eIF4E phosphorylation recruits β-catenin to mRNA cap and promotes Wnt pathway translation in dentate gyrus LTP maintenance

Summary The mRNA cap-binding protein, eukaryotic initiation factor 4E (eIF4E), is crucial for translation and regulated by Ser209 phosphorylation. However, the biochemical and physiological role of eIF4E phosphorylation in translational control of long-term synaptic plasticity is unknown. We demonstrate that phospho-ablated Eif4eS209A Knockin mice are profoundly impaired in dentate gyrus LTP maintenance in vivo, whereas basal perforant path-evoked transmission and LTP induction are intact. mRNA cap-pulldown assays show that phosphorylation is required for synaptic activity-induced removal of translational repressors from eIF4E, allowing initiation complex formation. Using ribosome profiling, we identified selective, phospho-eIF4E-dependent translation of the Wnt signaling pathway in LTP. Surprisingly, the canonical Wnt effector, β-catenin, was massively recruited to the eIF4E cap complex following LTP induction in wild-type, but not Eif4eS209A, mice. These results demonstrate a critical role for activity-evoked eIF4E phosphorylation in dentate gyrus LTP maintenance, remodeling of the mRNA cap-binding complex, and specific translation of the Wnt pathway.


INTRODUCTION
Neuronal activity-dependent synaptic plasticity is crucial for adaptive behaviors such as memory formation, 1 and dysregulation of plasticity is commonly found in animal models of neurodevelopmental and degenerative disorders. At excitatory glutamatergic synapses, the generation of stable structural and functional changes lasting hours or more requires de novo gene expression and protein synthesis. [2][3][4] Global gene expression profiles in various plasticity models have been elucidated using microarrays and RNAsequencing 5-9 as well as ribosome profiling. 10,11 In neurons, regulation of gene expression at the level of translation is critical for long-term synaptic plasticity. 12,13 Translation initiation, the multistep process by which the ribosome is recruited to mRNA, is tightly regulated and often rate-limiting for protein synthesis. 14 Eukaryotic translation initiation factor 4E (eIF4E), which binds to the 5 0 -terminal cap structure of cytoplasmic mRNA, plays a key role in both the process and regulation of translation. eIF4E enables assembly of a multiprotein translation initiation complex at the mRNA 5 0 end, which recruits the 40S small ribosomal subunit and scans to the mRNA start codon. Interaction of eIF4E with the scaffolding protein eIF4G is crucial for initiation complex formation. This interaction is obstructed by eIF4E-binding proteins (4E-BPs). Hypophosphorylated 4E-BPs bind to eIF4E, competes for eIF4G binding and represses translation initiation. Signaling to the kinase mechanistic target of rapamycin (mTORC1) enhances translation by phosphorylation of 4E-BPs, leading to their dissociation from eIF4E. 15,16 In a major convergent pathway, activation of extracellular signal-regulated kinase (ERK, aka mitogen-activated protein kinase; MAPK) signaling to MAPK-interacting kinases (MNK1 and MNK2) phosphorylates eIF4E on a single residue, Ser209. 17,18 Phosphorylation of eIF4E is usually, but not always, associated with enhanced translation initiation. [19][20][21][22] Thus, the molecular function of Ser209 eIF4E phosphorylation is unresolved. 23 Translational control in synaptic plasticity has been extensively studied in excitatory pathways of the hippocampus. A major question is whether translation mechanisms are differentially implemented to sculpt protein synthesis and plasticity in a pathway-specific manner. In the hippocampal CA1 region, ERK-dependent translation initiation regulates stable LTP formation. 24,25 Surprisingly, eIF4E phosphorylation is

Loss of eIF4E Ser209 phosphorylation selectively inhibits DG-LTP maintenance
We first examined the impact of ablating phospho-eIF4E on basal perforant path-DG transmission in adult anesthetized mice. Evoked field potentials were obtained across a range of stimulation intensities and input-output curves were constructed of the field EPSP slope ( Figure S1A), population spike amplitude ( Figure S1B), and plots of the EPSP-population spike relationship were made to evaluate synaptic excitability of granule cells ( Figure S1C). Homozygous Eif4e ki/ki mice were not significantly different from wild-type in any of these measures, indicating that ablation of phospho-eIF4E does not affect synaptic efficacy or granule cell excitability.
We then asked whether ablation of eIF4E phosphorylation impacts LTP induced by application of high-frequency stimulation (200 Hz, 4 trains of 15 pulses) ( Figure 1). In wild-type mice, HFS induced an increase in fEPSP slope which remained stable during 3 h of post-HFS recording ( Figure 1A). In Eif4e ki/ki mice, HFS induced an initial increase in fEPSP slope that was not significantly different in magnitude from wild-type control at 0-10 min post-HFS ( Figures 1A and 1B). However, in striking contrast to wild-type, the fEPSP increase declined completely to baseline by 2 h post-HFS. Already at 30-40 min post-HFS, Eif4e ki/ki mice exhibited a severe (60.6%) reduction in fEPSP potentiation relative to wild-type ( Figures 1B and 1C). Like homozygotes, heterozygous Eif4e ki/+ mice showed impaired LTP at 30-40 min post-HFS (Figures 1B and  1C). However, the inhibition in heterozygotes was transient, as fEPSP responses returned to the wildtype control level to exhibit stable LTP ( Figures 1A and 1B). The results show that phosphorylation of eIF4E is required for maintenance of synaptic LTP, but dispensable for basal transmission and LTP induction. Of interest, Eif4e ki/ki mice showed a stable increase in the population spike, indicating that mechanisms other than eIF4E phosphorylation regulate plasticity of granule cell excitability ( Figures 1C, S2A, and S2B).

Loss of eIF4E phosphorylation inhibits stimulus-induced release of eIF4E repressors and prevents initiation complex formation in DG-LTP in vivo
Next, we examined the biochemical function of eIF4E phosphorylation. Previous work showed that pharmacological inhibition of MNK prevents eIF4E phosphorylation and formation of the translation initiation complex during DG-LTP. According to this model, MNK activity triggers discharge of eIF4E repressors: cytoplasmic FMRP interacting protein-1 (CYFIP1) and the canonical eIF4E-binding protein (4E-BP2) 27,28 ( Figure 2A). However, the specific role of eIF4E phosphorylation is unknown, as MNK has multiple additional substrates with roles in translation and mRNA metabolism. 19,20,31 First, we asked whether ablation of Ser209 eIF4E phosphorylation impacts basal ERK-MNK signaling and initiation complex formation. In unstimulated DG tissue from naive mice, expression of total and phosphorylated (activated) ERK and MNK did not significantly differ between Eif4e ki/ki mice and wild-type (Figure S2B). eIF4E binds to the 7-methylguanosine (m 7 GTP) moiety of the 5 0 -terminal mRNA cap. To assess regulation of the eIF4E cap-binding complex, m 7  iScience Article immunoblotting was used to quantify levels of interacting proteins relative to eIF4E. Phospho-eIF4E was present in cap-pulldown samples from wild-type mice, but absent from Eif4e ki/ki mice. However, there was no difference between genotypes in eIF4G binding to eIF4E, indicating normal formation of the eIF4F initiation complex ( Figure 2C). Levels of cap/eIF4E-associated CYFIP1, FMRP, and 4E-BP2 in Eif4e ki/ki mice were also not different from wild-type ( Figure 2C). Immunoblotting of DG lysate (input) samples similarly showed no genotype differences in expression of the translation factors relative to GAPDH loading control ( Figure S3A). Thus, under basal conditions, absence of eIF4E phosphorylation did not alter ERK-MNK signaling or eIF4E protein-protein interactions.
Next, we analyzed ERK-MNK signaling in response to perforant path stimulation. At 40 min post-HFS, expression of phospho-ERK and phospho-MNK was significantly enhanced in ipsilateral DG relative to the contralateral, non-stimulated DG, with no significant difference between genotypes (Figures 2D and  2E). Thus, HFS-induced ERK-MNK signaling is intact and unaltered in Eif4e ki/ki mice. We then performed cap-pulldown assays in DG lysates to assess changes in eIF4E interactions. In wild-type mice, binding of CYFIP1, FMRP, and 4E-BP2 to eIF4E was significantly reduced while loading of eIF4G was enhanced, relative to the contralateral control DG ( Figures 2F and 2G). In contrast, HFS in Eif4e ki/ki mice failed to discharge the repressor proteins or enhance eIF4G binding to eIF4E ( Figures 2F and 2G). Rather, we observed increased recovery of CYFIP1 and 4E-BP2 in the cap-pulldown in Eif4e ki/ki mice, suggesting stabilization of the interaction complex in the absence of eIF4E phosphorylation. In DG lysates, CYFIP1 levels decreased in HFS-treated wild-type mice and increased in Eif4e ki/ki mice, consistent with degradation of CYFIP1 after release from eIF4E ( Figures S3B and S3C). Taken together, these results suggest that stimulus-evoked phosphorylation of eIF4E on Ser209 is required for the release of translational repressors, formation of the translation initiation complex and maintenance of LTP.
Translation of the immediate-early gene Arc is causally implicated in DG-LTP consolidation, 32 and regulated by MNK signaling in rats and mice. 27,33 In the present study, HFS-induced Arc expression was significantly reduced in Eif4e ki/ki mice relative to wild-type ( Figures S3B and S3C), whereas basal Arc expression did not differ between genotypes ( Figure S3A). Taken together these data suggest that Arc expression in LTP specifically depends on MNK catalyzed phosphorylation of eIF4E.
Ribosome profiling identifies phospho-eIF4E-dependent specific translation in LTP: Enhanced translation of Wnt signaling pathway Next, we used unbiased ribosome profiling to identify changes in translational activity linked to phospho-eIF4E-dependent maintenance of LTP. This analysis focused on 40 min post-HFS as a critical time in the transition to stable LTP. In the LTP experiments, mice received HFS and standard low-frequency test-pulse stimulation (LFS) to assess changes in the fEPSP. The ipsilateral, HFS-treated DG and the contralateral unstimulated DG were collected at 40 min post-HFS) ( Figure 3A). To control for effects of test-pulse stimulation, a control group received LFS only ( Figure 3A). To ascertain the effect of HFS, we normalized the data from ipsilateral HFS-treated DG to LFS-treated DG from respective WT and Eif4e ki/ki mice. We prepared RNA sequencing libraries from both ribosome-protected footprints (a proxy for translation) and total mRNA (a proxy for transcription) ( Figure 3A). Novaseq produced high quality reads for footprints and mRNA libraries because: (1) the distribution of footprint size (28-32 nt) is canonical ( Figure S4B, top panel), (2) the read distribution within the three frames is more abundant for the protein coding frame ( Figure S4B, bottom panel), and (3) the periodicity of ribosomal footprints across mRNA coding and non-coding regions is canonical ( Figure S4C). iScience Article First, we aimed to elucidate the translational landscape of LTP in wild-type (Eif4e +/+ ) mice 40-min after HFS.
In terms of global mRNA translation (TE was calculated by the RPKM reads of footprints normalized to mRNA abundance) and global mRNA abundance, there was no significant effect of HFS treatment relative to contralateral DG or LFS-treated DG ( Figure S5; Table S2). However, we detected an overall increase in mRNA-specific translation at 40 min post-HFS of differentially translated genes (DTGs) and an overall increase in transcription of differentially expressed genes (DEGs), including known immediate-early genes (IEGs) such as Arc, Junb, Npas4, Fos and Fosb ( Figure S5 and Table S2). Gene ontology analysis of DTGs and DEGs using DAVID, and Ingenuity pathway analysis (IPA) revealed LTP-related categories, such as calcium, post-synapse, excitatory post-synapse potential and key biological pathways, such as AMPK signaling, actin cytoskeleton and extracellular matrix ( Figure S5 and Table S3).
Second, we investigated the effect of the ablation of Ser209 phosphorylation, focusing on translational efficiency (TE) at 40 min post-HFS. TE was calculated by the RPKM reads of footprints normalized to mRNA abundance. Although we did not detect significant changes in global mRNA levels between HFS-treated wild-type and Eif4e ki/ki ( Figure 3B, R 2 = 0.981), there was a modest upregulation of translation ( Figure 3B, R 2 = 0.760). Consequently, analysis of log 2 of TE between wild-type and Eif4e ki/ki mice at 40 min post-HFS, as compared to LFS-treated control (ratio<0.667&ratio>1.5; p<0.05), identified that 471 genes in wild-type mice (414 upregulated and 57 downregulated) and 419 genes in Eif4e ki/ki mice (328 upregulated and 91 downregulated) were differentially translated (Table S4). We then performed GO analysis using the DAVID (database for annotation, visualization, and integrated discovery) platform 34 ( Figure 3 and Table S3). Key GO categories identified for upregulated DTGs both in wild-type and Eif4e ki/ki mice include memory, extracellular matrix organization, cell adhesion, cellular response to calcium ion and actin cytoskeleton (Figures 3C and 3D, Table S3). Strikingly, the canonical Wnt signaling pathway and its related genes were absent from the Eif4e ki/ki mice GO analysis, whereas they were significantly upregulated in wild-type mice ( Figures 3D-3F, Table S3). Within this GO Wnt signaling pathway category, we identified translationally upregulated Wnt4, Wnt receptors (Lrp5, Fzd2, Fzd4), and the receptor scaffolding protein disheveled 2 (Dvl2) ( Table S5). In addition, major b-catenin transcriptional targets (Adamts10, Bcl2l2, Ppard, Vegf, Hspg2, pkd1) exhibited enhanced translational efficiency at 40 min post-HFS relative to control mice given LFS only (Table S5). Key GO categories among downregulated DTGs in Eif4e ki/ki mice, as compared to wild-type, include protein synthesis, transcription, poly(A) RNA binding and ribosome ( Figure S4).
The list of mRNAs identified in ribosomal footprinting in Eif4e +/+ and Eif4e ki/ki mice is shown in Table S4. Many Wnt pathway components have long and structured 5 0 UTRs (Untranslated Regions). 35 Given the pronounced change in Wnt pathway mRNA translation, we further analyzed DTGs at 40 min post-HFS in the Eif4e ki/ki vs.Eif4e +/+ mice, focusing on 5 0 UTR mediated mechanisms. We find that the mRNAs of DTGs upregulated in Eif4e +/+ mice at 40 min post-HFS harbor 5 0 UTRs which are significantly longer, and more complex (higher %GC and high folding free energy) and are enriched in uORF, TOP and PG4 motifs compared with downregulated DTG ( Figure S7). Remarkably, HFS in Eif4e ki/ki mice did not elicit the same response as Eif4e +/+ mice, displaying a loss of function phenotype with no significant differences neither in length, % GC, folding free energy nor in the incidence of uORF, TOP and PG4 motifs in upregulated DTG compared with downregulated ( Figure S7). Taken together, these data suggest that eIF4E Ser209 phosphorylation is  Synaptic activity-evoked eIF4E phosphorylation recruits atypical b-catenin to the translation initiation complex In canonical Wnt signaling, b-catenin accumulates in the nucleus and activates transcription of Wnt target genes. 29,30 However, a study in vascular smooth muscle cell cultures shows interaction of b-catenin with FMRP in the eIF4E cap-binding complex. 36 In these cells, Wnt signaling triggers release of b-catenin from the complex, resulting in derepression of translation with nuclear accumulation of b-catenin.
We aimed to determine whether b-catenin is part of the eIF4E-cap complex of adult DG, and, if so, whether it is regulated by LTP-inducing stimuli and Ser209 eIF4E phosphorylation. In DG from naive unstimulated mice, b-catenin was detected in m 7 GTP cap-pulldowns and lysate samples, with no significant difference in expression between Eif4e ki/ki and wild-type mice (Figures 4A-4C). Thus, under basal conditions, b-catenin associates with eIF4E in a manner that does not depend on eIF4E phosphorylation. Following LTP induction, b-catenin in lysate samples from HFS-treated DG was increased 20% relative to the contralateral control, but there was no difference between genotypes ( Figures 4D and 4F). In contrast in the cap-pulldown assays, HFS in wild-type mice elicited a significant mean 78.3% enhancement of b-catenin levels whereas no change was found in Ei-f4e ki/ki mice ( Figures 4E and 4F). These data demonstrate recruitment of b-catenin to eIF4E that is evoked by HFS of perforant path synapses, dependent on eIF4E phosphorylation, and functionally linked to LTP maintenance and enhanced translation of Wnt signaling pathway (model shown in Figure 4G).
b-catenin stability and transcriptional activity are regulated by phosphorylation. We therefore asked whether the b-catenin that associates with eIF4E represents a distinct form. Immunoblotting was done using antibodies recognizing critical phosphorylation sites on b-catenin's N-terminal intrinsically disordered region and within its central Armadillo (Arm) repeats domain. 37 In the absence of Wnt, GSK3-catalyzed phosphorylation of the b-catenin N-terminus (Ser33/Thr41) promotes ubiquitination and proteasomal degradation. 38, 39 In the presence of Wnt, dephosphorylation of the N-terminal residues prevents degradation, allowing accumulation of b-catenin in the nucleus to regulate transcription. We probed with antibodies specifically recognizing phosphorylated (Ser33/Thr41) or non-phosphorylated N-terminal epitopes. Robust signals were detected with both antibodies in DG lysates from naive animals, with no difference between genotypes (Figures 4A and 4C). Following LTP induction, enhanced expression of phosphorylated and non-phosphorylated N-terminal b-catenin were observed, but again there was no difference between knockin and wild-type ( Figures 4D and 4F). Remarkably, immunoblots in cap-pulldown samples from naive and HFS-treated mice of both genotypes were negative for both phosphorylated and non-phosphorylated N-terminal b-catenin ( Figures 4C and 4F). This could mean that b-catenin in the cap complex lacks the N-terminus or is modified in a way that prohibits binding of the specific antibodies. The fact that total b-catenin is reliably detected at the expected molecular mass (95 kDa) in all cap-pulldown and lysate samples shows that b-catenin is not proteolytically cleaved.
Phosphorylation of Ser552 in the Arm domain regulates protein-protein interactions, with enhanced phosphorylation decreasing partner binding and promoting nuclear accumulation. 30,37,40,41 b-catenin Ser552 phosphorylation was detected in both cap-pulldown and lysate samples, but again, with no difference between genotypes at baseline (Figures 4A-4C). Following LTP induction, phospho-Arm was increased in HFS-treated DG relative to the contralateral DG in both genotypes ( Figures 4D and 4F). Strikingly, in cap-pulldown samples, Ser552 phosphorylated b-catenin was significantly increased 78.3% above contralateral control in wild-type mice and this increase was abolished in Eif4e ki/ki mice. However, normalization to total b-catenin showed that Ser552 phosphorylation state did not change ( Figure 4E). Thus, HFS induces a phospho-eIF4E-dependent recruitment of b-catenin to eIF4E where it maintains constitutive levels of Ser552 Arm phosphorylation.
Finally, immunoblotting in DG lysates was done to assess changes in protein expression of select, translationally upregulated Wnt pathway targets. In wild-type mice, HFS significantly increased expression of the key Wnt receptor scaffolding protein, Dvl2, relative to contralateral control, whereas no significant change was observed in Eif4e ki/ki mice ( Figure S8). Expression of Frizzled class receptor 4 (Fzd4) or secreted frizzledrelated protein 1 (Sfrp1) was unchanged, indicating differential impacts on protein expression in early LTP maintenance at 40 min post-HFS. iScience Article DISCUSSION This study elucidates a mechanism and function for Ser209 eIF4E phosphorylation in translational control of DG-LTP in vivo. Biochemically, eIF4E phosphorylation is required for synaptic activity-evoked remodeling of the cap-binding complex. The remodeling is bidirectional, with discharge of the translational repressors CYFIP1 and 4E-BP2, and recruitment of b-catenin. In the first translational profiling analysis of DG-LTP, we find that phospho-eIF4E is specifically required for synaptic activity-induced translation of the Wnt pathway. Functionally, ablation of phospho-eIF4E does not alter basal synaptic transmission or LTP induction but inhibits the maintenance phase of LTP. Previous work demonstrated critical roles for Wnt signaling in LTP at excitatory synapses. [42][43][44] Although b-catenin is known to mediate transcription downstream of Wnt, our results identify novel phospho-eIF4E-dependent recruitment of b-catenin to eIF4E and enhanced Wnt path translation specific to LTP maintenance.
Previous work demonstrated that acute pharmacological inhibition of MNK inhibits eIF4E phosphorylation and DG-LTP maintenance. However, MNKs have additional substrates which could impact translation and mRNA metabolism, including eIF4G, PSF (polypyrimidine tract-binding protein-associated splicing factor), and heterogeneous ribonucleoprotein A1. 19,20,31,45 Here, we demonstrate that activation of ERK-MNK signaling and LTP induction are intact in Eif4e ki/ki mice whereas stable maintenance of LTP is lost. Basal perforant path-evoked synaptic transmission and eIF4E protein-protein interactions were intact, with no difference between genotypes in eIF4E associated CYFIP1/FMRP, 4E-BP2, b-catenin, or eIF4G. Thus, the Ser209Ala Knockin mutation does not lead to compensatory upstream changes in ERK-MNK signaling or downstream remodeling of the eIF4E complex. Rather, our results demonstrate a crucial role for phospho-eIF4E in stimulus-evoked discharge of CYFIP1 and 4E-BP2 from eIF4E and enhanced loading of eIF4G to facilitate initiation. CYFIP1 can shuttle between the eIF4E complex and a Rac-WAVE1 complex involved in actin cytoskeletal remodeling and dendritic spine plasticity. 46,47 If such shuttling occurs in LTP, disruption of CYFIP1 release may impact both translation and actin dynamics.
Wnts have broad and diverse functions in embryonic patterning and development, including neuronal dendrite development and synapse formation, and continue to function in activity-dependent synaptic plasticity in the adult brain. 30,42,48 Neuronal activity-induced Wnt secretion and signaling through canonical b-catenin and the non-canonical planar cell polarity (PCP) and calcium pathways are implicated in LTP of excitatory synaptic transmission. [42][43][44]49,50 Roles for Wnt signaling in trafficking of AMPA-type glutamate receptors and structural plasticity of dendritic spines have been identified. 42,51 In DG-LTP, canonical Wnt/b-catenin signaling is associated with transcription of Wnt target genes. 7,43 Our ribosome profiling analysis uncovered phospho-eIF4E-dependent translation of Wnt-4 and Wnt receptors and scaffolds (Dvl2, Lrp5, Fzd-2 and 4) as well as endogenous inhibitors of Wnt receptors (Sfrp1 and 2). Targeting of the Wnt pathway as a class suggests a coordinate regulation which could serve to amplify Wnt signaling in LTP.
Whether recruitment of b-catenin to eIF4E is directly involved in regulation of Wnt pathway translation remains to be determined. Our analysis of mRNA features indicates that phospho-eIF4E preferentially promotes translation of mRNAs with long, structured 5 0 UTRs. However, as these features are not unique to Wnt family components, other factors must provide specificity. In the developing nervous system, b-catenin interaction with N-cadherin regulates dendritic spine plasticity. 52,53 N-cadherin stimulates Akt, which phosphorylates b-catenin Ser552. 54 Phosphorylation of b-catenin Ser552 has been shown to regulate protein-protein interactions. We show that b-catenin in the eIF4E complex is Ser552 phosphorylated, but there is no change in phosphorylation state to support phosphorylation as a mechanism of recruitment. Under reducing conditions on SDS-PAGE gels, immunoblotting showed that antibodies (Right). In wild-type mice, synaptic activation by HFS stimulates ERK-MNK signaling and eIF4E phosphorylation on Ser209. Phosphorylation triggers discharge of eIF4E-binding proteins, both CYFIP1 and 4E-BP2, and recruitment of b-catenin (b-cat) to the eIF4E cap complex. Release of CYFIP1 together with its binding partner FMRP is depicted. The b-catenin that associates with eIF4E is atypical as its N-terminal region (indicated by dotted line) is not detected by specific antibodies. Phosphorylation of eIF4E is required for Wnt pathway translation as a class and underlies LTP maintenance. In Eif4e ki/ki mice, ERK-MNK signaling is activated but loss of eIF4E phosphorylation prevents remodeling of the eIF4E complex, Wnt pathway translation, and LTP. See also Figure S7. iScience Article specific for the N-terminal intrinsically disordered region reliably and clearly detect b-catenin in lysates, whereas cap-pulldown assays are blank. This suggests that b-catenin in the eIF4E complex is a unique form that has undergone a structural change or post-translational modification that prevents detection of the N-terminal epitope.
Translation control can be general, affecting translation in a global manner, or more specific, impacting only subsets of mRNAs. 55,56 HFS in wild-type and Eif4e ki/ki mice increased expression of numerous IEG mRNAs and enhanced translational efficiency in gene ontology categories synaptic regulation and plasticity. Of these, the Wnt pathway was the only category of DTGs completely dependent on phospho-eIF4E. We compared our list of phospho-eIF4E regulated DTGs in DG in vivo with FMRP-target genes 57 and MNK1 targets identified using cortical neuronal cultures from MNK1 knockout mice 58 ( Figure S9). Although the preparations and methods are different, the main conclusion seems to be that there are few common mRNAs (less than 5%). This reinforces the importance of phospho-eIF4E in specific translation during LTP in the DG. DG-LTP consolidation has mechanistically distinct early and late phases of translation with different targets. We concentrated on the critical, early stage of translation (40 min post-HFS) and recognize that new patterns may emerge at later time points.
In cancer models, phospho-eIF4E regulates translation of targets involved in oncogenic transformation. 59,60 In the suprachiasmatic nucleus, phospho-eIF4E regulates translation of clock genes (Per1 and Per2) involved in circadian rhythms 61 and in dorsal root ganglion phospho-eIF4E promotes translation of bdnf involved in hyperalgesia and nociceptive transmission. 62 A recent ribosome footprinting analysis of forebrain tissue of adult Eif4e ki mice revealed regulation of mRNAs involved in inflammation (IL-2 and TNFa) and organization of extracellular matrix (Prg2, Mmp9, Adamts16, Acan). 26,63 Behavioral analyses of Eif4e ki/ki mice have revealed a role for phospho-eIF4E in regulation of depression-like behavior. 26 Collectively, evidence suggests that phospho-eIF4E regulates translation in a region-and stimulus-specific manner.
A previous analysis of the Schaffer collateral-CA1 pathway in hippocampal slices from Eif4e ki/ki mice showed normal basal transmission and LTP induction as well as normal late phase LTP maintenance (recorded for 2 h). 26 The ability to generate stable LTP in CA1 but not DG in Eif4e ki/ki mice could reflect differences in the balance and timing of mTORC1 and MNK-dependent translation in these circuits. LTP maintenance in CA1 requires mTORC1 signaling, which triggers removal of 4E-BP2. [64][65][66][67] Stable CA1-LTP induced by HFS or BDNF application is blocked by the mTORC1 inhibitor rapamycin. 66,68 In DG-LTP, inhibition of mTORC1 signaling by rapamycin does not affect LTP induction or maintenance. 27,28 It is therefore possible that ablation of phospho-eIF4E is compensated by mTORC1 signaling in region CA1, whereas MNK regulation predominates in DG-LTP and is not developmentally compensated in Eif4e ki/ki mice.
A recent study showed that inhibition of eukaryotic elongation factor 2 (eEF2) phosphorylation by conditional deletion of eEF2 kinase dramatically increases neurogenesis in the adult DG, enhances DG-dependent cognitive functions and decreases depression-like behavior. 69 The regulation of neurogenesis is specifically linked to proteostasis of mature DG granule cells, with increased expression of neurogenesis-related proteins (decorin, vimentin). The phenotype in Eif4e ki/ki mice is very different, with normal DG neurogenesis and hippocampal-dependent memory function, whereas depression-like behavior is increased. 26,70 Conceivably, eEF2 has a primary function in supporting neurogenesis-related translation whereas phosho-eIF4E supports synaptic plasticity of pre-existing inputs to mature granule cells. These different forms of translational control may cooperate in DG functions such as regulation of depression-like behavior.

Limitations of the study
Although genotypes did not differ in basal assembly of the translation initiation complex and specific defects in stimulus-evoked translation and plasticity were found in Eif4e ki/ki mice, compensatory developmental changes in the Knockin mice cannot be ruled out. The study uncovers functions for eIF4E phosphorylation in vivo but does not address the molecular mechanisms underlying discharge of repressors and recruitment of b-catenin. In vitro studies are needed to elucidate the molecular function b-catenin in complex with eIF4E.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
The authors declare no competing interests.

INCLUSION AND DIVERSITY
We support inclusive, diverse, and equitable conduct of research. One or more of the authors of this paper self-identifies as an underrepresented ethnic minority in their field of research or within their geographical location.

Materials availability
This study did not generate new unique reagents.
Data and code availability RNA-seq data will be deposited at GEO and made publicly available as of the date of publication. Accession numbers are listed in the key resources table. Original western blot images will be deposited at Mendeley and will be publicly available. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
Animals Eif4e S209A mice previously described 63

Antibodies used
Antibodies used for immunoblotting are listed in Table S1.
In vivo electrophysiology in mice Electrophysiology methods are as described 28 with minor modifications. Adult mice (12-weeks old) were anesthetized with urethane (injected i.p. 1.2 g/kg), which was supplemented throughout surgery and recording as required. Mice were placed in a stereotaxic frame and body temperature was maintained at 37 C. In one hemisphere only, a bipolar stimulation electrode (NE-200, 0.5 mm tip separation, Rhodes Medical Instruments, Wood hills, CA) was positioned for unilateral stimulation of the perforant path (3.8 mm posterior to bregma, 2.5 mm lateral to midline, and 1.6 mm depth from the brain surface) while an insulated tungsten recording electrode (0.075 mm; A-M Systems) was positioned in the DG hilar region (2 mm posterior to bregma, 1.5 mm lateral to the midline, and 1.5 -1.7 mm depth from the brain surface). The recording electrode was lowered into the brain in 0.1 mm increments while monitoring the laminar profile of the response waveform evoked by a 400 mA test-pulse stimulus. To generate input/output (I/O) curves, 7 stimulus intensities ranging from 80 mA to 400 mA were applied in randomized sequence. After generating an I/O curve, a stable 20 min baseline of evoked potentials was recorded using test-pulses of 0.1 ms pulse-width applied at 0.033 Hz. The test-pulses intensity produced a population spike of 30% of maximum. The high-frequency stimulation (HFS) protocol consisted of four trains of stimuli applied with an interval of 10 sec; each train had 15 pulses at 200 Hz (pulse-width 0.1 ms). The stimulus intensity for HFS was twice that used for baseline recordings. After HFS, test-pulse evoked responses were recorded for 40-and 180-min. After recordings were completed, the electrodes were removed, the animal was sacrificed, and the dentate gyri were micro-dissected and immediately frozen on dry-ice for later use. The maximal slope of the initial rising phase of the fEPSP and population spike amplitude were measured, and changes post-HFS were expressed in percent of baseline. In the electrophysiological experiments for ribosome profiling, mice received the standard LTP protocol consisting of low-frequency test-pulse stimulation (LFS) and HFS. DG tissue was collected at 40 min post-HFS.
Half of the lysate was used for mRNA extraction (total mRNA) while the remaining fraction was digested with TruSeq Ribo Profile Nuclease so that only the mRNA fragments protected by ribosomes were recovered (footprints). Both samples (footprints and total mRNA) went through a ribosomal RNA removal step using the Ribo-Zero Gold (Human/Mouse/Rat) Kit (Illumina). The footprint samples were then purified on a 15% TBE-Urea polyacrylamide gel (ThermoFisher Scientific) to select for fragments of 28-30 nucleotides. The total mRNA samples were heat-fragmented, according to the TruSeq Ribo Profile protocol, to yield small RNA fragments. Footprints and total mRNA fragments were used to prepare small RNA libraries, using the TruSeq Ribo Profile Kit, and were sequenced on an Illumina HiSeq 2500 System, at the Edinburgh Genomics facilities. Bioinformatics analysis was performed as previously described (Amorim et al., 2018). 26 Translational Efficiency (TE) was calculated as the ratio between RPKM of footprints and RPKM of total mRNA for each gene. Data was filtered to include only Differentially Translated Genes (DTGs) that meet the following criteria: FDR<0.05, p-value <0.05, and -1<Log 2 (TE)>0.585.

Gene ontology and pathway analysis
Gene Ontology (GO) and Pathway Analysis were performed using the online tool DAVID (Database for Annotation, Visualization and Integrated Discovery 73 version 6.8) and the Ingenuity Pathway Analysis Software (IPA; Qiagen; version 42012434), respectively. Differentially translated genes were submitted to IPA and subjected to Core Analysis with analysis parameters set to include Direct and Indirect Interactions and Experimentally Observed data only. For further analysis of relevant Canonical Pathways, a Molecular Activity Predictor (MAP) analysis was applied based on the differentially regulated genes belonging to each individual pathway. For GO analysis, filtered gene lists split to highlight genes differentially upregulated or downregulated in each dataset were individually submitted to DAVID and GO annotation gathered for Biological Function, Molecular Function and Cellular Component.

ll
OPEN ACCESS