Cpeb1 remodels cell type-specific translational program to promote fear extinction

Abstract

Protein translation is crucial for fear extinction, a process vital for adaptive behavior and mental health, yet the underlying cell-specific mechanisms remain elusive. Using a Tet-On 3G genetic approach, we achieved precise temporal control over protein translation in the infralimbic medial prefrontal cortex (IL) during fear extinction. In addition, our results reveal that the disruption of cytoplasmic polyadenylation element binding protein 1 (Cpeb1) leads to notable alterations in cell type-specific translational programs, thereby affecting fear extinction. Specifically, Cpeb1 deficiency in neurons activates the translation of heterochromatin protein 1 binding protein 3, which enhances microRNA networks, whereas in microglia, it suppresses the translation of chemokine receptor 1 (Cx3cr1), resulting in an aged-like microglial phenotype. These coordinated alterations impair spine formation and plasticity. Our study highlights the critical role of cell type-specific protein translation in fear extinction and provides an insight into therapeutic targets for disorders with extinction deficits.

Fig. 1.. The protein translation programs are required for fear extinction.

(A) Schematic diagram of the behavioral process for fear conditioning training, Tet-on system, and fear extinction experiment. (B) Percentage of freezing levels is shown in blocks of two sessions across all phases of the experiment (trial 4 during conditioning and trial 14 during extinction). Detailed experimental procedures are shown in the “Behavioral assays” section of Materials and Methods (*P < 0.05, ***P < 0.001, n = 5 to 10). (C) Percentage of freezing level in fear memory acquisition (Acq) and average percentage of freezing of 14 trials in the second stage of extinction learning from No-ext, Normal-ext, and Impaired-ext group mice (***P < 0.001, n = 21). (D) The average percentage of freezing was ranked in the 14 trials during the second stage of extinction learning compared to the percentage of freezing in the last three trials for the mice. (E) PCA for the average percentage of freezing in the 14 trials during the second stage of extinction learning (n = 19; Nor, Normal-ext; Imp, Impaired-ext). (F and G) Analysis of proteome and transcriptome on Impaired-ext and Normal-ext group. Proteome and transcriptome comparison between them was performed in (F). Scatter plot of the fold change of RNA and protein comparing Impaired-ext to Normal-ext is shown in (G) (mRNA, fold change >1.5; Protein, fold change >1.2; group A exhibits an inverse correlation between protein and mRNA levels; group B represents protein levels change while mRNA remains constant; group C represents a positive correlation between protein and mRNA levels). (H) Functional enrichment analysis of gene populations from groups A and B in (F) or (G) (fold change >1.5, P < 0.05, n = 3 per group). (I) Heatmap showing the expression profiles of protein levels (groups A and B) between Impaired-ext and Normal-ext mice (fold change >1.2, P < 0.05, n = 2 per group).

Fig. 2.. Cpeb1-dependent translational signatures in neuron and microglia are crucial for fear extinction.

(A and B) Immunofluorescence staining was performed in the IL region from Impaired-ext and Normal-ext groups by using anti-Cpeb1 (red) and Neun, Iba1, and glial fibrillary acidic protein (GFAP) (green) antibodies. The representative images are shown in (A), and relative intensity of Cpeb1 and the colocalized pixels in different cell populations were calculated in (B) (*P < 0.05, **P < 0.01, n = 6 cells from five mice per group). (C) Representative confocal image (right) for virus injection into IL of the mPFC from Cx3cr1-Cre mice. Coronal slices from virus infected (mCherry) neurons were stained with anti-Cpeb1 (cyan) antibody, and coronal slices from virus infected (EGFP) microglia cells were stained with anti-Cpeb1 (cyan) antibody. White dashed lines indicate neurons or microglia cells with virus infection, and yellow dashed lines indicate neurons or microglia cells without virus infection. (D) Schematic illustration of the workflow for the fear conditioning training and fear extinction learning assay after viral intervention of Cx3cr1-Cre mice. Hab, habitation; Cond, conditioning; Acq, acquisition. (E) Average freezing percentage of fear memory acquisition and average freezing percentage in the first and second sessions of fear extinction learning from M-shCon+N-shCon, M-shCon+N-shCpeb1, M-shCpeb1+N-shCon, and M-shCpeb1+N-shCpeb1 groups (*P < 0.05, **P < 0.01, n = 5 to 9 per group). (F) Percentage of freezing of 14 trials in the second session of extinction learning from M-shCon+N-shCon, M-shCon+N-shCpeb1, M-shCpeb1+N-shCon, and M-shCpeb1+N-shCpeb1 groups (n = 5 to 9 per group).

Fig. 3.. Distinct downstream effectors of Cpeb1 in neuron and microglia regulate fear extinction.

(A) Schematic outlining the Cpeb1 RNA immunoprecipitation sequencing (RNA-IP-seq) workflow from C57BL/6 J mice (n = 2 per group). (B) Scatter plot displaying genes from two independent Cpeb1 RNA immunoprecipitation experiments. Each dot represents one gene, with x and y axes indicating fold changes of IP/IgG for each Cpeb1 RNA immunoprecipitation experiment. (C) Venn diagram illustrating the overlap between CPE-containing genes, RNA-IP-seq genes (fold change of IP/IgG > 2), and proteins identified by mass spectrometry (fold change >1.2, P < 0.05). (D and E) Cpeb1 RNA-IP-seq data for Hp1bp3 [top of (D)] and Cx3cr1 [top of (E)] transcripts. 3′UTRs of Hp1bp3 and Cx3cr1, with WT or mutated CPEs, were inserted downstream of Renilla luciferase (hRluc) gene. These constructs were cotransfected into 293 T cells with the empty vetor pcDNA3.1(+) or pcDNA3.1(+)-Cpeb1, along with a firefly luciferase (Luc) vector as an internal control. After 48 hours, luciferase activity was measured using a dual-luciferase assay. Schematic illustration of WT or mutated CPEs in 3′UTRs of Hp1bp3 [middle of (D)] and Cx3cr1 [middle of (E)] are provided, and the corresponding luciferase activity [bottom of (D) and (E)] are shown (*P < 0.05, **P < 0.01, ***P < 0.001, n = 3 replications). (F and G) HT22 or BV2 cells were transfected with either empty vector (Con) or overexpress Cpeb1 (OE). After 48 hours, RNA was isolated and subjected to PCR poly(A) test. (F) Bands for Hp1bp3 poly(A) tail and primers. (G) Cx3cr1 poly(A) tail and primers (RT+, reverse transcription; RT−, no RT; n = 3 replications). Arrowheads indicate poly(A) PCR product. (H to J) Immunofluorescence of IL sections from Impaired-ext and Normal-ext mice using anti-Hp1bp3/anti-Cx3cr1 (red) (H) and anti-Neun/anti-Iba1 (green) (I) antibodies. Representative confocal images are provided. Normalized intensities of Hp1bp3 and Cx3cr1, Hp1bp3/Neun, and Cx3cr1/Iba1 colocalized pixels were calculated (J) (*P < 0.05, **P < 0.01, n = 7 to 9 cells from four mice per group).

Fig. 4.. Neuron- and microglia-specific Cpeb1-dependent translational signatures are necessary for synaptic function.

(A and B) Whole-cell patch-clamp recordings of IL neurons from Normal-ext and Impaired-ext mice. Representative mEPSC traces are shown (A), with quantification of mEPSC amplitude and frequency (B) (*P < 0.05, n = 7 to 8 neurons from four mice per group). (C to E) Whole-cell patch-clamp recordings of PL neurons from Normal-ext and Impaired-ext mice. Representative mEPSC traces are shown (C), with quantification of mEPSC amplitude (D) and frequency (E) (n = 8 to 9 neurons from four mice per group). (F) Paired-pulse ratios recordings of IL neurons from Normal-ext and Impaired-ext mice. mEPSC traces are shown in the right panel, with quantification of EPSC amplitude on the left (n = 9 neurons from four mice per group). (G and H) Golgi staining of IL sections from Impaired-ext and Normal-ext groups. Representative images are shown (G), with quantification of spine density and mushroom spine percentage (H) (*P < 0.05, **P < 0.01, n = 7 to 8 slices from four mice per group). (I) Golgi staining images showing the dendritic trees from Impaired-ext and Normal-ext groups. (J) Sholl analysis evaluating dendritic complexity in neurons from the Impaired-ext and Normal-ext groups (n = 8 to 10 neurons from four mice per group). (K to M) Dendritic spine in IL neurons labeled with AAV9-hSyn-mCherry and AAV9-hSyn-mCherry-shcpeb1 viruses from Cx3cr1-Cre mice with the neuron-specific, microglia-specific, and combined Cpeb1 deletion. Representative confocal images of spines are shown in (K), with quantitative analysis of spine density (L) and mushroom spine percentage (M) (*P < 0.05, **P < 0.01, ***P < 0.001, n = 6 neurons from four mice per group). (N to Q) Whole-cell patch-clamp recordings of IL neurons from Cx3cr1-Cre mice with neuron-specific, microglia-specific, and combined Cpeb1 deletion. Representative mEPSC traces are shown in (N), with quantification of mEPSC amplitude (O and P) and frequency (Q) (*P < 0.05, ***P < 0.001, n = 7 to 10 neurons from four mice per group).

Fig. 5.. Neuronal Cpeb1-Hp1bp3 signaling regulates fear extinction by affecting synaptic function.

(A) Confocal image for Hp1bp3- infected virus in the IL from C57BL/6 J mice (right), and image of Hp1bp3-infected neurons stained with anti-Hp1bp3 (red) is shown (left). White dashed lines mark virus-infected neurons, and yellow dashed lines mark noninfected ones. (B) Freezing percentages during fear memory acquisition and the first and second sessions of extinction learning were compared between control and OE-Hp1bp3 groups (**P < 0.01, n = 6 to 7 per group). (C) Freezing percentages across 14 trials in the second extinction session of control and OE-Hp1bp3 groups (n = 6 to 7 per group). (D) Whole-cell patch-clamp recordings of the IL neurons from OE-Hp1bp3 and control mice. Representative mEPSC traces (top), with quantified amplitude (left) and frequency (right) (**P < 0.01, n = 6 to 7 neurons from four mice per group). (E and F) 2D reconstructions of IL neurons from OE-Hp1bp3 and control mice (E), with Sholl analysis of dendritic complexity (F) (n = 5 neurons from four mice per group). (G and H) Dendritic spine density and mushroom spine percentage in IL neurons from the OE-Hp1bp3 and control groups. Representative images (G) and quantifications (H) are shown (***P < 0.001, n = 10 neurons from four mice per group). (I) Freezing percentages during fear memory acquisition and the first and second sessions of extinction learning were analyzed in Cpeb1 WT+Scr, Cpeb1 cKO+Scr, Cpeb1 WT+Sh-Hp1bp3, and Cpeb1 cKO+Sh-Hp1bp3 groups (*P < 0.05, **P < 0.01, ***P < 0.001, n = 5 to 7 per group). (J) Freezing percentages across 14 trials in the second extinction session in these groups (n = 5 to 7 per group). (K to M) Confocal images of dendritic spine in IL neurons from same groups (K), with quantified dendritic spine density (L) and mushroom spine percentage (M) (*P < 0.05, **P < 0.01, ***P < 0.001, n = 6 neurons from four mice per group). (N to Q) Representative mEPSC traces from these groups (N), with quantified amplitude (O and P) and frequency (Q) (*P < 0.05, **P < 0.01, n = 7 neurons from four mice per group).

Fig. 6.. Neuronal Hp1bp3 signal regulates synaptic function via a complicated miRNA network.

(A) Volcano plot showing the differentially regulated miRNAs in the mPFC from control and OE-Hp1bp3 groups with fold change >1.2 (pink dots) or <1.2 (cyan) and P value <0.05. The gray dots indicate the unchanged miRNAs (n = 3 per group). (B) Heatmap showing differential expression of miRNAs in the mPFC between control and OE-Hp1bp3 mice. (C) Quantitative PCR (qPCR) of 19 up-regulated miRNAs from control and OE-Hp1bp3 mice was performed in the mPFC between Impaired-ext (Imp) and Normal-ext (Nor) groups (*P < 0.05, **P < 0.01, n = 4). (D and E) qPCR was performed to measure the relative expression of pre-miRNA (D) and pri-miRNA (E) between control and OE-Hp1bp3 mice (*P < 0.05, **P < 0.01, n = 4). (F) An integrated analysis for the predicted targets of these 10 miRNAs. (G and H) The protein levels of Hp1bp3, Gnal, Rem2, Gria1, and Gria2 in the mPFC homogenates from control and OE-Hp1bp3 mice were examined by Western blotting. Representative images are shown in (G), and the quantitative analysis is shown in (H) (*P < 0.05, **P < 0.01, ***P < 0.001, n = 4). GAPDH, glyceraldehyde phosphate dehydrogenase. (I) The protein levels of Gnal, Rem2, Gria1, and Gria2 in the mPFC homogenates from Impaired-ext and Normal-ext mice were examined by Western blotting. The representative images are shown in right and the quantitative analysis is shown in left (*P < 0.05, **P < 0.01, n = 4). (J) The protein levels of Gnal, Rem2, Gria1, and Gria2 in the mPFC homogenates from Cpeb1 WT+Scr, Cpeb1 cKO+Scr, Cpeb1 WT+Sh-Hp1bp3, and Cpeb1 cKO+Sh-Hp1bp3 mice were examined by Western blotting. The representative images are shown in right and the quantitative analysis is shown in left. (*P < 0.05, **P < 0.01, ***P < 0.001, n = 4).

Fig. 7.. Microglial Cpeb1-Cx3cr1 signaling regulates fear extinction and induces an aged-like microglial phenotype.

(A) Representative confocal image (right) for virus injection in the IL of Cx3cr1-Cre mice. Coronal slices with infected (EGFP) microglia were stained with anti-Cx3cr1 (red). Arrowheads indicate virus-infected microglia, and arrows show noninfected microglia. (B) Average freezing percentage during fear memory acquisition and the first and second sessions of fear extinction in DIO-Scr and DIO-ShCx3cr1 groups (*P < 0.05, **P < 0.01, n = 4 per group). (C) Freezing percentages across 14 trials in the second extinction session of DIO-Scr and DIO-ShCx3cr1 groups (n = 4 per group). (D) Representative confocal image for virus injection into the IL from Cx3cr1-Cre mice. (E) Average freezing percentage during fear memory acquisition and the first and second sessions of fear extinction learning in Scr+Con, Scr+OE-Cx3cr1, Sh-Cpeb1+Con, and Sh-Cpeb1+OE-Cx3cr1 groups (*P < 0.05, **P < 0.01, ***P < 0.001, n = 5 to 7 per group). (F) Freezing percentages across 14 trials in the second extinction session for Scr+Con, Scr+OE-Cx3cr1, Sh-Cpeb1+Con, and Sh-Cpeb1+OE-Cx3cr1 groups (n = 5 to 7 per group). (G) Scatter plot showing linear regression of log₂FC of overlapping genes between ADEM gene sets and Cx3cr1-deficient microglia gene sets, with a positive correlation. Pink shading shows 95% confidence interval, and Pearson’s correlation coefficient and P value are shown. (H to J) Microglia from the Impaired-ext group display reduced branching and enlarged cell bodies. Confocal images are shown in (H), and Sholl analysis (I) and microglial cell size (J) were quantified by ImageJ (*P < 0.05, n = 6 cells from four mice per group). (K to M) Immunofluorescence on IL sections from Impaired-ext and Normal-ext groups using anti-Xaf1 (Red) and anti-Iba1 (Green) antibodies. Representative confocal images are shown in (K). Xaf1/Iba1 colocalized pixels (L) and normalized intensity of Xaf1 (M) were calculated by ImageJ (**P < 0.01, ***P < 0.001, n = 6 cells from four mice per group).

Fig. 8.. Disturbance of microglial Cpeb1-Cx3cr1 signaling leads to impaired phagocytic ability in aged microglia and the accumulation of perisynaptic ECM.

(A) Functional enrichment analysis of differentially expressed genes between Cx3cr1-deficient microglia and WT microglia (fold change >1.2, P < 0.05). (B) GSEA of phagocytosis-related genes in Cx3cr1-deficient microglia compared with WT microglia. (C) BV2 cells with knocked-down Cx3cr1 showed reduced phagocytic ability of fluorescent latex beads. Representative confocal images are shown in the top panel of (C), and engulfed beads per microglia were calculated in the bottom panel of (C) (**P < 0.01, n = 7 microglia cells from two replicates per group). (D and E) Aggrecan protein within microglial lysosomes from DIO-Scr and DIO-ShCx3cr1 group. Coronal slices from virus infected (EGFP) microglia cells were stained with anti-Aggrecan (cyan) and CD68 (red) antibody. The representative confocal images and 3D reconstruction are shown in (D). The number of engulfed Aggrecan in microglia and the normalized intensity of Aggrecan were calculated in (E) (**P < 0.01, ***P < 0.001, n = 5 microglia cells from four mice per group). (F) Aggrecan colocalization with dendritic spines in the IL labeled with AAV9-hSyn-mCherry and AAV9-hSyn-mCherry-shcpeb1 virus from M-shCon+N-shCon, M-shCon+N-shCpeb1, M-shCpeb1+N-shCon and M-shCpeb1+N-shCpeb1 mice. Representative confocal images of Aggrecan colocalization with dendritic spine are showed in the left of (F), and the Aggrecan positive dendritic spine was calculated in the right of (F) (***P < 0.001, n = 4 neurons from three mice per group).

Fig. 9.. A working model illustrating Cpeb1 bidirectionally drives neuron- and microglia-specific translational programs to promote fear extinction.

Our findings demonstrate that Cpeb1 in the IL regulates fear extinction by controlling Hp1bp3 translation activation in neurons and Cx3cr1 translation suppression in microglia. Neuronal Hp1bp3 translation activation increases a network of miRNAs affecting postsynaptic proteins, hindering spine formation and plasticity. Microglial Cx3cr1 translation suppression results in an aged phenotype impairing their ability to engulf ECM Aggrecan, promoting perisynaptic ECM deposition, and reducing dendritic spine density. Together, Cpeb1 bidirectionally drives neuron- and microglia-specific translational program to promote fear extinction.