PQN-59 antagonizes microRNA-mediated repression during post-embryonic temporal patterning and modulates translation and stress granule formation in C. elegans

microRNAs (miRNAs) are potent regulators of gene expression that function in a variety of developmental and physiological processes by dampening the expression of their target genes at a post-transcriptional level. In many gene regulatory networks (GRNs), miRNAs function in a switch-like manner whereby their expression and activity elicit a transition from one stable pattern of gene expression to a distinct, equally stable pattern required to define a nascent cell fate. While the importance of miRNAs that function in this capacity are clear, we have less of an understanding of the cellular factors and mechanisms that ensure the robustness of this form of regulatory bistability. In a screen to identify suppressors of temporal patterning phenotypes that result from ineffective miRNA-mediated target repression, we identified pqn-59, an ortholog of human UBAP2L, as a novel factor that antagonizes the activities of multiple heterochronic miRNAs. Specifically, we find that depletion of pqn-59 can restore normal development in animals with reduced lin-4 and let-7-family miRNA activity. Importantly, inactivation of pqn-59 is not sufficient to bypass the requirement of these regulatory RNAs within the heterochronic GRN. The pqn-59 gene encodes an abundant, cytoplasmically-localized, unstructured protein that harbors three essential “prion-like” domains. These domains exhibit LLPS properties in vitro and normally function to limit PQN-59 diffusion in the cytoplasm in vivo. Like human UBAP2L, PQN-59’s localization becomes highly dynamic during stress conditions where it re-distributes to cytoplasmic stress granules and is important for their formation. Proteomic analysis of PQN-59 complexes from embryonic extracts indicates that PQN-59 and human UBAP2L interact with orthologous cellular components involved in RNA metabolism and promoting protein translation and that PQN-59 additionally interacts with proteins involved in transcription and intracellular transport. Finally, we demonstrate that pqn-59 depletion reduces protein translation and also results in the stabilization of several mature miRNAs (including those involved in temporal patterning). These data suggest that PQN-59 may ensure the bistability of some GRNs that require miRNA functions by promoting miRNA turnover and, like UBAP2L, enhancing protein translation.


Introduction
Cell fate specification during animal development is tightly controlled to yield highly reproducible outcomes and avoid extreme variation. While individual cell fates can be described by quantifying the expression levels of RNAs that are expressed at any given time, the stable patterns of gene expression for a distinct cell type are often governed by the presence and levels of a relatively limited number of master regulator genes that function at the top of a gene regulation hierarchy. These master regulatory genes typically function in a concentration-dependent manner whereby expression levels above a critical threshold are sufficient to program downstream patterns of transcription in a dominant way. Changes in cell fate specification can occur in a switch-like manner when intrinsic or extrinsic changes lead to reduced expression or activity of a master regulator to levels below a discrete, critical threshold [1]. What common design features of GRNs set these thresholds and what cellular components ensure that the outcomes of sharp changes in master regulator expression are translated into distinct cell fates have been understudied.
The remarkable precision of C. elegans post postembryonic cell fate specification relies on GRNs that exhibit switch-like behavior to ensure normal temporal development. Larval development proceeds through four stages (separated by molts) where the timing of individual cell divisions and the patterning of temporal cell fates is invariant in wild-type animals [2]. Heterochronic genes, encoding transcription factors (TFs) and RNA-binding proteins, organize the sequence of temporal development through the control of stage-specific patterns of gene expression [3]. Importantly, these factors function in dosage-sensitive manners and exhibit sharp temporal gradients of expression (usually from high expression to low expression) that change during inter stage molts. miRNAs play a central role in promoting these sharp transitions from one stage to the next by curtailing the expression of discrete protein-coding genes of the heterochronic pathway that define individual temporal cell fates. This process occurs in at least three separate phases of development (L1-to-L2, L2-to-L3, and L4-to-adulthood) and is mediated by distinct miRNAs that regulate the translational output of independent target mRNAs [4]. Importantly, heterochronic miRNAs also function in dosage-sensitive manners whereby alterations in activity or the timing of their expression lead to the perdurance of target gene expression; resulting in the inappropriate reiteration of earlier patterns of cell division and cell fate specification at subsequent molts [4].
In this study, we aimed to identify candidate genes that modulate the ability of heterochronic miRNAs to elicit switch-like reprograming of temporal cell fates. Specifically, we sought to identify genes whose inactivation would enable hypomorphic alleles of miRNA genes to function normally. These efforts identified pqn-59, encoding a previously uncharacterized, cytoplasmically localized "prion-like" domain-containing protein, as a gene product whose inactivation suppresses loss-of-function phenotypes associated with ineffective miRNA-mediated repression. Consistent with pqn-59 functioning as a modulator of heterochronic miRNA function in temporal cell fate switching, pqn-59 cannot bypass the requirement for these regulatory RNAs in the heterochronic GRN. We demonstrate that the PQN-59 protein exhibits liquid-liquid phase separation (LLPS) properties in vivo and in vitro and that PQN-59 is important for the assembly of RNA/RNA-binding protein assemblies called stress granules during heat shock; a feature shared with its human ortholog UBAP2L. A comparison of proteins associated with PQN-59 indicate that PQN-59, like UBAP2L, is likely highly integrated with proteins involved in mRNA metabolism and translational activation. Finally, we show that pqn-59 depletion may suppress temporal patterning defects in miRNA mutants by promoting aspects of translation as well as increasing the steady state levels of some miRNAs including those involved in temporal patterning. In this capacity, PQN-59 activity functions to set and tailor gene expression levels of critical cell fate determining genes within normal physiological ranges; enabling them to function in GRNs that control bistable cell fate transitions in temporal patterning.

pqn-59 depletion suppresses the heterochronic phenotypes of lin-4 loss-offunction mutants
In order to identify gene products that modulate heterochronic miRNA-dependent developmental events, we performed a genome-wide suppressor screen to identify genes whose inactivation via RNAi could suppress the reiterative heterochronic phenotypes of a hypomorphic allele of the lin-4 miRNA gene. The ma161 allele of lin-4 is defined by a single nucleotide substitution in the mature lin-4 miRNA that reduces its ability to down-regulate its target mRNA, lin-14, which is normally turned off by the second larval stage [5]. As a consequence of these molecular defects, lin-4(ma161) animals continually express LIN-14 throughout larval development and reiterate L1 specific patterns of cell differentiation for each of the somatic blast cells at subsequent molts. Importantly, while the lin-4(ma161) allele generates a small regulatory RNA, its developmental phenotypes are indistinguishable from those completely lacking the lin-4 gene [5,6]. As a consequence of these temporal patterning defects, lin-4(ma161) animals lack vulval structures required for normal egg laying and also fail to induce the expression of an adult-specific col-19::GFP transcriptional reporter after the fourth larval molt (Fig 1A  and 1B). To identify suppressors, we exposed lin-4(ma161) animals harboring the col-19::GFP reporter to individual clones of a genomic scale RNAi library and identified dsRNAs that (A) Genetic screen used to identify suppressors of lin-4(ma161) heterochronic phenotypes. Wild-type animals express the adult-specific col-19::GFP transcriptional reporter at the end of the L4 molt and throughout adulthood. In contrast, lin-4(ma161) animals fail to express col-19::GFP and are unable to lay eggs due to the inappropriate reiteration of L1-specific developmental programs. Depletion of pqn-59 by RNAi suppresses both lin-4(ma161) phenotypes. (B) Pictographs of adult wild-type and lin-4(ma161) animals exposed to control or pqn-59 dsRNAs. See Table 1 for details. (C and D) Pictographs depicting the comparison of adult-specific alae and vulval structures of wild-type, lin-4(ma161), and lin-4(e912) animals exposed to control or pqn-59 dsRNAs. In panel C, solid line indicates continuous adult alae while a dashed line depicts the regions of the cuticle that lack these cuticular structures. In panel D, asterisks could restore normal col-19::GFP expression during adulthood. One of these clones generated dsRNA against pqn-59, a highly conserved and uncharacterized C. elegans gene, that robustly suppressed the reiterative heterochronic phenotypes of lin-4(ma161) mutants to a similar level as other previously described suppressors (including lin-14, lin-28, and lin-42) ( Table 1).
( � ) indicate the location of a normal vulval structures in RNAi treated animals at the L4 to adult molt. (E) Quantification of the wild-type, protruding vulva (pvl), and vulvaless (vul) phenotypes of the indicated animals treated with control or pqn-59 dsRNAs. (F) Images depicting the temporal expression patterns of an endogenous, GFP-tagged allele of the protein product of the major lin-4 target, LIN-14. (G) Quantification of normalized, stage-specific LIN-14::GFP expression in wild-type, lin-4(ma161) and lin-4(e912) animals. Each circle represents the average fluorescence from three seam cells per animal. (H) Diagram indicating the genetic lesions that alter the lin-14 3' UTR regulatory sequences. Black bars indicate the approximate locations of the complementary lin-4 binding sites in the lin-14 3'UTR. (I) Quantification of the levels of suppression mediated by depleting pqn-59 during the development of animals the express one of two gain-of-function alleles of lin-14. Animals were scored positive if any visible alae structures were present on the adult-stage cuticle. P values were calculated using unpaired t test with n.s = P > 0.05, � = P < 0.05; �� = P < 0.01; ��� = P < 0.001; ���� = P < 0.0001. Examination of adult lin-4(ma161) animals that had been exposed to pqn-59 dsRNAs exhibited normal temporal seam cell developmental division patterns and were now able to generate alae production on adult cuticles, indicative of normal seam cell temporal patterning ( Fig 1C). Furthermore, in contrast to control RNAi animals, pqn-59(RNAi) also suppressed the vulvaless (Vul) phenotypes of lin-4(ma161) animals, enabling these animals to lay eggs ( Fig 1D). Surprisingly, depletion of pqn-59 activity in wild-type animals did not induce precocious deposition of adult-specific alae at the L4 molt which distinguishes it from other previously characterized lin-4 suppressors) (Table 1). Furthermore, pqn-59 depletion in wild-type backgrounds only induced a mild, early expression of the col-19::GFP reporter in hypodermal cells found in the head and tail regions (H0, H1, and T cells) of approximately 17% of late L4-staged animals (L4.5 or later [7]).

Suppression of lin-4 loss of function developmental phenotypes by pqn-59 (RNAi) requires miRNA expression and lin-4 miRNA target sites in the lin-14 mRNA
Because the lin-4(ma161) allele generates some lin-4 miRNA, we asked whether depleting pqn-59 via RNAi could completely bypass the requirement for lin-4 during post-embryonic development. To accomplish this, we repeated RNAi-based experiments in animals lacking the lin-4 gene (e.g., lin-4(e912)). These experiments revealed that pqn-59 depletion had little effect on the col-19::GFP expression in lin-4(0) mutants (Table 1). Furthermore, adult-specific alae formation and suppression of the vulvaless (Vul) phenotypes were not restored in this genetic context indicating that pqn-59(RNAi) cannot bypass the normal requirement for lin-4 in temporal patterning (Fig 1C and Table 1). We then aimed to determine how pqn-59(RNAi) modulates the temporal expression of LIN-14 protein, encoded by the major lin-4 target mRNA [8].
We therefore examined the expression of a functional, CRISPR-tagged lin-14::GFP allele in wild-type, lin-4(ma161) and lin-4(e912) genetic backgrounds in experimental conditions where pqn-59 expression is altered. In wild-type animals, LIN-14::GFP expression begins during embryogenesis and is post-transcriptionally down-regulated at the L1 to L2 molt by lin-4 miRNAs [8,9]. In contrast, LIN-14::GFP expression perdures throughout development in both lin-4(ma161) and lin-4(e912) animals ( Fig 1F). When lin-4(ma161) animals were treated with pqn-59 dsRNAs, LIN-14::GFP expression was curtailed by the early L2 stage with similar kinetics as observed in wild-type animals exposed to either control or pqn-59 dsRNAs ( Fig 1F). In contrast, LIN-14::GFP expression is continuously maintained in lin-4(e912) animals in these conditions indicating that pqn-59 depletion does not bypass the requirement for this miRNA in development ( Fig 1F). Surprisingly, we found that LIN-14::GFP expression levels were consistently lower in animals exposed to pqn-59 dsRNAs regardless of genotype ( Fig 1G). Importantly, the pqn-59(RNAi)-dependent reduction of LIN-14::GFP expression is even evident during late embryogenesis, hours prior to lin-4 expression during the L1 stage ( Fig 1G). This suggests that pqn-59 may alter LIN-14::GFP expression independently of lin-4 function. We next examined alleles of lin-14 that generate mRNAs that harbor deletions in the 3'UTR of the lin-14 transcript and have a reduced ability to be targeted by the lin-4 miRNA. Specifically, we examined whether pqn-59(RNAi) could suppress the reiterative, gain-of-function phenotypes of lin-14(n536) and lin-14(n355) which harbor deletions of the lin-14 3'UTR that result in the deletion of five of seven or all of the lin-4 miRNA binding sites in the lin-14 3'UTR ( Fig 1G) [8]. Each of these mutants exhibit lin-4-like phenotypes (with different penetrance) due to the reduced ability of the lin-14 mRNAs produced from these loci to be downregulated the lin-4 miRNA [5]. We found that pqn-59(RNAi) was able to restore adult-specific alae formation in lin-14(n536) mutant animals that retain two of the seven lin-4 binding sites in the lin-14 3' UTR but not a lin-14 truncation allele, lin- 14(n355), that lacks all of the lin-4 binding sites (Fig 1H). These experiments suggest that the suppression of lin-4 reiterative phenotypes mediated by reducing pqn-59 expression requires the ability of lin-4 to both be produced and also able to physically interact with its target mRNA via complementary binding sites.

pqn-59 depletion can suppress other heterochronic phenotypes associated with reduced miRNA activity
We next aimed to determine if the genetic interactions between pqn-59 and lin-4 hypomorphic alleles are specific to developmental events that occur between the L1 to L2 stages of temporal development or whether pqn-59 plays a more general role in antagonizing the activities of additional miRNAs that function later in the heterochronic pathway. We depleted pqn-59 expression in two strains that exhibit distinct, reiterative alterations in the post-embryonic cell lineage due to a reduction in the activity of genetically separate sets of temporally regulated miRNAs. Animals containing the let-7(n2853) allele of the let-7 gene fail to properly down-regulate the expression of LIN-41 during late larval development and as a consequence, animals reiterate the L4 patterns of seam cell divisions during adulthood [10,11]. A single amino-acid substitution (S895F) in one of the two C. elegans microRNA-induced silencing complex (miR-ISC) Argonaute proteins, ALG-1, reduces the efficacy of the three let-7-family miRNAs (mir-48, miR-241, miR-84) that are required to down-regulate the expression of hbl-1, a transcription factor that is essential for L2-specific cell fates [12][13][14]. This regulatory defect leads to an inappropriate reiteration of L2-stage, proliferative seam cell divisions during the L3 stage [6,15]. Both let-7(n2853) and alg-1(ma192) animals fail to express col-19::GFP after the L4-toadult molt (Fig 2 and Table 1). Depletion of pqn-59 suppresses these phenotypes indicting that pqn-59 functions at multiple stages of post-embryonic development to control lateral seam cell fate specification (Fig 2A and 2B). Examination of the adult cuticles and the numbers of lateral seam cells of alg-1(ma192); pqn-59(RNAi) animals indicated that reducing pqn-59 activity during development can 1) restore the ability of these mutant animals to generate adult-specific alae and 2) limit the inappropriate reiteration of L2-stage seam cell division programs that leads to excessive proliferation of seam cells (Fig 2C).

pqn-59 encodes an essential protein that is localized in the cytoplasm throughout development
The pqn-59 gene is located on the left arm of chromosome I and encodes a predicted 712 amino acid protein with a N-terminal ubiquitin-associated (UBA) domain implicated structurally in various protein-protein interactions and in binding both ubiquitin and polyubiquitin chains ( Fig 3A) [17]. Additional analysis of PQN-59 protein structure indicates that it also harbors two separate additional unstructured regions. Immediately after the N-terminal UBA-like domain, PQN-59 harbors a stretch of arginine/glycine-rich sequences (RGG/RG) that are typically associated with RNA binding proteins ( Fig 3B) [18]. The C-terminal portion of PQN-59 harbors three "prion-like" domains composed of amino acid stretches disproportionately enriched for glutamine and asparagine residues (prion-like, Q and N, pqn-) [19] (Fig 3B). A comparison of the sequence conservation and domain architecture of PQN-59 to the proteins expressed in other model systems indicate that PQN-59 is evolutionarily related to a single :GFP phenotypes of let-7 hypomorphic alleles (let-7(n2853)) and antimorphic alleles of genes that encode core components of miRISC (e.g., ALG-1, alg-1(ma192)). See Table 1 for details on expressivity and penetrance of suppression. (C) alg-1(ma192) mutants reiterate the L2-stage or larval development one additional time and as a consequence, over-proliferate lateral seam cells and also fail to form alae structures on the adult cuticle [6].

PLOS GENETICS
PQN-59 antagonizes miRNA activity during C. elegans development Drosophila protein called Lingerer (Lig) and two human proteins named UBAP2 and UBAP2L ( Fig 3C). Lig functions as a growth suppressor that associates with several RNA-binding proteins implicated in the regulation of protein translation (including Rasputin(Rin)/G3BP1 a RasGAP SH3 binding protein (RasGAP-BP), Caprin(Capr), and FMR1 an ortholog of the Fragile X mental retardation protein 1) [20]. Human UBAP2 and UBAP2L proteins have been demonstrated to promote translation [21] and are implicated in the formation of stress granules, liquid-liquid phase separating RNA-dependent condensates that form in the cytoplasm in a variety of averse cellular conditions [22][23][24].
Homozygous pqn-59(tm2960) animals that segregate from a balancer strain develop very slowly and exhibit a severe reduction in brood size (Fig 3E and 3F) [25]. The reduction in fecundity includes a reduced capacity to fertilize oocytes as well as a reduction in embryonic and early larval stage viability [25]. The slow growth, sterility and larval lethality phenotypes of pqn-59(0) mutations can be rescued with a single copy PQN-59::GFP translational fusion targeted to chromosome II but not by related single-copy transgenes encoding PQN-59::GFP variants that lack the UBA-like or "prion-like" domains. This indicates that both of these domains are required for PQN-59 function (Fig 3F). While animals harboring PQN-59(ΔUBA)::GFP or PQN-59(ΔPrD1-3)::GFP transgenes as the sole expressed pqn-59 allele failed to exhibit normal brood sizes, they exhibited no alterations in adult-specific alae formation (S1 Table).
Proteomic analysis of whole worm lysates indicate that PQN-59 is a relatively abundant protein ranking in the top 5% of proteins expressed in embryos and larva ( Fig 3G) [26]. In order to examine PQN-59 expression, we generated single-copy, C-terminal GFP-tagged version of pqn-59 at the endogenous locus. Animals expressing this allele exhibited wild-type development and fecundity [25]. Examination of PQN-59::GFP indicates that it is expressed throughout development and in all somatic and germline cells (Fig 3G, 3H and 3I) (see below). At the subcellular level, PQN-59::GFP is localized exclusively in the cytoplasm (Fig 3H). We also noted that, during post-embryonic development, there is a transient increase in PQN-59 expression in all somatic blast cells during and immediately after cell divisions and that PQN-59::GFP expression is maintained throughout adulthood.

The "prion-like" domains of PQN-59 exhibit LLPS properties
Aside from the structured amino terminal UBA-like domain, a majority of the PQN-59 protein is predicted to contain intrinsically disordered regions (IDRs) and share features with IDRs of several proteins known to exhibit protein condensate formation (Fig 4A) [27][28][29]. Specifically, the carboxy-terminal~300 amino acids of PQN-59 are predicted to harbor three "prion-like" domains (PLDs) that are characterized by stretches of amino acids that are disproportionately enriched in glutamine (Q) and asparagine (N) residues when compared to most proteins in the C. elegans proteome (Figs 4A and S3) [19]. "Prion-like" domain containing proteins have also been demonstrated to exhibit natural "amyloid-like" properties which enable them to form fibril structures in vivo [30]. As a consequence, proteins harboring these IDR and or "PrD-like" domains exhibit a number of interesting biochemical properties including the ability to be co-precipitated by biotinylated-isoxazole (b-isox), which forms crystals in a temperature-dependent manner in aqueous solution and co-precipitates diverse proteins harboring low complexity domains (LCDs) from cell lysates [31,32]. We tested if PQN-59 is precipitated by b-isox by using two approaches. First, we demonstrated that b-isox precipitates PQN-59 from whole worm lysates (Fig 4B). Because PQN-59 could be co-precipitated with other endogenous C. elegans proteins that are directly precipitated by the b-isox compound, we purified a fragment of PQN-59 that includes the "prion-like" domains as a GFP fusion protein and demonstrated that these domains are also precipitated by b-isox compound (Fig 4C). The b-isox does not precipitate soluble GFP [32] or proteins from whole-worm extracts that cross react with anti-PQN-59 antibodies or the minor E. coli proteins that co-purify with GFP-PQN-59(PrD1-3) indicating the specificity of b-isox for proteins harboring LCD and PrDs.
To determine if the "prion-like" domains of PQN-59 also exhibit LLPS properties, purified solutions of recombinant GFP-PQN-59(PrD1-3) were mixed with Ficoll or dextran sulfate. When these crowding reagents were added, the solution became opaque and turbulent with an increased A 600 absorbance (Fig 4D and 4E). When these samples were examined with fluorescence and differential interference contrast (DIC) imaging, they exhibit micron-sized spherical droplets that freely move in solution and wet the surface of the glass coverslips ( Fig 4F). Previous studies have used the aliphatic alcohol 1,6-hexandiol to probe the material properties of proteins that exhibit LLPS [33]. 1,6-hexandiol is thought to disrupt the weak hydrophobic protein-protein interactions that are required for the formation and stabilization of protein condensates [34,35]. Incubation of GFP-PQN-59(PrD1-3) condensates with 1,6-hexandiol dramatically lowers the turbidity and absorbance of GFP-PQN-59(PrD1-3) solutions and eliminates visible condensate formation in microscopy samples (Fig 4D-4F). To fully demonstrate that GFP-PQN-59(PrD1-3) condensates exhibit liquid-liquid-like dynamic properties we performed fluorescence recovery after photobleaching (FRAP) experiments. As demonstrated in Fig 4G, GFP-PQN-59(PrD1-3) condensates formed in 10% dextran sulfate solution rapidly recover fluorescence when a portion of the condensate is bleached (n>10). This rapid recovery was also seen in condensates formed with 150mg/mL Ficoll (n = 20) (Movie S1) indicating that GFP-PQN-59(PrD1-3) forms phase-separated liquid droplets in vitro (Fig 4G). Interestingly, prolonged incubation of these condensates (> 1 week) at 4˚C leads to the formation of a hydrogel-like material that was incapable of being re-solubilized at room temperature in aqueous buffers (Fig 4H).
Under normal growth conditions, PQN-59::GFP exhibits reduced mobility in the cytoplasm and this feature requires the "prion-like" domains We next sought to examine the in vivo properties of PQN-59::GFP by comparing its diffusibility to that of a soluble, monomeric GFP. To accomplish this, we performed FRAP experiments using a strain harboring a single copy PQN-59::GFP (csh150) or a transgene driving soluble GFP driven from the glh-1 promoter. We focused on measuring fluorescence recovery in developing oocytes where both proteins are localized in large and accessible cytoplasmic compartment ( Fig 4I). As would be expected for a small, soluble protein, photobleached regions rapidly recover fluorescence signal in strains harboring soluble, monomeric GFP (Fig 4I). In contrast, photobleached cytoplasmic regions in oocytes expressing PQN-59::GFP recover fluorescence exceptionally slowly suggesting that PQN-59::GFP normally exhibits a limited diffusibility in vivo. We also performed FRAP on PQN-59::GFP in hypodermal cells and found that recovery was also slower than that observed for similar experiments with soluble GFP (S2 Fig). We then sought to determine if the "prion-like" domains contributed to this feature. We integrated a PQN-59::GFP transgene that lacked the three PrDs, csh152[pqn-59(ΔPrD1-3)::GFP], at the same site where we targeted full length PQN-59::GFP above. This fluorescent reporter recovered faster than the full-length PQN-59::GFP reporter indicating that the PrDs of PQN-59 contribute to its reduced diffusibility in the cytoplasm (Fig 4I).

PQN-59 exhibits LLPS in vivo and is required for efficient stress granule formation in developing oocytes
Eukaryotic cells respond to a variety of stresses by reducing protein translation and re-localizing mRNAs and many mRNA-binding proteins [36]. Distinct stresses, including heat and chemicals, induce the formation of ribonucleoprotein complexes, called stress granules, that are formed from pools of untranslated mRNPs. These granules are dynamic, require a number of core proteins for formation and, when formed, show liquid-like behaviors [36]. The human orthologs of PQN-59, UBAP2 and UBAP2L, share a similar domain structure, are predicted to harbor "prion-like" and IDR domains (S3 Fig) and have been demonstrated to localize to and function in the formation of stress granules in a variety of averse cellular conditions [22,23,37].
To determine if PQN-59 also functions in stress granule formation, we subjected animals expressing PQN-59::GFP to heat stress and examined PQN-59::GFP localization in developing oocytes. Heat stress (33˚C; 3hrs) lead to the robust redistribution of PQN-59::GFP from its normal marbled, cytoplasmic localization to a large number of cytoplasmic puncta that vary in size from approximately 0.25-3.5 μm (Fig 5A). To determine if the puncta observed in heat shocked oocytes were indeed stress granules, we also imaged the localization of the core stress granule component and sole G3BP1/2 ortholog in C. elegans, GTBP-1, that directly interacts with PQN-59 [25]. Under these conditions, PQN-59::GFP localization completely overlapped with the localization of GTBP-1::tagRED indicating that PQN-59 is a component of these structures during stress (Fig 5A). This same PQN-59::GFP transgene was localized to similar granules by a variety of stress treatments including arsenite treatment [25]. We used FRAP analysis to demonstrate that the PQN-59::GFP condensates that are induced by heat treatment exhibit LLPS properties as these cytoplasmic structures regain fluorescence rapidly after bleaching. As demonstrated in Fig 5B, photobleached PQN-59::GFP granules rapidly recovered fluorescence at a rate that is much faster than the recovery of bleached regions of PQN-59::GFP in normal growth conditions. We next sought to determine the genetic and structural requirements of PQN-59 and GTBP-1 for the localization of these proteins to heat-induced stress granules. First, we compared the ability of wild-type PQN-59::GFP to localize to stress granules to the dynamics of a version of PQN-59::GFP that lacked the "prion-like" domains. These experiments revealed that PQN-59(ΔPrD1-3)::GFP failed to localize to stress granules suggesting that the prion-like domains we had previously characterized as exhibiting LLPS properties in vitro are required for the LLPS properties of the full-length protein in vivo during stress granules formation ( Fig  5C). In other systems, G3BP proteins are important for the assembly and stabilization of stress granules during heat stress [37,38]. It is thought that G3BP1 in human cells is a central node of the RNA-protein network that triggers the formation of stress granules and this property is directly regulated by RNA binding [39,40]. We tested whether PQN-59::GFP could be efficiently localized to stress granules in the absence of GTBP-1 expression by performing heat shock experiments in gtbp-1(0) animals. Animals lacking gtbp-1 express normal levels of PQN-59::GFP and PQN-59::GFP localizes to stress granules upon heat treatment (Fig 5C). Since a significant portion of PQN-59::GFP is localized to stress granules in the developing oocytes of gtbp-1(ax2069) animals, we then asked if GTBP-1::tagRED association with stress granules requires PQN-59 activity. These experiments revealed that GTBP-1::tagRED localization to stress granules is reduced in pqn-59(0) animals ( Fig 5D) indicating that PQN-59 is required for the efficient localization of GTBP-1::tagRED to these membraneless organelles. A detailed quantification of the dynamics of PQN-59::GFP and GTBP-1::tagRED in stress granules is provided at higher resolution in Abbatemarco et al. [25].

PQN-59 and UBAP2L interact with similar cellular components that are associated with stress granule formation, translation and posttranscriptional gene regulation
To gain insight into the developmental function of PQN-59 in gene regulation, we sought to identify additional proteins that PQN-59 may function with to control developmental gene expression. To accomplish this, we immunoprecipitated PQN-59 from embryos using anti-PQN-59 antibodies and identified associated proteins via mass spectroscopy (S2 Table). We then performed gene ontology (GO) term enrichment analysis of PQN-59 co-precipitated proteins [48]. This analysis indicates that PQN-59 complexes with diverse sets of protein complexes including those enriched in RNA binding (mRNA and rRNA), protein translation (ribosomal proteins, translational initiation, elongation and regulatory factors), transcription (Direct DNA binding factors and co-activators), and transport (nuclear pore complex, motor proteins, microtubule binding) (Fig 6A). In addition to core complexes involved in translation, PQN-59 interacts with ALG-1, a core miRISC component whose mutation elicits phenotypes that are suppressed by pqn-59(RNAi) (Fig 2), and GTBP-1 which colocalizes with PQN-59 in stress granules (Fig 5).
The structural relationships between PQN-59 and human UBAP2 and UBAP2L (and Lingerer) (S3 Fig) as well as their relationship to stress granule formation suggests that these proteins may function in an orthologous manner to regulate gene expression and the formation of stress induced RNA condensates. To determine if PQN-59 and UBAP2L interact with similar types of cellular components, we compared our list of PQN-59 interacting proteins to the list of proteins that were found to physically interact with UBAP2L as measured by BioID assays [24]. This molecular approach utilizes a promiscuous biotin ligase fused to a protein of interest to covalently tag proximally associated proteins in vivo [49]. When this approach was used to identify proteins that function near UBAP2L in vivo, a total of 830 interacting proteins were identified above background [24]. To compare these two lists, we employed the Ortholist server to identify orthologous pairs of human and C. elegans proteins from these lists (S3 and S4 Tables) [50,51]. In addition to extensive, shared interactions between PQN-59 and UBAP2L with ribosomal subunits (large ribosomal subunits 2-7, 9-17, 18-22, 24-26, 28, 30, 33 and small ribosomal subunits 0-3, 7-13, 17, 20, 23, 24, 27, 30), these efforts identified 127 orthologous pairs of proteins ( Fig 6B); indicating that PQN-59 and UBAP2L are associated  [24]. Each group is organized by representative GO cellular compartment term as previously described [24]. The right network illustrates the location of PQN-59-interacting proteins that have orthologs that map to the indicated compartments.

PLOS GENETICS
PQN-59 antagonizes miRNA activity during C. elegans development with overlapping complexes in vivo. We also noted that a large number of PQN-59 interacting proteins that do not share orthologous interactions with UBAP2L are enriched in GO terms associated with germline-and early embryonic-specific functions; proteins that would normally perform tissues specific functions not employed in human cell lines that are derived from somatic sources (S4 Fig).
In order to both visualize PQN-59 interacting proteins into functional groups, we took further advantage of the large-scale application of the BioID proteomic approach outlined by Youn et al. that enabled an ultrastructural view of protein-protein interactions to be organized according to functionally and spatially related RNP-associated complexes (Fig 6C) [24]. When we overlayed the 127 PQN-59 interacting proteins that share orthologous protein-protein interactions with UBAP2 onto this map, we found that that a majority of proteins that are coprecipitated with PQN-59 are associated with other proteins that are found in two types of cytoplasmic condensates: stress granules and P-bodies (Fig 6C). While micron-sized stress granules are typically only visible during averse cellular conditions, multiple lines of evidence indicate that the protein-protein interactions that compose these bodies pre-exists in smaller forms during non-stress conditions [23,24]. Furthermore, core miRISC components are functionally associated with both P-bodies and stress granules where they are thought to function in post-transcriptional regulation of miRNA target mRNAs [52].

Depletion of pqn-59 via RNAi alters translation and miRNA metabolism
Metabolic labeling and polysome analysis of UBAP2L-depeleted human cells indicates that UBAP2L is important for normal translation activity [21]. Our analysis of LIN-14::GFP expression in animals depleted of pqn-59 activity via RNAi suggests that PQN-59 activity is required for normal LIN-14 expression during embryogenesis and L1 developmental stages. To determine if pqn-59 depletion may alter translational activation more generally we took advantage of a fluorescent translational reporter of ATF-4 expression that has been used to measure global reductions in translational efficiency (Fig 7A). ATF4/atf-4 genes in humans and nematodes encode orthologs of Activating Transcription Factor 4 that promote the expression of many genes including amino acid biosynthetic enzymes [53][54][55]. Both hATF4 and C.e.aft-4 mRNAs are negatively regulated at the translational level via 2-3 small open reading frames (uORFs) present in their 5' UTRs [56,57]. Under normal growth conditions where translational efficiency is optimal, these uORFs are efficiently translated [56,57]. Because the terminal 5'UTR uORFs in both human and C. elegans overlap with and are out of frame with the ATF4/ ATF-4 open reading frame, ATF4/ATF-4 expression is normally repressed. In contrast, when translational efficiency is reduced by stress, decreased mTORC1 signaling, or through the addition of translational inhibitors, the reduction in translational initiation enables scanning 40S initiating ribosomes to bypass the uORFs in the ATF4/ATF-4 5'UTR and initiate translation at the first ATG codon of the ATF4/ATF-4 open reading frame [56,57]. When we examined a GFP translational reporter of ATF-4 in wild-type C. elegans, Patf-4::uORF::GFP expression was very low throughout development. In contrast, depletion of pqn-59 via RNAi lead to the abundant expression of GFP (Fig 7B and 7C). The level of induction was similar to levels observed when mTORC1 signaling components are depleted via RNAi or when sublethal levels of cycloheximide are added to the growth media [57]. Therefore, we conclude that pqn-59 depletion in developing nematodes dampens translation.
Given that PQN-59 also interacts with miRISC components (Fig 6C), we reasoned that an additional way by which pqn-59 depletion could suppress the loss-of-function phenotypes associated with reduced miRNA activity would be to alter aspects of miRNA metabolism. To test this hypothesis, we measured the levels of multiple miRNAs in animals that were exposed  . (B and C) L4 or day-1 adult animals were placed on RNAi and GFP fluorescence was measured at the L4 of the offspring. Shown are three independent biological trials. Data is represented as mean + SEM. Dots represent individual animals (N>100). P-value was determined using an unpaired t-test two-tailed. (D) miRNA northern analysis of total RNA extracted from staged, wild-type and alg-1(0), alg-1(gk214), mutant animals. Animals were exposed to control or pqn-59 dsRNAs from hatching to the late L4 stage of larval development prior to RNA extraction. For calculation of relative expression of mature miRNA species, each experimental condition was calibrated to the expression level of 20 μg of total RNA derived from wild-type animals exposed to control dsRNA. SL2 RNA levels were used as an input control. Pre-miRNA to mature RNA was calculated relative to the ratio of the signal in indicated sample. (E) Model of the generic bi-stable switch that incorporates a miRNA to control temporal cell fates during development. Normally (left side), the miRNA target gene is expressed above a critical threshold early in development and is responsible for enforcing earlier cell fate identities. At a defined point in development, the induced expression and activity of a miRNA gene rapidly curtails the target expression below a critical threshold enabling a clear change in cell fate at an important developmental milestone (demarcated by vertical dashed line). In miRNA mutants that underaccumulate functional miRNAs (e.g., lin-4(ma161)), the expression and activity of a mutant miRNA lowers the expression of the target gene but not below the critical threshold to enable a change in the bistable switch. (F) We propose two non-exclusively mutual mechanisms by which depletion of pqn-59 may suppress the defects associated with ineffective miRNA-mediated gene regulation in bistable developmental switches. In Model 1, depletion of pqn-59 may increase the stability of the mutant miRNAs enabling the target gene to be efficiently dampened below the threshold at the developmental milestone. This model is consistent with data shown in above in panel A. Alternatively (Model 2), PQN-59 may function like UBAP2L and stimulate translation of some miRNA target mRNAs. Depletion of pqn-59 would lower the expression of the target gene independent of miRNA expression. This reduced expression would bring miRNA target to control or pqn-59 dsRNAs and extracted total RNA from late L4-staged animals. We then examined both miRNA processing and steady state levels using quantitative northern blots. As shown in Fig 7D, RNAi-mediated depletion of pqn-59 increases the relative abundance of multiple, fully-processed miRNAs including those that regulate temporal patterning (lin-4 and let-7). In addition, pqn-59 depletion altered the levels of other mature miRNA species that are expressed in distinct tissues and temporal expression patterns [58]. Increases in mature miRNA levels during C. elegans development have been demonstrated to arise by elevated expression of miRNAs at the transcriptional level [6,59] or by dampening mature miRNA turnover [60,61]. If pqn-59(RNAi) alters mature miRNA levels by increasing the transcription of affected miRNA genes without improving pre-miRNA processing rates, we would expect pre-miRNA levels to increase in the absence of PQN-59 expression. A comparison of the pre-miRNA levels in control and pqn-59(RNAi) conditions indicate that several pre-miRNA levels are not generally increased in the absence of PQN-59 ( Fig 7D). Because pre-miRNAs for many miRNAs are efficiently processed in wild-type animals (precluding their relative accumulation), we also measured how depleting pqn-59 expression alters pre-miRNA and miRNA levels in animals that harbor null mutations in alg-1, encoding one of the two C. elegans miRNA-specific Argonaute proteins required for normal processing and stabilization of mature miRNAs [62]. Animals lacking ALG-1 expression, alg-1(gk214), exhibit mild heterochronic phenotypes and accumulate the precursor mRNAs for a number of miRNA genes (Fig 7D) [47,62]. Depletion of pqn-59 in alg-1(gk214) animals suppresses the previously described and relatively weak heterochronic phenotypes (30% gapped alae in control RNAi conditions and 0% in animals exposed to pqn-59 dsRNAs). As demonstrated in Fig 7A, depletion of pqn-59 in alg-1(gk214) animals leads to an increase in mature miRNA species for a number of miRNAs without altering the levels of detectible pre-miRNAs. Changes in mature miRNA levels in alg-1(gk214); pqn-59(RNAi) animals were validated using TaqMan assays of independent RNA samples (S5A and S5B Fig). Similar changes in lin-4 and let-7 miRNA levels were also seen when pqn-59 was depleted in alg-1(ma192) antimorphic alleles (S6B Fig). Changes in mature miRNA levels for these miRNAs are not caused by an increase in the levels of two of the main miRISC components, ALG-1, ALG-2, AIN-1 and AIN-2, as levels of these proteins are not altered by PQN-59 depletion (Figs 6D and S6C). These combined results suggest that mature miRNAs are stabilized in the absence of PQN-59 and this increase may enable specific miRNA targets to be more efficiently regulated in these conditions.

Discussion
Bistability is a recurrent motif in developmental biology whereby distinct cell fates, defined by coherent patterns of underlying gene expression, can be switched by the activity of a key regulatory molecule within an established GRN. A fundamental property of bistable gene regulatory networks centers around changes in the temporal expression levels of specific, regulatory factors and the control of these levels around a critical threshold. Above this critical threshold, one cell fate is stable and below it, a distinct cell fate or expression program is enforced. In this manuscript we characterize the C. elegans pqn-59 gene, encoding a conserved RNA-associated protein, as a component that functions to assure that cell fate specification events mediated by miRNAs during post-embryonic development exhibit bistability. We identified pqn-59 in a reverse genetic screen aimed to identify suppressors of temporal patterning phenotypes in lin-4(ma161) animals. We subsequently found that depletion of pqn-59 could also suppress expression nearer to the critical threshold needed to initiate a bistable switch in cell fate transition. In this scenario, the residual activity of the mutant miRNA may now be enough to reduce target mRNA expression further and now below the critical threshold; enabling cell fate transformation in a bistable fashion.
https://doi.org/10.1371/journal.pgen.1009599.g007 retarded heterochronic phenotypes associated with hypomorphic allele of the let-7 miRNA gene and loss of function and antimorphic alleles of an essential miRISC component ALG-1. While pqn-59 depletion via RNAi can efficiently suppress these mutants, we also provide genetic evidence that pqn-59 is not a bypass suppressor of lin-4 and let-7 (i.e., depletion offers little or no suppression of phenotypes associated with null alleles of these genes); indicating that PQN-59 normally functions as a modulator of miRNA activity. Consistent with this interpretation, treatment of wild-type animals with dsRNAs against pqn-59 does not elicit detectible heterochronic phenotypes.
In order to illuminate mechanisms by which PQN-59 may function to modulate heterochronic gene expression, we analyzed phenotypes associated with a deletion allele of pqn-59 gene and also characterized the expression and molecular features of PQN-59 in vivo and in vitro. Several lines of evidence suggest that pqn-59 activities are not limited to regulating features of postembryonic temporal patterning and likely reflect a more widespread role for this protein in development. First, deletion of the pqn-59 gene results in pleiotropic phenotypes including an increase in the length of larval development and a dramatic reduction in fecundity. Furthermore, in contrast to the dynamic expression patterns of other heterochronic genes, PQN-59 is a very abundant and ubiquitously expressed protein. Third, depletion of pqn-59 leads to alterations in protein translation and also the relative stabilization of several mature miRNA species. Because miRNAs are predicted to regulate the expression of a large fraction of expressed mRNAs [63,64], inactivation of pqn-59 may disrupt large portions of gene expression of which some, for example the heterochronic pathway, are exquisitely sensitive to perturbation.
The human ortholog of PQN-59, UBAP2L, is an established RNA-binding protein implicated in multiple biological processes. Analysis of UBAP2L activity in human cells indicates that, like PQN-59, depletion result in a reduction of translational efficiency [21]. Large scale proteomic analysis of UBAP2L-associated proteins indicates that UBAP2L is highly integrated with other RNA-binding protein complexes implicated in stress granule formation and p-bodies, two membraneless cytoplasmic organelles formed through LLPS of RNAs and a multitude of RNA-binding proteins [23,24]. UBAP2L expression is required for the formation of stress granules in adverse conditions [22][23][24]. Both P-bodies and stress granules are highly integrated with RNA-binding proteins that play direct roles in controlling gene expression and RNA turnover; including proteins directly involved in miRNA activity (e.g., miRISC components) [52]. Furthermore, UBAP2L appears to be directly associated with translating ribosomes and can be crosslinked to ribosomal RNAs and the coding sequences of hundreds of mRNAs; indicating that it may play a direct role in controlling translation [21]. Consistent with this hypothesis, artificially tethering UBAP2L to target mRNAs stimulates translation and depletion of UBAP2L results in a decrease in overall translational rates and a reduction in the levels of polysomes [21].
While the level of sequence identity between PQN-59 and UBAP2L is limited (18.6% Identity, 28.0% Similarity (EBLOSUM62)) (S3 Fig), several lines of evidence suggest that these proteins perform orthologous functions. First, PQN-59, like UBAP2L, is predicted to lack overall strong secondary structure and is precipitated by biotinylated-isoxazole, a characteristic of proteins that harbor exhibit LLPS features in vivo [32]. We also demonstrate that the prionlike domains of PQN-59 themselves exhibit LLPS properties in vitro and that during heat stress, PQN-59 localizes to stress granules and exhibits LLPS properties in vivo. Structure-function analyses of PQN-59 domains indicate that the prion-like sequences in the C-terminus of PQN-59 are required for stress granule localization. Second, like UBAP2L, PQN-59 expression is required for the efficient recruitment of GTBP-1, the sole C. elegans G3BP ortholog, to stress granules [23,24,37]. Third, condensation of PQN-59 into stress granules is not dependent on GTBP-1; further mirroring the epistatic relationship between these proteins in human cells [22]. Finally, biochemical characterization of PQN-59 associated proteins indicate that UBAP2L and PQN-59 are physically associated with similar protein complexes in vivo; suggesting that they both may integrate aspects of mRNA metabolism, regulation and expression through these interactions.
Our genetic characterization of pqn-59 during C. elegans development and the potential conservation of functions between PQN-59 and UBAP2L in translation and RNA metabolism enable us to propose at least two non-mutually exclusive models for the function of PQN-59 in temporal patterning. A cornerstone of miRNA function in the heterochronic pathway centers on the rapid downregulation of their target mRNAs below a critical threshold (Fig 7E). In contrast to what happens in wild-type animals, animals lacking full activity of heterochronic miRNAs (as exemplified for lin-4(ma161) mutants) are unable to dampen the expression of their target mRNAs below critical threshold required for the bistable transition in temporal cell fate (Fig 7E). As a consequence, lin-4(ma161) animals exhibit heterochronic phenotypes that are indistinguishable from those in animals that completely lack lin-4 (as exemplified for lin-4(e912) mutants) [6]. As demonstrated in Fig 1F and 1G and Fig 7A-7C, pqn-59 depletion dampens translational output and this reduces the expression levels of the lin-4 miRNA target lin-14. In this context, depletion of pqn-59 may reduce LIN-14 expression to levels closer to the threshold that defines the bistable switch between cell fate specification (Fig 7F Model 1). The compromised lin-4 expression found in lin-4 (ma161) animals may be sufficient in this case to properly reduce LIN-14 expression below the critical threshold to advance development after the L1 stage (Fig 7F Model 1). As demonstrated in Fig 7D, pqn-59 depletion also results in a modest stabilization of many mature miRNAs. We hypothesize that the stabilization of mature heterochronic miRNAs elicited by depleting pqn-59 expression may enable the levels of these miRNAs to a level where they can now effectively dampen lin-14 expression (Fig 7F Model 2). Both of these models are consistent with pqn-59 functioning outside of the normal heterochronic GRN and furthermore explain the observation that pqn-59(RNAi) is an efficacious suppressor of multiple miRNA loss-of-function phenotypes but that pqn-59 depletion is incapable of bypassing the activities of these genes.
Overall, our study demonstrates that pqn-59 functions to modulate gene expression and cell fate specification during C. elegans development. Prior studies have indicated that UBAP2L and its functions with other stress granule and RNA-binding protein partners may play a complex role in a variety of human diseases. These include a role for UBAP2L overexpression in various types of cancer [65][66][67][68][69][70] and in pathologies that are related to protein aggregation in neurodegeneration [71][72][73][74]. We imagine that the genetic and experimental tractability of the C. elegans model will be instrumental in discovering the underlying mechanisms by which this conserved family of proteins functions.

C. elegans maintenance and genetics
C. elegans strains were maintained on standard media at 20˚C and fed E. coli OP50 [76]. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). pqn-59(tm2960) was obtained from Shohei Mitani the National BioResource Project (NBRP) at the Tokyo Women's Medical University. Hypochlorite treatment followed by overnight starvation was used to synchronize animals at L1. C. elegans strains used in this study are listed in S5 Table.

RNAi feeding
RNAi by feeding was performed using E. coli (strain HT115) expressing double stranded RNA corresponding to endogenous C. elegans ORFs (SourceBioScience, United Kingdom) using standard methods [77]. This library included plasmids producing dsRNA against pqn-59 or a control dsRNA expression plasmid (pPD129.36) [78] that does not contain sequence corresponding to any C. elegans gene [79,80]. Positive scoring clones were verified by replicate experiments and dsRNA-targeted gene product was confirmed by Sanger sequencing. To prevent contamination by E. coli OP50, late L4, P 0 animals were added to RNAi plates individually after removing co-transferred bacteria. Unless otherwise noted, F1 progeny were analyzed for RNAi-induced phenotypes following 3-4 days incubation at 20˚C.

Quantification of LIN-14::GFP expression
L4-staged transgenic P 0 animals harboring the endogenously-tagged lin-14::GFP allele were incubated on RNAi plates as above. LIN-14::GFP levels in F 1 progeny were quantified using Image J as in Stec et al. 2021 [81]. For quantifying LIN-14::GFP expression in lin-4(ma161) and lin-4(e912) animals, F 1 animals exposed to control or pqn-59 dsRNAs were cut with the tip of a 30-gauge needle to release the eggs from the vulvaless (Vul) P 0 mothers. For each animal, a total of three seam cells were measured and averaged.

Recombinant protein purification
pCMH1726, encoding the pHIS6-PQN-59(PrD1-3)_YFP protein, was constructed using GIB-SON cloning with a PCR fragment of the pqn-59 cDNA (encoding the terminal 305 amino acids of PQN-59) and pHIS6_Pararallele_GFP vector described in Kato et al.; replacing the Fus(LC domain) ORF in pHIS6-Parallel-FUS(LC) [85]. The His6::GFP::PQN-59(PrD1-3) protein induction and purification was accomplished using previously described protocols [85]. Briefly, BL21(DE3) + pRIPL cells were transformed with pCMH1726 (pHIS6-parallele-GFP:: PQN-59(PrD1-3) and selected on LB + ampicillin + chloramphenicol plates. A single colony was grown overnight in selective media and re-inoculated into 1L of LB medium and grown to an O.D.600 of 0.6. The culture was then cooled on ice for 20 minutes and 0.5mL of 1M IPTG was added. Cultures were then grown overnight at 16˚C. Bacteria were collected by centrifugation and lysed with a sonicator in 35mL of Lysis buffer (50 mM Tris-HCl pH 7.5; 500 mM NaCl; 1% Triton X-100; 20 mM β-mercaptoethanol (BME); 1 tablet of protease inhibitors (Sigma S8830, 1 tablet per 50 mL)). Samples were centrifuged at 35,000 rpm for 30 minutes at 4˚C. The supernatant was removed to a fresh 50 mL conical and further incubated with 2mL of Talon Beads (Qiagen) for 1 hour at 4˚C. Extract slurry was then applied to a column. Beads were washed with approximately 75-100 mL of lysis buffer (above) supplemented with 20mM imidazole. Samples were eluted in 2mL fractions of elution buffer supplemented with 250mM imidazole and quantified for concentration and purity using SDS-PAGE.

Microscopy and fluorescence recovery after photobleaching (FRAP) assays
For imaging of C. elegans larva in Figs 1 and 2, images were acquired with a Zeiss Axio Observer microscope equipped with Nomarski and fluorescence optics as well as a Hamamatsu Orca Flash 4.0 FL Plus camera. An LED lamp emitting at 470 nm was used for fluorophore excitation. For single images, animals were immobilized on 2% agarose pads supplemented with 100mM Levamisole (Sigma). Photobleaching of phase separated droplets or transgenic animals were carried out using a laser on a Nikon Ti-E microscope fitted with a Perkin-Elmer UltraVIEW VoX high speed spinning disk (Yokogawa CSU-X1) laser confocal microscope with live cell imaging capability, time-lapse microscopy, photokinesis, multi-position image acquisition, 6 diode laser lines (405, 440, 488, 514, 561 and 640nm). Imaging was performed at room temperature using spinning-disc confocal microscopy system (UltraVIEW Vox; Perki-nElmer) and a charged-coupled device camera (ORCA-R2; Hamamatsu Photonics) fitted to an inverted microscope (Ti Eclipse; Nikon) equipped with a motorized piezoelectric stage (Applied Scientific Instrumentation). Image acquisition and analysis was performed using Volocity version 6.5 (Quorum Technologies). For droplets, LLPS samples were spotted onto 22mm x 22mm coverslips which were rapidly placed onto the top surface of a slide. After droplets had wetted to the glass surface, individual regions were photobleached. C. elegans samples were prepared as above and subjected to photobleaching in a similar manner to LLPS droplets. Appropriate ROI outside and inside samples (adjacent to bleached regions) were taken as controls. The selected region of interest (ROI) was bleached with a 100% laser power. All the measurements were performed at room temperature and at least in triplicates. For C. elegans animals FRAP, the bleaching of ROI (5-10 μm) was performed at a 100% laser power and 50 iterations, after acquiring 2-20 images before bleaching. The recovery was monitored for 100-300 s. The images were corrected for laser bleaching by selecting a fluorescent region outside the ROI. FRAP data analysis as previously described [86].

Proteomic analysis of PQN-59 complexes
N2 worms were grown on OP50-seeded NGM plates, and embryos were harvested from gravid worms by bleaching (500 mM NaOH, 15% bleach). Embryos were resuspended in IP buffer (100 mM KCl, 50 mM Tris pH 7.5, 1 mM MgCl 2 , 1 mM DTT, 5% glycerol, 0.05% NP40, 1 mM EDTA, Protease Inhibitor Cocktail (Roche)) and frozen in liquid nitrogen. For protein extraction, the embryos were ground on dry ice using a mortar and pestle. The embryonic protein homogenate was thawed on ice and centrifuged at 14,000 rpm for 30 minutes. Equivalent amounts of PQN-59 serum or pre-serum, for control, were incubated with 10 μl of Protein G UltraLink Resin (Thermo Scientific) on ice for 1 hour and 30 minutes. After three washing steps of the beads with IP buffer, approximately 2 mg of embryonic protein homogenate was added and combined samples were incubated for 2 hours on ice. Beads were then washed three times with IP buffer and an additional three times with a "last wash" IP buffer (100 mM KCl, 50 mM Tris pH 7.5, 1 mM MgCl 2 , 1 mM DTT, 1 mM EDTA). PQN-59 complexes were eluted in 0.15% trifluoroacetic acid (TFA). Isolated samples were then frozen on dry ice and subjected to mass spectrometry analysis. Three IP samples from two separate protein homogenates were used in the analysis.

Quantifying preferential translation of ATF-4 reporter
About 8-12 L4 staged or one-day adults LD1499 [Patf-4(uORF)::GFP::unc-54(3'UTR)] [90]. C. elegans were put on RNAi plates containing either EV (empty vector, L4440) or pqn-59(RNAi) and placed at 20˚C. Three to four days later, F1 L4 animals were either inspected with a dissecting scope while on still on the plate. GFP intensity was scored in the following categories: 0 = none or very low GFP usually corresponding to untreated control, 1 = low, 2 = medium, and 3 = high GFP fluorescence visible or mounted on 2% agarose pads and anesthetized in 20 mM levamisole. An upright bright field fluorescence microscope (Tritech Research, model: BX-51-F), containing a custom-made triple band filter set to diminish the contribution of the intestinal gut autofluorescence and equipped with a color camera (model DFK 23UX236) were used for image acquisition at 20X magnification [91]. Image acquisition settings were kept constant across experiments. The total intensity per animal, intensity [a.u.], is calculated from fluorescence images using FIJI [92]. An unpaired t-test two-tailed was used for analysis with Graphpad Prism 9.2.0.

Northern blotting and miRNA Taqman assays
Total RNA was isolated using TRIzol (Invitroen) from staged populations of worms that were exposed to control or pqn-59 dsRNAs from hatching until the late L4 stage of development. Northern blots were performed as previously described [12,93]. Taqman analysis of extracted RNAs and calculations were performed as previously described [94].
Supporting information S1 Fig. pqn-59(tm2960)  The PLAAC: Prion-Like Amino Acid Comparison Server (http://plaac.wi.mit.edu) was used to query the primary amino acid sequences of PQN-59, Lingerer isoform A, UBAP2 and UBAP2L with the following parameters: Core length of 60 and a relative weighting of background probability set to the corresponding species of origin. In protein sequences below each PLAAC graph, red highlighted amino acids indicate glutamine (Q) and asparagine (N) rich sequences identified by this analysis as encoding "prion-like" domains. (TIF) S4 Fig. Depletion of pqn-59 via RNAi does not alter the localization of PGL-1, MEG-1, or MEG-3 to germ granule or CGH-1 localization to somatic P granules. (A) Confocal images of PGL-1mRFP and PQN-59::GFP in late L4 staged F1 animals exposed to control or pqn-59 dsRNAs. (B) Confocal images of CGH-1::GFP expression in the lateral seam cells of L4-staged F1animals exposed to control or pqn-59 dsRNAs. (C) MEG-1::GFP and MEG-3::GFP expression in P cells of F1embryos exposed to control or pqn-59 dsRNAs. (D) PGL-1::GFP expression in early L1-staged F1 animals exposed to control or pqn-59 dsRNAs. (TIF) S5 Fig. PQN-59 Interacting proteins that do not map to UBAP2L-BioID interactors are enriched for GO terms associated with germline and fertility. Relative GO term enrichment (molecular function) scores for proteins that co-precipitate with PQN-59 from early embryos. (EPS) S6 Fig. pqn-59 depletion corrects the reduction of mature lin-4 and let-7 miRNA levels observed in alg-1(0) mutants and has little effect on the expression of core miRISC components. (A) Taqman analysis of lin-4 and let-7 miRNAs isolated from wild-type or alg-1(gk214) animals subjected to control or pqn-59 dsRNAs. In each measurement was standardized by also quantifying the expression of U18 snoRNA in each sample. Error bars represent standard deviation (n = 3 biological replicates, two technical replicates). P-values were calculated using a Student's t-test, and corrected for multiple comparisons using a Bonferroni correction. (B) Taqman assays as outlined in panel A performed on young adult-staged lin-31(n1053) and lin-31(n1053); alg-1(ma192) animals subjected to control or pqn-59 dsRNAs. (C) In the eri-1 (ok2683) RNAi hypersensitive strain pqn-59 depletion decreases PQN-59 to 12% of its abundance in control eri-1(ok2683) animals. In wild-type animals pqn-59 depletion via RNAi decreases PQN-59 to less than half of its abundance in animals fed control RNAi. In contrast, ALG-1 and AIN-1 abundance are minimally affected by pqn-59 depletion. (D) Western blots of miRISC components of N2 animals grown on control or pqn-59 dsRNAs from hatching to adulthood. Western blots were prepared with LiCOR reagents and imaged with a Classic Infrared Odyssey imager. Quantification was performed in ImageJ. (TIF) S1