Main

A single sub-anaesthetic dose of ketamine elicits a rapid (within hours) and sustained (up to seven days) antidepressant response in patients with treatment-resistant major depressive disorder (MDD)4,5,6, serving as the basis for the approval of the enantiomer (S)-ketamine (esketamine) by the FDA for treatment of MDD. Ketamine may exert its antidepressant effects via one of its metabolites, (2R,6R)-HNK9, which may act as an inhibitor of NMDA (N-methyl-d-aspartate) receptors at certain concentrations9,15,16. Ketamine and (2R,6R)-HNK activate mTORC1 signalling and protein synthesis in the prefrontal cortex (PFC) and hippocampus (HPC)7,10,11,17,18,19,20. Furthermore, in rodents, the antidepressant response to ketamine and (2R,6R)-HNK is blocked by infusion of rapamycin, an allosteric inhibitor of mTORC1, into the PFC10,11. mTORC1 affects cellular functions as diverse as nucleotide and lipid synthesis, glucose metabolism, autophagy, lysosome biogenesis, proteasome assembly, and 5′-cap-dependent mRNA translation12,13. Therefore, it is important to understand the specific mechanism by which ketamine induces its antidepressant effect.

Members of the 4E-BP family are pivotal mTORC1 effectors that control mRNA translation (protein synthesis)21. The 4E-BP partner, eIF4E, binds to the mRNA 5′-cap structure (m7GpppN, where N is any nucleotide) as a protein complex termed eIF4F, which mediates the recruitment of the ribosome13. Hypophosphorylated 4E-BPs bind eIF4E to prevent the formation of eIF4F, thus impeding the initiation of mRNA translation. Following phosphorylation by mTORC1, 4E-BPs dissociate from eIF4E, thus promoting mRNA translation initiation21. The mTORC1–4E-BP signalling axis is crucial for synaptic plasticity21.

There are three 4E-BP-encoding genes in mammals. The most abundant isoform in the mouse brain is 4E-BP222, but the presence of 4E-BP1 has also been documented21. Before examining the consequences of ablation of 4E-BP genes on behaviours (Fig. 1), we confirmed the expression of 4E-BP1 and 4E-BP2 in the adult mouse PFC and HPC (Extended Data Fig. 1a). To our knowledge, the distribution of 4E-BP1 and 4E-BP2 in different types of brain cells has not been investigated. In the HPC, neurons (CAMK2α+ and GAD67+) expressed only 4E-BP2 (Extended Data Fig. 1b, c), whereas 4E-BP1 was restricted to astrocytes and microglia (Extended Data Fig. 1d–f). By contrast, in the PFC, 4E-BP1 and 4E-BP2 were expressed in excitatory (CAMK2α+) and inhibitory (GAD67+) neurons (Figs. 2a, b, e, f).

Fig. 1: 4E-BP1 and 4E-BP2 are required for the antidepressant effect of ketamine.
figure 1

a, Time immobile in the FST for wild-type, Eif4ebp1−/− and Eif4ebp2−/− mice treated with saline (Sal), ketamine (Ket) or (2R,6R)-HNK (HNK) and tested 1 h after treatment (two-way ANOVA, Fisher’s LSD, n = 10, 10, 11, 8, 9, 7, 10, 11, 10 from left to right). b, NSF performance of mice treated with saline or ketamine (two-way ANOVA, Fisher’s LSD, n = 15, 14, 11, 10, 15, 13 from left to right). c, NSF performance of mice treated with saline or (2R,6R)-HNK (two-way ANOVA, Fisher’s LSD, n = 9, 8, 9, 10, 5, 7 from left to right). d, Left, Slopes of CA1 fEPSPs recorded in slices from wild-type, Eif4ebp1−/− and Eif4ebp2−/− mice during treatment with ketamine (black bar indicates time of ketamine treatment). Top left, representative traces before (−2 min, black) and 90 min after (grey) application of ketamine (arrows indicate times of traces). Right, comparison of fEPSP slope before (−2 min) and 90 min after treatment with ketamine (two-way ANOVA, Bonferroni’s test, n = 8, 7, 8, from left to right). Mean ± s.e.m.; *P < 0.05, ***P < 0.001 versus saline-treated wild-type mice (a–c) or versus post-ketamine wild-type mice (d); $P < 0.05 versus post-ketamine Eif4ebp1−/− mice; $$$P < 0.001 main effect of treatment.

Source data

Full size image
Fig. 2: Differential effects of 4E-BP1 and 4E-BP2 in neuronal populations in response to ketamine and (2R,6R)-HNK.
figure 2

a, Immunostaining for CAMK2α and 4E-BP1 in the medial PFC of control and Eif4ebp1Ex mice (representative of n = 3). b, Immunostaining for CAMK2α and 4E-BP2 in control and Eif4ebp2Ex mice (n = 3). c, Time immobile in the FST for saline-treated control, Eif4ebp1Ex and Eif4ebp2Ex mice (1 h; one-way ANOVA, n = 47, 17, 17 from left to right). d, Effects of ketamine, (2R,6R)-HNK and fluoxetine (Fluox) on time immobile in the FST (shown as percentage of time immobile with saline treatment; one-way ANOVA and Dunnett’s test, n = 29, 22, 17, 14, 14, 14, 15, 16, 13 from left to right). e, Immunostaining for GAD67 and 4E-BP1 in the medial PFC of control and Eif4ebp1In mice (n = 3). f, GAD67 and 4E-BP2 immunohistochemistry in control and Eif4ebp2In mice (n = 3). g, Time immobile in the FST for saline-treated control, Eif4ebp1In and Eif4ebp2In mice (one-way ANOVA and Tukey’s test, n = 38, 28, 20 from left to right). h, Effects of ketamine, (2R,6R)-HNK and fluoxetine on time immobile in the FST in control, Eif4ebp1In and Eif4ebp2In mice (one-way ANOVA and Dunnett’s test, n = 30, 22, 21, 28, 10, 8, 16, 15, 15 from left to right). i, Effects of ketamine on NSF in control, Eif4ebp1Ex, Eif4ebp2Ex, Eif4ebp1In and Eif4ebp2In mice (two-way ANOVA, Fisher’s LSD, n = 17, 17, 6, 9, 10, 14, 10, 12, 4, 6 from left to right). Mean ± s.e.m.; *P < 0.05, **P < 0.01, ***P < 0.001 versus saline-treated mice.

Source data

Full size image

The effect of ketamine requires 4E-BPs

We injected (R,S)-ketamine (10 mg kg−1) or saline intraperitoneally (IP) into mice lacking 4E-BP1 (Eif4ebp1−/−) or 4E-BP2 (Eif4ebp2−/−) and their wild-type littermates and evaluated their performance in the forced swim test (FST). In wild-type mice, ketamine induced a decrease in immobility at 1 h (by 67.9%, Fig. 1a), 24 h (by 47.8%, Extended Data Fig. 2a), and 144 h (by 55.1%, Extended Data Fig. 2b) after injection. By contrast, ketamine did not induce any change in immobility in either Eif4ebp1−/− or Eif4ebp2−/− mice at any of the time points (Fig. 1a, Extended Data Fig. 2a, b). Even a higher dose of ketamine (30 mg kg−1, IP), which reduced immobility by 80.6% in wild-type mice six days after injection, did not decrease immobility in 4E-BP1 or 4E-BP2 mutant mice (Extended Data Fig. 2b).

To determine whether the antidepressant function of the ketamine metabolite (2R,6R)-HNK also depends on 4E-BPs, we treated Eif4ebp1−/− and Eif4ebp2−/− mice and their wild-type littermates with saline or (2R,6R)-HNK (20 mg kg−1, IP). Injection of (2R,6R)-HNK induced a 61.1% reduction in immobility in wild-type mice, but not in Eif4ebp1−/− or Eif4ebp2−/− mice (Fig. 1a). By contrast, a single dose of fluoxetine, a selective serotonin reuptake inhibitor (SSRI; 3 mg kg–1, IP; 0.5 h before testing), which reduces immobility in the FST, was similarly effective in wild-type, Eif4ebp1−/− and Eif4ebp2−/− mice (Extended Data Fig. 2h). Treatments with ketamine or (2R,6R)-HNK did not alter the distance travelled in an open field in wild-type, Eif4ebp1−/− or Eif4ebp2−/− mice (Extended Data Fig. 2c), indicating that potential differences in locomotion did not influence the FST results.

To corroborate these results, we tested mice in the novelty-suppressed feeding (NSF) task. In wild-type mice, ketamine reduced the latency to feed in a novel environment by 37.5% one hour after administration (Fig. 1c), without affecting the latency to feed in the home cage (Extended Data Fig. 2d). By contrast, in Eif4ebp1−/− and Eif4ebp2−/− mice, ketamine did not affect the latency to feed in the novel environment (Fig. 1c). Similarly, (2R,6R)-HNK reduced the latency to feed in a novel context in wild-type mice (by 61.1%), but not in Eif4ebp1−/− or Eif4ebp2−/− mice (Fig. 1c), without affecting the latency to feed in the home cage (Extended Data Fig. 2e) or locomotion during the NSF test (Extended Data Fig. 2f). Furthermore, ketamine reduced immobility in the tail suspension test (TST) in wild-type mice (by 17.0%), but not in Eif4ebp1−/− or Eif4ebp2−/− mice (Extended Data Fig. 2g).

Ketamine induces synaptic plasticity, which can be blocked by AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptor antagonists9,10,23. To determine whether potentiation of excitatory synaptic transmission in the HPC was impaired in Eif4ebp1−/− and Eif4ebp2−/− mice, we treated hippocampal slices from these and wild-type mice with ketamine (20 μM for 30 min; Fig. 1d). Ketamine increased the slope of field excitatory postsynaptic potentials (fEPSPs) in hippocampal region CA1 in all three strains, but it was larger in wild-type mice than in Eif4ebp1−/− mice (by 17.43%) or Eif4ebp2−/− mice (by 35.26%) (Fig. 1d). Consistent with the neuronal expression of 4E-BP2 in the HPC (Extended Data Fig. 1b–f), attenuation of ketamine-induced fEPSP potentiation was greater in Eif4ebp2−/− mice than in Eif4ebp1−/− mice (by 17.86%) or wild-type mice. Overall, these results demonstrate that both 4E-BPs contribute to the synaptic and behavioural effects of ketamine in mice.

Mutation of 4E-BPs in excitatory neurons

Ketamine may act directly on excitatory cortical neurons, or on inhibitory interneurons to increase excitability10,14,23,24, leading to the activation of mTORC1 in excitatory neurons14. To test whether 4E-BPs are required in excitatory or inhibitory neurons, we generated cell-specific deletions of Eif4ebp1 and Eif4ebp2 (Extended Data Fig. 3). Mice deficient for either 4E-BP1 or 4E-BP2 in excitatory neurons (Eif4ebp1Ex or Eif4ebp2Ex; Fig. 2a, b) showed no change in immobility in the FST compared to control mice (Fig. 2c). We treated the mice with ketamine, (2R,6R)-HNK or fluoxetine. Control mice responded to ketamine or (2R,6R)-HNK with a decrease in immobility in the FST (ketamine, 27.6%; (2R,6R)-HNK, 39% reduction; Fig. 2d). Consistent with the premise that ketamine stimulates mRNA translation in excitatory neurons, mice deficient for either 4E-BP1 or 4E-BP2 in excitatory neurons showed no response to either ketamine or (2R,6R)-HNK (Fig. 2d). Notably, fluoxetine reduced immobility in the FST in control (39.4%), Eif4ebp1Ex (60.2%) and Eif4ebp2Ex mice (37.7% reduction; Fig. 2d), showing that the acute response to SSRIs (the actions of which are mechanistically distinct from those of ketamine) does not require the 4E-BPs in glutamatergic neurons.

Deletion of 4E-BPs in inhibitory neurons

Dysregulation of GABA (γ-aminobutyric acid)-mediated neurotransmission contributes to the pathophysiology of MDD, and the effect of ketamine on these neurons has been proposed to be central in eliciting antidepressant responses25,26,27. Consistent with this, mice lacking 4E-BP2 in GAD2+ inhibitory neurons (Eif4ebp2In, generated using Gad2-Cre mice; Fig. 2e, f) showed decreased immobility when treated with saline (Fig. 2g) compared to wild-type or Eif4ebp1In mice. Furthermore, lack of 4E-BP1 or 4E-BP2 in Gad2-expressing cells conferred resistance to ketamine and (2R,6R)-HNK (Fig. 2h), and even to the acute effects of fluoxetine (Fig. 2h).

To determine whether the long-lasting effect of ketamine was also occluded in mice with cell-specific mutations of Eif4ebp1 and Eif4ebp2, we measured immobility in the FST six days after treatment with saline or ketamine (Extended Data Fig. 4c). Similar to the one-hour time point, ketamine decreased immobility only in control mice, and not in Eif4ebp1Ex, Eif4ebp2Ex, Eif4ebp1In or Eif4ebp2In mice (Extended Data Fig. 4c). In addition, in Eif4ebp2Ex, Eif4ebp1In and Eif4ebp2In mice, ketamine did not decrease the latency to feed in a novel environment, although it did in control and Eif4ebp1Ex mice (Fig. 2i). Neither treatment nor genotype altered latency to feed in the home cage (Extended Data Fig. 4a) or locomotion during the NSF test (Extended Data Fig. 4b). Thus, 4E-BP1 and 4E-BP2 in inhibitory neurons are required for the acute and sustained effects of ketamine; this highlights the importance of inhibitory neuronal circuit elements in the treatment of depression.

The absence of 4E-BP2 in inhibitory neurons resulted in diminished immobility in the FST (Fig. 2g); we therefore studied the role of 4E-BP2 in inhibitory neurons using patch-clamp recordings from hippocampal slices. First, we measured spontaneous excitatory postsynaptic currents (sEPSCs) in CA1 pyramidal neurons in slices, 24 h after treatment of mice with saline or ketamine (IP, 10 mg kg−1), to capture the long-lasting plasticity elicited by in vivo treatment (Fig. 3). In control mice, ketamine induced an increase in the frequency of sEPSCs (Fig. 3a–c), but not in their amplitude (Fig. 3a, Extended Data Fig. 5a, b). By contrast, this effect was absent in slices from Eif4ebp2In mice (Fig. 3a–c). We then studied correlative changes in miniature EPSCs (mEPSCs) and inhibitory postsynaptic currents (mIPSCs) in CA1 pyramidal neurons from control and Eif4ebp2In mice. The frequency and amplitude of mEPSCs were not altered by either treatment or genotype (Extended Data Fig. 6a–e). By contrast, the frequency of mIPSCs was increased in slices from Eif4ebp2In mice (Fig. 4a–c) and decreased by ketamine in CA1 pyramidal neurons from control mice (Fig. 4a, b). Moreover, the decrease induced by ketamine in control mice was absent in Eif4ebp2In mice (Fig. 4a–c). Neither treatment nor genotype affected the amplitude of mIPSCs (Extended Data Fig. 5c, d). These data indicate that long-lasting depression of inhibitory synaptic transmission by ketamine contributes to the elevated excitatory synaptic responses in CA1 pyramidal neurons in control mice.

Fig. 3: Ketamine-induced increase in excitatory synaptic transmission in pyramidal cells depends on the expression of 4E-BP2 in inhibitory neurons.
figure 3

a, Whole-cell recordings of sEPSCs in CA1 pyramidal neurons, 24 h after treatment with saline or ketamine in control or Eif4ebp2In mice. b, Cumulative distributions of sEPSC inter-event intervals as in a (Kolmogorov–Smirnov test, n = 7, 10, 6, 8 from left to right). c, Frequencies of sEPSCs as in a (two-way ANOVA and Bonferroni’s test). Mean ± s.e.m.; *P < 0.05, **P < 0.01, ***P < 0.001.

Source data

Full size image
Fig. 4: Inhibitory synaptic transmission in pyramidal cells and its depression by ketamine depend on the expression of 4E-BP2 in inhibitory interneurons.
figure 4

a, Recordings of mIPSCs from CA1 pyramidal neurons of control and Eif4ebp2In mice 24 h after treatment with saline or ketamine. b, Cumulative distributions of mIPSC inter-event intervals (Kolmogorov–Smirnov test, n = 8, 6, 8, 12 from left to right). c, Frequencies of CA1 mIPSCs (two-way ANOVA). Mean ± s.e.m.; *P < 0.05, ***P < 0.001.

Source data

Full size image

In summary, we have shown that the mTORC1 effectors 4E-BP1 and 4E-BP2 are necessary for the antidepressant response to ketamine (in the FST, NSF test and TST; Figs. 1, 2, Extended Data Fig. 2) and for synaptic plasticity in the HPC (Figs. 1, 3, 4). The ketamine metabolite (2R,6R)-HNK was also ineffective in mice lacking 4E-BP1 or 4E-BP2 (Figs. 1, 2). The 4E-BPs are important for synaptic transmission and for the induction of synaptic and structural plasticity21, such as that induced by ketamine in the HPC (Figs. 1e, 3, 4), the PFC10,28 and the lateral habenula29. Furthermore, the absence of 4E-BP2 has been shown to increase the frequency and amplitude of mEPSCs and the amplitude of mIPSCs in pyramidal CA1 neurons30,31, indicating that this signalling axis is involved in pre- and postsynaptic mechanisms to control the activity of both excitatory and inhibitory neurons. Activation of 4E-BP2 in excitatory neurons and of both 4E-BP1 and 4E-BP2 in inhibitory neurons is an absolute requirement for the antidepressant response to ketamine (Fig. 2). This is consistent with 4E-BP2 being the most abundant 4E-BP isoform in the brain21,22, and specifically in the HPC (Extended Data Fig. 1b–f).

Our finding that knockout of 4E-BP2 in inhibitory neurons induced a decrease in immobility in the FST supports the idea that GABAergic interneurons are important for the response to antidepressant treatments25,26,27. Indeed, the absence of 4E-BP2 in inhibitory neurons resulted in increased GABAergic synaptic transmission (Fig. 4a–c), which coincided with blocked enhancement of excitatory synaptic potentials by ketamine (Fig. 3a–c). Therefore, as part of its mechanism of action, ketamine triggers 4E-BP2-mediated plasticity events in inhibitory neurons that are likely to be central in maintaining the potentiation of excitatory synaptic transmission for days after ketamine administration.

Reduced activity in the mTORC1 signalling pathway has been identified in post-mortem tissue from individuals with MDD32. Thus, this pathway could be a target for treating MDD. Notably, in patients with MDD, one dose of orally administered rapamycin amplified the effect of ketamine33. This finding suggests the possibility that there may be divergent peripheral and central effects of rapamycin in modulating the effects of ketamine. Our results demonstrate that brain 4E-BPs, and downstream initiation of mRNA translation, are a pivotal target for ketamine and its metabolite (2R,6R)-HNK. Therefore, brain 4E-BPs are appealing targets for the treatment of depressive disorders.

Methods

Mice

Adult male mice (aged 60–180 days) were used. Eif4ebp1−/− and Eif4ebp2−/− mice have been reported22,34. Eif4ebp1f/f and Eif4ebp2f/f mice were generated by Ozgene Pty Ltd (Perth, Australia). To generate Eif4ebp1f/f and Eif4ebp2f/f mice, a targeting construct containing a neomycin selection cassette flanked by two short flippase recognition target (FRT) sites was used to generate a conditional allele (Extended Data Fig. 3). The targeting construct also contained two loxP sites between exons 1 and 2 and between exons 2 and 3 for both Eif4ebp1 and Eif4ebp2 genes (Extended Data Fig. 3). The neomycin selection cassette was removed after the generation of the ‘floxed’ lines. Conditional knockouts were generated by crossing Eif4ebp1f/f and Eif4ebp2f/f mice with either Camk2a-cre (Jackson Laboratories stock no. 005359) or Gad2-cre (Jackson Laboratories stock no. 010802) mice. Excitatory neuron knockout mouse lines are referred to as Eif4ebp1Ex and Eif4ebp2Ex; inhibitory knockout lines are Eif4ebp1In and Eif4ebp1In. For experiments using conditional knockout mice, control groups were a combination of either Camk2a-cre or Gad2-cre, Eif4ebp1f/f, Eif4ebp2f/f and wild-type littermates. Genotyping was carried out by PCR amplification of genomic DNA from tails using the set of primers shown in Extended Data Fig. 3. For Eif4ebp1, amplification of wild-type genomic DNA yielded a single 250-bp PCR product, whereas Eif4ebp1f/f genomic DNA produced a 347-bp PCR product; for Eif4ebp2, amplification of wild-type genomic DNA yielded a single 276-bp PCR product, whereas Eif4ebp1f/f genomic DNA produced a 379-bp PCR product (Extended Data Fig. 3).

Mice were housed in groups of 2–5 per cage with free access to food and water (except in experiments involving food deprivation), in temperature- and humidity-controlled rooms (21 °C, ~55% humidity), and a 12-h light–dark cycle. All procedures followed the guidelines of the Canadian Council on Animal Care and were approved by the Animal Care Committees of McGill University and Université de Montréal.

No statistical methods were used to predetermine sample size. The experiments were not randomized, but investigators were blinded to allocation during experiments and outcome assessment.

Behavioural procedures

Mice were handled once 24 h before each experiment. All behavioural experiments were performed between 10:00 a.m. and 3:00 p.m. Behaviours were recorded, stored, and analysed as MPEG files using an automated behavioural tracking system (Videotrack, View Point Life Science, Montreal, Canada) equipped with infrared lighting-sensitive CCD cameras. The analogue signals supplied by the camera were measurements of the luminosity of each point of the image scanned point-by-point and line-by-line at a rate of 25 images per second. These signals were transmitted to the Videotrack system and digitized on 8 bits by digital analogue conversion. Before the experiments, animal–image background contrast detection thresholds were calibrated by visual inspection to distinguish different behavioural patterns.

Forced swim test

Mice were placed in a glass beaker (24 cm tall, 14 cm diameter) containing 3 l water at 24 ± 1 °C. Mice were allowed to swim for 6 min. Immobility was quantified in the last 4 min of the test using the Videotrack system. Data were expressed as percentage of immobility from saline-treated mice of the same genotype.

Tail suspension test

Mice were suspended by their tail from a lever in a 30 × 30 × 30 cm white-painted enclosure. Movements were recorded and quantified for 5 min in a dimly lit environment.

Novelty-suppressed feeding

Mice were food-deprived for 48 h, then each mouse was placed in a brightly illuminated (100 W, 350 lx) open arena (40 × 40 white-painted floor with white walls 30 cm in height) containing lab chow (three pellets) placed at the centre of the arena. The latency to initiate feeding (in seconds) and total distance travelled (cm) were measured. The cut-off time was 600 s35. Feeding latency was also observed in the home cage containing three pellets spread on the floor. The session was terminated immediately after the mice began to feed.

Open field

Mice were placed in a brightly illuminated (100 W, 350 lx) open arena (40 × 40 white plexiglass floors with white walls 40 cm in height) and allowed to explore for 10 min. Total distance travelled was measured (cm).

Drugs

Ketamine hydrochloride (Narketan 10, Vetoquinol) was diluted in 0.9% NaCl solution (saline) and administered IP at 10 mg kg−1 (in some experiments 30 mg kg−1 was used). Mice were tested 1, 24, or 144 h after injection. (2R,6R)-HNK (Tocris Bioscience) was diluted in saline and administered IP at 20 mg kg–1; mice were tested 1 h after injection. Fluoxetine hydrochloride (Tocris Bioscience) was diluted in saline and injected IP at 3 mg kg–1; mice were tested 30 min after injection.

Mouse brain collection

Wild-type, Eif4ebp1−/− and Eif4ebp2−/− mice were briefly anaesthetized using 5% isoflurane and brains were immediately removed; PFC and HPC were dissected, flash frozen and kept at −80 °C until use.

Western blotting

Tissues were homogenized in RIPA buffer and western blotting was performed as previously described30. Antibodies against indicated proteins were: 4E-BP1 (Cell Signaling Technology, 1:1,000), 4E-BP2 (Cell Signaling Technology, 1:1,000) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), coupled to horseradish peroxidase (1:5,000; Abcam); secondary antibody was anti-rabbit (Thermo Fisher Scientific).

Immunohistochemistry

Control, Eif4ebp1Ex, Eif4ebp2Ex, Eif4ebp1In and Eif4ebp2In adult male mice (60–120 days old) underwent transcardial fixation using phosphate-buffered saline (PBS) and then 4% paraformaldehyde. Brains were cryopreserved, flash frozen and stored at −80 °C until use. Coronal slices were prepared (30 μm) and blocked for 1 h (at room temperature) in 5% goat serum (GS) and 0.5% Triton-X in PBS. Then, slices were incubated overnight (4 °C) in 1% GS in PBS with the following antibodies: rabbit anti-4E-BP1 (Cell Signaling Technology, 1:500); rabbit anti-4E-BP2 (Cell Signaling Technology, 1:300); mouse anti-CAMKII-α (Cell Signaling Technology, 1:1,000); mouse anti-GAD67 (EMD Millipore, 1:1,000); mouse anti-GFAP (Cell Signaling Technology, 1:1,000); rat anti-CD11b (BioRad, 1:500).

Next, sections were incubated in darkness (1 h at room temperature) in blocking buffer with AlexaFluor 488 goat anti-rabbit IgG (Thermo Fisher Scientific, 1:400), AlexaFluor 546 goat anti-mouse IgG (Thermo Fisher Scientific, 1:400), AlexaFluor 647 streptavidin conjugate (Thermo Fisher Scientific) or AlexaFluor 647 goat anti-rat IgG (Thermo Fisher Scientific 1:400), and DAPI (4′,6-diamidino-2-phenylindole dihydrochloride, Life Technologies, 1:1,000). Sections were washed with PBS (3 times, 5 min each) after each incubation step. Samples were visualized using the Zeiss LSM800 laser scanning confocal microscope and Zeiss ZEN 3.2 image acquisition software.

Field recordings

For brain slice preparation, wild-type, Eif4ebp1−/− and Eif4ebp2−/− mice (60–67 days old) were anaesthetized with isoflurane and the brain was rapidly removed. Transverse hippocampal slices (400 μm thickness) were allowed to recover for at least 2 h at 32 °C submerged in oxygenated artificial cerebrospinal fluid (ACSF; 124 mM NaCl, 5 mM KCl, 1.25 mM NaH2PO4, 2 mM MgSO4, 2 mM CaCl2, 26 mM NaHCO3 and 10 mM dextrose (pH 7.3–7.4; 295–300 mOsmol). After recovery, individual slices were transferred to a submersion recording chamber and perfused at 2.5 ml/min with ACSF at 27–28 °C. Field excitatory postsynaptic potentials (fEPSPs) were recorded in CA1 stratum radiatum with glass electrodes (2–3 MΩ) filled with ASCF. Schaffer collateral fEPSPs were evoked with a concentric bipolar tungsten stimulating electrode placed in mid-stratum radiatum proximal to the CA3 region. Baseline stimulation was applied at 0.033 Hz by delivering 0.1-ms pulses, with intensity adjusted to evoke fEPSPs with 30% of maximal amplitude. After a 30-min baseline recording, ketamine was added to ACSF (20 μM) for 30 min. Recordings continued for 1 h after ketamine wash out. fEPSP slope measurements were performed on digitized analogue recordings using the Clampfit analyse function (Clampfit 10.6). The slope was measured at between 10% and 90% of maximal fEPSP amplitude during an epoch defined by constant cursor placements, which excluded fibre volley and population spikes. The experimenter was blind to the mouse genotype and codes were revealed after data analysis.

Whole-cell recording of synaptic currents

Acute hippocampal slices (300 μm thickness) were prepared in ice-cold cutting solution containing 75 mM sucrose, 87 mM NaCl, 2.5 mM NaH2PO4, 1.25 mM MgSO4, 0.5 mM CaCl2, 25 mM glucose, and 25 mM NaHCO3. Slices were transferred to ACSF containing 124 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 24 mM NaHCO3, 2 mM MgCl2, 2 mM CaCl2, and 12.5 mM glucose. Whole-cell recordings were obtained from CA1 pyramidal neurons using patch pipettes (borosilicate glass capillaries; 3–5 MΩ). To record EPSCs, the intracellular solution contained 120 mM CsMeSO4, 5 mM CsCl, 2 mM MgCl2, 10 mM HEPES, 0.5 mM EGTA (EGTA), 2 mM QX-314, 10 mM Na2-phosphocreatine, 2 mM ATP-tris, 0.4 mM GTP-tris, and 0.1 mM spermine (pH 7.3–7.4; 285–295 mOsmol). To record IPSCs, the intracellular solution contained 135 mM CsCl, 10 mM HEPES, 2 mM QX-314, 2 mM MgCl2, and 4 mM MgATP.

Recordings were obtained in voltage-clamp mode using a Multiclamp 700A amplifier (Molecular Devices). Holding potential was maintained at −70 mV and series resistance was routinely monitored. Recorded signals were low-pass-filtered at 2 kHz, digitized at 20 kHz and stored on a PC. Data acquisition and off-line analyses were performed using 1440A Digidata acquisition board, and pClamp 10 software (Molecular Devices). Data were included only if the holding current was stable or if series resistance varied by less than 25% of initial value. To record miniature events, tetrodotoxin (TTX 1 μM; Alomone Labs) was added to the ACSF. In mIPSC experiments, 6,7-dinitroquinoxaline-2,3-dione (DNQX, 20 μM) and D(−)-2-amino-5-phosphonopentanoic acid (AP5, 50 μM) were added to block excitatory transmission via AMPA and NMDA receptors. In mEPSC and sEPSC experiments, GABAzine (5 μM) was used to block GABAA receptors, and in sEPSC experiments extracellular K+ concentration was 5 mM and TTX was omitted. For analysis, postsynaptic events were detected visually by Mini Analysis 6.0 software. The detection threshold was set at 3 pA and typically 200–300 events were sampled per neuron (range 125–200 events for EPSCs, 190–300 events for IPSCs). For the cumulative probability distribution analyses, EPSC and IPSC data were binned (bin size 5% of total number of events) for each cell to avoid oversampling from any neuron.

Statistical analyses

Data were expressed as mean ± s.e.m. and analysed using GraphPad Prism 6 (GraphPad Software Inc.). Data from Fig. 2c, d, g, h were analysed using one-way ANOVA, followed by post hoc multiple comparisons tests. Data from Figs. 1a–c, 2i, 3c, 4c, Extended Data Figs. 2a–h, 4a–c, 5b, d, 6c, e were analysed using two-way ANOVA (with treatment and genotype as factors), followed up by planned comparisons. Data from Fig. 1d were analysed using mixed-designed two-way ANOVA, with treatment time as repeated measured variable and genotype as independent variable; this was followed by Bonferroni’s multiple comparisons tests. For cumulative probability distribution analyses of EPSCs and IPSCs in Figs. 3b, 4b, Extended Data Figs. 5a, c, 6b, d, data were analysed using the Kolmogorov–Smirnov test. Values of P < 0.05 were considered significant. Details of statistical analyses are in Supplementary Table 1.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.