Fig 2: Electron micrographs of translation machinery in lateral amygdala axons.(a–b) Polyribosomes (black arrows) in axonal boutons (asterisks). A polyribosome in an astrocytic process (white arrow) is visible at the lower left of panel (a). (c–e) Axonal boutons (asterisks) containing immunolabeling (black arrows) for eIF4E (c), eIF4G1 (d), and eIF2a (e). White arrowheads indicate asymmetric synapses onto dendritic spines (a, d, and e) and shafts (b and c). (f) Myelinated axon (asterisk) containing immunolabel for ribosomal protein s6 (arrow). Scale bars = 500 nm.

Fig 3: DAP5 and eIF3d are both required for translation of DAP5-dependent mRNAs. a Overexpression of 4E-BP1 by stable cDNA transformation in MDA-MB-231 cells silenced with siRNAs: non-silencing (NSi), DAP5, eIF4GI, eIF3d, DAP5 + eIF4GI, and DAP5 + eIF3d. b Cap chromatography of cell lysates with or without 4E-BP1 cDNA overexpression. Representative immunoblots show almost total sequestration of eIF4E by overexpressed 4E-BP1. c Representative immunoblots of MDA-MB-231 cells silenced with NSi or DAP5 siRNAs as in (a) probed for proteins whose mRNAs were found to be reduced in polyribosomes by DAP5 silencing (Table 1). Lamin A, AUF1, and GAPDH served as DAP5-independent controls. Results are representative of three independent experiments

Fig 4: DAP5 interacts with eIF3d to drive selective translation independently of eIF4E and eIF4GI. a Distribution of mRNAs encoding c-JUN, ACTB, MMP1, and CDK12 in 48S ribosome-mRNA complexes in 293T cell in vitro mRNA programmed translation extracts (see Supplementary Fig. 5 flow chart for methodology). In vitro translation extracts lacking endogenous mRNAs were programmed with in vitro-synthesized capped and polyadenylated mRNAs. 48S and 60S ribosome subunit distribution profiles from in vitro translation extracts were plotted by relative absorbance at 254 nm against elution fractions. Non-programmed extracts served as controls. b Quantitative RT–PCR analysis of the in vitro translation fractions shown in (a) expressed as relative mRNA abundance as a fraction of total recovered mRNA. c Immunoblot analysis of initiation factors in isolated 48S ribosome-mRNA translation complexes fractionated by sucrose gradients formed with mRNAs encoding c-JUN, ACTB, MMP1, or CDK12. Total protein is a control from non-programmed 293T cell in vitro translation extracts. Ribosomal protein S19 (rpS19) is a loading control (n = 3)

Fig 5: Overexpression of eIF4E in microglia leads to social deficits.a Genotypes of control (Ctrl) and MG4E mice. b Microglial eIF4E overexpression and protein synthesis in cultured microglia isolated from both sexes of control and MG4E mice. n = 3 per condition. Two-sided t-test: puromycin, female **p = 0.0099 and male *p = 0.0217; eIF4E, female **p = 0.0088 and male *p = 0.0293. c Time spent in the central zone during the first 5 min of open field tests. Male: 15 control mice and 14 MG4E mice; Female: 23 control mice and 25 MG4E mice. n.s. not significant by two-sided t-test. d, e Sociability of MG4E mice. Male: 14 control mice and 14 MG4E mice; Female: 23 control mice and 25 MG4E mice. Two-way ANOVA with Fisher’s LSD post-hoc test for chamber time and social investigation: *p < 0.05; **p < 0.01; ***p < 0.001; n.s. not significant. Two-sided t-test for social preference index: male, **p = 0.0067; female, p = 0.7440. f Social habituation test. n = 16 male control and 14 male MG4E mice. Two-way ANOVA with Fisher’s LSD post-hoc test: *p = 0.0159 between control and MG4E mice in trial 1; ###p < 0.001 between trial1 and trial4 in control mice; #p = 0.0334 between trial 4 and trial 5 in control mice; n.s. not significant for comparisons of trial 1 vs. trial 4 and trial 4 vs. novel in MG4E mice. g Habituation index. **p = 0.0013 by two-sided t-test. h, i Novel object recognition. n = 12 control mice and 12 MG4E mice. Two-way ANOVA with Fisher’s LSD post-hoc test: *p < 0.05 and n.s. not significant (h). Two-sided t-test: *p = 0.0297 (i). j, Self-grooming. The 30-min self-grooming test was divided into three 10-min segments. n = 6 mice per genotype. Two-way ANOVA with Fisher’s LSD post-hoc test was used for each 10-min segment, whereas two-sided t-test was used for the whole 30-min test period. *p = 0.0276; **p = 0.0016; n.s. not significant. Source data are provided as a Source Data file.