Spatial and temporal translational control of germ cell mRNAs mediated by the eIF4E isoform IFE-1

Cell fate, proliferation and differentiation are dependent on gene expression, often at the level of protein synthesis (Kimble and Crittenden, 2007). Dysregulation of these processes is associated with human pathologies including infertility, birth defects and cancers (Graff et al., 2008; Song and Lu, 2011). Clinical studies have shown that the eukaryotic translation initiation factor eIF4E is overexpressed in multiple cancers including ovarian, esophageal, breast, thyroid and prostate (Furic et al., 2010; Ko et al., 2009; Kouvaraki et al., 2011; Salehi and Mashayekhi, 2006). Its overexpression upregulates the translation of mRNAs for cell cycle, growth and reactive oxygen species (ROS) regulatory proteins (De Benedetti and Graff, 2004; De Benedetti and Rhoads, 1990; Rosenwald, 2004; Rosenwald et al., 1993), leading to the cancer phenotype. However, eIF4E-1 depletion by just 50% prevents lung cancer growth in mice (Truitt et al., 2015). eIF4E is essential for cap-dependent protein synthesis in order to recruit mRNAs to the ribosome for translation. Specifically, eIF4E binds both the 7-methylguanosine 5′-cap of mRNAs and the scaffolding translation initiation factor eIF4G (Marcotrigiano et al., 1997). eIF4G assembles initiation factors, the 40S ribosomal subunit and mRNA to form the 48S complex, the rate-limiting step for protein synthesis (Gingras et al., 1999; Lamphear et al., 1995; Liu et al., 2003). The selection of cap-dependent mRNAs by eIF4E can alter cell fate by promoting growth and proliferation (De Benedetti and Graff, 2004; De Benedetti and Rhoads, 1990). eIF4E-mediated recruitment of mRNAs can be inhibited by eIF4E-binding proteins (4EBPs), which bind to the dorsal or lateral side of eIF4E to prevent association with eIF4G (Igreja et al., 2014). Cap-dependent recruitment is thereby inhibited, allowing cap-independent initiation to prevail (Contreras et al., 2008; Fukuyo et al., 2011). Growth factor signaling activates mTOR kinase to phosphorylate 4EBP, causing a dissociation from eIF4E and promoting cap-dependent protein synthesis.

Evidence suggests that eIF4E also regulates mRNAs in germ cells. Five isoforms of eIF4E (IFE-1–IFE-5) are expressed in the nematode C. elegans (Jankowska-Anyszka et al., 1998; Keiper et al., 2000). We have previously shown that individual isoforms regulate a unique subset of mRNAs in a tissue-specific manner (Amiri et al., 2001; Dinkova et al., 2005; Henderson et al., 2009; Kawasaki et al., 2011; Song et al., 2010). Given that all IFEs bind to mRNA caps, these findings suggest a selective cap-dependent regulation that goes beyond mTOR regulation. For example, although IFE-2 and IFE-4 are expressed in somatic tissue, they regulate different mRNAs whose products drive different developmental events. IFE-2 regulates aging, whereas IFE-4 regulates nerve and muscle tissue function (Dinkova et al., 2005; Syntichaki et al., 2007). IFEs 1, 3 and 5 (and to a lesser extent IFE-2) are expressed in the germ line and are involved in gamete development (Henderson et al., 2009; Keiper et al., 2000; Song et al., 2010). Germ line and embryonic development require essential proteins for cellular differentiation, maturation and viability. During gametogenesis stored mRNAs are utilized when transcription is largely silenced due to chromosomal condensation (Kelly and Fire, 1998; Seydoux et al., 1996). Studies in germ cells from multiple species have shown that available ribosomes translate these mRNAs as they become de-repressed, but recent evidence suggests that translation initiation factors take part in both the repression and subsequent recruitment (Dworkin and Dworkin-Rastl, 1990; Ghosh and Lasko, 2015; Goodwin and Evans, 1997; Macdonald and Smibert, 1996; Mendez and Richter, 2001; Wormington, 1993). In Xenopus laevis oocytes, for example, meiotic maturation is arrested by translational suppression of cyclin-B mRNA, which contains a cytoplasmic polyadenylation element (CPE) in the 3′-untranslated region (3′UTR). Repression of cyclin-B mRNA occurs when the eIF4E–maskin–CPEB (CPE-binding factor) complex forms. Maskin acts as a specialized 4EBP bound to mRNA that inhibits recruitment of an initiation complex through eIF4G (Barnard et al., 2005). To induce oocyte maturation, progesterone activates a signaling cascade that results in CPEB phosphorylation, Maskin dissociation from eIF4E and cap-mediated recruitment of the de-repressed mRNA. Translation of these products reinitiates the cell cycle and oocyte maturation (Barnard et al., 2005; Cao and Richter, 2002). Dysregulation of the first meiotic division in human oocytes is a large contributor to infertility and miscarriage (Hunt, 1998). mRNA translational control thus provides a substantive link to oncogenesis and reproductive health.

IFE-1 is a nematode eIF4E isoform bound to the 4EBP-like protein PGL-1 (Amiri et al., 2001). Together these proteins reside in P granules, which are germ-line-specific ribonucleoprotein (mRNP) complexes containing stored mRNAs (Kawasaki et al., 1998). Based on localization in these mRNPs, we hypothesize that IFE-1 uniquely catalyzes the subsequent translation of stored mRNAs during germ cell development. Supporting this hypothesis, ife-1 mutant worms have substantial developmental defects during gametogenesis and embryogenesis. These include a temperature-sensitive defect in cytokinesis in secondary spermatocytes, which results in a complete lack of mature sperm. Even under sperm-permissive conditions, ife-1 worms still show a reduced rate of oocyte production, viability and early embryonic lethality (Henderson et al., 2009). We have used a genomic approach called translation state array analysis (TSAA) to identify 77 mRNAs that rely on IFE-1 for positive translational control. Furthermore, IFE-1 carries out spatiotemporal recruitment of mRNAs involved in meiosis, mitogen-activated protein kinase (MAPK) activation, oocyte maturation and embryogenesis.

Our previous studies have shown that IFE-1 is required for the efficient translation of several developmentally controlled mRNAs in the C. elegans germ line (pos-1, oma-1, mex-1, pal-1 and glp-1) (Henderson et al., 2009). Here, we determined the entire population of messages that are IFE-1-dependent using TSAA (Chappell et al., 2013; Dinkova et al., 2005; Song et al., 2010). Using Affymetrix microarrays, we quantified the partitioning of every mRNA into ribosome-bound and unbound pools (Fig. 1A). We derived a measure of the relative change (R value) in polysome loading between wild-type and ife-1 strains (Fig. 1B). This method directly assessed the translational efficiency of each mRNA, independent of changes in the abundance, by quantifying both the polysomal (P) and non-polysomal (NP) content of mRNAs, rather than just the amount in polysomes. Calculating a mean fold change in R-value (MFR) allowed us to prioritize mRNAs with markedly altered translational efficiency. From the arrays, we also derived the mean fold polysomal loading change (MFP), as well as the change in total mRNA (MFT) from sums of the NP and P signals, for wild-type and ife-1 worms. The mRNA subset with the largest change in total signal (MFT) substantially overlapped with the subset of mRNAs with the largest change in polysome signal (MFP; Fig. S1B). By contrast, the relative translational efficiency group (MFR) shares just eight mRNAs with the polysome MFP group. Therefore, measuring only polysomal mRNA content would be a misleading assessment of translational control.

A prioritized list of 77 IFE-1-dependent mRNAs was derived from the 13,372 mRNAs reliably detected and statistically relevant in all three biological replicates. Each mRNA showed greater than 1.5-fold increase in relative polysome loading (MFR) when IFE-1 was present (Fig. 1B,C), which was largely maintained in the individual replicates (Fig. S2). Of the 77 candidates, 25 had known phenotypes or functions. Among these, only nine were relevant to the germ line or early embryogenesis (Fig. 1D), and these were pursued. Several (vab-1, vpr-1 and rab-7) encode genes that regulate MAPKs (likely MPK-1) in late stage oocyte maturation (Fig. 1D; Fig. S3A). Others (lin-26, rab-5, rnp-3, let-49, hmg-5 and grl-20) are essential for meiotic progression or embryonic development. The identification of these mRNAs is consistent with previous biochemical characterizations of the role of IFE-1 in protein synthesis, specifically in oocytes and early embryos, as well as the ife-1(bn127) phenotype (Amiri et al., 2001; Henderson et al., 2009). In contrast to the IFE-1-dependent mRNAs, most other mRNAs showed no change, as evidenced by their position along the R_wt=R_ife-1 diagonal. Their translational efficiency was unchanged in the ife-1(bn127) mutant worms (Fig. 1C; Fig. S3A). Many IFE-1-independent messages were housekeeping mRNAs (e.g. actin, tubulin and GAPDH), although some were cell cycle and kinase mRNAs (mpk-1 and cdk-4), and cell fate mRNAs (e.g. pgl-1; Fig. S3A). Remarkably, sperm-specific mRNAs, such as those encoding the 28 known major sperm proteins (MSPs) were unchanged in relative R value in the TSAA. Spermatogenesis is compromised in the ife-1 mutant even at the permissive temperature used in the TSAA (Henderson et al., 2009). MSPs have been shown by proteomic analysis to be deficient in ife-1 worms (Kawasaki et al., 2011). By segregating the MFR and MFT for all identified msp mRNAs, our data showed them to be underrepresented in the total population but unchanged in their translational efficiency (Fig. S3C,D). Decreased MSP protein in worms lacking IFE-1 is therefore due to lack of mRNA rather than translational control.

In order to determine the detailed translational efficiency of mRNAs, we analyzed the polyribosomal loading of mRNAs by sucrose gradient centrifugation. Based on known ife-1 mutant phenotypes, we focused on known mRNAs that regulate germ cell fates during meiosis and oocyte maturation. Polysome profiles of wild-type and ife-1 lysates that resolve mRNAs by incremental ribosome loading were recorded and fractionated (Fig. 2A,B). mRNAs at the top of the gradient, above the 80S peak, do not have ribosomes bound and are not translating. This includes mRNAs in mRNP structures such as P granules. A constitutive P granule protein, PGL-1, remained at the top of the gradient in non-sedimenting complexes (Fig. 2C,D), indicating that P-granule-bound mRNAs do not co-sediment with ribosome-bound mRNAs. mRNAs that sediment past the 80S peak have multiple ribosomes bound. These sediment progressively further with greater translational efficiency. Ribosome loading increases exponentially rather than linearly along the gradient. mRNAs that shift from ‘light polysomes’ (fractions 5–7) to ‘heavy polysomes’ (fractions 8–10) in the gradient have undergone a greater change in protein synthesis efficiency than a similar shift near the top or middle of the gradient. Only a modest decrease in the total polysome content was observed for ife-1. This is to be expected because global protein synthesis (even cap-dependent) is still being supported through the other eIF4E isoforms present.

The results of the TSAA led us to address the translation efficiency of germline mRNAs that contribute to the phenotype observed in the ife-1 mutant. TSAA results led us to discover new IFE-dependent mRNAs vab-1, vpr-1, rab-7 and rnp-3, translationally controlled in late oocytes. To determine changes in the translational efficiency of mRNAs identified in our TSAA (MFR>1.5), RNA from each sucrose fraction was subjected to real-time PCR (qPCR). Primers were used to detect vab-1 (MSP receptor), vpr-1 (MSP-like protein) and rab-7 (Rab GTPase) mRNAs. VAB-1 suppresses the MAPK activity required for late stage oocyte maturation. In the ife-1 worms, there was a reduction in heavy polysomal vab-1 mRNA (fractions 8–10) with a corresponding increase in non-translating and light polysomal vab-1 mRNA (fractions 3–5; Fig. 2E). This decrease in translational efficiency indicates that vab-1 utilizes primarily IFE-1 for initiation. mRNAs encoding VPR-1 and RAB-7 also showed a marked decrease in heavy polysome loading, with a corresponding increase in non-translating mRNA (Fig. 2F,G). More non-translating vpr-1 and rab-7 mRNA indicated that initiation events were less frequent in the absence of IFE-1. The decrease in translational efficiency of vpr-1 and rab-7 indicates that, in addition to vab-1, two mRNAs involved in VAB-1 signaling are also translationally recruited by IFE-1. Another mRNA strongly regulated by IFE-1 encoded RNP-3, a small nuclear ribonucleoprotein (snRNP) that was identified in the TSAA. Loss of RNP-3 activity has been shown to cause embryonic lethality (Saldi et al., 2007). Like vab-1, vpr-1 and rab-7 mRNAs, rnp-3 translation was much less efficient in the absence of IFE-1, confirming the 1.6-fold decrease observed in the TSAA (Fig. 2L). Much of its mRNA shifted from heavy polysomes and monosomes into the non-translating region of the gradient (fractions 1–3; Fig. 2H). General mRNA translation was not perturbed, as evidenced by the distribution of the housekeeping gpd-3 (GAPDH) mRNA (Fig. 2I). Therefore, other eIF4E isoforms support translation initiation of this mRNA and many others (e.g. nos-3, profile not shown; see Fig. 1C). These data show that proteins (VAB-1, VPR-1 and RAB-7) that regulate oocyte MAPK are strongly dependent on IFE-1 for their synthesis. The biological consequences of their translational control will be addressed later in this study.

Other mRNAs appear to be regulated only in part by IFE-1. RAN-1, similar to RAB-7, is a GTPase that regulates VAB-1 trafficking in maturing oocytes. The dependence of ran-1 mRNA on IFE-1, measured by TSAA, was considerably lower (Fig. 2L). However, when assessing the translational efficiency by resolved polysome gradients, a more complex distribution of ran-1 mRNA was evident. In the absence of IFE-1, ran-1 showed a modest decrease in heavy polysome loading. Most of its redistribution was among polysomes of smaller sizes (fractions 4–7; Fig. 2K), rather than out of polysomes altogether. Given that all polysomes were pooled for the TSAA, such changes in ran-1 translation would not have been reflected in the MFR. This points out the limitation of assessing simple ‘on–off’ state translation assays, and suggests that the TSAA underestimates the translational impact for some mRNAs and overestimates for others. Quantifications by both means, however, are generally consistent (Fig. 2L), and substantiate the use of TSAA to screen for some (but not all) IFE-1-regulated mRNAs. A complex polysome profile is also found for gld-1 mRNA, which encodes a crucial protein in germ cell development. GLD-1 is an mRNA-binding protein that suppresses mitosis and promotes meiosis by repressing the translation of key regulatory mRNAs, like glp-1 (Hansen et al., 2004b; Jones et al., 1996). The translational efficiency of gld-1 mRNA in ife-1 worms was diminished as indicated by a shift in the distribution of its mRNA, but it was not lost from polysomes altogether. The most efficiently translating gld-1 (heavy polysomes, fractions 9–10) was lost, and a strong peak in fractions 4–5 (light polysomes) appeared (Fig. 2J). The redistribution of just one peak suggests that IFE-1 recruits a limited subset of ran-1 and gld-1 mRNAs. Such a complex redistribution could indicate that other IFE isoforms weakly initiate their translation in all tissues. More likely, however, is that IFE-1 strongly recruits gld-1 and ran-1 mRNAs in particular regions of the germ line, whereas other IFEs initiate their translation elsewhere.

Several of the IFE-1-dependent mRNAs have also been found in mRNP structures (e.g. gld-1 and oma-1) (Detwiler et al., 2001; Jeong et al., 2011; Schisa et al., 2001), and it is possible that their sedimentation far into the gradient could be due to non-translating complexes. To test this, we added EDTA to lysates to chelate Mg²⁺ and separate all ribosomal complexes prior to centrifugation (Fig. 3A). mRNP binding does not require Mg²⁺, so bound mRNAs sediment as before. However, we found that all mRNAs tested, whether they were regulated by IFE-1 (Fig. 3C,D) or not (Fig. 3B), were displaced to the top of the gradient in the absence of Mg²⁺. Release demonstrated that their sedimentation was due to bound ribosomes rather than a dense bound mRNP complex.

Many germline mRNAs in C. elegans are regulated by repression through proteins binding to their 3′UTRs. These mRNAs need to be activated to bind ribosomes. We hypothesize that as such proteins release the 3′UTRs, IFE-1 plays an active role in recruiting these mRNAs for translation initiation in a spatiotemporal manner. To determine the ability of IFE-1 to recruit de-repressed mRNAs in the proper time and place, we used an in situ 3′UTR reporter assay. We visually monitored the translation of GFP-fused transgene mRNAs to observe changes in protein expression when IFE-1 was depleted by RNA interference (RNAi). As the reporter 3′UTR becomes de-repressed, GFP should be expressed in appropriate germ cells. If regulation of these mRNAs is mediated solely by de-repression, GFP expression will be unaffected by the loss of any one eIF4E isoform. However, if only IFE-1 is able to recruit these mRNAs as they become de-repressed, GFP expression from these mRNAs will not occur. This assay provides the advantage of being able to monitor translational control of an mRNA within single cells as they progress through the germ line in live worms.

GLD-1 is expressed in immature oocytes in the distal end of the gonad where it promotes meiosis, and its expression is translationally controlled by the 3′UTR (Francis et al., 1995; Hansen et al., 2004b). Having established that gld-1 mRNA was regulated by IFE-1 in the polysome analysis (Fig. 2J), we then determined that GLD-1 expression was regionally dependent on IFE-1 for its translation. Using an in vivo reporter assay with an N-terminal fusion of GFP to the gld-1 open reading frame (ORF), and the gld-1 3′UTR, fluorescence microscopy showed that gld-1–gfp mRNA became de-repressed in immature oocytes in syncytium (Fig. 4A). Upon depletion of IFE-1 by RNAi, the fluorescence in the whole distal region of the gonad was visibly diminished. Quantifying the mean GFP fluorescence intensity per exposure time (Fl/t), there was a reproducible twofold increase in GLD-1–GFP expression when IFE-1 was present (Fig. 4B). Our data show that in immature oocytes, IFE-1 uniquely promotes the translation of gld-1 mRNA.

For subsequent assays, we tested reporter constructs that fuse GFP to histone H2B and bear the 3′UTRs of various mRNAs, so that the fluorescence becomes localized in the oocyte nuclei. We previously demonstrated that pos-1 mRNA requires IFE-1 for translation (Henderson et al., 2009). pos-1 (posterior segregation) is natively repressed throughout most of oogenesis. It is de-repressed only in late stage oocytes, so newly synthesized POS-1 protein can help to establish early embryonic polarity. The pos-1 3′UTR reporter showed accumulation of the H2B–GFP product only in nuclei of the three largest oocytes of live worms. Following IFE-1 knockdown, H2B–GFP expression was nearly absent from oocytes of all stages (Fig. 4C). Quantification of GFP (Fl/t) showed a threefold increase in pos-1 3′UTR reporter translation by IFE-1 (Fig. 4D). This indicates that when the pos-1 mRNA became natively de-repressed in the −3, −2 and −1 oocytes, it was primarily IFE-1 that recruited the mRNA for translation. We also tested a tbb-2 3′UTR reporter. tbb-2 encodes β-tubulin and is expressed at high levels in the C. elegans germ line (Lu et al., 2004). Upon IFE-1 depletion there was no visible decrease in H2B–GFP expression from the tbb-2 3′UTR (Fig. 4E). Comparison of GFP (Fl/t) in the late stage oocytes (−3, −2 and −1 oocytes) showed no significant change (Fig. 4F). This was expected because tbb-2 is not a translationally repressed mRNA in germ cells (Merritt et al., 2008) and its translation is not dependent on IFE-1. Other germ cell IFEs can recruit β-tubulin mRNA (and presumably all de-repressed mRNAs) with equal efficiency. Thus, the observed temporally regulated mRNA translation is specific to IFE-1 and not the result of suppressed general cap-dependent initiation.

PGL-1 is an RNA-binding protein that localizes IFE-1 protein to P granules. An H2B–GFP reporter with pgl-1 3′UTR was fully active for translation in early meiotic germ cells (distal) from control RNAi-treated worms, and remained modestly abundant in late stage oocytes. Interestingly, when IFE-1 was depleted, GFP from pgl-1 3′UTR mRNA continued to be expressed in early germ cells and appeared to increase in all of the later stage oocytes (Fig. 4G). Quantification showed a more than twofold increase in GFP in the −3, −2 and −1 oocytes (Fig. 4H). Increased mRNA translation in the absence of IFE-1 suggests that other IFE isoforms more efficiently recruit pgl-1 mRNA. The effect might be due to pgl-1 mRNA stabilization, given that the levels of pgl-1 were elevated in ife-1 worms (Fig. S3B), but its translational efficiency was unchanged (Fig. S3A). By depleting IFE-1 in vivo from these transgenic GFP reporter strains, we identified a unique role of IFE-1 in positive translational control of certain mRNAs within individual germ cells at critical junctures during development. Furthermore, IFE-1 regulates these mRNAs in conjunction with their 3′UTR sequences and therefore must coordinate with the proteins mediating their repression.

Without IFE-1, oocytes grow slowly and mature inefficiently (Henderson et al., 2009). In C. elegans, growing oocytes naturally arrest in meiotic prophase before maturation and fertilization (McCarter et al., 1999). Meiotic arrest is due to signaling from VAB-1, which suppresses MAPK (Cheng et al., 2008; Miller et al., 2003). VAB-1 is the Ephrin receptor tyrosine kinase expressed both in germ cells and their somatic neighbors. MSPs released from sperm cause VAB-1 to be relocalized and degraded (Cheng et al., 2008). VAB-1 degradation restores MAPK activation, induces cyclin-B synthesis and causes resumption of the cell cycle (Miller et al., 2003). Phosphorylated MAPK is a marker of cell cycle progression and maturation. Wild-type gonads display activated MAPK only in the maturing (−1) oocyte when immunostained for diphosphorylated (diphospho)-MAPK (Fig. 5A). In worms depleted of IFE-1 by RNAi, gonads were devoid of activated MAPK (Fig. 5C). By contrast, vab-1 animals showed no suppression of MAPK from the −1 through to the −5 fully grown oocytes (Fig. 5B). As previously published, removing the suppressive role of VAB-1 permits a precocious oocyte maturation phenotype (Miller et al., 2003). When IFE-1 was knocked down in worms devoid of VAB-1, we observed even broader MAPK activation, earlier than the −5 oocyte, back to pachytene stages (Fig. 5D). This indicates that IFE-1 suppresses MAPK activity independently of VAB-1 in very early oocytes. In the absence of both IFE-1 and VAB-1, MAPK activation in all the later stage oocytes (−1 through to −5) was diminished relative to the −1 oocyte in gonads with IFE-1 (Fig. 5A), but greater than those lacking only IFE-1 (Fig. 5C). This indicates that IFE-1 suppresses MAPK activation in immature oocytes (both VAB-1-dependent and independent), while still promoting MAPK activation in the late stage (−1) oocyte. Alternating suppression and activation of MAPK likely involves recruitment of several mRNAs in the immature versus the maturing oocytes. Given that IFE-1 is expressed throughout all germ cell stages (Amiri et al., 2001), we suggest that IFE-1 exerts multiple phases of mRNA recruitment along the path of oogenesis (Fig. 5E).

The contribution of sperm versus cell-autonomous signals for maturation competence becomes important as oocytes become fully grown. Given that ife-1(bn127) worms do not make viable sperm at 25°C (Henderson et al., 2009), we sought to determine which signal is compromised. However, the loss of sperm was not observed in the ife-1(RNAi) phenotype (25°C; Fig. 5C). To rule out that loss of MAPK activation in the −1 oocyte was due to insufficient or crippled sperm, we visualized MAPK under conditions in which wild-type (viable) sperm could substitute for mutant sperm. vab-1 worms crossed with wild-type males (Fig. 6A) exhibited the same extent of MAPK activation out to the −5 oocyte as uncrossed vab-1 worms (Fig. 5B). The same immunostaining pattern suggested that the sperm present in the vab-1 mutant alone are sufficient to promote MAPK activation. We then depleted IFE-1 from vab-1 worms crossed with wild-type sperm. Again MAPK was activated beyond the −5 position (Fig. 6B), as exhibited by the vab-1 ife-1(RNAi) gonads (Fig. 5D). The lack of both VAB-1 and IFE-1 permits this broad, suboptimal MAPK activity throughout the whole proximal gonad.

To address whether sperm–oocyte signaling is altered when IFE-1 is completely absent, we also examined MAPK activation in gonads of ife-1 worms raised at 15°C (permissive for viable sperm) and 25°C (no sperm). Loss of IFE-1 prevented MAPK activation in both the presence and absence of isogenic sperm (Fig. 6C,D). However, this could be due to lower number of sperm (Henderson et al., 2009) or the under representation of msp mRNAs in ife-1 sperm (Fig. S3D). When wild-type sperm were mated to ife-1 worms at 15°C and 25°C, we observed a substantial ectopic MAPK activation in post-pachytene oocytes (Fig. 6I,J). This suggests that IFE-1-mediated suppression antagonizes the MAPK activation in early oocytes mediated by potent sperm. We compared MAPK activation changes due to IFE-1 deficiency with an established sperm-deficient mutant. The fem-2(b245) strain also fails to make viable sperm at 25°C. FEM-2 is a putative protein phosphatase that is known to function in the sperm–oocyte switch (Hodgkin, 1986), but it is unclear what role it plays in post-pachytene oocyte maturation. We have previously shown that unlike ife-1, which slows oocyte growth and fertility, fem-2 has no demonstrable effect on oocyte production or fertility (Henderson et al., 2009). Both ife-1 and fem-2 fail to activate MAPK in the mature (−1) oocyte when no sperm was present (Fig. 6D,F). However, unlike ife-1, fem-2 oocytes were competent for MAPK activation at the permissive temperature (Fig. 6E), indicating a potential difference in their sperm or the oocyte response to sperm. When exposed to wild-type sperm, both ife-1 ‘females’ and fem-2 ‘females’ elicited a broad but dampened MAPK activation (Fig. 6G–J) reminiscent of unregulated MAPK in worms lacking both VAB-1 and IFE-1 (Fig. 5D). These observations suggest that fem-2 acts downstream of or in parallel to ife-1 in the pathway to maturation. The simplest interpretation would be that fem-2 mRNA is likewise IFE-1-dependent; however, we have no evidence for such translational control (TSAA data showed fem-2 had an MFR of 0.99±0.25). Our findings support a model in which IFE-1-dependent mRNAs promote MAPK activation in the mature (−1) oocyte as well as suppress MAPK activation in immature oocytes (Fig. 7).

During germline development, proliferation and differentiation of the stem cells into competent eggs and sperm are regulated by mRNA translational control. Selective protein synthesis of germ cell determinants is governed temporally and spatially by both sequence-specific mRNA repression and translational activation. Much has been described about mRNA repression by RNA-binding proteins (e.g. GLD-1, FBFs, OMA-1, CPEB, Nanos and Pumilio) and how they are vital for germ cell development (reviewed in Friday and Keiper, 2015; Lee et al., 2015; Mendez and Richter, 2001; Nousch and Eckmann, 2013; Parisi and Lin, 2000). Repression, however, represents only the first half of the regulation. Our studies focus on positive translational control of mRNAs and the subsequent developmental activities. Distinct isoforms of translation initiation factor eIF4E carry out selective translational recruitment of mRNAs (Amiri et al., 2001; Dinkova et al., 2005; Henderson et al., 2009; Keiper et al., 2000; Miyoshi et al., 2002; Song et al., 2010). Here, we identified the entire complement of mRNAs recruited by the eIF4E isoform IFE-1 in germ cells. Our data implicate IFE-1-dependent positive translational control in several temporal phases of sperm and oocyte progression, including entry into meiosis and maturation (Fig. 7).

IFE-1 recruits a unique set of mRNAs, distinct from the four other eIF4E isoforms (IFE-2–IFE-5) expressed in C. elegans. The deletion of ife-1 is neither sufficient to substantially disrupt overall protein synthesis, nor the mRNA cap-dependent recruitment step. Three different eIF4E isoforms, IFE-1, -3 and -5, are enriched in the germ line (Keiper et al., 2000). Most individual ife strains are viable and display disparate phenotypes. Defects in ife-1 worms include temperature-sensitive secondary spermatocyte arrest, reduced oocyte production and viability, and embryonic arrest (Henderson et al., 2009). Such phenotypes suggest that IFE-1 recruits a defined subpopulation of mRNAs. Our TSAA identified 77 mRNAs that specifically require IFE-1 for efficient translation. The mRNAs encode GTPases (ran-1, rab-5 and rab-7) involved in cell signaling pathways required for mitotic spindle formation or vesicle trafficking, Ephrin receptor tyrosine kinase activity (vab-1 and vpr-1) that regulates oocyte maturation, and transcription/splicing factors (rnp-3, let-49 and lin-26) necessary for embryonic gene expression (Fig. 1D).

Our data suggest that IFE-1 promotes VAB-1 synthesis in non-maturing oocytes. This Ephrin receptor suppresses the MAPK signaling required for oocytes meiotic maturation (Miller et al., 2003). IFE-1 recruits not only vab-1 mRNA, but also those encoding several VAB-1 regulators. These include VPR-1, the VAP ortholog containing an MSP domain that binds VAB-1 in vitro; RAN-1, a Ran GTPase that promotes trafficking of VAB-1 to suppress MAPK activity; and RAB-7, a Rab GTPase predicted to traffic VAB-1 to lysosomes (Cheng et al., 2008). Thus, one eIF4E isoform acts through cap-dependent positive mRNA translational control to induce both positive and negative signaling for oocyte maturation (Fig. 7). Additionally, the patterns of MAPK activation showed that IFE-1 suppresses MAPK in young post-pachytene oocytes (prior to the −5 position) independently of VAB-1, suppresses MAPK in proximal (−5 to −1) oocytes through VAB-1, and plays an opposite role in promoting sperm-induced MAPK activation in the mature (−1) oocyte.

We propose that IFE-1 mediates the recruitment of several well-characterized translationally regulated mRNAs in late stage oocytes (e.g. vab-1, rab-7, oma-1, mex-1 and pos-1) (Henderson et al., 2009). Other mRNAs identified in our TSAA screen, and verified by polysome resolution of endogenous mRNAs, are also interesting in light of ife-1 phenotypes. rnp-3 mRNA, for example, encodes an snRNP-associated spliceosomal protein RNP-3 (also known as U2B) required for embryonic viability. Although redundantly active with RNP-2 (also known as U1A), knockout of rnp-3 induces embryonic lethality (Saldi et al., 2007). The inability of ife-1 worms to efficiently translate rnp-3 mRNA might contribute to the embryonic arrest and reduced fecundity (Amiri et al., 2001). These data indicate that IFE-1 has a substantially broader translational ‘reach’ than many characterized RNA-binding proteins. IFE-1 induces the selective synthesis of proteins that both promote and suppress cell differentiation, depending upon where and when IFE-1 encounters their mRNAs.

The preferential recruitment of an mRNA by one IFE isoform over another is not likely due to its sequence specificity. eIF4Es bind just the 5′-cap and first two nucleotides of an mRNA (Marcotrigiano et al., 1997). Even though IFE isoforms show individual preferences for either monomethylated or trimethylguanosine caps (Miyoshi et al., 2002), this level of discrimination is still insufficient to explain the mRNA selectivity we have now confirmed for IFE-1, IFE-2 and IFE-4 (Dinkova et al., 2005; Henderson et al., 2009; Song et al., 2010 and this study). Instead, preferential mRNA recruitment might come from interactions of eIF4Es with various 4EBPs (Igreja et al., 2014). The Drosophila Mextli protein is one such 4EBP that acts positively to recruit germ cell mRNAs into productive initiation complexes with eIF4E (Hernandez et al., 2013). Of the C. elegans eIF4E isoforms, IFE-1 uniquely associates with a 4EBP in P granules known as PGL-1 (Amiri et al., 2001). Notably, our gradient fractionation showed that P granules do not co-sediment or associate with ribosome-bound mRNAs suggesting that P granule mRNAs, which are repressed, must be ‘handed off’ to translation complexes. Protein–protein complexes involving other eIF4Es and other 4EBPs have been shown to recognize motifs in mRNA 3′UTRs and repress the translation of the bound message (Piqué et al., 2008; Stebbins-Boaz et al., 1999). As bound mRNAs become required, 3′UTR complexes become remodeled such that eIF4E can associate with eIF4G and recruit the message to ribosomes (Cao and Richter, 2002). In C. elegans, the gld-1 mRNA is bound and translationally repressed by a Pumilio homolog (FBF) in the mitotic region of the germ line (Marin et al., 2003). When germ cells migrate away from the niche, gld-1 mRNA becomes de-repressed and subsequently recruited for translation by IFE-1. As germ cells approach the meiotic region, newly synthesized GLD-1 protein represses the translation of glp-1 mRNA through 3′UTR interactions. The repression of glp-1 is an important event in the transition from mitosis to meiosis (Hansen et al., 2004a). We have shown that IFE-1 plays a direct role in both processes by recruiting each mRNA temporally in turn, as demonstrated by biochemical polysome analyses here and in our previous study (Henderson et al., 2009). The glp-1 and gld-1 mRNAs both require IFE-1 for their most efficient translation, but the exact role that IFE-1 plays in the transition from mitosis to meiosis remains unclear.

Biochemical polysome fractionation allowed us to directly assay the efficiency of individual mRNAs, but it provided no information on where or when each mRNA was being recruited by IFE-1. Instead, use of a histone H2B–GFP 3′UTR reporter assay permitted us to observe translational control events in situ and in vivo. We monitored spatiotemporal translational activation of each reporter mRNA within individual living worm gonads with single-cell resolution. Our data suggest that IFE-1 acts early in the shift from mitosis to meiosis by regional translation of gld-1 mRNA. Additionally, pos-1 mRNA became de-repressed in late stage oocytes, and this too was dependent on IFE-1. Similar late stage oocyte synthesis of MEX-1 protein in the −3, −2 and −1 oocytes was previously observed to be IFE-1 dependent (Henderson et al., 2009). Both observations suggest that other eIF4E isoforms are not sufficient to recruit either pos-1 or mex-1 mRNAs to ribosomes. Not all oocyte mRNAs require IFE-1, as evidenced by unaltered expression of the tbb-2 control. Interestingly, pgl-1 mRNA actually translated better in the absence of IFE-1, and its mRNA increased in abundance. This suggests that IFE-1 competes with the other IFEs (e.g. for initiation complexes or free ribosomes) that mediate PGL-1 synthesis. Thus far we can define three classes of germline mRNAs relative to eIF4E isoform IFE-1: those that uniquely require this isoform following de-repression, those that do not require the IFE-1 isoform at all, and those that use other IFEs to greater advantage. Collectively these data suggest that IFE-1 selectively activates mRNAs at the transition from mitosis to meiosis in the distal gonad (gld-1 and glp-1), at MAPK-induced meiotic maturation in the proximal gonad (vab-1, vpr-1, rab-7, ran-1 and oma-1), and in preparation for embryogenesis (pos-1 and mex-1; Fig. 7). As such, IFE-1 guides unique protein synthetic events that punctuate germ cell development throughout gametogenesis for both the sperm and egg.

The C. elegans strains used were Bristol, N2 wild-type strain, Caenorhabditis Genetics Center (CGC) strains AZ212 unc-119(ed3), ruls32 [Ppie-1:GFP:H2B:tbb-2 3′UTR +unc-119(+)],CZ337 vab-1 (dx31) II, DH245 fem-2(b245) III, JH2060 unc-119(ed3) III; axls 1498 [Ppie-1:GFP:gld-1 ORF:gld-1 3′UTR +unc-119(+)], JH2320 unc-119(ed3) III; axls 1677 [Ppie-1:GFP:H2B:pgl-1 3′UTR +unc-119(+)], JH2427 unc-119(ed3) III; axls 1751 [Ppie-1:GFP:H2B:pos-1 3′UTR +unc-119(+)]. SS712 [ife-1(bn-127)] was obtained from Dr Susan Strome (Molecular, Cell and Developmental Biology, University of California, Santa Cruz). All strains were maintained on normal growth medium (NGM) plates with E. coli strain OP50. For sucrose gradient fractionation and polysomal analysis, worms were grown on chicken egg yolk OP50 plates at 20°C. As previously described, mix-staged worm populations were isolated in M9 buffer, sucrose floated and pelleted in liquid nitrogen in the presence of RNase inhibitors (Dinkova et al., 2005).

Worm gonads were prepared for microscopy as previously described (Henderson et al., 2009), with the following exceptions: dissected gonads were fixed in formaldehyde (3% formaldehyde, 1× PBS) for 30 min, washed with PTW (1× PBS and 0.1% Tween 20), and stored at −20°C in methanol. To prepare gonads for immunostaining, samples were reconstituted and washed with PTW. Gonads were blocked overnight in PTWB (PTW plus 0.5% BSA) at 4°C. The anti-diphospho-MAPK (activated) mouse monoclonal antibody (catalog no. F7776; Sigma) was diluted 1:200 in PTWB and incubated with samples at room temperature for 2 h. Following another PTWB wash, goat anti-mouse-IgG conjugated to Alexa Fluor 488 (Invitrogen) at 1:400 was used as the secondary antibody. Samples were washed in PTWB and mounted on 1.2% agarose pads with 5 μl Vectashield (Vector Laboratories) with DAPI. Gonads were imaged on an Axiovert 200 M inverted microscope (Carl Zeiss) using DIC, FITC, and DAPI filter cubes and analyzers. Images were analyzed with Axiovision 4.3 software (Carl Zeiss).

Bacterial transformation and feeding techniques were as described previously (Contreras et al., 2008), with the following exceptions. The ife-1-specific plasmid (pT72id) was constructed by subcloning 153 bp (nt 589–741) of ife-1 cDNA sequence into plasmid pL4440. Double-stranded RNA was expressed in E. coli strain HT115(DE3) in culture containing tetracycline and ampicillin and was induced with IPTG. Induced cultures were grown on NGM agar medium plates supplemented with tetracycline, ampicillin and IPTG. Strains were synchronized and L3 or L4 hermaphrodites were transferred onto seeded RNAi plates at 22°C for 36–40 h. Hermaphrodites were transferred to RNAi plates, with a limit of three F0 adults per plate and incubated at 25°C for 24 h. F0 adults were then removed. The F1 generation was incubated at 25°C for another 24 h. Progeny were dissected or directly utilized for microscopy.

Adult hermaphrodites were transferred onto agarose pads with Vectashield mounting medium with DAPI for microscopy. AxioVision V4.8.2 (Carl Zeiss) was used to identify optimal exposure time (OET) for linear fluorescence output. Images were acquired along a time gradient of −40% OET, −20% OET, OET, +20% OET and +40% OET. A measurement for regions of interest (ROIs) was defined manually for each image. The densiometric mean was within each ROI and assembled on an Excel spreadsheet. Background fluorescence was defined from the y-intercept of the linear regression derived by plotting densiometric mean versus the percentage OET. Mean fluorescence at OET minus background fluorescence was divided by OET. Multiple values of mean fluorescence per exposure time were used to derive average values and standard deviations presented in the graphs in Fig. 4. P values were calculated using a two-tailed t-test. ROIs were defined for the distal region of immature oocytes (gld-1 3′UTR reporter) or for expression in the nucleus of individual oocytes (tbb-2, pos-1 and pgl-1).

Sucrose gradient fractionation (10–45%) was performed on wild-type and ife-1(bn127) worms as previously described (Dinkova et al., 2005) with modifications (Henderson et al., 2009). Profiles of the gradient were collected by continuous monitoring of the UV absorbance at 254 nm. It should be noted that profiles take on a different character from experiment to experiment, depending upon the amount of worm lysate loaded, the sensitivity of the UV monitor, and the exact centrifugation parameters of a run, as we have seen in previous publications (Dinkova et al., 2005; Henderson et al., 2009). This does not indicate changes in measured translation between experiments (e.g. polysome number), merely different placement (e.g. fraction number of a tetramer polysome or the gpd-3 control peak). For that reason, direct comparisons are made only between separate worm samples from the same sedimentation run (Fig. 2). For the TSAA (Fig. 1), however, three similar but separate sedimentation runs were used to normalize variability for general comparison.

Sample preparation for qPCR was identical to that previously described (Dinkova et al., 2005), except that RNA from gradient fractions was isolated in four volumes of Trizol (Invitrogen). RNA was isolated from one half of each 1-ml gradient fraction. cDNA was synthesized using Verso cDNA synthesis kit (Thermo Scientific). qPCR was performed on biological duplicates using an Sso Fast Evagreen Super mix (Bio-Rad) in an iCycler iQ5 Real-time PCR machine, following the manufacturer's protocol. Primer sequences used were: gpd-3 (forward 5′-GATCTCAGCTGGGTCTCTT-3′, reverse 5′-TCCAGTACGATTCCACTCAC-3′); rnp-3 (forward 5′-CGATCTATGTGAACAATCTCAA-3′, reverse 5′-ACCGAATTGTGTGAAAACC-3′); rab-7 (forward 5′-GCTCGGAGTCGCTTTTTATC-3′, reverse 5′-GGACAAACGGGAAATGGTCT-3′); vpr-1 (forward 5′-ACGAGGATAGTTTTGCTTCTT-3′, reverse 5′-ACTGTCGATTTCAACGATTT-3′); gld-1 (forward 5′-TTCAGGTCCAGTTTTGATGT-3′, reverse 5′-GACGTTAGATCCGAGAAGGT-3′); vab-1 (forward 5′-AAGAATATTGGACGGTTGG-3′, reverse 5′-GTCGCATATTCGGTAGTAAA-3′); ran-1 (forward 5′-ACTTGTCTTCCACACCAATC-3′, reverse 5′-GAGCGGTAACATCGAACA-3′). The real-time quantification of each mRNA was normalized to total RNA content. For detection of PGL-1 protein across the gradient, 20 µl of each fraction was mixed directly with an equal volume of SDS loading buffer and subjected to SDS-PAGE on a 10% gel, then immunoblotted with 1:2000 dilution of rabbit anti-PGL-1 antiserum as previously described (Amiri et al., 2001; Kawasaki et al., 1998).

Three independent sets of 10–45% sucrose gradients were prepared for wild-type (denoted by the designator w in equations below) and ife-1(bn127) worms (denoted by the designator i in equations below) as previously described (Dinkova et al., 2005; Henderson et al., 2009). Following RNA isolation from individual fractions, non-polysomal (NP) fractions 1–4 and polysomal (P) fractions 6–12 were pooled separately. Fraction 5 contained the 80S peak. These twelve samples were purified and hybridized to Affymetrix C. elegans arrays (UNC Functional Genomics Core Facility). We created formulas in Excel to compare raw microarray hybridization signal values with ratios that were normalized to total non-polysomal and polysomal mRNA signals. [Pw/(NPw+Pw)]=Rw, [Pi/(NPi+Pi)]=Ri, [Rw/Ri]=relative fold change in R (MFR). Summing the non-polysomal and polysomal signals (NP+P) from single replicates, we also accounted for the mean fold change in total mRNA (MFT, Fig. S4). Affymetrix returned a probe signal output value for each of 22,625 genes. Each gene was tagged as being statistically ‘present’ or ‘absent’ for each of the four samples (Pw, NPw, Pi and NPi). Signals flagged as ‘absent’ by Affymetrix in three or more of the four samples for a single replicate were removed. This analysis was performed in three biological replicates. Any mRNAs that did not pass the present or absent test in all three replicates were removed from the dataset, leaving a representative group of 13,372 mRNAs. The entire microarray dataset has now been submitted to GEO under the accession number GSE74459. To derive mRNAs translationally regulated by IFE-1 only, mRNAs with an MFR greater than 1.5 fold and an MFR−s.d.>1 were considered. This yielded a prioritized list of 77 mRNAs with decreased translational efficiency in the absence of IFE-1. mRNAs were identified by cross-referencing information provided by Affymetrix with WormBase and GeneBank online resources.

We thank Drs Kyle Mansfield and Chris Geyer for improving this manuscript. The fem-2, vab-1, and all of the GFP–3′UTR strains were provided by the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructure Programs [grant number P40 OD010440].