The ubiquitin-proteasome pathway mediates selective degradation of unloaded Argonaute proteins in C. elegans

Abstract

Argonaute proteins are essential effectors of small RNA-mediated gene regulation, yet the extent to which their stability depends on small RNA loading remains poorly understood. In Caenorhabditis elegans, we systematically disrupted the small RNA-binding capacity of multiple Argonaute proteins to assess their stability in the absence of small RNA partners. We found that while most Argonautes remain stable when unable to bind small RNAs, a subset, including PRG-1, HRDE-1, and PPW-2, exhibited markedly reduced protein levels. Focusing on the PIWI-clade Argonaute PRG-1, we show that its destabilization occurs post-translationally and is independent of mRNA expression or translational efficiency. Instead, unbound PRG-1 is targeted for degradation by the ubiquitin-proteasome system. Additionally, the failure to load piRNAs disrupts PRG-1 localization to perinuclear germ granules. We further identify the E3 ubiquitin ligase EEL-1 as a factor contributing to the degradation of unloaded PRG-1. These findings uncover a critical role for small RNA loading in maintaining the stability and localization of a subset of Argonaute proteins, and reveal a quality control mechanism that selectively eliminates unbound PRG-1 to preserve germline regulatory fidelity.

Introduction

Precise regulation of gene expression at the transcriptional and post-transcriptional levels is essential for fertility, proper development, and organismal longevity. One fundamental mechanism that contributes to this regulation is RNA interference (RNAi), a conserved biological process in which small RNA molecules recognize complementary target transcripts, resulting in mRNA degradation, translational repression, or transcriptional silencing. In addition to regulating gene expression, RNAi plays critical roles in defending against foreign genetic elements and preserving genome stability (Ghildiyal and Zamore 2009).

In Caenorhabditis elegans, small RNAs are produced via both exogenous and endogenous pathways. Exogenous small interfering RNAs (exo-siRNAs) arise from experimentally introduced or naturally encountered double-stranded RNAs (dsRNAs), a process first described in C. elegans (Fire et al. 1998 ). These dsRNAs are processed by the RNase III enzyme Dicer into short RNA duplexes (∼21 to 23 nt), which are then incorporated into RNA-induced silencing complexes (RISCs) (Zamore et al. 2000; Bernstein et al. 2001; Matranga et al. 2005). Endogenous small RNAs, in contrast, are transcribed from genomic loci or generated from mRNA transcripts. These include miRNAs, piRNAs, primary siRNAs, and secondary siRNAs, which are distinguished by their biogenesis pathways, structural features, and associated Argonaute proteins (Ketting and Cochella 2021).

The core effectors of RNAi are Argonaute proteins, which bind small RNAs to form RISC and mediate gene silencing in a sequence-dependent manner. Argonautes are a highly conserved family of proteins across eukaryotes, characterized by four conserved domains: the N, PAZ, MID, and PIWI domains (Hutvagner and Simard 2008). The MID and PAZ domains facilitate anchoring of the 5′ and 3′ ends of the small RNA, while the PIWI domain, structurally related to RNase H, can cleave target RNAs when it contains the catalytic DDH triad. In C. elegans, however, many Argonautes lack these catalytic residues and instead recruit other silencing machinery to repress gene expression (Yigit et al. 2006).

The C. elegans genome encodes 27 Argonaute genes, of which at least 19 are known to produce functional proteins that bind distinct classes of small RNAs (Yigit et al. 2006; Seroussi et al. 2023). These include members of the AGO, PIWI, and WAGO (worm-specific Argonaute) subfamilies, which each have distinct small RNA partners and biological functions. AGO-clade Argonautes include ALG-1, ALG-2, and ALG-5, which primarily bind microRNAs (miRNAs), and RDE-1, which binds primary siRNAs generated from double-stranded RNA and is essential for initiating exogenous RNAi (Tabara et al. 1999; Grishok et al. 2001; Brown et al. 2007). Additional germline-expressed AGO-clade Argonautes, ALG-3, ALG-4, and ERGO-1, bind 26G-RNAs and, along with RDE-1, function as primary Argonautes to trigger the production of 22G secondary siRNAs (Sijen et al. 2001; Conine et al. 2010; Vasale et al. 2010). The single PIWI-clade Argonaute, PRG-1, binds PIWI-interacting RNAs (21U-RNAs) and plays a central role in initiating silencing pathways in the germline (Batista et al. 2008; Das et al. 2008; Wang and Reinke 2008). WAGO-clade Argonautes act downstream of these pathways by binding 22G-RNAs synthesized by RNA-dependent RNA polymerases (RdRPs) and enforcing gene silencing (Ashe et al. 2012; Luteijn et al. 2012; Shirayama et al. 2012). Cytoplasmic WAGOs (e.g. WAGO-1, PPW-1, PPW-2) promote post-transcriptional repression, whereas nuclear WAGOs (e.g. HRDE-1 and NRDE-3) mediate transcriptional silencing and heritable epigenetic regulation (Gu et al. 2009; Guang et al. 2010; Buckley et al. 2012). Although structurally related to WAGOs, CSR-1 plays an antagonistic role by licensing gene expression and ensuring proper chromosome segregation (Claycomb et al. 2009; Seth et al. 2013; Wedeles et al. 2013). Together, this diverse Argonaute repertoire enables C. elegans to carry out complex, spatially regulated, and highly specific small RNA–mediated gene regulatory programs.

Among the endogenous small RNA pathways in C. elegans, the piRNA pathway is distinct in both its biogenesis and function. piRNAs are ∼21 nucleotides in length and are transcribed as single-stranded precursors by RNA polymerase II from thousands of discrete genomic loci (Ruby et al. 2006; Gu et al. 2012). This transcription requires dedicated machinery, including components of the small nuclear RNA-activating protein complex (SNPC) and the piRNA-specific factor PRDE-1 (Kasper et al. 2014; Weick et al. 2014; Choi et al. 2021). The resulting precursors undergo tightly regulated 5′ and 3′ end processing, including 2′-O-methylation at the 3′ terminus, to yield mature piRNAs (Billi et al. 2012; Kamminga et al. 2012; Montgomery et al. 2012; Tang et al. 2016; Podvalnaya et al. 2023). Mature piRNAs are then selectively loaded onto PRG-1, which uses partial sequence complementarity to scan for non-self or unlicensed targets in the germline (Bagijn et al. 2012; Lee et al. 2012; Zhang et al. 2018). While direct target cleavage is not required for piRNA-mediated gene silencing, piRNAs initiate a robust secondary siRNA amplification cascade that generates 22G-RNAs, engaging WAGO-clade Argonautes to reinforce gene silencing (Ashe et al. 2012; Bagijn et al. 2012; Lee et al. 2012; Shirayama et al. 2012). This multi-layered surveillance mechanism safeguards genome integrity and ensures transgenerational regulation of gene expression in the germline.

Numerous studies have highlighted the interdependence between Argonaute protein expression and small RNA stability. In C. elegans, small RNAs are frequently depleted when their corresponding Argonaute partner is mutated (Grishok et al. 2001; Batista et al. 2008; Gu et al. 2009; Conine et al. 2010). Conversely, the expression of some Argonaute proteins is reduced in the absence of their small RNA partners. For example, PRG-1 protein levels are significantly diminished in mutants lacking piRNAs or key piRNA biogenesis factors (de Albuquerque et al. 2014; Weick et al. 2014). However, the precise mechanisms underlying PRG-1 regulation remain unclear, and it is not known whether other C. elegans Argonaute proteins are similarly affected by the absence of small RNA binding.

In this study, we investigate whether Argonaute protein stability depends on small RNA loading in C. elegans. Using a systematic approach to disrupt small RNA binding across a panel of Argonaute proteins, we identify three, PRG-1, HRDE-1, and PPW-2, that are destabilized in the absence of their small RNA partners, while others remain unaffected. We focus on PRG-1 and demonstrate that its degradation occurs post-translationally via the ubiquitin-proteasome system when piRNA loading is impaired. Loss of piRNAs or mutations in PRG-1's RNA-binding pocket not only reduce protein levels but also disrupt its localization to perinuclear germ granules. Furthermore, we identify the E3 ubiquitin ligase EEL-1 as a factor that contributes to the degradation of unbound PRG-1. Together, our findings uncover a post-translational quality control mechanism that ensures PRG-1 stability is tightly coupled to piRNA loading, thereby safeguarding the fidelity of piRNA-mediated gene silencing.

Materials and methods

C. elegans strains

Caenorhabditis elegans strains were maintained at 20 °C on NGM plates seeded with OP50  Escherichia coli according to standard conditions unless otherwise stated (Brenner 1974). All strains used in this project are listed in Supplementary Data 1.

Plasmid and strain construction

All fluorescent and epitope tags were integrated at the endogenous loci by CRISPR genome editing (Dickinson et al. 2013; Friedland et al. 2013; Arribere et al. 2014; Ward 2015).

Plasmid-based CRISPR: For all CRISPR insertions of fluorescent tags, we generated homologous repair templates using the primers listed in Supplementary Data 2. gfp::3xFLAG::ergo-1 was assembled into pDD282 (Addgene #66823) by isothermal assembly according to published protocols (Gibson et al. 2009; Dickinson et al. 2013). To protect the repair template from cleavage, we introduced silent mutations at the site of guide RNA targeting by incorporating these mutations into one of the homology arm primers or, if necessary, by performing site-directed mutagenesis (Dickinson et al. 2013). All guide RNA plasmids were generated by ligating oligos containing the guide RNA sequence into BsaI-digested pRB1017 (Addgene #59936) (Arribere et al. 2014). Guide RNA sequences are provided in Supplementary Data 2. The gfp::3xFLAG::ergo-1 injection mix included 25 to 50 ng/μl repair template, 50 ng/μl guide RNA plasmid, 50 ng/μl eft-3p::cas9-SV40_NLS::tbb-2 3′UTR (Addgene #46168), and 2.5 to 10 ng/μl GFP co-injection markers. For wago-10::2xFLAG and 2xHA::prg-1 constructs, the injection mix included 50 ng/μl 2xFLAG or 2xHA oligo repair template, 25 ng/μl wago-10 or prg-1 guide RNA plasmid, respectively, 20 ng/μl rol-6 repair template, 25 ng/μl rol-6 guide RNA plasmid (pJA42, Addgene #59930), and 50 ng/μl eft-3p::Cas9 (pJW1259, Addgene #61251). All mixes were injected into the wild-type strain. F1 animals with the Rol phenotype were isolated and, for wago-10::2xFLAG and 2xHA::prg-1 constructs, genotyped by PCR and restriction digest to identify animals with the correct repair (Arribere et al. 2014). All repair template sequences are provided in Supplementary Data 2.

Protein-based CRISPR: For prg-1(cmp231[YK-AA]), prg-1(cmp301[YK-AA]), alg-3(cmp230[YK-AA]), ergo-1(cmp234[YK-AA]), csr-1(cmp244[HK-AA]), ppw-1(cmp335[HK-AA]), ppw-2(cmp331[HK-AA]), wago-1(cmp207[HK-AA]), wago-4(cmp332[HK-AA]), and wago-10 (cmp240[HK-AA]) mutations, we used an oligo repair template and RNA guide (Supplementary Data 2). All injection mixes included 0.25 μg/μl Cas9 protein (IDT), 100 ng/μl tracrRNA (IDT), 14 ng/μl dpy-10 crRNA, 42 ng/μl gene-specific crRNA, and 110 ng/μl of each repair template. The prg-1[YK-AA] injection mix was injected into USC1232  prg-1(cmp220[(mKate2 +loxP + 3xMyc)::prg-1]) I and USC1273 prg-1(cmp227[2xHA::prg-1]) I. The alg-3[YK-AA] injection mix was injected into USC1092  alg-3(cmp155[(GFP + loxP + 3xFLAG)::alg-3]) IV. The ergo-1[YK- AA] injection mix was injected into USC1046 ergo-1(cmp94[(GFP + loxP +3xFLAG)::ergo-1]) V. The csr-1[HK-AA] injection mix was injected into USC1137  csr-1(cmp173[(GFP + loxP + 3xFLAG)::csr-1]) IV. The ppw-1 [HK-AA] injection mix was injected into JMC225  ppw-1(tor119[GFP::3xFLAG::ppw-1c]) I. The ppw-2[HK-AA] injection mix was injected into JMC221  ppw-2(tor115[GFP::3xFLAG::ppw-2]) I. The wago-1[HK-AA] injection mix was injected into USC 1362 wago-1(cmp92[(GFP + loxP + 3xFLAG)::wago-1]) I. The wago-4[HK-AA] injection mix was injected into YY1325  wago-4(gg620 [3xflag::gfp::wago-4]) II. The wago-10[HK-AA] injection mix was injected into USC1191 wago-10(cmp204[wago-10::2xFLAG]) V. Following injection, F1 animals with the Rol phenotype were isolated and genotyped by PCR to identify heterozygous animals with the mutations of interest, then F2 animals were further singled out to identify homozygous mutant animals. The csr-1[HK-AA] animals are sterile and are maintained in a heterozygous state with the nT1 balancer.

Imaging and quantification

Live imaging of C. elegans was performed on 1-day-adult animals (∼72 h after L1 synchronization) for PRG-1, ERGO-1, WAGO-1, and CSR-1 samples or L4 animals (∼48 h after L1 synchronization) for ALG-3 samples in either M9 buffer containing sodium azide or immobilized with polystyrene microbeads (PolySciences 00876-15) and 25 µM serotonin. Imaging was performed on a DeltaVision Elite microscope (GE Healthcare) using a 60× N.A. 1.42 oil-immersion objective or on a Leica Stellaris 5 confocal microscope using the 63× plan apo N.A. 1.40 oil immersion objective. Images were pseudocolored using Adobe Photoshop.

Quantification of fluorescence intensity was performed in FIJI/ImageJ2 (version 2.14.0). 10 synchronized adult animals were imaged for each sample, and fluorescence intensity from the rachis of the pachytene region of the germline and from a background region was measured. A standardized region of interest (ROI) was used to compare samples: 127.162 pixels² for L4440 to lgg-1, 123.589 pixels² for L4440 to eel-1b and eel-1c, and 90.006 pixels² for L4440 to pas-5. The corrected total cell fluorescence (CTCF) was calculated using the formula CTCF = Integrated Density—(area of ROI × mean background fluorescence).

Bortezomib treatment

Bortezomib (LC Laboratories B-1408) was solubilized in DMSO at a concentration of 100 mg/ml. On the day of bortezomib treatment, plates seeded with OP50 were supplemented with bortezomib to reach a concentration of 5 µg/ml. Once the plates were dried, synchronized L4 animals were transferred to the supplemented plates and treated for 24 h before harvesting.

Western blots

C. elegans were synchronized at 20 °C by bleaching gravid adult animals and were maintained as starved L1 larvae for at least 24 h before plating on OP50. For sample collection, one-day-old adult animals were harvested ∼70 h at 20 °C after L1 synchronization for PRG-1, ERGO-1, CSR-1, PPW-1, PPW-2, WAGO-1, WAGO-4, HRDE-1, and NRDE-3 samples. L4 animals were harvested 48 h at 20 °C after L1 synchronization for ALG-3 and WAGO-10 samples. Approximately 400 L4s or 200 gravid adults were loaded per lane. Proteins were resolved on 4 to 12% Bis-Tris polyacrylamide gels (ThermoFisher NW04122BOX), transferred to nitrocellulose membranes (ThermoFisher LC2001), and probed with mouse anti-Myc 1:1,000 (ThermoFisher 13-2500), mouse anti-FLAG 1:1,000 (Sigma F1804), rat anti-HA-peroxidase 1:1,000 (Roche 12013819001), mouse anti-actin 1:10,000 (Abcam ab3280) or mouse anti-actin 1:5,000 (ThermoFisher MA5-11869), HRP-labeled anti-mouse IgG Secondary 1:10,000 (ThermoFisher A16078), and HRP-labeled anti-rat Secondary 1:10,000 (ThermoFisher A18871). For western blot following bortezomib treatment, synchronized L1 larvae were plated on OP50 for 48 h before being washed off and transferred to plates with OP50 supplemented with bortezomib. Following 24 h of bortezomib treatment, adult animals were harvested. Western blot quantification was carried out using FIJI/ImageJ2 (version 2.14.0). To isolate the protein bands, images were inverted, and the background signal was removed using the Subtract Background function with a rolling ball radius set to 50 pixels. Identical regions of interest (ROIs) were taken for the protein of interest within each blot, and a separate but identical ROI for actin was also used. For each sample, integrated density was measured and normalized to its corresponding actin band. These values were then further normalized relative to the wild-type lane (set to 1) to calculate expression levels.

Immunoprecipitation of ribosomes

A total of 100,000 synchronized adult animals (∼68 h at 20 °C after L1 arrest) were collected in biological triplicates by washing off plates with H₂O and resuspended in 0.15 M KCl and IP buffer (20 mM HEPES pH 7.4, 5 mM MgCl₂, and 15 mM KCl) supplemented with 0.5 mM DTT, 100 µg/ml cycloheximide (Millipore Sigma C7698-1G), 1% Nonidet P40 substitute, RNaseOUT (ThermoFisher 10777019), and protease inhibitors (ThermoFisher A32965). Samples were frozen in liquid nitrogen and homogenized using a mortar and pestle. After further dilution into 0.15 M KCl and IP buffer (1:10 packed worms:buffer), insoluble particulate was removed by centrifugation. Approximately 10% of the sample was taken as “input” and another 10% was used for a BCA assay to determine protein concentration. With the remaining lysate, identical protein concentrations across each sample were used for immunoprecipitation. Immunoprecipitation was performed at 4 °C for 1 h using anti-FLAG Affinity Matrix (Sigma Aldrich A2220), then washed at least three times for 10 min in 0.15 M KCl and IP buffer. RNA was then isolated from the matrix material as described below.

RNA isolation and RT-qPCR

For determining mRNA expression levels, synchronized adult animals (∼68 h at 20 °C after L1 arrest) were collected in biological triplicates. RNA was isolated using TRIzol reagent (ThermoFisher 15596018), followed by chloroform extraction and isopropanol precipitation. RNA samples were normalized to 10 μg/μl prior to DNase treatment (TURBO DNA-free kit, ThermoFisher AM1907) and reverse transcribed with SuperScript III Reverse Transcriptase (ThermoFisher 18080-051). All Real-time PCR reactions were performed using the 2× iTaq Universal SYBR Green Supermix (Bio-Rad 1725121), following the manufacturer's protocols, and run in the CFX96 Touch Real-Time PCR System (Bio-Rad 1855195). Samples were run with three technical replicates for each biological replicate. qPCR to check mRNA levels was normalized to rpl-32.

To determine expression levels of mRNAs associated with actively translating ribosomes, RNA was isolated from the matrix material, normalized, and reverse transcribed, as described above. All Real-time PCR reactions were performed using the 2× iTaq Universal SYBR Green Supermix (Bio-Rad 1725121), following the manufacturer's protocols, and run in the CFX96 Touch Real-Time PCR System (Bio-Rad 1855195). Samples were run with three technical replicates for each biological replicate. Technical replicates with no signal detected or greater than 35 cycles were excluded from analysis. Data was normalized to a control sample lacking the FLAG tag. Primer sequences are available in Supplementary Data 2.

RNAi assays

For RNAi experiments, control L4440, lgg-1, lgg-2, atg-18, epg-5, eel-1b, eel-1c, and pas-5 RNAi E. coli clones were sequenced-verified and cultured at 37 °C for 16 h. RNAi bacteria were then seeded on fresh RNAi plates. Synchronized L1 animals were transferred to seeded lgg-1, lgg-2, atg-18, epg-5, eel-1b and eel-1c RNAi plates and raised at 20 °C for ∼72 h. Adult animals on lgg-2, atg-18, epg-5 RNAi plates were collected for western blot analysis to assess protein expression levels, and adult animals on lgg-1, eel-1b and eel-1c were imaged to assess fluorescence intensity changes. For the pas-5 RNAi assay, synchronized L1 animals were plated on OP50 and raised at 20 °C for 24 h before being washed off the plates and transferred to pas-5 RNAi plates for 48 h, and then imaged to assess fluorescence intensity changes.

Results

PRG-1, HRDE-1, and PPW-2 protein levels depend on small RNA loading

Argonaute proteins contain two conserved RNA-binding domains that facilitate interactions with their small RNA partners: the MID domain, which binds the 5′ end of the small RNA, and the PAZ domain, which recognizes the 3′ end. The MID domain includes a binding pocket formed by a conserved tyrosine-lysine-glutamine-lysine (Y-K-Q-K) motif that anchors the 5′ phosphate of the small RNA (Ma et al. 2005). This 5′ binding pocket is highly conserved across species (Fig. 1a). Argonaute proteins in the WAGO clade often substitute the first tyrosine residue with histidine, resulting in an H-K-Q-K motif (Fig. 1a). This variation may reflect the difference in 5′ end phosphorylation: primary siRNAs are monophosphorylated, whereas secondary siRNAs are triphosphorylated (Ruby et al. 2006; Pak and Fire 2007).

Fig. 1.

Unloaded PRG-1, HRDE-1, and PPW-2 exhibit reduced expression. a) Sequence alignment of the conserved residues (highlighted) in the MID domain of Argonaute proteins across various species. Prefix Af, Archaeoglobus fulgidus; Aa, Aquifex aeolicus; Hs, human; Ce, Caenorhabditis elegans. Asterisks indicate the two residues that were mutated to alanine to generate small RNA-binding-defective Argonaute mutants. b) Western blot analysis of wild-type and small RNA-binding-defective Argonaute proteins. Anti-FLAG antibodies were used to detect GFP::3xFLAG-tagged ERGO-1, ALG-3, CSR-1, WAGO-1, WAGO-4, PPW-1, PPW-2, HRDE-1, NRDE-3, and 2xFLAG-tagged WAGO-10. Anti-Myc and anti-actin antibodies were used to detect 3xMyc::PRG-1 and actin, respectively. Band intensities were quantified using FIJI, and expression levels were calculated as the ratio of Argonaute signal to actin. These ratios were then normalized to wild-type to assess the impact of impaired small RNA binding on Argonaute protein expression. See Source Data file for uncropped western blots.

To investigate the functional consequences of impaired small RNA binding, we engineered small RNA binding-defective mutants representing various Argonaute clades. Previous work showed that mutating the first two residues of the YK/HK motif significantly reduces RNA binding affinity (Ma et al. 2005). Accordingly, we used CRISPR/Cas9 to endogenously tag each Argonaute protein and introduced point mutations replacing the first two residues of the Y/H-K-Q-K motif with alanines (resulting in YK-AA or HK-AA substitutions) (Supplementary Fig. 1a, Table 1). A semi-quantitative RT-qPCR assay confirmed that the HRDE-1[HK-AA] mutant is defective in small RNA association (Chen and Phillips 2024). Similar unpublished experiments from our laboratory indicate that WAGO-1[HK-AA] likewise fails to bind small RNAs (Rajeev and Phillips, in preparation). We then performed western blot analysis to compare the expression levels of wild-type Argonaute proteins with their small RNA binding-defective counterparts. Of the eleven Argonautes tested, most, including the primary Argonaute proteins ERGO-1[YK-AA] and ALG-3[YK-AA], as well as secondary WAGO-clade Argonaute proteins CSR-1[HK-AA], WAGO-1[HK-AA], WAGO-4[HK-AA], WAGO-10[HK-AA], PPW-1[HK-AA], and NRDE-3[HK-AA], maintained protein levels comparable to wild-type. In contrast, PRG-1[YK-AA], HRDE-1[HK-AA], and PPW-2[HK-AA] exhibited significantly reduced protein levels: 5 to 10-fold for PRG-1[YK-AA], 6 to 8-fold for PPW-2[HK-AA], and 2 to 4-fold for HRDE-1[HK-AA], compared to their wild-type counterparts (Fig. 1b, Table 1). The reduction in PRG-1 protein levels observed here is consistent with previous findings showing decreased PRG-1 expression upon loss of piRNA biogenesis factors such as PRDE-1 and PID-1 (de Albuquerque et al. 2014; Weick et al. 2014). Together, our results suggest that, while many C. elegans Argonaute proteins are stably expressed in the absence of small RNA association, PRG-1, HRDE-1, and PPW-2 require proper small RNA loading to maintain their protein levels.

Table 1.

Summary table listing all Argonautes examined, the fluorescent and epitope tags used in this study, the MID-domain mutants analyzed, and their effects on Argonaute localization and protein expression, as well as the degradation pathway responsible for reduced protein levels in the unloaded state.

Argonaute	Tag	Mutation	HK/HK-AA Mutant Localization	HK/HK-AA Mutant Expression	Degradation Pathway
ERGO-1	GFP/3xFLAG	YK-AA	No change	No change
ALG-3	GFP/3xFLAG	YK-AA	Cytoplasmic	No change
PRG-1	mKate2/3xMyc; 2xHA; GFP/3xFLAG	YK-AA	Cytoplasmic	Reduced protein level	Ubiquitin-proteasome system
CSR-1	GFP/3xFLAG	HK-AA	Enlarged germ granules and cytoplasmic	No change
WAGO-1	GFP/3xFLAG	HK-AA	Cytoplasmic	No change
WAGO-4	GFP/3xFLAG	HK-AA	ND	No change
WAGO-10	2xFLAG	HK-AA	ND	No change
PPW-1	GFP/3xFLAG	HK-AA	ND	No change
PPW-2	GFP/3xFLAG	HK-AA	ND	Reduced protein level	Ubiquitin-proteasome system
HRDE-1	GFP/3xFLAG	HK-AA	Germ granules	Reduced protein level	Ubiquitin-proteasome system
NRDE-3	GFP/3xFLAG	HK-AA	Embryonic granules	No change

For YK/HK-AA mutant localization and expression, “No change” indicates no detectable change relative to wild-type and “ND” indicates not determined.

piRNA biogenesis mutants recapitulate PRG-1[YK-AA] destabilization phenotype

Among the three Argonaute proteins with reduced expression upon failure to load small RNAs, we initially focused on PRG-1, which is a primary Argonaute that specifically binds piRNAs. piRNA biogenesis is a complex process involving both transcription of piRNA loci and processing of precursors into mature piRNAs. These include type I piRNAs, which contain a conserved Ruby motif, and type II piRNAs, which lack this motif (Ruby et al. 2006 ). Two key biogenesis factors involved in this process are PRDE-1, a nuclear protein required for the transcription of Ruby motif-containing loci and production of type I piRNAs (Weick et al. 2014), and PID-1, which functions more broadly in global piRNA biogenesis and piRNA precursor accumulation (de Albuquerque et al. 2014; Cordeiro Rodrigues et al. 2019).

To determine whether PRG-1 expression is reduced to a similar extent in small RNA binding-defective mutants compared to piRNA biogenesis mutants, we introduced prde-1(mj207), pid-1(xf35), or the prde-1(mj207); pid-1(xf35) double mutant into a strain expressing epitope-tagged wild-type PRG-1 and examined protein expression by western blot. To minimize potential effects of larger fluorescent tags on protein folding and stability, we used CRISPR to insert a 2xHA tag at the endogenous prg-1 locus. Notably, 2xHA::PRG-1[YK-AA] exhibited a dramatic reduction in protein levels, mirroring the results observed with mKate2::3xMyc::PRG-1 (Fig. 2a), indicating that the observed reduction in protein levels is not tag-specific. In the prde-1 mutant background, PRG-1 expression was reduced compared to wild-type but not as severely as in the small RNA binding-defective mutant (approximately twofold reduction in prde-1 vs 5-fold in PRG-1[YK-AA]). This result likely reflects the fact that prde-1 is required only for type I piRNA biogenesis, allowing PRG-1 to still bind type II piRNAs. In both the pid-1 single mutant and the prde-1; pid-1 double mutant backgrounds, PRG-1 levels were further reduced compared to prde-1 alone, though still not as depleted as in the PRG-1[YK-AA] mutant (approximately 2.5-fold reduction in pid-1 single and prde-1; pid-1 double mutants) (Fig. 2a). These results are consistent with prior findings that prde-1(mj207) and pid-1(xf35) mutants exhibit a strong reduction, but not complete loss, of piRNAs (de Albuquerque et al. 2014; Weick et al. 2014), and suggest that residual piRNAs in these backgrounds may be sufficient to partially stabilize PRG-1. Overall, these results support the conclusion that PRG-1 levels are reduced when it fails to load piRNAs, and indicate that in the absence of piRNAs, PRG-1 does not efficiently associate with other small RNA classes to maintain its stability.

Fig. 2.

Unloaded PRG-1 is mislocalized and depleted post-translationally. a) Western blot showing expression levels of 2xHA::PRG-1, 2xHA::PRG-1[YK-AA], and 2xHA::PRG-1 in prde-1(mj207), pid-1(xf35), and prde-1(mj207); pid-1(xf35) double mutant backgrounds. Actin is shown as a loading control. PRG-1 and actin were detected using anti-HA and anti-actin antibodies, respectively. Band intensities were quantified using FIJI, and expression levels were calculated as the ratio of PRG-1 signal to actin. These ratios were then normalized to wild-type to determine the effect of piRNA binding and biogenesis mutants on PRG-1 protein expression. See Source Data file for uncropped western blot. b and c) Live imaging of germlines from animals expressing wild-type PRG-1, binding-defective PRG-1[YK-AA], or PRG-1 in prde-1(mj207), pid-1(xf35), and prde-1(mj207); pid-1(xf35) backgrounds. Images acquired at identical exposure settings to compare fluorescence intensity b) and with individually optimized exposures to visualize subcellular localization of residual PRG-1 protein c). Scale bars, 15 µm. d) RT-qPCR of prg-1 transcript levels in untagged wild-type (N2), 2xHA::PRG-1, binding-defective 2xHA::PRG-1[YK-AA], and 2xHA::PRG-1 in prde-1(mj207) and pid-1(xf35) mutant backgrounds. Expression is normalized to rpl-32 and relative to expression in wild-type animals. Bar graphs represent the mean of three biological replicates. Error bars indicate standard deviation, and P-values were calculated using a two-tailed t-test. e) RT-qPCR analysis of prg-1 and pgl-1 mRNAs in translating ribosome fractions following immunoprecipitation of FLAG-tagged RPL-4 from 2xHA::PRG-1, 2xHA::PRG-1[YK-AA], and 2xHA::PRG-1 in the prde-1(mj207) mutant background. Enrichment is normalized to 2xHA::PRG-1 lacking FLAG-tagged RPL-4. Bar graphs represent the mean of three biological replicates, excluding technical replicates where the PCR failed. Error bars indicate standard deviation, and P-values were calculated using a two-tailed t-test and adjusted for multiple comparisons.

Altered localization of unloaded argonaute proteins

We next investigated whether Argonaute protein localization is altered when the proteins are unable to bind small RNAs. Previous studies from our lab and others (Guang et al. 2008; Chen and Phillips 2024, 2025) have shown that the nuclear Argonautes HRDE-1 and NRDE-3 mislocalize to the cytoplasm when unloaded. Specifically, unloaded HRDE-1 accumulates in the cytoplasm and germ granules, while unloaded NRDE-3 is enriched in the cytoplasm and, specifically during mid-embryogenesis, cytoplasmic foci in somatic cells.

To extend these findings to additional Argonaute proteins, we examined the localization of a subset of Argonautes whose protein levels remained stable in the YK/HK-to-AA mutants. We found that most Argonautes with germ granule localization in wild-type lost this association when unloaded. For example, both WAGO-1[HK-AA] and ALG-3[YK-AA] failed to localize to germ granules (Supplementary Fig. 1b, Table 1). In contrast, CSR-1[HK-AA] remained enriched in germ granules, though it displayed increased cytoplasmic localization compared to wild-type CSR-1. Germ granules in CSR-1[HK-AA] mutants also appeared enlarged and misshapen, a phenotype consistent with disruption of the CSR-1 pathway (Vought et al. 2005; Claycomb et al. 2009; Updike and Strome 2009). Notably, ERGO-1 was the only Argonaute examined that showed no localization defects when rendered small RNA binding-defective; both wild-type and YK-AA ERGO-1 remained fully cytoplasmic. Together, these results indicate that small RNA loading is broadly required for the correct subcellular localization of many Argonaute proteins, particularly those that normally reside in germ granules or the nucleus.

We next examined the localization of PRG-1, which normally concentrates in germ granules (Batista et al. 2008; Wang and Reinke 2008) (Fig. 2b). Under identical exposure conditions, neither the small RNA binding-defective mutant nor the piRNA biogenesis mutants showed detectable PRG-1 signal, consistent with reduced protein levels when PRG-1 is unloaded. To visualize the remaining PRG-1 protein, we increased exposure (Fig. 2c). The PRG-1[YK-AA] mutant showed a complete loss of germ-granule enrichment and instead exhibited diffuse cytoplasmic fluorescence throughout the germline, similar to the mislocalization observed for unloaded WAGO-1 and ALG-3 (Supplementary Fig. 1b). In the prde-1 and pid-1 single mutant backgrounds, PRG-1 retained partial germ granule localization accompanied by elevated cytoplasmic signal. These observations align with prior studies showing that prde-1(mj207) and pid-1(xf35) mutants retain a small but detectable subset of the normal piRNA pool (de Albuquerque et al. 2014; Weick et al. 2014), and suggest that this residual piRNA population is sufficient to promote limited PRG-1 granule association. In the prde-1; pid-1 double mutant, PRG-1 localization resembled that of the PRG-1[YK-AA] mutant, with predominantly diffuse cytoplasmic distribution and only occasional perinuclear foci (Fig. 2c). Although PRG-1 protein levels in prde-1, pid-1, and prde-1; pid-1 mutants are similar, PRG-1 localization differs markedly. This likely reflects the further depletion and altered composition of the remaining piRNA pool in the double mutant, consistent with PRDE-1 and PID-1 acting at distinct steps in piRNA biogenesis. Together, these findings demonstrate that PRG-1 requires small RNA loading for proper germ granule localization.

Unloaded PRG-1 levels are reduced post-translationally

To determine whether the instability of unloaded PRG-1 is due to degradation of the prg-1 mRNA, we first performed RT-qPCR on the small RNA binding-defective mutant and the piRNA biogenesis mutants. Compared to untagged wild-type, both the tagged prg-1 strain and the mutants showed a modest reduction in prg-1 transcript levels (Fig. 2d), suggesting that the endogenous tag may slightly interfere with proper transcription. However, when comparing prg-1 mRNA levels among the small RNA binding-defective mutant (YK-AA), the piRNA biogenesis mutants (prde-1 and pid-1), and the tagged wild-type strain, we observed no significant differences (Fig. 2d). These results are consistent with previous studies (de Albuquerque et al. 2014; Weick et al. 2014) and indicate that the loss of piRNAs or small RNA loading does not significantly impact prg-1 transcript abundance, supporting a post-transcriptional mechanism of regulation.

Although the reduction of PRG-1 levels appears to occur post-transcriptionally, these results could be due to impaired translation or post-translational degradation. To investigate whether prg-1(YK-AA) mRNA is being translated into protein at similar rates to wild-type, we performed a Translating Ribosome Affinity Purification (TRAP) assay. This method, previously shown to capture mRNAs associated with actively translating ribosomes, allows analysis of translation through RT-qPCR of ribosome-bound transcripts (Nousch 2020). We crossed a FLAG-tagged ribosomal protein (RPL-4) into strains expressing wild-type PRG-1, small RNA-binding-defective PRG-1[YK-AA], and PRG-1 in the prde-1 mutant. RPL-4-associated mRNAs were immunopurified, reverse transcribed, and analyzed by RT-qPCR. As a control, we used pgl-1, a P granule marker known to be actively translated in the germline.

We observed enrichment of pgl-1 mRNA in the TRAP pulldown relative to the negative control (2xHA::PRG-1 without FLAG-RPL-4) across all strains (Fig. 2e), confirming successful capture of translating ribosomes. Similarly, prg-1 mRNA was enriched in the pulldown samples from wild-type, PRG-1[YK-AA], and PRG-1 in prde-1 mutant strains, indicating that prg-1 is being actively translated in all cases (Fig. 2e). These findings suggest that in the absence of small RNA loading, PRG-1 is synthesized but destabilized at the post-translational level, likely through targeted degradation.

Degradation of unloaded PRG-1 is not mediated by autophagy

To maintain proteostasis, eukaryotic cells eliminate damaged, misfolded, or excess proteins through two major degradation pathways: the ubiquitin-proteasome system and autophagy. In Drosophila melanogaster, a previous study showed that loss of small RNA binding causes Ago1 degradation via the ubiquitin-proteasome system (Smibert et al. 2013), whereas in mice, unbound AGO2 is degraded through autophagy (Martinez and Gregory 2013). These findings suggest that different Argonaute proteins may be selectively targeted by distinct degradation pathways, and it remains unclear which mechanism is responsible for degrading unbound PRG-1.

To determine the pathway through which unbound PRG-1 is degraded, we inhibited each degradation system, either by RNAi or using pharmacological inhibitors, and assessed PRG-1 protein levels by western blot. If PRG-1 is targeted by either pathway when unloaded, its protein levels should increase following disruption of that pathway.

Autophagy is a multi-step process in which cytoplasmic components are engulfed by autophagosomes, which then fuse with lysosomes to form autolysosomes that degrade the cargo (Klionsky 2005). To assess whether PRG-1 is degraded via autophagy, we used RNAi to knock down key autophagy components involved at different stages of the pathway: atg-18, required for autophagosome formation (Takacs et al. 2019); epg-5, a metazoan-specific factor essential for autolysosomal degradation (Tian et al. 2010); and lgg-2, which facilitates autophagosome maturation and lysosome tethering (Manil-Ségalen et al. 2014). Western blot analysis revealed that depletion of these factors did not alter PRG-1 protein levels in either the small RNA binding-defective mutant or the piRNA biogenesis mutants (Fig. 3a, Supplementary Fig. 2a), suggesting that autophagy is not responsible for PRG-1 degradation. To further confirm this result, we assessed PRG-1 fluorescence in the germline of prde-1; pid-1 double mutants following RNAi knockdown of lgg-1, an essential autophagy gene involved in autophagosome formation (Manil-Ségalen et al. 2014). As with the western blot results, fluorescence intensity of PRG-1 remained unchanged following lgg-1 depletion (Fig. 3, b and e, Supplementary Fig. 2b), indicating that unloaded PRG-1 is not degraded through the autophagy pathway.

Fig. 3.

Unloaded PRG-1 is degraded through the ubiquitin-proteasome system. a) Western blot of 2xHA::PRG-1 and small RNA binding-defective 2xHA::PRG-1[YK-AA] following RNAi-mediated depletion of autophagy-related genes atg-18, epg-5, or lgg-2. L4440 serves as the RNAi negative control. Actin is shown as a loading control. PRG-1 and actin were detected using anti-HA and anti-actin antibodies, respectively. Band intensities were quantified using FIJI, and expression levels were calculated as the ratio of PRG-1 signal to actin. These ratios were then normalized to wild-type animals on the RNAi negative control (L4440) to determine the effect of knocking down autophagy-related genes on PRG-1 protein expression. b) Live imaging of GFP::3xFLAG::PRG-1 in a prde-1(mj207); pid-1(xf35) double mutant background following RNAi of either a control (L4440) or the autophagy factor lgg-1. Scale bars, 15 µm. See Supplementary Fig. 2b for longer exposures. c) Western blot of GFP::3xFLAG::PRG-1 in wild-type and prde-1(mj207); pid-1(xf35) double mutant animals treated with 5 µg/ml bortezomib to inhibit proteasome activity. Actin is shown as a loading control. PRG-1 and actin were detected using anti-FLAG and anti-actin antibodies, respectively. Band intensities were quantified using FIJI, and expression levels were calculated as the ratio of PRG-1 signal to actin. These ratios were then normalized to untreated wild-type animals to determine the effect of bortezomib on PRG-1 protein expression. d) Live imaging of GFP::3xFLAG::PRG-1 in prde-1(mj207); pid-1(xf35) double mutant animals following RNAi of a control (L4440) or the proteasome subunit pas-5. Scale bars, 15 µm. e) Quantification of GFP::3xFLAG::PRG-1 fluorescence intensity in prde-1(mj207); pid-1(xf35) double mutant animals following RNAi of lgg-1 (left) or pas-5 (right). L4440 serves as a negative control. Fluorescence was quantified from 10 regions of interest across 10 individual gonads per condition. Boxes represent the interquartile range (IQR) from the 25th to 75th percentile; whiskers extend to the most extreme data points within 1.5× the IQR. All individual data points are shown. Statistical significance was determined by two-tailed t-tests. See Source Data file for uncropped western blots.

Unloaded PRG-1 is selectively degraded through the ubiquitin-proteasome system

The ubiquitin-proteasome system targets proteins marked for degradation by ubiquitination and breaks them down via the proteasome (Baumeister et al. 1998). To test whether unloaded PRG-1 is subject to proteasomal degradation, we treated animals with bortezomib, a potent proteasome inhibitor (Lehrbach and Ruvkun 2016). Western blot analysis showed that while wild-type PRG-1 levels are not significantly changed with bortezomib treatment, the levels of unloaded PRG-1 (prde-1; pid-1 mutant) increased 2.5-fold following proteasome inhibition (Fig. 3c), suggesting that unloaded PRG-1 is indeed degraded through the proteasome. Western blot analysis of HRDE-1[HK-AA] and PPW-2[HK-AA] compared to their wild-type counterparts showed similar results, with levels of unloaded HRDE-1 and unloaded PPW-2 increasing 2- and 4-fold, respectively, following proteasome inhibition (Supplementary Fig. 2c), suggesting that unloaded HRDE-1 and unloaded PPW-2 are also degraded through the proteasome.

To corroborate these findings, we examined PRG-1 localization following RNAi depletion of pas-5, the C. elegans ortholog of a 20S proteasome alpha-type subunit (Papaevgeniou and Chondrogianni 2014; Kisielnicka et al. 2018). Because pas-5 knockdown leads to embryonic and larval lethality, RNAi was limited to 48 h and introduced to worms at the L2 stage. Fluorescence imaging of the germline revealed cytoplasmic accumulation of PRG-1 in the absence of functional proteasome activity (Fig. 3, d and e), consistent with impaired degradation. It is important to note that pas-5 knockdown did not restore germ granule localization of PRG-1 in the prde-1; pid-1 double mutant, consistent with PRG-1 remaining unloaded and with piRNA binding being required for granule localization. Together, these results demonstrate that in the absence of piRNA loading, PRG-1 is selectively degraded via the ubiquitin-proteasome system rather than autophagy.

EEL-1 contributes to the degradation of unloaded PRG-1

Proteins targeted for degradation via the autophagy or ubiquitin-proteasome systems are typically polyubiquitinated, with ubiquitin serving as a molecular tag for recognition and processing. Ubiquitination is carried out by a cascade of three classes of enzymes, with E3 ubiquitin ligases conferring substrate specificity by attaching ubiquitin to target proteins (Hershko and Ciechanover 1998). To identify factors that may mark unloaded PRG-1 for degradation, we examined previously published PRG-1 immunoprecipitation-mass spectrometry datasets for candidate interacting proteins (Chen et al. 2020; Suen et al. 2020; Singh et al. 2021). From these datasets, we identified 15 genes with potential roles in ubiquitination (Fig. 4a). To assess their role in PRG-1 regulation, we performed RNAi knockdown of each gene individually in the prde-1; pid-1 double mutant background and examined PRG-1 fluorescence in the germline. Among the 15 candidates, knockdown of eel-1 using two independent RNAi clones resulted in a marked increase in GFP::PRG-1 fluorescence, which was confirmed by quantitative analysis (Fig. 4b). Despite this increase in protein abundance, PRG-1 remained diffusely cytoplasmic and did not re-localize to germ granules, indicating that preventing degradation does not bypass the requirement for small RNA loading in granule targeting. eel-1 encodes an E3 ubiquitin ligase (Ross et al. 2011), suggesting that it could interact with unloaded PRG-1 and facilitate its ubiquitination, thereby targeting it for degradation via the ubiquitin-proteasome system.

Fig. 4.

EEL-1 promotes degradation of unloaded PRG-1. a) List of candidate prg-1 interactors involved in the ubiquitin pathway. Genes were compiled from multiple PRG-1 immunoprecipitation–mass spectrometry datasets. b) Live imaging of GFP::3xFLAG::PRG-1 in prde-1(mj207); pid-1(xf35) double mutant animals following RNAi of a control (L4440) or the E3 ubiquitin ligase eel-1. Two RNAi clones targeting distinct exons of eel-1 (eel-1b and eel-1c) are shown. Scale bars, 15 µm. Quantification of GFP::PRG-1 fluorescence is shown at bottom right. Fluorescence was quantified from 10 regions of interest across 10 individual gonads per condition. Boxes represent the interquartile range (IQR) from the 25th to 75th percentile; whiskers extend to the most extreme data points within 1.5× the IQR. All individual data points are shown. Statistical significance was assessed using two-tailed t-tests. c) Western blot of 2xHA::PRG-1, 2xHA::PRG-1 in the prde-1; pid-1 double mutant background, and 2xHA::PRG-1[YK-AA] animals following RNAi-mediated depletion of two RNAi clones of E3 ubiquitin ligase eel-1. L4440 serves as the RNAi negative control. Actin is shown as a loading control. PRG-1 and actin were detected using anti-HA and anti-actin antibodies, respectively.

Discussion

In this study, we identify a subset of C. elegans Argonaute proteins, PRG-1, HRDE-1, and PPW-2, that exhibit significantly reduced protein levels when unable to associate with small RNAs (Table 1). Using site-specific mutations to disrupt small RNA binding, as well as mutants lacking piRNA biogenesis factors, we demonstrate that the protein abundance of these Argonautes is dependent on successful loading with their cognate small RNA partners. Focusing on PRG-1, we find that its reduction in protein levels occurs post-translationally and is not due to differences in mRNA abundance or translation efficiency. Instead, unloaded PRG-1 is selectively degraded by the ubiquitin-proteasome system, as proteasome inhibition restores PRG-1 protein levels. These findings reveal that C. elegans employs a quality control mechanism to eliminate unloaded Argonaute proteins, potentially protecting germline function and safeguarding genome integrity.

Previous studies have shown that small RNA loading is essential for Argonaute stabilization across eukaryotes. In vitro analysis of human AGO2 revealed that small RNA binding induces a conformational rearrangement that locks the protein into a compact, protease-resistant state, whereas unloaded AGO2 remains conformationally flexible and accessible to proteolytic cleavage (Elkayam et al. 2012). This increased structural flexibility of unloaded Argonautes likely contributes to their reduced stability in vivo and provides a mechanistic basis for selective recognition by protein quality control pathways. Consistent with this idea, Argonaute proteins in multiple organisms are preferentially degraded when not bound to small RNAs (Suzuki et al. 2009; Johnston et al. 2010; Derrien et al. 2012; Martinez and Gregory 2013; Smibert et al. 2013; Kobayashi et al. 2019). In Drosophila, a specific lysine residue on Ago1 has been identified as the ubiquitination site that is selectively targeted in the unloaded state, and is proposed to be buried when Ago1 is small RNA-bound (Kobayashi et al. 2019), providing a compelling model in which conformational masking directly controls Argonaute turnover. Supporting this model, small RNA loading stabilizes Argonaute proteins by bridging flexible domains, particularly between the PAZ and MID domains, thereby shielding degradation-prone surfaces from recognition by ubiquitin ligases (Elkayam et al. 2012). While the specific ubiquitination sites on PRG-1 remain to be identified, it is plausible that, similar to Drosophila Ago1, key lysine residues may become exposed only in the unloaded conformation.

Importantly, our work demonstrates that not all unloaded Argonautes are destabilized. Molecular chaperones such as Hsp90 can bind and stabilize Argonautes in their unloaded conformation, protecting them from degradation (Johnston et al. 2010; Martinez and Gregory 2013), raising the possibility that C. elegans Argonautes differ in their affinity for chaperone-mediated stabilization. In addition, recent studies have identified regulatory regions within the Argonaute N domain that become preferentially exposed in the unloaded state. In plants and mammals, the N-terminal “N-coil” serves as a docking site for regulatory factors, including autophagy components such as ATG8-INTERACTING-PROTEIN1 (ATI1), and this region shows enhanced accessibility in unloaded Arabidopsis AGO1 and human AGO2 (Bressendorff et al. 2023, 2025). Disruption of the N-coil–ATI1 interaction significantly reduces, but does not abolish, degradation of unloaded AtAGO1 in vivo, indicating that this interface contributes to, but is not solely responsible for, Argonaute turnover. Consistent with this idea, additional regions within the N domain can recruit both autophagy- and proteasome-associated factors (Bressendorff et al. 2023), suggesting that Argonaute degradation is governed by combinatorial signals rather than a single determinant. Together, this framework provides a plausible explanation for our observation that only a subset of C. elegans Argonautes are destabilized upon loss of small RNA loading, depending on the exposure of regulatory surfaces, the availability of lysine residues targeted by ubiquitin ligases, and differential stabilization by chaperone interactions.

Beyond influencing stability, small RNA loading affects Argonaute subcellular localization. We find that PRG-1 lacking piRNAs fails to localize to perinuclear germ granules and instead accumulates diffusely throughout the cytoplasm. Similar mislocalization is observed for other Argonautes such as WAGO-1 and ALG-3, which normally localize to germ granules but become cytoplasmic when unloaded. Likewise, nuclear Argonautes HRDE-1 and NRDE-3 are excluded from the nucleus in the absence of small RNAs and relocalize to the cytoplasm, enriching in perinuclear SIMR foci for HRDE-1 or embryonic SIMR granules for NRDE-3 (Guang et al. 2008; Chen and Phillips 2024, 2025). Together, these findings suggest that small RNA loading promotes or stabilizes compartment-specific Argonaute localization, possibly by facilitating interactions with anchoring cofactors or nuclear transport machinery. However, this coupling is not universal; ERGO-1, for example, retains its cytoplasmic localization in the unloaded state, highlighting pathway-specific differences in localization control. Notably, for Argonautes such as PRG-1 and HRDE-1, mislocalization and protein destabilization co-occur upon loss of small RNA loading, raising the possibility that altered localization may precede and contribute to protein turnover rather than arising solely as a consequence of reduced stability. Dissecting the temporal order of these events will be essential for defining how small RNA loading coordinates localization and stability across distinct Argonaute pathways.

Collectively, our findings suggest that small RNA loading is an important determinant of protein fate for a subset of C. elegans Argonaute proteins, influencing their localization, stability, or both. For those Argonautes whose stability depends on small RNA association, this dependence is enforced by a surveillance mechanism that selectively degrades unbound proteins, thereby ensuring that only properly loaded complexes persist. Such quality control likely serves to maintain small RNA pathway fidelity and protect germline function by preventing the accumulation of nonfunctional or potentially disruptive Argonaute proteins. Future studies will be needed to identify additional E3 ligases, define the structural basis of degradation signals in the unloaded state, and determine why only a subset of Argonautes is regulated in this manner.

Limitations of this study

Numerous studies have established the Y/H-K-Q-K motif as a critical determinant of small RNA binding across diverse Argonaute proteins (Ma et al. 2005; Djuranovic et al. 2010; Webster et al. 2015), including our own prior work demonstrating loss of small RNA association for HRDE-1 and NRDE-3 upon mutation of this motif (Chen and Phillips 2024, 2025). However, in the present study, we did not directly assay small RNA binding for every Argonaute examined in Fig. 1b. As a result, we cannot formally exclude the possibility that some Y/H-K-to-AA mutant Argonautes retain the ability to load low levels of small RNAs. Nevertheless, multiple lines of evidence support severe functional impairment of these mutants: CSR-1[HK-AA] displays disrupted germ granule morphology and a fully penetrant sterility phenotype requiring maintenance as a balanced heterozygote, while ALG-3[YK-AA], WAGO-1[HK-AA], and PRG-1[YK-AA] lose normal germ granule association. In addition, unpublished small RNA binding assays from our laboratory indicate that WAGO-1[HK-AA] fails to bind small RNAs (Rajeev and Phillips, in preparation), further supporting the conclusion that mutation of the Y/H-K motif broadly disrupts Argonaute–small RNA interactions.

In this study, we also identify the E3 ubiquitin ligase EEL-1 as a contributor to the degradation of unloaded PRG-1, as PRG-1 protein levels are partially restored in the small RNA-binding-defective background upon eel-1 RNAi knockdown. These findings are consistent with a model in which EEL-1 participates in a quality control pathway that targets unloaded PRG-1 for degradation. However, we did not test eel-1 null alleles, nor did we directly demonstrate EEL-1-dependent ubiquitination of PRG-1, and therefore cannot conclude that unloaded PRG-1 is a direct substrate of EEL-1. Moreover, because eel-1 depletion does not fully restore PRG-1 protein levels, additional ubiquitin ligases or parallel degradation pathways may act redundantly to regulate unloaded PRG-1 stability. Importantly, we also do not know whether other Argonautes, such as HRDE-1 or PPW-2, are similarly degraded by EEL-1. Therefore, it remains unclear whether EEL-1 specifically targets PRG-1 or functions more broadly in the degradation of unloaded Argonautes.

Finally, we have not fully assessed the physiological consequences of the PRG-1[YK-AA] mutation. Although we observed no overt fertility defects at permissive temperature, we did not systematically quantify brood size or examine germline silencing defects, which may differ from those associated with classical prg-1 loss-of-function alleles. More broadly, the biological significance of selectively degrading unloaded Argonaute proteins remains unresolved. Future work aimed at identifying ubiquitination sites on PRG-1 and related Argonautes, and generating degradation-resistant mutants, will be necessary to determine whether unloaded PRG-1, HRDE-1, or PPW-2 is particularly detrimental to cellular or germline function. Such studies may also clarify why only a subset of Argonautes is subject to degradation upon loss of small RNA loading. One possibility is that certain unloaded Argonautes sequester shared cofactors or small RNA pathway components, thereby interfering with functional Argonaute complexes and causing defects more severe than simple loss of Argonaute activity.

Data availability

Strains are available upon request. The authors affirm that all data necessary for confirming the conclusions of this article are represented fully within the article and its tables and figures. A complete list of C. elegans strains, antibodies, and other reagents are provided in Supplementary Data 1, and all oligonucleotide sequences for strain generation and RT-qPCR are provided in Supplementary Data 2. The Source Data file contains all uncropped western blots shown in the article.

Supplemental material available at GENETICS online.

Funding

This work was supported by the National Institutes of Health [R35 GM119656 to CMP and T32 GM118289 to DHN].

Author contributions

J.M.Z.: Conceptualization, Investigation, Formal analysis, Writing–original draft, Writing–reviewing and editing, Visualization. D.H.N.: Conceptualization, Investigation, Writing–original draft. D.C.: Investigation. B.V.: Investigation. C.M.P.: Conceptualization, Formal Analysis, Writing–original draft, Writing–reviewing and editing, Supervision, Funding Acquisition.