl1 and Mtl1) form a complex that regulates the expression of genes possessing cryptic introns (Lee et al. 2013; Zhou et al. 2015). The underlying mechanism of this regulation is not known, but may involve siRNA-directed heterochromatin formation (Lee et al. 2013; Zhou et al. 2015). We show here that the NRDE-2/MTR-4 complex contributes to siRNA-directed H3K9me3 in C. elegans, suggesting that the gene-regulatory function of NRDE-2/MTR-4-like complexes may involve chromatin regulation and that this link may be conserved. On the other hand, a recent study in mammalian cells shows that the function of the mammalian NRDE2/MTR4 complex is not to inhibit gene expression, as it does in fungi and nematodes, but rather to promote gene expression (Wang et al. 2019). The mechanism by which the mammalian complex promotes gene expression may involve NRDE2 interacting with MTR4 on nascent RNAs to prevent MTR-4 from trafficking pre-mRNAs to the nuclear exosome for degradation (Wang et al. 2019). Thus, surprisingly, current data suggest that NRDE-2/MTR-4 complexes have different effects on gene expression in different eukaryotes: the complex inhibits gene expression in fungi and nematodes and promotes gene expression in mammals. Such opposing roles in gene expression could suggest that either the complex has radically different molecular activities in different eukaryotes or, more likely, that the complex has the same activity in all eukaryotes but the gene regulatory consequences of this shared molecular activity differs.
The mammalian NRDE2/MTR4 complex is thought to promote gene expression because it protects full-length pre-mRNAs from exosome-based degradation (Wang et al. 2019). The following model is put forth to try and explain how the C. elegans NRDE-2/MTR-4 complex might inhibit gene expression while, paradoxically, protecting pre-mRNA fragments from exosome-based degradation. In C. elegans, RNA-dependent RNA Polymerase (RdRP) use cytoplasmic RNAs as templates to generate secondary (2°) siRNAs, which bind to secondary AGOs to drive post-transcriptional and cTGS (Pitt et al. 2000; Vought et al. 2005; Phillips et al. 2012; Wan et al. 2018). If pre-mRNAs fragments generated during C. elegans cTGS were trafficked to the cytoplasm to act as templates for RdRPs, then the NRDE-2/MTR-4 complex might, paradoxically, inhibit gene expression by preventing the degradation of these fragmented template pre-mRNAs. This “template protection” model is particularly appealing as it may help explain several additional mysteries surrounding cTGS- and RNA-based heritable gene silencing in C. elegans. First, RNAi-based gene silencing can be inherited for multiple generations in C. elegans (termed RNAi inheritance) (Vastenhouw et al. 2006; Ashe et al. 2012; Buckley et al. 2012). This transgenerational epigenetic inheritance (TEI) requires the RNA pUGylase MUT-2/RDE-3, which adds alternative U and G nucleotides to 3’ termini of RNAi-targeted mRNAs during RNAi (Shukla et al. 2020). pUG tails then recruit RdRP enzymes, which use pUGylated mRNAs as templates for antisense siRNA production (Shukla et al. 2020). Current models posit that generationally repeated rounds of sense mRNA pUGylation and antisense RdRP-based siRNA amplification (termed pUG RNA/siRNA loops) underlie heritable silencing of genes during RNAi inheritance (Shukla et al. 2020). RNAi inheritance also depends upon the nuclear RNAi machinery (Ashe et al. 2012; Buckley et al. 2012). The reason the nuclear RNAi machinery is required for RNA inheritance is not known. The template protection model may help explain why: by protecting cTGS-derived RNA fragments from exosome-based degradation, the RNAi-directed NRDE-2/MTR-4 complex might help ensure that RNAs, which act as templates for RdRPs (after pUGylation by RDE-3 in the cytoplasm), are not degraded and, therefore, are available to help maintain gene silencing over generational timescales. On a related note, prevailing models of RNAi inheritance posit that genes undergoing heritable silencing are expressed (so that RNA fragments are available to feed pUG RNA/siRNA looping). In other words, TEI genes need to be expressed so that they are not expressed. The template protection model may help solve this obvious paradox. By helping drive cTGS, the NRDE-2/MTR-4 complex could prevent the production of full-length pre-mRNAs that have the ability to encode functional proteins, thus achieving gene silencing. At the same time, by preventing the degradation of these non-functional pre-mRNA fragments, the complex ensures a steady supply of the RdRP templates needed for maintaining gene silencing over generations. An important test of the template protection model will be to ask if mutations that disrupt MTR-4/pre-mRNA interaction during gene silencing also disrupt RNAi inheritance.
In addition to MTR-4, our NRDE-2 IP-MS analysis identified several additional NRDE-2 interacting proteins, which are also obvious candidates for novel nuclear RNAi factors. Interestingly, three of the top 10 NRDE-2 interacting factors were components of the U5 splicing snRNP (prp-8, eftu-2 and snrp-200) (Figure 1B). The U5 snRNP mediates the second transesterification reaction of pre-mRNA splicing (Turner et al. 2004; Nguyen et al. 2013). Interestingly, NRDE2 IP-MS in human cells also identified the same three U5 snRNP proteins as top 20 human NRDE2 interacting proteins (Jiao et al. 2019). Remarkably, IP-MS on the MTR-4 homolog Mtl1 in fission yeast also identified components of the U5 snRNP (Lee et al. 2013; Zhou et al. 2015). Thus, U5 snRNP and NRDE-2/MTR-4-like complex interactions are deeply conserved. Importantly, this interaction appears to be fairly specific as NRDE-2 IP-MS in C. elegans and mammals did not identify other snRNPs (e.g., U1/2/4/6) and mammalian NRDE2 preferentially coprecipitates with the U5 RNA over the other U RNAs (Jiao et al. 2019). Why might the NRDE-2/MTR-4 complex interact specifically with the U5 snRNP? Studies in C. elegans and in fungi suggest that nonconsensus splice sites may help mark specific loci as potential parasites (e.g., transposons and retrotransposons), and that cTGS-like systems may help silence such elements (Dumesic et al. 2013; Lee et al. 2013; Newman et al. 2018). Additionally, in mammalian cells, splicing of poor-quality splice sites depends upon NRDE-2 (Jiao et al. 2019). Finally, phylogenetic profiling studies hint at a deep connection between small-RNA based silencing systems and pre-mRNA splicing systems (Tabach et al. 2013). Together, these results suggest an ancient and conserved connection between cTGS and RNA splicing systems and hint that the cTGS/RNA splicing connection may, at least in part, be mediated by physical interactions between NRDE-2/MTR-4 and the U5 snRNP.
Finally, during the course of these studies, we made a perplexing observation that hints at a possible mechanism for how and why cTGS and RNA splicing may connect. We found that deletion of five of seven introns from the lin-15b gene (termed lin-15bΔintron) caused the lin-15bΔintron gene to become completely refractory to cTGS (Figure S8, A and B). The result shows that, in some situations at least, introns are necessary for cTGS. Why might introns be needed for cTGS? In fungi, the 3’ termini of the telomerase RNA is generated when the last intron of the telomerase pre-mRNA is subjected to the first, but not the second, splicing transesterification reaction (Box et al. 2008; Qi et al. 2015). Thus, “abortive splicing” (completion of the first, but not the second, splicing reaction) is used by some fungi to cotranscriptionally fragment and process nascent RNAs. It is possible that a related mechanism may be used by C. elegans to fragment pre-mRNAs during cTGS, with physical interactions between the U5 snRNP and NRDE-2/MTR-4 underlying this RNAi-directed abortive splicing mechanism. Assessing if 3’ termini of cTGS-derived pre-mRNA fragments coincide with 5’ splice sites will be a strong test of the “abortive splicing mediates cTGS” model.
Acknowledgments
We thank past and current members of the Kennedy lab for helpful discussions; This work was supported by NIH grants GM088289 and GM132286 to S.K; G.W. was supported by Basic and Applied Basic Research Foundation of GuangDong (2019A1515110744) and the National Natural Science Foundation of China grants (32070798); J.Y. was supported by the Ruth L. Kirschstein T32 Predoctoral NRSA (T32GM096911) and NSF Graduate Research Fellowships (DGE1745303); D.J.P. was supported by a Ruth L.Kirschstein National Research Service Award (1F32GM125345-01).
Footnotes
Supplemental material available at figshare: https://doi.org/10.25386/genetics.13028186.
Communicating editor: O. Rando
- Received April 23, 2020.
- Accepted October 5, 2020.
- Copyright © 2020 by the Genetics Society of America
