Distinct roles of two eIF4E isoforms in the germline of Caenorhabditis elegans

In germ cells, translation control of mRNAs that leads to new protein synthesis has been shown to be critical for stem cell maintenance, meiotic entry, completion of meiosis and gamete differentiation (Ciosk et al., 2006; Crittenden et al., 1994; Friday and Keiper, 2015). Current models suggest that RNA regulatory networks composed of RNA-binding proteins (RBPs), microRNAs (miRNAs) and translation initiation factors (eIFs) coordinate splicing, transport, storage, translation and degradation of mRNAs by the assembly of messenger ribonucleoprotein (mRNP) complexes. Repression of mRNAs by RBPs and miRNAs has been extensively studied in various developmental systems (Lai et al., 2011; Pillai et al., 2007; Zahr et al., 2018). However, repression is only part of overall translational control. Recent evidence suggests that selective activation of repressed mRNPs by eIF4 factors that recruit them to ribosomes represents a vital step in translational control (Ghosh and Lasko, 2015; Henderson et al., 2009). However, the mechanisms for the transition remain poorly understood. The prevalence of positive mRNA translational control by eIF4 factors in development has only recently begun to be appreciated (Friday and Keiper, 2015; Keiper, 2019).

Translation initiation factors in the eIF4 group are the first to bind mRNAs and recruit them to ribosomes (Rhoads, 1993). Two regulated components are eIF4E and eIF4G, which bind the mRNA 7-methylguanosine 5′-cap (m7G) and act as a scaffold for ribosome association, respectively. The eIF4E:eIF4G interaction is a critical step for mRNA recruitment (Keiper et al., 1999). This node is regulated by eIF4E-binding proteins (4EBPs) that compete with eIF4G for eIF4E interaction and usually inhibit cap-dependent translation (Gingras et al., 2001). We postulate that the eIF4E:eIF4G interaction may actively dissociate 4EBPs from repressed mRNPs and activate translation. To add greater complexity, multiple eIF4E isoforms and 4EBPs are expressed in all invertebrate, vertebrate and plant species studied (Hernández et al., 2005; Joshi et al., 2005; Patrick and Browning, 2012; Wilhelm et al., 2003). Five eIF4E isoforms (IFE-1 through IFE-5) are expressed from the C. elegans genome. Individual null strains are mostly viable, and, strikingly, loss of each IFE produces a unique phenotype (Keiper et al., 2000). IFE-1 and IFE-3 are the predominant eIF4Es expressed in the germline (Jankowska-Anyszka et al., 1998). We previously showed that IFE-1 is required for completion of spermatogenesis and, to a lesser extent, oocyte maturation (Friday et al., 2015; Henderson et al., 2009). IFE-1 also binds directly to PGL-1 (Amiri et al., 2001), a core component of P granules that sequesters germ cell mRNAs in vivo. PGL-1 is a combined 4EBP RNA-binding protein that is required for normal fertility (Kawasaki et al., 1998). P granules house a large cohort of proteins known to regulate mRNAs post-transcriptionally (Rybarska et al., 2009; Shimada et al., 2006; Voronina et al., 2012). P granules generally localize to nuclear pores and are uniquely poised to regulate the translation and stability of mRNAs exiting the nucleus (Sheth et al., 2010).

In contrast to the role of IFE-1 in the germline, we initially showed that IFE-3 was the only eIF4E essential for embryogenesis in C. elegans (Keiper et al., 2000). Mutations in ife-3 also disrupt germline sex determination (GSD) (Mangio et al., 2015). We hypothesize that two germ cell eIF4E isoforms (IFE-1, IFE-3) are differentially expressed to promote gamete maturation and sperm/oocyte fate, respectively. The isoforms localize to separable mRNPs that regulate distinct subsets of mRNAs; the identified mRNAs match with their prescribed germ cell functions (Fig. 1A). This study describes the unique germline roles of IFE-3 in the sperm-to-oocyte switch and oocyte growth and contrasts them with the very different roles of IFE-1. The GSD pathway relies heavily on mRNA translational control (see Fig. 3A). The pathway allows larval animals to transiently produce sperm, then switch to oogenesis. The GLD-1–FOG-2 complex is an upstream factor in GSD that binds the 3′-untranslated region (3′-UTR) of oocyte-promoting tra-2 mRNA and represses its translation, which initially promotes larval sperm production (Clifford et al., 2000; Goodwin et al., 1993; Jan et al., 1999). The FEM proteins (FEM-1, -2, -3) subsequently promote sperm fate by degrading TRA-1, the transcriptional repressor of the terminal sperm-promoting genes fog-1 and fog-3 (Starostina et al., 2007). In contrast, oocyte cell fate is promoted by TRA-2 inhibition of FEM-3, which allows TRA-1 to repress fog-1 and fog-3 (Mehra et al., 1999). fem-3-binding factors (FBFs) also promote oocyte cell fate by binding the 3′-UTRs and repressing translation of fem-3 and fog-1/-3 mRNAs (Thompson et al., 2005; Zhang et al., 1997). Translational control dominates GSD gene expression; however, the means by which opposing fate signal mRNAs, like tra-2 and fem-3, are coordinated are unclear. How are these signals integrated via negative and positive translational control to allow sperm production in larvae, followed by oogenesis in the adults? Are eIF4E isoforms and their cognate 4EBPs playing a role in this translational control?

Here, we distinguish the roles of IFE-1 and IFE-3, further demonstrating that eIF4E isoforms have different functions during development. We provide evidence that IFE-3 regulates mRNAs in a manner that is distinct from that of IFE-1 to promote oocyte fate and regulate GSD. We also describe the differential expression of IFE-1 and IFE-3 during hermaphrodite germline development using dual-fluorescent CRISPR/Cas9 tagging to show that their expression patterns correlate with their functions. IFE-3 associates with germ granules, but the timing, morphology and developmental pattern is distinct from that of IFE-1 in P granules. IFE-1 and IFE-3 localization depends upon their cognate 4EBPs. Our genetic, biochemical and expression data suggest differential dormant and active periods for IFE-1 and IFE-3 during the progression of germline development.

Comparing differential eIF4E isoform expression in the germline might provide insight into their divergent functions as well as the distinct phenotypes of each null mutant. To obtain a reliable expression pattern for each isoform, we edited endogenous ife-1 and ife-3 genes using CRISPR/Cas9 to generate N-terminally fluorescent-tagged mKate2::IFE-1 and GFP::IFE-3, respectively (Dickinson et al., 2015). Strains homozygous for each fusion were fertile and superficially wild type, indicating that each was functional as a translation factor. Both eIF4Es were robustly expressed in the gonad throughout development, but their expression and localization patterns were distinct. Both IFE-1 and IFE-3 localized to perinuclear germ granules, but were also soluble in the cytoplasm of germ cells in the distal gonad (mitotic to early pachytene region; Fig. 1B–D). Our previous study found IFE-1 in embryonic P granules (Amiri et al., 2001). Intriguingly, we observed separate perinuclear localization of IFE-1 and IFE-3 in a strain expressing both IFE fusions (Fig. 1B–D, insets, arrowheads). IFE-1 granules were larger (570±13 nm, n=81) than IFE-3 granules (366±7 nm, n=74) and more numerous. Another striking difference between IFE-1 and IFE-3 was seen in the rachis (gonad core), where IFE-3 associated with lattice-like structures (Fig. 1E–G). IFE-1 did not form similar structures (Fig. 1F). In the distal gonad, the soluble fractions of IFE-1 and IFE-3 were overlapping, unlike their respective granules (Fig. 1B,E, merged images, yellow).

In spermatocytes and oocytes, IFE-1 and IFE-3 localization patterns became even more divergent. Using nuclear morphology to stage spermatocytes (Fig. 1H–K), we observed that IFE-1 was upregulated in primary (1°) spermatocytes, where it formed perinuclear granules, then became largely soluble in secondary (2°) spermatocytes. After spermatid budding, IFE-1 was deposited into residual bodies (Fig. 1I). Conversely, IFE-3 was soluble throughout spermatogenesis and was diminished early in 2° spermatocytes (Fig. 1J). In oocytes, IFE-3 was strongly expressed and fully soluble, whereas IFE-1 formed granules (Fig. 1L–O, arrowheads). Our observations suggest that each eIF4E isoform forms distinct granules in early germ cells, but then each becomes soluble in the respective gametes in which they predominate, IFE-1 in spermatocytes and IFE-3 in oocytes.

The ok191 allele deletes the promoter and first exon of ife-3 (Fig. 2A). Stably balancing this mutation proved difficult because the ife-3 locus on the end of chromosome V is a hotspot for meiotic recombination (Mangio et al., 2015). It was therefore necessary to genotype worms from each generation for each experiment using a triple-primer genomic PCR to unambiguously confirm ok191 (Fig. 2A). Eggs from ife-3(+/−) mothers carry sufficient maternal IFE-3 for survival of some ife-3(−/−) offspring (Mangio et al., 2015). The escaping ife-3(−/−) hermaphrodites developed to adulthood but were sterile at 20°C. Their dissected gonads showed extensive sperm production and lack of oocytes, as evidenced by immunostaining for major sperm protein (MSP) and lack of LIN-41 (oocyte marker) (Fig. 2D). Wild-type hermaphrodite (Fig. 2B) and male (Fig. 2C) germlines are shown for comparison. These data demonstrate that ife-3 mutant hermaphrodites have masculinized germlines (Mog) and fail to produce oocytes at 20°C. Surprisingly, at elevated temperatures (25°C), at least 21% of ife-3(−/−) hermaphrodites were able to initiate oogenesis (Fig. 2E). Oocytes were evident in the proximal gonad by their expression of LIN-41 but did not grow normally. These results indicate that IFE-3 has roles in oocyte development beyond the sperm-to-oocyte switch. This phenotype differs strikingly from that of ife-1(bn127) worms, which arrests late spermatocyte differentiation and maturing oocytes (Friday et al., 2015; Henderson et al., 2009). Temperature obviously plays a role in the sterility caused by both ife-1 and ife-3 mutations. However, the sterility seen in ife-3 worms is alleviated at elevated temperatures, whereas the sterility seen in ife-1 worms is exacerbated by it.

We determined the prevalence of the ife-3 sterility at 20°C and 25°C. No heterozygous ife-3(+/−) hermaphrodites were sterile at 20°C (n=61) or 25°C (n=49) (Fig. 2F). However, 100% of ife-3(−/−) hermaphrodites were sterile (n=60) at 20°C, due to the Mog phenotype. Interestingly, only 79% of ife-3(−/−) hermaphrodites were sterile (n=62) at 25°C, indicating a ∼20% rescue of oocyte fate (Fig. 2F). The ife-3(−/−) mutants had smaller brood sizes (84±7, n=4) compared to wild-type animals (221±10, n=3) at 25°C (data not shown), and the embryos from ife-3(−/−) mothers never hatched (Fig. 2F). This observation supports our previous finding using ife-3(RNAi), which showed substantial embryonic lethality (Keiper et al., 2000). Curiously, the distal gonads of ife-3(−/−) hermaphrodites were occasionally bifurcated at 20°C (8%, n=60) and at 25°C (5%, n=62) (data not shown). These phenotypes are unique to ife-3 among the eIF4E isoforms, since neither the Mog phenotype nor the aberrant gonad morphology was ever observed in ife-1(bn127), ife-2(ok306), ife-4(ok320) or ife-5(ok1934) animals (Amiri et al., 2001; Dinkova et al., 2005; Henderson et al., 2009; Song et al., 2010). Lastly, we addressed whether IFE-3 was required for male-derived sperm to fertilize oocytes upon mating with fog-2(q71) females, as was previously found for IFE-1 (Henderson et al., 2009). Soaking males in SYTO17 labels their sperm for detection after mating (Fig. S2A–D). Although ife-3(−/−) males had somewhat decreased mating efficiency at 25°C, their sperm were able to fertilize eggs in 100% of fog-2(q71) females at both 20°C and 25°C (Fig. S2B,E). Males lacking IFE-1 produce very few sperm at 20°C (Henderson et al., 2009), but still fertilized eggs in 100% of fog-2(q71) females. By contrast, ife-1(−/−) males produced no sperm at 25°C, resulting in no fertilization events (Fig. S2D,E). These data establish that IFE-3 is dispensable for producing functional sperm regardless of temperature, in stark contrast to IFE-1, which is required for spermatogenesis at 25°C.

Many mutations that alter the sexual fates of germ cells have been identified in C. elegans (Barton and Kimble, 1990; Schedl and Kimble, 1988). Most of the corresponding GSD genes encode RBPs or mRNAs within a translational control network that culminates in the conditional synthesis of FOG-1/-3, which drives sperm fate (Fig. 3A). To understand where ife-3 might function in this network, we conducted epistatic analyses with mutations in components of the GSD pathway in conjunction with IFE-3 depletion. We chose to use ife-3(RNAi) for these experiments due to the embryonic lethality of ife-3(ok191) and the instability of the balanced mutation, both of which complicate the derivation of double mutant strains. Following ife-3(RNAi) treatment, 22% (n=157) of wild-type hermaphrodites failed to produce oocytes and developed the Mog phenotype, as evident by extensive MSP staining and apparent lack of oocytes (Fig. 3B,D). The penetrance of ife-3(RNAi)-induced sterility is lower than that of the null allele (22% vs 79% at 25°C, respectively). By comparison, 100% (n=129) of control(RNAi)-treated wild-type hermaphrodites showed both sperm and oocytes (Fig. 3B,C). To rule out the possibility of weak ife-3(RNAi) efficacy, we also performed ife-3(RNAi) in an RNA interference (RNAi)-sensitized strain eri-1(mg366), but obtained similar frequencies of Mog phenotype (20%, n=86; Fig. 3B).

To place ife-3 within the GSD pathway, we looked for reversion of feminizing mutations or enhancement of Mog phenotype by ife-3(RNAi). The fog-2(q71) mutation results in feminized hermaphrodite germlines (Fog phenotype) that never produce sperm (Schedl and Kimble, 1988) and is an upstream factor in the GSD pathway (Fig. 3A). Consistent with this, Fog gonads were observed in 100% (n=117) of control(RNAi)-treated fog-2(q71) hermaphrodites, as evident by the absence of MSP staining and the presence of large stacked oocytes (Fig. 3B,E). By contrast, 23% (n=127) of ife-3(RNAi)-treated fog-2(q71) hermaphrodite gonads developed as Mog, showing extensive MSP staining and no oocytes (Fig. 3B,F). IFE-3 depletion was thus able to revert the fog-2(q71) feminizing mutation at a similar frequency to wild-type animals, indicating that ife-3 promotes oocyte fate downstream of fog-2. Another feminizing mutation, fem-3(e2006), also causes hermaphrodites to develop Fog at 25°C but not 20°C (Hodgkin, 1986). As expected, none of the control(RNAi)-treated fem-3(e2006) hermaphrodites were Mog at 25°C (0%, n=102; Fig. 3B,G). None of the ife-3(RNAi)-treated worms (0%, n=116) developed as Mog, as shown by the lack of MSP staining (Fig. 3B,H). IFE-3 depletion was therefore unable to revert the fem-3(e2006) feminizing mutation, indicating that ife-3 promotes oocyte fate at a step genetically upstream or at fem-3, consistent with the findings of Mangio et al. (2015). FBF-1 and FBF-2 (fem-3-binding factors) are homologous RBPs that repress translation of fem-3 and fog-1/-3 mRNAs to promote oocyte cell fate (Thompson et al., 2005; Zhang et al., 1997). fbf-1(ok91) mutants are slightly masculinized (1% Mog), and fbf-2(q738) mutants are slightly feminized (1% Fog), for reasons that remain unclear (Lamont et al., 2004; Yoon et al., 2017). ife-3(RNAi) masculinized 43% (n=192) of fbf-1(ok91) hermaphrodites (Fig. 3B,J), whereas only 1% displayed Mog phenotype upon control(RNAi) (Fig. 3B,I). Therefore, IFE-3 depletion enhanced Mog penetrance 2-fold in fbf-1 mutants relative to ife-3(RNAi)-treated wild-type animals. Similarly, enhanced Mog penetrance was seen in fbf-2(q738) hermaphrodites (Fig. 3B). These data show that ife-3 and fbf genes act synergistically to promote oocyte fate, suggesting that IFE-3 works with FBFs in translational control.

A substantial number of ife-3(RNAi) hermaphrodites were not Mog, but developed small, stunted oocytes that failed to develop normally (Fig. 3B,H). This was first observed in some ife-3(−/−) null worms grown at 25°C (Fig. 2E) but was more consistent upon ife-3(RNAi). Small oocytes were observed in 39–80% of IFE-3-depleted worms, depending on the genetic background, and was most prevalent when masculinization could not be achieved (Fig. 3B,H; e.g. fem-3). Small cells were negative for MSP staining and were confirmed to be oocytes by LIN-41 expression (Fig. 4A). Staining with 4′,6-diamidino-2-phenylindole (DAPI) confirmed aberrant nuclear morphology, indicating defects in meiosis (Fig. 4A, insets). Both the RNAi and null mutation phenotypes indicate that IFE-3 must have roles beyond GSD in oocyte growth and completion of meiosis.

Because eIF4E isoforms bind to mRNA caps and catalyze translation initiation, we postulated that IFE-3 loss might adversely affect the translation of GSD pathway mRNAs. We directly analyzed the translational efficiency of several GSD mRNAs in control and ife-3(RNAi)-treated wild-type animals using high-resolution polysome profiling (King and Gerber, 2014). Our previous studies used identical methodology to analyze mRNAs that respond to IFE-1, IFE-2 and IFE-4 deficiencies in vivo (Dinkova et al., 2005; Friday et al., 2015; Henderson et al., 2009; Song et al., 2010). Each IFE isoform previously investigated positively regulates a unique subset of mRNAs. Large scale ife-3(RNAi) over a single generation provided sufficient material for sucrose fractionation of whole worm populations and allowed mRNAs to be resolved by their polysome loading (Fig. 4A). RNAi treatment induced Mog phenotype in a minority of the worms, whereas most (78%; Fig. 4A) produced primarily oocytes. IFE-3 depletion was confirmed under these conditions (Fig. S1A–C). Overall polysome content relative to 80S was modestly decreased by ife-3(RNAi), indicating that IFE-3 depletion does not dramatically alter bulk protein synthesis. Individual mRNAs were quantified across gradient fractions (Fig. 4B–H). The translation of fem-3, fog-1, fbf-1, gld-1 and tra-2 were altered upon ife-3(RNAi) treatment when compared to control(RNAi)-treated animals (Fig. 4B–F). Unexpectedly, several GSD mRNAs translated more efficiently after IFE-3 depletion. A substantial fraction of non-translating fem-3 mRNA (fractions 1–4) in control(RNAi) worms became shifted into light polysomes (fractions 5–8) upon ife-3(RNAi), indicating that IFE-3 deficiency allowed for more efficient initiation of fem-3 mRNA (Fig. 4B). Strikingly, the bulk of fog-1 mRNA also shifted from non-translating fractions into both light and heavy polysomes upon IFE-3 depletion (Fig. 4C). Polysomal loading of fog-1 mRNA was therefore greatly enhanced, particularly in the heavy fractions (9–11), where many ribosomes translate simultaneously. The synthetic output of FOG-1 is expected to be higher and might contribute to the Mog phenotype observed (Fig. 3). fbf-1 and gld-1 mRNAs experienced more modest increases in translational efficiency after IFE-3 depletion. For these, some non-translating mRNA (fractions 1–4) appeared to shift to light and heavy polysomes (Fig. 4D,E). Given the amount of fbf-1 and gld-1 mRNAs already translating in control(RNAi) worms, the output of these proteins may not be substantially changed. tra-2 mRNA was less efficiently translated after IFE-3 depletion, as shown by a shift to light polysomes (Fig. 4F, fractions 5–8) from heavy polysomes (Fig. 4F, fractions 9–11). The effect is modest, but a positive dependence of tra-2 translation on IFE-3 is consistent with the role of IFE-3 in promoting oocyte fate. Importantly, the translational efficiency of some mRNAs was unchanged upon ife-3(RNAi), including the housekeeping mRNAs GAPDH (gpd-3; Fig. 4G) and β-tubulin (tbb-2; not shown). Neither show preferential dependence on any single IFE (Dinkova et al., 2005; Friday et al., 2015; Song et al., 2010). More strikingly, mex-1 mRNA translational efficiency was also unchanged (Fig. 4H). We previously showed mex-1 mRNA translation to be strongly IFE-1 dependent (Henderson et al., 2009). By summing mRNA content across gradients, relative to gpd-3 mRNA, we found that, with the exception of fog-1, all germline mRNAs assayed were decreased ∼25–50% in ife-3(RNAi)-treated animals (Fig. 4I). This is consistent with ife-3(RNAi)-treated germlines being smaller than their control(RNAi)-treated counterparts (Fig. 4A). Therefore, changes in the translational efficiency seen were not due to alterations in the abundance of the mRNAs. These data suggest that IFE-3 is involved in the translation of several GSD mRNAs, but not housekeeping mRNAs, nor those regulated by IFE-1. IFE-3 acts to repress fem-3, fog-1, fbf-1 and gld-1 mRNA translation, and possibly enhance tra-2 mRNA translation. While a negative translational role for an initiation factor was unexpected, it is interesting that the mRNAs repressed by IFE-3 are also well-defined FBF targets (see Discussion and Kershner and Kimble, 2010).

Of the mRNAs assayed translationally, only fog-1 mRNA levels appeared to increase upon ife-3(RNAi). We wondered whether these changes were due to IFE-3 loss or the 22% of Mog animals in the population (Figs 3B and 4A,I). It is possible that the levels of sperm-promoting mRNAs (fem-3 or fog-1) increase following complete loss of IFE-3. Indeed, some studies suggest that steady-state fem-3 mRNA levels increase in certain Mog backgrounds (Zanin et al., 2010), while others demonstrated that fog-1 transcription is enhanced under conditions that favor spermatogenesis (Luitjens et al., 2000). We determined the relative fem-3 and fog-1 steady-state mRNAs levels relative to tbb-2 mRNA in ife-3(−/−) L4 hermaphrodites and compared them to wild-type animals. Early and late L4 hermaphrodites were analyzed in order to compare equivalent spermatogenic germlines (avoiding comparison of disparate oogenetic and Mog germlines). We found no statistically significant difference (P=0.328) in fem-3 mRNA levels between ife-3(−/−) and wild-type early L4 hermaphrodites (Fig. 5A). However, a large and statistically significant increase (9-fold; P=0.007) in fog-1 mRNA levels was found in early L4 ife-3(−/−) hermaphrodites compared to wild-type animals (Fig. 5B). We also examined late L4 worms to determine if the changes in fem-3 and fog-1 mRNA levels were maintained through the sperm-to-oocyte switch, since these animals are expected to have initiated oogenesis. The fem-3 mRNA levels were decreased by half in ife-3(−/−) worms relative to wild type (P=0.018; Fig. 5A), again consistent with having smaller germlines. However, fog-1 mRNA levels remained 6-fold higher in ife-3(−/−) worms (P=0.019; Fig. 5B). These data suggest that IFE-3 loss resulted in enhanced transcription/stabilization of fog-1 mRNA even prior to the sperm-to-oocyte switch; fem-3 mRNA levels do not change prior to the sperm-to-oocyte switch, but drop at later stages, likely due to the size of the gonad relative to the soma.

We confirmed IFE-1 residence in P granules of distal germ cells by colocalization with a PGL-1::GFP CRISPR fusion. We observed a high degree of overlap for IFE-1 and PGL-1 in germ cells (Fig. 6A–C, insets, arrowheads). IFE-1 expression was notably upregulated in 1° spermatocytes from L4 hermaphrodites, where it was enriched in PGL-1 granules (Fig. 6D–F). Importantly, IFE-1 left PGL-1 granules and became soluble when PGL-1 disappeared from 2° spermatocytes (Fig. 6E, dashed line). It is likely that PGL-1 stabilizes IFE-1 in P granules. Similarly, IFE-1 was present in PGL-1 granules that moved toward the cortex as oocytes approached the spermatheca (Fig. 6G–I, arrowheads). These data link IFE-1 tightly to PGL-1 in P granule dynamics. Furthermore, IFE-1 becomes completely soluble during late spermatogenesis, at a time and place during development that matches the ife-1(bn127) phenotype (Amiri et al., 2001; Henderson et al., 2009). We wondered whether the perinuclear IFE-3 granules in distal germ cells also colocalized with PGL-1, considering that IFE-1 and IFE-3 granules appeared distinct (Fig. 1B–D). Unlike IFE-1 foci, the IFE-3 foci were adjacent but clearly separate from PGL-1 (Fig. 6A vs J, insets, arrowheads). We often observed multiple IFE-3 foci decorating a singular PGL-1 granule (Fig. 6J–L, insets, top arrowheads). This was never observed for IFE-1. Also, in contrast to IFE-1, IFE-3 expression levels were not obviously upregulated during spermatogenesis but remained constant (Fig. 6N vs E) and largely soluble rather than enriched in granules (Fig. 6M–O). IFE-3 was strongly expressed in large oocytes, where it also appeared to be completely soluble (Fig. 6P–R, arrowheads). Considering that oocyte growth and embryogenesis have a greater requirement for IFE-3 (Fig. 2E) than IFE-1 (Henderson et al., 2009), soluble IFE-3 likely participates in protein-synthetic events that contribute to both. In healthy oocytes and spermatocytes, IFE-1 localization is tightly linked to its cognate 4EBP, PGL-1, whereas IFE-3 localization is not. Finally, the major soluble eIF4E isoform in spermatocytes is IFE-1, whereas IFE-3 is the abundant soluble isoform in oocytes.

IFET-1 is the worm ortholog of mammalian eIF4E transporter protein, which facilitates nucleoplasmic shuttling of eIF4E (Dostie et al., 2000). IFE-3 and IFET-1 are found in OMA-1 mRNP complexes in vivo (Spike et al., 2014). Others have shown that recombinant IFET-1 is able to bind each of the recombinant IFEs except IFE-4 (Li et al., 2009). Although IFET-1 shows the potential to bind several IFEs in vitro, isolation of native mRNP complexes provides evidence of the IFE-3:IFET-1 interaction only (Spike et al., 2014; M.S.S. and P.R.B., unpublished). Thus, IFET-1 is thought to represent a cognate 4EBP for IFE-3. We hypothesized that loss of IFET-1 would disrupt IFE-3 localization, while having no effect on IFE-1 localization. We depleted IFET-1 by RNAi and re-examined localization patterns. IFET-1 depletion was confirmed by loss of IFET-1::GFP expression (Fig. S1D–F). In our hands, ifet-1(RNAi)-treated animals had similar phenotypes (data not shown) to those reported for ifet-1(tm2944), including poor oocyte growth, occasional Mog and bifurcated distal gonads (Sengupta et al., 2013). Importantly, these are similar phenotypes to those of ife-3(ok191) (Fig. 2). Depletion of IFET-1 disrupted IFE-3 localization to perinuclear granules in distal germ cells (Fig. 7A vs B, insets, arrowheads) without disrupting perinuclear PGL-1 granules (Fig. S3A vs B, insets). Likewise, ifet-1(RNAi) prevented IFE-3 localization to lattice-like structures in the gonad core (Fig. 7C vs D, insets). However, the inverse was not the case. IFET-1 localization to perinuclear granules (Fig. 7E vs F) and lattice-like structures in the rachis (Fig. 7G vs H) were unaffected by ife-3(RNAi). These data provide evidence of hierarchical binding of IFE-3 to IFET-1, similar to that found for IFE-1:PGL-1 (Amiri et al., 2001). Therefore, localization of IFE-3 to perinuclear granules and rachis structures depends on its cognate 4EBP, IFET-1. IFE-3 and IFET-1 expression patterns were very similar throughout the germline (Fig. 7A,C vs E,G). In contrast to IFE-3, IFE-1 localization to perinuclear PGL-1 granules was maintained following depletion of IFET-1 (Fig. 7I,J vs 7A,B and Fig. S3E,F, insets, arrowheads). However, upon ifet-1(RNAi), aberrant aggregates formed in the rachis that contained IFE-1 and PGL-1 (Fig. 7K vs 7L and Fig. S3H, insets). These aggregates appeared quite dissimilar to IFE-3 or IFET-1 lattice-like structures and are likely unrelated. Similar PGL-1 mislocalization was observed by others in ifet-1(tm2944) germlines (Sengupta et al., 2013), which likely causes the IFE-1 mislocalization seen here. It is evident from these results that each IFE is localized by its cognate 4EBP. IFET-1 localizes IFE-3 to perinuclear granules and to structures in the gonad core, but does not contribute to normal IFE-1 localization.

Our data suggest that IFE-3 and IFET-1 work together in a functional complex to exert translational control. To address this directly, we compared the polysomal profiles of germline mRNAs following separate depletion of IFE-3 or IFET-1. Our model predicts that some GSD mRNAs will exhibit similar de-repression upon loss of either protein. Overall polysome content relative to 80S was similarly decreased by ife-3(RNAi) and by ifet-1(RNAi), indicating a modest decrease in bulk protein synthesis. Remarkably, the de-repression of fog-1 and fem-3 mRNAs was identical upon depletion of either IFE-3 or IFET-1 (Fig. 8B,C). The translation of daz-1 mRNA has been shown by others to be repressed by IFET-1 (Sengupta et al., 2013). We were also able to show that loss of either IFE-3 or IFET-1 caused identical daz-1 de-repression (Fig. 8D). Translation of IFE-1-dependent mex-1 mRNA and housekeeping gpd-3 mRNA was largely unaffected by loss of IFET-1 or IFE-3 (Fig. 8E,F). The three overlayed polysome profiles for each mRNA indicate an identical requirement for both subunits of the IFE-3:IFET-1 complex to exert translational regulation in GSD. Notably, IFE-3 is the nematode ortholog of canonical eIF4E-1, and would be expected to exert positive translational activity on some population of mRNAs, consistent with diminished polysome content following RNAi. Intriguingly, our polysome data also infer that IFET-1 (the nematode ortholog of human eIF4E-nuclear transport protein) also assists in that positive role.

In this study, we observed unique germline phenotypes, translational functions and expression patterns of eIF4E isoforms IFE-1 and IFE-3, as well as IFET-1, the cognate 4EBP that functions with IFE-3. We previously characterized the unique functions of IFE-1 to promote spermatogenesis and oocyte maturation (Friday et al., 2015; Henderson et al., 2009). Epistasis experiments placed ife-1 downstream of the terminal regulator of GSD, fog-1, indicating that ife-1 is not involved in GSD per se but rather promotes the final stages of gametogenesis (Cha et al., 2012). Spermatocytes lacking IFE-1 arrest just prior to spermatid budding; oocytes are deficient in Ephrin (VAB-1) signaling (Henderson et al., 2009; Kawasaki et al., 2011). mRNAs under the control of IFE-1 in oocytes have been identified and include pos-1, pal-1, oma-1, vab-1 and mex-1 (Friday et al., 2015; Henderson et al., 2009). For example, IFE-1 recruits vab-1 mRNAs to heavy polysomes, linking it to oocyte maturation. Differential polysome array analysis between wild-type and ife-1(bn127) worms identified all IFE-1-dependent mRNAs (Friday et al., 2015). We employed similar approaches to identify IFE-2- and IFE-4-dependent mRNAs (Dinkova et al., 2005; Song et al., 2010). Initial studies on IFE-3 showed that it was essential for both embryogenesis and the sperm-to-oocyte switch in hermaphrodites (Keiper et al., 2000; Mangio et al., 2015). Here, we extend the genetic characterization of IFE-3 and find that it has roles beyond GSD (Figs 2 and 3). IFE-3 depletion by RNAi and elevated temperature in ife-3(ok191) mutants allowed bypass of the sperm-to-oocyte switch but resulted in embryonic lethality, possibly due to defects in oocyte meiosis and growth. Thus, major roles for IFE-3 are to promote the sperm-to-oocyte switch, oocyte meiosis and embryogenesis, which is a clear distinction from roles of IFE-1 in sperm cytokinesis and oocyte maturation. For instance, IFE-3 is fully dispensable for spermatogenesis, whereas IFE-1 is required (Fig. S2). Remarkably, there appear to be non-redundant functional roles for three eIF4E isoforms in the germline, where each regulates its own unique subset of mRNAs (Fig. S5) (Friday and Keiper, 2015). The mechanism by which eIF4E isoforms discriminate among mRNAs is unlikely to be due to their protein structures, which are very homologous (Keiper et al., 2000). The structure of each mRNA cap-binding protein allows for recognition of only two to three nucleotides beyond the m7G cap (Marcotrigiano et al., 1997). Although IFE-1 and IFE-3 differ somewhat in binding to m7G and 2,2,7-trimethylguanosine (m2,2,7G) caps (Miyoshi et al., 2002), the modest affinity differences are unlikely to account for the selective mRNA translational control we observe (Dinkova et al., 2005; Friday et al., 2015; Song et al., 2010). In addition, genes subject to trans-splicing give rise to heterogeneity of m7G and m2,2,7G-capped mRNAs, with the exception of downstream operon mRNAs (Blumenthal, 2005). We suggest that mRNA selectivity is based on association with unique cognate 4EBPs; namely, PGL-1 for IFE-1 and IFET-1 for IFE-3. The fact that identical translational regulation of IFE-3-responsive mRNAs was observed upon loss of IFET-1 lends strong support to this mechanistic model (Fig. 8; Fig. S5).

Many of the GSD-disrupting mutations have been mapped to regulatory elements within the 3′-UTR of several genes (e.g. fem-3gf and tra-2gf alleles) (Barton et al., 1987; Goodwin et al., 1993), or were null mutations in RBP genes (fbf genes, fog-1, gld-1) (Barton and Kimble, 1990; Jan et al., 1999; Zhang et al., 1997). Each case pointed to mRNA repression events mediated by sequence-specific RBPs. Genetic characterization indicated a strong requirement for ife-3 during hermaphrodite GSD. Our epistatic analysis placed the role of ife-3 downstream of fog-2, and upstream or at the level of fem-3, and implicated synergistic involvement of fbf genes to promote oocyte fate (Fig. 3). Using polysome analysis, we identify mRNAs that IFE-3 and IFET-1 regulate in GSD and show that their role appears to include co-repression and/or activation of translation. Loss of IFE-3 led to enhanced translation initiation of repressed fem-3, fog-1, fbf-1/2 and gld-1 mRNAs. The fem-3 and fog-1 mRNAs show the most dramatic de-repression (Fig. 4), which may contribute to the involvement of IFE-3 in the sperm-to-oocyte switch. We found no evidence that IFE-3 depletion altered translational efficiency of other GSD mRNAs involved in oocyte fate (tra-1, nos-3, mog-1; data not shown). Importantly, translation of IFE-3-responsive GSD mRNAs was not substantially altered by loss of IFE-1 (Friday et al., 2015). However, they were de-repressed identically by depletion of IFE-3 or IFET-1 (Fig. 8). The similarity of translational outcomes and developmental defects strongly suggests a unified function for the IFE-3:IFET-1 mRNP in repression to activation switches (see Fig. 1A). Loss of IFE-3 also induced substantial new expression of fog-1 mRNA in larval germlines even before the sperm-to-oocyte switch occurred (Fig. 5). Additionally, tra-2 mRNA translational efficiency was partly dependent upon IFE-3 (Fig. 4F). However, it is also possible that enhanced gld-1 mRNA translation may contribute (Fig. 4E), as GLD-1 acts to repress translation of tra-2 (Jan et al., 1999). The translational efficiency of a known IFE-1-dependent mRNA, mex-1, was unperturbed by loss of IFE-3 or IFET-1 (Figs 4H and 8F), providing further evidence that each eIF4E works selectively with its own mRNA targets. Housekeeping mRNAs gpd-3 and tbb-2 appear to be impartial to any single isoform (Figs 4G and 8E) (Dinkova et al., 2005; Friday et al., 2015; Song et al., 2010). Interestingly, the mRNAs de-repressed upon IFE-3 or IFET-1 loss are also direct FBF targets (Kershner and Kimble, 2010). Combined with our genetic data, this suggests that IFE-3:IFET-1 may take part in FBF mRNP-mediated repression of some GSD mRNAs. Alternatively, IFE-3 may positively regulate an unknown factor that inhibits FBF expression or activity. Our polysome analysis showed that translation of both fbf-1 (Fig. 4) and fbf-2 mRNAs (data not shown) were de-repressed upon IFE-3 loss, supporting the latter interpretation. However, any interpretation is complicated by the finding that FBFs autoregulate their own translation (Lamont et al., 2004). If an IFE-3–FBF cooperation exists, it must exclude germline stem cell maintenance mRNAs, because we found no role for IFE-3 in regulating mitotic activity (Fig. S4) as FBFs are known to do (Crittenden et al., 2006). Patterns of IFE-3 expression place it in the right place and time to regulate GSD mRNAs.

How each IFE regulates a subset of mRNAs for unique germline roles may be best explained by subcellular localization patterns. Mammalian eIF4E isoforms were shown to differentially localize to P bodies and stress granules, suggesting that they differentially regulate mRNA trafficking, translation and/or turnover (Andrei et al., 2005; Frydryskova et al., 2016). Our studies begin to put C. elegans eIF4E isoform-specific regulation into the context of temporal and spatial mRNP complexes (Fig. S5). We characterized two eIF4E isoforms using CRISPR/Cas9 to fluorescently tag endogenous ife genes. Our early transgenic studies showed that IFE-1 localized to P granules in embryos via a direct protein–protein interaction with its cognate 4EBP, PGL-1 (Amiri et al., 2001). CRISPR-tagged IFE-1 likewise localized tightly to PGL-1 granules in early germ cells, oocytes and embryos (Fig. 6). We never observed any IFE-1 foci that were not coincident with PGL-1 throughout development. Previously we described the role of IFE-1 in regulating maternal mRNAs that are also enriched in P granules (Henderson et al., 2009). Clearly, both the translational function and subcellular localization of IFE-1 are intricately tied to PGL-1. Surprisingly, CRISPR-tagged IFE-3 also resided in germ granules, which were distinct from IFE-1:PGL-1 granules (Figs 1, 6 and 7). The cognate 4EBP of IFE-3, IFET-1, is suggested to be a broad-scale translational repressor and has important functions in embryogenesis, gonad morphology and GSD (Sengupta et al., 2013). IFET-1 depletion caused dissociation of IFE-3 from both perinuclear granules and rachis structures, but IFE-3 depletion did not disrupt association of IFET-1 with those same structures (Fig. 7A–D vs E–H). This suggests a binding hierarchy in which IFET-1 directs IFE-3 localization to both structures, where they prevent translation of their bound mRNAs in localized mRNPs, and that their functions are tightly linked, like IFE-1:PGL-1. Yet their localization and translational roles appear to be distinct from the IFE-1:PGL-1 mRNPs (Figs 7 and 8). A Drosophila 4EBP called Cup was similarly shown to regulate the localization and inhibit function of an oocyte eIF4E isoform (Nakamura et al., 2004; Wilhelm et al., 2003), while another called Mextli enhances eIF4E-1 activity in ovarian germ cells of the fly (Hernández et al., 2013). In worms, the biochemical isolation of in vivo OMA-1 mRNPs containing IFE-3 and IFET-1 also co-purify with polyadenylase (GLD-2) and CCR4–NOT deadenylase machinery (Li et al., 2009; Spike et al., 2014). OMA-1 mRNPs have roles in oocyte maturation (Spike et al., 2014). Such complexes are likely important for translational regulation by IFE-3. Curiously, the oma-1 mRNA relies heavily on IFE-1 for efficient translation (Henderson et al., 2009), suggesting separable roles for the two eIF4Es that ensure normal translational regulation of oocyte development.

Where IFE-1 and IFE-3 are localized in separate granules, they likely sequester their preferred mRNAs from active translation in repressed mRNPs. Where both are soluble in the germ cell cytoplasm, they likely associate with eIF4G (IFG-1) to participate in mRNA translation initiation (Friday and Keiper, 2015; Keiper et al., 2000). Each IFE moves to the soluble fraction in a different cell type – IFE-1 in late spermatocytes (Fig. 6E) and IFE-3 in large oocytes (Fig. 1H,L). In each case, the relevant IFE becomes soluble at a time when its positive translation activity is expected, based on phenotype. It is well known that 4EBPs inhibit translationally productive eIF4E:eIF4G interactions (Grüner et al., 2018; Haghighat et al., 1995). We propose a model in which IFE-1 and IFE-3 have active and dormant periods that are controlled by their regional expression and cognate 4EBPs, potentially explaining how each eIF4E isoform regulates its own subset of mRNAs (Fig. S5). Dormant periods are those when each eIF4E:4EBP is associated with a unique germ granule and acts as a translational repressor. In active periods, each eIF4E leaves the germ granule and the 4EBP to exert positive initiation activity on its cargo mRNAs, recruiting them to ribosomes for translation. We do not yet have direct demonstration of each regulated mRNA that resides in an IFE-containing mRNP. The mechanisms by which 4EBPs control eIF4E repression and activation in germ cells may also involve eIF4G competition or cooperation, as well as mRNP localization, shuttling or remodeling. Our current findings are novel in characterizing two distinct eIF4E:4EBP mRNP complexes in C. elegans that have independent dynamics and lead to differential translational control of germline mRNAs.

C. elegans Bristol var. N2 was used as the wild-type strain in all experiments unless otherwise noted. DUP71 pgl-1(sam33[pgl-1::gfp::3xflag])IV and DUP121 glh-1(sam24[glh-1::gfp::3xflag])I; pgl-1(sam52[pgl-1::mTagRFP::3xflag])IV were obtained from Dr Dustin Updike (MDI Biological Laboratory, Bar Harbor, ME). Strains JK2958 nT1[qIs51] (IV;V)/dpy-11(e224) unc-42(e270)V, CGC20 dpy-18(e364)/eT1 III; unc-46(e177)/eT1[umnIs9]V, JK574 fog-2(q71)V, CB3844 fem-3(e2006)IV, JK3022 fbf-1(ok91)II, JK3101 fbf-2(q738)II, GR1373 eri-1(mg366)IV, and DG3913 lin-41(tn1541[gfp::tev::s::lin-41])I were obtained from the Caenorhabditis Genetics Center (University of Minnesota, Minneapolis, MN) which is funded by the NIH Office of Research Infrastructure Programs (grant number P40 OD010440). The ife-3(ok191) allele was obtained from the C. elegans Knockout Consortium (OMRF), outcrossed ten times to N2 worms, and balanced with unc-34(e566) to make KX10 ife-3(ok191)/unc-34(e566)V. The ife-3 allele was then crossed into strain JK2958 to more stably balance it with nT1[qls51] to make KX98 ife-3(ok191)/nT1[qls51]V, and with eT1[umnIs9] to make KX140 ife-3(ok191)/eT1[umnIs9]V. Breakpoint of ife-3(ok191) deletion was determined by subcloning and sequencing of the genomic deletion PCR product, and found to be taattttcatattttccgct/tatcta/ttatcgattttttccagatg. KX143 ife-1(eu20[mkate2::tev::3xmyc::ife-1])III, and KX152 ife-3(eu21[gfp::tev::3xflag::ife-3])V, were made using CRISPR/Cas9-driven homologous recombination to make in-frame insertions at the respective endogenous gene loci in N2 worms. KX149 ife-1(eu20[mkate2::tev::3xmyc::ife-1])III, pgl-1(sam33[pgl-1::gfp::3xflag])IV was made by crossing KX143 and DUP75. KX155 ife-1(eu20[mkate2::tev::3xmyc::ife-1])III, ife-3(eu21[gfp::tev::3xflag::ife-3])V was made by crossing KX143 and KX152. KX157 ife-3(eu21[gfp::tev::3xflag::ife-3])V, pgl-1(sam52[pgl-1::mTagRFPT::3xflag])IV was made by crossing KX152 and DUP121 and outcrossing glh-1(sam24). KX159 ife-3(ok191)/nT1[qls51]V, lin-41(tn1541[gfp::tev::s::lin-41])I was made by crossing KX98 and DG3913. Strain PRB171 qfIs012(FL-IFET-1::ifet-1(tm2944): GFP::3XFLAG+unc-119(+)) was created by biolistic transformation of unc-119(ed3) worms (Praitis et al., 2001) with recombineered fosmid DNA generated by the C. elegans TransgeneOmics Project, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany (Sarov et al., 2006). All strains were maintained on nematode growth medium (NGM) plates at 20°C unless otherwise indicated, with Escherichia coli strain OP50 as described (Brenner, 1974).

Fertility of ife-1 and ife-3 mutant worms was determined at two different temperatures: permissive temperature (20°C) or restrictive temperature (25°C). To this end, ife-1(bn127), ife-3(ok191) and fog-2(q71) (control) mutant strains were grown from embryo stage at 20°C or 25°C. L4 males were transferred to new NGM plates and incubated for 24 h at 20°C or 25°C. Adult males were soaked in M9 solution with 50 µM SYTO17 red fluorescent dye (Molecular Probes) for 3 h, and each male was transferred to an OP50-seeded NGM with a single fog-2 female mutant worm. Twenty-four hours later, mating efficiency was determined by counting the number of fog-2 female worms with SYTO17(+) sperm in their spermatheca among fog-2 worms tested. Fertility was determined by counting the number of fog-2 females with SYTO17(+) sperm and embryos in the uterus.

To unambiguously determine the frequency of the temperature-sensitive rescue of ife-3(ok191) fertility and separate poorly balanced ife-3 genetic recombinants from true ife-3(−/−) animals at 25°C, we generated synchronized populations of KX98 ife-3(ok191)/nT1[qls51]V worms by using a solution of NaClO (1.3%) and NaOH (0.5 M) on gravid hermaphrodites, hatching F₁ embryos overnight in M9 at 20°C and seeding L1 larva onto NGM plates with OP50 to be grown at 20°C or 25°C. Twenty-four hours after L4 molting, animals were dissected with a 30-gauge needle under a stereomicroscope to determine if they were fertile (embryos) or sterile (no embryos), then individual worms were collected for genomic PCR using the following triple primer set: forward 1, 5′-TATGTGCATGTTGTGGAGGCTG-3′; reverse, 5′-TGTAGTCTCCGTACTCGTCG-3′; forward 2, 5′-CAGAATCGCTGGGCTCTC-3′, yielding a single 417 bp product in ife-3(+/+) worms, a single 756 bp product in ife-3(−/−) worms and both products in ife-3(+/−) worms. Results were pooled from three individual experiments.

Gene-specific RNAi was performed by feeding C. elegans with double-stranded RNA (dsRNA) expressing E. coli (HT115) as described, with minor modifications (Timmons et al., 2001; Timmons and Fire, 1998). Briefly, HT115 was transformed with either L4440 vector or gene-specific constructs and grown overnight at 37°C in 2xYT medium containing 50 μg/μl carbenicillin. Cultures were concentrated 5-fold by centrifugation, seeded onto nematode growth medium (NGM) plates containing 1 mM IPTG and 50 μg/μl carbenicillin, then induced overnight at room temperature before plating worms. dsRNA was extracted from bacterial plates after induction and run on denaturing 2% agarose gels with ethidium bromide to determine dsRNA expression when appropriate. For feeding, L4 hermaphrodites were seeded onto RNAi plates and F₁ offspring were evaluated for phenotype (Fig. 3) or processed for biochemical analysis (polysomes; Fig. 4). RNAi feeding experiments were conducted at 25°C, with the exception that CB3844 fem-3(e2006) and GR1373 eri-1(mg366) L4 mothers were grown at 20°C for the first 24 h of feeding, then shifted to 25°C, due to decreased fertility at higher temperatures. We generated the ife-3(RNAi) feeding construct (pT72-ife-3) by inserting a 475 bp restriction-digested ife-3 complementary DNA (cDNA) fragment containing only the coding region (Jankowska-Anyszka et al., 1998) into the L4440 vector backbone using standard cloning procedures. We also generated ifet-1(RNAi) feeding construct (pT72-ifet-1) by inserting a 600 bp PCR-amplified ifet-1 fragment into the L4440 vector backbone, again using standard cloning procedures (Maniatis et al., 1989). The fog-1(RNAi) construct has been previously described (Datla et al., 2014). The identity of RNAi clones was verified via Sanger sequencing (Iowa State Sequencing Facility).

CRISPR/Cas9 mediated genome editing was accomplished by using pDD162 (plasmid #47549, Addgene) to express single-guide RNA (sgRNA) driven by C. elegans U6 promoter and Cas9 DNA endonuclease protein under control of the eef-1A.1/eft-3 promoter, in conjunction with plasmid-based homology directed repair (HDR) template containing a self-excising cassette (SEC), as described with minor modifications (Dickinson et al., 2015, 2013). We generated multiple pDD162 sgRNA/Cas9-expressing plasmids for ife-1 and ife-3, using a Q5^® site-directed mutagenesis kit from New England Biolabs (NEB) as per the manufacturer's instructions. We also generated HDR templates for ife-1 and ife-3 by inserting PCR-amplified homology arms (∼550–950 bp) corresponding to upstream and downstream regions of each gene relative to the start codon (ATG) into pDD287 and pDD282 backbones (plasmid #70685 and #66823, Addgene), respectively, using a Gibson Assembly^® kit (NEB) as per the manufacturer's instructions. When possible, conserved mutations were made in protospacer adjacent motif (PAM) sites within the homology arms to prevent HDR template cutting by Cas9; when mutation of the PAM site was not possible, conserved mutations were incorporated throughout the sgRNA target site. sgRNA/Cas9-expressing plasmids and HDR template plasmids were isolated using a PureLink^® Miniprep Kit (Invitrogen) as per the manufacturer's instructions and confirmed via Sanger sequencing (Iowa State Sequencing Facility). We injected mated N2 hermaphrodites as young adults in order to increase brood sizes. Injection cocktails were as follows: ife-1 injection mix: pDD162-ife-1-256 (sgRNA/Cas9 50 ng/μl), pDD287-ife-1 (HDR SEC 10 ng/μl), pPD118.2 (GFP co-injection marker 5 ng/μl); ife-3 injection mix: pDD162-ife-3-181, pDD162-ife-3-7741, pDD162-ife-3-8300 (each sgRNA/Cas9 25 ng/μl), pDD287-ife-3 (HDR SEC 10 ng/μl), pCJ104 (RFP co-injection marker 5 ng/μl). Progeny of injected worms were selected with Hygromycin (final concentration 250 μg/ml) at 20°C and screened for survival, Rol phenotype and lack of co-injection marker, then heat shocked at 34°C for 3–4 h to remove SEC. Non-rollers from heat shock plates were screened by genomic PCR to confirm insertion at the desired locus. Animals containing the desired insertion were backcrossed at least three times to N2 males. N-terminal protein fusions did not appear to grossly affect IFE-1 and IFE-3 function as worms homozygous for both insertions were superficially wild type and had wild-type fertility.

Dissection of hermaphrodite or male germlines and immunostaining were performed as described with minor modifications (Cha et al., 2012). For MSP immunostaining, worms of the indicated genotype and treatment were dissected under a stereomicroscope using a 30-gauge needle, germlines were fixed in 3% paraformaldehyde in 1× phosphate buffered saline (PBS; pH 7.2) for 20 min at room temperature, washed in PBS+0.01% Tween 20 (PTW), then post-fixed in ice cold 100% MeOH and stored at −20°C. Following fixation, dissected germlines were rehydrated in 50% cold MeOH, washed in PTW, blocked for 1 h in PTW+0.5% bovine serum albumin (PTWB) at room temperature, incubated with mouse anti-MSP (1:100) primary antibody (Yoon et al., 2016) in PTWB overnight at 4°C, washed in PTWB extensively, then incubated in Alexa Fluor 488- or Alexa Fluor 568-conjugated goat anti-mouse-IgG (1:400) secondary antibody (Invitrogen) in PTWB for 1 h at room temperature. They then underwent a final incubation in DAPI for 15 min at room temperature before being washed extensively in PTWB and mounted on 2% agarose imaging pads in 90% glycerol/DABCO (ACROS Organics) anti-fade medium. For imaging CRISPR/Cas9 endogenously tagged fluorescent protein fusions, we replaced the 100% MeOH post-fix and rehydration step above with a 70% EtOH post-fix and rehydration step, and did not block in PTWB or use immunostaining; other steps for fixing and imaging these germlines was identical to those described above. We also imaged CRISPR/Cas9 fusions in live worms treated with NaN₃ (2 mM) and levamisole (5 mM) for immobilization. Germline imaging was performed on a Zeiss Axiovert 200M Differential Interference Contrast (DIC) microscope equipped with an Axiocam MRm CCD camera and corresponding DAPI/Hoechst, and FITC/GFP, and TexasRed/RFP optics with 40× air or 100× oil objectives. Image processing was performed using Axiovision 4.3 software (Carl Zeiss). Any image adjustments to contrast or brightness were conducted linearly.

Lysis and gradient buffers, gradient setup, centrifugation and fractionation were conducted as described with the following modifications (Dinkova et al., 2005; Henderson et al., 2009). L4-stage N2 hermaphrodites were seeded onto NGM plates with HT115 expressing either L4440(RNAi), ife-3(RNAi) or ifet-1(RNAi) and allowed to grow at 25°C for ∼96 h until plates were saturated with mix-staged F₁ and F₂ progeny. We estimate that young adult worms composed >75% of the biomass on each plate. Worms were then washed off plates using M9, allowed to purge gut contents for 30 min in M9, and floated in 35% sucrose at 4°C to remove debris and bacteria, then flash frozen in liquid nitrogen in the presence of 14 mM E64 protease inhibitor (Sigma-Aldrich) and 4 mM Vanadyl-RNC RNase inhibitor (Sigma-Aldrich). Worms were crushed in a mortar and pestle under liquid N₂ with an equal volume of buffer A [50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10 mM MgCl₂, 1 mM EGTA, 2 mM dithiothreitol (DTT), 800 U of RNAsin/ml (Promega), 4 mM Vanadyl-RNC, 5 mM PMSF, 0.4 mg/mL Cycloheximide (Calbiochem)], and centrifuged at 14,000 g at 4°C for 15 min. A 0.5-ml aliquot of the supernatant was applied to an 11-ml 10–45% linear sucrose density gradient made in buffer B (25 mM Tris-HCl pH 8.0, 140 mM NaCl, 10 mM MgCl₂, 1 mM DTT) and centrifuged in a Beckman SW41Ti rotor at 38,000 rpm at 4°C for 2 h. Gradients were fractionated into 1 ml fractions with continuous monitoring of absorbance at 254 nm with an ISCO UA-6 UV VIS detector. Sucrose fractions were stored at −80°C prior to use.

RNA from sucrose gradient fractions was isolated in four volumes of Trizol^® Reagent (Thermo Fisher Scientific) as per the manufacturer's instructions with the following additions: after isopropanol precipitation, fractions were extracted once with phenol-chloroform-iso-amyl alcohol (25:24:1), then extracted once with chloroform-isoamyl alcohol (24:1), followed by EtOH precipitation at 4°C overnight with 20 μg glycogen (Thermo Fisher Scientific). RNA was isolated from 500 μl of each 1-ml gradient fraction. RNA concentrations were measured on a NanoDrop^® ND-1000 spectrophotometer and assessed for RNA purity based on A260/280 ratios. cDNA was synthesized from each fraction using an iScript Advanced cDNA Synthesis Kit (Bio-Rad) using 0.4 μg of input RNA as per the manufacturer's instructions. qRT-PCR was performed in triplicate on a CFX Real-Time System (Bio-Rad) using Sso Fast Evagreen Super mix (Bio-Rad), according to the manufacturer's instructions. qRT-PCR quantification of each mRNA was normalized to total RNA content. Standard curves were generated for each primer set and used for absolute quantification. Similar traces were obtained for fem-3, fog-1 and gpd-3 mRNAs (Fig. 3) on two to three independent gradients. Primer sequences used were as follows: tbb-2 (forward, 5′-TCATCTCCAAAATCCGCGAA-3′; reverse, 5′-GAGGGATACAAGATGGTTC-3′); mex-1 (forward, 5′-AATGGATAAGCTAATGTTGGA-3′; reverse, 5′-GATATTGCGATGAGGAAGAG-3′); fem-3 (forward, 5′-AAGCTGACAGAGAAACGAGA-3′; reverse, 5′-AAAGGAATTCCAGATATTAAAGG-3′); fbf-1 (forward, 5′-GACCAATCAAAAATGCGCTAT-3′; reverse, 5′-GATTTCCAACCTCTGTAGATGTG-3′); fog-1 (forward, 5′-TCAGTGCCAGGATTTAGAGA-3′; reverse, 5′-AATATCGATGATACGGTTGTG-3′); gld-1 (forward, 5′-TTCAGGTCCAGTTTTGATGT-3′; reverse, 5′-GACGTTAGATCCGAGAAGGT-3′); daz-1 (forward, 5′-ACAAAAAGCCCAATCAAAG-3′; reverse, 5′-GCTGAAAGTGGAGAAAGATG-3′); gpd-3 (forward, 5′-GATCTCAGCTGGGTCTCTT-3′; reverse, 5′-TCCAGTACGATTCCACTCAC-3′); and tra-2 (forward, 5′-TTCAATTGCAACAAAACAAA-3′; reverse, 5′-TCTGCACCAAATTGTAGACC-3′).

We conducted micro-scale qRT-PCR on whole-worm lysates as described with minor modifications (Contreras et al., 2008). We generated synchronized populations of N2 and KX98 ife-3(ok191)/nT1[qls51]V worms using a solution of NaClO (1.3%) and NaOH (0.5 M) on gravid hermaphrodites, hatching F₁ eggs overnight in M9 at 20°C, and seeding L1 larvae onto NGM plates with OP50 to be grown at 20°C. Three independent biological replicates each containing ∼25 early or late L4 hermaphrodites (N2 or ife-3−/−) were collected into 25 μl of single-worm lysis buffer (15 mM Tris-HCl pH 8.0, 50 mM KCl, 1 mM MgCl₂, 0.45% NP-40, 0.45% Tween 20) containing 0.2 mg/ml Proteinase K (60 min at 60°C; 10 min at 95°C). Upon completion of lysis, RNA was isolated using the PureLink^® RNA Mini Kit (Invitrogen) with on-column DNAse I digestion, followed by EtOH precipitation at −20°C overnight with 20 μg glycogen (Thermo Fisher Scientific). RNA was resuspended in 10 μl of nuclease free H₂O for cDNA synthesis and qRT-PCR as described above. Reactions were conducted in triplicate from three independent biological replicates, quantified and normalized to tbb-2 (endogenous control) using the 2^−ΔΔCt method. Student's t-test was used to determine statistical significance between ife-3(−/−) and wild-type animals, with a P-value cutoff of 0.05.

Metabolic labeling of DNA synthesis was performed as described with minor modifications (Yoon et al., 2018). Animals were incubated in M9+0.1% Tween 20 with 1 mM EdU for 30 min at 20°C. Gonads were dissected and fixed in 3% paraformaldehyde solution in 0.1 m K₂HPO₄ (pH 7.2) for 10–20 min, followed by −20°C methanol fixation for 10 min. Dissected gonads were blocked in PTWB solution for 30 min at 20°C. EdU labeling was detected using a Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen) according to the manufacturer's instructions. After washing 3× with PTWB solution for at least 30 min at 10 min intervals, the dissected gonads were stained with a final incubation in DAPI for 15 min at room temperature before being washed extensively in PTWB and mounted on 2% agarose imaging pads in 90% glycerol/DABCO (ACROS Organics) anti-fade medium.

Frozen worm pellets used for polysome analysis (Fig. 3) were ground in liquid nitrogen in CL+ buffer [40 mM MOPS pH 7.5, 200 mM KCl, 2 nM EDTA, 4 mM EGTA, 2 mM DTT, 14 mM E64 and HALT protease cocktail (Thermo Fisher Scientific)], and centrifuged at 14,000 g for 10 min. The supernatant was batch bound to m7G sepharose (Sigma-Aldrich) beads for 1 h at 4°C, washed 3× with CL+, boiled in 2× SDS load buffer for 3 min, and 30 μg protein for each sample was resolved by 12% SDS-PAGE. Proteins were transferred to polyvinylidene fluoride (PVDF) membranes overnight and blocked with TST (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.05% Tween 20) containing 5% dry non-fat milk for 1 h at room temperature, then incubated with a 1:1000 dilution of an IFE-3-specific primary antibody CA82 (Jankowska-Anyszka et al., 1998) in TST containing 5% dry non-fat milk for 1 h at room temperature. Blots were washed 5× in TST containing 5% dry non-fat milk and incubated with a 1:10,000 dilution of goat anti-rabbit IgG secondary antibody conjugated to horseradish peroxidase (HRP) in TST containing 5% dry non-fat milk for 1 h at room temperature, then washed extensively. Detection was performed with an ECL+ kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Image acquisition was performed using an Amersham Typhoon Fluorescence/Phosphorimaging scanner (GE Healthcare) at the ECU PhIFI Core Facility.

We thank Dan Dickinson (University of Texas at Austin) for advice on CRISPR/Cas9 gene editing, Dustin Updike (MDI Biological Laboratory, Bar Harbor, ME) for fluorescent PGL-1 CRISPR strains, Sammy Chouffani for subcloning the ifet-1(RNAi) plasmid and Caitlin Palmer for preliminary experiments. We thank the C. elegans Gene Knockout Consortium at OMRF for the original mutant isolate of ife-3(ok191) and Dr Mihail Sarov of the C. elegans TransgeneOme Project for the IFET-1::GFP::3xFLAG fosmid. The fbf-1 and fbf-2 strains were provided by Judith Kimble (University of Wisconsin). The fem-3, fog-2, eri-1 and lin-41(tn1541[gfp::tev::s::lin-41]) strains were from the Caenorhabditis Genetics Center, funded by the NIH Office of Research Infrastructure Programs [P40 OD010440].

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.237990.reviewer-comments.pdf