Identification of FMRP target mRNAs in embryonic mouse cortex
To explore the target mRNAs of FMRP during corticogenesis, we performed RIP-seq analyses using cortical samples isolated from the WT mice at E14.5 (Fig. 2A). In total, we found 2288 candidate FMRP target mRNAs that were significantly expressed (measured by fragment per kilobase of transcript per RNA-seq read mapped, FPKM) in the FMRP-IP compared to IgG-IP or the negative control (FMRP-IP FPKM > IgG-IP FPKM). Next, we selected a stringent set of 947 mRNAs from 865 FMRP target genes based on gene expression values, fold changes in the logarithmic scale with base 2 greater than 1 (log2FC > 1), and FPKM greater than 10 against the FMRP-IP. (Fig. 2B; Table S1). The set of 865 FMRP target genes showed higher gene expression than the negative control, and therefore, could be validated as targets of FMRP in the developing neocortex.
To estimate the functions of the FMRP target mRNAs, we performed gene ontology (GO) analyses using the Visual Annotation Display (VLAD) – gene analysis and visualization analysis tool of the Mouse Genome Informatics (MGI) [25]. The top significant GO terms included biological processes related to early brain development, such as “Nervous system development,” “Generation of neurons,” and “Neurogenesis.” (Fig. 2C). This is quite reasonable because the identified FMRP target mRNAs were collected from developing cortices where massive neurogenesis is occurring (Fig. 1, Table S2).
Overlap among FMRP targets, neurogenesis, ID and ASD-associated genes
To obtain more insight for the significance of FMRP target candidate genes, we focused on the three criteria, i.e., neurogenesis, ID, and ASD since FXS patients often show ID and ASD symptoms [8, 9]. We first compared the identified 865 FMRP target genes with 1791 neurogenesis genes from MGI [25]. There was a highly significant overlap of 156 genes between the two groups, including those mainly assigned to GOs related to “Neuronal development”, “Generation of neurons,” “Neuron differentiation”, and “Cell morphogenesis involved in differentiation” (Fig. 3A). The results thus indicate that several targets of FMRP are important for neurogenesis during early brain development. We also found genes for “Axonogenesis” and “Neuron projection development”, i.e., the events after neuronal differentiation, as GOs for FMRP target genes, which may suggest the importance of FMRP in the establishment of neuronal networks.
We then examined the association between the identified FMRP target genes with 1088 ID genes based on Online Mendelian Inheritance in Man (OMIM) [26], which resulted in 126 genes (Fig. 3B), as expected, because ID is a core feature of FXS [27]. These overlapped genes included not only GOs such as “Brain development” and “Central nervous system development”, but also “Chromosome organization” and “Histone modification” unexpectedly. We also found 118 FMRP target genes that significantly overlapped with the 1025 ASD-associated genes using the public database Simons Foundation Autism Research Initiative (SFARI) [28] (Fig. 3C); again there came up with GOs such as “Histone modification” and “Chromatin organization”, as well as “Brain development”. As we expected, these GOs include several syndromic ASD-associated genes such as paired box 6 (PAX6) [29, 30],lysine acetyltransferase 6a (KAT6A) [31], mammalian target of rapamycin (mTOR) [32], Abelson’s helper integration 1 (AHI1) [33] and ubiquitin-specific peptidase 9 X-linked (USP9X) [34]. Overall, the overlap between FMRP target genes linked to ID and ASD could provide a correlation between loss of function of FMRP and the development of both ID and ASD.
Finally, we identified 17 genes as the “core” genes shared with neurogenesis, ID, and ASD gene sets (Fig. 3D, Table 1). The “core” gene set contained not only major transcription regulators such as Pax6, Myt1l, and Tcf4 but also components of Ras/MAPK (Nf1) [35], Wnt/β-catenin (Ahi1, Ctnna2, and Ctnnb1) [33, 36], and mTOR (mTOR, Ep300, Itpr1and Synj1) [36–40] signaling pathways (Fig. 3E). As mentioned above, the “core” gene set also included factors of the chromatin-remodeling complex [41], such as Nipbl, Smarcc2, and Smarca4. Besides, Usp9X has been thought to be involved in developmental processes through Wnt/β-catenin and mTOR pathways [42]. These common pathways can cause shared symptoms among FXS, ID, and ASD.
Table 1. 17 FMRP target genes associated with neurogenesis, ID and ASD. |
|
Gene symbol |
Gene name |
|
Ahi1 |
Abelson Helper Integration Site 1 |
|
Ank3 |
Ankyrin 3 |
|
Ctnna2 |
Catenin Alpha 2 |
|
Ctnnb1 |
Catenin Beta 1 |
|
Ep300 |
E1A Binding Protein P300 |
|
Itpr1 |
Inositol 1,4,5-Trisphosphate Receptor Type 1 |
|
Mtor |
Mechanistic Target Of Rapamycin Kinase |
|
Myt1l |
Myelin Transcription Factor 1 Like |
|
Nf1 |
Neurofibromin 1 |
|
Nipbl |
NIPBL Cohesin Loading Factor |
|
Pax6 |
Paired Box 6 |
|
Smarca4 |
SWI/SNF Related, Matrix Associated, Actin Dependent Regulator Of Chromatin, Subfamily A, Member 4 |
|
Smarcc2 |
SWI/SNF Related, Matrix Associated, Actin Dependent Regulator Of Chromatin Subfamily C Member 2 |
|
Synj1 |
Synaptojanin 1 |
|
Tbc1d23 |
TBC1 Domain Family Member 23 |
|
Tcf4 |
Transcription Factor 4 |
|
Usp9x |
Ubiquitin Specific Peptidase 9 X-Linked |
|
Expression of the “core” gene set in the developing cortex
To confirm the FMRP interaction with the mRNAs of the 17 “core” genes, we performed RIP-qPCR. All mRNAs were significantly enriched in the FMRP-IP, suggesting that these mRNAs included in 17 “core” gene set are targeted by FMRP (Fig. 4A). We further examined the mRNA amount of the genes in cortical primordial samples from E15.5 WT and Fmr1 knockout (KO) male mice. We found a significant increase of Nf1 mRNA and a significant decrease of Ahi1 mRNA in the Fmr1 KO mouse neocortex, while other genes showed no significant difference (Fig. 4B). These findings suggest that FMRP mainly functions as a post-transcriptional regulator of its target genes.
Among the “core” gene set, we highlighted three genes, Nf1, Ctnnb1, and Mtor, because these are involved in Ras/MAPK, Wnt/β-catenin, and mTOR pathways, respectively [40, 43, 44]. We first assessed the protein expression of Nf1, Ctnnb1, and mTOR in the cortical primordium at E15.5 (Fig. 5A, B). The Nf1 and mTOR proteins were widely expressed throughout the cortical primordium, while Ctnnb1 was concentrated at the apical surface. There seemed to be no change in these expression patterns in Fmr1 KO mice compared to that of WT (Fig. 5A). Immunoblotting analyses also showed that Nf1, Ctnnb1, and mTOR showed normal expression levels in Fmr1 KO mice corresponding to the immunostaining (Fig. 5A-D). This could imply that the translation of these targets was unaffected in the Fmr1 KO neocortex. Although the protein level of mTOR was unchanged in the Fmr1 KO, its phosphorylated form (p-mTOR) at Ser2448 was significantly elevated by 25.4% in the lysate of the Fmr1 KO neocortex compared to that of WT (Fig. 5B, E, F). This result suggests that mTOR signaling might be enhanced in Fmr1 KO mice during corticogenesis, which is similar to the result in the adult hippocampus [40] but the first evidence in the embryonic brain.