Cell-Type-Specific Translation Profiling Reveals a Novel Strategy for Treating Fragile X Syndrome

Abstract

Excessive mRNA translation downstream of group I metabotropic glutamate receptors (mGlu_1/5) is a core pathophysiology of fragile X syndrome (FX); however, the differentially translating mRNAs that contribute to altered neural function are not known. We used translating ribosome affinity purification (TRAP) and RNA-seq to identify mistranslating mRNAs in CA1 pyramidal neurons of the FX mouse model (Fmr1^-/y) hippocampus, which exhibit exaggerated mGlu_1/5-induced long-term synaptic depression (LTD). In these neurons, we find that the Chrm4 transcript encoding muscarinic acetylcholine receptor 4 (M₄) is excessively translated, and synthesis of M₄ downstream of mGlu₅ activation is mimicked and occluded. Surprisingly, enhancement rather than inhibition of M₄ activity normalizes core phenotypes in the Fmr1^-/y, including excessive protein synthesis, exaggerated mGluR-LTD, and audiogenic seizures. These results suggest that not all excessively translated mRNAs in the Fmr1^-/y brain are detrimental, and some may be candidates for enhancement to correct pathological changes in the FX brain.

Keywords: FMR1; FMRP; LTD; TRAP; autism; fragile X; m4; mglur theory; muscarinic receptor; protein synthesis.

Figure 1

TRAP-Seq Identifies Differentially Translating mRNAs in Fmr1^−/y CA1 Pyramidal Neurons (A) Confocal images show selective expression of GFP-L10a in pyramidal neurons of the CA1 region. (B) GFP-positive (GFP+) cells in CA1-TRAP hippocampus are enriched for CA1 neuronal markers (Camk2a: GFP− 0.683 ± 0.05, GFP+ 1.200 ± 0.138, ^∗p = 0.0046, n = 12; Wfs1: GFP− 0.370 ± 0.104, GFP+ 1.781 ± 0.224, ^∗p < 0.0001, n = 9) and depleted of glial markers (Gfap: GFP− 1.784 ± 0.650, GFP+ 0.054 ± 0.022, ^∗p = 0.0218, n = 12) compared to all cells. (C) Schematic representation of TRAP shows isolation of translating ribosomes (IP) from Input using anti-GFP coated beads. (D) Differentially expressed genes in CA1-TRAP versus Bergmann glia (BG)-specific TRAP are enriched in CA1 neuronal markers according to the Allen Brain Atlas enrichment algorithm. (E) Camk2a is significantly increased in Fmr1^−/y versus WT CA1-TRAP IP (WT = 1.00 ± 0.037, KO = 1.26 ± 0.054, ^∗p = 0.0004, n = 14). Total Camk2a is equivalent in Fmr1^−/y and WT FACs-isolated CA1 pyramidal neurons (WT = 1.00 ± 0.059, KO = 0.855 ± 0.047, p = 0.0734, n = 9). (F) TRAP-seq analysis reveals differential expression of 121 genes in the IP fraction and 3 genes in the Input fraction (FDR < 0.1). n = number of littermate pairs. Error bars indicate SEM.

Figure 2

Differentially Translated mRNAs in Fmr1^−/y CA1 Include FMRP Targets and mAChR Transcripts (A) Differential expression analysis shows gene changes in WT versus Fmr1^−/y Input fraction, with FMRP targets highlighted in blue. (B) FMRP targets were compared to differentially expressed (total) genes with the same level of abundance (normalized count between 10^2.5 and 10^4.25). (C and D) A cumulative distribution of FMRP targets shows a significant shift toward downregulation when compared to the distribution of total differentially expressed genes with the same level of abundance in both Input (C) and CA1-TRAP (D) fractions (K-S test, p = 9.23 × 10⁻¹⁴, p = 4.86 × 10⁻¹³). (E) Pfam analysis of enriched protein families reveals that six out of eight pClans enriched in the differentially expressed (DE) Fmr1^−/y CA1-TRAP gene list overlap with pClans enriched in the CA1-adjusted FMRP target list. (F) Heatmap shows the fold change of differentially expressed genes in each pair of Fmr1^−/y versus WT (IP and Input fractions). (G) GO analysis identifies G-protein-coupled acetylcholine receptor signaling pathway as the most enriched functional category in the upregulated Fmr1^−/y CA1-TRAP gene list. (H) Drug gene interaction database reveals that Chrm4 is the most amenable target pharmacologically. Upregulated genes are highlighted in green, and downregulated genes are highlighted in red. n = number of littermate pairs. Error bars indicate SEM.

Figure 3

M₁ and M₄ Are Excessively Synthesized and Overexpressed in Fmr1^−/y CA1 Pyramidal Neurons (A) Chrm4 and Chrm1, but not Chrm3, are enriched in the Fmr1^−/y CA1-TRAP IP (Chrm4: WT = 1.00 ± 0.124, KO = 1.72 ± 0.195, ^∗p = 0.0044, n = 14; Chrm1: WT = 1.00 ± 0.062, KO = 1.47 ± 0.079, ^∗p < 0.0001, n = 14; Chrm3: WT = 1.00 ± 0.085, KO = 1.184 ± 0.073, p = 0.192, n = 10). (B) Total mRNA levels of Chrm4, Chrm1 and Chrm3 are unchanged in FACS-isolated Fmr1^−/y CA1 pyramidal neurons (Chrm4: WT = 1.00 ± 0.209, KO = 0.98 ± 0.154, p = 0.926, n = 9; Chrm1: WT = 1.00 ± 0.150, KO = 0.87 ± 0.185, p = 0.602, n = 11; Chrm3: WT = 1.00 ± 0.232, KO = 1.14 ± 0.230, p = 0.668, n = 8). (C) Immunoblotting shows overexpression of M₄ protein in hippocampal slice homogenates (WT = 100% ± 5.7%, KO = 121% ± 5.7%, ^∗p = 0.0186, n = 12) and synaptoneurosomes (WT = 100% ± 7.10%, KO = 121% ± 6.89%, ^∗p = 0.0203, n = 9). (D and E) Hippocampal slice homogenates show no difference in M₁ (D) (WT = 100% ± 2.5%, KO = 105% ± 2.5%, p = 0.129, n = 12) or M₃ (E) (WT = 100% ± 9.5%, KO = 98% ± 12.9%, p = 0.94, n = 9) expression. (F) Schematic shows steps for FACS immunostaining. (G) FACS-immunostaining reveals significant increase in the expression of M₄ and M₁ (M₄: WT = 0.925 ± 0.045, KO = 1.075 ± 0.045, ^∗p = 0.0389, n = 6; M₁: WT = 0.920 ± 0.002, KO = 1.080 ± 0.046, ^∗p = 0.0092, n = 6), but not M₃ (WT = 0.962 ± 0.07, KO = 1.038 ± 0.1011 p = 0.547, n = 7), in Fmr1^−/y CA1 pyramidal neurons. n = number of littermate pairs. Error bars indicate SEM.

Figure 4

M₄ Synthesis Downstream of mGlu₅ Is Mimicked and Occluded in the Fmr1^−/y Hippocampus (A) Time course for DHPG stimulation experiments. (B) Analysis of transcripts encoding hippocampal mAChR subunits reveals a striking upregulation of Chrm4 mRNA in CA1-TRAP IP after mGlu_1/5 stimulation (Veh = 1.00 ± 0.12, DHPG = 1.72 ± 0.23, ^∗p = 0.0047, n = 15), with no changes seen in Chrm1 or Chrm3 (Chrm1: Veh = 1.00 ± 0.07, DHPG = 1.03 ± 0.06, p = 0.75, n = 16; Chrm3: Veh = 1.00 ± 0.12, DHPG = 0.82 ± 0.09, p = 0.209, n = 14). (C) DHPG stimulation of WT slices shows dramatic increase in Chrm4 mRNA in the TRAP IP fraction. In Fmr1^−/y slices, Chrm4 mRNA is already significantly elevated in the TRAP IP and does not increase further with mGlu_1/5 activation (WT vehicle = 1.00 ± 0.24, WT DHPG = 2.13 ± 0.41, KO vehicle = 2.28 ± 0.43, KO DHPG = 2.34 ± 0.44, ANOVA genotype ^∗p = 0.03, treatment ^∗p = 0.048, WT versus KO veh ^∗p = 0.03, KO veh versus DHPG p > 0.999, n = 7). Chrm1 mRNA is significantly elevated in the Fmr1^−/y CA1-TRAP IP, but DHPG does not increase Chrm1 in either WT or Fmr1^−/y CA1-TRAP IPs (WT vehicle = 1.00 ± 0.09, WT DHPG = 1.13 ± 0.08, KO vehicle = 1.50 ± 0.13, KO DHPG = 1.47 ± 0.16, ANOVA genotype ^∗p = 0.005, treatment p = 0.753, WT versus KO veh ^∗p = 0.04, WT veh versus DHPG p = 0.944, n = 7). Chrm3 mRNA is neither increased in the Fmr1^−/y CA1-TRAP IP nor elevated with DHPG (WT veh = 1.00 ± 0.14, WT DHPG, = 1.029 ± 0.18, KO veh = 0.946 ± 0.19, KO DHPG = 1.09 ± 0.19, ANOVA genotype p = 0.97, treatment p = 0.55, n = 7). (D) Immunoblotting shows a robust increase in M₄ expression in WT slices after 5 min of DHPG stimulation, which is maintained at 30 min and 60 min post-stimulation. In contrast, the elevated expression of M₄ in Fmr1^−/y slices is not further increased with DHPG stimulation (WT vehicle = 100% ± 6.63%, WT DHPG 5 min = 159.59% ± 9.37%, WT DHPG 30 min = 131.09% ± 13.23%, WT DHPG 60 min = 141.66% ± 12.08%, KO vehicle = 132.70% ± 7.31%, KO DHPG 5 min = 146.95% ± 7.94% KO DHPG 30 min = 131.04% ± 13.01%, KO DHPG 60 min = 155.75% ± 10.28%, ANOVA treatment × genotype ^∗p = 0.037, n = 7). (E) Time course for MTEP slice experiments. (F) Incubation with 10 μM MTEP reduces Chrm4 in the Fmr1^−/y CA1-TRAP to WT levels (WT vehicle = 1.00 ± 0.112, WT MTEP = 1.48 ± 0.230, KO vehicle = 2.63 ± 0.352, KO MTEP = 1.63 ± 0.161, ANOVA genotype ^∗p = 0.0014, treatment p = 0.1923, genotype × treatment ^∗p = 0.0119, WT veh versus KO veh ^∗p = 0.0024, WT veh versus WT MTEP p = 0.306, KO veh versus KO MTEP ^∗p = 0.0289, WT MTEP versus KO MTEP p = 0.880, n = 8). (G) Immunoblotting shows a significant reduction in M₄ expression in MTEP-treated Fmr1^−/y slices (WT vehicle = 100% ± 9.7%, WT MTEP = 97% ± 5.4%, KO vehicle = 181% ± 32.6%, KO MTEP = 103% ± 14.1%, ANOVA genotype ^∗p = 0.034, WT veh versus KO veh ^∗p = 0.0181, KO veh versus KO MTEP ^∗p = 0.0145, n = 6). n = number of littermate pairs. Error bars indicate SEM.

Figure 5

Enhancement of M₄ Normalizes Excessive Protein Synthesis in the Fmr1^−/y Hippocampus (A) Time course for metabolic labeling experiments. (B) Treatment with the M₄ antagonist PD 102807 (0.5 μM or 1 μM) significantly increases protein synthesis in both WT and Fmr1^−/y slices (WT vehicle = 100% ± 1.66%, WT PD 0.5 μM = 111.35% ± 7.16%, WT PD 1 μM = 136% ± 7.17%, KO vehicle = 114.59% ± 4.77%, KO PD 0.5 μM = 120.29% ± 5.02%, KO PD 1 μM = 129.05% ± 5.40%, ANOVA treatment ^∗p < 0.0001, WT veh versus KO veh ^∗p = 0.0379, WT veh versus WT PD 1 μM ^∗p = 0.0009, KO veh versus KO PD 1 μM ^∗p = 0.0065, n = 8). Example autoradiograph of slice homogenates shows upregulation of ³⁵S-labeled proteins with M₄ antagonist. Total protein stain of the same blot is shown for comparison. (C) Enhancement of M₄ with VU0152100 (5 μM) results in selective reduction of protein synthesis in the Fmr1^−/y hippocampus, but no change in WT (WT veh = 100% ± 3.12%, KO veh = 114.2% ± 3.47%, WT VU = 101.7% ± 2.33%, KO VU = 101.3% ± 3.05%, ANOVA genotype × treatment ^∗p = 0.0456, WT veh versus VU p = 0.6580, KO veh versus VU ^∗p = 0.013, n = 16). Example autoradiograph shows a reduction of ³⁵S-labeled proteins in Fmr1^−/y slices upon incubation with M₄ PAM. Total protein stain of the same blot is shown for comparison. n = number of littermate pairs. Error bars indicate SEM.

Figure 6

M₄ PAM Corrects Exaggerated mGluR-LTD in the Fmr1^−/y Mouse (A) Measurement of mGluR-LTD in hippocampal CA1 shows a significant elevation in vehicle-treated Fmr1^−/y versus WT (WT = 84.7% ± 3.4%, n = 16, KO = 71.2% ± 2.47%, n = 15, ^∗p = 0.0028). (B) Exaggerated LTD in the Fmr1^−/y is significantly normalized with 5 μM VU0152100 (KO PAM = 88.7% ± 2.76%, n = 13, ^∗p = 0.0003). VU0152100 treatment has no effect on WT LTD (WT PAM 87.6% ± 3.13%, n = 11, p > 0.999). (C) Comparison of all four groups (re-plotted from A and B). (D) Quantification of the last 10 min of recording shows a significant rescue of the LTD phenotype in the Fmr1^−/y with VU0152100 (ANOVA genotype × treatment ^∗p = 0.0191). n = number of animals. Error bars indicate SEM.

Figure 7

M₄ PAM Corrects the Exaggerated AGS Phenotype in the Fmr1^−/y Mouse (A) Time course for AGS experiments. (B) Injection of VU0152100 significantly reduces the incidence of AGS in Fmr1^−/y mice versus vehicle (Fisher’s exact test ^∗p < 0.0001; KO veh 15/21, KO VU 2/19, WT veh 1/14, WT PAM 0/14). (C) VU0152100 reduces severity of AGS in the Fmr1^−/y (KO veh wild running 4/21, clonic 11/21, tonic 3/21; KO VU wild running 1/19, clonic 1/19).

Figure 8

Potential Model for Correction of FX by M₄ PAM Our results suggest a model whereby M₄ is synthesized downstream of mGlu₅ in order to negatively regulate protein synthesis and LTD, similar to FMRP. In FX, the absence of FMRP leads to the excessive synthesis of M₄; however, this is unable to completely compensate for FMRP loss. By enhancing M₄ with VU0152100, protein synthesis, LTD, and other pathological changes are normalized.